caD1098'/j -
Otinadian Nuclear Society
INIS-mf—12827
Societe
C an'adienne
9th Annual Conference 9e Cong res annuel 1988 1988 13-15 June 1988 Winnipeg, Manitoba, Cana.da
13-15 juin 1988 Winnipeg, Manitoba. Canada
Proceedings
Comptes rendus
ISSN 0227-1907
Canadian Nuclear Society
Societe Nucleaire Canadienne
9th Annual Conference 9e Congres annuel 1988 1988 13-15 June 1988 Winnipeg, Manitoba, Canada
13-15 juin 1988 Winnipeg, Manitoba, Canada
Proceedings
Comptes rendus
Foreword The papers for these proceedings were prepared on standard forms supplied by the Canadian Nuclear Society and are generally published as submitted by the authors. Responsibility for the content of each paper rests solely with the author. These proceedings are copyrighted by the Canadian Nuclear Society. Requests for further information concerning these proceedings, permission to reprint any part of these proceedings, or orders for copies of these proceedings should be addressed to: Canadian Nuclear Society 111 Elizabeth St., 11th Floor Toronto, Ontario, Canada M5G 1P7 Printed by Heritage Publications, Peterborough, Canada
Acknowledgement The success of the 9th Annual Conference is due, in large measure, to the assistance of various CNS members who participated in reviewing papers; a job made particularly difficult because of the large number of submissions received. The secretarial assistance of Mrs. L.M. McCoy and Mrs. P.R. Drew is also gratefully acknowledged. Heiki Tamm Conference Chairman
Table of Contents Page SESSION 1 OPERATIONAL ENHANCEMENTS - I Chairman: R.E. Lewis, Ontario Hydro 1.
2.
3.
4.
5.
6.
ONTARIO HYDRO NUCLEAR GENERATION DIVISION INFORMATION SYSTEMS AT DARLINGTON NGS I.S. Hey*, Ontario Hydro
. ..
OPERATIONS DECISION SUPPORT SYSTEMS FOR CANDU NUCLEAR PLANT OPERATIONS H.E. Sills, J.W.D. Anderson, AECL - CRNL
...
ROUTINE TREND ANALYSIS PROGRAM FOR PLANT PROCESS PARAMETERS J.G. Comeau, New Brunswick. Electric Power Commission
... 13
ONTARIO HYDRO'S LOAD FOLLOWING REQUIREMENTS, ISSUES, EXPERIENCE AND STRATEGY, A.M. Lopez, R. Neuman, C D . Jin, J. Chada, A. De Santis, Ontario Hydro
...
FUELEM: A MICROCOMPUTER PROGRAM TO AUTOMATICALLY SELECT CHANNELS FOR REFUELLING, B. Rouben, D.A. Jenkins, C.R. Calabrese AECL - CANDU Ops
... 25
AN OPPORTUNITY FOR INEXPENSIVE NUCLEAR POWER, J.L. Bagshaw, C.J. Bromley, Ontario Hydro
... 32
3
8
20
SESSION 2 SMALL REACTORS: DESIGN Chairman: G.E. Gillespie, AECL - WNRE 1.
2.
3.
4.
5. 6.
THE NUCLEAR BATTERY: A SOLID-STATE, PASSIVELY COOLED REACTOR FOR THE GENERATION OF ELECTRICITY AND/OR HIGH-GRADE STEAM HEAT, K.S. Kozier, H.E. Rosinger, AECL - WNRE
... 41
CONCEPTUAL DESIGN OF A SMALL NUCLEAR REACTOR FOR GENERATING ELECTRICITY, J.F. Lafortune, D.A. Meneley, University of New Brunswick
__
48
NEUTRONIC DESIGN OF THE AMPS REACTOR CORE, R.E. Stone, A.F. Oliva, ECS - Power Systems Inc.
55
NEUTRONICS DESIGN OF A SMALL, SOLID-STATE, PASSIVELY COOLED REACTOR, THE NUCLEAR BATTERY, J.V. Donnelly, K.S. Kozier, G.R. Penner, AECL - WNRE
62
STARTUP OF THE SLOWPOKE DEMONSTRATION REACTOR AND LOW POWER TESTS, J.D. Irish, B.M. Townes, C M . Tseng, AECL - CRNL
67
CAREM: A SMALL ELECTRICITY PRODUCING REACTOR, J.P. Ordonez, J.J. Gil Gerbino, INVAP S.E., Bariloche, Argentina
... 74
(v)
SESSION 3 ACCIDENT BEHAVIOUR IN FUEL CHANNELS Chairman: R.A. Brown, Ontario Hydro 1.
2.
REFLOODING PHENOMENA DURING ECCS OPERATION, H. Mochizuki, Y. Hayamizu, O-Arai Engineering Center, Japan
... 83
THERMOSS-II: A MODEL FOR THERMOHYDRAOLICS OF CANDU FUEL CHANNEL WITH SUBCOOLED STAGNANT INITIAL CONDITIONS, P. Gulshani, AECL - CANDU Ops
... 89
3.
SENSITIVITY STUDIES OF CALANDRIA TUBE INTEGRITY IN THE EVENT OF PRESSURE TUBE FAILURE, P.S. Kundurpi, A.P. Muzumdar, Ontario Hydro ... 94
4.
FULL SCALE CALANDRIA TUBE BURST TESTS AT FAST PRESSURIZATION RATES, G.I. Hadaller, Stern Laboratories Inc., A.P. Muzumdar, Ontario Hydro ... 103
5.
AN EXPERIMENTAL AND ANALYTICAL APPROACH TO DETERMINE BEARING-PAD TO PRESSURE-TUBE HEAT TRANSFER, J.W. DeVaal, M.H. Schankula, V.D. Kroeger, AECL - WNRE, D.B. Reeves, A.P. Muzumdar, S.D. Sheridan, Ontario Hydro
... 110
THE EXPERIMENTAL MEASUREMENT OF CIRCUMFERENTIAL TEMPERATURE DISTRIBUTIONS DEVELOPED ON PRESSURE TUBES UNDER STRATIFIED TWO-PHASE FLOW CONDITIONS, P.S. Yuen, C.B. So, R.G. Moyer, D.G. Litke, AECL - WNRE
... 120
6.
SESSION 4 FUEL STORAGE AND WASTE MANAGEMENT Chairman: K. Nut tall, AECL - WNRE 1. 2.
3.
4.
5.
CANADIAN EXPERIENCE WITH THE DRY STORAGE OF USED CANDU FUEL, K.M. Wasywich, AECL - WNRE, C.R. Frost, Ontario Hydro
... 129
THE DOUGLAS POINT DRY SPENT FUEL STORAGE FACILITY, Robert R. Beaudoin, AECL - Montreal, Donald W. Patterson, AECL - WNRE
... 135
EXPERIMENTAL VALIDATION OF MODELS FOR RADIATION DOSE RATE FROM CANDU SPENT FUEL, K.M. Aydogdu, C.R. Boss, AECL - CANDU Ops
... 144
CLEANUP AROUND AN OLD WASTE SITE - A SUCCESS STORY, G. Vandergaast, D. Moffett, Eldorado Resources Limited, B.E. Lawrence, MacLarentech Inc.
... 151
PROCESSING OF LLRW ARISING FROM AECL NUCLEAR RESEARCH CENTRES, L.P. Buckley, V.T. Le, N.V. Beamer, AECL - CRNL, W.P. Brown, R.A. Helbrecht, AECL - WNRE
...
157
6.
CHEMISTRY RESEARCH FOR THE CANADIAN NUCLEAR FUEL WASTE MANAGEMENT PROGRAM, A.C. Vikis, F. Garisto, R.J. Lemire, J. Paquette, N.H. Sagert, P.P.S. Saluja, S. Sunder, P. Taylor, AECL - WNRE ... 164
7.
RADIONUCLIDE MIGRATION THROUGH FRACTURED GRANITE. LABORATORY STUDIES, D.M. Grondin, T.T. Vandergraaf, D.J. Drew, AECL - WNRE
(vi)
... 173
SESSION 5 REACTOR COMMISSIONING/DECOMMISSIONING Chairman: K.H. Talbot, Ontario Hydro 1.
2.
3.
4.
5.
6.
ONTARIO HYDRO NUCLEAR GENERATION DIVISION COMMISSIONING WORK AT DARLINGTON NGS A - IMPACT OF EXTENSIVE COMPUTER APPLICATIONS, D.R. McQuade, Ontario Hydro
... 183
DECOMMISSIONING OF NPD GENERATING STATION, R.E. Lewis, Ontario Hydro
... 194
DECOMMISSIONING OF NPD FROM AN AECL PERSPECTIVE, P. Pattantyus, Ontario Hydro
... 197
DECOMMISSIONING COST ESTIMATES A RATIONAL APPROACH, EMPLOYING A VALIDATED COMPUTER CODE, Joel Liederman, AECL - Montreal
. . 202
MINIMIZING OPERATING ERROR DURING A MAJOR UPGRADING PROGRAM AT A MULTI-UNIT NUCLEAR PLANT, G.A. Fowles, Ontario Hydro
... 210
ANNULUS GAS SYSTEM RESPONSE TESTS FOR PICKERING NGS A, UNIT 1 COMMISSIONING, J.M. Kenchington, P.J. Ellis, D.G. Meranda, Ontario Hydro
... 213
SESSION 6 NUCLEAR SAFETY EXPERIMENTS AND MODELLING Chairman: J.C. Luxat, Ontario Hydro 1.
2.
3.
4.
5.
6.
A MODEL FOR DROPLET SIZE DISTRIBUTION IN FLASHING JETS, Minoo Razzaghi, AECL - WNRE
... 223
REFILL STUDY OF A CANDU-TYPE HEADER/FEEDER SYSTEM UNDER NEAR-ZERO HEADER-TO-HEADER PRESSURE DROP, J.E. Kowalski, B.N. Hanna, AECL - WNRE
... 229
MIXING IN A VESSEL-PIPE ASSEMBLY, A.H.T. Lam, M. Ungurian, K.N. Tennankore, AECL - WNRE (Not Available at Time of Printing)
... 237
COMBUSTION BEHAVIOUR IN THE MODERATOR COVER GAS, G.W. Koroll, R.K. Kumar, C.K. Chan, AECL - WNRE
... 238
AEROSOL MATERIAL RELEASES FROM A ZIRCALOY-4 CLAD U0 2 PELLET AT TEMPERATURES UP TO 2000"C IN A FLOWING ARGON ATMOSPHERE, S.R. Mulpuru, F.B. Banks, M.D. Pellow, AECL - WNRE
... 245
EXPERIMENTAL MODELLING OF FLOW BLOCKAGE IN A CANDU FUELLING DUCT, I.G. Elphick, A.M.C. Chan, Ontario Hydro
... 252
(vii)
SESSION 7 THE NEXT GENERATION REACTORS Chairman: R.S. Hart, AECL - CANDU Ops 1. 2.
CANDU 300 DESIGN SAFETY IMPROVEMENTS, J.H. Hopwood, S. Pang, AECL - CANDU Ops
... 259
THE CANDU 300 DISTRIBUTED CONTROL SYSTEM, W.R. Vhittall, AECL - CANDU Ops
... 265
3.
CANDU 300 FUEL CHANNEL, N.P. Singh, AECL - CANDU-Ops
... 270
4.
COMPUTER SIMULATION OF FUEL CHANNEL REPLACEMENT, J.R. Candlish, K.C. Dalton, B. Marshall, AECL - CANDU Ops A CANDU DESIGNED FOR MORE TOLERANCE TO FAILURES IN LARGE COMPONENTS, N.J. Spinks, AECL - CRNL, F.W. Barclay, AECL - WNRE, P.J. Allen, F. Tee, AECL - CANDU Ops
••• 284
COURTENAY BAY - A STEP TOWARDS THE ADVANCED CONTROL ROOM OF THE FUTURE, J.E. Smith, Maritime Nuclear
... 291
5.
6.
... 277
SESSION 8 ADVANCES IN NUCLEAR ENGINEERING EDUCATION IN CANADA Chairman: H.W. Bonin, Royal Military College 1. 2.
ADVANCES IN NUCLEAR ENGINEERING EDUCATION IN CANADA, H.W. Bonin, Royal Military College
... 297
THE UNIVERSITY OF NEW BRUNSWICK NUCLEAR ENGINEERING PROGRAM, D.A. Heneley, E.M.A. Hussein, R.A. Chaplin, R. Girard, University of New Brunswick.
... 300
SESSION 9 SMALL REACTORS: SAFETY Chairman: P.M. French, AECB 1. TANK - A COMPUTER CODE FOR THE TWO-DIMENSIONAL MODELLING OF TRANSIENT BEHAVIOUR IN RESEARCH REACTORS, R.J. Ellis, H.J. Smith, P.A. Carlson, AECL - WNRE ... 305 2. 3.
DESIGN AND SAFETY FEATURES OF THE AMPS NUCLEAR ELECTRIC PLANT, A.P. Oliva, J.S. Hewitt, ECS - Power Systems Inc.
... 312
SAFETY-RELATED PARAMETERS FOR THE MAPLE RESEARCH REACTOR AND A COMPARISON WITH THE IAEA GENERIC 10-MW RESEARCH REACTOR, P.A. Carlson, A.G. Lee, H.J. Smith, R.J. Ellis, AECL - WNRE
... 320
4.
THERMALHYDRAULIC EXPERIMENTATION IN SUPPORT OF AMPS DEVELOPMENT, R.G. Gray, T.C. Currie, J.C. Atkinson, ECS - Power Systems Inc. ... 325
5.
PREDICTION AND MEASUREMENT OF THREE-DIMENSIONAL TURBULENT JETS UNDER SUCTION AND COUNTERCURRENT FLOWS IN A MAPLE-TYPE TEST FACILITY, S.Y. Shin, J.E. Kowalski, D.K. Baxter, R.L. Hmbroff, AECL - WNRE
(viii)
... 332
SESSION 10 OPERATIONAL ENHANCEMENTS - II Chairman: D.R. McQuade, Ontario Hydro 1. A METHODOLOGY TO REDUCE UNCERTAINTY ALLOWANCE IN CHANNEL FLOW VERIFICATION, K.F. Lam, Ontario Hydro 2.
3. 4.
5. 6.
... 341
DEFECT DETECTIVE: AN EXPERT SYSTEM FOR THE DETECTION AND EVALUATION OF FUEL DEFECTS IN CANDU-600 NUCLEAR POWER REACTORS, A.M. Hanzer, J.W.D. Anderson, C.V. So, AECL - CANDU Ops
... 347
END FITTING ROLLED JOINT INSPECTION APPARATUS FOR CANDU REACTORS, E. Cornblum, V. Yu, Ontario Hydro
... 353
PICKERING NUCLEAR GENERATING STATION - A MODERATOR INLET LINE HANGER REPLACEMENT, R.A. Kirkpatrick, Ontario Hydro, J.M. Bowman, Babcock & Vilcox Canada, V.R. Syauaons, S. El-Nesr, AECL - CANDU Ops
... 3 60
THE DEVELOPMENT OF ONTARIO HYDRO'S ON-LINE COMPUTERIZED WIRING INFORMATION SYSTEM, R.P. Lindsay, B.A. Rolfe, Ontario Hydro
... 365
COMPARISON OF THE RESULTS OBTAINED IN FUEL MANAGEMENT AT THE EMBALSE NUCLEAR STATION (C.N.E.) AND OTHER SIMILAR STATIONS, AND ALTERNATIVES FOR OPTIMIZATION, C. Moreno, A. De Pass, J.C. Vinez, Central Nuclear Embalse, Argentina
... 370
SESSIOH 11 FUEL CHANNELS: CURRENT POSITION AND IMPROVEMENTS Chairman: G.J. Field, Ontario Hydro 1. THE FABRICATION OF HYDROGEN SINKS FOR THE PRESSURE TUBES OF DARLINGTON UNIT 4, R. Degregorio, I. Grant, I. Inglis, E.V. Murphy, E. Price, AECL - CANDU Ops, M. Natesan, V.B. Stewart, Ontario Hydro ... 379 2.
3. 4. 5.
HYDROGEN INGRESS MECHANISMS IN ZR-2-5 vt% Nb PRESSURE TUBES AN OVERVIEW OF SOME RECENT PROGRESS, P.C. Lichtenberger, Ontario Hydro
... 384
SAG OF Zr-2.5% Nb PRESSURE TUBES, N. Badie, R.A. Holt, C.W. Schulte, R.G. Sauve, Ontario Hydro
... 391
PROSPECTS FOR IMPROVED Zr-2.5 Nb PRESSURE TUBES, B.A. Cheadle, AECL - CRNL
... 396
LEAK BEFORE BREAK AND LEAK DETECTION SYSTEMS OF CANDU FUEL CHANNELS, B.A. Shalaby, E.G. Price, G.D. Moan, AECL CANDU Ops, C.E. Coleman, AECL - CRNL
... 401
(ix)
SESSION 12 CURRENT ISSUES IN NUCLEAR SAFETY Chairman: D A , Meneley, University of New Brunswick 1.
2.
3.
4.
5.
6.
CURRENT METHODS IN QUALIFICATION OF CANDU HEAT TRANSPORT PUMPS FOR OPERATION UNDER LOSS-OF-COOLANT ACCIDENT CONDITIONS, A.N. Kumar, AECL - CANDU Ops
... 411
A COMPUTERIZED MONITORING SYSTEM FOR EMERGENCY PLANT OPERATING CONDITIONS AT THE POINT LEPREAU GENERATING STATION, Harry Storey, A.R. Johnson, B.K. Patterson, D.L. Stafford, New Brunswick Electric Power Commission
... 417
A REVIEW OF QUANTITATIVE CRITERIA FOR DEMONSTRATING NUCLEAR POWER PLANT DESIGN ADEQUACY, K.S. Dlnnie, Ontario Hydro
... 423
SOME RESULTS AND INSIGHTS FROM THE DARLINGTON PROBABILISTIC SAFETY EVALUATION, K.S. Dinnie, S.G. Lie, Ontario Hydro
... 428
A DISCUSSION OF INSTITUTIONAL FAILURE AND ITS IMPORTANCE TO NUCLEAR SAFETY, David Mosey, Ontario Hydro, Keith Weaver, Shaftesbury Scientific Limited
... 437
DEVELOPMENT OF A SAFETY ANALYSIS PROGRAM AT THE POINT LEPREAU GENERATING STATION, P.D. Thompson, D.F. Weeks, S. Alikhan, New Brunswick Power
... 447
SESSION 13 RADIATION APPLICATIONS: MEDICAL AND INDUSTRIAL Chairman: K.K. Mehta, AECL - WNRE 1.
2.
3.
TRANSPORT OF PROTOTYPE DEPLETED URANIUM CALORIMETER MODULES FOR THE ZEUS EXPERIMENT, J.J.G. Durocher, on behalf of the Zeus Collaboration, ZEUS/IPP Canada
... 453
COMPUTATION OF ELECTRON DOSE DISTRIBUTIONS IN TISSUE USING GAUSSIAN PENCIL BEAMS, G.A. Sandlson, W. Huda, D. Savoie, L. Papiez, J. McLellan, Manitoba Cancer Treatment and Research Foundation
... 459
RADIATION DOSES TO PATIENTS UNDERGOING IMAGING PROCEDURES IN MANITOBA, W. Huda, G.A. Sandison, Manitoba Cancer Treatment and Research Foundation
... 466
4.
PRODUCTION AND RADIOCHEMICAL SEPARATION OF ARSENIC RADIONUCLIDES FROM GERMANIUM, J.J.G. Durocher, D.M. Gallop, J.S.C. McKee, G.R. Smith, University of Manitoba, D.N. Abrams, M.W. Billinghurst, S.L. Cantor, Health Sciences Centre, Winnipeg ... 470
5.
THE APPLICATION OF ENGINEERING TO RADIOISOTOPE PRODUCTION, W.H. Pettipas, AECL - Radiochemical Company
... 4/6
THE CANADIAN IRRADIATION CENTRE, Y. Doyle, AECL - Radiochemical Company (Not Available at Time of Printing)
... 479
PHYSICAL AND MECHANICAL CHARACTERIZATION OF RADIATION-CURABLE CARBON FIBRE COMPOSITES, C.B. Saunders, A.A. Carmichael, V.J. Lopata, A. Singh, AECL - VNRE
... 480
6.
7.
AUTHOR INEEX
... 487
Session 1: Operational Enhancements - I
Chairman: R.E. Lewis, Ontario Hydro
CNS 9th ANNUAL CONFERENCE, 1988
k
1
ONTARIO HYDRO NUCLEAR GENERATION DIVISION INFORMATION SYSTEMS AT DARLINGTON NGS I.S. HEY Ontari o Hydro Darlington NGS-A INFORMATION SYSTEM ARCHITECTURE
SUMMARY
An information architecture charts the relati onshj p between the business processes (what the organization does) and the data produced and/or used by these processes. It also i dent i f i es the i n-f or mat i on •f 1 ows between vari ous busi ness processes. A very si mpl i -f i ed examp 1e architecture is shown in Fi gure #1. The boundaries outline which data will be produced/used within a particular information system and whi ch busi ness processes are lnvolved in this. Where data is used by other systems the data -flow relati on ships are also shown. Note that such an arch i tecture requir&s data to only be produced by one process (ie, avoi d data duplication) although there may be multiple users of any data.
The need -for a more e-f-ficient method o-f managi ng i n-f or mat i on 1 n Ontar 1 o Hydro's nuclear stations was initially identi-fied in the early 1900's. There f ollowed a Nuclear Generation Division (NGD) In-f or mat i Dn System Planning (ISP) Study in 1962, which i dent i -f i ed the maj or in-formation probl ems con-fronting the division and speci tied the existing and new i n+ormati on systems requi red to address these. In 1983/B4, the requirements f or these systems were de-fined. Detai led specifications were produced in 1985 and during 1986/87 the systems were desi gned, tested and installed, initially at Darlington NGS. The systems are now in -full product! on at Darlington and are in the process of being installed at Pickering NGS. This paper out 1 i nes an over vi ew o-f these systems, a brie-f description o-f the Work Management System, the way in whi ch they are all used, the costs associ ated wi th them and the bene-f i ts resul t i ng +ror,y them.
\
WTA
BUSNESsX
FCfiECAST HNMTEWWCC
INFORMATION SYSTEM NEEDS The maj or i n-f or mat i on probl ems existing in 19B2 were categari zed as •foil ows:
UWTS
DCIQJ3*
year /
A///////A///^/ WA yA E.?
1
^ u [ u; P
!
1
u
1
I "1
1
SWF
- Rapi dly increasing vol ume o-f information (estimated at 25"/. per
/
|
1
ETC
1 !
1
I
u - wthvxn
i
p -
O*T* ntcouczii
~ Information management getting less ef-f iczent despi te increases in appl ied resources
FIGURE 1: INFORMATION ARCHITECTURE - Su-f -f er I ng from data duplicati on among di -f -f erent systems INFORMATION - In-f or mat i on system i ncompat i bi 1 i t i es between di-fferent departments/statians i n Ontario Hydra - Inr -eased demand -for data m a m pulation/retri eval The ISP study developed a detailed l n-f or mat i on architecture and used this to spec i -f y the in-f or ma t i on systems requi red to address these problems.
SYSTEMS
Based on this architecture, a number of in-formation systems t common to NGD were identified, viz. i>
Staf £_Inf Drmat i.on_ - an exi sting system used to i dent iiy emp1oyees, produce t i mesheets, payroll, shift schedules, etc.
CNS 9th ANNUAL CONFERENCE, 1988 3
ii)
Oatgri ai_Manggg!t!ent_S5!-gtem_iMMSA - an existing system used to control the procurement, storage and issuing of stores material.
i i i)
igytemtDt_Se^re_Parts_iESPi - an existing system used to control the identification and specification o-f parts and components.
i v)
The mainframe computer and its peripherals have been upgraded twice since initial installation in 19B5 in order to keep pace with expanded station use.
Wor k_Managemgnt_ - a new system used to identify, plan, initiate and report on all preventative and corrective maintenance done within the station.
v)
course of their normal Bcties. At present, there are approximately 850 station users. The computer recognizes users and their associated access to different functions by inay of their badge tt and personal password, entered when logging onto the system.
COMPUTER ROOM
East _Reggrting_and_Cgntrol .System iQRCSl
TO HEAD OFFICE COMPUTERS
- a new system to gather and report or, labour and material costs. vi ;
C. P . U .
Document_Management_Sy.stem_.
. - a new system to produce, store, index, retrieve and display station documents.
PRINTER
(TAPE]-*-' COMM. CTL.
A number of specific systems were also identi-fied -for Darlington.
LOCAL CTL.
nal._Qf f i.ce_System__(PROFS>. - a new system, -for calendaring, electronic mail, meeting scheduling, etc.
REMOTE CTL.
ADMINISTRATION -JJJ ([TERM .tiI BUILDING /tl it III WORKSTATIONS
TCRM
- a new system t o produce station •flowsheets and miscellaneous drawi ngs. iii)
Chemistry._Lab_Sy.stgm_.
- a system to aid in the identi-fication, analysis and reporting o-f routine laboratory p; ocedures.
INFORMATION SYSTEM HARDWARE With the exception o-f Computer Aided Drafting and the Chemistry Lab System, all the information Systems reside on a central IBM mainframe computer located at site. (See Figure tt2) . Communication with this computer is via approximately 300 workstation's fpersonaj computers Una terminals} distributed within the station. All station staff are required to use one or more of these systems every day in the
4 CNS 9th ANNUAL CONFERENCE. 1988
FIGURE 2:
COMPUTER FACILITIES
WORK MANAGEMENT SYSTEM (WMS) The Work Management System is central to many of the key business processes involved in nucJear station commissioning anL operation. A detailed description of it, its implementation and use is given here. Many of the principles involved in WM5 also apply to the other information systems.
B«5ECiBtj.Qn_gf_WMS C-s can been seen i rom Fi g u r e 3, W M S is integrated with M M S , S I S and the E S P s y s t e m s , and t h e s e applicati o n s work dosely together.
Def i ciency Reports t DR's) are used to i dent i-fy correct! ve mai ntenance and commissi oni ng war k that has to be per-formed , eg , repai r o-f a defecti ve pump, commi ssi oning of a drier un it. Call- Up s are used to i dent i f y prevent i ve irai ntenance work that must be done on a regular basi s, eg, six month 1 y overhaul D-f a compressor .
EOUIPMEKT SPARE PARTS SYSTEM
The Projects sub-system is pri mari1y •for use try the technical staff to keep track of their mai or ongoi ng projects, iB. Desi c>n changes- DR's can be created automat i cal 1 y f rom this sub-funct i on i f -f i el d work i s requi red.
FIGURE
3:
NGD INFORMATION S Y S T E M S INTERFACE DIAGRAM
S I S i d e n t i f i e s the user when h e s i g n s on to W M S and a s s i g n s the security levels related to the position he h o l d s . M M S and ESP p r o v i d e material and s p a r e part d a t a t o u s e r s from wi thin W M S . L i k e w i s e , W M S p r o v i d e s information on material requi r e m e n t s -f or work p a c k a g e s to the supply staff using M M S . In a d d i t i o n to t h e s e on line i n t e r f a c e s , W M S inter-faces automatical ly with s e p a r a t e sch&duling s y s t e m s and with other Head O f f i c e s y s t e m s -for the p u r p o s e of cost r e p o r t i n g , control and equipment reliability data, etc.
The Work Package sub^system is the work horse of WMS. A Work Package i s created when a Def i c i e-ncy Report i s ver i f i ed , a call -up i s due to b.? per formed, or field work f or a proj ect is required. In this sub-system, information from the oths*~ systems i s oulled together to assist in the assessi ng, scheduli ng and execution of work. Work Reports, detai1ing the work done, are prepared duri ng work yxecuti on and became part o-f the Mark Package. On complet i on of the work, summary data of 1ang t&rm historical value i s transferred to the Equi pment Database. The Equi pment Database stores information relating to all t^e i ndi vi dual pi eces of equi pment that make up the station, including mai ntenance and defici ency hi story files. The desi gn of the WMS system al1ows for 3 years retenti on o-f detai 1 ed Work Package informati on and permanent on-1i ne retention of equipmenty mai ntenance and defici ency hi story files.
The Work Management System c o n s i s t s of 5 major s u b - s y s t e m s (see F i g u r e 4) all of which are h i g h l y i n t e r a c t i v e .
PREVENTIVE MAINTENANCE
CORRECTIVE MAINTENANCE
cu
DR SUB-SYSTEM
SUB-SYSTEM
PROJECT SUB-SYSTEM
WORK PACKAGE SUB-SYSTEM
The Equi pment Database was the first sub—system to be coded, tested and migrated to site in early 1986. This was -foil owed by the Work Package and Def i ci ency Report sub-systems in the middle Df the year. The Call~Ups and Proj ect sub—systems were mi grated i n late 19B6 and early 1987 respectively. Informal testing at Head Office of all the software took place during the development phase. Thi s was fol1 owed by a migrati on to a spec i al test environment for a formal test which lead up to the migration to site.
EQUIPMENT DATABASE
FIGURE 4:
WMS STRUCTURE
CNS 9th ANNUAL CONFERENCE, 1988 5
Implementation at site posed a number of challenges and was generally divi ded into the -fol lowing areas: ~ -
staff training stati on documentat i on data conversion and daca 1oading coord i nat i on o-f changeover to use of WMS - user support - back-up strategy development Staff training was begun in mid 1986. A speci al training envi ranment was set up on one o-f the Head Office mac hi nes and a training room set up at DNGS. The approach taken was to provi de detai i ed training to a -fairly smal 1 number o-f representati ves o-f all the major work groups at Darlington. These "Di sci pies" were then charged with scheduli ng and conducting all the hands-on training for their particular work group. Thi s approach was qui te successful and the -formal training was completed by the winter of 1986. Stat i on documentation had to be created or revi sed to ref1ect the new way of managi ng work. The creati on of a comprehensive User Manual was a fairly significant task, but the modi -f i cat i on of existing station procedures was •fairly straight forward as in general the process was not being changed, only the mediumData loading to the Equipment Database is an ongoing nrocess. It was started by extracting, converting, and then loading data electronically from a number of existing databases associ ated with the design and procurement of DNGS. Subsequent 1oadi nq i s 3 manual process based on physical checks of the equi pment in the station by operati ng personnel. Data conversi on of the existing acti ve Def i ci ency Reports and Ca11-Ups was carried out in the two or three weeks 1eadi ng up to the respecti ve i mplementati on of those sub-systems. Because o-f the relati vel y 1 ow vol ume of field work active at the time, this process was not nearly as major a task as it will be when operating stations such as Pickering implement WMS. The actual implementation of the system vas done in a phased manner. In Octoher 1986, the Equipment Database, DR, and Work Package sub-systems were implemented for a small area of the station (the Water Treatment Plant). This mini i mplementat i on permi tted a coup 1 e of maj or prob1 ems to be i dent ified and corrected with minimal
6 CNS 9th ANNUAL CONFERENCE, 1988
di srupt1 on, and also some conf i dence was built up prior to going stati an wi de. One month later, i mplementat i on was extended to the whole stati on. In January 1987, the Cail-Up sub-system was i mplemented stati on wi de. Subsequent releases of software such as the various interface packages have been imp1emented as they have been mi grated. The Project sub-system I s installed but has not yet been put to
User support has had a high profile and was seen as a key element in the success of i mplement i ng the system. In addi ti on to the provi si on of training and a User Manual, there have been a number o-f WMS i mp2 emBntst i on meeti ngs with the work groups, resources are avai1able to help users experi encing problems wi th the system, and an iof ormati on bulletin is issued on a regular basis to deal with problem areas Dr orovide new information. A simple bacK-up procedure that would see the f i 11d personnel usi ng a paper system for short term computer outages has been developed. A d i sastrer recovery pi an to use in the event of a long term computer outage is also being developed.
WMS has been i n servi ce at Darl i ngtan for approximately 19 months; and, as of Apri1 198B, data volume is approximately: 165,000 5,000 21,000 42,000
Equi pment Records Entered Call-Up Records Created Work Packages Raised Work Reports Written
There has been a high degree of acceptance of the system by the station staff. The wri ting of Def i c iency Reports and Work Reports by field staff, such as Operators, Mechanical Maintainers and Control Maintainers, is an integral part of their work. The previous paper systems have been totally replaced by the electronic system and typical funct ions (such as document preparat i on, ver i f i cati on, assessment, approval, i nqui ryf scheduling) are now performed at a keyboard. Maj or contr ibutors to user acceptance have been the many user friendly features of the WMS system and the high reliability of the system - no occurrences of data 1oss and system avai 1 abi 11 ty > 90"/..
Since implementation, a number o-f enhancements to WMS have been made to i n crease its -f unct i onal i t y. These changes were to a large extent identified by regular users of the system. An ongoi ng enhancement program is considered to be another key element in the success of the system. WMS BENEFITS The major identified bensfits of the system are associ ated with the availability to all station staff of up-to-date, on-1 in© i n + ormatiDn. Thi s all ows staff to perform their jobs more ef feet i veiy, i e, - Field Supervi sors obtaining up-to-date job status i nf ormation on work i n progress, - Techni cal staff obtaining i nformation an work i n progress on the equi pment they are responsible for, and also hi star i cal inforniati on about that equi pment.
INFORMATION SYSTEMS COST/BENEFIT ANALYSIS A cost/benef it analysis was done on the NGD Inf ormat i on systems at Darlington for the period 19B5 - 1992. The revi ew was based on site exper ience for period 1985 - 19BS with a projecti on to 1992. The concZ usi ons were as foilows: a>
The original cost estimates of M$ far the i nf ormati on system implementation (hardware and software) were achi eved.
7.25
b)
The total costs and quanti f i abls benefits are illustrated in Figure #5 and summarized as follows: Net Present Value
Total Costs Total Quantifiable Benefits Net Benefits
- Mai ntenance assessors f i ndi ng hi stori cal information about previ ously executed work Df a si mi 1ar nature. Also, the assessors have rapid on-line access to spare parts
33.49 M* 93i34_M* 59. B5 M*
Efficiency
- PIanni ng staff obtai ni ng up-to-date job status information and also bei ng not i f i ed of new work coming up for scheduli ng. - Station management staff able to review 'Performance Moni toring' statistics for the major work groups, •for analysis and corrective action. This informati on i s readi1y avai1able from off-1 ine reports and much o-f it was not avai1able in a ti mely manner or at sensible cost using the old manual systems. Some other areas where savi ngs are likely to be made include: - No 1ost paperwork
ItBt
FIGURE 3:
1987
1B90
1991
EFFICIENCY 81 EFFECTIVENESS IMPROVEMENTS
(i e, 1ost data) CONCLUSIONS
- No del ay due to mai1i ng ti me - Reduced data i nput requi rements - Less clerical/fi1 ing effort - Reduction in resources required to administer Call-Up program
The NGD Information Systems have been fully implemented at. Darlington. They have been generally well accepted by the Users and are .r-howi ng evidence of increasing efficiency and reducing costs in a number of areas. The key areas of note are the systems ability to provide on-line, up-to-date information quickly to all station users, and the fact that they are able to share information with the other systems.
CNS 9th ANNUAL CONFERENCE, 1988 7
OPERATIONS DECISION SUPPORT SYSTEMS FOR CANDU NUCLEAR PLANT OPERATIONS
H.E. SILLS, J.W.D. ANDERSON
Atomic Energy of Canada Limited
INTRODUCTION A number of ma jor accidents, such as Three Mile Island, Bhopal and Chernobyl, have highlighted the role of the process plant operators and the facilities to suppo rt them during complex disturbances. Increased plant complexity, tighter operat ing constraint^/ longer periods between upsets, and too much data, too fast, can result in high stress and unreliable performance. On most occasions, plant operators respond correctly to faults. On rare occasions, there has b e e n significant operator contribution to the course of the accident as well as equipment malfunction. In CANDU, as in the process industry in general, operators must bo provided with better information more quickly on the current status of plant components, actions to be taken, expert advice on complex situations, and the basic reason (s) for upsets. Studies in human-machine interaction emphasize the necessity of treating the designer, the operator and the plant as an integrated information system, "The introduction of new technology such as computer-based information processing necessitates a reconsideration of Che basis for design of industrial control systems. A gradual updating of previous des ign in response to operational experience is very likely to lead to systems that are suboptimal and unnecessarily complicated. As an example, the use of computers to analyze alarm signals in process pl^nt control rooms has been introduced to avoid ovei-load from the high number of alarms/ which in turn are a result of the traditional one-sensor-c>ne-indicator technology with normal range monitoring of measured variables individually" (1) . The availability Of inexpensive, powerful computing and interactive display hardware provides many opportunities f of increased integration of plant information handJ-ir.g and the development of operations decision support systems (2, 3 ) . This is facilitated in the case of CANDU (CANada Deuterium Uranium) which pioneered the application of computing technology to nuclear plant operations. The excellent performance and safety record of CANDU plants has been due, in part, to the success of this innovation. The long term objective for programs in operations decision support systems is reliable, automatic operation of the whole station in which advanced computing technology will form part of basic plant control and management systems (2) . This paper deals with the short term objective of better utilization of currently available plant information. COMPUTER ASSISTED OPERATIONS AND MAINTENANCE The decision to use digital control in CANDU power plants waa made in the 1960s. This pioneering decision led to the CANDU system being the world leader in computer control. In subsequent years, strong instrumentation smti tifesiq-r. IL«&IN& aV JteCL Vi&**e incorporated new technology, aa it became available, into the instrumentation and control design. For example, mimic boards are augmented by CRTs and the safety system logic has been computerized.
8 CNS 9th ANNUAL CONFERENCE, 1988
The recent advances in artificial intelligence technology, particularly expert systems, presents a further opportunity for significantly increasing the role of computers in nuclear plant operations (4-6) . As can be seen from references 7-12, there is a general worldwide trend on using advanced computing technologies in support of station design, maintenance, and operation. Systems developed using these technologies can enhance plant and operator reliability, increase plant availability, reduce the stress on operating staff/ increase plant automation, and improve the information flow to plant management.
PROTECT PUBLIC
FIGURE 1:
PLANT STATUS PROBABILITY
Figure 1 illustrates the probability of finding a CANDU plant operating in either the normal, upset or accident condition (13) . The automatic control system already assists the operator with many of the functions covered under normal operation. Although the need for operating assistance is perhaps more evident in the accident regime, the number of times this operating regime is encountered during the life of the plant are indeed few. This suggests that a significant portion of operations decision support systems should deal with normal plant operation and maintenance in order to realize the maximum benefit from the development effort and ease the implementation of such systems. More importantly, frequent usage of such systems away from crisis situations would instill operator confidence in their use and increase the likelihood such systems would be used in less probable situations (14). CURRENT PROGRAM AECL has identified computer assisted operations and maintenance as a strategic program. Discussions are in progress between AECt and the utilities to further refine the needs and concepts to sucessfully pursue identified initiatives as a national program, A major undertaking which is receiving support within the research and engineering groups of AECL is the Operator Companion project, an expert system intended to diagnose plant faults and advise the operator on appropriate restoring or corrective actions (13}. In its final form, Operator Companion is planned as a family of cooperating expert systems (and other advanced computing systems) communicating with each other and with the plant via a local area network as shown in Figure 2. This type of architecture offers a number
-
2.
the data from the system can be made available in preprocessed form matching the diagnostic strategy or strategies being considered,
- modules can be allocated to be dedicated, faster than real-tiire processors for plant data analysis,
Fault detection has t r a d i t i o n a l l y been handled by interpretation of the computerized alarm annunciation. However, diagnosis of the root problem is often done manually by the operator performing a t ime consuming search through flowsheets and manuals. The computer's ability to tirelessly and exhaustively search through data currently offers the best prospects for automated fault detection and diagnosis.
- real-time simulators or plant analyzers (5) can be incorporated for on-line power p l a n t decision making and "on-line" operator training, redundant workstations can offer equipment failure, a itiodular development and strategy can be used, and
On-line fault detection and diagnosis addresses the broad problem of on-line fault detection and diagnosis for any event and suggesting to the operator the most effective course of action.
recovery from
implement at ion 3.
obsolete models of workstations can be replaced as newer technology emerges.
Plant configuration and equipment status monitoring - to enhance the ability of the plant operators to monitor the physical status of the plant and i t s major equipment. The plant operator currently has to interpret the status of the plant from diverse information sources such as operat ions reports, manuals, drawings, control panel displays, and alarm indicators. On-line access to this information will provide a better indication of plant status on which to base operating decisions.
4.
Vital operating parameters - providing expert system tools that will enable the operator to rapidly acquire information on a l l the vital operating parameters, and normal and safety limits under any operating condition. Since the f i r s t requirement is to judge the overall consequences of a disturbance, this task is aimed at presenting the operator with concise picture of the overall plant profile based on vital operating parameters.
5.
Plant plant access events
operat ing procedures - provide the operator with convenient and rapid to relevant operating procedures as unfold.
Other technical broader 3pectrum of systems include: 1.
FIGURE 2 :
have been Companion 2.
1.
Improved CANDU alarm annunciation strategy an improved alarm processing and annunciation strategy for CANDU plants to assist the control room operator in obtaining a better and faster understanding of the plant status so that he can better respond to upsets and routine events (6). Some remedies, such as alarm conditioning, flight recorders, classification of major and minor alarms, and sorting alarms by system have already been implemented into the design. These have not entirely solved the alarm flooding problem or adequately improved the operators' ability to diagnose events.
relating decision
to the supports
Modules for specialized operational tasks to provide an operator assistance for a variety of specialized, routine tasks. An operator in a CANDU plant has to carry out a variety of specialized tasks such as fuel management, interpretation of plant chemistry data and identification of fuel defects . Expert system technology offers the opportunity of capturing scarce expertise and making i t readily available on every shift.
OPERATOR COMPANION ARCHITECTURE
The following technical a c t i v i t i e s identified as part of the Operator project:
activities operations
Automated testing of safety system components - for automating the routine testing o f safety system components in the current and future CANDU plants. Operators have to manually perform complex valving logic to isolate each channel for t e s t i n g and then return i t to s e r v i c e . Experience has shown that such testing i s labour intensive, creates a high maintenance load and decreases the overall safety system reliability. Automated testing should alleviate these problems by integrating selftest and diagnostic functions directly into the safety system components.
CNS 9th ANNUAL CONFERENCE, 1988 9
3.
Expert system t o o l s in e n g i n e e r i n g and c o n s t r u c t i o n - for e n g i n e e r i n g , d e s i g n , procurement, material control and c o n s t r u c t i o n a c t i v i t i e s t o make s c a r e e x p e r t i s e more widely a v a i l a b l e and freeing experts for more demanding t a s k s .
The level of abstraction and the search strategy are shown schematically in Figure 3. An estimate of the resource requirements of the various diagnostic strategies has been suggested by Rasmussen (1) and is presented as Table 1,
Success in these technical activities requires a careful match of the information processing done by computer and the information requirements of an operator during decision making. For example, Rassmussen (3) defines five levels of abstraction used to represent the functional properties of a system - functional purpose (highest level), abstract function, generalized function, physical functions, and physical form (lowest level). The importance of this to diagnosis is that identifying the causes of system malfunction requires information from the lower levels whereas the reasons for correct opcration are derived from higher levels. Another aspect is the depth of knowledge underlying human action related to skill-based behaviour, rule-based behaviour (compiled pattern matching) or knowledge-based (model-based) behaviour. This aspect is closely related to the choice of topographic or symptomatic search strategies for diagnosis. Topographic search tries to locate sources of deviation from normal operation whereas symptomatic search tries to match observations with atored templates of abnormal system conditions. Symptomatic search can be further subdivided into pattern recognit ion, decision table search, and search by hypothesis and test.
RESOURCE REQUIREMENTS FOR VARIOUS DIAGNOSTIC STRATEGIES
Performance factor
FIGURE 3 : RELATIONSHIP BETWEEN THE VARIOUS ABSTRACTION LEVELS OF FUNCTIONAL PROPERTIES AND THE TYPE OF CONTROL RELATED ACTIONS (ADAPTED FROM REF. 1 5 ) .
LOW
-
Low
Station operating manuals and operator training fit the decision table and pattern recognition strategies respectively and correspond to the recognition of design basis events. However, foi situations requiring significant diagnosis, such as unanticipated events, plant operators need access to a deeper knowledge representation of the plant available through topographic, and hypothesis and test strategies. This suggests that a nt/m'xr of diagnostic stratagies should be available to plant operators.
High
-
-
DEMONSTRATION SYSTEM
Low
High
High
Low
-
High
Topographic pattern search recogn- Decision Hypothesis it ion table and tegt
Time spent Number of observations High Dependency on pattern perception Load upon short-term memory Low Complexity of cognitive processes Low Complexity of functional model Low General applicability of tactical rules High Dependency on malfunction experience Low Dependency on malfunction preanalysis
Topographic
LOW
High
High
10 CNS 9th ANNUAL CONFERENCE. 1988
High
A demonstration system of an operator decision support system (ODSS) is currently being developed at Chalk River as a vehicle to realistically demonstrate how computing technology can improve plant operation, safety, and maintenance. The various functions are being implemented with general-purpose workstations coupled to a central plant database using local area network (LAN) technology.
Also attached to the LAN are Subsystem Experts that transparently access the plant database, apply rule-based algorithms to the station data, and offer recommendations in the event that a fault condition is detected. An individual Advisor will be dedicated to monitoring each of the major subsystems of the plant. Model-based reasoning for topological, and hypothesis and test diagnostic strategies relies on information from Subsystem Simulations also connected to the LAN. An object-oriented programming language is used to provide physical and functional representations of the subsystem. FIGURE 4: OVERVIEW OF DEMONSTRATION OFLRATOR DECISION SUPPORT SYSTEM
A small reactor has been selected for demonstration purposes as it is of appropriate scale while providing a sufficient number of components, sensors and alarms to adequately test Operator Companion concepts. Figure 4 shows this small reactor with its own dedicated data acquisition system providing high speed, jni-directional communication with the plant database. To demonstrate plant configuration and equipment status monitoring, cfte Operator Consols provides interactive plant schemat ica and component status as shown in Figure 5. Features currently incorporated include: - displays of the as-drafted flowsheets for the various operating areas of the plant, - display of the current device status of all operable devices as shown on the flowsheets, -
the ability to provide black and white hard copy prints of the flowsheets and operable devices,
- displays of the current device status including physical state as well as Work Protection Code tag status, - trend plots of component data, component status logs and specific information, and - dialogue with Advice File.
FIGURE 5 :
the Subsystem
Expe rt via
m
After many of the operations decision support system concepts have been successfully demonstrated, the next step will be to develop a stand-alone system and evaluate the system performance in a CANDU plant simulator before installing in a real plant. CONCLUSIONS Computer a s s i s t e d o p e r a t i o n and m a i n t e n a n c e w i l l yield major benefits in plant safety, reliability and a v a i l a b i l i t y through b e t t e r information utilization, improved diagnostic capability and improved surveillance, Advanced computer systems for operator support during normal operat ion provides improved protection of the significant investment in plant and reduced labour costs for station operation. Although the technical challenges are significant in achieving appropriate solutions, in meeting the plant operators information requirements, and in implementing operations decision support systems in the plant environment, success will help mainta-n or improve CANDU's technical leadership and performance record. REFERENCES (1)
RASMUSSEN, J . , " I n f o r m a t i o n P r o c e s s i n g and Human-Machine I n t e r a c t i o n : An Approach t o C o g n i t i v e Enginee- J n g , " North-Holland, 1986.
(2)
BROOKS, G . L . , "Advances i n Commercial Heavy Water Power R e a c t o r s " , P a c i f i c Rim Conference, B e i j i n g , China, 1987.
(3)
FUJII, K., SUTO, 0 . , KANEDA, M., AND KAWAMURA, A., "Application of H i e r a r c h i c a l Computer Complex Concept f o r Nuclear Power P l a n t s , IEEE T r a n s . Nuc. S c . NS-30, 1983, 806-810.
(4)
DUNN, J . T . , LIPSETT, J . J . , NOTLEY, M . J . F . , and SPINKS, N . J . , " F u t u r e Trenda in t h e Design of CANDU R e a c t o r s " , AECL-9179, 1986.
(5)
CHOU, Q . B . , "Applying "Expert Systems" Concepts to Advanced Power P l a n t C o n t r o l " , Instrument S o c i e t y of America Power Symposium, 19B6 May.
(6)
"Power P l a n t Alarm S y s t e m s : A Survey and Recommended Approach for Evaluating Improvements", MPR Associates, I n c . , EPRI report NP-4361, 1985.
(7)
"MAPI's K n o w l e d g e - B a s e d S y s t e m w i l l H e l p t o Deal With Abnormal C o n d i t i o n s " , Nuc- E n g . I n t . , 1987 J u l y , 2 8 - 3 4 .
the
INTERACTIVE SCHEMATIC FOR THE SECONDARY HEAT TRANSPORT SUBSYSTEM
CNS 9th ANNUAL CONFERENCE, 1988 11
(8)
I T O H , M . , T A I , I . , MONTA, K . ,a n d S E K I M I Z U , K . , "Artificial Intelligence Applications for Operation and Maintenance, Artificial I n t e l l i g e n c e and other Innovative Computer Applications in the Nuclear Industry", ANS Topical Meeting, Snowbird, Utah, 1987 August.
(9)
DELAIGUE, D., AND GRUNDSTEIN, M. , "A Survey of Framatom's Expert Systems A c t i v i t y " , i b i d .
110] SCHMIDT, F . , "Knowledge Based Systems Nuclear A p p l i c a t i o n s in Germany", i b i d .
for
(II.) ANDERSON, V., "NAK/INF: Advanced Information Technology for Accident and Emergency Management", Riso National Laboratory, i b i d . (12) UHRIG, R.E., "Applications of I n t e l l i g e n c e i n t h e U . S . Nuclear ibid. (13)
NATALIZIO, A., ANDERSON, J.W.D., and SILLS, H.C., " O p e r a t o r Companion: Advanced Support Systems for Plant Operators", AECL-9612, 1988.
(14) AUSTMAN, H.r.., P r i v a t e Manager, DNGS-A. (15)
12
Artificial Industry",
communication.
Station
MILNE, R., " S t r a t e g i e s f o r D i a g n o s i s " , IEEE T r a n s . Systems, Man, and C y b e r n e t i c s , 3 , 1987, 333-339.
CNS 9th ANNUAL CONFERENCE, 1988
ROUTINE TREND ANALYSIS PROGRAM
FOR PLANT PROCESS PARAMETERS
J. G. COMEAU New Brunswick Electric Power Commission Point Lepreau Generating Station
ABSTRACT Trend Analysis is the review of organized presentations of process parameter data over time for the purpose of understanding process control response to unplanned transients or system performance during normal operation. The concept of establishing a formal Trend Analysis Program for critical parameters of key process systems during normal operation is reviewed. Improved plant efficiency, reduced calibrations, potential avoidance of process transients, and the establishment of a more knowledgeable and responsive staff are all benefits from such a program. Easy access to personal computers and the availability of good analytical data base software have made this good practice a more viable and timely option for use by station staff as a predictive maintenance tool.
INTRODUCTION In most Canadian nuclear power stations there is a general requirement to review alarm summaries and process parame' -^r trends following plant upsets in order L O develop a comprehensive understanding of the cause of the trari ?ient and the event sequence. This practice is also used by station staff to troubleshoot process system abnormalities• It essentially consists of the review of alarms and graphs of process system data vs time, known as Trend Analysis. With dedicated effort from Operations and Technical staff, it is often possible to establish the cause of the unplanned transient and off-normal process response. The practice of trending process parameters during normal operation with a view to detecting incipient problems or to predicting maintenance requirements is, however, not commonly done except for a few specific areas, such as for station chemistry. Recent experience at the Point Lepreau Generating Station (PLGS) has illustrated the benefit of establishing a Trend Analysis Program for key plant process parameters during routine operation. Benefits from such monitoring, such as confirmation of appropriate system response, early identification of calibration problems, or the need for overhauling or cleaning equipment, have been achieved* This paper reviews the following aspects of a
Routine Trend Analysis Program for plant process parameters: a) data collection considerations for effective Routine Trend Analysis Program
an
b) current tools available to facilitate and enhance a systematic review of process system data c) routine Trend Analysis applications to illustrate its success in identifying system degradation and initiating corrective action to maintain optimum performance d) potential benefits to stations achievable by developing and applying such a program e) implementation considerations in establishing a Routine Trend Analysis Program A program requiring regular monitoring of process system performance beyond the continuous (gross deviation) monitoring served by the process alarms is practical and can enhance the performance and safe operation of the plant. If data trending is sot up so as to discriminate small changes in process state, it can indeed Identify calibration problems or the degradation of process equipment prior to system upsets or process alarms• In the former case trend analysis could complement normal calibration programs and permit a more rationalized approach to calibration frequency. DATA COLLECTION FOR TREND ANALYSIS There are several factors which should be considered when determining data collection needs for a formal Trend Analysis Program, namely: a) the identification assessment
of
parameters
for
b) the facility of data collection c) the method of data collection d) data anomalies from auxiliary operations e) the timeliness analysis
of
trend
production
for
CNS 9th ANNUAL CONFERENCE, 19BB 13
Knowledgeable station stat'f are able, through an undersCanding of the process system design features and expected response to transients, to identify the key parameters. In CANDU plants which have an abundance of process control programs, such as HTC, the key parameters normally include temperature, pressure, flow, level and mass > A basic approach to the selection of parameters key to process system health monitoring is to use alarm parameters or those critical safety parameters identified in Emergency Operating Procedures (EOPs). The facility of data collection is important to the selection of parameters for Trend Analysis. At PLGS data on various key parameters can either be called up automatically (Daily Station L o g ) , requested via Trend, Bar Chart, or Analog Input call-ups, obtained via data loggers, or read off of various station indicators. Since they are readily available, any such parameter can be used. Data that would require installation of new instrumentation might not be practical and would not normally be considered for such a program unless determined to be relevant (ie back-up parameter to one considered critical). The method of data collection is a key factor to a Trend Analysis Program. Station control computer inputs can easily be accessed for data collection. All important process parameter information can be accessed from one location. The use of station control computer generated Station Logs, regularly requested Trends, Bar Charts, or Test Procedure data are all common data sources • Analog process inputs tied directly to procass data loggers or monitoring computers make excellent candidates for routine trend analysis. The effect of normal routine operations on process system parameters are anomalies that should be considered when collecting data. For example, on-line fuelling affects certain parameters within the Heat Transport (HT) system, namely the pressurizer level or storage tank level, and could bias a trend if data is taken during fuelling on a few occasions without flagging this fact. Therefore a long term trend data base should normally avoid fuelling, as different channel flows or powers might affect the data. One "trick of the trade" used when only weekly data points were required was to use regularly called-up data taken on a day when fuelling is not normally done, such as Sunday at Point Lepreau.
• .it less than optimum response during the r.-ansient. For trend analysis of process parameters during normal operation the timing requirements are generally less restrictive. Data plotting and collection at weekly, monthly, or quarterly intervals may be adequate. Experience
14 CNS 9th ANNUAL CONFERENCE, 19BB
gained after such a program was put in place, as well as the facility of data collection, would dictate the required interval. TREND ANALYSIS TOOLS The primary tool for Trend Analysis is the graph plot which presents data in a forma; which facilitates interpretation by plant staff familiar with system performance. The typical graph consists of a plot of the desired parameter or parameters over time. Another method used is the comparison of data on established control computer Bar Charts or Trends (on routine call-up) with previous hardcoples of the Trend or graph over a pre-determined time period in order to observe any trend or significant change In the data. The most widely used method for trending data up until the last several years was to manually plot data on graph paper. This procedure Is still widely used and is generally very quick for producing graphs of small data bases. Hov7ever, if a data base consists of several different parameters and different variations of graph plots are required to facilitate or enhance analysis, this method is extremely time consuming for technical staff generally too busy to spend much time plotting data. This Is one reason why formal Trend Analysis has heen done for process or plant transients only, and long term surveillance of normal operation has been generally ignored. Another method used by staff at PLGS is a concept, hereby referred to as the reference System History File. This concept Involves the setup of a reference file of station control computer generated Trends, Bar Charts, and System Status Displays against which similar trends or dibplays are compared on a periodic basis (generally monthly) or subsequent to a system or plant upset. This method is a quick method of Identifying fairly significant changes (within alarm limits), but it is not sensitive enough to identify incipient or slow changing trends. The use of customized computer programs developed by analysts to plot data is not generally used for the trending of process parameters. The cost of developing the programs and the limited accessibility to most station staff has made this tool not very practical. It has however been used by staff from time to time. Over the past few years various data base software programs, developed for business applications, have been applied to process trend analysis with a good deal of success. Easy access by engineers to PCs and data base software has made the implementation of Trend Analysis Programs a more practical and less time consuming method for the monitoring of process performance. In fact these programs have facilitated development of database information
systems which can be used for easy retrieval of information as well as for the plotting of graphs or bar charts for interpretation by technical staff. At Point Lepreau data base software packages currently used by Technical stafr otter scale and range changing, time contraction and expansion, and multiple trends on t he same graph. Although not ideal, one commonly used at PLGS is REFLbX due to its facility of use and its accessibility at most PC stations. No doubt staff at other stations have found data base software capable of meeting their needs. New products will bring definite improvements. The key point made is that software now exists which can facilitate the task of plotting and replotting various parameter combinations to permit trend analysis and good presentation formats that aid in the uncovering of problems or causes of upsets that otherwise would go unnoticed. Their existence has virtually eliminated the task of Technical staff developing plotting programs and shortened the ti me to prepare graph plots for analysis. In addition, there are new developments now under consitle ration to make data base handling even quicker. The implementation ot hardware systems which would transmit on-line data to off-line computers loaded with various Trend Analysis packages would provide for a more reliable and timely monitoring capability that would ultimately lead to improved stat ion performance. A system called "Gateway" by the Computer Group at Point Lepreau is currently under review and falls in this category. The "Gateway" concept is essentially an interface between two computer network?, the station control computers and the PLGS VAX network (an off-line system accessible to station staff via "dumb" terminals or via PC links). It is designed to make on-line data available on the VAX network, and hence to station staff without disturbing Control Room operations. A full Trend Analysis Program would be unmanageable otherwise.
alerting .System Engineers of a problem. However, when such systems are developed at stations, it is most important that operations and technical staff co-operate in the "expert" system development; otherwise motivation to use such an aid and to trust it would be minimal. It Is now clear that the advent of improved tools for data base manipulation provides an opportunity to introduce improved system surveillance practices with existing manpower. The software tools, once staff have been familiarized in their use, offer excellent data retrieval, manipulation and analysis capability over the old manual graph-nlot technique. A concept which involves the rrans.nittal of "real time" data to off-line PCs with data base trending and "expert" interrogation programs will greatly enhance this practice.
ROUTINE TREND ANALYSIS APPLICATIONS The common application of Trend Analysis is the review of process parameter data for the various systems subsequent to plant upsets in order to assist in the understanding of the fault sequence or cause, to verify proper system response, and to assess system response to the transient for design improvement. Such trends are generally of a f-hort period in which process parameters fluctuate over a large range. Interpretation of information is often quite complex because of the effect of the various parameters on each other during a transient. An example of a hand drawn graph-plot used for assessment of a Load Rejection event at Point Lepreau is shown in Figure 1.
Once this network is established, possibilities with respect to data collection (off-line), its retrieval, and its analysis are greatly expanded. On-line data stored in VAX memory could be retrieved at any terminal, probably on a system basis. If Trend Analysis software routines are aet up so as to periodically call-up and trend process data, postmortem analysis of station upsets can be done shortly after the event. As a further extension, intelligent VAX terminals could be placed in the Control Room for the express purpose of collecting on-line data, analyzing that data for EOP governing conditions, and assisting the operator with the execution of the EOPs. liven t identification and early on-line diagnostics are also possible through "expert" systems currently under consideration for equipment health monitoring. Software could be developed in which deviation setpoints could be set within normal alarm setpoints and which could flag process deviations right at desktop PCs,
FIGURE 1 - LOAD REJECTION TREND
CNS 9th ANNUAL CONFERENCE, 198B 15
This giaph plot was used by tha System Engineer to determine whether the pegging steam valves functioned properly during a load rejection in July iy»7. The practice of trending process parameters during normal operation, has several dif ferer.c applications, namely; a) normal operation transient monitoring b) surveillance testing monitoring c) system and equipment health monitoring
with the HT system drained to header level. A station computer generated Trend was set up to monitor various HT parameters, namely SDC HX outlet and three Reactor Inlet Header (RIH) temperatures at the same header while the HT system was drained to header level. The Trend together with channel outlet temperature data indicated that many channels in a given core pass had reverse flow. Although cooling was acceptable, the condition was undesireable and SDC system valve positions were a Ltered to correct this. The station computer generated Trend used for this application is shown in Figure 3,
d) long term trending of process systems Normal Operation Transient Monitoring This application refers to the trending of process data of infrequent, although normal, operational transients or system states, such as hT cooldown, HT warm-up, or flow verification manoeuvres. It involves the plotting of data over time to provide a system response trend which is compared against a reference trend for such operations. This permits station staff to assess whether equipment or control programs responded acceptably to process demands during the operation. This data may also be useful for feedback to designers or safety analysts when verifying or modelling such system response. An example of such a trend is provided in Figure 2.
M(WS»
111! IS 15:#:»
$mc
sic mm 8IIBI
iiHi4 itnp aiSICC
8 ! » M TEUP J3.UXCC
RIHM TEB 33 2* KG C
5i ee
50 81
25 81
25 II
si as
se.n
25 Bl
25.09 IS 49
HT
COOLDOWN CURVE FIGURE
3:
SYSTEM STATE MONITORING (DCCX TREND)
Surveillance Testing Monitoring Besides providing procedural steps to demonstrate system logic availability, some tests require the recording of key parameter data to permit an assessment of equipmenc degradation. For example, the valve closure times for Containment Isolation or Dousing valves are recorded during tests. Trending of this data over time can flag equipment deterioration and would result in scheduled preventative maintenance of equipment rather than emergency maintenance when the test data revealed that an impairment level had suddenly been surpassed. This extra capability, achieved by routinely trending this data, might indeed be responsible for preventing an outage due to a special safety system impairment. FIGURE 2 - NORMAL OPERATION TRANSIENT (GRAPH) An example of this application used to monitor system performance during a non-standard, although normal, system state was the monitoring of the Shutdown Cooling (SDC) system performance
16 CNS 9th ANNUAL CONFERENCE, 198B
There was extensive use of data trending during the last Reactor Building Pressure Test which facilitated timely decisions regarding leak rate and repair requirements • Data was input to a floppy disc via a data logger. Data was then brought Co off-line PCs with REFLEX software to process it into trends for review.
System and Equipment Health Monitoring This application involves trending of process data for comparison against past or reference (expected) data so as to detect degradation of equipment or process control• Early detection of equipment degradation using trend analysis is a benefit of using this methodology as a predictive maintenance tool• If trends are scaled so as to be sensitive to s.tall changes in process state, they can identify the need tor calibration before spread limits are exceeded or design variables, such as heat transfer across HX, have reached undesirable levels. In the former case the use of routine trending could complement normal calibration programs and permit an approach to calibration frequency based on actual drift limits rather than on arbitrary intervals which may result in overcalibration of some instrumentation. The degradation of a HX or turbine performance can often be detected earlier than in current practice if Trend Analysis is applied as a consistent and regular program. Timely call-ups for cleaning or maintenance of such equipment, based on Trend Analysis results, would optimize station performance. Two examples are provided to illustrate this application. To establish optimum timing for condenser cleaning, station control computer generated Bar Charts, shown in Figure 4, and Displays are used by the System Engineer. The parameters of concern in this application are Condenser Cooling Water (CCk) flow, Condenser vacuum, and CCW temperature.
Ct» 5VSTE11
B7FEII8X CCUP IISCH, 1.(2 CUB TOUT H 81 CD1I TOUT 21.37 c»2» nn 21.37
I.N
CI2I TOUT
1.91
in 1.91
21.9!
i
1
1 25 91
1
1
1
1
1
1
1
1
1
1
1 59.89 M93 I 388 58.99 1 58.» 12598 nu 1 W It
1
1
1
1
1
1
•MM
i 3.89 n n PUS 3.89 i tiei PRESS PRESS CUTER FOR PRRailEIElS
Clll PRESS
"W
FIGURE 4:
!Ei C KG C DES C IEC C IES C 1CSC DEG C FP
1
CUB TOUT 21.tS 9.93 CD3I TOUT 9.89 1 REAC POKER 9.8988 SEN lODII CtH SFT-tl 3 895 CCV PlSUf 1.96 1.99 1 CUP 182 IP 83.87 188
The Pressurizer level trends plotted before and after the impulse line blowbaclcs showed a significant change in the spread. In addition the new spread of the instrumentation was so small that calibration of the transmitters was deemed inappropriate. This application is illustrated in Figure 5.
FRESScRiZtR LEUEL TSANSfliTTES TSEN 10.53
IMPULSE LIKE
10.50
CKVO RDfOVftZ.
10.45
COMPLETED
BLOWBAOC3
10. *0
M !
1 389 88 mm 389 91 KPfllO) 1 18.88 (FBIfll 1 1199 19 99 mm
mm
SYSTEM MONITORING (BAR GRAPH)
A Trend of Pressurizer level transmitter readings to monitor spreads in data over tine can be used Co indicate transmitter calibration needs
i i
10.35 10.30-N 10.35
|Yi
!
0
i
I
! V !*i * 6
- » - J!L.LT»
FIGURE 5:
I 5» IS
I.M
and problems. One such trend showed a spread to be well beyond the accepted allowance. Based on past experience, blowback of the transmitter impulse lines was recommended to remove any accumulated sludge which might be responsible for the poor response.
3
10
12
- * - 'J/i.ui
K
i i i
IS
18
cO
II
H
- f - LH..1.IC
EQPT HEALTH MONITORING (REFLEX GRAPH)
In future such trending will be continued as a predictive tool for Impulse line blowbacks and transmitter calibrations. Long Term Trending of Process Systems This application involves the long term trending of data with a view to understanding Jong term process response over time, to establishing seasonal changes, and to identifying and analyzing any short term anomalies and their causes. Such trending was done at PLGS for the Pressurizer level and RIH temperatures from first full power operation to the present (Reference 2 ) . The trends revealed an increasing Pressurizer level over time, corresponding with increasing RIH temperature. The cause of this increase is still not completely understood although secondary side fouling of boiler tubes is considered the prime candidate. Short term drops in pressurizer level and RIH temperature after outages, possibly due to short term removal of fouling material from the tubes during the outage transient, tend to support this assessment. Figure 6 shows the long term trends developed using REFLEX.
C N S 9th A N N U A L C O N F E R E N C E , 1988 17
d) Increased understanding of system performance during both normal and transient conditions by technical support staff; hence improved troubleshooting capability. e) Increased volume of system response information for transient review follow-up, for training, and for simulator modelling needs.
J f
•
f) Process response data for feedback to designers, safety analysts, and R&D staff. g) Information to develop procedures or to support improvement to operational procedures based on a better understanding of the process response.
Ri- "E^ERi'jRE TREND
1 j
u
yirmi
ir
(
P 1
!
i
,
;
1 i 11i 1 '
An illustration of the latter was clearly provided during PLGS's recent Annual Outage. A station control computer generated Trend was set up to monitor Shutdown Cooling System performance while the HT system was drained to header level for the Boiler Tube Leak search and repair.^ The trending of RIH and HX outlet temperatures revealed that certain channels in one loop of the HT system had reverse flow, A procedure which balanced headers by throttling specific SDC inlet and outlet valves was developed and tested during the outage. Work is ongoing to incorporate this proceiural improvement In a future revision of the Operating Manual-.
IMPLEMENTATION OF TREND ANALYSIS PROGRAM FIGUKE 6:
LONG TERM TREND (REFLEX PLOT)
This baseline data will assist in the assessment of system performance over the long term. Of current interest is the affect of the Condensate Polisher,which was installed about 2 years ago, on secondary side fouling. Trend analysis will facilitate interpretation of process performance data. BKM£FITS OF A ROUTINE TREND ANALYSIS PROGRAM The use of Trend Analysis of Routine Operation would offer the following benefits to stations: a) Early diagnosis of system component or control program faults based on process deviations outside expected or predicted operating ranges. b) Improved performance due to the optimization of maintenance routines, such as condenser cleaning. c) Optimized instrumentation calibration by adapting trends to monitor Instrumentation deviations against established calibration criteria limits. Such a system would use process Instrumentation monitoring to identify the need for calibration based on data spreads*
18 C N S 9th ANNUAL CONFERENCE, 1988
Currently regular Trend Analysis o\ process parameters during routine operations is not used by operational support staff on a wide scale, even though benefits of such a practice are generally positive. One of the reasons for the reluctance in increasing the use of Trend Analysis has been the lack of facilities to simplify the data collection and trending activities. The lack of evidence as to benefits to users and the plants Is perhaps another reason. Over the past few years hurdles to past difficulties have been overcome. The current status is that the concept of Routine Trend Analysis of Process Parameters Is being recognized as a worthwhile and reasonable program to implement, at least to some extent. Given t hat s tation management staff are convinced that the practice Is both practical and useful, the following are suggested as steps to implement a station wide Trend Analysis Program; a) Establish the use of process trending of routine operations as a normal station activity. b) Identify the key system parameters against which to monitor plant safety and performance.
c) Determine the required methodology (graph plots, trends, logs, bar charts).
Trend station
Analysis computer
d) Provide sufficient hardware and software support to facilitate data recovery and manipulation by operational staff. e) Establish a rationale and criteria for software selection. For example, data transferability to other software packages and simplicity of use should be nominal software requirements for this application. f) Encourage innovative ideas regarding the extent of process t rend analysis. For example, trending of the relative lift of feedwater valves downstream of a process pump over time could be a diagnostic method used to identify pump impellor wear. These are ^ry basic and general steps for consideration when developing such a program. Each station must rationalize further specific steps based on its requirements and expectations. CONCLUSION In summary, recent experience in the use o£ Trend Analysis at Point Lepreau has shown that there is merit to the establishment of a Trend Analysis Program for key process parameters during routine operations. With the improvement of trend analysis tools and the potential for using computerized data logging information and transmitting it to off-line personnel computers, it is now possible to increase the scope of trend analysis of plant systems beyond that of reviewing process response during significant transients. The initiation of a Trend Analysis Program for Routine Operations at stations is now practical and is a recommended good practice. Implementation of such a program by operational and technical staff should result in more optimized system performance and an enhanced preventative maintenance strategy. It might also minimize the occurrence of unplanned transients. ACKNOWELDGEMENT 1 would like to acknowledge efforts of various members of the PLUS Technical Staff, notably E. Sawatzky, B. Patterson, and A. Gillen, for their contributions and support in preparing this document. REFERENCES (1) INPO Document, (Jood Practice MA-302, "Trend Analysis", January 1982• (2) E. Sawatzky, "Pressurizer Level Operating History to End of 1987", PLCS Information Report, IR-33320-1, April 1988.
CNS 9th ANNUAL CONFERENCE, 1988 19
ONTARIO HYDRO'S LOAD FOLLOWING REQUIREMENTS, ISSUES, EXPERIENCE AND STRATEGY
A.M. LOPEZ, R. NEUMAN, C D . JIN, J. CHADA, A- DE SANTIS
Ontario Hydro Toronto, Ontario, Canada
ABSTRACT It was forecast that from 1986 to the late 1990s the nuclear plus hydraulic baseload generation capability within Ontario Hydro will frequently exceed the total system electrical demand. As a result, nuclear units will increasingly be required to meet changes in demand in order to minimize the operating costs. This paper discusses Ontario Hydro's electrical system requirements regarding nuclear generating stations and the key issues affecting their operation. It summarizes the economic impact of nuclear maneuvering. reviews the operating experience to date, and presents the long-term strategy that is being pursued. The results of our analysis indicate significant savings on fuel replacement costs due to nuclear maneuvering capability. This is achieved by eliminating the need for shutting down baseload nuclear units and then having to replace the required demand load from maneuverable but expensive coal-fired stations. Between 1986 and 1987, Pickering NGS-B and Bruce NGS-B demonstrated the nuclear maneuvering capability of CANDU" reactors and saved Ontario Hydro one million dollars in fuel costs. A typical load cycle at Bruce NGS-B consisted of reducing power to 5 0 % FP, holding at that power, then returning to full power when system required it. Deeper nuclear maneuvers, where power is reduced to 2 0 V 25% FP, have also been successfully demonstrated. A joint work program within Ontario Hydro is being pursued by Operations, Research, Design, and Systems Planning departments to anticipate potential areas of concern during the coming periods of significant unutilized energy. In particular, the potential long-term effects on equipment include a scheme to anticipate, monitor and evaluate the impact on equipment reliability and performance.
INTRODUCTION In 1987, 49.9% of Ontario Hydro's total generation of electrical energy was from 10 500 HW of installed CANDU nuclear capacity. By 1992, with the installation of 3 500 MW of additional capacity at Darlington NGS, it is expected that about 60.3% of Ontario's electrical energy will be from nuclear units.
20 CNS 9th ANNUAL CONFERENCE, 1988
Because of the increasing fraction of total electrical energy supply, Ontario Hydro 1 s nuclear units can no longer be regarded strictly as baseload generation. From 1988 through the 1990s, the nuclear plus hydraulic baseload generating capability will frequently exceed the total system electrical demand. As a result, nuclear units will increasingly be required to maneuver or shut down in order to help reduce the amount of unutilized baseload generation during times of low demand. Capabilities, constraints, and costs associated with such variable load operation are being addressed in order to operate the nuclear units in the most cost-effective manner within Ontario's bulk electrical system.
FORECAST BULK ELECTRICAL SYSTEM REQUIREMENTS Traditionally, nuclear units have been regarded as baseload generators, running continuously at maximum output. In Ontario, the other main sources of baseload energy are run-of-the-river hydraulic stations. The difference between the total power demand and total baseload supply is made up by power from fossil stations, peaking hydraulic stations Cie, where water can be stored and released later), and purchases. Occasionally, the nuclear plus hydraulic baseload capacity is more than sufficient to supply the total Ontario demand even with fossil and other resources reduced to minimum output or zero. The difference between forecast available basnload capacity and the total forecast power demand is called Unutilized Baseload Generation (UBG). This unutilized baseload generation provides a useful measure for forecasting the need to maneuver or shut down nuclear and hydraulic units. The unutilized baseload energy is expected to increase to a maximum in 1993 after the last Darlington unit is in-service, then decline as the forecast demand gradually increases. Figure 1 shows the expected and upper bound forecasts for UBG up to year 2000.
UNUTILIZED BASE LOAD ENERGY
NUMBER OF DEEP MANEUVERS (>10% FULL POWER) 2.000 •
ifloo
EXPECTED CASE UPPER BOUND CASE
1,600 1,400 1,200 1,000
400 200
TOTAL 12700 GWh EXPECTED UBO TOTAL 35200 GWh UPPER BOUND UBG
FIGURE 1
Figure 2 shows a 2-week forecast of UBG for May 1993, the forecast peak year. Up to about 5 700 MW of reduction in generation could be required on these Hay weekends, and up to about 3 800 MW on weeknights. Weeknight peaks occur from about 11 pm to 7 am. On weekends, there are blocks of unutilized energy which can be handled by 2-day shutdowns or by nuclear maneuvering. Superimposed on these blocks are early-morning 8-hour spikes of UBG.
UNUTILIZED BASE LOAD M A Y 1993
EXPECTED CASE: 5900 OEEP MANEUVERS UPPER BOUND CASE: 13700 DEEP MANEUVERS
FIGURE 3
KEK ISSUES AMD OPERATING EXPERIEHCE Load Cycle and Capability Ontario Hydro's nuclear units perform two general types of load cycle. The first type, called an economy maneuver, is a maneuver which is required to match low load demand and is performed to save nuclear fuel. The other type, called a security maneuver, is required to avoid violating system security criteria, eg, thermal, voltage or stability limits.
ZERO DAFOR (UPPER BOUND) CASE
The economy maneuver is characterized by a relatively slow ramp rate (typically 12 MW/min). Security maneuvers are also usually carried out at this slow rate, but when required may proceed at 40 to 60 MWYmin. During weekdays, generation from the nuclear units is reduced during the night and is restored to full load in the morning. On weekends, nuclear units are powered down on Friday nights and then restored to full power output levels by early Monday morning.
MS
433 HOURS OF THE MONTH
FIGURE 2
The maneuvering capability varies from station to station due to differences in design and operating constraints. For example, the Bruce A reactors are not normally used for deep maneuvers (>10% FP power reduction) to avoid the use of the expensive highly enriched U235 boosters. Furthermore, there are some concerns about delayed hydride cracking which could be worsened by nuclear maneuvering. Table 1 summarizes the current nuclear maneuvering capability for all Ontario Hydro reactors.
When the maneuvering capabilities of the different Ontario Hydro reactors are taken into account, the UBG forecasts can be converted into forecast numbers of nuclear maneuvers and shutdowns. A year-by-year forecast of deep maneuvers (greater than 1011 Full Power) is given in Figure 3 for the whole UBG period, from 1988 to year 2000.
CNS 9th ANNUAL CONFERENCE, 1988 21
TABLE 1
Most of the nuclear maneuvers occurred between March and Hay and again between August and September. During 1987, due to unseasonably low water levels and higher than expected demand load, the number of maneuvers was lower. Between 1986 and 1987, approximately 1 000 nuclear maneuvers have been performed by all Ontario Hydro reactors. Figure 5 shows the frequency distribution of nuclear maneuvers at each station.
ONTARIO HYDRO REACTORS MANEUVERING CAPABtUTV STATION
NET RUST RE fcOT DUC7X.W (MW) |IL"iM»
2ND REDUCTION (MW)
PNGS-A
515
SO
Nȣ
FWGS-B
516
90
100
COMMENTS
1) ONCE EVERY 2 DAYS ?) OELAYEO HYDRIDE CnACMNG CONCERN FOR UNITS 3 & 4 11 ONCE EVERY 2 DAYS 2] 9 HOun WAIT PERIOD AFTER 2ND neoucncw BEFORE RELOADING TOMCR
ONGS-A
75*
50
40 - 50
BNGS-0
837 360
4S0
150
1) UNITS CAN MANEUVER DAILY 2) SECONDARY REDUCTION TAKES a ro g HOURS 3) OHC CONCERNS FOR UNITS 1 « 2
ONTARIO HYDRO TOTAL MANEUVERS 1986-1987
4) PREFER NOT TO USE BOOSTERS 1> DEEP MANEUVER EVERY3 DAYS 2)8HOURWAIT BETWEEN 1ST *NP 2ND REDUCTIONS 3) FURTHER REDUCTIONS POSSIBLE WITH MORE WAtr PERIODS B81
CNSS
450
ISO
ASSUMED TOB5 THE SAME AS BRUCEB
Only nuclear maneuvers of greater than 1 0 % reactor full power are assumed capable of causing stress to a unit. Totals of deep maneuvers, together with shallow maneuvers and shutdowns, are estimated in Table 2 for the whole UBG period.
TABLE 2 FORECAST MANEUVERING REQUIREMENTS FROM 1988 TO 1999 PNGS-A
Expec t e d 12 .7 TWh
Unutilized Baseload Generation
PNGS-B
BNGS-A
BNGS-B
Upper Bound 3 5 . 2 TWh
Number of Deep Maneuvers
6 900
13 900
Number of Shallow Maneuvers
55 000 000
8 300
30
195
FIGURE 5
Economies Number of Shutdowns
Operating Experience Significant nuclear maneuvering started in early 1986. Figure 4 shows the power profile of one of the Bruce B reactors. DEEP LOAD CYCLE PERFORMED AT BRUCE-B CANDU REACTOR
The economic benefits from nuclear maneuvering arise from the differences in incremental costs between nuclear energy and hydraulic energy. The main economic benefit from nuclear maneuvering is avoided hydraulic spill. The cumulative savings associated with nuclear maneuvering is forecast to range between 63 and 171 million dollars (in 1988 dollars) from the present to the end of the century.
uou-
r
900 600 700\ 600500400-
vn 200-
v_ \
100010
12
14
1G
TIME(hr) FIGURE
22 CNS 9th ANNUAL CONFERENCE, 1988
1B
20
22
24
There is a concern that frequent and large nuclear maneuvers, or power changes, may increase cracking in the UO2 pellets. This may lead to an increase in the fission product free inventory. Since the free inventory of stable gases provides the driving force for strain failure of the fuel sheath, an increase in the fission product free inventory could potentially lead to higher radiological consequences in the event of postulated reactor accidents. Potential increase of free inventory was assessed for the Pickering NGS-B and Bruce NGS-B reactors, resulting in interim guidelines applied to nuclear maneuvering operations. To be conservative, it is assumed that the grain-boundary inventory within the UO2 volumes which exceed 1400°C during steady-state
operation is completely released into the free voids due to load maneuvering. Thus the resulting interim guidelines ensure that the free inventories of 1-131 and stable gases, including potential grain-boundary release, lu below the Ft'wts Inventory of the limiting cases in the safety submissions. Thus, it is expected that any increase in the free inventory due to power changes would be substantially lower than the bounding estimate.
F.U§A Performance The impact of nuclear- maneuvering can be assessed by reviewing the defect rate at the Bruce B reactors where most of the deep maneuvers were performed. By the end of 1987, a lifetime total of 42 180 bundles had been discharged. Of these, 33 bundles were confirmed defect (ie, defect rate = 0.08%). In 1987, following significant nuclear maneuvering, there were a total of 10 confirmed defect bundles out of 18 009 discharge bundles (ie, defect rate = 0.07%). Six of these bundles had defects caused by debris fretting, three showed sheath hydride and one revealed an outer element with a "crack" in the outboard end cap weld region. Delayed neutron scans confirmed that the defect existed shortly after the initial, insertion of the bundle into the core. Several bundles that experienced high powers during the cycling period were sent to AECL-CRNL for examination. For comparison, fuel bundles of similar burnups and power histories that were discharged from the reactor prior to nuclear maneuvering were also examined. The results indicated no significant, difference between the characteristics of these bundles. This confirmed that nuclear maneuvering did not increase the defect rate of the fuel bundles.
?_La..Dt Operation Ontario Hydro's experience indicated no major change in its operating practices or procedures due to nuclear maneuvering. Minor adjustments to smooth out operating practices are given below. First, a more systematic and coordinated communication between system control staff and plant operating staff was impLemented. Prior to anticipated nuclear maneuvering requirements, system staff and plant operating staff ensure that they are aware of t.he latest nuclear maneuvering forecasts and station capability. This ensures that System Operations are fully aware of current operating limitations or restrictions of the nuclear plants. On the other hand, the plant operating staff are aware of the expected power changes that may be required. Maintenance plans are arranged to fit within the overall operating plan. Secondly, in the area of fuel scheduling, it was necessary to select channels for fuelling more cautiously. The calculated power distribution has to be compared more carefully with the fuel channel temperature measurements. For the Bruce B reactors, an existing transient code based on flux mapping techniques has been enhanced and is currently being validated. This code would potentially remove the need for steady-state operation between nuclear maneuvering.
Thirdly, the calibration of the safety systems has to be re- evaluated. Following a maneuver, due to local flux transients, one or two of the safety system in-core detectors usually yield high uncalibrated detector readings. This results in reducing the operating margin to trip. This has no impact during the power reduction. However, during the recovery to full power, operations staff have to monitor these detectors more closely to prevent any spurious trips, This limits the rate of recovery to full power. To determine an accurate calculation of safety system calibration, it is also necessary to operate the units in a steady- state mode at least twice a week.
Effects on Long Term Equipment Reliabjlit_y_ Due to Iimited nuclear maneuvering experience to date, the impact on long-term effects on equipment cannot be fully assessed. Based on 2 years of nuclear maneuvering, there are no indications of degradation resulting from this mode of operation. To have a better assessment of the impact of nuclear maneuvering on equipment, a controlled test maneuver was performed at Bruce B. The purpose of the test was to obtain actual measured data on all affected systems that can be used to: (1) compare with design parameters to confirm that design envelopes are not exceeded, and (2) obtain parameters that indicate any potential problems resulting from maneuvering. These would be used to define supporting research programs. Preliminary results for the reactor vessel, fuel channels, inlet and outlet headers, feeder pipes, steam generators, and boiler inlets indicate that, relative to measured changes in temperature and pressure, nuclear maneuvering should have no significant impact. The original design specification for Bruce B allowed 10 000 power cycles in which pressure changed by 100 psi and temperature changed by 28.8°C in k minutes. The actual pressure change was 30-40 psi while the temperature change was a maximum of 30°C. The rate of change of pressure and temperature is of no concern as they are well within the 4 minute ramp specified in the design. Pressure tube concerns such as delayed hydride cracking and blistering, both affected by deuterium ingress, are being extensively studied under the CANDU Owners Group (COG) programs. Nuclear maneuvering concerns wi 11 be taken into account in the COG program. Power maneuvering does not affect the calandria parameters. The cover gas pressure varied by less than 1.2 psi. There seemed to be some correlation between dew point and power level, but there was no additional ingress of moisture during the maneuver. Rates are considerably less than the computer alarm setpoints and will not result in spurious alarms. Compared with current operation specifications and normal operation, the chemistry parameters measured were within those limits. The main problem in trying to assess the chemistry concerns is the lack of appropriate and sufficient samples.
CNS 9th ANNUAL CONFERENCE, 1988 23
WORK PROGRAM Although the current experience to date indicates no technical problem with nuclear maneuvering, there are three main issues that require further work before the peak UBG period when daily maneuvers are expected. First, there is a need for an improved and accurate transient simulation code to support fuel scheduling and calibration of the safety systems. Currently, fuel channel selection is based on a steady-state code. During maneuvers, this code is no longer accurate and it is more difficult to select channels such that potential long-term penalties are not incurred in the overall discharge burnup. The flux mapping code, currently being validated at Bruce B, looks very promising. However, since other reactors are not equipped with flux mapping detectors, an alternative mode of simulation has to be developed and validated. Secondly, the generic solution to the source term concerns has to be derived. The interim guidelines adopted for Bruce B and Pickering B may not be acceptable when daily load cycles are performed without significant derating. A program to provide the generic solution is being studied and involves getting actual data on high power bundles undergoing maneuvering. Thirdly, long-term affects on equipment reliability have to be resolved. As already noted, the first phase is to perform controlled test manc-uvers at the different nuclear stations to obtain operating data that can be compared with design specifications. Preliminary analysis of the Bruce B test data shows that the maneuvering transients fall within the design envelope. To assist in detecting any degradation which has not been foreseen in the design, a monitoring program will be implemented.
CONCLUSION Due to the increasing contribution of nuclear generation to the total Ontario Hydro installed capacity, it is forecast that there will he significant unutilized baseload generation during periods of low demand for the next decade. To meet the requirements of the bulk electrical system, it is necessary to perform a significant number of power maneuvers with the nuclear units. Experience between 1986 and 198 7 proved that these nuclear units can be maneuvered to meet system demand without significant operating problems. Compared with actual operating data, analysis confirmed that nuclear maneuvering is within the design specifications. Based on this observation, it is expected that nuclear maneuvering will not cause additional stress that will impact on long-term equipment reliability. However, due to unidentified degradation mechanisms, a prudent action is being planned to establish a monitoring program to assist in anticipating potential problems dJe to nuclear maneuvering.
24 CNS 9th ANNUAL CONFERENCE, 1988
FUELEM: A MICROCOMPUTER PROGRAM TO AUTOMATICALLY SELECT CHANNELS FOR REFUELLING
B. ROUBEN, D.A. JENKINS AND C.R. CALABRESE*
Atomic Energy of Canada Limited CANDU Operations Sheridan Park Research Community Mississauga, Ontario, L5K 1B2
ABSTRACT FUELEM is a computer program to assist Che fuelling engineer at a CANDU reactor in the selection of channels for refuelling. The processing performed by the program should reduce the human time devoted to this task. FUELEM is easy to use and quite versatile, It takes a negligible amount of execution time on a microcomputer. It can easily be further customized to a user's specifications. The code is described and the algorithm and rules which it uses are explained. Examples of channel selections are given for illustration.
INTRODUCTION We have developed FUELEM, a computer program to assist the fuelling engineer at a CANDU reactor in the selection of channels for refuelling within the next few days of operation. At present, this part of the fuelling engineer's :esponsibility requires his/her experience and human judgment. It is fairly time consuming, as it involves the close scrutiny of data such as power and burnup distributions, refuelling rates, zonal powers, zone-compartment water fills, recent refuelling history, etc. The idea underlying the development of FUELEM is that the judgment exercised by the fuelling engineer is amenable to being programmed on a computer. The processing performed by FUELEM should reduce the human time devoted to the task of channel selection.
artificial-intelligence medium will be evaluated later. We do feel that the use of FORTRAN 77 lends a high degree of portability to the code at this stage. It also enables the program to be installed on basically any computer, be it mainframe, mini, or micro. The first version of FUELEM has been designed for, and tested against, the case of a CANDU Model 6 with standard bi-directional eight-bundle-shift refuelling. However, the principle of the program is not restricted in generality or applicability.
PREREQUISITES FUELEM must have available to it, as input, files created by the fuel management code used to track the detailed operating history of the reactor. The code designed for this purpose by Atomic Energy of Canada Limited is RFSP (Reactor Fuelling Simulation Program). The computer files in question contain essentially the same information normally used by the fuelling engineer to establish a channel selection. The data transferred is of two types. The first part of the information consists of a "snapshot" of the reactor at the time when the channel selection is desired, with data such as: *
instantaneous power distribution
*
instantaneous burnup distribution
*
instantaneous zone-controller water fills
*
expected channel
*
value of Full Power refuelling, by channel.
reactivity
gain
on
re fuel1 ing, by
FRAMEWORK FUELEM is written in FORTRAN 77. It was developed and installed on an APOLLO workstation for convenience. Execution of the program on a microcomputer such as the APOLLO is especially important at the development stage, where numerous program features and different algorithms can be tested quickly, easily, and at low computer cost. While this type of program may be a good candidate for the application of an artificial-intelligence language such as LISP or PROLOG, FORTRAN 77 was used as the development tool. The j.racticality of writing the "final" production version of t h e p r o g r a m in an On attachment from Comision Nacional de Energia Atomica, Direcci6n de Centrales Nucleares, Gerencia de Ingenieria, Argentina
Day
(FPD)
of
last
The second part of the information consists of the desired or target long term time-average core picture, with the following data: *
power distribution with time-average lattice cross sections
*
target fuel exit burnup, by channel
*
target zone-controller water fills
*
target frequency of refuelling, by channel, or, equivalently, target dwell time (interval between successive refuellings), by channel.
CNS 9th ANNUAL CONFERENCE, 1988 25
ALGORITHM The algorithm in FUELEM consists of two distinct logical steps: 1. eliminate from further consideration channels which are deemed unsuitable for refuelling at the present time, ana? 2. establish short lists of channels recommended for refuelling within the next few days of operation. The user may be interested in the first step only, obtaining from the program lists of non-eliminated channels by zone number, thereafter personally narrowing th$ choice to a few channels in each zone. Or the user may wish to continue program execution in order to inspect the short selections of channels recommended by FUELEM. These two steps are now described in turn in the following subsections.
Channel Elimination Channels considered unsuitable for refuelling at the current time are eliminated according to the following set of rules: *
channels with power above a specified value (using different limits in different regions of the core)
*
channels with exit burnup below a specified fraction of their target exit burnup
*
channels which have been refuelled less than a specified number of FPD before the current time
*
channels which are neighbours to one or more channels refuelled less than a specified number of FPD earlier
*
channels which ate neighbours to channels with a power above a specified value
In addition, the user can indicate by input specific channels that he/she wishes to eliminate outright for any reason (most commonly, an operational constraint).
selected for refuelling, is an input specified by the user. At present, up to 14 channels may be requested. For a CANDU Model 6 reactor, this is sufficient for about one full power week of operation. The current limit of H channels is arbitrary and is not difficult to increase to, say, 20 channels. As a preliminary step, FUELEM sorts the candidate channels not eliminated according to the governing rules. Sorts are performe' by zone number (see Fig. 1 ) , by exit burnup, by reactivity gain on r e f u e l l i n g , and by d i r e c t i o n of refuelling. FUELEM then decides how many of the channels will be picked from each zone, and the "picking order" from the lists above. In this decision the program takes into account the total number of channels to be selected, the differences between instantaneous and time-average zone compartment fills, and the relative values of water fill in axially opposite compartments. As a rule, more channels will be selected in zones where the water levels are particularly low relative to the time-average values. Also, the direction of refuelling is chosen to reduce axial differences in water level. Channels are thus selected one at a time from the zones and lists of candidates identified. Each time a channel is selected, it eliminates some more channels from consideration, according to the neighbour-related rules listed earlier. In this way, a complete channel selection totalling the number of channels as requested by the user is established and memorized. The process however does not stop there. In practice many different channel selections are possible at a given FPD of operation. FUELEM therefore goes beyond the first selection to make other possible gelections. The program Execution time per •election being negligible - a. small fraction of a second a large number (maximum = 100) of possible choices are actually established. These are then sorted by total exit fuel burnup achieved as well as by total reactivity gain. All selections are a v a i l a b l e for d i s p l a y and inspection.
The "specified" quantities used in the above criteria are given built-in default values in the program; these however may be overridden by the user. For the purpose of eliminating neighbours, a distinction is made between the closest neighbours of a channel (the 4 channels immediately adjacent to it vertically and horizontally) and its second closest neighbours (the A channels diagonally adjacent to it). In the high-power central region of the core, both the closest and second closest neighbours of high-power or recently refuelled channels are eliminated. In contrast, in the lower-power outer region of the core, only the 4 nearest neighbours would be eliminated.
Channel Selection After having pruned the core of unsuitable channels, FUELEM establishes its short lists of channels recommended for refuelling. The length of the list, i.e., the total number of channels 26 C N S 9th A N N U A L CONFERENCE, 1988
FIGURE 1
SUBDIVISION OF CORE INTO 14 CONTROL ZONES
OVERVIEW
FUELEM can thus be used either in the roie of decision maker or in the role of assistant. In any case its results are subject to human monitoring, ver i ficar.ion , or j udgement.
The algorithm described in the previous section enables FUELEM to deliver a number of different channel selections for refuelling at any value of the energy clock (FPD of operation).
EXAMPLES
FUELEM will make two recommended selections: one which maximizes fuel exit burnup, and one which maximizes total reactivity gain. The fuelling engineer may choose either, depending on his/her overriding concern at that particular point in time.
Two examples are illustrated here. Bein are based on snapshots in the past operating history of the Point Lepreau reactor. The first example is at FPD 1667. The instantaneous zone-compartment water fills and zone powers are shown in Figure 2. FUELEM was requested to give a 14-channel selection. Figure 3 illustrates the channels remaining as candidates for refuelling after the first process of elimination, while Figure 4 illustrates the reasons for ignoring the eliminated channels. Figure 5 shovs the final selection recommended by FUELEM from the point of view of maximizing burnup. In this figure the channels selected are also numbered in the order suggested by the program for
However, the fuelling engineer is not; held to this or that particular choic?. If either recommended selection is not judged to be advisable tor some reason, alternative selections can be displayed to find a preferable choice. Also, as pointed out earlier, the user may wish to stop after the first step in the algorithm, estat-ishing 'he short list from the remaining candidate channels by personal judgement.
19
FIGURE 2
20
21
22
ZONE POWERS AND WATER LEVELS FOR CASE 1 MA? OF CANDIDATES
4
5
c
/
W * '
. -C
'
.
-C
9 -C
10 .
U .
-c
+£
-c
J
•L
-t
•c
-f
+0
1i .
1A . -C
1 ':•
1o ' ' .
. -C
!+C
-c
H
12 +C
'
17
)y
1 •<
^^
_I
^ ,1
.
-c
.
-i
.
-c
.
'C
K 1
i
-L
-C
M N I'I
1
•r
-c -r
-f
+c
.
-L
fC
.
-,-
.
F 'J R
• C
-c . .
-C
.
.
-C
FIGURE 3
. .
+1 -t +C
+C
'-C
+C -C
CANDIDATE CHANNELS FOR CASE 1 CNS 9th ANNUAL CONFERENCE. 1988
27
WAP OF tLlNrNAIED
9 . !LE LE ! . LB N RP N ' LE LB N . . N N N N N HF N RK N 'HP N
LE
Lb LB
J
L
LE Lb LE
N N
LB
K
Lb
LE
LB
LE
LB LB
LE LB
Lt N
N N
N
HP
LE N . LE . LE LE LE
N N N N
0 Lb
N LB N
Lt
.
N N
'
N
N
RR 1
14
14
N
N '
N
N
N I IJ N ' 14 HP ! 14
HP
HP HP
1* 11 LB LE N N N RR IJ W L6 L B N N HF N N N N
LE
LE
LE i L B LE !LE
LB
12 13 14 . LE L b N N
LB LB N N N HF N LB
15 li 17 ' LE !LE LE . 'LE LE RR ' 14 LE N ' N H !HP
18
Lt . b .
ii
Zi 21
LE
LE LE LE
IJ
N
N
14
IJ
Lt
N
N
RR
N
!HP
RR
N
i-B
. .
LB
LB
N
IJ
N
14 HP
N
LE
HP
'HP
HP
14
14
LE
LE
LE
N N
HP N
N N
HP HF
I HP
HP N
HP 14
N N
Lt
Lt
'HP
LE
N
HP
!HP
14
HP
14
RR
14
N HP
N
HF
14
LE N
N 14 ' N N HP N HF HP Rh N HP 14 N Lt
CHANNELS
N N HF N 14 N RR N N LB
I-B
HP ! N
14 N 'Hf N HP 'HF N 14 ! N N . ' . N Lb ' L E N LB ' L E
14
14
ut Lt
14 _b
IJ LB
Lb
LE
LE Lt
LE-
LE
HP = HIGH-POWER CHANNEL
RP = RECENTLY REFUELLED AND HIGH-POWER CHANNEL OUTSIDE CPPF REGION
RR = RECE14TLY REFUELLED CHANNEL
LB = LOW-BURNUP CHANNEL N
FIGURE 4
= NEIGHBOUR TO A HIGH-POWER OR RECENTLY REFUELLED CHANNEL
CHANNELS ELIMINATED IN CASE 1
l«
CHANNELS SELECTED 11 12 1.3 14 15
l»
17
IB
19 : o
p
u R
T I
3 1 1
V
.
J
•
1 BV AVERAGE EXIT 8URHUP REACTIVITY CHANNEL EXIT 8URNUP TIME SINCE LAST POMER 6AJN VALUE RATIO TO REFUELLING TARGET (BICRO-K) (KM) 123.3 5257.2 2 • 8.8449 144.6 145.6 3 • 161.6 6112.1 146.1 «.873» 132.8 I 183.3 5281.7 178.9 8.9978 176.8 4 285.6 5811.0 211.3 296.9 .8835 6 299.6 4765.3 234.2 .3673 300.9 7 + 118.6 5182.2 143.5 9.8363 153.7 5 62.0 3887.5 135.3 ).9478 296.3 2 1*7.1 3155.• 198.4 .3867 381.9 Z68.4 5916.2 199.9 3 .•476 192.1 1 * 1«3.2 3641.6 176.7 .»289 279.1 4 5714.9 282.1 215.9 .1275 219.2 5569.3 6 + S91.2 165.5 17».Z 0.9655 7 173.5 5573.3 158.9 8.9278 159.3 5 * 167.1 5207.1 165.6 •.9663 177.2 REACTIVITY GAIN 2526.* MICRO-K AVERAGE EXIT SURNUF 1/6.1 NUH/KGU SELECTION NUMBER
CHANNEL ZONE NO.
0*3 E12 LQ3 012 K2» R18 V«9 U*6 G13 E64 L«9 F17 316 T98 TOTAL
FH6LLINS DIRECTION
FIGURE 5 28
CNS 9th ANNUAL CONFERENCE. 198B
FINAL SELECTION FOR CASE 1
DMtU TINE RA1I0 TO TARGET 0.7545 0.8221 0.9769 1.1159 1.6176 0.7587 «.9616 1.620* 1.0659 1.0447 1.1558 •.957* •.8968 0.9741
b. for purposes of executing the above examples, operational constraints were not considered in FUELEM, and
refuelling. The average exit burnup in these 14 channels is 176.1 MVJ.h/kg(U), quite a good value. The second example is at FPD 17
c. even after the step of eliminating channels, the nu.nber of combinations of channels remaining is still quite large. Notwithstanding these cautions, the FUELEM selections do show a fair degree of similarity to the choices performed at the station. In the first example, six of the channels selected by FUELEM were actually refuelled within 15 FPD, while two could apparently not be refuelled on account of constraints. In the second example, four of the eight channels selected by FUELEM were actually refuelled within 5 FPD and a fifth was refuelled within 13 FPD. Considering the differences in procedure mentioned above, the similarities in results do indicate that the FUELEM selections are quite reasonable.
Direct comparisons of the FUELEM selections with i'.'tual choices of channels performed at the station are not truly pertinent because: a. rules and limit values in use at the station Cor eliminating channels may be different from those in FUELEM,
10
FIGURE 6
11
12
13
14
15
16
17
IB
ZONE POWERS AND WATER LEVELS FOR CASE 2
MAP OF CANDIDATES 1
Z
4
"J
o
.
. *C
' +C ' . . I
.
A B C
li t r b H J K
-C .
M N 0
+C .
-C
. .
7
? . *C
hi . .
K I .
.
.
•£
-C
.
.
. ! .
. .
. -C
-C
.
.
.
.
-C .
. -C
. . .
. . *C
-C . . +C
.
-C
.
.
.
b
i . '-C I I
11 -C .
.
11 l i +C. . + ( .
i I
14 15 l o . I . . l-C + 1 . 1
-C
.
17
IK
l'p
2.J
21
Li
. .
.
.
.
-C•
.
i ! . ! . !
. .
. *C
+L -C
.
! . !-C
-C .
. -I
-C *C
. -I
.
+C
-L
-C
*C .
-C
.
-C
. ' . '-C . ' . I
FIGURE 7
. . +C
-C +C .
. K
.
.
.
-C
+C
.
i+C »C ' . I . i
CANDIDATE CHANNELS FOR CASE 2 CNS 9th ANNUAL CONFERENCE, 1988
29
MAP UF tUNIt'Mrfc'D CH/MJNELS
9 i-E LB N LE LE N LE LE-
LB
LB LB
US uE LE
N
LB N N RF IJ LB
i_B
W fik N N IJ IJ
LE
. [ LE
is LE N N N
N RP N N N
LB IJ i.E LB
! IJ
Rfi N LB N R'R
N LB N
I.S
N '. N N 'HP HP !HF
HP IJ N
HF N
HF N IJ
N HP N
N N N
•-B IJ HF
,' N ' '
LE t-B
IJ N
> IJ
HF HP N
N LB N HP fcp N N N . LB L£
N .
! N ! IJ !
*
•
N 'HF HP N I. H N W N
LE' LE i_B
LE ' L E LE
FIGURE 8
H IJ
11
12
i;
17
uE L E uB LB LB LE H LB N LB W IJ IJ HF HP HF N
LE IJ HF N HF
N !Rh IJ 1 IJ N W. HF ! N IJ ' IJ
N N N IJ LE
IJ IJ N
IJ LE
IJ !LE uE i . !LE
LB
IJ N M IJ
IJ RR IJ
HI-'
IJ N N N N
RR IJ
N I M LE <-b N N HP H i-fc N N N N N N
H L-b
LE
IJ M N
H N
LE
LB
N N N
uE LE LE
' IJ > IJ !
N
13
Lb tfc i-E
JS
2.5
Lb L.B
i-b
^B
Lt uB uE
LE
21
ZZ
^b LE
LE i_E
i-L l_b
M
W
IJ
:.E
LE
14 IJ HP HF HP IJ
HF N HF
N N N
•E
LE i-k
1.6
LE
HF ' M N N ' N U LB ! N RR LB ! N N ) , (_£ 'LE uE L E 'LE LE
N N N
N W . i_t . LE
CHANNELS ELIMINATED IN CASE 2
CHANNELS SELECTED
1
Z
3
4
5
o
7
8
9
18
11
12
13
M
15
7
K14 H»3 T16 E»7 GI8 Oil 314 LI*
4 2 7 1 6 3 5 4
FUELLING DIRECTION
* • • + *
TOTAL REACTIVITY RAIN •
REACTIVITY GAIN (HICRO-K) 256.7 171.6 148.2 219.7 128.5 22*. 1 2*5.1 1S4.9
CHANNEL POMEf? (KU) S917.1
5555.9 5*76.1 5259.1 5554.9 5591.7 6*31.8 6*17.3
1 5 1 4 . 7 HICRO-K
FIGURE 9 30
CNS 9th ANNUAL CONFERENCE, 1988
17
IS
19
Zi
21
. !.
1 BV TOTAL R C i t C n v i r v G«1N
SELECTION NUMBER UUJfJEl. ZONE NO.
lo
EXIT VALUE
BURNUP RATIO TO TARGET
192.2 155.1 156.2 184.1 14*.2 176.6 162.9 168.1
I.M36 *.9*S9 •.922* l.*741 *.8179 l.*546 •.9527 9.8781
AVERAGE EXIT BURNUP>
TINE SINCE LAST REFUELLING (FPO) 188.8 158.8 175.8 2*5.5 134.8 179.7 148.4 134.S
1 6 / . Z MNH/KGU
FINAL SELECTION FOR CASE 2
WELL TINE RATIO TO TARGET 1 .*4*1 0.8816 ».S83* 1.1661 *.7358 1.M44 «.929S e.7319
SUMMARY AND DISCUSSION FUELEM is an operational tool designed to assist the fuelling engineer in making a good selection of channels for refuelling. It takes a negligible amount of execution time on a microcomputer work station. It is a highly portable program. FUELEM has been tested against realistic situations, using actual operating data from Pt. Lepreau. In these cases it has been found to recommend sensible selections of channels. The program is versatile, and can easily be customized to a user's specifications. Default values of quantities used in the process of channel elimination can easily be changed. Rules can be
changed, and required.
new
rules
can
be
incorporated
as
The application and field testing of FUELEM over an interval of operating time is the next step in the development of the program.
ACKNOWLEDGEMENTS The authors would like to acknowledge the availability of files of fuel-management operating data from Pt. Lepreau Generating Station. Thanks are due to R. A. Gibb and E. Young o f the New Brunswick Electric Power Commission for very useful discussion and criticism.
CNS 9th ANNUAL CONFERENCE, 1988 31
AN OPPORTUNITY FOR INEXPENSIVE NUCLEAR POWER
J. L. BAGSHAW C. J. BROMLEY
Ontario Hydro, Box 4000 Tiverton, Ontario
TABLE 1:
ABSTRACT The work to uprate the Bruce Nuclear Generating Station B Units from 845 MWe to more than 935 MWe is described. The benefits of this opportunity for inexpensive nuclear power are reviewed.
UNIT 5 UPRATING HISTORY
DATE
MWe
BOILER PRESSURE kPa(E)
Original Design
-
807
4275
88
In Service MCR
Mar 85
845
4185*
91
Phase I (Limited) HCR
Jul 85
865
4275
93
Phase I MCR
Jan 86
885
4395
95
Phase II
Dec 87
940
4645
100
REACTOR POWER %
INTRODUCTION The Bruce reactors were designed and built to be able to provide steam to the turbine/generators and to steam transformer plants to provide energy for heavy water production. It was decided that a steam transformer plant associated with Bruce Nuclear Generator station "B" was not necessary due to the cancellation of Heavy Water Plants. This meant that the reactors were relatively oversized compared to the turbine/generators and if a way to utilize this available energy through the turbine/generators could be found, then there was a real opportunity for inexpensive nuclear power. The original design had the turbine/generator rated at 807 MWe at 88% reactor power (RP). During the initial commissioning an output of 845 MWe was achieved due to the design flow margin in the turbine/generators provided by the turbine/generator manufacturer with the "Control Valves Wide Open". Two ways of producing more energy from the turbine/generators were studied in 19B1 to achieve 96% reactor power or more. These were to increase the amount of steam that could pass through the turbine (maintaining constant inlet pressure) by removing the second stage HP blades or by increasing the boiler steam pressure by 5%. In 1984 it was agreed to uprate the units by increasing the boiler steam pressure by 5%, called Phase I conditions, rather than the removal of the second stage blades because the increased pressure option maintains a better overall efficiency. This option maintains the capability to return to the original conditions and was less expensive than the blade removal option.
*Boiler pressure was reduced to match the T/G design inlet pressure. The pressure drop in the main steam lines was less than the design value and this caused the original T/G inlet pressure to be exceeded.
PLANT CHANGES Phase I In late initiated to of the unit pressure.
1984 a formal design study was investigate the uprating capability with a 5% increase in main steam
The turbine generator (T/G) manufacturer carried out a preliminary study and produced a turbine heat balance. This indicated a Valves Wide Open (VWO) output of 881 MWe (980 MVA). Reactor power was predicted to be 96% FP. Nuclear Design Study
In 1986 it was decided to conduct a test on Unit 6 with a further increase in boiler steam pressure to confirm that the steam from 100% reactor power could be handled by the turbine/generator. Following this successful test the modifications essential for operation at "Phase II" conditions were identified and implemented in Unit 5 during the outage in late 1987. Unit 5 then was run for a successful two month shakedown prior to declaring it "In-Service" at these new uprated levels on March 1, 1988. The history of Unit 5 Updating is shown in Table 1.
32 CNS 9th ANNUAL CONFERENCE, 19B8
The following areas were investigated and analyzed for possible impact on the Final Safety Report. The analysis was carried out for 98% FP to provide a 2% operational margin. Reactor Inner Zone Inlet Header Temperature (RIZIH). The increased secondary side pressure results in increased primary heat transport system temperatures. The increased inner zone fuel channel inlet temperatures would cause earlier channel dryout conditions during analyzed reactor
overpower events. The increase was predicted to be 2°C which could be accommodated within the initial safety analysis assumptions. Critical Channel Power (CCF). The CCP was further limited by 0.14 MWe. This limitation was offset by an updated safety analysis which provided more margin with no reduction in Neutron Over Power (NOP) trip setFoints. Reactor Trip Setpoints. The increased boiler pressure resulted in reduced trip setpoint tolerances on heat transport low pressure and pressurizer low level trips. These tolerances were considered acceptable. Pipe Breaks. There was no significant effect on the powerhouse or containment transients for postulated steam line or heat transport system pipe breaks.
Main Steam Safety Valves (MSSV's). The marg in between set and operating pressure would be reduced to just below ASME recommendations. There may be difficulties in resetting the values after operation if the blowdowns were greater than 5%. Phase I Updating Test Unit Ub was tested in March, 1985 but problems were immediately encountered with passing M S S V s. The maximum unit output was 892 Mwe at VWO and 95% FP. A heat rate test indi ^-ted an efficiency which was better than the T/G manufacturer's prediction. Several transient tests were performed, including a full load rejection, which indicated satisfactory unit response. Other tests included single exciter operation, single bank feedwater heater operation and a back pressure test.
Conventional Design Studies A formal design review was conducted by the T/G manufatturer to determine the ability of the turbine u^ operate at the uprated condition. Many facets of turbine generator design required study to determine their ability to function efficiently and reliably under the proposed conditions. The study examined the following areas: High Pressure (HP) Turbine. The effect of increased pressure drop across the stages was analyzed to determine the effect on stress levels, erosion protection and moisture removal.
At the end of the test, boiler steam operating pressure was reduced to 4275 kPa(g). A design study was initiated to increase the design pressure of the secondary systems to allow an increase in the setting of the MSSV's. Increased Design Pressure A total of 13 secondary side systems required detailed analysis and documentation changes to support the increase in design pressure from 4685 kPa(g) to 4892 kPa(g). Hardware changes included:
Low Pressure (LP) Turbine. An analysis similar to the H. P. turbine was performed on the three LP turbines. The effect of changes in extraction flows was also considered. Couplings. The effect of increased operating loads was checked to ensure stresses were acceptable on the coupling assemblies. Moisture Separator Reheaters (HSR* s ) . The effect of increased flows was checked to ensure efficiency, tube vibrations and steam distribution were acceptable. Relief valve design parameters and piping expansions were also checked. Turbine Control• Loading calculations, were examined.
rates
and
ovprspeed
Generator, The effects of increased output was checked to ensure insulation life, vibrations, auxiliaries, exciters and brushgear parameters were within expected design limits. The s tat or water flow rate was increased to maintain core temperatures at the same value as the original design. This was the only change required on the turbine/generator. Boiler Feed and Condensate Systems. All systems were analyzed to ensure that pumping capacities were adequate. The feedwater heaters and the main condenser were analyzed to ensure that tube bundle vibrations were acceptable. All systems were considered acceptable.
a) replacement of all MSSV springs so that the valve settings could be increased without over-compression of the springs. b) replacement of two small manual isolating valves on the high pressure feedwater heater vents and two flow orifice flanges on the high pressure feedwater heater drains. c) replacement of the level control valves on the high pressure feedheater drains. Following Ministry of Consumer and Commercial Relations (MCCR) and Atomic Energy Control Board CAECB) approvals, these changes were made and all units were uprated to Phase I conditions. Unit 5 was operated at these Phase I conditions during the record run of 4 75 continuous electrical generation. The factor during this record run was 97.3% capability factor was 99.7%.
uprated days of capacity and the
In November 1985, Bruce Engineering Department were requested to initiate a study to determine if additional uprating capability to 100% was possible by a further increase in operating steam pressure. This increase in operating pressure would again reduce the operating to set pressure margin on the MSSV's. This was considered acceptable because the MSSV's had exhibited improved performance following the Phase I spring replacement.
CNS 9th ANNUAL CONFERENCE, 1968 33
This study concluded that it was possible to increase steam pressure. The T/G manufacturer conservatively predicted an output of 921 Hwe (1025 HVA) at 100% RP. Nuclear Design Study Primary Heat Transport System CFHTS)• Based on experience at Bruce NGS"A", there was concern that increased PHT5 boiling, predicted to be 1% quality, could result in operating instability. This instability occurs due to the positive reactivity effects as channel boiling increases. Small changes in the amount of boiling causes reactor power to change which causes steam pressure and heat transport pressure to change. This cycie then repeats. Two design changes were made to reduce the effects of this instability. a) Increase PUTS operating pressure by 100 kPa(g) which would reduce the amount of boiling. b) Install a computerized pressurizer level control system which could counteract the instability. This computerized control system monitored boiler pressure, PHTS pressure„ reactor power and pressurizer level, and used appropriate feed forward terras to ensure stable reactor operation.
study. This indicated that the turbine generator could accommodate another 5% increase in steam pressure and provide a 5% increase in output. Critical areas such as HP blading loading, rotor stresses and moisture separator performance were reviewed to ensure the higher steam pressure and forces would not extend the machine beyond the normal experience values. This study indicated that the following changes were required; a) The MSR relief valve settings were required to be increased to provide additional transient operating margin. Additional relief value capacity was also required and four new relief valves were added to the MSR's. b) Revised electro-hydraulic were required.
control calibrations
c) increased stator water flows, which required larger pump impellers and motors, were proposed to maintain core temperatures at original design values. This change was not made because of concerns that the increased flows would cause erosion in the stator bars and the fact that subsequent testing indicated acceptable stator temperatures. Boiler Feed and Condensate Systems
The new pressurizer level control system also provided a modified pressurizer level setpoint control line. The original setpoint control line ramped up pressurizer level on a straight line basis as reactor power increased. The modified control line included a steeper ramp section above 96% FP. This accommodated the increased PHTS swell induced by boiLing. This action ensured that sufficient inventory was available in the pressurizer to accommodate reactor transients without causing a reactor trip on pressurizer low level. The stress and fatigue design requirements for the strain generators, preheaters, reactor headers, PHTS piping and pumps were reviewed and met the ASME code requirements. The fuel channels were also reviewed and found to be acceptable for the increased PHTS operating pressure. RIZIH Temperature. The increase in secondary side pressure was predicted to result in a further increase in inner zone inlet temperature of 2.5°C to 254*0. This was vGtry close to the safety analysis limit of 255*"C, Safety Analysis. A safety analysis for 103* FP was performed. This indicated that small changes were required to Shutdown System #1 CSDSl) and to Shutdown System //2 (SDS2) PHTS low pressure trips and to SDS2 low core differential pressure trips.
Boiler Feed. All three 50% main pumps were required to provide adequate flow to the boilers. Load reductions to Phase I output were required if only one of the two high pressure feedwater (HPFW) heaters was in service to respect tube bundle vibration concerns. Condensate. All three 50% main pumps were required to provide sufficient pressure drop across the deaerator level control valves for stable level control response. Both 100% level control valves were required. Electrical Systems Main Output Transformer (MOT). The MOT was considered capable of 1025 MVA generator output provided ambient temperatures were less than 19°C. Isolated Phase Bus (IPB) and Generator Current Transformers. The IPB design was reviewed and was found to be adequate. The generator current transformers were experiencing high surface temperatures at Phase I conditions and problems were predicted for continuous operation at 1025 MVA.
To support the safety Analysis, the maximum channel power was limited to 7.2 MWe and maximum bundle power was limited to 995 KW.
Environmental Limitations. The cooling water thermal discharges were checked to ensure that the Ministry of Environment limits were not exceeded. Design calculations indicated that both the maximum cooling water temperature rise across the station (limit of 11°C), and the maximum allowable temperature (limit of 32*C) would not be exceeded.
Conventional Design Study
Phase II Initial iterating Test - Unit 6
Turbine Generator. The T/G manufacturer performed a design review, which was similar to the Phase I
In order to confirm the design study parameters, an uprating test was proposed and was
34 C N S 9th /'•SINUAL CONFERENCE. 1988
approved by the MCCR and AECB. This one week test was conducted in Unit 6 in November, 1986.
Kain Transformer. Winding temperature taken to confirm design expectations.
The revised pressurizer level control system was not installed for the test. Transient tests (e.g. load rejections) were thus, not performed because reactor trips could result.
Delayed Neutron Scan. A scan was performed to ensure there were no fuel defects as a result of the power increase to 1007c. FP. No defects were observed.
The following design data:
HPFW Heater vibration Data. Acoustic data was collected to ensure that tube bundle vibrations were not increasing due to the increased flows, No changes were observed.
tests
were
conducted
to
verify
Maximum MVA TEST. The generator was loaded to its maximum (lag) rating of 1025 MVA to check core temperatures. All temperatures were acceptable. Heat Transport Stability Test. The heat transport system was "disturbed" to monitor for oscillations. Some minor oscillations were noted. Moisture Carryover Test. A carryover test was performed to ensure that the higher steam flows did not result in increased carryover. The results were well below design values. Final Feedwater Reduction Test. Extraction steam flows were throttled to the HPFW heaters to observe the effects on reactor inlet header temperatures due to reduced feedwater temperatures to the preheaters. This was done due to the concerns with the tight reactor safety report margins on RIZIH temperatures. The R1ZIH temperatures could be reduced but this resulted in an unacceptable overall reduction in unit efficiency. Turbine Generator Capability Test. Two tests were conducted by the Ontario Hydro Thermal Test Group. Accurate test equipment was installed to monitor generator output and various pressures on the turbine systems. The first test was performed at maximum output (turbine control valves wide open) at Phase I conditions. The second test was performed at maximum output at Phase II conditions. A maximum output of 934 MVJe gross was obtained at 99.5% reactor power. (Electrical power could not be increased above this value due to transmission line stability limitations). Plant Data The following data was collected that the plant systems were satisfactorily at 100% FP: Boiler Acoustic Monitoring, collected to ensure there internal boiler "noise". satisfactory.
to confirm operating
Acoustic data was was no change in The re su11 s we re
Pressure Data. Accurate data was collected on boiler pressures and turbine inlet pressures to determine pressure drops from the boilers to the turbine, Heat Balance• Several heat balance checks made to ensure reactor power calibrations accurate.
were were
data
was
Boiler Feed Pump (BFF) Vibration Data. The BFP' s were instrumented to monitor vibration levels. Vibration levels increased because the three pumps were operating "back on their curve". Chemical Data. Some key chemical parameters were monitored to determine the effects of increased power. No significant changes were seen. RIZIH Temperatures. Data was collected from all 8 temperature detectors due to concerns about accurate data. It was subsequently determined that the temperature indications were inaccurate due to their physical location. The test successfully demonstrated that the unit could produce 940 MWe/1025 MVA at 100% RP. Phase II Design Changes As a result design changes follows:
of this test, several additional were found to be required, as
a) RIZIH temperature elements required relocation and calibration to provide accurate indication. b) The Boiler Level Control program required modification to provide correct level control. c) The setting on the deaerator relief valves had to be increased to provide adequate operating margin. (They had previously been set below the maximum code allowable). d) The generator current transformers were conf irmed to be overheat ing and required replacement. e) An automatic boiler pressure reduction was required to ensure positive reseating of the M S S V s if they opened on a high pressure transient, f) A small increase in generator hydrogen pressure was seen as beneficial in maintaining generator temperatures as low as practical. These changes and the computerized pressurizer level control system were engineered and then installed during Unit 5's first major outage in September, 1987. Phase II Upratins Test - Unit 5 A Phase II Uprating commissioning performed on Unit 5 in December, 1967.
test
was
CNS 9th ANNUAL CONFERENCE. 1988 35
The unit was increased to 100% power in small steps. PHTS inventory measurements were made and the pressurizer level setpoint line was adjusted to match the PHTS shell. The gains were adjusted in the pressurizer level control program to compensate for observed swell changes when steam pressure or PHIS pressure was variei. Several transient tests were performed to demonstrate satisfactory operation. These tests
d) All three ma i boiler feed pumps are required i.jr o,;iT.it ion it l'Jirt FP. ro be acceptable iiimini r >jperut i.on.
The unit w.ir. returned to 100'!. FP and released for a 2 month shakedown run prior to declaring new maximum continuous rating for the unit. REVIEW
Loss of Main Boiler Feed Pump. 0.., main pump was tripped at 100% FP in a series of tests. A reactor power reduction to 90% FP was required to matrh feed flows with steam flows. Loss of Hain condensate Extraction Pump. One main pump was tripped at 100% FP. Two pumps were found to be adequate at 100% FP. Reactivity Transient. Gadolinium poison wan injected to ensure that the reactor regulating system would respond Qorrectly to this transient. No problems were encountered. PHTS Stability. Several transient tests were performed to ensure thai the new pressurizer level control system would pt-eveni. any oscillations. No problems were encountered. Maximum Generator MVA. The generator load was increased to 1025 MVA (lag). Exciter current limit portection circuits cut in (to limit output to 980 MVA) and the test was abandoned. Temperature data was collected on the main output transformers to determine their thermal limits. Single Exciter Failure. It was planned to fail one of the dual exciters at 1025 MVA. This test was performed at 980 MVA with no problems. Single Deaerator Levelcontrol Valve Failure. One of the two 100% valves was failed closed. No control problems were encountered. Minor Tests. Several minor tests were performed to ensure adequate system response. These were boiler blowdown, reheater drains pump failures, and HPFW heater drains pump failures. All tests were satisfactory. Load Rejection. The final test was a load rejection from 100% FP. Turbine overspeed control and overall unit response was excellent. The turbine generator was resynchronized to the system and reloaded to 100%.
TEST CONCLUSIONS a) PHTS boiling was l(>ss than predicted instability was observed. b) Two main Condensate adequate at 100% RP.
Extraction
Other areas that may limit the ability to run at the uprated conditions are: 1. In the summer, with warm lake water temperatures, it is expected that the output of the units will have to be. curtailed to maintain the moderator temperature within its reactor safety analysis limits. A revised safety analysis increased the maximum temperature allowed by 2°C to 64°C in the fall of 1987 but it instill expected that units will have to be backed off when the lake, water is warm. The cost of installing more moderator heat exchanger capacity has been reviewed and is not viewed as economically attractive at this time. 2. During the load rejection at the completion of the Unit 5 Uprating Test, the internal spray header in the deaerator failed. The deaerators are being closely inspected during the unit shutdowns and the headers are being rewelded using more extensive welds in the failure prone areas. 3. In order to run at 100% reactor power it is necessary to run all three boiler feed pumps. These pumps are nominally 50% pumps and two are sufficient for the Phase 1 conditions of 96% reactor power but are not enough for 100%. In the long term, to restore the ability to prevent process upsets, an additional small boiler feed pump is being considered to run in parallel with two of the main pumps when running above 96% reactor power. An alternative is to install an automatic reactor set-back to 96% reactor power on loss of one boiler feed pump.
and no
Pumps
were
c) 1025 MVA output could not be achieved due to exciter current limits. A re-designed circuit is required to allow continuous operation at 1025 MVA.with subsequent testing.
36 CNS 9lh ANNUAL CONFERENCE, 1988
One unit, Unit 5, has been successfully uprated and Unit 7 is being uprated during its current shutdown to utilize 100% of its reactor power. Unit 7 will be available in the Phase II uprated state by July, 1988 and it is expected that the transmission limits will be sufficient to be able to transmit the energy frOm two uprated units. Unit 6 and Unit 8 are planned to be uprated in time for a new transmission line which will be in service in 1990.
The investment necessary to achieve the uprating will be 6 M$ for all units. If it is decided to install extra boiler feed pumps the additional investment is estimated to be 6 H* for all four units. The return on the uprating investment from Unit 5, Phase II Uprating alone during the first three months Df this year was 1.4 M$. From the time of Lhe original uprating to the end of Harch 1988, a
total of 2 TWh of extra energy has been produced with a benefit of 25 M$ to Ontario Hydro, due to all this uprating. "prating will have increased the installed capacity of Bruce Nuclear Generating Station "B" by 280 MWe. A benefit of this uprating is the ongoing reduction in acid gas emissions. The present value of 280 MWe of clean power is in excess of 60 M$, when compared to the cost of acid gas emission control measures, such as scrubbers or purchasing low sulphur codl. CONCLUSION The benefits of uprating are expected to continue. Uprating the Bruce Nuclear Generating Station "B" Units is a golden opportunity for inexpensive power for Ontario.
REFEREHCES (1)
MOORE, Study"
R.H.,
11
(2)
MOORE, R. H. , Generator
lt
BNGS"B"1
Turbine
BNGS"B", Phase Uprating
Generator
11: Turbine Study",
CNS9ih ANNUAL CONFERENCE, 1988 37
Session 2: Small Reactors: Design
Chairman: G.E.Gillespie, AECL WNRE
CNS 9th ANNUAL CONFERENCE. 19BS 39
/ * .
THE NUCLEAB BATTERY:
A SOLID-STATE, PASSIVELY COOLED REACTOR FOR THE GENERATION OF ELECTRICITY AND/OR HIGH-GRADE STEAM HEAT
K.S.
KOZIER AND H.E. ROSINGER
Atomic Energy of Canada Limited Reactor Development-Program Responsibility Centre Whiteshell Nuclear Research Establishment Pinawa, Manitoba ROE 1L0
ABSTRACT This paper reviews the evolution and present status of an Atomic Energy of Canada Limited program to develop a small, solid-state, passively cooled reactor power supply known as the Nuclear Battery. Key technical features of the Nuclear Battery reactor core include a heat-pipe primary heat transport system, graphite neutron moderator, low enriched uranium TR1S0 coated-particle fuel and the use oi burnable poisons for long-term reactivity control. An external secondary heat transport system extracts useful heat energy, vhich may be converted into electricity in an organic Rankine cycle engine or used to produce high-pressure steam. The present reference design is capable of producing abouf. 2400 kW(t) (about 600 kW(e) net) for 15 full-power years. Technical and safety features are described along with recent progress in component hardware development programs and market assessment work.
BACKGROUND The Nuclear Battery is a small nuclear power supply designed to generate electricity and/or highgrade steam heat. It is being pursued by Atomic Energy of Canada Limited (AECL) as a complementary, follow-up product to the Slowpoke Energy System district heat source, but is at a much earlier stage of development. The Nuclear Battery p-ogram originated in 1984 as a joint project with the Los Alamos National Laboratory (LANL) to develop a small, 20-kW(e) nuclear power supply (1,2,3,4,5) for unattended short-range radar stations in the new North Warning System (NVS). however, the joint project was cancelled when it became apparent that the full-power demonstration of a prototype unit could not be completed in time to meet the demanding deployment schedule for the NWS application.
Nuclear Battery program mandate has been redirected to address small-scale, base-load electricity generation in remote communities that are at present served by Diesel generators. Consequently, the remainder of this paper largely concerns our progress toward this end. More recently, it has been recognized that the same Nuclear Battery reactor core that is being designed to produce electricity might also be used to generate high-pressure steam heat for industrial applications and thereby greatly expand the application market. Some preliminary thoughts regarding possible application to the specific case of the in situ recovery of bitumen from the Alberta Oil Sands are discussed.
Reactor Description The name "Nuclear Battery" was coined early in the program to highlight the passive and solid-state features of the concept and distinguish it from conventional water-cooled power reactor designs. The term "battery" reflects the function of its graphite core block as a thermal energy storage cell, or well, from which useful energy is extracted in a passive manner. Nuclear fission of a small quantity of uranium atoms located in fuel rods embedded directly in the graphite provides sufficient energy to maintain the core at a high temperature for many years. The basic features ol the Nuclear Battery reactor core module are snown schematically in Figure 1.
CONTROL HOD
The inherent technical attractiveness of the Nuclear Battery concept enables it to address a diverse range of applications that are especially suited to Canadian energy needs. Thus, Canadian development of the concept continued in an independent program that focussed for a time on use as an air-independent auxiliary power source for Diesel submarines as part of the Canadian Submarine Acquisition Project (CASAP) (6). Although the CASAP program ultimately opted for full-powered nuclear submarines of proven, conventional design, the design exercise vas useful in that it forced the consideration of more powerful versions of the concept, better matched to the needs of the commercial marketplace. AECL's long-term interest in the Nuclear Battery has always been its potential usefulness for broadbased commercial applications. Therefore, the
FIGURE 1 :
NUCLEAR BATTERY REACTOR CORE MODULE
CNS 9th ANNUAL CONFERENCE, 198B 41
Unlike conventional power reactors, no fuel is added or removed from the Nuclear Battery for the life o£ the system. Thus, skilled reactor operators trained in the handling of highly radioactive materials are not required and the core containment boundary need not be breached for refuelling. These attributes make the concept practical for use in remote locations, where maintenance costs would be prohibi tive.
the pipe by the evaporation of potassium from the liquid film lining its interior surface. The hot vapour passes into the central core where natural pressure gradients propel it at high velocity to the top or condenser region of the pipe.
The fuel for the Nuclear Battery is based on the TRISO (triply isotropic) coated-particle design developed for use in high-temperature gas-cooled reactors (HTGRs) and amply demonstrated in the Ft. St. Vrain reactor in the USA and the AVR and THTR reactors in West Germany. The fuel meat consists of tiny spheres, or kernels, of enriched U0 2 , about 0.5 mm in diameter. The Nuclear Battery would use LEU (low enriched uranium, less than 20 at.Z 2 3 5 U) fuel material. The fuel kernel is sealed in successive protective layers of low-density buffer graphite, high-density pyrolytic carbon, SiC ceramic, and a second layer of pyrolytic carbon to form finished particles about 0.9 mm in diameter. The TRISO coated fuel particles are mixed with a graphite matrix binder and formed into solid cylindrical compacts, a few centimetres in diameter and several centimetres long. A "fuel rod" consists of a vertical stack of fuel compacts, about 1.5 m long, inserted into predrilled holes in the graphite core moderator blocks. Approximately five hundred fuel rods are arranged in the core on a regular triangular lattice pitch. The required fissile inventory to sustain operation at 2400 kU(t) at a nominal care graphite temperature of 550°C for 15 full-power years is less than 35 kg of 2 3 5 U and may be reduced further as the core design is refined. Heat passes by conduction from the fuel rods, through the solid graphite moderator, to the heat pipes regularly dispersed throughout the fuel lattice. Each heat pipe is of identical construction and consists of a sealed metal tube, about 5 cm in diameter and 3 n*. long. The bottom end of the heat pipe terminates in the lower axial reflector while its top end protrudes above the reactor core. A schematic diagram of a Nuclear Battery heat pipe is shown in Figure 2. The heat pipe has a thin wall thickness and is constructed from a niobium alloy to minimize parasitic neutron absorption losses yet maintain adequate resistance to creep-buckling collapse at high temperatures. A composite wick structure lines the inside wall of the heat pipe and serves to distribute the liquid working fluid uniformly around the inner surface of the pipe while providing a permeable protective barrier to interaction with the fastmoving vapour core. Each heat pipe contains a few hundred grams of potassium working fluid, most of which fills the wick and lines the inside wall of the pipe as a liquid film or collects as a pool at the bottom of the pipe; less than a gram of potassium is in the vapour state in the central core. At the design operating temperature, the pressure of the potassium vapour within the pipe is sub-atmospheric. Individual heat pipes act as independent, localized cells for passive primary heat transport by natural processes. Thermal energy is extracted as latent heat in the in-core or evaporator section of
42 CNS 9th ANNUAL CONFERENCE, 1988
FIGURE 2:
A NUCLEAR BATTKRY HEAT PIPE/VAPORIZER ASSEMBLY
The potassium vapour gives up its heat through condensation on the slightly cooler interior surface at the top of the pipe. Liquid potassium returns by gravity to the evaporator section where the cycle is repeated. The evaporator and condenser sections of the heat pipe are separated by an adiabatic region where heat is neither added nor removed. The heat pipe transports large quantities of thermal energy in a near-isothermal manner for a small quantity of working fluid since it involves the energetic phase changes between the liquid and vapour states. Moreover, the absence of active pumps promotes overall efficiency by eliminating primary system pumping power losses. The size of the Nuclear Battery core is determined by both the neutronlc properties of graphite and the heat plp°s. The large mass ratio of carbon moderator nuclei to that of the neutron necessitates a large number of collisions to achieve adequate thermalization of fission neutrons. Consequently, the fuel rods are widely spaced, providing ample room for the heat pipes. Displacement of graphite by the heat pipes and neutron absorption losses in heat-pipe materials further increase the required core size. The present reference design has a cylindrical fuelled core region surrounded by radial and axial graphite neutron reflectors, so that the overall core dimensions are about 2.5 m in diameter by 2 m high. Active reactivity control is provided by four axial control/shutdown rods Inserted from above the core and arranged in an outer ring of three plus a single central rod. Passive long-term reactivity control is provided by burnable neutron poisons mixed with graphite and inserted in holes in the graphite moderator between the fuel rods in the inner core region. Besides limiting the available excess reactivity, the burnable poisons are highly effective at
flattening the cove radial power profile and thereby limiting the peak-to-average fuel burnup ratio. The graphite core is surrounded by a few centimetres on thermal insulation, 10 minimize thermal leakage losses, and rests on ceramic supports within a steel containment vessel. A helium cover gas system maintains an inert core environment while promoting heat transfer across clearance gaps. The vessel f s internal pressure is held at just slightly above one atmosphere* to minimize the potential for air ingress in the event that a leak occurs. The core containment vessel is suspended in a concrete shield vault that would typically be located below grade it'.W. Passage of air through the gap between the vessel and the concrete cools both surfaces so that they are maintained within their design temperature ratings. At full power, the normal heat loss through the vessel would be in the range of about 30 kW(t) and is comparable to the decay heat load at one hour after reactor shutdown from full power. Useful energy is extrac ted from the Nuclear Battery by the circulation of organic coolant in a coiled tube surrounding the condenser portion of the heat pipe. In a version of the Nuclear Battery to produce electricity, the organic working fluid is toluene and the coiled tube structure is referred to as a vaporizer. Each heat pipe has its own vaporizer, which is connected in parallel between inlet and outlet headers. The vaporizer housing forms an extension to the core containment vessel and is bolted to it at a flanged joint which has a metal ring seal. A narrow, helium-filled gap between the heat pipe wall and the vaporizer housing serves to drop the peak outlet temperature experienced by the organic fluid to about
The minimum spacing of the vaporizers above the core limits the number of heat pipes that can be inserted into a core of a given diameter. Thus, the maximum useful power output from the Nuclear Battery is constrained by the axial heat transport performance of its heat pipes.
Reactor Safety Features Although no nuclear reactor des ign can claim absolute inherent safety, reactors can be designed with safety attributes that will p revent the catastrophic release of radioactive fi ssion products in severe accidents, so that no evacu ation of surrounding areas would be required. Pas;; ive safety features that obviate the need for immediat e human action and that cannot be overridden by misgu ided intervention have been of paramount importance throughout the Nuclear Bat tery program. Containment of radioactive fission products begins at their source - the type of nuclear fuel selected. The TRISO coated-particle fuel chosen for the Nuclear Battery embodies perhaps the ultimate in containment principles because the fuel inventory is finely subdivided into 3 x 10 8 independently protected entitles. The small particle size enables its coating layers to withstand internal pressures and
' 1 atm = 101.325 kPa
stresses that would dest roy most macroscopic engineered structures. Indeed, even violent dismemberment and dispersion of the core for whatever reason would not negate the containment properties of the individual TRISO particles. The most outstanding safety feature of TRISO fuel is its demonstrated ability to withstand extreme temperatures. No failures of the particle coatings would be induced by exposure to a temperature of 1600"C for up to 100 h. In contrast, the peak fuel temperature In the Nuclear Battery during normal equilibrium full-power operation is about 600°C and it is not possible to increase this value by mote than about 100°C, even with all four control rods fully withdrawn. Many of the saSety arguments for the Nuclear Battery are strikingly similar to those of its TRISOfuelled, graphi te-moderated cousins, the prismaticblock and the pebble-bed HTGR designs. Amplification of these safety arguments is possible for the Nuclear Battery since it operates at a much lower temperature and power dens i ty, and because its maximum total fission product inventory (equivalent to that of about 77 irradiated CANDU fuel bundles) is quite small. A brief qualitative review of the response of the Nuclear Battery to various hypothetical "worst-case" accident scenarios will help to highlight some of its most important safety features. First, consider a complete loss of coolant accident (LOCA) in which all heat pipes or vaporizers abruptly cease to function. Second, suppose that the protective system fails to detect this event so that shutdown is not initiated. Because the Nuclear Battery possesses a strong negative core temperature reactivity coefficient of about -0-08 mk/°C, self-regulation would occur and the core power level would drop to match the available passive means of heat removal. Recriticality would occur after some hours at a stable power o£ about JO kl/(t), while the decline in the equilibrium xenon poison load (about 4,4 mk) would enable the core temperature to increase by about 55°C. No release of fission products would occur. Third, assume further that instead of merely fai1 ing to shut down the reactor, the control rods are fully withdrawn at their maximum permitted speed (loss of regulation accident coincident with LOCA). With burnable poisons to provide passive long-term reactivity control, only a maximum of about 7 mk of additional leactivity would be available, corresponding to a further temperature increase of about 88°C. Again, no release of fission products would occur. Fouith, assume that, coincident with the other events, breach of the secondary containment vessel occurs, exposing the core graphite to ambient air. At ihis point, the maximum graphite temperature would have increased to about 693°C while the temperature at the reflector surface would be less; a temperature drop of roughly 125°C from the core centre to its surface would be necessary to transport 30 k u ( t ) , assuming uniform heat generation and spherical georoe t ry. Burningr or self-sustained oxidation, of graphite in air is quite difficult to achieve. According to reference 7, a graphite temperature of 65O°C is the minimum necessary, but not sufficient, requirement to initiate burning. Additional requirements include an adequate supply of air, removal of the gaseous reac-
CNS 9th ANNUAL CONFERENCE, T988 43
tlon products, and a favourable geometry that Inhibits cooling by conduction, convection and radiation heat loss mechanisms. It seems unlikely, therefore, that exposure of the Nuclear Battery reactor core to air, e'en during a LOCA with all control rods withdrawn, coold generate sufficiently rapid oxidation of the graphite to raise the core temperature to values that would Induce fuel particle coating failures. Instead, erosion o£ the graphite through oxidation would reduce the available reactivity and thereby gradually reduce the core temperature that is sustainable neutronically. A second class of severe accident of importance to reactor safety is a large and rapid reactivity insertion, such as might be induced by rapid withdrawal of a control rod. For the Nuclear Battery, this type of accident is pertinent only during startup, because, once the reactor has achieved its full temperature and power operating condition, a maximum excess reactivity of only about 7 mk is available. Pulsed power tests performed many years ago in the KIWI-TNT experiments (8) at LANL using a variety of coated fuel particles (likely inferior to modern-day TRISO design) demonstrated that an energy deposition of at least 10 13 fissions/gU (equivalent to about 80 cal/gU) was necessary to induce coating failures in particles with 0.5-mm-diameter U0 2 kernels. This criteria was applied to the Nuclear Battery core in preliminary transient analyses and It was found that a corresponding instantaneous reactivity insertion in the range of roughly 20 mk would be required to induce coating failures. The calculated temperature response of the Nuclear Battery to a rapid reactivity insertion transient is qiite similar to that reported in simulations of the pebble-bed HTR-Module reactor (9). In general, the wide dispersion of the fuel throughout the core of the Nuclear Battery and its strong negative core temperature reactivity coefficient would cause a sharp power pulse response, somewhat analogous to that which is obtained in the homogeneous TREAT pulsed graphite reactor (10,11). The reactivity requirements of the control/ shutdown rods for the Nuclear Battery are about 43 mk from cold to hot, 4.4 mk for equilibrium poisons and 7 mk for long-term fuel depletion, for a total of about 54 ink. Thus, limiting the Individual rod worths to about 20 mk would be sufficient to ensure the integrity of the fuel during startup since only one rod would be withdrawn at a time. Also, this scheme would permit cold shutdown to be achievable with one rod stuck in its uppermost position. The neutronlcs features of the Nuclear Battery and its transient response during a control rod withdrawal accident are discussed in a companion paper (12).
FIGURE 3:
SCHEMATIC OF A 600-kW(e) NUCLEAR BATTERY FOR VILLAGE ELECTRICITY
Toluene is heated to about 37O°C In helical-coil vaporizers surrounding the condenser end of the heat pipes. Supercritical toluene vapour is collected from the parallel vaporizer streams and fed to the turbine inlet. Expansion of the vapour through the single-stage high-speed turbine converts a portion of the thermal energy into rotary mechanical motion to turn an ac generator and a Pi tot feed pump. The three rotating elements are usually combined on a single rotating shaft, the only moving component in the engine. The compact rotating unit is supported on twin hydrodynamic bearings lubricated by the toluene vorklng fluid. Exhaust vapour from the turbine is passed through a waste heat regenerator to improve the cycle conversion efficiency. The toluene vapour is then condensed In a water-cooled condenser. The Pltot feed pump draws toluene from the condenser and forces it through the regenerator where it is preheated to about 1 9 0 T . The output from the regenerator is then distributed to the individual vaporizers to complete the cycle. The net conversion efficiency is estimated to be about 25JS for a 2400 ktf(t) Nuclear Battery with a water-cooled condenser. Some of the key technical advantages of the toluene organic Rankine cycle engine for this particular application include 1. high reliability as a result of the lubricating qualities of toluene in the hydrodynamic bearings and very few moving parts;
Electricity Generation The Nuclear Battery is an effective energy source for small-scale electricity generation because its high temperature offers good Carnot efficiencies of about 54j. To produce electricity In a reliable manner for extended periods, it is coupled at the present time to a toluene organic Rankine cycle engine, as shown schematically in Figure 3.
44 CNS 9th ANNUAL CONFERENCE, 1988
2. low maintenance as a result of the non-corrosive behaviour of toluene and its good resistance to thermal decomposition; 3. extremely low activation of toluene in neutron radiation fields above the core; 4. good net conversion efficiency at modest peak temperatures for the supercritical, regenerative cycle;
5. relatively compact and lightweight engine components; 6- low vorking-fluid cost and its high availability; and
40. 35.
-~ : J TUEC I c = J Fuel PUEC 1
\
30. 7. adequate demonstration of the underlying technology.
25. ?n
Diesel Fuel Cost
-^. --•. _^_
Canadian Diesel-Generated Electricity Market The economic goal o£ the Nuclear Battery is to substitute for fossil-fuel energy sources wherever and whenever the delivered cost of such fuel becomes an excessive burden for the consumer. This need is likely to be felt first and strongest in the dozens of small, remote Canadian communities that are totally dependent on Diesel generators for their electricity. The total annual production of electricity from Diesel generators in Canada is about 1000 GWh, most of which is produced by units of less than 1 MW(e) capacity.
10. s. n
5
10
15
20
25
30
Core Life (FP Years) FIGURE 4:
NUCLEAR BATTERY LIFETIME UNIT ENERGY COSTS FOR ELECTRICITY
Steam Generation The cost of electricity in remote Canadian villages varies widely since it depends strongly on both the distance and means of fuel transport. A recent internal study by AECL has established the average cost of delivered Diesel fuel to be in the range of 0.15 to 0.20 S/ktfh and this has been adopted as the cost target for the Nuclear Battery. The estimated lifetime unit energy cost for electricity from a 600 kW(e) Nuclear Battery operated in a base-load mode (952 capacity factor) is shown as a function of core lifetime in Figure 4. It is assumed here that only the initial uranium fuel inventory is increased to maintain criticality for the required lifetime. Also shown in Figure 4 is the partial unit energy cost component associated with the uranium fuel inventory. The fuel partial unit energy cost is seen to be generally fairly flat and increases slightly at high core lifetimes as a result of both the present-value financial methodology employed and the gradual perturbation of the fuel-tomoderator atom ratio. Evident from Figure 4 is the need to achieve product lifetimes of 15 to 20 fullpower years and the fact that the Nuclear Battery might already be competitive today with Diesel generators in selected locations. The capital cost of the Nuclear Battery used for the analysis is preliminary. We expect further reductions as our program matures. In addition, further economics-ofscale will be achieved through the mass production of identical units.
Because the Nuclear Battery is a high-temperature heat source, the same reactor core module can be coupled through a secondary heat transport system to a steam generator, as shown in Figure 5. Since the choice o£ fluid is no longer constrained by tne thermodynamic requirements of a Rankine cycle engine, other working fluids, such as the Monsanto product, 0S-84, used in the WR-I research reactor (13) for 19 years at the Whiteshell Nuclear Research Establishment (WNRE), might be considered in addition to toluene. Also, a Nuclear Battery to produce steam would be less costly to construct since a steam generator is less expensive than an organic Rankine cycle engine conversion system.
Because remote communities vary in size and undergo significant seasonal variations in load requirements, only a portion of the Canadian Dieselelectricity market is addressable in a practical manner by base-loaded Nuclear Batteries. It is estimated that the mature Canadian market potential is for up to one hundred and fifty 600-ktf(e) units. In the most logical circumstances, it is envisioned that multiple Nuclear Batteries would be sited in a phased sequence at a common location. A new unit would be commissioned every five years or so, in perpetuity, as older units complete their output cycle. Diesel generators would be retained for peaking and backup power roles by virtue of their low capital cost.
FIGURE 5:
SCHEMATIC OF A 2400-k«(t) NUCLEAR BATTERY FOR STEAM GENERATION
A very large Canadian market for high-grade industrial steau1 heat exists in the Alberta oil sands
CNS 9th ANNUAL CONFERENCE. 1988 45
for the in situ recovery of bitumen using the proven "huff-and-puff " recovery process which requires highpressure injection steam of 80% quality. In situ recovery programs, such as the Esso Cold Lake project, are proceeding in gradual phases using natural gas to produce steam. In the past, AECL has evaluated the use of a large, organic-cooled reactor station to supply the -equired energy (14,13,16). There appears to be a plausible degree of technical fit of base-loaded Muclear Batteries for this type of application in that a single in situ recovery well requires roughly 6 MU(t) of steam for about 0.9 full-power years spread over eight calendar years. A cluster of twenty wells is at present fed from a single satellite station. Siting of Nuclear Batteries close to the wellhead should help alleviate steam distribution needs. In situ recovery of bitumen is very energyintensive and requires about 6.8 GJ/m3. The total resource that is recoverable by this method in Alberta is estimated to be 23 x 10' m3, or about 22£ of the world's total recoverable petroleum resources. The total energy necessary to extract this bitumen is equivalent to the lifetime output from about 140 000 Nuclear Batteries. Clearly, no single energy source technology vould be capable of supplying all of this energy and some mix will be required, particularly as natural gas pricey increase.
ed from a low neutron-absorbing zirconium alloy, vas made to transport about 10 kW(t) at 500°C in a carbon block in experiments at VNRE. Work at VNRE with composite wick structures in sub-scale heat pipes has demonstrated superior axial heat transport performance in comparison with knurled-tube designs. Notably, an axial heat flux about 202 greater than the average required for a commercial Nuclear Battery unit has been achieved and further improvements are anticipated. An experiment to study graphite oxidation under conditions that roughly simulate an air ingress accident in the Nuclear Battery was also performed. A small, electrically heated graphite block was consumed over a six-week period at temperatures progressively increased from 600 to 800°C, without observing a temperature runaway. A full-scale toluene vaporizer test loop, including a stainless steel/potassium heat pipe, is now being commissioned as the next important hardware development step. The near-term goal of the R&D support program is to demonstrate the overall technical feasibility of the Nuclear Battery concept in 5i non-nuclear, but otherwise comprehensive, test. Thus, plans are now under way to build a single-heal-pipe integrated test facility that will look much lil:e Figure 2, except it will use electric resistance heaters instead of nuclear fuel rods.
R&D Progress Significant technical progress has been made over the past several years to support the viability of the Nuclear Battery reactor power supply concept. Notably, the basic reactor physics features of the system have been confirmed by LANL in a full-scale mockup of the 2C-kU(e) core that achieved first criticality in 1987 July (17,18). A series of lowpowei core neutronlcr tests were successfully completed by LANL last year, including a measurement of the core temperature reactivity coefficient up to about 70°C. At VNRE, our core design efforts have concentrated largely on improving the economi- prospects for commercial-scale units and estaWlsWrig VcrcTia'ole neutron poisons as a plausible means of enhancing their passive safety. In hardware development programs at WNRE, a l-kU(e) REMCOM (Remote communications) (19) toluene organic Rankine cycle engine was commissioned and operated with a propane lieat source. This demonstration-scale unit was operated for a total of 784 h, including 465 h in a single continuous run with a simulated remote startup. It is now being converted to an electrically heated configuration that vill permit the determination of its conversion efficiency as a function of its operating state while providing data on toluene thermal degradation rates. Toluene thermal decomposition data have already been obtained over a wide range of conditions in staticcapsule and flowing-loop tests. Decomposition experiments concerning gamma radiolysis have been conducted with static capsules and are now being extended to flowing-loop tests. The heat pipe program at LANL successfully demonstrated the heat transport performance required for the 20-kW(e) core using a full-scale stainless steel/potassium heat pipe with a knurled interior surface. A similar knurled heat pipe, but construct46
CNS 9th ANNUAL CONFERENCE. 1988
CONCLUSION The Nuclear Battery high-tem;>erature power-source concept has evolved over the P>st fou,. years from the consideration of special-us; projects at lowpower levels into a commercial-scale unit capable of delivering either 600 kW of ele-tricity or 2400 kW of high-grade steam heat for 15 full-power years. This evolution has been accomplished without compromising the Nuclear Battery's technical feasibility or its superior passive safety features. Broad-based domestic applications for village electricity and for Industrial steam production have been identified that are in the long-term Canadian national interest and are sufficiently large to justify the further development and demonstration cf the concept.
REFERENCES (1) KIRCHNER, W.L., AND MEIER, K.-L., "Low Power Unattended Defense Reactor", 19th IECEC Proceedings, 1984. (2) MEIER, K.L., PALMER, R.G. AND KIRCHNER, W.L., "Low Power Reactor for Renote Applications", 20th IECEC Proceedings, l'>85. (3) MEIER, K.L., KIRCHNER, «.-•, AMD HLLCUTT, G.J., "A Compact Reactor/ORC Po'er Source", 21st IECEC Proceedings, 1986. (4) MB.IER, K.L. , KIRCHNER, W.L., AND HANCOX, W.T., "North Warning System Rea=tor Conceptual Design Report", L A - W J - M S , 1986. (5) PALMER, R.G., ANDDURKEE, J.W., Jr., "Neutronics Design Studies for an Unattended, Low Power Reactor", Proceedings of the Topical Meeting on Reactor Physics and Safety, NUREG/CP-0080, Vol. 1, Saratoga Springs, NY, 1986.
(6) "CASAP Update: Program Makes Definite Progress Amid Rising Controversy", Canada's Navy Annual, 1987. (7) SCHWEITZER, D.G., GURINSKY, D.H., KAPLAN, E., AND SASTRE, C., "A Safety Assessment of the Use o£ Graphite in Nuclear Reactors Licensed by the U.S. NRC", NUREG/CR-4981 BNL-NUREG-52092, 1987. (8) GEER, W.U., HOFFMAN, C.G., AND KERRISK, J.F., "The Behaviour of Coated Particles in a Large Nuclear Transient (Kiwi-TNT)", Los Alamos Scientific Laboratory, LA-3367, 1966. (9) LOHNERT, G.H. AND KINDT, T., "The HTR-Module, a Power Reactor W t h no Reactivity Excursions of Severe Consequences", ANS Topical Meeting on the Safety of Next Generation Power Reactors, Seattle, tfA, 1988. (10) FREUND, G.A., ET AL., "TREAT, A Pulsed GraphiteModerated Reactor for Kinetic Experiments", Proceedings of the Second United Nations International Conference on the Peaceful Uses of Atomic Energy, Geneva, 1958, Vol. 10, 1959. (11) GHATAK, A., AND NELKIN, M., "Microscopic PromptNeutron Kinetics of the TREAT Reactor", Transactions of the American Nuclear Society, Vol. 6, No. 1, p 110, 1963. (12) DONNELLY, J.V., KOZIER, K.S., AND PENNER, G.R., "Neutronics Design of a Small, Solid-State, Passively Cooled Reactor, The Nuclear Batcery", Proceedings of the 9th CNS Conference, Winnipeg, MB, 1988. (13) TURNER, D.G., "The UR-1 Reactor a General Description", AECL-4763, 1974. (14) PUTTAGUNTA, V.R., SOCHASKI, R.O., AND ROBERTSON, R.F.S., "Role of Nuclear Energy in the Recovery of Oil from the Tar Sands of Alberta", AEC1.-5239, 1976. (15) SOCHASKI, R.O., AND SMITH, D.W., "A Cost Comparison of Nuclear and Fossil Power for the Alberta Tar Sands", AECL-5682, 1977. (16) SOCHASKI, R.O., ET AL., "CANDU-GCR Power Station Options and Costs", AECL-6436, 1980. (17) "Compact Nuclear Reactor May Be Safer and Cheaper", Mechanical Design, p 24, 1987 December 10. (18) PITMAN, F., "The First Sustained Chain Reaction in a New 20-kW Modular Reactor", Nucleonics Week, p 6, 1987 October 29. (19) MONAHAN, J., AND MCKENNA, R., "Development of a 1-kW, Organic Rankine Cycle Pover Plant for Remote Applications", 11th IECEC Proceedings, 1976.
C N S 9th ANNUAL CONFERENCE, )98S
CONCEPTUAL DESIGN OF A SMALL NUCLEAR REACTOR FOR GENERATING ELECTRICITY J.F. LAPORTUNE and D.A. MENELEY Department of Chemical Engineering University of New Brunswick P.O. 4400 Fredericton, New Brunswick E3B 5A3
ABSTRACT A concept for a high-efficiency small nuclear reactor of 2 MWth is proposed for generating electricity. An integrated liquid-sodium/fuel cylindrical module is used as the basis to achieve high primary temperatures at low pressure while ensuring a high level of safety through the simplicity of the design. The reactor physics and primary thermal hydraulics are discussed. INTRODUCTION Small nuclear reactors (1-20 W t h ) have been investigated from the early 1950s to the early 1970s as heat and power sources for various applications ranging from supplying power to military bases in reriote locations to providing propulsion power for trains [1]. More recently, a number of countries, including Canada, have proposed to use small reactors to supply space heating and electricity to remote communities, raining towns, hospitals, industrial agglomerations, small submarines, etc [2-6]. The main asset of small reactors is their flexibility and independence; small units can be used in a wide range of conditions, are air-independent and do not need to be refuelled fof up to several years. Their main disadvantage has so far been cost. However, most recent concepts attempt to outweigh cost with a simple design based on well proven technologies. Sane aim at reducing operating costs by eliminating the requirement for full-time attendance by a qualified operator. In all cases, a strong emphasis is placed on inherent safety, which is defined as the system's ability to cope with abnormal situations without the intervention of engineered safety systems. Since most systems are based on light water technology, this emphasis often leads to low or atmospheric primary pressure reactors, in order to reduce the risk of a LOCA or to facilitate the management of such an accident. In return, low operating pressures lead to low conversion efficiencies when producing electricity. The objective of this paper is to present a concept of a small reactor operating at high temperature and low pressure. The aim of the design is to achieve high electrical conversion efficiencies without compromising safety. The results shown are based on a mostly theoretical study aimed at investigating the technical feasibility o£ -tine prepcesd concept. TVva basic concept is initially introduced. Some characteristics regarding the reactor physics and thermal hydraulics are then described. Finally, the main features as well as some possible improvements are discussed. CMS 9th AN. 4UAL CONFERENCE. 1988
THE CONCEPT System The fundamental concept en which the proposed reactor is based is illustrated in Figure 1. The primary coolant is liquid sodium. It is enclosed in a sealed vertical cylinder which constitutes a basic unit. The fuel is also contained in this unit. Heat transfer takes place by single-phase natural convection inside. The heat is removed by an external coolant drcuiatwg axtaMe. &r> insrt gas blanket is maintained inside the unit to accommodate for pressure fluctuations due to sodium expansion during heat-up.
'MAIM VESSEL WALL
rigure 1: The Basic Concept The actual unit is shown in Figure 2. The unit is 2 m high and 100 mm in diameter and made of stainless steel. It contains a bundle of 37 fuel rods of 10.2 mm diameter by 750 mm. Each fuel element contains a 100 mm Be reflector at each extremity. The total length of the fuel elements is 950 mm. The fuel is a U-Zr-Hj 6 dispersion alloy containing 15 w% uranium enriched to 16 % U 2 3 5 . A 0.5 mm thick stainless steel cladding is metallurgically bound to the fuel. The 37-element
vessel is shown in Figure 3. It consists of a 1.5 in diameter by 2.3 m high (without shielding) cylindrical tank divided into four sections by horizontal plates, namely (starting from the bottom): a. b. c. d.
!
op Berp|iecor
luel
bottom gas space, or gas outlet; calandria; top gas space, or gas inlet; and boiler region.
The units are inserted inside the calandria through the boiler region. The units themselves complete the seal between the boiler region and the top gas space. An annulus gap is maintained between the unit wall and the calandria tube to allow circulation of an inert gas and early detection of possible sodium leaks. The calandria contains the moderator; in this case, light water. The annulus gap around each unit maintains a thermal insulation between the high temperature units and the light water. An organic fluid circulates in the boiler region. The entire main vessel is maintained at near atmospheric pressure.
I 2m
A
the the the the
The system is a MODular, Sodium cooled and Thermalized Reactor; it has received the designation of MDDSTAR [7].
!
bortom Be refloeo'
Operating Principle • bundle support piato
Figure 2: Details of the Basic Uhit fuel bundle has a diameter of 80 mm, leaving an annulus gap of 10 mm between the bundle and the inner unit wall to allow circulation of the liquid sodium. The coolant fraction inside the fuel region is 40 %. The reactor core contains disposed in a hexagonal array. houses a control rod. Three more are used to insert shutdown rods.
33 such units, The central unit lattice positions The main reactor
During normal operation, the fuel produces 2 MW of thermal energy. This heat is extracted from the fuel by the sodium under natural circulation inside each unit. The sodium flows upward through the bundle and downward through the annulus, between the bundle and the inner unit wall. Heat is then exchanged to the organic fluid via the top portion of the unit, above the core. The organic fluid cools the unit by boiling at atmospheric pressure. Heat is then transferred to a conventional pressurised steam turbine loop used to generate electricity.
ftow gutfi shuKfcwn red insomon f>yslem
Figure 3:
Side and Top View of the Main Reactor Vessel CNS 9th ANNUAL CONFERENCE, 1988
49
The design of MDDSTAR is such that cooling of the units can be achieved via three paths:
The codes used for cell calculations were:
a. convection and boiling in the boiler region;
a. WIMS, with the 69 group Winfrith library [8]; and
b. radiation, conduction and convection through the annulus to the water moderator; and
b. XSDRN-PM, with the 16 group Hensen-Roach library [9].
c. forced convection to a through the gas annulus.
circulating
gas
Although path a. is the most effectiTO heat removal method, decay heat removal can still' be achieved within reasonable temperature limits using path b. or c. or a combination thereof. This design feature reduces the probability of total unavailability of cooling for the unity.
WIMS was used for all cross section collapsing. Whole core calculations were performed by the diffusion code CITATION [103, with three energy groups and in both azimuthal and planar twodimensional models.
,waiar calemdria tube (Zr) Partial or total replacement or redistribution of the units is made easy by the sijnplicity of the design and by the low power density of the core. There are two possible refueling matnods which could be envisaged.
annulus gap unil wall (SS) sodium 40 % sodium 10 % clad 50 % fuel
In the first, the annulus gas inlet and outlet would be sealed, and the boiler region as well as the top and bottom gas space would be filled with a heat transfer fluid. Cooling of the units would be assured by natural convection of that fluid while the units are transferred directly from the main vessel to a fluid-filled transport flask.
Figure 4: Cell Basic Unit Characteristics
The low level of decay heat for the nominal power considered and the thermal reserve provided by the sodium inside each unit would also allow the residual heat of the units remaining in the core to be removed exclusively by gas cooling. A unit being transferred could remain partially uncooled for several minutes before being introduced into a cooling-fluid-filled transport cask. Scope of the Work The concept presented is based on the novel idea of integrating the fuel and the primary coolant inside modular units. These units present some uncommon neutronic and thermal hydraulic characteristics: most of the federating material is located outside the fuel bundle and the primary heat transfer relies on natural circulation through a multiple fuel rod arrangement (the bundle) and inside a narrow enclosure (as opposed to inside a loop). Therefore, a major part of this design was devoted to the study of the concept's reactor physics and unit's thermal hydraulics. It should be noted that the objective of this work was to determine if the concept could work, and not how the system could work best.
REACTOR PHYSICS Reactor physics calculations were aimed at determining a suitable fuel type and fuel configuration, and estimating the major characteristics of the core, such as temperature coefficient of reactivity, control requirements and fuel lifetime.
50 CNS 9th ANNUAL CONFERENCE, 1988
The fuel unit is illustrated in Figure 4. The bundle radiuc was chosen so as to minimize neutron leakage through the sodium annulus surrounding the bundle and maintain a downward flow cross section approximately equal to the upward flow cross section through the bundle. Since sodium is practically transparent to neutrons, there is no inherent mechanism for neutron moderation inside the unit. Therefore, unless a moderator is present inside the bundle, all neutron slowing-dcwn takes place outside the unit, in the moderator annulus which surrounds it. This in itself leads to a well thermalized spectrum in the outside fuel elements but also to in^iortant self-shielding effects of the inside elements. This situation is not favorable sinoa a large portion of the fuel then becomes under-utilized. This is the reason why a U-Zr-Hj 6 fuel matrix is used. The presence of the hydrogenous medium increases moderation within the bundle and considerably improves the power density profile. Other options considered included a similar bundle with ua, fuel, an annular bundle with UO,, with and without a central Be or Zr-H rod. The choice made represents a compromise between adequate power density distribution, critical mass and thermal hydraulic configuration. Figure 5 shows the power and flux profiles in the bundle, for a hexagonal pitch of 119 mm.
Core Parameters The core configuration is shown in Figure The units are arranged in a liexagonal pitch order to minimize parasitic neutron absorption the light water. The core is surrounded by a 50
3. in in mm
0.0
0.2
0.4
0.6
0.8
NORMALIZED BUNDLE RADIUS
Figure 5: Neutron Flux and Power Profile in the Bundle beryllium reflector. Only one control rod is present in the core (the central rod). Three hafnium shutdown rods are used. During normal operation, with the tods withdrawn from the core, the shutdown sites are filled with light water. The active fuel core is equivalent to a cylinder with a 0.38 m radius and a 0.75 m height. The total mass of U 2 3 5 is 26 kg. The reactivity control requirements are given in Table 1. The core lifetime is presently 1660 MWdays, or slightly less than five years with a 50 % load factor. The negative reactivity provided by the shutdown rods is 96 mk. Because of their symmetrical arrangemerit, all rod worths are equal. With only two rods inserted, the total worth is approximately -44 mk. This is sufficient to scram the reactor and maintain a shutdown condition even after cool down.
The temperature coefficients of reactivity are given in Table 3 for fresh fuel. The main factors which affect the behaviour of the reactor are the strong negative coefficient for the fuel, and the comparatively strong positive coefficient for the water ntxkjrator. The fuel coefficient is mainly due to the prefer ce of hydrogen in the fuel matrix and is a well 'oiown characteristic of the TRIGA fuel [11]. Because of the inherent characteristics of the basic unit configuration, the water temperature coefficient increases with the hexagonal pitch. As the pitch is reduced, so is the absolute value of the coefficient, but so is also the available space between the units. The coefficient never becomes negative in the present configuration, partly because of the presence of water in the shutdown and control rod sites and around the core. Therefore, a pitch of 119 mm was selected as the minimum acceptable. Even at that value, the core arrangement allows only very small tolerances.
TABLE 3: AVERAGE TEMPERATURE REACTIVITY COEFFICIENT (mk/C)
Control Rod IN OUT Fuel (500-1000 C) Sodium (400-800 C) Water (60-100 C) Water void (0-5%)
-0.039 +0.0029 +0.101 -0.054
-0.041 +0.0058 +0.18 +0.025
However, although the high value for the water coefficient may, at first, appear to be unacceptable, it should ke noted that the light water is normally insulated and that transients inside the unit would not promptly affect the water temperature. Therefore, for power transients, the most iJT^ortant factor contributing to the overall power coefficient is the fuel temperature coefficient of reactivity.
TABLE 1: REACTIVITY CONTROL REQUIREMEOTS HTDRSULICS Bumup: Equilibrium Xenon: Hot/Cold swing:
27 mk 7 mk 21 ink
Table 2 shows the Core power peaking factors for a black control rod in and out of the core. These factors are high and suggest that options such as absorber rod distribution throughout the core, fuel zoning or burnable poison distribution may be advantageous.
TABLE 2: CORE POWER PEAKING FACTORS Maximum/Average Power Control Rod IN a/T Axial Radial Total
1.35 1.62 2.19
1.74 1.86 3.24
Natural convection in this type of assembly is a canplex phenomenon which has not received much attention in the past. In the present study, the thermal hydraulic behaviour of the sodium inside the unit was examined using the single-phase, three-dimaisional code CCf-Mix-lA [12]. The fuel bundle was modelled as a porous volume having the same porosity and directional permeabilities as the actual bundle. Hec3t_ Transfer Characteristics The heat transfer characteristics of the unit were obtained assuming a constant temperature of 300 C for the cooled portion of the wall. The following results are given in reference to that temperature. In most cases, a uniform power distribution war. assumed. However, some results were also obtained for a 6-region model approximc. ting the power profile calculated In the reactor pi.\^iJs section. Figure 6 shows the temperature distribution within the bundle calculated by COMIX for a unit power of 60 !*l. The hottest temperatures are found CNS9lh ANNUAL CONFERENCE, 1988 51
In abnormal situations, the unit can also be cooled through the bottom half via heat transfer to the moderator or forced convection to a flowing annulus gas. In this case, 10 % of the nominal heat can be removed with a maxinuni difference between the fuel and the inner unit wall temperatures ot only 21 C.
Intermediate Heat Removal In fact, given the excellent thermal properties of liquid sodium but the comparatively poor heat transfer characteristics of organic coolants, the limiting factor in heat transfer appears to be the heat removal from the organic side. The maximum heat transfer which can be achieved by boiling is of the order of 140 kW per unit, based on available correlations [13,14].
Figure 6:
Temperature Distribution in the Bundle
near the center top of the bundle. The central fuel element is also subject to the maximum axial temperature gradient. Although the radial temperature distribution is relatively flat near the bottom, there Is a significant temperature difference near the top between the inside and outside fuel elements. This difference is likely to affect the cell physics. Figure 7 shows the variation of the maximum fuel surface temperature, average fuel temperature and average sodium temperature as a function of unit power, for both power density distribution models. The major thermal characteristics of the unit for an average unit power of 60 kW are summarized in Table 4. Because of the low specific power and the small fuel element radius, the center line temperature is within 30 C of the surface temperature.
Nevertheless, using an organic coolant in the boiler region presents some advantages. The organic fluid serves as a buffer between the sodium and the water. Because of the high boiling temperature of most organic fluids, a high "source" temperature can be maintained without increasing pressure. Heat transfer to the intermediate could however be maximized by increasing the unit height, increasing the flow with the use of a pump or increasing the available heat transfer surface area with irregular shaping of the top half of each unit or with the use of fins.
TABLE 4: OPERATING CHARACTERISTICS Total Thermal Power Electrical Power Operating Pressure Average Temperatures Fuel Average Maximum fuel surface Sodium Organic coolant Moderator Average Linear Power Rating Average Linear Temperature gradient
-?000 kW 560 kW 101 kPa 400 C 450 C 346 C 260 C 60 C 2.18 kW/m 120 C/m
TRANSIENT RESPONSE
Figure '!: Overall Heat Transfer Characteristics of the Unit Assuming a cooling wall temperature of 300 C, it can be seen that fuel damage is unlikely below 500 kW of unit power and that no local boiling of the sodium occurs below approximately 450 kW (at atmospheric pressure, sodium boils at 850 C ) . Therefore, during normal operation, the conditions are kept well below critical. 52 CNS 9th ANNUAL CONFERENCE, 1988
The response of MODSTAR to external perturbations was examined using a simple point kinetics model of the reactor. Two situation types were considered: a step reactivity insertion and a moderator temperature increase. In the former, the cooling temperature on the intermediate side is assumed to remain constant throughout the transient and no heat is transferred to the moderator; the analysis considers the transient neutronic and thermal behaviour within the unit exclusively. In the latter, the flow of water is assumed to be suddenly interrupted while 10 % of the nominal power is transferred uniformly to the moderator inside the calandria. Boiling effects are not considered.
The results of two simulations are shown in Figures 8 and 9. For a step insertion of 2 mk, the strong negative fuel temperature coefficient stabilizes the reactor power at approximately 1.5 nominal power. The second example (Figure 9), shows that, even far the pessimistic case considered (10% of power to stagnant core water), the power increase i s slow and can be managed within a reasonable delay. The magnitude of the transient would depend on the margin to water boiling, since the reactivity increase due t o water would become negative upon boiling.
OPERATING O0NDITIONS The average operating conditions of the system are given in Table 4. Based on an available heat sink temperature of 10 C, with steam raised at 250 C in the power loop, the- results obtained thus far indicate that it is possible to reach efficiencies in excess of 25 % for conversion to electricity. Therefore, the total expected electrical power provided by M3DSTAR is approximately 560 kw. DISCUSSION The main features of K5DSTAR can be summarized as follows: a. The core is cooled by liquid sodium at atmospheric pressure; b. An atmospheric-fsressure organic heat transport loop is used to remove the heat from the basic units; this loop serves as a buffer by eliminating high-pressure boundaries on the units and also eliminating single barriers between sodium and water; c. Sodium and water are separated by at least two solid boundaries anywhere in the system;
with thermal toeobac*
d. There are no penetrations of the primary system (the basic units); this increases the integrity of the system but makes chemistry control more difficult; e. There are no pumps in the primary loop; °0.00
-1.00
3.00 12.00 TIME 'SI
Figure 8: Transient power Reactivity step
IS.OO
20.00
Response to a 2 mk
f. The organic coolant is kept outside the core; fluid degradation by radiolysis and pyrolysis Is reduced; radiation fields are liiiu_ted to shielded core leakage and gamma fields from sodium decay in the top half of the units; and j. Because of their size and simplicity, the units can be factory assembled and tested for integrity prior to their use. The major advantage of the M3DSTAR concept is the possibility to provide high temperature heat without the need to pressurize the primary heat transport loop. However, safety has not been compromised in doing so. The design combines simplicity with efficiency. There are no penetrations inside the units, and few penetrations inside the main vessel. Only one drive is needed (for the control rod) and there are only a few moving parts inside the core (four rods). The units are easy to insert and easy to retract in case of replacement. Because of the low system pressure and the absence of penetrations, the probability of a LQCA is expected to be reduced. The prompt negative temperature coefficient of the fuel confers a stable behaviour to the reactor during reactivity transients. Finally, at 2 With, the temperature safety margin is large (of the order of 500 C to prevent local sodium boiling).
"0.00
£0.00
40.0a
F.0.00 TIME (SI
B0.0O
I00.OO
120,00
Figure 9: Transient power Response to a Loss of Water Flow
However, negative aspects of the design ±- slude a high critical mass and a low clearance between the units. This latter point is due to the increased positive water temperature coefficient of reactivity which results from a greater pitch. This results in a tight arrangement of the units CNS 9th ANNUAL CONFERENCE. 1988 S3
which may render their insertion more difficult and may impair heat transfer on the organic side. Other problems of the present design relate to the large power peaking factors and the low sodium inventory. Although the small amount of coolant in the units improves the load response time, it provides only a small heat storage capability in case of power transients and necessitates constant cooling of the unit.
4. J.S. Hewitt, "The AMPS 1.5 MW Low-Pressure Compact Reactor", First int. Seminar on Small and Medium-Sized Nuclear Reactors, Lausanne, Switzerland, Aug. 1987. 5.
F.C. Foushee, R.W. Schleicher, G. Schlueter, J.S. Yampolsky, "Small TRIGA Power Reactors for District Heating", Nuclear Europe Vol 12 1984, pp. 33-36.
6.
C.W. Forsberg, "A process Inherent Ultimate Safety/Boiling-Water Reactor PIUS/BWR", Paper Presented at the Workshop on Inherently Safe Reactors in the Far East, Institute for Energy Analysis, Oak Ridge, Tn., 1985.
7.
J.F. Lafortune, Conceptual Design of a Small Nuclear Reactor for Electricity Generation, PhD Thesis, University of New Brunswick, N.B. 1988.
8.
J.V. Donnelly, "WIMS-CRNL, A User's Manual for the Chalk River Version of WIMS" AECL-8955 1986.
9.
Radiation Shielding Information center, "RSIC Computer Code Collection; SCALE-3, A Modular Code System for Performing Standardized Computer Analyses for Licensing Evaluation" Vol. 1-3, Oak Ridge, Tn., 1984.
Possible Improvements The current design of the MODSTAR concept indicates that improvements are possible in the areas of fuel inventory, power distribution and absolute power. It is possible to significantly reduce the critical mass required by, for example, reducing the diameter of the units, modifying the bundle configuration (to increase moderation in the center) and changing the unit pitch or the type of moderator. The power peaking factors could be reduced by using fuel zoning or burnable poison distribution strategies. Finally, it now appears that the current design could operate at higher power. However, implications on such aspects as control requirement and core configuration have not been examined. CONCLUSION An o r i g i n a l small r e a c t o r concept based on a n integrated fuel-sodium cylindrical unit and operating at atmospheric pressure can supply high temperature heat to produce electricity. The study of the reactor physics and thermal hydraulics indicates that the concept i s feasible. The proposed design combines simplicity, attractive safety features and a high conversion efficiency. Significant improvements to the current design can tie achieved with minor efforts. ACKNOWLEDGEMENTS
The help and support of the Reactor Physics Branch of Chalk River Nutiear Laboratories, for the reactor physics calculations, and of the Applied Physics Division of Argonne National Laboratory, from the thermal hydraulics section, are gratefully acknowledged. REFERENCES 1. R.F. Mann, L.G.I Bennett, "Analysis of the Operating Experience of Small, Military Nuclear Reactors", Final Report for DRBO, Task 25B0ZA, 1978. 2.
J.S. Glen, J.W. Hilbom, "The Canadian Slowpoke Heating Reactor", AECL-8257,. 1983.
3.
J.V. Donnely, "Safety Aspects of the Nuclear Battery Reactor Design", 14 t h CNS Annual Nuclear Simulation Symposium, Pinawa, Man., Apr. 88.
54 CNS 9th ANNUAL CONFERENCE, 1988
10. T.B. Fowler, D.R. VonQy, G.W. Cunningham, "Nuclear Core Analysis Code: CITATION" ORNLTM-2496, 1971. 11. S.L. Kbutz, T. Taylor, A. McReynolds, F. Dyson R.S. Stone, H.P. Sleeper jr., R.B. Duffield, "Design of a 10 kW fceactor for Isotope Production, Research ami Training Purposes", 2 n d Int. Conf. on Peaceful Use of Atomic Energy, Vol. 10, Geneva, 1958, pp. 282-286. 12. H.M. Dcmanus, R.C. Schmidt, W.T. Sha, V.L. Shah, "COMIX-1A: A Three Dimensional Transient Single-Phase Computer program for Thermal Hydraulic Analysis of Single and MultiComponent Systems", NUREc/CR-2896 Vol II 1983. 13. M.G. Cooper, "Saturation Nucleate Pool Boiling A Simple Correlation", First U.K. Conf. on Heat Transfer, Vol. 2, Cnem. Symp. Series, No. 86, 1984, pp. 785-793. 14. Dow Chemical Company, "Engineering Manual for Dowtherm Heat Transfer Fluids", Samia, 1987.
NEUTBOHIC DESIGN OF THE AMPS REACTOH CORE
R.E. Stone and A.F. Oliva ECS - Power Systems Inc.
1500-112 Kent S t r e e t Place de V i l l e , Tower B Ottawa, Ontario K1P 5P2 INTRODUCTION
Regulating Rods The Autonomous Marine Power Source (AMPS) i s a
nucleai—electric power plant designed for submarine application. The necessarily compact AMPS reactor (1) is water-cooled and fuelled with uranium-2irconium-hydride (J-ZrH^ 5) e u t e c t i c a l l o y , clad in s t a i n l e s s s t e e l . Erbium burnable neutron poison is incorporated in the fuel matrix to meet the design fuel burnup lifetime requirement while avoiding the need for more than an operating margin of excess reacti"loy. The reactor heat source (RHS) of the prototype AMPS plant supplies 1500 kWt to a low-temperature organic Rankine cycle energy conversion unit (ECU) which in turn generates 100 kWe, net. This paper describes the reactor physics studies performed in the AMPS reactor core design process. DESIGN REQUIREMENTS A number of requirements were e s t a b l i s h e d for the n e u t r o n i c design of the AMPS reactor core i n s u p p o r t of p l a n t performance and s a f e t y
objectives. Core Burnup Lifetime In order to meet the design target of 5 to 7 years between r e f u e l l i n g s in the proposed submarine application, a minimum core burnup l i f e t i m e of 1000 full power days to f i r s t r e f u e l l i n g was e s t a b l i s h e d . Based on 210 operational days per year, this represents a load factor of between 60? and 85$. Minimum Installed Excess Reactivity In the t r a n s i t i o n from zero power cold operating conditions to full power hot, both fuel and c o o l a n t t e m p e r a t u r e s increase, resulting in a reactivity loss. This is in addition to the operational reactivity loss associated with xenon-135 and the long term losses associated with fuel depletion and fission product buildup. The Installed excess r e a c t i v i t y , which i s compensated by the regulating rods, ensures that the reactor can remain c r i t i c a l at full power to the design e n d - o f - l i f e after taking into account both short term operational and long term reactivity losses. Minimizing excess reactivit. reduces the number of control rods required for reactivity control and reduces the peak steady s t a t e overpower achievable following a full control rod withdrawal, thereby supporting AMPS safety objectives.
In order to avoid the design complexity associated with installation of a system to perform boron reactivity shim in the coolant, the regulating rod system is required to have sufficient reactivity worth to compensate for short term operational reactivity changes and all long term changes due to fuel burnup until t h e core e n d - o f - l i f e . For p r o d u c t i o n r e l i a b i l i t y reasons, the r e a c t i v i t y worth requirement must be met even with one rod stuck out-of-core. Safety Shutdown Rods Safety shutdown r o d s , d i s t i n c t from the r e g u l a t i n g r o d s , are r e q u i r e d to have s u f f i c i e n t r e a c t i v i t y worth to bring the reactor to a cold shutdown condition following a t r i p with one rod stuck out-of-core and with the core in i t s most reactive s t a t e . Power Distribution The reactor core must be sized to ensure that for any expected core operating s t a t e , the associated core power distribution will not r e s u l t in any fuel element temperature exceeding the long term fuel steady state operating limit of 750°C. This level has been d e t e r m i n e d b a s e d on c o n s i d e r a t i o n of i r r a d i a t i o n and fission-product-induced fuel growth and deformation. Reactivity Coefficients Significant negative reactivity coefficients for fuel temperature, coolant temperature and coolant void are required. These play a key r o l e in p r e v e n t i n g s i g n i f i c a n t fuel o v e r h e a t i n g , following a l l postulated initiating events, by assuring that the reactor power production is at all times commensurate with the rate of heat removal from the core.
THE REACTOR CORE
The AMPS reactor core design i s based on Uranium-Zirconium-Hydride (U-ZrH^.g) fuel. In t h i s type of fuel, the majority of neutron moderation occurs in hydrogen homogeneously contained in the fuel material. This fuel has been successfully employed in 63 General Atomics TRIGA r e a c t o r f a c i l i t i e s in 23 countries and has logged over 300 reactor years of operation since 1958 under conditions of
CNS 9th ANNUAL CONFERENCE, 1988
55
fission power, heat flux, burnup and coolant pressure and temperature similar to those of the AMPS design. Benefits associated with the use of this fuel type include a large, prompt, negative temperature coefficient, of reactivity and the exceptional r e t e n t i o n of f i s s i o n products w i t h i n the fuel matrix. These c h a r a c t e r i s t i c s play s i g n i f i c a n t roles in o p t i m i z i n g the s a f e t y and o p e r a t i o n a l performance of AMPS (,?).
i n the g r i d i l a t e s . Figure 2 depicts the r e a c t o r v e s s e l assembly, showing the f u e l , r e f l e c t o r , i n t e r n a l supports and the c o n t r o l rod drive mechanism compartment located above the core r e g i o n . Neutron f l u x monitor tubes are located outside of the b e r y l l i u m r e f l e c t o r w i t h i n the reactor vessel.
The reactor core of the prototype AMPS plant consists of a hexagonal array of stainlesssteel Glad fuel-moderator elements arranged in an F - r i n g l a t t i c e as shown i n Figure 1. Seventy nine of the 91 available sites are reserved for fuel elements, six are reserved f o r f u e l - f o l l o w e d s h u t o f f rods and the remaining six sites are for clad boron carbide regulating reds. A radial beryllium reflector surrounds the core. Graphite sections integral with the f u e l elements above and below the f u e l l e d region serve as a x i a l r e f l e c t o r s . Fuel-followed shutoff rods each consist of boron carbide absorber integral with the fuel rod i n a follower arrangement. When called upon to t r i p the reactor, the entire absorber rod replaces the fuel rod as i t leaves the core region.
Figure ?:
AMPS Core (Elevation)
CALCULATIONAL METHODS AND DESIGN STUDIES
KOUUIIM MOSi OI.CK.O'.OIO.OI), MC O i l .
wen SK/raw urn, a.u.ao.v.lM. un i n . MTti ILL 0 I O 6 I M IN B
AMPS n e u t r o n i c design c a l c u l a t i o n s i n v o l v i n g urlticality, burnup and c o n t r o l r o d w o r t h s t u d i e s , were performed u t i l i z i n g multigroup d i f f u s i o n theory. This methodology, with s u i t a b l e d e s i g n a l l o w a n c e s , has been demonstrated to provide an adequate level of accuracy in predicting measured c r i t i c a l core size and reactivity requirements i n similar UZr-H fuelled systems. Calculations of control rod worth were performed by imposing nondiffusion boundary conditions in the form of neutron current-to-flux ratios at the edge of each conLrol cell simulated. Calculations of cell-averaged neutron cross-sections, whole reactor calculations and control rod parameters are described in more detail in the following sections. Cell-Averaged .Veutron Cross-Sections
Figure I :
AMPS CORE (Plan View)
The f u e l elements and the c o n t r o l rods are positioned by s t a i n l e s s s t e e l g r i d plates above and below the core region which i n t u r n are fastened t o g r i d p l a t e supports located above and below the b e r y l l i u m r e f l e c t o r w i t h i n the reactor vessel. Theau g r i d p l a t e supports also hold the b e r y l l i u m r e f l e c t o r i n place and are located so a3 not to obstruct the coolant holes
56 CNS 9th ANNUAL CONFERENCE, 1988
Cell-averaged neutron cross-sections used in the design analysis were generated for seven neutron energy groups (four thermal and three f a s t ) f o r r e p r e s e n t a t i v e AMPS operating conditions. Previous design analyses performed for similar U-Zr-H fuelled reactor cores have demonstrated that no s i g n i f i c a n t gain i n accuracy is obtained with the use of a more detailed energy group structure.(3)
Cell parameters for the heterogeneous fuelled r e g i o n of the core were generates d through cdlculations represent!ng each l a t t i c e ce11 in one-dimensional r a d i a l geometry. In the l a t t i c e c e l l model, the fuel and cladding dimensions r e p r e s e n t the rod e x a c t l y , and the outer dimension i s specified to represent the correct core water volume f r a c t i o n . A l l c e l l - a v e r a g e d f a s t broad-group c r o s s sections , representing energies above 1.125 eV, were generated using the f a s t portion (GAM) of the CGC-5 code, where 99 f i n e energy group c r o s s - s e c t i o n s are c o l l a p s e d i n the energy v a r i a b l e over a s p a t i a l l y independent f l u x derived by d i r e c t fine-group s o l u t i o n of the B-j n e u t r o n s l o p i n g down equations for each discrete reactor region. Nordheim1s resonance i n t e g r . i l methods were used to generate fast c e l l - a v e r a g e d c r o s s - s e c t ions for resonance absorber materials.
o p t i m i z a t i o n analyses - v a r y i n g r e f l e c t o r thickness, fuel composition and burnable poison loading to meet burnup, power d i s t r i b u t i o n and r e a c t i v i t y management requirements - were performed with the 1-D r a d i a l mode 1. Then, a ?.~D r-z model was used to converge on the optimized design, containing optimum weight fractions of erbium and uranium f u e l dispersed i n the fuel alloy necessary to s a t i s f y the 1OGC f u l l power day burnup target and simultaneously achieve the minimum excess r e a c t i v i t y required to ensure cr i t i cal i t y at f ul 1 power over the design burnup l i f e t i m e . A t y p i c a l model i n r-z g e o m e t r y i n c o r p o r a t i n g homogenized f u e l , r e f l e ; t o r , s t r u c t u r a l and shield regions is provided i n Figure 3«
C e l l - a v e r a g e d t h e r m a l broad-group c r o s s s e c t i o n s i n the f u e l region were generated u s i n g the multigroup cross-section code GTF. GTF computes the s p a t i a l l y dependent thermal spectra at each predefined mesh point in the l a t t i c e c e l l with the discrete ordinates method u s i n g P] s c a t t e r i n g and S^ q u a d r a t u r e approximations. Cell-averaged cross-sections are o b t a i n e d by performing a simultaneous 5 pace-and-energy collapse of the f i f t y - e i g h t ( f i n e ) thermal energy group cross-sections. N o n - f u e l region ce11-averaged thermal broadgroup cross-sections were obtained by averaging the 58 f i n e group c r o s s - s e c t i o n s over the s p a t i a l l y independent f l u x derived by s o l u t i o n of the Ej form of the i n f i n i t e medium thermal spectrum equations, using the thermal portion (GATHER) of the GGC-5 code. S c a t t e r i n g k e r n e l s were used to p r o p e r l y describe the n e u t r o n - m o d e r a t o r atom i nteract ions. The bound hydrogen kernels for h y d r o g e n i n water were generated by the THERM1DOR code, and those f o r hydrogen in zirconium hydride were generated by the SUMMIT code. These scattering '<• rnels have been used t o a d e q u a t e l y p r e d i c t the t e m p e r a t u r e dependent water and zirconium hydride spectra, measured at the General Atomics LINAC f a c i l i t y (3).
Reactor Calculations The b r o a d ^ g r o u p n e u t r o n c r o s s - s e c t i o n s , generated as described above, were used in whole reactor calculations performed usi ng the CITATION ( 4 ) code. CITATION, a Cinitedifference representation of neutron d i f f u s i o n theory, wa3 used in both the one-dimensiona 1 and two-dimensional formats i n the AMPS core des i gn ana 1ys i s. A l l re levant bur nup cha i ns were s p e c i f i e d i n order to c a l c u l a t e fuel depletion and f i s s i o n product buildup. F i r s t , a one-dimensional (1-D) r a d i a l model was formed i n c y l i n d r i c a l geometry w i t h a x i a l neutron leakage represented by group dependent bucklings obtained from an i n i t i a l c a l c u l a t i o n performed with a two-dimensional (2-0) r i g h t c y l i n d r i c a l ( r - z ) model. The m a j o r i t y of
Figure 3:
CITATION r-z Model of AMPS Core
Reactivity Coefficients Fuel t e m p e r a t u r e , coolant temperature and coolant density c o e f f i c i e n t s of r e a c t i v i t y were evaluated over the e n t i r e range of parametric v a r i a t i o n representative of the AMPS operating envelope. These e v a l u a t i o n s employed 1 -D r a d i a l CITATION models of the optimized core u t i l i z i n g b r o a d - g r o u p c r o s s - s e c t i o n sets e v a l u a t e d o v e r a wide range of m a t e r i a l
CNS 9th ANNUAL CONFERENCE. 1988 57
temperatures and densities. To determine each reactivity coefficient, criticality calculations were performed with cross-section sets representing variations in either fuel temperature, coolant temperature or coolant density w i t h respect to the reference condition. The p a r t i a l derivative of r e a c t i v i t y with respect to each independent variable was then determined. Control Rod Studies The AMPS c r i t i c a l assembly of Figure 1 contains twelve control rods - six aluminum followed boron car bi cte {Bq C) regu 1 at i ng rods operat i ng in two stagge"ed banks of three rods and si x fuel element-followed B^C safety shutdown rods. Under tvDical f u l l power operat i ng conditions, the sa y shutdown rod absorbers are poised approximately 3 cm above the core region and the oorresponding in-cort^ shutdown rod sites are f u e l - f i l l e d . One bank of regulating rods i s normally f u l i y withdrawn, leaving an alumi num follower in the correspondi ng l a t t ice site. The remaining regu 1 ating rod bank is p a r t i a l l y wi thdrawn to an extent determi ned by the available core r e a c t i v i t y . Although the r e l a t i v e l y large t o t a l number of control rods together with necessarily compact nature of the AMPS c r i t i c a l assembly dictated t i g h t packing of control rod sites in the core, the development of very com pa ct rod drive mechanisms permitted progression of the design. Control rod worth calculations were performed using a 2-D X-Y planar representation of the AMPS reactor. Axis I leakage was taken into account through the use of group dependent bueklings for each reactor region derived from the 2-D r-z calculations. Neutron c j r r e n t - t o i lux r a t i o s u t i l i z e d at the edge of each control c e l l were determi ned by performing onedimensional neutron transport calculations in seven energy groups using the code DTFX in onedimensional c y l i n d r i c a l geometry. In these calculations, the cell containing the control rod was surrounded by a homogeneous core region. Rod i n t e r a c t i o n e f f e c t s from neighboring control ce 11s have not been found to be significant in determining these currentto-flux ratios. Specific studies following:
undertaken
included
the
Control Rod Total R e a c t i v i t y Worth, Both r e g u l a t i n g and safety shutdown rod t o t a l r e a c t i v i t y worths were obtained, respectively, by comparing the AMPS nomi nal zero power co 1 d KEFF v a * u e ^ a 1 1 rods withdrawn) with KgFF obtained from 2-D X-¥ simulations in which either a l l BtjC regulating rods were deployed in-core or a l l BjjC safety shutdown rods were deployed in-core. Highest Worth Stuck Control Rod. These studies were similar to those described in the previous paragraphs except that various combi nat Lons of 5 regulating rods inserted or L> shutdown rods inserted were evaluated to determine the t o t ^ l c o n t r o l worth of e i t h e r the r e g u l a t i n g or
55
CNS 9th ANNUAL CONFERENCE. 1988
shutdown rods, assuming that the highest worth rod is stuck out of the core in the f u l l - o u t position. Power Peaking E.fects due to Control Rod Positioning. Regulating rod aluminum followers act as weak neutron moderators when inserted i n t o an o p e r a t i n g r e a c t o r core. This moderation effect results in power peaking i n fuel rods adjacent to the f o l l o w e r - f i l l e d s i t e . Ful1 or p a r t i a l regu1 at ing rod insertion also causes deviations from the reference power distribution. In the event of excessive power peaking, the core operating power level would have to be derated accordingly. Power peaking associated with regulati;ig rod positioning was evaluated for a l l regulating rod arrangements by compar i ng spat i a l power distributions of off-normal rod arrangements w i t h those distributions representative of tne reference c r i t i c a l assembly configuration. Similarly, for shutdown rods, operation with a s i n g l e rod stuck in-core would also lead to limited power peaking. Power peaking effects associated with various stuck rod c o n f i g u r a t i o n s were evaluated i n a manner similar to that used for the regulating rods. Flux Measurement Distortion due to Control Rod Pos i t ion i ng. R e a c t i v i t y c o n t r o l system behaviour during reactor operation depends upon signals provided by the neutron flux monitors. Neutron flux signals In IMPS are calibrated onl i n e to represent thermal power. However, changes i n regulati ng and safety shutdown rod position during reactor operation lead to flux shape changes whi ch can depress or elevate measured neutron flux signals, even i f reactor power remains constant. In compensating for flux shape changes, three classes of operating conditions have been addressed i n the r e a c t i v i t y control algorithm: C i) Single B^C control rod stuck f u l l y i n core, r e p r e s e n t ! ng a rod control mechanism failure. (ii) One bank of BjjC regulating rods f u l l y i n core, representing typical operation at power. (iii) Single Bi,C (for potentially any Bi,C absorber location) control rod stuck f u l l y outof-core, with other B^C rods f u l l deployed i n core. Again t h i s scenario is representative of a control rod drive mechanism f a i l u r e . The method used for analysis of flux peaking produced at the flux detector s i t e s , arising from the three previous classes of operating conditions, was analogous to that used in the power peaki ng stud ies. In this case, however, distorted-to-nominal thermal flux ratios were calculated at the flux detector locations shown in Figure 1. RESULTS Core_ Design Parameters The key destgn parameters characterizing the final design or the AMPS reactor core sized for a 100 kWe plant output are summarized in Table I.
Table 1 AMPS CORE DESIGN PARAMETERS FOR 100 KWe PLANT
Fuel Type Uranium Loading Enrichment Erbium Loading Uranium-23!) Mass Element Outer Dia. Cladding
U-ZrH! ft 23 wt % 19.7 wt % 0.81 wt % 9 Kg 3-716 cm SS 301
Reflectors Radial Reflector Nominal Thickness Height Axial Reflector
Beryllium 10 cm 38 cm Graphite (integral with fuel)
Core Fission Power Burnup Lifetime Nominal Diameter Height (Fuelled Region)
1.5 MW 1500 MWd 41 cm 38 era
Core Burnup Effects and Reactivity Control Requirements Figure 1 shows the profile of calculated core multiplication factor as a function of core . burnup at the full power operating level of 1.5 MW. Also shown is the value of Kgpp at zero power cold clean conditions. The effects of xenon and samarium poison are visible as well as the effects of long term fission product buildup and fuel depletion. It is evident from the figure that in view of the relatively flat profile of multiplication factor versus burnup, a minimum installed excess reactivity of only 18 mk at zero power cold clean operating conditions will be required to operate the reactor at full power over the design burnup lifetime of 1000 FPD. Figure 1 indicates a 10 mk design margin at 1000 FPD to allow for systematic errors i n t r o d u c e d by the calculational scheme, uncertainties in crosssection data and fuel manufacturing tolerances. Any available excess reactivity beyond that required for operational and burnup losses which e x i s t s when the core is built can be reduced by adjustments to the i n i t i a l fuel loading. Table 2 summarizes the r e a c t i v i t y budget arising from the core design anaJ"sis. The analysis demonstrates that the regulating rod system has sufficient reactivity worth to meet both the operational and long-term reactivity control requirements, even with one rod stuck out-of-core. In addition, the shutdown rod system has ample margin to bring the reactor to a cold shutdown condition, even if only five out of the six rods operate following a reactor trip.
cone BURNUP (FPO)
Figure 1:
AMPS Neutron Multiplication vs Core Burnup Lifetime
Factor
Table 2
AMPS REACTIVITY BUDGET (MILLI - K) O p e r a t i o n a l Components T o t a l Cold-Hot R e a c t i v i t y Loss Xenon-135 Xenon Override
22 16 1
39 Fuel D e p l e t i o n and F i s s i o n Product Buildup A d d i t i o n a l R e a c t i v i t y for 1000 FPD NET CONTROL REQUIREMENT
18
U n c e r t a i n t y i n Fuel Temp. Defect (Note 1) U n c e r t a i n t y i n Burnup Calculation ADDITIONAL CONTROL ALLOWANCE
13
Minimum S h u t d o w n
J
Margin
MINIMUM REQUIRED CONTROL RANGE REGULATING ROD CONTROL WORTH SHUTOFF ROC CONTROL WORTH
61 ( 6 r o d s ) 82 (5 r o d s ) 61 (6 rods) 99 (5 rods) 80
Note 1: The actual fuel element temperature i s sensitive to the size of the fuel-cladding gap. This gap can only be made uniform within the tolerances o<" the fabrication process.
CNS 9th ANNUAL CONFERENCE, 1988 59
Power and Flux Distributions The power distribution in the core varies as a function ot core burnup state and the position of the r e g u l a t i n g rods. Figure 5 shows calculated radial and axial power distributions at the core beginnlng-of-life condition with the r e g u l a t i n g rods fully withdrawn. Calculated core peaking factors are provided in Table 3 for the reference case, in which the regulating rods are f u l l y withdrawn, and for the control rod configurations leading to the highest distortion in the power d i s t r i but ion. While c a l c u l a t i o n s indicate that the axial power d i s t r i b u t i o n does show s i g n i f i c a n t changes next to a p a r t i a l l y inserted regulating rod, the peak fuel element linear rating in this case is no higher than with the rod f u l l y withdrawn. Calculations confirm that the peak fuel element temperature at f u l l pewer is below the defined steady state operating l i m i t of 750°C for a l l power distributions achievable in practice. This confirms that no derating of the power w i l l be required for any control rod configuration.
RELATIVE RAD/AL POWER DENSJTY
TABLE 3 POWER PEAKING FACTOflS
Mean Fuel Exposure ( F u l l Power Days) 0 FPD
AXIAL PEAKING FACTOR
Hot Element Average Element
1000 FPD
1.23 1.25
1.21 1-23
Reference Case Two Reg. Hods I n One S h u t o f f Rod I n
1.22 1.33 1.18
1.21
GLOBAL PEAKING FACTOR
1.51
RADIAL PEAKING FACTOR
Distortions i n n e u t r o n f l u x measurements a s s o c i a t e d w i t h c o n t r o l rod maneuvering are summarized i n Table 1 f o r selected cases. This information is used i n the design of the r e a c t i v i t y control system to minimize the e f f e c t of short-term flux shape changes on reactor power regulation. In addition, flux distortion factors for those detectors used in the safety shutdown system are used in the establishment of t r i p set points on the high neutron power t r i p parameter. TABLE 1 FLUX DISTORTION FACTORS (Flux at Detector f o r case J / Flux at Detector f o r reference case)
Case
HAOIAL ZONE LOCATION
RELATIVE AXIAL POWEfl DENSITY
Reference (All Rods Out)
Reg. System Detectors A B C
1.0
60
5:
AMPS C a l c u l a t e d R a d i a l and A x i u Power D i s t r i b u t i o n s a t 0 FPD
CNS 9th ANNUAL CONFERENCE, 1988
1.0
1.0
1.0
.87
1.08
1.03
.87 1.03 1.10
Two Reg. Rod In
.97
1.05
0.95
.97
One S h u t o f f Rod I n (E7)
figure
1.0
One Reg. Rod In
Or.? S h u t o f f Rod I n (C10) 1.11
AXIAL 70NE LOCATION.
1.0
S . D. S y s t e m Detectors D E F
1.19 1.92
1.01 0.8
1.11
.95 1.16
1.03 1.19 1.11
1.08 0.80 1.01
Reactivity Coefficients
CONCLUSIONS
Table 5 shows the calculated values for the AMPS fuel temperature, coolant temperature and coolant void coefficients of reactivity at 0 and 1000 FPD, at representative AMPS operating conditions. Large n e g a t i v e r e a c t i v i t y coefficients are found in a l l cases, thereby s i g n i f i c a n t l y c o n t r i b u t i n g t o operating s t a b i l i t y and safety of the AMPS core. The increase i n fuel temperature coefficient with i n c r e a s i n g temperature helps minimize the r e a c t i v i t y load associated with i n i t i a l system warmup w h i l e n o n e t h e l e s s p r o v i d i n g a s i g n i f i c a n t reactivity loss in the event of fuel overheating.
N e u t r o n i c a n a l y s i s h a s been performed t o achieve the f i n a l design of the 1.5 MW AMPS r e a c t o r core foi- t h e prototype AMPS p l a n t r a t e d at 100 kWe. The d e s i g n meets a l l s t a t e d performance r e q u i r e m e n t s . ACKNOWLEDGEMENT
The authors g r a t e f u l l y acknowledge the contribution of G. West of General Atomics in La Jolla, California in the preparation of the neutron cross-section sets used in this study. REFERENCES 1. J . S . Hewitt, "The AMPS 1.5 MW Low-Pressure Compact R e a c t o r " Proceedings of the F i r s t I n t e r n a t i o n a l Seminar on Small and Medium-Sized Reactors, Lausanne, Switzerland, (1987). 2. A.F. Oliva and J . S . H e w i t t , "Design and Safety Features of the AMPS Nuclear E l e c t r i c Plant", P r o c e e d i n g s of t h e Ninth Annual Conference of t h e Canadian Nuclear S o c i e t y , Winnipeg, (1988). 3. G.B. West, W.L. Whittemore, J . R . Shoptaugh, J.B. Dee and C O . Coffer, "Kinetic Behaviour of TRIGA Reactors", General Atomic Technologies, San Diego, C a . , U.S.A., (1967).
TABLE 5 COEFFICIENTS OF REACTIVITY FOR AMPS COSE Coefficient of Reactivity
0 FPD
1000 FPD
Coolant Void (Density) [mk/$void]
-1 .8
-1.6
Coolant Temperature (Note 1) [mk/°C]
.05
>4. T . B . Fowler and D.R. Vondy, "CITATION: Nuclear Reactor Core Analysis Code, (Rev.2)", Oak Ridge N a t i o n a l L a b o r a t o r y , Oak R i d g e , Tennessee, U.S.A., (1971).
-.03
Fuel Temperature [mk/°C] a 280°C
a 700°C
-0 .1 -0 .17
-0.07? -0.13
Note 1: T h i s i s e x c l u s i v e of changes i n coolant density with temperature.
CNS 9th ANNUAL CONFERENCE, 1988 61
NEUTRONICS DESIGN OF A SMALL, SOLID-STATE, PASSIVELY COOLED REACTOR, THE NUCLEAR BATTERY
J.V. DONNELLY, K.S. KOZIER AND G.R. PENNER
Atomic Energy of Canada Limited Whiteshell Nuclear Research Establishment Pinawa, Manitoba
ABSTRACT The current status o£ the neutronics design of the Nuclear Battery reactor is described. The selection of core geometry and in-core materials are discussed. The reactor safety characteristics of the Muclear Battery are briefly presented.
lattice cell calculations analyze the fine structure associated with fuel rods and in-core devices using neutron transport theory and provide homogenized cross section data for t.ie full reactor core calculations using neutron diffusion theory.
CONTROL ROD INTRODUCTION This paper reports the present status of the neutronics design of a small, solid-state, passively cooled reactor, the Nuclear Battery, to be employed for the small-scale production of electricity in remote locations. The economic pressure to achieve competitive total unit electricity costs has increased both the output power and the lifetime energy output from the Nuclear Battery to levels above those previously evaluated for special use applications. Thus, the main purpose of this study is to verify the neutronics feasibility of the revised design requirements and to define an initial core concept. *
•VAPORIZER
VESSEL
•HEAT PIPES
Nuclear Battery Reactor Description A schematic view of a typical Nuclear Battery core may be seen in Figure 1. The reactor core is cyllndrically shaped, vith fuel rods, heat pipes and control rods oriented vertically. Solid graphite within the core acts as the neutron moderator and heat transfer medium; surrounding the core are solid graphite neutron reflectors that reduce the fuel inventory. The core and reflectors are contained within a stainless steel vessel acting as a containment boundary. The reactor core is about 2 metres in diameter and 1.5 metres high. The mass of the Nuclear Battery reactor is approximately 17 Hg. The nominal core temperature is 500°C. The core of the Nuclear Battery is currently being designed for the production of 2400 kW(t) for a lifetime of 15 full-power years. Complete details are given by Kozier and Roslnger ( ] ) .
FIGURE 1:
NUCLEAR BATTERY CORE SCHEMATIC
The lattice cell calculations described in this report were all carried out using the AECL Research Company version of VIMS (2,3) vith ENDF/B-V nuclear data (4). The diffusion theory calculations for the reactor cores were all carried out using the 3DDT (5) code. For the analysis of TRISO coated particle fuel in WIMS, a particle fuel resonance treatment was developed.
Energy Conversion and Primary Heat Transport Neutror.lcs Calculation The neutronics analysis of the Nuclear Battery has been carried out to evaluate the reactor physics properties of the reactor core. The results of the neutronics calculations provide information such as the variation in fissile uranium requirements with changes in core materials and configuration, the changes in core reactivity with temperature and the dynamic properties of reactor. The neutronics calculations discussed in this paper are generally separated into two levels of detail: lattice cell calculations and reactor core calculations. The
62 CNS 9th ANNUAL CONFERENCE, 19BB
The cost and efficiency of production of electricity in small-scale applications from nuclear heat depends strongly on the energy conversion. In the energy range of the Nuclear Battery, 600 kW(e), and in the temperature range of 500 to 700°C within the reactor core, the most promising technology for producing electricity is a toluene Organic Rankine Cycle (ORC) engine. The 0RC engine will produce electricity at about 25X net efficiency and requires primary heat at about 400°C and a heat sink at about 20"C. A 1-kW electric ORC engine is being operated at the WhitesheH Nuclear Research Establishment of
AECL (WNEE) to gain operational experience and evaluate its performance. The primary heat transport technology selected for the Nuclear Battery are high-performance liquid metal heat pipes. The heat pipes are sealed tubes in which a fluid is vaporized in the evaporator zone and flows to the condenser zone. The condensed liquid returns to the evaporator zone by flowing in a film down the wall. The heat is transferred by the latent heat of the mass flow. In the range of temperatures in the Nuclear Battery, the liquid metal of choice is potassium. The operation of high-performance liquid metal heat pipes at the Nuclear Battary design point of 15 kW and 500°C has been demonstrated at VNRE, and continuing research is improving heat pipe performance and our understanding. The number and dimensions of the heat pipes and toluene vaporizers above the reactor core determine, to a large extent, the radial dimensions of the Nuclear Battery core. The selection of the heat-pipe wall material must take into account a number of multidisciplinary, and sometimes conflicting, constraints: - Under normal operating conditions, the heat pipe is a pressure vessel under approximately 1 atm" external pressure. The wall material and tliickness must be designed to resist fai.'ure by creepbuckling for the design lifetime of the Nuclear Battery. - During postulated accident scenarios, the temperature of the core and heat pip
TABLE 1:
HEAT PIPE WALL MATERIAL CONSIDERATIONS
Heat Pipe Wall Material
Limiting Temperature (°C)
Zircaloy-4
235 U Mass Penalty (kg)
1200
316-L
1050
0
12
1 atm = 101.325 kPa
Selection of Nuclear Battery Fuel Two fuel materials have been considered for the Nuclear Battery: uranium carbide and TRISO coatedparticle fuel. Uranium carbide fuel has been demonstrated for use at the temperatures and power densities required for the operation of the Nuclear Battery in the WR-1 research reactor, but is no longer in production in Canada. TRISO coatedparticle fuel was developed for high-temperature gascooled reactors, the operating temperatures and power densities of which significantly exceed those of the Nuclear Battery. The TRISO coated-particles consist of 0.05-cm-diameter U0 2 spheres with four coating layers: graphite buffer, inner pyrolytic carbon, silicon carbide and an outer pyrolytic carbon layer. The small particle size and robust layers give the TRISO coated-particle fuel very high retention of fission products and strength at high temperatures and high fuel burnups. For an otherwise identical core, the uranium-235 requirements for a uranium carbide and TRISO coatedparticle fuelled core were calculated to be very similar with no significant cost advantage for either fuel type. From reactor safety considerations, however, TRISO coated particle fuel is superior to uranium carbide because of its much greater resistance to the release of fission products.
Optimization of Reflector Thickness 'he axial and radial reflectors of the Nuclear Battery reduce the fuel inventory requirements of the core by reducing the leakage of neutrons from the core. Using assumed costs for uranium-235 and graphite, the size of the reflectors was varied and the amount of uranium required to make the reactor critical at end-of-life (EOL) was determined. The variation in relative cost with radial reflector thickness is shown in Figure 2, and with axial reflector thickness is shown in Figure 3. From the results shown in Figure 3, a radial reflector thickness of 30 cm was chosen. The variation of cost with the thickness of the axial reflectors in Figure 3 does not show a minimum over the range considered (up to 30 cm thick). A thickness of 20 cm was selected on the assumption that the costs components not considered (nor currently available), such as steel and concrete vessels costs, would penalize thicker axial reflectors. In any case, the variation of cost with axial reflector thickness is small beyond about 20-cm thickness.
Burnable Poisons
860
Nb-lZZr
Based on the type of information shown in Table 1, Nb-1% Zr was selected as the reference heat-pipe wall material for current design studies. The improved temperature limits of Nb-I£ Zr relative to Zircaloy-4 are considered a significant benefit during a temperature excursions, even at the cost of the increased fuel inventory.
The physics and design of the Nuclear Battery result in a relatively large amount of fuel being required to sustain the core through its lifetime relative to the minimum critical mass of a zero lifetime core. This large change in the fissile inventory of the core results in a large change in the excess reactivity available in the reactor during its life, as can be seen in Figure 4.
CNS 9th ANNUAL CONFERENCE, 1988 63
.8 -
\
.
\
"
"
•
•
•
•
•
'
"
"
.
-
"
•
-
•
•
•
•
'
,
•
, • - • • • '
•
ive
•
.
J5 .4
.2
and placed in the inner portion of the core alone. A number of burnable poisons were considered. With an optimized configuration of burnable poison, the total variation in the core k-effeetive over the core lifetime (after the initial fiss-on product transient) Is less than 7 mk. The maximum rate of change in reactivity Is about 3 ink per full-power-year. The small residual variations in reactivity could be compensated for with low-worth control by devices not connected with any active control system. They would be accessible only during reictor maintenance periods, and so do not introduce safety concerns. The small end-of-llfe reactivity penalty associated with the use of burnable poison will be minimized during design optimization and Is considered an acceptable trade-off for the Increase In safety.
Orantam QnphtU Tbtal
0
5
10
IS
20
23
30
33
40
45
50
55
60
Radial Reflector (cm) FIGURE 2:
VARIATION OF COST WITH RADIAL REFLECTOR THICKNESS
-
o 0
Uratam OrapUU ToUl
500 1O00 1500 2000 2500 3000 3300 4000 4500 3000 5500
Time (Days)
0) OS
FIGURE 4:
CORE REACTIVITY VARIATION
Reactor Temperature Coefficient
10
.. — — —
12
14
IS
" "
IS
20
22
24
2«
28
30
32
Axial Reflector (cm) FIGURE 3:
VARIATION OF COST WITH AXIAL REFLECTOR THICKNESS
Compensation for the variation In reactivity with fuel burnup could be carried out In a simple fashion by moveable control rods, as Is done In many power reactor designs. High-worth moveable control rods, however, add a large potential source of reactivity in the event of a failure in their control mechanisms and do not enhance the basic safety of a reactor design. Burnable poisons may be placed In the reactor along with the fuel at the start of life to minimize the use of moveable control rods. If the burnable poison is carefully designed, the variation of reactivity during core life can be minimized. A number of alternative materials and configurations for burnable poisons in the Nuclear Battery were investigated, including placing the poison uniformly in the fuel, lumped between fuel elements, 64
CNS 9th ANNUAL CONFERENCE, 1988
The variation of reactivity with reactor temperature is negative throughout the life of the core, as shown in Figure 5. Comparison of the fresh, mid-life and the end of life cores shows that the variation of reactivity with temperature does not change much over the life of the core. Assigning a nominal upper limit of 1900 K to the proven upper limit of persistent TRISO coated-particle fuel temperatuie with no increase in fission product release, the inherent temperature feedback will guard against fuel failures during reactivity excursions of up to about 65 mk from the normal operating condition. However, the use of burnable poisons limits the maximum possible excursion to only about 9 mk. Hence, the upper temperature of the TRISO particle fuel will never be approached. The negative temperature coefficient at the reactor operating point will allow the rea.tor to be used as a constant-temperature heat source during normal operation: with little or no artive control of the reactor, the fission power will adjust itself to balance power demands and return to the initial temperature.
range of postulated accident scenarios evaluate-) in time-dependent, rather than static, simulations. The accident scenario simulations provide information about such things as materials temperatures during these accidents. From these, safety margins may be calculated and design features may be evaluated. Accident scenarios have been analyzed that span the range of potential failures of reactor components and systems that would significantly reduce the heat removal capacity from the reactor and/or add significant amounts of reactivity in an uncontrolled fashion. The accident scenarios and results have been described in detail in anothev paper (6) and will not be discussed here.
293
493
893
»«
'09J
1293
'493
1093
1093
Temperature (K) FIGURE 5:
NUCLEAR BATTERY REACTIVITY VARIATION WITH TEMPERATURE
Moveable Control Rods in the Nuclear Battery The current design of the Nuclear Battery has four moveable vertical rods available for decreasing core reactivity. Three of the control rods are positioned on a circle at a fixed distance from the centre of the core, the fourth being located at the centre of the core. The size of the four control rods will be chosen so that when fully inserted their total reactivity depth will guarantee reactor subcriticality under all conditions. The shutdown depth requirement for the cold critical state of the reactor is about 47.4 mk (43 mk for cold to hot reactivity and 4.4 mk for xenon compensation). The total depth of the four nominal control rods is currently about 120 mk. This will be reduced as the details of the reactor control requirement and cold shutdown requirements are determined in detail.
An example of the behaviour of the Nuclear Battery during a simulated dual accident is shown in Figures 6 and 7. In this scenario, a loss of regulation from the cold condition occurred, and the control rods were withdrawn from the core at the maximum rate of 0.8 mk/s to an added reactivity of 51.4 mk. During the first four minutes of the transient, the reactor power has not increased sufficiently to raise the core temperature significantly- Between the fourth ind tenth minute, the core temperature increases and the power decreases until the core temperature is in equilibrium with the reactivity addition of the moving control rod. The core temperature continues to increase in equilibrium with the control rod reactivity addition until the control rod has been withdrawn completely from the core at about 80 min. After the reactivity additions have stopped, the reactor becomes sub-critical and the powei is reduced to decay heat until the core has cooled sufficiently for the reactor to become critical at about 300 min. After 300 min, the core power and temperature oscillate at low, benign levels because of the phase lags between power and temperature feedback.
The maximum rate of travel o£ the four control rods will be chosen as the minimum required for reactor control, i.e., the requirement to bring the reactor from a cold shutdown state to the hot critical state in an acceptable length of time. If we allow for the necessity of limiting the rate of change in temperature In the reactor components and the fact that the reactor would normally be operated in a base-load mode, approximately one hour vould be a reasonable lower limit, which translates to a maximum reactivity insertion rate of 0.8 mk per minute. During normal equilibrium operation of the reactor, the control rods will only be inserted to have a reactivity worth sufficient to allow power maneuvering and compensation for the residual variation of reactivity with burnup, approximately 4 mk (1 mk for operational control plus 3 mk for burnup between shim intervals).
Basic Dynamic Safety of Nuclear Battery Design To realistically assess the safety of the Nuclear Battery, we must evaluate its performance under a
Time (min) FIGURE 6:
MODERATOR TEMPERATURE VARIATION DURING POSTULATED ACCIDENT SCENARIO
Ihe lesiilts oi Itte stenat^o iiscusseA above are typical for all the postulated accidents studied and Indicate that the Nuclear Battery will remain safe without any intervention of operators or "engineered" safety systems.
C N S 9th A N N U A L C O N F E R E N C E , 1988 65
r
the production of electricity could be achieved, and that the basic strengths of the concept will be retained. The high level of safety of the Nuclear Battery can be achieved at the higher power and longer lifetime through the application of TRISO coated-particle fuel and burnable poisons.
<
REFERENCES 0)
(1) KOZIER, K.S., WD ROSINGER, H.E., "The Nuclear Battery: A Solid-state, Passively Cooled Reactor for the Generation of Electricity and/or HighGrade Steam Heat", Presented at the 1988 CNA/CNS Conference, Winnipeg, Manitoba, 1988, June.
o cu \
o (2) ASKEW, J.R., FAYERS, F.J., AND KEMSHELL, P.B., "A General Description of the Lattice Code VIMS", JBNES, 4(4), 564, 1966. •••I
10"'
io°
io'
to 2
io 3
-
•»
io*
Time (min) FIGURE 7; REACTOR POWER VARIATION DURING POSTULATED ACCIDENT SCENARIO
(3) DONNELLY, J.V., "VIMS-CRNL, A User's Manual for the Chalk River Version of WIHS", AECL-8955, 1986. (4) CRAIG, D.S., AND FESTAKLNI, G.L., "The ENDF/B-V WIHS Library", Atomic Energy of Canada Limited unpublished report, CRNL-27B4, 1985. (5) VIRGIL, J.C., "3DDT - A Three Dimensional Multigroup Dlffusion-Burnup Program", LA-4396, 1970.
Life-Cycle Considerations The Nuclear Battery is designed to have a high degree of long-term reliability and low-maintenance requirements, and so the long-term performance of the components within the core must be evaluated. The fuel performance requirements of the Nuclear Battery are within the demonstrated performance of TRISO coated-particle fuel. The maximum fuel burnup in the Nuclear Battery is approximately 85 000 Mtfd/tonne heavy element, and the fuel has been demonstrated to burnups of 140 000 MUd/tonne of heavy element. The operational temperatures of the TRISO coated-particle fuel in the Nuclear Battery are about 600°C, while the fuel has been demonstrated in longterm irradiations at 1100'C. The long-term performance of the heat-pipe vails within the Nuclear Battery will be determined by their resistance to creep buckling, and the designs will conservatively assure performance. The behaviour of the graphite under long-term irradiation in Nuclear Battery will be determined by the total neutron fluence and the temperature during the irradiation. In the Nuclear Battery, the graphite temperatures will be in the range of 500 to 600°C, and the total number of displacements over the core life was calculated to be 1.06 displacements per carbon atom (or a dose of 1.06 x 10 21 neutrons/cm2). Based on a limited assessment, the stored energy will not be a problem during irradiation at these temperatures. The changes in the structure of graphite during Irradiation should not be sufficient to cause mechanical or heat transfer problems in the Nuclear Battery, and in any case, the operation and performance o£ the Nuclear Battery will not be sensitive to small changes in the graphite properties.
CONCLUSIONS The neutronlcs calculations for the design of the Nuclear Battery indicate that the current targets for
CNS 9th ANNUAL CONFERENCE, 198B
(6) DONNELLY, J.V., "Safety Aspects of the Nuclear Battery Reactor Design", Presented at the 14th Annual Nuclear Simulation Symposium, Pinawa, Manitoba, 1988, April.
STARTUP OF THE SLOWPOKE DEMONSTRATION REACTOR AND LOU POWER TI-f;T'
J. D. IRISH, B. M. TOUNES, AND C. M. TSENr,
Atomic Energy of Canada Limited Research Company Chalk River Nuclear Laboratories Chalk River, Ontario KOJ 1J0
ABSTRACT This paper describes the approach-to-critical and the low-power physics and transient tests performed with the SLOVPOKE Demonstration Reactor and compares some of the test results with design calculations.
INTRODUCTION The SLOWPOKE Demonstration Reactor (SDR) attained first criticality on July 15, 1987. This reactor is a 2-MWt prototype of the larger 10-MWt SLOWPOKE Energy System Reactor presently being designed by AECL. Besides demonstrating the basic design principles and use of such reactors, SDR serves as a test bed for the design and analysis codes. This paper describes the approach-to-critical and the low-power physics and transient tests that have been performed and compares some of the test results with design calculations.
COMMISSIONING ION CHAMBER
Cf-252 NEUTRON SOURCE
ABSORBER PLATES
FIGURE 2: SDR CORE REGION
During some phases of commissioning, a 2 5 2 C f neutron source was located in the core region either at the centre, in place of the control rod, or in the Be reflector as shown in Figure 2.
DESIGN CALCULATIONS
SLOWPOKE DEMONSTRATION REACTOR Key lo culawey 1. Cc/e<4 bundles) 2. Control rod 9. Hoi tlamt duel 4. Primary heat exchanger 5. Coyer plat* 8. Waler purlllctllon lymm 7. Secondary heat exchanger I. Circulating pumpi 9. Control room
Reactor physics calculations were performed to support the design effort for SDR. These calculations were done in four energy groups using the finitedifference diffusion-theory code CITATION 1 . The four ?roup parameters for each core region vere obtained using the cell code WIMS-CRNL 2 . Where possible the predictions from the design calculations will be compared with the experimental results obtained during reactor commissioning. Since the design calculations were performed before commissioning to design the reactor and plan the commissioning experiments, they often do not simulate the experimental conditions exactly. However, in most cases, the design simulations are sufficiently close to the experimental conditions to allow comparisons to be made.
FIGURE 1; SLOWPOKE DEMONSTRATION REACTOR
SDR is a light-water, pool-type reactor with natural convection cooling which operates at ambient pressure. SDR and the reactor core are shown in Figures 1 and 2 respectively. The reactor core, approximately 30 cm x 30 cm x 50 cm high, consists of four fuel bundles containing U 0 2 fuel having an enrichment of 5 vtX ! 3 5 U in U. The core is surrounded by a radial beryllium reflector. Reactivity is controlled by a central control rod and four moveable absorber planes between the fuel bundles.
INITIAL POWER CALIBRATION An approximate measure of the reactor power was required for use during low-power commissioning. This vas obtained from two operational ion chambers located on opposite sides of the core outside the hot riser duct, and a third commissioning ion chamber located close to the outside surface of the beryllium reflector. Before fuel loading commenced these chambers were calibrated by measuring the ion chamber currents when the 2 5 2 C f neutron source of calculated effective power was placed at the centre of the
CNS 9th ANNUAL CONFERENCE. 1988 67
reactor-core rep ; on. The calculated effective power of the source is given by the formula: P * (» s X 2 M¥) / $R where
P = the effect; 'e power of the source, * s = the calculated flux at the ion chamber due to the source with no fuel In the reactor, and tR = the calculated flux at the inn chamber due to the reactor core operating at 2 MW.
A more accurate ion-chamber calibration will be determined later by recording ion-chamber currents while the heat produced by the reactor is measured using calorimetric methods.
FUEL LOADING AND APPROACH-TO-CRITICAL
Monitoring
During fuel loading and the approach-to-critical, the subcritical multiplication of neutrons from a Ct source, located at the centre of the reactor core region, was monitored wit): the commissioning ion chair'ior. The neutron multiplication factor is given by the formula:
to-critical was continued by removing the interstitial absorber pins. In order to appioach critical in a controlled mariner, the absorber plates were inserted a sufficient amount to more than compensate for the reactivity which would be added when a group of absorber pins vas removed. The group of absorber pins was removed and then an attempt to reach criticaiity was made by raising the absorber plates in steps. After 12 absorber pins had been removed, first criticality was attained with the absorber plates raised 702 of their full travel. After the removal of each subsequent group of 4 absorber pins an approachto-critical was made. Removal of absorber pins continued until criticality was obtained with all the absorber pins removed from the core, with the absorber plates raised 33% of their full travel. This compares with a predicted critical configuration with the absorber plates raised 292 or their full travel. This difference between the predicted and measured critical heights is equivalent to 17 mk. This is well within the uncertainty of 30 mk assigned to Ineffective values in the design calculations.
!b!
where M = the neutron multiplication factor, 1 = commissioning ion chamber current, and I s = commissioning ion chamber current with the source at core centreline, • the absorber plates raised, and no fuel in the reactor. As criticality is approached 1/M goes to zero. The fuel loading and approach-to-critical were designed such that, near critical, linear extrapolation of the plot of 1/M would predict criticality sooner than would actually occur. In this way a safe approach-tocritical could be performed.
Fuel Loading Before ftrei loadicg commenced, nine absorber pins were placed into interstitial positions in each of the four fuel bundles. Uith the control rod removed from the reactor and the pbsorber plates raised about halfway, fuel bundles were loaded sequentially into the core. During this period the subcritical multiplication of :ieutrons from the neutron source, located at the centre of the core, was monitored. After each fuel bundle was inserted or partially inserted the value of 1/M vas plotted against the amount of fuel in the core. Then the curve was linearly extrapolated, giving most weight to the points corresponding to the last two bundles loaded, i n order to predict a safe increment of fuel to load during the next step.
Approach-to-critical Criticality was approached by a combination of removing the interstitial absorber pins and raising the absorber plates. After the four fuel bundles were loaded, the approach-to-critical was continued by raising the absorber plates symmetrically. After the absorber plates had been fully raised, the approach-
68 CNS 9th ANNUAL CONFEHEMCE, 1968
CALIBRATION OF ABSORBERS Measurements were made to determine the reactivity worths of the absorber plates and control rod as a function of absorber position.
Method of Calibration A combination of neutron multiplication and reactor period measurements was used to calibrate the absorber plates as a function of position. After the absorber plates were calibrated, these two methods along with critical balance measurements were used to calibrate the control rod. To make a reactor period measurement, the control rod or absorber plates were raised to make the reactor between 0.5 and 1.7 mk supercritical. The resulting stable reactor period '-as then measured. The measured reactor period was then converted to a reactivity using a table of reactivity versus reactor period which was calculated using the inhour equation for SOS. For the absofber-piate reactivity calibration curve these reactivities were converted into reactiv' y gradients at the measurement point and a curve of reactivity gradient versus absorber plate position was plotted. The reactivity for any absorber plate position was then found by integrating under this curve. The subcritical multiplication measurements were changed into k-effective values using the formula k-effective = 1-(C/M) where C is obtained from a similar reactor configuration by equating reactivity gradients obtained just below critical, from subcritical multiplication measurements, to reactivity gradients obtained just above critical from period measurements.
Calibration of Absorber Plates The operating procedures for SDR require that the absorber plates be moved symmetrically so that after any reactivity adjustment all four absorb?!- plates should be at the same height. Thus neutron
multiplication and period measurements were made to determine a reactivity calibration curve as a function of the position of the absorber plr.tes for SDR, under the condition that at any given time all four plates would be raised the same amount. The results of these measurements ate presented as the points in Figure 3. The data points can be broken up into three distinct sets. The four points with the absorber plates raised furthest were obtained from subcritical neutron multiplication measuremer^s with 28 of the interstitial absorbing pins in the core. For this set of data the value of C used to calculate k-effectives was 1.38. The next set of data, for the central portion of the calibration curve, was obtained from period measurements with 24, 20, 10, 12, 8, 4, and 0 absorbing pins in the core. The ta",t set of data, for the absorber plates most of the way in the core, was obtained from subcritical neutron multiplication measurements with all of the absorbing pins removed from the core. For this set of measurements the value of C used to calculate Iteffectives was 1.13. For this data the left-hand scale in Figure 3 represents both reactivity and change in k-effective, depending on whether the data was obtained from period measurements or subcritical measurements.
multiplication measurements gi ve change in keffee five. In order to get react ivi ty, the changes in k-effective would have to be divided by k-effective.
Control Rod Calibration The control rod was installed in the reac tor• Then subcritical multiplication measurements, period measurements, and critical balance measurements were made in order to get information to enable the control rod to be calibrated. Previous subcritical multiplication measurements in SDR vere performed with the source located at the centre of the reactor. On^e the control rod was installed this pusi t ion was no longer available for the source. Therefore, the neutron source was moved to the source location in the Be reflector shown in Figure 2.
MM
1
-
_
£
/
i "6 -
A /
r
-X 5UBCR1T1CBL MULT.
HI
X
-4
> T- PEP.10D MEASUREMENTS
m u
-2
X
QSUBCRITlCflL MULT.
1 1
0
/ O
-8
-
x Q SUBCfllTlCflL HULT1PL1CBTION
—CflLCULflTlON MX
X PERIOD MEHSUHEHENTS CflLCULflTlON
t
-12 0
1 5
1
1
'
'
1
1
10 IS 20 25 30 35 40 45 SO 55 60 65 DISTANCE CONTROL ROD flfllSfO 1CMJ
y
-40 FIGURE 4: REACTIVITY VERSUS CONTROL-ROD HEIGHT
0
5 10 IS 20 25 30 35 40 45 SO 55 60 65 OlSTflNCE RB5QRBER PLflTES RfltSED
(CM*
FIGURE 3r REACTIVITY VS. HEIGHT OF ABSORBER PLATES
The results of these measurements are presented as the points in Figure 4. The solid line in Figure 4 is the previously calculated control-rod calibration curve. The calculation was done with the absorber plates raised 19.22 cm and the measurements were made with the absorber plates raised between 19.9 and 21.9 cm. While the agreement between calculation and experiment is reasonable, there is about a ten per cent difference in the reactivities predicted by the measured and calculated calibration curves. This may be due to the approximate method of model ling the control rod in the VIMS-CITATION calculation. This will be investigated further.
Comparison vith Predicted Absorber Plate Calibration Curve. The solid line fn Figure 3 is the previously predicted absorber plate calibration curve. The calculated reactivity values have been modified by subtract ing 17 mk from all reactivi ty values so that the curve will agree with the measured critical height for the absorber plates. For this curve the left-hand sc
REMOVAL OF ONE ABSORBER PLATE
There are several reasons for the differences between the measured data and the predicted curve. Two reasons are (1) that the measured data are for the reactor with various numbers of interstitial absorbing pins in the core while the calculated curve is for the core without any absorbing pins and (2) that only two constant values of C have been used to convert the measured multiplication factors to keffectives and the value of C probably varies somevhat vi th absorber p]ate position. Also the predicted curve gives reactivity while the data points whi ch were obtained from subcrltical
In order to show that SDR would remain shut down with one absorber plate fully wi thdravn and the remaining three plates fully inserted, measurements were made ot the critical position of thiee absorbers when each of the absorbers in turn was dtiven to its full-out pos11 ion. The minimum measured cri t ical height taken together wilh CITATION calculations show that with one absorber plate and the conttol rod raised to their full-out position and the other three absorber plates fully inserted, SDR will be at least 6 mk subcritical at a temperature of 24 degrees C.
CNS 9th ANNUAL CONFERENCE, 19B8 69
REACTIVITY WORTH OF SIMULATED VOIDS A measurement of the coolant void reactivity worth is required to contirm reactor physics methods used to predict this effect. To carry out this measurement, voids were simulated by a mixture of styrofoam spheres and vater contained in 36 acrylic tubes located interstitially in the fuel bundles. The hydrogen density in the mixture was about 37X of the hydrogen density in water.
To obtain temperature coefficients of reactivity relevant to the experimental situation, linear interpolation between the available data was used. The total calculated decrease in reactivity for a temperature change from 17.5 to 65 degrees C is 8.3 mk, which is 16% less than the measured value of 9.9 mk.
1
Another 36 acrylic tubes containing water only (water tubes), were used to determine the reactivity worth of the tubes alone so as to enable the void reactivity worth to be separated from any reactivity perturbation resulting from the acrylic tubes. The measured reactivity worth of the void tubes after the reactivity worth of the water tubes was subtracted was -9.8 mk. This is in agreement with a design calculation which predicted the worth to be -10 mk. The design calculation was a two-dimensional CITATION calculation which assumed that the absorber plates were fully inserted and that the void rods would be 50% void. In actual fact the absorber plates were not fully inserted and the void rods contained about 63% void. Because of this an accurate comparison between calculation and experiment will have to wait until three-dimensional CITATION calculations are performed for the actual experimental configuration.
i r
0
The reactivity worth of these 36 void tubes was determined from a series of period measurements and critical balances at regular intervals, while the 36 tubes vere removed from the reactor in groups of four, one from each bundle.
-1 -2 -3
I -" t -S > £ -6 or _ tr -7 UJ
-8 -9 i -10 -11 I 20
3D
40
50
60
70
TEMPEHflTURE IDEG Cl TEMPERATURE REACTIVITY EFFECTS FIGURE 5: REACTIVITY VERSUS TEMPERATURE The effect of temperature on reactivity was measured by using a 75-kV electric heater and forced circulation to raise the pool water temperature from about 17 degrees C to 65 degrees C over a period of several days. During this time the reactor was run at low power (~17 U ) . The changes in critical balance positions of the control rod and absorber plates were used to provide a measurement of the negative reactivity effect due to the increasing coolant temperature. In this way the reactivity effect of changing coolant temperature could be determined with reduced corrections for fuel temperature effects and no correction for xenon effects which would be required if nuclear heating were used to raise the coolant temperature. The reactivity change relative to an initial 17.5 degree C average core temperature is shown in Figure 5. The total decrease in reactivity over the range from 37,5 to 65 degree C was 9.9 mk, corresponding to an average temperature coefficient of -0.21 mk/degree C. Also shown in Figure 5 is a calculated estimate of the change in reactivity with temperature. This estimate uses coolant and fuel temperature coefficients of reactivity which were calculated before the experiment was performed using WIMSCITATION assuming a constant reflector (beryllium and pool water) temperature of 68 degrees C. The reactivity coefficients had been calculated for two situations: (1) the absorber plates withdrawn and (2) the absorber plates inserted. There is a large difference in the calculated coolant-temperature reactivity coefficients for tnese tvo conditions. In the actual experiment the absorber plates were at an intermediate position between full in and full out.
70 C N S 9th A N N U A L C O N F E R E N C E . 19B8
In viev of the approximations made in the calculations the agreement between experiment and calculation is reasonable. Further calculations with the absorber plates at the appropriate position, the correct reflector temperatures, and the change in dimensions as the reactor is heated are planned.
FLUX MEASUREMENTS The thermal flux shape in the core was measured by irradiating copper activation wires in one of the fuel bundles for one hour at a power of about 2 watts.
Experimental Details The 7x7 element bundle has 36 interstitial sites. Thirty-four lucite holders, each containing from 3 to 10 copper wires at distinct axial locations, were loaded into interstitial sites in one of the fuel bundles. Another lucite holder containing a full-length 500-mm axial copper wire was loaded into another interstitial site. Also two copper wires were wrapped around the corner fuel element closest to the core centre at positions above and below the bottom of the absorbers and two similar wires vere wrapped around the diagonally opposite corner fuel element at the same heights.
Details of Calculations The quarter core
three-dimensional CITATION model
used in the pre-startup design calculations was used to provide calculated flux distributions for comparison with experiment. In Ihis model the beryllium reflector is surrounded above, below, and outside by a 152.4 mm layer of water, corresponding to an infinite water reflector.
normalisation used, Figure 6 indicates that calculation overestimates the peak flux by 3.5%. Part or all of this discrepancy arises because the reactor simulation ignores the stainless-steel core support struc tures.
calculated flux peaking fuelled length.
The thermal group contributes more than 11% of the total calculated copper reactions at the cell boundary. Thus the experimental copper reaction rates provide a relative flux distribution dominated by the thermal (<0.625 eV) group. In this preliminary K comparison 20 of he experimental values were normalised to the calculated thermal flux distribution at a retctor power of 2 MU. The resulting normalisation factor was then used for the remaining flux comparisons between experiment and calculation. Since CITATION calculates average cell fluxes, a WIMS calculated cell-edge to cell-average ratio of 1.15 was used to de-ive interstitial fluxes in all fuelled regions. In the flux plots shown in Figures 6 and 7, the thermal flux unit is 10 I3 n/cm ? .s corresponding to a reactor power of 2 HW.
effects
above and below the
.BEBH l)lUM ,
FUEL
|
3 0
I
X.
I
\
m>
^ 2. 0
< f* * B
\
j
t)
f
CRLC
LOC
\A\
\\v
10
1 ° 1 10
\v \\\ \\\ "
6
+
> 0
f
EXPT
L
0. 0
\ \
/
P 1. 5 —i— i
•5
\ \
4
2. 5
IERHHL
Relative Neutron Flux Distributions
20
30
40
H0RIZ0NTHL DISTBNCE FROM CORE HX1S CCMI
FIGURE 7: RADIAL FLUX DISTRIBUTIONS
10
20
30
40
50
60
70
B0
90
DEPTH BELOM MflTER REFLECTOR
FIGURE 6: AXIAL FLUX DISTRIBUTION
Figure 6 shows the comparison between experiment and theory at the interstitial location nearest to the core centre. Note the good agreement between experiment and calculation including the large axial asymmetry caused by the absorber plates which come in from the top. Note also the large flux peaking above and below the fuelled region. The calculated peak element rating for this absorber configuration occurs in this element over the 2 cm mesh region centred at 57.2 cm below the top of the water reflector; this is just below the absorber tip elevation. The peak experimental flux also occurs in this element and as shown in Figure 6 is in good agreement with calculation, which provides confidence in the calculated peak element rating and therefore the corresponding power form factor. With the
Figure 7 compares calculated and measured radial flux distributions at three different heights along a line through the centre fuel row which extends perpendicularly from the absorber plates through the neutron activation site located in the beryllium reflector. The axial positions are numbered from 1 to 10 with 1 being near the top of the fuelled region and 10 being near the bottom of the fuelled region. Note that agreement between experiment and theory is good in the upper half of the core, locations 2 and 6, but near the bottom of the fuelled region, at location 10, the calculated results again are significantly higher than measured values. Other points of interest are the calculated thermal flux peaks in the beryllium reflector r-d near the core centre line below the absorber cip (axial location 10) caused by the water gap and the Zr absorber follower. The overall agreement between the measured and calculated flux distributions is good, indicating that predicted peak element ratings and power form factors should have an expected uncertainty of about + 5%.
Flux Gradient Across Fuel Elements The wires wrapped around the fuel elements were used to obtain azimuthal flux distributions around the fuel elements. The measured maximum/minimum flux ratio is 1,12 for the centre fuel element in the top section neighbouring the control rod and varies from C N S 9th A N N U A L C O N F E R E N C E , 1988 71
1,211 to 1.25 for the other three locations. The highest value of 1.25*.05 occurs at the corner fuel eleiaent top position. The peak flux occurs at the element surface farthest away from the bundle centre due to flux peaking in the water gap, thus there is a reduced maximum/minimum flux ratio for the centreelement top section, where the control rod removes any flux-peaking effect-
DETAILED STATICS CALCULATIONS Both the 17 mk over-prediction of the initial reactivity and the prediction of higher fluxes than measured at the bottom of the core indicated that structural material which had been omitted from the design calculations might be important. Because of this a new geometric model was developed for use with CITATION. This model incorporates all structural material within 15 cm of the top and bottom of the Be reflector and also uses a finer mesh spacing than was used in the design calculations when solving the neutron diffusion equation. A calculation for the initial critical case vith all of the interstitial absorber pins removed yielded a k-effective of 1.003. This is a significant improvement over the design calculations. It is planned to compare results from CITATION using this geometric model with the results of the other commissioning measurements.
simulation code, developed at the Chalk River Nuclear Laboratories, for modelling the dynamics and control system of SLOi/fOKE reactors. During all tests except 5 and 7, the absorber plafes were withdrawn in rotation to maintain approximately the same height. Tests 5 and 7 were conducted using a single absorber only. To produce 10% of full speed, the motor was switched on and off with a time ratio of 1:9 (i.e. 3 seconds on and 27 second? off). For 50% of full speed, a 1:1 ratio (usually 5 seconds on and 5 seconds off) was used. The withdrawal of the absorber plates was cont inued until the reactor power reached the 38 W overpower trip point, causing the mechanical shutdown system to drop all the absorber plates by gravity into the reactor and shut it down. Figure 8 shows the result of test #8 along vith a calculation of the transient using the computer code, TACLES.
„
I OOC-G'
•—
TACLES /
Expenm antal dat /
TRANSIENT TESTS
/
To demonstrate that the SDR neuIronic protection system can effectively shut the reactor down dining a startup transient, a series of tests was carried out by continuously withdrawing absorber plates until the reactor was t ripped by the neutronic protect ion system. Various initial condi t ions and absorber speeds were used. The trip point of the neutronic protection system was set at 38 W for all tests, with the log rate trip disabled.
J
\
j
TIME [SECONDS)
SPONTANEOUS FISSION SOURCE $ko - -19.2 mk REACTIVITY ADDITION- 1.4amWhWn(100%FUU. SPEED)
TABLE 1: SUMMARY OF TRANSIENT MEASUREMENTS FIGURE 8: SDR STARTUP TRANSIENT Test tt
25
Initial 'Cf Reactivity Source (mk)
-2 -2 -20 -20 -1.8 -19.2 -?.a -19.2
out out out in in in out out
Absorber Maximum Speed Rate-ol-iise (X of of Power full speed) 1 ( 7.1% ) TACLES Measured
10 50 50 50 100 100 100 100
4.1.1 14. n 14.0
B.n 14.1 14.2 25.2 24.0
4. 14. 12. 7. 13. 13.5 24. B 24.3
* Full speed = 8.382 mm/min.~
Eight separate tests were performed. In all cases, the reactor was initially shutdown with all absorber plates fully inserted. The absorber plates were then withdrawn to the appropriate level to provide the required initial reactivity. When the reactor power stabilized as indicated by the commissioning ion chamber, the test was started. Table 1 summarizes the eight tests that were performed along with the results of the TACLES simulation of the tests. TACXGS is a digital
72 CNS 9th ANNUAL CONFERENCE, 1988
The tests demonstrated that the neutrnnic overpower-ttip ptotection system can safely protect the reactor .vid effectively shut it down foi all these s t a M n p transients, including those starting from the lowest possible source level, spontaneous fission. For all tests, the reactor power did not overshoot the trip level. The experimental results agree within 10£ with the predictions of the computer code TACLES and thus validate the code iti respect of neutron kinetics and safety system dynamics. The results also confirm that the maximum rate-of-rise of power during the startup transient depends on the reactivity addition rate and neutron source level but is independent of the initial subcriticality. A higher rate of reactivity addition generates a higher maximum rate-of-rise of powe'. Note that at powers up to 38 V, tbermalhydraulic feedback effects are insignificant.
CONCLUSION The SLOWPOKE Demonstration Reactor was started up successfully. Low- power physics and transient tests were performed. The test results have been compared with design calculations. Neutron statics calculations using a more detailed model for the reactor have been started. The results of the tests were in reasonable agreement with design
calculations, thereby neutronics models.
increasing
confidence
in the
ACKNOWLEDGEMENTS J. W. Hilborn and R. H. McCarais participated in the supervision of the commissioning tests. The SDR operational staff ran the reactor and assisted with the experiments. Many people at both the Whiteshell Nuclear Research Establishment and the Chalk River Nuclear Laboratories contributed to the vork reported here. REFERENCES (1) FOWLER, T.B., VONDY, D.R., and CUNNINGHAM, G.W., "Nuclear Reactor Core Analysis Code: CITATION", ORNL-TM-2496, Rev. 2, 1971, July. (2) DONNELLY J.V., "WIMS-CRNL A User's Manual for the Chalk River Version of WIMS", AECL-895S, 1986, January.
CNS 9th ANNUAL CONFERENCE, 1986 73
CAREM : A SMALL ELECTRICITY PRODUCING REACTOR J.P. ORDONEZ AND J.J. GIL GESBINO INVAP S.E. EARILOCHE , ARGENTINA ABSTRACT
1.2.
The CAREM r5actor is a small e l e c t r i c i t y producing nuclear reactor. I t ' s power is 15 Mwe. I t uses natural convection cooling, a self-pressurized integrated primary c i r c u i t and passive emergency systems. I t is a modular reactor, each module being factory made and tested. In this way, the only work that must be car r i e d on site is the assembly and interconnection of the modules.
1.2.1. Modular design. The present nuclear enginee ring trend is towards standardization. Modern nuclear plants are b u i l t based on standard engineering, the licensing stages are standard and the big components are i d e n t i c a l . Reactors d i f f e r one from the other only by the modifications performed on the basic to adapt them to the s i t e .
The features of the CAREM reactor made i t an extre mely safe reactor, which is economically competitive in the range of 15-150 Mwe. 1.
PROJECT CRITERIA
1. 1. Objetives of the project The main objetive of the CAREM project is to allow the use of nuclear energy in ranges of lower power than the ones used at present.
Design c r i t e r i a
A low power reactor is f i t for a standard licensing and design, and enables, through the use of modu les, the planning of a real serial production. The CAREM design is modular, in a double sense : A. Each reactor is made up of twelve parts, nra nufactured and tested independently, and then taken to the site of the reactor to be assembled as a complete unit and interconnected with the rest. B. A plant is made up of several reactors, which together, provide the required power. The advantages of this c r i t e r i o n are :
The average module, in the current nuclear indust r y , is of about 1000 Mwe. A 600 Mwe plant is considered medium, and a 300-400 Mwe, small. The CAREM project is geared to the range of 15/150 Mwe, Such a power reactor could be used by any country and to different aims. -Electric energy production -Industrial steam production -Water desalination -Urban heating
-
Nuclear power has not been widely introduced in very low applications due to two facts: -The bigger the modules, the cheaper the nuc)ear energy. - I n order to use nuclear energy, a country requj res highly qualified people deciding, consequen_ t l y , upon large scale projects. Based on these »wo facts, the use of low power nuclear energy seems improbable. In the f i r s t stage of the project, the design c r i t e r i a , using the advantages of a very low power react o r , was to outdo the two i m p l i c i t limitations mentioned above. The answer was found making the simplest possible reactor, easy to operate and keep, and safe. This was accomplished by six design c r i t e r i a , which make possible a simple reactor, easy to b u i l d , easy to operate and safe, namely : A. B. C. D. E. F.
Modular design. Minimization of building and on-site work. Allowances in the design for power increase with the addition of more modules. Easy transport of components. Self-controlled design. Passive emergency systems.
74 CNS 9th ANNUAL CONFERENCE, 1988
-
Lower engineering costs. Lower Quality Assurance costs. Lower licensing costs. Lower interest rates while under construction. ( Lower radioctive inventory and less hazards, hence, higher public acceptability. Plants adjusted to the needs of the buyer, and the financial p o s s i b i l i t i e s . The p o s s i b i l i t y to add new modules in case of greater demand. Lower d i s t r i b u t i o n loss due to shorter dis tances between plants and consumers. Higher u t i l i z a t i o n of the installed production capacity. Lower spare-parts stocks and hence, lower maintenance costs. Lower interconnection network costs to more homogeneous d i s t r i b u t i o n of power.
1.2.2. On-site works Another advantage of a small power plant is that the on-site works can be minimized. Therefore, there is a reduction of dead times and costs of qualified personnel transport. This reduction in the CAREM project has been achie ved in two ways. On the one hand, modular design a Hows pre-assembly and testing of a l l the systems in the factory) once in the s i t e they are assembled and interconnected. On the ottwr hand, c i v i l works have been drastically reduced using pre-assenbled elements. 1.2.3. Growth of the plant. New reactors can be added to the plant, i f so is required by demand. All services which are in common are shared by the new modules and the existing ones. I t is foreseen, p a r t i c u l a r l y , only one control room for a l l reactors, only one effluent testing plant and only one fuel elements storage room.
] . .'. 1 . (.oiiijionen I '. I r a n s p o r I . M o d u I a r i | c \ I <|ii mi p l i e s t h e I r . m s p m -I t i l I m r . h r . I p . i r l s ; w h w h a r e i a it i IMI |>y r e g u 1.1 r mi M I i'. of I r a n s p u r I .1 l o n g I In- r o a d * . . I I p o s s i b l e , p . i r l s s h o u l d In- I r a n s p o r f e d i n M t r i f a r n e r s , o t h e r w i s e t h e y s h o u l d n o l -•/. e e l I.I) t o n w e i g h t , -.u I li.i I I h< *y ' -in In- '
Ifier'o f he
-.t.ile
I inn J.i J ihir,
fi'dii' '",
I 1 a i r . p o r I < us I s .ind a vo 1 ds
when 1hous i n g M i r •, 1 I <•.
0 I
t he
the
n.)t 111 ,i I
.1 v e t y
tt\
,K t i d i ' t t l : ,
( I c,ir
1 r i n d ' on'.e<|uoii' e ,
f ho p r o l i . i l n I i t. i e \
f.lu-rr
1;
<* n - ' i l r o d u / H o n
i>\ M w .
the ( tire
M
I ight
it1,
e)
Wiil.cr
I n I cgr'
gcfior.i I ing lo.it u r i " , ,
-\>re,'Air \
inorlu 1 c n.iiiiely
(.ARI M
( l
||
ji.jrl
(.h.inge*.
In
this
fiiis
,ind
f low
is
w.i'cr
f .K I
me-in i n g
pumps,
o p e n i t i on
this
.inrl
in
1 ndu< e d
in
difh-rcnt
fhe
,1
hy
rr.or
level',
f I tiv/,
he
ol
the
s y s I ems
r e-H l o r " . ,
I hrough
iis
Io
ye.irs
rind
g r - i d 1 en t
in
t he
(ore
and
stcim
tor-
f he
'.ore
Ih.if
a
Siifely
design l»y
of
I he
now
safefy
ol
iirnl
and
I'WU.
systems
s y s t cms
pn-v ious fi i g h e s t.
( onsensc
|»hiJosopny
aihievc more
sys-
t.fio
these
I hose
CIKI r g e i M . y
JIIIJSI
of
- o l d i ri(| now
news I
ri wor" I d w i d e
d o s i <|n
re,if l o r s
1 In- e i i i e n | e i i ( y t fie
1 mprov i ng
is
|»M > l o s o p h y
sale
I fie
ai h j i - v
Ihere
r'egrirds
M'-w
I lie
des i gns ,
:
ensure
-
of
between
su I f i'. ierit
n h t ,1 1 n o d .
emergent y
or'dor
I einpor'ii I u r e
dilfcr'-mc
< -in
indusfry
sd I i'l y
t fie
I fie
• 1 p p r u p 1 'i I o
s I iiriddnls .
/his
o(
hy
lundiiuioiil'i 1 In
in
SS i h I o
prim.iry
\\,r,
I h 1'.
(.nnvcf 1 on,
need
H M Ifor
I h i - < 0 0 I. i n I
>t 1 <)<>\,nit
type
INK i c i r
n.i I ur
densities
11 r e v I O N S
f el y
hy
•. i m p I c
iin
cvo I ved
lo
rig
'loos
very
-.y-.tcin
piping.
s I e.iin i j c n i ' t -i I o r
h i s s i vc
tems
s.i
in
t fie
is (MI-
s y s f cm'-,.
two
goals
re I i.ihlc
yr-l
:
111-
simpler
s y s I cms . f,AHl M m o d u l e
f out r o I let I,
natural As
niiiin
- for
upper*
n u -
f c,i I u r ' i " . . liy
r 001
Ihe
r i m ](•()••(• I c ( . t r i f.
1 n l e g r ,i I i-d I fie
r c g u i reiiien I s ,
I.AKI M MUDUI I
the
'.,
Iho
in
ft I
I i / 1 ng (ore
>' i \[ .
deliried
s t e,ii|y
pr'cssur'o
An
(oiner,
-Hid
] S ( iin
:
u c . i n 10111.
i i 1 ( ui I .
/etl.
is
so 1 f
(onveilion
regards
p.iss i vc
he
the
in I he
:
diffcrcnl
,1 r 1 • I d o w n
hy
V. I .
I',
< }i,iiiilnTy .ihsorhs
f.hr'ough
for"
drop
'i
r e d ic I 1 o n
self
;\
f hd I
pl'C'.'.ur'c
A v-i[ittr
vi".1,'11 ,
r"«Mi 1 in
ficrent ly
Redut i n g h o l h
ensure
pressure
'ore
this
r e s 111 t •,
ones ,
On I he o l h e r hand , •,.! I <• iind r e ! i a o | c dos i gns f.rir: he t d h ' f i up hci.ii u',c ( f i r power p r o d i n . e d hy f,AM M i ' . '.in. 1 I I . lies 1 d e s , i I I he so do s 1 (jrr. ii re I ot hn 1 t.a I ! y -ind ei 1 mom i (..11 1 y i o n v e u i c n I , t he I i < en-, i ng '. I -igc'. find iii ( op I (ih i I i f y o f ncii rliy r e s 1 den f s w i f I f ', i 111(11 i / i e / l .
to
* onvi-i I i(jn
',III,I I I
qcr/cr-ilor,
I. '.'. i>. I ini-ffjcrK y s y s I em. \ p c 1.11 cmphas i s WMS l a i d wh 1 'ii d ( " . 1 (|n i ii(| pd",', i vc ' . y . I cm', of '.fiut down arid emer >\rj\( y 1 t>i>} mi 1. ',111,1 I I net w o r l " , mean 1 i I I I c 1 e I 1 all Ic net w o r t *., I h e r e J o r e j I i -, v e r y i i i i j i u r l
iis
operjtiii'l
pttwers.
i o n l r ' o l (or1,
I. p r e s s u r e
so
d m H I M t r . i n ' . i e n l •..
Mi,11
1111'. ' 1 1 I et " i o n .11 l o w . -in e a s y rind I I c y 1I1 | c M M I 1111 o j i ' M i i t i o n , '.o .1-, I.n In- ii',cd i n n e l w o r l - s of -iriy t y p e .
indif'f
power',
|o I-V.IJI•(•
01 iii 111 e s
I.;1.1.. ',i'|f M i n i m i . D e s i g n inn'. I e n a b l e e a s y o p e r a I I DM a lid ',| |M] 1 111 I MM I Mj I S ys r 'MM', . I fir pOWCr Illll', I ,.IJ I IJ in. 11 11 .i I I / ad.j i r , I I n demand, so ' ha I I he rea< t o r ' .111 WMil- ( i ' , 'i base ,nid v i m base p I ant ,iiid 'i'Inn t power ' yt I i n g .
rin
MSI-,
npor.ifiri'i
ill
I I n '. 1 r i l i T I D I I
r.
n o IJ I n i n
Ihe
emorgern y
a
inherently
sale
pr'cssui" i / e d , arid
sei ond
very goa 1 ,
sys I ems,
re.ntor:
1 rit e g m f . c d ,
simple
t.o
too led
operate.
I.fie C A M M m o d u l e
t.fid I
is
:
s i inp I c ,
has
h i gl y
r e I i a h l c s y s I ems n o t d e p e n d i n g on hurridn i n f c r v o n l i o n , or' need of e |ef. t r it. i I y . A f t e r - s h u t d o w n i t i s p r a i . l i { .11 I y ini|ioss i h I o t o
r e I y on t hem, so a wool' was
I .IKMI
ris dctttl I tr 10 ; it 1 11' rWfirds , pos i I i vo uio.isui"es a r e r e i | u i rct\ \ar I h e s a l o l y o l tfie p l a n t . I lie (,AM M m o d u l e fias priss 1 vc omorgeru.y s y s I ems , wfi i 1 h a l s o pr'uve s ii.ipl o r f. ) M I I t he emerge in y s y s I ems of I fie p r e s e n t p| ,mt s .
• N.i 111 m I (.Oliver t i o n (.(JO I i ng . • I'll 1 , 1 , i v I'liierfjoMi y ',y'i I cur,. '{'.?. 'i) I i g h t Wiiler* ,in'I e n r i f h o d UTMII i 11111 : He.ivy w a t e r ' i n (i Mil.) l | power p K i n I imp l i e , >i ( omp 11 • x o p o r - i I i o n •ind .1 M i m p l o x d e s i g n due t o f.hc w.il.cr I r c . i l inerit., I r i l i u m p n i d i H . I . i o n find o n - l i n e r e l u e l i n ( | . Hint i 1 , why .1 l i ' j h l . wx< 00rl*. '1. 1 .. h) ! nt ofjr.i 1 n i l pr- inui r y 1 i rx.ui I. : Ih i \ me,in*, l.hii t. f h e ',le,ifii qiwr.tt.nr r, \u(.\\n\rt\ w i l . h i n l.hr> | i r e ' , ' , i j r o 1 v i " . ' , e l . I n f.hi , w
l i e n e r a I d('S( r' i | ' l ton .
I i g . I shows a (f i .igram of t he p r c - s u r - e v e s s e I ,( on ( d i n i n g t.he M I r e , t h e r i s e r 1 , t.he s I cam g e n e r a t u r rind t he down-( omor. [he 1 00 I ri nt fie,1 I s up i n 1 he f.ore , goes up t h r o u g h f h e r i s e r , \\u\Vf\ ,\ \V,\\'{im** huMflroil ,uni e n / f i l . y dc (jri'i' 1 .) I in n , goes dtjwn t fie s t ciim goner.] t or , w h e r e i I i s (.001 of I , tirid r e t u r n s t o l.hc 1 o r e I h r o u g h I he downf 011)1 T " .
I fie {(.ore p 1 us sures
d i I f erent.o p I us
r isor)
flowni o m e r )
in
dens i I ies
arifl
I he
p r o v i dos
t l i e o s f a l l I i shine n 1 o f
1.11 I d f he ,i
l>i'tween I eg
the
{ s t cam
d r i v trig I I ow h y
fon.c
hoi
leg
g o n e r si t o r t he I
en-
n.i t u r - i 1 t . o n -
v o i t, i o n . 1 fie ft i f I e r e r K e i n I e v e I s h o t w e e n f l i e < o r e a n d t fie s t o i l in g e n e r a 1 o r , p c r m i t s t fie a d , j u s (.men I o l I ho d e s i g n Ut inovifh* Jhr n o o d o d I low l o t o o ) (he M i r e .
C N H «*th A N N U A I
n O N M ' H P N C F . 19BH
75
in the tanks to keep the core cool for a week.
PRESSURE VESSEL — STEAM GENERATOR
CORE
Under loss of coolant accident condition, which in this reactor, due to the integrated primary circuit, is limited to a very small one, a safety injection system keeps the core covered by water. The core is cooled by steam generation. The steam passes through the break to the containment, where it is condensed. Eventually, the containment reaches its design pressure and temperature. At this moment, the neat loss through the walls is equal to the heat that is being generated in the core, so there is no net heat accumulation in the containment, and therefore, no further increase in the pressure. The injection system keeps the core covered by water for a week after an accident. Fig. 2 shows these systems in operation. 3. CAREM 15 3.1. General aspects
FIG. 1
The steam generator 1s a once-through, helically coiled type, with the tubes wrapped around the chimney that separates tha riser from the downcomer. The secondary fluid flows through the tube side, producing over-heated vapor. In the upper part of the pressure vessel, in the steam chamber zone, there are the shutdown heat exchangers. They are used to remove the core decay heat by natural convection. The external flow of the heat exchanger is driven by natural convection. There is a water flow through the secondary side of these heat exchangers, that is evaporated in them, and to a shutdown condenser. In this "intermediate" circuit, the flow is also driven by natural convection between the heat exchangers and the condenser. Two water tanks keep the shell-side of the condenser f i l l e d with water, which evaporates and is released to the atmosphere. There is enougn watsr
Progress ha^ been made as regards engineering to prove the feasibility of a 15 Mwe CAREM system. The following topics describe the core, the sceam generator, the pressure vessel, the containment, the emergency systems. Then, there is a reference to the CAREM system during transients and finally a general description of the plant. 3.2. Core The core is uranium enriched, light water cooled. I t is made up of hexagonal fuel elements. Each element has 127 rods, out of which 103 are fuel rods. The cladding is of Zircaloy-4. There are 37 fuel elements, Nineteen fuel elements have control rods.
SHUTDOWN CONDENSER
t
CONTAINMENT
EMERGENCY INJECTION SYSTEM
SHUTDOWN EVAPORATOR
PRESSURE VESSEL FIG. 2
76 C N S 9th ANNUAL CONFERENCE, 1988
WATER STORAGE TANKS
Control rods are made up of Ag-In-Cd control pins, whose guide tubes provide structural unity to the fuel assembly. The rest eighteen fuel elements have burnable poison rods instead of control pins. The fuel is U0?, less than 5" enriched.
treatment system, the cooling system, and the cold shutdown system. They lie inside a cylindric tank. The tank is assembled in the factory and then taken to the reactor, as the rest of the modules.
Under complete core rechange, complete power of the core lasts 300 days. Under partial rechanges the burn-up is improved. The reactor is practically self-controlled, therefore only the central control rod is used for control. I t keeps the reactor pressure constant through all the operation range. 3.3.
Steam generator.
The steam generator is a once-through, helically coiled type. The secondary fluids flows along the inner side of the tube and the primary one along the outer side. I t is made up of Incolloy-SOO tubes. In the upper part of the steam generator there are four entry tube plates, from which the tubes come out, and go in a straight position into the steam generator. Then, they rise, helically wrapped around the centrall wall to end up in the four exit tube plates; a hundred and twenty tubes come out of each tube plate. In this way, with only one bundle there are four independent steam generators, corresponding to each pair of entry and exit tube plates. The steam generator produces 45 kg of overheated vapor, which avoid the water separators and the steam dryers. 3.4.
Pressure vessel.
The pressure vessel comprises the whole primary circuit. The core is located inside the pressure vessel. The riser (central zone) and the downcomer (peripherical zone) are separated by a flue, and the steam generator is on the outer side of the flue, in the upper part of the vessel. On the upper part of the steam generator is the steam chamber, which absorbs the pressure and volume variations during transients. The vessel is made of SA-553B, and SS-316 lining. 3.5.
Containment.
A special containment system, flooded-type, was chosen. It contains only the upper part of the pressure vessel, where all the openings are located. The water level remains above the reactor vessel cover, so that any leak from the primary takes place under water, and is imnediatly condensed.
Shutdown cooling and emergency systems. Both are passive systems consisting, mainly, of four tanks. Two of them feed the condenser during nojr mal shutdown. The other two inject water in the pressure vessel in case of loss of coolant. The shutdown condenser is a heat exchanger, the intermediate fluid goes along the tube side and the water fed by the tanks flows along the shell side; the vapor produced here is released to the atmosphere. Both heat exchangers are helical, similar to the steam generator, but smaller. They sre situated in the steam chamber of '.he pressure vessel. 3.8.
Stability.
Stability and load following are accomplished by the reactivity feedbacks by means of fuel temperature, and void fraction in the core, which is produced by operoting with saturated coolant in the core outlet. As a power increase in the core produces temperature increases in the fuel, the cooTant and the void fraction,, and the three effects lead to reactivity decrease, any power disadjustment is quickly limited ensuring the stability of the s>3tem. The load following is achieved by the coolant temperature variation produced by the demand through the steam generator. If demand is greater in the secondary systems, water in the primary one is cooled and provides positive reactivity when reaching the core, thus, ther is a power increase in the core until a balance is reached between generation and demand. 3.9.
Safety.
The CAREM module is an essentially safe design. It has been achieved using the advantages provided by a small power core, natural convection refrigeration and on integrated primary circuit. Passive emergency systems are possible due to the low power. Since the reactor is cooled by natural convection, there is a flow of water through the core, not depending on mechanic or electric means. The integrated primary system altogether avoids a loos of coolant accident (LOCA) condition. Little damage can only take place in the water treatment piping. The great quantity of water kept in the vessel ensures its safety in case of an accident.
Besides the openings for the piping, there are three more : one to allow people in and out, one for wasted fuel elements transport, and the upper one for the equipment. A polar crane allows handling of heavy equipment inside the containment.
The decay power is low because the core power is small. In this way, the use of natural convection systems to remove it is possible. Therefore, the shutdown does not depend on external electric power.
3.6.
These reasons show that the probability of a major nuclear accident in a CAREM module is 10 times less to the one in a conventional nuclear power station; being the consequences as small, too. Summing up, to equal installed power in the CAREM model or a con-
Nuclear auxiliary system.
They consist of a water provision system, a water
C N S 9th ANNUAL CONFERENCE, 1988 77
ventional nuclear plant the risk is ten times smaller in a CAREM module. 4.
CONCLUSSIONS.
The CAREM module introduces a new proposal in the nuclear energy market. Its excellent characteristics as regards safety, easy construction, assembling and operation are a consequence of the effort that was made to fulfill the criteria which ensure its success when competing with plants ranging 15/150 Mwe power. Keeping every advantage of the nuclear energy utilj_ zation, such as the substitution of Fossil fuels and the access to a higher technological level, the use of the CAREM module is feasible in countries having a smaller installed power capacity than the one needed upto the present, for the installation of conventional nuclear power plants.
78 C N S 9th ANNUAL CONFERENCE, 1988
Session 3: Accident Behaviour in Fuel Channels
Chairman: R.A. Brown, Ontario Hydro
CNS 9th ANNUAL CONFERENCE, 1988
RERLOODING PHENOMENA DURING ECCS OPERATION H, MOCHIZUKI AND Y. IIAYflMIZU 0 arai Engineering Center Power Reactor and Nuclear Euel Development Corporation (PNC) 4002, Narita, 0 arai, Ibaraki 311 13, Japan
ABSTRACT Turn-around time, heat transfer coefficient after turn around, and quench temperature were measured and correlated for various kinds of heater bundles and ECC water conditions. Applicability of the quench temperature correlation to the safety analysis has been verified by the analysis of the system tests.
tic tests and system tests using the /ITS safety experimental facility. Hayamizu '" studied the thermocouple instrumen tation methods for measurement of heater rod surface temperature during reflooding. As a result, several instrumentation method were developed.
BOTTOM FLOODING EXPERIMENT Apparatus and Experimental Conditions
INTRODUCTION
The schematic flow diagram of the apparatus is illustrated in Fig. 1. Cooling water was pressureized to 1 MPa in an accumulator by nitrogen gas and injected into the test section through a quick opening pneumatic valve, a turbine floK meter, and a flow control valve. Tests were conducted using single channel or parallel channels in which 28-rod or 36 rod bundle heaters were housed. Not only normal shaped heater bundles but also a bundle simulating a ballooning assembly shown in Pig. 2 was used to study the difference in cooling characteristics. To simulate the ballooning heater, blockages of 0.222 is in length were mounted on heater rods. Figure 3 illustrates an example of the thermocouple Because a core of the ATR is vertical, there are two ,PRES5URE TUBE arrangement on possible reflooding modes after injection of ECC water in 36-rod heater. the case of loss of coolant accidents (LOCAs). When a pipe rupture occurs, the coolant of APCI is injected into a water Table 1 and 2 show drum which is a header of feeder pipes. In the case of a the size of the channel and heat downcomer, a main steam pipe, or a riser tube breaks, fluxes of the bottom flooding occurs. On the other hand, when a feeder heater in several pipe breakes, top flooding is taken place by water from the locations, steam drum. Moreover, the ATR has a characteristic that respec t i ve1y. the reflooding speed of bottom flooding is quite high compared with that of light water reactors (LWRs). It is because a fuel bundle is provided in a pressure tube with a concentric arrangement, and flow area is rather narrow. PNC has conducted reflooding tests which consist of character is The emergency core cooling system (ECCS) of the Advanced Thermal Reactor (ATR) consists of three types of systems, i.e. a High Pressure Core Injection (h'TOI) system, a Low Pressure Core Injection (LPCI) system, and an Accumulated Pressurized Core Injection (APCI) system. The role of HPC1 is to decrease the system pressure by spraying water into a steam region in a steam drum when a small break occurs and the system pressure hardly decreases. When the system pressure decreases to around 4MPa, the APCI system is always initiated and thereafter the APCI system is succeeded by LPCI.
AMAIN STEAM
© TURBINE FLOW METER
OUTLET PIPES
M CLOSED VALVE
FIG. 2 CROSS SECTIONAL VIEW OF BLOCKAGE HEATER
TABLE 1 CONFIGURATION OF TEST CHANNEL ITEM
—-_
AXIAL HEAT FLUX
HIGH PRESSURE ( \ INJECTION PUMP*- *
NUMBER OF HEATER
FIG. 1 ATR SAFETY EXPERIMENTAL FACILITY
LOW POWER CHANNEL
COSINE
UNIFORM
M
78
OUTER DIAMETER OF HEATER PINS (mm)
14 b
11 9
NUMBER OF TIE ROD AND DIAMETER (mm)
H 5
EFFECTIVE HEATED LENGTH [mm) INNER DIAMETER OF PRESSURE TUBE (mm) , ACCUMULATED PRESSURIZEO CORE INJECTION SYSTEM
HIGH POWER CHANNEL
FLOW AREA
4 9 i 700
117 3 4 6X10 <
110 B 5 65X10 '
9 J4xfO '
U 91X10 '
11 7X10 '
IS 7x10 '
CNS 9th ANNUAL CONFERENCE, 1988 83
(
OUTER
6.2 K W / P I N X 1 3 = M 2 K W
MIDDLE 5.55KW/PINX
2 = 67KW
INNER
6=
3.6 K W / P I N x
TABLE 2 HEAT FLUX AT EACH POSITION
2IKW POSITION
AXIAL PEAKING
1
0 63
INNER PINS
MIDDLE PINS
OUTER PINS
OCAL PEAKINC 1 00
1 12
1 3«59xlO*
2 0735^10*
2 3226x10"
0 6i
II
1 05
7 1*33x10'
3 « > 9 * 10'
3 8710x10*
III
1 16
2 4WXI0'
3 SIBOXIO*
4 276£*IQ-
IV
08
1 rOBIxTO*
? 6330>(0*
2 949*xID"
correlated as a function of reflooding volocity as shown in Fig.4. Turnaround time defined as the time required for the slope of the cladding temperature rise to reach zero after ECC water arrives at the inlet of the heated section is correlated by the following dimension less expression.
THERMOCOUPLE POSITION (RADIAL) BOTTOM
6
5
4
3
2
1
t • V
= 11 I
THERMOCOUPLE POSITION (AXIAL)
-0.6 j + 0.12 - (1) XH
FIG. 3 POSITIONS OF THERMOCOUPLES A downcomer break port illustrated in Fig.l was used to conduct system tests including blowdown and ECCS operation. The following test conditions were chosen operational condition of the real reactor.
to
t(s) is turnaround time, V(m/s) reflooding velocity, x(m) elevation from the bottom of the heated section, and X H (m) T U R N - A R O U N D T I M E (s) : 4 X I A L L E N G T H (m) XH : H E A T E D L E N G T H (m)
cover the
1) flooding velocity: 0.02 to 1.27 m/s pressure: atmospheric, 0.4-6.3 MPa initial cladding temperature: 350 to 850 'C 4)" coolant temperature: around 30TJ copped cosine 5) power distribution: 1.02 to 1.95 kW/m 6) linear heat rate:
•
2) 3)
**' NORMAL CLU3TEI
Results There are three typical characteristics related to re flooding. They are turnaround of cladding temperature, heat transfer after turn-around, and quenching, effected heated length. Figure 5 shows the comparison between experimental data for both normal and ballooning clusters and the correlation. Data scatter around the correlation within the error of 20-30%. Heat transfer in the reflooding process depends on the configuration of the fuel cluster as well as thermal hydraulic conditions. Turn-around starts from both sides, i.e. bottom and top, of the cluster in the ATR's fuel. It has been clarified that turn-around time at a given elevation is V (m/s)
• 0.029 o 0.057 • 0.295 t : TURN-AROUND TIME (s) V : REFLOOOING VELOCITY (m/s) : HEATED LENGTH (m) X : AXIAL LENGTH (TI)
0.5 LENGTH FROM BOTTOM, X / X H (-)
1.0
FIG. 4 NON DIMENSIONAL TURN-AROUND TINE VS AXIAL LENGTH 84
CNS 9th ANNUAL CONFERENCE, 1988
REFLOODING VELOCfTV. V (m/s)
FIG. 5 RELATIONSHIP BETWEEN TURN - AROUND TIME AND REFLOODING VELOCITY Heat transfer coefficients after turn around were calcu lated from rod surface temperatures using the one-dimensional heat conduction code "OLION" . The heat transfer coefficient after turn-around is nearly constant as shown in Fig.6. Time-averaged heat transfer coefficients are correlated as a function of time-averaged reflooding velocity for both normal and ballooned clusters. The average velocity is used because the reflooding velocity after turn around oscillates. The heat transfer coefficients at the middle part of the heater are lowest in the cluster, as shown in Fig. 7, and are correlated as follows.
Nu = 0.0304 Re 0 - 5 Pr°-
^fe^dffe-^s^ "0
TIME WHEN ECC WATER • REACHES TO THE TOP < OF THE CORE
(2)
where the Reynolds number and the randtI number are defined using physical properties in the saturated condition. The heal transfer coefficient of film boiling increasi as the 0.8th p o w - of the Reynolds number under forced connection flow regime. However, gradient ^ the film boiling heat transfer coefficient against flow rate in bottom flooding is more gentle.
FIG. 6 TIME HISTORY OK HEAT TRMSFK.K COKFFICIBT AFTER TURN AROUND
pressure, subcuoling, and velocity. Heat capacity and thermal conductivity of the heater rods simulate U 0 z pellets and the cladding of a real fuel pin, as shown in Table 3. Heater surface conditions have an effect on quenching from the stand point of wetness. However, its effect would be small. Subcooling may also have an effect on quenching. However, coolant reaches the saturation condition soon after coolant goes up. flfter all, heat flux, peak cladding temper ature, pressure, and reflooding velocity were chosen as effecting parameters on quenching in the study. Siege!t:il Carbajo correlated the quench temperature as a function of lht;su four parameters. But this correlation did not predict v.elI quench temperatures obtained in our experiment, (t is probably because they made their correlation based on the PWR FI.FCIIT "' test in which reflooding velocity was in the range of 0.003 0.03 m/s. It is quite slow compared with refloodinj; velocity of the ATR.
TABLE 3 THERMAL PROPERTIES OH HEATER AND PELLET ITEM
"0 01
0 02 0 OS 0 1 0 2 05 AVERAGE REFLOODIP4G VELOCITY. V(rr
FIG. 7 HEAT TRANSFER COEFFICIENT AFTER TURN AROUND Generally, there are two methods to express quenching. One is an expression using the quench fromt velocity. In this case, it is usually treated that a quench front moves from bottom to top. The other expression is to correlate the quench temperature as a function of several factors such as physical properties and testing conditions. In our experiment, it was observed that quench fronts moved from the bottom and top to the middle and met each other at the position between the top and the center as shown in Fig. 8. Therefore, a model that the quench front moves from the bottom to the top can not appropriately express quenching which occurs in the f!TR. Effecting factors on quenching by fuel rods are heat flux, peak cladding temperature, heat capacity, thermal conduc tivity, and surface condition:,, and coolant conditions, i.<>. TIME: 20 m in nn
HEATER (MgO)
THEOHETOL DENSITY (kg/m1:
3-S!0
HEATER (BN)
ACTUAL DENSITY (k|/m'i
3.000
1 B50
13.100
1 09
0 879
0 314
AT 973K
HEAT CAPACI1 , [ k j / m ' Kj
3.266
1.633
3.266
AT 973K
THERMAL CONDUCTIVITY fW/m Kl
3S
29.1
33
AT 973K
THERMAL DIFFUSIVITY (m'/sj
1 18XI0"'
I-7SX1O"!
10X10-'
ATOTK
r %
#
•
B00±20 ( Z
0 q = 1.74 (kV /n X
q = 1.5? (kW/nn) i
3
-0 I o
8 01
^ O
, . ,. I
0 1
FIG. 0 RELATIONSHIP BETWEEN QUENCH TEMPERATURE AND AVERAGE REFL00DING VELOCITV 180
240 SEC
QUENCHED REGION
H UUKNCII BEHAVIOR FOR M ROD liliNDI.E
(V 0. l.Wm/s)
I
AVERAGE PXFLOODtNG VELOCITY. V [m/s'
SPACER
CNS 9th ANNUAL CONFERENCE, 1966 85
NOTE
10.960
SPECIFIC HEAT (kJAg K)
.i
FIG,
PELLET IUC1,)
2.270
Figure 9 shows the relationship between quench temperature A T o ( TQ T 3 ) and average reflooding velocity. Quench temperature increases with reflooding velocity in the region where velocity is greater than 0.2 m/s. On the other hand, quench temperature decreases with reflooding velocity in the region where velocity is greater than 0.02 m/s and less titan 0.08 m/s. The effect of retlooding velocity on quench temperature is constant in the other regions. Figure 10 shows the effect of peak cladding temperature Tp on quench temperature A T r T (-TQ - T S ) as a function of reflooding velocity ranging 0.08 to 0.2 m/s for various linear heat rates. The data are correlated by the following expression. O.I
AT,T=231.0 +0.0752 TP
f o r l . O S q S26 kW/m
(3)
I PRESSURE, p (MPa)
350sTPs850 t FIG. 11 EFFECT OF PRESSURE ON QUENCH TEMPERATURE
500 0.68 - 0.32 loe.o P for 0.584
400 -
for
0.1SPS2 MPa 2
UJ
a. 300 a: UJ a
In this way, the following quench correlation is obtained. LINEAR HE:AT RATE ( k W / m )
200
LJ
o
I
1.95,
X
o
•
UJ
a
i
300
400
i
i
500
600
1.77,
TO = TS +
1.57 1.13,
1.02
i
700
A T P T • Fv • Fp
(7)
1.74
i
800
900
The comparison of quench temperature between measured and predicted is shown in Fig.12. fiuench temperature at the hottest point can be predicted within the error of ± 1 0 % . Most data including blowdown and bottoo, flooding using ballooning heater can be also predicted by the correlation hi thin the error of ± 1 5 %.
PEAK TEMPERATURE, TP (°C) FIG. 10 RELATIONSHIP BETWEEN QUENCH TEMPERATURE AND PEAK TEMPERATURE The dependence of quench temperature on reflooding verocity is correlated by dividing ATQ by ATrr.
F v =
AT0
—
ATPT
for 1.5 0 , 4 7 V - a - 3 for 1.0 for 1.38V ° - 2 for
V •2 0.02 m/s
0. 02
As the result, quench temperatures obtained by character i s t i c tests are predicted by the following correlation.
100 200 300 400 S00 600 MEASURED QUENCH TEMPERATURE, T o EXP ATp
(51
FIG. Quench temperature also depends upon pressure. Equation (5) is applied to blowdown data Co correlate the dependence of quench temperature on pressure. Figure 11 shams the factor of pressure on quench temperature. This shows that quench temperature decreases with pressure when pressure is less than 2 MPa, and is almost constant when pressure is higher than 2 MPa. M
CNS 9th ANNUAL CONFERENCE, 1988
700
12 COMPARISON OF aUENCH TEMPERATURE BETWEEN CALCULATED AND MEASURED
TOP FLOODING EXPERIMENT Apparatus and experimental conditions The same apparatus shown in Fig.
1 was used
to conduct
top floodir;; experiments. However, in this case a pressure tube made of glass was provided at the upper part of the test section Vj observe the arrival of coolant at the top of the heater. The inlet pipe connected to the test section was left open in the atmosphere to simulate the posturated double ended break of the feeder pipe. Coolant was supplied to the steam drum by an 1.PCI pump after initial bundle heater conditions were prepared. Experimental conditions listed below were chosen. 1) 2) 3) 4)
supplying water flow rate : 40 to 50 m V h pressure : atmospheric^ initial cladding temperature : 500 to 740 'C coolant temperature : 70 to 90 T.
Results It was clarified that top flooding phenomena was dominated by the countercurrent flow limiting (CCFI.). When water was supplied to the test section through the riser tube which consists of horizontal part having 2 degree gradient, and vertical part as shown in Fig. 1, vapor generated in the heated section prevented water from flowing into the heated section. The CCFL occurred at orificing position where a shield plug for neutron shielding was provided. The ECC water supplied into the steam drum was 125 to 140 liters/s simulating LPCI water injection flow rate in the real reactor. However, the coolant flowing into the break channel was restricted to 1-1.7 liters/s. Therefore, almost all water supplied into the steam drum contributed to raise the water level of the steam drum. Coolant arrival time to the core after the initiation of the reverse flow from the steam drum was dominated by the horizontal length of the riser. Because the velocity of the water front like a wedge is very slow in the horizontal region. We observed the flow regime of air water under atmospheric condition using a lucite mock-up. Figure 13 shows the relationship between coolant arrival time to the core and turn-around lime and injection flow rale. The coolant arrival time was measured by observation through a glass tube, and turn around time was measured by the outputs of thermocouples. This figure shows that there is a small delay between coolant arrival time and turn ut'jund time. Moreover, this figure shows that the effect of injection flow rate on these limes is exponential. Wedge shaped water front proceeding velocity V(m/s) is correlated as a function of injection flow rate Q ( m V h ) by the following expression when velocity in the vertical riser Is neglected.
1
(8)
2.48 e x p ( - 2 . 7 x l O 2 fl ) i l.C "
——•—_____ OBSERVATION SYMBOL HEATING POWER INITIAL CLAD TEMP
_
THERMOCOUPLE
O
•
70-100 kW 30-500C
70-100 kW 500-700 C
30
\, V)
t=23e»p(-0.9xl03Vo) + l0
r 20 UJ
" " • * - — So...<&._.
50 100 INJECTION FLOW RATE ( n ' / h )
FIG. 13 EFFECT OF INJECTION FLOW RATE ON WATER ARRIVAL TIME TO THE CORE
150
Therefore, coolant arrival time to the top of the core is estimated as follows by dividing the horizontal riser length U by V.
t,
!.„ (2.48 i!xp( 2.7 * 1 0 z U) < 1.08}
(9)
Turn around time for top flooding is vague, however it is in ihe order estimated by the Fquation (1). In this case, however, longth x(m) should be defined from the top of the heated section. Histories of heat transfer coefficients after turn-around for top flooding art shown in Fig.14. The heat transfer coefficients are nearly constant like the bottom flooding curves shown in Fig.6 except those near the top. In the case of top flooding, estimation of flooding velocity is very difficult. However, coolant flow rate drained through the riser is estimated by the equation of Wallis ! 4 ) as about ] liter/s. Therefore, flooding velocity is roughly estimated as 0.2 m/s. The heat transfer coefficient after turn-around is about 35 W/m z K, and is only about 20% of that of bottom flooding,
POWER INITIAL CLADDING : 700 C TEMPERATURE WATER FLOW RATE 41 m'/ WATER TEMPERATURE : 85 C
"0
100
200
300
400
TIME AFTER WATER INJECTION (SEC)
U!F PIG. 14 HISTORV OF HEAT TRANSFER COEFFICIENT AFTER TOP FLOODING Quench temperatures of top flooding are shown in Fig. together with those of bottom flooding. The quench tempera tures of low flow rate less than 0.01 m/s for both top and bottom floodings seemed constant. In the correlation of Siegel Carbajo which is applied upto 0.03 m/s reflooding velocity, the power of the velocity is small as well.
DISCUSSION In bottom flooding, it has been clarified that there are two types of quench fronts. One goes up from the bottom and the other descends from the top. The mechanism of the descending quench front seems to have connection with the structure of the fuel bundle and the power destribution. When ECC water is injected into the pressure tube, water is splashed by the high temperature claddings. Therefore vapor flow accompanies i lot of droplets. These droplets seems to be collected by t'.ie tie plate of the fuel bundle, ftoreover, power level of the upper part is low because of power distribution. Therefore, cladding temperature of the upper part of the fuel are apt to turn around. Once turn around occurs, heat transfer coefficient increases, and heat is removed by water collected beneath the tie plate. Then, quench occurs at the top of the fuel, and this promotes
CNS 9th ANNUAL CONFERENCE, 1988 87
incorporated into safety analysis codes to predict cladding temperatures al various positions during blowdown tests conducted using thi; ATR safety experimental facility, To verify the applicability of the proposed quench correfigure 16 shows the results of bottom floodong after a 150mm lation, comparison between the correlation and data measured downcomer break and a 150 mm main steam pipe break. In these by cheng et al. 'Tl> was carried out. They measured quench cases, 3MW heating power was continuously supplied for about temperatures Tor various conditions using a circular tube 30 or 22 seconds, respectively, and was decreased to the heater of 91.4cm in length, 1.1cm in inner diameter and 1.04 mm in thickness. Figure 15 shows the comparison for the level of decay heat power. Cladding temperatures increased after KCC water was injected into the water drum. Because data measured at 344 kPa in pressure and for about 0.05 to condensation of vapor generated by depressuri/.ation 0.4 tn/s in reflooding velocity. Quench temperatures are prevented coolant from flowing into the con'. Quenching had l>ri;ri!-ted to within ' 5% error. occurred when the cladding temperature was around 400'C. General behaviors of cladding temperature are predicted by the safety analysis code SENIIOR. Especially, figures show that quench temperatures at different locations and events are predicted with good accuracy. collodion of water beneath the tie plate.
CONCLUSIONS
200
300 400 500 MEASURED Q U E N C H TEMPERATURE (C)
15 COMPARISON OF QUENCH TEMPERATURE BETWEEN MEASUREMENT AND CALCULATION It is important to show the applicability of the quench correlations which is used together with many thermal hydraulic models. Correlations mentioned before were
A) DOWNCOMER BREAK
TIME (s)
B) MAIN STEAM PIPE BREAK
Correlations in regard to turn around time, heat transfer coefficient after turn around, and quench temperature associated with I'"CC water injection have been proposed. These correlations are valid for both normal and ballooned clusters. Moreover, the quench temperature correlation can be used for not only a reflooding process but also for a rewetting due to an increase in flow rate during blowdown phase, Applicability of the quench temperature correlation to the safety analysis has been verified by the analysis of system tests using the full mock up safety facility of the
NOMENCLATURE Dh h k l.n Vu Pr 0 Ro t Tp To Ts V x XH i>,
: thermal equivalent diameter (m) : heat transfer coefficient (W/m z K) : thermal conductivity (W/mK) : length of horizontal part of riser tube (m) : Nusselt number ( h • Di, /k) •' Prandtl number : volumetric flow rate (mVh) : Reynolds number ( V • D h /V,) : turn around time (s) : peak cladding temperature ("O : quench temperature ( D : saturation temperature C O : average refloodinp, velocity (m/s) : location (m) : effective heated length (m) : kinematic viscosity of saturated vapor. (mVs)
REFERENCES II) Hayaitiizu, V., Kitahara, T. and Adachi, J., "Bottom Flooding Heat Transfer in a Pressure Tube Type Reactor " , Proceedings of JUICR Meeting, Toronto, Canada, (1978). (2) Siegel, A.D. and Carbajo, J.J., "A New Experimental Correlation for the Rewetting Temperature " , Trans. Am. Nucl. S o c , 35, (1980). (3) Langerman, M.A., "Quick Look Report for Sum i sea It; MOD 1 Test S 0G 3 (LOFT Counterpart Test)", NUREC/CR 0251 (July J.978). M) Wai I is, n.B., "One dimensional Two phase Flow" Hill Book Company, (1969).
lfi
U
BF.II/IVIORS 0 l ; CLADDING TEMPERATURES ASSOCIATED WITH JnOmm DOWNCOMKR BREAK AND 1.00mm W I N STIiAM I'll'!! IIHEAK
CNS 9th ANNUAL CONFERENCE, 1988
McGraw
1.11 Cheng, C.C., l.au, P . O . and Poon, K.T., "Itaisuri.'im.'nls of True Quench Temperature of Suhcooloil Water l.'ndor Forced Oonveclive Conditions " , Int. .1. Heal Mass Transfer, 28, I, (1985).
THERMOSS-II:
A MODEL FOR THERMOHYDRAULICS OF CANDU FUEL CHANNEL WITH SUBCOOLED STAGNANT INITIAL CONDITIONS
P. Gulshani
Atomic Energy of Canada Limited CANDU Operations Mississauga, Ontario L5K 1B2
ABSTRACT A two-fluid model, c a l V u THERMOSS-II (for THERMohydraulic Model Of Si inding Start) has been developed to predict the dut :tion of flow stagnation and stratificatio:i and fuel heatup in a CANDU reactor fuel channel in the standing start phenomenon. This phenomenon refers to a subcooled, stagnant initial channel condition and may occur in CANDU fuel channel following a small loss-of-coolant accident and subsequent refill of the channel with emergency core coolant. From this condition, the channel coolant boils and stratifies and the steam flows Co the end fittings where it condenses heating up the end fittings. Eventually, the steam reaches a vertical feeder pipe resulting in channel flow and refill. THERMOSS-II generalizes the previous THERMOSS model to account in a simple manner for highly non-uniform temperature distribution in the end fitting body during the end fitting heatup. This temperature distribution and the consequently lower duration of channel flow stagnation were observed in a recent series of standing start tests conducted in a modified cold water injection test facility with reactor-like channel-end fitting geometry.
Section 2 of this paper, reviews the results of previous standing start tests conducted in the cold water injection test facility. Section 3 presents an analysis and interpretation of the results of a recent series of standing start tests conducted in the cold water injection test facility with the channel and end fittings replaced by ones which are identical to those in the reactor. Section 4 presents the model THERMOSS-II developed to predict the duration of channel flow stagnation. Section 5 presents a comparison between the predictions of THERMOS'- :i and the results of the recent series of standing start tests. Section 6 presents a summary of the results and conclusions.
2. CWIT FACILITY AND TEST RESULTS To stuc ':he standing start phenomenon, many tests wer conducted in the Cold Water Injection Test (CWIT) facility (Figure i) at Stern laboratories (formerly Westinghouse Canada) Inc. Gamma densitometers on the vertical section of the feeder pipes measured the presence of any void. Thermocouples on the channel heater elements measured sheath temperature. The channel water level could be deduced from this temperature measurement.
1. INTRODUCTION The standing start phenomenon may occur in a CANDU fuel channel following, for example, a small break in the core inlet piping in the reactor with subsequent emergency coolant injection and loss of forced circulation. For these conditions, a break size could exist for which, at the end of pump rundown, the pull of the break balances the natural circulation force and the flow in some of the fuel channels is reduced to zero. This subcooled, stagnant channel condition is referred to as the standing start condition. This limiting channel condition may be difficult to achieve in practice. FIGURE I SCHEMATIC OF CWIT FACILITY
Subsequently, channel coolant boils and stratifies exposing the upper fuel elements and part of the pressure tube to steam. These elements and the upper part of the pressure tube heat up. The steam from the channel flows to the end fittings and condenses, heating up the end fittings. Eventually, the steam heats up the end fittings to the saturation temperature a'nd reaches the vertical section of one of the feeders. The resulting buoyancy - induced flow refills the channel and restores good fuel cooling.
The tests were conducted as follows. The channel power was raised to the desired value after the loop was brought to the desired subcooled stagnant condition in pressure and temperature. The results of these tests were studied in detail in Referenced) and will not be described in this paper.
CNS 9th ANNUAL CONFERENCE. 1988 19
3.
MCWIT FACILITY AND TEST RESULTS
In the CWIT facility (Figure 1 ) , the end fitting simulators are significantly shorter than the end fittings in the reactor and are connected onto the channel side-by-side rather than end-to-end as in the reactor. To study the effect of end fitting geometry v. the standing start results, a new series of standing start tests was recently conducted in the CWIT facility with the channel-end fitting assembly replaced by one closely resembling a standard Pickering NGS channel-end fitting assembly. In this Modified CWIT (MCWIT) facility: i)
the pressure tube contains two-3 m long electrical heater assembly each consisting of six heater bundles and each bundle consisting of 28 heater elements rather than 37 elements as in CWIT,
Figure 2 shows the final fractional channel water level, i.e., the water level just prior to steam arrival in the vertical feeder, versus initial channel fluid subcocling for various values of channel power and header pressure in MCWIT tests. For clarity, uncertainty bars on the measured levels are not shown. To show the trend with subcooling, each pair of water level data points at 130 kW at each pressure and at various subcoolings are connected by a solid line segment. The line segments at each pressure indicate that the water level exhibits no definite trend with subcooling. It may therefore be concluded that the channel water level was nearly independent of subcooling as pointed out previously(1). Figure 2 also shows that the water level is lower at lower pressure for a given power and at higher power for a given pressure as ;nay be expected.
ii) the heater has a cosine axial power shape rather than uniform axial power shape as in CWIT,
SYMBOL HEADER PRESSURE (MPai
NOTE
NUMBERS ON SYMBOL INDICATE CHANNEL
iii)the end fittings are connected onto the channel end-to-end rather than . _ "e-by side as in CWIT,
POWERS IN kW
iv) the feeders are connected to the end fittings at angle rather than horizontally as in CWIT, and vj
the end fitting heat capacity is a factor of two smaller than that in CWIT. (Note that in the definition of the heat capacity the following components are considered: the end fitting and liner tube for the MCWIT facility whereas the horizontal feeder section, end fitting, the p ipe connect ing the end fit ting and channel and the unheated part of the channel for the CWIT facility.)
In the MCWIT facility, thermocouples are also placed on the outside surface of the end fitting body and in the fluid path in the annulus between the liner tube and the end fitting body. These thermocouples give an indication of the temperature distribution in the end fitting body and the progression of steam region along the end fitting during end fitting heatup.
INIT-.U CHANNEL FLUJD SUBCOOLING l ° O
FINAL FRACTIONAL CHANNEL WATER LEVEL VERSUS INITIAL CHANNEL FLUID SUBCOOLING AT VARIOUS CHANNEL POWERS AND HEADER PRESSURES OBSERVED IN STANDING START TESTS IN CWIT FACILITY WITH PICKERING NGS CHANNEL-END FITTING ASSEMBLY.
Figure 3 shows the duration of channel flow stagnation versus initial channel fluid subcooling at various values of channel power and header pressure in the MCWIT tests, To show trends with subcooling, the straight line segments are drawn
Typically, the tests showed that following channel coolant boiling and stratification, hot fluid followed by steam from the channel flowed into the end fitting annuli heating up the end fittings to the saturation temperature. In general, at a given axial location, the top part of the end fittings heated up to the saturation temperature and the bottom part of the end fittings did not heat up until much later in the transient whei, the channel water level was low. Steam condensation on the end fitting surfaces and steam generation in the channel and the resulting channel depressurization and represburization caused fluctuations in the channel flow. These fluctuations apparently resulted in asymmetric thermohydraulic conditions along the channol assembly. In a nuuibor of tests, these conditions resulted in more steam flow towards one of the end fittings causing this end fitting to heat up substantially more than the other.
90 C N S 9lh A N N U A L C O N F E R E N C E . 198B
SVMBOL
P <±
7 / 5 0 ~ A'Qo)
* /. /
§
HEADER PRES>SIJHC
(MP*]
I
MOH NUMBERS ON SYMBOL INDICATE CHANNEL
'
POWERS IN ' W
/
^f 0 1
,^130)
100
x; 100
100
1" INITIAL CHANNEL f LUlD 5U0COOLING I'Ct
FIG. 3
DURATION OF CHANNEL FLOW STAGNATION VERSUS INITIAL CHANNEL FLUID SUBCOOLING AT VARIOUS CHANNEL POWERS AND HEADER PRESSURES OBSERVED IN STANDING START TESTS IN CWIT FACILITY WITH PICKERING NQS CHANNEL-END FITTING ASSEMBLY
through each set of data pcints at a given power and pressure. Fcr a giver: ^ubccoling and power, the duration way 1onger at ::ver pressure in all tests as expected. Exception to this trend existed in asy:::n-.etric tost^, i.e., tests in which the degree cf ho-.itup of the- '..we end f itt ings was subir^ntirt I \y ai f: orer.t. :•'-•: a given sub cool ing and pressure, the- duration .: flow stagnation was 1 o:igsi at significantly 1 cv;Jr pewer (asynuuetrie tests exceptedj . For a gi'.v:. cover and pressure, the duration ci" H o w st-igii.-ji lor. increased with sub c JO ling as expected (asy :::..>-'trie tests
excepz*ad) . Tii« tests shewed that, :'ur similar test conditions, the sheath temperature increased with the duration of channel flew stagnation. In sorr.e high subcool ing/pressure and /or low power tests, the heater sheaths did not heat up above the saturation temperature and a significant channel flow was observed when the hot fluid (but no void) reached the vertical section of the outlet feeder. Fcr similar test conditions, the duration of channel flow stagnation in the MCWIT tests was significantly shorter than that in CWIT tests even for the most symmetric tests. This could be due to a higher degree of mixing between steam and water phases and the water phasa itself in the end fitting annuli in the CWIT facility caused by the ncn-collinear channel-end fitting arrangement. This increased mixing resulted in the heatup of a larger fraction of the end fitting mass in the CWIT facility. This more extensive heatup, the more massive end fittings and the horizontal feeder section connecting the end fitting resulted in longer duration of channel flow stagnation in the CWIT than MCWIT tests.
4.
THERMOSS-II:
MODEL OF STANDING START
To compute the duration of channel flow stagnation and stratification and the channel heatup time for the standing start condition, it is necessary to determine two quantities: Ci) the channel water level transient as it determines the steam production rate and, hence, the end fitting heatup rate, and (ii) the fraction of the end fitting mass that must be heated up to the saturation temperature before steam could reach the vertical feeder section. The two fluid model called THERMOSS_II (for THERmohydraulic Model Of Standing Start) has been developed to predict the duration of channel flow stagnation in the standing start phenomenon. THERMOSS-II generalizes the previous model THERMOSS(l) to account for the effects of cosine channel axial power shape and to quantify in a simple manner the fraction of the end fitting mass that must heat up to the saturation temperature before steam could reach the vertical feeder section. THERMOSS assumes that the entire end fitting mass and water contained within must heat up to the saturation temperature before steam could reach the vertical feeder section. It was pointed out in Reference 1 that this assumption was conservative and resulted in significant overprediction of the duration of flow stagnation for the test conditions in the CWIT facility.
Furthermore, it is shown in .v^cticn 3 that the
recent series: „:' ii?szz in •_:.•„- Y/.'.'/iJT tazi 1 ity exhibited d^rat: cr.z cf" flew ^tarnation significant.!/ s'r.crzer than those in the CWIT tests for similar te;;: L^ndit ion:;. The shorter duration is at t ri Luted i.\ Sec Lien 3, ro the heatup of a smaller fraction ~f the end fitting i:iass iri the MCWIT than CXI"' tests. THERJK32S-:: Seviai.icr: The f ol low; r.g 3£su.i.ripti&:.j and approximations are used in TilERKOSS-11. ij
Following channe: cooiar.'. boiling and stratification, THERMOSS- 11 assujT.es , as dees THERMOSS! I,1 , that the siea-i: flow in the channel and end fittings requires a pressure gradient zo overcome friction. Tr.is pressure gradient causes the water level to rise gradually along the channel until i: re-v-hc-s the top of the pipe in the end fittings. To compute the water level, TKERMCSS-I1 ss$-::;.e= that, at each instant in ti.r.e, the level is in quasi-steady state. In this state, tr.e level is computed by equating the frictional pressure drop in the steam phase to the hydrostatic head change in the water phase. In this computation THERMOSS-II models both axially uniform and cosine channel power shape. For simplicity, THERMOSS-II assumes a horizontal water level in the end fittings and channel. To capture the effect of the level rise in the end fittings, i.e., the lower level in the channel, the friction factor in the end fittings is adjusted to predict the water level in one of the MCWIT tests at the tiir.e just prior to steam arrival ir. the vertical feeder section. (Note that THERKOSS computed the actual water level profile(i). This level was found to be horizontal along most ci the channel length.)
ii) For simplicity, THERMOSS-II ignores the end fitting hear.up resulting from the initial flow of hot water from the cr.^r.r.e 1 (as the channel water toils and the channel water level falls). THERMCSS-II assumes that, at a given axial location, the part cf the end fitting that is exposed to steam is at the saturation temperature and the part of the end fitting that is submerged in watt?r is at the initial subcooled water temperature. This assumption would result in an overprediction of the duration of channel flow stagnation. iii)Figure 4 shows the vater level at two neighbouring instants of time. In a small time interval, the level slightly falls and extends further into the end fittings as the steam from the channel condenses on the end fitting surfaces and heats up the shaded parts of the end fittings to the saturation temperature. This mode of end fitting heatup assumes an infinite heat transfer coefficient from steam to the end fitting body and liner tube. This approximation is reasonable in view of relatively small thickness of the body and tube. Heat transfer frcn steam to the water in the end fitting annulus is ignored because the usual conduction heat transfer is relatively small and the vater is relatively hot (compare item {iij above).
CNS 9th ANNUAL CONFERENCE. 1988 91
Figure u z:.c;:i tr.e pu: _•. :. *-a J; c- c;ffere:.ccbetween, the durations of c;.-:.:;fcl t'.QV stag;:.= : lor: predicted by THERMOSS-II fc: the MJWI? test conditions and that observed ir. the MCWIT tests whicl: exhibited ;:.ost synsuci:-ic conditions, i.e., nearly equal degree of heaii:p ot the two end fittings. For the most sy::::.e:r:c ot the syr;_T.etric tests in Figure C, "HERliCSS-'. I cverpredicts the duration by at ::.ost 10%. This agreement between the predicted ar.d experiment] results is reasonably gecd.
"• \
S T E A M - * ^ 1 -*• STE.
STAGNANT WATER
SVMBOL MFADFF) «nESSu«C WPd.
"I Mass balance on the steam region and the above energy transfer between stea::. and the end fittings are used to obtain an equation for the rate of change of the steam volume assuming bulk boiling in the channel. This equation coupled vith the calculation cf the water level in item (i) above, is integrated to obtain the volume of the steam region as a function of time, channel power and pressure, initial fluid subcsoling and channel-end fitting geometrical dimensions. The duration cf flow stagnation is then equaled to the time at which the steam region extends to the end fitting feeder connection port. This time also determines the final char.r.el water level.
5.
COMPARISON CF THERMOSS-II PREDICTIONS AND MCWIT TEST RESULTS
Figure 5 compares the final channel water level predicted by THERMOSS-II for the conditions in the MCWIT tests with that observed ir. the MCWIT tests. Each bar indicates the up.ce: tair.ty in the location of the level due to the limited number of ther::,ocouple:j or. the heater element sheaths. Figure 5 shews that, within the experimental uncertainty, THERMOSS-II predicts the level reasonably well for some and underpreciicts it for other tests.
^ INDICATES UNCERTAIN^ IN MEASURED WATER LEVEL
MEASURED FINAL FRACTIONAL CHANNEL WATER LEVEL
FIG. 5
COMPARISON OF FINAL FRACTIONAL CHANNEL WATER LEVEL PREDICTED BY THERMOSS-II AND OBSERVED IN STANDING START TESTS IN CWIT FACILITY WITH PICKERING NGS CHANNEL-END FITTING ASSEMBLY
92 CNS 9th ANNUAL CONFERENCE, 1988
L
* C PSVMBOL NOTE i \ . aBEFIS ON INDICATE C H A N N E ; PD'.VETRS i', «k\
FIG. 4 WATER LEVEL MOTION FOR THERMOSS-lf
•3CO
60
#
80
T3C
IOC
CT31
TO
iNFTdU CM4NNE. ^;UiD SLJBCOOLINC • C
FIG. 6
6.
PERCENTAGE DIFFERENCE BETWEEN DURATION OF CHANNEL FLOW STAGNATION PREDICTED BY THERMOSS-II AND OBSERVED IN THERMALLV SYMMETRIC STANDING START TESTS IN CWIT FACILITY WITH PICKERING CHANNEL-END FITTING ASSEMBLY
SUMMARY AND CONCLUSIONS
The recent series of standing start tests conducted in MCWIT facility vith the reactor-like channel-end fitting geometry seems to show a smaller degree cf mixing ir. the water phase and between the water ar.d stea;:: phases in the end fitting annul us. This smaller degree of mixing rr.ay explain the heatu? of a srr.al~.er fraction cf the end fitting mass in the MCWIT tests than in the previous test conducted ir. the CWIT facility with the reactor unrepresentative end fitting geometry. A new two-fluid model c: the standing stare phenomenon, called THERMOSS-II has been developed. THERMCSS-II computes the (horizontal) channel water level following channel coolant boiling and stratification from a momentum balance on the steam and water phases similarly ^c that in the previous THERMOSS model C D . In the computation of the end fitting heatup, THERMOSS-II assumes that the part of the end fitting which is exposed to steam is at the saturation temperature and the part that is submerged in water is at the initial subcooled water temperature. This generalizes the approach used in THERMOSS which requires that the entire end fitting mass must heat up to the saturation temperature before steam cculd reach the vertical feeder section. Within the experimental uncertainty, THERMOSS-II predicts reasonably well the final channel water level observed in the recent MCWIT standing start tests series. For the most symmetric tests in this series, THERMOSS-II overpredicts the observed duration of channel flow stagnation by at most 10%.
REFERENCES
AKNOWLEDGEMENTS
(1) P. Gulshani, M.Z. Caplan and N.J. Spinks, THERMOSS: "A Thermohydraulic Model of Flow Stagnation in a Horizontal Fuel Channel", CNS 10th Annual Symposium on Simulation of Reactor Dynamics and Plant Control, St. Johns, New Brunswick, 1984 April 9-10; see also 5th European Nuclear Society, International Meeting on Thermal Nuclear Reactor Safety, Karlsruhe, West Germany, 1984 September 9-14.
The standing start tests reported in this paper were funded by COG-AECL. Analysis of the test data and THERMOSS-II development and verification reported in this paper were funded partly by the New Brunswick Electric Power Commission and Hydro Quebec,
CNS 9th ANNUAL CONFERENCE. 1988 93
"UENS1TIV1TY STUD11& 01' CALANDK1A TUBE INTEGRITY IN THE EVENT OF PRESSURE TUBK KA1LUKE" P.S. KUNDURPI AND A.P. HUZUHDAR Nuclear Studies and Safety Department Ontario Hydro, 700 University Avenue Toronto, Ontario ABSTRACT
transient to determine strain rate dependent material properties. In addition, the method did not account for plastic anisotropy effects.
The Issue of calandria tube integrity in the event of a sudden pressure tube failure is examined by considering the response of the calandria tube to the transient over-pressure in the annulus. A detailed sensitivity study of the main parameters Influencing the calandria tube response is presented based on the results obtained using a small computer code (FSTI). The details of the code validation are also included. The application of the code to Pickering NGS A and Bruce NGS A reaclors case indicates Lhal the calandria tube can withstand the transients with a large strength margin.
A new computer code for Fluid Structure Transient Interaction (FSTI) was developed to overcome these deficiencies and to account for the anisotropy effects. The latter effects are discussed in Reference 1 and is not covered here. This paper focusses on the main features of the new program and the sensitivity of the predicted results to the input parameters. The calandria tube strength margin tc failure in the event of a pressure tube failure in Bruce and Pickering reactors is also discussed.
INTRODUCTION
METHODOLOGY
In the licensing analysis of CANDU reactors a broad range of accidents are postulated and analyzed to evaluate their consequences. During certain postulated accident scenarios in which only the pressure tube (PT) ruptures, the calandria tube (CT) is subjected to severe transient loading conditions. Assurance of the calandria tube integrity reduces the economic consequences of these accidents. During such accidents the transient loading or. the calandria tube is separable into two distinct stages, i.e. the initial asymmetric loading stage and the overpressurization stage. The initial asymmetric loading causes bending of the fuel channel and ovalling of calandria tube (Reference 1 ) . In the later stage the calandria tube may experience a brief pressure transient exceeding the system pressure. The overpressure in the annulus Is caused by the surge of coolant into the annulus following the pressure tube rupture. The magnitude of the pressure rise is effectively dependent on the fluid discharge rate into the annulus which in turn is dictated by the hydraulic resistance characteristics of the primary heat transport system. The MINI-SOPHT computer code (Reference 2) which models the individual channel behaviour is used to estimate the coolant discharge rate and the associated pressure rise. The predicted pressure increase is an overestimate since volume changes in the PT/CT annulus and the feeder piping tend to reduce the pressure. The volume change becomes particularly significant if the pressure during the transient causes stresses to exceed the yield stress for the calandria tube material. A simplified method of correcting ihe pressure transient was developed earlier in Reference (2). This method uses an artificial instantaneous strain rate during each calculational time step instead of more appropriately using a mean strain rate during the
84 CNS 9th ANNUAL CONFERENCE, 1988
The method of correcting the pressure transients to account for volume changes, is as follows. Consider any particular node with a defined value of pressure (P), volume (V), fluid density lp) and temperature (T). Let the hoop stresses in the pipe due to the pressure (P) be
or 4p
4ra _ (*V) V V
P
(1)
The corresponding pressure rise can be obtained
*P
• "> " TP
(2)
The first term in the above equation is the change in pressure predicted by the MINI-SOPHT code (neglecting the volume change) and It is denoted by SPSOPHT- Tne volumetric strain (iV) is related to the hoop strain in the CT by V
V
(3)
where 4CQ = change in hoop strain when the stresses in the CT are within the elastic
limit the change jn hoop strain can be related to the incremental pressure by considering the biaxial state of loading as (4)
iP i\)
&ce
If the stresses in the CT are in the elastic-plastic regime, it is realistic to assume that the incremental hoop strain will still be given by equation (L>), provided that the elastic modulus E is replaced by the tangent modulus E t corresponding to the initial stress state. When the stresses in the CT are in the fully plastic regime, the material anisotropic effects become significant. However, the volumetric strain can still be related to the incremental pressure by relations similar to equations 3 and 4. The details of these expressions valid for plastic straining are given in Reference 1. For the rest of the formulation it will be assumed that the calandria tube is in the elastic-plastic regime.
where p o - 962.83 kg/m3 is the reference value used at 10 HPa and 100°C || = 0.794 - 0.0116T + 1.32 x 10" 4 T 2 - 5.978 x 10" 7 T 3 + 1.104 x 10~ 7 T 4
(7b)
nj = 1.26 For heavy water the corresponding properties are represented as
P = 9(T> + ff P
(8a)
where g(T) - 1122.18 - 0.449T - 0.98 x 10"3 T2 - 4.46 x 10"6 T3 | £ = 0.5291 - 1.119 Tj + 6.356 Tj 2
(8b)
- 9.956 Tj 3 + 6.596 Tj 4 where T = fluid temperature in °C
Hence in general the volumetric strain can be written as
P = pressure in MPa
(5) 280 From equations (2) and (5), the actual pressure rise is obtained as 4P
4P =
SOPHT
1 • 2p <£> (i
(6) -) (r)
From equation (6), it is seen that the MINI-SOPHT predicted pressure rise is reduced by a correction factor given by the denominator. The main parameters contributing to this correction factor are the instantaneous elastic modulus (Et = cto), the poisson's ratio \i and the bulk dc modulus of the fluid p(4P/4p). The bulk modulus of the fluid is proportional to the sonic velocity squared. The Instantaneous material properties such as elastic modulus E t can be obtained from the stress-strain relations for the Zr-2 calandria tube. These quantities can be either implicit in the code or they can be evaluated separately. In the present formulation, these quantities are evaluated implicitly in the code as follows. Representation of Fluid Properties The density data for light water (Reference 3) and heavy water (Reference 4) indicate that the variation of density with pressure is linear for a given temperature. These density data are represented in a polynomial form as follows for input into the code. For light water the density variation is represented as
Po
< p - 1 0 ) - °- 2 i (T-100)"1
(7a)
The above relations are valid for 4 < P < 40 HPa and 50 < T < 300°C. Representation of Material Properties The material properties of Zircaloy, such as yield stress, uniform elongation and ultimate tensile strength are temperature and strain rate dependent. The stress-strain relation for Zircaloy can be represented in Ramberg-Osgood form as, (Reference 5) e = + A <2-)» fly
«>
Oy
where e = the non-dimensional strain _ E_£ " "Y o ffy e A.n
= stress = yield stress = strain = empirical constants
It can be noted that the stress-strain curve in equation (9) has two numerical constants, i.e., A and n. To evaluate these two constants one needs to choose only two significant reference points through which the stress-strain curve passes. The yield stress is taken as the first reference point. The yield stress is generally obtained from the stress-strain curve by drawing a line with a 0.2 percent strain offset and with a slope equal to the elastic Young's modulus. Hence the strain corresponding to the yield stress is obtained as 0.002
(10)
CNS 9th ANNUAL CONFERENCE, 1988 95
Using the yield point as one reference point in the stress-strain curve, the value of constant A is obtained from Equations (9) and (10) as A =
2 x 10"3E
(11)
The value of index n is obtained by choosing a convenient second reference point on the stress-strain curve with known value of stress 0] and strain c\. The ultimate tensile strength is chosen as the second reference point in the present formulation. Substituting the known values of t\, a\ and A, the value of
VALIDATION OF THE CODE
i - cri
log [
(12)
log «rj EEI
where ei1 =
°Y
the actual mean strain rate is evaluated based on tie strain accumulated and the pulse width. If the actual strain rale does not agree with the assumed rate, the code iterates for the strain rate until the desired accuracy on the strain rate is obtained. At the end of the calculations, the actual strain accumulated in the transient, the strain rate and the calandria tube material properties are printed out.
To assess the accuracy of the FSTI code, the experimental results from full-scale rupture tests (Reference 7) are considered. In these experiments prototypical unirradiated calandria tubes were subjected to pressure transients by actual pressure tube rupture. The code is also used to simulate the conditions to which the Bruce N-6 calandria tube was subjected (Reference 8 ) .
°y Comparison with the Test Results
Using the above methodology, the complete stress-strain relation can be described by knowing the yield stress, the ultimate tensile stress and the failure strain (termed reference strain). The material data for Zircaloy calandria tubes, given in Reference 6, are again represented by a set of empirical relations for convenience of input into the code as shown in Table 1. These empirical equations implicitly account for the variation of material properties with strain rate and temperature.
FSTI CODE DESCRIPTION The main input variables tor the FSTI code are: (1)
Reference pressure (i.e. coolant supply pressure or header pressure).
(2)
MINI-SOPHT predicted peak pressure and pulse width.
(3)
Fluid temperature.
(4)
Calandria tube temperature.
Based on the fluid temperature, the density (p) and the sonic speed parameter (4P/Sp)are calculated using the expression given in equations (7) and (8). Based on the pulse width (which affects the strain rate) and calandria tube temperature, the appropriate values of yield stress and the elastic modulus are evaluated using the equations given in Table 1. At the start of the calculation the code assumes a mean strain rate to calculate the CT material properties. The corrected pressure transient is obtained by following the predicted pressure transient in small incremental steps. For each pressure increment, using equation (6). the corrected pressure increment and the corresponding stress and strain in the calandria tube are evaluated. At the end of each pressure increment the values of stress and the instantaneous elastic modulus E[ are updated. At the end of the transient,
96 C N S 9th ANNUAL CONFERENCE. 1988
The full-scale PT rupture test conditions and results are given in detail in Reference 7. The measured mechanical properties of the calandria tube at room temperature were 0 V £ T = 433 MPa and ouTSRT = 522 MPa. The simulation of the test loop and conditions by HINI-SOPHT predicts a peak annulus pressure of 20 MPa and a pressure pulse width of 70 ms as shown in Figure I for Test 6 in the series (Reference 2 ) . The calandria tube temperature for the test conditions is estimated to be 155°C. With these input conditions using the present code the corrected peak annulus pressure is estimated to be 9.07 MPa and the expected permanent hoop strain in the CT is around 1.08 percent. The measured values of peak pressure in Test 6 was around 8.7 MPa (Reference 7 ) . The measured strain was around 0.75 percent. The measured strain and the corrected annulus pressure agree very closely with the predicted results. Similar results are obtained for Tests 7 and 8 in the series.
Application to Bruce N-6 Case The ability of the code to predict the response of the Bruce N06 calandria tube is examined next by using the following input conditions (Reference 8 ) : (a)
PReference = 8.4 MPa.
(b)
Fluid and Calandria tube temperatures 50°C.
(c)
The predicted pulse width = 30 msec.
The material properties corresponding to the irradiated CT obtained from tests on calandria tubes (from Reference 6) are employed. With these known parameters It is required to find the maximum waterhammer pressure at which the
(.Twill fail, i.e., the MJNJ-SOPHT predicted annulus pressure is considered to be the variable for analysis. Using the methodology dJscussed above, the corrected peak annulus pressure, the burst pressure (dependent on the strain rate during transient) and the permanent strain during the transient are obtained for various maximum uncorrected waterhammer pressures. These results shown in Figure 2 indicate that the corrected peak pressure exceeds the burst pressure when the predicted Wdterhammer pressure is around 52 HPa. The MINI-SOPHT predicted annulus uncorrected waterhammer pressure is 54 HPa. This clearly shows thai the calandria tube in N06 ruptured due to the corrected annulus pressure exceeding the burst pressure. The predicted plastic hoop strain in the N06 calandria tube for these conditions is around 0.6 percent. This compares well with the measured hoop strain in N06 calandria tube on the order of 0.4 percent.
SENSITIVITY STUDS' From the list of input parameters to the code it is evident that calandria tube strain response will be sensitive to many parameters. The fluid temperature affects density (p) and sonic speed parameter (ip/ip) significantly, whereas the pressure (P) has very little effect on these quantities. The pulse width (through strain rate) and calandria tube temperature strongly influence the calandria tube material properties. A detailed sensitivity analysis has been carried out to assess the significance of the various parameters on the expected results. The main parameters chosen to assess the sensitivity of the results are the reference strain, the calandria tube temperature and the pressure pulse width. The effect of varying these parameters on the corrected peak annulus pressure and the resulting permanent strain is reported here. The effect of varying the calandria tube thickness and the material properties on the results is briefly discussed. The pressure transient in Test 6 (Reference 7) is taken as a typical transient for the sensitivity study. The following is the range of main parameters selected to study the sensitivity of the results: 2 - 12 percent Reference strain 115 - 195°C CT temperature - 30 - 110 ms Pulse width
Sensitivity to Reference Strain and Pulse Width It is to be noted that the reference strain is strictly a material property like yield stress and ultimate tensile stress which is dependent on strain rate (Table 1 ) . As the reference strain is a measure of the ductility of the CT material, it is considered as an input parameter in the present sensitivity analysis in order to assess the effect of tube ductility on the predicted results. The variation of the peak corrected pressure and the permanent strain for various pulse widths and reference strains are shown in Table 2 for a nominal calandria tube temperature of 155°C. From
these results it is seen that varying the reference strain Horn 2 to 12 percent causes a maximum of 6 percent reduction in peak corrected pressure and about 10 percent increase in predicted plastic strains. Similar results were obtained tor other calandria tube temperatures. Hence it is concluded thai the reference strain does not influence results significantly and the method of calculating the reference strain based on total elongation, given in Table 1 can be used with confidence for CT integrity assessment.
Effect of Pulse Width and Temperature To assess the sensitivity of the results to pulse width and calandria tube temperature, the reference strain has been assumed to be given by the relations given in Table 1. The predicted results for the range of calandria tube temperatures and pulse widths are shown in Table 3. These results indicate that with increasing CT temperature and pulse width, the corrected annulus pressure decreases while the permanent strain in the CT increases. Increasing the pulse width from 30 to 110 ms increases the permanent strain ty a factor of about 2.95 at high temperature and by 1.14 at low temperatures. The decrease in peak pressure for the same range of pulse widths is only about 6 percent for a given temperature. For a given pulse width, it is seen that both the corrected peak pressure and the plastic strain are very sensitive to CT temperature. Thus it is concluded that the water hammer pulse width and the CT temperature are critical parameters in assessing calandria tube integrity.
Effect of Calandria Tube Thickness and Material Properties The effect of varying the calandria tube thickness on the predicted result is examined next by using the conditions predicted in Test 6 as reference (Figure 1 ) . The results obtained by varying the thickness between 0.8 to 1.1 of nominal thickness are given in Table 4. These results demonstrate, as expected, that the thinner the tube the greater the reduction of pressure at the expense of increasing plastic strain, with the latter exhibiting a greater dependence on tube thickness. The effect of varying the material properties (i.e., yield stress and UTS) are studied next. The material properties of Zr-2 can be affected either by the amount of prior cold work or by the amount of irradiation. Any increase in the material strength is generally accompanied by a corresponding reduction in the ductility. In general, the yield stress increase is higher than the increase In ultimate tensile stress due to irradiation. To account for these effects in an analytical form, the material properties are assumed to vary as o-yRT "UTSRPT ci
= =
=
433 (1 + 0.08 y) MPa 522 (1 + 0.05 y> MPa 0.20 - 0.01 y
(13)
CNS 9th ANNUAL CONFERENCE. 1988 97
where y is the nii.nber of years in reactor. The variation of failure strain, yield stress and ultimate tensile stress wjth temperature and strain rate are still assumed to be similar to the unirradiated material data in Table 1 (as observed experimentally). With these properties, the predictec results, using the Test 6 pressure transient are given in Table 5. These results indicate that the plastic strain drops sharply with increasing irradiation with a corresponding moderate Increase in the peak pressure. Thus irradiated calandria tubes subjected to the type of water hammer transient shown in Figure 1 would be expected to survive with less than 0.5 percent plastic strain. MARGIN TO CALANDRIA TUBE FAILURE IN REACTOR The present methodology is used to evaluate the calandria tube strength margin in the event of a sudden pressure tube failure for both Pickering NGS A and Bruce NGS A reactors. The pressure transient predicted for the event In channel G-16 in Pickering Unit 2 is selected as a typical transient for the evaluation. This transient is similar to that shown in Figure 1 except that the peak pressure is lower (15 HPa) due to the hotter coolant entering the annulus. The strength margin during both the transient overpressure phase and the final steady-state loading is evaluated assuming a minimum biaxial strengthening' of 20 percent for irradiated tubes as observed in experiments (Ref 6 ) . Three different failure criteria are considered for the transient loading to cover a wide range of calandria tube ductility (i.e., 0.1% or 1% plastic strain limit during the transient or the hoop stress exceeding the ultimate tensile strength). Similarly to illustrate the tube lo lube strength variation the nominal values and ihe lower bound values of material properties are considered. The lower bound values of material properties are obtained by considering the mean measured values and by assuming that the spread of material properties In the irradiated condition is identical to that of unirradiated material. This Is very conservative as it ignores the observed trend in narrowing of the material properties towards the mean in the irradiated condition. The strength margins are defined as: Peak Waterhammer pressure needed to reach the selected Margin to failure = failure criteria -1 (transient phase) Actual waterhammer pressure during [he transient Margin to failure (steady state phase)
Predicted Calandria tube burst pressure Steady state annulus pressure
The results of these calculations, given In Table 6, show that large margin to failure exist for nominal irradiated calandria tube, with failure margins being larger in Pickering than in Bruce leaclors due to the increased tube thickness in the former. Note that for nominal tubes, the limiting failure margin is obtained for the steady-state rather than the transient phase.
98 CNS 9th ANNUAL CONFERENCE. 1988
Failure margins are smaller, but still positive, for the conservative lower bound tube strength assumed. However, for the low ductility failure criterion (0.1 percent plastic strain) the limiting failure margin is obtained for the transient rather than the steady-state loading phase. It Is reiterated that the lower bound strength is unrealistic since tube strengths approach approximately the same (nominal) strength after several years irradiation. CONCLUSrONS The method of accounting for the strain feed-back effects on the calandria tube response due to pressure tube failures has been improved to account for the strain rate effect and material anlsotropy. The results of the computer code FSTI compare well with the available experimental data and in-reactor experience. A detailed sensitivity analysis indicates that the waterhammer pulse-width, calandria tube temperature, thickness and strength (irradiation) are important variables in determining the resulting plastic hoop strain developed in the calandria tube during the initial pressure transient. This plastic strain is a crucial parameter is the assessment of calandria tube integrity since it can be compared with the measured tube ductility to assess the margin to failure. Application of the methodology to irradiated tubes in Pickering NGS A and Bruce NGS A reactors show considerable margins to calandria tube failure based on annulus pressurization following a sudden pressure tube burst. REFERENCES (1) P.S. KUNDURPI AND A.P. MUZUMDAR, 'Structural Response of Calandri'i Tube to a Spontaneous Rupture of Pressure Tube' paper presented at the 9th International Conference on 'Structural Mechanic;; in Reactor Technology', Lausanne, Switzerland, August 1987. (2) A.P. HUZUMDAR, J.K. PRESLEY, AND M. KWEE, 'Simulation of CANDU Pressure Tube Rupture' paper presented at the ANS/CNS Topical Meeting on 'Thermal Reactor Safety' San Diego, California, February 1986. (3) ERNST SCHMIDT. "Properties of Water and Steam in SI - Units" Springer-Verlag, Berlin, Germany 1969. (4) P.G. HILL, R.D. HACMILLAN AND V. LEE, "Table of Thermodynamic Properties of Heavy Water in SI Units", AECL-7531. December 1981. (5) tf. RAMBERT AND W.R. OSGOOD, 'Description of Stress-Strain Curve by the Three Parameters' National Advisory Committee for Aeronautics, NACA-TN902, 1943. (6) C.E. ELLS et al, 'The Behaviour of CANDU Calandria Tubes' paper presented at the CNS 8th Annual Meeting, Saint John, New Brunswick, June 1987.
(7) G.I. HADAU.ER AMD A.P. MUZUMDAR, 'Progress on an Experimental Program to Determine the Consequences of Pressure Tube Rupture jn CANDU Reactors', paper presented at the 8rh Annual CNS Meeting, Saint John, New Brunswick, June 1987.
(8) G.J. FIELD AND H.W. SHANAHAN. "The Failure of the Pressure Tube in Fuel Channel No. 6 of Bruce NGS A Unit 2 in March 1986", paper presented at 8th Annual CNS Heoting, Saint John. New Brunswick, June 1987.
TABLE 1: VALUES OF YIELD STRESS (GyR?) A N D ULTIMATE TENSILE STRESS (ffuTRT> F 0 R
Unirradiated Tube Ultimate Tensile St r e s s MPa
" U T S = ("UTSRT
+
Zr
~2
Irradiated Tube
•nl 2 0 . 1 6 _ j _ Q g T) ( - — ,
°UTS = ("UTSRT
+
22.18 _ l 14 f
10" 3
nj
= 0.0216 + 9.83 x 10' T
Yield Stress (oJ MPa
oy
= (^yRT) Exp 0.94
Failure Strain or Reference Strain CJ
z\
Typical values at room temperature and low strain rate
= 433 MPa <>yRT °UTSRT =
n2 0.06
°UTS
(-0.003 T)
0.2 - 0.04 logm (
10"
^
10' 3
= 0.01124 t 1.22 x 10~ 4 T
- 0.016 logjo e
cT
0.07 - 0.02 log {
^UTSRT = "yRT "
10-3'
MPa MPa
E = young modulus - (96 - 0.06 T) x 10 3 HPa "UTSRT " ultimate tensile stress at Room Temperature "yRT " yield stress at Room Temperature
CNS 9th ANNUAL CONFERENCE, 1988 99
TABLE 2: RESULT:; OF THE SENSITIVITY STUDY: VARIATION OF PEAK PRESSURE AND PERMANENT STRAIN WITH PULSE WIDTH AND REFERENCE STRAIN CALANDRIA TUBE TEMPERATURE - 155.0 P R e E = 6.5 MPa. P m a x - 20 MPa 2
Reference Strain Percent 6 8
4
Pulse Width ms 30 50 70 90 110
12
VARIATION OF' PEAK PRESSURE 9.704 9.578 9.501 9.400 9.399
9.479 9.289 9.174 9.107 9.068
2
4
9.402 9.194 9.092 9.017 8.971
9.356 9.156 9.048 8.969 8.895
Reference Strain Percent 6 8
Pulse Width ms 30 50 70 90 110
10
9.331 9.124 9.009 8.298 8.875
9.309 9.095 8.976 8.910 8.838
10
12
VARIATION OF PERMANENT PLASTIC STRAIN .869 .952 1.017 1.081
.874 .972 1.063 1.145 1.218
1.119
.873 .974 1.073 1.169 1.259
.873 .974 1.079 1.185 1.303
.872 .974 1.084 1.200 1.316
.872 .973 1.088 1.206 1.344
TABLE 3: RESULTS OF THE SENSITIVITY ANALYSIS - VARIATION OF PEAK PRESSURE AND PERMANENT STRAIN WITH PULSE WIDTH AND CT TEMPERATURE P R e f - 8.5 MPa, P m a x = 20 MPa VARIATION OF PEAK CORRECTED PRESSURE
Pulse Width ms 30 50 70 90 110
Calandria Tube Temperature °C
115 10.346 10.088 9.920 9.794 9.698
135
155
175
195
9.871 9.622 9.462 9.354 9.277
9.440 9.216 9.068 8.991 8.917
9.040 8.879 8.779 8.740 8.696
8.768 8.660 8.628 8.597 8.577
175
195
1.122 1.399 1.715 1.973 2.303
1.821 2.721 3.416 4.351 5.340
VARIATION OF PERMANENT PLASTIC STRAIN PERCENT
Pulse Width 115 30 50 70 90 110
100
.700 .736 .761 .782 .800
CNS 9th ANNUAL CONFERENCE. 1988
Calandria Tube Temperature °C 135 155
769 816 857 896 934
.873 .973 1.077 1.178 1.290
TABLE 4:
EFFECT OF CALANDRIA TUBE THICKNESS ON PREDICTED RESUI.iS P R e f = 8.5 HPa. P m a x - 20 MPa
C'f Nominal Thickness Assumed Thickness
0.8
0.9
1.0
1.1
Ppeak corrected
10.966
9.921
9.068
8.620
plastic strain
0.635
0.778
1.076
3.101
TABLE 5:
EFFECT OF CALANDRIA TUBE PROPERTIES ON THE PREDICTED RESULTS
P R e f = 8.5 MPa, P raax -- 20 MPa
^ in equn (13) ->
p
peak corrected
plastic strain
1
0
3
4
5
9 .068
9.656
10 .271
10.892
11 .506
12 .187
1 .076
U.821
0 .71
0.62
0.534
0 .44
TABLE 6:
Failure Criteria
2
CALANDRIA TUBE MARGIN TO FAILURE FOR VARIOUS FAILURE CRITERIA FOLLOWING G-16 TYPE EVENT IN PICKERING NGS A AND BRUCE NGS A REACTORS
Pickering NGS A CT Lower Bound Nominal Values of Strength Strength
Bruce NGS A CT Lower Bound Nominal Strength Strength
20
53
7
33
1% plastic strain during transient
100
127
80
106
hoop stress exceeds ffUTS during transient
267
273
233
247
hoop stress exceeds flyjg during steady state
25
42
0.1% plastic strain during transient
23
CNS 9th ANNUAL CONFERENCE, 1988
101
TEST 6
nn
D 07
0.,14
0.21
0.2B
9 . 5 MPA.
0.33
25RC
0. *2
0.49
0.5(5
FIGURE 1 The SOPHT Code Predicted Annulus Pressure Transient
Bruce N-6 Simulation P H e l -B.4MP2 T
Fl u ,a " T CT • 50°C
Pulse Width • 30 'n s«cs.
—r~ 40
60
—r~ so
,3k Maximum Prajst
FIGURE 2 ~
102
CNS 9th ANNUAL CONFERENCE, 19fc8
Variation of Corrected Pressurc, Rastic Strain and Burst Pressure With Peak Pressure
0. M
PULL SCALE CALANDRIA TUBE BURST TESTS AT FAST PRESSURIZATION RATES G.I. HADALLER Stern Laboratories Inc. 1590 Burlington St. East Hamilton, Ontario L8H 3L3
A.P. MUZUMDAR Ontario Hydro 700 University Avenue Toronto, Ontario M5G 1X6
ABSTRACT Four full scale Zr-2 CANDU calandria tubes with prototypic rolled joints and axial preload were burst while at 170xC. Calandria tubes which had been subjected to loading and strain during a pressure tube burst had essentially the same burst pressure as new tubes and no new failure mechanisms were identified. The burst pressures were above the primary system operating pressure of 10 MPa and thus offer some margin to failure in the case of pressurization due to pressure tube failure. 1.0
INTRODUCTION
Accident analysts of CANDU reactors includes the evaluation of the consequences of a postulated fuel channel failure leading to the discharge of steam and water into the calandria vessel containing the moderator. For this case it is conservatively assumed that failure of the pressure tube containing the high pressure coolant, results in simultaneous failure of the calandria tube which separates it from the low pressure water in the calandria vessel. The failure of a pressure tube at the Pickering Unit 2 reactor in August 1983 showed that the calandria tube could sustain the loading generated in such an event Uius iinuLiny the coolant discharqe to that which could pass through the small flow areas at the end-fitting bearings at the ends of the fuel channel. The present experimental program was begun with the objective of determining the margin to calandria tube failure under various postulated pressure tube failure conditions. In phase I and II reported previously [1,2] the objective was to determine tho pressure transient in the calandria tube annulus following pressure tube failure using progressively thinner stainless steel and then prototypical Zr-2 calandria tubes. This paper describes the first reported results of relatively fast pressurization burst tests of full scale calandria tubes. The target was to reach a pressure of 10 MPa in approximately 100 ms to simulate reactor conditions following a burst. The tests were carried out at Stern Laboratories Inc. as part o£ an on-going program with COG-CANDEV funding. Four test have been carried out with two new
calandria tubes and two calandria tubes subjected to the loading and strains in a pressure tube burst test from Phase II. 2.0
TEST OBJECTIVES
The objectives of these tests were to determine the burst pressures and failure mechanisms of new and previcasly strained calandria tubes while at temperatures representative of those following a pressure tube burst. These results will be used in assessing the margin to calandria tube failure that exists following various postulated pressure tube failures. 3.0
EXPERIMENTAL APPARATUS
3.1
Test Loop
The test loop consisted of high pressure (11 MPa) carbon steel piping complete with a circulating pump, flow and pressure control systems,an electrically powered boiler, a full scale calandria tube complete with prototypic rolled joints, a reentrant simulated pressure tube for heating, garter springs, and a system of axial supports to restrain the calandria tube and provide a mechanism for applying an axial preload and measuring the load during the test. The calandria cube annulus was connected to a supply tank charged with nitrogen through a line equipped with a double rupture disc to initiate the pressurization. The calandria tubes were rolled into 304 stainless steel flanges which were fastened to the end supports to rastrain the caxandria tube and provide a pressure boundary. Except in test 1, cooling coils were attached to the outside of the simulated calandria side tube sheets (hubs) to obtain a representative temperature distribution of the calandria tube rolled joint, the calandria tube end supports were fastened to a 50 cm I-beam which served as a support and a means of measuring the axial load. A tank with a c—vcd bottom and an open top was provided around the calandria tube to deflect coolant discharge upward. A schematic of the test apparatus is shown in Figure 1. CNS 9th ANNUAL CONFERENCE, 1988 103
The cnlandria tubes used in these test were made from Zr-2 with a nominal wall thickness of 1.37 mm. The i:,ilandri.i tubes were made to the Darlington reactor design and the tubes used in Tests 1 and 2 were new. The tubes used in tests 3 and 4 had been used previously in pressure tube burst tests 8 and 9 respectively [21 and had permanent strains of about 1% which occur in the initial pressurization and filling of the channel.
3.2
Instrumentation and Data Acquisition
The instrumentation used in each test consisted of thermocouples to monitor loop and calandrLa tube temperature, pressure transducers to monitor calandria tube pressure, strain gauges and linear variable displacement transducers (LVDT) to monitor calandria tube strain and load cells to measure the axial preload carried by the I-beam support. The signals were recorded on a high speed (1 kHz) computer data acquisition, converted to engineering units and plotted. A typical instrumentation scheme is shown in Figure 1.
4.0
TEST PROCEDURE
A typical test procedure involved instrumenting and assembling the calandria tube and installing it in the test apparatus. The piping was checked for leaks and the required axial preload was applied to the calandria tube by adjusting the screws on the tensioning mechanism and monitoring the load on the load cells. The loop, the calandria tube and the bump tank were filled with water pressurized to 4 MPa, the circulating pump was started, the electric boiler was energized and the system was heated to slightly above the required test temperature (170xC). The cooling on the calandria tube hubs was adjusted to provide a representative temperature distribution and the calandria tube annulus and the bump tank were isolated from the warm-up loop. The computer data acquisition system was started the bump tank was pressurized up to 18 MPa, the video and high speed movie cameras were started and the rupture disc assembly was burst rapidly pressurizing the calandria tube to failure. The computer data acquisition system was programmed to store data three seconds before and one second after the annulus pressure dropped below 3 MPa. Measurements of the calandria tube deformations and failure mechanisms were made and the data were plotted. 5.0
EXPERIMENTAL RESULTS
The test conditions and test results for the four tests conducted in this phase are summarized in Tables I and II. The detailed results of each test are discussed below. 5.1
Test 1
The first test carried out used a new calandria tube with an average temperature of 170xC (average of five top and five bottom measurements) and a relatively slow pressurization rate. The pressure in the annulus rises rapidly to about 10 MPa in 1 second and then slowly rises as the calandria tube strains thus increasing its volume. The rise in pressure and strain results in a shortening of the calandria tube and a corresponding rise in the axial load imposed on the rolled joint. The calandria tube failed after about 8.0 s with a
104 CNS 9th ANNUAL CONFERENCE, 1988
failure of the rolled joint at the meveable end. The pressure at this time was 10.9 MPa with an axial end load of 254 kN. The pressure and axial end load histories are presented in Figure 2. The bottom 180x of the calandria tube sheared in the rolled joint while the top half pulled out. The average permanent hoop strain was 6.81% with a maximum of 7.93% near the fixed end. This location corresponded to the maximum wall thinning in the seam weld which was 18.7%. The axial strain was .78%. It should be noted that in this test no end shield cooling was provided for the hubs.
5.2
Test 2
The second test was again carried out with a new calandria tube but with a somewhat faster pressurization after the initial pressurization. For this test the end shield hubs were cooled. The pressure rose rapidly in about 1 second from 7 to 10 Mpa and then more slowly to failure at 11.6 MPa. Pressure and axial load histories are shown in Figure 3. Failure occurred in the seam weld running over a 2.2 m length and terminating in the fixed end rolled joint and a circumferential break in the calandria tube. This section of tube was flattened in the bottom of the blast tank. The average hoop strain in the calandria tube was 4.0% with a maximum of 4.4%. The maximum wall thinning in the seam weld was -8.5%. The axial strain was -.48% and the maximum end load was 231 kN. The pressurization time was 2.0 s.
5.3
Test 3
The calandria tube used in Test 3 had been subjected to a transient pressure history in burst test number 7. This resulted in average hoop strains of . 8 % and larger local strains at the garter springs. The calandria tube had been fitted with bosses to allow pressure measurements and these were removed and the holes were sealed with internal patches which were welded to the calandria tube. The pressure rose rapidly to 10.3 MPa in about 50 ms and failure occurred at one of the repaired bosses. The resulting average calandria tube hoop strain was 0.5% with a maximum of 0.74%. The maximum axial strain was -.13% and the axial end load was 208 kN. Local deformations caused by garter springs in the previous burst t^st were smoothed with no indication of increased local strains. Pressure and end load histories are s's.ovn in Figure 4. In other respects this test result was essentially discounted due to the premature failure at the boss location. 5.4
Test 4
Test 4 used a previously strained calandria tube from pressure tube burst test 8 with average hoop strain of 0.8% and larger local strains near the garter springs with a maximum of 2.2%. The pressure rose rapidly to 10 MPa in about 50 ms and then slowly rose to 11.45 MPa over 800 ms to failure. Failure occurred in the seam weld over a 1.1 m long section near the fixed end. The failure terminated in the fixed end rolled joint and in a complete circumferential break in the calandria tube. The average hoop strain was 3.54% with a maximum of 4.6%. The itaximum wall thinning in the seam weld in the unburst portion of the tube was 14.0%. The axial strain was -.66% with an axial end load o£ 154 kN. The local deformations from the
previous testi were smoothed with no indication o£ increased local strain. Pressure and end load histories are presented in Figure 5.
6.0
CONCLUSIONS
Four full scale burst tests using new and previously strained prototypic calandria tubes were carried out with different East pressurization times at 170xC. The failure at the rolled joint vas probably due to the uncooled hubs in the £irst test. The burst pressures in test 2 and 4 are very close even though the pressuriEation rates are different and the tube in test 4 had been previously strained. No new failure mechanisms were identified from testing previously strained tubes. Since these burst pressures are above the primary system operating pressure of 10 MPa these test indicate that for new tubes there is some margin to failure and for irradiated tubes with an increase of 20 to 30% in strength [3] the margin will be even greater.
7.0
ACKNOWLEDGEMENTS
The Calandria Tube Integrity Program is funded via the COG-CMJDEV agreement. The authors acknowledge the efforts of the staff at STSRN LABORATORIES INC. where the experiments were conducted. 8.0
REFERENCES
(1) MUZUMDAR, A.P., HADALLER.. G.I., CHASE, R., " Experimental Program to Determine the Consequences of Pressure Tube Rupture in CANDU Reactors", presented at the Thermal Reactor Safety Meeting, San Diego, California, February 1986. (2) HADALLER, G.I., MUZUMDAR, A.P., "Progress on an Experimental Program to Determine the Consequences o£ Pressure Tube Rupture in CANDU Reactors", presented at the CNS Conference, Saint John, New Brunswick, June 19B7. (3) COLEMAN, C.E., et al, "Properties of a CANDU Calandria Tube", Canadian Met. Quarterly, Vol. 24, No. 3, p. 215-223, 1985.
CNS 9th ANNUAL CONFERENCE, 1988 105
TABLE I
TEST CONDITIONS
Test Number
1
2
3
4
Galandria Tube No.
613
G16
338
611
Calandria Tube Condition
new
new
bottom
top
top
top
Calandria Tube 0.2% Yield (kN)
421
435
438
413
Calandria Tube U.T.S. (kN)
520
525
529
529
28.0
26.5
26.5
26.5
16.5
15.6
16.5
16.5
yes
yes
yes
Seam Weld Location
Elongation
(%)
Axial Preload (kN)
no
Rolled Joint Cooling
strained*
strained*
* From Pressure Tube Burst Test tt7 •t From Pressure Tube Burst Test SB
TABLE II
TEST RESULTS
Test dumber
1
Burst Pressure (MPd) Average Temperature (xC) Average Permanent Hoop Strain Maximum Hoop Strain
(%)
Permanent Axial Strain {%)
3 10.3*
4
11.6
170
170
174
169
£.81
4.1
0.5
3.54
7.93
4.4
0.74
4.60
11.45
-0.78
-0.48
-0.13
-0.66
Maximum Axial Load !kN)
254
231
208
229
Pressurization Time (s) (7 to 10 MPa)
0.8
1.0
.05
.05
Pressurization Time (s) (7MPa to failure)
8.5
2.0
0.1
0.8
-18.7
-8.5
-2.1
-14.0
rolled joint
seam weld
weld@ boss
seam weld
Maximum Seam Weld Strain {%) (based on wall thickness) Failure Mode
* Failed at welded boss.
106
(%)
2
10.9
CNS 9th ANNUAL CONFERENCE, 1988
; SI M)O
oooa
CNS 9th ANNUAL CONFERENCE, 1988
107
POMU LOOP LEVL
Cat BNN UEST V-8461 311 COL TUBE END LOfli Illtl HinP TflNK LEVEL (HI 1881
t.ABORflTOf;CS I 19S? JULT 15 CPL INTC «1 T « t Pt 2S Plot I
FIGURE 2: Calandria Tube Integrity Test #1 History
STCRN LAB0RRT0R1ES INC. 198? NOV. 26 CPL TU*£. INTC «Z
Fil#j IVHSCS:s34 R»c»i 395,1496, 16888
.
-
5 CAL ANN VEST V-866S86SI31 - LOUD END LOfit STR3 3 HETCO DIAMETRAL STR 111
uses 12988
£t lasse I
- —
8809 O Z
^v* ~ ~v v ' "
isee
-ieee TIME
-s»«
e
INILLISCC)
FIGURE 3: Calandria Tube Integrity Test #2 History
108 CNS 9th ANNUAL CONFERENCE. 1988
Fill: CVMST3ii3< R»cjs 1282,1313, 1 P«HW CBL KHH WEST V-066S86S(281 - LOUD END LOUD (121 STR3 3.75 NETER DIOH STRfllHUl)
i
STERN LH»ORHTOR1ES INC. 1988 Ftt. 11 CBL TU»E INT5 »3
«ecte
32
4888
2080
-t«e T1NC
-56 IHILLI5EC)
FIGURE 4: Calandria Tube Integrity Test #3 History
STERN LRIORRTORIES
12B9B
E L O l END LORD CDLC ( 1 3 ) PflNU CflL RNN WEST 128) STR3 3 . 7 S ftETER B f l ( )
1988 FEt. 29 COL :NTS TEST •<
INC.
seee
zees
T1HE (niLLISEC)
FIGURE 5: Calandria Tube Integrity Test #4 History
CNS 9th ANNUAL CONFERENCE, 1988
109
AN EXPERIMENTAL AND ANALYTICAL APPROACH TO DETERMINE BEARING-PAD TO PRESSURE-TUBE HEAT TRANSFER
J.W. DEVAAL, M.H. SCHANKULA and V.D. KROEGER High-Temperature Chemistry Branch Vhiteshell Nuclear Research Establishment Pinawa, Manitoba
ROE 1LO
and D.B. REEVES, A.P. MUZUMDAB and E.D. SHERIDAN Ontario Hydro 700 University Avenue Toronto, Ontario M5G 1X6
ABSTRACT An experimental apparatus has been developed for investigating bearing-pad to pressure-tube heat transfer under simulated post-LOCA conditions. Results of experiments performed with this apparatus are discussed, and a procedure for extracting bearing-pad heat transfer coefficients from the experimental data is presented. Important factors affecting bearing-pad heat transfer during transient heating of the fuel element are also identified, and the approach taken for future experimentation is discussed.
INTRODUCTION FE/PT Contact
During a postulated large-break Loss-of-Coolant Accident (LOCA), rapid voiding of the coolant may occur in some fuel channels, followed by a period of steam cooling. During this degraded cooling period, changes in the geometry of the fuel and fuel channel components will occur if sufficiently high temperatures are reached (1). Among possible distortion modes, diametral straining (ballooning) of the pressure tube into contact with the calandria tube occurs if the pressure-tube temperature and internal pressure are high. Interaction between the fuel and pressure tube may occur through two different mechanisms. One that has been analytically investigated at Ontario Hydro using ihe computer code MINI-SMARTT (2) is the distortion of fuel elements into contact with the pressure tube (FE/PT contact) (3). If FE/PT contact occurs prior to pressure tube bal.iooning, it may create a hot spot on the pressure tube leading to local strain failure. A second interaction may take place as a result of contact between bearing-pad surfaces and the pressure tube throughout the transient. This mechanism is referred to as bearing-pad to pressure-tube contact (BP/PT contact). BP/PT contact may also lead to local pressure tube strain and under certain severe conditions to local strain failure. The LOCA conditions under which BP/PT contact occurs are highly varying. For example, the pressure, coolant conditions, surface contact conditions, oxidation, radiation, fuel element power, and resultant contact resistance between the bearing pad and pressure tube all vary throughout the transient. Figure 1 illustrates the differences between the FE/PT and BP/PT ccutact mechanisms. The MINI-SMARTT computer code is a two-dimensional finite-difference code capable of analyzing the heat
110 CNS 9th ANNUAL CONFERENCE, 1988
FIGURE 1:
COMPARISON BETWEEN FE/PT AND BP/PT CONTACT
transfer conditions in a cross-section of a CANDU fuel channel under both transient and steady-state conditions. This is ideal for the analysis of FE/PT contact where the axial contact length is large compared to the contact width. For BP/PT contact, the contact width is generally assumed to be the width of the bearing pad (2.5 ram) while the contact length is the bearing pad length {approximately 28 m m ) . Since the contact length is only a factor of 10 greater than the contact width, three-dimensional effects may have an important influence on tJte thermal gradients and resultant pressure tube strain. For this reason a new analytical approach has been taken by using the ANSYS (4) finite element computer code. This code has the capability to analyze threedimensional transient heat transfer conditions and solve for the resultant local strain. The recent release of ANSYS Version 4.3 also includes the ability to generate radiation view factors for all surfaces in the solid model. The purpose of this paper is to present the results of initial BP/PT experiments, and to discuss the analytical approach taken for determining bearing-pad heat transfer coefficients from the results of these experiments. The first section of this paper discusses the experimental approach taken at the tfhiteshell Nuclear Research Establishment (VNRE) to determine the contact heat transfer coefficient for BP/PT contact and the effects of this contact under LOCA conditions. The second section discusses the analytical approach taken at Ontario Hydro in simulating the experimental results. The paper
concludes with recommendations for future research and analysis.
EXPERIMENTAL APPARATUS AND METHOD Experimental Apparatus Figure 2 illustrates the experimental apparatus constructed for investigating bearing-pad to pressure-tube heat transfer. In operation, the electrically heated fuel-element simulator transfers heat through a contacting bearing pad to a small section of pressure tube located on top of a Zr2.5 vt.Z Nb block. As noted in Figure 2, thermocouples attached to this apparatus measure: heater core temperature, JEuel-sheath temperature, bearing-pad interface temperatures, and the temperature profile in the supporting Zr-2.5 wt.% Mb block. Bearing-pad interface temperatures are measured using three K-type surface temperature probes*, inserted through a series of 0.8-mm-diameter holes drilled through the support block and the pressure tube specimen. (See partial section in Figure 2 ) . With the exception of tvo of these contact thermocouples, all of the remaining thermocouples have been attached to the midplane of the experimental apparatus. This arrangement has been deliberately chosen to permit direct comparisons between two-dimensional computer simulations of the experiments and the measured thermal response of the midplane of the apparatus.
-L CooUhIG WATER.
J
FIGURE 2:
2-4
fuel-sheath temperatures
5-7
bearing-cud interface temperatures
S-12
temperatures in lover support block
DRAWING OF EXPERIMENTAL APPARATUS ILLUSTRATING THERMOCOUPLE LOCATIONS
Figure 3 presents a schematic of the instrumentation for the experimental apparatus and the environmental chamber surrounding the apparatus. This environmental chamber permits experiments to be performed in vacuum or gas environments up to atmospheric pressure. A linear motion feed-through arrangement at the top of this chamber permits
i
SuppLr IIO-A.C
FIGURE 3:
©®a>—s
J_
^
SCHEMATIC OF INSTRUMENTATION FOR THE EXPERIMENTAL APPARATUS
changes in applied load on the bearing pad to be made at any time or under environmental conditions during an experiment. The temperature of the fuel-element simulator is controlled through a 3000-VA AC power supply connected to the fuel element heater, and the pressure-tube temperature is independently controlled through a separate 300-W DC power supply. This smaller power supply controls a pair of 120-U cartridge heaters, mounted in the stainless-steel support block below the apparatus. A water-cooled sink mounted in the bottom of the test chamber removes heat from the experiment and ensures that a thermal gradient will result in the pressure-tube specimen and Zr-2.5 vt.% Nb block. Temperature data are collected during the experiments using an automated data-acquisition system. The main components of this system are 1) an Analog Devices uMAC-5000 computer; 2) an Apple lie personal computer, which acts as a terminal for the MMAC-5000; 3) an Epson LX-80 printer; and 4) a Hewlett Packard 7440A 8-pen plotter. In operation, the uMAC-5000 computer collects data from the tvelve thermocouples attached to the apparatus, saves this data in memory, and then later transfers this data to floppy disks through the Apple H e computer. Temperature data files stored on disk can be retrieved for further data reduction, and an extensive collection of programs has been written for outputting this data to the 7440A plotter or LX-80 printer.
' Product of Medtherm Corp., Huntsville, Alabama, U.S.A.
CNS 9th ANNUAL CONFERENCE, 1988 111
Experimental Method The experimental apparatus can be used to investigate bearing pad to pressure tube heat transfer under different heating rates, loads, and environmental conditions. R&sltlls ot these various experiments can be quantified and compared by considering how the overall fuel element to pressure tube thermal resistance changes as a function of different test conditions. This overall thermal resistance, R £ e , p t , is given by:
previous maximum fuel-sheath temperature and as a function of previous maximum contact thermocouple temperature. This information is also presented in Figure 4, revealing a significant decline in overall resistance whenever contact thermocouple temperatures exceed approximately 8Q0°C.
TABLE 1 MEASURED RESISTANCES FOR THE FIRST EXPERIMENT
- 7,.
-)
Sheath Temp.
(1)
CO where T f o is the fuel-element surface temperature avay from the bearing pad, T p t is the extrapolated pressure-tube surface temperature away from the contacting bearing pad, and 0 b l k is the rate of heat flov through the pressure tube and the lower support block. This heat-flow rate can be determined for steady-state conditions by: = -k
400 600 800 900 1000 1100
(2)
200 450 650 7^0 890 1000
Measured Resistance*
(°C/W) 59.6 44.5 32.2
3.3 3.4 3.3
Measured in vacuum, with a 40 N load, at steady-state fuel-sheath temperatures of approximately -400°C.
where k is the thermal conductivity of the block, (dT/dx| is the temperature gradient in the block, and A b l k is the cross-sectional area of the pressure-tube specimen and the lower support block. A computer program has been written for calculating overall thermal resistance (i.e., R fB _ pt values) from temperature data files collected during the experiments. Although this program only uses the simple steady-state resistance formulr.tion presented in Equations (1) and ( 2 ) , this elementary approach has provided a convenient "on-line" meLhod for quantifying changes in the overall resistance following transient heating. Calculating R fe _ pt has proved invaluable for quantifying the results of bearing-pad heat transfer experiments, and for indicating how the bearing-pad heat transfer coefficient changes as a function of test conditions during the experiments.
Max. Contact T/C Temp. (°C>
Concerns that long times at high temperatures may have caused this decline in resistance (6) were addressed through a change in test procedure for the second experiment. In this experiment, performed in an argon environment with an 80 N bearing-pad load, overall resistances were determined following rapid temperature excursions interrupted at successively higher fuel-sheath temperatures. Initial conditions for these transient tests were steady-state heat transfer through the contacting bearing pad with the fuel-sheath temperature maintained at 300°C. Steadystate resistances, determined following these individual temperature excursions, are presented in Table 2, and in Figure 4.
TABLE 2 EXPERIMENTAL RESULTS MEASURED RESISTANCES FOR THE SECOND EXPERIMENT Six bearing-pad heat transfer experiments have been performed since the apparatus was adapted for studying BP/PT contact in late 1986. The first four experiments in this series revealed that the overall resistance is a strong function of bearing-pad surface temperature when the experiments are performed in vacuum or argon environments. These experiments have been documented in detail elsewhere (5) and will only be discussed briefly here. Later experiments, which compare resistance behaviours in oxidizing and inert environments, (i.e., the fifth versus sixth experiment) have revealed that bearingpad surface oxidation may delay or prevent the transition to lower resistances seen in the nonoxidizing experiments. Results of these experiments will be discussed in more detail after the following review of the first four experiments. Results of Experiments Performed in Vacuum and Argon In the first experiment, performed in vacuum with a 40 N load acting on the bearing pad, overall thermal resistances vere determined following 1 to 6 h excursions to 400, 600, 800, 900, 1000 and 1100°C fuel-sheath temperatures. Table 1 presents these steady-state resistances, both as a function of 112 C N S 9th A N N U A L C O N F E R E N C E , 1988
Sheath Temp. (°C>
309 559 675 711 752 826 829 917 1002 1097 1177
b
Max. Contact T/C Temp. (°C)
231 278 341 349 365 397 405
428 455 504 612
Measured Resistance13
CC/W) 11.4 11.3 11.3 12.1 13.3 13.3 12.8 14.0 15.5 15.8 13.6
Measured in argon, with an 80 N load, at steady-state fuel-sheath temperatures of 300"C.
As noted in comparing resistances from the first and second experiments in Figure 4, the addition of argon (at atmospheric pressure) significantly reduces
R
io-pt a n d produces a more uniform resistance behaviour for contact thermocouple temperatures less than 600°C. This significant reduction in R fe _ pt with the addition of argon indicates that, at temperatures less than 600°C, more heat is transferred by conduction through the interstitial gas-gap between the bearing pad and pressure tube than is transferied by conduction through the solidto-solid contacts at this interface.
procedure was repeated at various fuel-sheath te.:|;eratuies to investigate the time required to reach steady-state, and to determine the steady-state resistance values generated at different fuel-sheath temperatures. Contact was maintained for a period of 3.5 minutes during each of these drop-te^ts and, as in the second experiment, this experiment was performed in argon with an 80 N load acting on the bearing pad. Table i presents the results of this experiment, and Figure 4 illustrates that the transition to lower resistances seen in the first experiment was also seen in this experiment.
90 ao -
s -70
EXP' T EXP- T EXP' T EXP' T EXP' T
NO. NO. ND. NO. NO.
1 2 3 A 4
(Vacuum) (Argonl (Argon) (Argonl (Vacuum)
TABLE 3 MEASURED RESISTANCES FOR THE THIRD EXPERIMENT Sheath Temp.
Max. Contact T/C Temp.
Measured Resistance 0 CC/V)
795 910
620 655 715
12.0 4.6 3.3
MO j
30
= 10 0
\ 100
FIGURE 4:
Measured in argon, with an 80 N load, after i.5 minutes in contact.
200 300 400 S00 600 700 600 900 1000 1100 MAXIMUM CONTACT T/C TEMPERATURE ("CI
MEASURED RESISTANCE BEHAVIOUR FOR ALL OF THE EXPERIMENTS
The resistance behaviour generated by this interrupted-transient technique indicated that, even when fuel-sheath temperatures near 1200°C were achieved, significant decreases in overall resistance failed to occur. This finding suggests that changes in resistance, which occur during rapid overheating of the fuel element, were probably controlled by the local surface temperature of the bearing pad during the transient and not by the fuel-sheath temperature as had initially been assumed. This hypothesis (confirmed in later experiments) helps to explain why extremely high fuel-sheath temperatures could be achieved in this second experiment with no resulting decline in resistance. Given that the contact thermocouples provide a reasonable estimate of the bearing-pad surface temperature, an examination of the resistance behaviour in the second experiment (Figure 4) revealed that bearing-pad surface temperatures probably never exceeded 65O°C during the transient tests performed in this experiment. This observation prompted further development of the fuel-element heater (to increase its maximum temperature capability), and underscored the importance of transient testing to simulate the BP/PT conditions expected during a large-break LOCA. In the third experiment, a different test procedure was used. Unlike the other experiments, where the fuel-element was heated in contact with the pressure-tube specimen, the fuel-element in this experiment was first heated to a steady-state temperature and then quickly lowered into contact with the pressure-tube specimen. This drop-type test
In the fourth experiment in this series, improvements were made to the fuel-element heater. These improvements generated higher contact thermocouple temperatures and prompted a return to the interrupted-transient test procedure used in the second experiment. This test procedure was also further modified to determine the effect of bearingpad load on solid-to-solid heat transfer at the BP/PT interface. This was done by determining the overall resistance behaviour in vacuum as a function of bearing-pad load, both before and after a series of interrupted-transient tests in argon (see Table 4 ) . The initial vacuum tests in this experiment (see first tour entries in Table 4) revealed that the overall resistance was initially quite high and very insensitive to changes in bearing-pad load. When these resistances were compared with the lowest temperature resistance in the first experiment (see Figure 4 ) , it became apparent that somewhat higher resistances were generated in the fourth experiment than in the previous experiments. This behaviour was suspected to be related to the initial fit between the bearing pad and pressure tube, which may not have been as good in this experiment as in the earlier experiments. This suspicion was confirmed when addition of the argon cover gas decreased the overall resistance to only 30.9°C/W (see first entry in the second section of Table 4) as compared with Rt,.pt values of 11.3 to 15.8°C/W in the second experiment. After the addition of argon, resistances were determined following transient heatings, where these transients were interrupted at successively higher fuel-sheath and bearing-pad temperatures (see remaining entries in the second section of Table 4 ) . This testing revealed a significant decrease in resistance for contact thermocouple temperatures approaching 800°C and, as noted in Figure 4, this behaviour is similar to the decreases in resistance seen in the first and third experiments. This common C N S 9th A N N U A L C O N F E R E N C E . 19BB 113
behaviour between the first, third and fourth experiments suggests that, although the initial resistance controls the local heating rate of the pressure tube, the transition to lover resistances depends only on some threshold bearing-pad surface temperature.
TABLE 4 MEASURED RESISTANCES FOR THE FOURTH EXPERIMENT Experimental Condi tions
vacuum, vacuum, vacuum, vacuum, argon, argon, argon, argon, argon, vacuum, vacuum, vacuum, vacuum,
d
80 60 40 20
80 80 80 80 80 80 60 40 20
N N N N
N N N N N N N N N
Sheath Temp.
Max. Contact T/C Temp.
Measured Resistance
(°C)
(°C)
(°C/W)
340 340 335 340
120 110 145 135
77.0 80.0 80.0 88.0
356 893 1058 1174 1177
306 473 533 731 796
30.9" 35.6" 34.4 a 12.0" 3.3"
326 325 335 350
270 270 270 270
22.0 25.0 35.0 66.0
Post-transient resistances, measured in argon, at 300°C fuel-sheath temperatures.
Information on the nature of the decline in overall resistance can be obtained from the fourth experiment by comparing resistances determined ip vacuum and argon before and after the series of transient tests. A comparison between the initial and final resistances in vacuum (i.e., 77 •» 88°C/W to 22 -» 66°C/W) reveals a moderate decrease in resistance, where this decrease can be directly attributed to an increase in the overall solid-tosolid contact area. This decrease in in-vacuo resistance, however, is much less than the associated change in resistance measured in argon (i.e., 30.9°C/tf to 3.3°C/W). These comparisons reveal that, although there is an increase in solidto-solid contact area, the decline in overall resistance is primarily related to a significant reduction in the resistance to heat transfer through the interstitial gas-gap between the bearing-pad and pressure-tube surfaces. Although substantially different test methods and environments were used in these four experiments, several common conclusions could be drawn from the resistance behaviours observed in these initial experiments: 1)
2)
Significant decreases in overall resistance were observed in these experiments whenever contact thermocouple temperatures exceeded 800°C. This similar behaviour between steady-state and transient experiments indicated that the transition to lower overall resistances vas controlled by the temperature of the bearing pad at the BP/PT interface. Examination of resistances measured in vacuum and argon before and after this transition (see Table 4) revealed that the reduction in overall
114 C N S 9th A N N U A L C O N F E R E N C E , 1988
resistance was primarily related to a significant drop in the resistance to heat transfer through the interstitial gas-gap. This observation indicated that the transition to lower resistance was associated with the collapse of the interstitial gap between the bearing pad and pressure tube at high bearing-pad temperatures. Collapse of the interstitial gap between the bearing pad and pressure tube is believed to be directly related to the high-temperature yield stress of Zircaloy-4. This suggests that oxidation factors that affect the high-temperature yield stress of Zircaloy-4 may significantly affect the temperature at which bearing pad collapse occurs. Oxidation and oxygen pick-up are processes known to cause an increase in the strength of Zircaloy-4 at high temperatures. This suggests that oxidation processes, which would naturally occur in the steam environment following the L0CA, could possibly increase the strength of the bearing pad and delay collapse of the interstitial gap to higher temperatures. To test this hypothesis, the fifth bearing-pad heat transfer experiment was performed in a 25%-oxygen 75%-argon environment, since this particular gas mixture has been shown by Uetsuka and Hofmann (7) to closely simulate the oxidation rate of steam at high temperatures. Effects of Oxidation on Bearing-Pad Heat Transfer Although the approach of using a series of interrupted-transient heatings worked quite well for determining bearing-pad heat transfer behaviour in inert environments, the cumulative effects of oxidation precludes using this approach in experiments performed in an oxidizing en/ironment. Consequently, overall resistances were measured in *.he fifth experiment both before and after a single fansier.t heating in the 25X-oxygen 75^-argon environment. Table 5 presents the results of this experiment, where the first entry in the table represents the steady-state resistance value before the transient, and the second entry represents the steady-state resistance value (measured at a 300°C sheath temperature) after the transient. As noted in Table 5, this severe overheating of the fuel element failed to significantly reduce the steady-state resistance measured following the transient.
TABLE 5 MEASURED RESISTANCES FOR THE FIFTH EXPERIMENT Experimental Conditions
Sheath Temp.
Max. Contact T/C Temp.
Measured Resistance
Pre-transient Ar-0,, 80 N
300
170
13.0
Post-transient A r - 0 2 ) 80 N 1080
775
10.7
The sixth experiment, performed in argon, was intended as a control experiment for the fifth experiment. In this experiment, attempts were made to reproduce the test conditions of the previous transient heating to determine if the absence of oxidation would affect the measured post-transient resistance. Table 6 presents the results of this
rppeat experiment in argon, and illustrates a significant decrease in steady-state resistance following the transient (i.e., 13.0°C/W to 5.5°C/W). Although attempts were made to reproduce the fuelelement thermal response recorded in the fifth experiment a comparison between Tables 5 and 6 reveals that somewhat higher maximum fuel-sheath and contact thermocouple temperatures were recorded in the sixth experiment. Although this indicates that the sixth experiment vas not an exact repeat of the fifth experiment, a comparison between all of the measured resistances in Figure 4 suggests that oxidation has an effect in preventing the transition to lower resistances seen in the argon and vacuum experiments.
TABLE 6 MEASURED RESISTANCES FOR THE SIXTH EXPERIMENT
Experimental Conditions
Pre-transient Ar, 80 N
Sheath Temp. (°C)
295
Max. Contact T/C temp. (°C)
200
Measured Resistance (-C/W)
13.0 PIGUKI-: b:
Post-transient Ac, 80 N
1180
825
EXAMPLE OF 3-D AND 2-D WIRE-FRAME MODELS OF THE EXPERIMENTAL APPARATUS
5.5
MODELLING OF THE EXPERIMENTS USING ANSYS The BP/PT experiments performed to date have shown that the bearing-pad heat transfer coefficient is not generally constant during transient heating of the fuel, but instead .significantly increases when some threshold bearing-pad surface temperature is reached. This finding indicates that modelling of the transient experiments is necessary in order to separate and quantify the various heat transfer modes operating during the fuel overheating process. Since the MINI-SKARTT computer code is used primarily for reactor licensing, it vas felt thai analysis of the experiments should be accomplished using a verified computer code other than MINISMARTT. A general purpose finite element code, ANSYS, was chosen for this purpose because it can model heat conduction in three dimensions, and because it can model radiative heat transfer between various SULfaces of the modelled solids. Before ANSYS is run, a scaled solid model of the system under analysis is constructed using the PATRAN (8) interactive solid modelling processor. This processor generates wire frame models of the various system components anJ allows boundary conditions to be applied to the surfaces of the modelled solids. These wire frames define the nodes and links that comprise the individual elements in the model. Figure 5 shows an example of a wire-frame model of the BP/PT apparatus. In developing the finiteelement mesh used in this analysis, certain nodes were placed in the same locations as thermocouples attached to the apparatus. This allows a direct comparison between calculated and measured temperatures. Radiative heat transfer between tile fuel element and pressure-lube surfaces becomes an important
compel iiitf mode of heat transfer at higher fuel sheath temperatures. Recognizing ibis, the ANSYS thermal model was constructed to include radiative heat transfei between the fuel-element and pressure-tube surfaces outside the bearing-pad contact area. The ANSYS code models gray-body radiative heat transfer between individual surface elements with the view factors for individual element pairs determined using a ray tincing procedure (4). Since radiation heattransfei relationships are quartic in nature, modelling radiation involves iteration within each time step in a linear finite element code. The ANSYS code performs this iteration and expresses the radiative heat transfer between surfaces in terms of an effective heat transfer coefficient, h,. Upon completion of the thermal analysis, the results may be printed in tabular form or plotted in a number of ways. For example, the heat flux and temperature distributions can be plotted as a threedimensional colour representation of the model in which the colours represent a scale of temperatures or heal flux. Figure 6 shows an example of this technique, plotted in gray-scale. The temperature distribution and any other necessary data generated in the thermal modelling is saved in a d.:ta file and may be used as boundary conditions for a thermal stress analysis using ANSYS. Stress relationships for the materials modelled may be supplied by the user if required. Analytical Procedure Several different approaches have been used in modelling the experiments with the ANSYS code. The first approach, chosen to investigate the shape of the "hot spot" in the pressure-tube specimen, modelled the heater, fuel-sheath, bearing pad, pressure lube and lower support block under steadystate conditions in a vacuum. Radiative heat transJei between the fuel element and pressure tube was neglected in this analysis. Figure 6 shows the temperature distribution predicted by this three-
CNS 9th ANNUAL CONFERENCE, 1988 115
995°C 1US4. 99S. 93b. 876.
In this approach, the heater model was neglected and fuel-sheath temperatures in Hie ANSYS model were fixed to the measured sheath temperatures from the experiments. In this vay the bearing-pad heat transfer coefficient could l)e determined without the added uncertainty of internal heat transfer conditions in the fuel-element heater. This twodimensional approach was also extended in later work to account for the effects of radiative heat transfer between the fuel sheath and pressure tube surfaces.
817. 758. 699. 639. S88. 521. 16?. 403. 343.
225°C
FIGURE 6:
2S4.
RESULTS OF 3-D THERMAL ANALYSIS UITH ANSYS (In-vacuo, h r ,, t = 0, h b = 1.0 kU/(m 2 -°C), Q i n = 30.3 V)
Figure 8 presents a flow chart outlining the procedure used to determine steady-state values of bearing-pad heat transfer coefficient (h^ p ) from the experiments using the ANSYS code. In this fitting procedure, the code is run with a given value of \ f to test if it produces the same extrapolated pressure-tube surface temperature (T t ) as in the experiment. If the estimated h b value is too high, the code will overpredlct 1 t relative to the experimental value. Uhen this occurs, the h b p estimate is reduced and the code is rerun to assess the reduction in (T p t ) c o d l ,. When convergence between ( T p t ) c o d e and (T t ) . , p t is achieved, h b p is established as the steady-state bearing-pad heal transfer coefficient for the modelled te^•.
For a given steady-state reading,
dimensional conduction analysis, where input heater power and lower block temperatures were chosen to simulate the results of one of the low temperature tests in the first experiment. The isotherms predicted in this initial analysis revealed that, although there was a three-dimensional effect, a twodimensional analysis for heat transfer would be valid in the bulk of the contact zone. Although this initial analysis successfully generated the shape of the "hot spot" in the pressure-tube specimen, f number of difficulties were revealed- The full three-dimensional analysis made the addition of radiative heat transfer a complicated task, and it was not possible to correctly match predicted temperatures with the experimental results due to unknown heat transfer conditions inside the fuei-element heater. These shortcomings prompted the development of a simpler two-dimensional model, which was also used in analyzing results from the first experiment (see Figure 7 ) .
Thermocouple 2 (Fixed Temperature) Bearing Pad
Boundary Conditions 1) Set sheath at T f> 2) Apply q to bottom of lower block 1) Apply known er., cF values id h r<( ,_ pt model
Fuel Sheath thermocouple A (tixed Temperature) Pressure Tube
- Zr 2.5 vt.Z Nb Block
Thermocouple 12 Tecujetiatucel
FIGURE 7:
2-D MODELLING APPROACH FOR SENSITIVITY STUDIES
116 C N S 91h A N N U A L C O N F E R E N C E , 1988
This approach for determining h b p assumes that all other modes oE heat transfer between ihe fuel sheath and pressure tube have been accurately modelled. As
comparisons given in the next section will shov, radiative heat transfer between the fuel sheath and pressure tube can have a significant effect on the fitted h b p value. For this reason, separate tests with no contact between the bearing pad and pressure tube have been performed (not reported here) for determining fuel sheath and pressure-tube emittances (i.e., E f s and E p t , respectively). Extending the fitting procedure shown in Figure 8 to d e ^ r m i n e the behaviour of h b p during a rapid overheating of the fuel element requires performing a step-by-s^p transient analysis of the apparatus with the ANSYS code. In this procedure, the initial steady-state h b value will be corrected as necessary, following the modelling of the behaviour during small increments in time throughout the overheating process. The corrected h b p values can then be considered as specific values of h b p (t) for the transient test.
QQO -
^ ^ - ^ ^ ^
*40 -
?""t - 400 u 300 a
Experiment 1. Sensitivity studies examining the effects of various bearing-pad heat transfer coefficient and emittance values have been carried out for several tests from the first and third experiments. The first study modelled a lowtemperature in-vacuo test from the first experiment with various assumed values of h b p . (Radiative heat transfer was neglected in this first study). Figure 9 shows a plot of temperature versus thermocouple position for the various assumed h b p values. Note that a very low value of 0.04 kU/(m 2 - D C) was required to adequately match the thermocouple readings in this test. This low h b p value is expected, however, since Rf e - p t measurements in vacuum have already indicated that relatively little heat transfer occurs through solid-to-solid contact at low bearing-pad temperatures.
u - measured
380 -
a. 340 |
• - « , . . £„ . 0.2
320-
• - E,. - e pl . 0.5
300 -
*-<•,.'
200 -
tBt
- 0.8
280 220 Thermocouple No.
FIGURE 10:
Sensitivity Analyses
h b p - O.OC kU/(m'- CJ
-30 *60 -
EFFECT OF ADDING RADIATIVE HEAT TRANSFER TO LOW TEMPERATURE IN-VACUO CASE PRESENTED IN FIGURE 9 (hb = 0 . 0 4 kW/(m 2 -°C), and tf. = e B t = 0.2, 0.5 and 0.8)
any definite conclusions, yet it does show that the ANSYS model gives a reasonable prediction. Experiment 3. In tests performed in the third experiment, fuel sheath and pressure-tube temperatures were higher than in the initial tests of the first experiment. These tests were also performed in an argon environment. This indicates that radiation and convection between the fuel sheath and pressure tube will play a more important role than in the previous analyses. Although convection is anticipated, this was not modelled in the following preliminary sensitivity studies. Figure 11 shows the results of a study that assumes an h b of 0.5 kU/(tn2-°C) and emittances of 0.1, 0.2 and 0.5. At this heat transfer coefficient, emittances of 0.5 and less underpredict the test results by approximately 20°C.
D
- neasured
Thermocouple Ko.
FIGURE 9:
MEASURED VERSUS CALCULATED TEMPERATURE PROFILES IN LOWER SUPPORT BLOCK FOR VARIOUS h b p VALUES (Low temperature, invacuo test from E:;pt. 1, h r , f t = 0)
The second study modelled radiation and examined the effect of emittance at a fixed h b p of 0.04 kW/(m 2 '°C). Figure 10 shows that an emittance of 0.2 matches the thermocouple readings better. An emittance value of 0,5 has very little effect on the prediction. An emittance value of 0.Bf however, considerably overpredicts the temperatures. Temperatures are quite low in this test preventing
TheraDcouple No.
FIGURE 11:
MEASURED VERSUS CALCULATED TEMPERATURE PROFILES IN LOWER SUPPORT BLOCK FOR VARIOUS EMITTANCE VALUES (High temperature test in argon from Expt. 3, h b p = 0.5 kW/(m!-°C),and c t . = e pt = 0.1, 0.2 and 0.5)
C N S 9th A N N U A L C O N F E R E N C E , 1988 117
The effect of varying h b at an assumed emittance of 0.2 is given in Figure 12. A close match with the experiment is found when h b = i.3 kV/(m 2 *°C). Raising the emissivity value would of course result in obtaining the same close match for a lover heat transfer coefficient. Since the two variables, emittance and h b , are interdependent in these sensitivity studies, it is not possible to determine n b P without knowing the emittance value. As mentioned previously, separate tests have been performed to determine emittance values, and future analysis will be directed at using these values to obtain better estimates of hht,.
These initial examples, in the form of sensitivity studies, have also illustrated that the value of heat transfer coefficient (h b ) determined by this procedure is quite sensitive to changes in radiative heat transfer between the fuel-sheath and pressuretube surfaces. This finding has revealed the need for a separate series of noncontact heat-transfer tests, before and aEter the transient tests in our experiments, to determine accurate values of emittance tor the fuel sheath and pressure tube surfaces.
FUTURE RESEARCH
Thermocouple No.
FIGURE 12: EMITTANCE OF 0.2 FOR HIGH TEMPERATURE CASE PRESENTED IN FIGURE 11 0.2, h b 0.5, 1.0 and
W/(m2-°C))
Experiments have shown that oxidation effects on both the bearing-pad and pressure-tube surfaces may be important during transient heating of the fuel element. This finding has led to a redesign of the experimental apparatus lo better simulate the postLOCA thermal response of the pressure tube. In this new arrangement, illustrated in Figure 13, the pressure-tube specimen is supported on six O.B5-mmdiameter alumina pins above the lover support block. This new arrangement preserves the entire vail thickness of the pressure-tube specimen, while maintaining a 9-mm gap between the specimen and lower support block (simulating the pressure tube to calandria tube g a p ) . Heat loss through the supporting pins is expected to be extremely small and the dominant heat transfer mode is anticipated to be radiative loss from tlie bottom of the pressure-tube specimen. To better simulate the radiative conditions that would exist in-reactor, a radius has been machined in the top of the lower support block matching the inside curvature of the calandria tube and a thin, stainless-steel heat shield has been spot-welded CJrcuraferentially around the gap to reflect incident radiation from the specimen and the lover block.
CONCLUSIONS An experimental apparatus has been constructed for investigating bearing-pad heat transfer under conditions similar to those expected following a large-break LOCA. Experiments performed in vacuum and argon environments with this apparatus have shown that the bearing-pad heat transfer coefficient is generally constant at lower bearing-pad temperatures, but rapidly increases when some threshold surface temperature is exceeded. This behaviour is believed to be related to a collapse of the interstitial gasgap between the bearing pad and pressure-tube surfaces, brought on by a softening of the Zircaloy-4 bearing pad at high temperatures. Results of an experiment performed in a 2-.?-oxygen 75%-argon environment suggest that oxidafion effects during transient heating of the fuel element increases the strength of the hearing pad and delays collapse of the interstitial gap to higher bearing-pad surface temperatures. Recognizing that the bearing-pad heat transfer coefficient is not constant during transient heating of the fuel element has prompted development of an analytical procedure for determining bearing-pad heat transfer coefficients from the experimental data. Simulation oi the experiments vith the ANSYS computer code forms the basis of this procedure, and several examples illustrating the determination of steadystate heat transfer coefficients using this procedure have been provided.
118 C N S 9th A N N U A L C O N F E R E N C E . 1988
FIGURE 13:
DRAWING OF REDESIGNED EXPERIMENTAL APPARATUS
An initial experiment performed using this redesigned apparatus has shown that the pressure-tube specimen tempeiatuie rises much faster than was the
case in our previous experiments. This faster temperature response more closely reproduces the thermal conditions expected during the accident scenario and ensures that bearing-pad and pressuretube oxidation will occur under conditions similar to those expected following a large-break LOCA.
(2) REEVES, D.B., AKCHINOPF, G.H. and MUZUMDAR, A.P., "Development of the MINI-SHARTT Code for Fuel Element/Pressure Tube Contact", International Conference on Simulation Methods in Nuclear Engineering, 1986 October, 2,id, Canadian Nuclear Society.
A more sophisticated modelling approach must now be used to extract bearing-pad heat transfer coefficients from the experimental data obtained with this new configuration. In this regard, further development of the ANSYS simulation package is required to better characterize heat transfer to and from the suspended pressure-tube specimen. Development in this area is continuing, and it is anticipated that future experiments performed with this new apparatus will lead to models for predicting bearing-pad to pressure-tube heat transfer under transient conditions in a steam environment.
(3) REEVES, D.B., ARCHINOFF, H., MUZUMDAR, A.P. and LOCKE, K.E., "Thermal-Mechanical Modelling of the Pressure Tube Follovi.ig Fuel Element Contact Under LOCA Conditions in a CANDU-PHVR", ANS/ENS International Topical Meeting on Thermal Reactor Safety, San Diego, CA., 1986 Feb. 2-6, American Nuclear Society. (4) ANSYS Revision 4.3 Volume 1 - Heat Transfer, DNT001-1, 1987 March 30, Swanson Analysis Systems Inc., P.O. Box 65, Johnson Hd., Houston, PA., 15342-0065.
REFERENCES
(5) DEVAAL, J.U., SCHANKULA, M.H., and KHOEGER, V.D., unpublished data.
(1) BROWN, R.A., BLAHNIK, C. and MUZUMDAR, A.P., "Degraded Cooling in a CANDU", Testing and Analysis of Safety/Relief Valve Performance, Second International Topical Meeting on Nuclear Reactor Thermal-Hydraulics, 1983 Jan. 11-14, American Society of Mechanical Engineers.
(6) DEVAAL, J.W. and CHOUBEY, R., unpublished data.
(8) PDA Engineering, 2975 Redhill Ave. CA., 92626.
Costa Mesa,
CNS 9th ANNUAL CONFERENCE, 1988
119
THE EXPERIMENTAL MEASUREMENT OF CIRCUMFERENTIAL TEMPERATURE DISTRIBUTIONS DEVELOPED ON PRESSURE TUBES UNDER STRATIFIED TWO-PHASE FLOW CONDITIONS
P.S. YUEN, C.B. SO, R.G. MOYER and D.G. LITRE
Atomic Energy of Canada Limited Reactor Development-Program Responsibility Centre Whiteshell Nuclear Research Establishment Pinawa, Manitoba ROE 1L0
ABSTRACT Three experiments have been conducted to measure the circumferential temperature distribution and deformation o£ heated horizontal pressure tubes under stratified coolant conditions. During the experiments, stratified coolant flow resulted in local hot spots at the top of the pressure tube. Experimental data and observations from these tests are presented.
from the first three experiments of the "Make-Up water" test series. The objectives of the three experiments were to measure the circumferential temperature distribution and the deformation of heated pressure tubes as a function of heating power and coolant level at a constant pressure-tube internal pressure. Data from these experiments will be used to verify the computer codes mentioned earlier.
INTRODUCTION APPARATUS In some postulated loss-of-coolant accidents (LOCAs), coolant flow in a horizontal fuel channel of a CANDU reactor may boil off and become stratified (1). The exposed part of the pressure tube will be heated up by radiant heat from uncovered fuel pins and steam convection. The pressure tube vill then become hotter at the top while it remains close to the saturation temperature below the liquid level. The result is a circumferential temperature gradient around the pressure tube. If the coolant voiding rate is high, the circumferential temperature gradient will be relatively small and the pressure tube will heat up uniformly. When the pressure tube deforms, it vill balloon uniformly into contact with its calandria tube and will cool down by transferring its heat to the surrounding moderator (2). If the voiding rate is low during flow stratification, a relatively steep circumferential temperature gradient may develop on the pressure-tube vail. The pressure tube may deform preferentially at the top and rupture prior to contact with the surrounding calandria tube. The integrity of the pressure tube in these situations depends on the pressure-tube circumferential temperature gradient influenced by the coolant level, fuel pover, and the fuel channel internal pressure.
The apparatus for all three tests was essentially the same, with minor modifications from one to the other. Figure 1 shows the horizontal test section which consists of a 2.29-m-long segment of CANDU-type channel. At one end (inlet), "make-up" water enters the test section through a connection welded to the bottom of the pressure tube. The other end (outlet) is connected to a vertical pipe (25.3 mm ID) for steam to exit. Pressurized water from a boiler of a given flow rate passes through a network of piping and is injected into the test section. The vertical pipe at the test-section outlet is connected to a condenser and a surge tank pressurized with helium. The test section pressure is maintained at. the desired value (1.1 HPa) during the experiment by controlling the helium back pressure in the surge tank.
Computer codes such as CATHENA/GENHTP (3,4) (WNRE), SMARTT (5) (OH), and AMPTRACT (6) (CANDU Ops) are capable of predicting the thermal and mechanical responses of the pressure tube during stratified flow conditions. However, these codes need to be verified against experimental data before they can be depended on to analyze postulated reactor accident scenarios. FIGURE 1: Two series of experiments have been performed at WNRE to study the thermal-mechanical response o£ hot pressure tubes under different stratified flow conditions, as well as to provide verification data for the various computer codes. The first series consisted o£ four tests in which water was boiled off from the channel without being replenished. The results were reported earlier (7). In the second series of experiments, saturated water was injected into the heated channel at a constant rate to maintain a constant water level ("Make-Up water" test series). This paper describes and presents results
TEST SECTION OF THE PRESSURE TUBE CIRCUMFERENTIAL TEMPERATURE DISTRIBUTION EXPERIMENT
Inside the channel, 36 electric fuel element simulators (heaters) (each 2.3 m long) are grouped into three separate heater-element rings. These heaters, together with a centre tube, have the same crosssectional dimensions as a CANDU-type 37-element fuel bundle. Each heater was made of two concentric zirconium tubes (7). The inner tube is the heater filament and the outer tube is the heater sheath. The heater filament is electrically Insulated from 120
C N S 9th A N N U A L C O N F E R E N C E . 1988
the sheath by small alumina rings spaced 100 mm apart. During the experiment, each heater is pressurized with helium to prevent the sheath from collapsing or ballooning. K-type thermocouples were spot-welded to the outside surface of the pressure tube and to the heater sheaths. They are grouped in three rings at axial locations labelled Rl, R2 and R3, as shown in Figure 2. The locations (labels) of various thermocouples in each axial ring are shown in Figure 3. Thermocouples to measure heater sheath temperatures (T/C 12 to 17) were available in axial rings Rl and R3 only. Moreover, T/C 18, an R-type thermocouple used in Test 3 to measure the heater-filament temperature, was available only in axial ring R 3 . Three Linear Variable Differential Transformers (LVDT) were installed (Figure 2) to measure the movement of the calandria and pressure tube. LVDT 1 measured the absolute movement of the top of the calandria tube. LVDTs 2 and 3 measured the gap width of the annulus between the pressure and calandria tube at the top and bottom, respectively.
TEST PROCEDURE At the beginning of each experiment, the test section was filled with water at room temperature and pressurized to 1.1 MPa. The water surrounding the calandria tube was heated to 75°C to simulate the moderator water. The pressurized water in the test section was gradually raised to the saturation temperature using six bottom heaters in both the outer and middle heater-element rings. This procedure reduced the top-to-bottom temperature gradient across the test section. After the water in the test section had reached the saturation temperature, valves at the inlet and outlet were opened, and slightly subcooled make-up water was Injected into the test section. After the make-up water was injected into the test section and all temperatures had stabilized (1-2 hours), the power supply was shut off and electrical cables to the rest of the heaters were quickly connected. Once test-section power was reestablished, the experiment was initiated and the transient time was referenced from this moment.
RESULTS
H 1 --- ses
J
FIGURE 2: AXIAL LOCATIONS OF THERMOCOUPLE RINGS AND LVDT (DISTANCE IN mm)
20° /
320°
Table 1 summarizes the test parameters. All three tests were conducted at a pressure of 1.1 MPa. In any of the tests, the pressure-tube circumferential temperature and the heater-sheath temperature distributions along the test section were quite similar. Therefore, for each test, we will describe only the temperature distributions measured at one of the axial-ring locations. After the description of individual tests, their results will be compared with one another to determine the effects of heater power and "make-up" water flow rate (water level) on the pressure-tube circumferential temperature distribution and deformation.
CALANDRIA TUBE
TABLE 1: TEST PARAMETERS
280°.
Test Number
Pressure (MPa)
Power (kW>
Flow Rate (g's)
100°
1
1.1
39
2
1.1
84
26
3
1.1
39 - 60
12
8.2
140°
PRESSURE TUBE 180°
FIGURE 3: THERMOCOUPLE LOCATIONS ON THE PRESSURE TUBE AND INSIDE THE HEATED CHANNEL
The test section was immersed In a pool of water kept at 75"C. The annulus between the calandria tube and the pressure tube was purged with C 0 z before each test. A garter spring vas positioned around the midplane of the pressure tube.
CNS 9th ANNUAL CONFERENCE, 1988 121
Test 1 The power history is shown in Figure 4. From 94 to 717 s, the power was raised slowly to 39 kW. Thereafter, it was maintained at approximately 38.5 kW until 1111 s when some heaters started to fall. Power was switched off at 1545 s and the test was terminated. During the experiment, the make-up water flow rate was 8.2 g/s. Water in the channel Initially boiled off, but measured heater-sheath temperatures indicated that the channel stayed about 1/4 full after 800 s.
BOO
1000
1200
1*00
200
1000
40O
600
BOO
1000
1200
1400
1000
: (SECONDS)
FIGURE 4:
POWER HISTORY OF TEST 1 (F INDICATING HEATER FAILURE AND END OF EXPERIMENT)
Figure 5 shows the pressure-tube temperature transient measured at axial ring Rl. The top of the pressure tube remained at saturation until 105 s, at which time a sharp increase in temperature indicates it was exposed to dry steam, and a top-to-bottom temperature difference (AT) started to develop. While the upper part of the pressure tubp was being heated up r the pressure tube started to deform (Figure 6 ) . From 0 to 640 s, the gap between the top of the pressure tube and the calandria tube narrowed slovly as a result of bowing caused by thermal expansion of the pressure tube. At 640 s, the temperature at the top of the pressure tube reached 640°C and thermal creep became significant, resulting subsequently in a rapid straining (deformation) near the top. As the pressure tube deformed, it moved away from the heaters and created a larger tlov area between the pressure tube top and adjacent heaters. This large flow area enhanced steam cooling along the top of the pressure tube and reduced radiant heat flux from the heaters. As the gap between the pressure tube and calandria tube decreased, heat transfer to the calandria tube by radiation and conduction through the C02 annulus became more efficient. As the pressure tube continued to balloon, these effects caused the heat-up of the pressure tube to slow down. Eventually the temperature at the top reached a maximum of 7]5°C at 895 s, yielding a maximum AT of 531"C. After 895 s, the temperature at the top of the pressure tube declined. The decline in the pressure-tube temperature could not be due to contact with calandria tube as the pressure tube continued to balloon substantially after 896 s and no contact was observed at this location when the apparatus was dismantled.
The power history is shown in Figure 7. The power was ramped quickly at 0 s and was maintained at an average value of 84 kV until 250 s at which time some heaters failed. At 259 s, the pressure tube ruptured at the top near the mid-plane of the pressure tube. Power was turned off and the experiment was terminated afterwards. The pressure-tube rupture was caused by heater failure and this will be discussed later. The make-up water flow rate in this test was 26 g/s. In the early part of the experiment, water in the test section boiled off rapidly. However, measured heater-sheath temperatures indicated the channel stayed about 1/2 full after 100 s. 122 C N SflthA N N U A L C O N F E R E N C E . 1988
FIGURE 5:
0
MEASURED PRESSURE TUBE CIRCUMFERENTIAL TEMPERATURES AT AXIAL RING Rl IN TEST 1 (F INDICATING HEATER FAILURE AND END OF EXPERIMENT)
200
«00
600 BOO 1000 TIME (SECONDS)
1200
1400
1«OO
FIGURE 6:
MEASURED MOVEMENTS OF CALANDRIA AND PRESSURE TUBES IN TEST 1 (F INDICATING HEATER FAILURE AND END OF EXPERIMENT)
FIGUPE 7:
POWER HISTORY OF TEST 2 (F INDICATING HEATER FAILURE AND END OF EXPERIMENT)
Figure 8 shovs the pressure-tube temperature transient measured at axial ring R2. The top of the pressure tube remained at saturation until 40 s, at which time a sharp increase in temperature indicates
It was exposed to dry steam, The temperature at the top of the pressure tube (T/C 1) reached 849°C when the heaters failed. Prior to the heater failure, a large AT (665"C) between the t op and the bottom of Immediately after the pressure tube was observed heater failure, thermocouples near the pressure-tube top registered sharp temperatu re increases, indicating that the top o£ the pressu re tube was heated by a secondary heat source shortly before it ruptured.
900
iV
,J\
800 700 '. 600
/ 1
500 100
A , '
MO
^^-^L^^—"— —
JO0
''
i
1» -iL. t \V •
5-8-7 100 100
ISO
200
T1UE (SECONDS)
FIGURE 8:
MEASURED PRESSURE-TUBE CIRCUMFERENTIAL TEMPERATURES AT AXIAL RING R2 FOR TEST 2 (F INDICATING HEATER FAILURE AND END OF EXPERIMENT)
The movement and deformation of the pressure tube can be inferred from Figure 9. From 0 to 204 s, the reduction in gap width between the top of the pressure tube and the calandria tube was due to upward bowing caused by thermal expansion. At 204 s, the top of the pressure tube reached 765°C, at which time thermal creep became significant and the pressure tube started to deform rapidly. However, the gap width reduction did not level off prior to heater failure, indicating that there was no contact with the calandria tube during the pressure-tube deformation. The LVDT signals surged at the instant of pressure tube rupture because the sensing core was moved by the pressure inside the annulus between the calandria and pressure tubes.
The power history is shown in Figure 10. It is divided into three periods: 0 to 1610 s, 1611 to 2565 s, and 2566 to 3506 s. The maximum power reached in Periods 1, 2 and 3 were 39, 46 and 60 kW, respectively. At 3507 s, some heaters started to fall, resulting in localized heating of adjacent heaters and the pressure tube wall. At 3582 s, the pressure tube ruptured near its mid-plane and at an angle about 60 degrees to the vertical. Heatersheath temperatures measured at axial ring Rl are shown in Figure 11. During Period 1, heaters in the upper half of the channel were quickly exposed to dry steam as water in the channel boiled off. T/C 14 started being exposed at 610 s, but the centrepin (T/C 11) remained just covered for the rest of this period. It is therefore inferred that the channel vas about 1/2 full after 610 s and for the rest of Period 1. The temperatures of the exposed heaters levelled off towards the end of this period, with the heater sheath near the top (T/C 12) reaching a temperature of 8 0 0 T . In Period 2, when the power was increased from 39 to 46 kW, the heater temperatures rose and the centrepin (T/C 11) was exposed. However, T/C 15 remained covered for the entire Period 2. It is thus inferred that the channel became slightly less than half full during Period 2. At about 2050 s, the temperatures of the exposed heater sheaths started to decline slightly, implying that the heaters were cooled. Cooling of heaters was due to pressure-tube deformation, which will be discussed later. In Period 3, the power was further increased from 46 to 60 kW in steps, and the temperatures of the exposed heaters Increased. T/C 15 started to be uncovered at 3100 s, but T/C 16 and 17 remained covered for the rest of this period. This implies the water level had decreased to about 1/4 after 3100 s. The heater-filament temperature (T/C 18 in axial ring R3) is shown in Figure 12. Towards the end of Period 3, the filament reached a very high temperature of 1600°C, close to the melting point of Zircaloy-4 (1760'C).
0
SCO
1000
1900
2000
2500
3000
3300
WOO
TIME (SECONDS)
FIGURE 1 0 :
0
FIGURE 9:
50
100
150 TIUE (SECONDS)
200
250
POWER HISTORY OF TEST 3 (F INDICATING HEATER FAILURE AND END OF EXPERIMENT)
Mo
MEASURED MOVEMENTS OF CALANDRIA AND PRESSURE TUBES IN TEST 2 (F INDICATING HEATER FAILURE AND END OF EXPERIMENT)
CNS 9th ANNUAL CONFERENCE, 1988 123
PERIOD 2
PERIOD 1
1 i 1
PEWOO 3
I 1 k-
12 ;•'-''«
_ +^
" !
,-r
"A k r i )\ A
ir'
/--
t 1
\J \ \ I
it
.[
0
500
1000
|500
2000
2500
3000
3300
WOO
TIME (SECONDS)
FIGURE 11:
0
FIGURE 12:
MEASURED HEATER-SHEATH TEMPERATURES AT AXIAL RING Rl FOR TEST 3 (F INDICATING HEATER FAIUJRE AND END OF EXPERIMENT)
500
1000
1500 2000 2500 TIME (SECONDS)
3000
3900
4000
HEATER FILAMENT TEMPERATURE (T/C 18) IN TEST 3 (F INDICATING HEATER FAILURE AND END OF EXPERIMENT)
Fi^ute W sVwus the us essu.se-tube transient measured at axial ring R2. In Period 1, the upper half of the pressure tube was quickly exposed to dry steam. The temperatures rose quickly, but the increase started to slov down at about 500 s. The pressure-tube temperatures levelled off towards the end of this period with the pressure-tube top reaching a temperature of 628°C. The maximum AT developed in this period was 444°C. In the same period, Figure 14 shows that only the calandrla tube moved. The movement was due to upward bowing caused by thermal expansion of the pressure tube. The pressure tube hardly deformed as the gap-vidth between the pressure-tube top and calandria tube was reduced by only 0.2 mm in this period. Obviously, with 39 kW power and a half-full channel, a steady state had been established towards the end of this period. Heat generation in the fuel element simulators was balanced by heat removal in the boil-off of make-up water, steam convection, and radiation to the pressure tube and moderator, without any significant pressure- tube uetotmattott- Iw fecl
124
CNS 9th ANNUAL CONFERENCE, 1988
of the pressure tube as detected by the LVDTs (Figure 14). The ballooning of the pressure tube enhances the steam cooling as well as heat transfer to the calandria tube (similar to Test J)> resulting in a decline in temperature. The pressure-tube temparature measured by T/C 1 started to fluctuate. This fluctuation is probably caused by the thermocouple sensing wires coming in contact with the calandria tube, and therefore is an indication that the top of the pressure tube has come close to the calandria tube. In Period 3, the power was further increased to 60 kW in steps. The pressure-tube top temperature continued to decline because the pressure tube continued to balloon. However, the temperatures measured by T/C 9 and 3 increased and tended to level off towards the end of this period. Moreover, the temperatures at these two locations (away from the top) were higher than at the top of the pressure tube. As the top and bottom of the pressure tube were being cooled, the sides became the hottest locations and the circumferential temperature distribution changed as the water level decreased. After the heater failure (at the end of Period 3 ) , T/C 3 (near the pressure-tube rupture) measured a sharper rise in temperature than any other thermocouple, and is an indication that the rupture was caused by localized heating due to heater failure.
O
FIGURE 11;
500
1000
1500 2000 2500 HUE (SECONDS)
3000
3500
4000
WEASJJRED 5KSS.UB.E TUBE CIRCUHFEREKEIAL TEMPERATURES AT AXIAL RING R2 FOR TEST 3 (F INDICATING HEATER FAILURE AND END OF EXPERIMENT) PERIUO 2
TOM, VS&,
FIGURE 14:
-StfflO 1500 HUE (SECONDS)
3000
MOO
WDO
MEASURED MOVEMENTS OF CALANDRIA AND PRESSURE TUBES IN TEST 3 (F INDICATING HEATER FAILURE ANP END OF EXPERIMENT)
DISCUSSION In all three tests, prior to heater failure, a large top-to-bottom AT in excess of 400°C developed on the pressure tube. Pressure-tube deformation occurred in some instances, buc the pressure tube neither contacted the calandria tube nor ruptured. Pressure-tube circumferential temperature distribution curves for the three tests (Period 1 in Test 3 is used in comparison) are given in Figure 15. The curves shown correspond to the circumferential temperature distributions at the Instant when the temperature at the top of the pressure tube has reached a maximum. Test 1 and Test 3 were conducted at the same power, but different "make-up water" flow rates resulted in different water level in the channel. The final water level in Test 1 was about I/A vs 1/2 in Test 3 (Period 1 ) . A higher peak temperature (715°C) at the top O L the pressure tube (a larger AT) was observed in Test 1, and this was caused by more uncovered heaters at the lower water level. However, the circumferential temperature gradients are comparable in both cases. This shows that a lower water level results in a larger AT but not necessarily a steeper temperature gradient. Tests 2 and 3 were conducted at different power values but comparable final water levels (both about 1/2). A larger AT with a steeper circumferential temperature gradient was developed in Test 2 (higher power). In this case, although the amount of exposed heater surfaces are comparable, the heat flux from the heaters is higher In Test 2 because of higher power. A higher heater power results in both a larger AT and a sleeper circumferential temperature gradient.
vuv "^^
TEST 1 (Pw-M
LCT*I
I/I)
BOO \
U..,,:4,
TEST 1 {P»»3
700
600
\ V
SOO
.
TEST
rlod 1) U'fl 12)
Ing in no significant pressure-tube deformation. This result shows that, depending on heater power and water level (coolant flow rate), heat generated In heaters can be transferred to the moderator without significant pressure-tube deformation. Deformation of the pressure tube alfects the heat transfer to the pressure and calandria tube, and thus the circumferential temperature distribution. The deformation of the pressure tube at the top can result in a decline of the pressure-tube top temperature due to enhanced steam cooling and more efficient heat transfer to the moderator. The decline in pressure-tube top temperature slows down the pressure-tube deformation rate. These results show that the pressure-tube circumferential temperature distribution and the deformation are Interdependent when the pressure tube Is surrounded by the calandria tube and moderator. Therefore, both these quantities depend on their past history. All three experiments were terminated after heater failure. In Tests 2 and 3, failing heaters also caused the pressure tube to rupture. Figure 16 shows measurements of the current in Test 3 at each of the three heater-element rings shortly before pressuretube rupture. At 5307 s (at the end of Period 3 ) , the onset of heater failure was evident. The current surge in the outer heater-element ting indicates short circuits were formed by the hot heater filament (1600°C, Figure 12) sagging into contact with the heater sheath. At 3566 s ("b" In Figure 16), more short circuits were formed. The short circuit results in current being partly conducted through the segment(s) of heater sheath in contact with the hot heater filament. This heater-sheath segment heats up rapidly and becomes much hotter than neighbouring heater sheaths. The exothermic chemical reaction between the Zircaloy sheath and steam was enhanced and sustained at this local hot spot, causing the sheath and filament to melt. At 3572 s (marked " c " ) , the current in the outer ring dipped, Indicating that some heater filaments in the outer ring did melt. Consequently, the near-by pressure-tube wall was heated intensely but locally by the heater sheath being oxidized. The pressure tube strained locally and, under pressure, ruptured. Thus, pressure-tube failure was caused by unrepresentative intense local heating of the pressure-tube wall caused by heater failure. Similar phenomena were observed in Test 2.
«.
300 •
\
e
Period 3
OOO •
:
I
1
200
1
ft
1
000
1 OUTER RING
nn 0
20
40
80
60
100
120
140
160
100
1
000
]
1
THERMOCOUPLE LOCATION (d«gr«fli from th» top)
I
MIDDLE RING
FIGURE 15: MAXIMUM PRESSURE-TUBE CIRCUMFERENTIAL TEMPERATURE DISTRIBUTION IN TEST 1, 2 AND 3 (PERIOD 1)
1
400
L
INNER RING
200
fr.v The peak temperature at the top of the pressure tube is an important factor governing the deformation of the pressure tube. A steady state was reached towards the end of Period 1 in Test 3, where the pressure-tube temperatures levelled off. In this case, the heater power was low and the water level was high enough to yield a relatively low peak temperature on the top of the pressure tube, result-
1
2400
3425
FIGURE 1 6 :
3450
3475 3500 3525 TIME (SECONDS)
3350
3575
3000
DETAIL OF HEATER CURRENT IN TEST 3
CNS 9th ANNUAL CONFERENCE. 1988 12S
SUMMARY Three experiments have been conducted to investigate the pressure-tube circumferential temperature distribution and deformation at a constant internal pressure as a function of heater power and coolant level (coolant flov rate). Results show that: 1. A lower coolant level results in a larger topto-bottom temperature difference (AT) but not necessarily a steeper temperature gradient on the pressure-tube wall; 2. A higher heater power results in both a larger AT and steeper circumferential temperature gradient; 3. Depending on heater power and water level, heat generated in heaters can be transferred to the moderator without significant pressuretube deformation; 4. When surrounded by the moderator, the pressure-tube circumferential temperature distribution and deformation are mutually dependent, and therefore are dependent on their past history; 5. Pressure-tube failure observed in Test 2 and 3 was caused by unrepresentative local heating of the pressure-tube wall caused by heater failure. Data collected in these experiments will be used to verify computer codes like CATHENA/GENHTP (WNRE), SMARTT (OH) and AHPTRACT (CANDU Ops).
ACKNOWLEDGEMENTS The experiments reported in this paper were funded under the COG program. The authors acknovledge the technical contributions of the COG members. In particular, the technical discussions with K.E. Locke (Ontario Hydro), P. Thompson (New Brunswick Electric Power), R. Hu (Hydro Quebec) and E. Kohn (CANDU Operations) are much appreciated. REFERENCES (1) HOBSON, N., RICHARDS, D.J., HANNA, B.N., and GRANT, S.D., "Status of the ATHENA Thermalhydraulic Code for CANDU LOCA Analysis", 11th Annual CNS Symposium on Simulation of Reactor Dynamics and Plan Control, Kingston, 1985 April. (2) GILLESPIE, G.E., MOVER, R.G., and THOMPSON, P.D., "Moderator Boiling on the External Surface of Calandria Tube in a CANDU Reactor During a Lossof-Coolant Accident", Proceedings of the International Meeting on Thermal Nuclear Reactor Safety, Chicago, 111., 1982 August, (3) RICHARDS, D.J., HANNA, B.N., HOBSON, N., AND ARDRON, K.H., "CATHENA: A Two-Fluid Code for CANDU LOCA Analysis", Proceedings of Third International Topical Meeting on Reactor Thermal Hydraulic, Newport, Rhode Island, 1985 October. (4) MALLORY, J.P., YUEN, P.S., and SO, C.B., "GENHTP - A New Heat Transfer Package for CATHENA", CNS 14th Annual Nuclear Simulation Symposium, UNRE, 1988 April. (5) LOCKE, K.E., MUZUMDAR, A.P., LUXAT, J.C., SO, C.B., MOYER, R.G. and LITRE, D., "Progress on SMARTT Simulation of Pressure Tube Circumferential Temperature Distribution Experiments - Tests 126 CNS 9th ANNUAL CONFERENCE, 1988
1 to 4", 8th CNA/CNS Conference, St. John, 1987 June. (6) GULSHANI, P., and SO, C.B., "AMPTRi.CT: An Algebraic Model for Computing Pressure Tube Circumferential and Steam Tempera'.ure Transients Under Stratified Chanm-l Coolant Conditions", 12th Annual Symposium on Simulation of Reactor Dynamics and Plant Control, Hamilton, 1986 April. (7) SO, C.B., GILLESPIE, G.E., MOYER, R.G., and LITRE, D.G., "The Experimental Determination of Circumferential Temperature Distributions Developed in Pressure Tubes during Slow Coolant Boildown", Proceedings of 8th CNS Annual Conference, St. John, NB, 1987 June.
Session 4: Fuel Storage and Waste Management
Chairman: K. Nuttall, AECL WNRE
CNS 9th ANNUAL CONFERENCE, 198B
127
'/tig
CANADIAN EXPERIENCE WITH THE DRY STORAGE OF USED CANDU FUEL
by K.M. Wasywich Atomic Energy of Cannda Limited and C.R. Frost Ontario Hydro
ARSTRACT In 1974, Atomic Energy of Canada Limited (AECL) began a demonstration program to investigate the feasibility of dry storing used fuel in concrete canisters (CCs). Because of the success of that program, CCs have been licensed for the dry storage of used CANDU fuel in Canada. Today CCs are being used at three reactor sites to store about 27 000 (390 Mg U) of used CANDU fuel bundles. Most of the fuel being stored In the CCs has been previously cooled in water pools for periods greater than 10 years. The largest quantity of fuel In dry storage In Canada is at the site of the decommissioned Douglas Point reactor. The peak and average burnups of that fuel are about 832 and 667 GJ/kg U, respectively, and the peak fuel temperature In the CCs was about 125°C. The radiation dose experienced by the operators while loading the CCs was 0-25 mSv per canister. In 1978, AECL initiated a research program to investigate the long-term behaviour of used fuel (both undefected and defected) during dry storage. The program is funded jointly with Ontario Hydro, a provincial government electricity-generating utility, and is scheduled to continue beyond the year 2000. Intentionally defected fuel that has been stored for about five years in 150*C moisture-saturated air has been examined recently. The results of the examination revealed no significant degradation of the fuel or cladding. Ontario Hydro has also designed a multipurpose concrete integrated container (CIC), with a capacity for 384 CANDU fuel bundles, for the dry storage, transportation and, possibly, the final disposal of used fuel. Two CICs are scheduled to be loaded with used fuel In 1988 for demonstration purposes, and two more are planned for construction In 1989 for transportation tests.
INTRODUCTION II 1TM A large q u a n t i t y of used C A N D U " fuel (about 540 000 bundles or ~10 000 Mg U) is now stored in water pools in Canada. When the decay-heat output of the bundles has fallen sufficiently, Interim storage may be continued in a dry storage facility. In 1974, Atomic Energy of Canada Limited embarked on a demonstration program to investigate the feasibility of storing used fuel dry In concrete canisters. Four demonstration CCs were constructed at AECL's WhtteBhell Nuclear Research Establishment (WNRE) In Manitoba in 1974/75. Because of the success of the demonstration program, CCs have been licensed for the diy storage of used CANDU fuel In Canada.
At present, about 27 000 bundles (390 Mg U) are being stored In CCs at three reactor sites In Canada: one in Quebec, one In Ontario and one In Manitoba. Ontario Hydro (OH), a provincial government electricity-generating utility, has also designed a multipurpose concrete integrated container for the dry storage, transportation and, possibly, the final disposal of used fuel. In support of these initiatives, a jointly funded AECL/OH research program was established In 1978 to investigate the long-term behaviour of used fuel (both undefected and defected) during dry storage. That program is scheduled to continue beyond the year 2000. This report reviews the AECL concrete canister program for the dry storage of used fuel, the OH concrete Integrated container, and the joint AECL/OH experimental dry storage program. The observations and results from the experimental dry storage program are summarized for th« interim examinations performed on used CANDU fuel bundles that have been stored dry for periods of up to ~5 years.
THE AECL CONCRETE CANISTER PROGRAM The evolution of the AECL concrete canister program started with a demonstration phase at WNRE and continued with the adoption of CCs for the dry storage of used fuel at WNRE, and at the Gentilly-1 and Douglas Point Nuclear Generating Stations. This evolution Is discussed in this section; future plans for the dry storage of used fuel In CCs and some results from the operation and surveillance of CCs are also discussed. Concrete Canister Experience at WNRE Atomic Energy cf Canada Limited began investigating various dry storage concepts In the early 1970s. In 1974, The Committee Assessing Fuel Storage (CAFS) recommended that Canada develop an alternative to water-pool Btorage 1 . Although water-pool storage has been accepted worldwide and has been used In Canada since the inception of our nuclear program, the pools require maintenance, and generate secondary wastes, primarily from the water purification system. These problems would not exist In dry Btorage. In the spring of 1974, AECL began a development and demonstration program at WNRE for the concrete canister dry-storage concept. During 1974 and 1975, four demonstration CCs (two cylindrical and two square geometry) were constructed at WNRE. One CC of each design was equipped with an electric heater to simulate the decay heat produced by the used fuel. In 1975/76, the other two CCs were loaded with used fuel. The cylindrical CC was loaded with 138 bundles from the WNRE research reactor,
CNS 9th ANNUAL CONFERENCE, 1988 128
WR-1 „ and the square canister was loaded with 360 bundles from the Douglas Point Nuclear Generating Station. The fuel bundles in the demonstration CCs were stored in a helium atmosphere to prevent UO^ oxidation in case any fuel cladding failures were present. A d d i t i o n a l details are available In Reference 2. Because of the success of the demonstrat ion program, CCs were adopted for the dry storage of used fuel from W R - 1 , The square canister design was abandoned because It was more expensive to build and the higher fuel packing density that it provided was not needed at WNRR. The major design change to the cylindrical canister was the removal of the lead shielding around the inner steel liner that was used in the demonstration canisters. To compensate for this shielding loss, the thickness of the concrete was Increased by 150 mm. The CC design used to stote used fuel from WR-1 is referred to as a "productIon canister" at WNRE, and I s shown in Figure 1 • It provides two engineered harriers against the release of radioactive materials. The first barrier is a seal-welded fuel storage container, called a basket, that contains the fuel bundles. After the fuel Is loaded Into the basket, the basket lid is seal-welded In position, the welds are inspected, and the baskets are then purged and backfilled with helium to ensure the integrity of the first barrier. The second barrier is the canister liner. After the canister has been filled with loaded baskets, the canister lid is positioned and seal-welded to the liner. The cavity containing the storage baskets has sample lines that penetrate the biological shield. These permit regular sampling of the cavity to determine If moisture or radioactive material are present. Moisture would Indicate a leak In the liner, while above-background radioactivity would indicate whether any baskets containing defected fuel had been breached.
SAFEGUARD SEALS
TEMPOfiAHV HOLDDOUN PLATES BEFORE SEAL WELQ1NG L I D
UFTIKG LUG(TYP.)
•— REINFORCING STEEL
HEIGHT 5.3m O.D. 2.3 m HALL 0.76 m
WNRE Is licensed to store two types of WR-l fuel under the conditions shown in Table 1. The product Ion CCs are licensed to a thermal rating of U .4 kW, compared with 2 kW for the demonstration CCs, and can hold up to 6000 kg U, depending on the fuel enrichment. To date, 18 CCs have been constructed and one I s under construction at WNRE for the storage of about 1900 used fuel bundles (~23 Mg U ) . At present, only 11 canisters are being fully used to store used fuel. Of the remainder, two are the original demonstration CCs that contained electric heaters, three are used for experiments and two are still empty; these latter two canisters will be loaded In the near future with used fuel from the decommissioned WR-1 reactor. Used Fuel Storage Tn Concrete Canisters At Gentilly-1 When the AECL-owned 250 MWe Gentllly-1 (fl-1) prototype CANDU-Botllng Light Water (BLW) reactor, located at Gentllly In the Province of Quebec, was decommissioned In 1984, a decision was made to store Its used fuel In CCs. Eleven CCs were constructed In 1985 to store 3213 bundles (67 Mg U ) - This was the first application of dry storage technology at a commercial reactor site In Canada.
130 CNS 9th ANNUAL CONFERENCE, 1989
— 6 STEEL BASKETS
C0NCBE7E BASE
FIGURE 1: The ordinary concrete surrounding the liner provides the biological shieldThe reinforcing steel within the concrete is designed to limit thermal and mechanical stresses in the concrete, and to control cracking of the biological shield to acceptable limits.
-
WNRE Production Concrete Cantster Used for the Dry Storage of Used Fuel From WR-1
The 103-mm diameter x 500-mm long G-l fuel bundle consists of 1
TABLE 1 APPROVED PARAMETER LIMITS FOR CONCRETE CAHISTER STORAGE OF USED FUEL AT WHRE'
PARAMETER
STANDARD WELDED BASKET
UC
FUEL TYPE Physical Form
Enrichment Burnup
Cooling Time Basket Loading
(1)
18-eleraent standard WR-1 bundles
(1)
14-element standard WR-1 bundles
(2)
Loose material in sealed stainless steel cans
(2)
Loose material in sealed stainless steel cans
S2.4 vt.% " » U in total U 477 GJ/kg U nominal
S2.25 vt.%
>SJ
U in total U
997 GJ/kg U maximum 745 GJ/kg U average
6 months minimum
IB months minimum
(1)
Maximum of 24 bundles, or cans containing bundles , in a single ring
Maximum of 24 bundles, or cans containing bundles, in a single ring
(2)
37 bundles in a double ring; 24 in outer ring with enrichment of si.8 wt.X 1 J 5 U , 13 in inner ring with enrichment of si.2 vt.% assy
Canister Loading
<6 baskets
<6 baskets
Decay Heat Loading per Canister
4.4 kW maximum
4.4 kW maximum
Used Fuel Storage Tn Concrete Canisters At Douglas Point
baskets in air, provided that the bundles have been dried thoroughly to maintain long-term storage Integrity-
Because of the success of the G-l CC program, the concept was adopted to store 22 236 used fuel bundles (300 Mg H) from AKCL's Douglas Point Nuclear Generating Station (DPNGS), a prototype 200 MWe CANTHJ-Pressurized Heavy Water (PHW) reactor located in Ontario. The reactor was shut down permanently in May, 1984. Forty-seven CCs (46 for used fuel storage plus one spare) were constructed at the site In 1986 in an area adjacent to the Turbine Building (Figure 2 ) . Concrete Canister Operation and Surveillance The DPNGS fuel bundle is smaller in diameter, but has one more fuel element than the G-l bundle. The bundle, which is 82 mm in diameter and 500 mm long, has 19, Zircaloy-4 clad, natural UOj fuel elements arranged in two concentric rings of 12 and 6 elements around a central element. The bundles had been irradiated for periods up to 297 KFPD, and achieved maximum and average burnups of 830 and 667 GJ/kg II, respectively. Most of the fuel bundles had cooled in a water pool for 3 to 20 years. The research done at WNRE indicates that the temperatures reached tn the basket, ~125°C, are acceptable for storing the
Visual Inspections, radiation field measurements, fuel bundle temperature measurements and canister air monitoring are performed routinely to monitor the operation of the CCs. These procedures are discussed in this section. Visual TnBpections. Hairline cracks In the concrete were First observed fn the electrically heated demonstration CCh at WNRF. at heat loads from 1.5 to ?..O kW. As the heat lo^d was increased to 3.5 kW, rhe cracks propagated and multiplied. Above 3.5 kW,
CNS 9th ANNUAL CONFERENCE, 1988 T31
GAWTtjV C E FOR LOADING
CONCRETE CANISTERS
CANISTER PLUG
FIGURE 2:
The Douglas Point Concrete Canister Site
no additional cracking was observed, but the crack width Increased with temperature, reaching 0.1 to 0.3 ram in some places at heat loads of 10 kW. The cracks closed as the heat load was reduced. Crackdepth measurements made by low-frequenry ultrasonic techniques indicated that the deepest cracks penetrated to about one third the thickness of the concrete. Quarterly visual inspections of the fuelled CCs at WNRE have revealed that no significant cracking or spallation of the concrete has occured to date- No unusual degradation of the exposed metallic portions of the CCs has been observed. Radiation-Field Measurements. Radiation-field measurements are made after the basket has been loaded with fuel and during loading of the baskettransfer flask and canister. Measurements are made regularly after the canister is filled with used fuel. At WNRE, radiation-field measurements are made on contact with the CC surface at a position about 2 m above the ground (the region with the highest radiation field in a fully loaded C C ) , and at 2 m from the canister surface at the same height. Radiation dosimeters located at the canister storage-area fence boundary are read every 6 months. The air between the storage baskets and the Inner liner of the CC is monitored to determine if any radioactive material has been released from the fuel storage baskets. At WNRE, the surface groundwater from a drainage ditch surrounding the CC compound is sampled periodically ~nd is analysed for radioactive contaminationSub-surface groundwater taken from deep wells near the fence boundary Is also analysed periodically for radioactive contamination. The on-contact radiation field on the outside of a freshly loaded used-fuel storase basket is typically 10 to 20 Sv/h. At WNRE, the peak on-rontact radiation field was about 0.1 mSv/h on tht outside of a freshly loaded demonstration canister. This dropped to less than 0.01 mSv/h after 10 years of storage. After loading, the peak on-contact radiation fields were 0.005 and 0.064 mSv/h for the G-l and DPNGS CCs,
132 CNS 9th ANNUAL CONFERENCE, 1988
respectively. The estimated total radiation dose received by all the operators during the CC loading operations at WNRE was less than 0.5 mSv per canister. The corresponding doses were 0.63 and 0.25 mSv per canister at G-l and DPNGS, respectively. To date, no radioactive material has been detected in either the canister Internal air space (WNRE, G-l, DPNGS) or in the surface or subsurface groundwater (measured only at WNRE). Temperature Measurements. Fuel-bundle temperatures are monitored periodically at WNRE In the demonstration CCs and continuously in fuel storage experiments. When the demonstration CC containing 360 DPNGS bundles was first loaded, the total decayheat output was 1.75 kW. At that time, the maximum fuel-bundle temperature was 1 1 0 D C After 10 years of storage, this dropped to 66°C. The calculated peak fuel-bundle temperatures were 125°C and <6G°C In the freshly loaded DPNGS and G-l CCs, respectively. Future Plans Additional CCs are scheduled to be built In 1988 to store about 5600 used fuel bundles (75 Mg U) from Canada's first demonstration CANDU power reactor (Nuclear Power Demonstration - N P D ) , which was shut down in 1987. New Brunswick Power, a provincial government electricity-generating utility, Is also considering using CCs to provide future used fuel storage capacity at its 600-MVe Point Lepreau nuclear generating station. In 1984, AECL signed ?n agreement with Nuclear Packaging Inc. (NuPac) of Federal Way, Washington, to provide that company with the sole licence to market AECL's concrete-canister technology for the storage of Light Water Reactor (LWR) fuel in the United States. Under this agreement, NuPac provides the engineering and marketing expertise and AECL supplies technical advice and assistance In CC development activities.
ONTARIO HYDRO'S CONCRETE INTEGRATED CONTAINER The concrete integrated container Is similar to the AECL concrete canister, but is designed to be a universal container that can be used for the storage, transportation and, possibly, disposal of Ontario Hydro 1 s used fuel. This concept is referred to as an "integrated system," and has been under development since 1981 6 * 7 . The current CIC design (Figure 3) is based on the following specifications: it will contain 384 used CANDU fuel bundles, can be loaded under water, has inner and outer steel liners for ease of decontamination, has separate storage and transportation closures, and is constructed from a durable highdensity concrete, which provides good radiation shielding. A two-stage CIC demonstration program has been initiated. Two CICs are scheduled to be built and loaded in 1988 in a used fuel storage demonstration at the Pickering nuclear generating site; two more are planned for construction In 1989 for transportation tests.
METALLIC SCAL
ITOHAOJLPD I
LID LIFTING | LUGS
BOLTS FOR TRANSPORTATION LJO
FIGURE 3:
The Ontario Hydro Concrete Integrated Container
LONG-TERM USED FUEL DRY-STORAGE RESEARCH PROGRAM In 1978, AECL Initiated a research program, funded jointly with Ontario Hydro, to investigate the longterm behaviour of used fuel (both undetected and defected) during dry storage. The program Involves three used fuel Btorage experiments", and Is scheduled to continue beyond the year 2000. The experiments consist of: the Easily Retrievable Basket (ERB) Experiment (dry storage In a if at seasonally varying temperatures), and the Controlled Environment Experiment » Phase 1 (CEX-1, storage In dry air at 150°C)
and Phase 2 (CEX-2, storage In moisture saturated air at 150°C). The ERB experiment involves the storage of two typical undefected used fuel bundles from Ontario Hydro's Pickering Nuclear Generating Station. Each phase of the CEX experiment involves the storage of eight bundles, four from each of Ontario Hydro's Pickering and Bruce nuclear generating stations (see Table 2 ) . In each phase of the CEX experiments, all the outer elements In two of the Pickering and two of the Bruce bundles were Intentionally defected before storage by drilling a 3-mm diameter hole through the cladding of each element. The objective Is to compare the relative storage behaviour of undefected and defected elements 8 . Although the storage temperature in a large dry-storage facility containing >10-yearold CANDU fuel bundles Is expected to be <100"C, a storage temperature of 150°C was chosen for the CEX tests to accelerate any detrimental effects that might occur. A moist environment is being used in the CEX-2 test to simulate a situation in which water might be transferred on the bundle and/or container surfaces, or within defected elements, Into the storage canister. To date, results from this program have Indicated that undefected and intentionally defected CANDU used fuel bundles can be safely stored in dry 150°C air for more than 41 months, and for up to 58 months In moisture-saturated 150°C air, without any loss of integritySome U 0 2 oxidation was observed in fuel samples from intentionally defected elements recovered for examination from both the dry and moist-storage tests. The degree of oxidation experienced by the individual UO2 grains was greater for fuel that had been stored in dry 150°C air than for fuel stored at 150°C in moisture-saturated air. The extent of oxidation in the dry-air experiment (CEX-l) appears to match closely one of the Arrhenius plots for dry-air UOj oxidation that was developed at AECL's Chalk River Nuclear Laboratories (CRNL) 9 . It has been predicted that a defected CANDU fuel element > 10 years out of reactor would not experience cracking of the cladding due to U0-, oxidation for at least 1000 years if stored continuously in dry air at 100aC10. Significantly more fuel experienced grain boundary oxidation in the moisture-saturated I50°C air than in the dry 150"C air. However, this has caused no loss of fuel element integrity after almost 5 years of storage. The moisture-saturated environment of the CEX-2 experiment is a much more extreme environment than that anticipated to prevail in any of AECL's CCr in Ontario Hydro's CIC. At this time, It is difficult Co make long-term predictions about the long-term Integrity of defected elements stored in moisture-saturated air at 150 D C since the oxidation of U0 2 in that environment appears to proceed by a different mechanism than in dry air. To ensure that the integrity of the used fuel bundles can be maintained for storage periods of 50 years or more, it Is recommended that the used fuel bundles and their containers be dried as thoroughly as possible prior to dry storage 1 0 . It Is planned that one used fuel bundle be retrieved for subsequent examination each year. In 1988, an Intentionally defected intermediate-flux (peak outer element linear power rating of 45 kW/m) fuel bundle will be retrieved and examined after being stored for ~70 months in air saturated with moisture at 15O°C.
CNS 9th ANNUAL CONFERENCE, 1988 133
FUEL BUNDLES USED IN THE DRY-STORAGE RESEARCH PROGRAM
Serial No.
Features
Discharge Date from Reactor
Average Bundle Burnup (GJ/kg U)
Average OuterElement
Burnup (GJ/kg U)
Peak OuterDecay Heat Eleraent e Start of Linear Dry Storage Rating (kW/m) IK)
Cooling Tun Before Dry Storage (a)
CEX-1 BRUCE BUNDLES (CANLUB) E04313C Plenum E04323C* Plenum F06584C non-Plenum F06605C* non-Plenum
1977 1977 1977 1977
Dec Dec Dec Dec
11147W 11I71W* 19558C 26244C*
CANLUB CANLUB non-CANLUB non-CANLUB
1973 1973 1975 1973
Dec Dec Jul Dec
F17267C* G18126H F18910C* G01355H
High-flux High-flux Low-flux Low-flux
1979 1980 1977 1980
Apr Jun Dec Jun
A08336H* CAKLBB A08313W CANLUB 3OO53C* non-CANLUB B00760W non-CANLUB
1976 1976 1973 1975
Dec Dec Oct Jun
863 867 675 614
1975 Dec 1975 Dec
625 636
672 672 780 766
607 607 704 690
33 33 39 39
20.7 20.7 23.4 23.0
2.8 2.8 ".8 2.8
48 50 43 48
6.6 6.6 7.8 7.1
6.8 6.8 5.2 6.8
53 48 22 22
27.5 <8.1 13.3 43.0
2.5 1.5 3.9 1.5
45 46
9.8 9.8 5.4 6.2
4.9 4.9 8.8 7.0
CEX-1 PICKERING BUNDLES 737 737 754 802
675 675 693 73 7
CEX-2 BRUCE BUNDLES (CANLUB) 863 841 654 704
775 7SS 589 636
CEX-2 PICKERING BUNDLES 939 946 733 668
•43 •43
ERB PICKERING BUNDLES A01789H
A01790W
CANLUB CANLUB
682 693
43
22.3 22.6
2.8 2.8
* All outer elements of these bundles were defected intentionally, except for one which was designated as a "control." NOTE:
The CEX-1 bundles were loaded into the canister in 1980 October. The CEX-2 bundles, 3OO53C and BO076OW, were loaded into the canister in 1982 June, and the remaining CEX-2 bundles were loaded into the canister in 1901 November. The ERB bundles were loaded into the canister in 1978 October.
REFERENCES 1.
W.W. Morgan ( e d i t o r ) , "Report by the Committee Assessing Fuel Storage; Part 1, Summary; Part 2 , Appendices," Atomic Energy of Canada Limited Report, AECL-5959/1 and /2 (1977).
2.
M.M. Ohta, "The Concrete C a n i s t e r Program," Atomic Energy of Canada Limited Report, AFXL-5%5 ( 1 9 7 8 ) .
3.
D.W. P a t t e r s o n and D. See Hoye, "Canadian Experience S t o r i n g I r r a d i a t e d CANDU Fuel In Conc r e t e C a n i s t e r s , " Presented a t the Third I n t e r n a t i o n a l Spent Fuel Storage Technology Symposium/Workshop, S e a t t l e , Washington, USA, April 8-10, 1986.
4.
I.M. Grant and W. Dicks, "On-Stte Dry Concrete C a n i s t e r Storage of Spent Fuel a t G e n t t l l y - 1 , " P r e s e n t e d a t the American Nuclear Society SPECTRUM '86 Meeting, Niagara F a l l s , New York, (ISA, September 16-18, 1986.
5.
P. Pattantyus and R. Beaudoln, "Commercial Applications of Dry Storage Technology," To: " I n t e r n a t i o n a l Conference on CANDU Fuel," Ch^Tk River, Ontario, Canada, October 6-8, 1986, Canadian Nuclear Society, Toronto.
6.
J . Freire-Canosa, S.J. Naqvl, G.S. Kelly and G.A. Mentes, "A Concrete Cask Design for Stor-
134 CNS 9th ANNUAL CONFERENCE, 1988
age, Transporcatlon and D i s p o s a l , " In: "Proceedings of the Second Internatlonal Conference on Radioactive Waste Management," Winnipeg, Manitoba, Canada, September 7-11, 1986, Canadian Nuclear Society, Toronto 7.
R.D. Hootan and N.C. Burnett, "Development of a Highly Durable Concrete C o n t a i n e r for the Storage, Transport and Disposal of I r r a d i a t e d F u e l , " Presented at the ACI Symposium on Cement and Cementltlous Materials for Radioactive Waste Management, New York, November 1, 1984.
8.
K.M. Wasywlch, J.D. Chen, K.I. Burns and D.G. Boase, "The Characterization of I r r a d i a t e d CANDU Fuel Bundles Stored In Concrete Canisters at WNRE," T_n_: " I n t e r n a t i o n a l Conference on Radioa c t i v e Waste Management," Winnipeg, Manitoba, Canada, September 12-15, 1982, Canadian Nuclear Society, Toronto.
9.
I.J. Hastings, R.J. Chenier, J.R. Kelm, E. Mizzan, J . Novak, D.H. Rose and A.M. Ross, Nucl. Tech. 1985. 68_, 40.
10.
C.R. Frost and K. M. Wasywlch, "UO2 Oxidation In Air and the Implications for Dry Storage of I r r a d i a t e d CANDU Fuel Bundles," Presented at the Workshop on Chemical Reactivity of Oxide Fuel and Fission Product Release at Berkeley, April 7-9, 1987.
THE DOUGLAS POINT DRY SPENT FUEL STORAGE FACILITY
ROBERT R. BEAUDOIN Atomic Energy of Canada Limited Montreal, Quebec, H3B 2V6, Canada AND DONALD W. PATTERSON Atomic Energy of Canada Limited WMteshell Nuclear Research Establishment Pinawa, Manitoba, ROE 1L0, Canada
ABSTRACT This Paper reviews AECL's experience with dry concrete canisters over the last 12 years and describes the storage of 388 MgU of spent fuel. The Canadian experience includes the storage of 23 MgU of both natural and enriched fuel at the Whiceshell Nuclear Research Establishment, the storage of 67 MgU of spent fuel from the decommissioned Gentilly-1 reactor, and the storage of 298 MgU of spent fuel from the decommissioned Douglas Point reactor. The storage of Douglas Point which represents the largest dry spent fuel storage canister facility built to date, is highlighted to describe its state-of-the-art licensing, safety, health physics and project management aspects. The Paper includes a general description of the storage site, the canister, the storage basket, the fuel handling equipment and the fuel handling operations. The project schedule is reviewed and compared with the actual dates achieved. The equipment, labor and total project costs are discussed. Two ongoing projects are described. The decommissioned Nuclear Power Demonstration (NPD) Plant, w h e n 75 MgU of spent fuel will be stored, and Point. Lepreau, where storage of spent fuel from on-going operations, at a pace of approximately 100 MgU per year will take place, starting in 1990. The status of canister storage for LWR reactor spent fuel is outlined. Future trends in the technology of dry spent fuel storage in concrete canisters are discussed. INTRODUCTION During the past few years, very important developments have taken place in Canada and U.S. in the field of dry spent fuel storage in concrete canisters. This paper will review the storage projects recently undertaken, with special emphasis placed on Douglas Point, the world's largest canister program, which was completed in 1987. Two other storage projects presently underway at the NPD and Point Lepreau stations are also described. An outline of the technology for LWR dry spent fuel storage in concrete canisters is included. AECL has already stored 388 MgU of spent fuel in canisters and the additional storage of more than 2,800 MgU in canis-
ters is underway as follows :
PROJECT
STATUS
QUANTITY OF CANISTERS
TONNAGE (MgU)
WNRE
Operating (1975)
11
23
Gentilly-1
Operating (1985)
11
67
Douglas Point Operating (1987)
47
298
NPD
13
75
Ongoing (1989)
Point Lepreau Ongoing (1990) Pickering A
Ongoing (88-89)
LWR (INEL)
Scheduled (88-89)
273*
4
T.B.D.**
2,790*
29
T.B.D.**
* Over a 30 year period ** To be determined EARLY CANISTER DEVELOPMENTS In both Canada and abroad, spent fuel has been stored mainly in spent fuel bays since the advent of nuclear power. Since 1975 the Whiceshell Nuclear Research Establishment (WNRE) has designed and tested several models of canisters for the dry storage of enriched uranium oxide fuel, enriched uranium carbide fuel, and natural uranium oxide fuel. The current canister inventory in the WNRE Waste Management Area consists of 17 canisters. 10 of these canisters contain enriched fuel, one contains natural uranium fuel and the other 6 are reserved for experiments or for future storage of fuel (see Figures la, lb and 2 a ) • The production canisters which are licensed for a maximal thermal rating of 4.4 lew can hold up to 6 MgU of spent fuel, depending on the fuel enrichment. Since the start of the program, WNRE has placed 23 MgU of spent fuel into storage. THE GENTILLY-1 STORAGE PROGRAM At the time the Gentilly-1 prototype CANDU Boiling Lfght Water reactor was being decomniis-
CNS 9th ANNUAL CONFERENCE, 1988 135
sioned in 1984, a decision was made to store the natural uranium spent fuel in concrete canisters because this was found to be the most economical approach for continued storage. The Gentilly-1 spent fuel had only been in service for the equivalent of 183 effective full-power-days and by 1984 had benefited from a minimum of 7 years of cooling. The decay heat and the radiation field conditions were thus low in comparison to the conditions that are typical at WNRE. The Gentilly-1 program stored 67 MgU of spent fuel in 11 canisters. Each canister contains 8 baskets (2 more than the WNRE canister) for a total holding capacity of 6.4 MgU of spent fuel per canister. The Gentilly-1 canisters are 2.6 m in diameter and 6.0 m high (see Figures 1-c and 2-b). Each basket holds 38 bundles, stored vertically, for a total of 304 bundles per canister. Unlike other CANDU fuel» the Gentilly-1 fuel has a central hole that accommodated a central structural tube when the fuel was in the reactor. This feature was exploited during the fuel handling operations and then was used for positioning the fuel in the basket, with vertical pins. The canisters are located indoors in an auxiliary wing of the decommissioned Gentilly-1 Turbine Building, in an area which had previously housed the standby power generators. Due to the low burn-up and long cooling period, only a minor fraction (10% to 15^) of the licensed canister heat release capability has been usedThe fuel heat release per canister averaged only 425 Watts and the calculated fuel temperature reached was only 20 °C to 25°C above ambient temperature (see Table 1 ) . TABLE 1 GENTILLY-1 & DOUGLAS POINT STORAGE PROGRAMS TEMPERATURE DATA FOR CANISTERS (°C)
LOCATION
DP G-l DP (calcu- (calcu- (mealated) lated) sured)
Daily Average Ambient Air
30
30
30
Canister Surface
36
46
32
Canister Liner
42
79
51
Basket Wall
47
101
67
*
Fuel Temperature
52 124 * extrapolated
84
*
Douglas Point, comparing continuing storage of fuel in the spent fuel bay and dry storage in canisters. The study showed canister storage to be less expensive than continued operation of a bay, for a site which is shut down, because of the need to keep services for the S.F.B. operation. A quantity of 22,256 bundles, representing 29S MgU of spent fuel is present ly stored in 47 canisters (one Is an empty spare canister), on a site adjacent to the Turbine Building. Each canister contains 486 bundles, representing 6.5 MgU of spent fuel. General
Methodology
for
Spent
Fuel
Storage
in
• l ". a . n A i ?. t - e - r 5 . Dry storage in canisters is based on placing baskets filled with spent fuel into pre-built concrete canisters» located either on or away from the power plant site. The fuel is first transferred (underwater) from the storage tra/s into specially designed baskets. Each basket is filled, inspected, closed with a cover, and lifted into a shielded work station located alongside the bay. The fuel in the basket is dried and then the basket base is seal-welded to the basket cover. The basket Is then lifted into a transport flask which mounts on the top of the shielded work station. The flask is transported to the canister site, using a suitable truck. A gantry crane is used to lift the flask to the elevation of the top of the canister. The canister plug is removed, the flask placed over the canister opening and the basket is lowered into the canister. These operations in reverse are undertaken to get the flask back to the shielded work station. The cycle is repeated until the canister Is full. The plug is then welded to the canister liner. An elevation view of a Douglas Point canister is shown in Figure 1-d. Safety and Licensing Considerations The primary Safety and Licensing which must be satisfactorily addressed spent fuel storage facility are: -
-
concerns in a dry
adequate fuel cooling, effective containment of radioactive inventory both during handling operations and during long term storage, adequate radiation shielding, adequate physical security and compliance with IAEA safeguards requirements, long term structural integrity.
All of the above considerations are factored into the design of the st orage elements and the fuel handling equipment, as described in the paragraphs that follow.
THE DOUGLAS POINT STORAGE PROGRAM
Basket and Canister Design
The Douglas Point station was permanently shut
The Douglas Point fuel bundle has a diameter of 82.5 mm, whereas the GentiUy~l fuel hae a diameter of 102.5 mm. This allowed more Douglas Point bundles to be stored per basket.
136 CfsIS 9th A N N U A L CONFERENCE, 1988
Various internal layouts were considered and a geometry having 54 vertically stored bundles per basket was retained. The Douglas Point basket is made of stainless steel and has the same external
4.8 MgU 2.4 m s 5.3 m (SQUARE CANISTER)
1.9 MgU 2.3 m x 4.9 in
2
i WNRE NATURAL FUEL 1 IN OPERATION
WNRE ENRICHED FUEL 10 IN OPERATION
OENTILLY-1 11 IN OPERATION
222
7.3 HgU 2.6 m X 3.6 In (HEAVY CONCRETE)
»*#3.I m i 6.1 m,/,.
1 DOUGLAS POINT t N.P.D. 46 IN OPERATION * 13 COMMITTED
POINT LEPREAU 273 COMMITTED OVER 30 TEARS
CONCRETE INTEGRATED CONTAINER PICKERING A 4 COMMITTED FlQ. 1 - I
4.0 MgU 2.6 m x 5.5 m
B:-":::
P.W.R. CASK DRY LOADING
P.W.R. CASK WET LOADING
CANISTER MODELS USED FOR DRY SPENT FUEL STORAGE
CNS 9th ANNUAL CONFERENCE, 1988
137
dimensions as the Gentilly-1 basket. By using a similar size basket, the same canister as Centilly-1 could be used again. However, at Douglas Point the canister height was (slightly) increased to 6.1 m and the available space used to store a ninth basket. Seismic, thermal and basket drop stress calculations were performed to show this simple modification to be acceptable. A reduction of six canisters was thus achieved. The fuel to be stored contained some three year cooled fuel (with higher gamma activity) and a larger quantity of fuel having cooling periods exceeding 10 years. This afforded the opportunity of providing for extra shielding within the basket itself by having an outer shielding ring of 24 fuel bundles of 10 years, or more, of cooling. The remaining 30 bundles, located in the inner rings positions, are fuel with 3 to 10 years of cooling. Canister Site_ Design A total quantity of 47 canisters (46 used, plus one spare) are located on the site. The spare canister is intended to facilitate loading operations and incident recovery, to permit leak testing of baskets or future canister testing, and for extra storage space should this become necessary. The canisters are arranged in four rows of 12 and are serviced by a gantry crane. The canisters are spaced approximately one meter apart and 12 base sla^s of reinforced concrete 7.3 m by 7.3 m (24 ft x 24 ft) and 61 cm (2 ft) thick serve to support 4 canisters each* The full canister array is only 14.6 m by 44 m (48 ft by 144 f t ) . The fenced area is of 57 m by 68 m (188 ft by 224 f t ) . The canister site is located outdoors, adjacent to the east end of the Turbine Building wall. The canisters were constructed using normal Portland cement for the concrete mix. A 5% air entralnment for enhanced freeze-thaw cycle resistance was specified. A 28-day compressive strength of 27.6 MPa is expected from such a mix. 10 of the 47 canisters were constructed using Silica Fume additives for improved strength. Two canisters (one normal concrete and the other Silica Fume) were equipped with a total of 34 embedded thermocouples and 5 embedded quarterbridge strain gauges. These served to measure the temperatures and rebar stresses reached after concrete pouring (due to the heat of hydration), and also after fuel loading operations. Fuel Handling Operations The Douglas Point fuel handling (F/H) operations were very similar to those of Gentilly-1. The spent fuel had been stored in trays, each holding 11 bundles. The early designs of fuel bund le s in the 60' s we re known to be bri 11 le and a quant ity of a few hundred loose "pencils" (detached from the main bundle but with sheath intact) were known to be present in the trays. This required very careful F/H operations, which were performed manually for the 22,256 bundles. The tray radiation levels were checked to ascertain the age of the fuel for proper basket positioning. The whole tray was then tilted to the vertical position. The fuel bundles were then
138 CNS 9th ANNUAL CONFERENCE, 19B8
moved with the aid of a bundle grappling tool y and placed into the outer ring or into the inner rings of the basket, depending on their cooling period. After the basket was filled, a basket cover was positioned on the basket base. The complete basket was then temporarily stored in the spent fuel bay until all baskets were loaded. The loose fuel pencils were placed in small perforated containers, approximately the size of a bundlet and each container stored in a basket just like a normal bundle. These operations were carried out by a crew of 6 (including a health physicist) and lasted from February to July 1987. No specific difficulties were encountered, except that one of the 423,000 pencils was sheared by the tilting table after it became loose from a bundle. Following the basket-filling phase, the basket welding and canister loading phases were undertaken. It would have been possible to undertake these operations in parallel to basket filling to accelerate the project, but it was decided not to. It was decided (as at Gentilly-J) to perform the canister loading after all of the baskets had been filled, to simplify operations, to reduce the congestion around the spent fuel bay, and to make optimum use of the highly skilled crew. A crew of 8 was used to perform the fuel drying, basket sealing, flask loading, transportation and canister loading operations. As described in section 4.1, all fuel drying and welding operations were performed in a specially designed, shielded work station and the fuel drying process was accomplished by blowing hot air into the basket and evacuating it by means of the active air discharge system; 20 to 30 minutes were required for basket drying. The basket welding was accomplished by two welds, made sequentially, by two remotely operated welding torches, located inside the shielded work station; 10 to 15 minutes were required to seal the basket. Few difficulties were encountered at this phase except for some spurious blockage of the welding wire caused by a spool of 'dirty' weld wire. The basket was transported from the shielded work station to the canister site in a siterestricted shielded flask, carried on a truck. At Douglas Point, the Gentilly-1 shielded flask was used together with a second. Identical flask, purchased to speed up the operations. The F/H speed improved from 3 baskets per shift at Gentilly-1 to 5-7 baskets per shift at Douglas Point. Few difficulties were encountered at this stage, except for periods when operations were delayed by (extremely) bad weather. Thermal and Shielding Considerations The maximum expected heat release (in 1987) from a Douglas Point canister was 2034 Watts. This represents about 45% of the licensed heat release limit of the WNRE canisters. This will decrease to below 15 00 Watts within 5 years. Detailed thermal calculations based on a windless > hot summer day, having an average ambient temperature over 24 hours of 30°C, determined an
Whlteshell Nuclear Research Establishment 13 Canisters Containing 17 MgU 1 976-1 988 Fig 2-a
GFNTII.LY -1 11 Canisters Containing 67 MgU 1 985 Fig 2-b
DOUGLAS POINT 46 Canisters Containing 298 MgU 1987 Fig 2-c CNS 9th ANNUAL CONFERENCE. 1988 139
expected maximum fuel temperature of 124°C In an average basket (see Table 1 ) . Following the loading of Instrumented canisters, actual measurements made of the temperature difference across the concrete Indicated readings of typically only 20°C (instead of 32°C). Extrapolations to obtain a maximum fuel temperature Indicated an actual value of only 84 °C would be reached with a 30°C ambient temperature, instead of 124°C. This Indicated there Is some conservatism in the thermal modeling and bundle heat release assumptions. Most of this conservatism Is attributable to the detailed spent fuel heat release calculations for the period of 3 to 20 years, the burn-up characteristics, the exact reactor history and the last core burn-up credit. Calculated and measured temperature data is given in Table 1. Shielding calculations, which included neutron dose rates, were also made to evaluate the dose rates which may be reached at various stages of the fuel handling sequence, for recently discharged fuel and older fuel. The neutron fields were found to be sufficiently high to require neutron shielding material on most leaded shielding structures. The concrete Itself has sufficient neutron shielding for the canister. Special Bubfa le Technology neutron dosimeters were used for workers' protection. Several improvements to the shielding system were made over Gentilly-1. The main one was the removal of the "L" shaped entrance opening- for the basket holding mechanism and the substitution of a totally-enclosed mechanism within the shielded welding station. Actual measurements of dose rates while fuel handling operations were underway are reported in Table 2. They show an excellent correlation with calculated values. TABLE 2 DOUGLAS POINT STORAGE PROGRAM DOSE RATE AT VARIOUS LOCATION (mSv/h) LOCATION
CALCULATED MEASURED (Gamma radiaGamma Neutrons radiation only)
Shielded Work Station (Contact) 0.05
- 0.076
0 06
0.05
Shielded Flask (Contact)
0.1
- 0.17
0.17
0.12
One canister (Contact)
0.025 - 0.064
0.026
0
In-between 4 canisters
0.1
0.012
0
0.0014
0
TABLE 3 DOUGLAS POINT STORAGE PROGRAM ACCUMULATED DOSE BY PERSONNEL (Person-mSv)
PHASE OF WORK
WHOLE BODY
Ba sket Loading
4.06
Ba sket & Canister Loading
7.58 * 11.64
Tocal Program
SKIN
TOTAL
8.3
12.36
11.0
18.58
19.3
30.94
* (includes 0. 73 person-mSv from neutrons > Douglas Point Program Schedule The storage program was scheduled for a duration of 19 months, from March 1986 to September 1987. The main program milestones are given below, together with the scheduled dates and the dates actually achieved:
Milestones
Scheduled
Actual
Start of Program
March 86
May 86
Construction License
September 86
September 86
Start Basket Loading
January 87
February 87
Start Canister Loading
May 87
July 87
Complete Canister Loading
September 87
November 87
It can be seen that the delay at the start of program (mostly attributable to financing approval) was carried to the end of the program. Despite the late start, the construction license was obtained on schedule. Delays attributable to extended commissioning of the F/H tooling and the shielded work station were carried to the end. The purchase of the second flask saved valuable time during the canister loading phase due to the enhanced flexibility that it provided. The fuel handling installation at Douglas Point achieved (with only one shift) a pace equivalent to 700 to 1000 MgU per year. THE NUCLEAR POWER DEMONSTRATION (NPD) PROGRAM
At Site Fence
- 0.26
0.0025
The total dose accumulated for the whole Douglas Point storage program was 31 person-Sieverts (3.1 man-Ren), including the neutron dose (see Table 3 ) . This corresponds to less than 0.1 person-Sievert (0.01 man-Rem) per MgU stored, which 1 B significantly lesa than other methods of storage.
140 CNS 9th ANNUAL CONFERENCE, 1988
The NPD plant was permanently shut down in June 1987, after 25 years of operation. A cost study was undertaken which, similar to Douglas Point, Indicated that spent fuel storage in canisters was less expensive than continued storage in a spent fuel bay, or storage in Tile Holes. The NPD storage program consists of storing 67 MgU of natural uranium fuel and of 8 MgU of enriched driver fuel. The NPD bundles are similar to the Douglas Point bundles. All of the remaining spent fuel from the NPD plant was moved to the NRX spent fuel bay in Chalk River Nuclear
Laboratories (CRNL), between September 1987 and December 1987. The storage program, which will be similar to that of Douglas Point t is scheduled for completion in 1989. The project is now at the planning stage. The same basket and the same canister will be used as at Douglas Point (see Figure 1-d). The commercializat ion of the technology at Gentilly-1 and Douglas Point will result In the NPD program being a straightforward production type of operation.
and uranium carbide fuels. For specific LWR fuel assemblies, the following studies have been conducted at WNRE during the past 5 years. LWR Canister Program - Studies Completed: I)
iO iii) iv) v)
Thermal Analysis
Shielding Analysis Criticality Assessment Fuel Handling Assessment Detailed Cost Analysis
THE POINT LEPREAU DRY SPENT FUEL STORAGE PROGRAM In 1987, a cost comparison between wet storage in additional spent fuel bays and dry spent fuel storage in canisters was j ointly undertaken by the New Brunswick Electric Power Commission (NBEPC), the Whiteshell Nuclear Research Establishment of A E C L and CANDU Operations of AECL. The scenarios studied were very exhaustive. They Included the building of additional 10 year and 20 year capacity spent fuel bays with epoxy or stainless steel linings. For the dry storage option, canisters of the Douglas Point type having 342 bundles (see Figure 1-d), and larger volume versions having 540 bundles (see Figure 1-e) were studied. All the scenarios were carried for 30 more years of plant operation and for 10 or 20 years of spent fuel production. The conclusion of the study was that spent fuel storage In canisters (having 540 bundles capacity) is significantly less expensive than construction of and storage in additional spent fuel bays. Based upon the recommendations of the study, which was completed at. the end of 1987, the NBEPC has undertaken a dry storage program based on the 540 bundle canister design, for implementation in 1990. This canister is shown in Figure 1-e. The program is now in the early project phase with most technical and licensing documentation scheduled for completion by the end of 1988. This program will store the yearly arisings of the Point Lepreau plant. These are nominally of 93 MgU per year but due to the very high plant capacity factor achieved by this plant the expected iate of storage is more likely to be arcund 120 MgU per year. Dry storage In canisters being the reference storage method for the entire lifetime of the plant, implies that approximately 30 more years of spent fuel production needs to be stored, assuming a 40 year plant life. This represents a quantity of 147,000 bundles at nominal capacity factors, or a total mass of 2,790 MgU. This amount of fuel can be stored in 273 canisters. These numbers may however be affected by the real plant lifetime and its annual load factor. Also, potential technological improvements increasing the quantity of spent fuel stored per canister could also have an effect on the final quantity of canisters used. After only 10 years of operation, the Point Lepreai* project, with 930 MgU accumulated, would be expected to be the largest dry spent fuel storage facility In the world, according to today's standards. LIGHT WATER CANISTER (CASK) PROGRAM Enriched fuel has been stored in canisters at WNRE since the raid L97O's. AECL canisters are fully licensed to store enriched uranturn oxide
These studies were conducted using PWR and BWR fuels of various burn-up and cooling periods (5 and 10 years), various fuel configurations and normal or heavy concrete shielding material. Several years ago Nuclear Packaging Inc. (NUPAC) a U.S. based company was granted a "sole license" to market the Canadian concrete canister technology In the U.S. market. This company has recently completed detailed assessments of: i) II) iii) Iv)
dry loading canister (cask), holding 9 PWR assemblies, dry loading canister (cask), holding 16 BWR assemblies, wet loading canister (cask), holding 9 PWR assemblies, wet loading canister (cask), holding 16 BWR assemblies*
These are shown in Figures 1-f and 1-g. A Topical Report In this regard was subir.itted to the US Nuclear Regulatory Commission, In November 1987. The licensing process with the NRC is expected to be completed in 1988. These canisters (casks) are about the same diameter as the 47 Douglis Point canisters but slightly shorter (5.5 ra instead of 6.1 m ) . A demonstration program (for these canisters (casks) at Idaho National Engineering Laboratory (INEL), is scheduled to start In 1988. THE ONTARIO HYDRO DRY SPENT FUEL STORAGE PROGRAM The Ontario Hydro Nuclear program is based on four unit stations, coupled u.^th very large spent fuel bays of 3,000 MgU to 6,500 Mg'J capacity. Due to the economy of scale, the construction cost ^ f these spent fuel bays Is typically well below $10 (U.S.) per KgU and total (overnight) storage costs slightly below $10 per KgU. Supplementary storage capacity will be requited at Bruce A in 1993-1994 and at Pickering A and Pickering B in 1994. A recent study carried out by Ontario Hydro has indicated that dry storage for future Pickering station needs would be competitive with building another spent fuel bay, because of the relative complexity of the fuel handling operations required for non-adjacent spent fuel bays. Ontario Hydro recently designed a 384 bundle (7.3 MgU), Concrete Integrated Container (CIC), which is shown in Figure 1-h. The constrjction of 2 demonstration CIC's at Pickering is scheduled for 1988, with two more units scheduled for 1989. Following the assessment of these demonstration units, dry storage may be implemented at Pickering at a rate of several hundred MgU per year.
CNS 9th ANNUAL CONFERENCE, 1988 141
COSTS OF DRY SPENT FUEL STORAGE IN CANISTERS The three principal generic advantages of dry storage are: the fact that cash flow requirements (as compared to building additional bay storage capacity) are relatively low, canisters have inherently low operating costs, and the static nature of the components require only infrequent attention. With canisters, very low life cycle storage costs can be realized because of the simplicity of the design and the inherent low cost of concrete as a shielding medium. For Gentilly-1 the storage costs were $51 (U.S.) per KgU because of the need to design and manufacture i.in= Fuel Handling equipment and because of the engineering development work required. For the WNRE development project, the storage costs were similar to Gentilly-1, (only the cose of the use of an existing hot cell was included) • The Douglas Point storage program was able to reuse the Gentilly-1 equipment. Because the quantity of fuel was much larger, storage costs of $18 (U.S.) per KgU were achieved. The NPD project is also expected to reuse Gentilly-1 equipment» but due Co the smaller quantity of fuel to be handled, costs of $22 (U.S.) per KgU are expected. The Point Lepreau storage p.ogram wi 11 use a larger canister. Its rate of storage of 93 MgU per year (930 MgU for 10 years) is much larger. Even with the inclusion of up-front costs, storage costs of components, operating expenses to the year 2020 and decommissioning costs, the predicted cost of the Point Lepreau storage program is only $10 (U.S.) per KgU for its total overnight life cycle cost. The storage costs for LWR fuel are somewhat higher than for natural uranium fuel. Recent evaluations indicate that costs of $30 - $35 (U.S.) per KgU can be achieved for a single unit plant. This is lower than steel casks ($60 $100 (U.S.) per KgU, or other concrete based systems with horizontal fuel storage ($40 - $60 (U.S.) per KgU). SUMMARY OF COSTS
Project WNRE
Genttlly-1
Tonnage
23
Costs (U.S.) S $50 (actual)
67
$51 (actual)
298
$18 (actual)
ling equipment adjacent Co the spent fuel bay, for direct dry spent fuel storage. The inclusion of this equipment signiFicantiy reduces the spent fuel bay nominal size (and hence the spent fuel bay capital construction cost), and also reduces the dry spent fuel storage cost by simplifying the fuel handling operations. Most future applications are expected to be in operating stations having an existing wet storage system with a nominal capacity of 10 years. The fuel will have a lower and better defined heat release, and larger capacity canisters will be possible. The proposed move from a 6.5 MgU canister at Douglas Point to a 10,3 MgU canister at Point Lepreau illustrates this point. Several technological advancements, presently in the design phase, are mainly intended to automate the fuel handling operations and to increase the storage capacity of the canister. Presently the 10.3 MgU, Point Lepreau type canisters are estimated to use only 50% of the licensed thermal heat release capacity of the canisters and the plans are to significantly increase this storage capacity. Future commercial applications for CANDU plants are expected to be centered on the four CANDU 600 stations, which will require new storage capacity as follows:
Plant
New
Stora ge Capacity Required
Point Lepreau
,990
- early 1991
Wolsong 1
1991
- 1992
Gentilly-2
1992
Embalse
1992
- 1993
The Point Lepreau dry storage program is already underway. Because the option of building an additional spent fuel bay has overnight lifetime operating expenses which are 50 to 60% more costly than those for dry storage, the other owners of CANDU 600 plants aie expected to opt for dry storage in canisters when their need arises. By 1993 an annual quantity of approximately 400 MgU of spen. fuel is expected to be stored at the four CANDU 600 plants. Canister applications for LWR plants and for other CANDU plants are being pursued. SUMMARY
Douglas Point NPD Point Lepreau
75 930 (over 10 years)
$22 (projected) $10 (projected)
FUTURE TECHNOLOGICAL AND COMMERCIAL DEVELOPMENTS Past experience has shown that dry spent fuel storage can be implemented, cost-effectively, Into an existing station. Future CANDU 600 and CANDU 300 stations will have built-in fuel hand-
142
CNS 9th ANNUAL CONFERENCE, 1988
AECL has already completed three projects t o t a l l i n g 388 MgU of spent fuel stored in 68 canisters. Ten of these canisters are loaded with enriched fuel, the remaining ones are loaded with natural uranium fuel. Two more projects and one demonstration project are ongoing. One project will store 75 MgU of spent fuel In 13 canisters. The other will s t o r e , over a 30 year period, a quantity of 2,790 MgU of spent fuel in 273 c a n i s t e r s . A t o t a l quantity of 1416 MgU in 175 canisters are thus committed. Engineering studies and licensing documents (Topical Report) are now completed for concrete canisters (casks) to hold BWR or LWR fuel. A demons t rat Ion of
concrete canisters (casks) holding LWR fuel Is scheduled at INKL and a demonst rat ion of Ontario Hydro designed Concrete Integrated Containers for CANDIJ fuel is also scheduled at the Pickering site. The advances in AECL1s canister technology displayed in recent proj e c t s , and the technological advances expected from committed demonstration an»i storage projects should assure the continued improvement of the econorai c advantage of concrete canisters (casks) for both CANDU and LWR dry spent fuel storage.
REFERENCES (1) Sochaski, R.O. 1984, "Canadian Experience with Concrete Canister Dry Fuel Storage", Presentation to the International Workshop on Irradiated Fuel Storage Operating Experience and Development Program, Toronto, Ontario, Canada, October 17-18, 1984. (2) D.W. Patterson, Whiteshell Nuclear Research Establishment and D. See Hoye AECL, CANDU Operations Montreal, 1986, "Canadian Experience Storing Irradiated CANDU Fuel In Concrete Canisters", Presentation to the Third International Spent Fuel Storage Tehcnology Symposium/Workshop, Seattle, Washington, U.S.A., April 8-10, 1986. (3) R.R. Beaudoln and P. Pattantyus, AECL, CANDU Operations Montreal, "Commercial Applications of Dry Storage Technology", Presentation to the Second International Conference on CANDU Fuel, 6-8 October 1986, Chalk River, Canada.
CNS 9th ANNUAL CONFERENCE, 1988 143
EXPERIMENTAL VALIDATION OF MODELS FOR RADIATION DOSE RATE FROM CANDU SPENT FUEL
K.M. AYDOGDU AND C.R. BOSS
Atomic Energy of Canada Limited CANDU Operations Sheridan Park Research Community Mississauga, Ontario L5K 1B2
ABSTRACT The standard calculational techniques and shielding codes ORIGEN, AUISN and QAD-CG have been used to calculate dose rates from the irradiated fuel bundles for the Douglas Point dry irradiated fuel storage facility. The measured dose rates from fuel trays under water were translated into an estimate of fuel decay times to identify 'cool' (eight years or longer) bundles in the bay. The radiation dose rates from the welding station, transport flask and concrete canisters containing the fuel basket(s) were measured and compared with those of calculations. INTRODUCTION The interim dry storage concept of irradiated fuel bundles has been developed since 1974 and some fuel has been stored in concrete canisters at the Whiteshell Nuclear Research Establishment since early 1 9 7 6 t l > 2 ) . When the Douglas Point station was shutdown in May 1984 a decision was made to use dry storage for the irradiated fuel. The records showed that 22,236 irradiated fuel bundles were stored in the spent fuel bay. These bundles have now been transferred to concrete canisters, built in an open area adjacent to and at the east end of the Turbine Building as shown in Figure 1.
Since the irradiated fuel bundles are highly radioactive, adequate shielding must be provided to minimize radiation exposure. The methodology used to calculate dose rates from irradiated fuel bundles with different shielding geometries (i.e., fuel basket, welding station, transport flask, etc.) are presented in this paper. The calculated 'dose rates are compared with those of available measurements. Most of the radiation released from an irradiated fuel bundle is gamma radiation from the beta decay of the fission products. In addition there is a small amount of neutron radiation but, for an unshielded fuel bundle, the dose rate from the neutrons is about 10" s of the dose rate from the gammas. As a result the shields on the existing equipment used for the fuel transfer operation were primarily gamma shields and were, consequently, built of lead and steel. These materials have such poor neutron attenuation characteristics that, as our study showed, the neutrons outside some of the fuel transfer equipment would contribute to the total dose rate. DESCRIPTION OF CALCULATIONAL METHODS The neutrons released by the irradiated fuel bundles are produced by spontaneous fission of the
FIGURE 1 DOUGLAS POINT ORY IRRAOIATEO FUEL STORAGE FACILITY
144 CNS 9th ANNUAL CONFERENCE, 1988
actinides " ° P u and ''•'Cm formed during irradiation. In addition the alpha particles released in the decay of the actinides " ' A m , 2 5 I P u and ! *°Pu have sufficient energy to emit a neutron from the fuel oxide via the l 6 0 (a, n) reaction. The isotope generation and depletion code was used to calculate the inventory,es of ORIGEN these actinides and the resulting neutron source strength of the fuel bundles. In addition the ORIGEN code was used to calculate the fission product inventories and the fission product decay gamma spectra. All the ORIGEN calculations covered the decay times of the fuel ranging from three to twenty years. The fission product decay gamma dose rates around various source geometries, i.e., basket in water, basket in the welding station, etc., were •calculated using standard techniques''1' and the QAD-CG' ' code. The neutron dose rates around the welding station and the transport flask were calculated by ANISbr , the one dimensional discrete ordinates transport code.
Radiation Sources from Irradiated Fuel Most of the fuel bundles discharged from the reactor had received full burnup (approximately 185 MW.h/kgU); the only exception was the last core load (3672 bundles) which was discharged prematurely when the station was shutdown in May 1984. The station operating history is given in Figure 2, in which the yearly integrated output is plotted against time. The station operated between 75% and 100% of full power. The power outputs in years 1972 and 1980 were low due to long outages.
-
r
1
/]
n
TABLE 1: GAMMA RELEASE KATES FOR DOUGLAS POINT RADIATED FUEL BUNDLES
Average Power, Average Exit Burnup
Haximuiu Power, Maximum Burnup
Energy Range (Mev)
Mean Gamma Energy (MeV)
Three Years decay
Ten Years decay
Three Years
(r/s)
(r/s)
(r/s)
(r/s)
0. 20-0.40 0. 40-0.90 0. 90-1.35 1. 35-i.80 1. 80-2.20 2 . 20-2.60 2. 6 -3.0 3. 0 -3.5
0.30 0.63 1.10 1.55 1.99 2.38 2.75 3.25
4.90E12* 2.42E13 1.17E12 2.26E11 1.41E11 2.01E10 1.55E9 4.89E7
5 .39E11 1 .12E13 2 .24E11 1 .21E10 5 .36E8 1 .62E8 1 .27E7 4 .01E5
7.33E12 3.43E13 1.79E12 3.53E11 2.16E11 3.14E10 2.42E9 7.64E7
6 .63E11 1 .43E13 3 .30EU 1 .81E10 7 .94E8 2 .52EB 1 .98E7 6 .27E5
decay
Ten Years decay
* read 4.90E12 as 4.90x10" contribute to the shielded or even unshielded dose rates, and the number of gammas emitted above 3.5 MeV at these decay times are essentfally zero. Therefore, seven or even six energy groups are quite satisfactory for the dose rate analysis given in the next section. The total neutron source in these irradiated fuel bundles was calculated by the same ORIGEN calculations. The total number of neutrons in the average power, average exit burnup bundle was calculated to be 6.5 x 10* n/s, 6.1 x 10* n/s and 5.9 x 10* n/s at three, five and ten years after discharge. The corresponding values for the maximum power, maximum exit burnup were 1.4 x 10' n/s, 1.3 x 10 s n/s and 1.2 x 10 s n/s respectively. Owing to the long half-lives of the neutron sources, there is a little drop with time in the neutron source strength.
Gamma Dose Rate Calculation
PEHMAN NT SHUTDOWN
FIGURE 2 DOUSLAS POINT POWER HISTORY
It is impractical to generate decay source spectra foi every fuel bundle versus time. To simplify the calculations, two scenarios of bundle power, bundle burnup were assumed. In one case, the bundles were assumed to have average values for 'both parameters viz., 221 kW bundle irradiated to an average exit burnup of 185 MW.h/kgU. In the other case, the bundles were assumed to have maximum values viz., 431 kW bundle irradiated to a burnup of 230 MW.h/kgU. For these operating conditions, the fission product decay gamma release rates at three and ten years after discharge are given in Table 1 in seven energy groups of the ORIGEN group structure. The gamma and X-ray groups below 0.3 MeV do not
The materials in the basket are treated as a uniform homogeneous mix of fuel bundles and steel. The source is treated as a self-absorbing cylindrical source ' while calculating dosa rates outside the welding station, the transport flask, and the concrete canister. Dose rates from the nine baskets of irradiated fuel in a concrete canister were calculated as a function of distance. This drop was used to find the dose rates from a row of twelve canisters and a .row of four canisters. The dose rates on contact with the fuel basket immersed in water were calculateu by the QAD-CG code. The dose rates at 35 mm f.nd 150 mm from the centre of a fuel tray containing eleven fuel bundles, were a.'so calculated by the same code. Lead dose buildup factors were used for the former distant •, while Broder's* 'multiple layer dose buildup factor formalism was considered for the latter distance. Neutron Dose Rate Calculation The one dimensional discrete ordinates transport code ANISN was used in spherical geometry to calculate the neutron dose rates arcund the welding
CNS 9th ANNUAL CONFERENCE. 1988 MS
station and transport flask. The neutron source in one fuel bundle given earlier was multiplied by 54 (the number of bundles in a basket) co obtain the total neutron source, viz, 3.5 x 10 s n/s for 3 year decay, average power, average exit burnup fuel. The neutron spectra resulting from spontan ritaneous Eission and 1 D 00 (a,n) (a,n) reaction reaction neutrons^* neutrons^ 5'' was ua fission JSI approximated by the C f neutron spectra given in Reference 9. This is given in Table 2 in ANISN energy group scheme.
IRRADIATED FUEL HANDLING Irradiated fuel handling techniques were the same as those used in Geutiily-1. The fuel baskets were loaded under water, drying and welding of the transfer of the sealed baskets to the concrete canister site was accomplished by means of the shielded fuel transport flask. The Gentilly-1 welding station and the fuel transport flask was re-used at Douglas Point after some refurbishing. Fuel Tray
TABLE 2: NEUTRON SPECTRUM USED IN ANISH CALCULATIONS FOR NEUTRON DOSE RATES OUTSIDE WELDING STATION AND TRANSPORT FLASK ANISN Grp.il>
Energy Range (MeV)
1 2 3 4
5 6 7 8
14.90 12.20 11.10 6.06 3.68
- 12.20 - 11. 10 - 6.06 - 3.68 - 2.23
2.23 1.35
- 1.35
0.827 0.498 0.302 0.183 0.067 0.041
9
10 11 12 13-27
-
Fraction Born 2.41 3.32 2.62 1.13 2.08 2.62 1.48 1.27 3.52 6.00 2.00
0.827 0.498 0.302 0. 183 0.674 0.041 0
E-4 E-4 E-2 E-1 E-1 E-1 E-1 E-1 E-2 E-2 E-2
The Douglas Point fuel bundle is a 19-eleinent design and it has a maximum bundle diameter of 81.7 mm while the bundle length is 495 mm. The bundle weights 16.7 kg of which 15.2 kg is UO ; and 1.5 kg is Zircaloy-2. Each fuel tray holds eleven fuel bundles. The fuel tray and tray pick-up tool •are sketched in Figure 3. Fuel Basket The fuel basket with 810 mm maximum diameter and 540 mm height is made of 340 L stainless steel as shown in Figure 4. Each basket has a maximum storage capacity for 54 bundles in four rings as shown in Figure 5. Bundles are placed vertically, in a basket, with six bundles in an inner ring, twelve bundles each in a second and third ring, and twenty-four bundles in an outermost ring.
—
Total = 1.000 Total number of neutrons born in basket = 3.5 x 10* n/s (average power, average e*it burnup fuel after thres years decay)
The element atomic densities were calculated for the basket and shielding materials. The cross section for the elements were represented by a Legendre expansion limited to P 3 , and S 1 6 angular quadrature were used. A 38 group coupled neutron gamma library of cross sections was used.
\ COVER SEAL WELDED TO CENTRE POST AN& BASE AFTGB FllllHG COVER INSTALLATION AND DRVINC
TOP VIEW , 100 mm HOLE IN PLATE
BUNDLE
O
o
C)
I ALUMINUM PLATE
CAPACITV M FUEL BUNDLES
FIGURE 3 A SKETCH OF DOUGLAS POINT FUEL TRAV AND TRAV PICKUP TOOL 146
CNS 9th ANNUAL CONFERENCE, 1988
FIGURE 4 DOUGLAS POINT FUEL BASKET
TOIAL OF M FJEi. BUNDLES BOUNDARV BEIW Ou'En RING AND INNER HINGS
?J w OUTER RIHG
JO IN INNER RINGS
FIGURE 5 TOP VIEW OF FUEL STOHAGE BASKET SHOWING ARRANGEMENT OF FUEL BUNDLES
Welding Station The welding station shielding walls consist of 152 mm lead slab sandwiched between 10 mm thick steel plates. The sketch for the shielding geometry used in dose rate calculations is shown in Figure 6. The welding station is shown in Figure 7.
\
WELDING STATION WALL
FUEL BASKET FILLED WITH FUEL
FIGURE 7 FUEL TRANSFER EQUIPMENT — SHOWING FUEL BASKET LOADING, ORVING AND WELDING FACILITIES AND OPERATIONS
PNEUMATIC HOSE
FOR BASKET GRAPPLE OPERATION
NOT TO GCALE 001 m_ STEEL
FIGURE 6
SHIELDING ARRANGEMENT USED IN DOSE RATE CALCULATION FOB THE WELDING STATION
SAFETY BOLTS
Transport Flask The transport flask has a similar shielding .angement as showi, in Figure 8. It is a cubical-shaped vessel, 1.16 m square and 1.12 m high, and is b^sed en the 1976 WNRE design.
B»riOM SHUTTERS
FIGURE 8
FUEL TRANSPORT FLASK - SECTIONAL VIEW
CNS91h ANNUAL CONFERENCE, 1988 147
Concrete Canisters The canister is a cylindrical reinforced concrete shell with a 34-irch diameter standard pipe carbon steel liner. It has a 2.59 m outside diameter, is 6.157 m in height and has an inner cavity with a diameter of 0.845 m. It provides shielding of 0.86 m of concrete (p = 2.4 g/cm 3 ) and 9.5 mm of steel. This is shown in Figure 9. The site contains 47 canisters, arranged in 4 rows as shown in Figure 1. The canisters are positioned on a 3.6 m square lattice.
MEASUREMENTS CALCULATIONS
DOSE RATE (S«/h|
FIGURE 10 COMPARISON OF MEASURED AND CALCULATED DOSE RATES 35 mm FROM A FUEL TRAV CONTAINING ELEVEN FUEL BUNDLES UNDER WATER
AVERAGE POWER. AVERAGE BURNUP
FUEL BASKET ID > 0 78 m HEIGHT . OM m (EACH BASKET HOLDS U 8L/NOLESI
MEASUREMENT RANGE CALCULATIONS CONCRETE BASE.
DOSE RATE (Sv/h)
FIGURE 11
EQUALLY SPACED
FIGURE 9 DOUGLAS POINT DRY STORAGE CANISTER SHOWING FUEL BASKET LOADING ARRANGEMENT COMPARISON WITH MEASUREMENT The measured dose rates from a fuel tray under water, a basket in water, a basket in the welding station and transport fJask, the nine baskets inside a concrete canisters are compared with those of the calculations. Ttfese are presented below. Fuel Tray The expected dose rates 35 mm directly above the centre of a fuel tray are given in Figure 10. The fuel tray is loaded with eleven fuel bundles. The measured dose rates (5 points) are also plotted in this figure. The calculated dose rates agree well with the measurements. 148 C N S 9th A N N U A L C O N F E R E N C E , 1988
COMPARISON OF MEASURED AND CALCULATED DOSE RATE 150 mm FROM A FUEL TRAY CONTAINING ELEVEN FUEL BUNDLES UNDER WATER
Figure 11 shows subsequent measurements involving about 50 trays containing bundles discharged between 1973 and 1984 according to the records. This time the dose fate was measured at 150 mm directly above the fuel tray containing eleven bundles. The measured points were joined together to form a horizontal bar and compared with those of the calculations in this figure. This figure shows that there is a factor of 1.5 discrepancy between the measurements and calculations. The discrepancy is attributed to: a) The treatment of fuel bundle as a homogenized cylinder of uniform source strength, (This assumption oj uniJorm strength underestimates the dose rates by, at most, 10% but this would be more than offset by the overestimate introduced by homogenizations - The latter effect could represent an overestimate of 30% so, overall, the calculations could be 20% too high.)
b) The Broder buildup factor treatment might introduce an overestimate of 10%. (Note that single layer dose buildup factors for lead kas quite suitable for dose rate calculations on contact or a fpw centimeters away from fuel baskets.) c) The fission product source strength calculated by ORIGEN could be overestimated by 10 to 20%.
TABLE 4: CALCULATED GAMMA AND NEUTRON DOSE RATES ON CONTACT WITH THE WELDING STATION AGAINST MEASUREMENTS DECAY TIME OF FUEL IN BASKETS
D O S E GAMMA
R A T E
(mSv/h)
NEUTRON
TOTAL
all 3 years
0.35*(0.54)** 0.028(0.060) 0.38(0.60)
all 5 years
0.084(0.13)
0.026(0.057) 0.11(0.29)
0.008(0.012)
0.025(0.051) 0.033(0.06)
0.06 (Avg.) 0.20 (Max.)
0.05 (Avg.) 0.08 (Max.)
Fuel Basker The dose rates on contact with several fuel baskets were measured under water. The measurements ranged from 10-40 Sv/h on the average to a maximum of 80-90 Sv/h. These values are compared with thoss of predictions, and they show a reasonably good agreement. This is given in Table 3.
TABLE 3: CALCULATED GAMMA DOSE RATE ON CONTACT WITH BASKET ON WATER AGAINST MEASUREMENTS
DECAY TIME OF FUEL IN BASKETS
DOSE: RATE (Sv/h) AVERAGE POWER, AVERAGE BURNUP
Measurements (Basket #282) NOTE: * **
68.2
98.4
ill 5 years
42.5
58.1
.11 10 vears
26.6
34.4
10-4U Sv/h (Avg. ,v 80-90 Sv/h (Max.)
Welding Station Measured radiation dose rates around the welding station were within the predicted levels for gammas, but for neutrons, although extremely low, they exceeded the predictions for some esses. Table 4 shows the measured dose rates from a "hot" basket inside the stations (Basket #282) against calculations. It was recommended that 25 mm thick polythene slabs should be used to reduce neutron dose rates around the welding station. Based on ANISN calculations, such a slab around the station reduces the neutron dose rate by a factor of 2.8. The predicted reduction factor was 4.8 with a 38 mm thick polythene slab. Transport Flask The calculated dose rates around a flask are compared with those of measurements in Table 5. Again, the gamma dose rates were well within the predicted levels and the neutron dose rates in most cases are within the expected levels. Concrete Canisters Measured dose traces sC 1 m and 2 m abovs Che base slabs for each of the 47 concrete canisters were compared with those of calculations. They averaged around 20 uSv/h and reached a maximum of 40 uSv/h, well within the expected levels. The comparison with the predictions are givon in Table 6.
: Avg. power, avg. burnup : Max. power, max. burnup
TABLE 5: CALCULATED GAMMA AND NEUTRON DOSE RATES ON CONTACT WITH THE TRANSPORT FLASK AGAINST MEASUREMENTS
DOSE RATE (Sv/h) MAXIMUM POWER, MAXIMUM BURNUP
ill 3 years
measurements
all 10 years
DECAY TIME OF FUEL IN BASKETS
D O S E
R A T E
GAMMA
(mSv/h)
NEUTRON
TOTAL
all 3 years
0.78*(1.20)*«
0.055(0.12)
0.84(1.32)
all 5 years
0.19 (0.29)
0.052(0.11)
0.24(0.40)
all 10 years
0.019(0.028)
0.050(0.10)
0.07 (0.13)
Measurements (10 flasks) 0.20 Avg. 0.10 Avg. Bask.#282 1 0.10 Avg. 0.08 Avg. 0.20-0.30 Max 0.20 Max. Note:
* **
: Avg. power, avg. burnup : Max. power, max. burnup
TABLE 6: CALCULATED GAMMA AND NEUTRON DOSE RATES ON CONTACT WITH CONCRETE CANISTERS AGAINST MEASUREMENTS DECAY TIME OF FUEL IN BASKETS
DOSE RATE (msv/h) Avg. power, Avg. exit burnup
all 3 years all 5 years ill 10 years
Me* surements*
Note
0.270 0.074 0.013
DOSE RATE (mSv/h) Max. power, Max. burnup 0.420 0.110 0.018
0.02 (Avg.) 0.04 (Max.)
Measurements were made at 1 and 2 m above the base slab for each of the 47 concrete canisters
C N S 9th A N N U A L C O N F E R E N C E , 1988 149
The measured dose rates at the fence were 1.5 uSv/h and 1.0 pSv/h opposite the twelve and four rows of canisters. The calculated dose rates for eight year decay bundles, 2.5 uSv/h and 2.0 nSv/h, show good agreement with these measured values.
CONCLUSIONS Standard calculational techniques and computer codes ORIGEN, ANISN and QAD-CG were used to calculate gamma and neutron dose rates around fuel trays, baskets, welding stations, transport flask and concrete canisters for the dry fuel storage program. Gamma dose rates from fuel trays under water were calculated by the QAD-CG code. The results obtained close to the source agreed well with measurements. However, at 150 mm away from the source, the calculated results were significantly higher than measurements. This is attributed to conservatism inherent i n the modelling. The same code was used to obtain contact dose rates from the fuel basket under water. Satisfactory agreement with measurements was obtained1. When the basket is inside the welding station, transport flask or the concrete canisters, satisfactory agreement with measurements was also obtained, considering the fact that the individual bundle histories were iiot simulated. Therefore, in general, the measured gamma dose rates were well within the range of th« calculated values. Neutron dose rates Vere calculated by ANISN in spherical geometry. The neutron dose rate, although substantially low, exceed the calculated dose rates from a few hot baskets.
ACKNOWLEDGMENT We would like to thank E.S.Y. Tin, Licensing Supervisor for G-l and DPNGS for providing us with the dose rate measurements and to A. Powaschuk, Radiation Protection Supervisor, for providing us assistance during measurements of dose rates from fuel trays in the spent fuel bay.
150 CNS 9th ANNUAL CONFERENCE, 1988
REFERENCES 1. OHTA, M.M. "The Concrete Canister Program" Atomic Energy of Canada Limited Report, AECL-5965, 1978 2. BOITLTON, J. , E.litor, "Management of Radioactive Fuel Wastes: The Canadian Disposal Program", Atomic Energy of Canada Limited Report AECL-6314, 1978 October. 3. BELL, M.J. "ORIGEN - The ORNL Isotope Generation and Depletion Code", Oak Ridge National Laboratory Report. ORNL-4628, 1973 May. 4. ROCKWELL, III. T., Editor, "Reactor Shielding Design Manual", D. Van NQstrand Company Inc., Princeton, New Jersey, 1956. 5. "QAD-CG. A Combinatorial Geometry Version of QAD-P5A, A Point Kernel Code for Neutron and Gamma-Ray Shielding Calculations", CCC-307, RSIC Computer Collection, 1979 May. 6. ENGLE, W.W. Jr., "A Users Manual for ANISN - A One Dimensional Discrete Orciinates Transport Code with Jinisotropic Scattering", K-i69S, Union Carbide Corporation, 1967 March. 7. BRODER, D.L. et al, Atnmnaya Energiya, Volume 12, p 30, 1962, reported in ESIS Newsletter, ISSN 0392-6591, Issue 41, 1982 April. 8. JACOBS, G.J.H. and LISKIE)}, H. "Energy Spectra of Neutrons Produced by a-Particles in Thick Targets of Light Elements1', Annual Nuclear Energy, Volume 10, No. 10, pp 541-552, 1983. 9. PROFIO, A.E. "Radiation Shielding and Dosimetry", a Wiley-Interscience Publication, John Wiley and Sons, New Vork, 1979.
CLEANUP AROUND AN OLD WASTE SITE A SUCCESS STORY
G. VANDKRGAAST AND D. MOFPKTT
Eldorado Resources Limited Ottawa, Ontario
B.E. LAWRENCE
HacLarentech Inc. Toronto, Ontario
ABSTRACT 42,500 m3 of contaminated soil were removed from off-site areas around an old, low-level radioactive waste site near Port Hope, Ontario. The cleanup was done by means of conventional excavation equipment to criteria developed by Eldorado specUic to the lana use arouno tYie Company's waste management facility. These cleanup criteria were based on exposure analyses carried out for critical receptors in two differe.it scenarios. The excavated soils, involving eight different landowners, were placed on the original burial area of the waste management facility. Heasures were also undertaken to stabilize the soils brought on-site and to ensure that there would be no subsequent recontamination of the off-site areac.
INTRODUCTION The Welcome Waste Management Facility is located four kilometers from the town of Port Hope, Ontario. Between 1948 and 1955, Eldorado placed approximately 17,000 ton-'-s of low-level radioactive wastes in a five hectare burial area inside the overall 36 hectare site. These wastes, primarily chemical precipitates at\4 residues from the processing of ores for the recovery of radium and uranium, also contain other metals naturally present in the ore such as arsenic, iron, cobalt, nickle, and manganese. They were placed on the ground surface of the burial area to a depth of about one meter or buried in trenches UP to five meters deep in the sand and gravel deposit which forms the burial area. The site was closed prematurely in 1955 because erosion and leaching of the waste materials was resulting in contamination of streams draining the burial area. In 1956, water collection ponds were excavated and a pumphouse was constructed for the pumping of contaminated effluent from the site through a 2500 m pipeline to Lake Ontario. In 1978, a chemical treatment system was put in place to remove radium and arsenic from the water prior to being discharged to the laVe, Substantial quantities of the wastes were removed from the site during the period 1956 to 1960 for metal recovery or burial elsewhere. Other than the above, however, the site was left largely as it was at the time of closure up to the 1980's - with wastes exposed at
surface and no definite idea on the extent to which contaminants had spread from the burial area. The facility was secured by chain-link fence and there was general evidence that the site was not resulting in any obvious environmental or haalth impacts. Gamma radiation levels at the facility fenceline were typically about 2 uSvAi, or less ton tine waste burial site, gamma radiation levels were in the range 5 to 500 uSv/h), radon concentrations at the nearest residences were within levels of normal background for north America, drinking water supply wells around the site showed no evidence of being influenced by contaminants from the site, and the water reporting to the treatment plant typically contained 10 ppM arsenic, 1 to 2 ppM uranium, and less than 1 Bq/L ratfium. In 1982, Eldorado began to plan for a new waste disposal facility in the local area. Amor.g other wastes, this new facility was to be designed for the disposal of the wastes and contaminated subsoils from and around the Welcome site, i.e., the Welcome site would be decommissioned and the wastes relocated to the new facility.
SITE Detailed site investigations at Welcome were carried out during 1983 and 1984 by HacLaren Engineers Incorporated of Toronto, Ontario. This consisted of an extensive, drilling and sampling program to determine the areal extent and depths to which above background concentrations of arsenic, radium and uranium existed. The focus on arsenic, uranium and radium was based on historical monitoring around the site and on preliminary work done in 1982. It had been determined that these were the only elements which had migrated from the site in significant quantitea and that the other metals associated with the waste were not present at unacceptable concentrations. Early on in the investigation, it became evident that substantial quantities of contamination had spread beyond the Eldorado fettteline
CNS 9th ANNUAL CONFERENCE, 1988 151
In addition, since there are in Canada tio criteria or standards for concentrations of arsenic, radium or uranium in soil which would allow the properties to be designated as clean, a major part of the work consisted of getting regulatory agreement and approvals for criteria developed and proposed by Eldorado for the off-site cleanup. Acceptable residual concentrations were established to allow unrestricted use of the cleaned up and restored properties. This was don" by specifying two different land use scenarios and estimating the resulting impact of each scenario on persons using the land for a number of likely uses. The potential impacts examined were total gamma radiation exposure to the individual for radium and uranium and toxicity to farm crops for arsenic. The two land use scenarios considered were:
1) Open Field.
Likely land uses are farming and rural residence.
2)
Creek channel. Likely land uses are occasional yisitors, primarily for recreational use, and use of stream water for irrigation of crops and watering of livestock.
The various exposure pathways considered for the two land use scenarios are shown in Figures 1 and 2. For scenario 1, two individuals were considered; the full-time farmer, living elsewhere, and the full-time on-site resident. The lerived cleanup criteria, based on the exposure pathways analysis are shown in Table 1.
FIGURE 2.
Table 1.
EXPOSURE PATHWAYS - CREEK.
criteria for off-site cleanup. Criteria
1. rial*
Sa-22i: 0.2 Bq/g* Uranium: 35 U£/£ Arsenic: 50 ug/g Rd-226: 0.8 Bq/fc* Uranium: 100 US^R Arsenic: 150 lig/g
Total Dose USv/annum Full time Fanner: 90 Full time Resident: 300
Typical Receptor:
90
*Above background
The overall investigation was conducted in several stages, with the level of detail and distance from the source increasing with each successive stage. The three phases of the investigation primarily concentrated on: - the original waste burial area, - the remaining areas of the Welcome site, and - the off-site areas neighboring the Eldorado facility.
FIGUKE 1.
EXPOSURE PATHWAYS - FIELD.
For each area, a gamma radiation survey was first carried out (using Eberline pHH-7 "Micro-R Meter") to direct the drilling and sampling program. By the time the investigation was completed, 57, 15 cm diameter boreholes, one to eight meters deep, had been drilled (primarily on the waste burial area) with a track-mounted drill rig and more than 1000 holes were drilled with a portable 75 cm ciameter pow^r auger and/or a 50 cm diameter hand auger to depths ranging from 15 to 280 cm. In addition, a small number of holes were excavated for sampling by a backhoe in areas too stoney to use augers. All holes were sampled at 15 cm intervals, resulting in a total of aboiit 5000 soil samples. The gamma activity of each sample was counted with a low-background scintillation counter (Eberline PRS-1 Portable Rate Meter/Sealer with SPA-3 Gamma Probe). In order to minimize analytical requirements, a correlation was sought between the net counts/minute Ccpm) and" radium concentration. Although a fair amount of variability was encountered, the resultant correlation, shown in Figure 3 was used to indicate the extent of radioactive contamination in the field. The net gamma activity
152 CNS 9th ANNUAL CONFERENCE, 1988
cot-responding to the two cleanup criteria for radium (Table I) were 75 cpm (field) and 160 cptn (creek). In addition to the gamma counting, selected soil samples were analyzed chemically for radium (about 200). uranium (about 2000) and arsenic (a"bout bOO) .
While these data indicate that contaminant concentrations did indeed exceed the respective cleanup criteria, they were not in the range of being an immediate health hazard. It was therefore considered possible to employ standard excavation procedures for t*ne removal and transport of the contaminated soils and only minimal precautions were required for the protection of workers. On the basis of the established cleanup criteria, and the results of the field investigations, it was possible to calculate the volume of contaminated soil and to map the extent of contaminant spread in the affected areas. The field investigations revealed a total of about 200,000 m3 of surface and subsoils which had become contaminated to levels above the respective area criteria. Approximately 1 5 % (31,000 m3) of this was found outside the fenceline of the Welcome site. This was therefore the amount targetted for cleanup. Figure 4 shows the areas of off-site contamination in L-elation to the Welcome facility.
Ha&iJtf5"
—"&ON"N"ELS a so Ra-226(pCi/g)
FIGURE 3.
NET GAMMA ACTIVITY VS RADIUM CONCENTRATION IN SOIL SAMPLES.
At the outset of the investigation, it was decided to do sufficient sampling so that the volumes and areas of contamination could be calculated to a level of detail to allow the actual cleanup to be carried out simply from \A\ese
160-2577 cpm 1-2 Bq/g 50-350 ppM 100-800 ppM
The distribution of contamination in the offsite areas further revea\eo ttoe •mec'nam\sm \>v which the contaminants were transported off-site. The burial area of the Welcome site is near a local topographic high point exposed to predominantly westerly winds. Run off from the burial area is in the westerly direction towards a creek channel about 1500 m away. The resulting pattern of contamination spread off-site to the north, east, and south was fairly uniform, covering a broad area (126,000 m*) up to about 335 m from the site boundary indicating areal deposition of wind eroded wastes. Furthermore, the contamination in these areas was found to occur in the near surface zone, ranging in depth from about 15 to 30 cm. To the west, contamination was generally restricted to drainage channels, occuring in narrow bands 5-30 m wide, and up to about 1000 m from the fenceline indicating water transport and deposition. Contamination was generally deeper \xrp to \ irt , Sttgfcestvnfc -mo-ce active mwemerrt. of contaminants by the surface streams.
Off-site contamination surrounding the welcome, site included the properties of eight private landowners, and road allowances belonging to the Township of Hope and the Ontario Ministry of Transportation and Communications. Eldorado contacted each landowner and negotiated an agreement for cleanup and restoration of the respective properties. These agreements specified the extent - are* and depth - to which the properties would be cleaned up, the cleanup objectives, the type of restoration work which would be done, and compensation provisions. Compensation was made for specific quantifiable items such as loss of trees, soil or fill removal and loss of farm crops during the period of awl cesticatioo..
CNS 9th ANNUAL CONFERENCE. 1988 153
FIGURE 4.
AREAS OF OFF-SITE CONTAMINATION.
One of the unique aspects of the cleanup program was that the restoration of the various properties was tailored to the particular type of land which was being cleaned up and also to the landowner's particular desires. In general, areas which were formerly wooded or non-productive were restored with sand fill. Stream channels were restored by shaping and grading alone. In this case, the landowners were compensated for the volume of soil or fill excavated. Agricultural fields were restored by replacing the top soil. In most cases, the cleaned up fields were parts of larger agricultural fields and agreement was made with the landowners that a thin layer of top soil from the areas adjacent to the cleanup area would be "borrowed" to restore the cleaned up area. The owners were paid at the going market rate for the borrowed top soil. Most areas were seeded with grass to complete the restoration. In addition to the landowner agreements, Eldorado also required the approval of the federal regulatory authority - the Atomic Energy Control Board and its advisors, Environment Canada and the Ontario Ministry of the Environment. This was required because it was intended that all of the contaminated soil and fill removed from the affected areas would be transported back to the burial area of the Welcome facility. As part of the approval process, the AECB, also agreed to act as the auditor for the cleanup. This audit included inspection of the cleanup work in progress, establishment of a compliance sampling protocol, supervising the compliance sampling and conducting independent analyses of these samples. The AECB also provided Eldorado with a letter of certification once each landowner's property had been cleaned up satsifactorily.
154 C N S 9th A N N U A L C O N F E R E N C E , 198B
The cleanup of the off-site contamination around the Welcome Waste Management Facility took place over four construction seasons (April through November) starting in 1984 and finishing in 1987. Removal of the contaminated soils was done by conventional excavation equipment; a backhoe with 1.8 m straight edge (ditching) bucket and tandem axle, 10 m3 capacity, dump trucks. Excavation of contaminated soil was "design driven". This meant that the perimeters of the areas of contamination were surveyed and marked on-site with stakes and within these areas, the excavation depths were strictly controlled to that specified for the particular area. There were to be no field determined conditions, i.e., gamma radiation levels, used to direct the excavation. In this way, the amount of soil actually excavated should be very close to that calculated. Only if subsequent compliance sampling showed that unacceptable amounts of contaminated soil remained, additional excavation would be carried out. This approach was selected because other remedial projects, relying on "field driven" criteria, often result in the excavation of soil volumes that greatly exceed the amount of material predicted. The "design driven" approach is immune to minor variations in the distribution of contaminants but still ensures that, overall, the cleanup will be successful. In practice, however, it was not always possible to stick to the design depth of excavation. In some cases it was physically difficult to maintain a specified depth, in other cases, compliance sampling results dictated that additional excavation was indeed required. Although most of the contamination occured at shallow depths requiring excavations of 15 to 30 cm, and on fairly level terrain, much of the area requiring cleanup was quite thickly wooded and/or littered with numerous boulders (up to 2 m diameter). This complicated the removal process somewhat in that the excavation of the tree stumps (after cutting and removing all the trees) and boulders made it difficult to maintain an even 15 cm soil removal depth. In these areas,
the actual, excavation depth more often was at about 30 40 cm, thereby increasing the volume of soil removed. Excavation of the contaminated soils from the various creek channels also provided some special challenges. Because the areas to be excavated were mostly narrow bands along the bottom of drainage channels which had heavily wooded banks, access to the areas of excavation was limited. Rather than removing all the trees from the creek banks (which were not contaminated), it was found that the contaminated soil could be removed by multiple handling of the soils by the backhoe towards specific access points spaced along the drainage channels. Furthermore, during the excavation of contaminated soils from the creek channels, it was occasionally found necessary (after compliance sampling) to go back and excavate additional soil from some areas. The water deposition of the contaminants is thought to have resulted in a more uneven distribution in these areas. All of the affected areas were successfully cleaned up and restored to the respective owners' satisfaction. A total of approximately 42,500 m 3 of contaminated soil (about 37% greater than the design) was removed and transported to the burial area on the Welcome facility. This material was graded into a set of gently contoured mounds over the entire burial area so that all previously exposed wastes received a fill cover at least 1 meter thick. These mounds were stabilized against erosion by a planting of grass seed. In order to protect the cleaned up areas from possibly getting re-contaminated by material transport from the Welcome site in the future, major on-site works were constructed for the containment of materials on-site. As mentioned, the contaminated soils brought on-site were seeded to provide a vegetative cover for protection against erosion by wind. A perimeter ditch system was constructed around the waste facility to convey all site runoff toward water collection ponds for treatment. New ponds were constructed which are of sufficient capacity to contain all the runoff from the site in all but the most extreme circumstances.
ENVIRONMENTAL AND HKALTH CONCERNS Prior to starting the cleanup program, environmental and health concerns were identified in relation to handling and working with contaminated soils. These consisted of: - gamma radiation exposure to workers on the waste burial area, - beta or gamma radiation exposure to workers by contact with contaminated soils, and - dusting of contaminants to the general environment during construction.
As a result, workers were issued with theunoluminescent dosimeters (TLD) and checked regularly with a contamination meter. High volume air samplers were set up at convenient locations near the construction areas. This monitoring indicated that the only concern with measurable impact was gamma radiation exposure while working on or near parts of the burial area in which waste materials were still exposed at the surface. Once all these areas were covered with the incoming fill, gamma radiation levels were not found to result in any measurable exposure. Contaminant levels in the off-site soils were generally low enough so tha . no criteria for exposure were exceeded in the monitoring of personnel or equipment. Similarly, while the high volume air sampling did indicate general dusting from the cleanup operations, analysis of the filter papers indicated extremely low levels of airborne contaminants. The monitoring program was therefore reduced substantially during the latter phases of the cleanup program. SUMMARY When the extent of contamination spread on properties around Eldorado's Welcome Waste Management Facility had been determined, the Company decided to cleanup and restore these properties immediately so that unrestricted use of the properties could be assured, even though there was not yet any prospect of a new, low level, radioactive waste disposal facility being made available. As a result, all of the contamination is now isolated and contained within Eldorado's property limits. The level of contaminants were not causing any immediate environmental or health impact. The cleanup was required to ensure that there would be no such impacts in the long-term. Because the levels of contamination were relatively low, standard excavation and transportation procedures were employed. Specialized containment of the wastes or protection of workers (other than exposure monitoring and contamination checking) were not required. The total cost of the cleanup operations and associated on-site remedial measures was approximately $1.1 million, or about $26/m 3 of contaminated soil excavated. The cleanup was considered a success for a number of reasons; 1.
The contaminated soils on the off-site properties were all satisfactorily removed, i.e. the cleanup criteria were met and certification for each was given by the AECB.
2.
All affected properties were restored to the satisfaction of the respective landowners.
3.
The cleanup of low-level radioactive material was carried out with no complaints or protests from the owners or other local residents.
CNS 9th ANNUAL CONFERENCE. 198a 155
A.
A high degree of cooperation was demonstrated between Eldorado, the regulatory agencies, consultants, analytical labs and the public.
5.
The "design driven" approach was successful in meeting the cleanup objectives, and keeping costs under control.
6.
Improvements in the environmental impact of the waste site itself were also realized. The covering of the waste site with fill has reduced on-site and fenceline gamma radiation levels and has also decreased the quantity of contaminants leaching from the site requiring treatment. Effluent discharges from the facility have been steadily improving in quality since the completion of the cleanup program.
15fl CNS 9th ANNUAL CONFERENCE, 1988
!'Koa;ssi.\"(; OF IJ.RW ARISING I-'ROM AECI, NUCI.KAK KKSKARCH CICNTRKS
1..I1.
Bl'CKl.KY, V . T .
N.V.
A t o m i c KnL 1 r j ; y o f ( ' a n a d a
(JKAMKR
l.im f t t ' d
Cha I k Kivt-T Nuc I car Laboratories Chalk RivtT, O n t a r i o , KOJ 1,10
-
W . I ' . IJROWN a n d K . A . H K L B R E C H T
At omit- Knurgy of Canada Mini ted Whitushel 1 Nuc 1 car Rt>search ICstabl i Pinawa, Manitoba, ROE 11.0
ABSTRACT
WASTE PROCESSING AT CRNL
Operation of nuclear research reactors and laboratories resul' in the generation of a wide variety of solid and w^uid radioactive wastes. This paper describes practical experience with processing of low-level radioactive wastes at two major nuclear research centres in Canada.
Waste Treatment Centre (WTC)
INTRODUCTION
Chalk River Nuclear Laboratories (CRNL) and Whiteshell Nuclear Research Establishment (WNRE) are two nuclear research centres operated by Atomic Energy of Canada Limited (AECL). Both research centres have been routinely processing a broad range of low-level radioactive wastes (LLRW) generated on-site, arising from the operation of research reactors, laboratories and various experimental facilities. A comprehensive program is required to manage these radioactive wastes to assure protect on of the public and the environment. To date, except for some wastes containing only short-lived radionuclides or very low concentrations which are diluted and dispersed, most of the wastes are stored. Routine radiological monitoring is done at waste management facilities and surrounding areas and at the property boundaries to ensure that environmental protection standards are maintained. Although storage of LLRW provides protection of the public and the environment at an acceptable cost, the need for surveillance, maintenance and land-use control to be continued for perhaps hundreds of years imposes responsibility and costs on future generations. To remove these future responsibi11 ties, a program Is underway to provide a transition from the current storage mode to permanent disposal. Thus wastes from both current and future operations will remain satisfactorily isolated as long as they remain hazardous [1-4]• Advanced waste treatment processes have been developed and brought into operation to improve waste processing so that the condit ioned wastes can be easily retrieved from storage and transferred to the ultimate disposal facility when a repository becomes available. Various solid and liquid wastes are volume reduced, immobillzed In stable matrices and/or properly packaged to faciIItate safe handling during current interim storage and future permanent disposal. This paper presents a review of the current waste processing practices to handle LLRW arising from the two AECL nuclear research centres. New treatment facilities being constructed and brought Into operation will also be briefly discussed.
AECL has constructed a waste treatment centre (WTC) to demonstrate volume reduction and immobilization techniques on a commercial scale and to improve management of CRNL wastes. The choice of processes, equipment and materials for use in the WTC was made after an extensive R&D program which included boCh laboratory and pi lot-plant tests [5-6]. The WTC integrates several processes to convert much of the waste generated at CRNL into a stable, compact, leach-res 1stant form which is suited for both storage and disposal. The basic flowsheet of the WTC is shown in Figure 1. It comprises: a control led-air incinerator for combustible solid and liquid wastes; a baler for non-incinerable solid wastes; membrane filtration, reverse-osmosis and evaporator systems for dilute aqueous wastes, and a ribbon-blender and a wiped-film bitumlnizer to immobilize the incinerator ash and liquid waste concentrates. The Incinerator and baler have been In service since 1982. The bituminlzation systems are ready for operation, but, as a result of equipment difficulties discussed below, the liquid treatment system is still being commissioned.
NOMNCMRASLE 5 « . C WASH
MCMERAftLC SOLS) AND I H U D WASTE
BASK F10WSHCET S U
FIGURE 1 :
TREATfi£r0
C£NTRE
FLOWSHEET OF CRNL WASTE TREATMENT CENTRE (WTC)
CNS 9th ANNUAL CONFERENCE, 1988 157
Solid Waste Treatment
Waste Segregation. Solid I.LRW at CRNL is segregated at source into incinerabie, nonincinerable and nonprocess ible categories. Specially marked polyethylene bags placed in containers of different colours and labels are provided at various waste collection points to facilitate waste segregation. Incinerable wastes are composed of combust ible materials, ideally free of items which might accelerate corrosion or interfere with safe operation of the incinerator. Materials having high chlorine-content such as PVC gloves and plastic suits generate highly corrosive HC1 gas if burned, and thus are disposed of in nonincinerable waste containers for compaction. Glass and metal objects are classified as non-processible wastes because they could form slag in the incinerator, interfere with the ash removal operation and could not be significantly volume reduced in the baler.
Wastes suspected of being contaminated with radioiodines and/or significant quantities of tritium are also currently not processed to minimize contamination of operators and the environment. The processtble waste bags, when received at the WTC, are monitored for radioactivity and inspected visually, but are not opened so as to minimize contamination. The current distribution of incinerable and nonincinerable wastes is 70% and 30%, respectively.
The inc
TABLE 1: PERFORMANCE DATA OF CRNL INCINERATOR
Year:
1982
1983
1984
Quantity: (m 3 ) (MS)
540
720 67
790
84
920 95
740 74
45
57
62
75
53
1070
1170
1350
1270
1400
No. of b u r n s Incineration. The incinerable waste category has been burned in a batch-loaded, two-stage, starved-air incinerator. The system (shown in Figure 2) consists of a vertical stainless-steel primary chamber, a horizontal refractory-lined afterburner and a dry flue gas treatment systemThe incinerator was designed to burn 10 m or about 1000 kg of solid waste in every 24 hour burn cycle. A more detailed descript ion of the incinerator design and operation has been presented elsewhere [7-9].
"«> f fl—J «s«
Average c h a r g e (kg/burn)
19B5 1986 1987
720 71
1620
• The incinerator has consistently produced a fully satisfactory inert ash product with an average volume reduction factor (VRF) of about 150:1 and a weight reduction factor of about 30:1. Fine ash is the most abundant material, constituting about 75% of the bottom ash product. It is of consistent q u a l i t y , free flowing and containing 1-3 wt% of fixed carbon. The contact gamma radiation field measured on an ash drum averages 3.5 mGy/h*, and rarely exceeds 10 mGy/h. • Since the system normalLy operates at slight negative pressure, a c t i v i t y containment is high. The r e l a t i v e l y low flow rate of the primary combustion a i r limits disturbance of the waste bed, resulting in low flyash carryover. The baghouse and roughing f i l t e r s have provided effective removal of the flyash particlesThe two downstream HEPA c i l t e r s , in service six years since s t a r t u p , did not require replacement until 1988 March when one of them fa 1 led the routine OOP f i l t e r efficiency t e s t . Particulate beta-gamma stack releases remained negligible, i . e . <37 kBq** per burn. Tritium releases averaged about 200 CiBq per burn and were normally less than 370 CBq per burn. • The starved-air combustLon process is able to cope with waste items having a wide variety of physica1 forms. A waste-segregatton-at-source program Is required, but no waste pratreatment is necessary*
SCHEMATIC DIAGRAM OF CRNL STARVKD-AIR RADWASTE INCINERATOR
1SB CNS 1th ANNUAL CONFERENCE, 1988
* **
10 mGy/h - I rad/h 37 kBq - L uCi
Segregated LLRW from a nuclear power demonstration (NPD) reactor has been routinely mixed and burned with CRNL waste without difficulty. The incinerator is also able to handle radioactive ol Is and solvents. A system to convert the afterburner Into a dual-fuel arrangement to burn some stockpiled organic liquid wastes has been installed and commissioned• Burning of 1iqutd scintillation fluids by Including the vials in the regular waste charge has also been successfully demonstrated. Tests with bales of both wet and dry incinerable waste were burned to demonstrate the ability ro satisfactorily burn compacted waste as part of total incinerator charge. This practice has been effectively used to deal with wastes accumulated during incinerator shutdown periods for repair and maintenance; routine incineration of baled waste has also been adopted as a cost' saving measure.
The baler has been operating with a VRF greater than 6:1, producing waste packages of about 0.45 m size. Table 2 provides some statistics on the performance of the baler. The average weight of a bale is 290 kg. The contact radiation fields from waste bales vary widely from 20 nGy/h to 20 mGy/hThe distribution was found to be log-normal wlt!i the average field of 250
• The air-to-air heat exchanger is the most critical piece of equipment because it has been susceptible to corrosion and deposition problems under high temperature cyclic operating conditions. Heat exchanger tube damage occurring at the hot inlet end is in the form of severe corrosion, embrittlement and impingement attack. Many heat exchanger tubes have failed and are progressively being replaced. A hard deposit forms rapidly on the inside of the tubes, mostly at the inlet end, and is difficult to remove completely, resulting in frequent tube cleanings. In addition to being used to treat site waste, the incinerator has been used as a development tool in a waste characterization program to estimate quantity and radiological characteristics of solid LLRW going to disposal * The incinerator ash is an ideal waste form which can be sampled and analyzed destructively to verify the data produced by a nondestructive waste monitor which measures gamma activity of Individual waste bags fed to the Incinerator. The incinerator also continues to be a teat bed to evaluate the performance of high temperature corrosion resistant alloys in the incinerator off-gas environment.
Bal.. ..h . The non -incinerable wastes, const i tut ing about 30% of the processible wastes, have been compacted in a 50-ton baler. The baler is basically a single-chamber downstroV 3 hydraulic baler, equipped with a hydraulic system to compact the waste and a hydraulic system to eject the completed bale. The baler is a relatively compact unit, having the overall dimensions: 3.2 x 1.8 x 1.7 m. It Is equipped with a ventilation system to prevent airborne contamination during baling operation. The baler is operated under a slight negative pressure and exhausted through HEPA filters to the bull ling ventilation system. Since more than 90% of the waste contains very little activity (less than 10 uGy/h on contact with a 40-11tre waste bag), baling operations such as emplacement of plastic liner, waBte loading and bale strapping are carried nut manually. Waste bags are loaded Into the compact Ion chamber without being opened. On average, It takes about 2 hours to complete one bale which contains about 70 bags of waste. The number of bags that can be put Into a bale vary according to the extent to which the bags are f M led and the rompactabi llty of the contents. Each bale Is strapped before being ejected from the compact Ion chamber. It la transferred by a two-tong Lifter to a reusable sheet metal box for shipping to the waste Htnragp area outfilde the plant-
TABLE 2:
PERFORMANCE DATA OF CRNL BALER
1982
1983
1984
1985
1986
1987
Quantity: (m 3 ) (Mg)
780 64
565 50
580 53
415 38
?80 34
360 35
No. of bales
222
173
184
128
110
122
The baling operation is quite simple, and has been virtually maintenance free. The only routine maintenance is to decontaminate the compaction chamuer by rinsing it with a detergent solution, carried out mostly for sanitary purposes. Occasionally wet wastes are compacted and the liquid released during compaction is orained and col leeted for processing In the WTC liquid waste treatment system. The baler has also been used routinely to compact a fraction of incinerable waste. Metal straps are replaced by plastic straps to facilitate the burning of baled waste in the incinerator. The baling/incineration combination has proved to be especially effective in deal Ing with incinerab.le wastes accumulated during maintenance shutdowns of the incinerator.
Shredding. The WTC has also been used as a test ground to develop treatment technology for wastes which currently are not processed. Shredding in combination with compaction can provide an effective volume reduction treatment for bulky nonprocessIble waste items. The shredding/compact ion combination may also provide a means to repackage wastes that have to be retrieved from existing storage facilities for disposal in an engineered facility. A slow-speed and high-torque shredder was purchased for the evaluation program. Thla type of shredder was chosen for study because: (i) slow-speed shredding (less thar 100 rpa) produces very little dust, an important conRide ration when working with active wastes; and (ii) high-torque allows shredding of most materials. Tne shredder was commissioned with inactive waste and then tested with noninclnerable and incinerable wastes to study the impact of shredding on compaction and incineration of LLKW. The operation of the shredder was Interrupted repeatedly due to knife failures despite changes made in the arrangement, configuration and type of shredder knives. It was identified that the knives broke from either impact with large unshreddable metal objects or from small quench cracks present In the knives. The cracks were likely caused by an improper case hardening heat treatment during the knife manufacturing process. To this point In time, the limited operating data indicate that;
CNS 9th ANNUAL CONFERENCE, 1988 159
clii? shredded waste can be compacted to a higher density than unshredded waste, resulting in an increase of up to 5CK in the overall volume reduction factor; shredding appears to have little Impact on the incinerator performance in terms of burning rate or ash quality; and waste shredding leads to more waste handling, and in turn, more exposure; thus, there is a need to develop techniques to minimize hands-on operation and reduce raan-rem exposure.
Liquid Waste Treatment SEMI-PERMEABLE MCMMANE
Low-Level Radioactive Aqueous Wastes. During the last 35 years, approximately 20 000 m of low-level aqueous wastes have been discharged annually into seepage pits in a sand knoll at CRNL. Extensive monitoring combined with environmental research has been underway for many years in the vicinity of the sand pits. Although the sand has only modest retention capacity for dissolved radtonuclides, only a small fraction of the radionuclides has migrated beyond the knoll. Although radioactivity releases via groundwater streams which drain Lhe area have remained below 0.1% of the Derived Release Limit*** for CRNL, direct discharges of liquid wastes to the soil are to be reduced to small values within the next few years through the use of waste treatment processes Installed at th£ WTC. The low level liquid waste treatment process at the WTC originally consisted of three interconnected recirculatlng membrane separation systems as shown in Figure 3 [11). This membfane process was designed to concentrate radioactivity and other impurities into a small volume, and discharge the bulk of the treated liquid into the Ottawa River. Liquid waste was to be fed to an ultrafiltration (UF) system where it was to be filtered. Water passing through the UF membrane, called permeate, which would be free of suspended solids but still contain all the dissolved iL.'terial, was to be passed into a spiral-wound reverse osmosis (SWRO) system. The pertfeate from the SWRO system will be low in b^th suspended and dissolved radioactivity. It will be stored, sampled, and, if below established limits, discharged to the Ottawa River. The relatively small volumes of concentrates from both the UF unit and the SWRO unit will be transferred to the concentrate stage. This system consists of tubular reverse osmosis (TRO) membrane modules because it must handle a slurry and have a reasonable efficiency for removing dissolved salts. The TRO permeate will contain considerable radioactivity, since the feed concentration will be high at this poin-. sid is recycled to the SWRO system for reprocessing. The TRO concentrate is sent to a wip.'d-film evaporator for bitumlnization.
***The Derived Release Limit (DRL) is defined as the upper limit for the release rate of a single radionucllde from a single source which is derived from the regulatory dose -uivalent limits by analytical models of all signi* jnt environmental pathways to an individual in the most heavily exposed group. In deriving the DRL, the intention is to establish a release limit such fhat acJTierence to it will provide virtual certainty of compliance with the recommendations of the International Commission on Radiological Protection [10).
160 CNS 9th ANNUAL CONFERENCE. 1988
IMMOBILIZATION
FIGURE 3:
FLOWSHEET OF THE LIQUID WASTE TREATMENT PROCESS OF W E WTC
Commissioning of the UF/RO plant has revealed generic design deficiencies in the original TRO and UF membrane modules. Membrane ruptures were encountered with the TRO units. The tubular membranes which were inserted into the supporting tubes could be easily separated from those tubes during shutdown periods and torn when they wei;e re-pressurized. On the other hand, the deficiency found in the original UF modules was the poor design of end seals which function to prevent the pressurized feed water from contaminating the permeate water. In addition to these design deficiencies, the costs to purchase replacement membranes for the original TRO and UF modules were also found excessive. These units were therefore replaced by membrane modules which are not only of better designs but also cheaper. The original TRO modules were replaced by modules which have membrane tubes much more resistant to ruptures because they are smaller in diameter and are bonded strongly onto a rigid siupport tube. The UF system has been replaced by a compact microfiltration system, equipped with 0.2 urn hollow-fibre modules which can be frequently cleaned by backwashing with air. The changes, however, involved effort in searching for and evaluating alternatives available on the market as well as in making modifications to the original systems. These changes therefore resulted in considerable delays in the commissioning program of the RO plant. Commissioning with inactive water has been completed, and a limited quantity of liquid waste (about 130 m ) has been treated during active commissioning.
Other Liquid Wastes. CRNL is currently storing small volumes of medium and high level liquid radioactive wastes. These wastes originated from ion exchange regeneration operations, spent fuel reprocessing carried out In the 1950s, and medical isotope production. They require treatment to convert Into a stable form for disposal. Some liquid wastes generated from the production of Ho-99 and Xe-133 isotopes
cnntaLn enriched uranium which CRNL has determined can be economically recovered. CRNL has prepared design studies of a facility, known as the Plant for Active Liquid Wastes (PAWL), which would consist of a batch vitrification process to Immobilize high-level liquid wastes into sodium borosilicate glass, and a uranium recovery process to recover enriched uranium from liquid wastes resulting from medical isotope production [12] .
WASTE PROCESSING AT WNRE
At WNRE, a research centre approximately half the size of the Chalk River facility, less wastes are generated and processedSolid wastes are compacted; low-level aqueous wastes are discharged directly to the Winnipeg River after being sampled and analyzed; intermediate-level aqueous wastes are concentrated tor immobilization of radioactivity in polyester resin; and organic liquid wastes ar^ Incinerated.
Drum compaction of Intermediate-Level Waste. A commercial 110 L drum ram compactor has been tiist? 1 led for remote operation In the enclosed decontamination area associated with the Hot Cells Facility (HCF) (14). The HCF provides shielded facilities for postirradiation examination of fuels and reactor core components, spent fuel management services, and remote handling services for other WNRE experimental programs. Tests have shown that garbage cans containing the compactible fraction of solid intermediate-level radioactive waste can be compacted by a ratio of 4:1 inside a standard 110 L drum. An average of four garbage cans will be contained in a drum. The crumpled garbage cans provide restraint against sprlngback of the compacted waste 'which has been a problem with this type of compaction processThe drum of compacted waste is capped with a sheet metal lid using a remotely operated lid crimper- The filled drum is then transported to a partially abovegrade concrete storage bunker in a bottom unloading lead shielded flask. Commissioning of the compaction, transportation and storage facilities has recently been completed and the system is now in service.
Solid Waste Treatment Liquid Waste Treatment Baling of Low-Level Wast£S. Low-level wastes were originally stored in shallow earth trenches, but since L986 they have been baled and stored in aboveground storage bunkers• A smal1 segregated fraction of the wastes, about 10%, which contains glass and metal objects is not processed but is stored in the same structures. Solid low-level radioactive waste is compacted using a baler identical to th^ unit at CRNL. The baler processes about 250 m fa of solid waste, achieving a VRF of 5:1 on average. Since the waste generation rate is relatively small, the waste bags collected are not processed immediately. They are accumulated in the low-level waste processing building which also contains the baler housed in a separate heated room. A baling campaign is carried out when sufficient waste has been collected, at a frequency of once every 4-6 weeks. Because the storage area in the building is unheated, the presence of frozen mopheads in the stored waste causes a large drop in the VRF during the winter. On average, the VRF achieved in the winter is only 4:1 while the VRF in the summer is about 7:1. The baler ventilation exhaust is discharged through a seven-metre high stack equipped with an absolute filtration system which contains both roughing and HEPA filters. The total radioactivity released annually to the atmosphere from the baler operation has been about 20 kBq based on a total fly analysis of a proportional stack gas sample [13]. A plan will be developed to segregate wastes Into Incinerable and nonincinerable fractions if it is decided to ship bales of incinerable wastes to CRNL for incineration. The compaction process at WNRE has permitted economical storage of low-level radioactive waste in aboveground engineered facilities and eliminated the need for trench storageThe experience to date in using the process has been similar to CRNL^s-
Low- and intermediate-level aqueous wastes at WNRE are handled in an Active Liquid Waste Treatment Centre (ALWTC). Low-level Aqueous WasteLow-level aqueous waste solutions from various buildings (e.g., laundry, decontamination, R&D, reactor) are collected and stored separately in holding tanks. The underground pi ping system to transfer waste from various sources to the holding tanks has recently teen replaced with a doubled-wall containment system. The plast ic transfer pipes for each source are contained in larger plastic pipes which are sloped to monitored leak collection sumps. Spare lines in the containment pipes can be used to replace leaking lines. When a tank is filled, its content is mixed, sampled, and analyzed before being discharged at a controlled rate into the process water effluent to the Winnipeg River. The effluent from the process drain outfall where it enters the river is continuously and proportionally sampled. Daily and weekly samples are routinely analyzed to confirm that release limits are not exceeded [15] . Table 3 summarizes volumes and activity of lowlevel aqueous waste discharged to the Winnipeg River. The average waste volume is about 4500 m /a, and the total radionuclide content released annually is about 70 GBq, which corresponds to less than 0-1% of the Derived Released Limit [16].
TABLE 3 :
1983 1984 1985 1986 1987
DISCHARGE OF LOW-LEVEL AQUEOUS WASTE AT WNRE _ _ _ _
Waste Volume (m 3 )
Total Isotope (GBq)
%DRL
6300 5000 4600 3300 J400
123 107 51 34 24
0.14 0.11 0.0? 0.04 0.03
CNS 9th ANNUAL CONFERENCE. 1988 161
Intermediate-Level Aqueous Waste. The intermediatelevel liquid waste collected from the Hot Cells represents less than 1% of the volume of liquid waste generated at WNRE but it contains more than 90% of the radioactivityThis waste contains about 300 GBq/m total fly, 40 g/m of total heavy elements and some fissile materials (<1 g/m 3 of u-233 + U-235 + Pu). A set oF two parallel climbing-film evaporators, each with a design evaporation rpte of 10 L/h, has been used to concentrate the intermediate-level aqueous waste. An average volume of 12 m per year is processed through the system (15). Concentration factors of about 10 and decontamination factors greater than 10 000 are routinely achieved. The concentrated liquid waste is immobilized in waterextendable polyester resin and packaged in 110 L mild steel drums. The decontaminated condensate is suitable for release thr.-j.gh the low-level waste holding tank system. The operation of the climbing-film evaporators has been plagued with fouling, plugging and entrainment problems. Fouling of the evaporator tubes has been dealt with by filtration of the teed solution to remove particulates, by pH control to prevent precipitation of salts In the feed solution, and by tube repljcement after processing 4000-5000 L of waste solution. Plugging of the concentrate return line from the steam/cnncentrate separator resulted in a poor decontamination performance. This problem was solved by increasing the pipe size of the return line. On the other hand, the evaporators have been assisted by natural evaporation occurring in the waste holding tank. The rate of water loss by this phenomenon is 3000-5000 L/a. Routine sampling of the building ventilation system confirms that the high evaporation rate does not result in significant release of radioactivity from the facility.
separate campaigns, and volumes of concentrated liquid produced by evaporation in one year may be carried over to be solidified in the next year. An average overall VRF of 3:1 has been achieved for the intermediate-level aqueous wastes, including all the waste volumes generated by the process such as tank and line flushing to remove particulate buildup.
TABLE 4:
TREATMENT OF INTERMEDIATE-LEVEL AQUEOUS WASTE AT WNRE EVAPORATION Volume Processed (m 3 )
1933 1984 1985 1986 1987
26.2 0.0 12.2 12.6
SOLIDIFICATION Volume Processed (m 3 ) 2.26 3.73 0.42 0.94 2.06
Number of Solidified Waste Drums 38 80 7 16 34
Organic Waste. An Industrial incinerator, equipped with a propane vortex burner, is used to burn spent organic liquid coolant which arose from the operation of WR-1, an organic-cooled research reactor which has been shutdown since raid 1985 for decommissioning. Small volumes of noncarcinogenic laboratory solvents and hydraulic oils free of pollutants such as nitrates, sulphates and halogens are also burnt In this incinerator. The designed operating temperature and residence time are 1000aC and 1 second, respectively. These conditions have been confirmed by tests. Gaseous effluents are discharged directly throug.i a 12-m high firebrick-lined stack. The incinerator has been burning about 40 m /a of organic liquid at a rate of up to 75 L/h [17j.
Solidification of the concentrated liquid in polyester resin has been performed smoothly without any incident. It is carried out batchwise in separate campaigns. When sufficient concentrated liquid waste has been collected for a solidification campaign, a representative sample is taken and a 0.2 L test block is made in the laboratory. If the test solidification is satisfactory, the solidification operation can begin. The drum mixer is a multicomponent assembly that Is used to: (i) retain the waste drum, (ii) provide ventilation, waste and catalyst addition connections, and (ill) provide high speed mixing for an emulsion of 60% waste in 40% resin. Solidification is carried out in-situ in a standard H 0 - L mild steel drum which has a large and a small bung opening on the drum top. The mixing blade is designed to allow Installation through the large bung opening. The blade is formed inside the drum by hydraulic actuation of the mixer shaft. After a catalyst, a methylethylketone peroxide solution, is added to promote resin polymerization, and after a short mixing period, the mixer blade and shaft are released from the drive head and fall into the waste drum. The volume of the solidified product is nearly twice that of the concentrated liquid waste.
For organic coolant wastes containing more than 15 Bq/mL, a distillation process is used to yield two products: a light product (about 60% of the feed volume) which Is decontaminated to less than 15 Bq/mL, permitting Its incineration; and, a heavy product containing most of the radioactivity which becomes a solid suitable for storage when cooled to room temperature.
Table 4 summarizes the waste volumes processed by the evaporation and solidification processes tn the last five years. It is worth pointing out that these two operations are carried out Independently In
Other Liquid Wastes. WNRE Is currently storing small volumes of medium and high level liquid radioactive wasteB resulting from a uranium recovery process and
162 CNS ath ANNUAL CONFERENCE, 1988
The incinerator operates without any off-gas treatment, but the radioactivity emission control is exerted by limiting the feed radioactivity to less than 15 Bq/mL. The stack gas Is continuously monitored for radioactive particulate releases during combustion operation. A proportional sample is run through a charcoal-impregnated paper filter that is removed and analyzed on a routine basis to monitor stack releases. The emission of radioactivity has been consistently less than 0.0052! of the derived release limit f..r the incinerator (16). The calculation conservatively assumes that all Che radioactivity present is due to Sr-90, which is the most restrictive isotope of those present, or potentially present, for a member of the critical off-site group.
discontinued fuel reprocessing experiments. These liquids will be immobilized with a suitable process to provide continued safe storage in a solid form.
[8]
BEAMER, N.V., "Experience with Low-Level Waste Incineration at Chalk River Nuclear Laboratories", AECL-8187, Atomic Energy of Canada Limited (1984).
CONCLUSIONS
[9]
LE, V.T., BEAMER, N.V. and BUCKLEY, L.P., "Experience with Radioactive Waste Incineration at Chalk River Nuclear Laboratories", presented at the Int- Conf. on Incineration of Hazardous, Radioactive and Mixed Wastes, May 1988, San Francisco, California (1988).
Atomic Energy of Canada Limited has been processing a broad range of low-level radioactive wastes at i t s nuclear research establishments. Its extensive background experience in the management of radioactive wastes and its continued effort to develop, improve and apply advanced waste processing techniques wtll contribute to a successful transition from interim storage to permanent disposal.
[10] PALMER, J.F., "Derived Release Limits (DRL's) for Airborne and Liquid Effluents from the Chalk River Nuclear Laboratories During Normal Operations", AECL-7243, Atomic Energy of Canada Limited (1981).
REFERENCES
[1]
CHARLESWORTH, D.H., "AECL's Activities In the Management of Low-Level Radioactive Wastes in Canada", AECL-9178, Atomic Energy of Canada Limited (1986).
(2)
DIXON, D.F. (Ed.), "A Program for Evolution from Storage to Disposal of Radioactive Wastes at CRNL", AECL-7083, Atonic Energy of Canada Limited (1985).
[3J
[4]
CHARLESWORTH, D.H., "Waste Management Activities at Chalk River Nuclear Laboratories", Proc. of the 6th Annual Participants' InfoMeeting - DOE Low-Level Waste Management Program, Denver, Colorado, 1984 Sept-, pp. 59-65 (1981). CHARLESWORTH, D.H., "Current Development Programs for the Disposal of Lnw- and Intermediate-Level Radioactive Wastes", AECL-6545, Atomic Energy of Canada Limited (1979).
[5]
BOURNS, W.T., BUCKLEY, L.F. and BURRILL, K.A. , "Development n( Techniques for Radwaste Systems in CANDU Power Stations", AECL-6534, Atomic Energy of Canada Limited (1979).
[6]
CHARLESWORTH, D.H., "The Canadian Development Program for Conditioning CANDU Reactor Wastes for Disposal", AECL-6344, Atomic Energy of Canada Limited (1978).
[7]
BEAMER, N.V., "Radwaste Incineration at CRNL", AECL-7437, Atomic Energy of Canada Limited (1981).
(11) BOURNS, W.T. and LE, V.T., "The Reverse Osmosis Plant in the CRNL Waste Treatment Centre - Description, Design and Operating Principles", Unpublished Report, CRNL-2352 (1984). [12J BURRILL, K.A., "Immobilizing Isotope Production Waste in Glass", AECL-8481, Atomic Energy of Canada Limited (1984). [13] HELBRECHT, R.A., "A Report to the Nuclear Safety Advisory Committee on the Operation of the WNRE Waste Management Area During 1986", Unpublished Report, WNRE-742 (1987). [14] CHARLESWORTH, D.H. (Ed.), "Status Report for the Period Ending 1986 September 30", Unpublished Report, TR-360-1 (1986). [15] BROWN, W.P., "WNRE Active Liquid Waste Treatment Centre Safety Analysis Report", Unpublished Repor1., WNRE-221 (1981). [16] WUSCHKE, D.H. and DUNFORD, W.E., "Derived Release Limits for Radtonuclides in Airborne and Liquid Effluents for the Whiteshell Nuclear Research Establishments", Unpublished Report, WNRE-406 (1982). [17]
HELBRECHT, R.A., "WNRE Waste Management Area Safety Analysis Report", Unpublished Report, WNRE-327 (1985).
NOTE: The unpublished reports referenced above are not formal publications, but may be obtained from Atomic Energy of Canada Limited, Chalk River Nuclear Laboratories, Chalk River, Ontario, Canada KOJ 1J0.
CNS 9th ANNUAL CONFERENCE. 1988 163
CHEMISTRY RESEARCH FOR THE CANADIAN NUCLEAR FUEL WASTE MANAGEMENT PROGRAM
A.C. VIKIS, F. GARISTO, R.J. LEMIRE, J. PAQUETTE, N.H. SAGERT, P.P.S. SALUJA, S. SUNDER and P. TAYLOR
Reseaich Chemistry Branch Uhiteshell Nuclear Research Establishment Atomic Energy of Canada Limited Pinava, Manitoba, ROE 1L0
ABSTRACT This publication reviews chemical research in support of the Canadian Nuclear Fuel Waste Management Program. The overall objective of this research is to develop the fundamental understanding required to demonstrate the suitability of vaste immobilization media and processes, and to develop the chemical information required to predict the long-term behaviour of radionuclides in the geosphere after the waste form and the various engineered barriers containing it have failed. Key studies towards the above objective include experimental and theoretical studies of uranium dioxiJe oxidation/dissolution; compilation of thermodynamic databases and an experimental program to determine unavailable thermodynamic data; studies of hydrothermal alteration of minerals and radionuclide interactions with such minerals; and a study examining actinide colloid formation, as well as sorption of actinides on groundwater colloids.
INTRODUCTION This publication reviews chemical research carried out in the Research Chemistry Branch of Atomic Energy of Canada Limited to support the Canadian Njclear Fuel Waste Management Program (CNFUMP). The overall objective of this research ' •* to develop tne fundamental understanding required to demonstrate the suitability of the vaste form, and to develop the chemical information required to predict the long-term behaviour of radionuclides in the geosphere after the waste form and the various engineered barriers containing it have failed. As it is now envisaged in the CNFUMP, used-fuel bundles will be sealed in corrosion-resistant containers (1). The containers will then be placed in a waste vault excavated deep underground in stable plutonic rock. The used fuel containers vill be surrounded by a clay-based buffer to minimize groundwater ingress, and the vault, once filled, will be backfilled and permanently sealed. Iti the long term, after failure of the containers, lelease of radionuclides contained in the used fuel would depend on the rate of dissolution of the U 0 ; matrix in groundwater. Thus, we are conducting experimental and theoretical studies to determine the U02 oxidation/dissolution mechanisms as a function of pH, redox potential, temperature, radiation fields, and groundwater anion concentrations. Furthermore, the solubility in groundwater of the various radionuclides contained in used fuel, and the transport of radionuclides through the geosphere, depend on chemical speciation. Thus, thermodynatnic data are required to determine the Important species
164 CNS 9th ANNUAL CONFERENCE, 1986
and their concentrations in groundwaters of various ionic strengths. For this purpose, we are compiling thermodynatnic databases for Key radionuclides (U, Pu, Np, Cs, Sr, Tc, I) and groundwater ions (Na*, K \ Mg 2 *, Ca 2 *, Cl", SOj"), and experimental programs are in place to determine unavailable data and to improve the accuracy of others. Also, the transport of radionuclides through rock fissures would depend on their interaction (sorption, chemical reaction) with the hydrothermally altered rock surface. Thus, we have in place a research program to study hydrothermal alteration reactions of common minerals and to define the interactions of radionuclides with mineral surfaces. Finally, in addition to transport of radionuclides as dissolved species, transport could occur either by sorption of radionuclides on existing groundwater colloids or by transport of pure radionuclide colloids. The latter can form, for example, by the precipitation of dissolved radionuclides on entering a lower-solubility regime. Thus, the sorption of uranium (VI) on magnetite and hematite colloids and the formation of uranium colloids, on reduction of dissolved uranium (VI), have been studied in detail as model systems representative of the behaviour of other actinides. For brevity reasons, only the more recent highlights of work in this area are summarized below. A previous review (2) summarizes earlier work, and, of course, specific publications quoted in this and the previous review contain additional details.
CHEMISTRY OF URANIUM DIOXIDE
Uranium Dioxide Oxidative Dissolution Electrochemical techniques and X-ray Photoelectron Spectroscopy (XPS) are being used to study the oxidative dissolution of U0 2 . Earlier vork (3-9) established the mechanism of oxidative dissolution of (JG2, at room temperature, as a function of pH, Eh, and groundwater anions (POj-, SOjj- and CO|-). It has been established in those studies that the redox potential is the most important parameter affecting \)02 dissolution. Specifically, for reducing conditions (< -100 mV vs SCE), U0, undergoes very little dissolution, and significant dissolution occurs only when V02 is oxidized beyond the U0 2 33 (U 3 O 7 ) stage. Also, it has been established that the uranyl ion U0|* is the active intermediate in the oxidative dissolution of uranium dioxide, and that COf", and other anions that form strong complexes with U0* 4 enhance U0 2 dissolution. Acidic conditions also accelerate U0, dissolution.
Although the disposal vault environment is expected to be reducing, radiolysis of groundwater by radiation emanating from used fuel may modify the redox chemistry of the system. In general, radiolysis of water by alpha, beta or gamma radiation produces various molecular species (0?, hV,, B20?) and free radical intermediates (OH, H, Oj, e " ) . It is thus necessary to ascertain the extent to which radiolysis products modify the oxidativi? dissolution of UOj, and a program is in place to address this (10-12).
C
C
c c
Lu
1/1
We found that hydrogen has no effect on the oxidative dissolution of UOj, except perhaps at temperatures > 100°C, in which case it appears to suppress oxidative dissolution. Oxygen, at a concentration of 10"' mol-dnr 3 and pH = 9, has no effect on U0 2 dissolution. However, in air or oxygen saturated solutions (102l - 1CM to 10" 3 mol-dnr 3 ) at pH > 5, U0 2 dissolution proceeds by the formation o£ a film of UO 2 .j, which is slowly converted to hydrated U0 3 or, in the presence of complexing ions, to UOj* complexes (11,13). Hydrogen peroxide (H 2 O 2 ) is about 200 times more effective than o 2 in oxidizing U0 2 in nearneutral (pH - 9) conditions (10). Our studies show that UOj oxidation in H 2 0 2 increases with increasing pH for pH < 6, is unaffected by pH in the range 6 to 10.5, and decreases with increasing pH for pH > 10.5. The oxidation of CTO2 (n the pg range 6 to 10, a range relevant to a waste vault in granitic environments, is significant when the H 2 0 z concentration is higher than 1O'J mol-dnr 3 (12). Electrochemical and XPS studies of U0 2 oxidative dissolution were also carried out in the presence of alpha- and gamma-radiation fields. We found that alpha radiolysis of water using radiation sources of alpha flux comparable to that expected at the fuel surface (- 200 kBq) at the tine of container failure (500 to 1000 years) did not oxidize U0 2 beyond the U0 2 3 3 stage. However, alpha fluxes greater than about 200 kBq lead to oxidation beyond the U0 2 33 stage. To assess the relative importance of radical intermediates generated by radiolysis of groundwater on the U0 2 oxidative dissolution, we developed an electrochemical system that enables studies in the presence of gamma radiation. Radicals such as OH, OJ and CO 3 were generated selectively t)y irradiating certain solutions as shown in Figure 1. According to corrosion potential measurements, the efficiency of these radicals in oxidizing U0, decreases in the order 0; > OH > CO;. Recently, we have begun to address the effect of temperature on U0 2 Qxidative dissolution, as the temperature o£ the vault could be as high as about 100°C, depending on vault depth and used fuel packing density. Studies of the redox potential at 55°C, pH 9.3 and 0.1 nol-dnr1 NaclO,, show that relative to room tempeiature the rate of dissolution appears to be higher (by a factor of about 2 ) , but the value of the redox potential beyond which U0 2 undergoes oxidative dissolution remains the same, i.e., - 100 mV vs. SCE.
Theoretical Analysis of Used Fuel Dissolution The safety assessment of used fuel disposal in an underground vault requires information on the rate of release of radionuclides from used fuel to the groundwater. Since more than 902 o£ the radionuclide inventory in used CANDU™ fuel is contained within the U0 2 grains, the major factor controlling the long-term release of radionuclides from used fuel is the rate of dissolution of the U0 2 matrix.
c c ^?-
z . -<*
•i -1.
Cr +
1 "•
: 1-
i 8. MOUI^S
1 ll.
i 16.
JO.
FIGURE 1- CORROSION POTENTIAL OF A UO 2 ELECTRODE IN THREE DIFFERENT SOLUTIONS UN'JERC-OING RADIOLYSIS: (A) 0 -SATURATED O.I MOL-DH"3 NaC10 4 + O.O1 MOL-DM"3 HCOONa ; (b) N 2 O-SATURATED 0.1 HOL-DM"3 NaCIO,, ' AND (C) NjO-SATURM'ED 0.1 MOL-DM"3 NaCIO, * 0.002 MOL-DH" S Na 2 C0 3 ,
Used-fuel dissolution rates are often determined using solubility-limited dissolution models (14). In these models, the rate of used-fuel dissolution is limited by the solubility of UO 2 , i.e., the dissolution rate decreases- as the concentration ot uranium in solution approaches C o , the U0 2 solubility. This is illustrated by the curves i,- Figure 2. The dissolution rate, which is proportional to - O c / 3 x ) x . 0 , decreases as T increases. At long timss, the used-fuel dissolution rate varies linearly wth the U0 2 solubility. The U0 2 solubility is an important parameter in solubility-limited fuel dissolution models. Thus, using thermodynamic principles, we have derived an explicit mathematical formula to calculate U0 2 solubilities as a function of temperature, pH, oxidation potential and anion concentrations (15). In addition, thermodynamlc reaction path calculations have been carried out to mod$l the dissolution of UO, (16). The qualitative agreement between these thermodynamic predictions and electrochemical experiments supports the use of thermodynamically derived source terms in the safety assessment of used-fuel disposal. More recently, we have begun to examine the impact of various phenomena, such as precipitation, on the rate of used fuel dissolution. The solubility-limited dissolution models that have been used to calculate used-fuel dissolution rates implicitly assume that the solubility of U0 2 is spatially invariant (14). However, the solubility of U0, could vary from location to location due to, for example, alpha-radiation-
CNS 9th ANNUAL CONFERENCE. 1986 t«5
induced redox potential giadlents, and precipitation of a uranium-containing solid could occur in the vault environment. After precipitation (at y - £ ), tl-e concentration profile of dissolved uranium is given by the dashed line in Figure 2. (In this example, the solubility of U0 2 is equal to Co for 0 < x < / and equal to 0 for x > / ) . This concentration profile remains constant as long as the solubility gradient, which caused the precipitation, does not change. Thus, after precipitation, -(3c/5x) x = 0 is a constant, whereas without precipitation -(3c/3x) r s 0 decreases with time. Therefore, precipitation enhances the rate of fuel dissolution. This effect is particularly important at long times (17).
FIGURE 2: CONCENTRATION PROFILES OF DISSOLVED URANIUM AS A FUNCTION OF THE DISTANCE FROM THE USED-FUEL/ BUFFER INTERFACE. FULL CURVES REPRESENT NONPRECIPITATION PROFILES AT VARIOUS TIMES T. THE DASHED LINE REPRESENTS THE CONCENTRATION PROFILE FOLLOWING PRECIPITATION OF A SPARINGLY SOLUBLE URANIUMCONTAINING SOLID, AT A DISTANCE i FROM THE FUEL/BUFFER INTERFACE
It should be pointed out lhat precipitation is not the only phenomenon that can increase the rate of fuel dissolution determined using solubility-limited dissolution models (IB). In fact, any phenomenon that decreases the concentration of uranium in solution, e.g., adsorp'ion or chemical reaction, would enhance the rate of fuel dissolution. The importance of all these phenomena in the vault environment must still be ascertained. For example, the magnitude and duration of any gradient in the U0 2 solubility must be determined before a realistic estimate of the effect of precipitation on used-fuel dissolution can be made using simple models.
dependent on the nature of the equilibrium chemical species. Vith good thermodynamic data, the important solids and solution species can be identified, and the behaviour of many of the actinides can be modelled. A critical assessment of chemical thermodynamic data for solids and aqueous species of uranium, neptunium and plutonium is continuing as part of the NEA/OECD Thermochemical Data Base pioject (19). For those elements most important to the CNFUMP, this project will provide internationally accepted, internally consistent chemical thermodynansic databases that are compatible with the CODATA key values (20). A new interim database for uranium has also been prepared (21) to update and supplement earlier databases (22,23) until the NEA/OECD database is available. For UO, to be a satisfactory waste form, it must be shown to be stable under reducing conditions for a range of groundwater compositions. Calculations have been done to show the effects of different ionic medi^(using activity coefficients and literature equilibrium constant data) on the solubility of uranium dioxide in model groundvaters (21). Three specific model groundwaters have been considered. Granite groundwater (GGU), which has a low ionic strength (I < 0.002 mol-kg' 1 ), Standard Canadian Shield Saline Solution (SCSSS), which is essentially a NaVCa 2 */Cl" 1 solution (I = 1.4 mol-kg" ) and SCSB/2, which is a high-chloride-concentration medium (I ^ 3.o mol-kg~ J ) in which again the predominant cations are Na* and Ca'*. Tne total carbonate concentration in GGW (1.0 x 10" ! mol-dnr') is higher than in SCSSS (1.6 x 10" 4 mol-dm- ! ) and SCSB/2 (1.0 x 10"* mol-dnr 3 ). In the latter two groundwaters the carbonate content is essentially limited by the solubility of CaCO,. Our calculations show that, in the absence of uranate formation, uranium solubilities in these three model groundwaters are similar for strongly reducing conditions under which U0 2 is the stable solid. For mildly reducing conditions (E = 0.50 - O.0592pH V) under which U.,0, is stable (Figure 3 a ) , the solubility also varies only slightly between the different model groundwaters (less than an order of magnitude within the pH range 6 to 11). This is coincidental. The stable solution species resulting from the dissolution of U\0 0 (or U,07) are primarily U(VI) species. The lower carbonate concentrations in SCSSS and SCSB/2 compensate for the increased stability of highly charged ionic U(VI) species, e.g., UO 3 (CO,)j", in the more saline groundwaters. As shown in Figure 3b, if GGU and SCSSS had the same lower total carbonate concentration (10"J mol'dm"3) as the SCSB/2. the calculated solubility of uranium in the more saline model groundwaters would be significantly higher (by as much as two to three orders of magnitude) than in the GCU. Actinide Solution Chemistry
THERMODYNAHIC DATA FOR KEY RADIONUCLIDES
Chemical Thermodynamic Solubilities of Aclinides
Databases
and
Calculated
A thermodynamic framework allovs the results of a moderate number of relatively simple experiments to be used T O estimate the solubility of U0 ? in groundwaters with widely different compositions and values of pH, temperature and electrochemical potential. The solubility can then be expressed as a function containing a limited number o£ parameters. Also, the transport behaviour of other aclinides in the vicinity of a nucJeai fuel waste vault m in the biosphere is
166 CNS 9th ANNUAL CONFERENCE, 1988
Work in this area is presently limited to the study of neptunium(V) chemistry. A knowledge of the aqueous solution chemistry of neptuniutn(V) is important in that it provides a model for the chemistry ot uranium(V) and plutonium(V). Uranium and plutonium in the " ( V ) " oxidation state disproportionate to the (IV) and (VI) oxidation states much more readily than neptunium(V). Therefore, the " ( V ) " oxidation state for uranium or plutonium is much less accessible experimentally than it is for neptunjum. Also, the predominant oxidation state of J''Np (flow nuclear fuel wastes) in solution in surface waters and mildly oxidising groundwaters would be neptunium(V) (23).
specific activity is low and its beta radiation weak, technetium-99 is considered a long-term hazard because of its long half-life and the large quantities involved. Moreover, the oxidized form of the element in water, TcO;, does not interact strongly with geologic materials. There are few chemical thermodynamic data for technetium(IV) species, although studies have indicated that technetium(IV) could be the stable form of the element in reducing groundwaters (28-30). Thus, the solubility of technetium(IV) oxides is being measured to develop a chemical thermodynamic database for technetium(IV) in order to model technetium behaviour in reducing groundvaters.
FIGURE 3: ACTIVITY EFFECTS ON URANIUM SOLUBILITY FOR MILDLY REDUCING CONDITIONS (E=0.5-0.592pH V, 25°C): GGW; SCSSS; SCSB/2. (a) CORRECTED FOR ACTIVITY EiTECTS; (B) AS (A) EXCEPT THE MODEL GROUNDt/ATERS HAVE BEEN CHANGED SO THE TOTAL CARBONATE J CONCENTRATION IN EACH IS 1Q- MOL-DM"3.
Dioxoneptunium(V) forms a series of anionic carbonate complexes that enhance neptunium solubility in neutral and moderately basin, aqueous solutions (2427). Calculations suggest that the stability field of the carbonate complexes may increase significantly at higher temperatures (23); however, there are essentially no data available for temperatures above 25~C. In order to determine how the formation constants of dioxoneptunium(V) carbonate complexes change with temperature, we have measured the 2J1 equilibrium concentrations of Np in aqueous carbonate solutions (1*1.0, NaCL0 4 ) over hydrated NaNp0 2 C0j and Na,Np0 2 (C0 3 ) 2 at 30, 50 and 75°C. At 30°C over NaNp0 2 C0j, a minimum solubility (~10~6 mol'dirr3) occurs in a plot of neptunium concentration versus free carbonate concentration at a free carbonate ion concentration of ~10" 3 mol'dnr 3 . This is similar to results at 25°C (2b). However, at higher temperatures the solubility minimum is less pronounced, and appears to be shifted to higher carbonate concentrations. It appears that above 25°C some conversion of hydrated NaNpO,CO3 to hydrated NajNpOj(COj)_, (and other carbonate-rich solids) occurs even at total carbonate concentrations considerably less than 0.1 mol'dm"3 ([Na*] = 1.0 rool•dnr3). This conversion is sluggish even at 75°C, but seems to occur more readily as the temperature is raised. Preliminary results indicate the formation constant for NpOjCOj increases slightly with increasing temperature.
We have measured the solubility of amorphous technetium dioxide obtained in situ by titration of an acid technetium(IV) solution with sodium hydroxide. We have also measured the solubility of crystalline technetium dioxide obtained by the thermal decomposition of ammonium pertechnetate. Figure U shows the results for amorphous TcO 2 as a function of pH. The solubility curve is typical of that of most transition metal oxides. The solubility is minimum in the pH range from 6 to 9. Figure 5 shows similar results for crystalline TcO 2 . Reliable data could not be obtained in acidic solutions due to the formation oi a wore insoluble compound with the acids used. Both forms of the oxide exhibit a similar behaviour as a function of pH, i.e., the solubility goes through a minimum of about 10~ a mol'dm"3 at near-neutral pH. More accurate data for the crystalline material would be required to assess reliably which of the two oxides is the thermodynamically stable form of technetium(IV).
- .3
Technetium-99 is produced in nuclear reactors with a fission yield of 6%. It is a long-lived beta emitter i/ith a half-life of 2.1 X 10 5 years. Although its
—r
1
1
1
1
1
1
-5 6 \
/
7
X
g
-
X
-
-9 r
10
1
1
1
1
1
.
1
8 PH
FIGURE U: SOLUBILITY OF FRESHLY PRECIPITATED AMORPHOUS TECHNETIUM DIOXIDE AS A FUNCTION OF pH AT 25°C.
From our studies, the solubility of freshly precipitated amorphous and crystalline technetium dioxide appears to be of the order of 10" 8 mol-dm"3 near neutral pH at 25°C. The oxidative solubility of amorphous technetium dioxide can be estimated by combining these results vi tH available thermodynamic data. The standard potential for the cell reaction TcO;
Technetjt/fn Chemistry
I
-4 -
+
3/2H,
* h"
=
TCOJ-2HJO
is -0.743 V, using the data given by Rard (31). Using this value, the activity of TcO, over solid Tc0j-2H,0 near neutral pH has been calculated as a function of the redox potential. The results are shown in Figure 6. From Figure 6, oxidative dissolution will be the C N S 9th A N N U A L CONFEFIENCE, 198B
167
Heat Capacity Data
-4t
The primary objective of Tlijs piogram is to develop methods for obtaining key Ihermodvnann> data (e.g., Gibbs energy, activity coefficient, enthalpy, etc.) for important fission products, actinides and groundvater ions a' waste vault temperatures (~ 100°C). Such information i.-^ required to model the equilibrium behaviour of key radionuclides (I. Cs, Sr, Tc. U, Np) in groundvaters under vault conditions.
O
12
FIGURE 5: SOLUBILITY OF CRYSTALLINE TECHNETIUM DIOXIDE AS A FUNCTION OF pH AT Z5°C.
dominant mechanism at pH 7 if the redox potential is above 0.45 V vs SHE at 25»c. Dissolved technetium(IV) species will predominate over TcO, only at lower potentials. (It should be noted that these calculations assume that the amorphous oxide contains tvo Jater molecules. Thfe results would differ if a different stoichiometry is used.)
SOLUBILITY OF Tc0 2 '2H 2 0 1
1
T
1
1
f
'
J
" t
P
-4K-
'TcO A /
• 6 -
z o
-10 00
/
TcUVKoq)
-8 1
O.I
1 0 2
J 0.3
/
I
1
0.4
0.5
1
06
0.7
E(V vs. S.H.E.)
FIGURE 6; CALCULATED SOLUBILITY OF AMORPHOUS TECHNETIUM DIOXIDE (TcO,-2H.O) AS A FUNCTION OF THE REDOX POTENTIAL AT pH 7 ANT 25°C. 16B CNS 9th ANNUAL CONFERENCE, 1988
WASTE-ROCK-WATER
INTERACTIONS
The sorption of radionuclides on mineral surfaces in rock fractures, and perhaps their incorporation into mineral alteration products, provide mechanisms to retard migration of the radionuclides through rock fractures. Applied research programs are in place to measure sorption coefficient; of representative radionuclides on a variety of mineral surfaces, and to characterize mineral alteration sequences on granite fracture surfaces. The following underlying research is being performed in conjunction with these applied programs. In addition to empirical measurements of sorption behaviour, it is desirable to understand the mechanism of bonding between a sorbed ladicnuclide and a mineral substrate, and hence gain greater confidence in the prediction of radionuclide migration.
/ O
Towards this objective ue have measured the heat capacity and density of aqueous radionuclide s-iutions as a function of temperature using a state-of-the-art flo:< microcalorimeter. An appropriate model was then applied to the measured data to develop equations for interpolating and extrapolating data. The temperature, pressure and composition dependence of the thermodynamic properties was represented by equations which use theoretical principles to combine the short-range (virial-type) and the long-r a nge (electrostatic or Debye-Huckel type) interactions for modelling aqueous radionuclide solutions. The experimental and theoretical wort has been completed for three fission products (Cs. I, Sr) and important groundwater ions (Na*, K* Mg 3 *, Ca ? *, Cl", and SO?"). Comprehensive equations have been developed for tabulating a number of tltetmodynami<- properties (Gitobs energy, enthalpy, osmotic and activity coefficients, etc.) as a function of both the temperature and the pressure. This york is described in detail in recent journal articles (3235). Currently, we are commissioning a new facility to obtain similar thermodynamic properties for radioactive solutions of technstium and neptunium.
Determination of the structure of species sorbed on a mineral surface is a major challenge. Most surface analytical techniques provide information on elemental composition, and in some cases oxidation state, but little or no bonding information. Conversely, spectroscopic and diffraction methods that do provide bonding information are relatively insensitive to processes localized on surfaces. Fourier-transform infrared spectroscopy has been used with some success to glean information on bonding at catalyst surfaces and some soil minerals (36-38). We are presently evaluating its use to determine the nature of sorption interactions of interest. It shows greatest promise for simple polyatomic sorbates, such as oxyanions, with wellcharacterized inftared spectra that are sensitive to the strength and symmetry of the surface interaction. For monatomic sorbates, such as Cs', Sr'', or I", which have no innate infiaied spectrum, information
might be obtained indirectly, from surface hydroxyl vibration features (see. for example, Figure 7B) in the substrate spectrum. Iron oxide minerals comprise a small fraction of granite, but they are important in maintaining the desired reducing conditions, and also are capable of sorbing both cationic and anionic radionuclides. The spectroscopic studies are therefore focusing on iron oxides and related minerals. Magnetite has been shown (39) to immobilize dissolved technetium by reducing soluble pertechnetate (TcOj) to an insoluble Tc(IV) species, which binds to the magnetite surface (Figure 7A). Ferric oxide minerals (e.g., hematite and goethite) are non-reducing and do not interact with pertechnetate, although they can sorb more strongly coordinating anions such as selenite.
Iodine-129 is one of the most problematic radionuclides, since the iodide anion is very soluble, and does not interact strongly vith any common minerals. Consequently, current models indicate that it will be the principal source of radiation dose arising from a high-level uaste disposal system. Recent experiments indicate that iodide may in fact be sorbed weakly by some iron-containing minerals. We are investigating this in more detail to determine whether this interaction may significantly retard iodine migration in the geosphere. This work will draw from our previous experience in the evaluation of candidate waste forms for iodine-129 (40). Work on waste forms for iodine-129 and carbon-14 required an understanding of alteration processes involving hydrolysis and the displacement of I" or CO^~ by other anions present in grounduaters (40,41). There are close parallels between these reactions and silicate mineral alteration reactions, which are largely a combination of hydrolysis and cation displacement. We are therefore using the methodology developed for waste-form reactions lo help understand mineral alteration processes. Cesium aluminosilicates, including the mineral pollucite (ideal formula CsAlSi 2 O 6 ), are among the few insoYtfole compoimos of cesium. TYie-j axe \UIOMTI \ O b« formed by reaction of cesium-containing solutions with a variety of aluminosilicate precursors, and have been proposed as host phases for radioactive cesium. We have examined their formation and alteration under hydrothermal conditions, and thereby estimated their thermodynamic stability limits (42). Results demonstrate that pollucite is unlikely to be formed in a waste repository, unless solutions are both alkaline and contain high concentrations of cesium. Again, the methods used are generally applicable to mineral alteration processes, although the experiments are time consuming.
FIGURE 7A: INFRARED SPECTRA OF (UPPER) NH,TcO., WITH HEMATITE, Fe,0,i (LOWER) MAGNETITE, Fe 3 O,, REACTION WITH NH.TcO.. LOWER FREQUENCY OF THE STRETCHING VIBRATION INDICATES REDUCTION TECHNETIUM, PROBABLY TO A Tc(IV) OXIDE (FROM (39)).
MIXED AFTER Tc-0 OF REF.
FIGURE 7B: INFRARED SPECTRA OF BOEHMITE, Y - A I O O H . MAJOR PEAKS AT 3295 AND 3090 CM"1 ARE DUE TO BULK HYDROXYLS. THE SHARP ABSORPTION PEAK AT 3671 CM"1 (ALSO SHOWN ENLARGED) AND THE BROAD, WEAK FEATURE AROUND 3560 CM"' ARISE FROM SURFACE 0-H VIBRATIONS.
In another study, we are characterizing mineral powder surfaces representative of the nuclear fuelvaste vault environments. This study is a prerequisite for interpreting mineral dissolution and for formulating predictive models for the adsorption of radionuclides onto mineral surfaces. For this work, we established a state-of-the-art sorption microcaloriw.e.t'at: ^.y^teia to Wfiasuce sutcace. areas and heats at adsorpt ion. Earlier, we developed a relatively convenient method for measuring surface areas of pure oxide powders, which was applied successfully to selected minerals (kaolinite, palygorskite, and feldspars). The principle, operating procedures and advantages of our calorimettic technique are described in detail in a recent publication (43). We have further developed this technique to characterize fracture-filling minerals with particular emphasis on the untreated minerals with surface areas less the 1 m:*g"-. Initially, we established a calibration curve using extensive saturation adsorption experiments on five alpha-alumina powders covering a wide range of surface areas (0.1 to 81 m ^ g " 1 ) . We have also measured surface areas for quartz, albite, chlorite and kaolinite covering a range from 0.11 to 10.2 mJ'g"'. Our results (0.11, 0.18, 1.2 and 10.2 m'-g'1) are in good agreement with the known BET values (0.15, 0.15, 0.8 and 9.5 m ^ ' g " 1 ) , respectively. We are now extending our measurements to several fracture-filling minerals (basalt, calcite, chlorite, gobbro, goethite, granite, gypsum, kaolinite, hematite, illite, muscovite and piomon t i t
CNS 9th ANNUAL CONFERENCE, 1988 169
COLLOID CHEMISTRY In a nuclear waste vault located in granitic rock, colloidal transport of actinides such as uranium could take place in two ways. Act.inides could be adsorbed onto ground™ter colloids and be transported in that maimei. Alternatively, it uctinide dissolution and reprecipitat ion occurred, stable actinide colloids could form. Therefore, to determine the possible role of colloids as transporters of activity from a nuclear vaste vaulc, it is important to understand the mechanism of colloid formation and behaviour in environments that are 3S similar as possible to those of an underground disposal vault. Thus, the research done in this program has focussed on the adsorption of uranium onto iron oxide colloidal particles and on the formation of uranium colloids by reduction of an aqueous uranium(VI) solution. Early in the program, we studied the adsorption of uranium(VI) onto hematite particles (44-46). The effects of pH changes, and the effects of bicarbonate ions and humic acid were studied. These studies lead us to conclude that the adsorbing species in 'he absence oC bicarbonate was likely a hydrated cation such as (UO,),(OH)t. -'hereus, in the presence of bicarbonate, a hydrar»(! carbonate anion. possibly, (U0,,)i,C0j(0H)i, was the main adsorbing species. The effects of humic acid were complex, hut it could both increase adsorption or decrease it, depending on conditions. More recently, adsorption oE uranium(VI) species onto colloidal magnetite particles was studied both in the presence and in the absence of bicarbonate. With bicarbonate present, adsorption was studied in the pH range of 7 to 9. The initial uranium uptake decreased as the pH increased, and an anionic hydroxy carbonate species is likely involved. At the highest pH and at uranium(VI) concentrations above 70 umol'dnr3, a precipitate formed on the surface of the magnetite. This precipitate increased the apparent adsorption enormously. but, of course, this is not true adsorption. The precipitate consisted of very thin plates and contained a great deal of uraniumIt did not form in control experiments where the magnetite -.-•as not present. A few studies were carried out of adsorption of uranium(VI) onto the same colloid, but in the presence of 500 umol'dnr-1 sodium chloride and in the pH range of 3 to 6. A small initial adsorption was noted at lower pH values, but this increased considerably vhen the pH reached 5.7. These adsorption results are roughly similar to those observed for adsorption of uranium(VI) onto a hematite colloid when adsorption is compared on a unit surface area basis. The largest adsorption corresponds to 2 mg-m"2, so the amount of actinide that could be transported by this mechanism is likely small. nil analysis of the Fe''/Fe*3 ratio in the surface of the magnetite sol was carried out using photoelectron ppe"troscopy. The Fe (3p) band was deconvoluted into H''- and Fe* ! components according to the procedure of Mclntyre and Zetaruk (47). The ratio was 0.08, suggesting that the surface does contain Fe + 2 t but perhaps not in the full stoichiometric ratio. The second phase of this work involved studies of uranium colloid formation by reducing aqueous uranium(VI) ions with sodium sulfide. Early work shoved that uranium colloids vere formed above i0°C (48), and that the colloids were either a mixed "alency (uranium(IV) + uranium(VI)) oxide or a complex uranium(IV) hydrated oxide. More recently the
170 CNS 9th ANNUAL CONFERENCE, 1988
reduction has been studied at temperatures as lov as 10°C (49). At these temperatures the reduction is slow, yielding a dispersion of black particles. In some cases, a mixture of black and purple particles is formed. These particles were characterized by X-ray diffraction, X-ray photoelectron spectroscopy, scanning electron microscopy, Fourier translorm infrared spectroscopy, photon correlation spectroscopy and microelectrophoresis. The black particles were found to be a mixed-valency uranium oxide, and the uranium oxide particles that formed at 10°C had a higher oxygen to uranium ratio on the surface than those formed at 25°C. There was some evidence that the purple particles were a reduced form of uranium. At higher pH or lower uranium concentration, a comple;* uranium(VI) hydrous oxide can also form. We have also studied the effects of bicarbonate, humic acid and sodium chloride on the reduction of uranium(VI) to uraniup(IV) by aqueous sulfide at 10' and 25T (50). Bicarbonate tends to stabilize uranium(VI) towards reduction, but sodium chloride has little effect. Humid acid enhances the reduction. In most cases, either a reduced form of uranium or a mixed uranium oxide is formed. Characterization of the reaction products by X-ray diffraction, X-iay photoelectron spectroscopy, Fourier transform infrared spectroscopy and photon correlation spectroscopy has been carried out. These studies have shown that uranium colloids could possibly form near a vaste vault if certain, perhaps restrictive, conditions are achieved. The movement of these colloids would contribute to the mobility of uranium. Other actinides have not been investigated in this program but they could potentially behave somewhat similarly.
CONCLUSIONS The above work highlights elements of an extensive program of in-depth chemistry research conducted by Atomic Energy of Canada Limited in support of the Canadian Nuclear Fuel Waste Management Program. Electrochemical and spectroscopic studies, and theoretical thermodynamic calculations are being used to delineate the chemical conditions over which the fuel matrix remains integral and dissolves the least in groundwater. Work to date has shown that chemical conditions provided by environments such as those encountered deep in granitic rock foimations, in the Canadian Shield, favour long-term fuel matrix stability with respect to dissolution. A large component of chemistry res'.^ch is also dedicated to assessing 'h.erntodyna,,i;.. data, and measuring unavailable data, in order to determine radionuclide solubilities and to define chemical forms required to assess the transport of the radionuclide through the geosphere. This component of research shows that, with the exception of iodine, most radionuclides would not be transported easily through the geosphere due to low solubility in groundwater and/or strong Interactions (sorption, chemical reactions) with the mineral surfaces. Also, a limited number of studies have been conducted to define the conditions under which sorption ol uranium on groundvaier colloids ami true tuaniun. colloid formation can occur. Such colloids could form under certain restrictive conditions near the vaste vault, but they are not expected to be an important carrier of radionuclides through the geosphere.
REFERENCES (1)
HANCOX, W.T. "Safe, Permanent Disposal of Nuclear Fuel Waste", Nucl. Journal of Canada 1, 256 (1987).
(2)
VIKIS, A.C., GARISTO, F., HAINES, R.I., HO, C.H., LEMIRE, R.J., PAQUETTE, J., SALUJA, P.P.S., SUNDER, s. and TAYLOR, P. "Chemistry Research for the Canadian Nuclear Fuel Waste Management Program", Proceedings of the Twentieth Information Meeting of the Canadian Nuclear Fuel Waste Management Program, Atomic Energy o£ Canada Limited Technical Record, TR 375 (1986).
(3)
SUNDER, S., SHOESMITH, D.W., BAILEY, H.G., STANCHELL, F.U. and MCINTYRE, N.S. "Anodic Oxidation of UO,, Pan I. Electrochemical and X-Ray Photoelectron Spectroscopic Studies in Neutral Solutions", J. Electroanal. Chem. 130, 163-179 (1981).
(4)
(5)
(6)
SUNDER, S., SHOESMITH, D.W. , BAILEY, M.G. and WALLACE, G.J. "Mechanism of Oxidative Dissolution of U0 2 under Uaste Disposal Vault Conditions", Proceedings of the 1st. Int. Conf. on Radioactive Baste. Hanag. Winnipeg, Canada; Canadian Nuclear Society, pp 398-405 (1982). SUNDER, S., SHOISSMITH, D.V., BAILEY, M.G. and WALLACE, G.J. "Anodic Oxidation of U0 2 , Part II. Electrochemical and X-ray Photoelectron Spectroscopic Studies in Alkaline Solutions", J. Electroanal. Chem. 150, 217-228 (1983). SHOESMITH, D.W., SUNDIiR, S., BAILEY, M.G. and OWEN, D.G. "Anodic Oxidation o£ U0 2 , Part III. Electrochemical Studies in Carbonate Solutions", in Proc. 5th Int. Symp. on Passivity, Bombannes, France, 19B3, M. Froment (ed.), Elsevier, Amsterdam, pp 125-130, 1983.
(7)
SHOESMITH, D.W., SUNDER, S., BAILEY, M.G., WALLACE, G.J. and STANCHELL, F.U. "Anodic Oxidation of U0 2 , Part IV. X-ray Photoelectron Spectroscopic and Electrochemical Studies of Film Growth in Carbonate-Containing Solutions", Applic. of Surf. Sci. 20, 39-57 (1984).
(8)
SHOESMITH, D.U., SUNDER, S., BAILEY, M.G. and WALLACE, G.J. ''Mechanism of Oxidation and Dissolution of CA.NDU Fuel", Proceedings of the Second Int. Conf. on Radioactive Waste Management, Winnipeg, Canada; Canadian Nuclear Society, pp 674-679 (1986).
(9)
(10)
(11)
SHOESMITH, D.W., S0NDER, S., BAILEY, M.G. and WALLACE, G.J. "Anodic Oxidation of U0 2 , Part V. Electrochemical and X-ray Photoelectron Spectroscopic Studies of Film-Growth and Dissolution in Phosphate-Containing Solutions", Can. J. Chem. 66, 259-265 (1988). SHOESMITH, D.W., SUNDER, S., JOHNSON, L.H. and BAILEY, M.G. "Oxidation of CANDU UOj Fuel by the Alpha-Radioiysis Products of Water", Scientific Basis for Nuclear Waste Management IX, L.0. Werme (Ed.), Material Research Soc. Symp. Proc. Vol. SO, pp X>9-})6 SUNDER, S., SHOESMITH, D.W., JOHNSON, L.H., BAILEY, M.G. and WALLACE, G.J. "Effect of Alpha-Rmllolysis of Water on CANDU™ Fuel
Oxidation", Proceedings of the Second Int. Conf. on Radioactive Uaste Management, Winnipeg, Canada; Canadian Nuclear Society, pp 625-631 (1986). (12)
SUNDER, S., SHOESMITH, D.W., JOHNSON, L.H., BAILEY, M.G., WALLACE, C.J. and SNAGLEWSKI, A.P. "Oxidation ol CANDU™ Fuel by the Products of Alpha-Radiolysis of Grounduater", Scientific Basis for Nuclear Waste Management X, J.K. Bates and W.B. Seefeldt (Eds.), Material Research S O Q . Symp. Proc. Vol. 84, pp 103-113 (1986).
(13)
SHOESMITH, D.U., SUNDER, S., BAILEY, M.G. and WALLACE, G.J. "Corrosion of Nuclear Fuel (UO2) in Oxygenated Solutions", in preparation.
(14)
LeNEVEU, D.M. "Vault Submodel for the Second Interim Assessment of the Canadian Concept for Nuclear Waste Disposal: Post-Closure Phase", Atomic Energy of Canada Limited Report, AECL8383 (1986).
(15)
GARISTO, F. and GARISTO, N.C. "A U0 2 Solubility Function for the Assessment of Used Nuclear Fuel Disposal", Nucl. Sci. Eng. 90, 103 (1985).
(16)
GARISTO, N.C. and GAHiSTO, F. "The Dissolution of U0 2 : A Thermodynamic Approach", Nucl. Chem. Uaste Manag. 6, 203 (1986).
(17)
GARISTO, F. "Solid Dissolution: Effect of Mass Transport-Precipitation Coupling", Chem. Eng. Sci. U , 3219 (1986).
(18)
GARISTO, N.C. and GAfclSTO, F. "The Effect of Precipitation on the Long-Term Release of Radionuclides from Used Fuel", Annals of Nucl. Energy 13, 591 (1986).
(19)
UANNER, H. "The NEA Thermochemical Data Base Project", Radiochim. Acta, in press (1988).
(20)
Committee on Data for Science and Technology (CODATA), "CODATA Recommended Key Values for Thermodynamics 1977, Bulletin 28", Paris, International Council for Scientific Unions, 1973,
(21)
LEMIRE, R.J. and TREhAINE, P.R. "Uranium and Plutonium Equilibria in Aqueous Solutions to 200°C", J. Chem. Eng. Data 25, 361 (1980).
(22)
LEMIRE, R.J. "An Assessment of the Thermodynamic Behavior of Neptunium in Water and Model Groundwaters from 25 to 150°C", Atomic Energy of Canada Limited Report, AECL7817 (1984).
(23)
LEMIRE, R.J. "Effects of High Ionic Strength Groundwaters on Calculated Equilibrium Concentrations in the Uranium - Water System", Atomic Energy of Canada Limited Report, AECL9549 (1988).
(24)
MAYA, L. "Hydrolysis and Carbonate Complexation of Dioxoneptunium(V) in 1.0 M Sodium Perchlorate at 25"C", Inorg. Chem. 22, 2093 <1983). —
(25)
VITORGE, P. "Mesure de constantes thermodynamiques de composes transuraniens pour prevoir leur geochimle", in Proc. of the Seminatre sur les Techniques d'Etude et les C N S 9th ANNUAL CONFERENCE, 1988 171
Colloidal a-, g- and Y- Ferric Oxide Hydroxides", J. Chem, Soc. Faraday Transactions I, 82, 2401-2410 (1986).
Hethodes d'Evaluation des Sites en vue du Stockage Definitif Souterrain des Dechets Radioactifs, IAEA, Sofia, Bulgaria 1984 (IAEASR-104/23). (26)
BIDOGLIO, G., TANET, «., CHATT, A. "Studies on Neptunium(V) Carbonate Complexes under Geologic Repository Conditions", Radiochim. Acta 38, 21 (1985).
(27)
INOUE, Y. and TOCHIYAHA, 0. "Studies of the Complexes of Neptunium(V) with Inorganic Ligands by Solvent Extraction with Thenoyltrifluoroacetone and 1,10 Phenanthroline. I. Carbonato Complexes", Bull. Chem. Soc. Japan 5B, 588 (1985).
(28)
(29)
(30)
(31)
(32)
(33)
PAQUETTE, J. and LAURENCE, W.E. "A of the Spectoeiectrochemical Study Couple in Technetium(IV)/Technetium(III> Bicarbonate Solutions", Can. J. Chen. 63. 2369 (1985). WALTON, F.B., PAOUETTE, J., ROSS, J.P.M. and LAWRENCE, U.E. "Technetium(IV) and -(VIII) Interactions with Iron Oxyhydroxides", Nucl. Chetn. Waste Manage. 6, 121 (1986). HAINES, R.I., VANDERGRAAF, T.T. and OWEN, D.G. "Technetium-Iron Oxide Reactions Under Anaerobic Conditions: A Fourier Transform Infrared, FTIR Study", Nucl. J. Canada 1, 32 (1987). RARD, J.A. "Critical Review of the Chemistry and Thermodynamics of Technetium and Some of its Inorganic Compounds and Aqueous Species", Lawrence Livermore National Laboratory Report, UCRL-53440 (1983) SALUJA, P.P.S., LEBLANC, J.C. and HUME, H.B. "Apparent Molar Heat Capacities and Volumes of Aqueous Solutions of Several 1:1 Electrolytes at Elevated Temperatures", Can. J. Chem. 64, 926 (1986). SALUJA, P.P.S., PITZER, K.S. and PHUTELA, R.C. "High-Temperature Thermodynamic Properties of Several 1:1 Electrolytes", Can. J. Chem. 64 1328 (1986).
SALUJA, P.P.S. and LEBLANC, J.C. "Apparent Molar Heat Capacities and Volumes of Aqueous Solutions of MgClj, CaCl, and SrCl2 at Elevated Temperatures", J. Chem. Eng. Data 32, 72 (1987). (35)
PHUTELA, R.C., PITZER, K.S. and SALUJA, P.P.S. "Thermodynamics o£ Aqueous Magnesium Chloride, Calcium Chloride and Strontium Chloride at Elevated Temperatures", J. Chera. Eng. Data 32, 76 (1987).
(36)
PARFITT, R.L. "Anion Adsorption by Soils and Soil Minerals", Advances in Agronomy 30, 1-50 (1978).
(37)
PARFITT, R.L. and RUSSELL, J.D. "Adsorption on Hydrous Oxides IV. Mechanisms of Adsorption of Various Ions on Geothite", J. Soil Sci. 28, 297-305 (1977).
(38)
ISIIIKAVA, T., NITTA, S. "Fourier-Transform Infrared
172 CNS 9th ANNUAL CONFERENCE, 1988
and KONDO, Spectroscopy
S. of
(39)
HAINES, R.I., VANDERGIIAAF, T.T. and OWEN, P.G. "Te-l-netium-Iron Oxide Reactions Under Anaerobic Conditions: A Fourier Transform Infrared, FTIR Study", Nucl. J. Can. 1, 32-37 (1987).
(40)
TAYLOR, P., LOPATA, V.J., WOOD, D.D. and YACYSHYN, H. "Solubility and Stability of Inorganic Iodides: Candidate Waste Forms for Iodine-129", Proceedings of ASTM 4th International Hazardous Waste Symposium, Atlanta. GA, May 1987, in press.
(41)
TAYLOR, P. "Solubility and Stability of Inorganic Carconates: An Approach to the Selection of a Waste Form for Carbon-14", Atomic Energy of Canada Limited Report, AECL9073 (19B7).
(42)
TAYLOR, P., DEVAAL, S.D. and OMEN, D.G. "Stability Relationships Between Solid Cesium Aluminosilicates in Aqueous Solutions at 200°C", Can. J. Chen)., in press.
(43)
SALUJA, P.P.S., OSCARSON, D.W., MILLER, H.G. and LEBLANC, J.C. "Rapid Determination of Surface Areas of Mineral Powders Using Adsorption Calorimetry", In: Proceedings of the Eighth International Clay Conference, Denver, L.G. Schultz, H. van Olphen and F.A. Mumpton, (eds.), The Clay Mineral Society, Bloomington, Indiana, p.267 (1987).
(44)
HO, C.H. and DOERN, D.C. "The Sorption of Uranyl Species on a Hematite Sol", Can. J. Chem. 63, 1100 (1985).
(45)
HO, C.H. and MILLER, N.H. "Effect of Humic Acid on Uranium Uptake bv Hematite Particles", J. Colloid Interface Sci. 106, 281 (1933).
(46)
HO, C.H. and MILLER, N.H. "Adsorption of Uranyl Species from Bicarbonate Solution onto Hematite Particles", J. Colloid Interface Sci. 110, 165 (1986).
(47)
MCINTYRE, N.S. and ZETARUK, D.G. "X-Ray Photoelectron Studies of Iron Oxides", Anal. Chem. 49, 1521 (1977).
(48)
HO, C.H. and MILLER, N.H. "Formation of Uranium Oxide Sols in Bicarbonate Solutions", J. Colloid Interface Sci. 11_3, 232 (1986).
(49)
HO, C.H. and MILLER, N.H. "The Reduction of Uraniuro(VI) Ions by Aqueous Sulfide at Lou Temperatures", submitted to Nucl. Sci. Technology.
(50)
HO, C.H. and MILLER, N.H. "The Effects of Foreign Ions on the Reduction of Uranium(VI) Ions by Aqueous Sulfide at Low Temperatures", submitted to Nucl. Sci. Technology.
RADIONUCLIDE MIGRATION THROUGH FRACTURED GRANITE. LABORATORY STUDIES
D.M. GRONDIN, T.T. VANDERGRAAF and D.J. DREW
Whiteshell Nuclear Research Establishment Atomic Energy of Canada Ltd. Pinawa, Manitoba, ROE 1LO, Canada
ABSTRACT Radionuclide migration has been studied in natural fractures in granite blocks of up to 30 cm in length. Results are reported for four migration experiments involving synthetic gfoundwaters containing tritiated water, 95 "Tc, 7 5 Se, 1 3 7 Cs, or 60Co-labelled natural colloids, which were injected into the fractures at flow rates of 0.1-0.45 mL/h, giving residence times in the fractures of up to 15 h. Also presented are the results of the post-experiment analyses, including an autoradiograph of one of the fracture surfaces, and the spatial distribution of the sorbed radionuclides determined by y-scanning and selective chemical extractions. INTRODUCTION Atomic Energy of Canada Ltd is assessing the concept of deep underground disposal of nuclear fuel wastes in plutonic rock at a depth of 500 to 1000 m. If a breach of the engineered containment surrounding the waste were to occur, the most credible way in which the radior.uclides could reach the biosphere would be via groundwater flowing through water-bearing fractures in the host rock. The movement of most radionuclides is expected to be retarded by their interaction uith the geological material lining the fractures. These interactions include a variety of processes such as ion exchange, chemisorption, precipitation, and mineralisation. The sorption of radionuclides on minerals found in plutons located in the Canadian Shield has been measured under static conditions (1-6). These sorption coefficients are used to predict the degree of retardation or radlom)clid.es migrating along natural fractures in granite and, hence, the eventual environmental impact of a used fuel waste disposal vault. Radionuclide migration experiments carried out under well-defined ceiditions are being used to determine the validity of the use of sorption coefficients in transport models. Results from these experiments help develop new, or improve existing transport models, for predicting the behavicmr of species migrating along natural fractures at different scales. Radionuclide migration studies are currently under way to assess the behaviour of radionuclides migrating through single fractures in crystalline rock of various dimensions. The residence time of the transport solutions in the fracture increases with the fracture dimensions. This residence time is an important factor in migration experiments, especially considering the slcrf kinetics o£ many radionuclide-rock interactions (7). Until now, research has been focused on rock samples, with, telatlvely short tlowpaths of up to 9 cm, providing residence time* u"ly a tev hours (8,9). This report presents results of experiments conducted on larger natural fractures providing flow paths of up to 30 cm in length, fracture volumes of up to 6 mL, and groundwater residence times of up to 15 h.
The main objectives of these experiments were (1) to determine if the results from relatively long-term dynamic experiments can be predicted using simple sorption coefficients (2) to develop an experimental methodology to be used in large-scale experiments (granite blocks of dimensions of up to 1 m x 1 m x 0.6 m or field studies) (10), including post-experiment analysis of the fracture surfaces to determine flow paths and radionuclide-mineral interactions and (3) to assess the effects of E h on cadionuclide migration. Four granite blocks, each bearing a natural fracture, were selected for this study. Sheet flow was obtained parallel to the long axis of the fractures and a variety of tracers, including 3 H,0, p5"Tc0"4, 75 60 Se05", Co-labelled colloids and '"Cs*, were injected. Sorption coefficients, determined from the elution profiles, were compared with those obtained under static conditions. Autoradiography, 2-D yscannirg of the fracture surface and selective chemical extraction of some sorbed radionuclides were carried out at the end of the migration experiments. THEORY When a dissolved radionuclide is transported by groundwater in a fracture, it usually reacts with the fracture surfaces and is removed, at least temporarily, from solution; hence, its movement is retarded relative to that of the groundwater. The rate of flow of the groundwater can be determined from the velocity of a non-reactive tracer, i.e., one that does not interact with the minerals lining the fracture. In this study, tritiated water or a dye, uranine (NaFluorescein), was used as a non-reactive tracer to monitor the flow rate of groundwater. The degree of retardation of the radlonuclldes was ofcxaineo fey comparing the radionuclide velocity relative to that of the non-reactive tracer. If the sorption process is reversible, then the retardation factor R is obtained from
R= U w /U H n . 1 + 2
k./b
[I)
where U u is the velocity of the groundwater, U. is the velocity of the radionuclide, k."is the surface distribution coefficient (cm), and b is the fracture aperture (cm). The surface distribution coefficient can be determined experimentally by exposing a radionuclide-tagged solution to a fracture surface similar to that used in an experiment. Values of k. based on the geometric surface area are obtained using the following equation:
k. - S/C S
is
the
sorbed
rtfdlonuclide
concentrniion
C N S 9th A N N U A L C O N F E R E N C E , 19BB 173
GGW Reservoir
(mol/cm 2 ), and C is the radionuclide concentration in solution (mol/cm 3 ). Again, it is assumed that the reaction is reversible.
n
The shape of the elution profiles reveals the effects of the irreversibility of the sorption process, dispersion of the tracer and diffusion into the rock matrix on the transport of the tracers. _ Peristaltic Pump
EXPERIMENTAL PROCEDURE Blocks of pink granite of medium grain size were obtained from the Cold Spring Quarry on the Lac du Bonnet batholith near Pinawa, Manitoba. The mineralogical and chemical compositions of this pink granite are given elsewhere (11). The intermediate-scale blocks were named ISB-1, - 2 , -3 and -5. The most abundant minerals identified by XRD analysis on the fracture surface of ISB-1 vere quartz and calcite, vith smaller amounts of K-feldspar minerals. Quartz, calcite, clinochlore and museovite were the major constituents of the fracture infilling material of ISB-2; again traces of K-feldspar minerals were present. Results for blocks ISB-3 and ISB-5 are not available but their surfaces appear to be similar to that of block ISB-1. Four experiments were conducted on blocks containing a single natural fracture. The dimensions of the blocks were as follows (L x W x H ) : ISB-1: 25 cm x 15 cm x 10.5 cm; ISB-2: 25 cm x 12.7 cm x 13 cm; ISB-3: 30.5 cm x 16.3 cm y. 15 cm and ISB-5: 30.5 cm x 16.3 cm x 15 cm. The fracture apertures vere determined from the elution profiles of the inactive tracers (uranine and/or tritiated water) at the time at which the concentration at the outlet was half of that of the initial solution and are given in Table 1.
MIGRATION EXPERIMENT EXPERIMENTAL CONDITIONS
1
ISB-1
Radionuclide
3
H,
137
75
Se,
Flow Fracture Residence Rate Aperture Time (mL/h) (um) (h)
0.4
160
15
137
Cs
2
ISB-2
Y
3
ISB-3
60
COcolloid
0.45
100
4
ISB-5
95
0.45
70
0.4
55
4.4 11 7.8
Blocks ISB-1 and ISB-2 were taken apart and channels were cut on the fracture surface of one of the halves, 1 cm away from the edges, to obtain a sheet flow between two opposite sides of the fracture plane (Figure 1 ) . This procedure was carried out to create an inlet and an outlet channel. The blocks were reassembled and stainless steel fittings were inserted at both ends of the channels. The remainder of the fracture aperture on the external surface of the block was sealed off using a silicone-based rubber (CGE, RTV-108). A peristaltic pump was used to circulate the radionuclide-tagged groundvater solutions through the inlet channel. The flow along the fracture was adjusted to 0.4 mL/h by a soleroid valve installed at
174
SCHEMATIC OF THE ARRANGEMENT USED IN THE MIGRATION EXPERIMENTS. NUMBERS 1 TO 5 DESIGNATE THE SOLENOID VALVES THAT CONTROL THE FLOW THROUGH THE FRACTURE.
one end of the outlet channel and controlled by a fraction collector. At the other end of the outlet channel, a reservoir filled with a synthetic groundvater solution (GGV, whose chemical composition is given in Table 2) was attached, so that the channel could be periodically flushed to avoid an accumulation of tracers. A solenoid valve and a timer were used to control the frequency of flushing and the volume of the flushing solution. When the flushing occurred, the flow in the fracture was interrupted by shutting valves 3 and 4 located at both ends of the inlet channel. TABLE 2
Concentration Ion Concentration Ion (mg/L) (mg/L) Na*
8.3
K*
3.5
Mg2*
3.9
Ca
Cs
Tc
FIGURE 1.
CHEMICAL COMPOSITION OF THE SYNTHETIC GROUNDWATER (GGW)
TABLE 1
Exp. Block No
^Tracer Solution Reservoir
C N S 9th A N N U A L C O N F E R E N C E , 1988
2+
Cl"
13.0
< NO" F~ HCO"
8.6 0.62 0.19 58.0
5.0 pH
6.5
A similar arrangement was used for blocks ISB-3 and ISB-5 except that the channels cut on the fracture surfaces were replaced by reservoirs covering the fracture where it intersects the end faces of the block. The reservoirs consisted of shallow channels (6 mm wide and 3 mm deep) following the contour of the fracture and cut in a Plexiglas plate. Again, the flow was controlled by a solenoid valve installed at one end of the outlet reservoir and controlled by a fraction collector. The flow rate was set at 0.45 mL/h. No flushing of the outlet reservoir was done in these experiments. Before a migration experiment was started, the fracture was flushed with GGW for several days to equilibrate the fracture infilling material with the synthetic groundwater solution, and to ensure the flow field was well established vhen the tracer solutions were injected.
RADIONUCLIDE TRACER SOLUTIONS 75
1!7
A solution containing tritiated water, Se and Cs was injected into ISB-1. Tritiated water and 1 3 7 Cs were used in ISB-2. The activities of tritiated water and cesium were both approximately 790 Bq/mL, whereas that of 75 Se was about 285 Bq/mL. In these experiments, 1 mg/L of inactive cesium (as CsCl) was 137 added to the solutions to limit Cs sorption on the fracture surface and obtain a breakthrough in a reasonable time. The above solutions were prepared using GGW and the pH was adjusted to 7.0 using 0.05 mol/L NaOH. Before use, the solutions were filtered through 0.45-um membranes. The detailed procedure is given elsewhere (3). A 95"Tc0~4 solution with an activity of 250 Bq/mL was made up with GGU and injected into block ISB-5. After a complete elution profile was obtained for ''"TcO^ (67.7 mL eluted), hydrazine (N 2 H 4 ) was added to the solution to a final concentration of 0.023 mol/L. An E h value of -350 mV was measured immediately after the addition of N 2 H 4 and it slowly increased to a value of -250 mV at the end of the experiment. The final pH value of that solution was 8.0. The radiolabelled colloid solution was prepared by sorbing 60 Co onto natural colloids collected from Fracture Zone 2 in the Underground Research Laboratory (URL). The colloidal material was obtained by collecting a 50-L sample of groundwater from borehole HC-8 and concentrating particles larger than 1 nm in a 2-L volume using a Pellicon tangential flow ultrafiltration system (Millipo-e). The colloid concentrate, with 73 mg/L of particulate matter, was radiolabelled by adding s 0 Co to 200 mL of colloid concentrate. During the addition of 6 0 Co, a pH stat was used to maintain the pH of the solution at 8.2. After a 3-month equilibration, dissolved 6 0 Co was removed from the colloid suspension using a Mini tan tangential flov ultrafiltration system (Millipore), which separated particles greater than 1 nm from dissolved substances. Filtered water containing 6 0 Co was removed and was replaced with 60 Co-free groundwater collected from Fracture Zone 2. A size distribution analysis showed that most of the colloids were in the 1- to 10-yn range. STATIC SORPTION EXPERIMENTS Static sorption experiments to measure ka were carried out using representative fracture surface material to compare with the results of the migration experiments. Coupons with a geometric surface area of 2 cm 2 were obtained from fracture surfaces of blocks similar to those usod in the migration experiments. All faces of the coupons, except the fracture surface, were coated with silicone rubber. Correction for the sorption of the radionuclides on the silicone rubber was made. The coupons were placed into polypropylene vials and covered with 20 mL of the tracer solutions used in the migration experiments. Contact periods varied from 1 to 24 d. All experiments were carried out in duplicate.
The distribution of cesium on the fracture surfaces was measured by two methods. The first approach involved y-scanning the fracture surfaces using a Ge(Li) detector. Each half of the blocks was enclosed in a plastic bag to avoid spread of contamination and a grid was drawn on the plastic over the fracture surface. A 5-cm-thick lead collimator with a l-cm! opening was positioned manually in front of a Ge(Li) detector and aligned with the grid. A distribution profile was obtained by counting at different points on the grid. The second approach used an autoradiographic technique specifically developed for rough surfaces (12). A liquid photographic emulsion (Ilford K-5) is brushed on a cotton cloth and allowed to dry in the dark. The cloth is then placed between the two halves of the block for times of up to 30 d and developed to reveal the distribution of the sorbed radionuclides. Selective chemical extractions are used to determine associations between heavy metals or radionuclides and components of soil, sediment or rock (7,13,14). These methods were used on the fracture surfaces of ISB-1 and ISB-5 to determine the various radionuclidemineral associations. Two locations (A and B, see Fig.4) were isolated from the rest of the fracture surface of ISB-1 v\sing a 4.5-cm-I.Q. acvylit; tube attached to the fracture surface with silicone-rubber. The extraction reagent was then poured in these reservoirs. In the case of ISB-5, the entire fracture surface was immersed face down in a Plexiglas tray containing the reagent. We identified four types of mineral-radionuclide associations using this technique: (i) Exchangeable. The extraction was performed at room temperature for 1 h with 13 mL of 1 M MgCl 2 , pH 7.0. (ii) Bound to carbonates. The rinsed (CDW) surface was leached at room temperature for 5 h using 13 mL of 1 M NaOAc adjusted to pH 5.0 with acetic acid. (iii) Bound to Fe(0H) 3 . The surface was leached with 32 mL of 0.25 M NH 2 OH.HC1 in 0.25 H HC1 for 0.5 h at 70°C. This extraction was performed in a convection oven. (iv) Bound to FeO0H/Fe 2 O 3 - 'Hie final extraction was performed using 30 mL of 0.3 M Ms-citrate, 4 mL of 1 H NaHCO, and 0.8 g of Na 2 S 2 0, for 0.25 h at 80°C in a convection oven. The extraction procedures were based on those developed by Walton (15). Autoradiography was used to detect any residual activity on the region of the fracture surface where the chemical extractions were performed. RESULTS AND DISCUSSION The experimental conditions for all four migration experiments are summarized in Table 1 and the results are discussed below:
ANALYSES The tritium activity was determined by liquid scintillation and that of "^Se, * 6 "Tc and v i 1 Cs by Yspectroscopy using a well-type Nal(Tl) scintillation detector (Harshaw Type 8SF8). Elution profiles were obtained by plotting the lelative activity (C/Co where C o is the initial concentration) of the radionuclides as a function of time or volume el;>tid.
Experiment 01: Intermediate scale block-1 (ISB-1) Solution containing tritium, 7 5 Se and 1 3 7 Cs was injected at a rate of 0.4 mL/h through the natural fracture in ISB-1, corresponding to a linear velocity of approximately 146 m/a and a water residence time in the fracture o£ approximately 15 h. The elution profiles are presented in Figure 2. Tritiated water and C N S 91h A N N U A L C O N F E R E N C E . 1988 175
selenium were eluted rapidly from the fracture with very little retardation of 75 Se. Injection of 1?7 Cs was done continuously for a total of 32 d until more than 300 mL were eluted. At this point, cesium was still absent from the eluates and the experiment was terminated.
1.0 -
a
a
• Tritium o Selenium-75
a
D
•
a* •
*• a
0.8 a
0.6 • 0.4 0.2 0
• »O«0»0«Q»l>
D
a
FIGURE 3. AUTORADIOGFAPH SHOWING THE DISTRIBUTION OF 131 Cs ON THE FRACTURE SURFACE OF ISB-1. 1
p-—?
8
1
1
12
20
Volume eluted ( m L ) FIGURE 2. ELOTION PROFILES FOR THE INJECTION OF TRITIUM AND 75Se IN ISB-1. NO BREAKTHROUGH WAS OBTAINED FOR n 1 C s .
The distribution coefficient can be calculated from the analysis of the elution profiles. Based on the velocities of tritiated water and selenium and an aperture of 0.016 cm, a k, of 7.2 x 10~ 4 cm vas obtained for1 5 Se. This value is much lover than those obtained under static conditions, which were 1.33, 1.77 and 2.50 cm after 2, 7 and 21 d, respectively. These static kB values are consistent with previous results obtained for the sorption of 15 Se on fractureinfilling minerals commonly occurring in granitic rock (16). However, to be able to adequately compare dynamic and static k# values, static k# values for shorter contact times are required. Kinetic effects have very likely affected the migration of ' 5 Se and further studies of the kinetics of 15 Se sorption on natural fracture surfaces are in progress to help interpret these results. Since an elution profile was not obtained for cesium, it was necessary to analyze the fracture surface to determine the distribution of cesium along the fracture. Figures 3 and 4 present an autoradiograph of the fracture surface and a 2-D y-scan respectively. Both techniques are in good agreement and show a front, indicating an average Cs transport of approximately 15 cm frora the inlet channel. Thus, the Cs front traveled 15 cm in 32 d, corresponding to an average linear velocity of 1.71 m/a. This gives a retardation factor of 146/1.71 = 85 and a corresponding k, of 0.67 cm. This value is lower than those obtained under static conditions (3.86, 4.62 and 5.87 cm after 2, 7 and 21 d, respectively). The difference in the ratios of surface area-to-solution volume, or in contact time, could account for the observed discrepancy.
176 C N S 9th ANNUAL CONFERENCE, 1988
FIGURE 4. -y-SCAN OF THE FRACTURE SURFACE OF ISB-1. FIGURES ARE IN CPH/CM2. ALSO SHOWN ARE THE LOCATIONS ISOLATED FOR THE CHEMICAL EXTRACTIONS.
The results for the sequential chemical extractions are listed in Table 3. They show that approximately 33£ of all the extracted cesium is associated vith the exchangeable fraction, 33X is leached in the carbonate extraction procedure and the rest is associated with the various forms of iron oxihydroxides. This observation is consistent with results obtained in other studies (4,5,15) and indicates that several complex sorption mechanisms are responsible for the fixation of this radionuclide.
TABLE 3
An autoradiograph of the fracture surfaces showed that 1 3 1 Cs was present across the entire width of the fracture, but its activity decteased with distance from the inlet channel. This is expected because a complete breakthrough did not occur.
SEQUENTIAL CHEMICAL EXTRACTIONS OF n 7 C s AND 9> 'Tc FROM NATURAL FRACTURE SURFACES
Fraction
% Total X Total 137 9 5» T c Cs Extracted* Extracted*
Reagents
Exchangeable 1M MgCl 2
38
37
Carbonate
1M NaOAc
34
25
Fe(OH) 3
O.25M NH 2 OH HCl in 0.25 HCl
22
24
Fe 2 O 3 /Fe00H
O.^M Na-Citrate 1M NaHC0 3 NajS 2 0 4 (s)
6
14
Experiment #3: Intermediate-scale block-3 (ISB-3) This experiment was designed to study the migration of colloids through a natural fracture in granite. The '"Co-labelled natural colloid was injected through the 100-um-aperture fracture at a flow rate of 0.45 mL/h, corresponding to a linear velocity in the fracture of 240 m/a and a residence time of 11 h. A total volume of 65 mL of tracer solution was eluted. When the experiment was terminated, 6 0 Co was still absent from the eluates and the initial 6 0 Co activity (C o ) had dropped significantly because of the colloid settling at different locations along the tubing and in the inlet reservoir.
irom 2 locations i.n fracture surface of ISB-1 from the fracture surface of ISB-5
Experiment It2: Intermediate-scale block-2 (ISB-2) Tritiated water and 1 J 7 Cs were injected into a natural fracture in ISB-2 at a flow rate of 0.4 mL/h (linear velocity of 500 m / a ) , giving a residence time of 4.4 h. Figure 5 shows the elution profiles for ll7 Cs. The breakthrough of 1 3 7 Cs was detected after approximately 1300 h (55 d ) . Because of mechanical problems encountered with the peristaltic pump, the experiment had to be terminated after 2900 h (120 d ) . However, the final 137Cs concentration in the eluate was larger than 0.5 C o . Since the last points on the elution profile were somewhat scattered, only approximate values for R and k, were obtained. They are 680 and 1.87 cm, respectively. After three days contact time under static conditions, a ka value of 5.6 cm was determined. Considering that the residence time in the dynamic experiment was relatively short, the agreement is quite good. The sorption coefficient calculated using the retardation factor is also significantly higher than that calculated for the ISB-1. This suggests that alteration minerals, namely, clinochlore and muscovite, present on the fracture surface of ISB-2 but not identified on ISB-1, might play a significant role in the sorption of u l C s and lead to a high value of the retardation factor. This suggestion would be consistent with observations made by several authors (4,5,15).
2
5
4
400
600
800
1000
Volume eluled (mLI FIGURE 5.
ELUTION PROFILES FOR THE TRITIUM AND 1 J 7 Cs IN ISB-2.
INJECTION
OF
Since no elution profile could be produced, the block was taken apart and an autoradiograph of the fracture surface was obtained (not shown). No activity was detected on the fracture surface except for a blurry band along the inlet channel indicating that the colloid had migrated less than 0.5 cm into the fracture. This indicates that the rather large colloids (1-10 um) settled in low locations and a higher water velocity would have been required to mobilize them. Additional experiments with colloids of smaller size are necessary to better understand colloid migration.
Experiment #4: Intermediate-scale block-5 (ISB-5) An experiment was conducted to determine the influence of redox conditions on the migration of 95l Tc through a natural fracture in granite. A "Tcff4 solution was injected at a flow rate of 0.45 mL/h through a natural fracture in ISB-5. The water residence time in the fracture, determined from the uranine elution profile, was 7.8 h, and the linear velocity was 340 m/a. 95m
The pettechnetate solution was injected through the fracture until a breakthrough was observed and a constant activity was measured at the outlet. At this time (150 h ) , hydrazine was added to the injected 95l "Tc solution to create a reducing medium. Separate experiments, conducted to study the reaction between 95» T c 0 - a n d hydrazine in the absence of granite, showed no decrease in 95 "Tc activity in solution due to the formation of a technetium precipitate, even though the redox conditions and pH were favorable to the formation of Tc(IV) species (17). The formation of Tc(IV) complexes (such as carbonates) or an incomplete dissociation o£ hydrazine at pH=B (18) might have prevented the precipitation of Tc(IV) species. Following tha addition of hydrazine, the 95»Tc activity in the eluate started to decrease after - 7.8 h as shown in Figure 6. The injection was carried out for a total of 490 h, until no apparent change in the 95 »Tc activity in the eluates could be observed. Using the calculated inventory of *5"Tc sorbed on the iiactuie sutiatds and assuming 95 "Tc is evenly sorbed on the entire fracture surface, a k. value of 0.09 cm was estimated. This k# value is very close to that of 0.11 cm obtained in the static sorption experiment for a contact period of one day.
CNS 9th ANNUAL CONFERENCE. 198B 177
(e) Chemical stripping techniques have been used to elucidate the sorption mechanisms involved in the retardation of migrating species. It has been determined that 137 Cs is retained on a natural fracture surface by cation exchange arid on iron oxihydroxides by theraisorption. It is possible that. Tc(.IV) c&i:tKK«s.t.e complexes were formed and sorbed through ion exchange and on iron oxides present on the fracture surface. (f) Static sorption coefficients can assist in determining the relative importance of various radionuclide-rock interactions- A good agreement was found between static and dynamic sorption coefficients for 137 Cs and Tc(IV). However, a discrepancy was observed for 75 Se and more work is needed to resolve it. no
no
Volume ImLI FIGURE 6.
ELUTIOM PROFILES FOR THE INJECTION OF 95"Tc IN ISB-5. NjH, WAS ADDED AT t = 150 h.
The results of the post-experiment analyses provide an explanation of the sorption mechanisms that led to the retardation of 95"Tc. The autoradiograph of the fracture surface (not shown) revealed that **Tc was mostly sorbed close to the outlet reservoir. This observation suggests th3t the reduction of TcO," to Tc(IV) is slow compared to the residence time of the solution in the fracture. Also, the chemical extractions (see Table 3) indicated that most of the extracted Tc was associated with the exchangeable and carbonate fractions. Less than 40X was associated with the iron oxides, which were expected to play a dominant role in the Tc sorption on fracture surfaces (17,19). It is conceivable that a complex such as a Tc(IV) carbonate was farmed after the addition of hydrazine and was later sorbed by ion exchange and by replacement of OH" groups of iron oxides.
ACKNOWLEDGMENTS The authors would like to thank P. Vilks for the preparation of the labelled natural colloid and K.V. Ticknor for the preparation of the synthetic solutions and producing the autoradiographs of the fracture surfaces. Also S. Ramsay for the typing of this manuscript. REFERENCES (1) VANDERGRAAF, T.T., ABRY, D.R.M. "Radionuclide Sorption on Drill Core Material From the Canadian Shield", Nucl.Tech. (1982), 57, 399-412. (2) KAMINENI, D.C., VANDERGRAAF, T.T., TICKMOR, K.V. "Characteristics of Radionuclide Sorption in the EyeDashwa Pluton, Atikokan, Ontario", Can. Mineralogist (1983), 21, 625-636. (3) ABRY, D.R.M., ABRY, R.G.F., TICKNOR, K.V., VANDERGRAAF, T.T. "Procedure to Determine Sorption Coefficients of Radionuclides Under Static Conditions", Atomic Energy of Canada Limited, Technical Record, TR-189, (1982).
SUMMARY AND CONCLUSIONS Our experiments have demonstrated that the migration of a radionuclide can be strongly influenced by chemical interactions with the material lining the fracture surfaces. The results obtained and the metht>*©\og} 4eMslop«t3 at« being used to kelp design larger-scale experiments, using large quarried granite blocks (1 n x 1 m x 0.6 m) (10). The results can be summarized as follows: (a) The migration of 7*Se along one natural fracture in granite was not retarded by interaction with the minerals lining the fracture in ISB-1. This observation is not consistent vith the results obtained for 75 Se under static conditions. This discrepancy might be due to the slow kinetics of 75 Se interactions under dynamic conditions. (b) 131Cs strongly sorbs on natural fracture surfaces and, therefore, its migration along a fracture is much slower than that of groundwater. Alteration minerals on the fracture surface play an important role in the sorption of 1 3 7 Cs. A greater retardation of 137 Cs in ISB-2 as compared to ISB-1 was attributed to the presence of clinochlore and muscovite. Cc) At relatively slow groundwater flows, large colloids show a strong tendancy to settle in low locations along a natural fracture. (d) Reducing conditions significantly increase the retardation of pertechnetate probably due to reduction to a more readily sorbed Tc(IV) species. 178 CNS 9th ANNUAL CONFERENCE. 1968
(4) TICKNOR, K.V., VANDERGRAAF, T.T., KAMINENI, D.C. "Radionuclide Sorption on Minerals and Rock Thin Sections Part I. Sorption pn Selected Minerals", Atomic Energy of Canada Limited Technical Record, TR365, (1985). (5) TICKNOR, K.V., VANDERGRAAf, T.T., KAMENENI, D.C. "Radionuclide Sorption on Mineral and Rock Thin Section Part II. Sorption on Granitic Rock", Atomic Energy of Canada Limited Technical Record, TR-385, (1986). (6) TICKNOR, K.V., VANDERGRAAF, T.T., JUHNKE, D.G. "The Effect, of Laboratory Time-Scale Hydrothermal Alterations on Igneous Rock Coupon Radionuclide Sorption/Desorption", Atomic Energy of Canada Limited Technical Record, TR-376, (1986)(7) WALTON, F.B., MELNYK, T.W., ROSS, J.P.H., SKEEV, A.M.M. "Radionuclide Sorption Mechanisms and Rates on Granitic Rock", Amer. Chem. Soc. Series No. 246, p 45-66, (1984). (8) DREW, D.J., VANDERGRAAF, T.T. "Small-Scale Radionuclide Migration Experiments", In preparation. (1986). (9) VANDERGRAAF, T.T., GRONDIN, O.H., VILKS, P., DREW, D.J. "Radionuclide Migration Studies in the Laboratory", Proc. 2nd CNS Conf. Winnipeg, 7-11 Sept. 1986, p 142-150.
(10) VANDERGRAAF, T.T., GRONDIN, D.M., DREU, D.J. "Laboratory Radionuclide Migration Experiments at a Scale of 1 m", Proceedings of the MRS Conf. Boston, MA, Dec, 1987. (11) TAMMEMAGI, H.Y., KEYFORD, P.S., REQUEIMA, J.C., TEMPLE, C.A. "A Geological Reconnaissance Study of the Lac Du Bonnet Batholith", Atomic Energy of Canada Limited Report, AfCL-6439, (1980). (12) TICKNOR, K.V., VANDERGRMF, T.T- "A Method for Producing Autoradiographs of Rough Surfaces", Nucl. and Chemical Vaste Management (1986) 6, 233-239. (13) CHAO, T.T., THEOBALD, P.K. "The Significance of Secondary Iron and Manganese Oxides In Geochemical Exploration", Economic Geology, 71, 1560-1569, (1976). (14) TESSIER, A., CAMPBELL, P.G.C., BISSON, M. "Sequential Extraction Procedure for the Speciation of Paniculate Trace Metals", Anal. Chem, 51, 844-851, (1979). (15) WALTON, F.B.
private communication.
(16) TICKNOR, K.V., HARRIS, D.R., VANDERGRAAF, T.T. "Sorption/Desorption Studies of Selenium on FractureFilling Minerals under Aerobic and Anareobic Conditions", Atomic Energy of Canada Limited Technical Record, TR-453, (1988). (17) PAQUETTE, J., REID, J.A.K., ROSINGER, E.L.J. "Review of Technetium Behaviour in Relation to Nuclear Waste Disposal", Atomic Energy of Canada Limited Technical Record, TR-25, (1980). (18) GALATEANU, I., BRATU, C , PETRIDE, A. "The Reduction of "Tc-Pertechtate by Hydra2ine", Radiochem. Radioanal. Letters 28(1), p 95-104, (1977). (19) VANDERGRAAF, T.T., TICKNOR, "Reactions Between Technetium in Containing Minerals Under Oxic and Amer. Chem. Soc. Symp. Series 246,
K.V., GEORGE, I.M. Solution and IronAnoxic Conditions", p 25-43, (1984).
C N S 9th ANNUAL CONFERENCE, 1988 179 I
Session 5: Reactor Commissioning/Decommissioning
Chairman: K.H. Talbot, Ontario Hydro
CNS 9th ANNUAL CONFERENCE, 198B
181
ONTARIO HYDRO NUCLEAR GENERATION DIVISION COMMISSIONING WORK AT DARLINGTON NGS A - IMPACT OF EXTENSIVE COMPUTER APPLICATIONS
By:
D.R. McQUADE
Ontario Hydro Darlington NGS A
ABSTRACT The extensive .utilization of computers has had a significant effect on commissioning at Darlington Nuclear Generating Station. The impact of computerization may be examined by presenting the challenges and benefits we have experienced in various applications of computers. Early problems with liroited documentation, vendor and design support are being overcome. The first major systems with computer application, Water Treatment Plant and Tritium Removal Facility, show that the computerization is a major contribution to their successful performance. We expect our other systems to be equally successful.
- Changes affecting design intent (ECN's): additional trips related to Moderator S Feedwater Systems, support systems for ECI piping and changes related to environmental qualifications. - Licensing Requirements: Request to load D O to Moderator-approved; approval to load fuel - request made, awaiting response. - Tritium Removal Facility-anticipated in-service date is October 15. The first runs to remove tritium from D O will be carried out in early June. - Unit 1 status: for operations, it is at the infant stage with a few system turnovers received. Construction is progressing well with the major structural and mechanical work complete.
DARLINGTON OVERVIEW Darlington Nuclear Generating Station has four reactors which are owned and operated by Ontario Hydro. Each reactor has a net electrical rated output of 881 MW(e). The first DO Tritium Removal Facility has also been located at Darlington to service all Ontario Hydro nuclear stations. Darlington NGS is located at Bowmanville, Ontario; 72 km east of Toronto, 10 km east of Oshawa. Commissioning Status: For the first unit, Unit 2, to start up we need 381 systems turned over from Construction in Units 0 and 2; as of the end of May, Operations has 290 of these systems. At present time approximately 50* of commissioning activities required for Unit 2 (critical) are complete. Unit 2 milestones are Fuel Load - June 22, 198B; Criticality September 20, 1988. The major issues to be resolved before Unit 2 critical: - System Commissioning Complete: We should be able to make September 20 date. - Licensed Shift Supervisors and First Operators Authorization Training: sufficient candidates prepared for present design but must keep, up with late changes which affect design intent,
- Units 3 and 4 status: Unit 3 is enclosed and reactor is in place. Unit 4 below grade work is complete, above grade work progressing well vith early pours of concrete to the reactor building and steel work for the turbine hall well underway.
INTRODUCTION Many papers have been presented in the past about commissioning activities at CANDU power plants. This paper will focus on the impact on commissioning of the extensive application of computers. Computer applications are not new to CANDU's and much has been developed and learned from earlier projects within Ontario Hydro. The growth in application has been steady and methodical. Initially at Pickering NGS A there were two computers for power control, annunciation and fuel handling. Bruce NGS A saw the separation of the fuel handling with dedicated computers. Pickering NGS B expanded the scope with more sophisticated CRT displays. Bruce NGS B went even further by computerizing more of the major process systems. At Darlington, the post Three Mile Island (TMI) era influenced greatly the man/machine interface and the full computer application to all the major process systems that include reactor power, heat transport pressure and inventory, moderator
CNS 9th ANNUAL CONFERENCE, 1988 183
temperature, boiler pressure and level, deaerator level, annunciation and extensive graphical displays. In addition to these, virtually all of the other plant systems have 'computers' which have had a major impact on commissioning. These impacts can be summarized briefly as follows:
the station processes; a. total of '6' Hot Spare Systems and Developmental Systems on site. This gives a grand total of '1450' computers to understand, operate, maintain and to keep the documentation up to dateSPECIFIC SYSTEMS
i) Complexity of the System - Computerization results from trying to manage complex systems. So far computers have not made life more simple - but they m\?.y, in timeii) Increased Manpower Requirements - To cope with commissioning and putting the various systems into service we require more people initially than with the more traditional systems. This initial group of people require considerable training before they can even begin to work. iii) Increased Interfacing Problems - In addition to the 'normal mix' of mechanical, electrical and I&C, add to this the vital element of computer hardware and computer software, then you have just about squared your interfacing problems. iv) Development/Testing Facilities - For the various types of computerized systems it is necessary to have off line systems upon which software can be run and tested. These facilities require extensive logistical support areas complete with air conditioned rooms and attendant administrative services, people and space. v) Ho_t Spares - In addition to (iv) above, Hot Spares are required to ensure the prompt ava?lability of spares and the repair and tes".ing of faulty components. vi) Software Mi nagement - Add to all of the above the need to maintain control over all software and the associated documentation, then it is obvious that along with the benefits of computerization there is some pain in the process. vii) Staff Training - A very significant effort is required to train the various management and trades staff. Like most training, this is usually not a one shot activity but an ongoing constant requirement to maintain expertise in the most volatile of all our human resources, the 'computer types'.
Appendix tt 1 lists the computers by type, Appendix # 2 lists the function and the type of computer used. The following is a brief description of each computer application. Water Treatment Plant The Water Treatment Plant (WTP) is the first in Ontario Hydro to successfully function completely on automatici It was also the first system commissioned at Darlington so we were /ery pleased with our early experience, particularly when it was recognized that the plant produces the best demineralized water in North America, according to our Chemistry Unit. The design, installation and start up was contracted to Degremant Infjlco Ltd. The equipment used is seven Modicon 481 Programmable Controllers with three different programs: i) Pre-treatment and Chlorination ii) Demineralization iii) Resin Regeneration and Neutralization These programs also provide the Common Processes Computer in the station Main Control Room with annunciation and trending. The various WTP graphics are also displayed at the WTP Control Panel. Problems There were no manufacturers drawings or schematics of the Modicon 484 provided for maintenance staff. There is a potential problem with loading software: there is no identification so a tape from any system can be installed in any controller, e.g. a Standby Generator tape can be put into WTP (request to Design to correct). There are potential problems with software change control; we must follow change control procedures. Two minor hardware problems were encountered: i) controller with defective CPU board ii) defective input module .
DARLINGTON - 'COMPUTER' APPLICATIONS Benefits The following provides a brief description of 'computer' applications at Darlington. Not all the applications are computers in the true sense. Some systems are a high concentration of electronic components with varying amounts of software and firmware, where they essentially replace the more traditional 'hard wired relay logic 1 , this paper has treated them as a 'computer'. Thera are a total of ' 94' computers and '1350' programmable controllers (PC) used in
1M
CNS 9th ANNUAL CONFERENCE, 1988
The benefits of programmable controllers: solid state devices with high reliability and fast response, simple on-linp programming and troubleshooting, wide ranqe of functions include logic, timing, calculating, sequencing and easy removal and replacement of components. There is reduced downtime: the train service cycle is 20 hours and automatic re-generation is 2.B hours compared to manual regeneration which was 6 hours. There is high availability of the
train. Chemical addition is automatic. It allows for operator intervention since control modes include automatic, remote manual, local manual. Standby Generators Each of the four, 24 MW Standby Generator has a Modicon 484 which has replaced traditional hard-wired relays for logic control. In addition, for fault diagnosis each Standby Generator (SG) has a Dranetz Sequence of Events Monitor which monitors the changes of state of selected input events during SG start up and run down. In the event of failures or major deviations from the norm, this monitor wi11 pin point the problem area. Benefits i)
ii)
With Modicon PC: -fast response an<3 inter tal cycle time -logic changes can be easily performed with P190 or P180 Programmer and logic can be monitored for faults. -the program is maintained on loss of power by battery back-up but is lost if there is both a power failure and battery rundown. With Dranetz Sequence of Events Monitor: -buffer to store events for later printout -easily programmable with front panel keyboard -useful in monitoring start ups and shutdowns and in particular for troubleshooting.
Problems i)
With Modicon PC: -as with WTP, lack of schematics/drawings for maintenance staff, potential problem with software change control, potential problem with loading software since no unique identification for each controller -input/output module failure attributed to infant mortality -one mainframe replaced due to faulty connector.
ii)
With Dranetz Sequence of Events Monitor; -no read out -one of four returned for repair under warranty.
Tritium Removal Facility The Heavy Water Management System Computer provides control and monitoring functions for the Tritium Removal Facility (TRF) and the Station Upgrader. The Control System consists of; i) Eight independent process control computers for automatic process control.
Five computers are dedicated to the TRF Sul2'.
CNS 9th ANNUAL CONFERENCE, 1888 1»5
Problems
Benefits
For Construction/ the Check and Test Support Services were not available, i.e. no power, no air conditioning, unfinished computer rooms, dusty construction environment. Computer cabinets did not nave adequate ventilation anc there were problems with computer system instrumentation grounds. Since no isolation devices were installed on computer I/O points, 1400 disconnect terminal blocks were retro-fitted.
DCC's are designed to be easy to change, fast, accurate and flexible. Features allow the computer to perform many functions to assist the operator and control the process. Use of light pens allows the operator to have greater access to displays and reduces required memorization (improved man machine interface).
Adequate and accurate maintenance documentation (for hardware and software) was not supplied and is difficult to obtain. The computer manufacturer is not under contract to Hydro. Sulzer is the prime contractor but does not have computer expertise- Computer experts are resident with Fischer & Porter parent company in the United States a id are not readily available to assist the customer in Canada. It is difficult to keep software media up to date and this requires control resulting in a major administrative exercise. A spare off-line computer for software maintenance is required. Failure to install updates to executive software of process computers made early commissioning very difficult by 1 rediscovering' problems with the software. Operators needed some adapting to get used to monitoring the plant via CRT's. The Annunciation System is poor. (Under review). There have been many revision levels of computer boards although there have been no problems to date. Calibration procedures were not supplied for much of the computer hardware and had to be locally developed. Indicator bulbs were burning out in the TRF mimic panel. The solution was to lower the DC voltage supply.
Digital Control Computers (DCC's) Each reactor has its own DCC to control and monitor main reactor processes at highest safety level and lowest production cost. The DCC's; -
1M
control reactor power - reactor regulating monitor channel temperature and flow control haat transport pressure and inventory control steam generator level and pressure control moderator temperature control deaerator level scan and monitor approximately 10,000 annunciation points control liquid zone/helium pressure provide step back monitoring provide channel power mapping monitor unit power regulator contains xenon predictor provide historical data collection provide hardware check provide electrical metering and display contain display and logging software (bar charts, graphical trends, sot points).
CNS 9th ANNUAL CONFERENCE, 1988
Historical information is logged on mag tape which will be convenient for information retrieval (post upset reviews). A record of operator changes to system parameters and software and hardware error logs are saved on separate disc files which saves the Technical Section time. Problems There is difficulty in securing appropriate support and cooperation from Design and from suppliers due to aging technology and fast changing staff complement. There are problems with software design and specifications. Applications software are specified by I&C and Electrical groups, software design and testing is done by Computer group with minimum input from others. Limited cross-experience, i.e. I&C unfamiliar with computer and Computer Group unfamiliar with I&C and Process Systems; It is difficult to use a single person or group for troubleshooting. Separate groups work on software and hardware. Few people understand both, but these groups should be working together. Design reviews were complicated and in some cases, untimely. They did not achieve the intended goal to the full extent. There were problems with software/hardware commissioning. Due to the extensive and diverse application of the process on the DCC's, it was not possible to treat them as one integrated system, comprised of hardware and software and process. The commissioning responsibility was divided into hardware, generic software and application control or monitoring software. This diversity led to multiple interfacing problems with overlaps and gaps, e.g. the Hardware Group were not familiar with software and bow the process system is supposed to function. Commissioning of process control/monitoring software is done by engineers with little software experience, therefore the Computer Software Groups must provide support and training. Control Maintenance lacks complete training; hardware training not complete and software training very limited. Hardware diagnostics is complex and requires software knowledge and therefore the Software Group must provide support. The Software Group has over 150 programs to commission as well as perform all other functions; originally the group had two people. With control/monitoring software, one person has one to four programs to commission.
Software turnover is still not complete and usually has some buqs in it. Software for all applications was turned over all at once. It would have been better if it was slowly built up and added to and worked on one program at a time to correct it. Due to late arrival of the software there was lit tie time to review chanqr-s and install packages on time. Operations has found over 400 problems/snags requiring a large amount of paperwork. The number of snags is not out of line with the extent of the software. A large number of software changes need frequent updates to procedures and manuals. Any software changes which were needed for criticality needed to be identified by February 1988 to be officially delivered and tested due to long lead time {approximately 9 months). Changes are easy but not quick. Troubleshooting tools do not work well because they are not fully tested by Design. If one proqr
There were many problems with the hard copier system. Frequent recalibrat ion was required duo to circuit sensitivity to power supply voltages. Historical information is stored on tape but there is inadequate recovery capability. Grounding details on most equipment were incomplete. Hardware design requiring improvements: there were problems with seating DEC circuit boards. I/O boards cannot be removed on power, therefore the computer must be taken out of service to swap a board. Closely spaced I/O boards can easily cause shorting when removing the board on power and may short- the power supply. It is very time consuming to locate a faulty device on UNIBUS. Most of the problems are currently under review and corrective actions planned or in hand. Common Process Computer Conditions are similar to the Digital Control computers (DCC's). Fuel Handling ine Fuel Handling computer controls, monitors and logs operation of three fuelling machines to allow on power refuelling of four reactors. There are 11 computer systems: six computers (one for each headj are assigned to control fuelling machines when they are are on Ihree trolleys, four computers are used to control Fuel Facility Auxiliary Area
CNS Otn ANNUAL CONFERENCE, 1908 1«7
Renefits
Operat i ons staff must be capable nf mak i ;ig MILchanges.
Benefits will occur during operation as opposed to commissioning. System operation and exceptions arc all logqed. Theie xs built in protective logic for equipment and easy interface with the operator via graphical displays.
There has boon limited document-f knowledge of fibre optics among the Tochnica1 and Maintenance staff.
Depaitment Business Computer The Department: Business Computer is used extensively station wide. All Operations staff use it daily. Work is done electronically instead of on paper. This is a topic of another paper at this conference produced by I.S. Hey.
-
The systems availablo are as follows: Work Management System M.^tnrial Management System Equipment Spurt) Parts Staff Information System Sparc Parts Froiessional •ifficn System Schoduling Systems Cost Reporting and Control BASIS: -houses all static) procedures and "locumer.tat ion -1-jq of all documentation received from o f f - s 11 e -w.>rd s e a r c h c a p a b i l i t y
The D e p a r t m e n t B u s i n e s s C o m p u t e r t i e s the i?r..id nfficcj C o m p u t e r .
into
Chore] :>r ry M o m t c r i n q Th<- C h e m i s t r y M o n i t o r i n g S y s t e m is ubed a s a '.-hemistry c o n t r o l analytical d a t a b a s e , a n d is locaf--"! ir. tt '• rh"iais*-ry L a b o r a t o r y . It h o l d s •i 1 ] r<-».)!] i r* d s p u n f i c a t i o n s for e a c h s y s t e m , • ;nnifnir".s -nc*" Jd 1 s aqa inst s a m p l e s and r e s u l t s ar>- dr-[ tctfd in n usoful g r a p h i c a l form. Access is .ilsn qainr-rj from the M a i n C o n t r o l R o o m t o p r o v i d e .KTuratf.- and t i m e l y i n f o r m a t i o n . •/HJ8O Pro'jr.immabLi' C o n t r o l l e r s (PC) T h e ^ ml 80 Pro
W<- must b e able? tn make f i r m w a r e c h a n g e s o n f.ite a s anon -in the fiold r e q u e s t s a c h a n g e , t h e f f o r e a p r o g r a m m i n g Kt/iiion is neodod (cot u p -it I).i r 1 I n'|t "") , Thin costr, monf?y a n d r e q u i r e s !ip.i'-<- .ind imnit tn- kept in good w o r k i n g o r d e r .
188 CNS9th ANNUAL CONFERENCE. 1968
There were several board failures, in particular with power dupply boards (5* failc-d; Operations and Construction} . These fai 1 urns -im under investigation by Ontari o Hydro Research and Design and it is a suspected 'burn in' problem. There have been delays in repairs because, loose parts were not purchased by Desiqn. Good boards were taken apart to repair failed boards. Some failures are duo to lack of <,-xporif.nce by maintenance st
Control
P r o g r a m m a b l e C o n t r o l l o r s (PC) are usi'd o n the G e n e r a t o r T e m p e r a t u r e Moni''oring (GTM) S / s t e m , the Moisture Separate Rcheatcr (MSR) control and Kotor Temperature Moni torinq (RTMR J. Commi ssioninq of these devices has not yet begun. Brown Boveri Canada's (BRC) Decont ic/Turbomat is a hard-wirod controller. Systems have now been developed by bBC to use programmable controllers and Darlington is BBC's lust generation of Doconti c. Control Log ic is implement ed vi a hard-wired logic and printed circuit boards. Turbomat is a sequential control lor used tu startup/shutdown the Turbine Generator Auxiliary Systems with minimal operator i nvolvcment, It places all auxi1i ary systems in service: turbine, generator, turbotrnl, excitation, etc. It runs up the t urbi ne tjenorator from stand still to full JIDWIT, monitors all important supervisory p.ir.imctcrs (vibration, trmpor.it ure, exp.msi cms) .mrl hits excel lent t rouble shoot i n<) c.ipabi 1 i t y U K i nq Incit 1 annvinic.it inn. Stmuhil inn moduli1 r. ,tr<' available tn .iKfijct in t roubl I'-ihcmf i n
Problems The Turbine Generator Controller is missing wiring information and has incorrect wiring information between BBC and Hydro. The BBC drawings required updating to Hydro's standard. BBC electronic modules do not have equipment codes, thus there is difficulty in using the Departmental Business Computer for tracking and repairs. Control Maintenance is developing a method to handle this. There was a problem with signal grounding and DC power supply. The supply was to be 24 Vdc but nominally it is 29,2 vdc and as high as 30.2 Vdc. The overvoltage trip setting of 30.5 Vdc has been exceeded on several occasions. Benefits Factory testing of the control panels is performed before shipping to site. A large portion of control logic commissioning can be accomplished before wiring is complete. This should greatly improve final turbine generator commissioning. The time required to troubleshoot is significantly reduced. Shut Down Systems 1 and 2 The two reactor Shut Down Safety Systems are SDSl and SDS2. Their purpose is to insert large amounts of negative reactivity to make reactor go subcritical whenever unsafe conditions exist. Darlington is the first nuclear station to fully computerize the reactor shutdown systems. Each of the independent Shut Down Systems has seven computers: Onp Monitor Computer annunciates abnormal conditions, stores historical data, generates status displays for operators and provides operator interface for control and safety s^ stem testing. Three generate displays Computer
Channelized Display/Test Computers channelized process and neutronic and respond to commands from Monitor to control testing of trip parameters.
Three Channelized Trip Computers read analog signals from trip parameter sensors, compare signals to set points and initiate a channel trip if set points are violated; or if there is a problem with the computer, detected via its self checks. The computers interface with relay logic to finally drop the shut off rods or open the liquid injection valves. Problems The design is incomplete: some software delivery is late, a maintenance computer was
not designed or delivered to meet Operations' requirements. Lete integration testing caused commissioning difficulties. Watchdogs time out because programs are slower than expected. Jumpers were installed to enable upward communication between computers. Termination modules meant to protect computers against transients introduced unacceptable time delays and induced failures on computer boards. Temporary regulated 48 Vdc power supply was needed because of failing digital output transistors due to too high station battery voltages. The voltage is not 48 Vdc as required but is between 53.5/55.9 Vdc. There was limited engineering support from vendors. Modifications were done by AECL, not the vendor because of vendor organizational changes. The manufacturer's manual was out of date, late in arriving, and lacking detail. The computer stalled without displaying error codes so cause of failure required extensive troubleshooting. There were materials shortages: incomplete board sets and limited component spare parts. Ground separation was needed between AC and DC grounds. Complex procedures are required for change control. There are licensing concerns related to the confidence in the software (currently under review). Benefits The computer allows for implementation of more complex logic than possible when using relays. There is additional information for Operators with use of computer displays. There is more advance warning of upset conditions with introduction of features such as Margin to Trip. Manpower requirements are lower for NOP calibrations. There is more effective control over execution and recording of safety system tests. It is easier to revise logic than make wiring changes. Displays can also be changed easily. Reliability testing is built into the systems. These tests are initiated and progressed by Operator prompts. Sequence of Events Monitoring Computer (SEM) This system is used mainly to monitor the operation of relays and circuit breakers related to the various electrical distribution systems. The SEM System carries out the following functions: - relates electrical disturbances to system disturbances - records status changes of circuit breakers and relays in chronological order on a magnetic tape - retains most recent 1000 status changes - provides hardcopy on printer
CNS 9th ANNUAL CONFERENCE, 1988 189
Problems Thert? is no historical retrieval for in format ion saved to tape. 11 is not use r friendly in assisting the Operator in determining the cause of an event. The software as delivered had many functional problems. It is not automated so the Operator must manually call up a sequence. The original purpose of SEM was to provide the Operator with automatic diagnostic capability but this is not provided at the present time. The wiring between the SEM and initiating relays was not complete. As relays send 125 vdc input to SEM, there is a shock hazard for personnel. There is shorting of integrated circuits due to poor design of back lit push buttons. Vibration Monitoring The Vibration Monitoring System provides alarms indicating high vibration Levels on major pump sets, e.g. Heat Transport, Moderator, Boiler Feed Pressure, Condensate Extraction Pumps, Low Pressure Service Water, etc. It has annunciation, monitoring and trending functions. It uses many pieces of equipment manufactured by different companies that were never assembled to form one system. Problems Due to involvement of several Design disciplines, difficulties were experienced in properly interfacing IRD instrumentationComputer hardware documentation was not supplied by OH Design or OH Research and had to be obtained directly from the manufacturer. Two chips on each decoder card of multi-point scanner were of the wrong type and were replaced. The system uier's guide is not too clear. Software commissioning has just started. Documentation must be kept up to date as software changes. The documentation received from designers and manufacturers was not sufficient.
Grounding - There is no clear and consistent specification of the grounding facilities to bo provided or definition of any special requirements. This seemed to fal1 between the crack of Electrical,, ISC and Computer Hardware organizations. Multiple Manufacturers - Different standards are adopted by different manufacturers wi th the extensive variety of equipment. This causes many problems with documentation, maintenance, troubleshooting, spares, etc. In days gone bye a plant would, in general, use one manufacturer's equipment. Board Revision Levels - There are multiple revision levels for boards on many computer systems. This has not caused a problem to date but keeping track of thousands of circuit boa "Ms is a challenge. Design/Supplier Support - Limited support from design and suppliers/manufacturers on many computers systems is a problem. The long project cycle time results in the fading of the suppliers 'corporate' memory. Operations are regularly greeted with news such as, the company is out of business, or no longer makes the model, and all who knew about the product have retired. Software Changes - Many software changes are needed, e.g. TRF over 150, programmable controllers over 150, DCC's over 400 snags. Most engineering changes now result in software changes. Implementing software changes is generally a protracted exercise when ensuring all parties are appropriately consulted and all documents are in order. Obsolescence - The design and construction cycle for a nuclear station is 13 years so that all computers are obsolete before the first unit starts up. This creates problems in getting technical support from the vendor. Experts now have moved onto other projects and no experts are available on old computers. There are problems in getting loose components for repairs and spares because they are no longer manufactured. Maintenance - Extensive maintenance facilities for software maintenance, testing and control are needed. The station cannot rely on outside sources for support as it does not exist.
Benefits It provides early warning of mechanical troubles or faults on major pumpsets. It provides alarms and diagnostic information and establishes signature condition on our major rotatinq equipment.
Computer Rooms - Computer rooms need to be relatively clean and air conditioned when computers are installed in order to avoid failures. The construction environment presents some problems. GENERIC BENEFITS
GENERIC PROBLEMS DC Power Supply - The various DC battery power supplies are causing board failures on many computer systems. This appears largely due to imprecise specification of the actual voltage ranges equipment will be subjected to.
190 CNS 9th ANNUAL CONFERENCE, 1988
Computers are easy to change, fast, accurate, flexible and reliable. There is ease in making changes; hard wired systems are not as flexible and take much longer to change than software.
Compar . T P can us' 1 mnre complex loqic and have .•rih.viciM rVaturos such as d i s p l a y s for the user (..•..;. n p ' r . i f n r ) . SoftuarD Jws a greater c a p a b i l i t y :or t i n s than hardware. Computers improve man-machi no i n t e r f e r e . hoi-.efits w i l l mostly occur during o p e r a t i o n . IMPACT
ACKNOWLEDGEMENTS Input from the following pr-ople is appreciated; Prank Amantea, Kingsley Butt, Dale Craig, Gary Cloghorn, Blair Crouse, Ian Driver, Ron Hong, Darek Kulczyn.ski, Ray Landry, Bill Oualtrough, Ed Sadok, Ernie Talnariu and Pob Thurier and Norma -Menis for assisting in preparing this paper and arranging for slides and photographs.
The impact of computer application has resulted initially in a greater need for resources, i.e. manpower, equipment, money. More time and manpower is required to do commissioning. More training is needed for personnel in maintenance, operation, :ommis«,ioninq and technical support. Additional procedures and more complex operating and maintenance manuals are required. Increased investment is required to pay for maintenance facilities, additional training and space for media storage. Additional sophisticated tools for repair and troubleshooting are r&quired, e.g. OH180 Field Mor.itor, Computer Boaird Repair Shop. Maintenance facilities and hot spares are required, e.g. Room A15 (0H180, Modicon, PROM programmer, Software Maintenance Computer, also Media Storage), Hot Spares (Modicon 484, Modicon 5S4, nulSO in shop, SDS).
CONCLUSION Darlington is the itiost complex CANDU station built to dstp. The extensive applications of computers has provided many unique commissioning challenges. The initial difficulties with limited documentation, vendor and design support have been addressed and significant improvements made. These improvements are reflected in the very good performance D f C J r first major systems to be commissioning with computer application, Water Treatment Plant and Tritium Removal Facility. The saving in time achieved by the 'ease of making changes' will be considerably reduced by the effort required to 'keep the documentation in order' and maintain the required services and infrastructure. At the beginning of this paper I commented on the significant success of the Water Treatment. I expect that this is the shape of things to come at Darlington and that ultimately each of our systems will be equally successful. Thus the CANDU System will remain at the leading edge of technology.
CNS 9th ANNUAL CONFERENCE. 19B8 191
ONTARIO HYDRO NUCLEAR GENERATION DIVISION COMMISSIONING WORK AT DARLINGTON NGS A - IMPACT OF EXTENSIVE COMPUTER APPLICATIONS
APPENDIX 1 COMPUTERS BY COMPUTER TYPE
Computer Type Totals
In Plant Use 20
DEC PDP-11/70 DEC PDP-11/44 DEC PDP-11/34 DEC PDP-11/23 DEC LST-11/23 DEC LSI-11/02 GEN. AUTOMATION GA-16/220 IBM 4381-Q13 Modicon 484 PC Modicon 584 PC Modicon 884 PC OH180 PC Brown Boveri Turbine Generator Control System
)
1
)
5 1 12
) ) t
10 44
y4 )— ) 'COMPUTERS
48 2 1 1275 24
) ) ) ) )
TOTAL .-
TOTAL: 1350 'Programmable Controllers' (PC)
Hot Spares/Test/Maintenance Computers DEC PDP-11/70 DEC DPD-11/44 for TRF Modicon 484 Modicon 584 for TRF OH180 GRAND TOTAL
192 CNS 9th ANNUAL CONFERENCE, 1988
2 1 1 1 1
1450
)
)— ) )
TOTAL:
6 Support Facilities
-
ONTARIO HYDRO NUCLEAR GENERATION DIVISION COMMISSIONING WORK AT DARUNGTON NGS A IMPACT "I 1 EXTENSIVF: COMPUTER APPLICATIONS
APPENDIX 2
APPENDTX 2
COMPUTERS KY FUNCTION
COMPUTERS P'< FUNCTION
COMPUTER TYPE*
FUNCTION* COMPUTER TYPE** Unit 1, 2, 3, 4: Channel G,I!,J T r i p Computer SDS2 (Yes)
Brown Boveri Turbine Generator Control System -Units 1,2,3,4: Auxiliary Systems (Yes) -Units 1,2,3,4: Excitation Control (Yes) -Units 1,2,3,4: Moisture Separator Rehoater (Yos) Rotor Temperature Monitorinq (No) -Units 1,2,3,4: Specialized Decontic (Yes) -I'nits 1,2,3,4: Electronic Governor (Yesl Units 1,2,3,4: Reactor Control (Ytfs) Hardware Mtce (No) Software Mtce (No) Uni t 0: Moni "_or Processes (Yes) FFAA-1,2: Irradiated Fuel Bay (No) FFAA-3,4: Irradiated Fuel Bay (No) FFAA-5,6 (No) Pressure Test Facility Fuel Handling (Yes) Fuel Handling TrolJey System s Heads System 1,2,3,4,5,6 (Yes) TRF: Man Machine Interface (Yes) TRF: Hardware/Software Maintenance (No) TRF Process Logic (Yes) D«0 Upgrader Process Logic (Yes) TRF s Upgrader Support Systems Annunciation (Yes)
Decontic (total: 4 1 per Unit) Excitation (total: 4, 1 per Unit) Programmable Controllers -(total: 4, 1 per unit) -(total: 4 , V P « * unit) Turbomat (total: 4 1 per unit) Turbotrol (total: 4 1 per unit) DEC PDP-11/70 (total: 8, 2 per unit) DEC PDP-11/7 0 (1) DEC PDP-11/70 (1) DEC PDP-11/70 (1)
DEC L S I - 1 1 / 2 3 ( t o t a l : 12, 3 p e r u n i t , 1 per channel)
General Automation GA-16/220 ( t o t a l : 12, 3 per u n i t , 1 per channel) General Automation Unit 1, 2, 3, 4: GA-16/220 ( t o t a l : Channel D,F.',F Display/ Test Computer SDS1 (Yes) 12, 3 per u n i t , 1 per channel) Ge"ne~Ea\ PttitomatAo'n \}nit \, 2, 3, h: GA-16/220 ( t o t a l : Channel G,H,J Display/ 12, 3 per u n i t , 1 Tost Computer SDS2 (Ye^l per channel) General Automation Unit 1, 2, 3, 4: GA-16/2 20 ( t o t a l : 4 Monitor Computer 1 per unit) SDS1 (Yes) General Automation Unit 1, 2, 3, 4: GA-16/220 ( t o t a l : 4 Monitor Computer 1 per unit) SDS2 (Yes) Design not Safety System Monitor finalized (1) Computer (No) Design not SDS1/SDS2 Hot Spares fi nalized (No) Unit 1, 2, 3, 4: Channel D,E,F Trip Computer bUSl (Yes)
Business Computer (Yes)
IBM 4381-C13 (1)
Standby Generators 1, 2, 3, 4 (No)
Modicon 484 (total: 4, 1 per generator) Modicon 484 (7)
DEC PDP-11/70 (1) DEC PDP-11/70 (1) DEC PDP-11/70 (1) OTC 5QP-LL/70. tU DEC PDP-11/70 (1) DEC PDP-11/70 ( t o t a l : 6, 1 p e r unit)
DEC PDP-11/44 (1) DEC PDP-11/44 (1) DEC LSI-11/02 (8) DEC LSI-11/02 (1)
Water Treatment Plant Vapour Recovery/Dryers (No) D,r' Upgrader Protective Controls & Process Logic (No) TRF Protective Controls & Process Logic (No) Off-Gas Management (No) Feeder Scanner (No) Spares, Test, Mtce (No) Spares, Test, Mtce (No) Unit 0: OH180 (No)
DEC LSI-11/02 (1)
Unit 1, 2, 3, 4: OH180 (No)
Vibration Monitoring (No)
DEC PDP-11/23 (1)
Spares, Test,
Unit 0 SEM: Common f.\cc«ica\ W
DEC PDP-11/34 (1) DEC PDP-11/34 ( T o t a l : 4, 1 per unit)
Mtce (No)
Modicon 484 (total: 36, 9 per unit) Modicon 484 (1)
Modicon 584 (1) Modicon 584 (1) Modicon 884 (1) Modicon 484 (1) Modicon 584 (1) Programmable Controller (175) Programmable Controller (275 per unit) OH180 (3)
Operator Interface iWailatsUitY iu Number of Computers i n Brackets
CNS 9th ANNUAL CONFERENCE, 1988 193
DECOTTUSSIONING OF NPD GENERATING STATION
R.E. LEWIS Ontario Hydro Rolphton, Ontario
BACKGROUND The Nuclear Power Demonstration (NPD) generating station at Rolphton, Ontario was committed in 1955 and began operating in 1962. It is owned by Atomic Energy of Canada Limited and Ontario Hydro, and operated by Ontario Hydro. It was intended to demonstrate the practicability of producing electric power with the CANDU reactor design, and to provide information for use in the design, construction, and operation of larger stations. Over its twenty-five years of operation hPD has prcduced a great deal of information that has contributed to the success of later CAWXJ stations; it is continuing to produce useful information in its decommissioning.
THE DECISION TO SHUT DOWN In 1984 two pressure tubes were removed from NPD to gain information useful in understanding the pressure tube failure at Pickering earlier that year. The relatively high dissolved hydrogen concentrations in these tubes led to a decision to remove more for analysis in 1987. The first tube removed in June 1907 had an even higher dissolved hydrogen concentration, causing such a significant worsening of its mechanical properties that the reactor appeared to need complete replacement of its pressure tubes if the problem was widespread. It was soon found that the problem was widespread, so Ontario Hydro and flECL decided to shut NPD down permanently rather than retube it.
"Ultimate Disposal" - either removing all radioactive material to a permanent repository or permanently filling the existing nuclear structure with concrete or sand or other material. The rest of this paper will describe the planning and execution of the Final Operation work. DECOMMISSIONING PLANS The first NPD decommissioning plan was written in 1985 to derive decommissioning cost estimates as part of a study of the wisdom of continuing to operate the plant. This plan was being revised jointly by Hydro and AECL in 1987 in the months before the decision was made to shut the plant down permanently. The plan (1) has fanned the basis for most of the work done so far. The key are:
It is the only reactor en the site. This means it would be expensive to have full time on-site staff for long term survei 11 ance. The nuclear part of the plant is underground. This makes it practical to remove almost all surface structures. It also means that seepage of groundwater into the plant is a potential problem. Most fuel was already off site. The NPD irradiated fuel bay was designed to ho!d only six months production of irradiated fuel (plus all the fuel in the reactor). Fuel has been regularly shipped to Chalk River Nuclear Laboratories since 1963. This meant that it was practical to remove all fuel from the site with a few months' work.
DECOMMISSIONING OVERVIEW The decommissioning of into three stages:
NPD
divides naturally
"Final Operation" - the removal of all fuel and heavy water, and shutting down all systems. The name was chosen to emphasize to the people doing the work that it is business as usual with the same procedures, policies, and licence as during normal full power operation. "Static State" - a period of about 50 years of radioactive decay. There is no on-site staff planned; monitoring will be done remotely from Chalk River with only occasional site inspections.
FIWL OPERATION ACTIVITIES The main tasks are: 1. 2. 3. 4. 5. 6. 7. B.
194
CNS 9th ANNUAL CONFERENCE, (988
features of NPD that shaped the plan
Defuel Ship all fuel Ship all heavy water Decontaminate Shut down all systems Salvage equipment Modify systems needed monitoring Experiments
for
long
term
1,
Work was begun when tne last fuel bundle left the site on all other systems.
Defuel
The reactor was defuelled by November 2 , 1987 (about a month ahead of schedule) by the regular staff using the existing fuel handling equipment. The design of the f\FD fuelling system meant that all fuel had to be displaced by dummy fuel bundles made of stainless steel pipe and plate.
2.
Ship all fuel
Ship all heavy water
This was completed uneventfully, using standard 200 kg stainless steel drums to ship the inventory of 85 Mg to CRNL.
4.
Decontamination
A study by AECL-CANDU Operations showed it was not beneficial to decontaminate the inside of any closed system, so no such work has been done. The Irradiated Fuel Bay was cleaned with high pressure water jets and manual scrubbing. The active drainage system includes a holding tank which contains severa2 cubic metres oi mildly radioactive sediment. ("Mildly" radioactive because it could all be pumped out to the Ottawa River wi th no viol ation of the Operating Licence,) This is all being removed and solidified for storage. The walls and floors of all accessible areas of the plant are being cleaned. The inaccessible areas have been surveyed as •far as practical. N o substantial amounts of fission product contamination have been found. No decontamination is planned for those areas. All this work is on schedule.
5.
Shut down all systems
The objective is to remove all hazards chemical, electrical, mechanical, radioactive to the extent practical. Work was begun August I , 1987 on nan systems, e.g. the turbine-generator.
6.
is on schedule for completion by
Salvage equipment
This is a substantial part of the total workload and will continue until September 1988.
For about 24 years N°D had shipped irradiated fuel to Chalk River in a cast steel flask holding 25 bundles. The licence for this flask had been suspended because of doubts of its ability to meet current standards. The flask that CRNL rented was much larger, which greatly speeded up the shipping. The old flask was also re-licen
3.
This work July 1, 1988.
All moveable rnaterial and equipment (stores inventory, tools, furniture, vehicles) is being distributed to users in AECL and Hydro. Installed equipment is being removed only if there is a user for it Dr if it is in the way. Because of NFD's historical significance as the first CANDU reactor a determined effort has been made to save i terns that may be of historical interest. The Pickering administration building is getting the handswitch that closed the circuit breaker to send the first nuclear electricity in Canada to the Ontario Hydro grid, and the meter that measured it. A 1arge storage area at the Wesleyville generating station is being "filled with all the control room panels, one of the three channeIs of reac tor regu1ating and protective instrumentation, parts of the turbine and its controls, numerous displays, pictures and models, one of the fuelling machine control computers, possibly one of the original fuelling machine heads and possibly other equipment. All this material will be kept for a potential future museum.
7.
Modify systems need for long term monitoring
One corner of the powerhouse, containing the main ventilation fans, is being gutted of all other equipment and fitted out as a Static State Control Area. This will house controls and power supplies for the ventilation, drainage, and surveillance of the nuclear areas of the plant. Those systems are being fashioned out of existing equipment to the extent possible. This work is on schedule for a hand-over of operating responsibility from Hydra to AECL on September 1, 1988. The nuclear and conventional areas of the plant are being physically separated (e.g. by block walls) to simplify the security of the nuclear section and to prepare for the demolition and backfilling of the conventional section. Hydro plans to remove the non nuclear buildings next year.
itial
Work was begun when the last fuel bundle left the reactor on the systems that had been needed to support it there, e.g. the main heat transport system.
CMS 9th ANNUAL CONFERENCE, 198B 195
B.
Experiments
kPD is the -first CANDU station to have an experimental program ^s a significant part of its decommissioning. The work done for this joint Hydro-AEQ_ program is; 1.
Removing three pressure tubes to determine the location and condition of their garter springs and to provide metallurgical sample-5 of zirconium-niobium, zircaloy-4, and zircaloy-2.
2.
Removing one calandria tube to provide samples of highly irradiated aluminum.
3.
Determining the location of the garter springs on about one third of the pressure tubes to provide information on the performance of "tight" garter springs,
4.
Taking numerous samples of irradiated and tritiated concrete, to investigate the effects of irradiation and the extent of tritium permeation.
5.
Sampling heat transport system pipes and welds to investigate the aging process of carbon steel.
6.
Destructive testing of electrical cables, Df the main generator stator and rotor, and of severa 1 1 arge> motors to he 1 p develop better predictive tests of insulation deterioration.
Apart from the concrete samples which ar& still being removed, all this work is complete, however none of the resul ts has yet been published.
T>€ NEXT STEPS At the time of the decision to shut down, NPD had a permanent staff of 188, ail of whom were guaranteed jobs at other hydro locations. There are now Bl still at fiPD. All of these will depart by September 1 or shortly afterward. Next year it is planned to J"»ave a Hydro construction team dismantle the surface buildings and backfill the underground rooms not required by AECL. Responsibility for control of the radioactive material at NPD will be turned over from Hydro to AHCL on September 1 of this year. At that point "Final Operation" will be over and the "Static State" will have begun.
REFERENCES U) R.W. POCKETT and Operations Plan", Ontario August, 1987.
R.E. LEWIS, "NPD Final Hydro Report NPDO-43,
196 C N S 9th A N N U A L C O N F E R E N C E , 1988
DECOfMISSIONING OF NPD FROM AN AECL PERSPECTIVE
P. ^ATTANTYUS (ATDMIC ENERGY OF CANADA LIMITED)
Ontario Hydra NPD NGS Rolphton, Ontario KOJ 2H0
Over the past 30 years, AECL has been developing the CANDLI nuclear generating stations in close col laboration with the Canadian utilities and manufacturers. Three prototypes reactors paved the road towards full commercial exploitation of the CAMXJ technology which were: - Nuclear Power Demonstration (NPD) reactor at Rolphton, Ontario. - Douglas Point reactor at the Bruce ryuclear Pokier Development in Ontario. - Gentilly-1 reactor in Quebec. AECL has devoted significant resources to develop solutions for handling, storage and disposal of the radioactive materials. As the prototype stations retired from service, AECL gained direct experience in decommissioning of - Gentilly-1 in 1983, followed by Douglas Point station in 19B5. Fol lowing shutdown of NPD, AECL assumed its obligations to decommission the nuclear portion of the NPD facility. The present NPD decnmmissioning program calls for preparing the site for an interim storage period. The intention is to place the plant into a hazard-free state so that it can be monitored remotely from CRNL located 40 km east of NPD, with only occasional cn-site inspection by CRNL staff. To this effect all irradiated reactor fuel and heavy water will be transported to CRNL. The NPD decommissioning program started in July 1987 and will be completed by 1970. The program includes upgrading and detritiation of the station's heavy water and storage of spent fuel in concrete canisters. This paper highlights the history of NPD and describes in detail the planning and implementation of the ongoing decommissioning activities at the NPD site with particular emphasis on the nuclear portion of the facility.
HISTORY OF ixFD In 1954 AECL investigated the feasibility of building natural uranium heavy water power reactors. In 1955 proposals were obtained from the Canadian industry to participate in the design and construction of a nuclear power station and the Canadian utilities were approached to operate the station and utilize the power. From the proposals received, the Canadian General Electric Company Limited was selected to design the nuclear systems and construct the station, while Ontario Hydro was chosen to provide the site, design the buildings and conventional systems and operate the station.
Construction at the site started in 1958 and the station was completed in 1962. Criticality was achieved on April 11, 1962 and full power was achieved on June 2 8 , 1968. The NPD station, although smal1 in size, demonstrated the technical feasibility and cost effectiveness of CANDU concepts and became a valuable facility for testing the developing technology and the training of personnel. In June 1987 during routine inspection of the reactor it was revealed that the pressure tubes had deteriorated to the point where further operation of the station was deemed unacceptable. Since replacing the pressure tubes was not cost effective both AECL and Ontario Hydro agreed that the station should be decommissioned. During the twenty-five years of operation, (from 1962 to 1987), the NPD station produced over 3 mi 11 ion megawatt-hrs of electrical energy. Its maximum continuous capability was 25 MW(e) and the average capacity factor of the station measured over its entire operating life came close to 70X f which is an impressive performance for a prototype power reactor.
DESCRIPTION OF NPD FACILITIES The NPD station is located on the west bank of the Ottawa river, about 230 km upstream from Ottawa. A E C L s research establishment at Chalk River (CRNL) is located 40 km east of NPD facility.
Aboveqround Structures and Facilities The powerhouse complex consists of the reactor and turbine halls (main section) control wing and service wing, which are housed in conventional aboveground buildings. The administrative wing on the west side houses the offices and lunchroom. The control wing between the administration wing and the main section contains the control room, chemical laboratory, change rooms and other service rooms. The service wing on the east side houses the facilities for equipment and maintenance, storage of spare parts, ventilation fans and the filters.
Underground Structures and Facilities The underground conventional portion of the station consists of the condenser and water treatment room, furnace room, air conditioning room and relay room.
CNS 9th ANNUAL CONFERENCE, 1988 197
The underground nuclear portion houses the reactor and associated systems which are enclosed within a thick reinforced concrete structure and shielding walls. The reactor vault is designed to a pressure of 10 psig, while the boiler room is designed to 5 psig. These rooms contain the reactor-boiler circuit. The other rooms house the •fuel handling facilities and the spent "fuel bay, moderator system, end access and the tube removal rooms. The underground nuclear facilities containing the majority of the radiological hazards, occupy a space of approximately 30 K 30 metres with a depth of 24 metres.
Miscellaneous Facilities The dousing tank, ventilation exhaust stack, pumphouse, guardhouse, north and south warehouses, storage sheds and emergency vehicle garage as well as trailers (attached to the administration wing) are the various independent structures within the site security fence.
DECOmiSSIONING OVERVIEW Decommissioning of a nuclear power generating station in Canada is carried out in different stages due to safety and economic considerations. An immediate dismantling and disposal of plant equipment following shutdown is neither desirable (from a radiation exposure point of view) nor feasible until an acceptable disposal site has been designated. It is AECL s policy to prepare its retired nuclear facilities for &n interim storage period, called "Static State" f rather than return it to "green grass" condition immediately following a shutdown. The Static State is a variant of long term "Storage with Surveillance" option (Stage 1 decommissioning as defined by IAEA). Complete dismantling and disposal is planned to take place 50 to 100 years in the future. During this period radioactive sources will decay significantly and the exposure to workers against radioactive hazards is greatly reduced. Dismantling, transportation and disposal (either in situ or elsewhere) can therefore be achieved in the safest and most cost effective manner. Thus the resulting three distinct stages of a typical overal1 decommissioning program, (also applicable to N°D) are: 1. Controlled and safe shutdown of plant systems and preparation for "Static State".
OVERALL DECLTTIISSIONING OBJECTIVES Reduction in AECL s Shutdown of NFD
Expenditures
Resulting
From
NPD has been a joint project between AECL (as the Federal Governments agent) and Ontario Hydro. AECL owns the reactor and associated systems, the nuclear section, while Ontario Hydra owns the site, most of the abaveground bui1 dings and strue tures, the conventional plant equipment and has operated nhe station under a contract to AECL. NPD was not considered as a commercially self supporting power plane. AECL's ccfets, however, have been partially offset by the sale of steam to Ontario Hydro. Since the permanent shutdown of the station, AECL's costs no longer have off-setting revenues and therefore must be defrayed from Treasury Board appropriations. Hence AECL s primary objective is to reduce expenditures as much as passible.
Reaching New Agreement with Obtaining a New License for NPD
Ontario
Hydro
and
Negotiations are underway with Ontario Hydro to amend the AECL - Ontario Hydro 1958 Operating Agreement to define responsibilities of each participant in the decommissioning program and during the static state period. Furthermore, the operating 1icense for the station will expire at the end of September 1988. An application to the Atomic Energy Control Board's Waste Management Division is being made to obtain an operating 1icense for the station as a Waste Management Facility.
Placing the Station into a Safe 5tate I^FD site shall be put safely into a state, which ensures that both the public and the environment are protected during the three distinct stages of decommissioning. This will be achieved by: - Removing or safely containing all conventional and radiological hazards in preparation for the static state. - Implementing the npcessary modifications to reduce the regui red services to meet security, surveillance and monitoring requirements. - Leaving site in such a state that it can be maintained with minimum costs and monitoring remotely with occasional on site inspection only.
2. "Static State" period.
Augment AECL's Expertise in Decommissioning
3. Ultimate or final disposal.
AECL has been a leader in decommissioning power reactors. This experience began with Gentilly-1 and was usefu11y app1ied to the decommissioning of Doug 1 as Point. Wi th each new decommissioning project AECL adds to its expertise and cost data base.
198 CNS 9th ANNUAL CONFERENCE, 1988
BeneTlciar ies or t-tCL s decommissioning i- now ledge are the nuclear plants at their design stage (design wi th deCDflyTu ssicrung in mind) , assessments of environmental impac t of decommissioning and the deccmrussioning cost estimates. For decommissioning projects AECL has therefore the capability to act as a contractor or consultant to nuclear facility owners or to the public sector.
PREPARATION FOR STATIC STATE Final Systems Operation In order to plan the decommissioning program in detail, the status of all nuclear and non-nuclear process systems and structures had to be defined. The sequence of system shutdown had to be developed allowing reactor defuelling, transfer of irradiated fuel to CRM_ ano draining and/or flushing of the heavy water from the nuclear systems. Nuclear waste such as represented by tr^tiated heavy water, and contaminated waste generated during the decommissioning period are to be transferred to CRM_ for storage. Subsequently, the various nonnuclear process and service systems were scheduled for shutdown. Fluids are planned to be removed from all systems and safely disposed of so that neither liquid nor toxic nor flammable materials remain on site.
The cont nment access areas wi 11 l^ave signif leant 1y lower levels of contamination and wi11 be accessed more often for surveillance purpose. The non-fixed contamination from walls, floors and the outside surfaces of piping and components will be removed. The unrestricted areas wi11 have no nan-fixed contamination on exposed surfaces, therefore thorough decontamination is to be applied whenever required. The general radiation fields shall be less than 2.5 micro Sievert/hr (0.25 mRem/hr). All aboveground facilities with the exception of the fan and filter rooms shal1 meet the criteria for unrestricted areas.
Definition of Containment Isolation Requirements (aj
It is intended to achieve as much independence from abo^'eground structures as possible so that the extent of isolation and modification work resulting from the pcssible complete Dr partial removal of aboveqreund structures is greatly reduced.
(bi
The segregation approach 1or radiological zoning shal1 result in keeping the number of penetrations which require isolation to a minimum.
(c)
The isolation methodology shall ensure that the inspection of containment barrier and the isolation points is feasible. This will ensure that if the breach of containment boundaries does occur, it can be identified and rectified.
Establishment of Radiological Containment Zoning The underground portion of the nuclear section containing most of tie radioactivity wil1 be isolated from the regaining structures (both above and underground) b/ designating a containment boundary. The majority of the radionuclides at NPD reside in the activated components in the vicinity of reactor core. Activation and fission products on the other hand have been carried over to auxi1lary systems and various parts of the station in the form of contamination. Based on the radiation fields (mainly emanating from activated componentsJ as well as the spread of contamination (concentration levels of fixed/contained surface contamination of alpha, beta and gamma emitters), radiological zoning criteria have been developed. It is feasible to subdivide the areas within the containment boundary into two zones. The third zone is represent areas which will have no significant radiological hazard, therefore the zoning areas tD be implemented are: (a) (b) (c)
the containment area C . the containment access areas 'CAA . the internal and external unrestricted areas 'IUA/EUA'.
The containment area has relatively high radiation levels with general fields above 25 micro Sievert/hr (2.5 mRem/hr). The reactor vault and dump pipe room are for example not accessible due to the high fields.
Decontamination Program Decontamination of systems and equipment and surfaces (floors, walIs and ceilings) shal1 be carried out based on the following criteria: (a)
To ensure dose exposures to radiation workers during the decommissioning operations are kept as 1ow as reasonab1y ac hievab1e ;ALARA principle).
(b)
To ensure dose surveillance in kept as low as principle).
(c)
To reduce the possibi1i ty of dispersal of radioactivity during the static state period.
The detailed corresponding to developed: (a)
exposures incurred during the the static state period are reasonably achievab1e (ALARA
decontamination strategy the described criteria was
A study was carried out to determine the extent of decontamination required on systems such as the Primary Heat Transport, Moderator, Reactor, Fuel Handling, and Auxiliary systems. It was decided, after assessing the various factors (cost of man rem exposures and ri sk of dispersal of presently fixed radioactivity), not to decontaminate systems but to merely allow the decay of radioactivety in the containment area during the static state period.
C N S 9th A N N U A L C O N F E R E N C E . 1988 199
• D•
tv .-t?^^
ujij!^:
Tur
-iur . P I J I J I I L H
^n, f^ce ac 11 ^ 11 v . u'u'e^ the
the
water .
sur faces
detection
water
proper
discharged
POE>«=.I h i 1 1U
ot
dbsesE^in
wiih
t (x>laminating
i.pm-ji .a i
Lir
r espec t
murpssed
s t a b i 1 I cat i o n
radioriuc i i d f & wJ i 1 be ( at f seel out Svstevns/f oom^ i n CAA, LUA attain
and
1 LJC»'P
wi[I
be
pr t?sc.r Ibed
the
accordingly.
decontaminated
radiu1ogica 1
t I'.ji .t.n«jr.<.
maintenance
Ber-/Lces arid s e c u r i t y
Static
to
define
the
of essential
the f a c i l i t y ,
requirements
i t was
for
services at s i t e ,
the
power
supply
and
lighting
Ham ter nance rcequifed
I^Jturitian
Static
the
such as:
except for
distribution
(both
State
or
breakdevjr. o"*
reduced
the dormancy period. CRNL.
basis.
of
Waste
services
required during
be carried out remotely by
and
periodic be
the
no permanent on-si te
The? s e c u r i t y , surveillance and
The? operating
consists
stage
wi11 have
monitoring functions w i l l
repairs w i l l -
routineP'bser^tiai
State Status
Manaqement Faci11ty
to execute the required functions during
long term storage period of
of
systems J shal1 0& provided on an ' as needed'
staff the
LOntist
^t <• uc t a r e s ,
5/9tems.
unpiannpd &-/€?nt
Dunne-
availability
lor
tr
Long Term lionltorinq and Maintenance Program
necessary
activities
roninq
c r i t e r j a.
In order
i.'^w i t a i
i he maintenance ^foqram 5ha! j
ground of
t h e zoninq areas d e s i g n a t e d as
UJA
the
•-.(-•
tu
for ic)
t
with
'-Jew ' ) -ed radiutui'. 1 ules on ' luor ft.-
operat ion
set ur i t v1 s •-• s T f "T't.
be decontaminated
t sdiusc l i \ i t y
wi I !
i^ater.
t he
nan - i 1 *ed
Cur t h p r m i r e p o t e n t i a l
r.;n,tt.^ and
to c o n t r o l
ifi-Elde-
bf3 decontaminated at
.. ..intai-iiTipcH -JI i i
maintenance
inspections
carried
out
on
program w i l l
of
an
the s i t e and "as required"
basis only.
outdoors and indoors). -
capability
for
ventilation
Iprior
to entering
the
IMPLEMENTATION OF STATIC STATE
CAA and C designate^ areas). Starting July -
ex temal
and
in terna 1
drainage
systems
(outside
and inside the underground section).
to
Hydro.
health physics for service,
visiting
radioprotection,
personnel
(dosimetry
obtaining 1icense
etc).
systems
wil1
security related were i d e n t i f i e d
provide
to respond
event.
CRNL
with
the
a and
waste
team was formed
for
monitoring
to plan and
the NPD static state,
management
providing
facility long
and
operating
term
securi tv,
maintenance
for
the
site.
to any particular
The security
The
requirements
execution
essentially
based on the need for:
detection of un-authonzed entry
through security
Ontario
Hydro
the
decommissioning
assisted by s
the dismantling
gate or doors.
of
work is
being carried out by Ontario H y d r o s f^PD
operating staff -
assumed its obligations
in collaboration with Ontario
implementation of
survei1 lance, Security
necessary information
f^PD
An AECL project
direct the -
19B7 AECL
decommission
various CRNL services.
construction of
the
staff will
undertake
aboveground
buildings and
of responsibilities,
scope of work
structures. -
detection of
-
detection
fire
inside
the structures. The division
of
high
water
level
in
the
sumps
and
duration
following
(drainage systems). -
routine visual (perimeter).
surveillance
-
remote access control
of
security
of
the
activities
are given in the
tables:
fence TABLE 1
-
transmission
of
to the
security
visual surveillance
to
facility. signals,
CRN_
alarms
and
Security Monitoring
MXLEflR AND
SECTION OF NUCLBPR
FACILITY
NON-NLCLE0R
Room. Periodic
monitoring
functions a,re
out to detect any unforeseen event that could
occur.
or
to be carried
RESPONS-
deterioration
IBILITY
integrity
of
OPERATIONS
The major monitoring parameters Scape
-
ONTARIO HVDRO AECL
concrete
structures
foundations, walls, c e i l i n g s ,
such
as
Transport & Storage of Irradiated Fuel,
etc.
heavy
Final Systems Operation
water and operational radwaste to CRM_.
-
integrity
of
carbon
steel
structural
the radiological
inventory.
and system
components. - s t a b i l i t y of
Static State Definition
Duration
200
CNS 9th ANNUAL CONFERENCE, 1988
Static
State
License Application to
Implement-
fiECB
ation July I9B7 - August 1988
As at September 1988, remain until spring 1989 responsibility far site surveillance.
no on-site staff will and AECL will assume security and remote
REFERENCES (1)
"NPD NGS Decommissioning Plan", a report to the Atomic Energy Control Board prepared by AECL, 14-01614-7 (R-0), 19SB March 11.
TABLE 2
RESPONSIBILITY
AECL
AECL (CM BEHALF OF ONTARIO HYDRO)
Scope
Security/Surveillance rtoni toring/Maintenance
Duration
September 1988/March 1989
In spring of 1989 it is envisaged that Ontario Hydro Construction staff will carry out the dismantling work. Responsibility for security and surveillance of the non-nuclear section during the normal working hours will be transferred to Ontario Hydro.
TABLE 3
RESPONSIBILITY
Scope
PECL
O N T W I O HYDRO CONSTRUCTION
Definition of dismantling requirements for nuclear structures.
Dismantling of aboveground nuclear and non-nuclear structures.
Sealing requirements for underground nuclear facilities.
Implementation of sealing of underground facilities.
Securi ty/Survei11ance Security/ Men a toring/Maintenance Surveillance only during working hours. Duration
April 19B9 - September 1989 (estimated)
CONCLUSION The NRD decommissioning project is progressing satisfactorily with the implementation of the Static St.ite. In September 198B the nuclear section will be put into Static State and the remote surveillance of the site will begin by Once the abcveground buildings and structures (with the exception of the ventilation exhaust stack required for purposes of remote surveillance) are dismantled, the nuclear section is properly sealed and the backfilled areas are landscaped, the entire site will have reached its Static State.
C N S 9th A N N U A L C O N F E R E N C E , 1988 201
DECOMMISSIONING COST ESTIMATES A RATIONAL APPROACH, EMPLOYING AVALIDATED COMPUTER CODE
JOEL LIEDERMAN
Atomic E n e r g y of Canada L i m i t e d M o n t r e a l , Quebec, Canada
ABSTRACT
Decommissioning cost estimating is receiving an increasing amount of attention. Estimates prepared to date, as drawn from the published l i t e r a t u r e , have di splayed a rather wide vari abil i t y , due, in great part, to the absence of a logical rational approach to cost estimation. This paper puts forward a suggested approach based on the Unit Cost Factor (UCF) method. It further suggests that computerization of the estimation process provides significant benefits and flexibi11ty. The AECL-DECOM computer code for decommissioning cost estimation is described in the context of the recommended approach.Two case studies, the Gentilly-1 and the Nuclear Power Demonstrator Reactor decommission!ng projects, are described, as are the efforts made to aate to validate the code on these projects. A reasonably good correlation has been achieved and improvements have been made, based on the experience gained. It is concluded that a rational» logical, validated code, such as AECL-DECOM, is a valuable tool for the preparation of reliable, defensible decommissioning cost estimates. INTRODUCTION In t h e l a s t s e v e r a l years t h e r e has been an i n c r e a s i n g l e v e l of i n t e r e s t displayed worldwide on the s u b j e c t of n u c l e a r f a c i l i t y decommissioning. A r e c e n t p o p u l a r i z a t i o n of the subject has resulted from at least three distinct factors. Firstly, several prototype nuclear facilities such as JPDR in Japan, Shippingport in the USA, and Gentilly-1 in Canada, among others, have been (or are in the process of being) decommissioned in the last few years. Secondly, many nuclear f a c i l i t i e s around the world arc approaching the end of their planned commercial li ves and, notwithstanding the possibility of life extension efforts, are candidates for decommissioning. Fi nally, as a consequence of increased Regulatory interest in the subject, the owners/operators of nuclear f a c i l i t i e s are being required to put in place decommissioni ng plans ( and collect funds for their ultimate decommissioning. When planning for decommissioning in the short or long terra, one of the most important tools needed to assist in the selection from among technically, politically, and socially acceptable alternatives is a cost estimate. This estimate must be as accurate and reli able as possible» since major decisions will be made using this data. Moreover, and especially when these e s t i mates are part of Regulatory submissions subject
202 CNS 9th ANNUAL CONFERENCfc, 1988
to intense scrutiny, these estimates must be credible and defensible. PRESENT SITUATION VIS A VIS PUBLISHED ESTIMATES Published estimates reported vzry si gni fi Cri.it ly one from the o t h e r . This v a r i a t i o n i s perplexi ng to those i nvolved in the technology field, d i s t u r b i n g to those responsible for f i n a n c i a l planning, annoying to Regulatory A u t h o r i t i e s , and confusing to the general p u b l i c . I t i s not the purpose of t h i s paper to attempt to explain the referenced d i f f e r e n c e s , although i t i s relevant to comment on the factors that are g e n e r a l l y agreed to be responsible for them. Scope of work for a gi ven de comrai ssioni ng plan represents the s i n g l e most important factor explaining the d i s p a r i t y . Even for a s i m i l a r choice of stage (per Reference 1) for a gi ven f a c i l i t y type (eg. PWR, PHWR, BWR of a s p e c i f i c power r a t i n g ) r a t h e r wide d i s p a r i t i e s have been noted. A survey of nuclear f a c i l i ty decommissioning c o s t s was conducted for the American Gas Association Depreciation Committee and presented a t i t s 1983 October meeting. This survey indicated t h a t of the 32 PWRs and 20 BWRs p o l l e d , the projected costs for a Stage 3 type decommissioning varied between US $55 t o $232 (1987 d o l l a r s ) per k i l o watt e l e c t r i c . The US u t i l i t y c o m m i s s i o n / u t i l i t y i n t e r f a c e i s q u i t e complex and confrontational and so some of these estimates may have r e p r e sented the "best that u t i l i t i e s could do under the circumstances". Examining only those f a c i l i t i e s in the 500 to 1150 MW(e) s i z e range, r e s u l t s in a s l i g h t l y l e s s di vergent band of e s t i m a t e s , being in the range of US $110 to $150 per k i l o w a t t e l e c t r i c . I t J s very relevant t o note an apparent c o r r e l a t i o n between the higher end of the s c a l e and the more recently prepared e s t i mates. Generi c co« t es t i ma t e s , such as the Elect r i c Power Research Institute (EPRI) sponsored Bathelle Labs study (Reference 2) appear to be f a l l i n g from favour, being replaced, to a great extent, by s i t e specific studies. In a recent study conducted by the Nuclear Energy Agency of the Organization for Economic Cooperation and Development (NEA/OECD), decommissioning costs reported by Canada, Federal Republi c of Ge rraany, Fi nland, Swede n, and the Uni ted States were compared. To f a c i l i t a t e comparison among widely varying reactor types and sizes the reported estimates were scaled to a common (1300 MW(e)) reactor size basis. This study reported costs between US $119 to $167 (1987 dol-
lars) million for an framedi ate stage 3 decommJ ssioning. For an immediate Stage 1 followed by a dormancy peri od of 30 years of storage wi th survei1 lance, and then a final Stage 3 (complete di smantli ng) the costs ranged from $120 to $134 mi H i on. The second raost i raportanc factor contri but ing to the di vers i ty of the estimaf amounts, i^ the absence of an accepted consi stent methodology for the preparation of these estimates. This paper will exami ne this subject more closely and suggest a possible solution to this problem. However, before moving on to do t h i s , i t is worthwhi le for the sake of completeness to very brief ly highlight other factors whi ch can collectively contribute to the diversity mentioned above. Since the various studi es have been done at different ti raes and in different countries, it i s usually necessary to bring the figures to a common base for comparison purposes, usually $US in a given year. Thus differences or variations in currency exchange and assumed inflation rates can have a major impact • It is necessary to dispose of or store the contami nated and a c t i vated components that composed the facili ty. In many countries the issue of radwaste disposal (especially for higher level wastes) has not been fully resolved, to the extent that costs of d i s posal can be compared one co another. Volume reduction and limits of releaseabili ty of very low level waste differ ftoin place to place (thus influencing total waste volume, and hence d i s posal c o s t s ) . Also, in the USA for example, changes in the national regulations (10CFR61 for example) and the formation of regional LLW compacts have had, and will continue to have an impact (escalation) on the disposal costs for decommissioning wastes. There are wide differences in labour practices (unions, e t c . ) and productivity from one part of the world to another i mpacting on the labour costs• Finally, there is far from universal agreement on the application of contingencies to various aspects of costs. I t would seem then that i t is important for those charged wi th the responsibi lj ty tot planning and/or executing decommissioning a c t i v i t i e s to have available to them the necessary tools to facilitate the preparation of estimates in a reliable and disciplined fashion. Given the typical engineering approach of considering mult i p l e options and given the inevitable tendency to "what if" when evaluating an estimate, i t is also sensible to automate the process of estimate preparation. A RATIONAL APPROACH TO COST ESTIMATING When engi neers and c o n t r a c t o r s cost out conv e n t i o n a l c o n s t r u e t i on j o b s , the common p r a c t i c e i s to depend s i g n i f i c a n t l y on databases of i n f o r mation b u i l t up over the y e a r s . These databases (see Reference 3 for example) a r e based on m u l t i plying m a t e r i a l takeoffs by t r i e d and proven u n i t c o s t s for a c q u i s i t i o n , l a b o u r , e t c . Drawing from t h i s p r a c t i c e , e f f o r t s a r e p r e s e n t l y underway in Canada and the USA to s t a n d a r d i z e the methodology for e s t i m a t i n g decommissioning c o s t s .
The G e n t i l l y - 1 Decommissioning
Project
G e n t i l l y - 1 , a 250 MWe, 833 MWth, CANDU (CANada J)euteriuBi JJraniura), type r e a c t o r i s located in Quebec, Canada. Thi s plant compri ses a n a t u r a l uranium f u e l l e d , heavy water moderated, b o i l i ng l i ght water cooled r e a c t o r . Gent i l l y - 1 was put in s e r v i c e in 1971 and operated I ntermi t t e n t l y u n t i l 1978. In 1982 a j o i n t committee of HydroQuebec and Atomic Energy of Canada Limited recommended a g a i n s t the rehabi l i t a t ion of the plant on economic grounds. P a r t s of 1982 and 1983 were devoted to the engi neering and econorai c studi es involving decommissi oni ng of G e n t i l l y - 1 . A wide range of decommissioni ng s c e n a r i o s ( s t a g e s ) were c o n s i d e r e d . These varied from safe s t o r a g e of a l l radioing* ---il hazards i n s i de the p l a n t wi th s u r v e i l l a n c e (Stage 1 ) , to prompt di s raantling of the t o t a l plant and d i s p o s a l of a l l equipment and s t r u c t u r e s for u l t i m a t e r e l e a s e of the s i t e for u n r e s t r i c t e d use ( S t a g e 3 ) . After caref u1 cons ide r a t i on, a de ci s ion was made to decommission the plant to a " s t a t i c s t a t e " which i s a v a r i a n t of IAEA's Stage 1 (Ref. 1 ) . Between April 1984 and April 1986, a c t u a l decommissioning of Gentilly-1 was completed wi t h i n schedule and budget. The scope of t h i s project i ncluded c u t t i ng and capping of a l l p i p i n g , cabling and p e n e t r a t i o n s to the r e a c t o r b u i l d i n g , d r a i n i n g and drying a l l systems i n s i d e the r e a c t o r b u i l d i n g . All equipment and s t r u c t u r e s from the s e r v i c e building except the e x t e r i o r w a l l s and main structural slabs were dismantled, decontaminated and disposed of. It became apparent that the methodology used to that time had rather serious shortcomings and that a computerized cost model was an essential tool to permit the analysis of numerous decommissioning scenarios, and for decision optimization purposes. The methodology and data presented in this paper are based on work done as part of the Gentilly-1 decommissioning studies as well as the actual results obtained during 1984 and 1986 field work. More -ecently, work has been done on the Nuclear Power Demonstrator Project (a 25MW(e) PHWR prototype reactor f i r s t placed into service in 1962) to f i r s t estimate the cost of decommissioning and then to collect actual data so as to validate these predictions. ASCL's Interest in Decommissioning Cost Estimating During the planning stages of the Gentilly-1 decommissioning project, AECL developed a computer code called PECOM to f a c i l i t a t e the process of preparing decommissioning cost estimates. A rigorous and highly disciplined approach to decommissioning cost estimating was followed. AECL was a pioneer in the approach tc standardizing decommissioning cost estimating methodology. For example, a significant amount of work has been done to date under the Coordinated Research Program (CRP) Agreement No. 3960/CF, in Decommissi oning and Decontamination, organized by the International Atomic Energy Agency (IAEA), Vienna. This approach has now been legitimized
CNS 9th ANNUAL CONFERENCE, 1988 203
by the i ssue of several i rapertant (North Ameri — can) standards on this subject.
period could vary (dependi ng on techni c a l , poli ti cal and other factors f rora 30 years to as much as 100 years or more.
Cost Est i mat i on Methodology Preparation stra i ght. forward known and the def i ned.
of cost esti mates can be a task if a l l cost components are esti mating methodology is well
The components that make up the total cost for decommi ssi oni ng a nuclear plant can be grouped i nto four categori es each of which needs to be hand Led in a slightly different fashi on. These are: a) b) c) d)
activity dependent costs period dependent costs speci al act i vi ty costs. dormancy period costs (if t l i ng is envi saged)
delayed di sraan-
Acti vj ty dependent costs are those associated with tasks that are d i s c r e t e , measurable and of a repet \ ?.i ve nature and can thus be analyzed by developing typical Unit Cost Factors (UCF) which can be applied to the category of equipment that they represent (e.g. cutting pi pe, removing pumps, dismantling structural s t e e l ) . Uni t Cost Factors are models whi ch take into consideration a l l the typical a c t i v i t i e s associated with, for example, the task of di sraantli ng piping. They consi der manpower requirements, duration of various tasks and special equipment involved. By factoring in labour costs, equipment rental or maintenance costs and so forth, the model then converts the whole range of a c t i v i t i e s into a cost, expressed in dollars per unit ($/m run i n the case of piping). The cost factor i s thus def i ned as the estimated amount of money requi red to remove one uni t of a component. Uni t Cost Factors have been developed for dismantling, packaging, transportati on, and disposal for a variety of categories of equi pment. Period dependent costs are those associated wi th the durati ons of different phases of the project such as engi nee ring, project and construction management, li censi ng, quality assurance and security. The p n j e c t schedule is then used as a base to deter mi ne the period-dependent cost. Special activity costs are spli t into two categories: special items that are non-repecitive such as the calandria or reactor vessel removal and miscellaneous items such as the operation and raai ntenance cost during the decommission! ng a c t i v i t i e s , purchase of heavy or speciali zed equipment, si te preparation and mobili zati on costs, and the cost of energy, for example. When a consideration of delayed dismantling i s '^eing made, i t is also essentiai to take into account the various operations, maintenance, security, and survei1lance costs duri ng the dormancy perlod between the terrai nation of the a c t i v i t i e s associated with putting a facility into a Stage 1 Cor " s t a t i c s t a t e " in the Canadian context) condi t i o n , and the commencement o£ the final dismantling activities. The dormancy
204
CNS 9th ANNUAL CONFERENCE, 1988
Once al1 these i ndi vidual components are deterwi ned, the next logi cal step is to combine a l l the casts together to develop the total cost of the decommi ssi oni ng program. Wi th sui table integrat ion wi th project schedules and manpower loadi ng, i t is possi ble to ascertain cash flow requi rements. The importance of planni ng for the £inanci ng of these costs requi res that consideration be given to the addi tional dimension of account)ng for possi ble vari at ions over time of factors such as:
.
.
inflation rate discount rate labour costs (includi ng productivity mates) li mi tations on cash flow i n any periods waste disposal costs
It is evident that computerization process is not only advantageous but recommended.
estigi ven
of the highly
Concept of Unit Cost Factors Uni t Cost Factors (UCFs) were mentioned earl i e r in this paper in the context of Acti vity Dependent Cos t s. The mos t ef fect i ve way to esti mate the t o t a l cost associated with these r e p e t i t i ve tasks is to u t i l i z e unit costs for disraantling and to multiply these unit costs by the quanti t i e s involved. The total number of possi ble unit costs can become exceedi ngly large and if taken to an extreme can make the estimation task unnecessarily complex. As a consequence i t is prudent and reasonable to group si mi lar tasks together so that in total no more than about 150 unit costs or UCFs are involved. I t must also be recognized that certain environmental influences can have a negati ve impact on the productivi ty, and hence cost, associ ated with certain disraantli ng related tasks. For example the presence of radi ation fields in the vicinity of a component to be dismantled adds to the cos t of the task be cause of the addi t i onal precautions, e t c . required. Similarly, if the component to be cut is located in a di fficult to reach location, scaffolding wi 11 be required, also adding to the cost. Of course scaffolding in a radioactive environraent is a possible combination. A simple way to handle this is to have variants of each UCF accounting for possible environmental compli cations. These UCFs must be validated based on actual field work and kept up-to-date based on advances in dismantling technology, radiation prelection practices, and labour productivity. AECL-DECOM COMPUTER PROGRAM DEVELOPMENT As pointed out e a r l i e r , the recommended approach to decommissioning cost estimating lends itself very well to the application of a compu-
ter. Not only does a computer faci li tate the process of keeping t rack of a l l the coraputati ons to be done, but it makes the task of "what if" evaluation extremely easy. Once the data is assembled and entered, it is remarkably easy to perform s e n s i t i v i t y analyses of say. .
#
vari at i ons in disposal cost changes i n labour rates alternative dismantling scenarios sequences the effect on cash flow and /or t o t a l result ing from schedule extensi on or p ress i on, or f rom an extensi on of the mancy peri od for a delayed dj smantling nari o.
or cost comdor.sce-
To facilitate the engineer!ng and project planning for the Gentilly~l Project, AECL developed a computerized decommissioning program called AECL-DECOM. The AECL-DECOH program makes use of an on line, i nteracti ve data base and was developed for estimating decommissioning activity dependent costs, waste volumes and worker radiation exposure. It is based on the Unit Cost Factor approach. The user keys in the plant inventory wi th the proper equipment code, site location, quanti ty, radloacti vity level and schedule acti vity. The computer program calculates the costs and the radiation exposure for the various operations based on the inventory and the builtin (user defined) Unit Cost Factors. The result of calculations can be sorted in a multitude of ways and at many levels, such as by bui ldings, rooms, schedule activities, radioactivity levels and equipment categories. The program is very flexible and allows for easy addition, deletion and rearrangements data. Furthermore, the data base spread sheet can accept up to hundreds of fields and as many inventory items as the user may des i re. This flexibility represents one of the many attracti ve features of the program, si nee it permi ts a user to carry out optiini zation studies for a large variety of decommassioni ng scenarios. The data base is divided into several subfiles (i.e. techni cal data of an equipment is in one subfile, cost data in another subfile, etc.). The mai n body of information on: 1. 2. 3.
the
Data
se
contai ns
Unit Cost Factors Physi cal inventory of the plant Radiological inventory
The Data Base maintai ns a l l the necessary Unit Cost Factors in a special f i l e and uses I t to calculate cost. The Unit Cost Factors can be updated as required to reflect changes in technology, actual hands-on experience and so on. They can also be adapted to local condi tions by factoring in local labour rates and productivity. Furthermore, new Uni t Cost Factors can be developed as required for di fferent types of nuclear plants. This is a powerful feature which allows the program to be kept up to date with a minimum amount of ef f o r t . The AECL-DECOM program requires the input of information on the total physical inventory
(equipment, materials, structures) of the plant. The physi cal properti es (si ze, volume, wei ght, type of materi a l , e t c . ) of each pi ece of plant component are entered i nto the Data Base uti li — 21ng specially designed input sheets. The information on component properti es can ei ther be obtained from the physical inventory records of each piece of equi pment or by extract! ng this from relevant system drawi ngs. It is preferable t if at a l l possible, to obtain this information from the physical inventory records. All plant components are classified by major types of equipment such as pumps, tanks, heat exchangers, valves and pipi ng. These are further subdivi ded, (for example, pumps and purapmotors ere classi f ied i nto six subgroups, to allow for the development of more representative Unit Cost Factors. Once the basic physical character!sti cs of a l l the major plant components have been entered into the Data Base, the program then calculates the t o t a l quant i ties for each type of coDponent and retains this information in a separate subfile for l a t e r use. Radiologi cal i nformation is required by the program to calculate predi cted radi ation exposure to workers, and to determi ne the quantities of radioactive wastes by category. AECL has developed procedures for generating radi ologi cal inventories by field surveys of a l l plant components and s t r u c t u r e s . The radiological data is entered into the Data Base, preferably at the same t i me as the physi cal i nventory data i s entered. Radiation fields (milli-rem per hour) at cont a c t , one meter from the component and the background radi ation level in the room are measured. The program can then calculate the t o t a l predicted radiation exposure to personnel in each room or area, and provides the cumulati ve value for the e n t i r e plant ( p r o j e c t ) . The AECL-DECOM program runs on any IBM-PC/XT (or compatible) microcomputer is fully menu d r i ven and user friendly, and can be used to prepare a range of cos*: estimates for any decommissioning project * Each of the components of costs can now be estimated by i nputti ng the relevant data into the AECL-DECOM program through submenus. Operation has been simplified even further by the p r e paration of a manual that walks the user through the program step by s t e p . A simplified Flow Diagram of AECL-DECOM is shown in figure 1. APPLICATION OF THE COMPUTER PROGRAM Gentllly-1 Plant During the studies associated with the G-l plant, decommissioning cost estimates for stages 1, 2 and 3 were prepared, using the AECL-DECOM computer program. The followi ng step by step approach was pursued. This approach can be applied to any nuclear power plant. Major steps were as follows:
CNS 9th ANNUAL CONFERENCE, 1988 205
ESTIKATE *KNH0tRS S COSTS KROJ.MGT
—
ENCG COKST. M3T TOOL DESIGN HEALTH PHY 5 . LICENSING O.A.
PERIoo OEPQtOENT COST
APPLY ESCAL. RATES
.OTAL COST .OfSMWTLIW .PACKAGIM3 .TRANSPORT .DISPOSAL (LLW) .STORAGE (U.W)
CASH FLOW I N NOMINAL $ CONSTANT *
IACTIVITY CQNTItt-
INTEGRATE
i
ADO
-CONTIN-j GENCY
I
-'DISCOUNT RATES
SPECIAL ITS* , COSTS ' ABBREVIATIONS! W.P>
ESTIMATE OPERATION t MAINTENANCE COSTS FOR DORMANCY PERI001
COSTS FOR SECURITY RADIATION SURVEYS, ENERGY, ETC.
.CONSTANT * .DISCOUNTS
'.tfPENO&fT COST
CENCY I
SCHEDULE
OTHER COSTS .INSURANCE .0 1 M .TRAINING -ENERGY.ETC.
CASH FLOW -
AGO -CONTINGENCY
* WOW PACKAGE
W3S
- WORK BREAKDOWN STRUCTURE
0*1
- OPERATION • WINTENMCC
DOftWNCY
UW
- LOW LEVEL WASTE
PERIOD
HtW
- HIGH L E V a WASTE
COSTS
FIGURE 1 - COST ESTIMATE FLOW DIAGRAM
. .
• •
Survey of equipment inventory Application of a computer code Survey of radioactive inventory Radi ologi cal exposure to workers (man-rera) Development of unit cost factors PERT/CPM network to determine c r i t i c a l path Manpower requi rements Integration of cost and schedule
. .
Suraraary of cos t s Financial analysis and cash flow
.
The physical inventory of a l l the plant components (equipment, s t r u c t u r e s , e t c . ) was obtai ned from a room by room survey. All component i terns were then entered into the database of the AECLDECOM computer program. Figure 2 is a tabulation of dose estimates (in manrem) for several selected rooms (T53L, T601), as gene rated by the AECL-DECOM code. Such outputs are very valuable in planning radiation protection program. Validation of the Code with Site Data In
206
order
to
test
the
validity
CNS 9th ANNUAL CONFERENCE, 1988
of
the
cost
estimates prepared using DECOM, a sample of actual cost and raanhour data from the Gentilly-1 Decommi ssi oning operation were processed through the DECOM code and i t was observed that the t o t a l cost figures were accurate within a 20% range, though costs for individual a c t i v i t i e s in some instances differed significantly due to changes in the project technical concepts, reduction in the production work day because of unantici pated clothing changes, showers or breaks, increased radi ation protection coverage provided, and special features such as asbestos removal. However, suffici ent conf idence has been developed in the c a p a b i l i t i e s of the DECOM Code through this experience. UCFs built into the code have been updated based on this validation exercise. The accuracy of the estimates will be f ur t he r imp roved by constantly reviewing the UOF's in the AECL-DECOM code and adjusting them as required. After 25 years of service, a decision was made in 1987 to shut down and decommission the 25 MW(e) Nuclear Power Demonstrator (NPD) Reactor at Rolphton, Ontario. This reactor was the demonstration project for CANDU PHWR pressure tube technology. (It was shut down primari ly
G-1
B
Rfi
CST COO
M5CTIPT
DPC0MMI5SIQHING
CONTACT RAO flELO 1MAWAT
T 531 363 HISC 5-IOK TOTAl 581 •
012
.014
T SOI 303 MI5C 5-1 OK
,030
.100
180000
20 000
T 601 322 TAHCI-5K
HANRffi
noon BKDND
.0(2 C?D .070
01*1 LAB
C.At
CULATtONS
PACK LAB
45
9
45
9
21
23
PACK
nm
ran
TOTAL
DOH
DOSE
MX
.001 .004 .336
run
.001 .005
.009
.092
.420
T 601 324 TANK 10-50K
.010
.070
.070
55
144
004
.002
.006
T 601 324 TANK I0-30K
.200
1.300
.070
55
220
.067
.MB
.935
406 13763
1.211 2.046
.167 .369
30833
210.197
TOTAL tOl' TOTAL " •
176 31000
GIUF" TOTAL
57946
J5.W2
I.37B 2.415
246 l «
K - 1000 LBS - 434 KG
because the costs of maintenance and operation of a unit of this low power rating were such as to make continued operation uneconorai c a l . ) As part of the engineer!ng studies with the decommissioning planning, the code was u s e d t o exami ne t h e c o s t s many alternative decommissioning Ultimately, the estimate associated selected scenario was further detailed the project estimate.
associated AECL-DECOM of a g r e a t scenarios. with the and became
This project is underway at the time of the writing of this paper. It is intended to collect actual costs of the various operations and to compare them with the AECL-DECOM predictions. This exercise will provide further code validation and the opportunity to further fine tune the UCFs as required. The code will also be useful as a project management tool during the implementation phase. Figures 3 and 4 are copies of printouts showing, for the NPD project, a summary report for the " s t a t i c state" phase (stage 1) of the project of the cash flow requirements (note that a l l the work, is accomplished in 2 years), and a typical page of the es tj ma te report f or the act 1vity dependent costs sorted by room. Figure 5 is a printout of a typical project (neither NPD nor Gentilly-1) of total decommissioning costs for a facility f i r s t placed into stage 1, then kept in this condition for 50 years (dormancy), and finally dismantled to a Stage 3 condltion.
CONCLUSION I t has been pointed out t h a t t h e r e has been great diversity in the cost estimates generated to date for decommissioning nuclear f a c i l i t i e s . Lack of a coherent and consistent approach is one of the key factor for t h i s . As the interest in decommissioning increases, the amount of scrutiny being given to decommi ssioning estimates will intensify, and the requirement to be able to produce a credible, r e l i a b l e , and defensible e s t i mate will be exceptionally important. The rigorous UCF based methodology in this paper is one way of achieving stated objective.
suggested the above
The large amounts of data to be considered, the need to estimate waste volumes generated and the large number of decommissioning scenarios that are to be analyzed make computerization using a code, like AECL-DECOM, represents a feasible and attractive alternative to preparing such cost estimates manually. The AECL-DECOM computer program, based on the widely accepted Unit Cost Factor approach is a versatile tool for applications in decommissioning studies. It can manipulate large amounts of data and make use of a large number of models that have been developed expressly for this purpose. Results can be generated for vari ous decommissioning scenarios and these can be reported in numerous ways which suit the specific requirements of the user.
CNS 9th ANNUAL CONFERENCE, 1988 207
...- ...
7--... msr j->
»
• •-••• / 3 V - BIBC iJOOF . .7 ••->(, JtBM PLUP <*iltir
i
--V.
HIM
M ?•
"5.
j
S rii
? *
^ t-L"
i*
i'
^
>«
3f
;: ;.'..'.;; S" ~ S S"-
?!
;::s: s xs ::t.-
5:
^ 'sr s|
WniHli [(MStMIII
<03ii
?MM
un
'JS71 BCO!
is:
•
:
;
°
;
P
<*M
7»»
?*O
0
•?"
.•
i"'o
re
t,
n
m
I CM51M1I KM I ML I CMIMTI W I M 1 4 CMtMIII HMIWI CWSIMTt MMIMLt COKIMTI MIMLf COWTMII KXIIHll CH1TUK H71*Mt CMSTMTI MHIMLI CtWSl
Z
J«5J
nmi
rtm
rwi
rc«M
I«IJ!
UU 1)71)1 5H(2
ENIiEitilN
tmmmm toioo Mill I I COST l i m STECITEnS
mi* dtlEI
tl 5«ltottl II I I I iDttl I I I
L Most important of all Js to ensure that whatever approach is used, the data base for the cost estimates is based on reliable field proven Information, and that the estimating methodology is validated by benchmarking it on actual projects.
208
«
CNS 9th ANNUAL CONFERENCE. 1988
iw«
nm
nm
nm
nm
n*i
nm? nm
Km
nm
nm
nmt
«
lOlrtl
»• C O S T
COST «
1DTAL COST TOST 1 YT E" wr M I I rINr INIK NI:V «
TYPE: ACTJVITV
TRANSrORIlir IMN DISF-OQftL
466Z llCriVITy ZZ4(,B9 ACTIVITV 077^8 flCrlVITv
pftcy ING
• • COST TVPE: PERlnD Eur.INEKRING DftfiFT[NG
*«*
Total
71517 FERIOD 6i.j.,n.i FERIOD
***
REFERENCES 1.
IAEA Technical 1983.
Report,
Series
No.
230,
2.
Updated Costs for Decommissioning Nuclear Power Facilities; Electric Power Research Institute, Palo Alto, California, Hay 1985; (EPRI NP-4012).
3.
Building Construction Cost Data, 1983, Robert Snow Means Company Inc., Kingston, Massachusetts, U.S.A.
4.
Decommissioning of Nuclear Facilities Feasibility, Needs and Costs; 1986, OECD Nuclear Energy Agency, Paris.
CNS 9th ANNUAL CONFERENCE, 1988
209
MINIMIZING OPERATING ERROR DURING A MAJOR UPGRADING PROGRAM AT A MULTI-UNIT NUCLEAR PLANT G,A. Fowles Ontario Hydro, Pickering Nuclear Generating Station Pickering, Ontario
ABSTRACT
TABLE 1
The extended forced outages for reactor retubing of Pickering Units 1 and 2 presented a unique opportunity to implement a major upgrading program. However, the program could not be quickly implemented on Units 3 and 4 and steps had to be taken to minimize the operational problems that could develop during the period of change. Operating staff who had trained on identical units would be faced with significant differences between original and upgraded system configurations. Preliminary results indicate that a successful program was put in place to manage the change and no significant increase J.I; error rates has occurred.
Plclt-UP unifications with tUior iwact on Unit operatic.
t 1
IMPACT ON OPEMTIOM
HODIFICATIOH 1(1)
|
coolant Injection Scheme
Kill (lv) lit
(11)
1.0
INTRODUCTION
Pickering Nuclear Generating Station A consists of four 540 MW(e) CANDU units, which were almost identical when they went into service between 1971 and 1973. The performance of these units was excellent with an overall capacity factor exceeding 80% in the ten years up to August 1983. However, these ten years saw significant advances in the fields of safety analysis and probablistic risk assessment, and life extension and equipment obsolescence wei'e identified as major issues in the continued viability of nuclear power. Also, 40 reactor-years of direct operating experience had revealed the need for plant modifications that would reduce radiation dose, improve maintainability and result in further increases in capacity factor. When the decision to replace the pressure tubes in Units 1 and 2 was finalized in March 1984, the resulting four year forced outages provided a unique opportunity to upgrade the units. The Pickering Upgrading (Pick-Up) program that was formulated was pared down from an original wish list of over 600 items. The 86- item Pick-Up program consisted of projects that ranged in size from the repositioning of control valves to ease maintenance to the installation of a $70 M high head injection scheme for emergency coolant injection (HPBCZ). Table 1 contains a list of those changes which had the greatest impact on unit operation, The Pick-Up program proceeded in parallel with the retubing, to improve operab i1i ty and maintainability, while increasing safety margins and extending the useful life of the units. However, it soon became obvious that managing the transition from original to upgraded units would be complicated by the extended period over which the changes would be implemented.
|3.
Upgraded annunciation System
(4| I
Conversion of ten control (adjuster) rods Into shutoff rods
5.
Deletion of Moderator Dump following d LOCK
6.
2.0
for LOCHs. CTianges to s t a r t - u p / s h u t d o w n procedures for heat t r a n s p o r t system. Hew crash cooldown p r o c e d u r e s .
Less restrictive pressure/ temperature envelope for neat transport operation. NO requirement for spec IdJ procedures to Inhibit delayed hydride cracking.
I , 1 |
Operator Interface with the | unit radically altered. | Enhanced monitoring capability 1 for equipment parameters. |
Reduced Xenon override capability. coarser control using remaining adjuster rods because of Increased reactivity worcft per rod. d i n New shutoff rod testing facilities.
1 | j | | 1 | |
in
Mew operating philosophy required mooerator to be fully pumped up for five minutes following a LOCK.
1 | | |
Reduced operating margins. New calibration and test procedures.
| 1 |
(i)
;
,
Mew Protective s y s t n Parameters | ( i ) and Trip Setpolnt changes / ni)
ASSESSING THE IMPACT ON OPERATING STAFF
Unit 1 was to return to service in 1987; Unit 2 in 1968 and Units 4 and 3 were to be upgraded in 1988 and 1989 respectively. As a resuIt, the previously identical units would be significantly different for a period of three years; requiring different operating and maintenance procedures but utilizing a common work force. If operating errors were to be minimized, considerable effort had to be directed at preparing the shift crews for this period of major change. The approach that was adopted included the following elements; (a) minimize differences between units by completing advance work on Units 3 and 4 wherever possible (b) major upgrade of operating and maintenance procedures Cc)
210 CNS 9th ANNUAL CONFERENCE, 1968
(1)
an
e x t e n s i v e on-pcwer t e s t i n g of new components.
institute a comprehensive training program
(d) full scope recommicsloni ng COL- restart of upgraded units (e) dedicated First Operators and Supervisors fur the restart of Units 1 and 2 (C) Control.led muvement of First between units after Unit L restart.
3.0
Shift
Operators
PICK-UP TRAINING
Before the restart of Unit 1, a complete set of upgraded operatLng and ma Lntenance procedures were developed. These procedures incorporated format, changes and operat*ional experience data in addition to the changes required as a result of new hardware. This was a major task requiring in excess of ten man-years of technical manpower. Many of the procedures remained common to aLI units but 17 systems were sufficiently different that separate operating procedures had tu be issued for the upgraded units. To prevent operating procedures specific to the upgrade units from being used on non-upgraded utii ta, n. colour cod ing scheme was used with the new procedures printed on green paper. Once the procedures were written, the training program could proceed. The training focused on the 40 licensed First Operators and Shift Supervisors but maintenance staff and f ield operators were also included. The licensed staff received four weeks of classroom and sinm ta t or tra in i.ng comb ined wi t.h a comprehens i ve field checkout program that included in-situ testing and operation of the newly installed equipment. Simulator training presented a difficult problem as the single Pickering NGS-A simulator could not accurately reproduce the response of all four units due to their system differences at any po i nt in time. A hybrid simulator was developed that could utilize either the old low head EC I system or the HHEC1 system and the two redundant annunciat ion channels were split between original and upgraded versions. A written examination was used to verify the adequacy of the training program, which was also extensively audited by the Atomic Energy Control Board.
4.0
significantly different design. This was desirable as an extensive program of commissioning tests was being carried out to optimize procedures that differed from those used on Units 3 and 4 in various degrees. A common on-shift maintenance work force was used at all times during the upgrade, as it was felt that well trained people using pre-prepared maintenance procedures could quickly switch from one reactor unit to another without problem. To date, this has proved to be a correct assumption, as no significant maintenance errors have resulted from the difference between units.
5.0
UNIT 1 IH-SERVICE
When Unit 1 had completed its commissioning tests and was formally returned to service in October 198 7, the shift crew structure reverted back to a single shift supervisor. By this time all operating procedures had been finalized and all shift supervisors had exposure to the new systems. The first operators remained dedicated to a single unit and only moved between units after a co-piloting period allowed them to become familiar with the details of a particular unit. This controlled movement reflects the significantly different actions that must be taken by the first operator in the first few minutes of some abnormal incidents; actions which must be taken before extensive procedure review can take place. Table 2 contains some examples of the different short term actions required on Units 1 and 4. TABLE 2
1
As Un it, t erit icality approached, the first operators were u Lvided into two groups. One group worked exclusive ly on Unit 1 and the other group supported the continued operation of Units 3 and 4. (Unit 2 wus being retubed at this time.) A separate Shift Supervisor was also placed in charge of Unit I restart. This allowed the unit to start up wi th dedicated 1icensed people, who dId not have to divide their attention between units of
UNIT * Che
1 I 1
I L«rge Loss o f Coolant | Accident
J
1
1 Ho manual i n t e r 1 vent Ion required.
1 Manual cont r o l of 1 recovery va lves may I be required wUhln I f i v e minute
1
! Spurious Heactoi
Trip
I b o i l e r s by manually 1 opening the steam 1 reject valv es.
b o i l e r s using HPECI logic.
J
UNIT 1 RESTART
Early in the retubing outages, the decision was made to return the units to service using a comprehensive commissioning program, very similar to the one used for new units. This ensured that al 1 systems were returned to ful 1 service in a controlled manner; identifying at an early stage any problems that had developed during the long outages and increasing the probability of reliable, trouble free operation after the restart. Full commissioning also allowed the operations and maintenance staff an opportunity to become familiar with the newly iustalled systems, which were also put through extensive test programs.
1 I
UNIT 1
ABNORMAL IMC1DEKT
1 Correct operation of 1 hydrogen i g n i t i o n 1 system to be 1 verified. 1 1 1 1 I Moderator pump up I required to clear 1 adjuster rod 1 Interlocks. 1 1 Trip logic to be 1 reset wltnln seven 1 •inutea to *«old I Xenon poison outage. 1
,
1 Ho hydrogen ignition j | t system i n s t a l l e d . t t 1 1
I I
1 HO noderato pump up | I required. I 1 I 1 T l i p logic t o be | 1 reset within twelve | I •Inutes to a void | 1 Xenon poison outage. | 1
To the maximum extent possible, common abnormal incident procedures are used on all units, utilizing logic branches to cover off the system differences between units. Only in one case, where the branching significantly complicated the procedure, were separate abnormal incident procedures developed for different units. Having common procedures allows training to focus on the overal1 objectives of each procedure and then demonstrate how the objectives are achieved in different ways on different units. Simulator results i ndicate that this method has been
CNS 9th ANNUAL CONFERENCE. 1988 211
till ec t i vc i n f between u n i t s .
6.0
ng
the
operators
fur
rotat ion
PHKLTHIWARY RESULTS
Picker.i ng has .i we I i deve loped report ing scheme f of opei at ing and ma int.nuance trrurs. S igti i f leant events report, c a if wri t ten and ana]y^ed fur human error bused un criteria that, rel ate back to the S tat ion 's key rt-sul t di'eas (safety, rel Lability , c it. Lzenslii p and cost) . Figure 1 shows the trend in annual error rates fur the tive years preceding the utart-up of Unit 1 . The data shows c 1 early how an upward trend had been reversed in 1985 by the application of other error teduc t Loti programs that were completely separate from the Pick Up program. FTGUPE 1
Human Error Rates at PNGS-A
OPERATOR
S3
84
85
86
I
ilL.I ill B4
85
6«
had
not
been
(c) The start-up procedure for the main heat transport system was modified and simplified when new pressure tubes were installed on Unit 1. However, several draft procedures were produced before the E inaI version was issued. As a result, i)i¥t main heat transport pump was started at the wrong point in the sequence, and a reactor trip occurred. The maintainer error rate did increase, and appears to have peaked in the period between Unit ] critieality and its return to full service. This was a period of intensive maintenance activity when the final phase of the Pick-Up modifications were implemented and comprehensive commissioning testing was completed. Five of the incidents reported in the third quarter of 198 7 occurred on Unit 1 and two more incidents involved Pick-Up changes that were being incorporated on Unit 3. The majority of these errors involved control maintainers; on two occasions relays were incorrectly installed and there were three incidents that involved inadvertent activation of adjacent equipment.
/.0
MAINTENANCE I
83
operators
CONCLUSIONS
87
MAINTENANCE
82
the
The consequences of all the above incidents were confined to lost production but they emphasize why this type of major upgrade program must be carefully managed to reduce operating error. Monitoring of the error rate trends will continue throughout the program to allow prompt corrective action, should an undesirable trend develop.
lllili .1 82
Units 3 and 4 and adequately prepared.
B7
Unit I was the first fully upgraded unit to be placed in service in October 198 7 following a per'iod of in tens Lve commissioning act Lvity. For this reason, more detailed information is included for 198/. Although the operator error rate has not shown any uignificant increase in 198V, three Unit 1 events have demonstrated the potential problems that are associated wiLh a major upgrading program, (a) Test c ircuit operations resulted in a reactor trip during Unit 1 low power operat ion. A Unit 3 temporary procedure had been erroneously applied to Unit I to overcome equipment problems. (b) A planned reactor trip and recovery test I rorii fu I I power was unsuccessful and the Unit 1 LuactoL poi:;oned out due l.o Xenon bui ldup. The post trip Lespouse way substantially different from
212 CNS 9th ANNUAL CONFERENCE, 1988
The implementation of a major upgrading program at a multi-unit nuclear generating station creates mi environment where the probability of human error can increase. At Pickering NGS-A, extensive system changes are being carried out over a relatively long period of time and the preliminary results indicate that an effective program has been put in place to prepare the operating staff. The operator error rate has continued to decrease through the period when the first fully upgraded unit was returned to service. There has been an increase in the maintainer error rate, but this appears to have peaked during a particularly maintenance intensive commissIoning period.
ANNULUS GAS SYSTEM RESPONSE TESTS FOR PICKERING NGS A, UNIT I COMMISSIONING
J.M. KENCHINGTON. P.J. ELLIS AND D.G. MERANtm Ontario Hydro. Toronto, Ontario. Canada Nuclear Systems Department Ontario Hydro 700 University Avenue Toronto. Ontario H5G 1X6
ABSTRACT This paper reviews the AnnLlus Gas System response tests carried out during the commissioning of Pickering NGS A. Unit 1 in Hay/June 1987 at the request of the AECB. It shows that the dynamic responses of the system (the station dew point monitors and moisture beetles) agree with the design predictions for adequate leak detection capability. As a result of the tests a new dew point trending range is recommended in the short term for Pickering NGS and a new hygrometer calibration (D2O in CO2) in place of the present calibration W2O in N2) is recommended in the long term. This will improve the accuracy of the hygrometers as the calibration environment (D2O in CO2) will match that of the AGS. The beetle blaster lines will be modified to avoid core by passing and tuning improvements to the dew point model programme will also be implemented. 1.0
(a)
Bruce NGS A, Unit 1 in December 1982: Argon injection with purge mode operation.
(b)
Pickering NGS A. Unit 1 in September 1983: CO2 tracer test with recirculation mode operation.
(c)
Bruce NGS B, Unit 8 in September 1984: CH4 tracer test on individual strings prior to station completion.
Further details are given in Reference 1. The test program was formulated by Ontario Hydro Nuclear Systems Department with input from Ontario Hydro Research Department who carried out the tests (Reference 2 ) . The program was then integrated with Pickering A Commissioning team who produced the operating procedure. 2.0
DESCRIPTION OF TEST PROGRAM
2.1
Description of the Annulus Gas System
INTRODUCTION
The following program of Annulus Gas System (AGS) response testing for dew point monitors and moisture beetles for Pickering NGS A. Unit 1 arose from an AECB request. These tests were part of the Commissioning Program following the retubing of Units 1 and 2. The AECB request was that measured amounts of moisture and water be introduced into selected annul! and that actual station instruments be used to monitor the responses. It was also requested that the lead unit be tested, if possible, and so Unit 1 was selected. This was the most extensive and closest simulation of an actual moisture leak into the AGS of an Ontario Hydro nuclear unit which has been carried out to date. Previous station AGS response tests are as follows:
The main features of the AGS are shown in Figures 1 and 2. In the internal circuit, the dry gas, CO2, is circulated through the annul! between the pressure tubes and calandria tubes which are arranged in a series-parallel configuration. There are 195 parallel strings of channels with 2 channels in series. The gas distribution network consists of an inlet and outlet 5.1 cm (2 inch) diameter gas header with 29-2.5 cm (1 inch) diameter manifolds (inlet and outlet) supplying 13 annuli each on average via 6.4 mm (1/4 inch) diameter (pigtail) tubing connections.
CMS 9th ANNUAL CONFERENCE. 1988
213
Internal Circuit
Keclrculation is maintained at a design flow of b.7 1/s (STP' by two Metal Bellows MBc-02 compressors in parallel with redundancy. Overpressure protection is provided to maintain the design pressure cc 103.2 kPa(g) (15 pslg). A cold finger loop can be valved :P to provide a liguid sample of moisture for isoiopic analysis as required. Gas make up is provided by a bulk CQj supply with a high mate up alarm and a bottle ges back up supply. A separate purge circuit is also provided with discharge to the Reactor Building Ventilation System. A separate beetle blasting circuit is also provider! which supplies dry CO2 gas to each beetle location. This allows all four beetles to be blasted simultaneously after one has alarmed to check for a spurious signal. The beetle will re-alarm after blasting if liquid is actually present.
j-
Reactor Building
II.™
Units 1.2 CO, &9L.'«ISTP)
Flow Circuit
Bottle Gas System
DESIGN OF TEST PROGRAMME
3.1
Philosophy of the Tests
The purpose of the response tests was to measure the actual dynamic responses of the system as indicated by the Panametric dew point monitors and the moisture beetles to simulated moisture leaks into the AGS. Heavy water (D2O) injection was selected as the most likely leak is from a pressure tube (PT) resulting in Primary Heat Transport (PHT) liguid or from a calandria tube resulting in moderator fluid appearing in the AGS. The test was designed to be as real as possible with moisture injection as near as practicable to the envisaged locations where leaks may arise.
IB 2 1.Pi 12 Gn on)
FIGURE i PICKERING NGS A ANNULUS GAS SYSTEM FLOW SCHEMATIC
Moisture beetles are located at the lowest elevations ot each S.I cm (2 inch) diameter header on both the north and south sides (4 in all). An air-cooled finned heat exchanger (HX1) ensures that the return qas is cooled below 66°C before leaving the reactor vaults. In the external circuit. the system temperature and pressure during normal operation are monitored and alarmed if the former exceeds 66°C and the latter is below 13.8 kPa(g) (2 pstg) or exceeds 48.2 kPa(g) (7 psig). The flow rate is monitored and alarmed below 282 1/mln (STP). Dew point is monitored by two on-line Panametric (600/700 series) hygrometers in parallel.
214 CNS Pin ANNUAL CONFERENCE. 1988
3.0
In the case of the dew point monitors, a small moisture injection rate in vapour form was envisaged for the AGS. It was introduced from the west reactor face ir the most remote quadrant of the reactor (see figure 1) to simulate the potentially slowest response. Vapour injection in the external circuit of the AGS (the accessible circuit outside of the reactor vaults) was also selected to simulate the potentially fastest response. In the case of the moisture beetles a larger moisture injection rate in liquid form was envisaged for the outlet gas header on the west reactor face. This refers to a scenario in which a PT or calandria tube leak has spread through the annuli and pigtails of the leaking string of channels and the associated 2.5 cm (1 inch) diameter manifolds to the b.l cm (2 inch) diameter gas header (see figure 2). The
y
2 in. Dia. Headers (Inlet and Outlet)
TW Bursting Disc Connections Feeder . Cabinet
Typical Arrangement of Annuli; Upper South .Side Quadrant
Typical Arrangement of Annuli; Upper North Side Quadrant
North Side Beetle Injection Test 2.3 kg/h
South Side
Beetle Injection Test 2.3 kg/h Manifold #22:
o»B= Typical Arrangement of Annuli; Lower North Side Quadrant
1 in. Dia. Manifolds
Internal Dew Point Injection Tests 5 , 1 0 g/h
FIGURE 2
PICKERING NGS A ANNULUS GAS SYSTEM TEST LOCATIONS FOR WEST REACTOR FACE
outlet gas header was selected rather than the Inlet header, to prevent accumulation of moisture in the annuli which would have required subsequent extensive vacuum drying. 3.2
Location of Injection Points
3.2.1
Dew Point Response
The external injection point is located downstream of the compressors at a temporary pressure gauge (see figure 1). The other injection point on the west reactor face is located in a 1 inch diameter manifold (number 22) in the south side lower quadrant of the reactor (see figure 2). This was accessed by 6.4 mm (1/4 inch) diameter instrument tubing about 10 m (33 ft) long to reach the centre of the reactor, inserted via an inspection plug in the 5.1 cm (2 inch) diameter header to the most inboard four pigtail connections.
3.2.2
Moisture Beetle Response
There are two selected locations on the outlet header at the north and south sides of the west reactor face at old bursting disc connections (now capped) on the 5.1 cm (2 inch) diameter outlet gas header. These locations are about 3.4 m (11 feet) aoove the floor elevation. Injections were carried out on both the north and south sides to test both of the outlet header beetles (M52i and M522) (see Figure 2 ) . 3.3
Identification of Test Parameters
3.3.1
Dew Point Monitor Testing
The main process parameters which required monitoring or setting during the tests; are as follows:
CNS 9th ANNUAL CONFE RENCE. 1988
215
(a)
Reclrculatlon Flow Rate
The recirculation flow rate with two compressor operation was regularly monitored as it determines the AGS recycle period hence system response time. Recirculation was employed during dew point monitoring. (b)
AGS Temperature and Pressure
These parameters were regularly monitored throughout the test as they effect the gas (CO2) density hence gas humidity/dew point. (c)
Gas Make-up Rate
This was monitored throughout the test as it represents the gas leak rate from the system. This affects the dew point response as It modifies the recirculation flow rate and gives rise to actual moisture loss. (d)
Initial System Dev.j Point
The initial dew point of the system prior to moisture injection is controlled by the purge operation. The dew point is typically controlled between -30°C and -40°C. For the tests, the initial station dew point was -37 to -40°C. (e)
Injection Rates
During normal operation. the station hygrometers are capable of monitoring D2O leak rates from 0 to 17 g/h which Is the maximum discriminable leak rate. This is the highest D2O leak rate above which moisture could condense out in the lattice tube areas of the end shields. The lattice tubes are outboard of the tube sheet where the temperature is controlled at about 60°C by the end shield cooling system. Thus, the maximum dew point will be that of the coldest part of the system which results in under estimation of the DjO leak rate. Injection rates of 5 g/h and 10 g/h were selected to obtain distinct response curves for fitting the predicted dew point response from the dew point model program. 3.3.2
Moisture Beetle Testing
The main process parameters which reguire monitoring or setting during the tests are as follows: (a)
Purge Flow Rate
It is considered that different mechanisms are involved in the accumulation of moisture at beetle M521 on the north side compared with beetle M522 on the south side. In the former case, moisture will be transferred as vapour in the return gas flow condensing to liquid
216 CNS 9th ANNUAL CONFERENCE, 1988
once outside the feeder cabinet. Conseguently gas flow rate is an important parameter in the time to beetle alarm. Purge flow on a once-through basis equivalent to the recirculation flow rate in Section 2.3.Ha) was selected to prevent moisture recycling to all the annuli to avoid subsequent vacuum drying and extended unit outage. This is a conservative test since moisture is not _.turned to the system and allowed to condense as with normal operation. As the injection will produce a two phase mixture (steam/water) the liquid component will run to the nearest beetle: M521 for north side and Mb22 for south side injection. In both cases, moisture as vapour will be carried in the return gas flow to alarm beetle M521. Moisture as vapour will have more difficulty in reaching beetle M522 as diffusion is the predominant meciianism. (b)
Liquid Injection Rate
The moisture beetles respond to a higher range of leak rates at which condensation has taken place. Experience has shown that a D2O collection rate of about 2 kg/h is a safe and suitable criterion for initiating unit shut down. For this reason, and other technical considerations such as system hold up uncertainties, the liquid injection rate was selected as 2.3 kg/h. 3.4
Cold and Hot Tests
The test program was divided into initial cold tests before f he unit was heated up followed by hot tests during normal operation. The cold tests were intended as preliminary ones to commission and confirm the testing procedure and .echnique. They would also provide an initial set of data to check the predicted responses, allowing for possible modifications and providing the assurance of a "dry run". The hot tests had to be carried out with PHT pump heat and Moderator heat providing simulated normal operating conditions as the unit was not ready for start up. The hot test data were intended to verify the predictions of the dew point model and confirm the estimated beetle response times. 4.0
RESULTS OF TEST PROGRAM
4.1
Significance of Core Bypassing
The cold tests carried out in May 1987 were concerned exclusively with dew point response and the results showed a significantly slower
response in all cases, i.e., 5 and 10 g/h injection rates at both internal and external locations. There was concern that the measured injection rate may be in error and so this was checked by directing Che flow to a drier bed for independent measurement. This provided a more reliable means of checking the injection rate for the subsequent hot tests. During the first part of the hot test program carried out in June 1987, it was discovered that annulus gas was bypassing the core section or internal circuit via the beetle blasting lines. Figure 1 shows a typical connection between an inlet header beetle and outlet header beetle via a beetle blaster line connection. Such bypassing reduced the effective gas flow rate and hence the dew point response. The appropriate valves were closed to prevent core bypassing and the more important test runs were repeated.
//' IJI
FIGURE 4 COMPARISON fjF MEASURED AND PREDICTED DUE POINT RESPONSES
D2O INJECTION RATE = 10 gin
4.2
Comparison of Experimental and Predicted Dew Point Responses
1
2 Time (hr/
4.2.1
Significance of Dew Point Calibration
Figures 3 and A show plots of dew point as recorded by the station (Panametric) hygrometers vs predicted dew point as a function of time for injection rates of 5 and 10 g/h respectively. These test data refer to external circuit injection only due to a limited time. In either case, the experimental data fall below the predicted values. There is good agreement initially, but the discrepancy increases with the system dew point. The predicted curves allow for the fact that the station hygrometers were calibrated for H2O in N2 whereas the AGS received D2O injection into CO2 gas. The theoretical corrections for H2O to D2O and N2 to CO2 as a function of the reference dew point (H2O in N2) are given in Figure 5.
3
2
1
_ °" o
I2
I- -" £
I -3-]
I
FIGURE 5 THEORETICAL CORRECTIONS FOR MEASURED DEW POINT
Q
-4-5-
N2 — ^ CO2
-6-
Predicted Response - 18 Nodes
-7
-5 -10 -15 -20 -20 -30 -35 -40 Calibration Dew Point (H Z O in N 2 @ 0 kPa(g))
Station Parametric Hygrometer FIGURE 3 COMPARISON OF MEASURED AND PREDICTED DUE POINT RESPONSES D2O INJECTION RATE = 5 g/h
Research Department connected their own Panametric hygrometer Into the AGS circuit downstream of the station hygrometers to obtain further dew point data. The Research hygrometer was calibrated for D2O in CO2.
4 5 Time (hr)
CNS 9th ANNUAL CONFERENCE, 1988
217
Figure 6 shows the plot of dew point as recorded by the Research hygrometer vs prediction as a function ot time for the 5 g/h injection rate. This showed good agreement between experimental data and theoretical prediction. However, the AGS always contained H2O prior to the D2O injection because of bulk gas make up moisture. This makes it difficult to match precisely the experimental data with prediction as the AGS moisture content changed from predominantly H2O to DjO. during the test.
average flow transport characteristic of the system. As the flow characteristics of the AGS are complex due to heating and cooling zones with complex flow paths through the bearings, annuli etc.. the exact flow regime between fluid mixing and plug flow is best found experimentally. Theoretically, if perfect mixing occurs, N - 1; if plug flow occurs. N -» «. In practice, 1 < N < ». In figures 3, A and 6, predicted dew point responses for N - 2 and N- 18 are given for comparison. It can be seen that in the
0 -2 Predicted Response — 2 Nodes„
-4 -6 -8
Predicted Response - 18 Nodes
-10-
(J
-14Research Hygrometer — Panametric
2 -16-18•20-
22•242628
7
V
FIGURE 6 COMPARISON OF MEASURED AND PREDICTED DUE POINT RESPONSES
3032D2O INJECTION RATE = 5 g/h
•34-
36 6 Time (hr)
4.2.2
Tuning of Dew Point Hodel
The experimental dew point response curves provided important data to "tune" the present dew point computer model. There is an adjustable parameter, N, which represents the
218 C N S 9th ANNUAL CONFERENCE, 1988
7
10
11
early stages (1 or 2 recycles of the AGS) there is a distinct difference between predicted curves for different N values and that N = 18 better represents the delayed response followed by a rapid dew point response. For longer periods of times, the
12
difference in predicted curves Is very little because recycled moisture dominates the dew point response. However, the first one or two recycle periods are the more Important for the action dew point range at the stations and consequently N - 18 Is more representative. 4.3
Comparison of Experimental and Predicted Moisture Beetle Responses
4.3.1
Worth Side Injection
ACTION
EXPERIMENTAL RESPONSE TIME
Pickering NGSA Station Dew Point High Alarm (-10°C) North Side Outlet Beetle (H521) Alarm
DESIGN RESPONSE TIME
28 min
< 2.9h
34 min
< 1h
Prior to the beetle alarm the average dew point rate of rise by 28 rain was greater than 40°C/h compared with the action set point of 7°C/h for shut down of the unit to zero power hot condition. 4.3.2
(a)
A calibration correction is required to relate the measured station dew point to the predicted dew point. This correction Includes both the theoretical adjustment due to the station probes being calibrated for H2O In N2 but operating in a D2O in CO2 annulus gas environment and a further empirical correlation to achieve an accurate prediction.
(b)
There is a closer agreement between the predicted and measured dew point if the hygrometers are calibrated for the AGS environment, namely D2O in CO2. This agreement is shown in Figure 6. Allowing for the manufacturer's quoted accuracy of +2"C only a small empirical correction Is required.
(c)
Figures 3. 4 and 6 show that "tuning" the dew point model to N = 18 improves the accuracy of prediction particularly in the first few cycles which are most applicable to station dew point leak detection.
(d)
The response times of the beetles for both the north and south side D2O injections were within the 1 hour design prediction.
(e)
The blaster line isolation valves must be closed to ensure no core by passing during normal operation until a more permanent solution has been implemented.
6.0
DISCUSSION
South Side Injection
ACTION
EXPERIMENTAL RESPONSE TIME
Pickering MGSA station Dew Point High Alarm <-io°C) North Side Outlet Beetle (M522J Alarm
44 min
51 min
DESIGN RESPONSE TIMS < 2.9h
< 1h
Prior to the beetle alarm the average dew point rate of rise was greater than 27°C/h compared with the action set point of 7°C/h for shut down of the unit to zero power hot condition. 5.0
down the unit safety allowing for certain adjustments. The specific points which have come to light are as follows:
CONCLUSIONS
Comparison of the experimental test data with design predictions shows that the present design parameters and operating procedures are adequate to determine PT leaks In time to shut
This test is consistent with previous tracer tests carried out at Pickering NGS and Bruce NGS but has provided more information on hygrometer calibration and beetle response time. Owing to core by-passing via the blaster line connections, the original set of test runs had to be repeated but there was not sufficient time to repeat all of them. The dew point response tests Involving Injection of moisture to the internal circuit (see Section 3.1) were not repeated but the repeated external circuit responses were considered representative of the internal circuit responses based on the results of the Pickering NGS A tracer tests in 1983 (Reference 1). The following changes to AGS operation have been adopted at Pickering NGS in consultation with AECB as a result of the tests.
CNS 9th ANNUAL CONFERENCE. 1988 219
(a)
(b>
220
The predicted dew point response will include a calibration correction relating the station hygrometer calibration for H2O in Nj to the AGS environment of D2O in CO2. The station dew point trending range and dew point rate of change set point will be reset accordingly. This
CNS 9th ANNUAL CONFERENCE. 1988
(c)
The adjustable parameter, N. in the dew point program will be adjusted to 18 for more accurate prediction of dew point response for Pickering NGS AGS.
7.0
REFERENCES
(1)
Kenchington, J.H. et al, "fln Overview of the Development of Leak Detection Monitoring for Ontario Hydro Nuclear Stations", CNS Conference, St. John, June 1987.
(2)
Singh, V.P. and Chang, S.D., "Pickering Annulus Gas Moisture Injection Tests", Ontario Hydro Research Report No. 87-254-K, February 1988.
Session 6: Nuclear Safety Experiments and Modelling
Chairman: J.C. Luxat, Ontario Hydro
CNS 9th ANNUAL CONFERENCE, 1988
221
A MODEL FOR DROPLET SIZE DISTRIBUTION IN FLASHING JETS MINOO RAZZAGHI Atomic Energy of Canada Ltd. Whiteshell Nuclear Research Establishment Pinawa, Manitoba ROE 1L0 ABSTRACT During some postulated accident scenarios for a CANDU PHVIR, high-enthalpy coolant is released into the containment as a two-phase jet. The liquid portion of the jet atomizes into droplets due to flashing, hydrodynamic drag and instabilities. This jet may contain dissolved fission products. The fraction of jet fluid airborne as aerosol-sized droplets limits the postulated release of fission products to the outside atmosphere, and so is of interest. A model has been developed to predict the droplet size distribution produced by such a jet. It incorporates the aerodynamic and thermal fragmentation of the jet and is based on a Konts Carlo technique. INTRODUCTION Loss-of-coolant accident analyses for CANDU reactors include a postulated scenario in which highenthalpy coolant at high temperature and pressure discharges into a containment system at atmospheric pressure. The liquid of the jet atomizes into droplets because of flashing (vapor formation caused by pressure -hange) and hydrodynamic instabilities. The breakup of the liquid in a superheated jet (liquid at a temperature above its atraospnericpressure boiling point) into drops can proceed by two mechanisms. One mechanism is by aerodynamic forces acting on the liquid surface to break it up and the other is by thermal fragmentation. The aerodynamic breakup mechanism is caused by the dynamic effects of the surrounding medium on the surface of the jet. It can be qualitatively explained as follows. When a disturbance (however small.), exists on the surface of the jet, the gas pressure over the crest of this disturbance is less than the average pressure, while the pressure at the base of the disturbance is higher than the average. Now consider an undisturbed streamline far from the surface of the jet. In the zone between the streamline and the jet surface, the gas velocity (considering a stationary jet and a moving gas) over the crest of the disturbance is higher than at the base. According to the Bernoulli equation, pressure at the crest must be lover than at the base. That is, the disturbance that has somehow formed on the liquid jet surface tends to increase. The higher the relative velocity between the jet and the medium, the more pronounced this effect w i U be. An increase in the size of the disturbance results in the detachment of a droplet from the jst surface. The thermal fragmentation mechanism results from the vapor bubble extending the liquid into a thin film. When the vapor bubble ruptures, the liquid film forms small dropiefs. The relative importance of these two mechanism depends upon the physical properties, particularly the degree of superheating.
These fragmentation processes may be accompanied by a process of an opposite character, coalescence of the drops. However, for superheated jets, the coalescence process is hindered to a great degree by the repulsive forces between evaporating droplets (1). Brovn and York (2) studied the sprays formed when hot water at pressures up to l MPa at different temperatures (360 to 378 K) was released in atmospheric conditions. The empirical correlation they developed based on their test results implies that both aerodynamic and thermal effects are important in determining the droplet sizes. This paper provides a model for jet breakup and estimates the droplet size distribution from a flashing water jet. It couples the two mechanisms of droplet formation - both aerodynamic and flashing fragmentation. Although the model is theoretical, it relies on available empirical correlations and experimental data. MODELLING APPROACH In order to account for both aerodynamic and thermal fragmentation mechanisms, our modelling approach is based on sequential coupling of the two mechanisms. This modelling Scheme is shown schematically in Figure 1.
HIGH ENTHALPY WATER JET
o
_ »
o—
O PRIMARY DROPLETS
FIGURE 1. MODELLING SCHEME The discharging jet is assumed to break up due to aerodynamic forces yielding a droplet size distribution. The droplets formed in this stage are called primary droplets. Their formation is assumed to occur so rapidly that thermal loses do not occur. That is, the temperature of the primary droplets is considered to be the same as the discharged fluid. C N S 9th ANNUAL CONFERENCE, 1988 223
The primary droplets are then assumed to reach thermodynamic equilibrium by developing a bubble in the droplet core due to evaporation during this stage. Single-bubble formation has been observed experimentally (3) within flashing butane droplets. The model takes into account the convective heat tra-nsiex iroui the suriace oi the primary droplets, which lowers their temperature. However, the effect of surface evaporation on the size of primary droplets is assumed to be negligible compared to bubble formation in the core of primary droplets. The combined bubble and liquid film are called "secondary droplets." The bubble growth and the thinning of liquid film in these secondary droplets continues until the growth of interfacial instabilities breaks the liquid film and forms tertiary droplets. If the tertiary droplets so formed are supetheated enough to satisfy the criterion for the formation of an internal bubble (discussed later), the sequence of internal boiling and breakup will reoccur. FORMATION AND DISTRIBUTION OF PRIMARY DROPLETS Levich (4) has shown that droplet formation as a result ot aerodynamic breath is iue to. disturbances whose wavelength X satisfies the criterion, X > o/(p, u 2 ) .
11
The discrepancy between equilibrium temperature, void fraction and pressure will go to zero as the spatial rate of pressure changes goes to zero. This has usually been observed for the situation where the fluid initially contained in a large reservoir is discharged through a pipe of sufficient length having a cdaine'iet rath smaller than the tesecioit diametec (9). The model describing the flow in this situation is called the Equilibrium Model (9). The other limiting model is realized as the length of discharge tube approaches zero (i.e., the orifice flow). At this extreme, vaporization is considered to be zero and the flow is described by the Frozen Model which uses the orifice equation for incompressible liquid flow G = 0.61 [5]
l
(9). Typical data (9) shows that the Frozen Model leads to the maximum flow. We have chosen the Frozen Model because it is consistent with our modelling approach - aerodynamic breakup of the liquid jet followed by internal l the pciwacy droplets•
INTERNAL DROPLET BOILING
We assume that the relative velocity, u, between the ambient gas and the jet is equal to the discharge velocity - that is, we are considering that the superheated jet flows into stagnant media. Disturbances on the surface of the jet are caused by random processes with tHe associated random variable being the wavelength of the disturbance. Since the wavelength is non-negative, one possible distribution function is lflgnormal (5), 1
1
1 exp[-
f(X) = 2 1 2
(2na g ) /
In2(
pressure (Pd = P>n,b + W D p ) drops below the saturation pressure corresponding to the temperature in the hottest part (usually the center; T c o n ) of the droplet. This condition in terms of temperature is p (10) (10)
X
1 )J . [2[
[61
t.
(
X
C
The parameters of this lognormal distribution are [3]
IWeJ 2
a I (p, u )
The extent of inequality required in equation 16] fat tb.e. onset at internal boiling depends on the mechanism of nucleation in the droplet. Based on the experimental data of Bushnell and Gooderum (11), this inequality is estimated as
14]
I1 where the critical Weber number (Weo) determine., the maximum stable drop size when surface tension force balances the hydrodynamic force on the droplet. The critical Weber number deper-is on the flow regime around the droplets and is a function of the drag coefficient (7). It has a value between 12 and 22 (8). Thus, given the discharge velocity u, the distribution of unstable wavelengths and primary droplets will be specified by equations [3] and (4J. The discharge velocity and flow rate of flashing fluid through a break depends on the spatial rate of pressure change within the reservior containing the U u i d . The dependence is due to the Unite cate, at. which evaporation and local void fraction can change. This results in nonequilibrium temperature and void fraction conditions.
224 CNS 9th ANNUAL CONFERENCE. 1988
0.9
17]
(1 • P. nb
C
for internal boiling to occur. If it is less, surface evaporation takes place until the droplet reaches thermal equilibrium with the surroundings. Internal boiling results in the formation of a bubble inside the droplet. Bubble growth starts from a nucleus with an initial size Db* (12) W(P v {T c . n ) -
Pd)
[81
Equation [8| is the Laplace equation which relates the pressures on opposing sides of a curved interface {bubble watlei) to its radii oi cur'Jatwce {diameter ȣ bubble nuclei). The bubble in the droplet and the relevant parameters are shown in Figure 2.
where D B is the diameter of the secondary droplet (=D b + 2H). As the bubble grows and D 5 increases, the droplet pressure for mechanical stability changes and so does the saturation temperature (T 6 a t ) according to T
s.t = T ( p d ) = T{P a m b
+
*o/D.)
.
[13)
Based on material balance, the thickness of the liquid film is H = 1/2 [Dp
+
D b (l - pv/p1)l1''3 - D b /2,
[14J and therefore, dD s /dt = D b
Pl)
[dDb/dtl Ds-
115]
The solution of equations [9] to [15) describes the bubble growth process inside a droplet.
BUBBLE COLLAPSE AND FORMATION OF TERTIARY DROPLETS FIGURE 2.
SECONDARY DROPLET
The bubble grows by 2 orders oE magnitude from its initial microscopic size VLb during a short time interval (t) (12). The duration (T) decreases as the superheat increases and is of the order of 20 to 1000 us (12). After the intial period ( T ) , the bubble growth rate is described by (12) [9]
dD b /dt
6 =+ (
L) (Td - T B . t )
[101
where
4> = 1.117 - 0.035 (Td - T B , t ) + 0.00043 (T,, - T B . t ) 2 for (Td - T 5at )<40 K and
= 0.4 otherwise.
LIQUID TEMPERATURE DURING BUBBLE GROWTH During the initial bubble growth period ( x ) , the droplet temperature (T d ) decreases due to convective heat transfer at the external surface. It is estimated from ( 6 ) , (Td - T.,b) exp(-12K£
(UJ After this time, the liquid film temperature decreases both by convection from external surface and by latent heat of evaporation at the interface of the bubble. This decrease is described by (6), <|Td
12K ( P.(Ta-T..b)
dt
PiCp(Dj - D b )
3 Trt W
3Lp v Dg dD b /Dt PlCp
dD./dt - D h
D= - D b )
D, - D b )
One of the uncertainties in the modelling of flashing jets is the specification of conditions at which bubbles would break. Based on photographic evidence of flashing jets, Lienhard and Day (14), assumed that a bubble breaks when its radius is equal to diameter of discharging nozzle. However, their analysis does not consider aerodynamic breakup. To model film breakup and bubble collapse, one might think of the secondary droplets as a pressure vessel (or a soap bubble). The bubble growth and film thinning would continue until a minimum film thickness is reached. This minimum film thickness depends on the yield stress of the liquid water film and pressure inside the bubble. However, we could not find any data regarding water film yield stress. Hence, we based our breakup mechanism on the stability of the liquid film. To determine the conditions for collapse of the bubble, the secondary droplet is considered to have no local slip velocity relative to the vapor around it. That is, the secondary droplet is considered to be in a stagnant media and falling under gravity. The maximum size of the secondary droplet in this case is governed by the Rayleigh-Taylor instability (15). The Rayleigh-Taylor instability occurs at the interface of two overlaying fluids when acceleration of the interface is toward the lighter fluid. Thus, when the secondary droplet is falling under gravity, the lower hemispherical part of the interface is prone to instability. If the instability that is initiated at the lower surface grows only slightly as it travels along the interface toward the upper surface, it will die out because the upper surface is stable. The criterion for instability-induced breakup is based on the comparison of a characteristic time, t., the time required for the amplitude of a small disturbance to grow by a factor 'e' (=2.718), with a characteristic time, t,, the time required for the disturbance to move along the interface to the equator. Grace et al. (15) postulated that breakup occurs if t. is sufficiently larger than the growth time t.. They presented the following empirical correlation for breakup,
CNS 9th ANNUAL CONFERENCE. 1986 225
RESULTS AND DISCUSSION
t. = 1-4 The equations for t. and to are given in Reference 6. The characteristic disturbance growth time t. depends on the liquid film thickness H and t, depends on the size of the secondary droplet. As the bubble grows, the Liquid, film, becomes ttiuuxer. and the size of the secondary droplet increases, ta increases and t, decreases. When equation 116] is satisfied, the liquid film breaks and the bubble collapses, creating the tertiary droplets. According to the experimental observations of Netfitt et al. (16), the number of tertiary droplets (n) varies from 1 to 10 per liquid film and bubble burst. In the present calculation the tertiary droplets are considered to be all of one size (the average size). Thus, mass conservation of liquid in the film after breakup yields Dt =
- D3b)l'3/n"3
Figure 3 shows the droplet size distribution for different discharge conditions.
s «H
[17]
The number of tertiary droplets per bubble burst, n, is considered to be a random integer distributed uniformly between 1 and 10. That is, the probability of getting any number of tertiary drops between 1 and 10 is equal.
METHOD OF COMPUTATION The computational scheme is based on the Monte Carlo method (method of statistical trials) (17). The droplet size distribution in a flashing jet is determined as follows. A random (statistical) initial instability wavelength X in the jet results in a primary droplet. Following a sequence of flashing in the droplet, the bubble growth and collapse are calculated deterministicaHyFinally, a second random number determines the number of tertiary droplets formed from the collapse of the bubble. The secondary droplets so formed are again subjected to flashing if they meet the flashing criterion (equation |7J). The size distribution of equilibrated tertiary droplets is determined by recording the number and size of final droplets for different values of initial random wavelength and the random number of droplets in the range of 1 to 10 formed by bubble collapse. The random variables, wavelength (X) and number of droplets per bubble burst (n) have to be sampled from their corresponding distribution. As mentioned earlier, X is distributed lognormally with o g and Xg given by equations (3] and [4], and n uniformly between 1 and 10. The computer code that implements the model takes 1000 random variables for X (the primary droplet size) and 20 random variables for n per X.
226 C N S 9th A N N U A L C O N F E R E N C E . 1988
10. 1520. Droplet Diameter fim
FIGURE 3.
AVERAGE OF FOUR RUNS FOR DIFFERENT DISCHARGE CONDITION FOR
t = 1000 vs and We,. = 12. discharge pressure = 2.5 MPa, discharge temp = 475 K discharge pressure = 5.0 MPa, discharge temp = 500 K o discharge pressure = 10. MPa, discharge temp = 550 K
We have taken the condition of the discharge fluid as slightly subcooled. This way we are consistent with our modelling approach so that the discharge fluid follows the Frozen Model and that there is no vaporformation before discharging into containment. This figure shows the average results of four runs of the code (each taking 1000 samples for X and 20 samples for n per \\. It is shown that reducing the amount of superheating (lowering the discharge pressure and temperature) increases the drop size. One of the parameters of the model is the period t during which the bubble grows from its microscopic size by 2 orders of magnitude. As mentioned earlier, it is of the order of 20 to 1000 us. This period affects the heat loss (droplet temperature) according to equation [11]. More specifically it determines how much heat is lost by convection compared to internal vapor generation. Figure 4 shows the results of the simulation for two values of t at the discharge condition of 5 MPa.
CONCLUDING REMARKS The model presented here is based on several hypotheses - a lognormal distribution function for instability wavelengths, a uniform distribution function for the number of tertiary droplets formed per bubble burst, a relation between the discharge velocity and the velocity of the ambient gas, the duration of inertial growth period of bubble nuclei, and a bubble collapse mechanism. The basis for these hypotheses were presented in the previous sections. However, alternative hypotheses are possible.
«.
«. 12. Droplet Diameter Jim
IC.
FIGURE A.
EFFECT OF INITIAL BUBBLE GROWTH PERIOD ON DROPLET SIZE distribution for a jet discharging from 5 MPa and 500 K to ambient pressure of 101 KPa. t = 100 us, We c = 12; X T = 1000 us, We, = 12
As T decreases, the droplet size distribution tends toward smaller droplets. This is expected because more internal vapor generation and consequently more bubble bursts are expected as the time for heat convection is reduced. The effect o£ critical Weber number is shorn in Figure 5. Evidently the droplet size distribution of a flashing jet is not a strong function of the critical Weber number.
The model presented here describes the droplet size distribution near the break. It considers the ambient gas to be solely vapor. As the jet grows radially by entrainment of air, the steam concentration around the droplets changes. This brings about processes of surface evaporation and condensation followed by a variation of the droplet size distribution. Another mechanism operating to change droplet size along the jet axis is coagulation. This process was neglected in the current model because evaporation near the break is expected to hinder coalescence (1). The processes are considered in the future development of the model. One possible improvement to the model is the calculation of the initial growth period (T). This calculation is complex and involves numerical solution of coupled momentum, heat and mass transfer processes. Future developments will consider this aspect.
NOMENCLATURE
c cPp
D G H Ke Kf L Mu n P pv R t
T u We
Dimensionless Clausius-Clapeyron number • MT..,,} M./) M w /
£ GREEK
n a: It. Droplet Diameter /im
Thermal diffusivity Parameter defined in eqn (10J Geometric mean of wavelength of instabilities Density Surface tension Geometric standard deviation Duration of first region of bubble growth
FIGURE 5.
EFFECT OF CRITICAL WEBER NUMBER ON DROPLET SIZE distribution for a jet discharging from 5 HPa and 500 K to ambient pressure of 101 kPa. T * 100 M S , We c = 22; X T = 100 ps, We c . 12 CNS 9th ANNUAL CONFERENCE. 1988
227
SUBSCRIPTS ambient bubble center droplet gas e liquid 1 primary P secondary droplet s t tertiary droplets SUPERSCRIPTS Ini tial i
amb b cen d
REFERENCES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
(11)
(12)
FEDOSEEV, V.A., "Dispersion of a stream o£ superheated liquid", Colloid Journal, 20, 463, 1958. BROWN, R., and YORK, J.L., "Sprays Formed by Flashing Liquid Jets", A.I.Ch.E. Journal, 8, 149, 1962. SHEPHERD, J.E., and STURTEVANT, B., "Rapid evaporation at the Super-heat Limit", J. Fluid Mech., 121, 379,1982. LEVICH, V.G., "Physicochemical Hydrodynamics", Prentice-Hall Inc,1962. WILLIAMS, M.M.R., "Some Topics In Aerosol Dynamics", Prog, in Nuc. Energy, 17,1,1986. RAZZAGHI, M., "Droplet Size Estimation Of TwoPhase Flashing Jets", Atomic Energy of Canada Limited Report, (in press) 1988. BRODKEY, R.S.,"The Phenomena Of Fluid Motions", Addison-Wesley, 1967. HINZE, J.D., "Fundamentals Of The Hydrodynamic Mechanism Of Splitting Dispersion Process", J. Am. Inst. Chem. Eng., 1, 289, 1955. FAUSKE, H.K., "Flashing Flows or: Some Practical Guidlines For Emergency Releases", Plant Oper. Prog., 4,132,1985. GYARMATHY, G., "The Spherical Droplet In Gaseous Carrier Streams: Review And Synthesis", in "Multiphase Science And Technology", Hewitt, G.F., et. al. Editors, Hemisphere Pub. Corp., 1982. BUSHNELL, D.M. and GOODERUM, P.B., "Atomization of Superheated Water Jets At Lov Ambient Pressures", J of Spacecraft and Rock., 5,231,1968. PLESSET, M.S., and SADHAL, S.S., " Void Volume Growth In Superheated Liquids", Am. Soc. Mech. Eng., Heat Transfer Div., Fundam. Phase Change, 38,43,1984.
(13) ABDELMESSIH, A.H., "Spherical Bubble Growth In A Highly Superheated Liquid Pool", in "Cocurrent Gas Liquid Flow", Rhodes, E., and Donald, S.S., Editors, Plenum Press, 1909. (14) LIENHARD, J.H., and DAY, J.B., "The Breakup Of Superheated Liquid Jets", J. Basic Eng., Trans. ASHE, 515, 1970. (15) GRACE, J.R., WAIREGI, T., and BROPHY, J., "BreakUp Of Drops And Bubbles In Stagnant Media", Can.J.Chem.Eng., 56,3,1978. (16) NEWITT, D.M., DOMBROWSKI, N., and KNELMAN, F.H., "Liquid Entrainment, 1. The Mechanism Of Drop Formation From Gas Or Vapor Bubbles", Trans.Instn.Chem.Engrs, 32,244,1954. (17) RUBINSTEIN, R.Y., "Simulation And The Monte Carlo Method", John Wiley, 1981.
228
CNS 9th ANNUAL CONFERENCE, 1988
REFILL STUDY OF A CANDU-TYPE HEADER/FEEDER SYSTEM UNDER NEAR-ZERO HEADER-TO-HEADER PRESSURE DROP
J.E. KOUALSKI AND B.N. HANNA
Atomic Energy of Canada Limited Whiteshell Nuclear Research Establishment Pinawa, Manitoba ROE 1L0
ABSTRACT The behaviour of CANDU header/feeder piping during refill in a near-zero header-to-header pressure drop determines the delivery of emergency coolant to the fuel channels in certain postulated loss-of-coolant accidents. This paper describes an experimental test facility and program to examine this behaviour. The header/feeder behaviour was examined to determine the effects of header preheat temperatures, injection flow rate, break size and break location. The thermalhydraulic behaviour during refill is described in detail for a selected experiment. The break size was found to have the least influence while the header preheat temperature and injection flow rate were found to have the largest. A preliminary simulation of one test with the two-fluid code CATHENA is shown. While the qualitative agreement between the simulation and experiment was good, the comparison shows that improvements in the code's constitutive relations are required.
experiment are compared with a preliminary simulation using the two-fluid code CATHENA.
TEST FACILITY Figure 1 shows a schematic of the M r g e - g c a l e Header (LASH) test facility at Stern Laboratories Inc., formerly Westinghouse Canada Limited , in which the experiments were conducted. The test loop consists of inlet and outlet horizontal headers connected by 30 feeder pipes. Blowdown and injection arrangements are attached to the facility together with two low and high pressure pumps and the connecting pipes.
INTRODUCTION In certain postulated LOCAs, a low pressure drop between the inlet and outlet headers of one core pass is predicted to occur early in the transient and persist for an extended period of time. The modelling of such situations requires that the header/feeder characteristics be known under transient conditions. This paper presents experimental data and a CATHENA simulation of refilling in a CANDU-type header/feeder system during water injection into both headers.
a0 - b^ ^~'2
Very little published information is available on h e a d e r b e h a v i o u r under transient conditions. Kowalski and Krishnan |1J studied the refill behaviour of a header/feeder system during water injection into the inlet header. Most experimental and theoretical investigations deal with accident scenarios in which two-phase flow is present in the inflow header. Header behaviour under steady-state two-phase inflow has been examined experimentally by Kowalski and Krishnan [2| and been modelled by deMan [3| and Gulshanl [4J. This paper describes the header/feeder experimental facility in which cold-water injection under near-zero header-to-header pressure drop conditions were performed. The experimental program and the parametric range of conditions are discussed. Experiments were divided into depressurization, header refill and feeder refill phases. The flow behaviour in each phase is examined in detail for a selected test. The header behaviour is examined with respect to the injection flow rate, break size and location, and header preheat temperature. The results from a d o u b l e - b r e a k , double-injection
FIGURE 1:
SCHEMATIC OF THE LARGE-SCALE HEADER (LASH) TEST FACILITY
The inlet and outlet headers are 325 mm ID and 4.2 m long, half-lt.igtb Pickering HGS headers Vertical turrets are attached to the North and South ends of each header. The feeders (each 50 mm ID) are connected to nozzles distributed along the
CNS 9th ANNUAL CONFERENCE. 1988 228
length of the headers In six banks of five (two at 90°, two ?• 45' and one downward oriented) nozzles each. Different angular arrangements are present In CANDU header/feeder systems but this is expected to pioduce similar results. Each feeder has an unheated horizontal section located about 10 m below Use Uea
e n v i r o n m e n t a l heat losses a small amount of condensate was trapped in the test loop before starting the transient. The blowdcwn and water injection started at approximately the same time. In some tests, a small amount oi Injected -water arrived at the headets before the onset of depressurization because of the manual sequence used to open the injection valve. Blovdown was initiated by opening either one or both of the 50- or the single 150-mm quick-acting ball valves, and subcooled water at approximately 30°C was injected into the North turret of both headers. The data acquisition system was initiated manually to record first the steady-state conditions, then the transient data. The experiment was terminated when conditions indicated that the test loop was refilled. Tests were conducted shown in Table 1.
TABLE 1:
over
the parametric
ranges
TEST CONDITIONS
Total Water Injection Flow Rjite: 30 - 60 kg/s Initial Header Pressure: 1-5 - 5 HPa(g) Loop Preneat Temperature: 200 - 320*C Break Size: 50, 100, and 150 mm Inlet/Outlet or Both Headers Break Location:
TEST RESULTS AND DISCUSSION While there has been insufficient time to analyze all of the tests in detail, representative tests have been selected and analyzed. Other tests have been examined only briefly to obtain general trends. Three phases were apparent in e a c h of the experiments: 1) a depressurization phase, 2) a header refill phase and 3) a feeder refill phase. A detailed analysis of each phase is presented here for the 50-mm double-break test conducted at a total Injection flow of 30 kg/s and an initial loot temperature of about 250"C. Depressurization Phase
FIGURE 2:
HEADER INSTRUMENTATION LOCATIONS
EXPERIMENTAL PROCEDURE AND CONDITIONS To achieve the conditions of near-zero header-toheader pressure drop, the loop was set up for most of the tests in a symmetric double-break, doubleinjection (i.e. injection into the North turrets of both headers) configuration. However, several tests vere conducted with break located at either the inlet or outlet header to investigate effect of the break's location on the header refill behaviour.
When the break opens at about 21 s, the inlet and outlet headers depressurize rapidly. Figure 3 shows both header gauge pressures as a function of time. A very steep initial depressuri^ation is observed, during which time the header pressures fall from 4 MPa to 3.3 MPa during approximately 1 s. This rapid depressurization causes flashing of the small amount of condensate trapped in the downwardorientated feeders and in the unused feeder nozzles. This flashing helps to lessen the rate of depressurization seen after 22 s. After the initial blowdovn; the pressure decrease is controlled by subcooled boiling within the headers and feeders as well as condensation on the cold injection water and the flow to the break. Header Refill Phase
The initial conditions for an experiment were established by cii "lating superheated steam in the loop with both header-break isolation valves closed. Vhen the m e a s u r e d h e a d e r and feeder surface Temperatures reached trie d e s i r e d initial temperature, the steam flow was shut off and both of the inlet and outlet header break isolation valves were opened. For the single-break tests only one isolation valve was opened. Because of
230 CNS 9!h ANNUAL CONFERENCE. 1988
Figure 4 shows the injection flow rates into the inlet and outlet headers as a function of time. Water enters into both headers almost immediately after opening ot the In'eak. ^ e time delay lot water reaching the inlet h e a ^ r and outlet headers is between 3 and 4 s, respectively. The sharp decrease and recovery of the injection flow rates after their initial increase is believed to be an
indication of water arrival at the headers. The flov distribution between the headers is fairly uniform. However, the injection flow rate increases from 10 to 16 kg/s during the header refill phase due to the continuous decrease in header pressure.
is little net movement of steam or water between the headers during this phase.
Time (s) 27
.
"-D—a—fl
- Inlet Header
12 11
10
u—D—n—a—n—a 9 8 7
6 5
o
u u
4 3
2
1
-_-- Outlet Header
32 ^—•—D—U 12 11 10 9
LJ—•—D—D—D 8 7 6 5 i
D 3
• 2
Q I
T3—•—o—•—a~rj—a—D—o 12 11 10 9 8 7 6 5 4
rj 3
rj 2
cr 1
37
FIGURE 3: INLET AND OUTLET HEADER PRESSURE HISTORIES
43
u u u JJ u u u—u~tf—n •—cr 12 11 10 9 8 7
6
5
4 3
2 1
FIGURE 5: INLET HEADER REFILLING SEQUENCE
il 30
FIGURE 4: COLD-WATER INJECTION FLOW RATE HISTORY The sequence of header refilling is shown in Figure 5 tor an inlet VieaiM • 'IVAs tea^t s t i l l i n g behaviour was inferred Erom the conductivity probes located at the South, Middle and North planes of the header. Flow stratification was evident during the refill phase and both headers followed similar patterns. In this test the refill time period varied between 15 and 25 s. The Mid-plane o£ the header refilled first followed by the North and South ends. It is presumed that the time deliy for refilling the South and North ends of the header was due to the time requir&d to remove storerf energy from the header wall through subcooied boiling. The steam produced ei'.her flows to the break or is condensed locally by the subcooied injection flow. Fluid temperatures measured vertically through both headers were uniform, indicating a homogeneous temperature distribution. It Is believed that the homogeneity results from subcooied nucleate boiling on the header wall and that it indicates that a bubble flow regime is present ai the bottom of the header, while a mist flow distribures energy in the upper part ot the taeaAet• The header-to-header pressure drop shown in Figure 6 was very small during the entire header refill phase. ThJb indicates that on average there
I
I -30
200 Time (si
FIGURE 6:
HEADER-TO-HEADER HISTORY
DIFFERENTIAL
PRESSURE
As injection proceeds, more subcooied water comes into contact with the steam generated leading to rapid condensation. No steam was detected by the conductivity probe rakes in the headers after about 90 s. However, two-phase flow was indicated by the three-beam 7-densitometer in the common blowdown line until approximately 150 s. The fluid density measured in the blowdown line is shown in Figure 7. The top and bottom header external surface temperatures were below saturation by 190 s and 140 s, respectively. •Ettect oi Initial Header temperature The header refill times inferred from conductivity probes are shown in Figure 8 as a function of the initial header temperature. For a given break size, the time required to refill the headers increases C N S 9th A N M U A L C O N F E R E N C E . 1988 231
with the Initial temperature. This Increase Is presumed due to the higher Initial stored energy in the header walls which is removed by subcooled boiling. The temperature effect appears to be more pronounced at the lower Injection flow rates.
20
• •
•
I
A AA
• • •
A
% 10
_ 1000
•
Heade
u
Total Injection Rate 30 kg/s J Toial Injection Rute 60 kg's
1
50
150
100 Break Size (mm)
FIGURE 7: COMMON BLOWDOUN LINE DENSITY HISTORY
Totaf (ajtctfon j
! SO ftgfi
Total Injection ]
! 60 kgis
FIGURE 9: EFFECT OF BREAK SIZE ON HEADER REFILL TIME
Inlet Header Outler Header
A
40-
20+
A A
Si
30'
* •
10
• O
20-
o DD AA
200
250
300
10-
Initial Loop Temperature (°C) FIGURE 8: EFFECT OF INITIAL LOOP TEMPERATURE ON HEADER REFILL TIME
O —i
1—
IB
OB Break Location
Effect of Break Size FIGURE 10:
The break size has a slight influence on the header refill time as shown in Figure 9. For break sizes of 100 and 150 mm, and total injection flow rates of 60 kg/s, the header refilled in 9 and 13 s. The refill time was longer (15 to 19 s) for a break size of 50 mm primarily due to the decrease In the injection flow rate to 30 kg/s. Effect of Break Location Figure 10 shows that more time was required to refill the Inlet header for the outlet-header break tests than for the inlet~header break tests. This Is due to tjie asymmetric injection vatet ilo-J distribution available to the headers. Much higher watsr Inflow was detected at the healer where the brea>. Is located.
232
C N S 91h A N N U A L C O N F E R E N C E . 198B
D8
EFFECT OF BREAK LOCATION ON HEADER REFILL TIME
Feeder Refill Phase As water flows into a feeder large quantities o£ steam are produced when it contacts the hot feeder wall. Feeder external surface temperatures are plottsd as a function of time in Figure 11 to show the feeder's behaviour during refill. These thermocouples are located on the inlet feeders about 0.3 m from the Inlet header. A sharp drop in temperature was observed in the 90° side feeders at around 50 s when the temperature dropped steeply towards saturation* Beyond 55 s the feeder temperature approaches the saturation temperature, indicating the flow of a boiling mixture. At around 105 s the temperature drops below saturation which Implies that the 90° side feeders are refilled with subcoolud water. The 45° side feeders showed a
lowei- r a t e of temperature d e c r e a s e . This is believed to be the r e s u l t of the lover surface preheat temperature and the energy removal by the presence of mist flow during the i n i t i a l stage of the depressurization phase.
:-
loo 200
follou this mode); feeder refill from the opposite direction is observed at bank #9. The sign of the &P was consistent with the direction of refill, as indicated in Figure 12 from the 7-densitometer readings. The multi-directional (i.e. the refilling of some feeders from the inlet side and some feeders from the outlet side) refining of feeders, even within the same feeder bank, may be explained in terms of the header behaviour. When a feeder (tor example, on the inlet header side) starts to refill the large amount of steam produced in the feeder quenching process pressurizes the opposite (outlet) header, which induces water inflow through the other feeders In the opposite direction. I- should be noted that the A" measurements in the horizontal feeder sections were not entirely consistent with the turbine flow meter readings. The extended period of flow stratification in these horizontal feeder sections (detected by the top and bottom surface thermocouples) .ray b« responsible.
Time Is 1 FIGURE 11: FEEDER SURFACE TEMPERATURE HISTORIES
= 100-
The single-beam 7-c/ensitometers, iocated* aootrc 0.5 m from the surface thermocouples, provide independent evidence of feeder refill behaviour. Figure 12 shows typical void fractions measured during the refill phas$. The obtained refill times determined from the votd measurement are consistent with those from the surface temperatures. Feeder i-5 Feeder 1-1
75 Inlet
Feeder V-? Feeder 9-1
I
Outlet
1
50 •
25-
•'
' '''
,,i
1 1 1
':
A 7< r 1
.1 1 " '"" [fftl/r :
)'\
0' ?'*•'
0
100 Time (s)
FIGURE 13:
Time (s)
FIGURE 12: FEEDER VOID FRACTION HISTORIES Differential pressure measurements between the headers and across the horizontal sections of the feeders are shown in Figure 13. In the feeder refill phase, the AP across the horizontal feeder section is large. Thi$ is understandable since a large water head is being formed in the refilling feeder, causing a preferential venting of steam through the other feeder. At feeder bank #1 the refill front progresses £rom the inlet feeders towards the outlet feeders (three of four feeders
0
100 T i m e
(s)
HORIZONTAL FEEDER SECTION DIFFERENTIAL PRESSURE HISTORIES
In summary, t h e LASH t e s t facility transient r e f i l l experiments show t h a t the p r o c e s s may be divided i n t o three phases, d ^ p r e s s u r i z a t i o n , header r e f i l l and feeder r e f i l l . The header r e f i l l process was c h a r a c t e r i z e d by n e a r - z e r o header-to-header p r e s s u r e drop and r e f i l l i n g from the injection turret location towards the North and South ends of the headers. The feedei r e f i l l phase showed multidirectional feeder quenches and r e f i l l s . A c o m p a r i s o n ot a selected header experiment with a preliminary simulation CATHENA two-fluid code is presented next.
refill by the
PRELIMINARY CATHENA SIMULATION The CATHENA code was developed a t the W h i t e s h e l l Nuclear Research Establishment (WNRE) p r i m a r i l y for the a n a l y s i s of p o s t u l a t e d l o s s - o f - c o o l a n t a c c i d e n t (LOCA) e v e n t s in CANDU r e a c t o r s . The code uses a staggered-mesh, one-step, semi-implicit, f i n i t e d i f f e r e n c e s o l u t i o n method, which i s not t r a n s i t
CNS 9th ANNUAL CONFERENCE, 1988 233
time-step limited. The comprehensive wall heattransfer package includes radial and circumferential conduction, thermal radiation, and the 2r-H 2 O reaction heat source. The CATHENA code includes c o m p o n e n t m o d e l s required for c o m p l e t e loop simulations, such as pumps, valves, a pressurizer, break discharge, an extensive control system, and a separator model. A more complete description of the code (formerly called ATHENA) can be found in Reference 5. A preliminary simulation of one of the Large Scale Header Test Facility near-zero header-to-header pressure drop experiments has been completed with the two-fluid code CATHENA. The simulation was performed to assess the constitutive relations within CATHENA under the conditions of subcooled water injection into a CANDU-type header/feeder system. This test forms part of the "component" validation program for the CATHENA thermalhydraulics code.
experimental results. This was necessary because a pumped injection system was used and Insufficient detail was available £or practical simulation of the injection network. The pressure in the discharge line downstream of the orifice was assumed to be atmospheric throughout the transient. Finally, all outer surfaces of the test facility were assumed to be adiabatic during the transient.
H T'I j1
UU
I Px ,
Dili JJU c\\
INI I T HlAMHl
[']
\
.turn
(]<.-••
o Test Facility Idealization iilru.i:!' III:AIM.I(
In a p r e l i m i n a r y c o d e a s s e s s m e n t , it w a s considered impractical to simulate the complete Large-Scale Header Test Facility by modelling each of the 30 feeder pipes individually, for a cold water injection transient. The CATHENA idealization chosen for the simulation is shown in Figure 14. In this idealization, the five feeders at each bank location have been lumped into a single "average" feeder. That is, the six banks, each with five individual feeders, have been replaced by six banks, each with one "average" feeder. In the averaging of the feeders the pipe surface area and thermal mass of metal have been conserved. As well, the feeder flow areas and volumes have been conserved. In contrast to the feeders, the inlet and outlet headers have been nodalized in detail. The inlet and outlet headers have each been divided into 10 thermalhydraulic branches with a total of 14 thermalhydraulic nodes and 13 links. The wall of the idealized inlet and outlet headers was divided into six circumferential segments (assuming lateral symmetry). A 2-D heat conduction calculation walls was performed within the header to capture any effect of stratification. The connection between the headers and the "average" feeder thermalhydraulic branches was accomplished through a "separator component" model in the CATHENA code. The model recognizes the upper and lower "edge" elevations of the feeder connections with the headers. These port elevations, together with the flow regime calculated in the header and the flow velocity in the feeder port, determine the liquid f r a c t i o n presented to the feeder port. The "separator component" model also recognizes vapour p u l l - t h r o u g h and liquid-en trainment phenomena through the empirical correlations of Kowalskl and Krishnan [2). Because the five feeders in a bank were simulated as an average feeder, the "lower" and "upper" edge elevations for the header connection were the bottom and the upper edge of the 90° side feeder entrance, respectively. The particular experiment chosen for simulation was a double-break, double-injection experiment with a 50-mm break and an initial loop preheat temperature of 250°C. The cold-water injection lines were nodalized for the simulation only as far as the T-junction just downstream of the check valve in the test facility. The cold-water injection process was modelled through flow boundary conditions taken from the
234 CNS 9th ANNUAL CONFERENCE, 1988
FIGURE 14:
CATHENA LARGE-SCALE HEADER TEST FACILITY IDEALIZATION
The CATHENA simulation's initial wall temperatures throughout the facility were inferred from the experimental results. During purging with steam, some condensation occurs within the header facility. This was apparent from the subcooled temperatures recorded in some feeders, particularly in T/S #3, and at the bottom of both inlet and outlet headers. An a c c u r a t e e s t i m a t e of the facility liquid i n v e n t o r y at the start of the transient is difficult. In this simulation it was assumed that the lower section of the downward-oriented feeders had been filled with saturated condensate. In addition, it was assumed that a small volume of liquid remained in the headers, principally in the blanked-off feeder ports at the start of the transient.
Simulation Results The inlet header pressure calculated by CATHENA is compared vith the experimental measurement in Figure 15. The general shape of the pressure response calculated by CATHENA is in agreement with experiment. H o w e v e r , the simulation's faster depressurization and lower "average" pressure are apparent. The calculation's faster depressurization and lower "average" pressure are the combined result of two modelling approaches in CATHENA: i) the flow regime map calculated within the header implied that the flow was stratified throughout the transient. During a depressurization, a more homogeneous flow pattern would be expected, allowing more flashing which would hold the pressure level up longer. 2) Subcooled boiling was not available for this calculation because of modelling changes being made to the heat transfer package in the present version of CATHENA. The lack of a subcooled boiling model results in the lower "average" pressure because condensation on the cold injection water is allowed to control the pressure. The small pressure spike calculated by CATHENA at approximately 50 s is an artifact of the discharge model. A two-phase
mixture of saturated steam and subcooled liquid Leached the discharge model location at that time in the simulation. The discharge model used assumes homogeneous flow a n d the s i m u l a t i o n ' s n o n equilibrium state caused the inconsistency. During the earlier part of the transient, the discharge model performed adequately since only high quality steam reaches the break.
the inlet and outlet sides (feeders 1 and 2 are the longest of the feeders in a bank). The trends shown in the CATHBNA simulation of an average channel per bank are in good agreement with experiment. The results also indicate that the phenomena of vapour pull-through and liquid entrainment are unimportant for this experiment since both headers are nearly liquid filled before the feeder-refilling process begins.
i §
0.75 H
M £
0.50 "
a
FIGURE 17: FIGURE 15:
INLET HEADER PRESSURE COMPARISON
The c a l c u l a t e d d i f f e r e n t i a l p r e s s u r e between t h e inlet and o u t l e t headers i s compared with e x p e r i m e n t a l d a t a in F i g u r e 16. The t r e n d s and m a g n i t u d e s c a l c u l a t e d by CATHENA a r e i n good agreement with t h e e x p e r i m e n t . During t h e i n i t i a l p e r i o d of header r e f i l l ( u n t i l about 50 s ) , t h e header-to-header pressure drop i s small and r e l a t i v e l y smooth. This time period has been identified as the period of header r e f i l l from the analysis of the experiments. In the later part of the experiment, between 50 and 110 s, the header-toheader pressure drop has large oscillations. This time period has been identified with the refilling of the feeder banks. Finally the header-to-header pressure drop becomes quiescent after the feeders are filled.
E>P«1mcnt
30
CATKENA 20 ID
E
0
d)k
A .jail
-10 -20
Inlet Feeder Outlet Feedci Expt.
* Inlet F outlet
A
AVERAGE VOID FRACTION COMPARISON FOR FEEDER BANK #1
0.50 -
o.:; o.o 100
FIGURE 18:
150
AVERAGE VOID FRACTION COMPARISON FOR FEEDER BANK #9
In summary, the trends predicted by CATIIENA in a preliminary siir.jlation of a double-break, doubleinjection test in the Large-Scale Header Facility are in a g r e e m e n t with e x p e r i m e n t a l r e s u l t s . However, It is clear that the inclusion of subcooled void g e n e r a t i o n is required to improved the simulation accuracy. Subcooled boiling should be included in the simulation before extending the idealization to the modelling of individual feeders.
= 1 CONCLUSIONS
FIGURE 16:
HEADER-TO-HEADER DIFFERENTIAL PRESSURE COMPARISON
The calculated inlet and outlet void fractions near the top of the "average" feeder channels for bank #1 and bank #9 are shown in Figures 17 and 18. These calculated void fractions Indicate that in the simulation bank #1 fills from the inlet side while bank #9 fills from the outlet side. In Figures 17 and 18 the average of the experimental measurements of void fraction for feeders 1 and 2 are shown for
It is clear from the transient refill experiments conducted under near-zero header-to-header pressure drop conditions in the LASH test facility that the injection flow rates and header preheat temperature have the largest influence on header refill times. The break size and location have less influence and appear to change refill times through their effect on the injection flow rate. The CATHENA simulation of a test shows qualitative agreement with the experimental results. However, some features that require enhancement for improved quantitative agreement are a move representative flow regime map for transient flow in a header, and
CNS 9th ANNUAL CONFEHENCE, 1988 235
the inclusion of subcooled boiling for more accurate simulation of the pressure response of a header during the injection of cold vater.
(3)
DEMAN, H.G.O., FEYCINBERG, V. and HIDVIDY, U . I . , "Steady-Slate Tvo-Pliase Header Flow Model", paper presented at (lie l
(41
GULSHANI, Stratified sented at Symposium,
[5]
RICHARDS, D.J., 1IANNA, B.N., IIOBSON, N. and ARDRON, K.H., "ATHENA: A Tvo-Fluld Code for C A N D U LOCh Analysis", P r e s e n t e d at 3rd International ConteL-er.ee on Reactor ThennalIiydiaulics, Oct. 15-18, Newport, RD, 1985.
ACKNOWLEDGEMENTS The experimental program described the CANDU Owners Group (COG).
was
funded by
REFERENCES (!)
KOUALSKI, J.E. and KRISHNAN, V.S., "Experimental Investigation of the Refill Behaviour in a CANDU-Type Header/Feeder System", proceedings of the 8th CNS Conference, St. John, NB, June 1987.
[2)
KOWALSKI, J.E. and KRISHNAN, V.S., "Two-Phase Distribution in a Large Manifold", presented at the 1987 AIChE Annual Meeting, New York, NY, November 1987.
236 CNS 9th ANNUAL CONFERENCE, 1988
P., "M1SSFIMCII: Models for Steady Flow in CANDU Headers", paper preibe J4lh Annual. Nuclear Simulation l'inawa, MU, April 19UU.
MIXiNC, IN A VUSSr.L-Pll'i; ASSI-MBLY A.II.T.
I,am, M. U n g u r i a n , K.N. T e n n a n k o r e A t o m i c I i n e r g y o f Canada L t d . W h i t o s h c l l N u c l e a r R e s e a r c h l i s t a b l i shmcnt
(Not
Available
at
Time
of
Printing)
CNS 9th ANNUAL CONFERENCE. 1988 237
COMBUSTION BEHAVIOUR IN THE MODERATOR COVER GAS
G.W. KOROLL, R.K. KUMAR and C.K. CHAN High-Temperature Chemistry Branch Atomic Energy of Canada Limited Pinawa, Manitoba ROE 1LO
ABSTRACT A combustible mixture o£ deuterium, oxygen, helium and steam may form in the cover gas space during postulated events such as an accidental loss of moderator. In order to analyse the likelihood of ignition and the potential consequences of combustion, should ignition occur, we need basic combustion data which we presently do not have. We have been doing experiments in our laboratory to determine some of the important combustion parameters such as flammability limits, ignition temperatures, burning velocities, detonation cell widths, transition distances and transition limits relating to the unique environment of the moderator cover-gas space of CANDU reactors. This paper describes some of the significant findings.
INTRODUCTION The presence of deuterium in the moderator cover gas and the concomitant combustion hazard was an early consideration in the safe design and operation of CANDU reactors (1). In the 1960's and 1970's, the research effort was aimed at understanding the radiation chemistry of the moderator (2). Today, deuterium transients during stprt-up and normal operation are better understood (3,4). Recent interest in the cover gas has come about from analyses of postulated accidents, such as a loss of moderator (5). During a loss-of-rooderator accident, it is envisaged that the decreasing cover-gas pressure may lead to a prompt release from the moderator of dissolved D 2 and 0 2 produced by radiolysis. If the release rate exceeds the capacity of the recombiners, a combustible mixture could form in the cover-gas space. Exposed adjuster rods, deprived of cooling, offer a potential ignition source. The increased cover-gas volume and the decrease in moderator level escalate the potential consequences of a combustion event, should ignition occur.
extraordinary, but sparsely researched, catalytic effect on combustion behaviour (6). Vith regard to internal geometry, exposed pressure tubes or guide tubes in the calandria offer arrays of repeated obstacles that can contribute to flame acceleration, possible transition to detonation and complex pressure loads. The complex geometry precludes simple calculations to predict pressure transients. The present research effort is broadly based, addressing several different aspects of combustion behaviour such as ignition, deflagration, detonation and transition to detonation. The scope of the individual research projects varies in technical and analytical rigor as appropriate to meet the aim of providing verified tools to predict combustion behaviour in the cover gas. The results obtained to date are presented and discussed in terms of flammability limits, ignition, burning velocities, detonability and cell sizes, and t'.e deflagration to detonation transition (DDT). Each pertinent experiment is described briefly and the sensitivity ot the phenomenon to the unique aspects of the cover-gas environment is discussed-
FLAMMABIL1TY
LIMITS
Flammability limits define the range of mixture compositions in which a flame can propagate. Mixtures of hydrogen and air are known to be flammable between U and 75X hydrogen by volume, the so-called lean and rich limits of f laminabili ty. Limits of flammability for deuterium have been determined experimentally to be narrower than those for hydrogen; lean and rich li..:its for deuterium in air are 5% and 75%, respectively (1,7). The flammability limit is about 9% hydrogen for scoichiometric hydrogen-oxygen-helium mixtures (8). Addition of steam to this mixture decreases the flammable range slightly.
IGNITION An.-lysts are assessing the likelihood of ignition by the adjuster rods and the nature and consequences of potential combustion. An integrated combustion research program is under way at the Vhiteshell Nuclear Research Establishment (UNRE) in support of these analyses. This paper describes that program and reports its significant findings to date. From a combustion standpoint, the cover gas is a unique environment vith respect to gas compositions and internal geometries. The gases that make up the cover gas are D a , C , helium and D 2 0 vapour. The key combustion properties of deuterium, such as ignition temperature, burning velocity, and detonation cell width, are unavailable in the literature-. Further, available data is limited for the effects of the helium diluent which, owing to its high diffusivity and low heat capacity, contributes to unusual combustion behaviour. Water vapour has an
238 CNS 9th ANNUAL CONFERENCE. 19B8
Ignition by a heated surface, such as the adjuster rods in the cover gas, is considered to be primarily thermal ignition. The spontaneous thermal ignition of a volume of combustible gas, so-called autolgnition, has been studied extensively and can be predicted theoretically by classical combustion theory. It is governed by the balance between the rates of the temperature-sensitive chain-branching reactions (i.e., H+D 2 •» OH+0) and those of the pressure-sensitive chain-breaking reactions (i.e., H+Oj+M -» HO 2 +M). At atmosphere pressure, the minimum auto-ignition temperature for H2-air mixtures is about 570°C, hovever, much higher temperatures are required of heated wires or rods to achieve ignition. Experimental measurements indicate that surface ignition temperatures of moist H2-air mixtures range from 65O°C to 7B0°C for 4-cmJ glow plugs (9), depending on composition. Surface ignition
temperatures are observed to approach the classical autoignition temperature as the heated surface area is increased. Thus, geometric considerations are critical. For the present purpose of demonstrating the safety margins oi ignition by the control roos in the cover gas, the geometry is known. Therefore, the central experiment in our program on ignition is an integrated experiment to simulate this geometry.
Large-Scale CTF Experiments An instrumented, electrically-heated full-scale simulator of an adjuster rod assembly was immersed in gases simulating the cover gas atmosphere in the 6.5 m 3 Containment Test facility (CTF) sphere at WNRE (see Fig. 1 ) . The temperature o£ the rods was increased until ignition occurred. The temperature at ignition was recorded for a matrix of compositions. Figure 2 summarizes the results for mixtures relevant to the cover gas atmosphere. Here, steam appears to inhibit ignition to a significant extent. This is due primarily to the high third-body efficiency of steam (6) in the chain-breaking
--
"
"
—
* M.
FIGURE 2: SURFACE TEMPERATURES AT IGNITION OF STOICHIOMETRIC H 2 -O 2 MIXTURES WITH HELIUM, STEAM AND C0 2 DILUENTS IN THE 6 mJ CTF SPHERE. CONTROL ROD ASSEMBLY SIMULATOR IN 'UP' POSITION
To date we have focused on two series of separateeffect tests on surface ignition behaviour as they relate to the moderator cover gas, deuterium substitution and surface recombination.
K - Kulice Pressure Transducer. P - Rosemount Pressure Transmitter P Z - Piezoelectric Pressure Transducer R T D - Resistance Temperature Detector
FIGURE 1: A SCHEMATIC OF THE 6.3 m 3 CONTAINMENT TEST FACILITY (CTF) AND INSTRUMENTATION
Separate Effect Tests It is impractical to simulate exactly all the relevant conditions of the cover gas environment in the large-scale CTF sphere. A versatile small-scale experimental assembly (see Fig. 3) was commissioned to study separate effects that are independent of geometry, and are impractical to study on a large scale or require large numbers of tests to elucidate. For example, large-scale combustion of deuterium is not practical but the isotope effect on ignition behaviour is not expected to depend on geometry. Likewise, we are unable [0 simulate on a large scale the intense radiation field and the resulting steadystate concentrations of gas-phase radicals produced by water vapour radiolysis.
Deuterium Substitution. Surface temperatures required to ignite mixtures of deuterium-oxygen-helium were measured in the small-scale experimental assembly, and are compared in Figure 4 with those for equivalent mixtures containing hydrogen. Mixtures containing deuterium ignite at surface temperatures 20-30°C lower than those for corresponding mixtures containing hydrogen. The result is counterintuitive, since deuterium is less reactive than hydrogen. Two other factors appear to override the isotopic difference in reactivity. First, owing to its greater mass, and hence lower collision frequency, deuterium is a less effective third body in chain-breaking reactions. Second, due to the lower thermal dittusivity o£ deuterium, heat is dissipated less readily from the near-surface layers. Both contribute to ignition at a lower surface temperature and offset the effect of lower reactivity of deuterium mixtures. Surface Recombination. The recombination of H 2 and 02 into H 2 0 at surface temperatures below the critical ignition temperature affects the mixture composition in the vicinity of the igniting surface. Locally, it reduces the proportion of combustible fuel while simultaneously enriching the near-surface gases with water vapour. Thus, surface recombination is likely to influence ignition temperatures. However, the real importance of surface recombination lies in its potential for mitigation of the combustion hazard by the consumption of fuel, without runaway to ignition. It is this aspect that was addressed in our tests. Stoichiometric mixtures of H 2 -O 2 in 702 He wnre exposed to the heated, 6-mm diameter, vertical, steel rod in the 2-L vessel and the depletion of H: and 0 5 was monitored. The surface temperature of the vod was kept constant throughout the test. The vessel walls were cooled to maintain a constant bulk-gas temperature. The results are summarized in Figure 5. The rate of depletion of fuel increased with higher surface temperatures. The extent and time scale of C N S 9th A N N U A L C O N F E R E N C E , 1988 239
recombination appear: to depend on the gas composition, igniter geometry and the history of the heated surface. We are continuing to investigate this effect.
BURNING VELOCITIES Laminar burning velocity is a fundamental physicochemical property that many computer codes require as an input parameter to determine the pressure and temperature transients resulting from combustion. It depends only on the thermodynamic state of the mixture, and is defined as the normal velocity of the flame relative to the unburnt gas. Burning velocity data for mixtures containing deuterium are unavailable, and data for hydrogen-oxygen mixtures with helium and/or steam dilution are sparse. In the work reported here, burning velocities were measured by the nozzle burner/schlieren cone angle method with particle-tracking of the unburnt gas velocity by laser-Doppler anemometry (6). Data obtained by this method were in close agreement with recently published data for hydrogen-air mixtures.
Burning Velocity of Deuterium
FIGURE 3: SCHEMATIC DRAWING OF THE 2 L INSTRUMENTED VESSEL FOR SEPARATE EFFECTS IGNITION EXPERIMENTS
Deuterium burns more slowly than hydrogen (see Figure 6 ) . Since the burning velocity varies with the square root of the thermal diffusivity of the mixture and the chemical rate of oxidation, and since the thermal diffusivity and the reaction rates of deuterium mixtures are lower than those for hydrogen mixtures, substitution of deuterium for hydrogen lowers the burning velocity significantly. The burning velocities of deuterium mixtures can be correlated with those of hydrogen mixtures by the expression
1/2
e
where S and a are the burning velocities and thermal diffusivities of the respective mixtures, and 0.85 is the square root of the average ratio of collision frequencies of deuterium and hydrogen with oxygen. This correlation predicts the burning velocities of deuterium mixtures within 52£ over a broad range of mixture compositions with burning velocities from 1 m/s to 12 m/s.
120 00
20 0
40 C
SCO
SOO
lOCC
% Helium in mixture
FIGURE A: SURFACE TEMPERATURES REQUIRED TO IGNITE STOICHIOMETRIC MIXTURES OF D 2 -O 2 -He AND H 2 -0 2 -He IN THE 2 L VESSEL. INITIAL TEMPERATURE 70°C. INITIAL PRESSURE 101.3 kPa.
n
100
c Hydrogen * Deuterium. o
'-.
i 3
3
"
\
0
8 0 o
6 0
Jl
o /
\
^
o
o
\\
4.0
:
0 2.0
7,,
•••"
^ " \
I
'T
1
J
J
1
1
1
O.I
0.2
0.3
0.4
0.5
0.6
0.7
0 6
0.9
Mole fraction hydrogen (deuterium) FIGURE 6: BURNING VELOCITIES OF H 2 - 0 j AND D^-0^ MIXTURES AT 25°C AND 1 0 1 . 3 kPa.
FIGURE 5 : HYDROGEN CONSUMPTION DUE TO RECOMBINATION AT A 7 cm ! HEATED STEEL SURFACE IN THE 2 L VESSEL. MIXTURE COMPOSITION, 0 . 2 H ; + 0 . 1 0 , + 0 . 7 He. INITIAL PRESSURE, 1 0 1 . 3 kPa. 240
CNS 9th ANNUAL CONFERENCE, 1988
It is interesting to note that there is no apparent contribution to the isotope effect on the burning velocity from the isotopic difference in bond energy. This supports the premise that the ratelimiting steps of the H 2 -O 2 reaction mechanism Involve atomic species (t-e.-, H + 0, -» QH. + Q arid H * 02 + M -> H0 2 + M) and not molecular forms of the isotopes (i.e., H 2 + 0 -» OH + H ) .
ii a
-—r—
1
-—I "I ' ~ i "" i Mcliiii n/sto.iiin ratio
\?. v —. IO O
I 80
• •
00 2.1
r>
0.9
(pure helium)
ii 0.!i " O.U
(pure steam)
f IJ
o7
>•
•
Effect of Helium and Steam Diluents Inert diluents influence the burning velocity to the extent that they affect the thermal diffusivity of the mixture or the heat capacity of the mixture (which ultimately affects flame temperature and thus the rates of reaction steps with high activation energy). Simple diluents lower the burning velocity more or less in proportion to the amount of diluent added until, at some high diluent fraction, the mixture becomes inert. The predictable effect of thermal diffusivity and the approach to the limit of flammability with dilution were exploited in a correlation (6)
C23
0 1
u 2
u 3
0')
U '•>
u f3
U9
Volume liaclion of diluent (He + steam) FIGURE 7: BURNING VELOCITIES OF STOICHIOHETRIC H 2 -O 2 MIXTURES WITH HELIUM/STEAM DILUTION. INITIAL TEMPERATURE 25°C, PRESSURE 103.1 kPa. detonation can propagate (3X) (12), the critical tube diameter, d c , for a successful transition of a planar detonation wave in an uncortfined volume (dc =13X) (13), and detonation limits.
where S o and a o are the respective burning velocities and thermal diffusivities of the undiluted mixture, J is the mole fraction of diluent and X L is the mole fraction of diluent required to inert the mixture. This correlation predicts the burning velocity of the H 2 -O 2 -diluent mixtures reasonably well over the entire range of hydrogen concentrations. In this respect, steam is extraordinary. Steam has a high third-body efficiency in the reaction, H + 0 2 > M -> HOj + M + heat, and increases the heat release in the preheat zone of the flame. This mechanism counteracts the effect of the decrease in flame temperature with steam to produce a burning velocity about 202 higher than the prediction of Eq. (2) for ordinary diluents. Helium, because of its low heat capacity and high thermal diffusivity, lowers the burning velocity least relative to other diluents. Since helium and steam are both somewhat unique, a database of burning velocities was obtained far a matrix of compositions of mixtures containing both. These are presented in Figure 7. Increasing the steam content in the He/steam mixture predictably lowers the burning velocity, but not as much as would be expected on the basis of thermal flame theory. Increasing the helium content of the He/steam mixture increases the burning velocity due to the increase in the flame temperature and the thermal diffusivity of the mixture.
DETONABILITY AND DETONATION CELL WIDTHS A detonation cell is a trajectory of the triple points formed by the leading shock, the Mach stem and the transverse shock of a detonation wave. Many of the detonation characteristics of a fuel-air mixture are related to the cell width. To cite a few, the detonation cell width W determines the detonation sensitivity of the mixture (- 1/X) as shown by Matsui and Lee (10), the energy required for the direct initiation (E ~ X") of spherical detonations (11), the minimum orifice diameter through which a detonation can propagate (3X) (12), the critical tube
The detonation test facility used in our program (14) consists of a 15-cm internal diameter, 7.5-m long, heated, stainless steel pipe, connected through a valve to a 2-m long, 2.5'Cm internal diameter initiator. A short length of spiral was located in the 2.5-cm diameter pipe to ensure that a fully developed detonation wave emerged from the smaller pipe into the bigger pipe. The cell imprints were obtained on a Mylar foil using the smoke-foil technique. The experiments reported here were carried out with stoichiometric hydrogen-oxygen mixtures with helium, argon, steam and carbon dioxide as the added diluents. The measured cell widths ranged from 1 mm to over 70 mm. Because of a considerable variation in the cell width from cell to cell in any given experitueat, an average cwec a. Large, micatiet at cells was determined. Figure 8 shows the variation of cell width as a function of hydrogen concentration for mixtures containing stoichiometric amounts of hydrogen and oxygen. The cell width increases as the hydrogen concentration decreases. Detonation occurred for H-~ O a -He mixtures down to 9.4X hydrogen. Experiments carried out in our laboratory (8) showed that the flammability limit of H 2 -O 2 -He mixtures at room temperature (22°C) and 106.6 kPa was around 9% hydrogen for both upward ami downward propagation of the flame in a 10-cm diameter, 1.8-m long, steel tube. Thus, it appears that detonation is possible down to flammability limits for dry H 2 -O 2 - He mixtures (and also H 2 -0 ; -Ar£on mixtures). Figure B also shows cell widths for mixtures with steam and carbon dioxide as added diluents, plotted as a function of hydrogen concentration. The presence of 20X steam, corresponding to the normal condition of the cover gas, significantly reduces the detonation sensitivity and 3ppears to narrow the detonable range. Cell widths for deuterium mixtures and hydrogen mixtures are compared in Figure 9. Cell sizes for
CNS 9th ANNUAL CONFERENCE, 1988
TRANSITION FROM DEFLAGKATION TO DETONATION If a combustible gas mixture is consumed by a slow deflagration, the maximum overpressure attainable is the constant-volume combustion pressure, which is roughly 6 times the initial pressure foe H2-O2 mixtures. However, if a transition to detonation occurs during the combustion process, shock waves with an amplitude of about 20 times the initial pressure of the mixture can be generated. Presently, the transition process is not veil understood. It is not possible to predict whether transition to detonation will occur under a given set of conditions (configuration of flow-obstructing structures and initial gas-mixture composition).
10
20
30
40
50
60
Hydrogen concentration {%)
FIGURE 8 : DETONATION CELL WIDTHS OF H ; - 0 2 - H e (SWICSIOMETRIC) MIXTURES WTH DIFFERENT STEAM AND
CO, CONTENT 100 80 60 40
.
Initial pressure 106.6 kPa
Transition Limits
TO = 22°C
Upon ignition by a glow plug the flame accelerated rapidly in the obstacle-filled duct and eventually became a detonation when transition occurred. Figure 10 shows the flame speed, averaged over 1.29 m near the end of the duct, for various stoichiometric H 2 -O 2 -He mixtures. In the range of mixtures containing 10Z to 19£ H2, all measured velocities were above 2500 m/s, indicating that transition to detonation had occurred. The Chapman-Jouguet (CJ) detonation velocities of these mixtures are also shown in Figure 10 for comparison. In tubes filled with obstacles, the detonation velocity is usually slightly less than the corresponding CJ values due to
H2 [or D2J - Oj - He system
10
I
8
S
6
4
20
As part of a program at VNRE to study transition from deflagration to detonation, experiments have been conducted with various H 2 -O 2 -He mixtures at atmospheric pressures in a 6.6 m-long, 28 cm diameter combustion duct. A set of obstacles was located in this duct to induce flame acceleration. These obstacles were spaced 28 cm apart and their blockage ratio (blocked area/total cross-sectional area of the duct) was 0.31. Six piezoelectric pressure transducers were mounted at various locations in the duct: to monitor the flame speed and the pressure development to determine transition limits and transition distances.
30
40
Fuel concentration (vol. %) FIGURE 9: DETONATION CELL WIDTHS OF DRY STOICHIOMETRIC H 2 -O 2 AND D 2 -O 2 MIXTURES WITH HELIUM DILUENT.
deuterium mixtures are larger, indicating less detonation sensitivity compared to hydrogen mixtures. The cell size is proportional to the post-shock induction time, which is determined by the rate of chemical reaction. If the difference in H 2 and D 2 rates of oxidation is assumed to arise through the isotopic difference in their collision frequencies with O j , then the isotopic ratio of cell sizes, Xp/X^, should equal J2. This is confirmed by our experiments.
tins, FIGURE 1 0 : MEASURED FLAME SPEEDS OF H , - 0 , - H e MIXTURES ACCELERATED BV REPEATED OBSTACLES, INDICATING QUASI-DETONATION
242 CNS 9th ANNUAL CONFERENCE, 1988
momentum losses and the combustion process is termed quasi-detonation (15) since the magnitude of the pressure wave is very similar to that of a CJ detonation (16). Ignition could not be achieved in these tests for mixtures containing less than 10% H2 since a glow plug was used as an Igniter. The upward and downward flammability limits £or these mixtures are 9% H 2 . These results suggest in agreement with our findings in the 15-cm diameter tube, that transition to detonation is possible for all flammable H2-O2-He mixtures.
SUMMARY The following summaries the significant findings on combustion behaviour in the moderator cover gas. O)
DeuSeriiM ignites more readily than hydrogen, but once ignited, burns more slowly and i." less likely to detonate.
(2) The presence of helium in the mixture narrows flammability limits, but. decreases ignition temperatures and increases detonation sensitivity. (3)
Transition Distances Transition distance, the distance between the point of ignition and the point of onset of detonation, was also measured in these experiments. It should be pointed out that transition distance is not an intrinsic property of the gas mixture. It depends strongly on the boundary conditions. However, transition distance is a useful parameter for comparing the relative level of hazard of different mixtures. As expected, the transition distance, J>T, increases rapidly as the mixture sensitivity decreases. For the 19% H 2 mixture, D T was measured to be about 2 m. Similar experiments with various H2-air mixtures were performed in the same apparatus using the same set of obstacles. It was found that the transition distances for H2-O2-He and H2-air mixtures could be correlated quite well with their laminar burning velocities. Figure 11 shows the relationship between DT and the laminar burning velocity S L . A single expression, D T = 3.7 x S L -°- 81 , seems to describe the results reasonably well. The two coefficients in this expression are expected to depend on both the boundary conditions such as the obstacle configuration and the properties of the gas mixture.
The presence of steam in the mixture mitigates combustion in that it narrows the flammability limits, increases th§ ignition temperatures, reduces the burning Velocity of the mixture and increases the detonation cell sizes (i.e., reduced detonation sensitivity).
(4) Transition distances correlate with mixture sensitivity as characterized by laminar burning velocity. ACKNOWLEDGMENTS Financial support from the CANDU Owners Group (COG) is gratefully acknowledged. REFERENCES (1)
HATCHER, S.R., "Estimated Gas Concentrations in the CAWDU Moderator System," Atomic Energy of Canada Limited Report, CEI-145 (1962).
(2)
H2AD, D.A. and SINGH, A., unpublished repott.
(3)
VAN BERLO, J.P., unpublished report.
(4)
VAN BERLO, J.P., unpublished report.
(5)
FUNG, K.K., unpublished report.
(6) KOROLL G.V. and MULPURU, S.R., "The Effect of Dilution with Steam on the Burning Velocity and Structure of Premixed Hydrogen Flames," Twentyfirst (International) Synposium on Combustion. The Combustion Institute 1986, pp. 1811-1819.
FIGURE 11: TRANSITION DISTANCES IN H2-O2-He AND H2AIR MIXTURES AND THEIR CORRELATION l/ITH THE LAMINAR BURNING VELOCITIES, S , OF THE CORRESPONDING MIXTURE The specific conditions leading to transition, such as local turbulence intensity and the leading shock strength of the accelerated flame, are not well understood. These are currently being determined.
(7)
PAYMAN, W. and TITMAN, H., "The Limits of Inflammability of Hydrogen and Deuterium in Oxygen and in Air," Nature 1J7, 1990 (1936).
(8)
KUMAR, R.K., "Flammability Limits of HydrogenOxygen-Diluent Mixtures", J. Fire Sci., 3, 245262 (1985), AECL-8890.
(9) TAMM, H. et al., "Effectiveness of Thermal Ignition Devices in Lean H2-O2-Steam Mixtures", EPRI NP-2956, RP 1932-14 (1985). (10) MATSUI, H. and LEE, J.ft. , "On the Measure of the Relative Detonation Hazard:; of Gaseous FuelOxygen and Air Mixtures," Seventeenth Symposium (Int.) on Combusti-w, p. 1269 (1978). (It) LEE, J.H.r , OTrSTAuTAS R. a nd GUIRAO, C., "The Link Between Cell Size, Critical Tube Diairstsr, Initiation Energy and De'.onability Limits," proceedings of ihc International Conference on Fuel-Air Explosions, McGill University, Montreal, Canada, p. 157 (1982). CNS 9th ANNUAL CONFERENCE. 1988 243
(12) THIBAULT, P., LIU, Y.K., CHAN, C , LEE, J.H. KNYSTAUTAS R. and GUIRAO, C , "Transmission of an Explosion Through an Orifice," Nineteenth Symposium (Int.) on Combustion, p.599 (1982). (13) fc^TROFANOV, V.V. and SOLOUKHIN, R.I., Soviet Physics - Doklady,Vol. 9, No. 12, p. 1055 (1965). (14) KUMAR, R.K., "Detonation Cell Sizes o£ H2-O2Diluent Mixtures," to be published in Comb, and Flame, (15) LEE, J.H., KNYSTAUTAS, R. and CHAN, C., "Turbulent Flame Propagation in Obstacle-Filled Tubes," proceedings of the 20th Symposium (International) on Combustion. The Combustion Institute, p. 1663 (1984). (16) CHAN, C. , and GREIG, iJ.R., "The Structures o£ Fast Deflagration and Quasi-Detonation", to be published in the 22nd Symposium (International) on Combustion (1988).
244 C N S 9th ANNUAL CONFERENCE, 1988
AEROSOL MATEKIAL RELEASES FROM A ZIRCALOY-4 CLAD UC12 PELLET AT TEMPERATURES UP TO 2000°C IN A FLOWING ARGON ATMOSPHERE
S.S. MULPUBU, F.B. BANKS and M.D. PELLOW
Atomic Energy of Canada Limited Uhitesbell Nuclear Research Establishment Pinawa, Manitoba ROE 1LO
ABSTRACT During some postulated loss of coolant and loss of emergency coolant injection accidents, vapours of fission products and structural (fuel and cladding) materials may be released into steam-hydrogen mixtures flowing in a CANDU fuel channel. These vapours will condense into aerosol particles in the cooler parts of the primacy heat transport system. As part of an on-going program to study aerosol formation, transport, deposition and associated Cission product retention in the PHTS, experiments were conducted to measure aerosol mass release rates from a Zircaloy-4 clad, U 0 2 pellet, inductively heated to temperatures up to 2000°C, in a forced-flow argon environment. A description ot these experiments and the obtained results, including fractional mass release rates of structural materials, are presented in this paper.
INTRODUCTION During Some postulated loss-of-coolant (LOCA) and loss-of-emergency-coolant-inject ion (LOECI) accidents, the temperatures in a CANDU reactor fuel channel may rise high enough to cause release of vapours of fission products and structural (fuel and cladding) materials into flowing steam-hydrogen mixtures. These vapours will condense in the cooler parts of the primary heat transport system (PHTS). Because the structural materials are relatively involatile, their vapours will condense readily into aerosol particles. These particles, in turn, provide sites for the condensation of the more volatile fission products. The aerosol transport of fission products in the primary heat transport system will thus be influenced by the structural material release rates. As part of an ongoing program to develop predictive tools for aerosol and associated fission product transport through the PUTS, experiments were conducted to measure mas-., release rates of chemical elements from a Zircaloy-4 clad-UO2 pellet at temperatures up to 2000°C in a forced-flow argon atmosphere. The release rates were measured at different temperatures, rates of temperature increase (ramp rates) and argon flow rates. Release rate constants for different chemical elements were calculated by fitting a simple, first-order differential equation that allows time variation of temperature to the data. The release rate data are useful for aerosol source-term model development.
EXPERIMENTAL ASSEMBLIES AND MEASUREMENTS Release rates of LWR core materials at high temperatures were measured by Albrecht et al. in SASCHA facility [1], There, a mixture of U0 z (35X by weight), Zircaloy-4 (102) and stainless steel (55%) was inductively heated in a crucible to temperatures
in excess of 2500°C in an argon/air/steam atmosphere. Natural convect.wa flow existed around the heated mass. The SASCHA data are not directly applicable to releases in CANDU channels for two reasons. First, there is no stainless steel in the CANDU channel, and second, forced (not natural) convective flow is postulated to exist around the fuel elements in a CANDU sub-channel. We have, therefore, avoided the crucible heating method. Horizontal Assembly The experimental assembly was designed to represent the geometry and flow configuration in a CANDU sub-ch.am\eL. In initial scoping tests, a Zircaloy-U clad, unirradiated U0 2 pellet (the sample) of approximately the same size as the CANDU fuel elements (see Figure 1) was placed in the centre of a quartz tube, using a zirconia support disc and a rod. The orientation of both the sample and the the quartz tube was horizontal. An argon flow was maintained in the quartz tube at a specified rate. The sample was heated inductively to a high temperature, within the field coil of an RF generator (LEPEL, 20 kU, 160 to 450 kHz). The surface temperature of the sample was measured by a MODLINE R series, two-colour pyrometer made by IRCON Inc.
PYROMETER
ARGON
1 QUARTZ TUBE
FrGURE 1.
FILO ziRCONia SUPPORT DISC A Schematic of the Experimental Assembly
Horizontal
A serious problem in reading the sample temperature was encountered with this experimental assembly. At temperatures in excess of 1750°C, blackgrey aerosol material had deposited on the inner surface of the quartz tube above the sample, Tapviev of the aerosol-clouded quartz tube is shown in Figure 2. Since the pyrometer was viewing the sample from above, these deposits interfered with the line of sight and, as a result, temperature could not be measured accurately. Further, the post-test chemical analysis revealed that the deposited material had reacted with the quartz and could not be totally extricated for its mass measurement. Attempts to C N S 9th A N N U A L C O N F E R E N C E . 1988 245
By manually controlling power to the field-coil, we heated the sample from room temperature to 1500°C at a slow rate, to avoid cracking o£ U0 2 as a result of thermal stress, then at a constant linear rate till the desired maximum temperature was reached. The maximum temperature was maintained for a specified period. The temperature of the sample was then brought down abruptly to values below 1500°C by cutting off power to the field-coil. Specimen temperature-time profiles obtained using the pyrometer are shown in Figure 4. Oscillations in the temperature can be noticed above 1850°C. These are caused by molten Zircaloy-U02 interactions. The melting point of Zircaloy-4 varies from 1760 to 1950°C, depending on its oxygen concentration. The appearance of a sample that was heated to 2000°C, at a rate of 0.5°C/s, is shown in Figure 5. The nonuniform sample surface resulting from melting and relocation of Zircaloy-4 can be seen in the picture.
FIGURE 2.
Aerosol Material Deposit on the Inner Surface of the Quartz Tube. The Pyrometer was Sighted on the Sample Situated in the Middle of the Quartz Tube Below the Deposit.
remedy the situation by increasing the flow rate and using tubes of larger diameter were not successful. The orientation of the tube was hence changed from horizontal to vertical. In the vertical orientation, no deposits were observed in the vicinity of the sample, the pyrometer line of sight remained clear, and the temperature measurement was reliable. The reason that the aerosol material did not deposit in the vertical orientation was that the buoyancydriven, upward, natural convective flow of the aerosol vapours was in the same direction as the forced convective flow of argon.
TIME
2000 C
Vertical Assembly A schematic of the vertical experimental assembly is shown in Figure 3. The data reported here were obtained in this assembly. The sample was placed on a zirconia base in such a way that it was concentric with the quartz tube. An argon flow was maintained in the quartz tube at a constant rate. The pressure in the system was slightly above 101 kPa (one atmosphere).
TIME
FIGURE 4.
Temperature-Time Profiles Recorded from the Pyrometer Reading. Figure 4-A Shows the Temperature Increase from 1500 to 2000°C, at a Rate of 2°C/s. Figure 4-B Shows the Same Increase followed by a Two Minute Constant Temperature Period.
HLTES ASSEMBLY
SPECIMEN Zr-UO 2 ^-i\IS
OF PYROMETER
ZIRCONfd 9dS£ TUBE QUARTZ
Hot vapours that emanated from the sample were carried by the flowing argon to the top portion of the tube. Here, the gas temperatures were low enough <100°C) for the metallic vapours to nucleate into aerosol particles on their way to the top of the tube, and hence they could exist only in the solid phase. These aerosol particles were trapped in a filter located at the top of the quartz tube. The aerosol mass and its chemical composition were determined by post-test chemical analysis. Filter Assembly
FIGURE 3.
A Schematic of the Vertical Experimental Assembly
246 CNS 9th ANNUAL CONFERENCE, 19B8
A special filter holder was designed, built and fitted to the top of the quartz tube to collect the aerosol particles with minimal losses. A schematic of the holder is shown in Figure 6. In the fitted position, the filter could be held on fop of the
quartz tube so that the aerosol laden gas passed directly through the filter as it exited the tube. Use of commercially available filter holders would have required channeling of the aerosol-laden gas into the holder through a narrow passage of about an order of magnitude smaller diameter than that of the quartz tube and would have resulted in a loss of some particles in the passage. The filter was made of quartz fibres and was rated to withstand temperatures up to 900°C. The collection efficiency of a fibre filter var ies with the particle size. It is minimum around 0.3 um and is higher at other sizes [2]. The minimum eff iciency o£ the quartz filter used in our experiments was rated at 99.992!. The appearance of a filter conta ining trapped aerosol material is shown in Figure 7.
FRONT
FIGURE 7.
BACK
FIGURE 5.
Appearance of a Zircaloy-4 Clad, UO 2 Pellet that was Heated to 2000°C, at a Rate of 0.5°C/s, in a Flowing Argon Atmosphere.
THERM'CPL.TYPE K COVER-
BODY (FILTER AREA CUT-AWAY]
VENT SUPPORT [ST. STL. MESH) FILTER (GLASS
FIGURE 6.
FIBRE)
A Schematic of the Filter Holder
Appearance Containing Material
of a Quartz Fibre Trapped (Black)
Filter Aerosol
Chemical Analysis Chemical elements expect'c to be present in the aerosol particles are U, Zr, Cr, Fe and Sn. The last three are the primary alloying elements of Zircaloy4. The intent here was to measure the released aerosol mass associated with each of the elements separately. Simple gravimetry is not suitable for this purpose, and therefore, we had opted for a spectrochemical technique. It involved dissolving the whole filter including the trapped aerosol material, in an acid (HNO 3 -HF-A1C1 3 ) solution and determining the concentration of the metallic elements in the solution by atomic emission/absorption spectroscopy. An inductively coupled plasma atomic emission technique was used to determine the concentrations of U, Zr, Cr, ar.d Fe. This technique involved nebulizing the solution into a fine mist, passing the mist through a hightemperature plasma zone, and detecting and processing the resulting atomic emission spectra using a sophisticated computer-based system. The detection limits for ihe concentrations of the elements vere better than a few hundred micrograms per litre of solution. These were adequate for analyzing the aerosol material trapped in the filter. Equipment to analyze Sn is not yet set up in this
CNS 9th ANNUAL CONFERENCE, 1986 247
system. Therefore, the concentration of Sn was determined by another spectroeheroical technique based on atomic absorption, which was more accurate but more time-consuming and expensive. Test: Matrix The test matrix (Table 1) covered a range o£ temperatures, rates of temperature increase (ramp rates) and flow rates. Scoping tests revealed that, when the maximum temperature was below 1800°C, the mass o£ the aerosol material collected in the filter was small and close to the detection limits. On the other hand, sheath material relocation became prominent above 2000«c (in argon atmosphere), resulting in a distorted sample geometry. Therefore, the maximum temperature was limited to the range between 1800 and 2000°C. Two linear rates of change (ramp rates) of the sample temperature, 0.5 and 2°C/s, were chosen to assess the effect of the temperature ramp rate on the aerosol material release rate. TABLE 1: TEST MATRIX Test
Temperature (C)
Ramp Rate (C/s)
Flow Rate (L/min)
1 2
1500-1800 1500-1800
0.5 2.0
12.7 0
3
1500-1800
2.0
12.7
4
1500-1900
0.5
12.7
5
1500-1900
2.0
0
6
1500-1900
2.0
12.7
7
1500-1900
2.0
25.0
8
1500-1900
2.0
50.0 12.7
9
1500-2000
0.5
10
1500-2000
2.0
12.7
11
1500-2000
2.0
25
12
1500-2000
2.0
50
13
1500-1900 and 120 s @ 1900
2.0
12.7
1500-2000 and 120 s @ 2000
2.0
14
12.7
RESULTS The mass of the released aerosol material and its elemental composition are presented in Table 2. The released amounts varied widely from a rov micrograms to a milligram. The magnitude of eccor in the chemical analysis varied for different elements. The error for a given element depends not only on its concentration in the solution but also on relative concentrations of other elements. Since these concentrations varied from one test to the other, so did the error. The released masses of some of the elements were small enough to be within their detection limits. This occurred especially in tests where lower temperature, lowei mass flow rate and higher temperature ramp rate prevailed simultaneously. The released mass for all elements was smaller at lower temperatures, while other conditions regained the same ( see Tables 1 and 2; the mass in test 1 < the mass in test 4 < the mass in test 9; test 3 < test 6 < test 10; test 7 < test 11; test 8 < test 12; and test 13 < test 14). This is expected because vapour pressures are lower at lower temperatures The released mass loir all elements was also smaller when the temperature ramp rate was higher (the mass in test 3 < the mass in test 1; test 6 < test 4; and test 10 < test 9 ) . This again agrees with the fact that the time available for the mass release is shorter at the larger la tes of temperature increase. The released mass was larger at higher flow rates of argon, while other conditions remained the same ( the mass in test 3 > the mass in test 2; test 8 > test 7 > test 6 > test 5; test 12 > test 11 > test 10). This is consistent with the fact that the rate of vapour mass transfer from a hot body to a gas flowing around it increases with the gas flow rate |4). An interesting result is that, when the flow rate was zero (test 2 and tsst 5 ) , no material was transported to the filter located approximately 0.5 m away from the sample. In test 5, where the maximum temperature was 1900°C, some evidence of material release could be found in the form of thin black coating of the zirconia platform on vhich the sample was located. However, no such coating was observed in Test 2, where the maximum temperature was only 1800°C. Fractional Release Rates To present the data in a form suitable for input into computer codes, we calculated fractional release rates for all the elements, by fitting the solution of a set of first-order differential equations of the form dm./dt
The mass flow rates of argon used in the experiments were 12.7, 25 and 50 L/min. These flow rates were chosen to maintain flow around the hot sample in the experiment similar to that postulated to occur in a CANDU sub-channel under LOCA-LOECI conditions. This flow similarity was attained by matching the flow Reynolds number in the experiment to that in the sub-channel. Sub-channel flow Reynolds numbers of a few hundred are of interest for analysis of severe LOCA-L0ECI events [3|. The three mass flow rates used in the experiments translate into Reynolds numbers of 78, 153 and 305.
248
CNS 9th ANNUAL CONFERENCE, 1968
-k.{T(t))m. ;
1,2,3,4,5
(1)
to the experimental data. Here, n^ is the mass of the ich element in the sample, T(t) is the sample's temperature as a function of time t, and k< is the fractional release rate (fraction/minute) for the ith element, which is defined by the following Arrfienius expression k. (fraction/minute)
(2)
TABLE 2: MASS AND ELEMENTAL COMPOSITION OF THE RELEASED AEROSOL MATERIAL Test
Released Mass (ug) Fe Sn
Cr 1
80+10
2
20+10
Zr
5+1.5
< 10
< 50
< 10
< 50
< 10
< 50
None transported to the filter
3
< 10
< 10
2 + 0.6
4
340 ± 2 0
5
None was transported to the filter, but black coating of the platform was observed
6
40+10
60 ± 10
< 10
80 ± 20
10 + 3
< 10
< 50
7
116 ± 7
32+3
80 ± 20
17+3
8
180 ± 1 0
35 ± 3
110 + 30
97 ± 5
426 ± 26
9
480 ± 20
80 + 10
380 + 50
10
110 + 10
30 ± 10
250 + 10 240 ± 7 0 53 + 16
138 + 8
230 ± 50
< 10
11
196 ± 1 0
69+5
90+30
86+5
530 ± 30
12
203 ± 1 1
76+3
110 + 30
102 ± 5
550 + 30
13
189 ± 9
36+3
80+20
12+3
176 ± 11
14
370 ± 20
110 ± 1 0
80 ± 20
50 + 10
1320 + 10
where, R is the universal gas constant, and Ai Et are constants pre-exponential factor activation energy of the release process for ith element, respectively. The solution Equations (1) and (2) is
and and the of
O "-
FLOW REYNOLDS NUMBER
S
305 (1-m ).= 1 - m.oexpl-A. J exp{-E./R T(t)) dt). 0 (3) Here, (l-n^) is the released mass of the ith species, m io is the initial mass of the ith element in the sample before it is heated and t, is the total period of heating. In Equation (3), the released mass (l-mt) and the temperature-time function T(t) are quantities measured in the experiments. At and E i are the tvo unknown constants. They were determined by nonlinear, least square fitting of Equation (3) to the experimental data. Results of tests in which the released mass vas less than the detection limit were not used in these calculations. The logarithm of the fractional release rates was calculated by substituting the values of the constants A 4 and E s into Equation (2). These aie plotted as a function of the inverse of temperature, in Figures 8 to 12, for all the elements and different flow Reynolds numbers. The release rates increased with the flow Reynolds numbers for all the elements, but the magnitude of increase varied from one element to the other.
o I O
-I REFERENCE •— [5]
S -3 -
-4 -
p-J^i ERROR
-5 3.2
FIGURE B.
3.7
4.2
4.7 (xlO'4)
5.2
5.7
Fractional Release Rate (see Equation (2)) for Chromium
CNS 9th ANNUAL CONFERENCE, 1988 249
J
-3
1
I
1
-4 --
305 FLOW REYNOLDS"
Z
a. o
NUMBER 78
or N
305 -2
.153
I -5
FLOW REYNOLDS NUMBER
t
-
REFERENCE
[5] -7
8-3
-
-8 -
-4
^
ERROR
ERROR _J
-5
3.2
3.7
4.2
I
-9
L
4.7
5.2
3.5
5.7
1
4.0
4.5
i
1
5.0
5.5
6.0
I/TIK) (xlO-4)
FIGURE 9.
Fractional Release Rate (see Equation (2)) for Iron
FIGURE 11. Fractional Release Rate (see Equation (2)) for Zirconium
-3 FLOW REYNOLDS NUMBER 153 78
-2 REFERENCE
[5]
-4 2
FLOW REYNOLD: 305 NUMBER REFERENCE[Ij
153
1 - 5 IS or
0: O
S-6
-4
-5
-8
-6 -
ERROR
ERROR
-9
-7
3.2
3.7
4.2
4.7
5.2
5.7
3.2
3.7
4.2
4.7
5.2
5.7
4
1/TtK) ( x l O ' )
FIGURE 10. Fractional Release Rate (see Equation (2)) for Tin
250 CNS 9th ANNUAL CONFERENCE, 1988
FIGURE 12. Fractional Release Rate (see Equation (2)) for Uranium
The release rates for the alloying elements o£ 7ircaloy-4 (Cr,Fe and Sn) obtained from this work are compared with those of our earlier work [5J in Figures 8 to 10. In our earlier work, 2g of Zircaloy-4 (no uranium) was magnetically levitated, while being inductively heated within the field coil of an RF generator, in a forced-flow (7.3 L/rain) argon atmosphere. Considering the differences in the experimental conditions, the release rates agree fairly well both in magnitude and slope. Ue could not identify any data in the literature for the release rates of Zirconium in argon atmosphere. One data point obtained by Albrecht et al. |1] i n the SASCHA facility was available for uranium at 27OO°C in argon atmosphere. This is plotted in Figure 12. The release rate obtained by Albrecht et al. [1| is about an order of magnitude lower than our extrapolated data. The experimental conditions in Albrecht's [1J work were quite different from ours. First, the composition of the material heated in their assembly was 35£ (by weight) V02, 55% stainless steel and 102! Zircaloy-4 (intended for LWR studies). The composition used in our work varied slightly for dif.irent tests and was approximately 12X Zircaloy and 882 U 0 2 . The chemical composition of the condensed (molten) phase can have large effect on the release rates. Second, convective flow conditions existed in their assembly as opposed to the forced-convective conditions in our tests. Forced convective flow enhances release rates. Hence, an order of magnitude larger release rate for uranium in our tests compared to that of reference 1 is not unexpected.
4. SHERWOOD, T., R. Pigf or d and C. Vilke,"Mass Transfer", McGraw-Hill, Nev York Inc. 1975. 5. HULPURU, S.R., D. WREN and R.K. RONDEAU "Aerosol Material Release Rates from Zircaloy-4 at Temperatures from 2000 to 22OO°C", American Nuclear Society Transactions, Vol. 55, pp. 433-434 and unrestricted, unpublished Whiteshell Nuclear Research Establishment Report, WNRE-770, 1988.
SUMMARY Experimental data on mass release rates of aerosol source (structural) materials from a hot, Zircaloy-4 clad, unircadiated U0 2 pellet situated in a forced-flow argon atmosphere have been obtained. The data covered a range of temperatures from 1500 to ?000°C and sub-channel flow Reynolds numbers from 78 to 305 postulated to occur in a CANDU channel under L.OCA-LOECI conditions. These data agree fairly veil with those available in the literature. They form Dart of a database useful for CANDU aerosol source-term model development. Release rate data in a forced-flow steam environment are currently being obtained.
ACKNOWLEDGEMENT This work was jointly funded by Atomic Energy cf Canada Limited, Ontario Hydro, Hydro Quebec and New Brunswick Power Commission under the COG safety program.
REFERENCES 1. ALBRECHT, H. , M.P. OSBORNE and H. WILD, "Experimental Determination of Fission and Activation Product Release during Core Meltdown", Proc. of Topical Meetii:sr o n Thermal Reactor Safety, CONF-770708, AUGUST, 1977. 2. S A W , O. ^Fundamentals of Aerosol Wiley-Interscience, New York, 1978
Science",
3. LAU, J., Ontario Hydro, Private communication
C N S 9th A N N U A L C O N F E R E N C E , 1988 251
EXPERIMENTAL MODELLING OF FLOW BLOCKAGE IN A CANDU FUELLING DUCT I.G. ELPHICK, A.M.C. CHAN Mechanical Research Department Ontario Hydro Research Division 800 Kipling Avenue Toronto, Ontario M8Z 5S4
ABSTRACT Flow losses were determined for fuel 1 ing trolleys i n a mode 1 CANDU fue11i ng due t. The mode 1 was representat ive of the Bruce and Darlington design. Operating conditions in the model were based on Mach numbers for a large break LOCA. A compressed air flow facility was constructed for these tests. The loss coefficients were based upon time-resolved measurements of pressure, temperature and flow.
number effects thus predominate. This is recognized as an adequate working assumption for flows with high local Mach numbers and was adopted in these tests.
TABLE 1: FULL SCALE LOCA CONDITIONS Fluid Pressure Temperature Velocity
100% saturated water vapour 180 - 220 kPa 120°C 3S - 110 m/s
INTRODUCTION To maintain structural integrity of a CANDU containment system, the pressure within the reactor vault should not exceed normal design 1imits. Of particular interest are the pressure histories following a large loss-of-coolant accident (LOCA). Consequently, the pressure history in the reactor vault must be determined for LOCA conditions. During a large LOCA, pressure builds in the reactor vault and containment system as steam and water escape from the ruptured pipe or header. Sensing the rise in pressure, the valves connecting the fuelling duct to the vacuum bui1ding open. Hot air, steam and aerosoIs flow from the reactor vault through the fuelling duct to the vacuum buildlng. Any obstruct ion in this path, inc1uding the fue11ing tro11eys, will add resistance to the flow. This wi 11 increase the overpressure in the accident reactor vault. Therefore, it is important to quantify these flow losses. Due to the complicated geometry and the three-dimensional nature of the flow, it is difficult to est imate these flow losses, Consequently, it was decided to determine the flow losses experimentally usi ng a scale model. This study focussed on the flow losses for the presence of one or two fuel 1 ing trolleys in the flow path. Following is a description of the experimental facility, the tests performed andthe results.
Mach Number Reynolds Number
0.071 - 0.223 25.3 x 10 - 79.5 x 10
Operating conditions in the mode] were obtained by equating Mach numbers. An appropriate scaling factor was selected by satisfying the condition that the Reynolds number in the model exceed 2 x 10 . It was expected that wi th the above Iimit satisfied, loss coefficients would be independent of Reynolds number. The modeI parameters for the same reference 1ocation are shown in Table 2.
TABLE 2: MODEL PARAMETERS Fluid Pressure Temperature Velocity Mach number Reyno3ds number Scaling Factor
dry air 180 - ?.0O kPa 20 °C 24 - 75 m/s
ij
0.071 - 0. 223 6 x 10 - 2 x 10 1/40
! j
J Air Flow Facility
MODEL AND FACILITY Design Consi derat ions The model was representat 1 ve o€ the Bruce and Darlington design. Full scale LOCA conditions wer * used to develop the model parameters (see Table 1). in particular, the flow condi t ions are given at thr; fuel 1 ing machine duct (FMD) opening located between the reactor vault and the fuelling duct (1). With compressible flows, the important modelling parame t ers are t he Mach number and the Reyno1ds number. Flow losses can bo attributed to form drag and skin friction. At Reynolds numbers > 10 , skin friction becomes small compared to form drag (2). Mach 252 CNS 9th ANNUAL CONFERENCE, 19BB
The air flow facl lity is shown in Figure 1. Two hign pressure vessels stored 4 m of air at 5.5 MPa. The pressure was reduced from B. 5 MPa to 220 kPa in 2 stages. The first pressure control valve following the high pressure tanks was sized to 10.5 kg/s and 11 reduced the pressure to 700 kPa. The second pressure control valve reduced the pressure to 220 kPa. An accumulator, with a 0.5 m capaci ty was located between the pressure controllers. Air can enter the model either through the top of the reactor vault or through the fuelling duct via the 10 cm pipe and the dlffusers. The dlffusers produced a uniform velocity distribution at the entrance to the model. By means of shutoff valves, air could bo directed to elthor one of the ontrances,
BACKPRESSURE
FIGURE 1: COMPRESSED AIR PLOW FACILITY
FFUSER A
)
FIGURE 4: F.ffi VIEW OF TROLLEYS
FIGURE 2: CPCSS-SECTION OF MODEL
FIGURE 5: TUNNEL INTERIOR
FIGURE 3: SIDE VIEW OF TROLLEYS
FUELLING MACHINE
..REACTOR SUPPORT
The model included one reactor vault and a portion of the fuelling duct. A cross section Is shoun In Figure 2. The trolleys, rails and rail supports were modelled In considerable detail (Figures 3,4,5). A grid of 9 pltot tubes was located over each fuelling machine duct (FMD) opening as shown In Figure 6. Another 2 grids of 9 pi tot tubes were located 0.3 m upstream and downstream of the reactor vault In the fuelling duct. The total mass flow rate was measured with a venturl flow nozzle. It was installed downstream of the model.
LOCATION OF PITOT TUBE
's/,'-•'••''A •' < i ' • . . ' •
m i i I J
I 1 FUELLING MACHINE OPENING
FIGURE 6: LOCATION OF PITOT TUBES OVER FMD OPENINGS CNS 9th ANNUAL CONFERENCE, 19BB 253
Determining the Flow Losses TABLE 3 : TEST MATRTX air flow directed through # of trolleys
reactor vault
none one two
reactor vault + deflector
t unne1
X X X
X X X
The instrumentation provided the required data from which flow losses could be calculated. In part icular, the flow losses were determined for the fuel 1 ing trolleys located below the FMD opening. The flow loss coefficient was defined as the difference between the upstream and downstream total pressures divided by the upstream dynamic pressure. Mathematically:
X
- P 1/2 P V K
= the flow loss coefficient L
P
= the upstream total pressure
For one series of insta 11 ed as shown in simulate a pressure calandria by diverting mode 1.
tests, a deflector plate was Figure 7. Its purpose was to rise from one side of the air flow to that side of the
tu P
= the downstream total pressure td
p \
= the fluid density = the* fluid velocity.
The total pressures, upstream and downstream, were based on pitot tube measurements. Distributed pressure measurements were necessary due to the 3-dimensional nature of the flow. For the case of air flow originating from the reactor vault, there uere two grids of pitot tubes, one above each FMD opening. A total pressure for each FMD opening was calculated by averag.ng individual measurements. The upstream total pressure was calculated as a weighted average of the total pressures for each FMD opening. The weighting factor was based upon the mass flow rate through each FMD opening. For all cases, the downstream total pressure was calculated by averaging individual measurements located in the fuell lg duct downstream of the trolleys. For the case of flow originating in the fuelling duct, the total pressure was the average of measurements upstream of the trolleys. The density was determined from the stagnation temperature and static pressure in the reactor vault. The flow velocity was calculated from total mass flow rate, density upstream of the trolleys and the cross-sectional area. This was defined as the area of the FMD openings or the tunnel depending on where the flow originated.
-Reactor Vault
FIGURE 7: DEFLECTOR PLATE INSTALLATION
To acheive different Mach numbers while maintaining the same mode1 pressure, t he back-pressure valve was adjusted. Typically three valve settings were used for each test case.
Data Reduction and Error Analysis
Instrumentation and Data Acquisition The measurement of distributed pressures and total flowrate required a large number of pressure transducers and the means for scanning and storing the data. All the pitot tubes were connected to a bank of pressure transducers. They were housed in a microproressor-based pressure scanning system. This unit electronically scanned the pressure transducers. The data was transmitted to a microcomputer via the serial port. During a test, the microcomputer retrieved the data and stored it in random access memory (RAM). After the test, the data was tranferred to floppy disk for storage.
For each time-resolved set of data, the Joss coeff iclent, Mach number and Reyno1ds number were calculated. The loss coefficients were plotted as a function of Mach number or Reynolds number. Errors in the calculated loss coefficient, Mach number and Reynolds number were estimated. Relative errors were obtained using the standard procedure. The three measured quant i t i es in t he tests were gauge pressure, differential pressure and temperature. Table 4 shows their error bands. The relative errors for the calculated parameters are shown in Table 5.
TABLE 4 : MEASURED QUANTITIES
OPERATING PROCEDURE Description
Thn tests wore divided into three series with a total of soven casos. Air flow was directed through the roactor vault or through the fuelling duct. The number of fuelling trolleys located below the FMD uponlngs was one, two or nono. Table 3 shows the test ma t r I x. 254 CNS 9th ANNUAL CONFERENCE, 1988
Pressure Temperature Differential Pressure
Units kPa °C kPa
Error Band ± 0. 35 t 0.5 i 0.03B
!
TABLE 5 : REDUCED DATA Description Loss Coefficient Mach Number Reynolds Number
Relative Erroi 12. OX 1.1% 2.3%
Error Band ± 3.3 ± 0.0016 ± 16,600
15 -
Loss Coefficients 10 -
The dependence of the loss coefficients on Mach number €or the three series of tests are shown in Figures 8,9,10. Generally, each test run yielded a set of loss coefficients close in value and clustered around one Mach number. For a given number of trolleys there were two or more clusters. Each one represented a different setting of the backpressure valve. The est imated relative error of 12% for the loss coefficient is shown ' the error bands. Notice that the data falls within the error estimates.
5 -
0.04 NO TROLLEYS
* Z TKtll.LGYH
FIGURE 1 0 : REACTOR VAULT OVERPRESSURE, DEFLECTOR PLATE SERIES
The Tunnel Series included all data collected when the air flow entered the model through the fuelling duct. There is a set of three runs each for the cases of 2 trolleys and 1 trolley; and a set of four runs for the case of no trolleys in the fuelling duct. Notice that for the case of one trolley and no trolleys, the loss coefficients increased very little with increasing Mach number. Conversely, there is a significant increase of loss coefficient with Mach number for the 2 trolley case.
' NO THOLLEYS
FIGURE 8: FUELLING DUCT OVERPRESSURE, TUNNEL SERIES
The Reactor Vault Series included all data collected when the air flow entered the model through the reactor vault. Two trolleys were located in the model. It can be seen that the loss coefficient is essentially constant over the range of Mach numbers. It was observed that the pressures at the downstream FMD opening were less than those at the upstream FMD opening. There are fewer obstructions between the downstream FMD opening and the tunnel exit as compared to the upstream opening. Hence, there would be a lower pressure drop on the downstream side. The Deflector Plate Series was similar to the Reactor Series in that air flowed through the reactor vault during a test. However, it differs in that a deflector plate was mounted between the bottom of the large diffuser and the top of the reactor vault. The deflector plate allowed air to enter the model over the upstream FMD opening. The loss coefficients with two trolleys can be compared with the Reactor Series results. Common to both curves Is an end point of K = 27.9 at a Mach number of 0.09. However, the loss coefficients are less in the Deflector Plate Series than the Reactor Series for lower Mach numbers.
MACH NUUE1RR
FIGURE 9: REACTOR VAULT OVERPRESSURE, REACTOR SERIES
The loss coefficients are also plotted as a function of Reynolds number for the Tunnel Series (Figure 11). For the cases of one trolley and no trolleys, the loss coefficients were Independent of Reynolds number. With two trolleys In the model, the loss coefficients Increased slightly with Reynolds number. The Reynolds numbers ranged from 9.S x 10 to 1,5 X 10 . Similar results were evident for the Deflector Plate Series (not shown).
CNS 9th ANNUAL CONFERENCE. 1988 255
0.15 -
o.i -
0.05 -
0.7£
1.3 (Millions) REYNOLDS NUMBI2R 1 TROLLEY
i
NO THOLLEYS
x 2 TROLLEYS
FIGURE 11: LOSS COEFFICIENT VS REYNOLDS NUMBER, TUNNEL SERIES
o.oe r NO TROLLEYS
FIGURE 12: CHOKING INVESTIGATION, TUNNEL SERIES
Proximity to Choking rt Ss of interest to know if choking occurred in the model for any of the tests. To determine the proximity to choked flow, the mass flowrate was plotted as a function of the pressure drop across the trolleys. For isentropic flow, the slope of this curve would be zero for choked flow (3). However, it should be noted that the pressure drop is not isentropic in these tests. Consequently, the critical backpressure ratio for choking will be less than the isentropic case (0.528). Nevertheless, a near-zero slope will still indicate conditions are very close to choked flow. Since the upstream temperature and pressure changed during a test, the mass flowrate had to be normalized. This was also necessary for comparison with other tests. The mass flowrate was normalized by the critical mass flowrate for isentropic choked flow. The equation follows:
CONCLUSIONS 1.
Loss coefficients in a model fuelling duct have been determined. The model is representative of the Bruce and Darlington design. The flow conditions are representative of a large break LOCA.
2.
Scatter in the loss coefficients for a given Mach number was 12% or less. This is in good agreement with the error analysis.
3.
In no case did choking occur with the total mass flow.
4.
The loss coefficients show little dependence on the Reynolds number.
M M
0.0404 A
P o
where
M = the total mass flow rate [kg/si T = the stagnation temperature I°CJ a A=
the area of the FMD openings or the fuelling machine duct, depending on the flow origin, [m I
P = the stagnation pressure
The ratio of mass flow to choked mass flow ( M/M ) was plotted as a function of the ratio of static pressures upstream and downstream of the fuelling trolleys ( P /P ). Figure 12 shows the plot for the Deflector PJate°Series. The results indicate that the two trolley cases are much closer to choking than the other cases. This is consistent with the reduced flow area when two trolleys are in the model.
258 CNS 9th ANNUAL CONFERENCE, 1988
ACKNOWLEDGEMENT This project was funded by the Nuclear Studies and Safety Department (NSSD), Ontario Hydro. The authors wish to thank W.W. KOZIAK and N. J. PENN, NSSD, for their valuable input.
REFERENCES (1)
KOZIAK, W. w.. Nuclear Studies and Safety Department, Ontario Hydro, private communication. 1983.
(2) VENNARD, J.K., R. L. STREET, "Elementary Fluid Mechanics", 5 ed., McGraw-Hill Book Company, Toronto, 1970. (3)
REYNOLDS, W.C. , JjLC. PERKINS, "Engineering Thermodynamics", 2 ed., McGraw-Hill Book Company, Toronto, 1970.
Session 7: The Next Generation Reactors
Chairman: R.S. Hart, AECL CANDU Ops
CNS 9th ANNUAL CONFERENCE, 1988
257
CANDU 300 DESIGN SAFETY IMPROVEMENTS by J.M. HOPWOOD AND S. PANG Atomic Energy of Canada Limited, CANDU-Operations Sheridan Park Research Community Mississauga, Ontario Canada 15K 1B2 Tel: (416)823-9040 Ext. 4580
ABSTRACT The safety design of the CANDU 3 is an evolution of the successful CANDU approach, building on features from previous CANDU plants. Close attention has been paid to safety throughout the conceptual design process. This provides confidence in our ability to meet safety goals with the final product, without excessive design changes.
(a)
Group 1 - includes buildings, structures and equipment required for the production of electrical power.
(b)
Group 2 - includes buildings, structures and equipment required to mitigate accident consequences.
Some of the key features of the CANDU 3 layout include separation of buildings by function; segregation of high energy systems from plant control INTRODUCTION systems and from safety systems; and segregation of safety and safety-support systems from process The CANDU 300, now designated the CANDU 3, is the systems. most recent in the line of CANDU designs trom Atomic Energy of Canada Ltd. As shown in Figure 1, the CANDU 3 nuclear generating station consists of six principal The safety design of the CANDU-3 is an evolution of buildings. The distribution of equipment and services the successful CANDU approach, building on features among the buildings is primarily by function. from the CANDU-6 design and other CANDU plants. One of the objectives for the CANDU-3 is to further enhance plant safety while maintaining an economic, ~^CANDU 300 reliable plant design. This is accomplished through attention to safety throughout the design process. Careful choice of plant layout, and system grouping and separation, are emphasized in protecting against external events. Safety considerations are built into the design of both process and safety systems (the two shutdown systems, the emergency core cooling system, and containment) at the conceptual stage, to maximize both capability and availability of system response for abnormal events. This paper reviews and summarizes the salien' design safety features in the conceptual design for the CANDU-3. The overall grouping and separation of plant systems is covered first. Safety-related improvements to process systems, heat removal and operator interface are discussed next. Finally, the special safety systems are discussed in turn, with comments on the accident analysi» approach used in support of their design. GROUPING AND SEPARATION Canadian safety practice requires physical and functional separation betwe.r systems used to mitigate accident conditions, and systems used for normal plant operation, as wel.\ as separation between safety systems. These safety considerations play a significant role in the plant layout and distribution of systems and components in the CANDU 3 design.
;*CTO» HUXILIARr BUILDING
1>J6 COOLINO CC AND DjQ flECTJVERr I V B I L I W l AND PERSONNEL 4CCCSS CONTROL ROOM NEW FUEL L O U M M MAINTENANCE BUILDINO TURBINE BUI10INO ADMINISTRATION BUILDINO
To implement the above separation requirements, the following grouping philosophy "'as adopted: FIGURE 1 PLANT LAYOUT
C N S 9th ANNUAL CONFERENCE, 1988 259
The reactor building (RB) contains the main components of the nuclear steam supply; i.e. the reactor structures, the moderator system, the heat transport system, the fuel handling system, and the auxiliary systems.
The group 2 service building is connected to the reactor auxiliary building through an above-grade enclosed structure. This structure includes a personnel passageway at grade level and a cable/process umbilical area above the personnel passageway. The cable/process umbilical area is vertically divided into two sections by a wall, designed such that the consequences of a fire or a process failure are confined to one section and not transmitted to the other side.
The reactor building is surrounded by the reactor auxiliary building, vhich interfaces with the other principal buildings and accommodates all of the connecting umbilicals. The reactor auxiliary building also houses the Emergency Core Cooling system in the Group 2 area, and the irradiated Fuel Storage Bay and The four safety systems are located in two Main Control Room in the group 1 area. Other Group 1 physically separate areas within the Group 2 Service areas include the Turbine Building, the Group 1 Building: SDS1 and emergency core cooling system Service Building and the Maintenance Building. (ECCS) in one area, and SDS2 and containment system in the other (see Figure 2). This arrangement provides separation between SDS1 and SDS2 as well as between The layout of the Group 1 area is ba3ed on the following rationale: ECCS and containment. (a)
The turbine building is separated from the main control room and the support services for the reactor systems, to reduce the consequences of any fire which may originate in the hydrogen cooling system or in the turbine lubricating oil system. The turbine is oriented in a direction which virtually eliminates the possibility of turbine missiles affecting the main control room, the reactor systems and the safety systems.
Cb)
The main control room area is located away from the high-energy piping such as main steam and feedwater.
(c)
The maintenance building, which includes the waste management systems, is separated from the main control room and the group 1 service building to minimize the spreading of activity to "clean" areas of the plant.
, C A N D U 300
There are two service buildings for Group 1 and Group 2 functions respectively. The group 1 service building houses the services required for normal operation of the plant: the chilled water system, the group 1 recirculated cooling water system, the group 1 un-interruptible power supplies, class III power distribution for all group 1 areas, and class IV power supplies tor the nuclear steam plant (NSP) loads.
UJ
The group 2 service building contains the safety systems and safety support systems necessary to mitigate accident conditions.
REACTOR BUILDING
FIGURE 2 GROUP 2 SERVICE BUILDING
(a)
Control equipment for shutdown systems 1 and 2, containment and ECC In summary:
fh)
Grcjp 2 Diesel generators
(a)
(c)
Secondary control area (including the group 2 control equipment rooms)
Greater physical separation between buildings has enhanced the plant's protection against any design basis events.
(b)
(d)
Group 2 feedwater system
(e)
Group 2 recirculated cooling water system.
Grouping all safety-related equipment in one area which is sheltered from the external events and from events which could occur in the group 1 areas of the plant, provides hxcellent protection to these systems.
(c)
The physical separation between safety systems in the group 2 service building effectively produces three separate groups of systems, which is an enhancement over previous designs.
The location of this building is shown in Figure 1. Structures and systems for this building are seismically qualified. The group 2 service building structure is strengthened to resist the tornado hazard.
260 CNS 9th ANNUAL CONFERENCE, 19B8
PROCESS SYSTEMS FEATURES The heat transport system includes an enlarged pressurizer capacity. The pressurizer controls the pressure of the heat transport system (HTS) and accommodates the HTS DjO volume change from zero power cold conditions to iujj power operation, Minor perturbations in HT D 2 0 inventory are controlled by feed and bleed as in previous designs. A large pressurizer reduces the requirement for feed and bleed during rapid changes in reactor power and reduces the severity of transients in the heat transport system due to depressurization. Particularly, in the event of secondary side failures, the large pressurizer inventory keeps the heat transport system liquid-filled, providing optimum thennosiphoning conditions, if the hea.t transport pumps are not available. Cooling spray from the D 2 0 feed pumps is used for pressure control in the pressurizer, instead of steam bleed. Elimination of the pressurizer steam bleed valves minimizes potential leakage and reduces the number of outflow paths from the heat transport system. It also facilitates the use of a small bleed condenser Cdegasser condenser for CANDU-6), which serves as a receiver for heat cransport relief and purification bleed. A smaller vessel reduces the total discharge of inventory from the heat transport system in TMI type scenarios. As a defence in depth, the moderator system is used as a heat sink in the unlikely event of a LOCA coincident with the loss of emergency core cooling. To satisfy this requirements, the moderator heavy water temperature is maintained at a certain degree of subcooling. In the CANDU 3 design, circulation of the moderator heavy water in the calandria is improved by locating the outlet nozzles slightly above the inlet nozzles in the upper half and along both sides of the calandria. The inlet nozzles inside the calandria are directed downward close to the calandria shell to promote a flow pattern which eliminates hot spots. As a result, the desired moderator subcooling is achieved while operating at a higher moderator outlet temperature. This allows the sizes of the moderator heat exchangers and pumps to be reduced. In certain scenarios such as steam generator tube leaks, it may be necessary to isolate the secondary side of the steam generators. This is accomplished in the CANDU 3 design by having a separate steam line leading from each of the two steam generators directly to the turbine steam chest. The steam balance header is eliminated in the CANEU 3 and is replaced with a relatively small pipe connecting the two steam lines, with isolation capability. Hence, each steam generator can be isolated by remote manual operation of the main steam stop valves and the valves on the interconnect line.
HEAT REMOVAL CAPABILITY The reactor heat removal function is provided by two independent groups of systems. Group 1 systems perform the heat removal function for normal power production and during normal reactor shutdown. Group 2 systems perform the reactor decay heat removal function during accident conditions and common cause events which may disable the group 1 systemsj Group 2 systems are seismically qualified.
The Group 1 systems consist of the feedwater system and the recirculated cooling water(RCW)/raw service (RSW) water systems. The feedwater s»"»"- which supplies normal feedwater to the steam generators comprises the main feedwater pun>r>; on class IV power and two auxiliary feedwater pumps, one motor driven on class III power and one steam-turbine drivenThe recirculated cooling water system is a closed loop system supplying cooling water to the moderator, shield cooling, reactor building air coolers and other group 1 users in the plant. The raw service water system in turn supplies cooling water to the RCW heat exchangers. The reliability of the above cooling water systems has been maximised by careful choice of components and control. From past experience, a major contribution to the failure of the cooling water systems is from failures of rubber expansion joints and check valves. Therefore, for CANDU 3, rubber expansion joints have been eliminated in the RCW system and their use in ths RSW system has been minimized. Special attention has been given to proper design of chei.k valves or other valving arrangements to preclude flow bypass at the pumps. In addition, a two-train RSW system is provided. Each train has independent piping from the RSW pump house to the RCW heat exchangers and is capable of full load operation providing redundancy to cater to pipe failures. The capability to remove heat during accident conditions has been enhanced by providing an ir-dependent source of high pressure feedwater to the steam generators. This group 2 feedwater system consists of two 100% high pressure pumps, one motor driven supplied by group 2 class III power, and one direct diesel driven. The diesel driven pump is independent of Class IV and Class III power. This assures a supply of feedwater if the main electrical services become unavailable. These pumps take water from demineralized water storage tanks located in the group 2 Service Building. The system is completely separated from the group 1 feedwater system located in the turbine building. The system is capable of supplying water to the steam generator at full pressure and is automatically initiated on a low-low steam generator level signal. The Group 2 feedwater system ensures a continuous supply of feedwater without a requirement to depressurize the steam generator, if normal feedwater supply is lost. Water is introduced into the steam generators before it empties or before the tube bundle is uncovered, which is beneficial from a thermal stress point of view. Another feature of grrup 2 heat removal systems is the group 2 RCW system which supplies cooling water to: (a)
the ECC heat exchangers r.nd pumps, and
Cb)
the shutdown cooling heat exchanger and pumps.
The group 2 RCW system is functionally independent and physically separated from the group 1 RCW system, and is seismically qualified. The system is located in the group 2 Service Building. The group 2 raw service water system is located in its own pumphouse, and supplies cooling water to the group 2 RCW heat exchangers. The above systems are supplied by group 2 Class III power.
CNS 9th ANNUAL CONFERENCE. 1988 261
One notable feature in the heat removal capability of the plant is the provision of group 2 services (cooling water, power supply, etc.) to the shutdown cooling system. The shutdown cooling system is normally used to remove decay heat after reactor shutdown and provides cooldown of the HT system for me intenance. The shutdown cooling system is seismically and environmentally qualified and is supplied with group 2 services to provide a long-term heat sink following a s-iismic event. It also has the capability to cool down the HT system from zero power hot conditions. It can bring the HT system to a cool and depressuri.'.ed state following a seismic event. In addition, in the event of loss of group 1 RCW (which supplies the moderator or shield cooling heat exchangers) ( the shutdown cooling system on group 2 RCW can be used to cool down the heat transport system so as to minimize the effect on the reactor structures. Normally isolated connections to group 1 RCW are also provided for the shutdown cooling system as back up.
MAN-MACHINE INTERFACE Improvements to the display of plant data to the operator enhance plant safety as veil as providing for greater ease and reliability of normal operation. The Plant Display System (PDS) will provide an integrated presentation of both normal control system and safety system information on CRT displays located in operator consoles. The system provides for normal operation of the plant directly through the console CRT interface. Conventional stand-up control panels are used for specialist functions such as maintenance shutdown conditions. The control displays in both the main and secondary control areas are standardized, according to common format for required systems. In addition, Group 2 systems monitoring and control functions will be grouped together in a Post-Accident Management display to aid operator action under accident conditions. Safety Systems are designed to minimize operator actions. For instance, the routine testing of special safety systems is carried out by computer control, to reduce the operator effort required, and to minimize the possibility of error. Other examples are given below.
SHUTDOWN SYSTEMS Shutdown system features have been chosen operability, maintainability and response speed.
for
Operating experience with programmable controller based shutdown systems on CANDU 6 plants has demonstrated positive gains in several a r e a s , including the selection of trip signals, higher reliability than conventional modules, elimination of potentially unsafe failures through self checking, and greater facility for incorporating changes as a result of pre-installation testing. The CANDU 3 design retains these features and includes automatic calibration of flux detector signals and improved display and trip testing. Testing of shutdown system functions (including liquid poison injection system gadolinium concentration) can be carried out on line. Shutdown
262 C N S 9th A N N U A L C O N F E R E N C E , 1988
system test computers are used for automating the overall te ' process. This decreases operator burden and reduces channel rejection time. In addition, self-checking by the computer provides specific information on the location of maloperating components, aiding maintenance trouble-shooting. Trip coverage for abnormal events is achieved by using signal parameters well-established in the design and operation of previous CANDU reactors. For neutronic trip coverage, ion chambers are located on each side of the core, for each shutdown system, to detect high rate of change of flux. This provides further depth in detection of abnormal power increases, and also provides faster response for loss-of-coolant accidents (LOCA's). Neutronic trip signals are transferred directly to the shutdown mechanism for speed, while process trip signals are transferred through the trip monitoring system for flexibility. The speed of the shutdown system reactivity mechanisms has also been improved. For shutdown system no. 1, mechanical 3hut-off rods "se a lighter design with a stronger accelerating spring. The design of the calandria and shield tanks also allows the shut-off rods to be positioned closer to the core, so that there is a shorter distance to travel before the rods are effective. The liquid poison D 3 O interface is closer to the core. Optimising the position of the reactivity elements closer to the core provides a simple, inherent means of increasing shutdown system speed. SDS2 poison concentration is monitored on line, using a recirculating sampling system.
EMERGENCY CORE COOLING The Emergency Core Cooling System (ECCS) includes LOCA detection, emergency coolant injection, and steam generator cooldown. LOCA detection, and ECCS initiation, is achieved automatically for all pipe breaks down to the level of leaks managed by normal process systems. To provide unambiguous detection, for high injection reliability without excessive risk of spurious operation, a LOCA is indicated by the combination of a main parameter signal (low coolant pressure) and one of three conditioning signals: high containment presure (large LOCA), high moderator level (in-core LOCA) and very low pressurizer level (small L O C A ) . The low pressurizer level signal provider detection of all pipe breaks which cause an uncontrolled drain from the heat transport system, regardless of break location and siz<=. In a related approach, the pressure and inventory control system has been designed to bottle up on any internal valve failure from the heat transport system to the pressure and inventory control system. All ECCS responses, including the switchover from the high pressure accumulator stage to the low pressure stage, are controlled automatically. Together with the automation of LOCA detection, this minimizes the reliance on operator action following a LOCA, freeing operator attention for monitoring and accident management. The emergency coolant injection system consists of an initial high pressure accummulator stage, followed by pumped injection, first using water from an external storage tank, then recycling the water
"YCANDU 300
spilled from the core following a LOCA. The LOCA detection signal initiates not only ECC injection but also rapid boiler cooldovm via the main steam safety valves. The injection system is designed for core refill matched to the heat transport pump post-LOCA operating strategy. The layout of the injection supply comprises parallel injection to each reactor inlet header, and to each reactor outlet header. This arrangement avoids by-pass of flow around the steam generators following ECC valve operation, and improves HT pump operating conditions during injection. To enhance refill of the Zvel channels after a large loss of coolant, feeder pipes are laid out to ensure a small downward slope to the upper horizontal pipe sections. This allows more efficient water inflow to the fuel channels. For improved long-term accident management, the KCCs recovery pumps and heat exchangers are located in an enclosed area in the Group 2 portion of the reactor auxiliary building, with facilities to collect leakage from pumps and heat exchangers. Availability of the ECCS is maximized by careful attention to redundancy requirements, and by the use of fault tree analysis to evaluate the design reliability at the conceptual stage. CANDU reactor plants have the capability for decay heat removal via the moderator, even for the situation where ECC is hypothesized to fail following a LOCA. The CANDU 3, includes features to further enhance this capability. Calandria tubes will be pre-treated to obtain a high-eaissivity inside surface to enhance radiative heat transfer to the moderator during temperature excursions. A duplicated boiler cooldown signal ensures that the PHTS reaches low-pressure conditions independently of ECCS operation. Provision to supply make-up water to the moderator after a LOCA provides additional assurance of heat sink capability.
CONTAINMENT The containment system provides an envelope around the nuclear steam supply system which can be made leak-tight in the event of an accident. The envelope is provided primarily by the reactor building (see Figure 3). This building will have a complete steel liner for assurance of low leak rate. Penetrations are grouped into a small number of specific areas In the reactor building for easier access. As well as on-line penetrations testing, overall containment leak rate will be monitored on line. Thjj will reduce the reliance on expensive full-pressure tests of the containment envelope. Pressure excursions within the reactor building are accommodated by a choice of design pressure to cover the whole range cf heat transport system pipe breaks. This means that a dousing system is not required. This simplifies containment system testing, improves containment availability and eliminates operating problems due to spurious dousing operation. In addition, the use of passive means to accommodate containment overpressure emphasises the inherent safety aspects of containment. The layout of structures within the reactor building features a large accessible area, enclosing a relatively smaller vault which includes the reactor, fuelling machine and steam generators. This makes for effective contamination and tritium control while allowing the fullest degree of access for equipment maintenance.
PERSONNEL ACCESS ROUTE
CONNECTING STRUCTURE TO GROUP 2 SERVICE
FIGURE 3 REACTOR BUILDING ELEVATION
In addition, the reactor vault would enclose any fission products released from an accident, providing a short term holdup of fission products before they could reach the actual containment pressure boundary. In the long-term, containment mixing between the reactor vault and the accessible area is achieved by operation of local air coolers. Mixing of the containment atmosphere provides the first level of defense against hydrogen burn following severe accidents, by diluting the Concentration of hydrogen to acceptable levels. A secondary level of defense is provided by hydrogen removal using igniters.
ANALYSIS SUPPORT TO DESIGN Assurance of the design safety enhancements is developed by carrying out supporting accident analysis, reliability analysis, and scoping probabilistic safety assessment (PSA) studies at the earliest stage of design. This analysis is intended to support design decisions, as opposed to later accident analyses carried out to evaluate the final design. As such, conventional, well-established methods are used in most cases, with an emphasis on sensitivity studies to confirm design margins and to compare design alternatives. As an example, containment analyses were carried out (a) To establish the desired containment design pressure with dousing eliminated, and b) to confirm the performance of the containment envelope in limiting fission product releases. Well-established modelling methods including previously-applied fission product release methods (reference 1) were used. This allows a comparison of the design performance with presently-operating reactors vhich have been analysed using the same methods.
CNS 9th ANNUAL CONFERENCE, 1986 263
using a simple limiting source term, as a test of the containment dose mitigation capability. Based on previously-used weather assumptions for a typical Canadian site, releases would lead to no more than a few percent of the dual failure individual guideline dose, even for assumed containment leak rates up to an arbitrary, 'high value t?E 5% atmosphere/day at design pressure.
•00
—
Simple tests such as this provide timely confirmation to the designer of the applicability of the design choices made.
so to -
SUMMARY
*C —
FIGURE 4 REPRESENTATIVE CONTAINMENT PRESSURE TRANSIENTS
Sample results for containment overpressure following LOCA are illustrated in figure 4. Significant containment overpressure is only seen in the long-term for the dual failure accident scenario of a LOCA with ECC assumed unavailable. In this situation, the absence of energy removal by the ECC heat exchangers or boiler cooldown allows the maximum post-LOCA energy release into containment. This result therefore provides the basis for defining the containment design pressure. Fission product release from containment was calculated for this same case,
264
C N S 9th A N N U A L C O N F E R E N C E , 1988
The conceptual design of the CANDU-3 has been developed with an emphasis on enhancing plant safety. This is done focusxng on the whole plant, both norm-itl operating and safety systems. Experience and feedback from operating CANDU plants has been a significant factor, as well as the drive to simplify operator action and to make high safety system availability easier to achieve. Building safety into plant design at the conceptual stage provides confidence that the final product will meet safety goals without excessive changes.
REFERENCES (1) "Gentilly-2 Safety Analysis for Higher Containment Leakage" by R. Hu and M.S. Quraishi. Paper presented at 8th annual CNS Conference, Saint John, N.B., June 1987.
THE CANDU 300 DISTRIBUTED CONTROL SYSTEM
W.R. WHITTALL
Atomic Energy of Canada Ltd. CANDU Operations Sheridan Park Research Community Mississauga, Ontario L5K 1B2 Telephone: 416-823-9040
ABSTRACT One of the significant design improvements in the new CA.VDU 300 nuclear plant design is the use of a modern microprocessor-based "Distributed Control System" (DCS). This system consists of a distributed network of local stations linked by dual-redundant data highways. The local stations are strategically located throughout the plant to reduce the amount cf plant cabling and wiring. The system performs the basic device control functions previously performed by logic relays, timers and electronic analog devices, as well as the higher level control functions previously performed by a dual-redundant computer system.
INTRODUCTION In the current CANDU 600 plants, the basic control functions are performed by a variety of devices, including logic relays, timers, comparators and analog controllers. Most of these devices are now obsolete. These devices are connected to each other and to field devices via multi-conductor cables and cross-connection wiring in a large Control Distribution Frame (CDF). More complex control functions (see Table 1) are performed by a central dual-redundant Digital Computer Controller (DCC) system. The system consists of two mini-computers, X and Y. Input signals are connected to X and Y input multiplexers in parallel via the CDF. X and V output signals are connected via the CDF to transfer relays controlled by X and Y watchdogs. Normally X is in control, and Y is operating in hot standby mode. In addition to performing control functions, the DCC system also performs monitoring, alarm message display, graphic data display and data recording functions. The display functions are performed using a shared colour-graphic display system which drives twelve colour c.r.t. monitors in the main control room.
Several hundred OH 180 programmable controllers are used in each unit. A dual-redundant minicomputer DCC system performs major control functions, monitoring functions, alarm message display and graphic data display functions, as in previous CANDU plants. Conventional electronic analog comparators and controllers are used, and control device interconnections are mostly implemented by multi-conductor trunk cables and distribution frames, as in previous CANDU plants.
DCS CONCEPT The DCS concept for CANDU 300 plants is based on the use of programmable controllers, linked by data highways, for basic binary and analog device control functions, as well as the higher level group control functions listed in Table 1. The system design is based on the use of standard digital control system modules, selected from the PR0C0NTR0L P13/42 product line manufactured by Brown Boveri and Company. This product was specifically designed for use in power plant distributed control and monitoring systems. The data highways consist of standard 50-ohm coaxial cable, in a multi-drop bus configuration. Input signals and local stations which locations to reduce Figure 1 ) . Because
output signals are connected to are located in strategic plant plant cabling and wiring (see of the difficulty of providing
CANDU 300
TABLE 1: CANDU 600 COMPUTER CONTROL FUNCTIONS Reactor power regulation Heat transport system pressure Heat transport system inventory (pressurizer level) Boiler pressure Boiler level Unit power regulation Moderator temperature Fuel handling system Turbine run-up
In the four-unit Darlington station, under construction in Ontario, programmable controllers have been introduced to perform the binary control logic previously performed by logic relays and timers.
FIGURE 1 SIMPLIFIED LAYOUT OF THE DISTRIBUTED CONTROL SVSTEM
CNS 9th ANNUAL CONFERENCE, 1988
265
adequate environnu n t a l protection for accident conditions, local stations for devices inside contai nwent are located out5 ide the containment penet rat ions. Logic relays, electronic analog devices, control computers, mu]ticonductor trunk control c a b l e s , and CDF c r o s s - c o n n e c t i o n s , are eliminated. Operator i n t e r f a c i n g functions are performed by a s e p a r a t e computer system c a l l e d the Plant Display System (PDS), which is linked to the DCS data highways.
BENEFITS CANDU 300 DCS provides many significant benefits compared with previous control systems. These benefits include: - reduced cabling and wiring, - reduced centre.
cable-tray
congestion
in the control
- reduced equipment cost, reduced engineering cost due to simplified procuiement and the use of block diagram programming, in
system
• Logic relays and timers Electronic analog comparators • Electronic analog controllers DCC process input/output subsystems Contact scanner
The variety of e q u i p m e n t , s u p p l i e r s and documental ion is significantly reduced, leading to a reduction in engineering costs. The use of block diagram programming supported by a computer graphic work station is significantly easier than the production of wiring information, in the case of basic device control logic, or computer programs, in the case of group control functions. The DCS has comprehensive self-checking and fault diagnosis features which allow immediate detection and replacement of failed components anywhere in the control system.
DCS CONTROL FUNCTIONS
- elimination of obsolescent technology,
more consistency documentation.
TABLE 2: EQUIPMENT REPLACED tJY THE DCS
design
and
- increased design flexibility to satisfy requirements at minimum overall cost, «
new
The DCS performs the control functions for most of the plant systems, excluding the electrical power system, the special safety systems, the safety support systems, the fuel handling system, and certain specialized or independent balance-of-plant systems such as the electro-hydraulic governor system and the water treatment plant. The fuel handling system is controlled by a similar but separate distributed control system. The plant systems controlled by the DCS are listed in Table 3.
TABLE 3 : PLANT SYSTEMS CONTROLLED BY THE DCS
- more comprehensive self-checking of the complete control system, - faster location of faults, - easier a n d more accurate up-dating documentation after changes (as-builts).
of
The most significant benefit is the reduction in plant cabling and wiring. lit the CANDU 600, the equivalent control systems involved about 80,000 cross-connections via the Control Distribution Frame. In the CANDU 300 the DCS data highways eliminate the need for a CDF. Cabling and wiring associated with logic relays, ccntrol loops, etc. is also eliminated. The substantial reduction in the time required for installation and checkout of plant cabling and wiring is a major factor in reducing the overall project schedule. Since the control logic is implemented by control processor programs which are thoroughly tested in the design office, the commissioning schedule is also significantly shortened. Studies have indicated that without this design improvement, the project schedule would be extended by at least 3 months. The equipment previously used to implement the functions performed by the DCS is listed in Table 2. The cost of the DCS is significantly less than the cost of an equivalent system using the type of equipment in Table 2. Some of the equipment used previousJy is obsolete.
266 CNS 9th ANNUAL CONFERENCE, 1988
-
Reactor power Moderator system Heat transport system (excluding shutdown cooling) BoiIer pressure Boiler level Shield cooling Recirculated cooling water Heavy water management Irradiated (used) fuel cooling Reactor building cooling Reactor building ventilation Access control Active drainage Liquid waste handling Balance-of-plant process systems
The control functions performed by the DCS include the relatively simple control of individual devices, such as control of motors via contactors and switchgear, control of isolating valves via solenoid valves, and control of modulating control valves via analog signal converters, as wel 1 as the more complex control functions, such as reactor power regulation, heat transport system pressure and inventory control, boiler pressure control, and boiler level control. {The flux mapping part of reactor powor control is performed by the PDS, because of the 1 arge number of calculations involved and the slow updatn requirement. The plant can continue to opera to for some time without flux mapping.)
DATA ACQUISITION In addition to scanning input signals for control functions, the DCS scans additional input signals from the same systems and other systems, for transmission to the PDS for various monitoring and data acquisition functions. These functions include a general purpose analog alarm scan, a general purpose contact event scan, flux mapping and channel temperature monitoring. The additional input signals include flux mapping detectors, channel outlet resistance temperature detectors, pump vibration measurements, control valve positions and radiation monitoring signals. Contact scanning by the DCS provides a plant-wide general purpose facility for over 2000 field contacts, including fast scanning (<20 milliseconds) of some inputs for high accuracy event sequencing.
completely independent communication paths linking the local stations of each channel. The same data is transmitted independently on both highways. Each receiving station reads and checks the signals from both highways. Program logic in the control processors is used to select good signals from the two sets of signals. The highways are linked to the Plant Display System by separate relatively simple fullduplex data links.
, CANDU 300 FK 8K
PR
BK
DCS ARCHITECTURE BV
The overall Figure 2.
DCS
configuration
is
shown
POWER
CHANNEL « LOCAL STATIONS
1
I
CHANNEL 8
I
1
CHANNEL
c
1
INPUTOUTPU
MODULES
in
POWER
• LOCAL 6US CONTROLLER • HIGHWAY BUS COUPLER - CONTROL PROCESSER
)
• 1 1 1 1 1 1 1 1 *
FIGURE 3 TYPICAL LJOCAL STATION
A H I G H I V ys
The three channels are linked by transfer stations, to allow inter-channel signal communication for multi-channel functions, such as median selection for triplicated sensors and switchover logic for redundant process equipment. Separate transfer stations are used for each of the two highways in each channel, to maintain redundancy.
r
LOCAL STATIONS
TRANSFER STATION
TRANSFER STATION
1
[
ADDRESS TRANSMITTERS
fl if
ADDftESS TRANSMITTERS IFVOH
f \l L
!
] i] i ] i ] \ i 1\ t ± i \
PLANT D SPLAY SYSTEM
m -BUS COUPt-f fl< T THANSCIEVER •"r1 (BK02/FK01)
FIGURE 2 DCS CONFIGURATION
The system is divided into three separate channels, to match the channelization of redundant sensors, actuators and power supplies. Each channel consists of a number of local stations, installed at strategic locations throughout the pi ant, and linked by dual-redundant data highways. A block diagram of a typical local station is shown in Figure 3. Separate communication modules are used to link a local station to each of two highway cables, so that there are two
Each signal source and sink device is connected directly to a local station. The use of junction boxes is minimized. Electrical power for the signal source and sink devices is provided via the local stations. For example, the power supply for 4-20 mA transmitters is provided by the analog input modules - there is no need for the separate power supplies, fuses, resistors, and all the associated wiring previously used for current loops. The transmitter is simply wired directly to the analog input module. Important signal source devices are triplicated and connected to separate local stations in channels A, B and C. Dual-redundant process devices, such as pumps and va]ves, are powered by "odd" (A) and "even" (C) station power buses, and are controlled by separate local stations in channels A and C respectively. The control functions are distributed among a number of dual-redundant control processors installed in the local stations. Only one type of control processor is used. This processor (PR03) implements binary and analog control functions. The control processors for basic device control are located in the same local stations that contain the output modules connected to the control device?. The control processors for higher level group co-ordination functions are located in local stations in channel B. Only three or four pairs of processors are needed to implement these group control functions.
CNS 9th ANNUAL CONFERENCE, 1988 267
"\ CANDU 300
SIGNAL COMMUNICATION Signal comrrunicat ion via the data-high ways uses base-band digital pulse duration signalling on RG23 3U 50-ohm coaxial cable. Signals are broadcast cyclically in source addressee "telegrams". The telegram format is shown in Figure 4. The average telegram duration is less than 60 microseconds (maximum 85 microseconds) , including protocol timing intervals. This provides a useful information throughput of about 270,000 data bits per second.
MEDH =
V 3 MED DEV BAD
n 1
L
X UPUM K>t(M
AND
2 OF A MTC ROHZ (.O P MTC «OH3 LOP MTC ROH4-LO-P ^
___ ADDRESS TYP 236 |;S
OOOOOOl
_
INFORMATION
__
^INIERVAL_ "" TYP 9 4 us PARITY BIT
2 3
r
INTERVAL 2 OF 3 2 3
' 0 0 I D O i OOO
NOT. I 0 0 0
1 0
1
I 000
I
I
1 0
11)
OR MTCLUCA INSIDE MTC-LOCA-OUrstDE
FIGURE 4 TELEGRAM FORMAT
Addresses are transmitted in 8 different cyclic sequences, with periods of T, 2T, AT. . . 128T milliseconds, from 8 address lists stored in dualredundant address transmitters (FV01). In our application T is less than 10 milliseconds. Binary signals are updated every T or 4T milliseconds and analog signals are updated every 8T, 16T or 128T mi 11 i seconds. This last category is used for slow measurements such as temperatures and flux mapping detectors. The local station cycle time is 5, 10 or 20 mi 1liseconds (set by a jumper) and the control processor cycle time is 20 milliseconds, so the total system delay is less than 50 mil li seconds for fast binary control functions, and less than 160 mil 1iseconds for fast analog control functions. The signal transmission technique provides very high immunity to electro-magnetic interference (EMI) and incorporates very effective error detection. The undetected error rate in a power plant environment has bef:n shown to be less than 1 telegram in 3D 1 8 which corresponds to ] telegram in 10 s years. The frequent cyclic updating of signals provides constant active signal checking and immunity to transient data errors. Data words which have detected errors are simply flagged and ignored by control processors and output modules. Persistent lack of good data is detected by bus couplers, processors and output modules, and results in fail-safe control action. Cyclic signal updating avoids the need for message acknowledgement or retransmission, and eliminates any possibility of highway overload under all plant conditions.
PROGRAMMING Control functions arc configured from a library of standard function blocks using the P10 function diagram language. An example of a P10 function diagram is g i v e n In F i g u r e 5, Design and
268 CNS 9th ANNUAL CONFERENCE, 1986
2 2
FIGURE S P10 FUNCTION OIAGRAM
documentation of a control function program is performed relatively easily using a computer graphics work station. Control function programming is done by control system designers. No special computer programming knowledge is required and the documentation is easy to understand. The processor internal machine language is represented by hexadecimal code which is a direct one-to-one translation of the function block symbols. The hexadecimal code is stored in user EPROMS and interpreted by bit-slice processor firmware in the control processor. Program execution is very fast, and no additional code is needed for initialization, input-output or program flow control. Program interrupts and multi-tasking operating software are not used. Signal communication is controlled by simple address lists in the local station communication modu]es (BK02 bus couplers) and highway controllers (FV01 address transmitters). The transmission of signals via the data highways is essentially transparent to the control system designer.
CONCLUSION The DCS concept for the CANDU 300 combines proven programmable controller and data highway technology with CANDU fault-tolerant control system concepts. This advanced integrated control system concept has substantial cost and schedule advantages relative to previous systems. Very high reliability is assured by the use of proven industrial digital control equipment configured in a fully redundant channelized system architecture.
REFERENCES (1)
E. SIDDALL, "Computer Control of Nuclear Power Plants", Nuclear Congress, New York, June 1962, AECL 1539
(2)
J.E. SMITH, "Experience With Process Control Computers In The Canadian Nuclear Power Program", IEEE Transactions on Nuclear Science, Vol. NS-20, No. 1, February 1973
(3)
W.R. COOPER and W.R. WHITTALL "Direct Digital Control of the Gentilly Nuclear Power Station", ANS Topical Meeting, Ann Arbor, Michigan, April 1973
(4)
W.R. WHITTALL and E.M. HINCHLEY "Canadian Experience in Computer Control of Nuclear Plants", University of Tennessee/ANS/ISA, Knoxville, Tennessee, October 1973
(5)
N.M. ICHIYEN and N.YANOFSKY "Computers' Key Hole in CANDU Control", Nuclear Engineering International, August 1980
(6)
W. FIEGUTH and T. McNEIL "Computer Systems Used in CANDU Nuclear Generating Stations", Canadian Conference on Industrial Computer Systems, McMaster University, May 1982
(7)
L.M. PANIS and M. SALM, "PR0C0NTR0L P - A Modern Control and Monitoring System for Power Plants", Brown Boveri Review, Vol. 71, No. 8, PP 327 to 335, August 1984
(8)
M. SALM and F. LAURIJSSEN, "Power Plant Experience with Intelligent Remote Multiplexing" IEEE Paper 84SM529-4 (1984)
(9)
W.E. WHITTALL and G.A. HEPBURN "Canadian Nuclear Power Plant Control and the Application of Microprocessors" EPRI Seminar, Phoenix, April 1985
(10)
J.R. POPOVIC, B.C. CHAN, D. BURJORJEE, B.K. PATTERSON, "Computer Control in CANDU Plants", Symposium on Advanced Nuclear Services, CNA/CNS International Nuclear Conference, June 1986
CNS 9th ANNUAL CONFERENCE. 1968 269
CANDU 300 FUEL CHANNEL
N.F. SINGH
Atomic ETier^y «I Canada Sheridan Park Research Community Mississauga, Ontario L5K 1B2
ABSTRACT The fuel channel design in CANDU 300 is governed by the station requirements for a fast construction schedule, high channel power and a high capacity factor while achieving significant savings in capital and operating costs over current designs. The shop-assembled, composite fuel channel-calandria tube module offers significant improvements in schedule for large-scale channel removal and replacement at the end of channel design life. The number of channel components are also substantially reduced over current designs. This channel design is expected to be effective and reliable to match the increased demands specified for CANDU 300 stations.
1.0 INTRODUCTION In CANDU reactors, fuel channels containing fuel bundles are supported by the reactor assembly. The channels form part of the nuclear Class 1 pressure boundary of the Heat Transport System as illustrated in Figure 1. Figure 2(a) illustrates the arrangement and key features of a typical fuel channel in a CANDU reactor. The fuel channel essentially comprises a zirconium alloy "pressure tube" connected to stainless steel "end fitting" at the ends by means of
"^ CANDU 300 Steam generators
Hex transport
rolled joints. End fittings form the out-of-core extensions of the pressure tube and provide connections to the fuelling machine, feeder pipes, annulus gas system and end shields. The pressure tube containing the fuel bundles is located inside a Zircaloy-2 calandria tube which extends between the calandria tubesheets of the reactor assembly. A calandria tube separates the hot pressure tube from the relatively cold moderator inside the calandria, and spacers provide separation between the pressure tube and calandria tube. The annulus between ttie pressure tube and calandria tube is filled with dry COj gas at a low pressure. This annulus gas serves to minimize corrosion of components and heat transfer from the pressure tube to the moderator. The moisture content of the gas annulus is monitored to detect any leakage of the moderator or heat transport coolant entering the annulus. The CANDU 300 reactor Bas 232 fuel channels. Figure 2(b) illustrates a schematic of the proposed CANDU 300 fuel channel. The general layout of this design is similar to fuel channels in current generation CANDU reactors illustrated in Figure 2 ( a ) , but this design uses a mix of new and modified features and components as described in Section 2.3. The arrangement of heat transport system steam generators and pumps provides uni-directional coolant flow through the core and enables fuelling to be performed from one end of the reactor using a single fuelling machine, as illustrated in Figure 1. This single-ended refuelling feature allows simplifications to be made to the current fuel channel design to achieve a fuel channel-calandria tube composite assembly. Fuel channel design in CANDU 300 is discussed in Section 2 and the replaceability features are described in Section 3.
2.0 FUEL CHANNEL DESIGN
2.1 Design Considerations The fallowing design considerations were used to develop the fuel channel design:
fuelling machine
Single fuelling machine at channel outlet end (tingle-ended en-power refuelling) in CAN0U 300
Fuel channel
:! Moderator ( 0 ) 0 1 Moderator beat exchanger
FIGURE 1 NUCLEAR STEAM SUPPLV SYSTEM SCHEMATIC 270 C N S 9th A N N U A L C O N F E R E N C E , 1988
" Maximize channel life by incorporating materials and design improvements identified by research and development programs 0 Minimize construction schedule as follows: ° maximize access for construction and equipment * minimize number ot components ° minimize in situ activities • maximize shop fabrication 0 assure flexible construction sequence. ° Minimize channel replacement schedule as follows:
2(0) SIMPLIFIED ILLUSTRATION OF FUEL CHANNEL FOR CANDU 300
POSITIONING . ASSEMBLY n — —~\)
7 1
FUELLING L- MACHINE INTERFACE
|CTX|
CALANORU TUBE ROLLED INTO CALANDRIA TUBESHEET
IfcH
END FITTING
2(a) SIMPLIFIED ILLUSTRATION OF A CANDU FUEL CHANNEL
FIGURE 2 °
minimize number of parts handled in situ minimize work done in active areas minimize number of tools required in situ 0 minimize in situ disassembly Minimize channel cost by simplifying the design and by reducing the number of components 0
*
0
At the inlet end, the features of smaller diameter inlet feeder pipes and a reversed feeder pipe layout outboard of the end fitting provides a more flexible feeder pipe layout over current designs. This flexible layout allows higher axial creep elongations to be accommodated at this end. This layout is also
2.2 Impact of Single-Ended Refuelling on Fuel Channel Design
"^ CANDU 300
The single-ended refuelling feature in CANDU 300 allows the following changes to be made to the current fuel channel d e s i g n , to a c h i e v e a shop-assembled channel module: 1.
The single-ended refuelling feature allows elimination of the end fitting sideport, channel closure and 1 iner tube from the inlet end fitting as illustrated in Figure 2. This allows the outside diameter of the inlet end fitting to be reduced sufficiently to enable the channel to be pulled through the reactor assembly. This reduction in inlet end fitting diameter is essential to the concept of a shop-assembled channel module capable of being inserted into, or removed from, the reactor assembly from one end.
2, With no fuelling machine interface at the inlet end it is possible to have co-axial feeder pipe connections and to arrange the feeder pipes in a reversed layout outboard of the end fittings, see Figure 3(a), This arrangement allows the end fitting length to be reduced because feeder pipes do not cross over the end fittings between the end shield and the feeder connection, as is the case with the conventional sideport arrangement, as shown in Figure 3(b). A 40% reduction in end fitting length is achieved.
3(1) REVERSED FEEDER PIPE LAYOUT-INLET END
\\ I CONVENTIONAL FEEDER PIPE LAYOUT-OUTLET END
FIGURE 3 FEEDER PIPE LAYOUT CNS 9th ANNUAL CONFERENCE. 1988
271
more favourable in accommodating differential creep between channels. This is because feeder pipes connected to the taster growing high-power channels generally move away from feeder pipes connected to the low-power channels instead of moving closer together as in conventional layouts with a sideport connection. A. Feeder pipe flexibility at the inlet end allows the outlet end to be fixed permanently so that axial creep and thermal elongation of the channel is directed to the inlet end. This arrangement eliminates the positioning assembly hardware at the two ends and eliminates the bellows at the outlet end. These changes result in a s i m p l i f i e d end fitting configuration at the outlet end. 5. Permanently fixing the outlet end also benefits the feeder connection design at this end by minimizing feeder pipe loads. 6.
Fuel is pushed out of the channel into the fuelling machine clamped to the outlet end fitting by flov-assisted fuel pushers located upstream of the fuel in the inlet end fitting. The effectiveness of the fuel pusher is maximized by having the flow holes in the liner tube located close to the outlet feeder port in the outlet end fitting. This allows the liner tube length to be reduced significantly over current designs. The shorter liner tube allows the outside diameter of the outlet end fitting to be reduced to just inboard of the sideport region.
2.3 Design Features The arrangement of fuel channel components in CANDU 300 is similar to that in current CANDU reactors as illustrated in Figure 2. However, the CAKDU 300 channel incorporates a mix of new and modified features and components. The principal features and components of the fuel channel in CANDU 300 described below are explained with reference to the simplified channel schematic of Figure 2(b) and the more detailed Figure 4, Figure 5 is a cut-away perspective view of the channel outlet end. The composite fuel channel-calandria tube module is achieved by means of the simplifications made possible by single-ended refuelling (see Section 2.2) and by the use of calandria tube extensions to connect the calandria tube to the fuel channel assembly. This modular assembly has advantages both for initial installation and for channel replacements. The advantages of this modular concept are described in Section 2.A. Calandria tubes are rolled into stainless steel calandria tube extension CCTX) tubes at each end instead of being rolled into calandria tubesheets as in all current channel designs. These calandria tube extensions are located in the end shields (between the lattice tube and end fitting) and extend outside the end shields where they are welded to the lattice tube extensions. This weld forms the pressure boundary for the moderator which flows in the annulus between the CTX and lattice tube. At the outlet end, the CTX is also attached to the end fitting to seal the gas annulus at this end and to fix this end of the end fitting to the reactor
CANDU 300 CHANNEL INLET
/
CAUMOftl* TUBE PRLSSUHETUBE
CHANNEL OUTLET END
FIGURE 4 272
CNS 9th ANNUAL CONFERENCE. 1988
CANOU 300 FUEL CHANNEL
current designs to accommodate the larger feeder pipe size (89 mm instead of 73 m m ) . This larger size is maintained within the outside envelope of the current (73 mm) feeder connection size for 2J4 inch nominal pipe (73 mm outside dia.) but the seal ring and hub have larger bore sizes.
CANDU 300
' The outlet end fitting is a modification of the current end fitting design resulting from shortening the liner tube, elimination of bellows, positioning assembly and bearings, and changes to the rolled i inr hub.. Channel closure: The channel closure is similar in its sealing arrangement to previous designs except that it incorporates a new design of latching mechanism to improve its performance and to reduce its cost by using fewer components. This closure design is fully interchangeable with closures used with CANDU 600 or Pickering type channels. A slightly smaller version of this channel closure was developed previously for the Douglas Point reactor but it was not used.
The following new features/components are used in this channel design: • FIGURE 5 CANDU 300 FUEL CHANNEL-OUTLET END DETAILS assembly. Attaching the outlet end fitting to the outlet end shield directs axial elongation of the channel due to creep/grovth and thermal expansion towards the inlet end.
•
The inlet end has a bellows to accommodate this elongation, whereas in most operating CANDU reactors axial creep elongation is shared between the two ends as illustrated in Figure 2. This bellows connects between the end fitting and the CTX extension outboard of the end shield and it also serves to seal the gas annulus at this end. As explained in Section 2.2, the feeder pipe layout at the inlet end is outboard of the shortened end fittings, which results in a more favourable layout for accommodating axial and differential creep elongations of the fuel channel. The following summarizes the design features and components incorporated in the CANDU 300 fuel channel assembly: •
Pressure tubes for CANDU 300 are Zr-2.5wtSNb tubes but they are made to a slightly modified specification (Task Group 3 Route 1) to minimize axial creep elongations. These tubes are currently under production for Ontario Hydro reactors.
•
The reference design uses standard calandria tubes. Thick-ended calandria tubes may be used to improve the calandria tube rolled joint strength and to reduce channel seg.
•
Outlet feeder connection: The connection is a modification of one
feeder of the
Pressure tube-to-end fitting thin hub rolled joint at the inlet end: A thin-hub rolled joint using a 6 mm hub is used at the inlet end because of geometry constraints of the module. Preliminary feasibility tests conducted on prototype thin-hub joints demonstrated that this joint would be acceptable based on criteria of leak-tightness, pull-out strength and residual tensile stress levels. Additional tests are required to qualify this design.
• Inlet feeder pipe connection: This connection to the inlet end fitting may be a co-axial flange connection as shown in Figure 3 or it may be a welded connection. • Inlet end fitting: The configuration and size of the inlet end fitting differs from previous CANDU designs as described in Section 2.2. These modifications provide a shorter, simplified end fitting with a small outside diameter and a co-axial feeder connection. This makes it a much lighter component and reduces its cost. The material is modified Type 403 stainless steel as in previous designs. A carbon steel transition piece shop-welded to the inlet end is required if a welded feeder pipe connection is used at this end, for compatibility with the feeder pipe material. •
• The snug-fitting spacer made of Inconel X75O is the same as used in current channel designs.
CTX and CTX connections; The CTX connections include the calandria tube rolled joints, the welds between the CTX and lattice tube extensions outboard of the two endshields, the CTX connection to the outlet end fitting and the CTX connection to the bellows at the inlet end.
Bellows: The bellows at the inlet end is much longer and of a smaller diameter than the bellows used in previous channel designs and has to accommodate 275 mm of axial elongation caused by irradiation creep/growth and thermal expansion of the fuel channel.
CNS 9th ANNUAL CONFERENCE, 1988 273
Fuel Channel internal hardware: The new internal hardware comprising the inlet shield plug, fuel pusher, outlet shield plug and latched spacer plug are currently being tested. These features are described briefly below: (a)
(b)
U)
Inlet and outlet shield plugs: These flow-through shield plugs are designed to allow single-ended fuelling. The upstream or inlet shield plug is attached to its end fitting, but is capable of being removed to gain access to the upstream end of the channel for maintenance work or as an emergency access to the fuel channels, if required.
Fuel pusher: A flow-through fuel pusher resides upstream of the fuel bundles in the inlet end fitting and partially envelopes the inlet shield plug. During on-power refuelling the pressure drop across the fuel pusher is used to provide the push force on the fuel string. Unlatching the latch spacer plug allows the fuel string to move into the fuelling machine attached to the outlet end fitting. Latched spacer plug: A latched spacer plug is required in the outlet end fitting in addition tc the shield plug. It latches into the baffle liner in the sideport region to support the fuel string after the channel closure is removed.
2.4 Advantages of Shop-Assembled Channel Module The shop-assemMed channel module design adopted for CANDU 300 and described in Section 2.3 has the following advantages over current fuel channel c'^signs:
required at the reactor face, after accounting for the fewer channels in CANDU 300.
2.
Calandria tube replacement concurrently with pressure tubp replacement. This is important in performing repeated channel replacements to achieve the 100 year station life.
3.
Shop assembly minimizes delays in the schedule due to non~conformance of components or assembly.
4.
Installation of replacement channels will use essentially the same tooling and procedures used in the initial installation. However, during the initial installation of channels, some tooling operations may not use the remote feature because of tight schedule constraints. In most cases the knowledge and experience gained during initial installation will be of direct benefit to channel replacement work because the operations in the two cases are very similar.
5. The same tooling and procedures are required for single channeL replacement and lar&e scale channel replacement, with some exceptions (e.g. shielding provisions *nd working platform access). 6.
Significantly faster initial channel installation and large scale channel replacements because of minimizing activities at the reactor face, as follows: •
«
All phases of work during channel removal, channel disposal, and channel replacement requires the handling of complete channel modules instead of many components and subassemblies. The plan is to use the Fuel Channel Replacement Cradle (see section 3) to remove and replace 3 channel modules simultaneously to reduce the number of times the reactor vault is cleared. All activities for assembling the channel, including insp^^ion and leak-tightness activities, are transferred to the shop instead of the traditional method of assembly at the reactor face. For example, annulus spacer locations and rolled joint leak tightness are confirmed during final inspection in an off-reactor facility prior to bringing new channels into tte seacxot: vault £ot
Reductions by a factor of 3 to h are anticipated in the number of tools and inspection equipment and in the activities
274 CNS 9th ANNUAL CONFERENCE, 1988
Removal and disposal of the one-piece module minimizes the contamination to be dealt with during channel replacements.
2.5 Design Improvements The following summarizes the important design improvements to be incorporated in the fuel channel design for CANDU 300, to extend fuel channel life: 1.
1.
site the
Pressure tube leak detection: Fast leak detection and leak location in any channel for any size leak is achieved by continuous monitoring of moisture content and rate of change of dewpoint of C03 gas in the pressure tMbe-t;2iTi&Tidfi.& tibe s m » l > » . This ensures detection and appropriate operator action well before a crack from a pressure tube defect grows to an unstable size (critical crack length). The annulus gas system configuration links pairs of channels directly to the leak monitoring system to achieve rapid leak detection. The annulus gas system improvements are described in Reference 1.
2. Pressure tube-to-ealandria tube contact: The snug spacer design, optimized spacer locations, factory installation afld final off-reactor inspection of spacer locations prior to installation provide assurance that the tubes do not contact. 3. Reducing hydrogen ingress into pressure tubes: Very large research and development programs are underway to seek ways of reducing hydrogen ingress into pressure tubes, for example: use of. ytttium sinks (2J •, coatings £ot ewi fitting, rolled joint hubs; chemistry control of annulus gas and the heat transport system coolant; etc. Recommendations from these programs will be considered for implementation in the CANDU 300 program.
4. Improving pressure tube design life: A number of research arid development programs sponsored by the CANDU Owners Group (COG) are aimed at developing pressure tubes with a longer life. Any recommendations from these programs that are shown to improve channel life will be considered for implementation in the CAND'J 300 program. 5.
Calandria tube integrity: Increasing the probability of calandria tube survival following a pressure tube failure is a priority in CANDU 300. The following improvements are under investigation: Providing oxide coatings and roughening the calandria tube inside diameter to improve heat transfer properties, to increase the probability of calandria tube survival under certain accident scenarios. Fuel channel bearing clearances to be defined on the basis of improving calandria tube survival under accident scenarios of pressure tube failure.
to work vault.
The station design integrates the following aspects of channel replacement: ° access and vault requirements ° replacement tooling conceptual design ° facilities required for channel replacement ° requirements on interfacing systems and components ° minimizing operations in active areas. The following briefly describes the provisions integrated into the CANDU 300 station design to facilitate repeated, rapid, large scale fuel channel replacements: 1) Shop-Assembled Channel Module: The advantages of the shop-assembled channel module in improving channel replaceability are discussed in Section 2.4. 2) Fuel Channel Replacement Cradle: The removal of irradiated channels from the reactor and the transportation of new channel modules and aligning them with the reactor face is to be performed by a remotely-operated fuel channel replacement cradle. This cradle will be mounted on the fuelling machine carriage in place of the fuelling machine head and will utilize the normal fuelling machine carriage drive and locating mechanisms. The cradle will be capable of alignment with three adjacent fuel channel positions on the reactor face and to simultaneously carry three channel modules, b o t h for channel removal and c h a n n e l replacement. 3) Full-Width Shielding in Vaults: Provision of full-width shielding connected to the feeder cabinets in the inlet and outlet vaults will provide a large work area and allow two crews
at any level
in each
4) Inlet and Outlet Vaults: The plan is to have the inlet vault accessible at all times during normal channel replacements and to provide it with fresh air. The outlet vault will not be accessible for the period between channel modules being pushed into the replacement cradle and the cradle returning to the vault with new modules. Normally, the inlet and outlet vaults will be isolated from each other but an access between the two vaults is being considered. 5) Working Platforms in Inlet and Outlet Vault: Light-weight working platforms spanning the full reactor face will be used in the inlet and outlet vaults to enable working at different levels of the reactor face. These platforms can be lightweight because they do not have to support shielding (see Item 3 ) . 6.
Minimizing Airborne Contamination: During large scale channel replacements the calandria will be maintained under a small negative pressure to draw airborne contamination (released during the replacement operation) into the building exhaust ventilation system. This will be done by installing temporary ducting between the calandria relief lines and the building's exhaust ventilation system.
7)
Disposal Facility for Irradiated Channels: The plan is for the replacement cradle to discharge thf: irradiated channels directly into the disposal facility through a port in the reactor building wall.
8)
Mandrel Operation: Hydraulically-actuated mandrels connected end to end will be used to push irradiated channels out of the reactor and to pull in new replacement channels to be installed.
3.0 FUEL CHANNEL REPLACEMENT Fuel channel replaceability has the highest priority in the CANDU 300 design. Fuel channel replacement requires the removal and replacement of 10 meter long channel modules. The design target for performing large scale channel replacement is 3 months.
side-by-side
9) Channel Removal Operations: Channel removal only requires detaching AGS tubing connections, cutting CTX welds and disconnecting the feeder connections at the inlet and outlet ends, before pushing the channel module into the replacement cradle as illustrated in Figure 6. 10) Channel Installation Operation: The operations for positioning and attaching new channel modules in the reactor assembly essentially reverses the sequence of the channel removal operations described above. The welds are to be made with the same automatic welding equipment used in the original installation. Feeder connections are connected in the same manner as in the original installation, except that the feeder pipe connection at the inlet end is to be made with manipulators operated from the inlet vault. 11) Computer Simulation of channel Replacement: Computer simulation methods used to optimize industrial processes are being used to model the work environment and the channel replacement scheme to "test" the feasibility and duration of the replacement operation early in the design p h a s e , as discussed in Reference 3. These computer simulation techniques together with data from mock-up studies will be used to improve the efficiency of the channel replacement operation. CNS 9th ANNUAL CONFERENCE. 1988 275
CANDU 300 INLET END FITTING
i
BELLOWS
INLET CTX
CALANDRIA TUBE
I
/
1
PRESSURE TUBE
/
ACL INTERNALS IN PLACE INCLUDING NEW FUEL DURING INSTALLATION
JL
CHANNEL MODULE READV FOR INSTALLATION IN REACTOR
INLET END SHIELD
'/ f f
CAUNDniA
"T V T. 1 ° " T L E T
OUTLET FEEDER.. PIPE
E
ALL INTERNALS IN PLACE (NCLUDJNG FUEL DURING INSTALLATION
Ivf/
^-V-^
AO TUBI
FUEL CHANNEL ASSEMBLY INSTALLED IN REACTOR DISCONNECT FEEDER PIPE AND SPRING FE60ER TO SIDE
*£\ [2J REMOVE CTX TO LATTICE TUBE WELD
( I ) MANDREL ATTACHED V / TO CTX TO PHOTECT ; BELLOWS \ ^
i CUT CTX TO \S3> / LATTICE TUBE WELD DISCONNECT FEEDER CONNECTIONS AND SPRING FEEDERS TO SIDE
,.,.-..•-- - r r
IT--
-
SHIELD PLUGS AND FUEL PUSHER IN PL>r.E
EXTENDABLE TUBULAR MANDREL V ^ H - i ' •_ ' • -Ui CARRIES END OF E/F THRU TO DISCONNECT ( \ ) OUTLET CALANDRIA TUBE SHEET AG TUBING ' — ' IN 3 SECTIONS 3 0 m LONG SUPPORTED IN INLET LATTICE TUBE OPERATIONS ON INSTALLED CHANNEL ASSEMBLE PRIOR TO CHANNEL REMOVAL
F/c
REPLACEMENT CRADLE pOS|T|ON T o RECEIVE IRRADIATED CHANNEL |N
FIGURE 6 FUEL CHANNEL REPLACEMENT
4 . G SUMMARY
5.0 CONCLUSION
The fuel channel design in CANDU 300 is governed by station requirements for a fast construction schedule, high channel power and a high station capacity factor while achieving significant reductions in capital and operating costs over current designs. The proposed pre-assembled channel module design comprising a fuel channel-calandria tube assembly that can be inserted into the reactor structure from the outlet vault is achieved because of the single-ended refuelling feature adopted for CANDU 300. Other design improvements are being incorporated in this channel design to enhance both reliability and channel life.
The fuel channel design for CANDU 300 is expected to be effective, reliable, and be capable of faster replacement to allow repeated channel life extensions, to achieve the 100 year station life specified for CANDU 300 stations.
The shop-assembled module concept reduces the number of components in the channel assembly and simplifies channel installation and replacement. It offers significant cost savings during initial construction. The major savings of this modular design are realized in the dramatic reduction in station outage achievable for large scale fuel channel replacement. The shortened schedules for initial installation and for large scale channel replacement flow from the modular construction feature which requires handling single channel modules for all phases of work and minimizes the operations and activities in the reactor vault. The reduction in activities at the reactor face and the much shorter schedule also significantly reduces man-rem exposure during large scale channel replacement.
276 C N S 9th ANNUAL CONFERENCE, 1988
6.0 REFERENCES (1) B.A. SHALABY, E.G. PRICE, C D . MOAN, C.E. COLEMAN, "Leak-Before-Break and Leak Detection Systems of CANDU Fuel Channels", Ninth Annual Canadian Nuclear Society Conference, Winnipeg, Manitoba, June 12 to 15, 1988. (2) R. DeGKEGORIO, I. GRANT, et al, "The Fabrication of Hydrogen Sinks for the Pressure Tubes of Darlington Unit A", Ninth Annual Canadian Nuclear Society Conference, Winnipeg, Manitoba, June 12 to 15, 1988. (3) J.R. CANDLISH, K.C. DALTON, B. MARSHALL "Computer Simulation of Fuel Channel Replacement", Ninth Annual Canadian Nuclear Society Conference, Winnipeg, Manitoba, June 12-15, 1988.
COMPUTER SIMULATION OF FUEL CHANNEL REPLACEMENT J. R. CANDLISH K. C. DALTON B. MARSHALL ATOMIC ENERGY OF CANADA LIMITED, CANDU OPERATIONS Sheridan Park Research Community Mississauga, Ontario L5K 1B2 (416] 823-9040
ABSTRACT Fuel Channel replacement can extend the life of existing CANDU reactors but has previously been time consuming and expensive. Simulation of CANDU fuel channel replacement can reduce the cost of developing faster methods of replacing channels in new and existing reactors. Computer simulation software for modelling the work environment and replacement process is now available on v=ry powerful medium cost votV.st.&t.iOTis. Thfc t«;teiqi«s ot model. VLng ami expected results are described in this paper in the context of the planned retubing process for the CANDU 3 reactor. INTRODUCTION Many Nuclear power stations will, during the next 20 years reach design lifetimes. Rebuilding key plant components within the nuclear island to extend the life of these stations has tremendous potential compared to building new facilities. It also presents a considerable challenge to utilities and nuclear engineers to determine safe and adequate measures for rebuilding or reconditioning existing plants. The CANDU reactor is uniquely suited to life extension because its most highly irradiated components, fuel channels, are small enough to be disassembled and replaced within the reactor building. While the feasibility of this process has been well established, there is still much to be done to reduce the cost of this process. Components which initially cost less than 20 million dollars will cost the utility ten to twenty times as much to replace in labour, tooling and station downtime. To minimize this overhead portion of fuel channel replacement cost, it is usual to build realistic simulation facilities to prove tools and procedures. These simulations while invaluable for training and tool proving do not provide early feedback to the designer about deficiencies in the planning of work flow during fuel channel replacement.
ADVANTAGES OF COMPUTER SIMULATION The following discussion of the advantages of a computer simulation is paraphrased from the AutoMod user manual by Auto Simulations Inc. A simulation is the imitation of the operation of a real-world process or system over time. A system is studied over time by developing and exercising a simulation model• A computer simulation model usually takes the form of a combination of logical/mathematical assumptions, graphical or physical representations that define the relation between system elements and completely describe system operation.
The simulation model can be thought of as a laboratory model on which the modeler can conduct experiments: - on a proposed or designed system, before building it; - on an existing system, without disturbing it. The great advantage of simulation is that a system design or a potential change in system operation can be tested prior to its implementation. In almost all cases, developing and exercising a simulation model is less disruptive and much less expensive than experimenting with the real system or with an accurate physical replica. The following simulation:
are
some
uses
of
computer
1.
Simulation enables the study of, and experimentation with, the internal interactions of a complex material handling or manufacturing system.
2.
Simulation can be used to test a design of a new system to identify problem a r e a s , thus suggesting design improvements, before the system is built.
3. Simulation can be used to test and improve management strategies, production schedules, control methods such as algorithms, job sequencing and workforce scheduling, before they are implemented. h. Simulation can be used to study the effect of changes in the manufacturing process due to alternative tools and work methods. 5. By varying system parameters, simulation can be used to identify the critical elements and factors that are most important for efficient system operation. 6.
For the design of new systems, simulation modelling allows the design to be tested, add improved, before the real system is constructed. Without simulation, a system design can lead to unforeseen problems that result in an inevitable redesign requiring expensive changes to the system after it is built. CNS9lh ANNUAL CONFERENCE. 1988 277
7.
The knowledge gained in developing a simulation model may be of great value to the client and modeler in gaining an understanding of, and suggesting improvement in, the total system. In large-scale complex systems, rarely does one find a single individual who understands how the whole system operates. The mere act of bringing together diverse sources of information and knowledge in order to make a simulation model possible and realistic is of great value in itself in educating personnel concerning the complex interactions in a large system.
Computer simulation tools which are used to optimize industrial processes can be effectively adapted to study fuel channel replacement.
FACTORS AFFECTING CHANNEL REPLACEMENT DURATION. The process of channel replacement as depicted in Figure 1 can be broken down into the tasks o£ moving men equipment and materials, assembly and disassembly tasks and inspection and work control activities. There is also a preparation and a later commissioning phase srtiich have not been addressed in this study. The basic principles of time and motion study can be applied to assembly and disassembly tasks to make the tasks as efficient as possible and cools can be optimized and perfected to be as easy to use and as reliable as possible, but the overall duration can not be substantially reduced unless the work flow is studied and planned at the basic level. For example, how can the number of tasks be minimized, what is the maximum number of tasks which can be effectively done in parallel? What are the tradeoffs between complexity of tasks and numbers of operations or between time saved by parallel operations and time wasted by congestion of the workspace? Our ability to answer these kinds of questions is limited by a lack of experience in performing these tasks. Accordingly, unless we can differentiate between different ways of performing tasks and estimate the times accurately, there is no way to make good decisions in the early planning stages of a project as to the best approach to the replacement task.
CANDU 300 BELLOWS
rH
, •mirew,SH,a?
.
f-ia
CAL.
. • o u ^ T t m SHIELD ;
f H
L INTERNALS IN PLACE INCLUOING FUEL
u-:•••-. FUEL CHANNEL ASSEMBLY INSTALLED (N REACTOR
DISCONNECT FEEDER PIPE AND SPRING FEEDER TO SIDE wDEl
TTACHED
TO CTK TO PROTECT BELLOWS
I r t : - ; - , - - •-...-• ••••rr SHIELD PLUGS AND FUEL PUSHER IN PLACE
.l
EXTENDABLE TU9ULAH MAIMOREl CARRIES END OF EfF THRU
•
TO
OUTLET CALANDRIA TUBE SHEET IN
^
^
V4-
••'
'
.
.
*
•
••:•:
•
' : : • • ; ; • • • ' I
•
\HORt* TUBE
INLET END
I
/ FIVllNG
ALL INTERNALS IN PLACE AhD WITH NEW FUEL
CHANNEL MODULE FOR INITIAL INSTALLATION AND CHANNEL REPLACEMENT
FIGURE 1
CNS 9th ANNUAL CONFERENCE, .
FUEL CHANNEL REPLACEMENT
I
' •H - 4 -
OPERATIONS ON INSTALLED CHANHEl. ASSEMBLY PRIOR TO CHANNEL REMOVAL
SUPPORTED IN INLET LATTICE TUBE
278
'
OISCONNECT AG TUBING
] SECTIONS 3 0 m LONG
METHODS AVAILABLE PLANKING
FOR
FUEL
CHANNEL
REPLACEMENT
The appropriate methods for making good decisions about optimum work methods are the same techniques which are applied to the planning of manufacturing processes. These include time and motion studies, process charts and flow diagrams which optimize the choices of work layout, work methods, inventory management and transportation. It is also necessary to determine and minimize the effect of equipment breakdown. Although in a manufacturing environment considerations such as inventory costs are important, for the nuclear maintenance application the emphasis shifts to minimizing downtime and minimizing radiation worker exposures and making best use of the limited available space. The first stage in designing the fuel channel replacement process is to describe the process in terms of a sequence of operations. The initially proposed list of operations for the CAND'J 3 fuel channel is presented in Appendix I. In addition to the operations there are a series of setup and transport activities which involve moving from channel to channel, row to row and in and out of the vault with equipment, materials and men. To define these setup and transport activities a workspace layout is required. The proposed work layout must be within the constraints of the reactor vault and at the same time provide maximum space for the work crews. Computer modelling of the reactor vault allows the engineer to design work layout in a three dimensional environment and to visualize the utility of the space in conjunction with tools, materials and crews. Modern computer graphics hardware in conjunction with software designed to take full advantage of the hardware capabilities will permit interactive walk through of the r oposed work layout. The shaded perspective and wire frame presentations allow for realistic and see through representations of the model. Slides accompanying this presentation show the results of modelling of the CANDU 3 vault using the DENEB IGRIP software and a Silicon Graphics workstation. The next stage in the planning process is to define the work flow by combining the operations and the set up and transport activities. In this example the assumption is that the preparation and commissioning phases of the work are planned separately independent of commencement and completion of work on the reactor face. An example of part of the proposed work flow for the CANDU 3 retubing is shown in Appendix II. To perform a computer simulation of this activity this work flow was converted into program code used in the AUTOMOD programming language. The AUTOMOD language is geared to simulation of automated industrial processes. The terminology used in the AUTOMOD code refers to loads, movement systems resources and processes. A typical model deals with a process in which a load representing any component is being moved between different points in a factory or warehouse. In our application, the component is fixed and the resources must compete for the space in the vicinity of the part. Of course the channel is ultimately moved out of the core and replaced by a new channel but the logistics of most of the tasks are concerned with working on the channel in place.
The movement systems are equated to either the workers or the cranes used to move heavy loads around on the platform. For the inlet vault operations the loads are equated to the tools or replacement parts which are carried to the work location. Since the worker is both a movement system and a resource, the concept of a resource is not used. The operations performed on each channel are defined as processes having a certain duration which may be random and having a probability of needing to be repeated. While an operation is being performed on one channel, it is not practical to work on any of the adjacent channels. Shielding panels and workers would block the access to three or four channels on either side of the present work location. The AUTOMOD simulation also provides the facility to claim or block a certain physical space while a process is ongoing. This part of the model coding is done non-interactively and requires the design team to understand the concept of the system and its boundaries. In this study the inlet vault work platform has been looked at rather than the overall fuel channel replacement process. Its interactions with outside processes are treated as Inputs and outputs. The possible difficulties with a simulation analysis process are that the modelling process can be time consuming and costly both in manpower and computing resources. To reduce this difficulty the model must be planned in a way that addresses the specific simulation needs but does not add unnecessary complexity to the model.
RECOMMENDED MODELLING STEPS 1.
Formulate a problem statement which is clear to both the modeller and policy makers based on a detailed description of the problem and its background.
2.
Ask what questions are to be answered by the simulation. At this point it should be decided whether simulation is an appropriate tool.
3.
Collect data concerning the factors that must be specified to define the simulation model.
A.
Code a simple model of the movement system and processes and gradually increase the complexity of the model. Where detail is not required, treat a process as a black box always avoiding unnecessary complexity.
5.
Verify the model to ensure accuracy and to ensure that code executes as intended and that the model adequately addresses the problems it was designed to solve.
6.
Validate the model by comparing the model with the real life process it wfs constructed to represent.
VALUE OF INTERACTIVE MODELLING While simulation coding is a specialist activity, the validation of the model is best done by tooling designers, operations staff and the cor-1 "uctlon staff with direct experience in the jl channel replacement. The simulation that is provided by AUTOMOD using suitable workstation hardware is an animation which can be run in real time or be accelerated or dec .erated to suit the viewer. This
CNS 9th ANNUAL CONFERENCE, 1988 279
greatly facilitates reviev of the model by the non-specialist as veil as providing an excellent debugging tool. The graphic display of the simulation is an important element because for many participants in the review it allows them to use their ability to recognize a correct or incorrect process. This can significantly increase the productivity of the model validation process and promote input to a simulation from those who understand the process best. The validation of a model represents its acceptance as a real measure of our ability to perform fuel channel replacement on a CANDU reactor during an outage. The cost of unnecessary outage time is an important issue to all existing nuclear plant operators. One design objective for the CANDU 3 is to ensure that methods of minimizing outage time are carefully considered during the design phase - a time when subsequent outage times can be influenced significantly. To do this we will need methods of measuring outage time which are both acceptable to clients and allow them to have input to the design. Graphic display of a simulation will be very useful in this respect. RESULTS OF SIMULATION The questions that are to be answered by this simulation can be best summarized by asking what is the shortest practical outage duration which can be achieved with the lowest practical radiation dose to workers. If we are to evaluate the merits of different schemes for fuel channel replacement, this is the basic question that needs to be answered. It is really two interrelated questions; what is the outage duration and what is the radiation dose to workers, both of which need to be minimized. These same questions will be addressed by rurrent methods of planning reactor retubing. The computer simulation's chief advantage is thtt it allows tb.3 cesi^ner to address questions which involve a lot of detailed computation and tabulation which the computer d'-es very well - questions such as: 1.
What are the accumulated radiation doses to each worker based on an assumed radiation field,
2. What is the effect on duration of outage of random failures of tools or processes ort in other words, what is the value of varying degrees of reliability of equipment,
280 CNS 9th ANNUAL CONFERENCE, 1988
What is the optimum number of workers in a given space to minimize the congestion of the work area 4.
What is the effect on outage of different shift and break schedules.
REPORTING FACILITIES As part of the simulation results, reports can be generated summarizing statistics about the durations of processes and time spent moving from point to point. Radiation dose estimates can be tabulated for each worker by assigning dose rates for different locations in the work area and creating reports for cumulative duration of each worker in the active area which take into account movement within the work area and breaks in the work schedule. CONCLUSIONS If reducing reactor outage is an important issue to utilities and designers alike (and we think it must be) then an incentive exists for simulation of reactor maintenance to reduce outage time in practice. Little attention has been paid to mathematical simulation. Host emphasis has been placed on hardware simulation. While this is the most convincing and useful approach for fine tuning the replacement tasks it usually occurs after many basic design decisions concerning the replacement process have already been made. What is needed is a tool which can measure replacement efficiency economically and with some acknowledged authority early in the design phase when the greatest potential for reducing the retubing schedule exists. The validity of the model is very important since design decisions will need to be made based on the results. Because of this a graphic display of the process is needed to permit non specialists in computer modelling to understand and evaluate the simulation results. As the design progresses computer simulation combined with data from mockup studies will provide an increasingly sophisticated, powerful and realistic model for evaluating process alternatives. The CANDU 3 design has established a target of 90 days as the replacement time for any major piece of equipment including all 232 fuel channel assemblies. Computer simulation techniques are helping us ensure that these targets can be achieved.
APPENDIX I CANDU 300 FUEL CHANNEL REPLACEMENT
VAULT OPERATIOHS:
VAULT OPERATIONS; OUTLET VAULT
INLET VAULT
1.
Remove insulation panels
1.
Remove insulation panels
2.
Cut inlet feeder pipe
2.
Disconnect gas annulus tubing interconnects
3.
Spread feeder pipe
3.
Disconnect feeder coupling
4.
Disconnect gas annulus tubing
A.
Spread feeder pipe
5.
Cut CTX weld slot-rotary milling cutter remote
6.
Cut CTX weld undrrcut - freeing outlet end
7.
Shift channel first 0.2m
5.
Install first mandrel section
6.
Cut CTX weld
7.
Shift channel 0.2m
8.
Installraandrelsections and shift channel in mandrel length increments
8.
Attach removal can Withdraw channel in mandrel length increments Install replacement channel in mandrel length increments
9.
Receive replacement channel and remove mandrel sections
9.
10.
Unspring feeder pipe
10.
11.
Attach feeder pipe with cryfit sleeve
12.
Reweld CTX to lattice tube
13.
Attach gas annulus tube
14.
Replace insulatior panels.
11.
Fine position to attach feeder coupling
12.
Unspring feeder pipe
13.
Connect feeder coupling
U.
Weld CTX to Lattice Tube
15.
Connect gas annulus tubing
16.
Replace insulation panels
CNS 9th ANNUAL CONFERENCE, 1988 2«1
APPENDIX I I CAUDU 3 FUEL CHANNEL REPLACEMENT PROCESS FLOWJilMiRMi:
IHLET VAULT OPERATIONS
TO SHEET
SHIFT CHANGE AND CLEAN UP
n REMOVE IMSULATION PANELS ARE CUTS CONFIRMED?
TRANSPORT FEEDER CUTTING & SPREADING TOOLS TO WORKSTATION
CURRENT ROWS COHPLETE BOTH LEVELS? DISCONNECT GAS ANHULUS TOBINO
CURRENT ROMS COHPLETE BOTH LEVELS?
SHIFT CHANCE AND CLEAN UP AS REQUIRED
ALL DONE?
SHIFT CHANCE AND CLEAN UP
TYES REHOVE CTX CUTTING EQUIPMENT
TRANSPORT HANDREL SECTIONS TO WORKSTATION ON UPPER PLATFORM ONLY BEHOVE FEEDER CUTTING AND SPREADING TOOLS
ri TRANSPORT REQUIRED" NO OF FIRST MAWDREL SECTIONS TO WORKSTATION FOR SHIFT UP TO 16 FOR EACH PLATFORM
PULL CHANNELS INTO REMOVAL CANS IN 2.0M INCREMENTS
INSTALL HANDREL OH EHD OP F / C •; JHPLETE WITH CTX WELD CUTTIHG TOOL LOWE;III PLA.TORH 1 ROW
PUSH NEW FUEL CHANNELS INTO PLACE
I
CUT CTX HELD AND REMOVE CUTTEK
R£MOVE ACS TUBING AHEAD OF CURHEHT DETACH MANDREL SECTIONS AND PLACE ON CARRIER
OUTLET CTX WELD CUT COMPUTE)
REPLACE CTX WELD COTTER
PUSH CHANNEL FROH INLET TOWARD OUTLET JDO MH TO CONFIRM CUTS CUTTING TOOL
UOVS TO NEXT CHANNELS
SHIFT CHANGE
ALL CHANNELS III ROW COMPLETE?
2S2
CNS 9th ANNUAL CONFEHENCE. 1988
ULTRASONICALLY INSPECT CTJt WELD
IDENTIFY AND TAG DEFECTIVE WELDS
REMOVE FI8ST MANDREL SECTIONS FROM REPLACEMENT CHANNEL (PERMANENT SHORT MANDREL REMAINING
LOWER PLATFORJf I SOW
SHIFT CHANGE
CURRSBT ROW COMPLETE? BOTH LEVELS? UNSPRINQ FEEDER PIPES
I ALL ROWS COMPLETE? EOTH LEVELS?
REPOSITION CHANNELS FOR FINAL INSTALLATION AND INSTALL CRYOFIT SLEEVES
REPLACE GAS ANHULUS TUBIHG
SHIFT CHANCE AS REQUIRED
LOWER PUTFORM 1 HOW
SHIFT CHANGE AND CLEAN UP
TRANSPORT HAJJDREL SECTIONS TO DISPOSAL/STORAGE IN CROUPS OF UP TO 16
*—NO
CURRENT ROH COMPLETE BOTH LEVELS?
CURRENT ROW COMPLETE? BOTH LEVELS?
YES ALL ROVs COMFLETE BOTH LEVELS?
CUT AND REWELD DEFECTIVE CTX WELDS
YES
LOWER PLATFORH 1 ROW
ALL ROWS COMPLETE? BOTH LEVELS
I YES RAISE PLATFORM TO ROW A RAISE PLATFORM TO ROW A
WEto CTX TO U T REPLACE INSULATION PANELS
LOWER PUTFORM 1 ROW
VISUALLY INSPECT WELD REMOVE ALL EQUIPMENT FROM VAULT ROW COMPLETE? LOWER PUTFORH TO STORAGE POSITION
CNS 9IH ANNUAL CONFERENCE. 1988 2 ( 3
A CANDU DESIGNED FOR MORE TOLERANCE TO FAILURES IN LARGE COMPONENTS
N.J. SPINKS-, F.W. BARCLAY**, P.J. ALLEN***, F. YEE*
Atomic Energy of Canada Limited -'Advanced CANDU Project Chalk River Nuclear Laboratories Chalk River, Ontario KOJ 1J0 Telephone: (613) 584-3311 X 3573 "Whiteshell Nuclear Research Establishment Pinawa, Manitoba ROE 1L0 ***CANDU Operations Sheridan Park Research Community Mississauga, Ontario L5K 1B2
ABSTRACT Current designs of CANDU reactors have several groups of fuel channels each served by an upstream coolant supply-train consisting of an outlet header, a steam generator, one or morg pumps in parallel and an inlet header. Postulated failures in these large components put the heaviest demands on the safety systems. For example, the rupture of a header sets the requirements fcr the speed of shutdown and for the speed and capacity of emergency coolant injection, and it has a large impact on containment design.
The fuel channel and all piping between headers is of small diameter (up to 10 c m ) . The supply train piping can be larger (up to 40 cm or more, depending on the size of the reactor). Postulated failures in the larger-diameter piping set many of the requirements for design of the safety systems: the speed of shutdown, the speed and capacity of emergency coolant injection and the leak tightness of containment.
STEAM GENERATOR
A CANDU design is being investigated to reduce the impact of failures in large components. Each group of fuel channels is supplied by more than one train so that if one train fails the rest continue to work. Reverse flow limiters reduce the loss-ofcoolant from the unbroken trains to a broken supply train. The paper describes several design options for making the piping connections from multi supplytrains to fuel channels. It discusses progress in design and testing of flow limiters. A preliminary analysis is given of affected accidents.
•^FEEOERS INTRODUCTION A change in the design of the CANDU primary heat transport system is being investigated that could improve the tolerance of the plant to large pipe failures and could reduce the reliance on the special safety systems. The CANDU reactor has four special safety systems that are independent of the process systems and independent of each other. These are the tvo shutdown systems, the emergency coolant injection system and the containment system. The CANDU primary heat transport system is comprised of several groups of fuel channels, ea<-'h group served by an upstream coolant, supply-train consisting of an outlet header, a steam generator, one or more pumps tn parallel and an inlut header. There aro about 100 fuel channels per group connected to the headers by end fittings and feeder p tpes.
284 CNS 9th ANNUAL CONFERENCE, 1966
FIGURE 1:
A CANFLG TWO-TRAIN CIRCUIT
In the new design, called CANFLO, each fuel channel is supplied from more than one train so that if one train fails, the rest continue to provide at least some coolant to the channel. Reverse flow limiters at the inlet reduce the loss-of-coolant from the unbroken trains to a broken supply train. The loss Is intended to be less than the flow from the unbroken trains so as to maintain forward flow in the affected group of fuel channels, Figure 1 Illustrates the concept in a simple-circuit, twosupply- train design. The cross-connection downstream of the main inlet headers could be a new set of headers, of smaller diameter than the main headers, each supplying a small number of fuel
channels. Another option is to supply each channel with two inlet feeders, each being connected to a different supply train. The latter approach was adopted for the CIRENE pressure-tube reactor near Rome. The concept of minimizing reverse flow by the use of special devices in coolant supply-lines has been developed for the advanced gas-cooled reactors in the United Kingdom (4). Figure 1 shows the use of small headers and forward-flow limiters near the outlet headers. Theje are intended to further reduce the rate of coolant loss to a broken supply train.
PUMPS INLET HEADERSINLET FEEDERS -
By reducing the core voiding rate for the postulated large-break loss-of-coolant accident (LOCA), the CANFLO design would have a reduced rate of void reactivity addition and would therefore require a reduced rate of reduction of reactivity via the shutdown systems. Fuel temperatures during a postulated largebreak LOCA can be high enough to put a limit on core-power density drring normal operation. By maintaining forward flow, even for the large break, the CANFLO design avoids the conditions which lead to the high temperatures. Consequently this restraint on power density can be removed to allow the reactor power to be uprated with consequent reductions in specific capital cost. Reduced LOCA. temperatures, even for an uprated reactor, lead to reduced releases of radioactivity into containment. With CANFLO, the reduced rate of loss-of-coolant would permit a reduction in the speed and capacity of emergency coolant inject ion. Also the reduced rate of release of energy and radioactivity into containment would reduce the release of radioactivity from an impaired containment. In the following two sections of the paper, options are discussed for the design of the heat transport system and for the reverse-flow limiters. A test program has been initiated to assess various designs of flow 1imiters and this is described. The paper concludes with some performance predict ions.
OUTLET FEEDERS
/CHANNELS
FIGURE 2:
THE CONVENTIONAL CANDU FIGURE-OF-EIGHT CIRCUIT
A CANFLO system with four pumps and four steam generators could have several arrangements: a simple circuit with four supply trains in parallel, a single figure-of-eight circuit with two supply trains in parallel as shown in Figure 3 or two simple circuits like Figure 1 with two supply trains in parallel. The design with four trains in parallel would be most effective in maintaining a forward flow to the core in the event of a LOCA but the header to feeder connections could be very complicated. The two-circuit design should have a residual advantage from the void-reactivity viewpoint. The figure-of-eight design, being a single circuit, could be the simplest from the design viewpoint.
DESIGN OPTIONS A conventional CANDU heat transport circuit, shown in Figure 2, has two supply trains and associated channels in series, the so-called figure-of-eight design. It provides pump redundancy for events involving failure of a single pump and, compared to a parallel-pump arrangement, (reduces the positive void-reactivity effect of a LOCA by ensuring that half the channels of the circuit are less directly effected by a break. The CANFLO design of Figure 1 achieves pump redundancy with two pumps in parallel and the flow limiters reduce the rate of core voiding, hence the rate of addition of reactivity during a LOCA. The above two-pump two-steam-generator designs of Figures 1 and 2 are suited to reactors with smaller power output. Larger numbers of pumps and steam generators are needed for larger outputs to 1imit the size of these large components. The CANDU 600 usos four pumps and four steam generators In an arrangement of two figure-of-eight circuits. The division Into two circuits further reduces the positive void ro/JCtivity effect of a LOCA.
FIGURE 3: A CANFLO FIGURE-OF-EIGHT CIRCUIT The benefits of a single circuit design are in simplifications to the pressure and inventory control, emergency core cooling, shutdown coo A ing and purification systems. The two-circuit design requires interconnecting piping and valves to service the two circuits with common systems. These can be eliminated with a single-circuit design.
CNS 9th ANNUAL CONFERENCE. 1988 285
Another CANFLO option is shown in Figure 4. It is called the figure-of-800 because it has features common to the simple circuit of Figure 1 and the figure-of-eight circuit of Figure 2. Its advantage over other four-supply-train options is that, in any pair of supplies to a group of channels, the two sources come from divergent parts of the cireuit. This reduces the effect on the good train of a break in the other train of a pair. Figure 4 is drawn without flow limiters at the outlet. They are expected to be less effective in the figure-of800 configuration.
o
MAIN HEADERS
o
FLOW LIMITERS~ SMALL HEADER^
FEEDERS TO FUEL CHANNELS
FIGURE 5:
FIGURE 4:
THE CANFLO FIGURE-OF-800 CIRCUIT
A CANFLO arrangement of two trains in parallel permits a selection of two pumps plus four steam generators or four pumps plus two steam generators, if such options prove to be desirable given the required reactor power output and given the availability of these large components. Similarly a three-train design with three pumps and three steam generators is an option that is not available with conventional CANDU figure-of-eight circuits. The conventional figure-of-eight CANDU design avoids the need for return piping from one face of the core to the other. Refuelling is done at both faces. However, to reduce fuel handling equipment costs and to facilitate fuel-channel replacement, refuelling at only one face is being considered for advanced designs and return piping is needed as in the CANFLO design of Figure 1. All channels would be refuelled at the outlet so that all outlets are at one face, the refuelling face, and all inlets are at the other face. This leaves the inlet face relatively clear for the extra CANFLO inlet piping. There are two arrangements of the CANFLO inlet connections currently under consideration. One design uses a set of small headers downstream of the large inlet headers to make the common connection from several supply trains. The inlet feeders are in turn connected to the small headers. A typical arrangement showing five inlet feeders connected to a small header is shown in Figure 5. Cooling flow would still be provided to the fuel channels if a large inlet header were to fail since the flow limiter would limit the flow rate to the broken header. However, there is a possibility of flow stagnation in the channels connected to a small header (the five channels served by the five feeders of Figure 5) if the small header were to fail.
286 CNS 9th ANNUAL CONFERENCE, 1988
A TYPICAL SMALL HEADER ARRANGEMENT
The other inlet arrangement under consideration utilizes two inlet feeders for each channel. The two feeders for a particular channel would be connected to different inlet headers as shown in Figure 6. Each feeder would carry one half of the required channel flow. A reverse-flow limiter would be located in each feeder.
o
HEADERS
o
—FLOW LIMITERS— FEEDERS -
TO FUEL CHANNEL FIGURE 6:
DUAL FEEDERS TO EACH FUEL CHANNEL
The ideas of small inlet headers and dual feeders might be combined in order to minimize the total length of feeder piping. Figure 7 shows fuel channels connected via short lengths of feeder to small vertical headers positioned near the reactor inlet face. However, consideration must be given to the axial elongation of fuel channels which occurs due to fast neutron irradiation and due to thermal expansion. This elongation requires a certain
min imum lengt h of t eoder in t he. plane perpendicular to the fuel channel . Tin; arrangement of Figure 7 would havt1 t o havt* t ht? requi red f lexibi 1 ity Lo accommodate This movement.
FROM MAIN INLET HEADERS I I SMALL T .HEADER
o- -o -o o- -o- -o -o- FUEL
The vortex diode i s morti complex than the ventur i but is more ef feet ive in pert ormance and does not rely on moving parts like the check valve. Pure fluidic devices that utilize vortex flows to induce large d ifferences between thei r forward and reverse~ resistance characteristics have been known for many years (1, 2 ) . A vortex diode of the type developed by Zobe1 (2) is i1 lustrated in Figure 8. Grant and Wright (3) discussed the possible applications of fluidic devices in reactor circuits, and vortex diodes have subsequent ly been i nstal led a.s protect ive devices in the coolant c i rcuits of AGR reactors in the U.K. (4).
NORMAL FLOW DIRECTION
CHANNEL FIGURE
SMALL VERTICAL HEADERS AT THE REACTOR INLET FACE
FLOW L1MITER OPTIONS The purpose of an inlet-flow limiter is to limit the reverse flow to a pipe break but not at the expense of a high resistance to flow in the forward direction. The purpose of an outlet-flow 1imiter in to limit the rat.e of forward flow to a pipe break But again not at. the expense of a high resistance to normal forward-flow rates. Flow 1imi ters must not impact on power production so they must essentially be maintenance free- This means, for example, that they must be resistant LO vi brat ion-induced damage and to erosion/corrosion at high flow velocities. Flow limiters must, be testable periodically for their effect on flow in both the normal and reverse directions but the reliability must be high so that this testing can be infrequent. The normal CANDU channel out let-temperature monitoring system provides the means to check normal flow but an addi t ional test would be needed for reverse flow. If a pump were turned off at low power, the channel out let-temperature monitoring system might again be an adequate indication of excessive reverse flow to that pump, or existing header-to-header pressuredrop instrumentation might be useful. The venturi would seem to be the best device for a flow limiter at the outlet. The impact on the norma1 flow is minimal and choking at the throat would limi t the forward flow. The venturi, the vortex diode and the check valve have been considered for use as reverse flow 1imiters at the in let. The attraction of the ventur L is i ts s implic ity, which should lead to reliability. However, it is not as effective as the other opt ions and would require a heat transport system des i gn with more t nan two supp ly trains in para I Inl.
FIGURE
A ZOBEL-TYPE VORTEX DIODE
For the most part, the work cited above has focussed on the single-phase characteristics of vortex diodes, whereas for a CANDU system we wish to compare the reverse resistance in two-phase (possibly choked) flow with the normal single-phase forward resistance. Mori and Premoli (5) investigated this aspect with a series of experiments conducted in pressurized water. Their results showed an increase in the reverse-to-forward flow resistance ratio when the reverse flow was accompanied by flashing in the diode (6). More recently, diode designs offering increased performance have been reported (7). Thus, there is good reason for optimism that a fluidic diode can be designed with suitable characteristics for use as reverse-flow limiters in a CANFLO system. The check valve, or rather the class of devices that restrict reverse flow by mechanical movement, would be very effective but may require excessive maintenance. Bearing in mind that the leak tightness to a reverse pressure drop is not important, a simple robust design is being sought. An important considerat ion is the frequency of failure leading to reduct ion of normal flow. This has to be extremelv low.
CNS 9th ANNUAL CONFERENCE. 1988 287
In order *o gain further insight into achievable flow-limiter characteristics in a CANFLO system and nlent i fy prospect ive designs, an experimental program is being pursued at the Whiteshell Nuclear Research Establishment. Since we are interested not only in the performance of the device per se, bur a 1 so in its ease of manufacture and compat ibillty with other system requirerents (highpressure operation, avaiIable space, etc.), we expect to test a range of geometric characteristics, including some devices which are more compatible with axial flow geometry than the vortex diode. Testing will initially be done at low temperature and pressure in a transparent facility. Those devices that appear most promising in these tests will then undergo a more rigorous test program, using conditions representative of a CANDU coolant system.
NUMERICAL MODELLING Calculations have been made to compare the LOCA response of several CANFLO options with each other and with a convent ional CANDU reactor. All simulations assumed the same number and size of major components (pumps, steam generators, core) as in the CANDU 600 except for the small increase in pump head required to supply the pressure loss through the flow limiters. The injection of emergency coolant was modelled. Most CANFLO cases used the figure-of-eight geometry of Figure 3. One, case 5, used the f igure-of-800 geometry of Figure 4. All CANFLO designs were thus singlecircuit designs, whereas the conventional CANDU 600 has two figure-of-eight circuits. Details of each case analyzed are shown in Table 1, Several combinations of inlet- and out let-flow limiters were studied. Cases with check valves at the inlet were assumed to eliminate flow lost to a broken train at the inlet. Case 6 not only has check valves at the inlet but is also assumed to have the means (called DP valves) to eliminate flow lost to a broken train at the outlet.
TABLE 1:
SUMMARY OF CALCULATIONS
Case Reactor No. Des ign
Break Size %
InletFlow Limiters
Outlet Ci rcuit Flow Config. Limiters
1 2 3 4
100 10U 100 100 100 100 35
Diodes Diodes Chk valves Chk valves Diodes Chk valves None
Venturis f-8 f-8 None Venturis f-8 None f-8 Venturis f-800 DP valves! f-8 None 2xf-8
5 6 7
CANFLO CANFLO CANFLO CANFLO CANFLO CANFLO CANDU 600
For CANFLO, the 100% in let-header break was expected to give the highest fue I temperatures and so was chosen for the study. For comparison, the .15% in let-header break was chosen for the conventional CANDU 600. This leads to a period of flow stagnation giving maximum fuel temperatures in the downstream fueI-channels. The main tool for the calculations is the CATHENA thermohydrauMcs code (8). It was developed primarily to analyze I.OCAs in CANDU reactors. 288 CNS 9th ANNUAL CONFERENCE, 1988
For subsonic flows through Venturis and for forward flow through vortex diodes, head-loss factors derived from Reference (9) have been used. For reverse flow through vortex diodes, the headloss factor is based on the reverse-to-forward flow characteristics given by Syred and Roberts (7) for a SI DO diode discharging steam-water f1ows from high pressure. (The SJDO diode is an improved version of the vortex diode.) The flow is constrained, where necessary, to remain at or below the sonic velocity. For this purpose, an effective flow area is speci fied, which may be less t han the physical throat area of the component. This provision permits the model 1 ing of choking in components such as vortex diodes. The CATHENA models include a high-power channel in parallel with the group of channels downstream of the broken inlet header. This is expected to have the highest fuel temperatures. Temperatures are calculated for the highest-power fuel element in this hot channel: the hot-channel, hot-pin sheath temperature. Because of the posit ive void coeff ic ient of reactivity, the shutdown power transients include an overpower component due to the partial core voiding which occurs to a varying extent in all cases. In case 7 (the conventional CANDU) the overpower transient had been calculated using the threedimensional neutron kinetics code CERBERUS (10). For the CANFLO cases, CATHENA runs were done initially with no overpower component in order to evaluate the voidage transient for each case. The voida^e thus calculated was used as input to a point k inetics code. The result ing overpower transient was then used in a final CATHENA calculation,
RESULTS Three major safety-related benef its from the CANFLO concept:
are expected
(1)
a reduction in peak fuel sheath temperatures following a loss-of-coolant accident,
(2)
a reduction in overpower react i vity, and
(3)
a reduction in the rate of energy discharge to the containment.
from
reduced void
We will examine the results of the calculations in terms of each of these benefits. Figures 9, 10, 11 and 12 show the results for all cases except for cases 2 and h. These are similar to cases 1 and 3: the outlet Venturis give little benefit. The temperatures for case 7 are siightly lower than expected from earlier CANDU 600 simulations, probably because the 35% break is not quite the worst with the CATHENA model. This means that CANFLO improvements could be somewhat better than indicated by the following compar isons. The calculated hot-channel, hot-pin sheath temperatures for the CANFLO cases are compared with those for the conventional CANDU in Figure 9. In all cases, a substantial benefit is indicated for CANFLO. The calculation for the figure-of-800 circuit, using f Luidic di odes at the in let ,
indicates a reduction in peak sheath temperature of 300°C. As expected, the use of check valves as inlet-flou limiters (cases 3 and 4) confers a decided advantageThis advantage is further enhanced if DP valves are introduced at the outlet U a s e fe>.
FIGURE 12:
20 1
CASE 6
FIGURE 10:
PREDICTED HOT-CHANNEL FLOW
PREDICTED ENERGY TO CONTAINMENT
The lower sVieath t«s»pei;&t\n:«:S obtained wvth CANFLO are directly related to the ability to maintain forward flow through the downstream core pass during the early part of the transient. Figure 10 shows the flow in the hot channel for each case. In the conventional CANDU, since it Is a worst case, stagnation occurs very quickly. For ^ANFLO, a significant forward floV is maintained for some 10 or more seconds depending on the case. Cases 1 and 2 have a small negative flow at 20 to 30 seconds that causes a reduction in temperature: a smaller break, say 90%, might give a worse result. Cases 3, 4 and 5 show a small positive flow up to 50 seconds. These need to be continued past 50 seconds until the channel is refilled by emergency coolant. Case G, with DP valves, shows a strong forward flow throughout the transient. The hot-channel power transients shown in Figure 11 are caused by coolant void in the channels immediately downstream of the broken inlet header. T'noMg'n tVra tKMTLO designs Vi-Je reduced -acid pet channel, twice as many channels are directly effected. (Recall that the conventional CANDU 600 has two figure-of-eight circuits and the CANFLO designs have only one circuit.) The result is that the power transients for CANFLO cases 1 and 2 are worse than for the conventional CANDU 600 and the power transient for the figure-of-800 is about the same. The peak temperatures are lower, as we have seen, because of the ability of the CANFLO systems to maintain forward flow during the early part of the transient. For the cases with check valves, the power transient is greatly reduced. Finally, Figure 12 shows the energy released to containment for each of the systems analyzed. In this Figure, the reference case is a CANDU 600 1002 outlet header break. Ag3in, CANFLO case 1 is worse than the conventional CANPU and the case with DP M&bjes is su&staj\tAaA.L^ improved. T*\e atKer cases give results similar to the conventional CANDU.
FIGURE 11:
PREDICTED HOT-CHANNEL POWER TRANSIENTS C N S 9th A N N U A L C O N F E R E N C E . 1988 289
CONCUSSIONS
(10)
(J i V'ITI t fit* f o r l r x <) i *>iie i li.ir.n t er i st i rs report ed here, I he CANFi.O t w o - s u p p l v - 1 r.iin figure-of-80u d e s i g n s i gn i i U'.int 1 v r e d u c e s l.Ot'A t u e i 1 e m p e r a t ure.s as compared to flu- c o n v e n t \oiui] CAN1M; f>00. With c h e c k vaJ vi>s in a CAN'l-'l.U f igure-nf -e ight or f i j^urt'of - 8 0 0 , flit1! 1 ewperat H I T S wuul il be t'urt her r e d u c e d a n d t ho p o w e r I r a n s ient t r^m vo i
REFKRENCKS (1 )
HKIM, R. , "An Invest, igat ion of the Thoma Count erf low Brake", Trans 1 at ions of the Mun ich Hydrau lie: I nst. i tute , 1929 , ASME TransLation by M.P. O'Brien, 1935.
( 2)
7.OBE!,, R. , "Kxpur ine.nt s ow a Hydrau I ic Reversing F.I bow". Mitt. Hyd. Insf. , Tt-cli. Hochsrhulo Muncrhen, 8, 1, 1930.
(3)
GRANT, J., WRICiUT, .1., "Potential Applications of Fl iiiii i cs in NIK: lour Plant " , Proceedings of r lie 2nd Cr.-mf i t> Id Conference, Paper K2 ( Pub 1 i.she.d by I he Br i t ish Hydromechan ics Research Assorial ion), 1967.
(A)
t.EORCE, I'.T. , WARD, J.R. , MITCHELL, F.M. , "Vortex D iodn CIiarat:t:er ist ics at. High Pressure Rat ios", Proceed i ngs of the Conference on Process Control by Power Fluidics, Institute of Measurement and Control, London, Paper 23 1075.
(5}
MORI, G., PRKMOLl A., "Vortex Diodes in TwoPhase Klow", European Two-Phase Group Meeting, Brussels, Paper A5, 1971, June 4-7.
(6)
GRANT, J., "Powe r fluidics and Env i ronment", Cr i t icaI Rev iews Environmental Control, CRC Press, 1977.
(7)
SYRED, N., ROBERTS, P.J., "Vortex Diodes App 1 ied to Post-Ace ident Heat Removal Syst ems", K hi idics Quarterly, 1989, June.
(8>
RICHARDS, D.J., tlANNA, B.N., HOBSON, N., ARDRON, K.H., "ATHENA: A Two-Fluid Code for CANDU LOCA Analysis", Third International Top i caI Meet ing on Reactor Thermalhydraulics, Newport, Rhode Island, \9H5y October 15-18.
(9)
"Fiowmc-fer Compitt.at ion Handbook", Prepared by the ASME Research Committee on Fluid Meters, The American Society of Mechanical Engineers, 1961.
290 CNS 9th ANNUAL CONFERENCE, 1988
the in
DASTUR» A.R., ROUBEN, B., BUSS, D.B.„ "CERBERUS - A Computer Program lor Solving the Space-Dependent Neutron Kinet ics Equat ions in Three Dimensions", TDAI-176, 1979, December.
COURThlNAY
BAY -
A S T E P TOWARDS TUB ADVANCED CONTROL ROOM OF THE
J.E.
FUTURE
SMITH
ATOMIC ENERGY OF CANADA LIMITED
INTRODUCTION Over the past three decades power plant operation have seen first a move from localized control to centralized control, then extensive automation followed by the development and use in the nuclear industry at least of full-scale operator training simulators. These successive developments have all be justified on the basis of safe, efficient, and reliable operation. They have been made possible by equally revolutionary changes in the technology of control and of control equipment. Recent developments in microcomputers and graphics have made possible at reasonable cost the completely computerized control room with integral control and training facilities. The new control console for Courtenay Bay Unit 4 is, in the author's opinion, a major step in this direction. Courtenay Bay Unit 4 is a 100 megawatt oil-fired unit, part of a four unit electrical generating station located in Saint John, New Brunswick. The station is owned and operated by the New Brunswick Electric Power Commission. N.B. Power required a low-cost solution to the problem of relocating the control of unit 4 from its existing isolated location to the combined control centre used by ufiits 1, 2 and 3. Although the main motivation was 3 saving in direct operating costs the request stemmed at least in part from N.B. Power's endeavour to introduce to conventional power station design the highly successful technology developed for their AECL designed nuclear power plants. System Description The system configuration is a modernized, distributed version of AECL's standard dual redundant DCC system, in which processing is distributed on a channelized basis. Data gathering and control functions are carried our. by dual IBM 7552 industrial computers that access plant data through a dual redundant Computer Products High Security Multiplexer. These devices are located In the former Unit 4 control room where all plant control signals are available. The control computers communicate with the Operator Interface Units (IBM 7532 industrial computers) via dual redundant high speed serial data links. These data links are standard off-the-shelf local area networks transmitting at 2.5 megabits per second using a token ring topology.
The process I/O system handles: - 256 analog inputs - 320 discrete inputs - 32 analog outputs - 192 discrete outputs Historical data storage requirements are met by dual redundant, 20 megabyte hard disks. The operator interface console shown in Figure 2 is completely CRT based using interactive graphic techniques, with the exception of the turbine runup panel, two flame monitors, and a backup boiler level meter. Fully dual redundant, the graphics are designed to allow full control of the unit in any phase of operation, with only two CRTs operable out of the four. Except for the startup phase, adequate control is possible from only one CRT. The display computers are mounted in the console. Twin 20 megabyte hard disks are provided for data storage. The display system is based on the MATROX PG-640 high resolution graphics controller. All computers are powered by uninterruptible power supplies.
battery
backed
AECL's standard fault detection and transfer of control (0CC-X to DCC-Y) strategy has been implemented in this system. A watchdog timer in each computer provides transfer of control from the controlling to the corresponding standby computer and fail-safe action in the event of failure of both computers. The watchdogs are updated on an exception basis i.e. they are not updated unless all checks have been passed and all tasks completed on time. Pollers are provided to test the state of the highway and of all nodes on it. Failure of a node or a highway results in transfer of control to the standby. DESIGN CONSIDERATIONS In this design we aimed to use automatic control techniques in combination v>ith trained human operators to provide safe, efficient, and reliable operation under all loads and transients. The design would minimize hardware, engineering, and construction costs. Maintainability and obsolescence were other major considerations. Cost We aimed for a payback of all costs in three years or less from an estimated saving of five operating shift positions. By using a CRT operator interface and computer network technology, installation costs were minimized- The powerful software
CNS 9th ANNUAL CONFERENCE, 1988 291
deve lopment and project rmin^geint-'iit tools available for the IBM PC-AT were inst rumental in keeping engineer ing costs down. The avai l a b i l i t y of lou cost but high performance i n d u s t r i a l grade IBM PCAT computers and we 11 proven PC network hardware kept botfi component and system integration costs down. Reliability KeLiabi Lity and tolerance of failure are major design requirements for the Courtenay Bay systern since total loss of this system must result in a plant shutdown. The dramatic increase in intrinsic equipment reliability achieved by the electronics industry over the past decade has sharply reduced the need for redundancy as a means to achieve high digital sysreios reliability. Nevertheless redundancy is still employed at Courtenay Bay in part to minimize raaintenance d ownt ime• Fault Tolerance and Self-checking Regardless ot the degree or type of redundancy employed1, a real-time system must incorporate mechanisms for the detection of faults and the minimization of their effects. The system, therefore, incorporates certain techniques for the detection ot" faults, and provides for the graceful degradation of tlie system by the progressive shedding of functions as they are found to be affected. Sustained failure is used as the criteria for abandoning a function. Most failures in computer and process control equipment are transient in nature. Even those software faults that persist into the operating phase tend to manifest themselves only under rare combinations of circumstances. Thus a retry mechanism has been incorporated in all AECL failure response software since 1963 and is one of the main reasons for our success. Obsolescence The process control industry has been hit by major shifts in hardware technology every five to seven years whereas the average Life of a power plant is thirty years or more. Consequently most power plants still in operation are operating with control equipment that is no longer supported by the manufacturers except on a special (and costly) basis. This is becoming a major concern as utiLities look to plant life extension. Consequently there is considerable pressure for guarantees of long term availability of components and for downwards compatab!lity of new designs. Considering the small segment of the electronics market represented by the power industry, this could be quite expensive to achieve unless the control system suppliers change their strategy of using proprietary equipment. This was one of the major factors in AKCL' s choice of equipment for its Courtenay Bay design. The computers are IBM PC-AT's, a de facto industry standard, compatible graphics equipment is available from numerous suppliers, and the process Input-output equipment is in widespread use because of its compatibility with every major microcomputer and most mini-computers. The software, too, has been designed to be easily portable to other compu-
292 CNS 9th ANNUAL CONFERENCE. 1988
ters and even to same other operat ing systems. Almost all the software I K written in the C Programming Language and the sou ret; code is available, so that the utility or his consultant can provide long term software support. By sharing component and software designs wi th the huge micro-computer market we are assured of long term low cost component availability and downwards corapatibility with new equipment. Control System Compatibility There are sound project management and financial reasons to favour an island approach to power plant procurement. In this approach the major sections of the plant, such as the turbine and boiler, are bought complete with all controls and auxiliaries. If this approach is used, some way must be found to integrate the control systems so that the interface with the human operator is both comprehensive and consistent. This problem was faced at Courtenay Bay where the new system had to incorporate existing control equipment of different vintage and design ranging from pneumatics to a modern tiailey Net-90 burner management system. The result was a system that can readily communicate with several of the proprietory distributed control systems and PLC's on the market today. The only major limitation is the willingness of the other suppliers to cooperate but most provide adequate computer interfaces. Operator Training With the average age of power plant operators in North America increasing rapidly there is a growing recognition in the industry of the need to train new ones. Also modern digital control equipment will require retraining of the experienced operators. For this purpose training simulators have many advantages over conventional on-the-job training but in the past such facilities have been prohibitively costly except for the training of nuclear plant operators. With the advent of: the totally computer based operator interface such as will be used at Courtenay Bay Unit 4, the situation is radically changed. The development and checkout of both the interactive operator displays and the control software requires the parallel development of process simulation as an integral part of the system. Thus an operator training simulator is available for use during plant shutdown without the need for extra expenditure. Retrofit Situation Courtenay Bay is an early 19bO's vintage plant with a variety of electrical and pneumatic instrumentation that is not directly compatible with modern digital controls. Special signal conditioning and buffering units had to be provided to overcome this difficulty. Functional Requirements Reliance on an interactive CRT based control console for operation of a power generating unit is relatively new. Most digital control installations
4. 5. 6. 7. «.
to Hate have continued to rely on panel mounted controllers and hand-switches or pushbuttons. Conventional control rooms present all the operating information available on a continuous basis. It is important to note, however, that while everything ts on the panel simultaneously, the operator cannot simultaneously absorb it all. The designer must lay out the panel to meet the operating requirements. Since these are different for startup, normal running, and shutdown the panel layout is usually a compromise. Once built it is inflexible. CKT consoles have quite different characteristics. Usually it is not good practice to attempt to spread out all the plant controls and displays across the console. Rather the controls and displays are brought to the operator as needed. The trick is to accomplish this in an easy, natural roanne r. To this end the CRT displays are grouped as to function, area of the plant, and phase of operation. Total loss of the display system would result in a plant shutdown so the system is designed to be fully dual-redundant and the graphics have been designed to allow full control of the unit in any phase of operation wirh .>ny two of the four CRT's operable. In steady state operation adequate monitoring of the unit is possible from only one CRT. Flexibility in the use of screens is a must. Therefore in this system all displays can be located on any of the four MATROX screens. In order to promote coordinated control, however, most control displays are allowed to exist on only one screen at a time. The exception is for certain on/off controls for valves or pumps where, for convenience, control of the same devices should be allowed from more than '-ne schematic display. Such schematics cover a va iety of information. They may overlap In cerms c." content and control functions, with one screen designed for one phase of operation such as pre-start preparation and another the warmup phase. During the transition period both screens may be needed. If the process of selection of screen functions Is to become instinctive and errors minimized, menu selection should not normally involve more than eight items. When dealing with large numbers of items or choices the normal human approach is to classify, going from the general to the particular. Consequently at Courtenay Bay screen functions are selected via a hierarchial system of menus. There is a Main Menu of eight primary functions as listed below and associated with each of these are appropriate submenus. The Main Menu items are displayed horizontally at the bottom of the screen with their associated function keys Fi to F8. Above this menu is displayed the "Alarm Summary Screen" giving the latest alarm status information. MAIN MENU 1. 2. 3.
CONTROL FUNCTION SCHEMATIC DISPLAYS 'A1 SCHEMATIC DISPLAYS 'B'
TREND DISPLAYS BAR CHART DISPLAYS ALARM ANNUNCIATION STATION LOG SPARK
Selection from the main menu is by single stroke of the indicated key. This results in display of a primary screen for this function. At the bottom of the screen is a submenu arranged horizontally listing the other screens associated with this function. Pressing the indicated key will select that display. The submenu is retained at the bottom of each display and the operator can easily move from one to the other within that function by pressing the appropriate keys. To return to the main menu the operator simply presses Q for 'quit'. While information is organized according to the main selections listed above, regardless of the function performed, any further breakdown into subgroups is functionally dependent. A study of operator interaction wich the existing Courtenay Bay control panels supported by other man/machine studies indicated that the activities are divided into discrete phases; -
item selection, confirmation o£ item selection, action selection, confirmation of action selection, consideration, and action or rejection.
For success, provision must be made In the design to carry out all these phases, smoothly, quickly and naturally. Working with representatives of the Courtenay Bay operating staff, Maritime Nuclear developed the following method. Item Selection and Confirmation. Item selection is made using a moveable cursor on the screen. When selection has been confirmed by pressing an (ENTER) key an acti on box appears at the bo11om ri^ht hand corner of the screen. This box contains the device name, its current state and the action choices to be made. Selection and confirmation of the action to be performed is made using the cursor and (ENTER) key and the action to be taken Is highlighted in the appropriate colour but no action is taken. If after consideration the operator decides to go ahead he presses the (ENTKR) key again. If he wishes to reject the action he simply moves the cursor away from the selected device- The action box then disappears. All this sounds cumbersome but experience with the simulator confirms that, in fact, the process is quick, natural, and soon becomes instinctive. EXPERIENCE TO DATE The system will be delivered to site in late June of this year. Installation and commissioning will most likely extend throughout the summer. Thus a report on commissioning and operating experience will have to wait.
CNS 9th ANNUAL CONFERENCE, 1988 293
Nevertheless, some valuable lessons have already been learned. Perhaps the most important is that using the software development tools available today it is possible to develop from scratch in less than one year, a cost competitive high integrity, distributed digital control system of advanced design using of t'-t he-she If process I/O and personal computer equipment. Another important lesson is the need to integrate simulation software with the rest of the system early in the development phase and to make this available to operating personnel. The ability to realistically demonstrate the operator interface as it was being developed allowed prompt feedback, from the plant operating personnel and promoted an early understanding of the nature of what they would have to deal with. This resulted in very effective teamwork between designers and operators. During the design phase on this project AECL was called upon to assist the client in solving electrical interference problems with a new control system from another supplier that was being installed on an almost identical unit at Dalhousie G.S. This brought home to all the need for greater care in power supply, as well as instrument grounding, where distributed control equipment was concerned. As an alternative to jumping the cursor from one item to the next when selecting items from a schematic screen a "mouse" has been provided to allow direct movement of the cursor to a selected item. Experience to date with the two methods is inconclusive. Both methods have some clear advantages and some drawbacks but both work well and it is likely that both methods are equally satisfactory. Clearly, this is a matter of customer preference. Inadequate equipment supplier documentation was often a serious problem causing costly delays.
294 CNS 9th ANNUAL CONFERENCE, 19B8
Session 8: Advances in Nuclear Engineering Education in Canada
Chairman: H.W. Bonin, Royal Military College
CNS 9th ANNUAL CONFERENCE. 1988
•/
295
T\T NUCLPAR
ENC-TNFERTNC
EDUCATION
IN
UugueR W. T^oni n Department Chemistry and Chemical Engineering Royal Military College Kingston, Ontario K7K 5L0 Tel. :{613)541-6fin
INTRODUCTION
AT THE CRADITATE LEVEL
The following Is a summary of rhe presentations made at the session entitled "Advances in Nuclear Engineering Education in Canada" at the 9th Annual Conference of the Canadian Nuclear Society, IS June 19RR, Winnipeg, Manitoba. The papers presented were: "Nuclear Engineering Education at McMaster University" (Wtn. J. Garland), "The University of New Brunswick Nuclear Engineering Program" (D. A. Meneley, F..M. A. Hussei n, R -A. Chaplin, R. Glrard), "L1Enselgnement du Genie Nucleaire a l'Kcole Polytechnique" (0. Rozon), "Evolution of the Nuclear Engineering Programmes at Royal Militrary College" (H.W. Ronln, L.C.I. Sinclair, presented by Dr. O.J.C. Runnalls).
All of tie five universities grant post-graduate degrees either in nuclear engineering, or in chemica 1 or mechanics 1 englneering, but with an evident nuclear engineering emphasis In the programme. Tn add!tion, nuclear science post-gra^uate programs are available at two of the univers11ies: McMaster and University of Toronto• Generallv, strong collaboratlon with the industry is a characteristic of these programmes. ThIs ranges frcn thesis topLcs which are parts of on-going research at major nuclear research centres to internship programs wlthin tnc Industry for the graduate students and even to ioint ventures with the industry (e.g. Croupe d*Analyse Macleal re at Pcole Polytechnique).
As a first attempt, the purpose of the session was mostly to provide a platform for comparing the nuclear engineering programmes given at five major universities. The number of invited speakers was kept at five, in order to provide enough time for discussion after the presentations. Other Canadian universities are active in nuclear engineering, notably at the graduate level where thesis projects sre offered as part of research in collaboration with the nuclear industry.
The courses offered at the post-graduate leve1 cover the breadth of nuclear science and engineering: advanced fission reactor physics, numerical methods, heat transfer, two-phase flow, thermal hydraulics, health physics, radiation protection, nuclear phvsics, radiation detection and measurement, radioIsotopes, nuclear medicine, nuclear materials, nuclear chemistry, plant control, instrumenrat ion, expert systems, neutron radiography, fuel and waste management, nuclear safety and reliabi11ty, thermonuclear fusion, etc.
ACTIVITIES AT THE UNDERGRADUATE LEVEL
The thesis topics chosen by the students also cover the domains just mentioned, but In recent years, a shift of the interests was observed from the large nuclear reactor design and engineering to other important domains such as the nuclear materials science, corrosion, nuclear medicine, radiation damage, and small nuclear reactors for cogeneratlon and district heating. This shift of interests parallels a similar shift within the Industrv, as orders for new large CANDU are withheld temporarily.
All of the universities represented at the session have this point in common: none offers a Bachelor's degree in nuclear engineering. However, several courses in nuclear engineering, or in areas closely related to nuclear engineering^ are offered within several programmes such as Engineering Physics, Chemical Engineering, Mechanical Engineering and Fuels and Materials Engineering, to name a few. The undergraduate students can enroll in courses In nuclear reactor physics, thermohydraulies, heat transfer, nuclear safety and reliability, health physics, radiation protection, numerical methods, radiation detection, Instrumentation, and control, among others. Tn this area, McMaste*" University stands apart with a strong programme In nuclear medicine, in addition to the topics mentioned above. Another common point to the five universities is that nuclear engineering is a multi-disciplinary activity, involving not only the trad I ttonal branches of engineering, but also such as Civil, Electrical and Metallurgical Engineering, plus other departments 1 Ike Computer Reience, Geology, Industrial Engineering and even Aerospace Engineering.
The administration of the nuclear engineering programmes varies among the university. At McMaster University, the programmes are focused at the Department of Engineering Physics: at University of New Rrunswlck, the programme is within the Department of Chemical Engineering, and at RMC, the adminf stration Is doe by the Department of Chemistry an-1 Chemical EngineeringAt University of Toronto, several departments are Involved with nuclear engineer?n&: chemical engineering and mechanical engineering In particular. To help coordinate the various activities related to nuclear engineering, the I'ntversltv of Toronto has established the Nuclear Engineering Center, funded entirely by industrv. Its role Is to facilitate comrounicat ion between the students,
CNS 9th ANNUAL CONFERENCE, 1988 297
professors and experts from the tndus cry, rind ensure the best opportunities for rosenrch and ioh opportuni t ies for the graduates . <\t Ecol e Po 1 vtec'iTi t quo dp Mont ren 1, the sItuation was quite s h i l n r fn the i^fiO's, with nuclear engineerfng being the affair of the Department of Engineering Physics. However, the Tnstltut de Henie Nuc leal re was created In 1 ^) "7 0 to administer and expanded programme at the Master's decree level first, further expanded to offer the doctorate degree in 1979. A low number of undergraduate service courses were offered s I nee the start of the ICN. With a lart>e part of thp RS.D activities in major computer codes, the Tnstttut created the "Croupe d'Analvse Nuclea i re" (CAN), with finanei ng from Hvdro-^uebec and the support of AECL, In order to perform the technological transfer of computer software and develop new codes, for nuclear reactor safetv and performance analysis. Tn 1QR7, the CAN became an tndiistrial-onlv firm, as its activities had shl fted outside the norma 1 academic framework.
Each representative reported an Interesting wealth of enui pment, ranging from a Van de Craaff accelerator (McMaster) to fission reactors (S MW pool-type at McMaster, SLOWPOKE-2 reactors at University of Toronto, Ecole Polytechntque and Royal Military College, to heat transfer loops (Ecole Polvtechnique), to subcritlcal reactors (II of T, Ecole Poly technique), to neutron generators (Hntversity of New Rrunswick). When the equipment is lacking, the universities can usually compensate by close collaboration with the industry• For example, ffniversity of Vew Rrunswlck works closely with the Point Lepreau Generation Station, with the students spending several davs for performing experiments. Before the acquisition of the SLOWPOKE reactor, students at RMC used to spend a week in May to perform several experiments at the MeMaster Muclear P.eactor.
CHANCES AHFAD It seems that fill of the universities are consi dering or implement Ing some changes to their programs in order to ad just to the evolution of the work market In the nuclear 1ndustry. A common point Is that the student populattons are either at steady state or declining. The obvious reason lies in the perception of the nuclear i ndustry within the pubIIc as a -norIbund industry. As a consequence, students are lead to believe that the future Is hleak, and only a low number of die-hards enroll in the nuclear programs.
became the Inst 11 do Genie- Knerget I quo. The latest step in this evolut i on is the proposal of -i new Master's degree In Engineering programme, whfrh w])1 have the Nuc lear EngI nee rIng o^t i on and the Energy Management option sharing a common trunV mado of the f ol lowing courses : Indtistr f a 1 Process Energy , TwoPhase System ThermohydraulIcs and Fnergv System RelIabiItv and Sa fety. ^ happy exception Is the rase of Royal Military Co Liege at Kingston, Ontario. Extensive mod if i cations to the nuclear engineering programmes are necessary to cope with a sharp Increase of the graduate student population. As a result of the l^h I te Paper on defence, the Department of National Defence decided to build a strong team of experts In nuclear engineering to work on the Canadian Subraari ne Acnu isIt Ion Programme. "or the next several vear=, i t i s foreseen that about six naval officers will register each year In the Master's nr>gree In Nuclear Hngi neering programme. Thls programme was extensively modifted to Include courses more suited, to the needs of the Canadian Navy, such as Nuclear Reactor KngineerIng, Advanced Thermodynamics, Heat Transfer, Safety/Control ( Numerical -iethods and Radiation Protect Eon and Shielding. At the undergraduate level, no changes have yet occurred, but two proposals have been drafted and are ready for submissIon tii the Department of National Defence when It Is officially decided to acquire a fleet of nuclear submafinesThese proposals are for the Introduction of either a Bachelor's degree In Nuclear Engineering, or a oneyear Engineering Diploma of Nuclear Engineering programme. In 19Rfl, RMC has introduced a new ::afessional de i/eloptnent course called Nuc lear Kami 1 iari zation Short Course, given in May and September to military and civilian personnel assigned to the submarine acquisition programThis course is given in addition to the "Advanced Radiation Safety" course gtven once a year to Army personn°I.
PROBLEM AREAS Unfortunately, the time left for the discussion and the questions period at the conference was too short for a sound discussion of the problem areas, and those could only be ment toned: the recrultment of new students at both the undergraduate and graduate levels, the financing of research projects and operat ions of the n.iclear programmes fin the light of recent cutbacks in the industry and in the grants from the traditional agencies), the role university In H&D programmes, the various means of collaboration with the industry, the place of university In nuclear reactor operators' training and the means of collaboration between universities, nther more i'3wn-to-earth issues were briefly addressed: the CNA/ONS Student Conference, the development of computer software and the means of sharing th*» resul ts between the universities, the development of commnnlcation skills among the students, the computer-atded teaching techniques, the need for a CanadI an textbook in nuclear engineer!ng and the best way to exchange comments and InformatIon among the professors (newsletter, CNS Bulletin, . . . ) , or evt n computer software.
The low enrolIment problem has forced at least one university to modify its Master's degree program. At ^cote Po1ytechnfque, the problem Is not so much In the total student population, but rather in the declining of the relative number of students regl stored in the M. Rnt». programme • Li nked with a reduction of the length of the programme from ° vears to 1* vears, this resulted (n a decrease of the number of courses offered, eventually leading to financial difficulties. Having assumed the responsiCONCLHsrONf. bility of the complementary v.nerzv or I ent.i t Ion at the undergraduate level In 1977, the Tnstttut de Oenie Mhilp this particular session rann.it claim havNuciealre was asked In 1 QflS t<> develop a new post ing solved the problems ont1Ined above, it has the ?,rfidufitf> program In energy management, and then merit of having permitted a comp, r\son of the ma 1or
298 CNS 3th ANNUAL CONFERENCE, 1988
programs in a structured way and, quite important,nroblem areas have been Ident1fierf and addressed. \n interesting fact was revealed at the session: at that moment, three positions for professors were open (one at McMaster and two at H of T ) , and rir. Runnalls ment toned that discussions were underway to fund T chairs at University of Toronto. From the brief discussion with several representfitives from the industry, it appears that they are generally satisfied by the formation given to the nuclear engineers by the programs at the universities. The session has been a valuable exercise, and should be repeated In the future, although focusing on one or a few of the problem areas, now that the Reneral overview has been done.
CNS 9th ANNUAL CONFERENCE, 1968 299
THE UNIVERSITY OF NEW BRUNSWICK NUCLEAR ENGINEERING PROGRAM D.A. MENELEY, E.M.A. HUSSEIN, R.A. CHAPLIN, R. GIRARD Department of Chemical Engineering, University of New Brunswick Fredericton, New Brunswick E3B 5A3
ABSTRACT A new program of nuclear engineering education was started at the University of New Brunswick in 1984 to serve the needs of the Maritime region. The program is based on the long-standing close relationship between the Faculty of Engineering and local industry. It is a small program, in keeping with the modest needs of the region for nuclear engineers. Nevertheless it includes an undergraduate option and a graduate program as well as training assistance to NB Power,, owner-operator of the 600MW station at Point Lepreau.
INTRODUCTION Nuclear engineering education in Canada has been carried out for the past forty years largely through in-house programs in research laboratories and engineering firms. More recently. Universities have contributed through graduate studies augmented by a limited number of undergraduate courses. No complete undergraduate program exists in Canada. There are three established programs in Ontario {U of T, McMaster, and RMC) , one in Quebec (Ecole Polytechnique) , and now one in the Maritimes (UNB). These programs have contributed a considerable number of graduates; nevertheless, the majority of staff now working in the nuclear industry do not have formal education in the nuclear engineering field - their training ranges through all the engineering disciplines plus several science fields. In 1983 the Faculty of Engineering at the University of New Brunswick approached the New Brunswick Electric Power Commission {NB Power) regarding the possibility of establishing a Chair in Nuclear Engineering. The proposal included an undergraduate option program of 20 credit hours {out of 18 0 required for graduation) plus a graduate program. In addition, it was proposed that the operator training program for Point Lepreau staff be continued and expanded. 7ollowing agreement by NB Power, the proposal was taken to the Natural Sciences and Engineering Research Council
300 CNS 9th ANNUAL CONFERENCE, 19B8
The research program is based on (ai development of small-scale experimental and theor°r ical work in support of the CANDU energy syst< m, and (b) development of instruments which use nuclear radiation in various forms as a diagnostic probe for measurement of the physical properties of matter. These objectives can be met within the framework of a small University research organization, especially with the support of nuclear power utilities and outside research laboratories.
UNDERGRADUATE PROGRAM The undergraduate option program was designed so that some of the courses could be substituted for Mechanical Engineering or Chemical Engineering required courses. The remainder of the courses necessary to make a total of 21 credit hours for the option were assigned to the technical elective category. The first premise of the undergraduate option program was that the course material should provide educational content equivalent to that of the courses replaced. In this way, it is possible for the student to graduate with a regular degree in Mechanical or Chemical engineering. The undergraduate courses are designed and offered as a vehicle for teaching the fundamentals of mathematics, physics, and numerical analysis, rather than to present details of nuclear reactor designs. Nuclear Engineering courses now available at UNB are shown in Table 1. These are all one-term courses. The option program begins in the fourth term of a eight-term schedule, with most of the courses begin taken in terms five to eight. As can be seen these courses provide a broad range of subject matter relevant to the field. Perhaps the most important omissions are metallurgy and electronics. Some of the courses developed attract technical elective students from other Departments of the Engineering faculty as well as the Department of Physics. The Nuclear Practice School involves an assigned project to be initiated at an industrial site over a period of two weeks, followed by preparat ion and presentation of a project report at the University. This course follows a precedent which has been followed successfully in the Chemical Engineering Department at UNB for several years. The Practice School has been enhanced this year through addition of a four month industrial internship program. Project assignments are made by agreement between the University and the industrial sponsor as is the case with practice school projects; the student then works for a summer term with regular summer student pay and prepares a report on which the course is graded. When the program is fully developed the student will spend an additional four months at the industrial site working on jobs defined by the industrial sponsor. Following the first year of the undergraduate option program, class sizes have been rather small -
between three and sevenThis is to be expected in a specialized field such as nuclear engineering, also considering the relatively inactive state of the industry at this time. Enrolment was at a peak (16) at the beginning of the program, most likely because o£ the strong prospect oE a second unit at Point Lepreau at that time. The small classes offer an advantage in stronger interaction between students and faculty, with the result that many option graduates pursue graduate studies in the field. Of the 13 option graduates to this date, four are now enrolled in graduate school at UIIB and one at the University of Michigan. With the creation of the Chair in Power Plant Engineering, specific courses in this field have been offered to complement the nuclear engineering program. Along with existing courses in mechanical and chemical engineering, students can now select a package of •ourses giving a high degree of specialization in .ower plant engineering. Class sizes have ranged from :ive to nine with numbers split approximately evenly between chemical and mechanical engineering iepartments.
TABLE 1 NUCLEAR ENGINEERING COURSES AT UNB* Atomic and Nuclear Physics Phys 2962/2965 Nuclear physics for Engineers Phys 4961 Nuclear Engineering Ch.E. 5724 Nuclear Safety and Reliability Ch.E. 5844 Nuclear Practice School Ch.E. 3823 Ch.E. S744/M.E. 5744 Steam Supply Systems Ch.E. 5754/M.E. 5754 Steam and Gas Turbines Nuclear Chemical Processes Ch. E. 5804 Nuclear Design Laboratory Ch.E. 5814 Nuclear Radiation Engineering Ch.E. 5834 Two-Phase Thermal-Hydraulics Ch.E. 6754/6854 Special Topics - The Monte Ch.E. 6504 Carlo Method Courses beginning wit.i a "5000" designation are available to both senio- undergradate and graduate students; courses witii "6000" designation are tifteced only at the graduate level.
GRADUATE PROGRAM The general purpose of any graduate program is to instruct students in the art of original investigation. The specific subject being studied is secondary to this goal. Course work is necessary to provide the knowledge and techniques required to solve the assigned research problem. Some specific courses related to nuclear engineering have been developed to meet the needs of graduate students; they also have access to all other courses nresented on the UNB campus. There is, at present, no accreditation of the nuclear engineering graduate program at UNB. Instead, the students must meet the basic requirements of either the chemical Engineering or Mechanical Engineering departments in order to receive a graduate degree. This usually involves enrolment in a few undergraduate courses to fill gaps in the knowledge of the student in the particular discipline. The requirement has caused no problems to date.
As referred to in the Introduction, the UNB graduate research program in nuclear engineering centres on small-scale projects for nuclear power plants and on nuclear instrumentation development. A list of present graduate students and their projects is given in Table 2. These projects are indicative oi the scope of research work to date; most projects require a combination of theoretical and experimental work. A further project not on this list is nuclear station dynamic simulation using the DSNP simulation shell. In aridit io-, '• - a large IBM 30 90 computer system, students have available to them a number of mini- and micro-computers of various types. Laboratory equipment consists of a number of neutron and gamma sources, shielding, and instrumentation. A large 14 MeV neutron generator will be installed in the new laboratory space within the next few month3. The equipment installed is limited both by funding and by available laboratory space. For graduate experimental work, appropriate students are attached foe several months to an existing nuclear engineering laboratory. Other than gaining access to a much broader range of equipment than would be available at tfte University, the student also is exposed to professionals working directly in the research field of interest . In this way the student broadens understanding of research in general and his or her field in particular. Up to this date, UNB students have visited Point Lepreau Nuclear Station, Chalk River Nuclear Laboratories and Argonne National Laboratory. Several other possibilities are under consideration. Research funds to support graduate studies in nuclear engineering amount to a total of about $225,000 per year. The sources of funding include NSERC, NB Power, AECLf and Che Terry Fox Fund. More than sixty publications have been produced as a result of this funded research. The graduate program has attracted students from a variety of backgrounds, including chemical and mechanical engineering, engineering physics, metallurgical engineering, and computer science. Some students have achieved national and international recognition by winning awards from NSERC, the Canadian Council of Professional Engineers, and the IEEE Nuclear and Plasma Sciences Society. The graduate program in Nuclear Engineering at UNB is now well established. The prospects for the future are very encouraging.
OPERATIONS TRAINING Over the past six years the University has provided instructors to NB Power for the conventional general course series for First Operator in Training and Shift Supervisor in Training candidates, as well as for technical unit staff. Since 1984, when the UNB program was expanded, nuclear general courses also have been given for the same classification of staff. Course material for the above training is based largely on the Ontario Hydro system. Several of these courses have been updated and rewritten by UNB staff. More recently, the CANDU Owners Group has decided to standardize the basic course material for all plants in Canada. It is expected that some of the remaining
CNS 9th ANNUAL CONFERENCE, 1988 301
rewriting necessary will be done by UNB faculty. This type of training has been very effective, as evidenced by the high marks of candidates in their AECB examinations. It also is a very useful experience for the instructors because they gain a deep understanding of the plant during the process of developing and presenting the courses.
TABLE 2 CURRENT GRADUATE RESEARCH TOPICS J.F. Lafortune - Design of a Small Nuclear Reactor for Electricity Production. A.M. Shalabi - Crack Initiation in zirconium - 2.5% Niobium Due To Hydride Formation. D.R. O'Connor - Gadolinium Concentration Monitor for CANDU Stations. J. Zhang - Stochastic Modelling of Fluid Flow. T. Whynot - Compton Scattering for Nondestructive Testing of Concrete. R. Brown -.Numerical Solution Method for Two-P3ia.se FJow in Steam Generators. X. Wu - Modelling of Steam Turbine Dynamic Response. A. Djordjevich - Tissue Radiation Therapy.
Compensator
Design
for
G. Noye - Display System for Nuclear Power Plant Simulator.
nuclear physics. The reason is that the major development phase for existing nuclear power reactors is over; unless a completely new development is initiated the engineering of these plants will become quite routine. Most of the engineering work fits logically into one or more of the traditional disElpJ-Loes- The more specialized aspects of nuclear engineering will likely be concentrated in graduate education. Associated with mechanical engineering, an expansion of undergraduate power plant engineering education and development of an integrated package of courses in this field is expected in the future. The driving force for this development is the strong impetus for increased efficiency of thermal units, both in-service and in design. within the existing program, diversification of graduate research topics is expected, to the extent possible given the limitations in expertise of faculty and the interest of graduate students. The reason is the broadening applications of nuclear processes in industry and medicine as well as the broad range of topics which arise during power plant operation. Closer association of grayuate education directly with industry is expected as a result of the strong impetus from granting agencies to support collaborative research with industry and technology transfer to the commercial sector. Support of programs completely independent of industry likely will become more and more difficult. If the results of the UNfc - NB Power operations training system are taken up by other utilities, one can expect an increase in the fraction of operator training covered by University faculty.
V. Wu - Vectorization of Monte Carlo Algorithms.
FUTURE PROSPECTS It is expected that undergraduate nuclear engineering education will be slowly integrated with that of other disciplines, especially for items which are not specifically telated to reactor physics or
302 CNS 9th ANNUAL CONFERENCE, 1988
In conclusion it is expected that the detailed education and training needs of nuclear engineering students will change drastically in the future as the industry swings further away from the large laboratories which were needed to develop the technology in the first place, and toward commercial applications. The constant aspect of these changes will be a continued need for graduates who excel in the basic sciences, mathematics, and engineering disciplines.
Session 9: Small Reactors: Safety
Chairman: P.M. French, AECB
CNS 9th ANNUAL CONFERENCE, 1988 3 0 3 / .
'3o
TANK - A COMPUTER CODE FOR THE TWO-DIMENSIONAL MODELLING OF TRANSIENT BEHAVIOUR T.H RESEARCH REACTORS
R.J. ELLIS, 8.J. SMITH and P.A. CARLSON
Atomic Energy at Canada Limited Uhiteshell Nuclear Research Establishment Pinawa, Manitoba ROE 1L0
ABSTRACT Computer simulations of transient behaviour are needed in the safety assessment of a nuclear reactor. TANK (1,2,3), a two-dimensional two-neutron-energygroup space-time reactor kinetics computer code, is being developed at the Vhiteshell Nuclear Research Establishment as a tool to simulate transients in the MAPLE class of research reactors. TANK was used to analyze a variety of postulated reactivity insertion transients in both U02- and metallic-fuelled MAPLE Research Reactors (2). The results of preliminary transient simulations are discussed in this paper.
INTRODUCTION The behaviour of a nuclear reactor during transient operation has important safety and design consequences. Transients may occur during reactor power level changes or during accidents. All transients are initiated by a change in the reactivity of the nuclear reactor, the insertion of positive reactivity being potentially large in accident situations. Depending on the severity, an accident-induced transient may damage a nuclear reactor because the increased temperatures arising during nuclear excursions may lead to fuel material or cladding failure. Computer simulations of transient behaviour in postulated accident scenarios are therefore necessary as part of the safety analysis of a proposed nuclear reactor. To properly describe a reactivity insertion transient, the simulated time-dependent behaviour o£ the following is required: reactor power level, inverse reactor period, reactivity, and fuel, cladding and coolant temperature distributions. The space-time reactor kinetics computer code TANK (Transient Analysis with Neutron Kinetics) calculates this information during simulations of transients in the Atomic Energy of Canada Limited (AECL) MAPLE class of research reactors. Furthermore, details in a TANK simulation are calculated on a site-by-site basis, giving a two-dimensional description of the transient behaviour. This provides for a far more realistic transient simulation than a simplistic point-kinetics treatment.
THE REACTOR KINETICS COMPUTER CODE TANK The two-dimensional two-neutron-energy-group space-time reactor kinetics computer code TANK is being developed at the Whiteshell Nuclear Research Establishment as a tool to simulate reactivity insertion transients in MAPLE research reactors. TANK is used to model reactor behaviour in a twoneutron-energy-group representation, in two dimensions. Some neutronics considerations are made for effects in the third (axial) dimension. The two designated neutron-energy groups are fast neutrons
with kinetic energy j 0.625 eV and thermal neutrons with kinetic energy < 0.625 eV. Among the transient situations that can be analyzed are reactor power level changes, loss-ofcoolant accidents, channel blockages, and loss-ofregulation accidents. The versatility of TANK allows for the simulation of many asymmetric reactivity insertion transients, important as part of a safety analysis study. One option that can be selected for a TANK transient simulation run is the actuation of the shutdown system by tripping. Trip set points are input for coolant temperature, absolute neutron flux level, power, inverse period (also referred to as the rate log of pover) or time. The effectiveness of various shutdown trip poi»it values and the shutdown actuation delay time can thus be assessed for different transient conditions. For TANK simulations, nuclear reactors are modelled in two dimensions on a hexagonal mesh of up to 900 cells. The actual number of cells depends on the size and complexity of the reactor. Each cell in the hexagonal mesh representation of the reactor is characterized by a set of nine kinetics parameters: fast and thermal neutron macroscopic absorption and fission cross sections, fast-to-tbermal macroscopic removal cross sections, fast and thermal axial diffusion coefficients, and the cell-averaged fast and thermal neutron velocities. These parameters are obtained from computations with the multigroup transport code U THS (4,5) which uses the ENDF/B-V (6) cross section library. The kinetics parameters (cross sections) for the cells corresponding to fuelled sites in the reactor core are determined for a range of average fuel temperatures and bulk coolant void fractions. This information is used in the simulation of the reactivity feedback effects of fuel temperature changes and void formation. Parameterized equations for the cross sections are fit up to the third degree in fuel temperature and void fraction. These equations are incorporated into TANK subroutines, allowing us to update the cross sections as conditions change in the reactor. The three-dimensional neutron diffusion code 3DDT (7) is used to calculate the axial power and flux level distributions. With weighting based on the 3DDT axial power distributions, TANK is used to determine the kinetics parameters for the control rods and the shutdown rods as a functio.i of the axial position of their hafnium absorbers. TANK uses numerical kinetics equations for factorization approach separate the space and neutron flux level,
methods to solve the neutron the two energy groups. A flux (8) is used in TANK to time dependence of the total
CNS 9th ANNUAL CONFERENCE, 1988 305
0 (r,t) = 0 (r,t) N(t) where is the space-time neutron flux,
Reactivity Feedback Mechanisms Reactivity feedback (8,10) is a phenomenon that greatly influences the behaviour of a nuclear reactor during a transient. The state of a nuclear reactor, as characterised by neutron flux levels, power level or temperatures, changes when a reactivity insertion takes place. These changes result in an incremental change in reactivity that feeds back and alters the total reactivity, thus affecting the transient behaviour. When the incremental changes in reactivity are negative, the total reactivity is reduced, either slowing dovn or stopping the transient. This is referred to as negative reactivity feedback; from a safety viewpoint, many reactors are designed to take advantage of this effect. Subroutines in TANK were written to account for reactivity Eeedback mechanisms in the simulation ot transients in MAPLE research reactors. The effects of coolant temperature and density and fuel temperature on the total reactivity are simulated. Research reactors of the MAPLE class are light-water cooled and moderated, and operate in an under-moderated condition (1,2). As the coolant temperature increases during a transient, the coolant density decreases. The formation of void in the hottest fuel channels further decreases the coolant density. Seing under-moderated, the reduction of coolant density in a MAPLE Research Reactor results in a negative reactivity feedback effect. Fuel temperature reactivity feedback, on the other hand, is primarily the result of the Doppler broadening of the 238 U resonance capture cross sections (8). As the fuel temperature rises, so does the probability that a neutron will be captured by "aU. This capture does not result in the release of any fission neutrons, so the net effect is a drop in the neutron population. It is a negative reactivity feedback etrect most evident in low-enrichment uranium (LKU) fuels.
change. The cross sections are re-calculated in TANK subroutines, as mentioned above, at each time step during the transient simulation. The new values are fed back into the kinetics algorithm of TANK; a new dynamic reactivity and power levels for each fuelled site are determined. Since temperature and void reactivity feedback effects have a strong bearing on the transient behaviour of a nuclear reactor, measures are taken to determine as accurately as possible temperature distributions and void fractions in the reactor core. The next two subsections introduce the methods used in TANK.
Temperature Determinations. Temperature distributions in the fuel, cladding and coolant are continuously updated for each individual fuelled site. This process involves numerous thermalhydraulic considerations, such as the geometry and composition of the fuel pinsi coolant flow rates, density and pressure, helium gap effects (if applicable), and cladding-coolant heat transfer coefficients. Cladding-coolant heat transfer coefficients are determined in TANK for steady-state operation using the Dittus-Boelter (1,11) correlations for single-phase turbulent coolant flow. Alter the start of a simulated transient, the heat transfer package of the thermalhydraulics code SPORTS-M (12,13) is used to determine the cladding-coolant heat transfer coefficients for subcooled and saturated boiling. A set of coupled differential equations, described elsewhere (3), governs the transient behaviour of the temperatures of a representative fuel pin. The set of differential temperature equations is solved for the transient temperatures using the AECL code package STIFFZ (14) in conjunction with TANK subroutines. Thermal properties for materials used in the reactor are required in the solution. Tempera'ure-dependent thermal properties are evaluated at the appropriate temperatures (2) and, where needed, their derivatives with respect to temperature are calculated. STIFFZ is used to perform an error-controlled integration of the ordinary differential equations. During a transient simulation with TANK, STIFFZ is called for each fuelled site at each time step.
Void Determinations. The production of coolant void near the cladding surface of hot fuel elements is modelled in TANK using the heat transfer package of SPORTS-M. Void fractions are determined using the Zuber-Findlay correlation method. A weighted convolution technique is used in TANK to account for bubbles carried downstream by the coolant. This approach allows for the presence of both detached bubbles and bubbles that adhere to the hot cladding surface. The axial power distribution, as calculated by the code 3DDT, is used in determining the "worth" of the void bubbles as a function of position. In essence, the negative reactivity feedback effect of void formation is greater in the hot mid-plane of the reactor than at the end of the fuel channels.
TRANSIENT SIMULATIONS WITH THE CODE TANK Reactivity feedback effects are simulated in TANK analyses by updating the temperature- and voiddependent cross sections as conditions in the reactor 306 C N S 9th A N N U A L C O N F E R E N C E , 1988
In a TANK simulation run, the initial steady-state condition of the nuclear reactor is determined before
the; transient is started. This critical information is stored in files for later access, and provides the starting conditions of the transient. Included in these steady-state results are total reactor power, distributions of temperature and individual fuel channel power levels, average cladding-coolant heat wavisi^t c^ettitierAs i7 mk), the reactor can become super-prompt critical, resulting in a nuclear excursion with a very short period. The accident scenarios simulated below also assume that the shutdown systems fail to actuate. These accident scenarios are highly improbable, but are analyzed as part of the safety assessment of the MAPLE class of research reactors. The ramp rates considered correspond to the withdrawal, at maximum "olocity, of all three control rods; initially about 0.5 mk/s. The step insertion considered below in a zero-power MAPLE research reactor is 10 mk, sufficient to cause a super-prompt excursion.
A Transient in a DQ^-Fuelled MAPLE Research ReactoiUranium dioxide is an optional fuel mateiial for use in MAPLE research reactors with relatively low power density requirements. A ramp insertion of reactivity, initially at 0.57 mk/s, with the reactor at a full power of 1 Mtf, is simulated with TANK. The simulated reactor power and reactivity time histories during the first 18 s of the transient are shown in Figure 1. The power increases exponentially as reactivity is added to the reactor. The slope of the reactivity vs. time curve decreases because of the negative reactivity feedback effect of the increasing transient fuel temperature. Then, at a time of 7 s, the onset of subcooled boiling in the shutdown rod sites results in void formation. The resulting negative reactivity feedback effect is seen to slov dovn the power excursion. LustatjiUUes In void production rates, particularly between 13 and 15 s (Figure 1) and 19 to 20 s (not shown), cause rapid changes in the reactivity. The instability in tht reactivity curve at 19 s is too erratic to plot.
Reactor Power (MW)
c D T3 0) *-i
(0
o •u
c
040
FIGURE 1:
10 12 14 16 18
Time (s)
SIMULATED TRANSIENT BEHAVIOUR OF A 1 MW UO 2 -FUELLED MAPLE RESEARCH REACTOR DURING A RAMP REACTIVITY INSERTION OF 0.57 mk/s
A good description of a transient includes plots of the thermal behaviour of the hottest fuel channel. Accordingly, Figure 2 shows the simulated variations with time of the average fuel, cladding vail, and bulk coolant temperatures for the hottest channel. The hottest fuel channel is one of the three undeployed shutdown rod sites. The fuel temperature rises rapidly during the early stages of the transient, as shown in Figure 2. After the formation of void bubbles at a time of 7 s, the fuel temperature curve shows a gradual increase with time. At 18 s into the transient, the average fuel temperature \s ©TIVJ 190° C. 1VreOTCtfts'jKtfvJlngm a x i m a £\sel centre-line temperature is 5A0°C, far below the U0 2 melting point of 2840-C. The effects of the reactivity instabilities at 13 s and at 19 s are seen as rapid variations in the temperatures. By 20 s, the fuel temperature seems to remain constant in this simulation.
Some Transients in a Hetallle-Fuelled MAPLE Research Reactor For applications requiring higher power densities, metallic fuel materials are used in MAPLE research reactors. U3Si-Al is used in the fabrication of fuel pins for the 36-element driver assemblies. U-Al is the fuel material in the 12-element assemblies, situated in the shutdown and control rod sites and in the central fuel channel.
Rama Reactivity q at Full fow.ec. Figure shows the characteristic temperatures of the hot central channel in a simulation of a 0.42-mk/s ramp insertion transient at a steady-state power level of 10 MW. The temperatures rise monotonically until CNS 9th ANNUAL CONFERENCE, 1988
307
9.2 s when the onset of void formation slows the runaway excursion. Rapid fluctuations in poutr (not shown) result in the rapid temperature variations seen in Figure 3; the peak average fuel temperature is 335"C. These instabilities are caused by variations in void formation, enhanced by the high thermal conductivity of the metallic iuel. The UO;-fuelled research reactor discussed above (see Figure 2 ) is less prone to rapid changes in the average fuel and cladding temperatures because the thermal conductivity of U0 2 is only about 5X that of the metallic fuel material. The heat generated in a U0 2 fuel pin during a sudden increase in power takes longer to conduct to the coolant.
350-
300-
400-
Bulk Coolant Temperature
10 "To"
5
15
20
Time (s) FIGURE 3:
SIMULATED AVERAGE TRANSIENT TEMPERATURES FOR THE HOTTEST FUEL CHANNEL IN A 10 My METALLIC-FUELLED MAPLE. THE RAMP REACTIVITY INSERTION IS INITIALLY 0.42 mk/s.
Time (s) FIGURE 2:
SIMULATED AVERAGE TRANSIENT TEMPERATURES FOR THE HOTTEST FUEL CHANNEL IN THE U0 2 -FUELLED MAPLE
4Ramp Reactivity Insertion at "Zero" Power. Figure 4 shows the simulated time history of the dynamic reactivity during a reactivity ramp from "zero" power (nominally, 6 VI) with the coolant under full forced convection. Initially, the reactivity rises linearly at 0.44 mk/s until 15.2 s, when the negative feedback from the sudden increase in fuel temperature causes a down-turn. At 16.8 s, the rapid repeated formation and collapse of void in the hottest channels cause instabilities in the reactivity. The inverse period of the simulated transient is shown in Figure 5. An invetse period greater than 0.05 s 1 would trip the reactor, so the transient would have been halted at the beginning by a properly functioning shutdown system. In Figure 6 the average temperatures in the central channel reach plateaus, the maximum average Euel temperature being 29O"C. Observe that all the temperatures are constant at 20"C until 14.5 s. It is important to note here that simply monitoring the temperatures and the power level would not indicate that a transient was under way until after nearly 15 s into the transient.
308 CNS 9th ANNUAL CONFERENCE, 1988
•X3 V
(0 O
3-
c
4
6
8 10 12 14 16 18
Time (s) FIGURE 4:
SIMULATED RAMP REACTIVITY INSERTION OF 0.44 mk/s IN THE METALLIC-FUELLED MAPLE INITIALLY AT A POWER LEVEL OF 6 V
Figure 7 displays an important feature of TANK'S two-dimensional spatial considerat ions. During the above-0.44-mk/s ramp insert ion transient, the power in the central site levels is 1.25 Mtf, but the overall reactor power continues to climb (14.8 MW at 19 s ) . An explanation is that void format ion is occurring mainly in the central 12-element assembly, thus inhibiting its power excursion, but the power generation in the rest of the core is on the rise. This is one example of the improved realism of the sitnulation when modelled in two spatial dimensions, with TANK, rather than in zero dimensions with a point kinetics model. n O
2.5-
16
Power (*10) in the centra) fuel channel
C
2
4
6
8
10
12
1*
16
T 1 ME: ( S ) FIGURE 5: INVERSE PERIOD DURING THE SIMULATED TRANSIENT FROM 6 V. A TRIP SET POINT OF 0.05 S-1 IS USUAL FOR THE INVERSE PERIOD.
14
Time (s) FIGURE 7:
BUIK COOLANT TEMPERATURE
15
16
17
IB
19
T I WE (S) FIGURE 6:
SIMULATED AVERAGE TEMPERATURES FOR THE HOTTEST FUEL CHANNEL DURING THE RAMP INSERTION AT 6 V
COMPARISON OF THE SIMULATED TRANSIENT BEHAVIOURS OF THE TOTAL POWER AND THE POUER IN THE HOTTEST CHANNEL
Step Reactivity Insertion at "Zero" Power. As mentioned above, a very rapid reactivity ramp at near-zero reactor power results in a step insertion of reactivity. Figure 8 shows a simulated 10-mk step insertion transient at "zero" power (6 V) caused by a 4-cm "jerk" withdrawal, in 0.02 s, of the three control rods. The coolant flow in this scenario is assumed to be nearly stagnant. A nominal roolant velocity of 1 cm/s accounts for natural circulation only, and the entire reactor is at room temperature, 20°C. The reactor power pulses to 1.6 MV causing the average fuel temperature in the central site to jump to 260°C. The average bulk coolant temperature in the central fuel channel rises to 46°C from 20"C. Negative reactivity feedback caused by void formation is the major mechanism in limiting the power excursion. The core cools as the power pulse subsides; the void bubbles collapse, resulting in a rapid insertion of reactivity. Figure 9 shows thai the TANK simulation of this situation predicts a series of power bursts. The average fuel and bulk coolant temper.iC N S 9th A N N U A L C O N F E R E N C E , 1988 309
tures in the hottest channel appear to level asymptotically. This behaviour is consistent with certain subcooled excursion tests performed with the BORAX-I (Boiling Reactor Experiments) and the SPEBT-I (Special Power Excursion Reactor Tests) reactor (16-19). To demonstrate the validity of TANK when used to simulate transients in metallic-fuelled research reactors, the code will be benchmarked against the results of certain SPERT I transient excursion experiments (19).
FUEL TEMP (#C X 100) 2.5
1/1
2
S
FUEL TEMP (*C X 100)
,.»H
a 2
COOL TEMP ('C X 100)
a Ld <
1.5
POWER
o
(MW)
Z TIME (S) FIGURE 9: COOL TEMP (•C X 100)
SERIES OF POWER BURSTS RESULTING FROM THE INITIAL STEP REACTIVITY INSERTION OF 10 mk
REFERENCES
FIGURE 8:
SIMULATED TRANSIENT FOR A 10 mk STEP REACTIVITY INSERTION IN A METALLIC-FUELLED MAPLE RESEARCH HEACT0R AT 6 V. NEARSTAGNANT COOLANT, INITIALLY AT 20°C.
(1)
ELLIS, R.J., "Two-Dimensional Analyses of Accident-Induced Transients in the MAPLE Research Reactor," talk presented to the 3rd McMaster Symposium on Nuclear Sciences and Engineering, 1987 October.
(2)
ELLIS, R.J., CARLSON, P.A., AND SMITH, H.J., "Comparisons of the Steady-State and Transient Thermal Behaviours of \}OZ, U-Al and UjSi -Ai MAPLE Research Reactor Fuels," proceedings of 14th Annual Nuclear Simulation Symposium, Pinawa, Manitoba, 1988 April.
(3)
ELLIS, R.J., "Physics Considerations in the Code TANK," unpublished data.
(4)
ASKEW, J.R., FAYERS, F.J., AND KEMSHELL, P.B., "A General Description of the Lattice Code WIMS," Journal of the British Nuclear Energy Society, 4, (4), 564, 1966.
(5)
DONNELLY, J.V., "HMS-CRNL - A User's Manual for the Chalk River Version of WIMS," Atomic Energy of Canada Limited Report, AECL-8955, 1986 January.
(6)
CRAIG, D.S., AND FESTARINI, G.L., "The ENDF/B-V WIMS Library," unpublished Atomic Energy of Canada Limited Report, CRNL-2784, 1965 April, available from SDDO, Chalk River Laboratories, Chalk River, Ontario KOJ 0JO.
(7)
VIGIL, J.C., "3DDT, A Three-Dimensional Multigroup Diffusion-Durnup Program," Los Alamos Scientific Laboratories Report, LA-4396, 1970.
CONCLUDING SUMMARY Computer simulations of transient behaviour during accident scenarios are required in the assessment of the safety of a nuclear reactor. Two-dimensional modelling of transient behaviour provides more detail and greater realism than point kinetics methods. To this end, the two-dimensional tvo-neuuon-energygroup space-time reactor kinetics computer code TANK is being developed for use in the transient analysis of MAPLE research reactors. Using TANK, plausible simulations of transient behaviour were generated for a variety of postulated accident scenarios in MAPLE resea-ch reactors. Regardless of the fuel material, the TANK simulations show the MAPLE research reactors to be safe from fuel or cladding failure under some extreme conditions.
310 CNS 9th ANNUAL CONFERENCE, 1988
(8) OTT, K.O., AND NEUHOLD, R.J., "Introductoiy Nuclear Reactor Dynamics," American Nucleai Society, La Grange Park, Illinois, 1985. (9)
OTT, K.O., AND MENELEY, D.A., "Accuracy of the Quasistatic Treatment of Spatial Reactor Kinetics," Nucl. Sci. Eng., 36, 402-411, 1969 June.
(10)
KEEPIN, G.R., "Physics of Nuclear Kinetics," Addison-Wesley Publishing Company, Inc., Reading, Massachusetts, 1965.
(11)
TONG, L.S., AND WEISMAN, J., "Thermal Analysis of Pressurised Water Reactors," American Nuclear Society, Hinsdale, Illinois, 1970.
(12)
LEUNG, H.K.H., AND SHIM, S.Y., unpublished data.
(13)
HILLS, P.J., SHIM, S.Y., AND HOUR, S.E., unpublished data.
(14)
CARVER, H.B., AND STEWART, D.G., "STIFFZ (Subroutine Package)," written in 1978, listed in AECL FTN LIBRARY (Rev. D ) , 1983, June.
(15)
HETSRONI, G., "Handbook of Multiphase Systems," McGraw-Hill Book Company, New York, 1982.
(16) KRAMER, A.W., "Boiling Water Reactors," Addison-Wesley Publishing Company, Inc., Reading Massachusetts, 1958. (17)
NYER, W.E., FORBES, S.G., BENTZEN, F.L., BRIGHT, G.O., SCHROEDER, F., AND WILSON, T.R., "Transient Experiments with the SPERT-I Reactor," Nucleonics, U , (6), 44-49, 1956 June.
(18)
SCHROEDER, F., FORBES, S.G., NYER, W.E., BENTZEN, F.L., AND BRIGHT, G.O., "Experimental Study of Transient Behaviour in a Subcooled, Water-Moderated Reactor," Nuclear Science and Engineering, 2, (1), 96-115, 1957.
(19)
WING, A.P., "Transient Tests of the Fully Enriched Aluminum Plate-Type B Cores in the SPERT I Reactor: Data Summary Report," Phillips Petroleum Company Report, ID0-16964, 1964 June.
CNS 9th ANNUAL CONFERENCE, 1988 311
DESIGN AND SAFETY FEATURES OF THE AMPS NUCLbiAR ELECTRIC PLANT
A. F. O l i v.i and J . i J . Ht'wi I. ECS - Power S y s t e m s inv. 1^00-11J Kent Street Place do Ville, Tower b Ottawa, Ontario KIP 5P2
INTRODUCTION The Autonomous Marine Power Source (AMPS) i s a small-scale nuclear-electric power plant i n i t i a l l y concoi ved for commercial submar i ne a p p l i c^tions[1j r e q u i r i ng a long-endurance a i r independent supply of e l e c t r i c a l power. Refinement of the concept permits adaptation to p o t e n t i a l a p p l i c a t i o n s in m i l i t a r y nuclear.'dies e l - e l e c t r i c submar i nesC^j. The i nitial development e f f o r t has focussed on an extremely compelct 100 kWe p r o t o t y p e power p l a n t , but design studies indicate that on r e l a x i ng the size constraint the AMPS p r i n c i p l e s may be r e a d i l y applied i n the 400 t o 1000 kWe range. The AMPS p i ?int ut i 1 izes a low-pressure, lowtemperature reactor heat source coupled to a low- temperature organic Ranki ne cycle engi ne g e n e r a t i n g e l e c t r i c a l power. The dimensions and weight of the plant are necessarily cons t r a i ned by the scale or the submari ne, and the design incorporates f e a t u r e s of intrinsic safety, r e l i a b i l i t y , s i m p l i c i t y and ease of ope r a t i o n conditioned to the intended mobile marine a p p l i c a t i o n . The use of intrinsic s a f e t y p r i n c i p l e s leads to s i m p l i f i c a t i o n of plant systems, and thus to a c o s t - e f f e c t i v e and compact design w i t h improved l i k e l i h o o d of s o c i e t a l acceptance. This paper describes i.he key design features of the AMPS p l a n t , while placing special emphasis on safety aspects.
DESIGN FEATURES Basic Requirements and Design Approach The r e q u i r e m e n t s g o v e r n i n g the AMPS system d e s i g n are determined by the nature uf the intended a p p l i c a t i o n , namely, t o provide l e v e l s of e l e c t r i c a l power i n the 100-1000 kWe range on board small air-independent manned submari ne vehicles. These power l e v e I s are s u f f i c i e n t to meet propulsion and h o t e l loads of vessels from 600 up to approximately 2500 tonnes d i s p l a c e ment . The s t r a t e g y employed i n the design of AMPS has been t o i d e n t i f y d e s i g n features which are conducive to minimizing design and o p e r a t i n g complexity and enhancing s a f e t y and r e l i a b i l i ty. To achieve these goals, the design approach has emphasized: ( i J the i n c o r p o r a t i o n of large design and s a f e t y margins, ( i i ) the use of i n t r i n s i c safety p r i n c i p l e s , ( i l l ) r e l i a n c e , whore poss Lble, on passive rather than act i ve system concepts to meet key plant performance 312 CNS 9th ANNUAL CONFERENCE, 1988
,snd s j f e L y requi rt--m<-'nt3 , and ', i v ) provis i on of a high 1^-ve 1 of automat ion f o r pi ant operations . The i ncorporat ion of p r i nci p l - s of intrinsi c safety reduces the dependency on complex a c t i v e safety systems in assuring safe reactor operation. Reliance on physical p r i n c i p l e s inherent i n t h f r e a c t o r do s i g n , r a t h e r than on the act ion of electro-mechani cal components i ncorporated i n complex systems, renders proof of safety less e q u i v o c a l . Enhanced s a f e t y , r e l i a b i l i t y and c o s t - e f f e c t i vvness o f d e s i g n are d i r e c t outcomes of having fewer, less complex systems to operate. The simpler plant i s ^ l s o monj r e a d i l y automated, and * high degree of automation reduces the requirement f o r operating s t a f f . Operating s a f e t y i s a l s o improved through automat i o n , s i nee hurmn operators are o f t e n a weak l i n k i n the s a f e t y c h a i n . A number of i nnovat ions v i t a l to meeti ng the special demands of the AMPS a p p l i c a t ion have been i ncorporated i n i t s design. A particular challenge has been to configure i t s components t o achievF- a compact, l i g h t w e i g h t plant w i t h i n a pressure h u l l extension ^hich i s required to be n e u t r a l l y buoyant. Also, the design of an uncondi t i o n a l l j a v a i l a b l e means for passi vely c o o l i n g the r e a c t o r i n the dynamic marine e n v i r o n m e n t , under c o n d i t i o n s o f l i m i t e d thermosyphon head a r i s i n g from vessel dimensional c o n s t r a i n t s , has spawned a novel t e c h n i cal s o l u t i o n . The design of r a d i a t i o n s h i e l d ing to provide acceptable dose rates w i t h i n t.ie s h i p ' s manned spaces, despite l i m i t e d o v e r a l l system dimensions and a s t r i n g e n t weight budg e t , i s -also a considerable challenge.
Plant Design The AMPS comprises four p r i ncipa1 subsysterns, namely, a low-temperature, low-pressure, 1ight water cooled reactor heat source (RHS); a lowtemperature organic Rankine cycle (ORC) engine, an i ntegrated f a u l t - t o l e r a n t c o n t r o l and moni t o r i n g system and an e l e c t r i c a l generation and d i s t r i b u t i o n system. The i n t e r r e l a t i o n s h i p between the subsystems i s i l l u s t r a t e d i n Figure
Table 1 shows the main c h a r a c t e r i s t i c s of the AMPS n u c l e a r - e l e c t r i c p l a n t , s i zed to del i vt-r 100 kWe n e t . Also shown are the main c h a r a c t e r i s t i c s of a 400-kWe plant designed for the somewhat larger submari nest2 J.
Each of the prinvi pcil subsystems i s descri bed in the following sections.
TABLE 1 KEY AMPS PLANT
The Reactor Heat Source
PARAMETERS AMPS PROTOTYPE
General. The AMPS design features L. reactor core of relatively low power density. Cons i d e r a t i o n s of the overall si ze, weight and shielding configuration determine that the core contain enriched fuel and hydrogen moderator. The light: water c o o l a n t , together wi th the hydrogen content of th r : uranium zirconium hydride fuel, support these core design r e quirements .
AMPS
400
Configuration: Min. pressure hull diameter (m) Typical pressure hull extension (m)
3.7
7.3
11
Plant Rating at Nominal Temperature:
The reactor primary cooling system is of a lowtemperature , low-pressure design. This property o f f e r s s i g n i f i c a n t advantages since i t supports the design objectives of i n t r i n s i c safety, and he lps reduce the complexity, cost and weight of many of the pressure boundary components . A low-temperature , low-pressure design also simplifies system fabrication and nai ntenanct and can significantly reduce the riSK of failure of piping, vessels or other components over the l i f e of the plant. The s e l e c t i o n of the operating pressure and temperature for the primary heat transport system i s highly dependent on the specific application. The philosophy adopted is that, for each AMPS plant design, the lowest primary coolant operating temperature and pressure are chosen, consistent with practical conversion to e l e c t r i c a l energy. This approach minimizes p 1 ant complexity and avoids compromising the inherent safety principle cf design. The key ohallenge in the design process becomes that of oalaneing the need to enhance plant efficiency and minimize size and weight, versus the need to maintaii. the efficacy of intrinsic safety design features for mitigating accident consequences over the e n t i r e spectrum of design D^sis accidents.
Core thermal power (MW) Net power to supplied to batteries (kWe) Net efficiency (.%) Reactor Heat Source Fuel type
Fuel U-235 enrichment ( J ) Cladding Coolant No. of fuel elements Fuel element OD (em) Refuelling i n t e r v a l ( y r ) Norn. oper. p r e s s . CMPa) Design p r e s s u r e (MPa) Coolant o u t l e t temperature (°C) Coolant mass flow (kg/s) Core power density (kW/1)
1.5 100 6.7
O-Zr-H 20 SS30« H2O 85 3-75 5 to 7 0.16 1.0 95 36 26
3.5 H00
11 . 1 (see note 1)
V-Zr-H 20
Alloy 800 H20 480
1 .27 8 to 10 1.7 3.0 166 141
15
Organic Rankine Engine: Working fluid (Freon) Turbine inlet press.(kPa) Vapour temperature (°C) Condenser pressure (kPa) Condenser temp. C°C) Seawater temperature (°C) nominal range
H-11 550
83
R-113 t?10 1t9
66
80 18
35
5 0-20
20 0-30
Note 1: A net output of 500 kWe and a net pla"t efficiency of 1^.3$ i s achievable at a seawater temperature of 0°c for this plant. £ L 1 C T B I O U . FX
1
I
PEACTCH
CFKUM1C oismuunn EMJINE
JflF>CTM COCUWT ( C O a l [ CtWTJCL 4 K>1TCR]NO SIGNALS
CtKTAX 4 KMITGRING SYSTEM
LJ > WTCT 3XL1N0
ELECTnrcM- PI
H1KM ELECTRI&U.
KKH
Figure 1 ;
The /IMPS sub-3ysteras.
J
ELECTRICAL • « M « TO
Primary Cooling System. The primary coolant i s circulated through the primary heat transport (PHT) system by two pumps operated in p a r a l l e l . As indicated in Figure 2, the reactor core i s supported within a reactor vessel which, in turn, i s mounted inside a reserve of cool water that forms part of the passive cooling system. A distinctive feature of the reactor design i s the s e g r e g a t i o n of hot c i r c u l a t i n g primary coolant from the cool reserve water, except during operation of the passive cooling system. Key elements of the passive cooling system are s p e c i a l l y designed p r o p r i e t a r y hydrooynamic p o r t s r e p r e s e n t i n g openings from the core region into the reserve tank above and below the r e a c t o r core. These components prevent significant exchange of hot primary coolant and cool reserve coolant under conditions or hydrodynamic balance extant during normal pumpedflow operation. The design of these components takes into account the effect on the hydroCiMS 9th ANNUAL CONFERENCE. 1988 313
dynamic balance of the worst conditions of varying g-foroe that occur as a consequence of both dynamic vessel motion and ship attitude.
refuelling interval of between 5 and 7 years, depending on reactor operating profile. As indicated in Table 1, the core design for the tOO-kwe AMPS plant employs fuel elements 1.27 cm in diameter. This fuel design permits increased power density witnout increasing the peak fuel operating temperature, relative to that of the 100-kWe plant.
Figure 2:
Simplified AMPS flow diagram.
If pumped flow ceases or system temperature rises significantly in a departure from normal operating conditions, cool reserve water enters the PHT system via the lower openings, Is driven through the core by natural circulation, and e x i t s into the reserve coolant tank. Operation of this system occurs automatically and without operator intervention or the operation of active components. The passive cooling system also provides for the ultimate safe dissipation of decay heat to the sea, regardl e s s of the v e s s e l ' s attitude, even in the absence of on-board electrical power. The key function of the large mass of reserve coolant i s to help slow thermal transients and tempora r i l y s t o r e any excess heat generated under normal post-shutdown and accident conditions. Ultimate dissipation of this heat to the environment is effected by means of a passive reserve tank cooling loop, or alternatively, by thermal conduction through the wall of the reserve coolant tank. This process also provides a means of maintaining the reserve coolant at temperatures considerably lower than those of the main coolant circuit during pumped flow operation, despite modest amounts of heat addition arising from small levels of inadvertant exchange flow as well as radiation-induced heating. Reactor Core The core of the 100 kWe AMPS prototype p l a n t [ 3 ] c o n s i s t s of a hexagonal array of 85 uranium-zirconium hydride fuel elements clad with s t a i n l e s s s t e e l . Each element has an active length of 38 cm and a diameter of 3.75 cm. The fuel matrix contains approximately 23 wt % uranium, enriched to 20$ in uranium-235. Long term reactivity management is achieved by incorporating erbium burnable poison homogeneously interspersed in the fuel matrix. A radial beryllium reflector of nominal thickness equal to 10 cm serves to flatten the power distribution. The approximate overall dimensions of the core region, including the reflectors, are 60 cm in height by 65 cm in diameter. This configuration provides a core lifetime to f i r s t refuelling In exces3 of 1000 full power days, and yields a 314
CNS 9th ANNUAL CONFERENCE, 1988
Reactor power regulation in the AMPS is performed by the automatic control system, which calls for the insertion or withdrawal of the six neutron-absorbing regulating rods in response to the power demand signal. The safety shutdown system deploys six separate springloaded absorber rods as the basic shutdown mechanism to rapidly shut down the reactor, independent of the control and monitoring system. Fuelled follower rods are attached to the absorber elements so that the shutoff rod locations in the core are occupied by fuel when the rods are poised in their normal fully-out position. Either of the regulating and shutdown rod systems are separately capable of bringing the reactor to cold shutdown with one of their rods inoperable in the out position. Thus two independent and reliable means of rapid shut down are provided. Figure 3 is an a r t i s t ' s representation of the principal reactor assembly of the AMPS reactor, showing the reflector and various shielding components together with the inlet and outlet coolant channels. The outlines of the control rod mechanisms located within the reactor service shaft are also shown.
Figure 3=
A r t i s t ' s r e p r e s e n t a t i o n of AMPS reactor assembly.
the
Shielding The design of the shielding system is judiciously integrated with the primary heat transport system and the reserve tank to attain the general objective of achieving adequate radiological protection, while minimizing the impact of shield mass on the overall ship design. The main shielding components are internal to the reserve coolant tank boundary.
However, additional shielding is provided as necessary, depending on specific dose rate objectives established at particular locations external to the reserve tank. Lead shielding internal to the reserve coolant tank provides gamma ray attenuation, while the water shielding (which also serves as reserve coolant) reduces fast neutron leakage. Borated s t e e l p l a t e s diminish secondary gamma rays emitted in thermal neutron capture in other shield materials. Provision in the design is made for a delay tank in order to allow for radioactive decay of activation products (e.g., N1°) borne by the primary coolant prior to its entering the r e l a t i v e l y unshielded coolant piping located outside the reserve coolant tank. Energy Conversion Unit The conversion of the low temperature heat to electrical energy at reasonable efficiency is achieved by means of an organic Rankine engine u t i l i z i n g organic Freon refrigerant as the working f l u i d . The r e l a t i v e l y low primary coolant temperature of the AMPS plant precludes the use of a conventional steam-based Rankine cycle conversion unit normally found in nuclear-electric plants. Figure 2 shows the key components of the organic Rankine cycle. In this system, the hot water of the primary heat transport system is circulated through a heat exchanger, the secondary side of which serves as an evaporator ( b o i l e r ) for the Freon. The Freon vapour passes through the turbine alternator to generate e l e c t r i c i t y . The turbine exhaust vapour is then passed through a seawater-cooled condenser and the condensate is then returned to the evaporator to complete the cycle. The choice of Rankine cycle working fluid depends mainly on the temperatures of the heat source and heat sink, but parameters such as toxi c i ty , f lammabil i ty and corrosi vi ty play significant roles in determining the suitabili ty of a p a r t i c u l a r fluid in the submarine environment. In addition, rates of thermal and r a d i o l y t i c decomposition expected under the anticipated operating conditions must be considered in long-endurance applications. Ultimately, however, plant efficiency, si ze and cost are the key determinants of optimal fluid selection for a particular plant. Plant Control and Protection Control and Monitoring System. All plant control functions are implemented through the AMPS control and monitoring lystem which features on-line control, system monitoring, datalogging and reporting o p e r a t i o n s . A high degree of automation has been adopted to minimize the requirement for operator intervention in routine operation. During normal operation at significant power levels, the reactor regulating function adjusts reactor thermal power to s a t i s f y e l e c t r i c a l power requirements while maintaining process variables within prescribed operating ranges. Additional features incorp-
orated in the control system i nclude: (i) provision for operation when the load is to follow reactor power, ( i i ) automated reactor warmup and cooldown, and (.Hi) p r o t e c t i v e functions such as reactor power setbacks and turbine runbacks. The protect! ve functions normally obviate the need for shutdown system intervention. Also, in the event of single component failures, the systern is designed to effect a smooth automatic transition to "offnormal" power production modes. Central to the control and monitoring system is a fault-tolerant computer which performs digital and analogue scanni ng of all AMPS operations, input and output signal error checking, execution of control algorithms and alarm annunciation. Alphanumeric and colour graphic displays and data base generation are also provided. Tne operator interface is implemented with specialized function keyboards to permit timely and effective response to all r o u t i n e and off-normal plant conditionsCri t i c a l inputs, outputs and peripherals are duplicated or triplicated for r e l i a b i l i t y and connected to different computers. In this way, failure of a single sensor, computer or interface module can be tolerated without interruption of the functions c r i t i c a l to plant safety or availability. Tne system is ruggedized and qualified to operate over a wide range of environmental c o n d l t i o n s . It is configured to permit on-1ine replacement of hardware or software modules witliout affecting on-going control functions. Safety Shutdown System In addition to the plant control and monitoring system, the AMPS has a t o t a l l y independent safety shutdown system. The system is designed to protect the reactor from a number of postulated initiating events and t r i p parameters capable of detecting these conditions are i nstailed. The spri ngloaded shutoff rod drive mechanisms are designed such that the ability of the rods to drive into the core is independent of system orientat ion. The t r i p l i c a t e d , channelized logic system fails to a safe condition upon the loss of power. Electrical Generation and Distribution System The e l e c t r i c a l generation and distribution system for the AMPS plant is designed to convert the AC output of the two turbine alternators to a DC supply suitable for charging the submarine battery. The DC output supplies both the submarine battery, and i series of DC/AC inverters. These pulse width modulated device? provide a close control of the frequency and voltage supplied to the AMPS equipment loads. The DC busbar also supplies power directly to the uninterruptable power supply CUPS) required for the plant control and safety systems. SAFETY ASPECTS
General The AMPS possesses a number of intrinsic safety features designed to minimize the likelihood of fuel failures or core damage as a consequence CNS 9th ANNUAL CONFERENCE, 1988
315
of fuel overheating following postulated design basis accidents. These features are implemented to enable the AMPS to respond passively and without rapid operator and safety system intervention to accident conditions which threaten the release of radioactive material to the environment. The design rationale for the innovation of such features is described below, as are the expected responses of the plant to various accident scenarios.
control computer technology, with suitable redundancy at both the component and ."system l e v e l s , reduces the likelihood that control system failure w i l l lead to loss of regulation. Mojt importantly, the simplified plant design with few components and few moving parts results in fewer modes of failure, irrespective of whether they are of the random, common-mode or cross-link type. A plant of simplified design is easier to operate; thus there is also less likelihood of an operator-induced accident.
Safety Design Philosophy
AMPS Safety Design Objectives
In general, the key safety goal for the AMPS is to avoid or minimize radioactivity releases both within the ship and to the environment. The plant is designed to support the objective of ensuring a reduced occurrence probability (frequency) for at. idents having high severity of anticipated consequences in the form of radioactivity releases. This aim is consistent with standard nuclear safety and licensing practices. For the smaller AMPS reactors i t is anticipated that the permissible dose limits for accidents within each frequency category would be lower than those typical of large nuclear-power plants. This recognizes the link between the concept of permissible dose (consequence) and the benefit conferred on society by the p l a n t , which in turn is related to power output and the intended application.
In addressing the basic safety requirements described above, the following derived safety design objectives have been established for the AMPS design. These are:
As may be deduced from the foregoing plant d e s c r i p t i o n , the AMPS design incorporates safety features providing ( i ) reactor shutdown, ( i i ) heat removal from the core, ( i i i ) ultimate long term heat dissipation and (iv) containment of f i s s i o n products. Without precedent in land-based n u c l e a r - e l e c t r i c plants, these features are designed to function effectively, despite the unique constraints posed by the d i s t i n c t l y mobile marine environment. Amongst these constraints are the limitations of ( i ) head space as defined by the submarine vessel diameter, ( i i ) the ambiguity of gravitational and tnertial forces within the submarine and ( i i i ) the design for shock loadings from the effects of collisions and explosions internal or external to the vessel. A number of generic categories of accident are considered in the AMPS plant design. These include: ( i ) loss of primary coolant flow events arising, for example, from pump stoppage; ( i i ) secondary side events culminating in loss of the normal heat sink; ( i i i ) loss of reactor power regulation events and (iv) loss of coolant accidents a r i s i n g from pressure boundary failures. These event categories are considered both with and without shutdown system i n t e r v e n t i o n , although not a l l such events necessarily w i l l form part of the plant design basis. A number of measures are taken in the design to reduce the frequency of occurrence of these events. For example, the incorporation of a large margin between operating pressure and design pressure in the heat transport system s i g n i f i c a n t l y reduces the likelihood of pipe breaks. The use of advanced fault-tolerant 316
CNS 9th ANNUAL CONFERENCE. 198B
(i)
Transients which threaten fuel cladding integrity shall not require immediate shutdown system intervention to assure safety. To the greatest extent feasibl e , long periods of time on human response time scales should be available between the i n i t i a t i o n of an accident and the time of possible core damage.
(ii)
Passive cooling of the reactor core shall be available without the action of e l e c t r o - m e c h a n i c a l components, moving parts or any action by the plant operator, and should operate independently of vessel orientation and remain e f f e c t i v e despite i n e r t i a l forces arising from vessel motion.
(iii)
Loss of coolant accidents arising from piping f a i l u r e s should not rely on pumped emergency coolant to keep the reactor core covered with water, or to ensure adequate rates of heat rejection to the ultimate heat sink (the sea).
(iv)
Coefficients of reactivity should be established by design to automatically, and without any external control action, shutdown the reactor in the event of overheating.
(v)
Reactor safety should not rely on the correct functioning of hardware interlocks or administrative controls on operator action.
(vi)
No e l e c t r i c a l power sources, active safety systems or safety support systems should be required to operate after the reactor has been shutdown.
The fulfillment of these design objectives is the clue to the simplification or elimination of the complex engineered safety systems which are otherwise required. To f u l f i l l these objectives, also makes i t possible to contemplate automated unattended reactor operation. AMPS Safety Features Specific
AMPS design features which help
tn
meeting the defined safety design objectives are described in the following paragraphs: Fuel Uranium-zirconium-hydride (U-Zi—H) fuel imparts a large, prompt negative temperature coefficient of reactivity to the core, due to the characteristic neutron spectrum shift occurring with fuel heating. This feature has a number of advantages. It can compensate for accidental reactivity changes (even step changes exceeding prompt critioality) through modest but rapid changes in fuel temperature. This effect is the basis of the pulsing capability of the General Atomics TRIGA research reactors fuelled with V~Zi—H. In such pulses reactor power typically rises from 100 watts to 1000 MW and returns to its original level within 100 milliseconds. The negative temperature coefficient of reactivity ensures that the reactor will be safely and automatically shutdown in response to reactivity insertions which would otherwise threaten cladding integrity. Not only does this eliminate the need for engineered shutdown systems with hairtrigger response, it alleviates concern for events leading to rapid reactivity insertions such as uncontrolled rod withdrawal or cold water accidents. Another feature that can be shown by virtue of the large negative temperature coefficient is that, by designing for minimal installed excess reactivity and large margins between operating fuel temperatures and the safety limits for fuel cladding integrity, the reactor core is guaranteed to be subcritical at elevated fuel temperatures well below the safety limit beyond which cladding failure is expected to occur. It follows that the case for guaranteeing cladding integrity depends on the ability to demonstrate removal of decay heat alone, irrespective of the functional capability of the shutdown system in a particular accident scenario. Analytical air cooling studies of the fuel in the AMPS core for the 100-kWe plant suggest that no cladding failure will occur even following a sudden instantaneous loss of reserve tank water. This event bounds a number of accident categories in AMPS. The operant heat transfer phenomena in this event may be evaluated experimentally in future AMPS testing programs. U-Zr-H fuel offers the additional feature of an exceptionally high fission product retention capability. This characteristic of the fuel matrix reduces the fraction of the core's radionuclide inventory which is available tor release following random or accident-induced cladding failure. It therefore helps ensure, in any event, that the primary system remain relatively free of fiction products, thereby keeping total radioactivity releases low in the event of pressure boundary failure. Core Design. The low power density of the AMPS core design minimizes the rate of decay heat production in each fuel rod, thereby facilitating passive removal of decay heat even under severe cooling conditions. The large margins between the operating fuel temperatures and the defined safety limits described in the previous section are also a direct outcome of low power density in design.
The AMPS core design philosophy utilizes the negative prompt fuel temperature coefficient of reactivity, together with a low installed excess reactivity, to limit the maximum possible fission rate. This approach contributes significantly to the safe, self-limiting power excursion behaviour of the AMPS reactor. The low installed excess reactivity is implemented through careful core design optimization^]. For the core of the 100-kWe plant design, the self-limited fission power level with all control rods withdrawn is less than 200$ of full power. This quasi-steady state level is one at which the fuel can remain adequately cooled in an acceptable operational mode for many minutes before power reduction is necessary. As a result, there is no requirement for hardware interlocks to limit the rate or amount of positive reactivity addition. Passive Cooling System. The design of the AMPS passive cooling system ensures immediate and unconditional availability of a cool and massive water supply to the core, thus providing certain safety advantages typical of a pool type reactor. The same design feature also provides for a means of dissipating reactor heat to the ultimate heat sink. This unique design feature, in concert with the large negative temperature coefficient of reactivity, is the key element governing the ability of the reactor to sustain, without risk of fuel overheating, a failure of the primary pumps or a loss of the normal process heat sink, even in the absence of a rapid and immediate reactor shutdown. The capabilities of the system may be appreciated through the use of the following example. If the primary pumps should fail, exchange flow driven by natural circulation is established immediately between the core and the reserve coolant tank. Although the flow rate through the core is somewhat reduced in this cooling mode, the effective core inlet temperature is now determined by the lower reserve coolant temperature and an adequate departure from nucleate boiling ratio (DNBR) can be sustained without reactor shutdown. This scenario is illustrated in Figure 1 for the 100-kWe AMPS plant design. If reactor shutdown fails to occur in the period immediately following this event, the reserve coolant temperature rises at a rate commensurate with the rate of excess heat addition and the mass of reserve coolant. During this extended period, corrective action on the part of the operator, the control system or the shutdown system can be implemented at any time to terminate the transient. Consistent with the rise in system temperature, intrinsic mechanisms, such as those associated with the negative fuel temperature coefficient and the negative coolant void coefficient, act to continually reduce reactor power. This trend will proceed until the heat generated in the reactor core matches the heat removed from the reserve coolant tank and deposited in the sea.
CNS 91h ANNUAL CONFERENCE, 1988 317
r i p id shutdown system act ion and permi ts ampl'j time for an o p e r a t o r - i n i t i a t e d t r i p thereby reducing the number and complexity of the installed t r i p parameters.
Figure M:
Styl ized representation of AMPS passive cooling system operation following loss of pumped flow.
Departures from normal PHT system operation i n i t i a t e d by less frequent events, such as uncontrolled regulating rod withdrawal, lead to exchange flow which augments thro ugh-core pumped action. This increased cooling effect extends the a c t i o n time that can be safely allowed before correct!ve act ion needs to be implemented. The passive cooling system also plays a s i g nificant role in ensuring safe passive response to loss-of-coolant accidents. In the event of primary pressure boundary f a i l u r e , only a small fraction of the t o t a l PHT inventory is l o s t . The remaining water, at temperatures of up to 50°C, remains in the reserve coolant tank to keep the hydrodynamic ports (and therefore the core) covered so that the core decay heat can be removed by the normal shutdown heat removal method. This approach, of r e s t r i c t i n g the fraction of the t o t a l primary inventory which i s a c t u a l l y at operating temperature, and keeping i t open to a large mass of cooler water, is an effective means of ensuring a continuously a v a i l a b l e source of emergency cooling water and a guaranteed heat removal path. Safety Systems. The dependence in the AMPS design on active safety system intervention, to ensure that plant safety goals are met, is minimized through the incorporation of the inherent safety features described above. Thus, the performance requirements placed on those active systems which are installed are considerably less onerous than those typical of large power reactors. The result is that fewer and less complex systems are required to meet the safety objectives of the design. As noted above , the AMPS is equipped with a safety shutdown system which detects the i n i t i a t i o n of an accident and i n i t i a t e s reactor t r i p via the release of the safety shutdown rods. Due to the unconditional a v a i l a b i l i t y of passive decay heat removal, no safety system or operator action is required after a successful reactor trip. The AMPS is designed for short-term independence of ( i ) s p e c i f i c operator a c t i o n , ( i ) e l e c t r i c power or ( i i i ) operation of any active components. This clearly obviates the need for 318
CNS 9th ANNUAL CONFERENCE, 19B8
For cases i n which the shutdown system is impaired or i noperative, the descri bed processes, which i nherently reduce reactor power in response to a tendency toward fuel overheating, play a key r o l e i n achievement of ultimate safety. However, in order to ultimately terminate these less frequent (and possibly nondesign-basis ) events, whi ch myy represent the more severe challenge to fuel cladding i nt e g r i t y , implementat ion of a backup means of ensuring reactor shutdown i n the event of shutdown system impairment may be appropriate. Two options under consideration include a passive emergency boratlon system i n i t i a t e d by operator action or the implementation of i n dependent t r i p l o g i c which would dri ve the regulating rods into the core. The f i n a l choice w i l l be governed by r e l i a b i 1 i t y uonsiderations for such systems of low deployment frequency requirement. Due to configurational limitations arising from submarine dimensional constmints, i t may not be feasi ble to configure primary coolant pi pi ng to avoid d r a i n i n g the reserve coolant tank under a l l credible pipe break scenarios. While a i r - c o o l ing analysis can potentially demonstrate adequate fuel cool ing in this event, i t is prudent design practice to minimize the extent of f u e l overheating which would be experienced. Therefore, in certain AMPS plant designs, f a i l - s a f e primary heat transport system isolation is implemented to mi nimi ze the loss of primary heat transport system inventory following such breaks and to f a c i l i t a t e continued operation of the passive cooling system. This avoids the requirement for pumped emergency cool i n# to ensure that the f ue 1 rema i ns well-cooled under this rare category of event.
AMPS DEVELOPMENT
PROGRAM
The AMPS development program has been underway s i nee 1985. Engi neeri ng effort to date has focused on: ( i ) d e f i n i t i o n of plant systems and components, ( i i ) the development and appl i cation of plant safety and nuclear licencing c r i t e r i a and ( i i i ) the integration of plant systems into candidate submarines. To establ i s h the overall plant design and to determine operating and safety characteristics, analyses have been undertaken i n a number of areas including reactor physics, radiation transport, t h e r m a l h y d r d u l i c p , thermodynamics and the dynamic and s t a t i c c h a r a c t e r i s t i c s of components and structures. The p r i n c i p a l elements of the reactor heat souroe, the energy conversion unit and the control system of a 100-kWe AMPS prototype have been defined and engineered in d e t a i l . Work is proceeding on the engineering d e f i n i t i o n of AMPS plants suitable for naval applications in submarines of the 2000 tonne class at power levels approaching 1000 kWe. These studies w i l l serve to identify the upper l i m i t on plant thermal r a t i n g , and the p r a c t i c a l i t y of applying the AMPS p r i n c i p l e s within the config-
u r a t i o m l , space and weight constraints charact e r i s t i c of candidate submarines. To confirm that the AMPS design possesses a very low core melt probability, work is continuing on the detailed assessment of design basis and non-design basis events over an entire spectrum of accidents and plant operating modes. Test programs have been established to verify and demonstrate component and system performance and to validate the methodologies employed. To date a number of tests have been completed including: in-core materials comp a t i b i l i t y tests at the General Atomics TRIGA reactor i n s t a l l a t i o n in California, thermalhydraulic component t e s t s at the National Research Council in Ottawa, materials research at the University of Toronto and in-core tests of proprietary reactivity control concepts at the McMa.jt.er University research reactor. In order to demonstrate compliance with the objectives of the thermalhydraulic design, ECS has built a full-scale replication of an AMPS submarine reactor design using electricallyheated elements in place of the fuel rods. The f a c i l i t y f a i t h f u l l y reproduces the thermalhydraulic conditions of the AMPS reactor assembly operating at a nominal thermal power of 1.5 MW within a 3.7-metre diameter reserve coolant tank. The electrically heated fuel elements permit r e a l i s t i c simulation of the design distribution of heat flux throughout the core at all operating power levels and over a range of accident conditions. This prototype demonstration, located at Stern Laboratories in Hamilton, is capable of validating all aspects of the reactor core thermalhydraulics and the a c t i v e and passive cooling systems at power levels of up to 3 MW. The prototype is mounted in a manner to permit a change in orientation through 90°, thereby simulating boat change of attitude. However, the r i g does not simulate dynamic motion. Experiments on a t e s t r i g incorporating a single fuel element began in September 1987 and studies of all overall system with the complete full scale prototype are to be completed in October of 1988[t]. Following the completion of these t e s t s , the r i g will be used as the heat source in the prototype testing of an organic Rankine cycle (ORC) engine with a gross power output of 160 kWe. The ORC engine prototype will demonstrate plant performance and verify system design over the entire range of operating conditions and will be invaluable in providing data on the operation, maintenance and r e l i a b i l i t y of the plant. Tests of the ORC engine prototype will begin in the Fall of 1988.
CONCLUDING REMARKS
The design of the AMPS nuclear-electric generating plant seeks to deliver significant levels of e l e c t r i c a l power while meeting nuclear safety requirements at a high acceptance level. These objectives are to be met while also accommodating the severe space, weight and dynamic constraints imposed by marine propulsion applications. The intensive program of plant design and systems integration a c t i v i t i e s to date have led to the definition of a generic AMPS design covering the power range from 100 kWe to 1000 kWe. The design supports a unique safety approach which has been outlined in both general and specific terms in this paper. The AMPS power plant design and development program is providing early demonstration of the f e a s i b i l i t y of a simple, compact, safe, and c o s t - e f f e c t i v e e l e c t r i c a l power generation system for mobile marine applications in the 100-1000 kWe range. Based on progress to date, a prototype AMPS system for incorporation in a submarine could be available within the f i r s t half of the next decade. REFERENCES 1. J.S. Hewitt, P. Wilkins, G.A. Kastner, "A Nuclear Powered Submarine Vehicle for Commercial Operation", pp 20-32 - 20-36, Proceedings of the Sixth Annual Conference of the Canadian Nuclear Society, Ottawa (1985). 2. R.J. Gosling, A.F. Oliva, K. Church, "The AMPS Nuclear Reactor Based Air-Inaependent Power Source for Diesel Electric Submarines", Proceedings of the International Symposium on Conventional Naval Submarines, Volume I I , The Royal Institution of Naval Architects, London (1988). 3. R.E. Stone and A.F. Oliva, "Neutronic Design of the AMPS Reactor Core", Proceedings of the Ninth Annual Conference of the Canadian Nuclear Society, Winnipeg (1988). *). R. Gray, T. Currie, J. Atkinson, "Thermalhydraulic Experimentation in Support of AMPS Development", Proceedings of the Ninth Annual Conference of the Canadian Nuclear Society, Winnipeg (1988).
Following successful completion of these two programs, a planned AMPS nuclear reactor prototype will serve to confirm the neutronic design and provide v e r i f i c a t i o n of the nature and magnitude of the nuclear/thfcrmalhydraulic system interactions during normal operation and under simulated accident conditions.
CNS 9th ANNUAL CONFERENCE, 1986
319
SAFETY-RELATED PARAMETERS FOfi THE MAPLE RESEARCH REACTOR ANO A COMPARISON WITH THE IAEA GENERIC 10-MW RESEARCH REACTOR
PA CARLSON. A G LEE, H J SMITH AND R J ELLIS
Atomic Energy of Canada United - Research Company Whrteshell Nuclear Research Establishment Pmawa, Manitoba ROE 1L0
ABSTRACT
A summary is presented of some of the principal satety-relaled physics parameters lor the MAPLE Research Reactor, and a comparison is given to the IAEA Generic 10-MW Reactor This provides a means to assess the operating conditions and fuelling requirements for safe operation of the MAPLE Research Reactor under accepted standards
- control rod reactivity worth. - shut-down margin. - fuel temperature coefficient of reactivity, - coolant temperature coefficient of reactivity. • coolant void coefficient of reactivity and - reactivity balance
This paper compares the performance and safety characteristics of the MAPLE Re-search Reactor with the IAEA Generic 10-MW Reactor INTRODUCTION MAPLE RESEARCH REACTOR Atomic Energy of Canada Limited has developed a state-of-thean mutt
The M A P L E Research Reactor has a reacior assembly immersed m an open poo) of water The reactor assembly consists of an inlet plenum, a core gno structure surrounded by a heavy water tank (see figure T) and a chimney The core gnQ structure has 19 lattice sites arranged m a hexagonal array (see Figure 2) Table i summarizes the key features of the MAPLE Research Reactor The fuel is based upon the aiurnmum-clad uranium-silicidealummum (U SiAi) d
Many versions of the MTR-:ype pool reactor are in operation around the world and these reactors have some small but significant differences Consequently the International Atomic Energy Agency iiAEAi has developed a 10-MW generic reactor model to generally oesenbe these MTR-type poo' t&aciors This reactor model was developed to allow various laboratories to benchmark their physics codes and to establish a standard for MTR-type pool reactor perio/mance Malos and Preese |1) have analyzed the performance and saiety characteristics of the IAEA Generic 10-MW Reactor A design criterion used to guide the development of the MAPLE Research Reactor was that imporiam safety-retaied parameters should be at itast similar m magnitude or exceed corresponding safely-related parameters of typical MTR-type reactors The safety-related parameters evaluated from the physics calculations are
320
CNS 9th ANNUAL CONFERENCE, 1988
FIGURE 1: MAPLE HEAVY WATER TANK
ci
TABLE 1
MAPLE ReteTch Re»clor Key Fe»ture»
Reacior Type Steady-Stale Power Level Number of Standard Fuel Elements Number of Control Fuel Elements Number of Shutoff Fuel Elements Irradiation Channels Active Core Geometry Grid Plate Lattrce Pitch (mm| Moderator. Coolant Reflectors Coolant inlet Temperature (°C)
Pool-Type 1-iOMW
12 3 3 up to 3-in core up lo 15-m reflector 19 hexagonal sues 19 positions 801 HO D.O
CR -
CBn t r , i
s
. ,,
FIGURE 2: MAPLE CORE GRID STRUCTURE
IAEA GENERIC 10-MW REACTOR
The key features of 1he IAEA Generic 10-MW Reactor are summarized m Table 3 Figure 3 shows tfie core arrangement The core contains 23 MTR-type standard plate-type fuel elements, five control fuel elements and two irradiation elements The core is reflected on two opposile faces with graphite and is surrounded by light water
TABLE 2
LEGENO Fuel Element Deiion Description for MAPLE Research Reretor
OtTMHOMD ruCL (LEnfHT
Sundtrd Assembly Shape NumDer of rods assembly Flo* tube diameter (flat to liati immi - inner • outer Mass of uranium ig) Mass of' '0 (g) Initial linear fissile content ig ? " \ l mmi Rod pitch immi
Hexagonal
-|. ._,. . n - a • , • ; . - !
36
74 d 77 6 2126 8 4190 0 698 120
FIGURE 3: IAEA GENERIC 10-MW REACTOR CORE
TABLE 3 IAEA Generic 10-MW Re»c1or Keu FeJIurei
Control Fuel Assembly Shape Number of rods assembly Flow tube diameter |mm) - inner •outer Mass of uranium (gi Massof"" 3 \l(g) inn;ai hnear fissile content ig'^U.mm) Rod pitch circle radii |mm| - inner • outer
Cylindrical 18
60 0 62 5 1063 4 209 5 0 349 120
24 0
Reactor Type Steady-State Power Level Number ol Standard Fuel Elements Number of Control Fuel Elements Irradiation Channels Active Core Geometry Grid Plate Lattice Pitch (mm2) Moderator. Coolant Reflectors Coolant Flow Rate |m3/h) Coolant Inlet Temperature (°C)
Pool-Type MTfl 10 MW 23 5 l at Core Centre 1 at Core Edge 5 x 6 Positions 8 x 9 Positions 77x81 HO C. H 0 1000 38
CNS 9th ANNUAL CONFERENCE. 1988 321
Table 4 summarizes the fuel element specifications The standard fuel elements have 23 plates. U Si -Al Suel and 390 g U The control fuel elements have 17 plates and 288 g ! 5 U Each control fuel element has four aluminum spacer plates The fork-type absorber blades are located between pairs of aluminum spacer plates
COMPARISON OF CONTROL HOD WORTHS
The following tables contain a comparison of the safety-related parameters for the MAPLE Research Reactor with the equivalent values for the IAEA Generic 10-MW Reactor
Fuel Element Deacriptioru for IAEA Generic 10-MW Reactor Fuel Type . ^ Urcnium Enrichment iwt% '3~UI 3 Element Dimensions (mm ) Plate Thickness (mm) Water Channel Thickness (mm) Plates'Standard Element Plates'Control Element Fuel Meat Dimensions (mm3) Clad Material Clad Thickness (mm)
U Si.-AI 19 7& 76 x 80 x 600 1 27 2 19 23 17 - 4 Al Plates 0 51 x 63 x 600 Al 0 380 inner 0 495 outer 4 45 390 288
Uranium Density in Fuel Meat (gem 3 ) 1JJU Standard Fuel Element igi ' " U Control Fuel Elemem (g)
Table 6 compares ttie control rod worths for the two reactors at beginning of core (BOC) for equilibrium cores in the case of the IAEA Generic 10-MW Reactor, the equilibrium core is the one obtained after a simulated transition (Ref |1)) of 14 fuellings with a gradual transition from HEU (93 w/o) to LEU (19 75 w/o) After this, a number of LEU refuellings are required to reach equilibrium In the case of the MAPLE Research Reactor, the only fuel used is the LEU and the equilibrium is achieved by a gradual replacement of the zirconium block irradiation sites m the initial fresh core with fuel, until only one of these sites remains in the central position for the equilibrium core The results show that tne reactivity worths of the control systems of the two reactors are very similar There is one important difference All five of the reactivity shim rods on the IAEA Generic 10-MW Reactor are used for control while only ihree out of the six reactivity shim rods in the MAPLE Research Reactor are used tor control The other three are used strictly as a fast shutdown system Notice that the three control rods in the MAPLE Research Reactor, tutly inserted, have about the same worth as the five control rods m the IAEA Generic 10-MW Reactor Notice that the IAEA Generic 10-MW Reactor does not have a separate shutoff system
CODES USED IN MODELING EACH CASE TABLE 6 A summary of the computer codes used in the reactor physics modeling of each of the two reactor systems is given m Table 5 The numerical results for the IAEA Generic 10-MW Reactor are taken directly from Matos and Freese 11 ] in the comparisons given in this work The results for the MAPLE Research Reactor are obtained by using the current AECL-RC versions of WIMS |2| and 3DDT (3| The cross-section oata library used was an 89-group ENDFB V in the case of the MAPLE Research Reactor while the EPRI-CELL calculations tor the IAEA Generic 10-MW Reactor were based on ENDFB.'IV A comparison of the two data sets with experiment has recently been completed (4) Aside from a few nuciides agreement is good
Control Rod Wortrn in LEU Equilibrium Cores «1 BOC
IAEA 10-MW GENERIC REACTOR Control Rod Configuration
k^
MAPLE RESEARCH REACTOR
ip(mk)
All Rods Out
1 0767
0
1 0764
0
All Rods In
0 9105
169 5
0 9097
170 2
All Rods 50°c Out
1 0101
61 2
1 0309
41 0
Comparison ol Codes U»«d
IAEA Generic 10-MW Reactor
MAPLE Research Reactor
Detailed • ansport
EPRI-CELL lENDFBIVi
WIMS (ENDFB/Vl
Core c luSion
DIF3D
3DDT
Fuel ournup and management
REBUS-3
FULMGR
322
CNS 9th ANNUAL CONFERENCE, 1986
The IAEA Generic 10-MW Reactor uses all five shim rods for control whereas the MAPLE Research Reactor uses three out of six shim rods for control
REACTIVITY CONTROL REQUIREMENTS Certain safety requirements need to be met for reactivity level control of a research reactor The following are a summary of some of the typical requirements |5|:
Sufficient reactivity worth shall be provided in the reactivity control mechanisms so that the reactor can be shut down and maintained shutdown under aJ! conditions
and this accounts for the difference from the IAEA 10-MW Generic Research Reactor
Two independent reactivity reduction systems shall be incorporated in the design One o( these systems shall be (ast acting, and one may be the normal reactivity control system
TABLE 7 Reactivity Balance Tablet lor the LEU Equilibrium Coret at BOC
A single tailure in either reactivity reduction system shall not prevent the system from completing its satety (unction when required Reactor loading changes including fuelling shall follow specific procedures covering core assembly, disassembly and modifications The reactor management should insure that all reactor loading changes have been properly reviewed An up-to-date running record should be maintained of all reactor loading changes showing the current state of the reactor reactivity balance During normal operations the reactor protective system should have a reactivity worth greater than the available excess reactivity ol the reactor It is common practice for many research reactors to have a safety margin of two on available excess reactivity
Reactivity Component
IAEA
Burnup Xe Poison Experiments Control Reserve Cold-to-Hot Swing Total Excess Reactivity Excess Reactivity x 1 5
317 150 50 30 71 2 106 8
MAPLE Ap(mk)
130 26 5 20 0
165
100 15 71 0
106 5
TABLE 8 The total rod worth less the most effective safety rod shall be gieater man the maximum excess reactivity plus the reactivity worth of the worst single luel-loading accident
Shutdown Margin* In LEU Equilibrium Cores at BOC Basis
IAEA Ap (mk)
MAPLE Ap(mk)
REACTIVITY BALANCE COMPARISON
In establishing reactivity balance tables and shutdown margins, the aforementioned reactivity contro1 guidelines must be kept in mind ftn«n selecting any fueling scheme for a research reactor In the case of me MAPLE Research fieacto' ano the IAEA Generic 10-MW Reactor, the comparison ,s made between the LEU equilibrium cores in each case The total excess reactivity in Table 7 is the reactivity worth of the beginning of core iBOCl k^ injable 6 with all rods out This is broken down into components for ' "Xe poison load, fuel burnup. reactivity loading from experiments control reserve and cold-to-hot swing The shutdown mergins are then displayed in Table 8 The first line m this table is the reactivity worth of all shutdown rods fully deployed minus the total excess reactivity from Table 7 in the second line one subtracts 1 5 times the total excess reactivity from the shutdown margin The factor of ' 5 is a safety margin (Condition Si some have suggested as minimal shutdown margin for all shutofl roas deployed In the third line of Table 8 the figures represent the reactivity worth ol all shutoff rods deployed excepi one minus the total excess reactivity The one rod assumed excluded is the rod of greatest worth This figure should be at least 10 mk for a safe shutdown margin in the event that one rod fails to deploy iCondition 7| One facf to observe is that the MAPLE Research Reactor uses three out of six reactivity shim rods tor shutdown and has separate control and shutoff systems The IAEA Generic 10-MW Reactor uses the same system of five reactivity shim rods for both control and shutdown in spite of using three instead of five, the shutdown margins using all rods (or the MAPLE Research Reacior are equivalent to those (or the IAEA Generic 10-MW Reactor Even if one rod is lacking the margin is still above 10 mk for MAPLE and therefore acceptable It is natural that losing one out ol three will be more significant than failing one out of five.
Toial Excess Reactivity and All Rods In
98 3
99 2
Excess Reactivity x 1 5 and All Rocs in
62 7
63 7
Total Excess Reactivity and Max Worth Rod Out
28 9
181
COMPARISON OF KINETICS PARAMETERS
Table 9 shows a comparison of kinetics parameters along with reactivity coefficients for the IAEA Generic '0-MW Reactor and the MAPLE Research Reactor The first parameter is the neutron generation time m microseconds This is the ratio of neutrons present in the core to the generation rate of fission neutrons It equals the prompt neutron lifetime divided by the neuiron multiplication factor k The values for the IAEA Generic 10-MW Reactor and the MAPLE Research Reactor are both in the 40 to 45 range The next factor is the effective delayed neutron fraction This is given in percent of total fission neutrons Even for a uniform reacior core, the effective fraction differs from the actual fraction 16) because allowance is made in the former for the tact that the delayed neutrons have lower energies than the prompt fission neutrons Thus the delayed neutrons usually have a greater importance in thermal reactors than do the prompt neutrons, and in some cases the difference may be as much as 20% The value given in Table 9 lor the IAEA Generic 10-MW Reactor is trie effective delayed neutron fraction, while the value given tor the MAPLE Research Reactor is the actual delayed neutron traction Initial estimates of the effective delayed neutron fraction for MAPLE give a
CNS 9th ANNUAL CONFERENCE, 1988 323
value of 0 736^ Thus both reactor systems have reactivity margins to prompt critical of about 7 3 mk per dollar oi reactivity Following this, the reactivity coefficient figures are given The MAPLE Research Reactor figures are (or a fully fresh core, where extra Zirconium blocks, each wnh three H.O filled holes, are included This is •ntenced to provide a worst case as adding extra driver fuel increases the undermoderatioo Although not as large as the IAEA Generic 10-MW Reactor values, the MAPLE Research Reactor reactivity coefficients are still significantly negative for fuel temperature, coolant temperature and •Old
MW Reactor *. Applications in Nuclear Data and Reactor Physics. 0 £ Cullen. R Muranaka. and J Schmidt, ed , World Scientific Publishing Co Pte Ltd. Appendix A-2. p796, 1986 (2) Askew. J R and Fayers. f J . 'A General Description of the Lattice Code WIMS". Journal of the British Nuclear Energy Society, «(4).564.1966 (3|
Vigil, J C . '3DDT • A Three-Dimensional Multigroup Diffusion Burnup Program'. Los Alamos Scientific Laboratories Report. LA4396, 1970
(4)
Okazahi, A . 'Trip Repoa IAEA Advisory Group Meeting on Nuclear Data tor the Caicul&tion of Thermal Reactor Reactivity Coefficients *. Private Communication. IAEA. Vienna. 7-10 December 1987
TABLE B Klr.otict Parameter! for LEU Equilibrium Corel Parameter
IAEA
(5) Axford D J . 'Draft IAEA Safety Guide Research Reactor Safety Criteria", private communication from Consultants Group Meeting. IAEA May 4-7. 1987
MAPLE (6)
42 4
Generation Time. A IJJS) Delayed Neutron Fraction. 0
(%)
Water Temp Only(mk°C.
45 0
0 7311
0 687
-0 0737
-0 0426
Fuel Temp Only imk'°C)
-0 0247
-0 0103
Void Coetf (mk % Voidi (O-iCe Voidi
-2 7
-2 0
CONCLUSIONS AMD SUMMARY
The MAPLE Research Reactor control and shutotl system has significant reactivity kvonn • one of the systems 13 out of 6i worth is the same as eroire 5 rods in the IAEA Generic 10-MW Reactor This allows similar reactivity balance conditions There are different control and shutoff systems for the M A P L E Research Reactor - each with the same worth because of symmetry Out differently instrumented Each of me separate systems
REFERENCES
[i|
324
Maios J E and Freese K E . "Safety Analyses for HEU and LEU Equilibrium Cores and HEU-LEU Transit/on Core IAEA Generic 10
CNS 9th ANNUAL CONFERENCE. 1988
Bell. G I . and Glasstone. S . 'Nuclear Reactor Theory' Van Nostrand Remhold. New York. 1970. p472
THERMALHYDRAULIC EXPERIMENTATION IN SUPPORT OF AMPS DEVELOPMENT R.G. G r a y , T . C . C u r r i e , J . C . A t k i n s o n ECS - POWER SYSTEMS INC. Ottawa, O n t a r i o
ABSTRACT
An experimental program has been initiated to investigate thermal hydraulic behaviour of the reactor heat source of the Autonomous Marine Power Source (AMPS). The program has been designed in phases to support a progression which culminates with full-scale integrated systems tests of an electrically heated AMPS reactor core at thermal powers of up to 3 MW. Significant progress has been made to date and preliminary r e s u l t s of the separate effects t e s t s are very encouraging: the core has demonstrated thermalhydraulic s t a b i l i t y and a high resistance to dryout. This paper outlines the phases of the program as currently envisaged and presents preliminary results from the i n i t i a l testing. INTRODUCTION The Autonomous M a r i n e Power Source (AMPS), which i s under a c t i v e development by ECSPower S y s t e m s , i s a new g e n e r i c t y p e of
n u c l e a r - e l e c t r i c plant for ai r-independent submarine a p p l i c a t i o n s ^ . The p l a n t incorporates a low temperature, low pressure, water-cooled reactor coupled to an organic Rankine cycle heat engine for the generation of electrical power. The primary heat transport system of AMPS has been designed to incorporate both active and p a s s i v e cooling c i r c u i t s . This approach accommodates the operating temperatures that a r e necessary for obtaining an acceptable conversion efficiency under normal forced cooling conditions. It also allows for an exchange flow, which i s driven by natural convection, for the removal of decay heat under reactor and pump shut-down conditions, and for a d d i t i o n a l cooling that may be necessary because of off-normal operating conditions (e.g., impairment of primary system pumps). To provide a sufficient reservoir of coolant at lower temperatures for use during passive cooling, the reactor assembly is immersed in a large inventory of water contained in the r e s e r v e c o o l a n t tank (RCT). Special hydrodynamic components (HDC's) in the upper and lower legs effectively isolate the reserve coolant from the primary system under normal o p e r a t i n g c o n d i t i o n s , without a c t u a l l y providing a physical barrier to exchange flow. For normal operation, the pressure drop in the core assembly is balanced by the hydrostatic head in the RCT. For operation under offnormal or shut-down conditions, a pressure
imbalance will exist in the c i r c u i t . These same components then default by instrinsic means to the passive mode thereby allowing exchange flow with the RCT. Experimental and full-soale testing programs have been initiated to support all aspects of the AMPS integrated research and development effort. This paper will highlight only the f u l l - s c a l e , n o n - n u c l e a r thermal hydraul i c experiments of the reactor heat source design which are currently underway. The reactor systems simulated for these tests are those of the AMPS 1500 kW^ design, with over-power provisions to accommodate testing at power densities up to twice those expected in the actual reactor. The o b j e c t i v e s of the thermal hydraul ic experiments are to: 1) demonstrate the operational and safety objectives of t h e AMPS r e a c t o r thermalhydraulic design. 2) provide data for design feedback, 3) a l l o w computer code development and verification, and M) provide design verification of the active and passive cooling systems. PROGRAM PHASES
Although c o s t l y , f u l l - s c a l e t e s t s of AMPS hardware are affordable and have been employed throughout the thermalhydraulic t e s t program to d e m o n s t r a t e the technological and safety features of AMPS. For the most part, the geometries of the core, the cooling systems, and the key thermalhydraulic components have been accurately represented. Thermalhydraulic experimentation has been planned as a program encompassing three major phases. Phase I, which will not be described in this paper, was a preliminary experimental study of the passive cooling concept and was performed at the hydraulic testing laboratory of the National Research Council in Ottawa. Phase I results were used in finalizing the prototype design of the hydrodynamic components for more advanced tests during Phase II oC the program. Phase I I , currently in progress, consists of f u l l - s c a l e separate effects t e s t s of the c r i t i c a l components of the reactor he.it source. Specifically, this phase is to include complete thermalhydraulic t e s t i n g of the following configurations: a s i n g l e simulated fuel element and flow channel, a simulated reactor
CNS 9th ANNUAL CONFERENCE, 1988
325
core assembly, and the HDC's which are critical to both the forced and passive cooling systems of AMPS. Phase III will consist of comprehensi ve systems t e s t i n g of the fully integrated components, i ncluding the RCT. This test plan includes both normal operation and simulated accident conditions. Phases II and III are to be completed at Stern Laboratories I n c . , formerly the System Test Laboratory of Westinghouse Canada, in Hamilton,
pressure drop across the simulated grid p l a t e s . The behaviour of core hydraulic r e s i s t a n c e i s p a r t i c u l a r l y i m p o r t a n t in the AMPS design because of i t s effect on the pressure balance which inhi bi t s exchange fl ow between the core and RCT under normal operating c o n d i t i o n s . Secondary o b j e c t i ves of the s i n g l e element t e s t s i n c l u d e d e v a l u a t i o n of heater element d e s i g n and s h e a t h thermocouple attachment method, as well as f u r t her development and verification of the d a t a a c q u i s i t i o n dnd control software.
Ontario. PHASE I I : FULL-SCALE COMPONENT TESTS Single Element
Tests
The s i n g l e element t e s t s included hydraulic r e s i s t a n c e and c r i t i c a l heat flux (CHF) t e s t s over a range of conditions r e p r e s e n t a t i v e of both normal and abnormal operation within the AMPS c o r e .
3-*^
I*
H6
FIGURE 1:
Schematic diagram, test condition ranges, and instrumentation for the single element test facility (Phase II).
The flow channel was sized to be representative of a single subchannel within the AMPS core. These t e s t s t h e r e f o r e provided an early o p p o r t u n i t y f o r an e v a l u a t i o n of the thermalhydraulic characteristics of the AMPS core. Several c h a r a c t e r i s t i c s were of particular interest: nature of sheath dryout (location, rate of rise of sheath temperature, drypatch propagation), the boiling regime along the channel, the pressure drop dependence along the heated length, and the magnitude of the
326
CNS 9th ANNUAL CONFERENCE, 1988
FIGURE 2:
Test
Facility
and
E l e c t r i cally Simulated Fuel Elements.
Procedure.
The loop
and
v e r t i c a l upflow t e s t section of the single e l e m e n t t e s t f a c i l i t y , i n c l u d i n g the e l e c t r i c a l l y simulated fuel rod, are shown schematically in Figure 1. A heated surge tank
was uMIized for pressure control and for the i n i t i a l deaeration of the loop water. The diasolved oxygen coit ent was reduced to 1 ess than 20 yug/l by o o i l ing the loop water at atmospheric pressure in the surge tank before the s t a r t of the t e s t s . Adjustment of the secondary side cooLing water flow was used to vdry the loop operating temperature. A main l o o p v a l v e provided flow c o n t r o l w i t h a solenoid-operated rewet valve connected i n par a l l el to i n i t i ate rewetti ng of the sheath during the CHF t e s t s . Pressures, temperatures and flows were measured at various locations around the c i r c u i t as shown in Figure 1. The a I emen t was instrumented w i t h sheath thermocouples, the majority of which were semiburied i n the sheath w a l l . The outside surface of the element was grooved to accommodate the thermocouples and the t i ps were brazed i n place t o ensure good thermal contact and mi nimi ze f 1 ow perturbat ions . The sheath thermocouple l o c a t i o n s were staggered both a x i a l l y and circumferentially as shown in Figure 1. The t e s t section was also i nst rumen ted wi th sensitive p r e s s u r e t r a n s d u c e r s to y i e l d absolute pressure measurements, as well as differential pressures across the simulated g r i d p l a t e s and the heated l e n g t h of the element. Polycarbonate tubing was used to provide a transparent test section for some tests . I n more severe t e s t s , a steel test s e c t i u i i was necessary w i t h a polycarbonate viewing s e c t i o n above the upper grid plate s i mul at or . The transparent sections allowed visual observation of the boil ing regime under different operating conditions for the q u a l i t a t i v e determination of the extent of void formation w i t h i n the flow channel. A video camera wi th recorder, and " s t i 11** photography were used to record the b o i l i n g regime. Figure 2 i s typical of the e l e c t r i c a l l y heated elements f a b r i c a t e d i o r the experimental program, including the single element t e s t s . As shown f o r the disassembled element, the f ilament r i bbon is wound on a cerami c core. The pi tch of the r i bbon i s vari ed to achi eve tht? required axial power d i s t r i b u t i o n . Signals from the loop, test SQction and heater element sensors were patched into the lab 1 s computer-based data a c q u i s i t i o n system for moni t o r i ng , r e c o r d i ng and control purposes . Capabi1ities of the system include parameter calculation ( e . g . , calorimetric power, dryout t h r e s h o l d based on temperature averaging), conversion to engineert ng uni t s , and on-line video display of data with alarm annunciation. The data were stored on magnetic tape for data reduction o f f - l i n e . S t r i p chart recorders were also connected to the sheath thermocouple analogue signals to obtain conti nuous records during the CHF t e s t s .
Test conditions (power, flow, i nJ e t temperature, o u t l e t pressure) were set t o appropriate values, and allowed to s t a b i l i z e for each teat poi nt pri or to the capture of sensor out puts with the data a c q u i s i t i o n system. For the CHF t e s t s , the power supplied to the element was slowly increased u n t i l dryout occurred or the power l i m i t of HU kW was reached. Tests were performed at sevoral combinations of mass flowrate, outlet pressure and i n l e t temperature. (The complete range of test oondi tions for the single element t e s t s , as well as the nominal design points, are shown i n Figure 1 . ) The data acquisition system sampled the sheath thermocouples automat i cai L y in the transient mode and checked for sheath dryout. At detect ion of dryout, the rewet valve was tripped open permitting cold water to flood the test section. Power reduction to the element, under computer c o n t r o l , was also used to aid the rewet process. P r e l i m i n a r y Results and Conclusions. The i ns trurnented, e l e c ' r i c a l l y heated el ement demonstrated i t s robust qualities duri ng the si ngle element t e s t s , having survi ved a very severe series of tests. The s h e a t h thermocouples performed well as did the modif i c a t i ons whi ch were made to the upper end f i t t i n g t o minimi ze v i b r a t ion and possible fretting. The s i n g l e element t e s t s , w h i l e o p e r a t i n g within the design envelope, demonstrated stable thermalhydraulic v e r t i c a l flow c h a r a c t e r i s t i c s . The heated length pressure drop was measured under adiabatic and diabatic conditions during the hydraulic resistance t e s t s . The variation of pressure drop Cat desi gn condi tions) wi th element power was found to be minimal . At des i gn condi tions, the element operates wi th subcooled nucleate b o i l ing over the e n t i r e heated l e n g t h . Under more severe abnormal condi tions, the sheath remains effectively cooled despite very high voiding. Sheath dryout was d i f f i c u l t to i n i t i a t e during the CHF t e s t s . Dryout i n i t i at ion generally required low flow and low system pressure at high power. Figure 3 shows the data from a typi cal t e s t . At an outlet qual i t y of about \0%, an incremental increase i n power i n i t i a t e d p a r t i a l sheath dryout at thermocouple 3(T3), l o c a t e d 110 mm from the end of the 381 mm heated l e n g t h . As expected, dryout was i n i t i a t e d further upstream at lower flows. G e n e r a l l y , s h e a t h temperatures increased rapidly following dryout. The dryout c r i t e r i o n used by the data acquisi t i o n system often permitted sheath temperature spikes of greater than 10C^C . The highest sheath temperature measured during the CHF tests was about 700°C,
Various steady-state runs were completed at element powers up t o 31 kW (where 17.6 kW represents the average design power per element ) i n the hydraulic resistance t e s t s .
CNS 9th ANNUAL CONFERENCE. 1986 327
compared. As shown, the dryout powers are s l i g h t l y higher at 200 kPa than 110 kPa.
T II, i
90 70 50 30
Sin,l
FIGURE
CJ a T (u» I'CJ
Sr>tt<
30 5D 70 90
O A
• •
F i w . l . (»(/.>
P l o t of s i n g l e element CHF t e s t data showing t h e e f f e c t s of mass
velocity and inlet subcooling at constant outlet pressure.
FIGURE 3 :
Results from CHF test 162 of the single element test s e r i e s .
The dryout power was found to be greater than the i4UkW power l i m i t in a l l of the t e s t s performed at flowrates above 0.15 kg/s, which is less than to" of the single element design flowrate of 0.39 kg/s. Typical results of t e s t s performed a t lower flowrates {<_ 0.1 kg/s)and various inlet temperatures (T in) are summarized in Figure 1 for an outlet pressure (p o u t ) of 200 kPa. (Similar plots were produced from data at 110, 160 and 240 kPa.) The single element mass flowrate, ma3S velocity and corresponding full-core mass flowrate are p l o t t e d on the abscissa while the single e l e m e n t power, a v e r a g e heat f l u x and .orresponding full-core power are plotted on the ordinate. The data plotted in Figure 4 indicate an almost linear relationship between dryout power (CHF) and mass velocity for a given i n l e t suboooling (AT sub) and outlet pressure. The CHF d a t a a l s o e x h i b i t a p r e s s u r e dependence, as can be sec-, in Figure 5 wherethe data obtained at 110 KPa and ZOO kfa with subcoolings of 32°C and 30°C respectively, are
32B
CNS 9th ANNUAL CONFERENCE. 1088
FIGURE 5:
Plot of single element CHF test data showing a pressure dependence at constant inlet subcooling.
Assembled Core Tests
The primary objectives or the full-scale core assembly tests are similar to the single element tests hence a complete test matrix of hydraulic resistance and CHF test points was specified. Other key objectives of the assembled core tests were to determine the distribution of frictional pressure losses throughout the assembly, and the flow distribution within the core. Similarly, the CHF test portion was
expanded to Include evaluation of the effect of t e s t s e c t i o n i n c l i n a t i o n , which effectively s i m u l a t e s p o s s i b l e l i s t i n g conditions in a submarine application.
shape and seam welded, is abuut 58 cm in height with a diameter of about ^3 cm.
Test Facility and Procedures. The feeder loop for the assembled core t e s t s includes loop hardware and instrumentation similar to that already described for the single element t e s t ; e . g . , heated surge tank, main heat exchanger, pump, flow control valves. As well, condensing spray headers were added to the outlet lines to accommodate the higher outlet qualities of the full-core t e s t s . A partial section of the t e s t section assembly with pressure instrumentation is shown schematically in Figure 6.
FIGURE 7:
FIGURE 6:
A p a r t i a l section of the full-scale core assembly (Phase I I ) showing one of the four pairs of inlet-outlet l e g s , the cooling flow path, and differential pressure transducer locations. Not shown are the flow measurement l o c a t i o n s for each i n l e t - o u t l e t l e g , and the absolute pressure transducer location at the s t a r t of the heated length.
The outer 2-m s h i e l d i n g tank,which i s an integral p a r t of the assembly, provides radiological protection in the actual nuclear application. The simulated r e a c t o r core assembly i s installed in a concentric inner core pipe welded inside the tank. Figure 7 shows the simulated r e a c t o r core assembly with the upper flow modifier plug (which includes the upper plenum) removed. The core, which was formed into a 6-sided hexagonal
Simulated r e a c t o r core assembly showing the upper grid plate, the hox-.ijonal shaped core, and the lower core support with i n l ° t plenum.
Electrically heated elements are installed in the 79 fuel rod positions, 33 well as in the 6 f u e l - f o i l owed shut-off rod positions of the core l a t t i c e . Separate, computer-controlled DC power supplies feed each ring of elements to duplicate the reactor radial power distribution. Full-length tubes are used to simulate the regulating rods, some of which are transparent t o permit borescopic visual examination of the core. Core instrumentation includes a d i f f e r e n t i a l pressure transducer connected across the c o r e , and 78 sheath thermocouples (staggered both circumferentially and a x i a l l y on the elements) located in p r o b a b l e dryout l o c a t i ons throughout the lattice. Flow velocities are measured in five subchannels with axially transversing p i t o t s t a t i c tubes . I n s t a l l a t i o n of the simulated reactor core assembly into the shielding tank is shown in Figure 8. Visible in the figure are the 3-m long calming lengths connected to each of the four outlet legs. A similar piping arrangement was also required for the inlet legs to permit a c c u r a t e measurement of the differential pressure between the inlet and outlet legs.
CNS 9th ANNUAL CONFERENCE. 1988 329
The CHK tests demonstrated that the core i s even more resistant to dryout than indicated by the s i n g l e element t e s t r e s u l t s . With a vertical core, for instance, the dryout power exceeded the hardware-imposed power l i m i t of 3MW at a l l of t i e coolant i n l e t temperatures and pressures examined when the f l o w was greater than 3 '
FIGURE; 8:
I n s t a l l a t i o n of the s i m u l a t e d ru.ictor core assembly ( w i t h the uppm- flow plenum attached) at Stern Laboratories in Hamilton, Ontario.
The proc:dur^;j for the hydraulic resistance and CHF tests for trw assembled core were similar to those described for the s i n g l ' element tests. The ranges of pressure t nd temperature examined were also s i m i l a r . However, the ranges of mass flowrate and power used (up to 50 kg/s and 3MW r e s p e c t i v e l y ) were representative of those for a f u l l - c o r e , as opposed to a single "hannel . Additionally, CHF tests included Lest section inclinations of 0 ° , 22 1/2° and yo° from the v e r t i c a l . Preliminary Results. The assembled core tests were completed during the preparation of this pape . Preliminary analysis of the hydraulic resistance test data has shown the magnitude of the f r i c t i o n a l pressure drops throughout the test section to be in agreement with expected values, w i t h the possible exception of the pressure drop between the outlet plenum (above the core) and the outlet legs. This pressure drop i s higher than expected and i s being investigated f u r t h e r . (Ttie expected values are based on theoretical estimates and the results from the single element tests.) The variation of heated length with reactor power was found to minor at d e s i g n c o n d i t i o n s , agreement with the single element
pressure drop be r e l a t i v e l y which i s in results.
Reasons for the improved core CHF performance, which are assumed to include the effects of cross-flows, turbulent mixing, and the very low annulus r a t i o (outer diameter /inner diameter) of the annular test section used in the single element t e s t s , are currently being assessed. The same CHF dependence on mass velocity and pressure observed for the single element t e s t s , was also apparent in the f u l l - c o r e , low flow data. The dryout location for a l l v e r t i c a l tests was in the c e n t r a l , high power density region of the uore. As expected, the Uryout location spreads out from the central region towarcs the top as the test f a c i l i t y is i n c l i n e d . At an i n c l i n a t i o n of 22 1/2° to the v e r t i c a l (the maximum permanent l i s t design p o i n t ) , dryout occurred at thermocouples in the outer rings at the top, as well as i n the central core region. The data i n d i c a t e t h a t t h i s e f f e c t i s more pronounced at low flows. At an i n c l i n a t i o n of 90°, the dryout location i s predominately at the top of the core, i n the outer element r i ngs. For the most part, the dryout powers ranged from about 0 to 20? less for' the 22 1/2° inclined core t e s t s , when compared to results f o r the v e r t i c a l core under similar coolant conditions. As expected, for the 90° inclined core, dryout powers were s i g n i f i c a n t l y less. For example, at 90° the dryout power was 1125 kW (at an outlet pressure of 165 kPa, i n l e t temperature of 75°C, mass flow of t k g / s ) , compared to 2330 kW at 22 1/2° (at 160 kPa, 65°C, 3 k g / s ) , and >3000 kW for the equivalent v e r t i c a l core. The highest dryout power measured during the 90° inclined-core tests was 2523 kW (at 160 kPa, 75°C, 28 k g / s ) . Hydrodynami o Component (HDC? Tests
The i n i t i a l analysis of the dynamic pressure measurements i n the core has shown (at design c o n d i t i o n s ) a r e l a t i v e l y f l a t flow-velocity d i s t r i b u t i o n at the core i n l e t . The flow d i s t r i b u t i o n becomes progressively more peaked as the probe.'i are moved further up the core. As expected, the d i s t r i b u t i o n shows a Reynolds number dependency as i t also becomes more peaked at lowur Reynolds numbers.
330
CNS 9th ANNUAL CONFERENCE. 1988
The HDC's are an integral part of both the forced and passive cooling systems of AMPS. The HDC t e s t s , which are currently underway, are intended to v e r i f y the f i n a l design of the components, and t o determine their detailed pressure l o s s c h a r a c t e r i s t i c s . This i s important i n AMPS because the segregation of the hot primary f l u i d from the cool RCT water
during normal operation r e l i e s on a pressure balance condition. Similarly on departure from normal operation or for shut-down cooling, the c i r c u i t pressure loss must be low to enable h e a t removal from the c o r e t o the reserve coolant by natural convection. For submarine a p p l i c a t i o n s , final heat d i s s i p a t i o n from the RCT to the sea i s by conduction through the pressure hull or by means of a passive reserve tank cooling loop. A secondary o b j e c t i v e of these t e s t s i s to v e r i f y thermalhydraul i c computer codes which have been developed at ECS, and which are used t o predict exchange flows with the RCT under various c o n d i t i o n s . The experimental setup for these t e s t s will include f a c i l i t i e s already fabricated for the previous t e s t s ; e . g . , the instrumented feeder loop. The t e s t s e c t i o n will be instrumented with t e m p e r a t u r e , flow and pressure sensing transducers. The t e s t conditions will include a full r a n g e of s t e a d y - s t a t e o p e r a t i n g conditions. PHASE I I I : FULLY INTEGRATED SYSTEM TESTS Phase I I I of the program w i l l c o n s i s t of comprehensive systems t e s t i n g of the f u l l y i n t e g r a t e d components, including the RCT. The t e s t s e c t i o n i s shown schematically in Figure 9.
9. There are a c t u a l l y eight HDC's, one i ;i each of the four i n l e t and four o u t l e t l e g s . The hydrodynamic components are i n s t a l l e d between the nozzles and the surfaces of the mounting pads. They provide openings into the iKT while allowing unimpeded through-flow during forced cooling. For forced cooling, the flow path i s from the t e s t loop, into the i n l e t r o t a t i n g coupling, and through the i n t e r n a l piping (installed within the RCT) which directs the flow to the inlet of the HDC's. The flow then passes into the inlet plenum, through the core and out through a similar route to the test loop return l i n e . The passive cooling flow path is also shown in Figure 9. During passive cooling, a thermo-si phon head is established through the core, with cold water being drawn from the RCT, through the lower HDC and into the core for cooling. The core water is returned to the RCT via the upper HDC's. For the system t e s t s , cooling coils are surface mounted on the outside of the tank to cool the reserve coolant. The p r e l i m i n a r y t e s t plan for the f u l l y integrated system includes tests under normal operating and selected accident conditions. Applicable normal operating modes will include plant start-up, run-up to full power, constant power operation, power cycling, plant shut-down and equivalent decay heat removal during passive cooling. System testing is also to include analysis under simulated accident conditions - both accidents leading to shutdown and accidents without shut-down. Specific accident scenarios to be tested will include 1oss-of-primary system pumped flow, loss-ofreactor power r e g u l a t i o n , loss-of-secondary side heat sink and may also include simulated loss-of-cool ant accidents. ACKNOWLEDGEMENT
The authors wish to acknowledge the significant contribution made by Stern Laboratories in the design and construction of the test f a c i l i t i e s , and in performing the tests. REFERENCES 1) Hewitt, J . S . , "The AMPS 1.5 MW Low-pressure Compact R e a c t o r " , Proceedings of the f i r s t IAEA I n t e r n a t i o n a l Seminar on SMALL AND MEDIUM-SIZED NUCLEAR REACTORS, Lausanne, S w i t z e r l a n d , 1987 August 2H - 26. FIGURE 9:
S c h e m a t i c r e p r e s e n t a t i o n of the f u l l y i n t e g r a t e d systems test f a c i l i t y (Phase I I I ) showing p a r t i a l s e c t i o n s , and the forced and passive cooling flow p a t h s .
T h e s h i e l d i n g water t a n k , including the core assembly, i s i n s t a l l e d within the t-m diameter r e s e r v e c o o l a n t tank. The support for the f u l l - s c a l e assembly allows i n c l i n a t i o n through 90°, simulating permanent l i s t conditions of a submarine h u l l . The location of only one HDC i s shown in Figure
CNS 9th ANNUAL CONFERENCE, 1988 331
PREDICTION AND MEASUREMENT OF THREE-DIMENSIONAL TURBULENT .JETS UNDER SUCTION AND COUNTERCURRENT FLOWS IN A MAPLE-TYPE TEST FACILITY
S.Y. Shim, J.E. Kowalski, D.K. Baxter and R.L. Hembroff Atomic Energy of Canada Limited Whiteshell Nuclear Desearch Establishment Ptnawa, Manitoba, Canada ROE 1L0 rel.:(2O4)753-23U Telex:07-57-m
ABSTRACT
INTRODUCTION
Flow patterns In the 1/5-scale hydraulic model of the MAPLF, (Multipurpose Applied Physics Lattice Experimental) research reactor were observed experimentally and were simulated numerically. The study was necessary to ensure that the core jet, which contains short-lived high-level radionuclides, is contained in the open chimney of MAPLF, under normal operating conditions. Two selected experiments were simulated using a three-dimensional flow model, MAPL3D, to study the effects of chimney height and core velocity on the flow patterns in the chimney. The numerical results correctly predicted these effects and the unique flow patterns in the chimney.
A new multipurpose research reactor called MAPLE has been developed by Atomic Energy of Canada Limited [1]- The MAPLE class of reactors is designed to generate a maximum thermal output ranging from 1 to 30 MW. It is a light-water-cooled research reactor with an open-chimney-in-pool arrangement.
NOMENCLATURE A a g p r S T t At u v Vol w z
Area Convection-diffusion coefficient Gravitational acceleration Pressure Radial distance Source term Temperature Time Time step Velocity component In the circumferential (}) direction Velocity component tn the radial (r) direction Control volume Velocity component in the vertical (z) direction Vertical distance
Greek Symbols
*
r e Vp
E T
Ceneral dependent variable Thermal dlffusivity Circumferential distance Viscosity Fluid density Summation Stress
Sub!jcripts e 1 nb p,u t
Effective Laminar Neighbouring nodes of p Split component of source term Turbulent
Superscript o
Old
332 CNS 9th ANNUAL CONFERENCE, 1988
The primary heat transport system (PHTS) of a typical MAPLE research reactor is shown in Figure 1. During normal operation, the reactor coolant is forced upward through the core and then enters the chimney, which is open to the pool for core accessibility. The coolant in the chimney is drawn into the heat exchanger by the pump. The discharge from the pump is split into two components - a flow to the inlet plenum and a bypass flow. The flow to the inlet plenum is directed to the core to cool the fuel. The bypass flow, a small fraction of the total flow, returns to the top of the chimney and flows down through it to combine with the core flow. The bypass flow is used to cool the pool water and to contain the core flow in the chimney. The combined bypass and core flows exit from the chimney via two suction outlets leading to the heat exchanger. The high-flow core jet in the chimney is subjected to the suction (due to the suction outlet flows) and countercurrent (due to the returning bypass flow) flows. Understanding this flow behaviour is of particular Importance in the open-chimney system in MAPLE. Under these conditions, the core jet, which contains short-lived high-level radionuclides, must be contained in the chimney to be recirculated in the PHTS. To show this, the flow patterns in the chimney must be such that the bypass flow in the chimney fs downward and of sufficient magnitude to overcome any upward diffusion. Such flows are quite complex and highly tliree-dimensional. The effects of various geometric and physical conditions on the flow patterns in the chimney were studied numerically using the assumption of a twodimensional planar jet (2}. A visual study was made to determine the effects of geometric and physical conditions on the core jet confinement in the chimney [3 ]. Based on the visual study, scaling laws governing the flow were derived and demonstrated using a two-dimensional flow model [4]. Three-dimensional components of velocity and turbulence In the chimney were measured using the split-film anemometry technique [5] • A quantitative comparison between the predictions from MAPL3D and the experimental data Is given in [6J.
Schematic Flow niagram of 1/5-Scale Buoyant Jet Test Facility chimney modules, inlet and outlet modules, a core module, a reelrcularlng pump, flow and temperature measuring devices, and dye Injection systems. The core and chimney assembly consists of up to five stacked hexagonal modules. The upper three modules are made of transparent acrylic to permit flow visualization. To date, the experiments performed at WNRE have covered a wide range of Isothermal high-flow conditions. The database generated from the experiments includes information about: (I)
FICItRE 1:
Schematic flow niagram of MAPLE Primary Heat Transport System
This paoer presents numerical results of MAPL3D for the high-flow core-let behaviour in the chimney, anrl compares them with experimental data. A threedimensional flow model, MA P L3D, was used to simulate some selected experiments in the 1/5-scale hydraulic model of MAPLE at the Whlteshell Nuclear Research Establishment (WNRE). The numerical results correctly predicted the unique flow patterns in the chimney and the effects of chimney height and core velocity on the chimney flow patterns. The MAPL3T) code is also described.
the geometric effects of suction angle, chimney height, and suction location on the effectiveness of core-jet confinement In the chimney. These effects were measured in terms of the bypass flow ratio required for full confinement of the core tet in the chimney; (II) the effects of physical conditions such as core velocity and bypass flow ratio on the flow patterns in the chimney; and (ill) the velocity and turbulence distribution in the chimney for various flow conditions. The results of (i) and (ti) were obtained by the hulk and local dve injection technique [3,4] which provided information on the overall and local flow patterns in the chimney. The results of (iii) were obtained using the split-film probe connected to a constant-temperature anemometer f5J. The low-flow buoyant jet experiments are being performed at WNRE. As well, a full-scale test rig is being built by AECL.
MATHEMATICAL MODELLING VALIDATION EXPERIMENTS A l/">-scale hydraulic model of a typical MAPLE research reactor was constructed at WNRE. Experimental results are useful not only to validate various modelling aspects of three-dimensional flow predictions, but also to better understand the flow mechanism affecting jet behaviour under suction and countercurrent flow conditions. The main components of the 1/5-scale model are shown schematically in Figure ?. The components include a pool tank, with two large acrylic windowg,
This paper describes the detailed flow patterns predicted in the chimney and pool. For high-flow conditions, where the buoyancy force Is negligible when compared to the inertial forces, isothermal conditions represent the normal high-flow operations of MAPLE adequately. However, for low-flow conditions, the buoyancy force begins to affect the flow patterns In the chimney and an isothermal assumption ts not valid. Although the three-dimensional flow model described Is capable of simulating low-flow buoyant jet conditions, the simulations presented in this paper have been limited to Isothermal conditions where experimental data have been available. CNS 9th ANNUAL CONFERENCE. 1988 333
The modelled area Included the chimney and pool above the core exit. The hexagonally shaped chimney was approximated as a cylinder so that a cylindrical coordinate system could be used. Because of the two planes of symmetry in the design, only one quadrant (90°) of the experimental system was simulated to reduce fhe computational effort. ^ I D U E V E T , tYre computer model can handle the full 360' plane for an asymmetric design. Conservation Equations The governing equations for three-dimensional, turbulent flow of an incompressible fluid are the conservation equations of mass, momentum (in the circumferential, radial and vertical directions) and energy in c y l i n d r i c a l coordinates. Using Boussinesq's eddy viscosity concept [7], the turbulent fluxes of momentum and energy are related to the mean-velocity and mean-temperature gradients via a turbulent viscosity and a thermal diffusivity, respectivelv. The final set of conservation equations solved is given in Appendix 1.
NUMERICAL MOnEU.ING Finite-Difference Formulation The entire domain of Interest is discrettzed into finite control volumesA staggered grid is M = * 4 sft A a t t\w •iel.otitv wjrtea lie Uetueeo. the nodes of the scalar variables. The governing equations are written In a finite-difference form for every node in the calculation domain. The convection flux terms are written in the hybrid upwind-central difference form [11). This formulation ensures the diagonal dominance for the matrix by limiting the grid Veclet number. The second derivatives In the equations are written in a typical central difference form. The resulting finite-difference fora of the governing equations at a node, p, surrounded by six neighbourIng points, nb, can be written in the following general form [12): [a p + o(Vol)/At - R * ^nb a nb*nb
+
PCVoD«°/At
(3)
Turbulence Model
Solution Procedure
The turbulent viscosity and the thermal diffusivity are not fluid properties, but oe-ptrin on the state of turbulence. Several turbulence models are available in the literature, and a good review of these models may be found in [8).
Rqn. (3) is an implicit expression in terms of dependent variables. The variables In the convection-diffusion coefficients and the source term (including the pressure gradient term) are evaluated explicitly from the previous step.
MAPL3D uses a simple turbulence model derived from the scaling arguments based on our experimental data [A]. Unlike the free-jet analysis, the turbulent mixing near the source and sink would be considerably higher than elsewhere. To account for this effect, the turbulent viscosity was calculated at each node based on the local values of the velocity, density and a length scale from:
Eqn. (3) is solved using a procedure similar to the SIMPLEC algorithm [13). This procedure eliminates the mass equation in favour of the pressure correction equation that couples velocity and pressure. In the procedure, the momentum equations are first solved to obtain a velocity field from the prevailing pressure field. Mass conservation is then enforced by solving the pressure correction equation, thereby determining the adjustments to r V velocities and the pressures. This process is repeated until the solution converges.
pc(local velocity)(inlet diameter)
(1)
where the constant, c, for the free jet ["»] was used. The thermal diffusivity" was evaluated from: r
= |it/c.
(2)
•a'neTe o c = 0.9 is taV«fi f-rwm ( W V IViese atnple models allowed for realistic mixing of the three interacting streams (core, suction and countercurrent flows) in the chimney. Initial and Boundary Conditions The symmetry of the flow allows only one quadrant of the system to be modelled, and thus two boundaries of the calculation domain consist of symmetric planes. The Initial flow velocity fields were set to zero for the simulations. A converged steadystate solution was independent of any specified Initial conditions. The inlet and outlet flow conditions were modelled as a mass-momentum source and sink, respectively, so that global mass conservation is always maintained. Uniform velocity profiles were assumed and explicitly specified at these flow boundaries. W.I tHe itvtecwaV walls. In. the pool ware modelled by modifying the corresponding convection-diffusion coefficients so that the neighbouring nodes across the wall did not influence each other. Wall shear stresses were determined using the logarithmic law formula.
334 CNS 9th ANNUAL CONFERENCE, 1988
A time-marching scheme rather than a strictly steady-state one (i.e., At = •») was used, and was found to be very effective In ensuring the convergence of the solution. The strictly steady-state scheme lor our application very often resulted In a diverging solution stemming from a node near the strong-vorticity region. However, the time-marching scheme prevented a solution divergence, especially during early iterations. For later iterations when the solution is close to converging, the transient terms no longer influenced the solution stability. However, for our application, too small a time step in this scheme resulted in an Incorrect solution since the transient terms became larger than the other physical convection-diffusion coefficient terms in the diffusion-dominated regions. The finite-difference equations are solved cyclically for their respective variables while the contributions of the other variables in these equations are assumed to be sources for that solution (an alternating-direction implicit method). For a given plane, the method sweeps in the other two directions sequentially on a line-bv-line basis uslnR a trldiagonal matrix algorithm. The solution algorithm marches tVirovis'n tVitj vftiole family of sucn pia-nes in that coordinate direction. This process is then repeated for the plane" normal to the other two coordinates. This fully elliptic nature of the solution scheme was necessary since the flow under consideration exhibits strong reel rcul.it ton In every
plane In the chimney. This procedure continues cover all the dependent variables.
to
To accelerate convergence, the finite-difference equations are solved by a line over-relaxation method, developed by Van Doormaal and Ralthby [13], applied to the pressure correction equation In twodimensional flow. For the present application, the method was extended for a three-dimensional flow and applied to all dependent variables as shown In Appendix 2. Convergence and Accuracy The complex nature of flow and the coupled and nonlinear nature of the governing equations necessitated that a converged solution passed several acceptance tests. First, the value of the norm of the residuals over the initial value must decrease ifonotonically to a value less than 1 0 " 3 . ||e||k < 10- 3 X||E||° at k'th iteration
(4)
where ||e|| = the Euclidean norm of the residuals = [£ (RHS-LHS of Eqn- (3) £,or each equation for all dependent variables) 2 !' o v e r t h e e n t i r e calculation domain. Secondly, the solution must not change as the number of nodes increases. Finally, a trend for convergence was monitored through the output of the dependent variables at the nodes to see If the values and flow patterns remain unchanged with successive Iterations.
COMPARISONS BETWEF.N SIMULATION AND EXPERIMENTAL RESULTS The equivalent diameters of the suction outlet, the core, the chimney, the reflector tank and the pool in the experimental model are 0.052, 0.096, 0.16, 0.23 and 0.74 nt, respectively. The suction height (from the core exit to the centreline of the suction outlet) is 0.285 m. The heights of the chimney (for the high chimney case) and the pool are 0.76 and 2.0 m, respectivelyAll these dimensions were taken exactly the same except that the diameters of tt\E cota awl the svicttan- au.tT.et Eat the. numerical, simulation were 0.0R9 and 0.059 m. For the vector plots presented, each arrow represents a velocity vector at a given node. Since the calculation domain contains a wide range of velocity magnitudes, only the arrow tips are shown to indicate the directions for very low velocities. Effect of Boundary Velocities After the effect of a grid spacing on the solution was examined from preliminary simulations using different grids, a nonuniform grid layout of 8x17x35 (in the circumferential, radial and vertical direction, respectively) shown In Figure 3 was chosen. The grid used is finer In the chimney than in the pool in order to capture vortex flow patterns in the chimney. For high-power operation of MAPLE, the flow through the core must Increase to keep the fuel cool. The operating limit for high-power operation must ensure that the jet from the core does not leak out of the chimney as a result of the Increased momentums of the core Jet, the suction flow and the bypass flow.
FIGURE 3:
Grid Layout Used for MAPL3D Simulation for High Chimney Case
Figures 4 and 6 show th
CNS 9th ANNUAL CONFERENCE. 1988 335
m FIGURE 4:
Predicted Flow Fields on Constant 1-Planes for High Chimney (Core Velocity of 2.5 m/s and Bypass Flow Ratio of 10%)
FIGURE 6:
Predicted Flow Fields on Constant l-Planes for High Chimney (High Core Velocity of •i m/s and Bypass Flow Ratio of 10%)
Effect of Chimney Height Figure 7 shows a nonunlform grid layout of 8x17x35 (in the circumferential, radial and vertical direction, respectively) used for the low chimney case (reduced 'o a half height of the high chimney case of Figure 3 ) . For the design of a MAPW- chimney, the chimney height must permit full containment of the core jet in the chimney, or the bypass flow ratio must Increase sufficient enougTi to match the core-jet momentum In the chimney. As seen In Figure 8, the core flow Is predicted to be contained in the chimney for the bypass flow ratio of 45%. Successive simulations with reducing bypass flow ratio showed that the core jet leaks out of the chimney for a bypass flow ratio less than about 40Z. The experimental results [3] Indicated that the required bypass flow ratio was about 34%.
SUMMARY AND CONCLUSIONS
J-Srclta, . 5 v,,,,, » 2.42 m/s FIGURE 5:
336
j-S«Uon = 8 V1M
Predicted Flos' Fields on Constant j-Planes for High Chimney (Core Velocity of 2.5 ro/s and Bypass 'low Ration of 10Z)
C N S 9th A N N U A L C O N F E R E N C E , 1988
This paper described the three-dimensional flow model, MAPL3D, and the validation experiments performed in support of MAPLE reactor development. The high-flow jet behaviour study has proven to be useful not only for assuring the viability of the openchimney concept of MAPLE, but also for understanding a complex flow behaviour under suction and countercurrent conditions. The measured velocity n r o f M e s In the chimney were well predicted from MAPL3D. The parametric effects of core velocity and chimney height on the chimney flow behaviour were studied ustng MAPL3D and a reasonable agreement was obtained In comparison with experimental data.
ACKNOWLEDGEMENTS The authors are grateful to A.C.L. Holloway, A.O. Campagna and W.D. Warnlca for their earlier contribution to the MAPL3D development.
REFERENCES (1)
R.F. LlnSTDNE and J.I. SAROUDIS, "MAPLE: A New Multipurpose Reactor for National Nuclear Development In the 1990s," Proceedings of an International Symposium on the Significance and Impact of Nuclear Research in Developing Countries, Athens, Greece 8-12 September 1986, IAEA-SM-291/19, 1986.
(2)
S.Y. SHIM and D.K. BAXTER, "Numerical Simulation of a Confined Jet under Suction and Counter-MomentuTii for the Canadian MAPLE Research Reactor," Proceedings of the 7th Annual Conference of the Canadian Nuclear Society, Toronto, 1986 June.
(3)
P.T. WAN, S.Y. SHIM and V.S. KRISHNAN, "Experimental Investigation of Core Flow Jet Confinement in a MAPLE Flow Test Facility," Proceedings of the 8th Annual Conference of the Canadian Nuclear Society, Saint J o h n , 1987 June.
(4)
S.Y. SHIM, P.T. WAN, D.K. BAXTER and R.L. HF.MBROFF, "A Study of Submerged Confined Turbulent .lets under Suction and Countercurrent Flows," Proceedings of the 5th International Conference on Numerical Methods In Laminar and Turbulent Flow, Montreal, July 1987.
(5) FIGURE 7:
Grid Layout Used for MAPL3D Simulation for Low Chimney Case
Dynamics C o n f e r e n c e , July 24-28.
Ohio,
U . S . , 1988
(6)
S.Y. SHIM, P.J. MILLS, J.E. KOWALSKI, D.K. BAXTER and R.L. HEMBROFF, "Thermalhydraulics Studies of the MAPLE Research Reactor," to be presented in the 3rd International Topical Meeting on Nuclear Power Plant Thermal Hydraulics and Operations, Seoul, Korea, 1988 November 14-17.
(7)
P. BRADSHAW, T. CEBECI and J.H. WHITELAW, "Engineering Calculation Methods for Turbulent Flow," Academic Press, New York, 1981.
(8)
D.B. SPALDING, "Turbulent Models: A Lecture Course," Imperial College of Science and Technology Report, CFD/82/4, 1982.
(9)
H. SCHLICKTING, "Boundary-Layer Theory," McGraw-Hill Book Company, New York, 1979.
(10) W. KOLLMANN (editor), "Prediction Methods for Turbulent Flow," Hemisphere Publishing Corporation, pp. 259-349, Washington, 1980. (11)
D.B. SPALDING, "A Novel Finite-Difference Formulation for Differential Expressions Involving Both First and Second Derivatives," Int. J. Num. Methods Eng. 4, 551, 1972.
(12)
S.V. PATANKAR, "Numerical Heat Transfer and Fluid Flow," Hemisphere Publishing Corporation, Washington, 1980.
(13) J.P. VAN D00RMAAL and G.D. RAITHBY, "Enhancements of the SIMPLE Method for Predicting Incompressible Fluid Flow", Numerical Heat Transfer 7, U7-163, 1984. FIGURE 8:
Predicted Flow Fields on Constant i-Planes for Low Chimney (Core Velocity of 1.9 m/s and Bypass Flow Ratio of 45%) CNS 9th ANNUAL CONFERENCE, 1988 337
APPENDIX 1: CONSERVATION COORDINATES
EQUATIONS
APPENDIX ?.:
IN CYLINDRICAL
Continuity Equation at
r
I 8(prv) S(pw) = „ r 8r BE
ae
LINE OVER-RELAXATION METHOD
The finite-difference equations can be expressed in the more convenient line torm: =
b
l]k*i+l]k 0
Momentum Equations
e
0
ijk*ij-lk
+
f
ijk*ljkl
u-component (in the 9 direction): h
o(pu) + 1_ o(pu^) + 1_ d(prvu) + d(pwu) at r B6 r 3r 9z = i. ^£. - P v u + i. r a9
r
To solve Eqn. (2.1) along planes of constant j or k, []BE is defined as a better estimate of *,
99 + 1 a ( r T 8r) _j_ 9z
r 38
r^
6r
( 2 'D
ljk
o
bz
no
(9 " U
v-component (in the r direction): + (6 - 1) (•,
(2.2)
a(pv) + 1_ a(puv) + 1_ 5(prv2) + 5(pwv) at r 89 r ar 3z =. 5P + £«i + I !ll£ l+a(r^r>.va'Er2j99 8r r r 89 r 3r 8z r
where 0 Is a relaxation parameter such that for 0=1 [
w-component (in the z direction):
l^iik+llBE' above form.
3(pw) + 1_ B(puw) + 1_ B(prvw) + B(pw 2 ) 3t r 88 r Br 3z az
r ae
r ar
' f*iik-l^BE
3z
P r e s s e a " as in the
taljk "(d ijk +e n = bljk*i+ljk + c i
(2.3) o o f ijk*ljk+l + 8ijk*ijk-l'
3r
( ) () r 58 r d8 Br dr
are ex
Substituting Eqns. (2.2) Into (2.1) results the following equation.
Energy Equation 3(pT) + l_ a(puT) 1 at r 88 r
anc
8z
<) 8z
-
(6 -
+ eIjk +
where Along planes of constant j or k, Eqn. (2-3) can be rewritten in a trtdiagonal matrix form. 8r
r 36
[r()
8r r
5z
],
r 88
(Sji + I 5i), and oz r 88
338
CNS 9th ANNUAL CONFERENCE, 1988
u(Zv_+
BW) Bz 3r
A
n n n ljk*ijk = B I jk0i+l jk+ Cijk
(2
*A)
A similar estimate Is made for solutions in the j and k directions.
Session 10: Operational Enhancement - II
Chairman: D.R. McQuade, Ontario Hydro
CNS 9th ANNUAL CONFERENCE, 1988 339 A
'3ko
A MKTHODOI.OGY TO REDUCE UNCERTAINTY AU.OWANCK IN CHANNEL FLOW VERIFICATION
K.F. I.AH Nuclear Studies & Safety department Ontario Hydro. Toronto H5G 1X6 (416) 592-7837 ABSTRACT This paper is a continuation of a previous CNS publication which described a methodology for flow monitoring in CANTO reactor channels. The work, undertaken by design and operations personnel in a multi-unit operating power station, which has resulted in a significant reduction in spurious alarms without compromising reactor safety, is reported here. A methodology to reduce the uncertainty allowance in channel flow verification is also presented. INTRODUCTION Flow obstruction causing fuel dryout. may lead to fuel damage, and may consequently affect fuel channel integrity, and thus should be avoided. The flow conditions in a CANDU reactor channel under boiling ri subcooled conditions may be inferred from the differential pressure ac-oss the fuel channel, as measured during fuelling. This method of channel flow verification is widely used in CANDU power reactors where channel boiling cannot be precluded during normal operation. Overview A methodology for the evaluation of the allowable limits, referred to as the "Channel Delta-P (flP) Alarm Setpoints" was presented in [1], The paper described how the safe operational limits of a fuel channel under flow obstructed conditions can be assessed by means of a FLOW BLOCKAGE MAP. However, safety analysts who evaluate these setpoints are confronted with the uncertainty allowances which are necessary to accommodate the large variation due to tolerances in fuel channel design and reactor operation under different conditions. Tight operating tolerances on these setpoints very often result in alarms under normal reactor operation. These alarms, real or spurious, have to be confirmed by other means to ensure continued safe reactor operation. Very often, the reactor power has to be derated and the channel exit temperature monitored to determine if adequate coolant flow is available in the channel. Spurious alarms not only add an extra burden to operators but represent a potential economic penalty on the overall performance of the reactor. This paper describes the progress made on the subject of channel flow verification since the issue of [1] in 1982. A methodology to reduce the uncertainty allowance in channel flow verification in a multi-unit CANDU power station is presented.
The objectives of this paper are:
(a) to develop a methodology uncertainty allowances of setpoints.
to define the channel delr^-P
(b) to discuss the progress made on channel delta-P measurement including - reduction of uncertainty allowance of differentia] pressure measurement - refinement of channel delta-P limits by well-calibrated ultrasonic flow measurement during commissioning. - refinement of channel delta-P limits through in-service channel delta-P measurement. (c) to derive channel flows in a multi-unit station by using in-service channel delta-P data in lieu of ultrasonic flow measurement. (d) to apply the methodology developed in (a), the progress made in (b) and the flow derived in (c), where applicable, to refine channel delta-P setpoints in a multi-unit CANDU power station. DERIVATION OF CHANNEL DELTA-P SETPOINTS The criteria for deriving the operational limits, termed channel delta-P limits, of a CANDU reactor channel were presented in [1] by means of a FLOW BLOCKAGE HAP. These limits are derived on the basis of prevention of flow obstruction causing fuel sheath dryout which could lead to fuel failure or pressure tube damage. Two types conceivable.
of
flow
obstruction
are
considered
(a) inlet feeder obstruction - The location of the flow obstruction is arbitrarily taken at the inlet feeder exit. Its precise location is inconsequential to the analysis since the inlet feeder is subcooled and there is no significant density and viscosity variation along the inlet feeder. Any inlet feeder obstruction results in a reduction in channel delta-P. Therefore, the lower channel delta-P setpoint should be capable of detecting inlet feeder flow obstruction which could lead to fuel sheath dryout. (b) channel obstruction - The location of the flow obstruction is assumed to be at the channel exit. Because density and viscosity vary along the channel as a result of heat addition and rne effect of two-phase multiplier as quality increases along the channel, the location of the flow obstruction which gives the highest channel delta-P (for the same pressure loss coefficient) is most limiting at the channel exit. Any flow obstruction in the channel in between inlet and outlet end fittings results in an increase in measured channel delta-P. Therefore, the upper CNS 9th ANNUAL CONFERENCE, 1988 341
channel deita-P setpoint should be capable of detecting channel Flow obstruction which could lead to fuel sheath dryout. Outlet feeder obstruction, notwithstanding having a similar effect on channel delta-P as inlet feeder obstruction is considered to be highly unlikely since any debris, if present, in the outlet feeder, would be small enough ro pass through the holes in the liner tube, the annulus space between the liner tube and the pressure tube or the inlet feeder bends of various curvature radii. Furthermore, such debris has an insignificant effect on the channel flow and would likely be transported into the reactor outlet header.
UNCERTAINTY ALLOWANCE IN CHANNF.I. DELTA-P LIMITS A list of major uncertainty allowances which could affect the channel delta-P limit was presented in [1J. For a multi-unit CflNDU power station, there would be additional variations due to minor changes in fuel channel design and reactor operation under different condition.
Sensitivity Analysis of A Typical CANDU Reactor Channel
TAfll.F. 1:
Parameters
Parametric Sensitivity in AP (kPa) Variations Nominal Upper Lower (One Sigma) Condi', i ori Limit Limit
. Design Tolerances Fdr. Res. 10% IEF* Res. 10% Channel Res. 10% 10% OEF** Res. Total
. Reactor Operation Bundle Res. Inlet Temp. Hdr. Press, niff ROH Press. Flux Shape Total
1.53% 0.51°C 23 kPa 71 kPa 0.1 HW
28.3
2\ .2 1.0 10 .6 3 .1 ?3 .9
1.8 1 .8 Ifl.O 9.2 20.4
2.9 --
1.6 1A
15.8
22 .4 2 .5 5 .6 23
2.8 4.0 0.2 5.9 13.5 15.5
20.8
-19.2
--
--16.)
. Thermalhyciraulic It is well recognized that the measured delta-P values depend on a number of parameters which could vary from the predicted values used in the calculations. Also, uncertainties in empirical modelling must be taken into account. For this reason, the effects of each parameter which could affect the delta-P limits evaluated in [1] are considered. These parameters can be grouped into five main categories in accordance with the cause and nature of the uncertainties. To simplify tht present analysis, all parameters are assumed to hove a normal distribution. The magnitude of all variations guoted are based on a one-sigma standard deviation. unless stated otherwise.
Modelling U-l CHF Corr. L-L TPF Corr. C-H TPF Corr. FS TPF Corr.
The effect of design tolerances on channel delta-P limit is most significant. Uncertainties Due to Reactor Operation During normal reactor operation, there are minor temperature and pressure perturbations in the heat transport system. Also, normal fuelling and shim operation of the reactor result in variations in axial heat flux distribution along the channel. Furthermore, changes in bundle alignment during fuelling usually result in changes in the channel resistance. These variations affect the flow and pressure distributions of the reactor channels and therefore affect the delta-P limits. A header-to-header differential pressure variation of +;2% can be assumed. The outlet header pressure
342 CNS 9th ANNUAL CONFERENCE, 1988
— -— —
2..1 .3 ?.. 2..4 1.8 4.8
5.0 14.4 9.2 1.2 17.8
3.5
10.8
1.0
Instrumentation Diff. Press. Transmitter
20.0
20.0
20.0
Overall Uncertainties (one sigma)
38.3
40.7
37.1
Total Grouping Channel Elvn.
Uncertainties Due to Design Tolerances As given in Table 1, parameters such as feeder resistance, end fitting resistance and fuel channel resistance fall under this category. An uncertainty allowance of +10% is assumed for each of these parameters. The resulting variations in delta-P values of a representative channel at 100% FP normal condition and critically flow-restricted conditions are also provided.
(1) 3.18% (2) 5.6% (3) 10°, (4) 10%
4.86 m
Notes: * ** (1> (2)
Inlet End Fitting Outlet End Fitting U-l Critical Heat Flux Correlation Lawrenc-Leung Two-Phase Flow Correlation
(3) Chenoweth-Hartin Two-Phase Flow Correlation (4) Fitzsimmons Two-Phase Flow Correlation variation is estimated to be +71 kPa which reflects the pressure measurement and control accuracy. The inlet header temperature variation is estimated to be +0.51°C, which reflects the dependence on the accuracy of boiler pressure measurement and control. Assuming the fuel bundles are randomly aligned, the variation in total bundle resistance is estimated to be +1.53% which is evaluated from the standard deviation of +5.29% on the overall single bundle Junction resistance of a 37-element bundle. The variation of axial flux tilt on dryout channel power is assumed to be +0.1 MW.
signi ficdnt1y reliability of transmitter.
Uncertainties Due to Thermal hydraulic Correlations Based on the goodness-of -f i t of the thermalhydrrtulic correlations obtained from U-I experiments, the uncertainties in the Critical Heat Flux (CHF) correlation and the two-phase pressure drop correlation are evaluated to be +3.18% and +5.6% respectivejy. The uncertainties in two-phase multipliers of Chenoweth-Martin and Kitzsimmons are assumed to be +10%.
(b)
The original design of differential pressure transmitters, used in channel delta-P measurements, utilizes a full range centre-zero differential pressure transmitter to cater for bi-directional channel flows in a CAWDU reactor. Since the uncertainty allowance of a differential pressure transmitter is dependent on its calibrated range, a full range centre-zero differential pressure transmitter has a much higher uncertainty associated with its calibrated span than a single range differential pressure transmitter. To reduce the uncertainty allowance on channel deita-P measurements the latter differential pressure transmitter is selected.
(c)
To cater for the bi-directional channel flow in a CANDU reactor, two single range differentia] pressure transmitters arranged in "parallel-revrrse" configuration are used. Their output signals are fed to a current discriminator which selects the rational (4-20 mA) high current signal and rejects the irrational (below 4 mA) low current signal. The rational signal is then processed, displayed and connected to the channel delta-P alarm circuit.
Uncertainties Due to Channel Elevations As discussed in [1], the elevation of the fuel channel with respect to the centre line of the reactor has an effect on the measured delta-P values due to densimetric pressure differences between the inlet and outlet feeder fluid columns. Since the representative channels have different elevations from the rest of the channels in the same group, this effect has to be considered in the evaluation of delta-P limit. It is recognized that the variations of channel elevation are dependent on the distribution of the channels within the group and are generally not in the normal distribution. For conservatism, the uncertainties in channel elevation can be taken to be equal to the reactor core radius, the maximum possible channel elevation in a reactor.
improved the measurement the differential pressure
ULTRASONIC FLOW MEASUREMENT DURING COMMISSIONING Uncertainties in Differential Pressure Transmitter Measurement The measurement uncertainty of a differential pressure transmitter can be evaluated from the manufacturer's specifications based on the effects of linearity, repeatability, hysteresis, stability, temperature, pressure and power supply variations. The typical value of uncertainty of a good quality differential pressure transmitter is +40 kPa (two s i gma).
DIFFERENTIAL PRESSURE TRANSMITTER DESIGN MODIFICATION In order to improve the reliability of flow monitoring applications, design work was initiated in 1983. Three types of differential pressure transmitters were installed on a fuelling machine trolley on a trial basis. The transmitter best suited for this application was selected. The durability and accuracy of the "best" transmitter is only marginally adequate. Further review of differential pressure measurement procedures has concluded that over-ranging effects are responsible for the premature failure of the transmitter. Three design changes were undertaken to improve the accuracy and reliability of the differential pressure transmitters. (a)
A software change to synchronize the pressurization of the pressurizing pumps on either side of the fuelling machine was implemented. The design change ensures that the pressure drop across the differentia] pressure transmitter does not exceed its calibrated range during leakage testing and closure plug removal. This design change has
Ultrasonic flow measurements of individual channels in Unit 6 at Bruce NGS fl were undertaken in November 1983 as part of the commissioning program. The ultrasonic flowmeters used in the commissioning tests were calibrated against a test loop in the National Research Council of Canada laboratories and are therefore considered to be one of the most accurate and practical measurement method from a measurement standards point of view. The uncertainty allowance on ultrasonic measurement on channel flow is assessed to be less than +1% (one sigma) and includes the accuracy of the test loop, uncertainty in pipe diameter and pipe wall thickness, and velocity profile variations. The average channel flow in Rruce Unit 6 prorated to the operating condition is found to be 2.5% higher than the evaluated flow based on design data. The higher measured flow is also confirmed by on-power flow measurements of the fully instrumented channels (FINCH). Since channel flow and its pressure drop are related, a higher channel flow results in a higher pressure drop across the channel and corresponds to a higher channel delta-P limit if the assumed maximum channel power (AMCP) remains the same. Therefore, the upper channel delta-P limits derived from design data are under-estimated since the measured channel flows are higher than the evaluated flows. It should be noted that a reduced upper delta-P setpoint diminishes the tolerance to flow obstruction in a channel and does not pose safety concerns. However, from an operating standpoint, a reduction in upper channel delta-P setpoint would result in a decrease in operating flexibility and is more likely to result in spurious alarms during norma] fuelling.
C N S 9th A N N U A L C O N F E R E N C E , 1988 343
To resolve this operational concern, it is necessary to re-evaluate channel de'rs-P Jimits to account for the differences betv~~n the design flow and the measured flows. Channel delta-P limits of each channel can be assigned ba'.ed on the measured flows with the aid of a chfnne) delta-P versus nominal subcooled flow plot derived in [1].
REASSESSMENT OF EMPIRICAL CONSTANTS BY USING IN-SliWICR MEASUREMKNT OF CHANNEL DELTA -P During 1985/86, over one thousand channel delta-P measurements were recorded in Rruce NGS h Unit 6 during fuelling. Analysis of these channel delta-P measurements indicates that the end fitting pressure drop contribution constants (EFPDCC) [)) used in the assessment of channel delta-P setpoints are under-estimated. An assumption used in the evaluation of channel delta-P limits is that the EFPDCC's are the same for end fittings with identical geometry. When channel delta-P limits were evaluated for Bruce NGS B Unit 6. it was recognized that these contribution constants may be higher due to longer end fittings because of a different fuel channel design. These longer end fitting bodies result in higher contribution constants because the pressure drops as "seen" by the differential pressure transmitter of the fuelling machine are higher. Theoretically, EFPDCC is related to the pressure drops of the tee-junction and the horizontal portion of the end fitting including the liner tube and the shield plug. The latter portion is "seen" by the differential pressure transmitter while the tee-junction loss is not.
REASSF.SSMF.NT OF CHANNEL DELTA-P SKTPOJNTS BASKD ON COMMISSIONING 6, IN-SERVICE MEASUREMENTS Since the values of KFPDCC are 7-]8% higher than the corresponding assumed values used in the original computation of channel delta-P limits, incorporating the revised RFPIXC in the channel delta-P limit calculation can result in an increase in both the upper and the lower channel delta-P limits. ." t is possible to further reduce the uncertainty in channel delta-P limits by making use of the tesults of the ultrasonic flow measurements during commissioning. Table 1 shows that the uncertainty contribution of channel delta-P measurements at nominal flow conditions due to design tolerances alone amounts to +28.3 kPa. This value is considerably higher than the uncertainty allowance of 1% from ultrasonic flow measurements corresponding +16.4 kPa. Therefore. it is advantageous to use measured channel flows rather than the design calculations and so minimize the uncertainty. Table 3 shows a revised sensitivity analysis of channel delta-P and the corresponding limits. This table also includes a revised uncertainty allowance of the improved differential pressure transmitters arranged in "parallel-reverse" configuration.
TABLE 3:
Sensitivity Analysis of A Typical CANDU Reactor Channel Based on Ultrasonically Measured Flow
Parameters For this reason, the EFPDCC is expected to be dependent on the velocity pressure of the fluid entering and leaving the end fitting. Accordingly the channel delta-P data can be segregated into groups in accordance with the ii. ?t/outlet feeder connection sizes and inlet ..wtiperatures, if applicable. The values of EFPDCC are given in Table 2. They are 7-J8% higher than the corresponding ones used in the "original" computation.
TABLE 2:
Parametric Sensitivity in AP (kPa) Variations Nominal Upper Lower (One Sigma) Condition Limit Limit
Flow Measurement Ultrasonic Method
Reactor Operation
50 mm 63 mm
63 mm 63 mm
50 mm 63 mm
63 mm 63 mm
EFPDCC Used in "Original" Computation
0.65
0.69
0.65
0.69
Revised EFPDCC
0.77
0.77
0.69
0.77
Slumber of Data Points Ll5-;d
344
208
695
CNS 9th ANNUAL CONFERENCE, 1988
169
105
16.4
15.4
6.5
16.1
23.3
15.5
4.8
17.8
N/A
End Fitting Pressure Drop Contribution Constant (EFPDCC)
Inlet Temp In jet Temp 268 °C 254°C Group A Group B Group C Group D
Conn. Size Inlet Feeder Outlet Feeder
1%
4.86 m
Instrumentation Diff. Press. Transmitter Overall Uncertainties (one sigma) Observed AP variation (one sigma)
23.
10.e
1.0
10.0
10.0
10.0
25.3
31.9
3.5
26.5
Jt should be noted that the design tolerances in Table 1 were estimated conservatively in an attempt to provide the same setpoint for channels with the same lattice location of different reactor units. Following a revision of fuelling machine computer software, the channel delta-P setpoint program has the capability to accept different setpoints for the same lattice location of a different reactor unit. Accordingly, the pair of values for each channel can be replaced by values based on individual channel flow measurement.
FREQUENCY
FIGURE 2: OCISASONICALLX MEASURED CHANNEL FLOW BRUCE
120
UNIT
6
100
so 60 40
CHANNEL FLOW CALIBRATION BY MEANS OF DIFFERENTIAL PRESSURE MEASUREMENT Having established the feasibility of individual flow assessment from Unit 6, it was decided to apply this principle to other units but using delta-P measurements as a means of calibration. The appropriate EFPDCC's derived from in-reactor measurements can be applied (see Table 2). To provide an added degree of assurance that this method of channel flow measurement is accurate for the present application, the channel delta-P data of Bruce NGS B Unit 5 are compared with the corresponding flows measured from the fully instrumented channels. The results are given in Figure i.
20
I
27
25
28
29
C8ANNEL FLOW KG/SEC
FREQUENCY
FIGURE 3 :
120
DELTA-? DERIVED CHANNEL FLOW BRUCE
UNIT
5
1OO SO'
60
1
25
AP 1000 vo*
FIGURE 1 MEASURED AP vs. FINCH FLOWS
900 800
7ocr
?5 MEASURED FLOWS
26 KG/S
The graph confirmed that channel delta-P measurements can indeed be used for channel flow "Calibration". Furthermore, Figures 2 and 3 show the similarities of channel flow distribution between Bruce NGS B Units 5 and 6 even when using two different methods of channel flow measurement. The channel delta-P limits can then be assigned based on the measured flows which are derived by differential pressure measurement. The uncertainties in channel flows estimated by differential pressure measurements can be evaluated as discussed below: (a) A sensitivity analysis has shown that a flow variation of 1% (approx. 0.25 kg/s) produces a channel delta-P variation of 16.4 kPa under nominal power condition. Therefore, a 10 kPa variation of the latter parameter corresponds to 0.61% change in flow.
IMT 27
23
C:-:A:;.SSL F L O W
23 KC/SE
(b) The observed channel delta-P uncertainty (based on in-service pressure drop measurements) is 23.1 kPa. Therefore, a channel delta-P variation of 23.1 kPa in the "calibration curve" corresponds to 1.41% variation in flow if no error is assumed in the differential pressure transmitter. (c) The total uncertainty in flow measurement, by using a differential pressure transmitter with an uncertainty of 10 kPa, Is therefore 1.53%, which is slightly below the accuracy of ultrasonic flow measurement. Table 4 shows a revised sensitivity analysis of channel delta-P and the corresponding limits for Bruce NGS B Unit 5.
EFFECTS OF REVISED CHANNEL DELTA-P SETPOINTS ON SPURIOUS ALARMS During 1985-86, over two thousand channels had been fuelled in Bruce NGS B Units 5 and 6. A total of 124 alarms, amounting to about 6%, have been received and all confirmed to be spurious. Figure A shows the frequency of spurious alarms during this period. The channel delta-P setpoints were derived based on the design flows with uncertainty allowances given in Table 1. Figure 5 portrays the expected frequency of spurious alarms should the channel delta-P measurements be compared with the revised setpoints which are evaluated from the measured flows with uncertainty allowances given in CNS 9th ANNUAL CONFERENCE, 1988 345
TAfll.F. 4: Sensitivity Analysis of A Typical CANDU Reactor Channel Based on In-Service Channel Delta-P Measurements Parameters
Parametric Sensitivity in AP (kPa) Variations Nominal Upper Lower (One Sigma) Condition Limit Limit
. Flow Measurement niff. Press. Method 1.53%
25.1
23.4
9.8
. Reactor Operation
16.1
23.3
15.5
. Modelling
N/A
4.8
n .8
3.5
10.8
1..0
10.0
10.0
10. 0
. Grouping Channel Elevation
4.86 m
. Instrumentation Diff. Press. Transmitter Overall Uncertainties (one sigina)
31.6
36.4
27. 5
SUMMARY AND CONCLUSIONS A methodology to define the uncertainty allowance of channel delta-P setpoints for CANDU reactor channel has been presented for the purpose of channel flow verification during fuelling operations. This methodology is applicable to all channels under both subcooled and boiling conditions. The progress made on channel delta-P measurement in the past years has been discussed. These include design changes in the fuelling machine software, improvement in differential pressure measurement techniques and modelling of pressure drop in the end fittings. Further reduction of uncertainty allowance is made possible by means of commissioning data and in-service differential pressure measurements during fuelling. The methodology developed here and the progress made in the past years allow channel delta-P setpoints to be defined more precisely to preclude flow obstructions that could lead to fuel sheath dryout while minimizing spurious alarms to enhance operational reliability. Channel flow in a multi-unit CANDU power station can be accurately derived from in-service channel delta-P measurement during fuelling. The channel delta-P-derived flows then allow channel delta-P setpoints to be evaluated with minimum uncertainty allowance.
Tables 3 and 4. Following a revision of the channel delta-P setpoints in these two reactor units, the freguency of spurious alarms would be dramatically REFKRKNCES reduced to less than 1%. This modification has now been installed with satisfactory results. [1] LAM. K.F., "A Methodology For Flow Blockage Detection in Boiling or Subcooled CANDU Reactor Channels", Proceedings of the Third CNS Annual Conference, June 1982, A-31 to 40. ACKNOWLEDGEMENT
FIGURE 1: AlAM FKMCWr - - M l t l l W l ' CIIAMIU BE11A-P SETPOIMS IDIH AIAINIS
|;'i
»n oi M I I u n i t s
?;n
ALARM f R10UCHCV
5.6 1
_rq-n n . nCASURCO FLO* KG/StC
FIGURE 5i ALARH FREOOEIICf
LLL
r
,., 1
- HIVISCD CIIMKU
>til/iL Aunits mi I
JLL
346 CNS 9th ANNUAL CONFERENCE, 1988
K I I A - P SEIPOIIIIS
10 «n II.' •
The author wishes to express his sincere thanks to the staff of Nuclear Studies and Safety Department of Ontario Hyd-o for their assistance during the course of this work.
DEFECT DETECTIVE:
AM EXPERT SYSTEM FOR THE
DETECTION AMD EVALUATION OF FUEL DEFECTS IN CANDU-600 NUCLEAR POWER REACTORS
A.M. MANZER, J.W.D. ANDERSON and C.W. SO
Atomic Energy of Canada Limited CANDU Operations Sheridan Park Research Community Mississauga, Ontario L5K 1B2 CANADA ABSTRACT DEFECT DETECTIVE is an expert system computer program that is used by AECL to detect and evaluate fuel defects on the basis of data received from the four operating CANDU-600s. By applying the rules of data interpretation, the program evaluates the condition of the fuel defect in each suspect channel and recommends the removal priority. The recommendations are designed to minimize the costs associated with premature refuellings and the occupational exposures associated with the buildup of gamma radiation fields. These fields are due to depositing fission products that are released from some defects. Since both factors can have an impact on station economics, it is important to routinely analyze all available data that provides clues on the number, the condition, and the location of fuel defects. This expert system is intended to speed up routine analysis of fuel defect detection. It is written entirely in the PROLOG programming language which offers several advantages beyond expert system shells.
INTRODUCTION In the late 1960s, the prototype CANDU reactor, Douglas Point, experienced unexpectedly high gamma radiation fields within the reactor building. These fields contributed to high occupational exposures. In the 1970s, the designers aimed for a much lower exposure target of less than 500 man rem p e r year for the CANDU-600 r e a c t o r 1 1 ' . In the 1980s, the four CANDU-600 plants now operating in New Brunswick, Quebec, South Korea and Argentina, are achieving annual exposures of less than 25% of the target, a'^out 100 man rem per year. Many factors contribute to these low exposure levels, particularly the unique capabilities of CANDU reactors for on-power refuelling, and on-power failed fuel detection and location' . After many years of CANDU operating experience, other factors that contribute to low exposures are becoming apparent: good fuel performance and good post-defect behaviour. Despite the good record of the CANDU-600s, it is still possible to m a i n t a i n or lower the occupational exposures while lowering the costs of defective fuel to the station. This can be achieved by applying all the available techniques for detecting, locating and evaluating fuel defects. The analytical techniques developed over the past few years are numerous and site staff do
not always have time to apply them. The routine analysis of defect detection is well suited to the analytical techniques of an expert system. The advice generated by DEFECT DETECTIVE is intended for the fuelling engineer who must decide when to refuel suspect fuel channels. Table 1 provides some indication of the typical costs associated with the refuelling of a suspect channel. By refuelling a fuel channel before it achieves its target burnup (Case 1) , good fuel is discarded and the burnup loss could be as high as 520,000. Also, a premature refuelling operation could introduce a local high flux region in the core which could lead to a reactor overpower trip (Case 2) , The replacement energy co;;t due to a rsactor trip could be as high as $30,000 per hour of down time. On the other hand, by leaving a fuel defect in the core for too long, the gamma radiation fields could increase significantly on the primary circuit (Case 3 ) . These fields are caused by the depositing fission products that are trapped within the uranium that is released from certain d e f e c t s ' • '. If an outage is extended because of the high gamma fields, th<>n the costs could be about $600,000 per day. The important point is that an incorrect decision for removing a defective fuel bundle may cost several hundred thousand dollars to the station.
TABLE 1 SOME
CASE
ECONOMIC CONSIDERATIONS ASSOCIATED REFUELLING SUSPECT CHANNELS
DESCRIPTION
1
Burnup loss due to premature refuelling
2
Energy replacement due to reactor trip (1 hour downtimeJ
3
Maintenance outage extension due to high gamma fields (1 day)
WITH
MAGNITUDE
ASSUMPTIONS
• $10,000 $20,000 • •
2-'t$K per bundle % burnup loss 8 bundles replaced.
*> $30,000
• $40 per MWh(e) for replacement energy.
•v$600,000
• $40 per MWh(e) for replacement energy.
CNS 9th ANNUAL CONFERENCE. 1988
347
The risk of high gamma fields with small defects is minimal because they tend to release only dissolved fission products that arc: removed by the •••iffication system or by radioactive decay. These :. -icts do not normally contribute to long term gamma fields until the hole opens up and releases uranium which contains depositing fission products (5) Ideally, the fuelling engineer should postpone refuelling of a suspect channel until the channel achieves its target turnup, or until there is some sign of post-defect deterioration or uranium release. For the above cost considerations, it is important that the fuelling engineer receives early warning: •
when a fuel defect must be located,
•
when a defect needs to be removed,
•
when several defects are present, and
•
when additional actions are needed to assess ^yc cottivca the detect, status if the cccs.
Good defect detection is essential for providing early warning to the fuelling engineer.
TOOLS FOR EVALUATION
FUEL
PEFECT
DETF.CTION/LOCATION/
The CANDU-600s come equipped with systems that help detect and locate defective rods while the reactor is at power, The systems tap into the primary circuit and indirectly measure the activity concentrations of various fission products in the coolant, as described below. •
The gase<*:s fission product monitoring (GFP) system is a computer controlled high resolution gamma ray spectrometer. It operates continuously, repeatedly measuring gamma ray activity 1 R ttve primary citcu.it sa that transient releases of fission products can be detected.
•
The delayed neutron monitoring (DN) system counts the delayed neutrons emitted from short-lived fission products (1-137 and Br-87) in coolant immediately downstream of each of the 380 fuel channels. It is the primary tool for locating defective fuel in the core.
•
The gamma ray spectrometer in the chemistry laboratory that analyses the gamma ray energy spectrum of coolant "grab" samples, (radiochemical data). The analyses determines the activity concentrations of fission and activation products in the coolant.
The procedure for detecting and locating CANDU-600 fuel defects normally involves interpretation of the gaseous fission product monitoring and delayed neutron monitoring data. An evaluation of the defect condition depends on the data generated by all three systems. This is done by applying the basic principles of defect detection as outlined in the next section.
DEVELOPMENT PRINCIPLES
AND
APPLICATION
OF
DEFECT
DETECTION
An e x p e r i m e n t a l program with defective DU-type UO 2 elements was carried out at the
damage were irradiated in separate tests in a experimental loop of the NRX reactor. Thi fundamental program combined with the reacto
release behaviour, and of post-defect deterioration behaviour. The program has had a profound effect on how defective fuel is npw detected and located within ithin CANDU Dower power stations' . When most fuel rods fail, the noble gases are the first of the fission products to escape to the coolant with Xe-133 being the most abundant. The gaseous fission product monitoring system is programmed to spot any sudden increase in Xe-133 activity concentrations since this is the main indicator of a fuel defect in the c o r e 1 " . The control room is normally alerted within an hour or two depending on the rate of increase of Xe-133 activity concentration. When this happens, a delayed neutron scan is usually planned in order to search for the suspect fuel channel haying the highest increase in delayed neutron signal' '. In other cases, a defective fuel rod may not release noble gases for reasons that are fuel temperature dependent. At low power or temperature the UOj matrix will retain a large portion of the noble gases causing the defect to escape detection by the gaseous fission product monitoring system. Due to their water-soluble p r o p e r t i e s , trie radioiodines can, at times, escape before the noble gases. These defects are detected and located during the routine delayed neutron scans that are performed 2-4 times per month. The task of locating a fuel defect becomes more complicated when certain defects do not behave like the above cases. There is a growing database of CANDU-600 fuel defects that are not easily detected before they are discharged from the core according to the normal refuelling schedule. With hindsight, we now identify two categories: 1) high powered bundles with a small does not open up with time, and
defect
that
2) low powered bundles with a large defect. These systems meet the basic requirements of detecting and locating defective fuel. Once the fuel channel with defective fuel has been identified, the ch&tK^L can be refuelled during routine on-power fuelling operations. In fact, some fuel defects that are discharged from the core, are so small that they cannot always be visually confini ad in the fuel inspection bay. The absence of significant damage suggests that the fuel defects are being removed prematurely. 348 C N S 9th A N N U A L C O N F E R E N C E . 1988
The first type of defect allows noble gases to escape but not the halogens except during reactor power transients. The small hole size combined with the high inside temperatures likely prohibits water ingress and subsequent release of the halogens. These defects are detected by the gaseous fission product monitoring system but can not be located by the delayed neutron system using standard techniques.
The second type of defect does not allow detectable quantities of noble gases nor halogens to escape primarily due to its low power. The hole is large enough to allow water ingrtss, U 0 ; oxidation and subsequent release to the coolant. At low power and temperature the fission product inventories are relatively low and remain mostly trapped within the U0 3 matrix. The uranium is released which deposits within the primary circuit. These defects are not detected by the gaseous fission product monitoring system nor located by the delayed neutron system, especially if these defects are located near the channel outlet where there would be very little uranium deposition in core. Their presence is indicated by uranium buildup in other channels. This buildup becomes a source of fission products which is primarily detected by increasing levels of iodine-134 and increasing loop average delayed neutron signals* 3 '. There is another difficulty in locating fuel defects that reside in one of the fuel channels that have low delayed neutron signals due to hardware p r o b l e m s . All four o p e r a t i n g .CANDV 600s have up lo 26 of these low signal channels that are located near the core periphery. Due to the low flows in the sampling lines, the standard techniques using the delayed neutron system can not locate defects in most of these low signal channels. Their presence is detected by the gaseous fission product monitoring system. Many rules for analysing CAND1I-600 data have been generated and some of them are being applied during the routine analyses the data received from the sites. Some rules have been published in the open-literature'-*" 8 ' 1 0 > 1 1 ' , whereas others are still under development and being tested against data in ongoing jobs. Today, the manual interpretation and recognition of patterns in the data are becoming time-consuming and inefficient, to the p o i n t that not all the rules of interpretation can be applied.
delayed neutron scan files are simply updated with the current data and the program terminates (see Figure 1 ) . If consultation is selected, the program proceeds to generate two lists of suspect channels. Figure 2 shows the flow sheet for defects that do not release uranium and Figure 3 for ones that do release uranium.
DN SCAN DATA 380 SIGNALS ! DATA NORMALIZATION
I UPDATE
CONSULTATION A
* FILES
REMOVE LOW SIQNAL~\ CHANNELS FROM DATA \
URANIUM RELEASE t N
FIGURE 1: MAIN FLOWSHEET FOR DEFECT DETECTIVE
FLOWSHEET A ^S, JNO URANIUM RELEASE}/
PROGRAM DESCRIPTION There are two phases of DEFECT DETECTIVE: the first has been developed and is described in this paper, and the second is currently being developed and is briefly mentioned. Phase I will locate and evaluate the fuel defects that are easily found using the current techniques. Phase II which is an extension of Phase I, will help locate and evaluate defects that are difficult to find as described in the previous section. These defects include: high powered bundles with a small hole, low powered bundles with a large hole, and defective bundles residing in channels that have low delayed neutron signals. Phase II will also be used to search for a common pattern among several suspect channels that appear simultaneously. For example, high signals for a group of channels may be due to reasons that are not fuel related like neutron detector malfunction, reactor power transient during delayed neutron scanning, and so on. DEFECT DETECTIVE begins processing the delayed neutron scanning data by accepting as input, "normalized" signals from the 380 fuel channels in the core. This initial normalization removes certain factors that influence the measured signals, such as neutron counting time, reactor power, and detector sensitivity. If the user chooses not to consult DEFECT DETECTIVE, the
FIGURE 2 : FLOWSHEET FOR FUEL DEFECTS THAT DO NOT RELEASE URANIUM CNS 9th ANNUAL CONFERENCE, 1988
349
A channel is en the first if its delayed neutron signal is significantly higher than normal, a "step-wise" increase. Channels on the first: list only are called new suspects(see Figure O . A channel is on the second list if the signal increases slowly with time. These are called suspected small defects. The fuel defect |n channel L22 at Point Lepreau is a good example* '. In this case, the signal increased at a rate cf 3% per 10 days. Channels on both lists are called confirmed small defects.
f FLOWSHEET a ^ ^^UJRANHJM PELEASEy1 IT
•V APPPOX. LOCATION THtN g>V^|VjVf4 I WITHIN CHANNEL]
The recommendations p r o v i d e d by D E F E C T DETECTIVE, shown in Figures 2 and 3, include; -specifying the refuelling priority of a suspect channel, specifying the number of bundles needed to remove the defect, and listing additional actions that may be needed to confirm the presence of a defect. These recommendations are partly has 3d on data available from other sources such as the gaseous fission product monitoring system, radiochemistry data, and fuelling records. If a defect can not be found in Phase I, Phase II is recommended. Phase I of DEFECT DETECTIVE is written entirely in the PROLOG programming language. Execution of a PROLOG program formally corresponds to trying to establish the truth of logical assertions. Assertions are defined in the program to be unconditionally true (facts) or to be true if a sequence of other assertions is true (rules),
FIGURE 3: FLOWSHEET FOR FUEL DEFECTS THAT RELEASE URANIUM
NORMALIZED DELAYED NEUTRON SIGNALS FOR 8USPECT FUEL CHANNELS
Given a top level assertion to be proved, PROLOG proceeds to try to prove other assertions that support the truth of the top level assertion. I*: proceeds recursively until all of the remaining assertions are proved to be facts or until one of the assertions can not be shown to be true. In the former case, the top level assertion is true, and in the latter case, it is false. This process is called backward chaining inference. If alternative ways of proving an assertion are provided, PROLOG will try each one in turn until the truth of the assertion is established or until the alternatives are exhausted. This process is called backtracking.
NEW SUSPECT /
(LARGE STEP INCREASE)
SMALL DEFECT (SLOWLY INCREASING AITH TIME)
CONFIRMED SMALL DEFECT (CONTINUOUSLY HI3H S'QNAL)
TIME FIGURE h: PATTERN RECOGNITION FOR SUSPECT CATEGORIES 350
CNS 9th ANNUAL CONFERENCE, 1988
These features make the implementation of expert systems and parsers for natural language processing very simple in PROLOG. PROLOG allows the use of rich and complex data structures which can be used to describe relationships among data objects. In conventional languages with relatively simple data structures, such relationships must be embodied in code which complicates coding. Finally, PROLOG source code can usually be read as a specification of what the code is to do rather than as a detailed set of instructions about how to perform the job. This is referred to as having a declarative as opposed to procedural syntax. This makes PROLOG programs easy to write, debug, modify and maintain. An expert system shell was not used for DEFECT DETECTIVE because the program, in addition tc acting as an expert system, is required to perionr. several tasks outside the expert system mode]. Simple shells are generally not powerful enough to perform these tasks. The tasks include reading files of delayed neutron scan data generated by a FORTRAN program, doing linear regression calculations and plotting trends for fuel channels of specific interest. It is notable that the logic programming paradigm which PROLOG uses is easily
capable of accomplishing these procedural functions. As shown in Figure 5 PROLOG needs only a few lines to implement linear regression compared to FORTRAN which would need many steps.
• • • ; • • • . • • :
"
"
"
'•• =
'
• • • • •
" i ' 1 :•
• • . .
. . . .
.
.
:-
:.
»
,
;
.
<
/
.
i
v
.
.
.
.
DISCUSSION PROLOG is proving to be a useful tool for analyzing complex data associated with delayed neutron scanning. The clarity of PROLOG source code allows quick dad easy modifications of the rules used to examine the data. For example, it is possible to adjust the threshold value for identifying channels with increasing signals. Also, it is possible to fine tune the "step increase" threshold for the delayed neutron signal of each specific fuel channel.
i l ' l ' l . . .
CONCLUSIONS •• ."•.-•: • •-;-••••) i; • •:
•
.
.1.!.n.IT.i.i.r..'l.!:.J
, r I s . •. ' s . 1 } . y . . ( I T . I F . * : . ( . ' . « I . f i . \ :
:-
= 1" • i : . = H . - i
J
j .
1. The DEFECT DETECTIVE expert system for fuel defect detection offers potential for reducing the costs associated with the handling of defective fuel in the CANDU-600s. Its usefulness will become more apparent as experienced staff are replaced by inexperienced staff. 2.
FIGURE 5: LINEAR REGRESSION ROUTINES IN PROLOG
At this point, the program gives the user the option of proceeding with the session or exiting. This is done in cage the user merely wants to update the knowledge base with a new delayed neutron scan and consult it later. If a consultation is chosen, as it would be in the event of a Sudden increase in Xe-133 concentration in the coolant, the program proceeds to produce two lists of suspect channels (Figure A). After the suspect channels have been identified, the program proceeds to consider each suspect in turn. The suspects are grouped according to being on one of or both of the lists. As each channel is considered, the channel name is displayed as well as the primary heat transport loop that the channel is in and its identification in the delayed neutron scanning roomThe n m p r a m then initiates a series of questions about the channel to the user. By answering the questions, the program obtains additional information about the channel such as the date of the last refuelling, for example. This information is used to evaluate the condition of the defect and to try to locate it more precisely within the channel. The last point is important for deciding the number of bundles needed tc> remove the defect. The questions that are asked depend on the answers to previous questions. When the questions have been answered, the program displays an assessment of the defect condition and recommends a course of action. The program then proceeds to the next suspect channel. When all of the suspect channels have been processed, the program prompts the user to indicate whether or not plots of the data trends for any channels are desired. If so, plots are displayed going back over the previous ten scans.
Phase I of DEFECT DETECTIVE is now ready to be tested at AECL durir ring routine analysis of CANDU-600 delayed neutron data. Phase II is currently being developed. The usefulness of Phase II of the system will be demonstrated when it finds a fuel defect before expert staff can.
ACKNOWLEDGEMENTS The authors gratefully acknowledge the efforts of the following people for sending data to AECL-CO during the past few years: R.W. Sancton of NBFEC (Point Lepreau) , N. Macici of HO. (Gentilly), •J.C. Vinez and D. Castillo of CNEA (Embalse) , and D.Y. Eom of KEPCO (Wolsung).
REFERENCES
(1) D. Barber, J. Van Berlo, "How the CANDU 600 Meets the ALARA Concept in Design and in Practice", International Meeting on Nuclear Power Plant Operation, Chicago, 111., (1987 September).
(2) J.
Levy nd T. Mitchell, "Determination of Location of Failed Fuel in CANDU Power Reactors" IAEA Specialist Meeting on the Behaviour of Defected Zirconium Alloy Clad Ceramic Fuel in Watej Cooled Reactors, Chalk River, 1979 September.
(3) R.L. daSilva, D.R. Mccracken, K.J. Monserrat, "The Release of Fission Products From Defective Fuel Elements - A Study of Deposition Behaviour", AECL-8705, 1986 March. d ) A.M. Manzer, "Transport Mechanism of Uranium Released to the Coolant from Fuel Defects", Proceedings of the International Conference on CANDU Fuel, Chalk River, Canada, 19B6 October, (AECL Report 9330). (5) A.M. Manzer, N. Macici, "Fuel Detection by Radioiodine Monitoring", 8th Annual CNS Conference, St. John, New Brunswick 1987 June.
Finally, the program displays a summary of the conditions of the defeet(s)and recommendations for refuelling each suspect channel.
C N S 9th A N N U A L C O N F E R E N C E . 1988 351
(6) M . Floyd, B.J. L e w i s , A.M. Mar.zer, R.D. MacDonald, P.T. Truant, "The Detection, Location and Identification of Failed Fuel in Canadian Power Reactors", IAEA Coordinated Research Program o.i the Examination and Documentation Methdology for Water Reactor Fuel (ED-WARF) (to be issued as AECL Report 971+). (7) A.M. Manzer, "In-Core Assessment of Defective Fuel in CANDU-600 Reactors", IAEA Specialist Meeting on Post-Irradiation Examination and Experience, Tokyo, Japan, 1984 November. (8) A.M. Manzer and R.W. Sancton, "Detection of Defective Fuel in an Operating CANDU-600 MH(e) Reactor", ANS Conference on "Fission Product Behaviour and Source Term Research", Snowbird, Utah, 1984 July. (9) M . A . Shad, "Fuel Management and Fuel Performance at Point Lepreau Generating Station, International Conference on CANDU Fuel", Chalk River, Canada, 1986. (10)A.M. Manzer, R.W. Sancton, N. Macici, "Canadian CANDU-600 Perspective on Fuel Integrity Performance Indicators", International Meeting on Nuclear Power Plant Operation, Chicago, Illinois, 1987 September, AECL Report 9602. (ll)A.M. Manzer, "Defective Fuel Detection in CAHDU-600s", 6th Annual CNS Conference, Ottawa, Canada, 1965 June.
352 CNS 9!h ANNUAL CONFERENCE. 1988
END KITTING ROLLED JOINT INSPECTION APPARATUS FOR CANDU REACTORS E. CORMBLUM AND V. yil Ontario Hydro Toronto, Ontario ABSTRACT A system to rapidly and reliably inspect a large number of rolled Joints in the pressure tubes of the Ontario Hydro NGS Reactors was required. This paper describes the system as it was designed and installed, and its operation. INTRODUCTION In 1987, Ontario Hydro decided to inspect a minimum of 83 end fitting rolled joints at Bruce NGS. The purpose of this sampling was to obtain assurance that no manufacturing or installation defects are present. Although inspection tools and associated delivery system existed, they required approximately 8 hrs per fuel channel, and thus were adequate for the inspection of a small number of joints only. For a large scale inspection, the economic penalty resulting from the use of these tools was unacceptable. Consequently, it was decided to develop new equipment, which would shorten the time required for the inspection and retain the high performance standards of the existing equipment. A completely new system for inspecting the rolled joints was developed. Named PIPE (Packaged Inspection Probe), it is composed of three main elements: The inspection tool proper The tool delivery and operating system The data acquisition system (DAS)
process is repeated until inspection is completed.
the
rolled
joint
Th" inspection tool is shown in Figure 3. The Delivery and Operating System The inspection had to be performed without prior defuelling of the fuel channel. The delivery system, thus, had to be able to visit tne various channels of the reactor, handle the closure and shield plugs, and operate the inspection tool itself. It was decided that all these functions can be performed by the existing Fuelling Machines, once they are modified to handle the Inspection Tool. These modifications are described below. See also Figures 4-6. The Ram Head The Fuelling Machine Ram Head had to be redesigned, to allow an electrical connector to be installed on its front face, and provide passage for the coaxial cables leading to this connector. Guiding features, matching those of the inspection tool were added. This enabled the ram to pick-up the inspection tool from its storage station in the Fuelling Machine macazine, while, at the same time, performing the extremely delicate operation of mating its electrical connectors with its counterpart on the inspection tool. The electrical connectors had to be able to withstand repeated connection/disconnection operations while submerged in heavy water. The Spline
The PIPE project was funded and controlled by the Nuclear Systems Department of Ontario Hydro. The project was managed and executed by a design team with members from Ontario Hydro Design, Research and Operations Departments. Testing was done using the facilities of General Electric Canada. The Inspection Tool The actual inspection is performed using ultrasonic probes. Eight such probes are installed in a block, which in turn is attached to the front end of a unit called "The Transducer Carrier". These probes transmit their signals via individual coaxial cables to a connector located at the rear of the carrier. The block containing the probes is capable of both axial and rotary motions, and so an effective scanning of the rolled joint area can be performed. In order to minimize the inspection time, the ultrasonic probes are arranged so that when the block containing them rotates, the probes scan a section of the rolled joint with an axial length of 8 mm. Once the rotation is complete, the inspection block is moved axially 6 mm (to obtain an overlap of the areas being scanned) and the
In the Fuelling Machine, the ram head is attached to a ball bearing spline shaft. This allows the machine to move the ram axially, and also to apply torque to the ram head. Due to different travel lengths (the PIPE operations requested longer travel than presently available), a new spline shaft was regulred. This had to be hollowed to allow the passage of the cables. The original splines were ball splines, and no supplier could be found for PIPE bail splines. A new spline shaft/torque nut assembly was thus designed. Both the spline shaft and the nut are made of hardened stainless steel. The keys which run in the spline shaft grooves, are made of hardened stainless steel. The lateral surfaces carrying the bearing leads are covered with plates of a synthetic material. This greatly reduces the friction, and still allows the spline assembly to transmit the full torque required. Ram Extension Tube and Ball Screw Two more main internal F/H parts had to be redesigned - a ball screw, and an extension tube, which is installed between the spline and the ball screw. Both had to allow for passage of the cables
CNS 9th ANNUAL CONFERENCE, 1988 353
through their centres, all the way to the back of the fuelling machine. The Cable Rewinding Mechanism The total travel of the ram head is 330 cm. This is then the amount of cable that the ram will pull after it while travelling forward, and lias to be pulled back on some storage device when the ram is retracting. The cable has to be under constant tension to prevent contact with other fuelling machine moving parts and subseguent damage. A rewinding mechanism was designed and installed on the Fuelling Machine. It consists of a drum, on which thp cable is wound for storage, and a constant orgue motor, which controls the cable tension. The drum's shaft is hollow, and contains a cup. This serves as the penetration of the cable through the pressure boundary. The internal cable and the external cable are connected inside this cup, and then a potting material is used to encase the connection and seal the passage. A smaller drum is located on the shaft, external to the pressure boundary, for the storage of the external portion of the cable. Space constraints dictated that the individual coaxial cables are held together forming roughly a round bunch while located inside the spline/ball screw assembly. Once outside however, they are attached to a flat cable, which makes it easier to store it on the rewinding mechanism. In order to relieve the electrical connectors from the cable tension, a stainless steel wire rope carries the tension load, bypassing the connectors. Fuelling Machine Computer Software A copy of the Fuelling Machine software was modified to carry out the inspection of the fuel channel rolled joints automatically. An interface is also provided between the computers of the Fuelling Machine and the data acquisition system to synchronize the ultrasonic data collection with the motions of the Fuelling Machine during the rolled joint inspection. The Fuelling Machine computer also monitors the cable rewinding mechanism control circuits such that the Fuelling Machine operations would stop immediately if a cable breakage is detected or the motion of the cable is not synchronized with the charge tube or ram motions. Ultrasonic Inspection System and Cable Layout The ultrasonic signals from the probes in the transducer carrier are transmitted through a wet connector to the ram cable in the fuelling machine and through sealed in-line connectors to the flat cable in the cable re-wind mechanism. The cable from the external drum of the cable rewind mechanisms on both Fuelling Machines are routed to a common reactor vault penetration. Outside the reactor vault, the cables are connected to a switch unit which will switch either one of the two transducer carriers to the Ultrasonic Inspection System (UIS). The UIS consisting of ultrasonic instruments and data acguisition equipment is located in the control room at Bruce NGS. The ultrasonic instruments include remote pulser preamplifiers, switch units, balance potentiometers
354
CNS 91h ANNUAL CONFERENCE. 1988
and KB 6000 ultrasonic instruments. The data acquisition eouipment js based on a Compaq 386 personal computer with optical and hard disk data storage. The UIS is checked before and after each rolled joint inspection to ensure the ultrasonic probes and associated equipment are working properly. Eight circumferential 45 shearwave probes spaced one millimeter apart axially are used for the rotary scan. The ultrasonic signals from each probe generated by the calibration notch are checked and graphically displayed. The operator will be advised to take corrective actions if the signal from any one of the probes is below a predefined range. The data acquisition system collects the ultrasonic data during the rotary motions of the transducer block. The data is processed and displayed during the axial repositioning of the transducer block of 6 mm for the next rotary scan. Hardcopy of the ultrasonic data can be obtained in the form of a series of colour isometric plots. Each roiled joint inspection is uniquely identified and permanently archived in full resolution on an optical disk, allowing further analysis at a later time. Ultrasonic indications observed during the automatic inspections can be reassessed by manually commanding the Fuelling Machine to position the transducer block at the desired location, and repeat the inspection. If necessary, more detailed information can be obtained on the indication by performing an axial scan using one normal beam, two axial shear wave and two circumferential shear wave probes. KOLLED JOINT INSPECTION USING PIPE The Bruce reactor is placed in guaranteed shutdown conditions arid the D2O inside the fuel channels is kept at 28 psig and 37°C by placing the Primary Heat Transport System on maintenance cooling. Two transducer carriers and one defuelling carrier are loaded into each magazine of the two fuelling machines which have been modified with PIPE components. Five shield plugs are also loaded into each fuelling machine/magazine to be used as counter weights to balance the weight of the cable rewind mechanism which is located at the end of the fuelling machine. The fuelling machines are locked onto the fuel channel to be inspected and the channel closures and shield plugs are removed from both sides of the fuel channel. The charge tube picks up the transducer carrier at the channel outlet and at the same time the electrical connectors in the ram head and transducer carrier mate. The charge tube engages the transducer carrier in the liner tube lug and then performs an ultrasonic system check against a calibration notch located on the front sleeve of the transducer carrier. The charge tube at channel inlet picks up the defuelling carrier and engages it to the liner tube lug. The ram at channel outlet advances and pushes the transducer block out of the carrier body. The transducer block in turn pushes the fuel out of the inspection area while the other end of the fuel string is received by the defuelling carrier. The ram retracts to separate the transducer block from the fuel and the charge tube performs the ultrasonic inspection by rotating the
transducer block in a series of alternating 363° rotations with the transducer block positioned at a different axia) position for each rotation. AKer the rolled joint area has been inspected, the transducer block is returned into the carrier and another ultrasonic system check is performed. The transducer carrier is returned to the magazine at the channel outlet and the ram at the channel inlet pushes the fuel clear of the defuelling carrier and the charge tube returns the defuelling carrier to the magazine. The channel inlet rolled joint area is then inspected by repeating the above procedures but reversing the operation of the two fuelling machines. The average channel inspection time is two hours. Figures 1 and 2 depict the computer displays of calibration check and actual inspection. CONCLUSION PIPE has performed exceptionally well, meeting and
90 180 270 530 M L SB? SIGNALS FIGURE 1
exceeding the specifications. Compared to the previously existing systems, the inspection time was dramatically reduced - from 8 to just over 2 hours per channel. The goal of inspecting 83 Bruce rolled joints was exceeded - 108 Joints were inspected in the time period allotted. The inspection time reduction, and the reliability of the components and of the system as a whole have resulted in substantial cost savings for Ontario Hydro. During March 1988, the system was used again, with the same reliability and inspection time per channel. A similar system was designed and manufactured for Pickering NGS. Its testing was completed, and the system is now ready for operation. Due to the similarity between the components of Pickering NGS and the 600 MW CANDU reactors, this system can easily be used by them.
FULL SEP SGHALS FIGURE 2
CNS 9th ANNUAL CONFERENCE, 198B
355
CO
2
o m
TRANSDUCER CARRIER. FIGURE 3
Ultrasonic Transducer Carrier
Ram Head with Electrical Connector
Ram Spline Shaft
o
2 CO >
I 8 General Arrangement of P.I.P.E. Components in Fuelling Machine Head
z o & Co
FIGURE 4
Ram Extension Tube
Bam Ball Screw
Cable Clamp
Flat Coaxia Cable
> z c >
z n m
Ram Spline
Ram Head
Pin Connector
S.S. Wire
tS Individual Co«xi»l Cables
Schematic of Electrical Cable Arrangement Inside F/M Ram FIGURE 5
Ram Ball Screw
Cable S.S. Wire Clamp
Guiding Pulleys
Fuelling Machine (ReU
o z CO
External Drum
Bearing Holder
Io
o 2 O
Cable Rewinding Mechanism Assembly FIGURE 6
PICKeRING UUCLEAR GENERATING STATION - A MODERATOR INLET LINE HANGER REPLACEMENT
R.A. KIRKPATRICK (ONTARIO HYDRO) J.M. BOWMAN (BABCOCK & WILCOX CANADA] W.R. SYMMGNS, S. EL-NESR (ATOMIC ENERGY OF CANADA LIMITED)
INTRODUCTION
Atomic Energy of Canada Limited
Ontario Hydro's Pickering Nuclear Generating Station (PNGS), Units 1 and 2 were shutdown for large scale fuel channel replacement. In conjunction with this work, otheir nonroutine inspection and maintenance activities were performed to determine the overall condition of! the units. One inspection was a visual examination o£ components within the calandria vaults. During these inspections, it was seen that a moderator inlet line hanger (identified as HR-29) had failed in both units. Subsequent inspections during planned maintenance outages of Pickering N G S Units 3 and 4 revealed that hanger HR-29 hat5 failed and reguired replacement. It was determined that the failed HR-29 would need to be replaced ptior to each unit's restart and the replacement technique development should begin immediately.
Atomic Energy of Canada Limited (AECL} is Canada's national nuclear organization. It was established in 1952 by the Government of Canada with a mandate to develop peaceful applications of nuclear energy. AECL is responsible for research and development for nuclear related activities; designs and markets CANDU nuclear power stations; and provides engineering services to electrical utilities throughout the world. Babcock and Wilrox Canada Babcock end Wilcox is a major supplier of nuclear and fossil fuelled steam generating equipment and related services. Babcock and Wilcox Canada is the major supplier of st^am generators for the CANDU system and supplier of replacement steam generators for PWR systems. CANDU
To meet this objective Ontario Hydro Central Nuclear Services contracted Atomic Energy of Canada Limited (AECL) and Bahcock & Wilcox Canada (BWC) to supply tooling and manpower to effect these replacements. A strategy was developed to address the unique problems posed by this project. These problems included accessing tooling through small inspection ports, manipulating tooling from a significant distance and the high radiation fields within the vault.
The Canadian power reactor system is called CANDU, a design characterized by natural uranium fuel, heavy water moderator, pressure tuoe containment of primary coolant, bundle fuel and on-power refuelling. CANDU is an acronym derived from the words Canada, deuterium and Uranium. The most significant feature in the design is the use of a multiple pressure tube configuration rather than a single large pressure vessel. (Fig. 11.
AECL and BWC, working i n parallel, each developed tooling capable of operating in conjunction with the other to facilitate removal of the failed hanger and installation of a replacement. This paper describes t? replace hanger HR-29.
the
program
undertaken
Ontario Hydro Ontario Hydro is a Public Utility serving the nearly 9 million people of Ontario, Canada, in a geographical area approximately twice the size of France. Installed capacity in 1987 was 30,000 MW divided amongst three generation types. Nuclear Power accounts for 10,500 MW or 35% of Ontario Hydro's capacity contributing almost half of the total electricity demand. Ontario Hydro currently has 16 commercial CANDU nuclear units in service and 4 units under construction. Central Nuclear Services support the operating units. The group supplies specialist knowledge in specific areas, assisting the stations with long range studies, and interacting with Ontario Hydro design and research and outside agencies. 360
C N S 9th A N N U A L C O N F E R E N C E . 1988
FIGURE I PICKERING NGS A' VAULT
The horizontal fuel channel design facilitates the on-power fuelling operation which, in turn, is a major contributor to the consistently high CANDU rankings in lifetime performance listings of the world's power reactors. The design employs a heavy water primary heat transport system which transfers the heat from the reactor core to steam generators. The low pressure, low temperature moderator circulates in a system fully separate from the primary coolant system. It is contained within a horizontal cylindrical calandria vessel in which the fuel channels are supported. The conventional side of a CANDU unit is similar to that of other types of reactors. Background At the time the failed hanger (HR-29) was discovered in Units 1 and 2, the reactors were about to be recommissioned at the end of their retubing outages. The required schedule allowed two weeks for preparation of a replacement strategy, six weeks to develop and commission tooling, and two weeks to prepare procedures and train the replacement teams on a full scale mockup. The Pickering NGS-A calandria vault is a concrete box-type structure located approximately in the centre of the reactor building. The internal dimensions of this structure are 16.75 metres high, 12.2 metres wide and 6.1 metres deep. Located within the vault is the reactor calandria vessel which sits above the dump tank into which the moderator may be dumped in order to achieve a guaranteed shut-down state. The moderator inlet line is a 254 mm, schedule 10 stainless steel pipe which transfers the heavy water moderator between the moderator pumps and the calandria vessel This line is supported by
the hangers HR29 and HR30 attached to stainless steel lugs weldod to the base of the calandria. These basic components formed the geometric environment in which the replacement was performed. (Fig. 2 ) .
The environmental limitations included: (1) High radiation fields: measured at 150 to 500 R/hr. gamma. (2) Access restricted to 200 mm diameter inspection ports. Thus remote tooling W3.s essential to accomplish the replacement. General Replacement Concept To provide the most effective approach to the replacement, Ontario Hydro's Central Nuclear Services group devised a conceptual procedure that integrated the developed tooling of each organization, utilizing a combined team approach for the replacement of HF29. All tooling and procedures were developed fail safe operation as a primary consideration.
with
A great deal of the operation's success was due to the fact that the entire hanger replacement operation was rehearsed on a full scale mock-up in AECL's Sheridan Park Engineering Laboratory. The mock-up duplicates relevant portions of the biological shield penetrations, the calandria shell, the dump tank and the moderator inlet pipe. This mock-up was also used for tool and procedure development as well as for operator training. Four (4) inspection ports were utilized throughout the replacement. Tooling systems were permanently placed in the East Upper (EU), Center Upper (CU) and West Upper (WU) ports while the East Lower (EL) port wa s ut i 1 i zed for old hanger remova 1 and new hanger installation.
CALANDRIA
EXTENDED TRACK 1
> '
DUMP TANK
TUBE SUPPORT
e
FIGURE 2 MODERATOR INLET PIPE - PIPE HANQER REPLACEMENT CNS 9th ANNUAL CONFERENCE. 1988 361
The tooling in the EU and CU ports combined to effect the replacement while a camera system in the WU port provided an overview. To view all replacement activities, v total of five (5) cameras were employed. Cameras were situated in the EU, CU, WU and EL ports to provide both general and close up views of the vault work. Each tooling operator was provided with appropriate views for the specific operations by use of a central switching station. In addition, a central communications system was utilized to ensure safe and efficient operations- (Pig- 3 ) .
The main support tube consisted of a cut away aluminum pipe with remotely operated pneumatic support feet fitted with suction cups to hold the tube on the upper dump tank surface. (Fig. 2) . Into this tube was placed an I-beam track to allow the manipulator arm trolleys to be driven into .he vault. The main vault access trolleys which ride this track had mounted to them a hydraulically operated actuator with a manipulating arm which carried the end effector trolley. The end effector trolley was driven up and down the manipulator arm delivering a variety of end effectors for various operations such as grinding, cutting, gripping, and bolt extraction. In addition to the main operations, tooling vision was provided from vidoo cameras mounted on the articulated camera arm which accessed the vault through the EU inspection port,
Central Upper (CU) Port Tooling The tooling developed to access the hanger location through the CU poit vias a Vvydraulicallv operated manipulator named "ROBERT" (Remote Operated, Bent-Arm Examination and Retrieval Tool). ROBERT consisted of main boom sections of aluminum construction. Hydraulically operated elbow (90 degree) and wrist (135 degree) joints, hydraulic rotational joints in the forearm and wrist, and hydraulic insert and rotation at the CU port entrance were utilized to position the end effectors to the work location. (Fig. 3 ) .
The wrist consisted of a rotary actuator and mounting table for various end effectors. The rotational movement was primarily utilized for final tooling alignment although there was some load carrying capacity. FIGURE 3 ROBERT The forearm {between the tfrist and elbow joints) provided the necessary reach to access the hanger area Cron the CU port. The second rotary ioitit was located in the forearm. The tooling system and a basic description of the approach used are explained in the following sections.
There were two extension pieces which attached to the main boom on the port side of the elbow joint via a keyed coupling.
East Upper (EU' Port Tooling The tooling developed to access the hanger location through the EU port was based on gantry robot principles using 3 supported track installed in the vault along which could be driven manipulator arms to locally access the replacement site. By making use of end effectors attached to the manipulator a^ms, various inspection, cutting, bolt removal or holding operations could be performed. The basic tooling consisted of a main support tube which carried the access track. Mounted on this track were trolleys with single plane rotational joints (rotation in the other plane being obtained from access track rotation as required). Connected to the joints were tracked manipulator arms carrying end effector trolleys to hold the tools necessary for replacement operations.
362 CNS 9th ANNUAL CONFERENCE, 19B8
The entire ROBERT manipulator slides through an aluminum guide tube. This guide tube supporting the manipulator provided a smooth installation surface. ROBERT had on board two T^nual focus, auto iris, black and white video cameras >' one mounted on the wrist for tooling alignment and one mounted on the forearm for an overview of the wrist operation. Lighting of the area was provided integral with the arm. Both cameras were mounted on fai1-safe pneumatically or electrically actuated pivots which retracted during insertion/retraction through the guide tube. ROBERT had provision for pneumatic, or electrically actuated end effectors.
hydraulic
West Upper (WU) Port Tooling A third port, port WU, was used by Ontario Hydro inspection staff utilizing thei r Bent Arm Remote Nuclear Inspection Equipment (BARNIEI. BARNIE is a remotely operated inspection arm used to provide visual and 1ighting assistance to the repair teams during the hanger replacement.
The ROBERT arm attached a pneumatically operated retrieval clamp to the hanger. This clamp was connected to ROBERT with a stainless steel Safety cable to prevent loss of the hanger components in the vault. The grinding end elf ector was attached to the EU manipulator arm. The grinder was then actuated and the bolt cut between the lugs. (Fig. 2 ) .
Replacement Procedures The replacement procedure began of tooling into the calandria vault.
with
placement
At completion of the cut, the bolt head fell into a retrieval cup attached to the manipulator arm end effector and was removed with the grinding head.
Tooling Installation The tooling was installed through port EU with placement of the two piece main support tubes using a roller bearing pad in the outboard end of the port. Once installed, support legs were lowered to the dump tank and using a vacuum pump the suction cups were fixed in place. In this way a stable working platform was provided. The trolley tracks, manipulator arms and camera booms were then installed. The ROBERT arm was installed through port CU. The arm was insta 1 led in f o\ii sections. P,t each section joint all pneumatic, hydraulic and electrical connections were made using quick connect couplings. Before the final section was coupled, a reaction force sleeve was installed to react the linear and rotational loads applied by the arm on the guide tube at the outboard end.
The remaining bolt piece was then removed by a bolt extraction end effector which consisted of a pneumatic plunger and locating assembly attached to the trolley tf.hie on the manipulator arm. The bolt was pushed out of the bracket assembly into a catch container attached at the end of the bolt extractor. The hanger was then removed from the 1 ug by the ROBERT arm which lowered the hanger to the vault floor. An articulated arm then reached through port EL, two feet above the vault floor carrying a TV camera and lights, snared the hanger safety cable, and removed the hanger through port EL.
Replacement Hanger Design Hanger Removal Hanger HR-29 is a carbon steel pipe support anchored beneath the ca.landria vessel supporting the stainless steel moderator inlet line. Figure 4 depicts the failed hanger HR-29. The steps followed were as follows:
in
the
removal
procedure
The original hanger installed as HR29 was a standard carbon steel pipe hanger. To develop a replacement hanger a number of requirements had to be taken into consideration. The new hanger was specified to resist stress corrosion cracking and had to be replaceable using remote tooling. To facilitate these requirements, the hanger wdS designed as shown in Figure 5.
LOCK WASHERS
FIGURE 4 FAILED HANGER HR 29
FIGURE 5 REPLACEMENT HANGER CMS 9th ANNUAL CONFERENCE, 1988
363
The features incorporated into the new hanger started with the replacement of the upper eye by an open hook since tho roinstallation procedure was easier with the upper bolt i nstalled prior to the hanger itso If. The hanger rod was changed to a stainless steel rod into which was designed with a turnhuckle to allow final adjustment of tho hanger length. Finally the hanger strap, again manufactured from stainless steel, was designed to include a captive bolting arrangement to allow ease of assembly once inside the reactor vault. Replacement Hanger InstaHat ion The steps followed were as follows:
in the installation
procedure
The replacement stainless steel nut and bolt assembly were attached to a special end affector on the EU manipulator arm. This end calandria inserted. completed (Pig. 6 ) .
effector was then placed around the hanger lug, and the nut and bolt Activation of the pneumatic nut driver the replacement bolt installation.
Once in position the lower saddle half was raj sed into the closed position by the ROBERT arm and the captive nut and bolt were torqued by the nut runner end effector on the roani pulator arm to complete hanger replacement.
Analys_is_ o£ _the Fai led_ Hanger Prior to the replacement operat ion, vi sual examination and know]edge of the vault environment had led to the suspi cion that there could be two failure mechanisms. The two suspected mechani sms were either stress corrosion cracking or corrosion fatigue. The hangers that were removed from Units 1 , 2, and 4 at Pi ckering NGS-A were exami ned at AECL's Chalk River Nuclear Laboratories. It was concluded that the failure mechanism was intergranular stress corrosion cracking. Figure 5 shows the fracture in the Uni t 1 hanger. It was located at the bracket collar which was the area of high residual stress. Typically, hangers removed from Unit 3 and 4 showed similar failure modes and locations. The hanger removed from Uni t 2 fractured at the saddle bolt hole location. It was determined that high residual stress was also present in this area due to cold punching. Conclusion The replacement of HR29 m Pickering Units 1 to 4 was successfully completed. The team approach employed by Ontario Hydro proved to be a successful method of resolving a complex problem in a 1imi ted time.
\
FIGURE 6 REPLACEMENT BOLT AND INSTALLATION TOOL
The replacement hanger was attached to a retrieval cable and installed into the vault through port EL. Once in the vault the retrieval cable was retracted through port EU to r£»ise the hanger up the side of the dump tank to be picked up by the ROBERT arm. The ROBERT arm then positioned the hanger between the calandria bracket lugs, and the EU articulated arm placed the hanger hook over the stainless bolt. The ROBERT arm then released the hanger and an end effector on the manipulator arm adjusted the turnbuckle to place the hanger saddle in tho correct position.
364 CNS 9th ANNUAL CONFERENCE, 1988
THE DEVELOPMENT OF OKTUUO HYDSO'S ON-LUE OCMPUTERIZED VHKOG ncoRHftrrcN SYSTEM R.P. LIHJSAY B.A. ROLFE Ontario Hydro, 700 University Avenue Toronto, Ontario
ABSTRACT
reduced software maintenance and support. It was then possible to consider the development of features which could not have been justified with a smaller user base. The various user groups are listed in table 1.
An on-line database containing control wiring records and associated information has been developed and placed in service for operating nuclear power plants. The maintenance and modifications of the electrical and control systems of the plants are facilitated by the use of this electrical engineering Information system.
TABLE 1: USER GROUPS USER ORGANIZATION
POPULATION
Design Engineering
40
omaxcnoN Design Drafting Ontario Hydro recognized the advantages of computerized information systems early and introduced these systems where possible for the design of Pickering G.S. "A". Station wiring and cabling records were one of the first applications of a computerized system. The original programs were written for Pickering G.S. "A" in 1965 and were also used for Pickering G.S. nB". Based on the experience gained on Pickering G.S. "A" an improved version of the system was developed and used on the Bruce and Darlington plants. Over time these wiring systems have been modified to accommodate differing requirements as well as to incorporate changes in computer technology. Repeated patching of the old softvrare led to difficulty in achieving reliable and economic operation of the system. Many of the decisions taken in the original design of the wiring system were no longer valid due to changes in the relative costs in the work environment. The difficulty of maintaining the obsolete software was also a concern. In 1984 a decision was made to develop a new advanced on-line wiring system to replace the existing systems.
DEVEUXHENT Prior to commencing the re-development work, a comprehensive study of the technical and the financial viability of the project was undertaken. This study was carried out by functional engineering and software development staff. Input from user groups was sought and utilized. The major conclusion of this review established the technical direction for the future and financial constraints. Also, at that time certain basic requirements were established. It was decided to utilize an on-line database which would Increase costs in the area of computer processing and storage and decrease the use of hardcopy reports and other labour intensive practices. This direction was chosen to achieve a cost split which would load the cost stream in areas where future escalation in cost are anticipated to be low (i.e. computer costs) and to reduce as much as possible the costs which are subject to escalation (i.e. labour costs). Another important goal was to have one standard system for all generating stations. This goal would result in the benefits of having cannon operating methods and also
150
Construction
900
Operations
400
Protection & Control*
40
* Switchyard operation and maintenance.
Early in the project a detailed analysis of the old system was performed with input from all the user groups. This led to the development of detailed specifications which included the basic functions of the old system plus additional features which could readily be incorporated into the new system. In many cases these additional features would remove manual paper support systems which had grown up around the application. It was possible in many cases to better utilize data already in the records but not readily accessible in the old system. SYSTEM FEATURES
The key functions provided by the new on-line wiring system ar«: as follows: Management Tool -
Security system to control usage Cable inventory control Work tracking Problem reporting, on-line reference manual, & news bulletin
Design of Electrical Control Systems -
Wire and cable termination records Cable schedules Equipment coding and schedules Cable tray allocation and loading Power supply allocation schedules Channel and voltage separation Automatic features to assign terminations and conductors CNS 9th ANNUAL CONFERENCE, 1988 365
handling of this information. The new system has several key features which allow such productivity gains to be made in plant operation. A significant consideration in good staff utilization is to provide people with the information they need to do their job. The on-line wiring system is designed to achieve this.
Construction of Electrical Systems - Record of construction progress - Identification of outstanding work and "changes only" - Record of turn-over status - Variety of reports to suit the job requirements
Access to the Information Operation and Maintenance Tool - Graphical representation of connections for a circuit - Work control package for plant modifications - Load and power supply listings - Complete list of equipment locations SYSTEM HARDWARE ARRANGEMENT The database is resident on the corporate mainframe computers located in the head office complex in Toronto. User remote terminals are located at each generating station and also at head office design engineering and consultant locations. The head office facility supports volume printing and micro fiche production. See figure 1.
The on-line environment provides several opportunities to improve the access to the stored wiring information by the general user population. The requirements of the users is to provide access to the data on a 24 hour a day 365 days a year basis. In addition to timely access, the location of the terminal equipment is critical in the effective application of an information system. A multi unit nuclear power plant occupies a large physical area. Some stations are approximately lkm in length; thus the amount of time taken in internal travel to and from the job site is considerable. The wiring system has removed the necessity to travel between the job location and the shop to retrieve design documentation used in many maintenance activities.
Issue System APFlICKnCH FOR OPERATING IMNIS The new system is utilized during the design, construction and operating phases of a plant life cycle. Vast amounts of documentation are required to record the complex electrical and control sytem designs in contemporary nuclear plants. Substantial benefits accrue in productivity enhancements in the
The wiring system contains not only the latest approved record of the plant design but also the preliminary and historical data of the detailed information. This allows the users to view by means of a variety of reports not only the latest design requirements but also the historical records of previously issued levels of design information. The process by which the creation of this "issued" information is generated is governed by a date
FIGURE t 366
CNS 9th ANNUAL CONFERENCE, 1988
stanping of the various detailed records for each connection. Once a record has been issued it cannot be removed from the system. To revise the issued information a copy is transferred to the design level and modified. This revised information is then issued to the field. Once this occurs both the original issue ana tne new information are available for access by the field staff. This process is illustrated in figure 2.
TD!
CESIGN J FLAG INDICATES RECORD RECORD | TEMPOpARY IDESIGN STARTED! DESIGN ID) I AUG 10 1987 I I
! i
DESIGN RECORD
ID
IDESIGN STARTED1 I AUG. 24, T9B7
TEMPORARY COPY OF ISSUED RECORD AUG. 20. 1987
1
ISSUEU AUG 20. 1987
r- LAG INDICATES RMANENT MECORD ISSUED in
Li ISSUED SEPT. 4. 138 7
PERMANENT RECORD
Construction Status A nuclear plant contains millions of terminations for control wiring. The accurate record of the installation of the control wiring is necessary to maintain safe and reliable operation. The system has the facility to record the installation status of each cable and termination. The construction status routines also verify that a logically correct construction sequence has been followed i: the execution of the installation procedure. Reporting System Information can be retrieved from the databases tyj means of on-line asvd batch reports. Theie ate 29 different on-line report types and 25 batch reports. Within each report type there are several options based on the issue level or the construction status which can be selected to supply information to perform a specific task. By sorting the information to best suit the task at hand significant gains in productivity can be made. The system can produce reports on a physical basis (i.e. all the connections for a specified panel) or on a system basis (i.e. all the connections for a specified process system.) Changes Only Reports. These reports provide detailed instructions to the tradesman to perform certain specific tasks. The reports are arranged to show deletions on the first line and the new connection on the line below. There are three basic types of changes only reports; 1. Outstanding This report identifies the work which is required to take a portion of the installation from the as installed condition to the latest issued revision.
2.
Specified Issue Date This report identifies the work to be done to implement an installation to a specific revision which may not be the latest.
3.
Changes Between Two Dates This report identifies the differences between two specified dates. Work outstanding is not identified as this report includes all changes between two dates and does not identify any activity with respect to the installed condition.
An example of a changes only report is shown in figure 3. It should be noteo that the determination of work outstanding is performed as part of the reporting system. The determination of work outstanding must be done at the reporting stage as the work to be done is a function of the specified installation to be achieved, combined with the actual condition in the field at the time of the performance of the work. Graphical Output. The system has over 1000 users within the organization. These users vary from individuals TM> are employee, essentially full time in the use of the system to those who utilize the system for only a relatively small portion of their job duties. To make the system as effective as possible, it is important to make the output easily understood. The report format which is most frequently used lists all the connections and cables etc. for a single electrical path. The text version of this report can on occasion be very large and difficult to assimilate. For plant operation it is necessary that maintenance staff quickly identify the correct electrical circuit from the listing of various conductors. To assist the users a graphical display of this report has been provided. This report is created by a PC application which interrogates the mainframe to acquire the necessary information. The data are downloaded onto the PC where a local application program analyzes the data and creates a graphical representation of the text. Hardcopy output of the report is provided by means of laser or dot matrix printers attached to the PC. A sample of this report is shown as figure 4.
As described previously a large number of users are allowed access to t.ie database This creates concern that misapplication of some of the business functions of the system will occur. Many different groups with different job duties are simultaneously attempting to go about their duties. To control the operations within the data base an extensive security system has been developed. This system controls access to the levels of information that various users can view. Typically the field users are not allowed access to the design level information prior to it being released to the field via the issue procedure, in addition each of the 131 transactions is controlled individually as to who can perform each of the add, modify, delete, and display functions. An example of where this capability is used is tine update of cable tiay data. This capability is reserved for the drawing office with the field having only the capability to view the data. The submission of and selection of batch run times is also controlled by the security system for economic reasons. CNS9th ANNUAL CONFERENCE, 1988 367
GTCL/ZSSUE REPORT DATE 87/05/01
GROUP TERMINAL CONNECTION LIST (STCL) PAGE NO CSTATUS (HOSK OUTSTANDING: .ATEST CONSTRUCTED TO 87-04-30) GROUP ID:
DEV TO OEV
//SB5155A
(IN)
UNIT 2
COLOUR BIND HIRE
TERMINAL
CABLE NUMBER
1
PL351
INCOMING DEVICE OR STRUCTURE CHAN TERMINAL IDENTIFICATION
*»» OLD: 9 C (Hut NEW: 9
B H B H
02-5153-050<*0
/5153///PS
3
B H B H
02-5153-05040
/51E3///PS
3
«** NEH: 22
B W W 0
02-5153-05040
/5153///PS
•*» NEH: 13
B H G H
02-S153-0E040
*** NEH: l
B W H S
**» NEH: 15
COMMENT
MIRE NUMBER
02,5153,000320 RHV
STATUS CONSTR. TO ISSUE
ISSUE DATE
85-10-15 85-10-11
85-10-11
02,5153,000320
» 85-10-20
6
02,5153,000323
» 85-10-11
/5153///PS
7
02,5153,000324
* B5-10-11
02-5153-05040
/5153///PS
8
02,5153.00032S
» 85-10-11
B H BR N 02-5153-05040
/5153///PS
9
02,5153,030326
» 85-10-11
*** NEH:
16
BHHBfl 02-5153-05040
•51S3///P5
10
02,5153,000327
* 85-10-11
*** NEH: 17
B H S W
02-5153-05040
/51S3///PS
11
02,5153,000328
• 85-10-11
**« NEH: 18
B H H S
02-5153-05040
/5153///PS
12
02,5153,000329
• 85-10-11
LOCATION: PUHPHOUSE DEV TO OEV (OUT)
TERMINAL •»• OLD: 3 C •«* NEH: 3
& IHHI
COLOUR BIND HIRE
UNIT 2
CABLE NUMBER
/5153///PS
DESTINATION STRUCTURE
CHAN TERMINAL
B H B H
02-5153-05040
//PL351
9
B H B H
02-5153-05040
•VPL3S1
9
B H H 0
02-5153-05040
//PL351
B H G H
02-5153-05040
//PL351
COMMENT
HIRE NUMBER 02 ,5153,000320
RMV
STATUS CONSTR. TO ISSUE
85-10-15 85-10-11
ISSUE DATE
85-10-11
02 ,5153,000320
• 85-10-20
12
02 ,£153,000323
• 85-10-11
13
02 ,5153,000324
* 85-10-11
NEH:
7
FIGURE 3
As-built System In the movement from the batch environment to the on-line environment it became necessary to provide a rapid method of incorporating any field initiated changes in the documentation as quickly as possible. The system can only be used in the operational plant if the data reflects accurately the actual field condition. To achieve this accurate reflection of the as installed condition it was necessary to develop a scheme which would allow the updates to be made in the field in a timely manner while providing all the checks and balances required by the quality control process. This is done by the application of the security system and the creation of capability to record these field modifications as unapproved. In addition a systematic log is maintained of all field initiated modifications. Work Plan The continued operation of a nuclear power plant requires that modifications to the plant control systems be incorporated in planned and correctly sequenced order. A s many changes may be in various stages of implementation at any one tine and the
368 CNS 9th ANNUAL CONFERENCE, 1968
installation of these changes must be co-ordinated with the plant operational requirements, it is necessary to carefully control the revised design documentation to ensure that the status of the change is known at all times. The work plan system which is an integral part of the wiring system allows the field staff to assemble discrete portions of the outstanding design work in work packages which may be individually scheduled and monitored. Each work package is assigned to a "work plan holder" who is responsible to control and monitor the progress of the change. After the specific work items have been identified by the work plan holder any further revision of the design information is automatically brought to the attention of the work plan holder. Other features within the work plan system facilitate the reporting of the work to be done and the ultimate feedback of the completed work after the installation has been made.
CORRQir SQUIDS AID FUTORE FUNS At this time the initial software development and testing is complete. The conversion phase is underway, with 3 units of Ontario Hydro's 20 nuclear
units transferred to the new system. It is intended to transfer the remaining units during the balance of 1988 and in early 1989. It is intended to continue to develop the application to further integrate the design information for future facilities. Plans are beinq prepared to develop a system to automate the recording of equipment specifications and the setpoints. This system also will track the installation and calibration of the equipment in the construction and operation phases of future projects.
(1) N.H. Thompson, J.E. Saul, B.G. Overton, and S. Godfrey, "A Computer System for Station Wiring". Paper presented at the IEEE PES Summer Meeting, San Francisco, Cal., July 9-14, 1972. (2) R.P. Lindsay, R.T. Friedmann, "CADS/Computer Wiring Interface". IEEE Trans.-Energy Conversion Vol. EC-1, pp.82-87.
/•JB6193 7172/L503LS2 05000063663 BWOU 35 I 05800043603 BWGV 7172/L503LS1 "CIIF194U 1
235
BWEW "JB6103 ••CBF194V 05000063603 BUOY 233 33
"CDFIS2V /5 95088861191 25
M2
••CDFI78H
™
"CDF184H
494 05000060007 BWBKO
15 65980868071 BUGR
34
16?
0S89ee6ei2i BUVG
34
'7172/P563R2 4
D23
^7I?2/P583R1 • 7
024
FIGURE 4
CNS 9th ANNUAL CONFERENCE, 1988
369
COMPARISON OF THE RESULTS OBTAINED TN FUEL MANAGEMENT AT THE EMBALSE NUCLEAR STATION (C.N.E.) AND OTHER SIMILAR STATIONS, AND ALTERNATIVES FOR OPTIMIZATION
C. MORENO, A. DE PA2, J.C. VINEZ
Comision Nacional de Energia AtoT.ica Central Nuclear Emhslse Republica Argentina
^.
SUMMARY
This worK d e s c r i b e s the operating experience of the Embalse Nuclear S t a t i o n fuel management from fresh core to d a t e , and a comparison i s made with the performance reached by the other CANDU 600 s t a t i o n s (Point Lepreau, G e n t i l l y - 2 and Wolsung) from the p o i n t of view of fuel management with the i n t e n t i o n of understanding the d i f f e r e n c e s , in order to a s s i s t in the achievement of optimization a l t e r n a t i v e s . The data for the mentioned comparison with regard to the other s t a t i o n s were obtained from the CANDU Owners Group Information Exchange System (COG)• The Embalse average core burnup i s 4050 MWd/t.U (97.2 MWh/KgU) and the monthly average e x t r a c t i o n burnup i s 7800 MWd/t.'0. (187,2 MWh/KgU) with an average consumption of 14.2 E/C per f u l l power day. Up to d a t e , and for the t>aine thermal energy generated Embalse shows a consumption in the order of 180 0 fuel elements l e s s than i t s s i s t e r s t a t i o n s . At an estimated value of USS 4000 p e r bundle, t h i s means a saving of USS 7,200,000 in the approximately 4 years of commercial o p e r a t i o n . A comparison i s also made of the burnup and fuel consumption of the 4 stations, standarizing s i g n i f i c a n t parameters, which also shows a s l i g h t l y s m a l l e r consumption for the Embalse S t a t i o n .
- Operating power - Uranium weight per fuel element - Poisons dilutedin the moderator, as well as on the c r i t e r i a channels to be refuel led.
For the paid to burnup. channels
above reasons r special attention should be find the possible ways to maximize extraction In p a r t i c u l a r , a careful selection of the to be refuelled i s of help.
In Embalse, the time average (Ref. 1 ) contemplates two burnup extraction zones: an internal zone with 132 channels, averaging 7 75 7 MWd/t.U. (186,2 MWh/KgU) and an external zone with 248 channels, averaging 6830 MWd/T.U. ( 163,9 MWh/KgU) . The value for the whole core i s 7190 MWd/t.U.{ 172,6 MWh/KgU) • The safety limits are 7.3 M W for channel power and 935 K W for bundle power.
FUEL MANAGEMENT PERFORMANCE
AT CNE
INTRODUCTION
The basic o b j e c t i v e of fuel management is to maintain the r e a c t o r c r i t i c a l while maximizing fuel e x t r a c t i o n burnup and c o n t r o l l i n g power d i s t r i b u t i o n to s a t i s f y channel and bundle powers safety and operating limits. The most significant parameter to assess fuel management performance is extraction burnup/ as i t represents a measure of the use of the fuel, since i t i s the fission energy generated by uranium mass unit.. in turn^ extraction burnup depends on several which affeet neutron economy, such as: - Moderator purity - Parasite absorptions within the core
370
the
The effort applied to increasing the extraction burnup has direct economic effects, as fuel is one of the significant elements forming part of the cost of e l e c t r i c i t y generation. The elimination of 1 tnk of absorber implies a burnup increase of 116 MWd/t.U., which brings about a saving of approximately US$ 250,000 in fuel costs in one year of operation, at a typical load factor of 85%.
3.
2.
adopted for selecting
CNS 9th ANNUAL CONFERENCE, 1088
factors
From f i r s t c r i t i c a l i t y to march 31/88, the Embalse reactor generated energy equivalent to 1 146.0 FPD {full power days). F igure 1 shows the powe r h is to ry of t he Embalse• I t may be observed that during the f i r s t years the ope r a t ing powe r was redu - M due to nat io nal gr id r e s t r i c t i o n s , but in tne last years we have had a high load factor, especially in 1987 and 1988. As from 1985, a j o i n t study was perfomed between AECL and CNEA on load following of a CANDU s t a t i o n , whereby several cycles (power reduct ions} were carried out during weekends (Ref.2).
HMHAI.SK
From May to early July 1987 the opt=»ration of the CJE was a t y p i c a l . A malfunction ing of the fuel ling machine bridge side "C" transmission and indication sys tern, added to a labour eonf l i c t , forced to ope r a t e for an extended period without refuel l i n g . After 3y days in such c^ndition, the CNEA a u t h o r i t i e s decided t o take the • ' .Ion oat of s e r v i c e . At that time, 17 of the 2 t a d j u s t e r rods had been removed, and power was a t 74%. After overcoming the problem, ncminal condit ions were restored by an intensive refuelling program. Up to 6 r e f u e l l i n g operations per day were made, and the average ref uel ling value during the recovery stage (24 days) was 3-5 refuel lings/day (Ref.3).
I'. I'.-JKR
1984
Figure 2 shows the evolution of average core burnup and average extract ion burnup (monthly averages), from fresh core to March 1988. Month-to-month f l u c t u a t ions respond to d i f f e r e n t causes- There are always v a r i a t i o n s due to the channels selected, operating power, moderator purity, etc.. In p a r t i c u l a r , the low value observed for October 19 86 (729 fpd) i s due to the fact t h a t on t h a t date 8 channels were emptied to measure garter spring p o s i t i o n s between the pressure tubes and c a l a n d r i a tubes of such channels. In May-July 1987 (875 t o 919 fpd) t h e e x t r a c t i o n burnup exceeds 8000 MWd/t.U. as a consequence of reduced power operation and adj u s t e r rods removed from the core as described above.
1986 E j J U L . I AUG. i
Figure 3 shows fue 1 cons umpt ion for the same p e r i o d , represented in bundles.
SET,
Table 1 summarizes the main fuel management parameters since 1984 (beginning of commercial operation) u n t i l the f i r s t q u a r t e r of 1988 a s annual averages.
mm
FIGURE 2
1987
tf£BJ££ COHS AND BUT BURMP VS. PPD
1988 A.\
JAN. I FEB. I HAHCK [ APRIL ( fttl"
M
I
I
\
Jffl
1
1
1
^W/WX//////Y/////A ^'/////,y///.- V/////A *>&/////&//• W/A *••<•'//W/'/'Y/////A lit''-'' "•%•/ ' '
-Y/////4
•<--'////A///'/-X/////A •\s/7//W- Y////A <*/.'////&./I/X/////A 1
I
I JUNE
I
AUG.
SETT.
OCT.
NOV.
|
DEC.
1 I I
0
100 200 300 400 500 600 700 M0 900 1000
nru porn puts
CNS 9th ANNUAL CONFERENCE, 1988 371
T A B L E
-
MAIN FUEL MANAGEMENT PARAMETERS ANNUAL AVERAGES
1 1 SfEAR 1 1 1 1984
1 CHANNEL REFUELLING
BUNDLES DISCHARGED ANUAL ACUM.
FULL POWEii DAYS ANUAL ACUM.
BURNUP
BUNDLE PER DISCH. F . P . D .
240
1 1 1
1928
1928
154 . 4
248. 5
3681 o| 5714 .0
12 . 4 9
1 1985
531
1 4233
6161
27 1 0
519. 5
3881 0| 6953 . 1
15 . 6 2
1 1986
394
1 3 160
9321
214 8
734. 3
4099 0| 7428 . 5
14 . 7 1 |
1 1897
568
| 4544
13865
319 5
1053. 8
4066 T\ 7812 . 2
14 . 2 2 |
1 1988*
52
1 1292
14281
92 2
1146. 0
4118 2| 7818 .C
14 . 0 1 |
CORE |
I
(*) first quarter of 1988.
Another parameter assisting the fuel management assessment is the failed fuel rate. As at march 19 88, of a total 19725 irradiated bundles, with 15061 discharged bundles/ 11 failed bundles were found.This value corresponds to foreign origin fuel and i s wi thin acceptable limits. Power distribution was at a l l times maintained within the maximum allowed power limits both for channels and for bundles • This was s t r i c t l y observed even during the intensive restoration of reactor nominal conditions carried out after the operation with removed adjuster rod banks described above. The fuel management code used at Embalse is the PUMA-C code developed by CNEA (Ref. 4). The same was compared with Canadian codes and subsequently adjusted with measurements taken at low power after f i r s t c r i t i c a l i t y , and is periodically verified with the readings from the 1U2 in-core vanadium detectors. These comparisons are producing good results t the mean quadratic error between the measurements and the values calculated with PUMA-C being in the order of, but less than, 2% (typical value 1.8%). Embalse is carrying out the irradiation qualification of the Argentine-origin fuel bundles. A failure epidemic took place in late 1985. This originated a post-irradiation study to determine the causes of such failures. The problem was caused by a deficiency in the plug-sheath weld. After overcoming this shortcoming, qualification was continued, without incidents up to date.
372 CNS 9th ANNUAL CONf ERENCE, 1988
As a summary, we present the typical values reached at present:
GENERATED ENERGY ACCUMULATED TO 0 3 / 3 1 / 8 8
1146 FPD
TOTAL NUMBER OF DISCHARGED FUEL BUNDLES TO 0 3 / 3 1 / 8 8 15061 N a t , - U 104 D e p . - U TOTAL NUMBER OF IRRADIATED FUEL BUNDLES TO 0 3 / 3 1 / 8 8 19725 b u n d l e s NUMBER CHANNELS REFUEL TO 03/318C AVERAGE CORE BURNUP
2077 C h a n n e l s 4050 MWd/t.U 9 7 . 2 MWh/KgU
AVERAGE EXTRACTION BE/RWUP . . . . 7800 MWd/t.U 18 7 . 2 MWh/KgU NUMBER OF DISCHARGED BUNDLES PER FULL POWER DAY
1 4 . 2 Bun/FPD
FAILED FUEL/TOTAL IRRADTED FUEL ( C a n a d i a n - b u i l t f u e l ) . . . 0 . 0 6 %
FIGURE 4
CUXULATIVB SUi/BEP
CUMULATIVE NUMBER OF BUNDLES FUELLED VS FULL POW£R DAYS FOR THE FOUR STATIONS
OF BUNDLES REFUELLED VS. FPD
16,000' I
0
4.
100 200 300 400 500 600 700 800 900 10C0 TOLL JVttS PITS
I I I I I I I I I I I I I I I I |
| 1
1200
COMPARISON WITH OTHER STATIONS
In o r d e r t o a s s e s t h e performance of o u r s t a t i o n with r e s p e c t to fuel management and r e f u e l l i n g strategy, and u n d e r s t a n d t h e d i f f e r e n c e s t o i d e n t i f y t h e most significant reactivity parameters with a view to an optimization, a comparison was made with the cither CANDU 600 s i s t e r stations (Point Lepreau, Gentilly-2 and Wolsung).
1 I I f I I M
I I I I I I I I I 1000
1300
lull pomr days
T A B L E
-
2
AVERAGE REFUELLING RATES AT THE CANDU 6 0 0 ' s The data for the other stations were taken CANDu Owners Group (COG) Information System. Figure 4 shows the consumption of fuel function of full power days.
from the Exchange
elements as a
I t may be observed that the Embalse curve is below that of i t s sdster stations in a value which for 650 f pd was approximately 1000 bundles. In Embalse's current situation (1146 fpd), the difference exceeds 2 000 bundles with regard to Point Lepreau and i s in the order of 1500 bundles when compared to the other two plants.
REACTOR
| Point Lepreau
| AVERAGE REFUELLING RATE | 1 (BUNDLES/FDP) ( 1 FROM| 8 7 / 0 4 / 0 1 | 8 7 / 0 7 / 0 1 | 8 7 / 1 0 / 0 1 | \ TO | 87/12/31 | B7/12/31 | 87/12/31J
This is also reflected in the average extraction burnup, which is also a measure of specific consumption. Table 3 shows average extraction burnups for the 4 s t a t i o n s . To supplement the comparison, i t is useful to note that the fuel failure rates are of the same order for the four stations (for foreign-origin fuel). Table 4 shows such comparison
I
15.6
|
I 15.1
15.1
14.7
|
16.3
16.0
15.9
I
Gentilly - 2
I
Wolsung 14.1 Embalse
Table 2 shows fuel consumption per full power day of the four stations for the last three quarters of 1987. The smaller specific consumption of Embalse may be noted.
15.4
15. b
i
|
15.2
I 13.0
I
Though the four stations are CANDU-600's, there are some differences between them which explain some of the differences in the comparisons. Bnbalse differs from the others in the following: -
Less total thermal power MW-th of the others).
(2105 MW- ;h versus 2155
-
Less parasite absorption in the lighter weight of the adjuster detectors.
sore rod
due and
to a less
CNS9thANNUAI CONFEHENCE. 19B8 373
T A B L E
-
Using the standard condi t ions mention fid in Ref. ^ to standardize the extract ion burnups of the 4 s t a t i o n s , and thus leave aside design and ope rat ing pa ramete rs differences, we obtained the values sliown in Table 5, which compares the strategies applied in each s t a t ion
3
BURNUP ASSESSMENT PERIOD: FROM 8 7 / 0 4 / 0 1 TO 8 7 / 1 2 / 3 1
1
REACTOR
Q
I
1 I 1 FPN IN 1 PERIOD I 1 1
o i n t Lepreau
1 1 Gentilly - 2
1 1 1 1
1
1
i
i
1 1 Wolsung
1
247
I 7462.5 | 179. 1|
2 9 7
265 3
7412.5 177. 9|
16.3
68C2.5 I 164. 7|
14.3
7302.1 I 187. 2|
1 |
1 231. 5
1 1
7190.0 | 172.0)
5
Standars parameters Moderator p u r i t y 99.9 % a t Coolant p u r i t y 99.722 % a t Boron in moderator . • * . . 0.2 ppm Liquid zone level 50.0 % average Operating power 100.0 % (2155 MW-th)
I I
15. 1
-
BURNUP COMPARISON CORRECTED TO STANDAR CONDITIONS
7254.2 174. 1|
If..5
7691.6 | 184.6| 1 I 6912.5 | 16^.91
TABLE
I I
I
1 1 1 1
Embalse
„
1
I
PREDICT.| AVERAGE BURNUP TIME AVG( REFUEL. 1 ( * ) BURNUP | RATE OVEK| I MWd/t.U MWd/t.U | PERIOD MWh/KgU] BIND/FPD MWh/KgU
I
1
87 FIRST BURNUP | MWd/t.U | MWh/KgU|
I
(*) Burnup assessed from actual refuelling
rate.
QUARTER | 87 FOUTH QUARTER | EQUIVAL.I BURNUP I EQUIVAL.| REF.RATE) MWd/t.U | REF.RATE) BUN/FPD I MWh/KgUl BUND/FPEj
I TABLE
-
Point Lepreau
7324.9 175.8
Gentilly - 2
7216.6 | 173.21
4
15.4
7191.7 | 172.61
15.7
15.6
7508.3 | 180.2|
14.9
15.4
7195.8 ] 172.7|
I CANADIAN-BUILT CANDU-600 FUEL PERFORMANCE BY STATION (1987 September)
TATION
Point Lepreau
| NUMBER OF BUNDLES | | | | % DEFEC, | I IRRAD. | DISCH. | DEFEC.*| «« I
I I
I
i
I
I 30005 I 25445 I
Gentilly - 2
I
I
|
19722 |
17
15162
13
I rfolsung
I
22620 |
32
20154 I
Embalse
8
89604 |
73458 |
70
1
I
|
0.06 |
I
I
|
0.07
I
|
0.14 |
I
I
|
0.05 |
I I
I I 0.09
|
I
I
parameters
which
* V i s u a l l y confirmed or suspected ** Defective/irradiated There are differ:
also
Purity of Bmbalse) -
some
operating
(99.82%
for
The weight of the uranium in national origin element is less than foreign-origin elements.
fuel
Folicy of moderator.
moderator
operating
heavy
water
without.
374 CNS 9th ANNUAL CONFERENCE, 1988
boron
in
|
7270.0
|
175.9|
15.4
I Embalse
7362.5 | 176.7|
I
15.3 I 7587.5 | | 182.1|
I
14.5 I
I
|
I
|
I
I I Wolsung
the
From the standardized values i t may be seen that the refuelling strategy performance of the 4 stations i s comparable, with Embalse's specific consumption 's values being s l i g h t l y lower. I t should be noted that under standard conditions, the uranium weight of the fuel elements is not considered. If taken into account, i t would cause the reduction of Embalse's specific consumption due to the lesser weight per bundle of the Argentine origin fuels. On the other hand, the epidemic of Argentine fuel fai.lures mentioned above led to the refuelling of fuel channels suspected of containing failed elements with less burnup than the time average. The overal 1 effect of t h i s was additional fuel consumption, estimated at about 50 bundles.
!i.
SIGNIFICANT PARAMETERES FOR BURNUP OPTIMIZATION.
ALTERNATIVES Table 6 shows a s e t of p a r a m e t e r s and t h e i r e f f e c t on ex t r a c t ion burnup.
TABLE
-
6
FACTORS AFFECTING EXTRACTION BURNUP
CURRENT VALUE
MOD.PURITY (% at.)
COMPARIS VALUE
99.9
BUiWUP | INCREASE INCREASE (%) I MWd/t.U | | 253.54
The r e f u e l l i n g selection criteria also plays an important role in t h i s a n a l y s i s . Though the a b i l i t y of the s t r a t e g i s t to s e l e c t the channels to be refuel led i s very d i f f i c u l t to e v a l u a t e , we find t h a t not to everdimention the nurnber of candidate channels for refuel ling b r i n g s about an i n c r e a s e in average extraction burnup which is not compensated by channels which would be r e f u e l l e d with a higher burnup.
+ 3.52
COOL,PURITY (at.)
- 2.76 |
BORON IN THE MODERAT.(ppm)
5. The l a s t two f a c t o r s were included to provide a q u a n t i t a t i v e idea of t h e v a r i a t i o n s wh ich may be caused in t h e average e x t r a c t ion burnup. In particular, operating power cannot always be def ined i n t e r n a l l y , s i n c e i t depends on e v e n t u a l r e s t r i c t ions imposed by t h e n a t i o n a l g r i d .
We wish to mention t h a t the use of s l i g h t l y enriched uranium i s a good o p t i o n , in the middle term, to improve the use of uranium in CANDU r e a c t o r s . This would mean a saving in the order of US$ 4,500,000 p e r year of operation at 100% power according to e s t i m a t e s ( R e f . 6 ) . This i s an area t h a t should be s t u d i e d in the near f u t u r e . 6. CONCLUSIONS
FUEL DENSITY (kg) LIQUID 2ONES
92.8
+ 1.30 |
<*> OPERATING POWER (%)
We note
+ 4.46 |
that:
1.
Moderator p u r i f i c a t i o n brings about a s i g n i f i c a n t i n c r e a s e in burnup and may be c a r r i e d out with the s t a t i o n upgrader. An i n c r e a s e of 0.01% atomic i n c r e a s e s e x t r a c t i o n burnup in the order of 0.5%.
2.
Coolant purification does not cause s i g n i f i c a n t g a i n . I t would not be a d v i s a b l e .
3.
The use of boron in the moderator i s a d e t r i m e n t a l factor for burnup. The Embalse experience in o p e r a t i n g without boron in the moderator shows t h a t i t s use i s not j u s t i f i e d , because in almost 3 y e a r s the boron would have prevented a power decrease due to lack of r e a c t i v i t y on only on 2 o c c a s i o n . On such occasion, 12 hours a f t e r the power reduction f u l l power operation was r e s t o r e d .
4.
any
The uranium weight of the fuel i s a s i g n i f i c a n t f a c t o r . We have to d i s c u s s t h i s with the n a t i o n a l s u p p l i e r , who i s going to be Embalse 1 s s u p p l i e r , to obtain t h e value of the foreign s u p p l i e r . We have no e s t i m a t e s on the coat of t h i s enhancement.
The r e a c t o r of the Embalse Nuclear s t a t i o n reached 1146 fpd of energy generation on 0 3 / 3 1 / 8 8 . Fran the beginning, the PUMA-C code developed by CNEA was used for fuel management o p e r a t i o n s . The c r i t e r i a used for the r e f u e l l i n g s t r a t e g y were developed and adapted as i n d i c a t e d by experience, by the p e r t i n e n t station department. At p r e s e n t , we have an average e x t r a c t i o n burnup of 7800 MWd/t.U. (187.2 MWh/KgU}, which exceeds in about 8.5% t h e time average burnup (7190 MWd/t.U. 6 172.6 MWh/KgU). The s p e c i f i c fuel consumption i s 14.2 bundles/fpd. At no time have the channel and bundle power l i m i t s been exceeded. From the comparison made between t h e four CANDU 600 stations. Point Lepreau, Gentilly-2, Wolsurg and Embal s e , i t may be se en t h a t , for the same th.i imal energy g e n e r a t e d , the l a t t e r has consumed abou: 10% less than i t s s i s t e r stations. At present, this represents a difference of more than 2,000 bundles with regard to Point Lepreau and about 1,500 bundles with respect to the other two. From the standardized comparison of a set of significant parameters, i t i s seen that the Embalse consumption i s s l i g h t l y less than that of i t s s i s t e r s t a t i o n s . The most significant component of the differences of the last paragraph is due to construction factors and to different values of some operating parameters• The four stations show failure rate values.
similar
foreign-origin
fuel
CNS 9th ANNUAL CONFERENCE, 19BB 375
In the middle term, to improve the use of uranium in CANDU reactors a good way will be the use of slightly enriched uranium.
7 - REFERENCES 1 - CAI.ABRESE C. - FINK J. UPDATE OF THE REFUELLING STRATEGY AT THE C.N.E. CNEA-RE-CA-85-17, July 1985. 2 - VINEZ J. - KEIL H. - HANZER A. - KARGER J. LOAD FOLLOWING IN CENTRAL NUCLEAR EMBALSE C.N.A. 1986 CONFEKENCE 3 - MORENO C. - OE FAZ A. OPERATING EXPERIENCE WITHOUT REFUELLING FOR AN EXTENDED PERIOD AT THE C . N . E . A . A . T . N . 1937 CONFERENCE 4 - GRANT C. PUMA - NUCLEAR REACTOR OPERATION SIMULATION SYSTEM CNEA-RE-163, JUNE 1980 5 - CANDU FUEL AND REACTOR PHYSICS NEWSLETTERS CANDU OWNERS GROUP (C.O.G.) 6 - MINUTES OF MEETING WITH AECL EXPERT (A. T-L^NZER) ON 87/10/02. 7 -
GENERAL REFERENCES - DM 1 8 - 0 1 1 0 0 - 1 PHYSICS DESIGN MANUAL - C . N . E . MONTHLY OPERATION REPORTS
376
CNS 9th ANNUAL CONFERENCE, 1988
Session 11: Fuel Channels: Current Position and Improvements
Chairman: G.J. Field, Ontario Hydro
CNS 9th ANNUAL CONFERENCE. 1988
hi
377
THE FABRICATION OF HYDROGEN SINKS FOR THE PRESSURE TUBES OF DARLINGTON UNIT U R. DEGREGORIO (1) I. GRANT (I) I. INGLIS (1) E.V. MURPHY (1) E. PRICE (1) M. NATESAN (2) W.B. STEWART (2) (1)
Atomic Energy of Canada Limited CANDU-Operations Sheridan Park Research Community Mississauga, Ontario L5K 1B2
(2)
Ontario Hydro 700 University Avenue Toronto, Ontario M5G 1X6
ABSTRACT
Research activities related to the use of yttrium metal hydrogen sinks (hereafter called yttrium sinks) as hydrogen getters in pressure tubes. This has involved computer model ling of the yttrium sink effect, diffusi on measurements, gettering experiments and corrosion testing. Both full size rolled joint specimen:; and smaller sections were used in the experimental part of the research program. Research activities were carried out by AECL Whiteshel1 Nuclear Research Establishment.
A development program has been ongoing for thvi past eighteen months with the prime objective of identifying a practical method of installing hydrogen sinks in Darlington pressure tubes. This paper provides a description of the overall program from the basic concept definition through to prototype development, testing and the selection of a fabrication method.
1.0
INTRODUCTION
—
Development of prototype fabrication methods, qualification of selected methods and detailed design of the end fitting and yttrium sinks. In this phase of the program various fabr-" cation methods were evaluated and two were chosen for qualification to fully proven techniques. Two detailed designs were developed, one for each fabrication method. AECL-CANDU Operations were responsible for this part of the program.
During reactor operation the CANDU pressure tube picks up hydrogen from the primary heat transport fluid. Analyses of removed pressure tubes showed that the hydrogen pick-up rate was greater in the rolled joint area than that occurring by corrosion in the bulk of the pressure tube. Thij additional pickup is attributed to galvanic coupling between the stainless steel end fitting and the pressure tube which releases hydrogen on the end fitting which then diffuses into the pressure tube. This hydrogen ingress wil 1 eventually result in the terminal solid solubility (TSS) of hydrogen in the Zr-2.5% Nb alloy being reached in the rolled joint area. As a result, the material is eventually embrittled by the formation of zirconium hydride. This is an essential condition for delayed hydride cracking (DHC) and should be avoided to eliminate the possibility of cracking of the pressure tube at service temperatures. One method of preventing this buildup of hydrogen is to use a chemical getter, such as yttrium, to remove hydrogen at a rate sufficient to retain hydrogen in solid solution in the Zr-2.5% Nb.
2.0
The concept of using yttrium as a hydrogen getter in this application originated at Whiteshell Nuclear Research Establishment. Therefore, in January 1987 Ontario Hydro authorized a research and development program at AECL aimed at producing a practical design of chemical getter which can be installed at the end of the pressure tube to pull hydrogen away from the highest stressed region of the rolictd joint. Overall project management responsibility rested with Ontario Hydro while AECL has been responsible for the following aspects of the development program:
Yttrium can getter hydrogen from the zirconium because it has a higher chemical affinity for hydrogen than does zirconium. Yttrium has the highest chemical affinity for hydrogen of all the metallic elements. By placing a ring of yttrium at the end of the pressure tube the incoming hydrogen will be gettered by the yttrium, thus preventing the formation of zirconium hydride within tho Zr-2.5% Nb pressure tube. If a sufficient quantity of yttrium is encapsulated within the end of the* pressure tube, tha hydrogen levels at the highest stressed zone of the rolled joint (i.e. the burn is li mark) can be
The program target was to develop a practical method of installing yttrium sinks at the end of the pressure tubes for Darlington Unit U. This paper deals with the activities associated with AECL-CANDU Operations' portion of the development program. Following is a description of the overall concept, the design requirements, the options evaluated, a summary of various test results and the recommended fabrication method.
HOW A GETTER WORKS
CNS 9th ANNUAL CONFERENCE. 10BB 379
maintained below terminal solid solubility at operating temperatures for 30 years of reactor life, thus eliminating concerns with D?1C at the rolled joint. The effectiveness of an yttrium insert to getter hydrogen from the rolled joint will depend upon a number of factors, such as its position, size and fabrication method and also, deuterium diffusion rates and operating temperatures. Figure 1 illustrates the effect of an yttrium insert on keeping the hydrogen levels below TSS in the rolled joint region of the pressure tube.
Low Probability of Encapsulation Failure Yttrium exposed to primary heat transport system (PUTS) fluid corrodes rapidly and gettering will tlicn stop. Furthermore, failure of the encapsulation wi]) cause the yttrium sink to increase in volume to the point where the passage of fuel may in some cases be impeded. •
Minor Impact on End Fitting_^esj_gn_
Installation of the yttrium sink must not necessitate a major re-design of the end fittings or changes to the rolled joint design which would in turn require a full re-qualification program. No Impact on Fuel Handling System or Internal Fuel Channel Components The Fuel Handling System design and the design of fuel channel internal components must remain unchanged as a result of installing pressure tubes which incorporate yttrium sinks.
Although non-bonded designs were explored, tests to 3ate have indicated that metallurgical bonding between the yttrium and the zirconium encapsulation provides superior diffusion of hydrogen from the pressure tube to the sink. Minor Impact on Construction Schedule The installation of yttrium sinks must not result in an unacceptable delay to Che Darlington Unit 4 construction schedule.
4.0
GENERAL CONFIGURATIONS
Two basic configurations were considered development of fabrication methods:
for the
1. Non-integral: wherein the yttrium sink is fabricated as a separate ring and then welded to the pressure tube, and
FIGURE 1 HYDROGEN CONCENTRATIONS IN R/J AFTER IS YEARS WITH « WITHOUT SINKS
3.0
DESIGN REQUIREMENTS
2. I»t«gxaV. wVrf^ftb-j a g^oo^e- is rc,&ct\irad in each end of the pressure tube and the yttrium is then inserted in this gfoove and sealed by welding. The yttrium sink becomes an integral part of the pressure tube without a weld zone between the sink and the rolled joint. Figure 2 identifies two methods yttrium sinks based on the above-
The criteria that any design of yttrium sink must meet in order to be acceptable for installation in Darlington Unit U ate listed below.
5.0
PROTOTYPE DEVELOPMENT
•
5.1
Conceptual Approach
Minimal Risk of Creating A New Problem
It is essential that incorporating yttrium sinks does not create new problems. For example, the installation of yttrium sinks must in no way impair the leak tightness of the rolled joint. •
of
installing
Based on the design requirements of Section 3.0 three main steps had to be achieved in fabricating sinks, namely: -
Encapsulation water.
- to isolate yttrium
from the PHTS
Reasonable Probability of Success
There must be reasonable assurance that the yttrium will keep the hydrogen levels inboard of the burnish mark belou TSS lor 30 years of operation at 80% capacity.
380 CNS 9th ANNUAL CONFERENCE, 1988
Bonding - to provide a diffusion path pressure tube to the yttrium-
Attachment - to attach the yttrium the pressure tube. This applies non-integral design.
from
the
sink ring to only to the
TABLE 1 - SUMMARY OF FABRICATION METHODS
A. INTEGRAL END GROOVE
Koute
EBW
VF
EBW
- erosion of cladding material due to eutectic reaction - schedule
VF
LBW
- d i s a p p o i n t i n g LBW r e s u l t s from i n i t i a l we Id t r i a l s - n o d o m e s t i c LBW faci1i ties available for production
Zr-2 5«t N6 P T
B. NON-INTEGRAL Zr ENCAPSULATED Y RING
EBW
HIP
KBW N / A
EBW (in billet)
CEX
EBW
- difficult weld preparation - difficult to control location of 'Y' in cladding - three welds required to seal and attach
GTAW
IH
N/A
- schedule
GTAW
EBW
N/A
- erosion of cladding material - *Y' contamination of seal weld -vertical E B W facility
N/A
- inadequate seal - schedule
EXW
N/A
- schedule
MMF
N/A
- schedule
FIGURE 2 ALTERNATIVE CONFIGURATIONS FOR ATTACHING YTTRIUM RINGS TO THE ENDS OF PRESSURE T " 3 E S
5.2
Development of Fabrication Routes
Within the limits imposed by the design requirements, geometry and materials, there were still several processes applicable for encapsulation, bond ing or attachment. Processes that were considered are listed in Table 1. Combined with the two basic configurations, the variety of processing techniques gave many potential fabrication routes for sinks. The routes judged most commercially feasible were explored by fabricating prototypes and subjecting them to simulated rolled joint fabricat ion, corrosion testing, non-destructive and metallographic examinations. Because of the pressing time schedule, emphasis in these tests was on processes which would most 1ikrly be usnd for production. Weaker prospects were simply e1iminated.
RSW
WONBONDED
Many routes were in fact eliminated at this stage. Suppliers were in some cases unable to make satisfactory prototypes within the time available. Explosive welding and magnetic metal forming for instance were quickly ruled out for immediate applications. In other c a s e s , prototypes were fabricated successfully, but features were encountered on evaluation which led to their being eliminated. Eutec tic melting ar a bonding process for example caused irregular thinning of the encapsulation wall. Encapsulation wal 1 thinning resulted in failure of a prototype of this route in autoclave testing.
5.3 Corrosion Testing and Heat Treatment Development It was quickly recognized also that the thermal cycles associated with bonding and welding would affect the corrosion resistance of the Zr-2.5% Nb pressure tube material. Testing the effects of various manufacturing techniques on the corrosion resistance of zirconium alloys, and developing remedial measures, occupied a large part of the prototypo effort.
Main Reasons for Discontinuing Devclopmnnt
Encapsulation Bonding Attach.
N/A N/A
Electron Beam Welding Laser Beam Welding Vacuum Furnace Hot Isostatic Pressing Co-Extrusion Gas Tungsten Arc Welding Induction Heating Resistance Seam Welding Explosive Welding Magnetic Mocal Forming Notes:
1) Routes \~h are non-integral type; Routes 5-10 and integral type 2) Routes 3 and 10 chosen for further development 3) Where "schedule" is shown as a reason for discontinuing development, this means that this method could not be developed in time for implementation in Darlington Unit A.
CNS 9th ANNUAL CONFERENCE, 1988 381
simulates the most severe pressure and temperature cycles that a fuel channel will experience during 30 years of operation.
Corrosion testing was carried out by preparing specimens which accurately simulated potential fabrication routes. For example, rings of zirconium alloy were subjected to typical bonding thermal cycles, then were electron-beam welded to virgin pressure tube, to simulate the non-integral designs. Other pieces of pressure tube were given bonding treatments and simulated gas tungsten arc seal welds, t'- represent various integral designs.
Three double ended rolled j<- int assemblies were fabricated for each of the two designs chosen for complete qualification (six rolled joints per design). The joints were rolled using the total range of interference and clearanee applicable to Darlington Unit 4. When the eye A ic test program is concluded each joint will have experienced a total of 6770 cycles. Each cycle consists of a concurrent AT of 236°C and AP of 11.8 MPa. Approximately 2000 cycles have been completed to date with no sign of rolled joint leakage.
Aging heat treatments between 500°C and 600°C are reported in the literature as beneficial for restoring corrosion properties of welded Zr-2.5% Nb alloys. Coupons cut from the specimens were heat treated at a variety of times and temperatures in this range. These coupons have been exposed to PHTS water chemistry in a static autoclave to test their corrosion resistance and so establish the optimum heat treatment. Some of these samples now exceed 3,000 hours exposure. In addition, actual prototypes of th^ most promising routes have been exposed in the autoclave for similar periods of time.
PRESSURE TUBE
ROLLED JOINT GROOVES
PRESSURE TUBS IN END FITTING AFTER ROLLING
EB ATTACHMENT WELD \
Equipment to perform heat treatment in accordance with AECL's strict temperature uniformity and cleanliness requirements was developed for the prototypes by a local supplier. A precisely defined temperature zone was necessary to minimize the length of pressure tube affected by temperatures above that at which autoclave stress relief is carried out. High standards of cleanliness and inert atmospheric purity are necessary to avoid contamination of the reactive zirconium alloys at temperature.
£6 SEAL WELOS
DETAILS OF YTTRIUM SINK AREA NOTE ARROWS ^ ^ INDICATE AREAS OF DESIGN WHICH HAVE BEEN CHANGED TO \NCORPOHATE YTTRIUM SINKS
FIGURE 3 ROLLED JOINT WITH YTTRIUM SINKS ATTACHED 5.4
Preferred Options
Upon completion of prototype development, the bonded non-integral design was chosen for further qualification testing. Another configuration, the non-bonded integral design, was subjected to qualifying tests but has since been eliminated on the basis of inferior gettering.
6.0
ROLLED JOINT QUALIFICATION PROGRAM
6.1
Requirements
6.1.2 Leak and Pull Testing. The purpose of the leak and pul 1 testing program is to demonstrate that the addition of yttrium sinks does not cause any deterioration in the leak resi stance or axial strength of the rolled joint. The main elements of the test program are; helium leak testing
The installation of yttrium sinks in pressure tubes will require minor changes to the design of the pressure tube/end fitting rolled joint. Figure 3 identifies various aspects of the design which have changed. A number of tests were carried out to ensure that installation of yttrium sinks has no deleterious effects on the integrity of the rolled joint. Following is a summary of these tests and the results to date.
6.1.1 Cyclic Testing. The purpose of cyclic testing is to demonstrate that yttrium sinks will not affect the capacity of the rolled joint to resist repeated tempo rat lire and pressure cycles. The test program
382 CNS 9th ANNUAL CONFERENCE, 1986
thermal cycling (5 cycles), helium leak testing -
followed
by
repeat
hot pressurized pullout testing fluorescent sink area
penetrant
inspection
of
the
yttrium
profiling of the rolled joint A total of 24 rolled joints were used for this part of the development program, of which 12 were pulled to failure. This phase of the development program complete and all results arc acceptable.
is now
7.0
RECOMMENDED FABRICATION ROUTE
The recommended method of installing yttrium sinks on Darlington Unit 4 is the non-integral HIP bonded design using electron beam welding for both and attachment Figure 4 shows a cross-section of an early prototype of this route.
needed to weld full-length pressure tubes. After heat treatment of the welded ends and final machining, the pressure tubes with sinks attached are autoclaved and inspected. Manufacture by this basic route has been successfully demonstrated on over 40 prototype and qualification specimens. Several non-destructive examinations have been specified for full-scale manufacture. These include ultrasonic checking of bonding, measurement of encapsulation thickness using special ultrasonic techniques developed at AECL-CRNL, ultrasonic examination of the electron beam attachment welds to ASME Code requirements, and visual examination after steam autoclaving. After passing these examinations, the sinks are expected to exhibit excellent reliability in-reactor.
8.0
FIGURE 4 PHOTOMICROGRAPH OF PROTOTYPE YTTHIUM SINK (SCALE 4 x >
The essential features of manufacture are as follows. Electron beam (EB) welding is used to seal the assembly in view of the small weld size, also because the encapsulation is evacuated during EB welding. Evacuation is necessary for the hot isostatic pressing (HIP) bonding process. The HIP process, which is performed commercially, subjects the assembly to intense pressure at elevated temperatures, creating diffusion bonds at the interfaces with little dimensional change. Electron beam welding was selected for attaching the bonded sink also, because the small weld zone minimizes the axial space required. Special equipment will be
CONCLUSIONS
As a result of the yttrium sink development program, an acceptable fabrication method has been identified for installing yttrium sinks in Darlington Unit 4 pressure tubes. Based on the test results to date, yttrium sinks will be effective getters and wi! -->t affect the integrity of the rolled joint. Fui.nermore, all of the design requirements identified in Section 3.0 will be met with the possible exception of the schedule for installation. This aspect is still unresolved since it depends on bids which are due shortly that cover the actual modification of Unit 4 pressure tubes.
9.0
ACKNOWLEDGEMENTS
The authors wish to acknowledge the efforts of C D . Cann. The basic concept originated at WNRE. Mr. Cann vas of great assistance in the preparation of this paper.
CNS 9th ANNUAL CONFERENCE, 1988 383
HYDROGEN INGRESS MECHANISMS IN ZR-2.5 wt% Nb PRESSURE TUBES - AN OVERVIEW OF SOME RECENT PROGRESS P.C. LICHTENBERGER Ontario Hydro, Toronto, Ontario, Canada
ABSTRACT
ROLLED JOINT PROGRAM
This paper provides simplified illustrations of some of the Important physical processes involved, and selected examples of progress toward identifying and understanding hydrogen ingress mechanisms associated with Zr-2.5 wt% Nb (Zr-Nb) alloy pressure tubes (PTs) operating in CANDU PHW nuclear power reactors. (1) The material is drawn mostly from the current CANDU Owners Group (COG), Fuel Channel R&D programs, which are being carried out principally at the research laboratories of Ontario Hydro (OH) and Atomic Energy of Canada Limited (AECL). A brief status of the current h*' '^theses for explaining the increased hydrogen content observed in the recently removed surveillance PT from Pickering Unit 3 Channel L09 (P3LO9), is also given.
,
J
UPDATE MATHEMATICAL MODEL
DETERMINE WGREES RATES | FROM SURVEILLANCE
1
( ASSESS SUPERSATURATE, —W DIFFUSION AND J PRECIPITATION PHENOMENA
-I 1
J *J
STUDY BARBERS TO HYDROGEN INGRESS
DETERMINE ROUTES OFOTGRESS
1 |
1 j
|
h
RECOMMEND BEST-ROLLED JOINT
|
T J " "^j
STUDY 2, AND Ti ALLOVENDFITWGS
_ J ^ |
STUDYCHEWCAL INHIBITORS
v_
| |
J H
RE0OMMEN0 OPTJMUU METHOD OF REDUCWG INGRESS
OETWHI«SOURCES OFWCRESS
—J
t 1 |
]
FULL-LENGTH PRESSURE TUBE PROGRAM
INTRODUCTION R&D programs i n s u p p o r t of hydrogen* i n g r e s s mechanism s t u d i e s , can be divided into two sub-program areas; a) hydrogen ingress at rolled joints (RJs) and b) hydrogen ingress along the full length of the PT. Until 1983, the ongoing work program in these areas was small, fragmented, and l i t t l e effort on in-flux effects was accommodated. Since then, the scope has been expanded to include a full program of in and out of flux work. Figure 1 charts the key program elements combined in an iterative work process. Zirconium alloys have two important features in relation to their hydrogen ingress behaviour. On the one hand, they easily form thin protective p a s s i v e oxide films in aqueous or gaseous environments containing O2. These oxides (ZrO2) are usually extremely good barriers (ie, low permeation films) to hydrogen ingress. The hydrogen permeation r a t e r barriers can be simply described, by:
through
-Q/RT atoms
such
(I)
cm2s where [H"]
-
concentration of hydrogen atoms a t the surface of the b a r r i e r the e f f e c t i v e barrier
e"Q/ RT
-
thickness
of
the
a factor controlling diffusion through the barrier according to an a c t i v a t i o n energy Q and temperature T (R-the gas constant).
* hydrogen Includes Che Isotopes protiuo^deuteriun, and critiura 3M
CNS 9th ANNUAL CONFERENCE, 19B8
d
EVALUATE FACTORS OTNTROLLING CORROSION ANDHYDRIQiNG
r-
DEVELOPAUMFIED MODEL FOR LONG-TERM CORROSION AND HYDHID1NG OF PRESSURE TUBES
ASSESS METHODS FOR REDUCING CORROSION ANOWWWWG
n
-is FIGURE 1 :
R&D PROGRAM ELEMENTS/PROCESS
On the other hand, Zr and its alloys have a low solubility limit for hydrogen given by the following relation: (2) C - 1.2 x 10 5 exp (- 3 5 9 0 ° ) ppm RT
(II)
Hydrogen concentrations (C) above this terminal solid s o l u b i l i t y (TSS) limit, form b r i t t l e Zr hydrides which can Impare PT material Integrity. The rate controlling step for hydrogen permeation, in (I) is associated with the presence of a "barrier oxide" of thickness x. The problem, then, becomes largely one of understanding the nature of and mechanisms for forming protective oxides versus the nature of and mechanisms for forming or evolving oxides that are non protective, damaged, or of reduced protectlvity. These studies are complicated by the need to examine synerglstic processes involving hydrogen diffusion as a function of oxide morphology, interface effects, radiation effects, electrochemistry, e t c . This makes accelerated testing difficult and necessitates a mechanistic approach to the detailed study of the problem.
uo 3550 OPERATING DAYS (1934)
I
60 -
J
60
•10
V-—(
" \
10
20
t 20
1 N
T
FIGURE 2 :
T
T
Ti alloy small diameter EF hubs are being used to study tiie effects of galvanic coupling on hydrogen ingress at RJs. Exposures are being carried out with double ended PT/EF assemblies in an out-reactor loop under PKTS chemistry conditions. The objective is to compare the hydrogen build-up at the tube end rolled into a Ti alloy EF, with the tube end rolled Into a standard 403 SS EF. Galvanic effects between the Ti alloy EF and the FT, are expected to be low because of the s i m i l a r i t y in electrochemical potential of these materials, thus reducing the source of cathodic hydrogen for ingress at the RJ.
1-" 1
30
DISTANCE FROM INLET (cm)
effect of hydrogen concentrations over the range 0 to 30 era /L. Reactor operating guidelines recommend a range of dissolved hydrogen in the coolant of 3 to 10 cm5/L.
DISTANCE FROM OUTLET (cm)
DEUTERIUM BUILD-UP IN PICKERING UNIT ROLLED JOINTS
These investigations are being supplemented using permeation c e l l experiments to quantify more precisely the source term under a variety of PHTS conditions.
Routes The follovsi^g. s e c t i o n s d e s c r i b e selected highlights and key project objectives of ongoing hydrogen ingress R&D studies in RJs and along the PT length. This work i l l u s t r a t e s the processes occurring at and near the PT surface and caused by environment-surface oxide interactions as well as the influences of the bulk alloy on the metal-oxide interface.
The three possible hydrogen ingress routes, leading to build-up at RJs are the EF route (diffusion through ths EF, followed by direct absorption by the PT where the normally protective OX i de TiaR h p p n i^amaci^-A i n t"Vif» R.T p r n n v p r p n i n n - a
HYDROGEN INGRESS AT ROLLED JOINTS The presence of hydrogen in RJs (Figure 2) , In combination with high tensile stresses (caused by an improper rolling procedure), has resulted in PT leaks by the Delayed Hydride Cracking (DHC) mechanism at Pickering 'A' and Bruce 'A' reactors. (3) in order to predict the probability of future leaks at RJs, that could occur at operating temperatures, there is a need to identify the mechanisms and to control this hydrogen build-up through-. - the establishment of an understanding hydrogen sources and ingress routes,
of the ¥£':-:':•':•'&:•:•':•':$$:-'£
ENDFITTING
;•;%•;':'#:-:-;$£%%£
- the development and updating of a quantitative predictive model, and - the exploration of practical methods and concepts for reducing hydrogen ingress. ^ — Figure 3 schematically depicts RJ hydrogen ingress sources and routes.
Sources The two possible sources of hydrogen ingress at RJs are dissolved hydrogen in the coolant, (added to suppress radiolytic decomposition), and galvanic coupling between the endfitting (EF) and PT. Tests are in progress to determine the magnitude of the effects of dissolved hydrogen. Small diameter joints using standard 403 stainless steel (SS) EF hubs, are being exposed under primary heat transport system (PHTS) conditions in an out-reactor loop. Results, so far, indicate no discernible
0 (7)
0
0
- D2 O + DISSOLVED D2
HOUTFS
SOIIRCFS
1. D j UPTAKE FROM CORROSION
4. 01SSOLVEO C^ IN THE COOLANT
2. O 2 UPTAKE VIA THE OUTBOARD CREVICE 3. D2 UPTAKE VIA THE END FITTING
5. GALVANIC COUPLING BETWEEN STEEL END FITTING AND Zr-Nb PRESSURE TUBE
FIGURE 3 :
SCHEMATIC ILLUSTRATION OF PRINCIPAL HYDROGEN INGRESS SOURCES AND ROUTES AT ROLLED JOINTS
CNS 9th ANNUAL CONFERENCE. 1988 385
Double-ended small diameter RJ assemblies are being tested in an out - reactor loop under PHTS chemistry conditions where the bore of the EF hub on one end of each assembly has been Cr plated. Chromium has a low permeability to hydrogen. If the major source of hydrogen ingress at RJs is di £fusion through the steel endfitting, then chromium plating should reduce ingress at the RJ . Sealed crevice tests to investigate the influence of the crevice en hydrogen ingress are being planned. Corrosion contributes a smal1 uni form hydrogen contribution over the length of the RJ region.
In addition to experiments to investigate the use of Cr plating or Ti alloy EFs to reduce hydrogen ingress at RJs by controlling the route and source respectively for the hydrogen, other protective measures are being evaluated. Chemical inhibitors to promote hydrogen recombination reactions or poison hydrogen absorption on Che steel endfitting surfaces are being considered. Tests using AsO3 and boric acid are being planned. Yttrium, which has a high gettering a b i l i t y for hydrogen, compared with Zr alloys, is also being evaluated as an "absorption sink" by physical incorporation into the PT.
Figure 4 (a) i l l u s t r a t e s the range of hydrogen profiles for all PTs removed and analyzed; Figure 4(b) shows the maximum hydrogen measured as a function of reactor operating time for all Zr-Nb PTs removed so far, and those recently micro-sampled.**
ZIRCALOY-2[Pt.2) HIGH POWER CHANNELS^ (3600-3710 EFPD)
O 200 | 180 Z 160 UJ
Z 140
200 -400 DISTANCE FROM INLET (cm)
FIGURE 4 < a ) : A mathematical model and computer code have been developed to analytically predict hydrogen build-up and redistribution at RJs. The model uses the finite difference method to solve the governing diffusion equations with source terms representing hydrogen ingress via the groove region of the RJ and along the PT via the corrosion reaction. The model handles hydride precipitation as well as spatial and time variations in ingress rates.
ALLOY; Zr-2.5Nb
Hydrogen b u i l d - u p along t h e PT l e n g t h , with a peak toward t h e o u t l e t end, o c c u r r e d i n t h e Z i r c a l o y - 2 (Zr-2) tubes, of P i c k e r i n g 1 and 2 r e a c t o r s , c o n t r i b u t i n g t o t h e f a i l u r e i n one tube (P2G16). (4)
In order to predict future hydrogen build-up, Zr-Nb PTs, there i s a need to identify mechanisms and to control uptake through:
in the
9 '
• Removed PT's
I
In-situ Micro-sampled PT's
A data base of 18 RJs from Pickering 'A' and Bruce 'A' reactors has been analyzed. There is considerable tube to tube variability. Best fits are obtained by adjusting parameters to include: a declining ingress rate, an extended ingress region, and/or hydrogen supersaturation (hysteresis in TSS). Predictions of end of life hydrogen near RJs in Pickering and Bruce reactors can be made. A declining ingress rate suggests the possibility of the repair of oxide at the groove region of the RJ. FIGURE 4 ( b ) : This would support the EF route for hydrogen ingress.
HYDROGEN INGRESS ALONG THE PRESSURE TUBE
DEUTERIUM AXIAL CONCENTRATION PROFILES FOR Zr-2 AND ZR-Nb REMOVED PTs
TIME (Ful Power Years)
MAXIMUM DEUTERIUM CONCENTRATION IN Zr-Nb PTs (AT 5.5 TO 6.0 m FROM INLET)
The Ingress of hydrogen along Zr-Nb PTs i s normally observed to be low during the early life of tubes (see Figure 4(b) - the anomalously high P3-L09 result and a few other higher than normal micro-sample results are discussed later in the text). However, as the tubes age there are three generic effects that are being actively studied, in order to assess their potential for leading to future changes in rates of ingress: (5)
-
the establishment of a quantitative understanding of the long term hydrogen ingress behaviour,
-
the development of a predictive model to end of life conditions, and
forecast
Coolant Chemistry Effects: The interaction between the coolant and the oxide might result in processes such as accelerated corrosion (thereby increasing the production oT corrosion hydrogen - this may be
-
the s p e c i f i c a t i o n of p r a c t i c a l methods and procedures for the reduction and control of hydrogen ingress.
**Small (100-300 mg) samples scraped in-situ from the inside of the PT for subsequent hydrogen analysis.
386
CNS 9th ANNUAL CONFERENCE, 1988
more probable as the oxides increase in thickness), or a reduction in the effective thickness or protectiveness of the barrier oxide (possibly by local anodic dissolution and/or oxide morphr^ogy changes - again this might be enhanced by thick oxide effects such as. impurity concentration or stress relaxation processes). in equation (I) these processes could bo modelled by increasing [WJ and decreasing x.
Bulk Alloy Effects: Kicrostructural evolution, as the PTs age, can be expected to result in alloy phase changes and the subsequent formation of precipitates such as ZrjFe. If these precipitates intersect the metal-oxide interface, a process that might be enhanced as corrosion on the water side proceeds, they may alter the effective barrier oxide protectiveness by introducing lower permeability gaps or windows inco this normally protective barrier oxide. In equation (I) this would be modelled by local effective reductions in x. Annulus Gas Effects: The normally protective 2rO2 that is formed by autoclaving PTs prior to installation in p o w e r reactors is n o t thermodynamically stable. The oxide is continually dissolving by diffusion into the metal lattice because the stable state of oxygen in Zr is that of oxygen dissolved in the Zr lattice. Since diffusion rates of oxygen ions are highest at crystal defects such as g-ain boundaries, it is at these local sites that oxide degradation is expected to occur first. It Is therefore, necessary to continually repair this oxide by exposure to an environment containing a minimum critical oxygen concentration. This effect is again modelled in equation (I) by alteration of the thickness of barrier oxide x. Figure 5 schematically shows the above three processes that could enhance hydrogen ingress in the PT. Highlights of some of the work in progress to quantify their specific nature are now briefly described.
ACCELEHATED HYDBOGEN INGRESS
Coolant Chemistry
Effects
Carrier bundles*** are being used to expose test coupons of PT materials to power reactor '.'oolant conditions. Coupon variables are oxide thickness (by laboratory prefilming), surface preparation, alloy type, and residual alloy Impurities. Reactor variables are, boiling v s non boiling channels, water chemistry (Pickering vs Bruce, vs Pt. Lepreau), and channel power (high flux vs low flux). Preliminary results after exposures for 14 months in a Pickering reactor are being analyzed. These first results will provide an early indication of the potential for increasing corrosion and hydrogen ingress rates with increasing oxide thickness for the Zr-Nb PTs - an effect which is believed to have contributed to u n a c c e p t a b l y high hydrogen concentrations in Zr-2 PTs. Additional exposures are underway at Bruce and Pt. Lepreau. Comprehensive studies are underway to understand and model radiation chemistry processes occurring in the PHTS coolant, and with porous/cracked oxides, so that coolant chemistry can be optimized to minimize the rate of hydrogen ingress. Data that will assist in this optimization will be obtained from sample lines that will soon be added to several channels in a Bruce 'B' reactor. I n i t i a l experiments are being carried out in in-reactor loop tests to determine the c r i t i c a l dissolved hydrogen concentration necessary to surpress radiolytic oxygen production, to determine the influence of LiOH concentration (used to c o n t r o l coolant pH) on r a d i o l y s i s reactions, and to determine the effects of boiling on the oxidizing character of the coolant. The more oxidizing the coolant, the higher the rate of corrosion - a possible precursor to eventual higher hydrogen ingress rates in Zr-Nb PTs. Extensions of this work are aimed at detailed quantification of reaction mechanisms and rates. In addition, specific yrograms are in place to study the effects In anc1 out of flux of H2/O2, temperature and fluence on corrosion and hydrogen ingress. Studies also include microstructural and microchemical c h a r a c t e r i s a t i o n of oxides and metal-oxide Interfaces, and fundamental studies of the interaction of irradiation with ZrO, and Zr alloy metal surfaces.
oxide dissolution vs formation AGS EFFECTS -
Bulk Alloy Effects
COOLANT EFFECTS Water Chemistry
FICURE 5 :
accelerated corrosk>n
In order to Investigate effects of agir.g of the bulk alloy, analytical electron microscopy studies are in progress. These are being applied to samples of PTs removed from power reactors after up to -12 full power years. Unirradiated samples containing a range of Impurity types and contents are also being artificially aged by thermal treatment to simulate a condition of PTs in late l i f e , and subsequently exposed to PHTS chemistry and temperature conditions. Preliminary results to date indicate that Fe and Nb are redistributed and precipitated by such aging treatments. Ho critical effects of aging on corrosion or hydrogen Ingress in Zr-Nb PT materials have yet been deduced from these ongoing experiments.
ACCELERATED HYDROGEN INGRESS THROUGH ZrO 2 COVERED PT SURFACES Standard fuel bundles containing one empty fuel element into which are loaded multiple test coupons. CNS 9th ANNUAL CONFERENCE, 1988 387
Annulus Gas System (ACS) Effects R&D programs are In place to carry out testing as a function of t e m p e r a t u r e , gas composition, operating stress, effects of irradiation, and oxide structure/thickness, in order to q u a n t i t a t i v e l y descri'be t"he process oj oxide 'orea'Kdown vs maintenance of a protective oxide barrier. Current recommendations #re to maintain nominal concentrations of 0.1 to 2 volume % O2 in the annulus gas recirculating systems for a l l CANDU reactors. (6)
200
As with the RJ sub-program, the ultimate objective is to provide a detailed analytical predictive model for hydrogen ingress along the length of the pressure tube. The framework for such a model has been developed for the corrosion of Zr-2. Over the next several years the scope of planned R&D (see Figure 1) should enable a similar model to be formulated for Zr-Nb. In the interim, and u n t i l additional mechanistic trends are identified, the modelling approaches will be extended by adding to and re - organizing a comprehensive data base and by developing curve Sitting and s t a t i s t i c a l methods "to extrapolate current trends and deduce key functional relationships. (7)
P3LO9 HYPOTHESES As a r e s u l t of t h e high d e u t e r i u m c o n t e n t found toward t h e o u t l e t of P3LO9, micro-samples were o b t a i n e d from a t o t a l of 30 PTs from P i c k e r i n g U n i t s 3 and 4 (P3 and P4) i n o r d e r t o h e l p e s t a b l i s h t h e e x t e n t of high deucerium c o n c e n t r a t i o n i n t h e PTs of these r e a c t o r s . Tfte f o l l o w i n g summarizes t h e preliminary results of these analyses. Investigations a r e c o n t i n u i n g i n a l l t h e program a r e a s d e s c r i b e d below i n order to e s t a b l i s h t h e p r e c i s e reasons for t h i s high deuterium observation.
Rolled Joint
Effects
Hydrogen i n g r e s s a t tbe toVletl juiTits is consistent with that observed in tubes previously removed from Pickering. Current evidence does not support a migration of hydrogen from the RJ region as a cause of the unusually high hydrogen concentration at the 5-9 m location.
Coolant Effects If hydrogen ingress from the coolant Is the cause, then we would expect accelerated aqueous corrosion and or a significant change in the fraction of corrosion hydrogen picKed up. This would require a correlation of measured hydrogen in the PT with several of: fuel channel power, boiling, thick oxides, evidence of o*Ide spalling or concentrated chemicals (such as LiOH) in the oxide. To date no substantial evidence or correlation has been found for thick oxides (Figure 6) extensive spalling of oxide, Li concentration, or boiling. Therefore, coolant «l£ftcts do M t a?jeat to he causing acceleration in hydrogen ingress rates.
38B
CNS 9lh ANNUAL CONFERENCE. 1988
300
400
500
DISTANCE FROM INLET END (cm)
FIGURE 6:
Bulk A l l o y
COMPARISON OF DEUTERIUM CONCENTRATION AND OXIDE THICKNESS PROFILES FOR P3LO9
Effects
If an alloy aging effect is the cause, then we would expect microstructural and/or microchemicai differences to exist and to correlate with the hydrogen profile. Although some evidence has been found for precipitation and redistribution of Fe and Nb, there is to date no substantial evidence that these features are any different than those observed in other removed tubes with low hydrogen pickup. Therefore, bulk alloy effects do not appear to be causing acceleration in hydrogen ingress rates.
Annulus Gas System Effects If ingress from the annulus gas side is the cause, then we might anticipate evidence of degradation of the normally protective oucside (ie, annulus side) oxide, and dry low oxygen annulus gas conditions.
Observations to date are: (a>
Ihe outside, ws.ii3* af>^*a*s degraded as iv>t«i:e.d by anodic polarization t e s t s . This technique (8) provides a relative measure of barrier oxide effectiveness (from i t s resistivity) which is interpreted by i n f e r e n c e to characterize the oxide on the outside (annulus gas side) of the PT, as either protective (high r e s i s t i v i t y ) or non protective (low r e s i s t i v i t y ie, revealing "near metallic" behaviour). Figure 7 compares the extent of outside oxide degradation ( i e , fraction of measured sample spots testing "near metallic") with measured hydrogen near the observed hydrogen peak location In P3L09. The apparent correlation here suggests that hydrogen entry into the pressure tubs via local degradation of the oxide ( i e , x tending to 0 in equation II) could be a contributor to the high hydrogen measured. Similar test results on other Zr-Nb pressure tubes removed from Pickering reactors and the NPD reactor are gl*!
oxides on other PTs removed Pickering and Bruce reactors. (b)
from
both C RANGE OF MOST P3 & P4 MICRO SAMPLES
The observation, in P3 and P4 reactors, of typical measured annulus gas dew points, often less than -4 0 ° C (implying very dry conditions), and measured [C^] frequently < detectable**** suggests a possible low oxidizing environment in the gas annulus of these reactors.
(c)
Figure 8 indicates that the hydrogen analyzes from most mien., samples from PTs (25 of 30) fall within a narrow band (shaded area). However, data from four P4 tubes and one P3 tube indicate Somewhat higher results. Some of these data show a temperature profile that is similar to the profile for P3LO9 (also
• FRACTION METALLIC • DEUTERIUM
Z68
=
FIGURE 8 :
276 AVERAGE TEMPERATURE ALONG THE CHANNEL (C)
DEUTERIUM CONCENTRATION VS TEMPERATURE FOR P 3 / P 4 MICRO-SAMPLES WITH HIGHER RESULTS (INCLUDING P3L09) IDENTIFIED
shown f o r c o m p a r i s o n i n F i g u r e 8 ) . Of t h e s e 5 tubes showing higher hydrogen concentrations, one P3-M09, is in the same annulus gas "string" as P3LO9, and the other four are in the same sequential position within other annulus gas s t r i n g s . These apparent correlations are being further investigated.
SUMMARY AND CONCLUSIONS 300
400 500 600 DISTANCE FHOM INLET END (cm)
FIGURE 7 :
ffvrfrogen fnpress at Rolled
BREAKDOWN OF PROTECTIVE OXIDE ALONG THE OUTSIDE OF P 3 L 0 9 COMPARED TO DEUTERIUM CONCENTRATION
1.
Ingress rates are variable.
2.
Dissolved hydrogen in the PHTS coolant not appear to affect pickup at the RJ.
3.
Galvanic effects between the steel end-fitting and the PT are suspected to be the major source of hydrogen entering the RJ regions.
4.
Diffusion through the endfitting is to be the p r i n c i p a l route for entering the RJ region.
5.
Cr plate and Ti alloy endfittings are being studied to evaluate their potential for protecting the ends of the PT from hydrogen build-up.
6.
An a n a l y t i c a l computer model has been developed chat provides practical predictions of hydrogen build-up and redistribution at RJs.
EVIDENCE FOR REDUCTION IN PT OXIDE PROTECTIVENESS ON THE GAS ANNULUS SIDE
FULL POWER YEARS
MAX % METALLIC BEHAVIOUR
MAX MEASURED DEUTERIUM (mg/kg)
P3LO9
11.9
80
135
P4K10
10.6
30
28
P3J09
9.6
0
5
11.6
0
5
PRESSURE TUBE
NPDG05
does
suspected hydrogen
(AGS AT NDP IS OPEN TO AIR; AGS AT P3 AND P4 IS N2)
**** detection limit i s < 0.01 vol % CNS 9th ANNUAL CONFERENCE. 1988
389
Hydrogen Ingress Along the Pressure Tube (Zr-Nb) Carrier bundle tests, are being carried out to evaluate key effects such as the potential for increased hydrogen ingress rates with increased coolant side oxide thickness, and the effects of bulk boiling, under real PHTS conditions. Minimizing the corrosion rate by the addition of a minimum c r i t i c a l dissolved hydrogen Is expected to minimize the long term rate of hydrogen ingress associated with coolant effects. In-reactor tests to confirm this are in progress. 9.
There is evidence that as PTs age in the reactor, redistribution and precipitation of alloy constituents increases. This may affect future hydrogen Ingress rates.
10.
In order to ensure that hydrogen ingress from the gas annulus is minimized, a critical minimum oxygen level in the recirculating gas. Is required to maintain a protective barrier oxide. Oxygen additions around 0.1 to 2 volume % are expected to be sufficient in practice, for current reactor systems.
ACKNOWLEDGMENTS This work is p a r t i a l l y funded by the CANDU OWNERS GROUP (COG). I t is being undertaken by many groups and i n d i v i d u a l s Involved in Fuel Channel R&D activities, at OH and AECL. Their efforts are gratefully acknowledged. The author wishes to especially thank a l l of the Contract Officers and support staff involved In COG WP#35 - Corrosion and Hydrogen Ingress, at Atomic Energy of Canada Research Laboratories and Ontario Hydro Research Laboratories, whose work is discussed and in some cases has been reproduced here.
REFERENCES (1)
WARR, B . D . , LICHTENBERGER, P . C . , " C o r r o s i o n Performance of Z r - 2 . 5 wt% Nb P r e s s u r e Tubes In O n t a r i o H y d r o ' s O p e r a t i n g Nuclear R e a c t o r s " , p r o c e e d i n g s of t h e Canadian Nuclear S o c i e t y Annual Conference, S t . John N . B . , Canada, June 1987.
(2)
KEARNS, J . J . , " T e r m i n a l S o l u b i l i t y and P a r t i t i o n i n g of Hydrogen i n t h e Alpha Phase of Z r - 2 and Z r - 4 " , J . Nucl. M a t e r . , Vol. 22, pp. 292, 1967.
(3)
PERRYMAN, E.C. , "An A p p r a i s a l of Delayed Hydride C r a c k i n g i n Z r - 2 1/2% Nb P r e s s u r e Tubes, CRNL-1461, February 1976.
(4)
FIELD, G . J . , DUNN, J . T . , CHEADLE, B . A . , " A n a l y s i s of t h e P r e s s u r e Tube F a i l u r e a t P i c k e r i n g A NGS-Unit 2 " , AECL-S335, June 1984.
(5)
COX, B . , "Mechanisms of Hydrogen Absorption by Zirconium A l l o y s " , AECL-8702, January 1985.
(6)
PRICE, E . G . , p r i v a t e communication.
(7)
UARR, B.D., "Modelling Corrosion and Deuterium Pickup i n Z r - 2 . 5 wt% Nb P r e s s u r e T u b e s " , OHRD Report No. 87-73-K, May 1987.
(8)
RAMASUBRAMANIAN, N . , p r i v a t e cormnunication.
P3LQ9 Hypotheses 11.
There is to date, no substantial evidence from examinations of removed PTs or micro-samples from P3 and P4 reactors, that hydrogen ingress rates are, at present, accelerating due to coolant or bulk alloy effects.
12.
y p same position in annulus gas strings as P3L09, suggests that hydrogen ingress from the gas annulus may have contributed to the high concentration observed i n PT P3L09. Additional work (in progress) is required to evaluate further the l i k e l i h o o d of this Ingress process. 13.
390
Recommendations to add O2 to the gas annulus of a l l r e a c t o r s are be ing implemented. Surveillance by periodically micro-sampling and removing PTs will continue in order to monitor the effectiveness of these measures.
CNS 9th ANNUAL CONFERENCE, 1988
SAG OF Zr-2.5% Nb PRESSURE TUBES
N. BADIE, R.A. HOLT, C.W. SCHULTE and P G. SAUVE Ontario Hydro Toronto, Ontario, Canada
This paper presents a comparison of predictions of pressure tube sag and pressure tube/calandria tube (PT/CT) contact with measurements in power reactors. The absolute sag deflection and the rate of deflection in the channels studied in Pickering Unit 3 and Bruce Uni t 2 are predicted equally well by both PT design equations currently in use. However, inreactor sag deflection data cannot be used to test the validity of PT deformation models. PT/CT contact data do provide such a test. The more recently developed PT design equation is in better agreement, on average, with the small available set of contact data from power reactors, and is likely to give a better estimate of the closest possible approach of contact toward the channel outlet.
supports with an initial gap. CDEPTH calculates rrude flection profile of the PT and CT at specific-Li operating intervals, and the extent of contact, between the tubes in channels --.-he-re the spacers .ire greatly displaced from their design locations. Two models for PT deformation were used to calculate strain rates. Both models use the concept of separable, additive strain rates due to thermal creep, irradiation-induced creep and irradiation growth Cdimensional change in absence of stress, without any change in volume). Thus
€
th
where €j
is the strain rate in direct ion 'd'
et^
is the strain rate due to in-reactor creep
e-r
is the strain rate due to irradiation- induced creep, and
£Q
is the strain rate due to irradiation growth
INTRODUCTION PTs in CANDU (CANada Deuterium Uranium) reactors deform with continued operation as a result of thermal and irradiation-induced creep and irradiation growth. There are three modes of deformation in these tubes, namely sag, elongation and diametral expansion. Concerns regarding PT sag were initially centered around operational issues such as being able to move the fuel through the tube, or the fuel channel interfering with horizontal reactivity mechanisms. In recent years, however, sag has come to be considered as an important factor in determining the reliability of PTs in CANDU reactors. As a result of its lower creep stiffness, the PT sags faster than the CT and, depending on the locations of the spacers, may contact the latter. This contact produces a local cooling effect in the wa.ll of the PT and, if the hydrogen isotope concentration is above a certain threshold level, causes precipitation of hydrides and formation of hydride blisters at the point of contact. This reduces the fracture resistance of the PT. The ability to predict the extent of PT/CT contact in a fuel channel based on that channel's operating conditions and spacer locations is therefore e s s e n t i a l for reliability assessments of reactors. In this paper, we assess two existing PT deformation models by comparing each model's predictions of PT sag deflection and PT/CT contact with measurements made in some Pickering NGS-A and Bruce NGS-A fuel channels. The sensitivity of sag deflection and contact predictions to the CT creep rate is also determined, and used in the assessment.
PREDICTIONS The predictions were obtained using the Ontario Hydro finite element program CDEPTH ( 1 ) , which was developed to evaluate the creep response of fuel channel assemblies. The PT and CT were modelled as straight beams, and the channel spacers as spring
thermal
The models relate the three components of strain rate at a point in the structure to the stress, temperature and fast neutron (E > 1 MeV) flux at that point, and to the microstructure and crystallographic texture of the material. The first model, referred to as PT-1, was developed in 1979 according to the understanding of the mechanisms of creep and growth at that time, and the constants in it were adj usted to fit che early measurements of average axial and c ircumferenti a 1 strains in some Pickering NGS-A Unit 3 channels (2). PT-1 is currently the standard model for PT deforma tion p r e d i c t i o n s . The second, more recently developed model, referred to as PT-2, is different from PT-1 in that it includes an empirically derived mathematical function to represent the variation of thermal creep along the length of the tubes (as observed in diametral strain profiles), uses growth anisotropy factors which were derived from experiment , and was fitted to a wider range of experimental data (3). The predictive equations for strain rates based on the deformation mode Is are re ferred to as design equations.
MEASUREMENTS In-reactor measurements of tube sag are carried out by the Inspection and Maintenance Department (IMD) of Cntario Hydro's Central Nuclear Services. Channels K05, J17, Mil and M12 in Pickering 3 and channels K07, L08, 013 and T12 in Bruce 2 are gauged at regular intervals as part of the Periodic Inspection Program (PIP). Other Pickering 3 and Bruce 2 channels are selectively gauged as part of the In
CNS 9th ANNUAL CONFERENCE, 1988 391
I S 1 i
i-
i l . c t . . . r •, i i
'•.,,; i! . • 1 ! > • , :1 i "f. :-.il B.' I ' ! . 1 . . , t i t l u - K C ,-t:.,J:f . , . ) ,
>•
i 1 " . • . n :
l :
I r o n ,
] . TS
Tal.l
,.
.'.p.,i
. M i t v
•. i
RKSI
,ii
in .„.,
„•••
f'il,,-
:
S i i x r
I l \ S i
''li.-iiLj'.i'! t o r s .i
s . u \ i : V d S ' i r , i : y : : ^
I n s p . - i ' r
n'U-'AK''.
K M I .-irrri
>
ri:**
H . -r v n - i i u - M i i o m c i
; viiif p r o
. i t
{ i I o
.'()(> i s
;,r.,|
,,ii,|
t ' f
*• J i - . A H
t h r u
! •.
h i - ,i ;
iuNp----:
.i i njii1,
j u t t\/,i"^ -
i * '"•
t J (
m.ule
A p p . s r . i t
•-.•}; i ( ' ) ! i r f i . s u r . - ' ;
1 o c a .' i ( > , ! . • •
l
ne
i!.r, i r S : !.,): i
rn
i'-.~.i ,i , , i
f o r
i n n lu-;id i h«-
' " l u - i ' i
,,n'. • !",-
Ml.,.
]
'
'
P
o f
!
i
' K
t
1
I1S
i l
ll..!,
i-
1 \ - i -
;!,, licit i-li
li
.s;i
•« r.n,'
: I n -
T i n - s ) i > \ y 1
i .i v e
TABLE 1 DISPLACEMENTS OF SPACERS FROM THEIR DESIGN LOCATIONS IN SOME PNCS-A AND BNGS-A CHANNELS
c in;
•
[1
•
'.-K'l
tl i e
p,(,ti .m
and
0
0
P l - J l .'
-11
-40
P3-M11
-i;
0
K
-t!23
'»
•3
U
•.!,.
AVERAGE SAG DEFLECTION AND DEFLECTION IN P3 PIP CHANNELS
1,
*] (jnm)
(mm i
MEASUREMENT
20.1
2r)
PT-1
20.1
23-5
PT-2
22 it
28.6
01T1.KT ORIENTATION
P3-KU5
)
TABLE 2
u 1 t r a s n n i c i n d i c a t i o n s on tlie o u t s i de s u r f a c e of ( hti'T , n e a r the* s i x o ' c I ock p o s i t l o n , d u r i n g f h e CIGAR 1 I J V do tree t i on s c a n .
INLET SI'ACER (cnO
•
i mi,, . •-•hi 1 (1 i ' , p 1
Thi- FTs i n c h a n n e l s P 3 - J 0 Q . P'i-LO l ». P6-K10 a n d B 7 P] -1 wt'i-e r e moved from t h e r e i i c t o r s Cor P o s t I r r a r i i i i lion F.x.-niii n n r i o n ( P I E ) >i t Cha 1 k R i v e r N'uc 1 en r Lahcirat o r i es . The m e a s u r e d l e n g t h s o f r o u t a c t i n t tuPIF. c h a n n e l s a r e h a s e d on w h i t e m a r k s on t h e o u t s i d e s u r t a c e of t h e PT In r h e r a s e o f c h a n n e 1 ?i•Ml?,
OLTLET Sl'ACHR
•
Ri<..<* •
(IffliH't i o n p r o f i } * • . The 1 or,it i mis of s p a c e r s i n i-nrh rh;inru*l a r p also d e t e n» i « p d d u r i n g i n s p t - c t i o n hv ii!etins of edriv c n r r f i u p r o ^ c j s .
C1I.ANNK1.
\
i n 1P.. 1
i n . ,
•
:
i
r o h W i i u s
s; 1 o ] > o
t o
,•
T! , ,
•••.•.-•.:
.1.,' ;,,,
1,_
\\\\
C-M;
. i h .i i I-.II
u:;ins', t ! n -
u s
i t-iii'.: ii .
V ',
1
It-u;
iI.-
h-r'."«-
( , i u ; ; i n , :
•. .in,! ,-s :l.i V 1 In • i r 111-': ,!•. in ; lii'. stu.lv
, • • . , , . , 1
i-
1
(10''
mm•KK
b ;
>.:•
•'
' • '
EFPH - Equivalent Full Power Hours - Deflection at a862O EFPH • Deflection at 72480 f.FPH " Rate of deflection
K
TABLE 3
P3-M12
ND
P3-JO9
-70
+ 128
B2-KO7
+ 22
B2-O13
ND
+ 12
w
+1
-17
w w
B2-P12
-7
ND
P3-LO9
+ 16?
-15
u
+42
w
-6
Notes:
1 2. i.
The sign convention used is positive outboard. ND - Not Detected F. - East. , W - West
CNS 9th ANNUAL CONFERENCE, 1988
h
(mm)
w
+6
P4-K10
392
w
B2-L08
B2-T12
AVERAGE SAG DEFLECTION AND DEFLECTION RATE IN B2 PIP CHANNELS
h (n>m)
i (10'4
mm/EFPH)
17.9
25.8
4.18
PT-1
18.A
25.3
3.h2
PT-2
21.6
29.1
4 .06
MEASUREMENT
1 NOTE: The data for channel B2-O13 is currently under review and is therefore not included in the above averages. il - Deflection at 35740 EFPH
rtj.)^onab It* a^rc- MKcnt wi Ch the* measurement s for bofh o be in be t ter agreement rcdCtors. FT- ) appears ',-:i rh the P.3 mcivcment s for 'ioth t ho ahsolutf .s;j£ 1 whi 1 < ? PT? -seffps ro lie in bertiT and trho sat., rn 1 h'?jnpnsuremeiu s of saj\ ra tt .
t i me of contact of successive CDKPTH v.nrlcn. Tin- : r u n v s in Fi f,ur<- ? wr-re both obi ,-i i ucd w i t h !'T - .• . with d i f f t- rent c a 1 andr i a t ubr- r rw- p f.'i t c- s . before- , Both the predicted r i me- of initial i mi: ,i. and t he subsequent- spreading of coat ac t art- s.ei- n be r,"i t.her insensiti ve to "he CT cn-f-p ivt f <•.
The predict ion:; in Tali les ?. and .1 were obtai ned using, the current ly accepted v,i lue of Che C T creep race K c ,. This value was derived from a small number of c vsV re<"ici:oi" stress - relaxation data on specimens taken from one CT and exposed to a sraal1 fracti on of 1 ifetime finenee in power reactors. Hence, there is an uncertainty of about a fac tor of two assoc iated with the CT creep race ( 6 ) ,
—. PIE AT B3.700 EFPH a PT-;;. LOW Kcc °
P T - 2 HIGH Kcc
This has to be compared to the PT creep rates which were based on a large amount of experimental data from small specimens and full scale PTs tested under a variety of conditions in test reactors, and normalised co the diametral creep rates observed in P3 PIP channels and the P3 and P4 elongation measurements . The PT creep rates are therefore known wich greater certainty than the C T creep rate. The effect of the uncertainty in the CT creep rate on the predicted deflection of the PT is illustrated in Figure 1. which shows the measured and predicted sag profiles in channel P3-J09. The profiles predicted by each PT design equation are shown with two values o f the CT creep rate both of which fall within the band of uncertainty of this parameter. (The profiles marked "high K c c " are those obtained with the current value of K c c . ) This variation in the CT creep rate clearly affects the amount of sag deflection by more Chan the difference between the PT design equations. Therefore, the predicted sag deflections and deflection rates could be adjusted to match the measured values independently of the FT design equation used. A s a result, sag deflection data cannot be used Co assess the PT deformation mode]s. Rather, they could be used to adjust the CT creep rate once Che PT design equation is selected based on more significant tests.
V 100
<&? „ NX."
• » •
o PT-1. HIGH Kcc '
" PT-2. HIGH Kcc • PT-2. LOW Kcc
as o-
The effect of the PT design equation on the contact predictions for the same channel is shown in Figure 3. The two curves in this Figure represent the two design equations, and the horizontal bar indie a :e.s the extent of contact observed on PIE. In this case. both the time of inicial contact and the subsequent spreading of contact are affected. PT-2 predicts earlier initial contact and faster spreading of contact than FT-1.
^ a ^ ^ T ^ B
'I
iff
WEST
f DISTANCE FROM INLET (ml
DISTANCE FROM [NLET (ml
FIGURE 1:
SPREADING OF CONTACT IN CHANNEL P'i-.lu' ACCORDING TO CDEPTH, WITH THE PT-J DESI< EQUATION AND TWO VALUES OF THE CT CREKATE.
PT-1. LOW Kcc
EAST
SPACER
FIGURE 2:
f
o MEASURED
1,«^»
-SO
DISTANCE FROM INLET (m)
PREDICTED AND MEASURED DEFLECTION PROFILE OK PT IN CHANNEL P3-J09 AFTER 83,700 EFPH OF OPERATION, SHOWING THE EFFECTS OF CT CREEP RATE AND PT DESIGN EQUATION.
Pressure Tube/Calandria Tube Contact Figure 2 shows a typical plot of the predicted spreading of contact, in this case for channel P3J09, The data points in the figure correspond to the
FIGURE 3:
SPREADING OF CONTACT IN CHANNEL IM-.HI'i. ACCORDING TO CDEPTH, WITH BOTH DF.Sli-'.V EQUATIONS.
The above-mentioned effects of the CT creep rate and of the PT design equation on contact predictions ai'f similar lor all channels in this study. Based on these observations, contact data can provide a pood test of the validity of the PT deformation models.
CNS 9th ANNUAL CONFERENCE. 1988 393
}•
PREDICTED AND MEASURED LENGTH OF CONTACT
i:ilANNKl.
MEASURED
(m)
PT-1 (ra)
T i n -
c
tlit-
H M > :im u n i
;
o I I K H U - U
o n l l i ' l . .•it i d
PT-,! (m)
T h i s
p r e d i c t
I a b u l a t
t ' d
o c c u r s
o
c a s e s
.
w i t h
(
P T - 1
PJ -JO-T
0. iii.
0. 32
0.51
P3-I.0")
I. I f
1. 10
1 .38
P3-M12
2 . 9=>
2.T?
2.88
P4-K10
0 . 19
0.23
O.?7
B2-P12
1.29
0. ?7
0.87
Average
1.21
0.99
1.18
j
o
r i l i e
i
s
i
n
d t - v c l o p f d
t
} ' T , w i nIf-
c r e e p
r a t e
i i l m ^ ;
,.<• ;
n T n b l i r > 'i t ] i - 1
r h r < - ( -
n u t ] [ • [
<\ i ! i < • ] - . • : > :
r o n :
i
d nI .-1:-: i m u n i
o m o a s u r e r n e n i
i ' i c - k o r i n p I h t
[
j
n t h e
W ; J K
,-,ivc u f
i > ;s h ' i w n i
P T -} h
. i l ' . u
t-:-:t < - n t
c
o
i
:. ,
T
o f
i
h
: h
e
•(,«•
••
h
i
!v
i u t - l u i l . - - ,
n - - . , ' • :
n w i ii i - f i
\ n y
,
•'>>',,- .
.
) > / • : I i-y
e .•iVfi-.-ij-,,-
P T ,
i'-:.' c ) i •:].:
<>'. • n i , : , , , ' ;
t i n -r l i , i n i i < - 1
e r w ^ i n r i
; t
F T - ;>
v l i t - i v
<> ,
s i \ \ wi I i r.-itir
•
: I n -
* > : • : : >• u :
C I K H U . I - 1
s id
p j v < l i .
:nw.-tvt]
!n ' : , . . , > • , : ' < • < •::.<
r ! , i •,
; •,
hr-h.v.'i'/.u-
• ] <•:
t i n - •.-..!• i . . - ! ' • : :
, i s» < - u t
i tin.•('.
,-.i;• [ ;.
As a r e s u l t . P T - 2 p r e d i c t s ,i d i M * - r ^ n : IK-II I - ' : < d e p e n d i n g o n w l i e t l i c r .i P T i s i t : - '.-. 1 ] i'>! v ; i : ] j i * s I.. end a t t h e i n l e t o r t \w o u t l c t y-v.d nf a r K . - m n . - 1 .T ] 1 r h r e e i - h a n n e 1 s s ! : o v n i n T a b 1 f- l~> .1 i"i- " b , - i ' - h i o u t l e t " r l i . t r i n t ' ] s . : lit- h i ^ h t - r I i-T.j-.i.-ra: Mr.- ,,r.ci h i i-! matter ia 1 crec-p r a t e a t r h e o u t 1 (•*. .•-, i d<- o i : l..c o m b i n e i o c a u s e ;i s i g u i f i c . i n ? 1 v ti i f'.ln- r s : r . i i n r t h e r e . This eftect i s included if! J ' T - / b u . # i , ; i : d i n PT - 1 , h e n c e t h e - d i f [ t - r t - n c e i n i he- ' M ' f - d i c * i o r : s
TABLE 6
PREDICTED AND MEASURED MAXIMUM OUTLET EXTENT OF CONTACT Tables 4 and 5 show the p r e d i c t e d and measured l e n g t h and c e n t r e of c o n t a c t i n c h a n n e l s P3-.109. P3-L09, P3-M12. P4-K10 and B2-P12. Note t h a t in a l l of the above channels, the back end of the PT ( i . e . the end which e x i t s the e x t r u s i o n p r e s s l a s t ) i s a t the channel o u t l e t . On a simple a v e r a g e , PT- 2 und e r p r e d i c c s the measurements of length of contact by 3%. while PT-1 under/predicts by 18%. The p o s i t i o n of the c e n t r e of c o n t a c t i s p r e d i c t e d e q u a l l y well by both e q u a t i o n s , although the f a c t t h a t the p r e d i c t ions do not take tube elongation i n t o account causes the p r e d a t e d c e n t r e of c o n t a c t t o be a few centiinetres in error.
PREDICTED AND MEASURED POSITION OF CENTRE OF CONTACT
CHANNEL
.MEASURED (m)
PT-1 (m)
f'T -.'
P3-JO9
1.70
1.55
; "fi
P3-LO9
1.47
1.45
l.bl
P3-M12
1.92
1.45
i. n
Average
1.70
1.48
1 .68
CHANNEL
Note:
MEASURED (m)
PT-1 (m)
PT-2 (m)
P3-JO9
+ 1.48
+1.39
+ 1.45
P3-L09
+0.87
+0.90
+0.92
P3-M12
+0.45
+0.19
+0.29
P4-K10
-0.34
-0.25
-0.26
B2-P12
•0,89
-0.62
-0.62
Average
+0.31
+0.32
+0.36
i mj
The maximum outlet extent of contact is given relative to the centre 1ine of the PT.
Recent obse rva '.ions of deut P r i uni c one c-nt r.i f i on profiles which increase toward the out 1e t ends o f Pickering Unit 3 & A PTs (7) indicate chat tin- estimate of the iraximum outlet extent of c o m ;u-* is important in assessing the reliability of PT.s i n CANDU reactors. As shown in Table ft. the prrnr.s in the predict ions of this distance show a si p,m f i c-.sn: scatter. For this reason , es t ima f cjs of the max i muir outlet extent of contact based on ei thc-r des i>;n equation should include a measure of rhr* unrprt-ii nt •,• in the contact predictions of these equat ions. Al .';o. as more contact data becomes ava i1 able, we will be able to be t ter test our predic t i ve capah i1i t v and improve confidence in contact predict ions ma dp wi i h design equations.
CONCLUSIONS 1. Note:
Position is given with respect to the centre line of the channel, the sign convention being positive toward the outlet,
394 CNS 9th ANNUAL CONFERENCE, 1988
The absolute sag defleetion and the rate ot deflection of PTs in Pickering I'ni r . 3 .nul Bruce Uni t 2 are predic ted wi th ri'.ison.iM *• ,-JC curacy by both PT design equal i ons rurrcti! ] v in use,
2.
3.
4.
5.
Sag deflection data do not provide a good test of the validity of PT deformation models because of the sensitivity of deflection predictions to the CT material creep rate which is known with much less certainty than the PT material creep rate. Predictions of PT/CT contact are sensitive to the PT design equation, but relatively insensitive to the CT creep rate, and therefore provide a good test of PT deformation models, when used in conjunction with other in-reactor data. Based on the limited amount of contact data presently available, the PT- 2 deformation model, on average, is in better agreement with power reactor measurements than the PT -1 model. PT-2 predictions of the outlet extent of contact in "back-end-outlet" channels are in significantly better agreement with power reactor data than the predictions made with PT-1.
REFERENCES (1)
SAUVE, R.G, , " P r e d i c t i n g Creep Response of CANDU Fuel Channel Assemblies", Proceedings of t h e 1 1 t h C a n a d i a n C o n g r e s s of A p p l i e d Mechanics, Edmonton, A l b e r t a , Canada, Jum1987.
(2)
IBRAHIM, E . F . , HOLT, R.A., "Anisotropy of I r r a d i a t i o n Creep and Growth of Zirconium Alloy P r e s s u r e Tubes" , J . Nuc. M a t . , v o l . 90, pp. 311-321, 1980.
(3)
CAUSEY, A.R. , FIDLERIS, V. , MACEWEN. S.R. . SCHULTE, C.W., " I n - R e a c t o r Deformation of Zr2 . 5 we* Nb P r e s s u r e Tubes", ASTM STP 956, pp. 54-68, 1987.
(4)
PETTIGREW, M . J . , "An Instrument to Measure the D e f l e c t i o n of Pressure Tubes I n - r e a c t o r " , AECL Report CRNL-588, January 1971.
(5)
BARON, J . A . , DOLBEY, M . P . , ERVEN, J . H . . BOOTH, D., MURRAY, D.W., "Automated I n s p e c t i o n and G a u g i n g of F u e l C h a n n e l s i n CANDU Reactors 1 1 , 1. Mech. E. , C139/82, 1982.
(6)
FIDLERIS, V . , CAUSEY, A . R . , HOLT, R . A . , "Factors Affecting In-Core Dimensional S t a b i l i t y of Zircaloy-2", Optimizing Materials for Nuclear Applications, TMS AIME, pp. 35-50, 1985.
(7)
LICHTENBERGER, P.C., This Conference.
ACKNOWLEDGEMENTS
The a u t h o r s wish t o thank t h e I n s p e c t i o n and Maint e n a n c e Department of O n t a r i o Hydro f o r p r o v i d i n g the i n - r e a c t o r measurements, and t h e CANDU OWNERS' GROUP (COG) for funding t h i s work.
CNS 9th ANNUAL CONFERENCE, 1988 395
I'lfl'SI'HTH RW
IMI'hVAFli .'K'-'J.r. Ml PK'KSSiM
H. \ .
Moniie I'hnlk
Km-i-i*\
I ' i v c r
\u<
Ch.tll. K i w r ,
T? |i|-.-
CUKMH.l'
of ir'nr
C;in:i»t:i
-'ftiliir-ii.
M'STRMT 'tin i rii'i'vu'i' ! i f e rs i'i I h f'i iiii" g a r t i •[' s[ it i MiTs , t h e i r* s c t ' \ ifore~breaU is a fniii'l ion o f Ir.ilrngcri t '<>n« 'enl r a t i o n , s u s c e p t i hi 1 i f y t o • !'•!,• iv-i I h \ ' d r \>U • c r a c k in.u arc) frvict ui'e t nugtine:;::. P f ' r ' s s l / r e t lihen a r e )«• j iii^ i !• -\ e I t))x-'(l t h a i s h m i l d h:»\'e ) i 'i.. >r r ; i ! e s < il" h> droL^en i nj*T'ess , yt t f i uni s i n k s t o (-t il ]e\ -I t h e hvi I n i!es.
I. i HI i ' • ••
hilx>r:ii
m
Kii.i
it -
].](>
ftirry t he u-.-ji.-h I oi t lie f u e l a m i 1 In h e a \ ••• u a t e t c o o Ian! . feMititf'K i . f c h a n g e s i i i 11 i.- urn • 1 '•!• a n d lenuth u s i n g tinciiir.-nt d e s i g n "ijuat ion ( 1 ) show t h a i in t h e nci.oi r e a c t u r . - ilier.> shouic' net U- a n ; (troblem:f o r n \ e r 10 y e a r s , K i g t m - 2. S a g < vui r j i u s " a p r o b l e m in a feuIulws as t hey will c o n t a c t h m i ;--t .nl a 1 r e n d i \ 11 \ iu< •! -bar i i sni I n l i e s . l l i ) U - e \ e i , t h e s a g i s nn u ' e a I'linel i o n ..f I h e p r « i [ v r t i e w v<\' t t i e c . - i l a n d r i a f u U t h a n I ht • pr < ' s s u r e I u l n - .
REACIOR V.S5EL 1CALANDRI*)
HXED END SHIELD
I
CALANDfllA TUBE
PRESSURE TUBE
FREE (NO SMIEIO
[
END FITTING
IVfROllKTION'
ELONGATH
= = 7 =
f)n.- of the .-juhrintrttffH of* (lie CANTO; m a t ' t o r i s that t he |j;-f '.ssui-i • t ubos can bx? replaced when t.hey ha\'e reaeh'-d the end of t h e i r useful ] iff. Hnwover, loni* shi)tdou7is ai-f* r e q u i r e d t o r e p l a c e a i l the p r e s s u r e I lilies iuvi t l i e r e f c r e the length nf t h e i r sorviro \ii'r is \ i'1'v i nijKiT't ant to t he *•( "ori'Hrjii's of t he r e a c t o r . \P\> KI-IS the f i r s t
r-WDl' r e a c t o r
The
prfssure
T
GAS ANNULUS [ 1 ! END SHIELD D I S T O R T I O N ^
i
and the p r e s s u r e
t ubes were mado from 7 i r c a l o y - 2 by a si m i l a r I'aliri -aI ion process t o (hnRe developed for the Ifanford N-Reactor. Dotigla.s Point a;id P i c k e r i n g I'nit.s 1 find 2 al.so had Z i r r a l o y - 2 p r e s s u r e I ubt s s, but a l l subsetjuent f e a f t n r s hnve had I he h o t t e r Zr-ii. S Nb j-,r essur-e I UIM-S ,
=j
GARTER SPRING
t ubes
- I:
SIMl'I.JKII-'l) DR-WISfi 0I-" fANDV-IHK VRHSKI.
-|260
in Pir-ker ins
I ' n i t s 1 ;md 2 have recent I y ht-en replart^i wi I h 7x~'l.\> \ b and with the c l o s i n g a\' NTD, t h e r e a r e no |c;trtl*-r HIW 7.\ ri"i) o.v-2 p r e s s u r e tubes in s e r v i c e . M Uioijgh the s e r v i c e record of the 7.V-1. fi N'b preRKure (HIK'S is very good, 77 have had t o be removed e i t h e r l»-cause t.hey contfiined d n f w t s or bf*catjKO they de\-eloj)ed c r a c k s . Therefore H tube is r e q u i r e d that is more- t o l e r a n t lo the s e v e r e ojoerat infi c o n d i t i o n s and hence w i l l have a longer s e r v i c e l i f e .
/Too
3
E L
. Prfi,B,D-
2.5
- 150
Tljf s e r \ i c e l i f e of a p r e s s u r e tulx- depi-nds on i hantjes i ti s h a p e r e ] a t . i vo t.o d e s i jen a l lowajices i t - n b i l i t y ttj 1 f ; i l ; - l > e f o r e - b r e a k i f i( r-i-nrks.
its and
*
Tli< • st r< s s and t em|x*r-at \ir^ from t h<* p r e s s u r i ;'.(•iiiriM of the p r e s s u r e tubes during s e r v i c e due 'n c r e e p and tfrov.-th. The tubes i n c r e a s e in dt.-uneter and lenirt h, and \ hey a l s o sag in 1 he middl e, I-i tli ire I, ;is they a r e only support.si at each end and CNS 9th ANNUAL CONFERENCE, 1986
N
i ?
SKRYTfT I.IFF
396
RF.-VTOR
0.5-
0
10 20 30 EFFECTIVE FULL POWER 'EARS
AN
n ni-
40
i-rah-lVd 'iv-lVr:ik durinn Servjc" '['ho nw>«t probable way for a pressure tube to fai 1 is for- .1 ft"i'-l< to initiate at a small detVrt bv delayed hydride cracking; it then propaeatns both alonK llie lube ;ind through the wall and eventually a l l o t s \ lie heavy water coolant to leak. To ensure leali-lvfarc-break, Hi*1 leakage must be- detected and l he mm-tor shut down before the crack a rows to the c r i t i c a l length at winch i t becomes unstable and the lulie i-upiun-s. If a crack is long before t.herp i s iietI^H-I;.ible ]f>n}i;ie assiireii. Hence, the Ker\-ice l i f e of the \u\n> is limited by 1 he m a t e r i a l ' s s u s c e p t i b i J i t y lu (.-i-rtfU i n i t i a t i o n , the r a t e of crack iirowth and the
In Iti a l l y the pressure tubes have excel lent fracture toughness and resistance to cracking. However, during s e r v i c e , the neutron flux introduces damage to the c r y s t a l lattice that increases the tensile strength and decreases the d u c t i l i t y of the tubes (2), Figure 3. The damage t o the c r y s t a l l a t t i c e MISO probably increases the s u s c e p t i b i l i t y t o crack i n i t i n t i o n , increases the r a t e of erack growth I 3 ) , Fi si11re 4, and decreases the c r i t i c a l crack l**ngt.h 14), Figure 5. Tn the r e a c t o r s that, havt sufficient bearing travel for axial elongation and four g a r t e r springs ni t h e i r design location, changes in dimensions and shape should not limit the service l i f e of pressure tubes. Their service l i f e will probably be limited by confidence in t h e i r a b i l i t y to leak-before-break.
& D 0
FIGURE 4: TIE EFFECT OF TEMPERATURE ANT) IRRADIATION ON DELAYED HYDRIDE CRACKING CRACK VELOCITY
200
D
1
°a°
t
40
P > » 0 l HPi IP.1 R = 41.3 i m f T > 4» i « J
°
*
• Q
l
i
3
FAST NEUTRON
XX'SHAIL SPECIMENS + +'BURST TESTS
AXIAL TRANS.
5 FLUENCE
8RITTLC fBACTUflE
•
20
Ci4B UNIRRADIATED Clie IRRADIATED IN NflU P3F13 1.32 in FROM INLET END UNIflflAOIATED Zr-2.5 Nb
DUCTILE FAILURE
X !0~'3n/tn' O t f SMALL SPECIHENS
100
CONSERVATIVE RESULTS,
»800
CCL LIMITED BY SPECIMEN LIGAMENT
50 -
• AXIAL o TRAN5. ! FAST
'•',:
I NEUTRON
)
1 FLUENCE
5 x
6 Il)"
2 5
«
150
200
300
TEMPERATURE.0:
7 n/m
2
THE F.FFHT (*" K.VST VKlr.MW HJ'FNff: (IN 'ffl TKNSIIJ-: PW1PKRTIFS OF Z R - 2 . S Mi PRESHI.RK •H 'hKS
FIGIRF. n :
TtfE EKFHlT OF SERVICE TIME ON T t E C R I T I C U . CRACK LENGTH OF ZIRC.AIJDV-2 PRESSl.KE TVBKS UI-MVED FROM NPD
CNS 9th ANNUAL CONFERENCE, 1988 397
: v !iTt -\ IV.; 111! M-'|,'\ H 1\ I.II 1 i
i
i..•:iK~U fi •i-i>-t>i (\*ik is a f'unct KJII of i r i i t i:i! i n n . c r a c k g r o w t h ;unJ c r i ! i c i l c r a c U The ini(h >f( :trtt iw1 -i! '\irn)t'n\ \urinbles arc:
PARAHETERS
i-v.n-U lentil K.
EFFECT OF IRRADlATrON
Crack Crouch
Stress. Hydrides
Increasas Velocity
C r i t i c a l Crack
Stress. Hydrides. Hydride
Dflcr
If delayed h>dt ide cracking fuuld not occur or if it was much niut^ difficult for it to occur, then the pressure lubes would be loss likely to crack and shouId hi vp 1 onger service 1 ives. For delaypd hydride cracking to occur the material must be susceptible, hydrides must be present and there must be a minimum combination of flaw and tensile stress.
Susoqptibl1ity
\X*UH* 0.27 ;it% then hydrides would not be present ?it nvii'lof (ljx'rat inii temperatures and dolaverl hydride ti;nliiiiii '-ctuld mil < if-cur during re;ictnr operation.
to_Dela^ed_Hydride_Cracking
In pressure tubes most of the a grains are oriented uith their basal plane normals close to the circumferential direction. A smaller proportion of grains are oriented with their basal plane normals r-loso to the radial direction and very few with their hasaJ plane normals close to the axial direction, Figure G. All of the cranking in pressure tubes has occurred on (he radial-axial plane because the highest operating stress is in the circumferential direction. However, tests have shown (5) that it is much more difficult for cracking to occur on the radial-circumferential plane than on the radial-axial plane. Rolled sheet has a simpler texture than pressure tuljes and has most of the a grains oriented with their basal pJane normals perpendicular to the rolling plane. Cracking occurs much more easily when the crack :?lane was parallel to the basal plane rather than the prismatic plane, Figure 7. Therefore, if pressure tubes had ?&< a grains oriented with their basal plane normals close to the circumferential direction they would be much more resistant to delayed hydride cracking on the radialaxial plane. However, a new pressure tube (TG3 Route -i\ which has a more radial texture had a susceptibility to cracking similar to standard tubes, Table 1. This was probably because there were still ibany of the a grains oriented with their basal plane normals close to the circumferential direction - the resolved fraction of grains in the circumferential direction was still 0.48. Experiments have been conducted to develop a fabrication route that should produce a more radial texture similar to rolled sheet.
RESOLVED FUNCTION OF GRAIN'S Radial
Ci rcumferent.iai
Longitudinal
O..3L>
0.62
0.06
FIGURE 6:
TEST DIRECTION!
A
B
C
Resolved F r a c t i o n of Grains
0.67
0.16
0.16
0.4
0.4
Crack V e l o c i t y mm/day
3 9.1
KIH MPaVm
FIGURE 7:
Prese-noe of H>drides The solubility of hydrogen in zirconium is very Tube Jtrt.- at. room temperature but increases with increasing temjjerai urn. For hydrides to be present in the Hiicrostrur-ture the hydrogen concentration must be Standard threat t;r than the solubi 1 ity limit. New pressure TC3 Route 4 fulx-'s i<;rii:iin about 0.1 at% hydrogen; above about ?.()['. r .'ill t hn hydrogen is in solution and therefore Potential I1- Inyed h.vdride cracking cannot occur. During SI-IS iff- the hydrogen concentration increases from the -
398 CNS 9th ANNUAL CONFERENCE, 1988
THE ORIENTATION OF a GRAINS IK' STANDARD ZR-2.5 NB PRESSURE TUBES.
15
THE EFFECT OF TEST DIRECTION ON THE DELAYED HYDRIDE CRACKING OF ZR-2.5 N"B ROLLED SHEET AT 2 5 0 T
ta««pc»i l i t f
eo DHC of
Zr-2.
t t u r a Tuba*
Velocity
0.32
0,63
0.06
0.49
0.48
O.Ofc
0.67
0.16
0 16
1-6
would nut foe a f O e t e d . The first pressure tubes with these sinks Kill jirohably be installed either m Darlington 1 11 i< 1 or in Pickering I'nit S when it is. ret.uU'd. We are now developing the sink concept for (he whole lM«!th .if the tube. Two appl ications are ly-iiu' t'\»liint«il, external rings on the outside of the tulie similar In the rings at t he ends and wires in tin- middle ,-f I he i.:il I . Figure 1 1.
HYDKIDEO RING SECTION or PRESSURE TUIE
FIGL'HF 8:
DELTERIIM PinCTP AT TIIE FNDS OF ZR-2.5 NT? E TJ^KS WRINU SERYTCF
YTTRIUM INSERT -
MC«T 1313 C<* 60 * 18
•
• HIIIIUWl / 10 Ofl'l x m*iUMD 1 110 M l S MOtL ' 10 OKU MML , ,io M r l 3o
g
c
2B-2S.C/. Kb
P
z o
1 X
Effffllvf FUU POWeR DAYS
h9
FWRE
9:
•
•
HYDROGEN PICKIP OF ZIRfAI.OY-2 .ANT) ZR-2.5 NT? PRESSURE TIRES DURING SERVICE IN KJWER REACTORS S-
Tin1 SP>". iff time for the hydrogen concentration to n-H.-h O.Z7 at% could be increased by reducing the r:.li- of in_Srpsf.. Research programs are investigating e.'H-h of the threp ingress routes to either reduce or eliminate the ingrfiss. Ingress from corrosion with the heavy water coolant may be reduced by small changes to the chemical composition of Zr-2.5 Nb, barriers between the pressure tube and end fitting are being evaluated and changes to the annul us gas have been recommended to ensure that the oxide Ijarrier on the outside of the tube is maintained.
-
,/» - -*• - " " ^ ID. n. )o. < OlSfONCC FHaH I END IMHI
FIGLKE 10:
THE m'DROGEN CONCENTRATION ALONG A PIECE OF ZR-2.5 NB PRESSURE TUBE WITH AN YTTRIUM INSERT AT ONE END AFTER HEATING .AT 313"C FOR 60 AND 180 DAYS
FIGURE 11:
SCHEMATIC OF A FUTURE PRESSURE TUBE WITH INTERNAL AND EXTERNAL YTTRIUM SINKS
An alternative is to collect the hydrogen in a manner i-herf it could not cause cracking and will not afOct the integrity of the tubes. Yttrium has a groatpr affinity than zirconium for hydrogen (7) and in a diffusion couple will collect all the hydrogen. 1.4 at% yttrium was added to Zr-2.5 Nb as an alloying •-lenient but it only prevented delayed hydride i-racV. LriK up to hydrogen concentrations of about 0.72 at% and it lowered the corrosion resistance and fracture toughness of the alloy. However, yttrium in contact with zirconium works very well but it has \.T.\ poor corrosion resistance and must be encapsulated. The first application of yttrium sinks will be at the ends of pressure tubes where the ingress is the highest (81, Figure 10. The advantage of this nppl irat ion is that the ends of the pressure tube are mithoHtd of the grooves in the rolled joint and are not carrying any load. Hence, if anything does Bo wrong with the sink during service the end of the iul«> could be machined off and the rest of the tube
C N S 9th A N N U A L C O N F E R E N C E . 1988 399
S I !<• St i',-.;-. a n d J
UEFKNF.NCRS
T h e r e li:is t o U- ri minimum c o m b i n a t i o n o f t e r i s i I*1 s t r e s s ami flaw d e p t h f o r a r r a r k I n i n i t i a t e . The hno|) s t f f s s from t h e h e a v y wat or < -on l;uit. p r e s s u r e i s Kill Ml'a I-IIKI a f l a w must fx- > 0 . 5 mm d e e p V> i n i t i a t n a n-ii k. Htit.c^er, i V there a r e residual I onsi le st r^'SHcs in t h* • I ube, foi" e\;uiiple a^ < -an ooeur in 1 he i- il l-'d Muni rt-uion, t hen t In- i -ombined tensi \o s t r e s s i -an i-ausc crack intf at \er.v smal I f lavs. Menre, residual st n'srii-s, manuffu't uring Haws and flaws that voulii form during s e r v i c e .should be minimized.
A.M. Causey, V. f-idl r i s , S.H. HicHwen and r . k . Sfhiilfe, " In-re ct or DofcmvHt ion of /.t-2.-i N'h 1'rossure Ti hcK" , ASTM STP 95fi,
(2)
V. F. Ibrahim, "Mochariical I'roper t i<».s of Cold Drawn Zr-2.n Mi 1'r-essure Tubes a f t e r up I o Twelve Years in CANTO.1 Rear-tors", proceedirms-; of t he International Conference, on Materials for Nui-lnfir fieaetnr Core Application**, 1987 October, Bris1<»l , Kngland.
i-\nkir\TU>\ n\- VW]XVY\VV TTBKK
CU
S. Saga! , unpublished work, AKCL-CItNL.
The program has t wo phases.
(^1)
C.E. Coleman, B.A. Cheadle, A.R. Causey, C.K, Chow, P.H. Davies, M.D. MrManus, I). K. Rodgers, S. Saga t and G. Van Drunen, "Rxamination of Zircalay-2 Pressure Tubes from \PD", ASTM I n t e r n a t i o n a l S\Tnposium on Zirconium in the Nuclrar Industry, 1988 June, San Diego, Cali fornia.
(5i
C.F., ToJeman, "P.ffeet of Texture on Hydride Reorientation and Delayed Hydride Cracking in Cold Worked Zr-2.5 NTD" , ASTM-STP 754, pp. 393-411, 1982.
(61
V.F. L'rbanic, R. Cox and G.J. F i e l d , "Long-term Corrosi on and Deuterium Uptake i n CANDl'-PHW Pressure Tubes", ASTM-STP 939, 1985, St rasbnurg, France.
(71
W. SpaJthoff and H. Wilhelm, "The Use of Hydrogen Getters for Prevention of Hydrogen Kmbri111 ement in ?.irconium-Al 1 oy Fuel Cans" , ASTM STP 458, pp. 338-344, 1969.
(8)
C D . Cann, unpublished work, AECL-WNRE,
(9)
B-D. Warr, M.W. Hardie, V.F, L'rbanic and D.O. Northwood, "Corrosion Performance of Zr-2.5 Mb F'ressure Tubes in Ontario Hydro' s Operating CANDU Nuclear fteactors", proceedings of the 5th Asian-Pacific Corrosion Conference, 1987 November, Me 1bourne, A u s t r a l i a .
Hiase 1: Susoept ible
rrrsKyre Tubes_ Th_at_ _Wl.l.l_ Delays] Hydride Cracking
Tin- a l l u y composition and the fabrication route haw been specified for- pressure tubes t h a t may ha\e H lower r a t e of doute.rium ingress and should be less suscf-pt ibl«» ti> i 1Playcd li>{jride crackingAnalysis of thf. Kydro^en ingress of pressure tubes removed from U ^ jKiwer re«ctors i n d i c a t e s that, the ingress may be pn>]*n"t ional t o the Ye concentration ( 9 ) . The Fe concent rat ion in Xr-2.5 Ml) used to be specified as 1fiOO ppm maximum and has l>een lowered to 650 ppm in the hojy* that i t will reduce the hydrogen i n g r e s s . The f a b r i c a t i o n route and inspection procedures have txjen improved to minimize the poss i bi1i ty of fabrication flaws. The extrusion and cold working processes have been njx^cified that should produce a radial t e x t u r e and reduce the s u s c e p t i b i l i t y to delayed hydride cracking on the r a d i a l - a x i a l plane. Two tubes have been ordered and should be delivered this f a l l . If the t e x t u r e produced does not give the 11 fiiu< -1. i on in susce.pti bi 1 i ty t o de 1 ayed hydride cracking that we feel i s possible then we will try again with another f a b r i c a t i o n r o u t e .
Phase 2: Eliminate Reactor _Oper_a_ti_pri
Delayed Hvdride Cracking during
Those tubes will probably be q u i t e d i f f e r e n t t o c u r r e n t t u b e s . The composition of t h e a l l o y may be modified to reduce the hydrogen i n g r e s s . This may be by redue-int? the consent rat ion of some of the impurity Hemenl.fc or by adding small tunmints of elements as alloying a d d i t i o n s . Ceramic coatings and surface i ivatiwnt K arr- a l s o being tested. The tubes will ha\e yttrium sinks t.o keep the hydrogen concentration bHow 0.27 at% for 30 years. The yttrium will e i t h e r l>e HinbpdtWl in the wall or as rings an the outside of the. tube or possibly both. A tube has been ordered that K'i ] J be extruded with yttrium rods in the b i l l e t to produce small wires in the middle of the wall of the finished tube. We hope to have sections of t h i s lube, and a section of tube with external yttrium rings on t e s t in a loop in about nine months' time.
7.1-2.5 MJ pressure tubes that will be less l i k e l y to i'ftirU and have longer service l i v e s can be developed. These tubes wi11 have smal1 changes to t hei r rhemiral composi t.ion and c r y s t a l lographic texture and will have, si nits to nolleet the hydrogen.
400
CNS 9th ANNUAL CONFERENCE, 1988
LEAK BEFORE BREAK AMD LEAK DETECTION SYSTEMS OF CANDU FUEL CHANNELS
E.A. E.G. G.D. C.E.
(J)
Shalaby (>) Price (') Moan C1) Coleman C)
Atomic Energy of Canada Limited Sheridan Park Eesearch Community Mississauga, Ontario L5K 1B2
(') Atomic Energy of Canada Limited Chalk River Nuclear Laboratories Chalk River, Ontario KOJ 1J0
ABSTRACT In the CANDU reactor each of the approximately 400 hot pressure tubes containing the fuel bundles and the pressurized heat transport water is surrounded and insulated from the cold moderator by a calandria tube. The pressure tubes are made from cold worked ZT-2.5%Sb, and the calandria tubes are made from annealed Zircaloy-2. The annulus between these two tubes contains a gas vhose dewpoint is measured to ascertain if leaks have occurred into the annulus. Leak-Before-Break provides the operators of CANDU reactors with advanced warning of potential unstable failure of pressure tubes. It relies on matching the characteristics of the leak detection procedure to the characteristics of the cracking phenomena. A procedure for leak detection and reactor response has been built up from the use of the annulus gas. The characteristics of the crack - its shape, its length at wall penetration and its growth rate by delayed hydride cracking, and the expected leak rate at various lengths are used to establish the response time for leak detection. The reactor is required to be shutdown when the crack has a length much less than the critical crack length. This critical crack length, determined using burst tests on slit tubes, is the crack length at which the crack growth becomes unstable. New CANDU reactors will continue to emphasize the importance of the annulus gas system for leak detection. Emphasis has been placed on the design of this system to increase its sensitivity and to shorten the response time. These requirements have led to investigation of new means and concepts of detecting pressure tube cracks or leaks.
sample and record moisture at varioua positions. The use of a bulk moisture differential monitoring technique using current instrumentation circuits promises to increase system sensitivity and alleviate the problem of instrument drift. Analytical tools have been established to evaluate system response characteristics to different leaks at different positions in the systems. Combining Leak-Before-Break technology and improved techniques of leak detection provide increased margins between the time at which a crack penetrates a pressure tube and the time the operator receives the information and takes corrective action to prevent unstable failure. 1.0
The CAl.'DU reactor has a controlled atmosphere space surrounding each of its pressure tubes which is suitably monitored to detect leaks from the primary system. Thus if a crack develops in a pressure tube, penetrates the tube wall, is detected and located by fluid leakage and action taken before the crack becomes unstable, then Leak-Before-Break has been achieved. The key elements are the ability to detect the leaking fluid and the time available to take action relative to the time taken to detect the leak. The earliest possible detection of a leak is an obvious advantage and current systems rely principally on the measured dewpoint. However other techniques have become available since the currently operating designs were installed. This paper will concentrate on improvements in the detection of leaking fluids.
2.0 Current developments of an upgraded detection system have focused on in-situ and remote moisture measurement on individual channels with sensors that can detect all possible ranges of leaks but particularly small ones. The performance of various detectors for in-situ application in high temperature and radiation environments is currently being evaluated. Examples are fibre optic spectrophotometry, and ceramic plate detectors where the capacitance varies with moisture. The characteristics of new detectors such as accuracy, sensitivity and drift in the gas annulus environment must be established. Leak location or identification practice will be improved by a modified system configuration and the use of multiport indexing valves to intermittently
INTRODUCTION
THE CANDlf REACTOR
Figure 1 depicts the main features of a CANDU reactor. A calandria containing moderator heavy water at 70°C is penetrated by about 400 horizontal fuel channels. Each channel consists of a pressure tube (PT) of length 6m, containing the natural uranium fuel and heat transport heavy water at a pressure of 10 MPa and at a temperature ranging from 250 to 300°C, surrounded, and insulated from the cold moderator, by a calandria tube (CT). (Figure 2) The space between the pressure tube and the calandria tube is filled with gas and is described as the annulus gas system (AGS). The pressure tubes are made from cold worked 2r-2.5Nb, with a wall thickness of A.2mm and inside diameter of 103mm, and the calandria tubes are made from annealed Zircaloy-2, with a wall thickness of 1.4mm and inside diameter of 129mm. C N S 9th A N N U A L C O N F E R E N C E , 1988 401
The AGS is a closed recirculating loop, with headers and intermediate tubing connections, that ensures a uniform flow through each annulus (Figure 3). Most CANDU reactors have parallel connection with up to 12 channel annuli in series per string between headers. During normal operation the system is on recirculation mode. Compressors are used to recirculate CO, (or N,) through the system which is normally operated at a pressure of 103 kPa(g) and a devpoint ranging from -30°C to -18"C. When the dewpoint reaches 0°C due to chronic leakage, the system is purged at the rate of 6NL/s until the normal devpoi^t of -30°C is reached. Typically purging for moisture removal is administered once a week. The system dewpoint is continuously monitored using devpoint sensors, (Fig. 3 ) . Alarms on "high devpoint" and on "high
Figure 1 (OANDU 600 MW reactor
The pressure tubes are attached to end fittings at each end of the channel. The end fittings contain connections to the heat transport system and closures to enable fuelling to be done on power. The connection of the pressure tubes to the end fittings is a rolled joint. The rolled joint fabrication produces residual stresses in the pressure tube due to the wall thinning and tube expansion. In early power reactors an incorrect rolling procedure produced excessive tensile residual stresses in the pressure tube. These stresses, particularly the hoop tensile, had a high value over a length of lOum on the inside surface just inboard of the burnish mark representing the inboard limit of roll penetration. The high tensile hoop stress was the cause of crack initiation that led eventually *.? detectable leakage of the primary heat transport vater itico the channel annulus. A modified installation procedure has eliminated the high stresses in current reactors.
L END f
GAS ANNVLUS
OUTLET Figure 2
I FJEL
BUNDLES |l?l
CO] Boulc 51,
Figure 3 Schematic diagram of gas flow in current Annulus Gas Systems
rate of dewpoint rise" indicate the presence of a leak into the system. The Electronic water detectors termed "beetles" indicate the collection of liquid vater. These have a slower response than the dewpoint sensors for small leaks. Cold finger traps are used to obtain samples of moisture to establish the source of the heavy water (water from the moderator contains more tritium, than the primary heat transport water - thus the source can be determined).
SPACPRS i4i FACERS w
PRESSURE T U6E
Simplified description ol a F U el Channel - The gap. or annulus, between Ihe pressure tube and the surrounding Zircaloy-2 calandna tube is filled with an insulating g^s. The annulus is sealed at each end by bellows thai accommodate relative axial movement between the fuel channel and the reactor end shields
The annulus gas system (ACS) has been developed into a system that is sensitive to the presence of any moisture resulting from a breach of the primary heat transport pressure boundary through a PT leak or a CT leak (although the Utter type of leakage has not been significant). 402 CNS91h ANNUAL CONFERENCE, 1988
3.0
CRACK GROWTH IN PRESSURE TUBES
Experiments and reactor experience have shown that the principal crack nucleation and growth mechanism in Zr-Nb pressure tubes is delayed hydride cracking. There have been 24 instances in which pressure tubes have leaked from cracks which nucleated at the rolled joints. In all of these cases the cracks originated at the regions of the rolled joints with high tensile residual hoop stresses. The cracks grew through the wall permitting leakage of the heat transport water into the annulus; they also grew axially.
CHiTlCAL CRACK i f N G T H i C C U
- C C L T
Figure 4 Schematic: diagram to show crack dimensions at onset of leakage
The time available to the operator to detect the leak and to take action is the time needed for the crack to grow from its length at wall penetration to the length at which it becomes unstable (its critical crack length, CCl). This available time can be estimated from the model shown in Figure 4. In the model the crack nucleating at the inside surface has grown radially and axially by delayed hydride cracking such that its axial length at wall penetration is L o , and with inboard and outboard grovth at a velocity V, the time, t, required for the crack to grow to its critical length, CCL, is (CCL - L 0 )/2V.
The corresponding tests on small specimens prepared from irradiated material have given crack velocity data which are greater by a factor of 2-5 than those from the tests on unirradiated material. Finally the data from a series of crack velocity measurements carried out on a leaking full sized rolled joint after removal from reactor have given crack velocity data which are smaller than those from the unirradiated small specimens. These velocities are shown as individual points in Figure 6. The leaking rolled joint used in the tests was pressurized by hot water, and crack growth was by delayed hydride cracking. There is thus a wide variation in the crack growth velocity data that could be used in equation (1). The critical crack length for Zr-2.5Nb pressure tubes has been measured in slit burst tests using irradiated and unirradiated material at different temperatures. A statistical survey of the data has shown that the 95X lower confidence level CCL for the inlet end at 250C at a hoop stress of 150 MPa is
(1)
This is the time t available to the operators to take action; it clearly depends on L o , C and V.(l) An examination of crack shapes suggests the crack lengths at wall penetration L o were close to A times the wall thickness, but the detailed examination of three of them indicates that the crack length may have been 7 times the wall thickness when the leak was detected. Thus a conservative value for L o in equation (1) would be 28mm. The delayed hydride crack growth rate in cold worked 2r-2.5Nb depends on the stress intensity factor at the crack tip as indicated in Figure 5; for the stress intensity factors found in pressure tube flaws the crack velocity is mostly independent of the stress intensity factor. This velocity has «n Arrhenius' dependence on temperature, shown in the graph in Figure 6. The band of data in this figure was obtained in tests on small specimens prepared from unirradiated material.
Figure 6 Graph showing the dependence ol delayed hydride cracking velocity on temperature (or Zr-2-5Nb. The band includes the scatter observed in tests carried out on small specimens The other data were measured in tests on a leaking crack in a pressure tube during heating and cooling cycles when trie crack length was 25-50 mm
A conservative estimate of the time available for the operator to take action following the onset of leaking can be obtained using a) CCL - critical crack length (95% L C D " 50 mm
( UNSTABLE
1
CRACK QROWTH
b) V - crack velocity in irradiated material • 2.7 x 10-'m/sec c) L o - crack length (7 tiroes wall thickness) / f
STABLE CRACK GROWTH
- 28m.
K
>C
NO CRACK GROWTH
The calculated time is about 10 hours. From the lower crack velocity data determined in the component experiments, the time available could be about 100 hours. 4.0
STRESS INTENSITY rACTOR |K||
Figura 5 Schematic diagram of the dependence of crack velocity on strE intensity factor
IMPROVEMENTS IN THE ANJTOLUS GAS SYSTEM
The above Leak-Before-Break scenario with the short time available for taking action ha3 prompted an increased emphasis on the performance requirements of the Annulus Gas System (AGS) to: C N S 9th A N N U A L C O N F E R E N C E . 1988 403
1.
Minimize leak detection time to PT leaks.
2.
Improve leak location capability using the system configuration.
and action must be less than the time taken for a PT crack to grow to its Critical Crack Length (CCL). b)
This prompted redefining such requirements and investigating new means and concepts in detecting PT leaks. Current development has focused on the following aspects: 1. New Annulus Gas System configurations for improved response to such leaks. 2.
Hew dew point monitoring systems.
3.
Mew dewpoint sensors for both in situ and remote applications.
The developments in these methods are discussed in the following sections. 4.1
Mew Annulus Gas System Configuration
Two improved system configurations were considered based on the existing CANDU designs. They stem from two basic requirements: a) Minimize leak detection time. The leak detection time and time for operator decision
Minimize leak location time using the system configuration.
4.1.1 First ConfiRUration (Figure 7). This configuration involved changes to the AGS on the reactor face in conjunction with increased recirculation flow rates (Figure 7 ) . This scheme entails having 4 subcircuits each monitoring one quarter of the core. All channel strings containing 7 (or more) channels in series are subdivided such that the maximum number of channels per string is five. Each subcircuit has 30 inlet lines (9.5mm tubing) connected to 120 channels and two compressors for a normal recirculation rate of 6NL/s. Using this scheme the estimated leak detection time will be between one half and two thirds of the current Darlington detection time. Amongst the various schemes considered this offered the optimum in improvement to the leak detection time, no cascading changes to the AGS design pressure (currently 103 kPa) and the minimal number of additional components and modifications to the existing AGS configuration. A feasibility and optimization study of this scheme is ongoing.
ENLARGED VIEW
VALVE (TYPICAL} ROTAMETER fTYPICALJ
LEGEND N - NUMBER OF INLETS NC - NUMBER O f CHANNELS/STRING CP - COMPRESSORS DP p DEW POINT SENSORS
Figure 7
404
CNS 9th ANNUAL CONFERENCE 1988
Proposed modified configuration (or Annulus Gas System
(OP)*
%(DP)*
LEGEND * DP "J\
0
ADDITIONAL COMPONENTS DEW POINT MONITOR MULTIPOINT INDEXING VALVE
Figure 8
Second configuration considered for Annulus Gas System
4.1.2 Second Configuration (Figure 8 ) • The second configuration considered entailed two modes of sampling; Bulk and Selective. As per Figure 8 the system is divided into two circuits; a)
Circuit A is used for bulk moisture monitoring. The circuit will be configured to allow each reactor half to be monitored separately for bulk moisture using the existing channel annuli configuration.
b)
Circuit B is used for selective moisture monitoring. The circuit is configured to allow a group of rows to be monitored separately and selectively on a row by row basis using a multiport indexing velve.
The principle of circuit B is based on the introduction of new 6mm outlet tubes inserted through the bellows attachment ring of each channel. This configuration offered, for any size leak, a maximum time for leak detection and location of 1.3 hours. This configuration was not pursued further due to the practical difficulty of effecting such changes on the reactor face at some stage of the reactor life. Moreover improvements in leak detection time could be realized by changes to the existing system recirculation rate and time period over which the dew point rate of rise is calculated.
switching the gas streams between both probes at predetermined time intervals and computing the differential on each cycle (Figure 9 ) . Thus a deviation from the steady state (No Leak) situation can be readily detected. The differential monitoring scheme caters for drift of probes and electronics and if the first-rise-of-dev-point philosophy is used there is no characteristic difference between the circuit with a small leak and a healthy one with a high reading moisture from a chronic leak. The second confirmative way of detecting a leak at its occurrence besides the "high rat:- of dew point rise" is the "high differential". The criteria for a probe used in a differential system is that it should be flow independent and have an adequate response time. The higher the sensitivity of the probe, the better the signal to noise ratio, the smaller the leak that can be detected. The differential probes used need not be selected for good absolute measurement properties. The gain and response of the differential monitors should be matched to better than + 10% during calibration and this matching needs to be maintained in service. + .3
New Dewpoint Sensors For In Situ and Remote Application
A.3.1 In Situ Detectors 4.2. New Dew^ioint Monitoring System: Bulk Differential Moisture Monitoring The bulk differential measurement technique was developed to increase the overall system sensitivity, particularly to small leaks. This is achieved by alleviating the inherent features of instrument and electronics drift. The scheme developed entails
The feasibility of inserting moisture sensing probes in the bellows region of the channel has been investigated. This in situ mode of moisture measurement would readily monitor the dew point in each channel and detect a leak in a given channel instantaneously. Insertion of millimetre sized probes through the bellows attachment ring of a C N S 9th A N N U A L C O N F E R E N C E . 1988 405
GAS SAMPLE PROM CIRCUIT A1
GAS ~ *
GAS SAMPLE PROM CIRCUIT A2
(1
GAS
DEWPOINT SENSOR 1
DEWPOINT SENSOR 2
SIGNAL 1
SIGNAL 2
ANALOGUE TO DIG TAL CONVERTERS
COMMON MICRO- PROCESSOR
DIGITAL OUTPUTS
NTERNAL OR EXTERNAL GA. SWITCHING
GAS SAMPLE FROM CIRCUIT B i
O FROM CIRCUIT B2
SENSOR 3
DEWPOINT SENSOR 4 SWITCHING MAY ONLY BE DONE TO QUANTIFY A LEAK Figure 9
Bulk and selective moisture monitoring circuit
channel is considered „ mechanically feasible option. Based on the relatively high temperature (300*C) and radiation (10 R/h) fields in the bellows region the following type of detectors are currently being evaluated.
to the analyzer. Ultra-violet, visible and near infra-red light is transmitted through an encased fibre cable that has been inserted into the gas stream. Some of the light is absorbed. The light that is not absorbed is transmitted back to the analyzer where the fluid's composition is determined from the absorption spectrum.
4.3.1.1 Thermocouples and STDs. Thermocouples and Resistance Temperature Detectors (RTDs) were considered as potential devices for in situ application. They are robust and can readily withstand the temperature and radiation field in the region. With a recirculating AGS, a leak in a PT will result in a mixture of DjO vapour and CO, flowing to the detector. Under a leak situation the mixture temperature will be different from the normal steady state temperature and can be detected by the thermocouple or the RTD. Thermocouples and RTDs are based on the principle of converting thermal energy directly into an electric voltage when a temperature gradient exists between the hot and cold junction of the thermocouple. Temperature at the hot junction is determined by measuring the voltage appearing at the cold junction. Some preliminary tests were done to determine if pigtail temperature measurement could be used as a viable moans of detecting leaks into the AGS. The experimental results (Ref. 2) obtained were in doubt due to problems encountered in measuring and controlling the small CO, flow rates and in measuring the *mall temperature difference. Indications were that the very small ATs expected could not produce a viable leak detection system. A summary of the tests is contained in Ref. 2.
4.3.1.2 Fibre Optic Spectrophotometry This system consists of a scanning spsctrophotometer, a fibre optic probe that is imerted into the gas stream to be analyzed and fibre optic cables that transmit the light from the sensor 406
CNS 9th ANNUAL CONFERENCE, 1988
Using a sample of air saturated with 0,0 at 23*C a near infra red spectrum of deuterium oxide vapours were recorded. The spectrum (Figure 10) was collected using a spectrum analyzer (Guided Wave Inc. Model 200-40). The instrument was purged with dry nitrogen for these measurements and the vapour was prepared by drawing a stream of dry air over heavy water to saturate it prior to entry into the gas cell. The spectrum of the light water vapour is overlaid on the same chart of Figure 10 for comparison. The spectra of light and heavy water are essentially identical in shape, although there is a significant difference in amplitude. The capability of the fibre optics to withstand the radiation fields and remain clear is currently being investigated. If the fibres cannot withstand the fields and this detection system is proven viable, consideration will be given to using detector tubes to bleed the CO, gas to a less harsh environment where the fibre optic probe can be located.
~r~ i" i ~ i i i i i i i \ 1206 1267 1333 1400 1467 1533 1680 1667 17J3 I860 IU? 1933 2MWf
IWPM or m AND m 1 (era FATHLEHCTH Figure 10
Absorption spectra of D2O and H2O
A.3.1,3
Ceramics Humidity Sensors.
The principle of these devices is based on preliminary tests which indicate changes in the capacitance or resistance of some materials as a function of their moisture concentration.
reactors. New CANDU reactors will continue to emphasize the importance of the annulus gas system for leak detection. Current developments of an upgraded detection system have focused on (1) an improved AGS configuration, both on and off the reactor face, for a minimum leak detection/location time,
These sensors have exhibited good radiation resistance characteristics with minimal effect on sensitivity. Evaluation tests continue to provide the cumulative effect of radiation/temperature to the sensitivity, accuracy and drift characteristics of the plates.
(2) application of in situ and remote moisture measuring devices and (3) a new improved moisture monitoring system for increase system sensitivity and improved performance.
A.3.2 Remote Application Detectors. Based on the performance requirements of the improved AGS a review of sensors was carried out with respect to the following defined specifications. o Accuracy: +2°C for the dew point range of -70°C to +20°C o Long term drift: calibration to be carried out once every A-6 months.
6.0
ACKNOWLEDGEMENTS
The development work on improved leak detection system was carried out in collaboration with Mr. G. Soraogyi and funded by the CANDU Owners' Group under WP6A09. Feedback on AGS configuration and moisture detectors specifications was provided by P.J. Ellis of Ontario Hydro. Some of the data presented in the Figures were obtained in experiments funded by the CANDU Owners' Group under WP690.
o Speed of response for wet up and dry down scenarios to be within +2°C in 3 minutes and j^2°C in 10 minutes respectively. Potential sensors with suitable specifications are currently being reviewed, and tested in reactor conditions to determine their suitability. 5.0
CONCLUSION
Leak-Before-Break is an established methodology for pressure tube failure in CANDU reactors and is incorporated into the operating procedures for power
7.0
REFERENCES 1. PRICE, E.G., MOAN, G.D., COLEMAN C.E., "Leak Before Break Experience in CANDU Reactors, AECL 9609, 1988 April 2. CHAN, A.M.C., "Temperature Measurements in Pigtails" memorandum to P.J. Ellis, 1987 June 18.
•h
C N S 9th A N N U A L C O N F E R E N C E . 1988 407
Session 12: Current Issues in Nuclear Safety
Chairman: D.A. Meneley, University of New Brunswick
CNS 9th ANNUAL CONFERENCE, 1988
409 .
CURRENT METHODS IN QUALIFICATION OF CANDU HEAT TRANSPORT PUMPS FOR OPERATION UNDER LOSS-OF-COOLANT ACCIDENT CONDITIONS
A.N. KUMAR
Atomic Energy of Canada Limited CANDU Operations Sheridan Park Research Community Mississauga, Ontario L5K 1B2
ABSTRACT The Heat Transport. (HT) pumps in a CANDU reactor circulate heavy water coolant through the reactor core •inder all normal operating conditions. For a I,oss-of-Coolant-Accident (LOCA) in the heat transport 5 ystem, depressurization of the system occurs. It is d?sirable to establish that the HT pumps will continue to operate under two-phase flow conditions resulting from a LOCA for a short duration. Evaluating the operational behaviour of large centrifugal pumps under two-phase flow conditions is complex. Various techniques have been used with different degrees of reliability and associated costs. This paper presents the current methods used for evaluating the post-LOCA operational capability of CANDU HT Pumps. Conclusions from the recent research studies suggest the possibility of using the LOCA test data generated by Ontario Hydro for qualification of future CANDU HT Pumps. A methodology for such a cost-effective approach is also presented.
1.0
INTRODUCTION
The main function of CANDU heat transport (HT) pumps is to circulate heavy water (D 2 0) coolant through the reactor so that the decay heat from the reactor core is continually removed. For any Loss-of-Coolant-Accident (LOCA), either due to a pipe break or due to spurious failure of a valve in an auxiliary HT system, the HT pumps may experience very low suction pressure or varying levels of void. From the safety consideration, it is desirable to have the HT pumps capable of operating for a specified duration even under LOCA conditions. During the post-LOCA operation of the HT pump-motor set, the severity of loading is mainly on the pump components. The general design approach is to ensure that the most critical pump components, i.e. the pump bearing and the mechanics 1 seals, retain their capability throughout tiie pump operation under LOCA conditions.
The CANDU heat transport pumps for all the Ontario Hydro nuclear stations have been qualified for post-LOCA operation by testing full scale pump-motor sets in a test loop. Thus the capability of the pump-motor set to operate under various postulated LOCA conditions was quantified and demonstrated. The qualification of Bruce 'B1 HT pumps for post-LOCA operation represents a typical example of high costs involved (about $2.5 million - 1982) to produce reliable and conclusive data (Reference 1 ) .
On the other hand, the heat transport pumps for CANDU 600 nuclear stations were shown to have adequate capabili cy of operating under LOCA conditions by analysis (Reference 2) . This approach involved building a sophisticated computer model of the pump-motor set and evaluating its response to various dynamic loads simulating those under the LOCA conditions. The reliability of the analytical results are largely dependent upon the accuracy and validity of the various assumptions made. Cost of such an analysis is typically an order of magnitude less than that for qualification by full scale pump tests. Recent research studies (References 3,4,5) have concluded that (a) the affinity laws are equally valid for either a single or two-phase flow operation conditions and (b) the test results from a scaled down pump model can be used to rredic~ the performance of a full scale pump even under two phase flow conditions. However, the two pumps should have similar hydraulics, rotor dynamics and the pump support arrangement. Thus the existing two-phase flow data from the Ontario Hydro full scale pump test program can be combined with the relatively low costs involved in a sophisticated analysis to qualify the next generation of CANDU HT pumps. This paper presents such a technique for qualification of future CANDU HT pumps. A summary of existing methods used for qualification of CANDU HT pumps is also presented.
2,0
DESIGN VERIFICATION OF CANDU HT PUMPS
The heat transport pumps used on CANDU reactors are all vertical single stage, single suction double volute centrifugal pumps. The Bruce and Pickering HT pumps have a single discharge while the HT pumps on all other CANDU reactors have a double discharge arrangement. The CANDU HT pumps were supplied by three different pump manufacturers, with Byron Jackson pumps providing about 35% of CANDU HT Pumps.
The HT pump is connected to a vertical electric motor by means of a spacer type coupling. The pump motor is provided with two radial and two thrust bearings. The motor bearings on al] CANDU HT pumps are oil lubricated tilting pad bearings. The motor
CNS 9th ANNUAL CONFERENCE. 1988 411
speed is e i t her 1 BOO rpm or 1500 rpm depending upon the supply frequency (60 Hz or 50 Hz). The pump motor set is ei ther rigi dly supported ai the futfiip rasing (Figure 1) or the pump discharge pipes and directly connected to the Reactor Inlet Header (RIH) (Figure 2} . In the latter case, the RIH is
3.1.1
Fump internal components such as: the pump bearing journal assembly the- mechanical seals
3.1.2
All thu motor bearings
3.1.3
Tump s uppor t. arrnngemont attachment bolts.
including
3.1.4
Pump bowl and
chment bolts.
the
The i terns 3,1.3 and 3.1.A are structural comporcnts whose integrity was established by stress and fatigue analysis to meet the respective ASME Code allowablps for accident condition.
MOTOR MOUNT VENT
SECONDARY CONTAINMENT
3M\ L I T / S MOTOR MOUNT DRAIN ^|§=| IJ /Q (io leakage collection)
The capabi)j ties of item 3.1.1, to a large extent, and of item 3.1.2 have to be established for successful operation of the HT pump motor/set. Hence a brief description of the pump bearing is provided before proceeding to the LOCA qualification methods.
3.2
Types of CAN'DU Pump Bearings
As already mentioned, the CANDU HT f.umps incorporate either a hydrostatic or a hydrodynamic hearing. A brief description of each type of bearing is provided. 3.2.1
Hydrostatic Bearing
This bearing design feature is unique to the HT pumps supplied by Byron Jackson Pump Co. The metallic
FIGURE 1 BRUCE B HEAT TRANSPORT PUMP SUPPOR'l ARRAKGEMENT
anchored by a number of pipe restraints to the building structure. In addition, the mass of the motor (about 50,000 kg) is supported by two spring hangers which are attached to the steel structure, surrounding the pump motor set and embedded in the building concrete. The motor is also provided with seismic restraints (either 3 or 4) to accommodate the earthquake loads. The pumps are designed and manufactured to ASME Code Section III, Class 1 requirements. The internal assembly for a typical CANDU HT pump-motor set is as shown in Figure 3.
3.0
LOCA QUALIFICATION PRINCIPLE
In this section the genera 1 principle and controlling parameters used in LOCA qualification of a CANDU HT pump motor set are presented.
3.1
Critical Pump-Motor Components
During the post-LOCA operation of the HT pump motor set the severity of loading is mainly on the pump components rather than on the motor components. The main aim was to establish that these components, together with the other items listed below, retained their design capability to enable the HT pump operation under LOCA conditions. These components are:
412 CNS9fh ANNUAL CONFERENCE, 1988
FIGURE 2 TYPICAL 600 M W HTP SUPPORT ARRANGEMENT
BRAKE DRUM . UPPER GUIDE BEARING PADS J l u *- " UP THRUST BEARING PADS -•/ \L~« D O W N THRUST BEARING PADS - 5m
M O T O R SHAFT INERTIA PACKET • ROTOR ASSEMBLY
INERTIA PACKET
io'crLOWER GUIDE BEARING PADS
2m
pulsations. These pressure pulses are due to alternating slugs of water and steam entering the pump. The frequency and magni tude of pressure pulses depend upon the temperature {50°C - 265°C) pressure (<2.5 MPa) and void fraction (0 to 90%J existing at the pump suction. There are no reliable analytical methods which predict the magnitude and/or frequency of these pressure pulses. The pulsating load acting on the pump impeller causes a dynamic unbalance of the pump-motor rotor assembly. This results in high vibrations and pump shaft run-outs, their magnitude dependent upon the severity of unbalanced loads. Thus vibrations and pump shaft run-out are indirect measure of the severity of two phase flow loads on the HT pumps. Hence, the vibration levels experienced by the pump motor set was taken as the major governing factor in qualifying the CANDU HT pumps either by full scale pump tests or by analysis.
SPACER COUPLING
HYDROSTATIC PUMP BEARING PUMP IMPELLER
FIGURE 3 HEAT TRANSPORT PUMP MOTOR ROTOR bearing is located just above the pump impeller and is energized by the pump head. The D z 0 from the pump discharge is provided under pressure to the bearing through 8 pockets around the bearing diameter. The bearing journal is also metallic and is shrunk fit on the pump shaft. There is a large diametral clearance between the bearing and the joarnal (0.86 to 0.96 mm). Under normal operation, the inherent self centering principle of this bearing coupled with large diametral clearance ensures that metal-to-metal contact does not occur. Under two-phase flow conditions the pump head whicii energizes the bearing may not exist. Hence, metal-to-metal is likely. However, the journal and the bearing material arc compatible to avoid galling under this condition. The bearing details are shown in Figure 4,
3.2.2
MAIN IMPELLER FIGURE 4 BYRON JACKSON HYDROSTATIC PUMP BEARING
Hydrodynamic Bearing
The hydrodynamic pump bearing is of carbon-graphite material provided with a single helical groove (Figure 5 ) . This bearing relies on the supply of a minimum specified quantity of D s 0 being provided at all times and at a controlled temperature for its successful operation. The bearing clearance and stiffness in a hydrostatic bearing are higher than those in hydrodynamic bearing by a factor of about 2.5 and 6 respectively. Under pulsating loads during the two phase flow, a significant water film in the bearing is necessary to provide the damping and cushioning effect. Hence, a guaranteed external supply of D z 0 at controlled temperature has to be provided to the bear ing under two-phase flow conditions. The bearing load capacity under the highest LOCA loads also should be evaluated to provide adequate bearing clearance and the rotor dynamics.
2.3
TOTAL FLOW ~ 0 0 3 m 3 ' s e i
GLAND INJECTION INTO PUMP CASE
3 25 Igpm
Controlling LOCA Qualification Parameters
When the HT pumps are operated und^r two phase flow conditions, the pump impeller experiences pressure
FIGURE 5 CANDU 600 HEAT TRANSPORT PUMP HYDRODYNAMIC BEARING CNS 9th ANNUAL CONFERENCE, 1988
413
4.0
LOCA QUALIFICATION METHODS FOR CANDU HT PUMPS
A. 1
Qualification by Full Scale Pump Testing
assessment was based on ability of the pump to continue operating for 15 minutes following a LOCA before being tripped by operator action.
This method involves testing the actual pump-motor set in a test loop simulating the designed pump support arrangement as closely as possible. The test loop permits the pump operation at the rated design conditions. The temperature and the pressure in the test loop can be varied during the tests. The water can also be drained from the test loop until void conditions are produced at the pump suction. The pump motor set is instrumented for various measurements such as: the pump suction temperature and pressure, volume of liquid drained from the loop, the pump shaft run-out at the pump coupling and frame vibrations on the pump and motor.
D
SATURATION
FIGURE 7 FRAME VIBRATION (TOP OF MOTOR) VS. SUCTION CONDITIONS
A typical test involves running the pump at the normal rated conditions. The water is then slowly bled from the test loop over a period of about an hour. ThiJ produces a slow transient in which the pump suction pressure is gradually reduced until it reaches the saturation pressure and then continues into two phase flow with progressively increasing void. This voiding is continued until the pump performance completely degrades (i.e. no flow through the pump). The loop temperature and flow control valve setting are maintained constant, within the practical limits, for each test.
SHAFT BUNCUT imm PEAK TO PEAKi
-06
*
100
MD'C
/ -OS
03 / Wfl*C
A series of tests is carried out for various valve settings and for various constant temperatures over the entire range say 50°C to 266°C.
-0 2
r
100%
SMPl
PUMPSUCTICN
The full range of postulated transient conditions arising from various postulated LOCA's (predicted by SOPHT code - Reference 6) can thus be evaluated. The pump performance curve, quantifying the pump operation in the high vibration region, corresponding to each constant test temperature can be established. Hence, the operational capability of the HT pumps for any predetermined LOCA transient condition can be quantified.
FIGURE 8 SHAFT RUN-OUT AT THE PUMP COUPLING VS. SUCTION CONDITIONS The Ontario Hydro test data was on HT pumps with: - single suction, single hydrostatic bearing.
discharge
with
a
The pump motor set was rigidly supported below the casing (Bruce HT Pumps)
Ontario Hydro have Completed a range of LOCA tests quantifying the post-LoCA operational capability of Pickering, Bruce and Darlington HT pump motor sets.
- single suction double hydrostatic pump bearing.
discharge
with
a
The pump motor set was rigidly supported at the Typical pump performance curve, vibration and the pump casing (Darlington HT Pumps) pump shaft run-out variation with the pump suction conditions for Bruce ' B ' HT pump motor set obtained - single suction single discharge with a from the OH tests are shown in Figure 6, 7 and 8. hydrodynamic bearing. The pump was supported on Further details of these tests are given in an anchored header and the motor by beams of two Reference 1. In this case the pump integrity hangers (Pickering HT Pumps). 1 HEAOrm) f IN HEIGHT OF LIQUID |
AT TEST TEMPERATURE
4.2
;co
r* 1
This method of qualification involves evaluating the response of the pump motor rotor assembly to the pulsating loads experienced during the LOCA conditions. The capability of the pump bearing in particular, and entire rotor bearing assembly, in general, to withstand such LOCA loads has to be established.
i1 m i
•100
m
—
—
— " " " " " " " " "
Qualification by Analysis
100*C(TEST 137|
^•'fy
The following steps are involved in the analysis:
100%
5UP]
PUMP SUCTION
SUB-COOLING A T BUMP SUCTION
FIGURE 6 PUMP PERFORMANCE CURVE 414
CNS 9th ANNUAL CONFERENCE, 1988
- Developing a sophisticated mathematical model wherein all the masses, bearings and supports are represented.
predicting the magni tude and frequency of pressure pulses generated during LOCA conditions are not yet available. The pressure pulse magnitude for the CANDU 600 HT pumps were estimated from the Ontario Hydro tests on Bruce HT pumps.
- Calculation of the pump and motor bearing stiffness and damping coefficients in required directions preferably under LOCA conditions (i.e. non-linearities to be included). - Critical frequency rotor assembly.
analysis
of
the
pump-motor 5.0
- Unbalance response analysis due to an oscillating radial force acting at the pump impeller over a frequency range of 0 to 33 Hz. The load and frequency of the oscillating load is considered to be r e p r e s e n t a t i v e of the w o r s t load experienced during the two-phase operation of the HT pump.
FUTURE TRENDS IN LOCA QUALIFICATION OF PUMPS
Recent studies in Germany and the USA (References 3, 4) showed that scaling down the size of the pumps while maintaining the same specific speed produces very similar performance characteristics both in single and two phase flows. This is reinforced by Japanese studies (Reference 5) which concluded that the similarity rule is equally applicable to the performance of centrifugal pumps under single or two phase flow conditions.
The CANDU 600 pump motor set similar to that at Point Lepreau Nuclear Station was analysed by this method. The analyses showed that the pump bearing maintained adequate clearance even when the pump head degraded up to 30 percent of its rated value. On further degradation of head, the vibration amplitude exceeded the bearing clearance resulting in metal-to-metal contact between the bearing and journal. The summary of thr unbalance response results is presented in Table 1.
Thus there is an increasing trend of utilizing the test data from scaled down pump tests to full sized pumps. Analytical techniques are also being used in combination with test data from scaled down pumps.
5.1
The unbalanced load was that corresponding to 7 percent vapour content by volume and was considered to be acting at a frequency which produced peak shaft deflection.
Conclusions
Recent research studies clearly suggest the possibility of using the existing data from the Ontario Hydro full scale pump test for the following:
Non-linearities in estimating the bearing coefficients were not included. Hence all values above 50% head degradation may not be accurate.
5.1.1
In order that the Ontario Hydro test data is directly applicable, the new pump hydraulics, rotor dynamics and the support arrangement should be similar to one of the qualified O.H. HT pumps. This aspect should be considered in the design of future CANDU HT Pumps. A potential cost saving in the range of $2.5 million is possible.
It must be noted that the pump bearing and journal material are compatible so that the galling or seizing of the bearing is unlikely even when metal-to-metal contact occurs. Furthermore, the impeller loads are grossly reduced when the pump head degrades to 50% of the rated head. This is because the loads imposed by the vapour are much lower than those imposed by the liquid.
5.1.2 Thus CANDU 600 HT pumps with hydrostatic pump bearing were considered to have adequate capability for post-LOCA operation. This method of (about $250,000).
Qualification of Future CANDU HT Pumps
Qualification of CANDU 600 HT Pumps
Pickering HT pump LOCA test data may be used for qualifying the CANDU 600 pumps with hydrodynamic pump bearing. Similarly the Darlington HT pump LOCA test data may be used for qualifying the CANDU 600 HT pumps with hydrostatic pump bearing.
qualification is relatively cheap However, reliable techniques of
Table 1 Unbalanced Response Analysis Results
"Va" (Predicted Vibration Amplitude at the Pump Brg.) x 0.025 cr.m
Pump Head available for bearing
"Cr" (Designed Bearing radial
Amplitude Ratio
Rotor Dynamic Evaluation
energization
clearance) X 0.025 mm
100% rated
17
2
0.12
Excel lent
50%
17
8.5
0.50
Good
30%
17
15
0.8B
Marginal
25%
17
17
1.00
Poor
2 OX
17
20
1.18
Bearing Rub CNS 9th ANNUAL CONFERENCE. 1988 415
In both the cases, the forcing functions and the forcing frequencies corresponding to various LOCA conditions may be obtained from the O.H. test data. Using this data, the LOCA qualification of the HT pumps could be carried out by a sophisticated analytical technique. This semi-emperical approach is cost effective, reliable and quick.
ACKNOWLEDGEMENT The author wishes to thank AECL Management for giving permission to prepare and present this paper at the conference.
REFERENCES
4. Kastner, W.. Seeberger, G.J., "Pump Behaviour and its Impact on a Loss-of-Coolant-Accident in a Pressurized Water Reactor". Nuclear Technology, Vol. 60(2), P.268-277, 1983 February.
2. Allaire, P.E., et al, "Heat Transport Pumpset Analysis of Response to Dynamic Impeller Forces". Final Report, VIBCO Research Inc. 1988 February. 3. Kamath, P.S., Swift, W.L., "Two-Phase Performance of Scale Models of a Primary Coolant Pump. Final Report". EPRI-NP-2578, 1982 September.
416 CNS 9th ANNUAL CONFERENCE, 1988
5. Fujie, H. et al, "A Study of Applicability of Similarity Rule to Performances of Centrifugal Pumps Driven in Two-Phase Flow". Nuclear Engineering Design, Volume 85(3), P.345-352, 1985 March. 6. Chang, C.Y.F., Skears, J., "SOPHT-A Computer Model for CANDU PHWR Heat Transport Networks and their Control". Nuclear Technology, Volume 35, PP.591-602, 1977.
\ 11*11-1 I ' K H I / K I ) MONITORING SYSTI-IM FOR KMKRUKNn (AIIM,
roMUMuNS HAHHTr A.H.
VI 'IHF: P O I N T
|'|.\M
LKHKKAl' (JKNKIiATlNU
SI-YIIu\
Mt>K.F.\ JOHNSON
D.L. STAFFORD
New Brunswick H l e r t r i c Power t 'onuiii s s i on Point. Lepreau Generating S t a t i o n
BHIi-.lf.rke am iiftu-r a n n u n c i a t i o n program, the monitoring tin1 SvmpLoms-Urient.ed lunerge.ncy Opera) ing Procedures 1fcXJP1s) i s hcinu incorporated in I he Stat. ion c o n t r o l computers ( DlX's ) . I ho pros'rams: - I'ro\-uk' I-Vent Kecogni tjon and a n n u n c i a t e tho in i i i r t i ' t i t f • 111 a w|H'C l a I "t% i n d o w " \ n i i u i i i - i . i i niti
()K IMF Klip S\S'IKM
Poirjt J,pjjreau'.s Kmereency uperating F'rotfHJuros ex i st jn !.' - Z . •> < ni ( 1 " I .i-r i iiii hi nd< •!".. utn • hi rider has boon provided for- each BJl'. \ proamblf *•:•• i s t ^ iii each section that is fn! loufii 1A "l.oii I ^rvun:;" (!• i-^uri If, which in turn a r e t'oliow&i bv "JARs".
o n I he*
fKT's.
- Monitor and aiuiunciate when ini|*">rtant c o n d i t i o n s ini IJIIK- i mat .i-|ii atilo t o r any given HOP a f t e r t ho t-vtMit has been acknowledged by t h e o p e r a t o r . - IV;.'1. ido thn c a l c u l a t i o n s and d i s p l a y of parameters r e q u i r e d t o p r o p e r l y monit,or the event in p r o g r e s s . - Provide d e d i c a t e d S t a t u s Displays for each EOP. a s welt a s " r h a i n s " of o t h e r rovolant d i s p l a y s . - P r o v i d e o p e r a t o r c o n t r o l over what i s t o be ••[ i si*I. !•• cd and what, important oond i L i ons a r e t o bo monitored.
I \IHt>IMiTlct\ l-jiiergoncv o p e r a t i n g Procedures (KOP's} have been developed for the Point Lepreau Generating" S t a t i o n and have neon approved for operational u s e . The development of FJOP's lias been on-going since t h e TMI-'J accident in 1979 and more emphasis has been nlaced on W)P*R s i n c e the Ohernobvl a c c i d e n t in lffHfi. SuiiK' of t ho monitoring ami r>rocedura I .'isuects of the wr i 11 on K.( )P' s can nou be computeri zed and made .-.niinlor and more e f f e c t i v e with the use of Mir st^-it j I.-U i.'onl.i'oi i.'ompiitor's for the benefit of the Control Hoom Our-rator. KiL-ht KUP's h a \ e been .Joveloped t o d a t e (Table I ) , the (.ioncrif W)P i s not y e t complete, Kaoh of t h e s e wr 111 en H )(''•. i-ont-ui ris t.ho «*ond i I i UDK roqui red to reroeni ;:e a a j \ e n e v e n t . For a l l of t h e BOP'K , t h e r e art- ;if>pr!j\imat,ely )()0 parameters and subse*juent combi n a t i o n s of t h o s e parameters t h a t must be recotfniBed bv the oue.rat.or t o deduce which EOP must be used. After an KOP has been i d e n t i f e d , t h e o p e r a t o r must take the ;«-! iunx I h a t a r e iliven in the -sj»ecific R')l''s, and lie must c o n t i n u a l l y monitor c e r t a i n "
[.o<»ititrams. Lo^mrnms arc diagrams that pro*, ide: - The set of alarm conditions that must e \ j s t to invoke the I\OF - those art- known ;\y. F.nlrv i ondit ions. - ijovorninii Condition diamonds, which r e q u i r e condi t i o n s t o lie cont iriuaJ I v mon i torvd anij 11 "in alarm", i n t e r r u p t the o|>orat o r ' s presi-ni a c t i v i t i e s and force him to take c o r r e c t i v e actions. There nvi\' be a number of valid Governing Conditions in effect a t anv time, and they a r e assigned s^iecific p r i o r i t i e s nn each Logi^ram page. - tether diamonds which r e q u i r e monitoring and subsequent d e c i s i o n s , hut a r e onlv addressed once. - The panel and instrumental ion devices that must be used t o a s c e r t a i n what t h e a c t u a l plant condi t.ions a r e 1 o answer1 t he q u e s t l n n s provided in the Ijecision Diamonds. - Action Boxes t h a t d i c t a t e o p e r a t o r a c t i o n s t h a t a r e required a s the r e s u l t of c e r t a i n d e c i s i o n s t h a t have been made. TAIts. TARs a r e simnlv **\t ens ions tu the l.oiii^rams. They d e t a i l the more complicated operator a c t i o n s and remove t:J ut t.er and potent, ia 1 coril'us i on from I In* logigrams.
Loeigram llierar<"hy. l.uEjisrams h a \ r .i h i o r a r c h i i a I order. The Governing Conditions at the ton of t h e pace take, ijrocodonce over those that a r e |owrr on t hi page, (ffn-f a nasro has been changed. Governing Conditions from t h e formor- page no longer- a p p l \ . If a Governing Condition from a former page i s r e q u i r e d , i t i s copied onto t h e n e \ t page, hence tho need t o turn back pages when e\ecut.ins ;m B>P i s ej iminated. If >* h i g h e r level (icnerning Cond i I ion changes t o t h e A1,ARM s t a t e , t h e o p e r a t o r must stoi< what he i s doing and a t t e n d t o t h e higher l o \ e l Governing C o n d i t i o n .
TARLF 1: LIST OF EMRKGENCY OPHHATiNG a)NOIT10NS VIA' 1 2 •i I "i H 7 H H 1
nTT.K (ieneric iXJP (not yet complete! Dual Station Control Computer Failure * Loss of Feedwator Loss of Instrument Air Loss of S e r v i c e Water I.OS;K of Class I K l e c t r i c a l Power Larsje Loss of Coolant Accident Snt'i I I Loss of Coolant. Accident. Steam (jonorator Tube F a i l u r e
W ft-' 'J. r a u i i u t t .'onI i ' o |
b e imp lemon I rtI o n t h e :->laliun
i .Diiijiut.ci—. for
ob\ i o n s
rnnsoii!-..
CNS 9th ANNUAL CONFERENCE, 1988 417
Annunciat ioti M s l c n i . I
OR
|
C!
When an aUi-rn t-cruti I tut) j.s
det.oc'tecJ, i t I M - / H t m l t o >i i n n * on t h e a n n u m : i a U o n (.'HJ' w i t h an as.soc iat tnj f l a s h i n g " \" at th<- I n u i rm : r:u
.OSS
7 - 1 2 PS SOT IKSTJt AlN PRESSURE V
of t h e l i n t * . At t h e same t.inu- tht- aiuiuriciat i o n tiorri sound:-: ;m a l a r m ^ o n d i l i o n . Uhf'ti an a f a r n i • u n d i i j i j n
P L U 1 CUE OW MORg O f
clears,
U K * "A"
is
tunwii
j n t - j a I l a s h i MJ
t h e atinujir- lat i on hot'n xuunds a r e t u r n cond 11 u r n . —•"•
«•'«
To ?tckn
t h e h o r n hv p r r - s s i n i l
butt.on.
Klashinu:- " \ ' s " An-
with
V'KN'ulVtJ-.rK.K
Hashing
the operator
o oooc oo o
;m ;I|;IJHJ,
sileni-p
t h e Al.WM
Kverv
it
i h ( ' i .[x-rat o r inu:-.l
! w-;i
t h e ||( iHN S H . K M K
nush-
then :,tu[>p.>d h'.
pushhuf t m,.
is
tht- Vl.AK'M KKsKI
satin. will
lodged on t h e I x V ' s
These d e v i c e s .
been a c k n o w l r d ^ e d
iitim
rate.
t u ntjrnvil
occurs,
HUkN S I M M r ,
1970's.
to
system a t
and t h u s
t h e programming
they a l l for this
V a n a n V-73 assembly
KF.VSONS FX)H U S I N G Deficiencies
Instrument
- \ i r l.oqieram
operator
of
LKVKL LO.
exist
method of Haidwai,-.
Iwo ( ont.rol
i omputers c o n t r o l
M-:'iv:m iji-iti •\-:\\ n i i ; M a t ion . lxf"t.
ixifV. \
.Norifti I J \ , ] * v \
is:--uin«s l o n t r u l
uhen IX'CX
i s unavailable?. smaller in
Analog e\ist I'hf
:>ii') I'NJ's p r o v i d e almost a l l
interface
for
the svnt.om.
I n p u t s and in e\cexs of on the sxstein t o i, >ri i t or
DieiUil
IH ' . U I H I I Contact
Inputs are d i v i d e d
plant
i •-, i-onrK'f-t.ffi t o arils
: HThnliis.
418
Or
-
or
i •.' - i it •
on t h*
PnriuH i f
.r-.
..("T if')/
i n j c s i p o n s i . 1 t o (}}XTri1.or' k"W«/.ir
CNS 9th ANNUAL CONFERENCE, (988
to h r « i n
conditions
is
alarms are of
Mien upset c o n d i t i o n s
it
npressarv to be in
is
the
o p e r a t i n g at
surh as a
alarms are gentakes
to determine the cause of respond to
the
This
an cmer-tfenev
occur,
hundreds of
Under such c o n d i t i o n s , to
indicate
the s i t u a t i o n
time
the
for
trip,
and
correctly.
I n a d d i t i o n , when a large number of p o i n t s art m alarm, i t becomes d i f f i c u l t to discern t h a t , for example a Loss of instrument Air l'Jivmonc-\ Condition exists.
time.
1
(such as whf*n
. m - i-i^iiir-i'il i n out |JU1. messages), and Demand
• •! • '•-• i 'mi
erated.
tlie
the
Inputs
ono IX'C at. a
programs e x i s t
pouer.
the operator
1 SOU
conditions.
i oiniH,scrs and Demands.
ITOKJVUIIS .lutoiiwit i ivi I I \
•• l i
the man-
two c a t e g o r i e s
''..n>!»j-,I-I • in • mi .mis «••.(•! Hi.- when r(H|Uirt'd Hi--
of
I n p u t s which a r e monitored by a separate
->• >t t ware. Thrt-H' types cji
plant
The alarms are a l s o designed
Shutdown System T r i p ,
These
In excess of
lifiOO D i g i t a l into
invar. that
a r e i unnoc I F»I] t.o each IX.V, and i . ' l ' s
seanm r ' i h i i j i
I" '
Two
CRT's
the sysleiu.
inform
the alarm t o the operator-.
non-alarmine stat.e when the s t a t i o n
IX'CT
to
such as DiiO SIXJR.-\(ib; TANK
is
in control.
present In p a r t i c u l a r ,
designed
nature.
have a s s o i l a i ed lu vhoards rv;ist kt'\lr.),tni->
the
has heen itkuie t o
alarming plant
because not a l l
,\1 i in
Annunciation.System.
wilh
T l i e i r desi gnat ions ; i r r
rUtfnuK*i;i( ion i'Rl ' : - , , and several
machine
point
such
atvi
K ) K KOI' M O M l O l i | \ ( J
system.
a s i n g l e event, No attempt
s i t i n i f icance o f
Svstem,
programs were not
liari t o be d e v e l o p e d .
With The Present
messages are d e l i b e r a t e l y Loss of
Ihest'
svstem hits been done
THK I X ' C ' S
Various d e f i c i e n c i e s
]:
that- h a w been
lan^uaee.
s i n g l e - t i e r e d annunciation
l-ikii-.h (JKSCKlfilu.N uK TUK STATION CONTROL COMPUTERS
t lie items
Point
i n I he e a r l y
t h e KOI1 M o n i t o r i n g
llouevt-r.
fchen
J l^sh al
\\ ] I 1 uo o u t .
was . i o s i ^ n e d
some o f
implement
and
by ALARM
['he computer rit.ul.jon
r
available, of
followed
t h e MENUS program mav seem conunonpiace
obvinush
:i},tsm
lit.
litiht
Consequent 1 \
installed
rem.in
t h e window'--;
Dated T e c h n o l o g y ,
as
cfr't-'nu
-t wind^u
(he t-. iriiiow u i i l
When HORN SILKNTK
Leprcvui tfrticrnliruZ
for
Wht^n t l j e a l a r m c o n d i t i o n has
by p r e s s i n g
RKSf-T a r e prt-sKc<%i,
\u
printer.
provided
:\LAKM ACKNOW1.KIXJE t h e wjndciu w i l l the alarm returns,
pushbut.ton.
each h a w an enaraved mes-
When an a l a r m r c j n d i t i o n f l a s h 'imckly
a slow
pressing messaires
nii.o a l a r m o r r e t u r n s
Uiiuiow a l a r m un i t:-. . • r • t - a l s o alarms.
htnnl\\,
" N ' s " ; i r e i-l jininat ed frr^wi the CKf when presses
t ime- a iro ssacn cniw"-.
normal,
ke\
!• i sure
" N " , and
rcvfiK'S ts
It is not cvisv to reduce the number of alarms hecause there are u s u a l l y very good reasons f u r each find evor> alarm. Kor e\ample, alarms are o f t e n r e quired to perform post mortem a n a l y s i s of an event. Also, i t i s not easv to e o n d i t i o r i out nori-ossenl i.t I alarms under c e r t a i n plant o p e r a t i n g s t a t e s such as iJurini; shutdouriH, tiec-ause f o r i^amplr, shutdowns are the only time that some alarms e,ui be l o s l e d . Alsu, under Kmeriie'io T o n d i t i o n s , i l i :•; not easv (a d e t e r mine w i t h ahsuhitc t orUi'mi t.y wbil alarms w i l l he generated d u r i n g the emervlt^ru-y,
^ i Mi t tic- j , i i - y iiirnit •• -r 111 11> J u! -. ' a a l a r m u n d e r u p s e t L>r sluiiiiouii . mid i 1 i m.-. xi l i e s i rattle- l o h a v e t h e COMIJ 'ill f\- nic-n 11 ci c c r i a i n («,• >\ a ( a r m s a n d t o g e n e r a t e " h i l i h o r - ie\.< •! " a i a r i i u . M.ci, • c c i . i i n i M -iinnl-a t i o n s a n d c o m b i i i a t i o n - . •'' ! mi- . e n d it u u i s c \ i M . l o r e x a m p l e . Mien .i [ v i r t n - u l a r i •;iuhifi?jt i o n oT ].i-u I n s l rumen! \ i r Pre.-.sm ,• .ii-irin^ I - M S I t h r o u g h o u t t h e s t a t i o n , i t r a n :» i-Mil! I \ di-duced U (ut*K- I ha l a L o s s o f l u s t r u men I J.ii' ix-i id i t i o n e \ i s | > . e t |K-I \c. *ri».(cin\i Jo Cnjiijujter i_*-:iIJ£_EOP' s . When tin- di-cc-.u.ii u a s made t o u s e c o m p u t e r s t o a s s i s t , t h e o i - c i . i i . : i i n inonit.-i n i i (-jiiej-tfeiK ^ i until t i o n s , i t was CC-IK lude.l t h a t t,i he- u s e f u l , it was ne< i > s s a r \ t o nidi i ; nj J i M1- p l a n t c o n d i t i o n s , a n d t h a t i t w a s v e r y di-s i r.d>h - Dun ? }jr- in-ii -.\:-.! .'in would b e i n t e g r a l ed
d e a i i l ^ into the Control Room. Attempts had been made t o implement the systems iri stand .ilorh' compnipi -. that ut j t nut couneiiod to the o n - l i n e comput t-rs .
!I K - ^ - m - i - i -
i i . h t ' t i ' i i i l •• i l l t ' f i c u l t
t o u s e
i i \ .tiLse uf the neerJ j'ur the operatur to continously etil ei
1
t- \ i . T ! ! t i t ;
|.!.:h'
. i - n d i I i • jfiK
i l i t (.-
t h a I. l l l a c h i n e .
!t i s a n t i c i p a t e d that in the f u t u r e , an Kthernet I.*.-, a l \i •. •, i '.(•• i-1 TK .v: I 1 tit- i n s t a 1 1 ed \ La a y;a t eway coiuputer UJ pi e\ ide J i \ e d a t a lo cNternai computer .'•:-.: t-.'iis. i:ii .sucjj a svxi.fMii h.as no I > r t been implcmeiiieii. I'urt.hennore, t h e s e l e c t e d a]>proacli hns t h e adxarn.asi'e of heirul on Lht? computer system thiat t h e o p e r a t o r use;-, a l l ihe tune !u c o n t r o l the s t a t i o n . ^vs.teni. Reyujrejut'iiLis Foi Jl»e BJ^' Monitoring. Svsteni.. The programs t.liat h a \ e been pro\ uleci to d a t e for monitoring 1 £).")>'*s can e a s i l y riui on the S t a t i o n i,ontrul tumputera - the- l o i a l memory rE^qmred i s about 6-1 ki lo-bvi---. u ^ U t l U1110 UOO wordsl. AlJ of t h e memory • ist'd J\:J t h e new .system i s r e s i d e n t in the Bulk Memory Lnit (BML) which a t Point Lepreau Generating S t a t i o n is a I me&itivfe HA>Hi i s k . It : s aril, i c i pat ed t h a t implementation of t h e EuH Moiul.orinsJ System w i l l not ari\ e r s e l \ aff pet svstem load] ne'.
Lrsonomies was one of the l;ev design consider— a t j o n s wit hi t.he impJemcnUition of FX')V Monitoring .'-.v^tem. Jt was decided that the svstem must i n t e i f ; i ! r mi., 1'i/iul J.t-fjrPrjii's control room en\ ironinent in :i natm,;J ntiiinei1. liius i t un% desirahir. 1 t h a t the ne^ >»>st.t'm be run:-£»sserf tiy the same keyboards and djspifAK t h a t a r e used to monitor and control t h e s t a t i o n . l-oat.ures U'uit. minimiimize l.he number of Ke\sL/okss to a r u e s s B)K-related d i s p l a y s have been
i ri£ bvstem n - l a u - d >i i spja\ s u 11 h a minimum of kevst.rokes. Ch;uns a r c a |u\>i;r auum-d -.i-itui-ncf- ui ilispia\ and fund ions thai can he access'-.i simply by p r e s s i n g member of a ^ u v n c/iai/i has ln-rn ,K cessfij \ ja t h e MHNl'S program. Special Slatu:, |Ji-j»|a\x. Spei ial s l a U l s d i s p l a y s were c r e a t e d t u provide Ihe o|x-rator with o n - l i n e di KplavK o/ (he (**i/"iii« f * r-. !!:.)! »jii.st In- ijioni l.ort'd during t h e execution of each KOI1. These• d i s p l a v s con t .HI n e s s e n t ia 11 v a l l i*u-amot e r s I-I-UIII reii f'or* execution of an Btf' on a s i n t i l e s c r e e n . The HA' d i s p l a y s pro\ ide rr-ums i t.e dai.i in .i coiisist t-nt manner for onrh rt>P (with t h e e x c e p t i o n of tX)l-' 2) .
OVl-J?V]l->" (JK nil- ••O-ll'l TK1,' MfMTORIV, hYsTKM the pmuram (if*s i titu^i (*> T't-citWi; •.> • i-XA' ent rv cond i t i o n s , aiii.1 to provide subsequent monitoring once a s p e c i f i c KJP has been turned oi. i * c t i l o t i i.CM I &nt»rgency Conditions Monitor). GCDSh' i s the (.ioverninsj Conditions Display prourain llmt i s ust^i in cuii.Htnctioti with Et.'M to turn on and turn off EOP's, and to d i s p l a y l i v e l i s t s of Cioverning Conditions. These l i s t s h a \ e a one t o one correspondence u i t h page numbers t h a t e x i s t in each B)P and thie\ cuiit-ain a l l the Govex-nins Conditions that t h e operator must monitor during the execution of each KOP (Kiijure I ) .
.Annunciation Of.EOF Entry Conditions. When an EOP entry condition has been i d e n t i f i e d , an alarm i s output t o the window s e c t i o n of the annunciation CRT with the following format tFigure 2): ECM EOPnn xxxx ENTH\ CONDITIONS MET. Also, an a d d i t i o n a l alarm w i l l be generated on t h e normal p a r t of the d i s p l a y i n d i c a t i n g t h e parameters t h a t t r i g g e r e d t h i s alarm {Figure 2 ) . This secondary alarm i s required to inform t h e o p e r a t o r exactly what t r i g g e r e d the Entry Condition, and t o allow him t o determine for" himself whether or not the Entry Conditions Met alarm was s p u r i o u s . Note t h a t when a message i s output t o the window, the borders of the window will flash u n t i l the message has been acknowledged. Also, the border of the window v% i 11 not appear unless a message e x i s t s in the window. fl 7512 PT90? INSTRUMENT RIR PRES R3175 LO fl ECfi EOP 4 LOIR R 3 I 7 5 LO Y Y Y
fhe h p e t j a j Messa^f:- Bo\. A five l i n e bo>. on the annunciaLion ("H'l:s ua:-, createtJ in l i e u of .••iJ'J 11 i uriH i h ) ndfjw a I a inns t o r t he* f'o I lowintJ r e a s o n s : - Alai'm messages in the box cannot lie " r o i l e t i off" trie .st.roen durJnC ttn n\hivm b u r s t . - I t w/j,s c:onsiderabJ \ ewsier and cheaper t o implement than s e p a r a t e u ) ndow al;jr-nis. - It i s f a r more f l e x i b l e tlian window a l a r m s . - It i s the; place where I ho o p e r a t o r ' s a t t e n t i o n i s focused during ims'i ro/)dj t. i o n s . sep.-irat.f- Mcthrxi Of Vkriowjed^irig AiarmK. A -*t-pai'rjt i- nift.hod of Hcknov%led^jng key alarms v i a t h e :j|*-rai :\- Kcvhciard was \>\
fl EOP 4: Loin — ENTRY CONDITIONS MET n EOP 4 TURNED ON
2:
HJP H i t . r v I ' o n r t i l u . n '•Irt ^Ir The Sppc'irtJ MessMSP Ho-. CNS 9th ANNUAL CONFERENCE. 1988 419
nnn^ iJn \ n KOI1. Mien . m I-lri t r v C o n d i t i o n s Met. mm u v r w r - . i IK- o p e i v i t UJ t a n o n I \ ru-know l e d g e t . h o :,.;.IL'I' i<\ i . i i l i n i i HI- i h i - l i u M - m i i i i ! Cuu.li t i u n s f ) i s .i\ i..np t:n .tint f i i l i - i ' i i i g ;i " 0 " f o l l o w e d b y i-.Kiim' IvMKK a m i HUMI KM-XVfK (KNTKN . m d KNl-.ri.TK i' t u i K - t i o n k e \ s t h a i t - M s t w j l h .••ill t h e o p e r a t o r \limrtts t. I h i s !>,-itmr u;is provided t o p r e v e n t Ill itx'r.it ur l( )j» f j o s s i b l v m i s s i n g a n K C M - r e l a t o d « l .-inu v . l , , 1 i' an i l a r m burv.1 u a . s i n p r o g r e s s . Knt e r i n i s done, t h e .1 "()' ' I l l l - l l s on I b e KOI'. When S h i s i s n o i i o r r i t i - d i n t h e W i n d o w : ECH BOP I I I i u ) u i M i ; ri larn c)\. In .-ujdi! I U I I , f h e f i r s t d i s p l a y p a g e 1 1 1 1 11 HM-:D l-jiiersieniA 0 | x > r a t i n £ C o n d i t i o n s b e c o m e s " l i v e " • '• I .
EOP 04
LOSS OF INSTRUMENT fllR GOVERNING CONDITIONS
1339:99 PflGE # 0 2
1 I S S B S REQUIRED?
OK
2 I S STEFBRCK RCTION REQUIRED?
OK
3 flRE RLL BLR LVLS RESTORING TO STPT OK 4 I S PRS LVL RESTORING TO STPT?
EHTER. OROR OP
OK
Lr«E f / f l ^ I TO StJ CNfPV 1 8 15flCHHOWEBGE- 1URN W ON/OFF PflGE I TO IISPLflV RMOTHER FflGE 'FflGE < TO RCTlVflTE BHOIRtR FflGE
Kigure .3: Loss Of Instrument Air Governing Condi! ions Display GCnSP S t a t e s . Two s t a t e s exist for each lioverniiis; Condition diamond - they a r e ALARM and OK. The U.AKM sU>U> corresponds t.o the c o n d i t i o n t h a t will r e s u l t in a sideways branch in the Governing Condition diamond in the t a b l e of the i>rrit.ten EOP, n.tid the OK s t a t e corresponds to the branch t h a t e x i t s through t.o the bottom of t h e Governing Condi t i o n s diamond. ALARM and OK were chosen in l i e u of \EK and NO because not a l l d e s i r a b l e answers t.o the Governing Conditions diamonds a r e c o n s i s t e n t l y \\& or NO. Mien a i l Governing Conditions have been met., the e n t i r e Governing Conditions d i s p l a y for a migtis can lie t-luuigt'iJ \ i.-» th' 1 (io\criutm ( omt i t j oris Display. When ihc-i is dom-, a l l iKj\'*iriini< (*ondi t.iortK on lire (ifi^ious |*iiip no longer apf/Ji. 'Ylw Uo\erniijrf i iriil.tiuns tit' (.lie ti(-\* i>fiU(w will now be monitored,
420
CNS 9th ANNUAL CONFERENCE, 1988
T h e w r i t t e n KOf w i j j inclmit' a s t e p t J i / i l iv i I I ":-,i-t ' t h e G o v e r n i i i t ? c o n d i t i o n s o i s p l a v t o a n e w pa*4e a s t h e C i r s t ;ut.iuti u l ' C - K h iu'\% K i i t l i n i « i ^ i - .
UTAH'S AM) CHAINS Two r.'i t her global priluiriccmrTits ivorv mmlt- lo 1 ho tcjjjjpvitt-r system. Thev we/e the cretiliQn of a system ul" "Rxlcrided DT.\)J'.s", and I ho trcnt ion .)t "Chains". Kxtendpd UTAIi's. Tht- lonn DTAH stands \\,v \)n\:\ TABle. \ system of ortal ();')()() mc'mor'V resident UTAH's existfxl before tin- KM crid<>d UTAH's were created . These var11 nbl CH are tfefif-rw (.(^i li\ VHI' IOUS programs on tin- svst.om unti can be ijst^i to produt;e Trends, Hars and Status Displays. The e x i s t i n g UTAH t a b l e could not be used because it was almost Cull, and not enough core-resident memorv existed t.o create a second such t a b l e . Also, additional requirements were placed on the new variables in that some of the l'X)P Decision Ro\es ask if a u;tr/impter- is fihcjve or belo\% some setpoint , and r i s i n g or f a l l i n g . Thus, i t was necessary to c r e a t e fi system that could discern if a parameter was r i s i n g or f a l l i n g . I>ie approach that is used Lo determine il' a parameter is r i s i n g or f a l l i n g i s to f i l t e r the present value of a given parameter wit.h a short term f i l t e r , and with a lonff term f i l t e r . Then by comparing the two, it. is possible tu determine if the parameter is r i s i n g , f a l J i n g , or s t a b J e . The r e s u l t i n g condition of t.ho parameter wi 11 be indicated on the \'T)P sUitus dispJa\-.s to the r i g h t of the parameters that are be i r\a mon i t.oreii. Expansion of the e x i s t i n g DTAB system, and provision of the r e q u i s i t e memory for the f i l t e r e d values, and the b i t s t o indicate whether a variable i s r i s i n g or f a l l i n g r e s u l t e d in the creation of the Kxtended DT.Mi system. Extended UTAB's begin at. address 0500, and currently extend to address 0777. The upper limit., however is 07777 DTAIi's. Kxtended DTAR's a r e stored in the BMl\ and each have four words of memory associated with them. They q u i t e nicely provide the memory needed to }iave Die computer determine if a c e r t a i n parameter is r i s i n g or falling. Implementation of Extended DTAB's required mod i f ir:fit ions t o : - A1C1N/AIOUP - Analog Input S p e c i f i c a t i o n s I n i t i a J izeA:\xinto Program - HAUlU/HAHVy - Bar I n i t i a i i z e / B a r Update Programs -
>fTBAR
-
Modify
Trend
and Bar Program
- MJDIS
- Numerical V a r i a b l e s Display Program - SMLST - Summary L i s t l ^ o g r a m - STA1N/STUP2 - Status I n i t i a l i z e / S t a t u s display Update Program #2 - TRLOG - N u m e r i c a l T r e n d Log Program - CKMN/TKNl.'P - Trend I n i t i a l i z e / T r e n d U p d a t e Programs
Chains. Chaining i s a f e a t u r e thnt was added to allow the o p e r a t o r t o quickly change d i s p l a y s of ke\ parameters. Without, c h a i n s , the o p e r a t o r has to cancel a d i s p l a y , remember the code and t-nJl up a subsequent dis-pj.-iy. Since t Ju* I*X>P syst em invol ves working with s e v e r a l d i s p J a > s , it w;is de
IIAGI- KuRMkli .ui.l i'-VJK BACKWARDS a r e two of puslifmf t o n s Uiat h a \ e be:Mi p r t n i d e d on t h e ki\\ h o a r d s ) .
the operator
Mil1 IM-H.V-. Mil CIl UK I'UOGIUMS Se\rral programs !iad to be created or mollified to ai i onmioilaI.e t.he integration of the I-JOP Monitoring System into the Station Control Computers. VlMX.i. MNix) is the program that displays special alarms on t.he annunciation CRT's. The bottom five lines on the annunciation CRT's were reserved for MMA). Normally, nothing appears in this area; but v>'hen a program such as ECM outputs a message to VINIX), Ihe message appears m the area. WINDO is programmed to surround the five line area with a cyan bo\, uhen messages are present.. Also, when an alarm is in the non-acknowledged state - i.e. when the "A" is flashing, the border of the cyan hox also flashes. Thus, an alarm appearing in this area is very difficult for the operator to miss. 3'be possible alarms tliat. can be output to WINDO •iff small in number, and they are prioritized, thus message flooding of this box is not anticipated to be a concern. The intended use of the box is to quickly get the operators attention. Then, additional information can be obtained by calling up key displays. Keactor_Sub-Cooling. Margin Monitor,. Since no existing program provided it, and there was an EOP requirement, it was necessary to provide a Reactor Sub-Cooling Margin Monitor. This logic used the reactor header temperatures and pressures to determine the number of degrees C, and the number of MPa(d) that each header is from saturation conditions. Alarms are pro\ided on Sub-Cooling Margin Temperatures being less than 10 and 5 degrees c. These alarms are conditioned out at high power. ECnST, Eight dedicated status displays have been c reateii for the (OOP Monitoring System. Oie display exists for each fcX>P (except for Dual Station Control Computer Failure). ICach of these displays has been programmed to provide essentially all the data that is require! by the operator to ascertain existing plant conditions vis-a-vis the EOP. Figure 4 is a copv of the status display that is associated with Bjp -I - Loss Of instrument Air. 88IRV27Z
EOP 04 LOSS OF INSTRUMENT RIR 13 39 28 WIIEP PSESSUPE »«
H!5 PPES IPPOH) titlRP, «PB(G| f
N I
LEVEL
FPCSSURIZEV LEVEL SEIP0IK1
SIflTUS
PPESSUP.WEP. LEVEL EP»OP.
10ILEP LEVEL StBllJS
LEVEL l « l
801 LEVEL STPt (HI •••[RR LEVEL EPP lltl
MENUS. A menu display program was provided on the system that enables the operator to very quickly call up certain displays. The operator can obtain the Governing Condi t.ions Display from the MENUS display wi t.h three keystrokes followed by the EOP number. Mien the Governing Conditions display ha.s been called up in this manner, a CHAIN is automatically invoked that allows the operator to call up displays that are associated with each HOP simply by pressing PAGE FORWARD or PAGE BACKWARDS. The first display following the Governing Conditions display is always the dedicated status display for the selected KOP. This is followed by other displays that may be needed to execute the EOP. CPSUM. CPSUM is the control program summary display program. This program provides a contitiousl y updated list of alarms that are associated with a given control program. Modifications have liad to be made to CPSI.'M to have jt accommodate chaining and to accommodate ECM.
ADDITIONAL USES OF THE EOP MONITORING SYSTEM The EOP Monitoring System can be used to monitor other plant conditions in addition to the eight EOP's listed in Table 1. Two can^!.i^Les for monitoring by Et.'M are Reactor Heat. Sink Monitoring during plant shutdowns, and Condenser Salt Leak Detection. UKfit.__girik_Mgni_tOTIn^. A version of ECM was installed in the Station Control Computers Just prior to Point Lepreau's 1SS8 Annual Outage to provide a Heat SinkMonitoring feature while the plant was shutdown. The Heat. .Sink Monitor was written to monitor the station during three states of operation: 1. Heat Transport System operating on Shutdown Cooling in Shutdown Cooling Pump mode with the Heat Transport System full and depressurized. 2. Heat Transport System operating on .Shutdown Cooling in Shutdown Cooling Pump mode with t.he Heat Transport System partially drained, but not breached. 3. Heat Transport System operating on Shutdown Cooling in Shutdown Cooling Pump mode with the Heat Transport. System partially drained to the headers and breached.
HIS PRtS STPT •••IPB KPI1IGI
HE5SWI2EP P8ESSIIRI2ER LEVEL
"UKR
CSTAT\ CSTAT stands for CaJculations btetus Monitor. This program existed prior to the creation of the KOP Monitoring System. It, however, lias been reprogrammed to accommodate the neu system. In particular, this is the program that computes most of the new D'TAB's that, are required by KCM and by the HJP Status MspJays.
•••IRK
BCM B02 B03 '•'IK I •••IRR I •••IRS
OUTLET TEKPEP.RTUPE
flU>ILIRPV HES1 EHtSHlELJ MTLET TEHPERflTOPE WEPflllIRE EB5T EHDSHIEL1! OUTLET TEKPEPflTUPE
I: Loss Of Instrument Iji w)^la\
tHIPR
Condens^er..SaiLLeak_Deteciion_. Point Lepreau Generating Station is cooled by sea water. Consequently, the rapid detection and isolation of condenser leaks is mandatory to prevent the possible shutdown of the station. II has tieen proposed to provide Jogic in ECM monitor for condenser salt leak Condi tions.
PRESENT STATUS OF THE SYSTEM
" M B R DEG C I « M R R DEC C • " ' I R B «EC C I
\ir H;P Status
At the time of writing of this paper, the entire EOP Monitoring System liad not been completely implemented on t.he Station Control Computers at. Point, l.epreau. Much of the infrastructure had been installed, as well as t.he Heat Sink Monitoring function of liCM. Demonstration versions of GC'DSP and tt'M that monitor CNS 9th ANNUAL CONFERENCE. 198B 421
r'jnersencv plant, c o n d i t i o n s had boon w r i t t e n , but t h e a c t u a l cud ins; programs for t h e o n - l i n e system had not been ful it i>rot£nuiwtcd and debugged. THE !t i s envisioned t h a t a f t e r t h e HOP Monitoring System has been f u l l y implemented, s e v e r a l addi t.ioricii i'e;»t.urr& wi]J Ije .-idded to the svstem. It, has been proposed t h a t a "gateway" computer be added t o l.hr S t a t i o n <..ont.n.)l Computers ho e n a b l e t h e remote a c q u i s i t i o n of d a t a v i a t h e p l a n t Ethernet Local Area Netuork. When t h e new communications system i s in pltur-, it w i l l be p o s s i b l e t o f u r t h e r automate t h e d e t e c t i o n , monitoring and execution iif Emergency P l a n t Operating C o n d i t i o n s / P r o c e d u r e s .
HKl'KKKNCKS (I)
422
H. COL.QUMOUN, A.R. JUINSON/J.F. McCALLUM/D.F. WEEKS, "Bnergency O p e r a t i n g P r o c e d u r e s Based On Thermodynamie S t ^ i t e " , Canadian N u c l e a r S o c i e t y 8 t h Annual C o n f e r e n c e , J u l v / A u g u s t 1987,
CNS 9th ANNUAL CONFERENCE. 1988
A REVIEW OF QUANTITATIVE CRITERIA FOR DEMONSTRATING NUCLEAR POWER PLANT DESIGN ADEQUACY
K.S. DINNIE
Ontario Hydro 700 University Avenue Toronto M5G 1X6
ABSTRACT The paper examines some current and previous attempts to establish quantitative safety standards for nuclear power plant designs. The implications of these standards to reactor design are discussed using the results of the Darlington Probabilistic Safety Evaluation for illustration. It is concluded that standards based solely on health risk may not be appropriate for demonstrating design adequacy. Some additional design goals are suggested for discussion.
quantitatively by comparison of predicted plant performance against deterministic, event-based criteria. In Canada this refers to the single/dual failure limits of the AECB Siting Guide [2] or, more recently, the dose limits contained In the consultative document C-6 [3 I . While this kind of approach dees provide an effective basis for system design and assessment of system capability under accident conditions, it cannot by itself be regarded as a demonstration of the overal1 adequacy of the design for the following reasons,-
INTRODUCTION The level of safety at a nuclear power station is determined by many contributing factors; good design, good construction, good operation, good regulation. It is axiomatic that safety can always be improved, albeit at some cost. The question remains as to how it can be demonstrated that the plant design aspect of safety is adequate, where "design adequacy" means that it is no longer necessary to spend money on additional safety features. Design adequacy is, of course, not to be confused with "public acceptability" which involves wider issues of a technical, sociological and political nature. The demonstration of adequacy is more clear cut where quantitative expressions of plant performance can be compared with appropriate standards of acceptability. The use of quantitative targets or limits has long been an integral part of the licensing process in Canada. The advent of risk assessment has provided a method of estimating the risks due to accidents posed by the operation of the plant and provides an alternative means of quantifying design adequacy. The problem then becomes one of setting appropriate standards against which adequacy can be assessed. This paper examines some current and previous attempts to establish numerical safety standards. The results of the Darlington Probabilistic Safety Evaluation (DPSE)[1J are used to examine the alternative approaches and draw some conclusions as to their effectiveness and implications. The objective is to identify those standards which set appropriate limits on plant design such that, if met, no further design changes need be considered. The purpose of this paper is to identify some of the issues and promote discussion related to the setting of standards to demonstrate design adequacy.
EVENT-BASED SAFETY LIMITS The determination of design adequacy for purposes ">f plant licensing is accomplished
the
(a)
There is no limit on the number of sequences in any dose category, and hence there is no integration of risk contributions.
(b)
Multiple system failures outside the single/dual failure definition are excluded.
Exan.ples can be found where design changes have been required, even though all dose limits have been met. Discussion related to the need for such changes takes place in something of a void in the absence of any criteria to determine whether an adequate level of safety has already been achieved. In addition, in order to achieve a large margin of safety in licensing analysis, it has been the practice to require significant conservatism at many stages during the analytical process. Much of the safety system design is predicated on highly stylized accident assumptions. This approach has been effective in achieving an accident-tolerant design. However, there have been some side-effects. The estimation of dose consequence places great importance on radioactive releases during the first hours of a postulated accident sequence. As a result, attention is focussed on accident sequences which maximize this component and design features which mitigate these releases. The outcome is a design which is very sophisticated in dealing with the kinds of events envisaged in the safety analysis; for example, the design provides a very large, complex and expensive containment overpressure relief system to handle large LOCAs. In comparison, relatively less attention has been given to the implications of dealing with more slowly-developing events and to the related aspects of accident mitigation and control which have generally been found to dominate risk. This is, at least in part, due to the relative lack of importance of tiiese aspects in meeting current quantitative safety limits.
CNS 9th ANNUAL CONFERENCE, 19B8 423
PROBABILISTIC SAFETY GOALS The advent of modern probabilistic safety analysis (PSA) offers a potential solution to the limitations of the deterministic approach. In principle, integrated plant risk can be compared against objective risk-based standards ("Safety Goals") and design change recommendations can be evaluated on the basis of the potential for risk reduction and tha predicted level of risk imposed compared to the relevant goal. Safety goals are usually expressed in terms of mean risk, which eliminates some of the bias implicit in conservative deterministic analysis. If it is accepted that the output of a good quality PSA is a meaningful representation of the accidental risk posed by the facility, the question remains only to develop suitable safety goals.
smaller on a per unit basis. For these goals are equivalent to or than the USNRC gcals.
multi-unit site more restrictive
There are currently no equivalent collective goals, which are more difficult to derive and justify. However, if the most exposed individual is adequately protected, it is the expectation that other members of society will be adequately protected.
Significance of Risk-Based Safety Goals The purpose of defining safety goals is to establish criteria against which the adequacy of any actual or proposed nuclear plant design may be judged. If such goals do not represent an "adequate" level of safety then they will be of only limited use in the design process.
Safety goals are usually based on a qualitative principle such ss; "Due to the operation of a nearby industrial facility, individual members of the public should not be presented with additional risks to life and health which are significant in comparison to other risks to which they are normally exposed."
The prompt fatality goal under consideration Ontario Hydro is expressed in such a way that, meet it, any accidental release which is predicted result in at least one prompt fatality must occur
This principle can also be applied to population groups (a collective gcal) as well as to individuals (an individual goal). Separate goals are usually applied to prompt and delayed fatality risk arising from operation of the facility concerned, representing the primary potential health effects of radiation exposure.
7.5x10 /reactor-yr. The DPSE predicts the mean frequency of a "large release" to be 8.2x10 /reactor-yr. However, it is considered highly unlikely that the release from any of the events which dominate the contributors to this frequency would result in any prompt off-site fatalities.
Two sets of proposed risk-based safety goals currently under discussion in North America will be examined in more detail.
The USNRC Approach In 1986 the USNRC published its proposals for probabilistic safety goals in final form. The prompt fatality goal was defined as 0.1% of the background fatal accident risk applied to the average individual within 1 mile of the plant boundary. The latent fatality goal was set at 0.1% of the background cancer risk averaged over the population within a 10mile radius of the site. Quantitatively they translate to about 4xZO /yr prompt fatality and 2x10 /yr latent fatality respectively. They apply to the average member of the population group rather than the most exposed individual and to each reactor on a given site.
A Canadian Approach At Ontario Hydro preliminary safety goals for internal use are under consideration. These goals are based on the qualitative principle stated above, where insignificance is interpreted as a risk less than 1% of the average accident fatality risk and cancer risk in Canada. They are intended to represent mean values and apply to the most exposed individual in each case. Expressed quantitatively, t^ey correspond to a prompt fatality goal of 3x10 /yr and a latent fatality goal of 1x10 /yr for the site under scrutiny. For a multi-unit station the goals apply to the site as a whole and would be correspondingly
424 CNS 9th ANNUAL CONFERENCE, 1988
at to to at
Recent studies in the US [4 ] have calculated the prompt fatality risk for a number of plant designs. The mean risk f alls approximately in the range 5x10 /yr to 5x10 /yr to the average individual within one mile of the site boundary. The mean frequency of a large release resulting in at least one fatality is estimated to be in the range 1x10 /yr to 5x10 /yr, indicating that individual goals are generally more restrictive than collective or averaged goals. Although the DPSE did not explicitly estimate the overall latent risk, it may be approximated by;
Mean latent risk = + + k
5x10 _/yr (normal operation) 1x10 /yx (DPSE quantified risk) 7x10 7 /yr (large release risk) 7x10 /yr
The risk from normal operation is based on measured releases from existing stations (1% of the derived emission limit). The large release risk is calculated from the DPSE large release frequency, conservatively assuming 2 Sv to the most-exposed individual. It can be seen that there is a comfortable margin between latent risk and the corresponding suggested safety goal of 1x10 fatalities/yr. Similarly, the US estimates place collective latent risk several orders of magnitude below the corresponding goal. Latent fatality goals appear to be significantly less restrictive than prompt fatality goals. For example, to pose a risk to the individual of 1x10 /yr, the predicted frequency of an accident resulting in a dose of 1 Sv (a very severe accident) would have to be 1x10" /yr.
It is expected that improved methods and safety research will further reduce the predicted likelihood of prompt fataLities from a large release. If th; r, is the case, it wi 11 be more evident that existing reactor designs already meet current proposals for safety goals by a s ignif Leant margin. A safety goal which permits an accident resulting in up to 1 Sv to some individuals with a frequency of 10 /reactor-yr intuitively would not represent an adequate design target, The question remains as to whether this intuition can be confirmed quantitatively by using other criteria. In the following sections some alternative considerations which might prove to present more intuitively appropriate limits on plant design are reviewed. Results from the DPSE will be used to illustrate their possible significance.
ECONOMIC CRITERIA Off-site Economic Risk The occurrence of an accident that caused a large offsite release of radioactivity could result in a very significant economic penalty. Costs of such an occurrence have been estimated for Southern Ontario [5] to be in the range 1 to 10-billion dollars using 1981 economic data. Assuming, for current purposes, the higher value, at what accident frequency would the economic risk approach unacceptability? One possible approach is to examine what annual insurance premium might be considered reasonable to obtain full coverage for offsite economic risk. Ontario Hydro is currently required to obtain $7 5million coverage for which it pays of the order of S100,000 per reactor per year. In the hypothetical situation that full coverage could be obtained, 5100,000/reactor-yr would clearly be an acceptable premium. It is judged that premiums up to perhaps ten times higher might be considered reasonable. Thus, it is proposed that an acceptable level of economic risk would lie in the range of $100,000 to Sl-million/reactor-yr. To clearly meet this target, the estimated frequency of a large release should lie towards the lower end of the range 10 to 10 / reactor-yr.
On-site Economic Risk Accidents which cause damage to reactor fuel and result in a release of radioactivity inside containment have the potential to cause significant economic penalty mainly due to the cost of obtaining replacement power. The economic risk is found to be dominated by high-frequency/low cpnsequence events which would not meet the definition of an accident being used in this paper, because they do not present a significant risk to the public. Losses from this type of event can be absorbed into the general costs of power production. Of greater concern is the potential consequence of more severe accidents. On-site costs at an Ontario Hydro multi-unit station have been estimated to be in the range 51-bill ion to about SlO-billion [1], depending on the magnitude of core damage.
On-site risks are not normally insurable so, theoretically, a utility is free to establish t helevel of economic risk at which it wishes to operate based on internal economic cons iderations. The onsite economic risk from reactor accidents involving fuel damage could be regarded as acceptab]e if, when treated as a cost of production, its contribution can be regarded as insignificant. In line with the Ontario Hydro approach to health risk safety goals, it is postulated that a contr i but ion of Less than 1% to cost of production can be considered insignificant. This translates to an acceptable risk in the range SI-mil lion to SlO-million/reactor-yr depending on the station. It is not unreasonable for the acceptable level of self-insured risk (i.e., onsite) to be higher than that for insured (i.e., offsite or third party) risk. Acceptability might also depend on the size and economic resources of the utility. In order to clearly pose an acceptable level of risk, a 'Sl-billion' accident should be predicted to occur with a frequency below about 10 /reactor-yr and a ' $10-billion' accident below 10 /reactor-yr. bearing in mind that this does not involve containment reliability, it represents a somewhat more restrictive criterion than off-site economic risk. It is also the only criterion which applles to core damage rather than offsite release.
LOSS OF LIVELIHOOD Most efforts to derive safety limits, targets or goals are based on the reasonable presumption that the principal concern of safety design i<= to protect the life and health of the affected population. Thus, most are derived by estimating the likelihood of death arising from radiation exposure and comparing this with background mortality levels. The presumption that mortality risk is the limiting consideration needs to be questioned, at least as far as the derivation of quantitative goals are concerned. It is proposed here that an issue of major and potentially more limiting concern may be that of "loss of livelihood". There is some evidence to suggest that, at least in the public's mind, risk of loss of livelihood (for example, as the result o£ a prolonged evacuation due to ground contamination) may be of equal or greater importance. Indeed, people have been known to accept significant health risks in defence of their way of life. Concerns with respect to the broader impact of a severe accident have led the USNRC to propose a general guideline that "the overall mean frequency of a large re lease of radioactive materials to the environment from a reactor accident should be less than one in one million per year of reactor operation". If this is defined as a release sufficient to cause at least one prompt fatality, then it may or may not be the limiting criterion on plant design depending on the outcome of safety research. Currently, it is more 1imi ting than the proposed prompt and latent fatality goals. However, if the concept of a release sufficiently large to result in loss of livelihood were adopted, a more rigorous and, it is believed, more realistic criterion would result. If the prompt criterion is used, accidents of the magnitude of Chernobyl could
CNS 9th ANNUAL CONFERENCE, 1988 425
occur at a frequency of up to 1x10 /reactor-yr and still meet safety goals, because there were no offsite prompt fatalities and the latent fatality goal is limiting. If loss of livel ihood were used, the predicted frequency of such accidents would have to meet the large release goal for the design to be considered adequate (i.e., less than 10 / y r ) . Suggesting such a safety goa.L presents two problems; defining what is meant by a large release sufficient to cause loss of livelihood, and establishing an acceptable frequency target. In qualitative terras, any accident sequence involving severe core damage and containment bypass or loss of the containment function would be a candidate for inclusion in the definition. A decision to evacuate an area on a long-term basis would depend on many situation-specific factors {including health risk considerations), but the avoidance of a mean dose to individuals of the order of 0.25 Sv or greater resulting from the pathways associated with the long term effects of ground contamination (ingestion, groundshine, resuspension) might constitute a practical working definition. This figure is suggested by emergency evacuation criteria. Clearly, there is no meaningful "background" frequency against which to assess acceptability related to loss of livelihood. One possible approach is to aim for a high degree of assurance that no such event will occur during the lifetime of the current generation of nuclear power pi.ant designs. A simple way of arriving at a target frequency is to use the following equation;
P = ( 1 - Np ) , • 'lere P = Target probability that no accident will occur during life of program, N = Number of reactors in program, p = Target mean frequency of a large release per reactor-year, y = Average lifetime of a reactor
There would not be much merit in applying such a criterion to a single reactor or site. Assuming a 2o-reactor program, a 40-year life and a target probability of .99, the frequency goal would be 1.3x10 /yr. This argument can be extended to include all commercial power reactors. It does not seem reasonable to establish design adequacy standards based on the size of a domestic nuclear program, especially where reactors are sited close to international boundaries or where several countries use the same basic design. Whatever level of safety is claimed through predictive analysis, the safety of nuclear power plants will ultimately be demonstrated by performance. Further occurrence of severe accidents will undermine the credibility of the technology, wherever they occur geographically. The potential international impact of severe accidents suggests that there should be some unanimity in setting design goals. Figures 1, 2 and 3 show the results of varying the equation parameters for target probabilities of .9 .95 and .99 respectively for *up to 600 reactors with average lifetimes up to 50 years. A frequency goal
426 CNS 9th ANNUAL CONFERENCE, 1988
of the order of 2x10 /reactor-yr would provide a degree of assurance in excess of .95 over the range considered. A 400-reactor program, a 4 0-year life and a_ target probability of .99 requires a goal of 6x10 /reactor-yr. Thus, it appears that a "loss of livelihood" goal in the area of 1x10 /reactor-yr would provide a high degree of assurance that such ar. accident would not occur in the foreseeable future. It is noteworthy that the above result is very similar to the prompt fatality goal under consideration in Ontario Hydro and the VSNRC lorqe release proposal, in that they all apply to a large offsite release of radioactivity, with the difference that the consequence of concern is no longer related directly to prompt fatalities. In practice, the analytical uncertainties are such that the estimated frequency of a large release may not be very different, whichever definition of a large release is used. it is also noteworthy that "compliance" can be demonstrated in terms of accident frequency alone,
CONCLUSIONS A number of potential design criteria have been reviewed to attempt to identify those which might determine design adequacy and to suggest some suitable numerical values for use as design goals, These goals are express-ed solely in trrms of accident frequency, Although safety goals have traditionally been based on comparisons to other societal mortality risks, it is concluded that such considerations may not be sufficient to fully determine design adequacy. Three additional goals are suggested for consideration: (I)
The predicted mean frequency of accidents leading to significant damage to fuel in the core should be about 10 /reactor-yr or lower.
(2)
The predicted mean frequency of accidents leading to severe core damage should be about 10 /reactor-yr or lower.
Here, "significant" damage implies fuel overheating leading to fission product release but core structural integrity basically remaining intact. "Severe" damage implies loss of core structural integrity leading to the release into containment of a substantial fraction of core inventory. These two criteria are set by on-site economic ri.sk considerations. (3)
The predicted mean frequency of a accident leading to a large offsite release of fission products sufficient to cause loss of livelihood in the local population should be about 10 /reactor-yr or lower,
"Loss of livelihood" is assumed to be caused by the need for prolonged evacuation (months) to avoid the possible effects of 1 ong term dose received through the pathways associated with ground contamination. Adoption of such a goal vould Lequire the definition of the cause of loss of 1 ivelihood to be expressed quantitatively to achieve consistency and avoid misuse. Avoidance of a mean individual dose in excess of 0.25 Sv is suggested in this paper.
If these goals are met, any other rationallyderived goals based on health effects of radiation exposure or economic risk would also be met. Reactor designs which are assessed to meet these goals would not require further design changes to improve safety. These goals would be considered aspirational. Failing to meet the goals would not necessarily imply unaccentability, only that design changes should continue to be considered on a cost-effectiveness basis.
REFERENCES
Deterministic practices could be derived from these goals and be directly used in design and licensing. Good construction, operation and regulation would still be needed to achieve a safe and viable industry. All quantitative targets or goals, whether based on health risK or otherwise, are derived from the same basic objective of establishing a safe and viable nuclear industry. M l involve a degree of subjectivity in establishing an acceptable level of risk. This paper suggests some design goals which set standards with the objective that, if met, the design could be considered "adequate" and the allocation of significant resources to provide additional safety features would be unnecessary.
FIGURE 1 :
Target P r j b a b i l t t y
[1]
"The Darlington Probabilistic Safety Evaluation - Summary Report", Ontario Hydro, December 1987.
12]
Hurst, D.G. and Boyd, F.C,, "Reactor Licensing and Safety Requirements", Paper 72-CNA-1O2, presented at the 12th Annual Conference of the Canadian Nuclear Association, June 1972.
[3]
"Requirements for the Safety Analysis of CANDU Nuclear Power Plants", Atomic Energy Control Board Proposed Regulatory Guide, June 1980.
[4]
"Reactor Risk Reference Document", NUREG-1150, U.S. Nuclear Regulatory Commission, Draft for Comment, February 19B7.
[5]
Lonergan, S.C. and Goble, R.L., "An Estimate of the Off-site Economic Consequences of a Severe Accident at the Pickering Nuclear Power Station", Submission to the Ontario Nuclear Safety Review, November 1987.
= 0.90 -3
5 30-yr W-yr E.0-yr
life life life
<•
/
30-yr 40-yr
50-yr
lire l i f e
life
Number Q> P l d n t i
CNS 9th ANNUAL CONFERENCE, 1988 427
SOME RKSU1.TS AND INSIGHTS FROM THE DARLINGTON PROBABILISTIC SAFETY EVALUATION K.s. Dinnie. S.G. Lie Ontario Hydro 700 University Avenue Toronto, Ontario H5G 1X6
ABSTRACT
FUEL DAMAGE CATEGORIES
The Darlington Probabilistic Safety Evaluation (DPSE) represents the first comprehensive study of a muiti-unit CANDU nuciear generating station using up-to-date risk assessment methods. A previous paper [1] described the methodology used in the DPSE and provided some preliminary results. In this paper we present some of the final results of the study and provide additional information to identify the major contributors to accident frequency and risk. Some Insights are offered into the relative significance of these results.
The event tree analysis generates a large number of event sequences which could result in economic or public health consequences of potential significance. Categorization is carried out to simplify the evaluation of risk. Each fuel damage category (FDC) represents a collection of event sequences judged to result in a similar dsgree of damage to fuel in the reactor core. All event tree end-states which cannot be shown to be inconsequential must be allocated to one of the FDCs. A brief description of the ten FDCs is given in Table 1.
THE DPSE The primary objective of the DPSE was to provide a thorough design safety verification of the 4-unit Darlington Nuclear Generating Station using probabilistic methods. To achieve this, accident frequencies were estimated for a comprehensive range of internal plant initiating events covering the full range of accident consequence. Accident consequences were calculated to permit the evaluation of economic and public health risk for all but some of the more unlikely accident sequences. The process of estimating accident frequency permits the identification of the most likely ways in which accidents leading to various levels of fuel damage and having the potential for off-site release of radioactivity can occur. Design adequacy is determined by demonstrating that the frequencies of the various accident categories and the calculated risks are acceptably low. In the absence of established risk-based criteria for adequacy, some judgment is required to draw such conclusions. The primary results of the DPSE are expressed in terms of accident frequency and economic and public health risks. Accident sequences were grouped into categories to facilitate the integration of the results and it is around the two sets of consequence categories Cone set for fuel damage, one for ex-plant release) that the study results are structured. The role of consequence categories in the DPSE is shown schematically in Figure 1. From their relative contribution to the overall results, the most important categories can be identified. By using suitable measures of importance the major categories can be further evaluated to determine those initiating events, plant systems, component failures and human errors most significant to the study results. In the following we will present the quantitative results of the DPSE and provide interpretation and discussion on their significance.
428 C N S 9th ANNUAL CONFERENCE, 1988
Two estimates for the fission product release-to-containment source term were generated for each FDC (except FDCO, for which only the frequency was calculated'/. Best-estimate (BE) values were obtained from parametric studies, using some judgment. Probable-maximum (PH) values correspond closely to conservative safety analysis results used for licensing purposes. The range of event sequences covered by the FDCs is defined by the scope of the DPSE Study itself. The lower consequence threshold for significance is deemed to be the occurrence of a loss of heat transport system integrity resulting in emergency coolant injection system initiation. Although significant fuel damage is not likely, the event is considered to have the potential for significant economic consequence and is designated FDC9. At the other extreme, all events which have the potential to result in a "severe accident" as defined in the DPSE have been assigned to a single category, FDCO. FDCO contains all those sequences which have the potential to result in loss of core structural integrity such that fuel cooling cannot be assured. The calculation of the mean frequency of the ten fuel damage categories is a major result of the probabilistic safety evaluation effort, the results are summarized in the third column of Table 1. The unit of frequency used is occurrences/reactor-year. Relative Importance of Fuel Damage Categories One indication of the relative importance of each FDC is the mean fuel damage "risk", expressed as frequency of the category times the approximate fraction of the equilibrium inventory of fission products released from the core. Columns 4 and 5 of Table 1 summarize these quantitative estimates for best-estimate and probable-maximum fuel damage for each FDC, respectively. Figure 2 shows the relative contributions to fuel damage "risk".
Initiating Events
Event Trees
Fuel Damage
Containment Event —• Tree
Categories
Ex-Plant Release Categories
^
Dose Calculations
IE1
EPRC4
EPRC5
EPRC3 IE2
EPRC1
IE3 etc. FIGURE 1 Logic Development Process
TABLE 1:
FDC No. 0
Category Description
.Loss of Core Structural Integrity
1 Moderator as a heat sink required within 200 s of reactor t r i p
FUEL DftHAGE CATEGORY DESCRIPTIONS
Mean Frequency (per reactor-yr)
Mean Fuel Damage Risk (10"4% core inv/ Mean Pop Health Risk Contribution reactor-yr) 4 2 • 1 (10Person-Sv/yr) PM BE
Mean Economic Risk (M$/reactor-yr)
3.8 x 10-6 2 x 10 *>
.3
.6
2 Moderator as a heat sink 8 x 10 -5 required between 200 s and 1 hour after reactor t r i p
4.1
8.1
5.6
.3
3 Moderator as a heat sink required >1 hour after reactor trip
9.4
14.1
118.0
1.0
Large LOCA. Early Stagnation 3 x 10" 5
.6
1.2
2.2
.06
4
1.0
2.0
7.4
.2
4.6
23.0
2.3
2.2
.03
32.0
.2
1.1
8.0
1.1
.2
.0
.0
5.2
4
5 Large LOCA, Delayed Stagnation 6
5x
1 x 10"
Single Channel Event, Containment Overpressure
2 x lO'3
7 Single Channel Evont, No Containment Overpressure
3 x 10" 3
8
Loss of Cooling to Fuel in a Fuelling Machine
2 x 10" 3
.002
9
LOCA. No Fuel Failures
2.3 x 10-2
.0
0.01
NOTES: 1. Best-estimate prediction of fission product release. ?.. Probable-maximum fission product release. CNS 9th ANNUAL CONFERENCE, 1988 429
FDC 6 25.8% FDC 2 20.4%
FDC1 0.7%
FDC 5 5.0%
FDC 6 22.9%
FDC4 1.4%
FDC 4 FDC1 FDC 7 FDC 8
FDC2 9.1% FDC 8 9.0%
3.0% 1.6% 0.2% 0.01%
FDC 7 35.9%
FDC5 2.3%
FDC 3 15.8%
FDC 3 46.9%
Probable Maximum Source Term
Best Estimate Source Term
FIGURE 2 Relative Contributions to Fuel Damage "Risk"
Because the degree of fuel damage associated with FDC9 is essentially zero, it is not significant as a contributor to risk of fuel damage. If the probable maximum estimates of fuel damage magnitude are used, the risk is dominated by FDC6 and FDC7 which describe the so-called "single channel events". When more realistic fuel damage estimates are adopted, FDC3 becomes the most important category. It is worthy of note that large I.oCAs appear to be unimportant. If severe accidents (FDC0) were to be included and associated with a large magnitude of core damage, FfiCO would contribute a few percent to overall fuel damage risk. Importance of FDCs to Public Health Risk A pre-requisite to any radioactive release to the environment is the occurrence of an event represented by one of the FDCs. Release to the environment depends on the interaction between the characteristics of the FDC and containment. The relative importance of FDCs with respect to public health risk is determined more by the associated containment response than the characteristics of the FDCs. In the DPSE, dose consequences are analyzed for events representative of each ex-plant release category (EPRC), the constituents of which are obtained by merging FDC logic with the containment failure events. In the following we estimate the individual FDC contributions to quantified public health risk. First, the mean frequency of each EPRC sequence involving a given FDC is multiplied by the mean population dose appropriate to the EPRC in which the sequence belongs. These products are added to give the individual FDC contribution to the quantified public health risk. The results are given in column 6 of Table 1, under the title of mean population health risk contribution in units of person-Slevert per reactor year. The relative contributions of FDCs to quantified public health risk are shown in Figure 3a. A portion of the risk, that due to severe but low-frequency accidents (EPRCO). was not quantified.
430 C N S 9th ANNUAL CONFERENCE, 1988
The emergence of FDC3 as the most important category is continued. In this case it occurs because. included within the broad ran^e of sequences covered by the category, is a subset which involves I.OCAs outside containment (e.g. via HT pump seals or the D2O storage tank). The sensitivity of risk to event sequences involving containment bypass is a function not only of the frequency and consequence of these particular sequences, but also of the effectiveness with which containment suppresses the consequences, hence risk, from the remaining sequences. FDC0 also contains some sequences with the potential to bypass containment or result in its impairment. The DPSF. study has estimated the frequency of FDC0 and its constituent sequences but not the risk. FDC3 and FDC0 are considered to be the most "important" categories from a public risk standpoint. Important is used in a relative sense: the total risk, even allowing for a contribution from FDC0. is still judged to be small. Importance of FDCs to Economic Risk The impact on society from an accident at a commercial nuclear power plant is more likely to be economic rather than public health related. Because of this, the objective of the DPSE economic consequence analysis has been to develop and implement a method of assessing on-site economic consequence for each FDC. The total cost is found to be dominated by the cost arising from the need to provide replacement power. When combined with frequency estimates for each FDC, as shown in the third column of Table 1. a means of estimating economic risk is obtained which may be used to review the adequacy of station design. Mean risk estimates are given in the last column of Table 1 in the units of million dollars per reactor year. The relative contributions of FDCs to the economic risk are shown in Figure 3b. The dominant contribution stems from FDC9, which contains those events and event sequences that lead to a loss of coolant requiring the emergency coolant injection (ECI) to operate. With successful ECI there is no
FDC 8 1.9% FDC1 0.1% FDC 7 10.7% FDC 4 0.6% FDC 8 FDC4 FDC 6 FDC 5 FDC 7 FDC 2 FDC1 FDC 3
0.84% 1.6% 1.7% 5.4% 0.13% 4.1% 0.06% 86.2%
FIGURE 3a Relative Contributions of FOCs to Quantified Health Risk
significant fuel overheating and hence no possibility of any significant adverse public health effects. However, the initiation of ECI leads to downgrading of the heat transport system heavy water by the light water of the ECI system. Also, because the ECI system is common to all four units, a multi-unit shutdown for an extended period of time is possible. This can entail significant economic cost for the utility. Thus, the use of economic risk as an importance measure produces a somewhat different result from other measures, in that relatively high-frequency/low consequence FDCs dominate. Initiating Event Analysis of Important Fuel Damage Categories The major task of DPSE study was to identify various sequences of events that lead to the release of radioactivity and to calculate their frequencies of occurrence. Typically, each such sequence of events is the result of an initial malfunction, or initiating event, followed by failures of other functions or systems designed to mitigate its effects. Two of the most important fuel damage categories will be broken down to examine the principal contributors in terms of initiating events. FDC3 has one of the largest collection of cutsets (a cutset is a combination of primary event failures that can lead to the occurrence of the top event) and, according to Figure 3a, is the dominant contributor to the quantified public health risk. We will also take a closer look at FDCO, potentially the most severe of all FIKs in terms of accident consequence. FDC3. FDC3 is characterized by fuel failures in multiple channels in the long term (more than one hour after trip), with the moderator as heat sink for a prolonged period and with the primary pressure low. The time delay leads to a significant reduction In decay heat and, hence, a relatively minor degree of fission product release of the order of the core free inventory, about 1% of core inventory. The category includes five general types of event sequences: (a) Very small loss of coolant events combined with failure of emergency coolant injection (ECI) on demand;
_
FDC 6 21.4%
FDC FDC FDC FDC
5 3 2 9
1.9% 9.7% 2.9% 50.6%
FIGURE 3b Relative Contributions of FDCs to Economic Risk
(b) Loss of coolant events pressure ECI recovery:
and
failure of low
(c) Small loss of coolant events and failure to reduce steam generator pressure and thereby, HT pressure, to a value below the high pressure ECI pump shut-off head; (d) Loss of main and auxiliary feedwater, shutdown cooling and emergency service water to the steam generators, leading to multiple pressure tube failures (so-called "loss-of-heatsink" sequences); (e) Failure of heat transport circuit cooling when operating in the long-term shutdown cooling mode. FDC3 contains a total of 5062 cutsets generated. For this discussion the cutsets were further truncated at 10~*>, which reduces the number of cutsets to 104 which contain 57 failure events. The maximum cutset frequency is 4.1 x 10"5an( j t n e sum of the 104 cutsets' frequencies is 3.3 x 10"^. Thus the 104 cutsets represent 70% of the total frequency of 4.7 x 10~ 4 for FDC3. These cutsets involve 18 initiating events (IE), which can be grouped for convenience into pressure tube rupture (IE-PTF), all out-reactor small loss of coolant accidents (IE-SLOCA), small main steam line break (IE-SSLB), large channel flow blockage (IE-LFB), feedwater line break (IE-FLB). and shutdown cooling system failure while shutdown (IE-SDC). Some initiating events appear in more than one cutset, e.g., IE-PTI* is involved in 25 cutsets. The importance of an initiating event can be measured as a ratio of votal frequency of cutsets with and without the initiating event (i.e., with the frequency of the initiating event set to zero). Figure 4 shows the relative importance of the initiating events to FDC3. We note that Figure 4 is based only on the flrsc 104 cutsets of FDC3 but, because most of the truncated cutsets involve low-frequency combinations of these same events, the result obtained is considered representative of FDC3. The most important initiating events are found to be small LOCAs (IE-S1.OCA and IE-PTF) which can lead to sequences involving failure of ECI recovery. Small steam and feed line breaks are also CNS 9th ANNUAL CONFERENCE, 1908 431
s ign i f icarit . leading to loss of heat sink cis described in (cl) above.
OT cutsets,' frequencies is 3.0 x 10 6 . Thus the OT cutsets represent 7-i whole.
sequences
FIXX). FIXO is potent i
up
of
the
following
These < niseis involve )") initiating events, which can be grouper! into pressure tube rupture (IE-PTF), loss of low pressure service water (IE-LOI.PSW), large flow blockage (IF.-I.FR), moderator failure (1KMODF), loss ot switchyard (IF.-l.OSWYD), main steam line break (IF SSI.H2), loss of end-shield coolinq (IF i.OKSC). loss of reactivity control from the guaranteed poisoned shutdown state UF.-GPSS), failure on demand to shut down reactor (JE-SD-AI.L), shutdown coolinq failure while in the shutdown mode UF-1.O1MV SIX) and end fitting failure (IE-EFL). For FDCO, the IF-PTF is the most prevalent event, involved in /!9 cutsets. Figure 5 shows the relative importance of the initiating events in FDCO.
generic
(a) Failure of the moderator system to maintain core cooling foJlowing a I .OCA coincident wir^ loss of KCf, in the short or long term; (b) Failure to shutdown the reactor by all possible means following events requiring reactor shutdown to prevent core damage; (c) Failure to keep the reactor subcritical in the long term, or prevent uncontrolled reactivity insertion when the reactor is started up after a long shutdown;
As in F1X1. pressure tube failure (ffc'-PTF. 47%) is found to he important, in this case because it is postulated that some modes of pressure tube/calandria tube failure can lead to moderator leakage. Sequences involving Joss of Jow pressure service water (IK-I.OI.PSW, 12%) may induce a LOCA and fail tu provide cooling to the moderator should F.CI fail. A loss of switchyard buses (IF.-LOSWYD, 4.4%) accompanied by other electrical system failures could cause total loss of heatsink and ECI failure. All these sequences are of type (a) above, clearly the most important type of severe accident.
(d) Failure to prevent loss of core structural integrity following losses of end-shield cooling and moderator cooling; and (e) Failure to maintain HT inventory when operating in the long-term shutdown cooling mode. FDCO contains a total of 2036 cutsets. For this discussion the cutsets are further truncated at 10~8, which reduces the number of cutsets to 69 containing 105 basic events. The maximum cutset frequency is 2.0 x JD"^ and the sum of the
SLOCA SSLB LFB PTF SDC FLB
Out-reactor Small Loss of Coolant Accident Small Main Steam Line Break Large Flow Blockage Pressure Tube Failure Shutdown Cooling Failure Feedwater Line Break
IE-SLOCA 45.5%
—
IE-PTF 17.7%
Others 1 % IE-SDC 6% IE-FLB
6.5%
IE-LFB 10.8% IE-SSLB
12.5%
FIGURE 4 Relative Importance of Initiating Events to FDC 3 432
CNS 9th ANNUAL CONFERENCE, 1988
Other significant contributors are moderator system initiating events (IE-HODF, 13%), which are somewhat speculative sequences involving the possibility of a VIJ deflagration in the moderator
GPSS LFB MODF LOINV-SDC LOSWVD LOESC SSLB2 LOLPSW SD-ALL PTF
Loss of Reactivity Control from the Guaranteed Poison Shutdown State Large Flow Blockage Moderator Failure Loss of HT Inventory while in Shutdown Cooling Mode LOSS of Both Switchyard Buses LOSS of Endshleld Cooling small Main Steam Line Break LOSS of Low Pressure Service Water Demand to Shutdown the Reactor Pressure Tube Failure IE-LOLPSW 12.0%
IE-GPSS
2.0%
IE-LFB
6.7%
Others
1.0% —
IE-MODF
IE-LOINV-SDC
'
/
IE-SD-ALL
1.0%
12.8% —
1.0%
IE-LOSWYD 5.0% IE-LOESC 7.8% IE-SSLB2
4.0%
/
\
IE-PTF 46.7%
FIGURES Relative Importance of Initiating Events to FDC 0
cover gas volume. It is noteworthy that sequences involving reactivity excursions (IE-SD-A1.L, TE-GPSS) Tontribute only a small fraction to the total frequency. KX-P1.ANT RELEASE CATEGORIES The study used six EPRCs spanning the range from "negligible release" (EPRC5) to events with the potential to cause a "large release" (EPRCO). Each EPRC contains a number of different sequences all assessed to exhibit the same potential for public dose consequences. The calculated results for the EPRCs are given in Table 2. No detailed calculation of consequence and risk was performed for EPRCO and EPRC5. Although the estimated frequency is relatively high compared to other ERRC frequencies, the dose consequence from events in EPRC5 is comparable or less than expected normal operating releases and can be deemed negligible to risk. The significance of the EPRCO frequency result with respect to overall risk is discussed in another paper [2]. The category that dominates the quantified risk is EPRC1, mainly through sequences involving core damage (FDC3) and containment bypass. These sequences do not appear as important contributions to the total FDC3 frequency. The mechanisms for containment bypass include small loss of coolant accidents outside containment either through the heat transport pump seals or via the D2O storage tank. Some of the important considerations in the study affecting public dose confluence and risk are discussed below. Importance of Containment Failure Modes To Risk Each Ex-Plant Release Category is made up of sequences of events representing combinations of FDCs and one or more containment failure modes. The relative importance of containment failure modes to risk is dependent on the characteristics of the various FDCs and cannot easily be separated and expressed numerically.
TABLE 2:
EPRC
Mean Frequency (/Reactor-yr)
The containment failure modes represented in the containment, event tree are shown in Table 3. These can be divided conceptually into four general groups: loss of the containment function (containment bypass, CFS, C F H ) ; loss of envelope integrity resulting in increased leakage (CIA1, CIA2): failures of overpressure suppression (PRV, DOUS, V A C ] , VAC2); failure to minimize radionuclide concentrations in the long-term air discharge (AFPC, l.TPC, KTR1, FTR2). The relative importance of these modes can be assessed qualitatively and is discussed below. The first group, involving loss of the containment function, is the most important to risk. Sequences involving these failure modes are allocated either directly to the most severe consequence category, EPRCO, or are found to dominate the quantified component of risk. The next most important failure mode is failure of long-term pressure control (LTPC), the effect of which is assumed to cause the EFAD (Emergency Filtered Air Discharge) system to be permanently unavailable. This conservative assumption results in the prediction of a slow, unfiltered release from containment in the long term and is particularly Important for accident sequences which can cause initially high levels of airborne fission products inside containment. The importance of this failure mode is further enhanced by conservatism in the modelling of long-term airborne paniculate behaviour inside containment. For similar reasons, less severe failures of the EFAD system (FTR2) also appear prominently. Loss of containment envelope integrity does not result in large releases from containment if the release pathway through the EFADS can be maintained, even though it causes rapid loss of the vacuum reserve. Loss of integrity does not appear to result in significant consequence unless accompanied by other containment failure modes. Combinations involving envelope integrity failure and failures of overpressure suppression (e.g. CIA2*VAC2 or PRV) or filtration (CIA2*LTPC) are required to make these failure modes significant in terms of consequence,
PUBLIC HEALTH RISK ESTIMATES
Mean Ind. Dose (Sv)
Mean Pop. Dose (Person-Sv)
Mean Ind. Risk (Sv/yr)
Mean Pop. Risk (Person-Sv/vr)
0
4.4xlO" 6
(1)
(1)
(1)
(1)
1
9.2xlO-6
2.4X10" 1
1.3xlO3
2.2xlO" 6
1.2x10-2
2
5.7xlO" 6
5.9x10-3
3.2xlO 2
3.4x10-8
1.8x10-3
5
3
1
3
1.7xlO-
2.0x10-8
4.9x10-1
4
1.5x10-"
7.OxlO-5
1.9x10°
l.OxlO" 8
2.8x10-4
5
3.1xlO" 2
(1)
(1)
(1)
(1)
1.2xlO-
Total Quantified Risk (1 unit) (4 units)
2.9X10
2.3x10-6 9x10-6
].5x10 ' 2 6xl0" 2
NOTES: (1)
Not estimated quantitatively. CNS 9th ANNUAL CONFERENCE, 1968 433
TABI.K 3: CONTAINMENT EVENT TREE TOP EVENT FAILURE DEFINITIONS CIA1
An impairment exists which, if detected at the pressure test and linearly extrapolated, would predict a leak rate of up to 3% of containment volume per hour at full positive design pressure.
CIA2 An impairment exists, greater than CIA), up to the equivalent hole size associated with a failed personnel airlock seal (0.113 m 2 ) . CFS
Failure of containment envelope integrity resulting in effective loss of pressure control (any hole size greater than 0.113 H I 2 ) .
Containment will remain effective for a wide range of containment system impairments, including a substantial envelope impairment. Relative Contributors To Public Dose Consequence For dose calculation purposes each EPRC is represented by a single sequence chosen to be representative of the category. This choice determines the associated fission product release characteristics. An indication of the approximate relative contributions of pathways and radionuclldes to dose for each EPRC can be obtained from output options of CRAC2 [3], the consequence analysis code used in the DPSE. However, these options are limited and restrict the forms in which data can be obtained.
VAC1 Vacuum building main chamber pressure between 15 and 39 kPa absolute. VAC2 Vacuum building main chamber pressure between 39 kPa absolute and atmospheric.
TABLE 4a: RELATIVE CONTRIBUTORS TO POPULATION DOSE BY PATHWAY EPRC
DOUS Failure of dousing when vacuum structure main chamber pressure exceeds 39 kPa absolute. PRV
Failure of sufficient PRVs minimize IOP adequately.
CII.
Failure of containment integrity resulting in significant contaminated liquid release to the RAB or ESW system. Failure of containment integrity caused by an overpressure transient that peaks above 350 kPa absolute (2.5 times containment design pressure) as a result of a hydrogen burn.
FTR1
EFAD filter efficiency within the range 99.8 percent and 99 percent (for all nuclides).
FTR2 EFAD filter efficiency within the range 99 percent and 90 percent (for all nuclides). AFPC
2 4 27 45
to operate to
CFH
Failure of chemical addition to minimize airborne fission product concentrations inside containment.
L.TPC Failure to maintain an adequately filtered flow from the reactor vault to the EFAD stack enough to maintain containment at a sufficient negative pressure to prevent out-leakage for the duration of significant fission product release (30 days). although, in general, they remain insignificant to risk because of low frequency.
Cloudshine
Groundshine
Inhalation
21 10 3 2
77 86 70 53
Table 4a indicates the relative importance of each pathway to the individual dose at 20 km. This is considered to be representative of the relative contributions to population dose. Table 4b gives the relative contributors by pathway to the individual dose at 1 km. In all cases, inhalation from the passing plume is a dominant mechanism. TABLE 4b: RELATIVE CONTRIBUTORS TO INDIVIDUAL DOSE BY PATHWAY EPRC
Cloudshine 1 3 14 26
Groundshine
Inhalation
22 1 1 0
77 96 85 74
Table 5 breaks these contrilutors down further to identify individual radlonucl
Failures of overpressure suppression only result in significant consequence if the envelope integrity Is lost independently and the overpressure is accompanied by significant release of fission products from the core. Because of the improbability of these combinations, overpressure suppression failures are not significant to risk.
In the DPSE, the radioactive food chain component is assumed to be interdicted such that the food chain component contribution to dose is negligible. Other than this, no emergency protective actions were credited in the calculations.
Only for FDC0 and FDCl is the release from containment considered non-negligible with no containment failure mode. An intact containment envelope and a long-term release path through the EFADS are sufficient to effectively minimize dose consequences, irrespective of the presence of any other containment failure modes, for all FDCs.
For EPRCs 2-4, which have a significant long-duration component to the release, it was necessary to compress the total month-long release into a single puff to obtain relative data. This introduces further uncertainty as to the precise interpretation of the significance of the groundshine component. However, groundshine does
434 C N S 9th ANNUAL CONFERENCE, 1988
TABLE 5: RELATIVE CONTRIBUTORS TO INDIVIDUAL DOSE BY RAniONUCLTDE AT 1 km KPRC
Cloudshine
Groundshine
Inhalation
Kr-88 I-J34
Ba-140 (36) La-140 (20) Zr-95 (12) Nb-95 (9.5) Others (22.5)
Ce-144 Ru-106 y-91 1-131 Sr-89 Others
(26) (12)
l.a-140 (12) r-)32 (11) Others (39)
1-132 (92) Others (8)
Ce-144 Ru-106 y-91 Sr-89 1-13] Sr-90 Others
I-J32 (65) Others (35)
(40) (15) (12) (11) (9)
(13)
(44) (20) (10) (7) (7) (5) (7)
Total Ce-144 fla-140 Ru-106 Y-91 1-131 Sr-89 La-140 Others Ce-144 Ru-106 y-91 Sr-89 1-131 Sr-90 Others
(28) (11) (10) (9) (8) (6) (6)
(22) (42) (19) (10) (7) (7) (5)
(10)
Xe-J33 Others
(94) (6)
I-J31 (43) Others (57)
1-131 (93) Others (7)
1-131 (80) Xe-133 (14) Others (6)
Xe-133 Others
(95) (5)
1-131 (65) Others (35)
1-131 (97) Others (3)
1-131 (72) Xe-133 (25) Others (3)
Others 10% Sr - 89 7%
Ce -144 42% • Ce-144 28%
Y-91 10%
Ba-140 11%
1-131 7%
Ru -106 10% Y - 9 1 9%
Ru-106 19% Sr - 90 5%
EPRC1
EPRC2
FIGURE 6a Relative Contributions to Individual Dose by Radionuclide at 1 Km
not appear to be a significant dose contributor for these categories. As expected, the representation of EPRCs 1 and 2 as uncontrolled, unflltered releases results in the predominance of long-lived particulates via the inhalation pathway as the main contributor to dose. The presence of some radlonuclldes (e.g. l.a-140, Ce-144) Is believed to be more a result of simplifying assumptions with respect to radionuclide release from fuel and modelling of paniculate transport inside containment, rather than any new insights into the volatility of these species. It also indicates that attention must be paid to modelling assumptions which represent the transport of fission products In paniculate form. For the more probable sequences represented by EPRCS 3 and 4, the main release mechanism is long
term exhaust of containment atmosphere via the EFADS. In these cases the more familiar nucildes Xe-133 and 1-131 dominate, indicating the effective removal of particulates from the discharge. CONCLUSIONS This paper has presented some of the major quantitative results of the DPSE. Emphasis has been given to Identifying the types of sequences which dominate accident frequency and risk. In addition, some important aspects of the related containment behaviour and radiation dose calculations have been identified. The dominant contributors to accident frequency and risk are found to be slowly-developing sequences mostly involving small loss of coolant initiating events. For fuel damage, the most important C N S 9th ANNUAL CONFERENCE, 1988 435
Others 3%
Xe-133 14%
Xe -133 25%
Others 6%
1-131 80%
1-131 72% EPRC4
EPRC3
FIGURE 6b Relative Contributions to Individual Dose by Radionuclide at 1 Km mitigating system Failure is a loss of emergency coolant recirculation some time during its required period of operation. For risk, it is those loss of coolant accidents whose leak path bypasses containment which, in the event of fuel damage, allows direct release of fission products outside containment.
REFEREMCES
If the accident sequence results in fission product release into containment, the subsequent contribution to risk is small, even for a wide range of containment system impairments. If the predominant release pathway is through the filtered air discharge system, the radiological releases are small and involve only the most volatile and inert fission product species. If the release pathway is such that filtration does not occur, the calculation of risk is very sensitive to assumptions regarding particulate fission product release and transport.
(2) Dinnie, K.S., "A Review of Quantitative Criteria For Demonstrating Nuclear Power Plant Design Adequacy", Paper to be presented at the Canadian Nuclear Society 9th Annual Conference, June 12-15. 1988, Winnipeg, Manitoba.
436 CNS 9th ANNUAL CONFERENCE. 1966
(1) King, F.K., Raina, V.H. and Dinnie, K.S. "The Darlington Probabilistic Safety Evaluation - A CANDU Risk Assessment", presented at the Canadian Nuclear Society 8th Annual Conference June 14-17, 1987, Saint John, New Brunswick.
(3) Richie, L.T.. Johnson, J.L. and Blond, R.H., "Calculations of Reactor Accident Consequences Version 2: CRAC2", NUREG/CR-2236, February 1983.
A DJSCUSSION OF INSTITUTIONAL FAILURE AND ITS IMPORTANCE TO NUCLEAR SAFETY David Mosey Nuclear Studies and Safety Department Ontario Hydro Keith Weaver Shaftesbury Scientific Limited ABSTRACT In this paper, we will attempt to show from a review of selected past accidents, both nuclear and non-nuclear, that there exists a class of failures which is important enough that it should be identified and dealt with as a separate group in its own right. We have called this class of failures "institutional failure". We define It in general terms as being the absence or malfunction of some corporate activity which is necessary for nuclear safety. Such absence or malfunction results from human error (in the general sense) where the error can be remote in time, place and corporate hierarchy from the point at which the physical threat to safety occurs. We will offer evidence In supvort of both our contention that this phenomenon exists and our definition of it and we will distinguish it from other recognized classes of failure. The operation of institutional failure In past accidents will be indicated and relevant recent literature will be cited and discussed. We will also identify some mechanisms which exist in one organization (Ontario Hydro) and which could be applied to the management of institutional failure and its causes. While our work is preliminary, we are able to draw some limited conclusions about institutional failure and how it might be dealt with. THE NATURE OF INSTITUTIONAL FAILURE Hardware reliability has been a major concern of all industries which use complex equipment or demand high levels of safety. The concern with highly reliable equipment in space and aeronautical applications led directly to the development of techniques such as fault and event trees which are now used widely to study the safety of nuclear systems [1]. The availability of such powerful analytical techniques, the need to explore events with remote probabilities of occurrence and the acknowledged need to be concerned with component reliability have all led to a concentration of effort on equipment failure and the roles of these failures in accidents. Practical experience and the study of hypothetical nuclear accidents have led to the Identification of three main classes of basic failures which are broadly deemed to be the causes of accidents: mechanical failure, human error and catastrophic natural events [1]. These classes of failure have been used to categorize, according to cause, the major nuclear accidents which have been experienced (for example in [2], [3J and [4]). These three classes of events would appear to
include all the possible initiators. We consider that they do not reflect fairly all the causes of accidents. Hazardous situations or accidents cannot occur without the action of some initiating event. To keep plants free from such events, and to mitigate the effects of these events when they do occur, a number of activities have been established which can have a direct physical impact on the plant. Examples of these are the functions of control room operators, maintenance and inspection crews, and safety anjlysts. These depend In turn on a number of corporate functions such as recruitment, training, and the provision of effective management, reporting and supervision structures. It is our contention that failures in such corporate functions can result in a significant, even a major Impairment in plant safety, since they can have a direct Impact on the activities which serve to maintain or enhance safety. The following section will provide evidence for this contention based on a review of several accidents both within and outside the nuclear industry. SURVEY OF PAST ACCIDENTS Eight accidents will be reviewed, including two non-nuclear high-consequence accidents. It is important to emphasize that, particularly with respect to the earlier reactor accidents, there is no intention to imply incompetence on the part of those involved. Our objective is, while applying the advantages of hindsight, to use these events to illuminate a current potential problem. The NRX Accident In December 1952 the NRX research reactor at Chalk River was severely damaged in a power transient. In the course of preparing for a low-power experiment a combination of human and mechanical failures caused the reactor to become divergent and a number of the shutdown rods failed to insert when invoked. The experiment required that a number of fuel rods (including several symmetrically arranged in a central region of the core) be provided with temporary cooling connections and consequently lower than usual coolant flows. When reactor power reached about 17 Htfth these rods boiled dry resulting in a reactivity insertion of about 2.5 mk. Moderator draining was initiated and the reactor was brought to zero power, but not before an estimated peak power of about 90 HWth had been reached. The reactor vessel was damaged with fuel melting
CNS 9th ANNUAL CONFERENCE. 1988 437
occurring in ?.?. locations [5]. The most frequently cited lesson of this accident is that relating to the provision of a poised, fast-acting shutdown system, independent of other reactor control functions. It is indeed true that the accident led to the identification of one of the fundamental tenets of CANnu reactor safety philosophy: however it is also true that the accident ijlustrated a form of failure that can neither be described as simple "operator error" nor "mechanical failure". As Lewis noted in (5]:
concluded that the fire began during the second nuclear heating which was too soon and too rapidly applied. This caused the failure of one or more uranium fuel cartridges which smouldered and gradually caused the fire. More specifically, the Committee pointed out that: (a)
the fuel temperature thermocouples were positioned to record maximum fuel temperatures under normal pile operation. Under Wigner release conditions, control rod positioning to maximize temperatures towards the lower front of the pile 'jvould mean that the thermocouples would not be seeing the highest fuel temperatures in the core;
(b)
the Pile Physicist had available to him no Operating Manual of any description -- the only written document on Wigner release operations was a 12-line Minute relating to uranium temperatures at which certain actions were to be taken to control air flow through the pile:
(c)
that while some thermocouple readings were falling and some gave no indications of Wigner release, a "substantial number" of thermocouples showed steadily increasing temperatures;
(d)
second nuclear heatings had been applied during three previous Wigner releases, in 1954, 1955 and 1956. On the first two occasions the second heating had not been applied until 24 hours after the last regional temperature rise and after all graphite temperature readings were falling. In the 1956 operation all graphite readings except one were falling.
"To reduce the risk of human error and mechanical failure, no doubt a better system of review and inspection should be established. This should relate the design considerations to the current practice". In the case of the NRX accident existence and application of the appropriate structures and procedures might well have: (a)
precluded operation of the reactor in a degraded condition, with the shutoff rod removal interlocks out of service;
(b)
established formal written procedures for manipulation of the air control valves of the shutoff rod system;
(c)
given greater consideration to the reactivity implications of operation with a symmetrical concentration of fuel rods with very low coolant flows in a central region of the core.
Structures and procedures to meet the needs identified by Lewis were in fact instituted following the NRX accident with the formation of a "Safeguards Group" at Chalk River >o review proposed reactor experiments, a group which subseguently became the Reactor Technology Branch. The Windscale Reactor Fire On October 10. 1957, Windscale Pile No. 1, an air-cooled graphite moderated plutonlum production reactor caught fire as a result of operations to release Wigner energy. Following initial preparations the reactor was made divergent to generate nuclear heat for triggering the Wigner release. Nine hours later (October 8) the Pile Physicist felt that the release was not proceeding as it should, so the reactor was made divergent again at 11:05 and maintained at low power until shutdown at 17:00. During October 9 graphite temperatures varied considerably but showed a general increasing trend. Attempts to control temperature were initially partially successful but there were soon indications that some fuel had failed. Eventually visual inspection revealed that in a total of about 150 channels the fuel was at red heat. Use of COj on the morning of October 11 had no appreciable effect. At 08:55 water injection was begun and by 15:10 the following day the pile was cold. In their report to the Chairman of the United Kingdom Atomic Energy Authority, the Committee of Inquiry chaired by Sir William Penney [6] 438 C N S 9th ANNUAL CONFERENCE. 1988
The principal technical defect contributing to the accident, the Committee determined, was the inadequacy of instrumentation for the "safe and proper operation" of a Wigner release. In addition, however, the Committee identified the almost complete lack of operating documentation as a "serious defect" and pointed to a number of "deficiencies and inadequacies" of organization including a lack of definition of division of responsibilities and "undue reliance on technical direction by committee". As well, it was noted that the operations staff at Windscale were "not well supported in all respects by technical advice", which had resulted in a tendency to push pile temperatures higher without complete realization of all the technical factors involved. The SL-] accident Located at the National Reactor Testing Station, Idaho Falls, SL-1 was the prototype model of a 3 MWth boiling water reactor designed to provide electricity and space heat for military installations in remote arctic locations. The reactor was destroyed in January 1961 as a result of a power transient caused by the almost complete manual withdrawal of a central control rod. A feature of this reactor was that withdrawal of its central control rod was sufficient to make the reactor critical.
While the cause of the accident was determined to be the inexplicable action of an operating crew member [7] it is important to note that a number of undesirable conditions had developed with respect to the reactor and its operation. Designed by Argonne National Laboratories, the reactor's fuel and cladding used a new aluminum-nickel alloy (X-8001; -- the first power reactor application of this material. Thompson [8] notes that "developmental problems" with this alloy resulted in the burnable boron poison (reguired to ensure the specified three-year core life) being incorpoiated in discrete strips tack-welded to fuel elements, rather than dispersed throughout the fuel matrix. The evidence by F. Pittman in [9] notes the novelty of this approach. SL-1 achieved first criticality August 11, 1958 and, under ANI. direction, testing continued to February 5, 1959 when the reactor was handed over to Combustion Engineering for operation as a test, demonstration and training facility. Early questions were raised by the Army Reactors Branch of the USAEC about the adeguacy of the technical documentation provided by ANI. and subsequently it was noted that reactor operating procedures "completely satisfactory to the ARC have never been completed" (Nelson in [9]). While operation of SL-1 was the contractual responsibility of Combustion Engineering, actual operating crews were military personnel. These crews operated the reector on an around-the-clock basis, but with supervision by CE staff only during conventional working hours. The normal operating complement was two. Training of operators was provided by the military, examinations were administered by the military and CE and formal certification was via memo from CE staff. On duty at the time of the accident were: one shift supervisor ("Chief Operator"), one certified operator and one trainee. (W. B. All red in [9]) Between January and April 1959 some control rod sticking was experienced (rods failed to drop cleanly into the core). Modifications were made to the rod drive seals and the seal water circuit (R. Morgan and V. Hendrix in [9]). During an inspection in August 195?, it was noted that the boron strips were noticeably bowing out from the sides of the fuel assemblies. At about this time. Combustion Engineering made the following recommendations to the AEC concerning SI.-l: (a)
For "field application, use of aluminum was not sufficiently advanced" (H.C. Schrader in [9]) and recommended stainless steel as the preferred material;
(b)
A new SL-1 core should have an adequate shutdown margin with any one rod withdrawn:
(c)
The control rod drive mechanisms should be redesigned.
These recommendations were endorsed by the AEC and arrangements were made for a new core to be available in Spring 1961 (F. Pittman in [9]).
In an August I960 inspection serious deterioration of the boron strips was observed. The deterioration was such that fewer fuel assemblies than planned could be removed for inspection for fear of further boron loss [10]. Calculations and tests suggested that 18 percent of the boron had been lost and that withdrawal of the central rod by 14.3 in would establish criticality. This was reported by Luke and Cahn in [10] and reviewed by CE's Nuclear Division Safety Committee in November 1960. It was also noted in the November 1960 issue of the Industry periodical "Nucleonics". To compensate for the boron loss, cadmium strips were inserted in two unused control rod positions in November. After this modification, operation of SL-1 at a higher power level (4.7 MWth versus 3 HWth) began (Nelson in [9]). In November and December ]960 marked deterioration in control rod performance was experienced, with 33 cases of sluggish or sticking rods being recorded in the operations log. Full rod travel exercises were instituted by the plant superintendent in December in an effort to ameliorate the situation. The SL-1 Project Manager remained unaware of this problem. The accident took place at about 21:00 on January 3, 1961. The centra] control rod was withdrawn about 20 in. (criticality would have been reached at 16.7 in.) and reactor power abruptly rose to an estimated 20,000 MW. Violent destruction of the central core region terminated the transient [11]. There were a number of occasions in the course of SL-1 operations when a safety review would seem to have been indicated (a)
the 1959 recommendations by CE (accepted by the AEC) for a redesigned SL-1 should have prompted a review of the reactor and its continued operation;
(b)
the continuing problem with loss of boron poison, and the decision to overcome this by the use of cadmium absorber strips, should have been subjected to a full safety review; subsequent operation of the reactor at a higher power level than before was permitted without "adequate hazards review" (A.R. Luedecke in [9]);
(c)
the control rod problems in late 196U were not reported to project management although they are recorded in detail in the SL-1 Operations Log. Allred in [9] has noted that this deteriorating rod performance should have prompted shutdown of the reactor pending resolution of the problem.
In [8] and [9] the organizational complexity of the SL-1 operation -- through design, construction, commissioning and operations — is cited as a key contributor to the lack of response to indications of a deteriorating safety situation. In [9] attention is drawn 10 the difficulty in ascertaining the allocation of responsibilities between the many organizations and departments, both civilian and military,
C N S 9th A N N U A L C O N F E R E N C E , 1988 439
involved. Thompson notes in [8] that while there were numerous people with, in principle, some safety responsibility at various periods in the reactor's life, there was no single group with continuous safety responsibility, totally cognizant of the situation with the authority to take decisive action. The same author also emphasizes the importance of unequivocal definition of safety responsibility -- "a line organization should be used, not a committee". The Fermi-1 Fuei Melting Accident Located near Munroe, Michigan, the Fermi-1 reactor was a 100 HWe liquid metal cooled fast breeder reactor. On October 5, 1966, during operation at a power level of 34 HWth, increased radioactivity levels were detected in the sodium coolant's argon cover gas system and subsequently the building ventilation system [12]. The reactor was shut down and subsequent investigations revealed that two fuel assemblies had suffered extensive melting damage. By the end of 1968 it had been determined that a zirconium liner plate had broken free of a conical flow guide, and caused a local coolant flow blockage [13]. The zirconium liner plates were intended to provide added protection against melt-through in the event of a loss of coolant accident. They were installed, in response to concerns raised by the regulatory authority (the Advisory Committee on Reactor Safeguards of the USAEC), late in the construction phase (in 1959) and did not appear on the "as built" plant drawings. From discussions in [14] it seems that installation of these plates was carried out because it appeared to be the most economic method of obtjjning licensing approval from the regulatory authority. The failure to document this last-minute design change was a notable omission. o p greater seriousness was the fact that, as h. Shaw notes in [14], the prospective operators accepted an SCRS suggestion as a "mandate" rather than "debate the issue on a technical basis". The decision to install the liner plates without rigorous technical review suggests that there existed a lack of clear perception as to who had responsibility for safety. Three Mile Island Unit 2 On 28 March 1979 the TMI-2 reactor suffered a total loss of boiler feedwater. Auxiliary feedwater supplies were not forthcoming as isolation valves had been left closed. A relief valve on top of the pressurizer failed open allowing a steady discharge of primary coolant water to containment. Misdiagnosis of the situation by the reactor operators led to their throttling injection water flow with the result that core cooling was seriously impaired and the core destroyed. Numerous factors contributed to the accident, including mechanical failure (the failed open PORV), design problems (for example. Inadequate instrumentation and poor control room layout) and human factors (auxiliary feedwater isolation
440 CNS 9th ANNUAL CONFERENCE, 1988
valves left closed, injection water flow throttled etc.). It appears that the training provided the reactor operators was inadequate. Taught to use pressurizer level as an indicator of RCS level in the core their responses were in accordance with their training but not appropriate to the actual reactor conditions. However, as G.M. Ballard points out in [3], a PORV had failed open et the Davis Besse plant two years earlier under very similar circumstances (although the reactor was at relatively low power at the time) and it was recognized that, this could result in misleading inferences being drawn from pressurizer level readings. That the implications of the Davis Besse event were not effectively communicated to the operators of the very similar Three Mile Island reactor was a serious communications failure within the overall organization of the nuclear power community in the US. Chernobyl Unit 4 The Chernobyl reactor was destroyed in the course of a power transient during an attempted test. The Chernobyl accident is frequently discussed in terms of design vulnerabilities and operator error (for example in [4] and GiUus in [15]). The use of "operator error" in this context is highly misleading. At Chernobyl, the sequence of events leading up to the power transient involved a series of systematic and deliberate violations of operating poli-ies and principles in an effort to carry out the planned test. As noted in [16], the work plan for the turbogenerator run-down test was not properly prepared, had not received requisite approval and paid essentially no detailed attention to the safety implications of the test. The established approvals process was effectively bypassed. This breakdown points to problems in nuclear safety management which, as Tanguy argues in [15] must have pervaded the plant up to senior management levels, and "maybe going even higher". The Fllxborouqh Explosion Just before 17:00 on June 1. 1974 the Flixborough works of Nypro (UK) Limited were virtually demolished by an explosion of cyclohexane vapour. The report of the formal investigation [17] found that the cyclohexane release was most probably due to the failure of a temporary pipe connection between two cyclohexane reactors. The pipe connection had been installed to bridge the gap left when ore (No. 5) of a series of six reactors was removed from service after a crack was discovered in it. The report drew attention to the following: (a)
No cons.deration was given to inspecting the other five reactors for similar cracks;
(b)
There was at the time no properly qualified works engineer at the plant, who would have been in a position to insist on such Inspection!
(c)
(d)
No engineering drawings of the temporary pipe connection had been prepared and as constructpd It was of unknown strength and did not comply with the relevant British Standard;
Uncertainty as to the actual load being carried by the ships and the actual number of passengers on board:
(c)
Inability to read draught marks upon leaving harbour and failure by the Company to provide draught gauges.
The connection was not pressure tested to the full safety value pressure of the system.
In U s conclusions, the Report noted that Fllxborough demonstrated how through "accident, mishap and misadventure the stage may be unconsciously set for disaster" and the blame for the accident "must be shared between the many individuals concerned, at and below Hoard level".
Loss of mv "HeraId qf_ Free Enterprlse" ShortJy after leaving Zeebrugge the Townsend Thoresen ferry "Herald of Free Enterprise" capsized as a result of flooding through the bow loading doors which had been left open. The assistant bosurj, whose duty it was to secure the doors, had not done so because he was asleep in his cabin. The Wreck Commissioner's enguiry [18) noted that: (a)
In October 1983 both the bow and stern doors of the "Pride of Free Enterprise" had been left open when the ship sailed from Dover because the assistant bosun had fallen asleep;
(b)
In October 1984 the Master of the "Pride of Free Enterprise" wrote to all deck officers, bosuns and assistant bosuns to note that on two occasions when sailing from Zeebrugge stern or bow doors had been left open and attention should be paid to seeing that this "dangerous situation does not occur";
(c)
Jn June 1985 Captain Blowers of the same vessel wrote to the Chief Marine Superintendent (later Director) suggesting the installation of lights on the bridge mimic panel to indicate the status of bow and stern doors, a suggestion which was treated extremely dlsmissively by the company managers. This matter was raised again by two other Captains in 1986.
The Wreck Commissioner commented that "by the autumn of 1986 the shore staff of the Company were well aware of the possibility that one of their ships would sail with her stern or bow doors open. They were also well aware of a very sensible and simple device which had been suggested by responsible Masters". In addition to drawing attention to this specific instance of an apparent failure to appreciate the significance of a clear and present hazard, the Wreck Commissioner's Report points to other problem areas, particularly;
(a)
(b)
"Unsatisfactory" and ambiguous Company standing orders;
All the accidents discussed above involve some human failures that do not appear to originate in either the control room or the design office. The interpretation of what was the "major cause" of each of these accidents can be disputed. The presence of significant contributing factors in each case (in addition to key factors or events) seems to be beyond doubt. Table 1 summarizes the situation.
FORMULATING A DEFINITION From the foregoing review, it can be seen that there are a number of ways in which an institution might fail to provide the types of safety-related support activities which it shoujd provide. A list (illustrative only) of some of these failures appears in Table 2. The items in this list are grouped into two very general classes. A fuller analysis couid be expected ro provide a more detailed classification. The examples in Table 2 can be described as Illustrating the absence or inadeguate functioning of appropriate organizational structures, training programmes, communication channels etc. In each of these examples the shortcoming is ultimately a human failure. Attributing the source of the problem to "poor organizational structures" or other generalized concepts does not point the way toward a solution. From the insights of the previous section and from the foregoing discussion, our working definition of institutional failure is as follows: Institutional failure is the impairment or absence of a corporate function which is necessary for the safety of an Installation. Such a failure is the result of human error in activities which may not be acknowledged as important to safety and may occur far from the man-machine interface. There are two objectives in formulating this definition: (1)
To question the fairness of restricting "human error" as a cause of accidents to those individuals located at the man-machine interface and to draw attention to the continuing risk of misdiagnosing the causes of accidents as a result of this restriction. A non-specific example of this is in citing operator error as a cause or contributing factor, when in fact the operators may have reacted reasonably, and when they could be expected to have reacted otherwise if some missing function had been present.
C N S 9th A N N U A L C O N F E R E N C E , 1988 441
{?.)
To movo toward increased safety by identifyinq more exactly all the causative factors in accidents and enabling the faults thus identified to be corrected.
SAFKTY FUNCTIONS OF THE INSTITUTION If the concept of institutional failure is to be of any use, it will be necessary to define more clearly Just what it is that fails and how that failure occurs. To be able to do this a much fuller understanding is needed of the ways in which corporate functions can affect activities that are clearly related to safety, and of the ways in which decisions affecting these corporate functions are made. General Characteristics There is a wide range of responsibilities necessary for ensuring the success of a large project. These range from highly specific technical functions, through project management and engineering to the senior functions of planning and strategy, maintaining an appropriate corporate structure, deciding on overall allocation of resources and so on. In the specific case of a corporation with responsibility for a nuclear installation, any function which is immediately necessary for the design, construction or operation of the installation has the potential to influence nuclear safety. The absence or impairment of any of these functions might give rise to situations in which damaging accidents become more likely. The distincrion between "initiating events" and causes of an accident is well recognized but it is of such importance to an understanding of the role of institutions in safety that this distinction is worth stressing. In a complex organization such as a nuclear utility, there are a great many functions and responsibilities which have to be co-ordinated. This immediately suggests two general characteristics which may be desirable in an institution: <])
The various functions and responsibilities and their importance should be characterized as clearly as possible. The significance of this point seems to be borne out by the review of accidents presented earlier.
(2)
Good communication is essential, since no functions are completely independent in a large organization. From our review of accidents, the term "communication" means more than open channels between individuals, departments, functional groups, etc. It should also be understood in a wider senso, including clarity and completeness of procedures, the timely availability of various types of internal and external information to people likely to need it, ready availability of accurate and up-to-date engineering and technical data, and so on.
ft.NNUAL CONFERENCE, 1988
Characteristic:; Particular to a Specific Institution Some examples of institutional failure were identified in the review of accidents, and presented in Table 2. The entries in this table may have general applicability to all organizations engaged in a similar activity but, since organizations vary, determining the significance of any such general examples for a specific institution will require some "Interpretation". This would reflect local variations, such as the different technological, political and regulatory factors involved. APPROACHES TO THE MANAGEMENT OF INSTITUTIONAL QUESTIONS IN A NUCLEAR UTILITY Organizations such as Ontario Hydro do have mechanisms which are probably capable of dealing with institutional factors, even though these mechanisms may not have been explicitly designed to do so. Some of the established mechanisms which could be used to manage institutional questions are: (1)
Nuclear Integrity Review Committee (NIRC)
(2)
Various safety analysis functions
(3)
Quality Engineering programmes
(4)
Human factors programmes
(5)
Operational audit
(6)
Significant Event Report (SER) system.
DISCUSSION As the above review of accidents indicates, there have been numerous and explicit references in reports on accidents to factors which could be called institutional. The references to these factors can be found both within and outside the nuclear industry. The importance of the phenomenon of institutional failure is at odds with the lack of a general recognition that the phenomenon exists or of a standard term to identify it. In the past few years, some recognition of the importance of the institution in safety can be inferred from the frequent appearance of the term "safety culture". As far as we can determine [19], the first use of this term in the nuclear energy context can be credited to Edmondson, but in this case it was used with a specific and restricted definition attached to it, i.e. "an environment in which safety and operation become synonymous and where safety systems and procedures are perceived by the operators as both necessary and useful and not as hindrances to the efficient discharge of their operational responsibilities". In subsequent use "safety culture" has been invoked in an increasingly generalized sense with decreasing meaning and usefulness (for example in [20] and [i]
TAB1.K ) : REVIEW OF ACCIDENTS SUMMARY OF KKY EVENTS AND CONTRIBUTING FACTORS ACCIDENT
KEY EVENTS
CONTRIBUTING FACTORS
NRX
- reactivity insertion due to dryout of fuel rods
- lack of dedicated fast shutdown system - Insufficient safety review of proposed reactor experiments
WINDSCALE
- human error: pile physicist misjudged core temperatures and initiated second nuclear heating too soon
- inadequate fuel temperature instrumentation - inadequate operating documentation - organizational deficiencies: unclear division of responsibilities: lack of appropriate reporting structure - lack of technical support and advice for operating staff
Si.-J
- human error: gross reactivity Insertion
- various design deficiencies - documentation deficiencies - lack of a clear reporting line - unclear division of safety responsibilities - general organizational complexity
FERMI-1
- mechanical failure of in-core component
- insufficient safety analysis - responsibility for safety apparently not clearly assigned
TMI-2
- mechanical failure: PORV stuck open
- inappropriate operator training
- human error: valve left In incorrect position by maintenance workers
- deficiencies in instrumentation
- human error: operators misdiagnosed reactor condition
- human factors short-comings in control room design - failure to communicate implications of Davis-Besse incident
CHERNOBYL
operator error: violations of operating principles and policies
- insufficient safety evaluation of proposed test - unqualified personnel directing reactor operations - inadequate plant management structure
FLIXBOROUGH
- mechanical failure: pipe rupture
- lack of qualified engineering staff - engineering responsibilities improperly assigned - inadequate documentation - excessive work stress
HERALD
- operator error: bow doors left open
- general disregard for safety by senior management - lack of instrumentation - unclear assignment of shipboard responsibilities - insufficient staff - excessive work stress - poor communication C N S 9th ANNUAL CONFERENCE, 1988 443
TABLE 2: REVIEW OF ACCIDENTS ILLUSTRATIVE I.TST OF INSTITUTIONAL FAILURES GKNKRAi. DESCRIPTION OK FAILURE
SPECIFIC INSTANCE
COMMENTS
U ) Organization and Management: - Organizational Complexity
Sf.-l
- Assignment of Safety Responsibilities
Windscale S1.-1 FERMI-1
too many organizations with direct involvement in the project
unclear assignment of important safety responsibilities
Herald of Free Enterprise - Inadequate Reporting Structure
Windscale S1.-1 Chernobyl Herald of Free Enterprise
- Inappropriate Staffing
- Failure to Appreciate Safety Concerns
lack of effective mechanisms for communicating safety concerns through management
Windscale
• lack of technical support and advice for operating staff
Chernobyl
• unqualified personnel involved in reactor operations
Flixborough
lack of qualified engineering staff
Herald of Free Enterprise
apparent staff shortage
Chernobyl
apparent general inattention to safety
Herald of Free Enterprise
apparent disregard for safety by management
(2) Communication: - Unclear or Inadequate Documentation
Windscala SL-1 FERMI-1 Flixborough Herald of Free Enterprise
Inadequate In-plant Communication Methods or Procedures
NRX Herald of Free Enterprise THI-2
- Inappropriate Training
444 C N S 9th ANNUAL CONFERENCE, 1988
TMI-2
documentation incomplete or absent
Recognition of the importance of institutions to nuclear safety is desirable. However, the significance of the phenomenon demands clear definitions, and the use of a term such as "safety culture" in its present portmanteau sense is ultimately not acceptable, just as a licensing requirement for reactors o be "safe" would not be acceptable. The present work has been directed toward showing that institutional failure exists and providing a general description of it. The greatest need at the present time, it seems to us, is to develop a more complete and formal statement identifying those corporate functions which have important safety implications and of the ways in which an institution can fail to provide those functions adequately. We see this as proceeding from a study of past accidents such as is reported here, but explicitly incorporating an understanding of the specific situation in which the institution finds itself and the safety-related constraints to which it is subject. In general terms, such a course of action might involve consideration of points such as the following: (a)
Any past and current weaknesses in equipment or components and how the institution responds to these situations;
(b)
The nature of the demands placed on personnel in safety-related activities:
(c)
The effects of major changes within the corporation on the provision of corporate functions which may be necessary for plant safety.
Only wher. the actual and potential failure modes of an institution are better understood can one reasonably expect to determine what can be done to avoid these failures, i.e., to define the elements of a "safety culture". It was noted earlier that an institutional failure results from human failure and that one of the chief motivations for postulating institutional failure is to be able to identify it, isolate it and find ways of preventing it from occurring. To that end, it may be desirable to consider treating such failures as a type of significant or reportable event. The only purpose in this would be formally to pinpoint the problem as closely as possible to give the best chance of correcting it.
CONCLUSIONS From this preliminary study we draw the following conclusions: (1)
There appears to be good evidence to support the postulate that what we ter.;> institutional failure exists. As it is used here, the term means that some function which is necessary or desiratle for safety is either absent or not functioning to some minimum criterion
(2)
A study of a limited number of nuclear and other accidents seems to indicate that institutional failure i.-an be identified in a significant fractior. of these accidents. Institutional failure seems to be important.
(3)
Additional study is needed to characterize the nature of institutional failure more precisely.
(4)
Mechanisms and procedures exist which appear to be applicable to the identification of actual or potential instances of institutional failure. Further work is necessary to determine how such mechanisms could be applied to the problem in a systematic way.
REFERENCES [1]
HENLEY, E.J. and H. KUMAMOTO. "Reliability Engineering and Risk Assessment", Prentice Hall, Englewood Cliffs, 1981.
[2]
"Report to the American Physical Society on Radionuclide Release From Severe Accidents at Nuclear Power Plants", American Physical Society, February 1985.
[3]
"Nuclear Risks: Reassessing the Principles and Practice After Chernobyl", Complete Proceedings of Conference held in London, England, December 1986, IBC Technical Services Limited, 1987.
[4]
SNELL. V.G. and J.Q. HOWIESON, "Chernobyl, A Canadian Technical Perspective", Nuclear Journal of Canada. Volume 1 No. 3, September 1987.
[5]
LEWIS, W.B.. "The Accident to the NRX Reactor on December 12, 1952", AECL-232, Atomic Energy of Canada Limited, Chalk River, 1953.
[6]
PENNEY, W., B.F.J. SCHONLAND, J.M. KAY, and J. DIAMOND, "Report to the Chairman, UKAEA on the Accident to the Windscale No. 1 Pile on October JO, 1957", October 26, 1957.
[7]
NELSON, C.A., "Final Report of the SI.-l Board of Investigation". United States Atomic Energy Commission, September 1962.
[8J
THOMPSON T.J. and BECKERLEY, J.G.. (eds) "The Technology of Nuclear Reactor Safety", MIT Press, 1964. Chapter 11.
[9]
"Hearings Before the Joint Committee on Atomic Energy, Eighty-Seventh Congress, First Session on Radiation Safety and Regulation, June 12-15, 1961", US Government Printing Office 1961.
[10] LUKE, C.W. and H. CAHN, "Evaluation of the Loss of Boron in the SL-1 Core". CEND-1005, Combustion F.ngineering Inc, 1960. [11) "Final Report of the SL-1 Recovery Operation May 1961 through July 1962", IDO-19311, General Electric Company, 1962.
C N S 9th A N N U A L C O N F E R E N C E , 1988 445
[12] DUFFY, J.G. and W.H. JKNS. "Investigation of the Fuel Melting Incident at the Fnrico Fermi Atomic Power Plant", Proceedings of the American Nuclear Society National Topical Meeting on Fast Reactors. San Francisco, ANS-101. April \0-l?.. 1967. [13] SCOTT, R.L., "Fuel Melting Incident at the Fermi Reactor on October 5, 1966", Nuclear Safety. Volume 12 No. 2. March to April. 1971. [14] "Hearings Before the Joint Committee on Atomic Energy, Congress of the United States, Ninetieth Congress, Second Session on General and Reactor Development Program", US Government Printing Office, Washington, 1968. [15] "Understanding Chernobyl", The Uranium Institute, London, 1986. [16] USSR State Committee on the Utilization of Atomic Energy, "The Accident at the Chernobyl Nuclear Power Plant and its Consequences", Information compiled for the IAEA Experts' Meeting, August 25-29, 1986. [17] PARKER, R.J.. J.A. POPE, J.F. DAVIDSON and W.J. SIMPSON. "Report to the Secretary of State for Employment of the Formal Investigation into Accident on June 1. 1974 at the Nypro Factory at Flixborough", HHSO, 1975. [18] HONORABLE MR. JUSTICE SHEEN, "mv Herald of Free Enterprise, Report of Court No. 8074", HMSO, London. 1987. [19] BROWN, R.A., personal communication to the authors. May 26, 1988. [20] "Summary Report on the Post-Accident Review Meeting on the Chernobyl Accident", IAEA, Vienna, 1986. [21] "The Safety of Ontario's Nuclear Power Reactors: A Scientific and Technical Review", Report jf the Ontario Nuclear Safety Review, Toronto. February 1988.
446 CNS 9th ANNUAL CONFERENCE, 1968
DEVELOPMENT OP A SAFETY ANALYSIS PROGRAM AT THE POINT LEPREAU GENERATING STATION
P.D. THOMPSON, D.F. WEEKS and S. ALIKHAN New Brunswick Power Point Lepreau Generating Station P.O. Box 10 P o i n t Lepreau, N.B. EOG 2H0
ABSTRACT Safety analysis is traditionally viewed as a des Lgn related activi ty that is performed to obtain th operating licence tor a new plant, Its primary f unction is consldered the means by which the adequacy ot the special safety systems are evaluated, and the method by which the overall plant design is re. . wed to assure that safety objectives are met. Such analyses require large computational resources, specialised technical staf t ana must be completes on a tignt time frame consistent with the project schedule. Given these cons t r a i n t s , this activity is best done by a la rge group at a cen tral oft-si te Loca tion.
- the Large volume ot" work, involve.! - the r e l a t i v e l y short time scale aval lable work to the s a t i s f a c t i o n of the AhCB complete of experienced detailed - th e w i de va r i e ty s p e c i a l i s t s required designers (hence - proximity to the i nformation) - the large computational and associated administrative resources required - awareness of relevant research and development work - familiari ty with past licensing issues and safety a nulys is approach.
In the cast* ot an operating plant where the station licence must be renewed routinely, dafety Ana lysis plays an equally important, uut somewhat modified role. Hence a slightly different approach is required to meet the need for this onyoing Satety Analysis. Besides maintaining the original set oi safety analyses and the various associated computer codes, there is an additional requirement to provide timely and cost-eftective analytical support as each station encounters i t s own particular set of design and operations related issues . The es tab 1L shine nt of a safety analysis group at Point Lepreau fuItilied these needs while at the same time optimizing the use of off-site consulting services m supporting the large scale technica1 ef tort required to address on-going topical licensi ng and safety analysis issues.
In addition, during the licensing period the a t t e n t i o n of s i t e staff was focussed on those areas in which they had the primar/ responsit>i ix ty. These areas included design review, operator training, development of the commissioning and associated Quality Assurance programs, and coordination with cons truetion on scheduling for testing and turnover of plant systems. In addition, s i t e statf were recfuired to prepare operational documentation, and manage the overall a c t i v i t i e s necessary to ontain the Ope-rating Licence .
This paper the tomation discusses the providing the enhancement of
outlines the overal1 philosophy behind of the onsite analysis group, and be net its of such a group in terms ot essential technical support and the real operational satety.
BACKGROUND Atomic Energy of Canada Limited (AECL) performed the original set of satety analysis requi red to obtain the i n i t i a l operating licence for the Point Lepreau Nuclear Generating S t a t i o n , This was done over the tune period from 1974, when s i t e approval was obtained, to 19B3 when the s t a t i o n was granted i t s full power operating licence. Although each submission to the Atomic Energy Control board (AECb) was reviewed and approved by the experienced staff of the on-site New Brunswick Power license ng group, the analysis i t s e l f was performed o f f - s i t e by the design orcjarursa t ion as pa r t of i t s def ined scope ot work. This approach waa necessary for a variety of reasons which include the to 1 lowing:
DEVELOPMENT OF ON-SITE ANALYSIS CAPABILITY As the station entered into i t s operating phase, i t became apparent that the previous process which served i t s purpose during the design and construction periods was no longer optimum for the following reasons: - the high cost of consultant services compared to in-house staffing costs - the limited benefit of work in terms of developing the u t i l i ty staf f to unders tand and properly manage the analysis activity and the associated technology fcr lonq term technical support - the difficulty in adequately controlling the scope, direction and quality of work performed off site - inaccessibility of the designer to provide the technical support in a timely manner - inabili ty of the off-site group to eas ily access station specifIC data and information on design changes, testing and operating practices etc. The f i r s t two points are cost related, while the remaining 4 items deal with the relevance or quality of the work i t s e l f .
CNS 9th ANNUAL CONFERENCE, 1988 447
'J PA(. In lee
(1)
(2)
(.11
;
. in
\ ^ i on
^
(1)
Starting in 1985, an i n i t i a l team was formed. This consisted of a small group of s i t e staff experienced in commissioning and licensing, supi»cted by expect safety analysts on secondment from the design organization.
(2)
Although eiich group membiad his own field of expertise, they tended to have a general familiarity of plant design, operations, and safety analysis, as opposed to being highly specialized in one particular discipline.
(31
Principal analysis codes were transferred at the time the group was formed. The attached staff from the design organization were experienced in the use of these codes, and were used to transfer this knowledge to s i t e staff.
(4)
A close liaison with operating staff and other members of the technical unit was established. In addition, close contact with the day to day operation of the plant was maintained through representation at the daily planning meoting, review of shift log and reviewing unplanned e«nt reports, pecvodtc teuotts on safety assessment of plant operation and design change proposals.
d)
Careful consideration was given to the nature and scope of jobs to be handled in-house. These were based on cost/benefit factors such as; will the job need to be repeated regularly? How much of the local resource is taken up with a given job? Is availability of operations staff/information an important factor? Is performing the work going to significantly contribute to s i t e staff expertise/experience? What time scale i s the work required in?
(6)
To maintain control and ensure cost effectiveness in cases where off-site consultants were used, the scope, analysis assumptions, method)logy, schedule, and l i s t of deliverables were fully defined and approved before the s t a r t of tach analysis.
(7)
To minimize the cost to HB Power where work of generic nature was involved, every effort was made to coordinate the scope and thereby benefit from associated cost-sharing with other u t i l i t i e s and in some cases, the design agency.
on-site analysis the group were to
Providing rtnjlytlc.il support in defining the operjrinq envelope ul various process parameters consistent with safety analysis. Tms l nrorm.it: i on is requJ-re.J tu
(4)
(5)
Analysi,.-J performance capabilities for primary and alternate heat sinks under specific sets ot operating conditions related to outage situations, tactonrig in such aspects as tuel cooling, recall times, and personnel protection. ijeveloping an analysis basis for preparing the necessary operating procedures to cover abnormal operation and accident situations. These procedures cover a carefully defined set ot (allure conditions with plant-wide consequences. The analysis base i s required to sinulate the expected plant response to identify the symptoms based response characteristics and define the allowable range of the critical satety parameters.
HAJOR CHALLENGES (6)
(7)
Assessing the need for research and development required to support safety and licensing issues, developing the required program, reviewing the results to develop an understanding of the phenomena involved arid i t s Impact on the safety analysis. Providing information/training services to operations and other technical unit staff on subject matters related to reactor aafety and associated analysis.
OVERALL APPROACH To provide such services* was adopted.
448
the
CNS 9th ANNUAL CONFEReNCE, 1988
following
approach
The biggest challenqe fof t h e newly formed a n a l y s i s group was t o b r i d g e t h e gap between t h e needs of t h e s i t e t e c h n i c a l s u p p o r t / o p e r a t i o n s s t a f f and what was r e a d i l y a v a i l a b l e i n terms of s a f e t y analysis. A few key factors exacerbating the situation included: (1)
For the most part, analysis was performed for hiqhly unlikely events. In the rare case where i t examined an event to which the station staff could more easily r e l a t e , highly conservative or stylized assumptions made trie analysis Hignif icantly alien to the real life scenario. In most instances these assumptions were not properly documented and hence not easily understood by people other than the safety analysts themselves.
{2)
The terini no logy employed in reporting the analysis differed to that used in the field. This ana in lead to a dif t iculty in extracting what an operations person wouId view as "useful" inf orrtidtion,
(3)
In some cases, the rationale as to why certain equipment/systems were designed in such a way, or had to meet specif ic requi rements was not adequately documented. Hence when there was a s l i g h t deviation from a given requirement, i t s associated implication on plant performance can Id not be readily assessed. As several ot the original designers had moved on to other jobs, the answers to some questions arising from plant operation coulJ not be easily addressed.
(4)
In general, the analysis codes did not model the as-bui11 plant in de t a i 1 , and si nee on ly a ]imited number of s e n s i t i v i t y studies were done i n the t i r s t place, the safety analysts could not easily respond to s i t e questions regarding potential impairments.
In an at tempt to reso1ve thes e ty pes of dif E i c u I t i e s , the newly formed s i te analytical group adopted a different approach to analysis.
NEW APPROACH TO ANALYSIS Analysis codes that could model the detailed s teady s t a t e and dynamic behaviour of both the process and control aspects of the plant were act]ui red, Major areas of the plant which were s linn lated inc Luded the reactor core, prima ry he* t transport system and i ts auxiliaries, steam/feedwater/condensate systems, and the special safety systems and several safety related systems. A development program (including documentation and verification) for each code was i n i t i a t e d . The development work was controlled so that a successive code version was derived from the previous one, so that the code (or model) evolved monotonically. The plant specific codes so developed became more useful, robust, powerful and flexible. In addition code maintenance wag made much easier. In para Llei with the code development ef forts, a data base of key plant parameters was established. This allowed for both short term and long term tracking of those parameters which can have a significant effect on analysis. This information was us*»d in two direct ways. Firstly, the variation in process parameter was factored into t r i p coverage related analysis, such that the natural variation cou Id be accommodated wi thou t compromis ing sat ety. The nature of variations cons idered include fluctuation within the normal control tolerances, variations caused by specific actions (eg. fuelling) and long term d r i f t . The aspect of process paramecer verification is important when one realizes that th so called "end of life" condition is not necessari)/ the most limiting initial condition for r.lL accidents. The second area where the data base was used, covered the code tuning and lock-on to actual plant conditions. This allowed for a better quantification of uncertainties as well as providing a wore useful tool for operations support analysis* Although developing versattle and accurate codes is an important element to performing meaningful analysis, the method by which these codes are applied is equally important.
For events which might be of interest to operations staff, the following approach was adopted. The analysis is f i r s t performed with the process and control systems responding as expected. The documentation highlights the various actions ot the major systems and attempts to identify the important alarms. If the analysis is being performed in support of a licensing issue, then the various process system act ions are reviewed to determi ne whether or not they contribute to the mitigation of the event and as to whether they would likely s t i l l be availab Le, Subsequent cases are then performed along the more classical lines. This approach allows for more useful intormation to be gi ven to the opera tor, provides meani ngful support to the development of £bnormal ^lant Operating Procedures (APUP's), and j u s t i f i e s the conservative nature of key analysis assumptions. This later point is important since, in some ins tances, process action can contribute to making the scenario more severe than otherwise. Each analysis attempts to following s i t e related information: -
incorporate
the
data base of key plant parameters data from periodic and mandatory testing most recent design changes oj>erating procedures experience from the field, including significant events
CURRENT PROJECTS This section l i s t s the major a c t i v i t i e s that are either being conducted within the group, or being performed off-site but under the di rection of the analysis qroup. i)
Operations Related Support - Analysis of Sept. 1987 liquid zone control upset where relief valve 3481-RV9 3 opened spuriously, causing a disruption of helium c i r c u i t pressure leading to a loss ot spatial control. As some of the zones began to f i l l , reactor power beyan to f a i l . In an attempt to minimize the power error, RRS called for a 2-bank adjuster outdrive. This lead to a significant top to bottom flux tilt, sufficient to activate both shutdown systems on regional overpower. - Review implications associated with the Dec. 1987 failure of Emergency Water Supply sys tern expansion joint 3461-EJ2 to determine whether this failure, and subsequent method of repair, lead to an impai rment of ECC. In addi tion, the basis for statements In OP & P concerning implications of BWS system unavailability, are being reviewed and documented. - Determine the length of time required following reactor shutdown, before the FHTS should be drained down to header level and opened, to ensure adequate alternate heat s inks. - Analysis to determine more appropriate stroke timing requirements for dousing valveH and containment isolation valves,
CNS 9th ANNUAL CONFERENCE, 196B 449
- SDS1 impairment analysis to determine the trade-off between rod drop i n i t i a t i o n delay anil number of shutoff rods required to he avai lable. - APUP support analysis. - Assessment of one-pump-per-loop, hot commissioning test results operation in this mode. ii)
zero power to support
AECB Action Item HeJUtea_ - Analysis arising out of the review of the CHERNOBiL accident (review of past methods for determining LOCft power pulses, and studying shutdown system effectiveness under distorted flux shapes). - Review of post I.OCA PHT.'J pump operation and design analysis for an automatic pump t r i p . - Analysis of Small LOCA - Loss of ECC events to further investigate pressure tube i n t e i r i t y . - Analysis of consequences arising from steam and feedwater lir>e failures in the turbine hall. - Recalculation of consequences from outlet header breaks in which Class assumed unavailable.
large IV is
l i t ) R&D Related follow-uP (Generic Safety Items) - Comparison of rtO-14 test results FIREBIRD-III Mod 1 predictions.
with
- Comparison of FIRgBIRD-III Mod 1 and CATHENA predictions for 100% ROH break. - Assessment of pressure tube circumferential temperature experiments, and generic assessment of pressure tube inteyrity. - Assessment of calandria tube steam impingement experiments and generic study of calandria tube integrity under LOCA conditions. - Prediction of 3 dimensional moderator temperature in the
steady-state
CANDU 6 0 0 . - Assessment of standing start. - Participation ISP-23. iv)
recent
CWIT
in OECU/CSNt
experiments
standard
on
problem
General
Continued development and documentation of Point Lepreau s p e c i f i c models used with SMOKIN, HUCIRC, SOPHT, FIREBIRD and PRESCON. Documentation of design d e t a i l s a s s o c i a t e d s p e c i a l s a f e t y systems I C o r i n c o r p o r a t i o n t°90 Safety Report R e v i s i o n ) .
450
CNS 9th ANNUAL CONFERENCE, 1988
with into
CONCLUSION BY involving operations stat* awl s j a t w TO?VM«S safety to t n e maximum extent possible, the resulting analysis is not only more accurate and meaninytul, awareness and knowledge of but that t h e operatocs analysis is increased through improved saEety communication and understanding. They are therefore more cognizant of the impact of the fatlurf.-i in safety related equipment/systems and or operating procedures/practices which Itfads to improving the operational safety. real
Session 13: Radiation Applications: Medical and Industrial
Chairman: K.K. Mehta, AECL WNRE
CNS 9th ANNUAL CONFERENCE. 1988
451
TRANSPORT OF PROTOTYPE DEPLETED URANIUM CALORIMETER MODULES FOR THE ZEUS EXPERIMENT J.J. GUY DUROCHER * ON BEHALF OF THE ZEUS COLLABORATION ZEUS/IPP Canada Department of Physics University of Manitoba Winnipeg, Manitoba, R3T 2N2 ABSTRACT A major portion of the Canadian contribution to the ZEUS experiment is the design, construction and transportation of modules for the front and rear calorimeter sections -.if the ZEUS detector. This paper describes the transport of the prototype depleted Uranium modules to Switzerland for testing at CERN.
Funding for this stage of Canadian involvement in ZEUS came in the form of a major installation grant from NSERC awarded in April 1987. Design, construction and testing of calorimeter components is underway at Carleton University, the University of Manitoba, McGtll University, the University of Toronto and York University with assembly of the modules taking place at a light industrial site in Markham, Ontario.
INTRODUCTION THE ZEUS DETECTOR ZEUS is one of the two approved experiments for the HERA electron-proton collider facility under construction at DESY in Hamburg, FRG [1]. It is an international collaboration involving over 300 scientists from 10 countries. Canada is a major partner in this effort, providing ten percent of the personnel and paying nine percent of the cost of building the detector to be used in the experiment. Canadian scientists are presently working on the depleted Uranium-scintillator calorimeter and the third level trigger of the data acquisition system. Along with N1KHEF (the Netherlands), the Institute of Particle Physics (IPP) of Canada is primarily responsible for the engineering design, construction and transport of the forward and rear calorimeter sections of the ZEUS detector. Support is being provided by groups in Germany, Spain, Japan, Israel and the United States.
The primary design consideration of the ZEUS detector [2] (fig.I) was accurate' identification and measurement of leptons and jets of hadrons produced in e-p collisions. It includes as part of its design, a state-of-the-art sampling calorimeter [3,4,5] to aid in particle identification. The calorimeter region of the detector is in the shape of a closed cylinder and consists of forward (FCAL, proton direction) and rear (RCAL, electron direction) end caps along with the barrel (BCAL, covering the lateral sides of the cylinder) unit. Depleted Uranium (DU) is used as an absorber and plastic scintillator, read out via plastic wavelength shifter bars, is the detector material. Calorimeter sections are constructed from alternating stacked layers of scintillator and DU, assembled in long modules (fig.2) and organized into towers to allow for readout of the scintillator signals by photomultiplier tubes (PMTs).
FIGURE 1. The ZEUS Detector CNS 9th ANNUAL CONFERENCE. 1988
453
sized modules at Ihe European Centre for Nuclear Research (CERN). A second test site will subsequently be established at DES Y.
TABLE 1 PARAMETERS OF THE ZEUS CALORIMETER Module Type Number Readout Channels DU Plates DUMass Scint. Mass
Irnglh thlrttri
FCAL 24 4264 4200 171t 8t
BCAL 32 5184 3808 230t lOt
RCAL 24 3392 2230 98t 5t
TRANSPORT OF THE PROTOTYPE MODULES Schedule One of the main criteria affecting the shipment of the prototype units was the overall project schedule. In particular, two factors came into play: availability of beam time at CERN and the start of delivery of the full sized modules to DESY in late 1988. Beam time had been scheduled on the CERN PS for testing the prototypes in mid-November 1987. It was vital that the first of the modules be in Geneva prior to that time.
Figure 2. ISOMETRIC VIEW OF FCAL MODULE Each DU plate is encapsulated in a 0.4 mm thick, hermetically sealed stainless steel jacket which serves to reduce the level of the natural radiation from the DU, thereby protecting other parts of the calorimeter and people nearby the unit as it is being assembled. The complete calorimeter module is held together by stainless steel tensioning straps which wrap around the whole device and attach to the iron frame. PMTs and electronics mount in the back beam of the iron frame and will be shipped in situ with the modules. Approximately 40% of the 24 FCAL and 24 RCAL modules will be constructed and assembled in Canada at the Markham site. Table 1 shows some of the parameters of the various calorimeter sections. These data indicate some of the problems which might be encountered in transporting modules from their assembly site to DESY. To lest the integrity of the calorimeter design four prototype modules were assembled at York University [2]. Each module measured 200 x 1210 x 2130 mm and contained approximately 1.7 tonnes of depleted Uranium. They were designed as fully functional prototypes of the full size FCAL modules using the same materials and electronics to be used in the larger units. The modules were assembled for use in a test facility for the full
454
CNS 9th ANNUAL CONFERENCE. 198B
Assembly of the full size units was scheduled to begin in early 1988. This dictated the prototype modules being shipped to CERN quickly to allow thorough testing to be carried out in late 1987. In this way any last minute improvements to the calorimeter design could be incorporated prior to the start of manufacturing. Another factor came into play in the spring of 1987 when plans for shipping of the prototype modules were being made. Problems in the delivery of the DU plates to York University meant a delay in the final assembly of the units. It appeared likely that the first module would not be ready for shipping before September; leaving only a few weeks for its transport to Switzerland. These considerations dictated the selection of air freight as the means by which the modules would be shipped to CERN. A search was begun for a commercial carrier able to transport the units and work began on the design of a container suitable for air freight. Radiation Considerations The decision to ship the modules via air brought with it a number of constraints on the design of the shipping container. One of the most significant was the allowed level of radiation at the surface of the container. Since chartering a trans-Atlantic flight to
transport each of the modules was economically unfeasible, radiation levels had lo conform to those acceptable for transportation on regularly scheduled flights. This meant the packaged module had to conform to a Group 1 classification as defined by the regulations of the Atomic Energy Control Act (6J. Depleted Uranium may be shipped under Group 1 categorization provided that it is packaged in such a way as to ensure that the radiation levels at the surface of the container do not exceed 5 microSieverts/hour at any point.
prototype modules were reasonably fragile units. The normal shocks and vibrations associated with transporting a package of this size and mass could do serious damage to the assembly. The container had to be designed in such a way as to ensured the delicate electronics, photomultiplier tubes, scintillator material and wavelength shifter bars would arrive undamaged. It was also necessary to maintain the close external and internal tolerances under which the modules are constructed and which are crucial to their operation. 1)1 Utr\it\ u . llc(j;h(
The radiation hazard presented by DU is mostly in the form of gamma rays with a small contribution from neutrons emitted as a result of spontaneous fission. The alpha and beta radiation present can easily be eliminated with a small amount of shielding. In the present case, however, the problem arises due to the quantity of radiation, rather than the quality. The total activity due to the 1.7 tonnes of DU in each of the prototype modules was estimated to be between 40 and 70 GBq [7]. Fortunately, the task of shielding for this large amount of radiation is made easier as a result of a materials used in the assembly of the modules. Initially, the alpha and beta radiation from the DU is eliminated by the 0.4 mm stainless steel (SST) cladding each piece is encased in. Figure 3 illustrates the effectiveness of the layer of SST. Radiation along the sides and back of the units is reduced to manageable levels by the structural iron frame (see Figure 2). Finally, the DU itself plays a major role in radiation shielding. Since it has such a high density (approximately 19 g/cm ), Uranium is very efficient as a radiation absorber. Thus DU plates near the surface of a module will act as a shield to radiation emitted by plates at a greater depth. As part of studies aimed at determining the amount of shielding required, measurement of radiation levels from the encapsulated DU were conducted at York using c< !i jonents of the calorimeter modules. Plates were stackud to determine the level at which selfabsorbtion begins to play an important role. A layer of 10 mm (about 3 plates) of DU was able to filter out the radiation from plates deeper in the stack. When the DU plates were interleaved with the scintillalor and wavelength shifter material the effect of selfabsorption was diminished somewhat as a result of small angle scattering in the scintillator/WLS gap. The reduction was small, however, in terms of the total effect. A number of shielding materials were also tested fci a variety of configurations to determine their suitability for use in the container. The aim was to design a container that would be as light and compact as possible so as to reduce the freight charges while still providing shielding to meet the Group 1 classification. Suspension System Although the iron back and C beams were adequate lo support the DU and scintillator material, the assembled
_,
,
" •' '
'
T •
.
• 1
Figure 3. PLOT OF DU ACTIVITY VS. DISTANCE ABOVE PLATE ILLUSTRATING THE EFFECT OF THE SST CLADDING To protect the module during transport, it was decided to mount the module and/or its primary shielding on a suspension system that would be able to effectively damp out vibrations. A number of shock absorber systems were studied with a synthetic rubber spring system chosen for the final design. The high density spring rubber [9] material chosen was in the form of a cylinder with a outer diameter of 100 mm and an inner diameter of 25 mm. Th<. material was cut into lengths of approximately 180 mm. The length of the springs was chosen to give the assembly as low natural frequency as possible when the module was in place. To determine the spring constant of the material, pieces of the rubber cylinder of various lengths were compressed using a hydraulic press and their change in length recorded as a
CNS 9th ANNUAL CONFERENCE. t988 455
function of the applied force. Figure 4 shows the force vs distance curves for three cylinder lengths. The spring constant for each length was taken as the slope of the central, approximately linear section of C3Ch curve. As one would expect for material of this type, shorter springs have higher spring constants and thus higher natural frequencies under load. Using the mass of the module encased in the primary layer of shielding material, it was possible to determine the maximum length of rubber spring material that would provide a mechanically stable base using four springs centred on the unit's centre of mass. Calculations showed that four springs with uncompressed lengths of 180 mm would result in the assembly having a natural resonance frequency of about 4 Hz. This was considered acceptable for the present case.
fasten to these top and bottom plates. Depending upon the area of the calorimeter to be covered, the thickness of the shielding varied from 6.3 mm to 38.2 mm. Since the C-arms and back beam provided effective shielding, only relatively thin material was required for that area (4B, 5B and 6B). Near the from of the calorimeter however, where only the stainless steel cladding and the self-shielding of the DU came into play, a 38.2 mm plate (3B) was needed to reduce the radiation to a safe level. The second major section of the container was an exoskeleton made up of steel angle and "1" beam and fitted with rubber shock absorbing springs (springs are not shown in Figure 5). The inner box containing the calorimeter was suspended about its centre of mass by the four springs attached to the piece labeled IS. A symmetric section of the exo-skeleton, 2S, fits on top of the inner box. Four springs attached to 2s are meant to damp out any upward excursions of the box which might occur in transit. A further 6 springs attached to sections 3S (2 each) and 4S (1 each) damped out lateral motion of the calorimeter. The exo-skelelon was assembled about the inner box with the rjbber springs under slight compression. This resulted in the assembled system having a resonant frequency of approximately 4 Hz. The lower section of the exo-skeleton. piece IS, was permanently attached to a large wooden palette (not shown in Figure 5) via six more rubber springs. The length of the 6 cylinders of rubber used was chosen to provide a coupling with a resonance frequency of less than 4 Hz, thereby providing further isolation of the calorimeter unit during transit. All lifting of the unit was done through the palette thus avoiding any direct pressure on the module. Eye bolts along the perimeter of the palette were available for lifting the unit with a crane in the event a sufficiently powerful forklift was not available.
0.00
0. 01
003
0.02
DISPLRCEfiENT
( m)
Figure 4 FORCE VS. COMPRESSION CURVES FOR THREE LENGTHS OF SPRING RUBBER Shipping Container
Shipping
The container used to ship the prototype calorimeter units was designed at the University of Manitoba and York University and constructed in the Faculty of Science Shops at the University of Manitoba. The two main sections of the container, the inner shielding box and the supporting "exo-skeleton", are shown in Figure
Upon completion of construction in August 1987, the container was assembled using large steel spacers in place of the C-arn'.s on the module. The empty, assembled container had a mass of approximately 2400 kg. In late August 1987 it was shipped by truck to York University in Toronto where the first of the prototype modules was crated for shipping to Geneva. Arrangements had been made with Swiss Air to transport the assembly in the cargo section of one of its regularly scheduled 747 passenger flights. Upon arriving at the Geneva airport, the container and module were transferred to a truck and driven to the CERN test site.
Primary radiation shielding •**.« provided by an inner box fabricated from plates of hot rolled steel. The two largest plates, labeled IB and 2B in the figure, attached directlv 'o the iron C-arms of the calorimeter. The other plates making up the box then
456
With the module encased in the shielding box and the exo-skeleton assembled around it, 19 mm plywood panels lined with 5 mm lead foil were attached lo the outside surfaces of the exo-skeleton, providing the final shielding of the calorimeter. The pjywood also served to cover the top I beams and thereby reduce the chances of an attempt being made to raise the unit using these beams as lifting points.
CNS9th ANNUAL CONFERENCE, 1988
I..
z&'-i'-' .•;•-• L i l i t . .Vii-^.-_Ea-tt_ s:
FiguieS. EXPLODED VIEW OF THE MAJOR COMPONENTS OF THE SHIPPING CONTAINER
Radiation measurements carried out at York prior to shipping and at CERN when the module arrived [7] found the levels at the surface of the container Co be well within the Group 1 classification limits. Careful inspection of the prototype module after arrival revealed only slight misalignment of a few spacers and wavelength shifter plates. After the module was removed, the empty container was reassembled using the steel spacers and returned to York. This procedure was repeated for the remaining three prototype units. SUMMARY The techniques developed &nd the container designed for shipping the prototype DU calorimeter modules bom York University to CERN were shown to be well suited to the task. All four modules arrived in CERN with no detectable mechanical damage due to shipping. That the container provided adequate vibration isolation was illustrated when shipping the fourth module to Geneva. At a point during transit the container received a blow hard enough to severely bend one of the main I beams running the length of the exo-skeleton. In spite of this, the rubber springs damped the blow adequately and the module was not damaged.
Tests have been carried out at CERN on the four prototype calorimeter modules and the results are presently being analyzed. Work has started at the Markham site on construction of the full sized FCAL and RCAL modules. The first complete module should be ready to ship by late 1988 with a production schedule of one module a month for the remainder of the project. Work has begun on the design of an appropriate shipping container for the full size modules. Since the largest modules will have a mass of over 9 tonnes, they will be shipped by land and sea rather than by air. To keep shipping costs at a minimum, it is hoped to ship two or three modules at a time in a single container. Since the modules are so massive it is hoped to minimize the mass of the container. Given the large amount of DU in each of the modules, however, minimizing the container mass will likely result in the shipment not meeting the standards for Group 1 classification under the Atomic Energy Control Act ACKNOWLEDGEMENTS This work is supported by the Natural Sciences and Engineering Research Council of Canada, and the Institute of Particle Physics of Canada
CNS 9th ANNUAL CONFERENCE, 1988
457
Members of the ZEUS Collaboration directly involved in this work as coauthors were: G.R. Smith, JJ.G. Durocher, J.K. Mayer and J.G. Lancaster, Department of Physics, University of Manitoba; W.R. Frisken, D.K. Hasell and D.R. Mosscrop, Department of Physics, York University; and G.G. Stairs, Department of Physics, University of Toronto. Member institutes of the ZEUS Collaboration are: Carleton, Manitoba, McGUl, Toronto and York of Canada; Bonn, DESY, Freiburg, Hamburg©, Hamburg(II), Julich and Siegen of Germany; Weizmann Institute of Israel; Bologna, Cosenza, Florence, Frascati, L'Aquila, Lecce, Milan, Padua, Palermo, ENEA-Rome, Rome, Trieste and Turin of Italy; Tokyo-INS and Tokyo-Metropolitan of Japan; NIKHEF-Amsterdam of The Netherlands; Cracow and Warsaw of Poland; Madrid of Spain; Bristol, lxmdon(IC), London(UC), Oxford and Rutherford of the United Kingdom; and Argonne, Columbia, Illinois, Ohio State, Pennsylvania State, Virginia and Wisconsin c.' jie USA. REFERENCES [1] "ZEUS and HERA", CERN Courier, 26. p. 16, July/August 1986. [2] KRUGER, J., Editor, "The ZEUS Detector, Status Report 1987", PRC 87-02, DESY, November 1987. [3] KLANNER, R., "Test Program for the ZEUS Calorimeter", Nucl. Insti. and Meth. in Phys. Res., A265.200-209,1988. [4] WIGMANS, R., "On the Energy Resolution of Uranium and Other Hadron Calorimeters", Nucl. Instr. and Meth. in Phys. Res, A252,389-429,1987. [5] BRUCKMANN, H., ANDERS, B., BEHRENS, U., CLOTH, P. and FILGES, D., "On the Theoretical Understanding and Calculation of Sampling Calorimeters", DESY Report 87-064, ISSN 0418-9833, July 1987. [6] "Regulations Respecting the Packaging and Safety Marking of Radioactive Materials Preparatory to Transport", Atomic Energy Control Act, SOR/83-740, Extracted from the Canada Gazette Part C, October 12, 1983. [7] ROS, E., Editor, Minutes of the ZEUS Weekly Meeting, 9/11/87, Hamburg. [9] Catalog 93, McMaster-Carr Chicago, IL,pg 2085.
458
CNS 9th ANNUAL CONFERENCE, 1988
Supply
Company,
COMPUTATION OF ELECTRON DOSE DISTRIBUTIONS IN TISSUE USING GAUSSIAN PENCIL BEAMS
G. SAND1SON, W. liUDA, D. SAVOIE, L. PAPIEZ and J . McLELLAN
Department of Medical P h y s i c s Manitoba Cancer T r e a t m e n t a n d R e s e a r c h F o u n d a t i o n 100 Olivia Street Winnipeg, Manitoba.
ABSTRACT M a g n e t i c a l l y scanned t h e r a p e u t i c e l e c t r o n beams from tlie S a g i L t a i r e Therac 40 a c c e l e r a t o r can be modelled u s i n g a c o l l i m a t e d i s o t r o p i c s o u r c e i n which the emitted electrons s c a t t e r according to Fermi-Eyges smal1 angle multiple scattering theory. This theory predicts a Gaussian spatial and angular spread of an electron pencil beam with depth in t i s s u e . A semiempirical method based on this theory can be used to derive the standard deviation J of this Gaussian with depth in a tissue-equivalent medium from broad electron beam penumbra. The results obtained with this semiempirical method at 16 and 22 Mev beam energies are compared to Fermi-Eyges theory and a range straggling modification to this theory (5), for homogeneous tissue-equivalent media corresponding to muscle, lung and bone. The semi-empirically derived values of a demonstrate that neither Fermi-Eyges theory nor the range FJ._raygling modification to this theory possesses universal validity and this may lead to significant dose computation errors in the treatment planning of radiotherapy patients. A ' f r i c t i o n ' term is i n t r o duced into the Fermi-Eyges electron transport equation to account for the effects of range straggling. This friction component i s successful in modelling the measured variation of mean square scattering angle with depth in homogeneous media. INTRODUCTION T h e r a p e u t i c e l e c t r o n beam t r e a t m e n t p l a n n i n g r e q u i r e s dose computation a l g o r i t h m s s u i t a b l e f o r i m p l e mentation on t h e small computers commonly e n c o u n t e r e d in r a d i o t h e r a p y c e n t r e s . C u r r e n t l y , most p e n c i l beam a l g o r i t h m s a r e based on the s m a l l a n g l e m u l t i p l e s c a t t e r i n g t h e o r y developed by Fermi (1) and Eyges ( 2 ) . This theory s a t i s f a c t o r i l y accounts for the p e n e t r a t i o n of e l e c t r o n s in the a i r space above p a t i e n t s ( 3 ) , b u t i t has major l i m i t a t i o n s in t i s s u e media. These l i m i t a t i o n s are due t o the theory being r e s t r i c t e d t o small anqle m u l t i p l e s c a t t e r i n g and n e g l e c t i n g the major secondary i n t e r a c t i o n p r o c e s s e s of large angle s c a t t e r i n g , d e l t a ray p r o d u c t i o n , bremsstrahlung production and e l e c t r o n range s t r a g g l i n g in t i s s u e . Large angle s c a t t e r i n g and d e l t a ray production l i m i t the success of Fermi-Eyges theory a t shallow depths in t i s s u e for reproducing measured dose d i s t r i b u t i o n data, whereas e l e c t r o n range s t r a g g l i n g becomes the important f a c t o r near the e l e c t r o n s ' range in t i s s u e . To overcome these 1 i m i t a t i o n s , e m p i r i c a l data i s u s u a l l y incorporated i n t o the Fermi-Eyges t h e o r e t i c a l framework t o modify the c a l c u l a t i o n s so as t o achieve c l o s e r agreement with measured dose d i s t r i b u t i o n d a t a . Previous workers have obtained values of the e l e c t r o n p e n c i l beam spread parameter ii from Monte Carlo c a l c u l a t i o n s {4,5) o r from semi-empirical methods, and these are then used as input data t o e l e c t r o n dose computation a l g o r i t h m s . One semi-empirical method combines the s o l u t i o n for a r e c t a n g u l a r broad beam dose d i s t r i b u t i o n computed from
the convolution of the Fermi-Eyges p e n c i l beam over the broad beam area with measured broad beam dose d a t a . Values of o a t depth in a homogeneous medium may th^n be obtained from the measurement of broad beam penumbra formed beyond the edge of a c o l l i m a t o r ( 3 ) . The p r e s e n t work provides an intercomparison over an energy range 7 t o 22 Mev between Fermi-Eyges theory, an e m p i r i c a l modification to t h i s theory to account f o r range s t r a g g l ? ng (5) and t h e broad beam --enumbra semie m p i r i c a l method of d e r i v i n g values of e l e c t r o n p e n c i l beam spread o with depth in homogeneous t i s s u e - e q u i v a l e n t media. In a d d i t i o n , a ' f r i c t i o n ' term i s i n t r o duced i n t o the o r i g i n a l Fermi-Eyges e l e c t r o n t r a n s p o r t equation i n an attempt t o t h e o r e t i c a l l y account for thu e f f e c t s of range s t r a g g l i n g . METHODS OF DERIVING PENCIL BEAM SPREAD IN TISSUE (a)
Fermi-Eyges Theory and Range S t r a g g l i n g cation
Modifi-
The Fermi t r a n s p o r t equation i s a second o r d e r d i f f e r e n t i a l equation which d e s c r i b e s t h e diffusion of e l e c t r o n s under s m a l l - a n g l e s c a t t e r i n g c o n d i t i o n s . Advantage can be taken of the r o t a t i o n a l symmetry of the problem to quote only the parameter values p r o j e c t e d onto the (z,x) plane for the t r a n s p o r t e q u a t i o n , as fo1lows :
¥*z («.x.o x, .x -c i<)x£+ £4 iff 2
i
rftj
where F(z,x,0 ) is the distribution function for the electrons, z is the depth in tissue, x is the l a t e r a l coordinate, 0 is the polar angle projected on to the {zTx) plane and k i s the linear scattering power (6) of the absorbing t i s s u e . The solution to this equation for a pencil beam boundary condition given by F = & (x). 6(0 ) at z = 0 is F(z,x,0
2(A
oVV
where with i = 0,1,2 and value of k has been considered constant. Integrating this distribution function over a l l 0 resul.s in a spatial Gaussian distribution with variance o given
CNS 9th ANNUAL CONFERENCE. 1988
459
Also integration over a l l x r e s u l t s jji, an angular Gaussian d i s t r i b u t i o n with variance ••)" givon by
d i s t r i b u t i o n over the broad beam a r e a . D (z) D(x,y,z)
After t r a v e l l i n g through the a i r d r i f t space the e l e c trons from an a c c e l e r a t o r enter the p a t i e n t ' s t i s s u e . The density of soft tissue i s about 1000 times the density of a i r and the electrons lose energy in t h i s medium a t a rate of about 2 Mev cm . The energy loss of electrons in t i s s u e may be; taken into account within the framework of Fermi s c a t t e r i n g theory (I) by u t i l i z i n g the solution due to Eyges ( 2 ) . This solution assumes that the loss of energy for the electrons comprising the pencil beam i s a well defined function of depth only in t i s s u e and i s i d e n t i c a l to Equation 2 except that the parameters A. are redefined as z 1 A. = =r 2 j k(z)(z-O d. 0
"(0)
(0)z
= - ^ — I SSD/(SSD + z) J » |ERF( ( x + a ) / / 2 o ) - E R F ( ( x - a ) / / 2 o i l
' l E R F ( ( y + a ; > / / 2 o ) - ERF( i y - a ) / fl u ) I H where D(x,y,z} i s the dose d i s t r i b u t i o n for a d i v e r ging square f i e l d of length a a t depth z in the t i s s u e , x and y are l a t e r a l coordinates, SSD is the source surface d i s t a n c e , o i s the root mean square pencil beam spread a t depth z, D^z^ is t n e central axis dose d i s t r i b u t i o n for an i n f i n i t e sized field and ERF is the e r r o r function. The c e n t r a l (x z) plane of t h i s dose d i s t r i b u t i o n is given by DJ D(x,0,z) =
with i = 0 , 1 , 2 .
In general, the d i s t r i b u t i o n of s p a t i a l spread at depth for beams which have an a r b i t r a r y Gaussian d i s t r i b u tion a t the t i s s u e surface is given by (6) + 2" x(0)z
2
ccn 2 SSD 'SSD+z'
' [ERF! (x+a)//2~o) - ERF( (x-a) / J a ) x 2-ERF(a//2*0)
9
Sandison and Hud;, (j) have derived the r e l a t i o n s h i p between the penumbra width w(z) of t h i s beam p r o f i l e and the value of o(z) as w(z) = /Tit o-ERFIa/^ 3)-(l-EXP(-2a 2 /o 2 ))
1
10
.
+• - J k ( f ) U - -
^o where the linear scattering power k(z) i s a function of depth z in the tissue due to the electron energy loss and computed according to the procedure of Jette et al (7), o (0) i s the spatial variance of the pencil beam at the tissue surface, TP (0) i s the angular variance of the pencil beam at the tissue surface, 0 x (0) is the cross correlation coefficient between l a t e r a l position x and projected angle of travel 8 to the 2-axis at the tissue surface and the integral term is the contribution of multiple small angle scattering in the tissue-equivalent medium. The theoretical values of o computed using Equation 6 do not account for range straggling. However, empirical modifications to the theoretical results have previously been suggested (5,8) to account for this effect. The modification suggested by Lax et al (5) was derived by f i t t i n g an empirical function to Ntonte Carlo data for a monodirectional pencil beam incident upor water at energies of 5, 10 and 20 Mev. The empirical multiplicative formula provided includes the bremsstrahlung dose component and is given by 0 12)
Equation 10 may be solved i t e r a t i v e l y (c)
for o ( z ) .
Fermi Theory with Friction
A friction term may be introduced into the Fermi transport equation which may be considered to have the physical interpretation of a constraint to the unlimited growth of I 7 w.lth depth in tissue. Equation 1 then becomes
|£
,1(BO F) x
(z,x,e
3Z
2
2F
X
where B i s a constant f r i c t i o n c o e f f i c i e n t . The solution of Equation 11 for a pencil beam i n i t i a l condition (see Equation 2) is also Gaussi.-n and given by )
-1
02
r
° < ^
12(1.5-9), where p(z) = expi-S*"'*"' " ) and S = O.95U/R,) and R, is the empirical range of the incident electrons. 02 = ~- [1 - EXP(-2Bz)J x 4B (b)
Semi-Empirical Method
Theoretically derived values of o based on FermiEyges theory only include small angle multiple scattering collisions. The use of such theoretical values for describing penumbra shape or the variation cf central axis depth dose in homogeneous tissue-equivalent media is limited because other interaction effects are important. To overcome this limitation, some dose computation algorithms based on Fermi-Eyges theory use values of o(z) as input data which are derived from broad beam penumbra. This method is based on the dose distribution D(x,y,z) resulting from a uniform broad rectangular incident planar fluenee of electrons which is computed by convolution of the pencil beam dose
460
CNS 9th ANNUAL CONFERENCE, 1988
= - ^ - + - i - [4-EXP(-B?) -EXP(-2Bz) - 3 ] 2B 4B
12b
and I 2-EXP(-Bz) - EXP(-2B?:) - 1 ].
According to Equation 12a, n approaches a limiting value of k/4B with depth z. XTr-ese a n a l y t i c expressions have been derived assuming k i s constant.
MATERIALS AND METHODS
The Saqittainj Thcrac 40 l i n e a r a c c e l e r a t o r was used to produce e l e c t r o n beams in the eneryy range 7 to 22 Mev. Jn-phantom beam p r o f i l e measurements were obtained both p e r p e n d i c u l a r i l y and p a r a l l e l to the c e n t r a l axis of the thurapy e l e c t r o n beam using Kodak Industrex M film. The film was retained within the manufacturers opaque paper c a s s e t t e when placed in the perpendicular o r i e n t a t i o n . However, for p r o f i l e measurements with the film o r i e n t e d p a r a l l e l to the c e n t r a l axis of the beam, the paper c a s s e t t e was r e moved in a photographic darkroom and the bare film placed in a s p e c i a l l y designed holder made from t i s s u e equivalent m a t e r i a l . Throe film holders were cons t r u c t e d of the same material as each of the t i s s u e equivalent phantoms ussd for dosimetry. These were hiqh impact polystyrene muscle s u b s t i t u t e , a r e s i n based lung t i s s u e s u b s t i t u t e known as LN1 (9) and a resin based bone t i s s u e s u b s t i t u t e known as SB3 ( 9 ) . The physical p r o p e r t i e s of these and other m a t e r i a l s are given in Table 1. The holders were painted matt black t o prevent l i g h t transmission and sealed using black i n s u l a t i n g tape. Film placed in the holder f i t such that one of i t s edges was p e r f e c t l y aligned with the top surface of the phantom. The holders were placed in the appropriate phantoms, and these phantoms were then squeezed i n a vice to exclude air-gaps and i r r a d i a t e d . Films were developed using hand processing and read manually by a Sar^ont-Welch densitometer (Densichron mode PDD) with a 1 mm light a p e r t u r e . TABLE 1:
Scattering Power* Cons cane ko c-1
1.00
0 .55509
3.3428
47.2689
PolysCyren*
1.06
0 .53768
3.4322
40.8859
Teroex
1.01
0 •5*716
3.3070
39.9887
LS
0.30
0 52>92
0.9537
12.4764
SB 3
1.84
0 5U84
5.7047
113.8676
1.205x10"* 0. 49S75
3.6264x10-3
FIGURE 1: VARIATION OF o IN POLYSTYRENE FOR A 22 MEV BEAM. FULL LINE SHOWS FERMI-EYGES THEORY AND DASHED LINE THE LAX ET AL (5) MODIFICATION. (b)
Hacer
C Z/A ) = Z F (Z/A)
56.257«10"3
" b e r e r ' " fr " ctl<>]1 by "»i»i>t of eletaenc Z t of atomic weight A^
I Linear scattering power !<(*) - k 0 . I E(E+2»OC2)
(a)
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 DEP1H. z/Ro
PHYSICAL PROPERTIES OF DOSIMETRIC MATERIALS
Density a an"3 1.1023
Air
variation of o with depth z for a point monodirectional pencil beam of 22 Mev enery incident normally upon the surface of a semi-infinite water phantom. The axes have been expressed in dimonsionless units by scaling distance in terms of the continuous slowing down approximation range r . The simple Fermi Eyges theory predicts values of o which always increase with depth z. It cannot predict the effects of range straggling which reduce pencil beam spread close to the electron range. An empirical modification of the theory suggested Lax et al (5), and given in Equation 7 to account for this effect is also shown in Figure 1.
Fermi-Eyges Theory and Range Straggling Modification
2 The general expression for the spatial variance o (z) of a Gaussian pencil beam at depth z in a medium is given by Equation 6. The f i r s t three terms of this expression depend upon parameters evaluated at the surface. These parameters define the contribution to the spread of the pencil beam provided by a i r scattering from the electron source to the tissue surface. It is necessary to evaluate these parameters since their contribution to pencil beam spread can be largt at shallow depths. Figure 1 shows the theoretical
Semi-Empirical Method
The s e m i - e m p i r i c a l values of a{z) d e r i v e d from broad beam penumbra depend upon t h e c o l l i m a t i o n geometry d e f i n i n g t h e f i e l d . Figure 2 shows semiempirLcal values of o d e r i v e d from t h e penumbra in p o l y s t y r e n e of a 10*10 cm2 f i e l d c o l l i m a t e d 11.5 cm above t h e phantom s u r f a c e for beam e n e r g i e s of 7 and 22 Mev a t an SSD o f 100 cm. only o values up co t h e p r a c t i c a l range of the e l e c t r o n s a r e computed. The full curves in Figure 2 indicate the theoretical variation of o predicted by Fermi-Eyges theory, a were derived from the penumbra of the broad beam in lung-equivalent and bone-equivalent materials for electron beam energies of 7, 16, and 22 Mev as shown in Figures 3a and 3b respectively for the 7 and 22 Mev beams. The electron range r of a 2^ Mev electron in the lung-equivalent material is about 35 cm. This is larger than the dimensions of the phantom employed and accounts for the incomplete data set shown in Figure 3a at the 22 Mev beam energy. Again the full curves represent the predictions of Fermi-Eyges theory. The semi-empirical results presented in Figures 2, 3a and 3b agree closely with the predictions of FermiEyges theory up to a depth of about 0.7 r for the incident electrons. Beyond this depth there is a marked divergence between theory and measured data. This observation has been noted previously HO) and is due, at least in part, to range straggling. Random errors associated1 with the measurement of penumbra width are relatively large for low beam energies and high density materials. For instance, measurements on a 7 Mev beam show a variation in ii(z) from the surface of the phantom to i t s maximum value CNS 9th ANNUAL CONFERENCE, 1988 461
0
FIGURE 2 :
1
2
3
*
5
6
7
y
9 - 3 11 -2
VARIATION OF -1 IN POLYSTYRENE DERIVED FROM BROAD BEAM PENUMBRA FOR 7 AND 2 2 MEV BEAMS
FIGURE 3
VARIATION OF J IN BONE DERIVED FROM BROAD BEAM PENUMBRA FOR 7 AND 22 MEV BEAMS
CHCM SIM 7
*1
?:<;•/
FIGURE 3 a :
VARIATION Of (" IN LUNG DERIVED FROM BROAD BEAM PEWUMBfW FOR 7 AND 22 MEV BEAMS FIGURE 4:
of o n l y 2 mm i n t h e bontf phantom and 4 nun i n t h e p o l y s t y r e n e phantom. This v a r i a t i o n may be compared t o a s t a n d a r d d e v i a t i o n i n tfre measurement of o under t h e s e c o n d i t i o n s of 2 mm. The a i r s c a t t e r i n g component t o t h e d a t a shown i n F i g u r e s 2 , 3a and 3b may be s u b t r a c t e d t o compare t h e experimental data to the predictions of the range straggling modification suggested by Lax et a l . Figure 4 compares the a i r - s c a t t e r i n g subtracted experimental data to this modification for a 22 Mev beam incident upon po)ystyrene . The ordinato and ,-uscissa in this plot have been made dimensionless by dividing by tho continuous slowing down approximation r . Figures 5 and 6 make the same comparison in lung and bone cgulvalcnt media for a 16 MPV beam. (r:)
Theory with Friction
Riin«j(? straqylinq affects both tho variation of 462
CNS 9th ANNUAL CONFERENCE, 1988
COMPARISON OF MEASURED VALUES OF 0 IN POLYSTYRENE WITH PREDICTIONS OF FERMIEYGES THEORY (FULL' LINE) AND EMPIRICAL RANGE STRAGGLING MODIFICATION OF LAX ET AL (DASHED LINE) ( 5 )
and the variation of <"' with det-th in t i s s u e . Roos et al (II) have measured tho parameter K2* with depth in a number of homogeneous materials of differing atomic number and density. Their results are reproduced in Figure 7 for hr-irgeneous media of different atomic number. I t is apparent that a saturation value of ~0^ , equal to about 0.65 radians , is achieved within X the range r for high atomic number materials. Equation 12a can be used to determine the of 7^ and this is shown i n Figure 7. To b e h a vvior i determine a value of the parameter B, the following approximation was used: ^ - = 0.65.
0 «
C.S
08
VO
TIGURE 7 : FIGURE 5 :
COMPARISON OF MEASURED VALUES CF ,i IN LUNG WITH PREDICTIONS OF FERMI-EVGES THEORY IFULL LINE) AND EMPIRICAL RANGE STRAGGLING MODIFICATION OF LAX t T AL (DASHKD LINE) (5)
0.0
MEASURED VARIATION OK :" WITH DEPTH IN HOMOGENEOUS MEDIA OF V/ifirOUS ATOMIC NUMBER (11)
0.1
0.2 H.3 0A
0.5 0.6 0.V 0.8 0 9 1.0
PREDICTED VARIATION OF "" USING FERMI-EYGES THEORY (FULL LINE) AND T^E FRICTION MODEL (DASHED LINE) COMPARED TO THE MEASURED DATA OF ROOS ET ALL (11) FOR CARBON AND ALUMINUM
00
3.2
0 4
COMPARISON OF MEASURED VALUES OF 0 IN BONE WITH PREDICTIONS OF FERMI-EYGES THEORY (FULL LINE) AND EMPIRICAL RANGE STRAGGLING MODIFICATION OF LAX ET AL (DASHED LINE) (5)
T h e o r e t i c a l p r e d i c t i o n s of the v a r i a t i o n of 0 , using the abovi. value of B and Equation 12a, t o r carbon and aluminum arc compared to the measured values in Figure 8 for a 10 Mev beam. The value of k was taken to bo c o n s t a n t and equal t o the l i n e a r s c a t t e r i n g power of 5 Mev e l e c t r o n s ( i . e . , the average e l e c t r o n energy over the r a n g e ) .
The variancr- of the spatial Gaussian -) for this friction model can be obtained using Equation 12b. The so re s u1ts a re s hown in Fi gu re 9 fo r a monodi ro ctional pencil beam incident upon polystyrene at 22 Mev and compared with both Fermi-Eyges theory and the semi-empirical values. DISCUSSION
The spatJ al and angular distribution of an electron pencil beam scattering in tissue may be obtained from Fcrmi-Eygcs theory. However lim { tations to this theory include the neglect of second order electron interactions. To overcome these limitations experimental or Monte Carlo derived data may be incorporated into the theory to reproduce measured dose d i s t r i butions in tissue equivalent media.
CNS 9th ANNUAL CONFERENCE. 1986 463
s a t u r a t i o n bf-I.Livior L i c h i c v i i i q •y.Ajd
• -1 ? ni
18 1 6 1 -I 1 ? 1.0 0.8 0.6 0.-1 0?
CONCLUSIONS
oo n i 0.2 •:::•> n-1 r s 0.6 0.7 n.a n 9 i n 1.1 nrpin. ?/Ro J:
PREDICTED VARIATION Of •' USING FERMIL'YGCS THEORY (FULL LINE) AND THE FRICTION MODEL (DASHED LINE) COMPARED TO THE MEASURED DATA FOR POLYSTYRENE AND A L'J MEV BEAM
The semi-empirical method of deriving values of pencil bo am spread .• is to use broad beam penumbra width measurements arid Equation 'J. Pi qure.s 2, 3a and 3b show that for d i f f e r e n t typou of media therei s c losr- agreement between Fermi -Eyqes theory and the semi -empi ri cal vul ues de te riniiied from broad beam penumbra up to a depth corresponding to about 0.7 times the cont inuoun y. lowi nq down approximation range r of t h e f-Jertrons. At • j re • a t o r de p tl i s 1 a r go d i s c re pancies a r i s e due to e L' -r.-tron range s t r a q q l i n g e ffects. r
An empi r ical formula { i) has previous ly been suggested to describe the- effects of range straggling on the pencil beam spread parameter • and thereby achiL-v closer agreement with experimental dose d i s t r i b u t i o n dat? for broad beams. Figure 4 compares the suggested modi f i cation to Form1' -Eyges theory with the measured semi-empirical values of o derived from broad beam penumbra in polys Lyreno for a 22 Mev beam. The modification gives good agreement with the measured data for t h i s p a r t i c u l a r beam energy and medium. A change of media density to lung is seen from Fiqurc b to cause the Lax et al (5) modification to ovo r e s t imatc the measured data. The bone data of Fiquro (> demonstrates that for a change in both density and atomic number the empirical modification i s worse and appears to be unable to adequately predict the measured v a r i a t i o n . This indicates that the empirica1 modi fication suggested is limited. The introduction of a f r i c t i o n term into the Fermi transport equation i s an attempt to account for the effects of range s t r a g g l i n g . The advantage of t h i s approach is that the Gaussian behavior of the pencil beam is retained making doso computation r e l a t i v e l y simple (12). Figure 7 shows that tho value of - ^ tends to saturate- at a value of 0.05 radians in high atomic number material. This behavior can be reproduced using the. f r i c t i o n model by determining the value of B from the saturation value of ~!i^ obtai nod from Equat ion 12a in the 1imit of large z. Figure 8 compares tho measured data of Roos ct al (11) to tho t h e o r e t i c a l predictions of the fri ction model and Formi-EyiH'S theory. F'.-rmi-Eyges theory p r e d i c t s an incroasing value of
464 CNS 9th ANNUAL CONFERENCE, 1988
F'-rmi -\:yi]i s Uivory i H 1 imi ti-d in i t s use for the computation of duse in t i s s u e madia because only sma 1 l-iUKjlc mu 11 ij'l '• sea Ll. i_r ir.q j nte ractions are taken into act.-ount. All other porcessos includinq larrje anqle scat tori mi, de I t.i ray production, bromsstrahlunq production and electron range- s t r a g g l i n g are i quo red . Tin- use of an empi r ical modi fication to t h i s thu-ory ti account for range straqql ing does not appear to offer s a t i s f a c t o r y improvements. A thooret ical approacli to range straggling i s 1 ikely to offer the required improvements to Fermi-Eyges theory. The introduction of a friction term in^o the Fertni transport equation accounts f^r the —y measurcd saturation beha\ r ior of t:hf- [.aram-'Lor v*", in homoqennou:; scat te ring ^ediun1. Unfortunately , t h i s f r i c t i o n model does not^predir/t the measured variation in the parameter : *". A mode] incorporat inq electron absorjJtion is currently bcin'? pursued to overcome t h i s d i f f i c u l t y . REFERENCES
(1)
ROSSI, B. and GREISEN, K-, "Cosmic-ray Theory", Review of Modern P h y s i c s , Vol. 13, pp.240-270, 1941.
(2)
EYGES, L . , " M u l t i p l e S c a t t e r i n g with Energy Loss", P h y s i c a l Review, Vo1. 74, pp.1534-1035, J'J-IB.
(3)
SANDISON, G. and HUDA, w . , " A p p l i c a t i o n of Fermi S c a t t e r i n g Theory t o a Magnetically Scanned T h e r a p e u t i c A c c e l e r a t o r " , Medical P h y s i c s , Vol. 15, No. 4, 1988.
{A)
ANDRE0, I', and BRAHME, A., "Mean Energy in E l e c t r o n Beams", Medical P h y s i c s , Vol. 8, pp.682-687, 1981.
(5)
LAX, I . , BRAliME, A. and ANDREO, P . , " E l e c t r o n Beam Dose Pi am i ng u s i n g Gaussian Beams : Improved Radial Dose P r o f i l o s " , Acta Radiolog i c a Supplement, No. 364, p p . 4 9 - 5 9 ,
(6)
INTERNATIONAL COMMISSION ON RADIATION UNITS AND MEASUREMENT, " R a d i a t i o n D o s i m e t r y : Electron Beams w i t h E n e r g i e s Between 1 and 50 Mov", ICRU Report 35 ( B e t h e s d a , M a r y l a n d ) , 1 9 8 4 .
(7)
JETTE, D . , PAGNAHENTA, A . , LANZL, L. and ROSENFELD, M., "The A p p l i c a t i o n of M u l t i p l e S c a t t e r i n g Theory to T h e r a p e u t i c E l e c t r o n Dosimetry", Medical P h y s i c s , Vol. 10, p p . 1 4 1 L46, 1983.
(8)
WERNKR, B., KHAN, F. and DEIBEL, F.( "A Model for Calculating Electron Beam Scattering in Treatment Planning", Medical Physics, Vol. 9, pp. 180-187, 198;?.
(9)
WHITE, D., "Tissue Substitutes in Experimental Radiation Physics", Medical PhyRics, Vol. 5, Wo. 6, pp.467--179, 1978.
(10)
PERRY, D. and HOLT, J., "A Model for Calculating the Effects of Small Inhomoqeneities on Electron Beam Dose Distributions", Medical Physics, Vol. 7, No. 3, pp.207-215, 1980.
(11)
ROOS, II.r DREPPER, P. and HARDER, D., "The Transition from Multiple Scattering to Complete Diffusion of High Energy Electrons", In Fourth Symposium on Microdosimetry - Euratom Eur 5122 d-e-f., 1974.
(12)
STORCHI, P. and HUIZENGA, H., "On a Numerical Approach of the Pencil Beam Model", Physics in Medicine and Biology, Vol. 30, pp.467-473, 1985.
CNS 9th ANNUAL CONFERENCE, 1988 465
W. HUDA G.A. SANDISUN
Department uf Medical P h y s i c s Manitoba Cancer T r e a t m e n t and R e s e a r c h W i n n i p e g , Manitoba
foundation
to 1987. All these data are presented below in summary form in Tables 1-3. can be con The e f f e c t i v e d o s e e q u i v a l e n t . E' veniently used to characterize the radiation delivered to patients undergoing all types of diagnostic examinations. By combining lh^ patients for two elimination categories (Computed tomography (CT) and unclear medicine) are presented, together with the corresponding per caput radiation closes. Demographic data of patients undergoing CT scans show that this population is generally older than sn average "adult" population, and that the corresponding radiation detriment is approximately 45,'; of the value expected in a similarly irradiated "adult" population.
INTRODUCTION The effective dose equivalent, H , was introduced by the Internationa] Commission for Radiological Protection (1CR11) in 1977 to take into account nonuniform radiation exposures of botli occupationally exposed personnel and members of the public. (1) The H has also been increasingly utilised as a radiation risk parameter in Computed Tomography (2) and Nuclear Medicine C3) and has been adopted as the most appropriate dose parameter in all diagnostic medical procedures by th~ United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR). (4) In this paper, the number of diagnostic procedures using ionizing radiation in the province of Manitoba (population of 1.0 million) Is combined with representative estimates of the H values for each procedure to obtain the corresponding per caput radiation dose. In addition, variations in the annual patient throughput of selected examination categories (CT and Nuclear Medicine) are investigated and the radiological .significance of the population exposure for CT is assessed.
METHOD Data on the number of diagnostic examinations which use ionizing radiation in the Manitoba health care system are collected by individual departments, and by the Manitoba Health Services Commission, for administrating purposes and payment of physicians. A comprehensive set of statistics for the year 1979 have been published (5) and lorm the basis for most of the collective doses to Manitobans. Supplementary data was also obtained on the number of patients undergoing radionucllde scans (6) for the years 1981 to 1985 and CT scans (7) for the years 1977 4flfl
CNS 9th ANNUAL CONFERENCE. 1988
TABLE 1: DISTRIBUTION OF RADI0LOGK EXAMINATIONS FOR THE PROVINCE OK MANITOBA IN 1979.(5) THE TOTAL NUMBER OK STUDIES PERFORMED IN THE SPECIFIED CATEGORIES WAS APPROXIMATELY 920,000. CT AND NUCLEAR MEDICINE ARE NOT INCLUDED IN THESE TOTALS.
EXAMINATION CATEGORY
Chest Extremi ties Spine-pelvis Head-neck Abdomen Obstetrj cal Gastro-intestinal Genito-urinary Neuroradiological Vascular Cardiac Mammography Other
NUMBER OF STUDIES
36 7,000 221,000 88,000 45,000 27,000 3,000 80,000 30,000 1,800 2,800 1,800 5,500 45,000
TABLE 2 : NUMBER OF RADI0NUCLIDE SCANS PERFORMED IN THE PROVINCE OF MANITOBA IN THE PERIOD 1981 TO 1985 (6)
NUMBER OF RAD1ONUCL1DE SCANS
1981 1982 1983 1984 1985
23,600 24,100 25,200 25,500 24,800
TABLE J: NUMBER OF CT SCANS IN MANITOBA FOR THE PERIOD 1977 TO 1987 (7)
TABLE 5: AVERAGE H . VALUES FOR THE THREE TYPES OF CT SCANNERS USED IK MANITOBA (EMI 5005; GE 9800; h SIEMENS DRH) EMPLOYING TYPICAL CLINICAL TECHNIQUE FACTORS. ALL H . VALUES ARE IN mSv.
NUMBER OK CT JCANS 1977* 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987
800 4, 200 4, 700 4, 600 5,500 7, 200 8,200 11, 200 14, 800 15, 300 18, 200
Values for the H 's associated with most of the examination categories listed in table 1 were taken from the results of a detailed survey of radiation doses to patients in the UK performed in 1983 by the National Radiological Protection Board (NRPB) (8). The doses received by patients undergoing similar examinations in England in 1983 and in Manitoba in 1979 are assumed to be broadly comparable. The results of the NRPB survey are summarized below in Table 4
TABLE 4: AVERAGE OF EFFECTIVE DOSE EQUIVALENT (H ) VALUES FROM A RANDOM SURVEY OF 20 RADIOLOGY DEPARTMENTS IN ENGLAND IN 1983
AVERAGE
Lumbar spine Chest Skull Abdomen Thoracic spine Pelvis
2.15 0.05 0.15 1.39 0.92 1.22
I.V.U. Barium meal Barium enema Cholangiography Cholecystography Cholangiography/ Cholecystography
4.36 3.83 7.69 2.59 0.95 1.17
HEAD SCAN
CHEST SCAN
ABDOMINAL SCAN
PELVIC SCAN
EMI 5005 GE 9800 SIEMENS DRH
0.84 1.9 2.3
4.8 15.7 16.0
2.6 6.3 9.5
?.l 6. 7 13.4
To obtain an estimate of the collective patient dose for the province of Manitoba in the year 1979, the following additional data, and assumptions, were required.
" Four month period only
EXAMINATION
SCANNER TYPE
(mSv)
It is important to note that the values presented in table 4 account for the whole examination procedure (i.e. multiple views and combinations of radiographic/ fluoroscopic procedures) and also include any need for repeat exposures. Data for patient doses arising from radlonuclide scans show that an average procedure is associated with a patient H of about 5 mSv (3,9). Variations about this mean value are generally of the order of factors of two to three. The data for radiation doses to patients undergoing CT scans on aecond and third generation scanners are presented in summary form in Table 5. (7)
a) The extremities are not generally included in the evaluation of effective dose equivalents, and in radiation protection practice ^re generally covered by a non-stochastic dose limit of 50 mSv/year for occupationally exposed personnel. (1) These exposures have therefore been neglected in the collect ive dose estimation, and this is deemed to be appropriate from a radiation risk perspective. b) The 88,000 spine/pelvis examinations performed in 1979 were broken down into 44,000 lumbar spine exams, 15,000 thoracic spine exams, and 29,000 pelvic exams, based on current estimates of the relative frequencies of these three procedures. c) The obstetrical exams were all taken to be the equivalent of an average abdominal study, from a radiation risk perspective. d) The 80,000 CI studies were broken down into b8tQ00 barium meal exams and 32,000 barium enema exams, again based on the current relative frequencies of these two procedures. e) Neuroradiological, vascular and cardiac studies, which only totaled 6,400 in 1979, are relatively high dose procedures and were allocated an average H of 10 mSv based on estimates made of typical values of the total energy imparted in these procedures. (8) f) For the mammographic studies, a mean H of 0.3 mSv was used, based on the typical glandular dose values currently suggested by the NCRP in their report (No. 85) on mammo^raphy. (10) g) To deal with the remaining 45,000 miscellaneous studies, the average value (weighted for relative frequency) of the procedures listed In table 1 was arbitrarily adopted. Although this procedure has little formal justification, it is considered most unlikely that it could lead to any serious errors in the resultant population collective doses from medical exposure. The total number of nuclear medicine procedures performed each year in Manitoba lias remained almost constant for the period 1981 to 1985, as have tno corresponding patient radiation doses. (6) It lias been assumed that both these parameters can be used for 1979, which is supported by anecdotal evidence
CNS 9th ANNUAL CONFERENCE, 1988 467
: rum s t a l l w u r k i n j ; in tin.- r e s p e c 1 i ve d . p a r I men I s . Dct.i i ] i-d .ma I y.>,(.•.-> o( p a t u n t ilusr:^ in nuc I e a r :ned ic h i e t o r I he y e a r s I {iS 1/ lliH:> a n 1 p r e s e n t e d e 1 M t • u' f 11 • r e ( " ) as :i r e tin1 cor n-.s/ujjul i iij*. d e l a i 1 s p a t i e n l d o s e s rt'CL-ivcii I rum Cl exams d u r i n g t h e [HTiod I " / / / t < < 8 7 . <,/)
t e r m s of a p e r c a p e ' v a l u e c x p r t ' i . s u d in JJSV ( N . l i . t o La I p o p u l a t Km ul Mani t o b a ]\a± b e e n v e r y s t a b l e Che I a s t d e c a d e a t I . 0 mi 11 i u n i n h a b i t a m s ; . lor TABLE f>: "LK CAPLT RADIATION UOSL IN MANITOBA (1979)
liie a
R.'KP
typical
del iiu-d i IK- r M e c t i v e
".ulu 1 t " p o p u lat imi
tlit.1 ! if Id o!
pjt ietu
p r o b l e m a t ical . Jt.-i:;u.-\raph iv
I'.ivun t-lnst, >'i ni;i ikfvi 1 y
in
'i he
.-t
: i i 1 i i u It y L*xami 1
i.s imliki-'lv
1
t e r n s o i .u;i', si :: ami
impact
Lo
h.ivt
l-iml also
a major
lilt
r! U-ct
i:u- urj;nn r i s k
s iy.n i t icant 1 >" i »i l w
>;uuetic d i s o r d e r s
in
depeudy
ain'v ol
exposed
p n p u I a L ton
have a 1
rad ica 1 I v l o w e r
"youiii. ," p o p u I at ion
on
may
EXAMINATION
it
this
irradiated child expectp u p u lat ion
risk"
than a
radiological
Bi't h tin- J^c re ) ai ed L M e its described in the p recei'd i iii* pa ra>;raph are 1 ike 1 y to be o t" ma jor concern in the ease ur" mediiM J exp-j^ure, since ill pat lent s are must likely to be o Ul than youn^. To account for this I actor, it. ih useful Lu define- .3 Population Irradiation iactur U'li") which is ^iven by: ;; ot detrimental etiects in patient _ pyp.u lat. ion •'• ol del r iniental el 1 ects in 1CRP adult population
To evaluate the I'll' lor tne pafienL population under»;uinji CT examinations in the province of Manitoba t lie basic demographic data on the aex/a^u distribution were obtained. (1\) Tlie pojiulation was otherwise assumed to have a normal life expectancy, although tiii s nay be undujy opt imist ic si no.' individual s ref]u i r i n^ med ica ] at tent ion may be e. xpec ted to have a shorter lift: expectancy than average. On this basis, it is straijU'ti orward to calculate the PIK factors lor the lour sub-categories uf CT exams uaed in the study described in detail elyewhere (11) (head scans; chest scans; abdominal scans and pu Jvie scans). Although ir. is possible to generate PIF values for the nenetii and somatic components ul the IT. for any class o! rad ioloj;ica 1 exaininat ion, in this report only the overall 1'ir is presented, which is a weighted (by Z ol the Lotal 11 ) sum ot the genetic and somatic I'LK's. It is clearly a'lso possibJe to combine PIF'y for differing clauses "f radiological examination tc obtain a weighted (by /, of Che total 11 ) overall PIF. ihis is performed bejow in the case of the four subcategories ol CT examination to generate an overall Plf- (or CT.
KLsms Pupulat ion dose i n Man i t o b a (19 79) ' l a b l u h blmws a l l I lie e s t i m a t e d contributors to Lh. pMiJul.it Inn dnse f o r the p m v . l m ' - o l llnnUi>l>.-i In 197'). 'l|i«- I I U M ' S In enct! .-aUv.orv a r e pi i ... , , o . in
468
CNS 9th ANNUAL CONFERENCE, 1988
RADIOLOGICAL
SUMMED TOTAL 1.0
HXAM1K-
I'OK
THE
mSv
I'EH
CAI'UT H .
(/JSV)
fact
vi 1 1
exami nat ionb. in addi t ion, the ]on& laU'i.t period asisoi1 iat ed wi tli mub t radiat ion induced cancel's aJ so imp lies that the rad io \o-.\ Leal det riment ul a ^iven rad iat iun exposure iu an aged populat iun is Iikely Lo he smal1 or than t iiaL assumed by the ICRP because many individuals woujd be expected to die before any nuUtu ion induci-d t-ancer cou] d be expressed.
1
IS
any
intiuction til"
- .in " o l d "
n)
THE
the
tin- v f i R h r i n j ;
relative
u n d e r g o i\i& identical
PROVINCE
uxprctanvy
Tlu-
"v,euet if
FKUM ALL
ATKJNS-
is
dilUT
Al t h o u g h mi
tlif i.i)"l.spring oi
cU-arly
could
that
taftora"
areas.
individuals the
is
nnt ion
l roi:i t hi' \Ci\V " a d n 1 t " p o p u 1 at imi. i at-tors
rad iolu^y
a p o p u l a i E L?U uiKk-r>:u inj.'
ra.lioN.^ieal
lor
anil i t * app I i cat iiMi in
(iv-.-.i:, in di.ij;iu)st ie ha^i^
UMtur.-b
dust* t<<[uiwi lent
the over
Barium
enemas
B a r i u m mual.4 Hen H o - u r i n a r y Nuclear Lumbar
m e d i c int1 spines
Abdomens Pelvic
246 184 I'il 125* 95 38 3D
Vascular
28
Neuroradio logical Cardiac Cheat Thoracic spine Hcad/necU CT Obbtotrical Mammu^raphy Miace] 1 aneous
18 18 18 IU 7 b A 2 A'J
* Extrapolated est iraate h)
Nuclear medicne H,'s for 1981/1985 The values For "the per caput H 's from the 25,000 Nuclear medicine examinations performed each year in Manitoba are shown in table 7.
TABLE 7: PER CAPL'T RADIATION DOSES l:0K RAD10NUCL1DE SCANS IN MANITOBA FOR 1931 TO 1985. ALL DOSES ARE EXPRESSED IN pSv.
YKAK
1981 I1,.'8 2 1983 198^ 1985
c)
PER CAPL'T H p (pSv) ['_^ 122 126 1JA 13A 127
CT H ' s in Manitoba for 19 77/1987 The per caput radiation doses from CT in Manitoba for the eleven year period (1^77 to 1987) are shown in summary form in TabJe 8. Over this time period the number of CT sianners in the province has grown from the first unit installed in 1977 (EMI 5005) te w M r h have been added to (IE 9800's (1984 and 1987) and a Siemens DRH (1986).
PER CAPUT RADIATION DOSES FROM CT SCANNERS IN MANITOBA FOR 1977/ 1987. ALL DOSES ARE EXPRESSED IN juSv.
VT H
1977* 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987
tlie real value .
(pSv)
0.7 4.2 4.7 5.8 7.3 9.5 10.8 23.5 40.8 43.5 81.0
The major radiological procedures which coiuribulLto this total are barium enemas (25%), barium meals (18%), genito-urinary examinations (13Z), nuclear medicine (12.5%) and lumbar spines , y.5%). The pur caput radiation dose from Nuclear Medicine has remained relatively stable at about 123 jiSv over the last oecade. By contrast, the per caput radiat ion dose from CT has increased from a value oi 4.2 ,uSv in 1978 to a value of 81 uSv in 1987. The reasons for the marked increase in the per caput C'T do.se include increased examinat ions, higher radiat ion doses f rom modern CT scanners, and an increasing proportion of (high dose) body CT studies. Tho patient population undergoing CT scans is generally older and the radiation detriment is estimated to be 452 of that which would be predicted for a standard ICRP "adult" population.
REFERENCES *Four months only d)
I'IK factors for CT in Manitoba (1983) Population Irradiation Factors were computed on the basis of the demographic factors (i.e. age and sex) of the patients undergoing the four classes of CT examination considered in this report. The resultant PIF values were 0.49 for head CT scans; 0.53 for chest CT scans; 0.44 for abdominal CT scans and 0.17 for pelvic CT scans. The computation of the overall CT PIF for Manitoba in 1983 is shown in Table 9.
TABLE 9:
EXAMINATION SITES FOR CT
Head Chest Abdominal Pelvic
Total
COMPUTATION OF PIF VALUE FOR CT IN MANITOBA FOR 1983
TOTAL II OF COLLECTIVE 7. OF DOSE S E TOTAL PROCEDURES (person-Sv) S F 6,100 300 1,200 600
8,200
5.2 1.4 3.0 1.2
10.8
48 13 28 11
100
(1) INTERNATIONAL COMMISSION ON RADIOLOGICAL PROTECTION, Publication 26, Per^amon Press, Oxford (1977) (2) HL'DA, W. and SAXDISOX, C.A., "The use of the effective dose equivalent, H , as a r'sk parameter in computed tomography" British Journax o( Radiology, 59:1986;1236-1238 (3) JOHANSSON, L., MATTSSON, S. and NOSSLIN, B., "Effective dose equivalent from radiopharmaceuticals" European Journal of Nuclear Medicine, 9: 1984;485-489 (4) UNITED NATIONS SCIENTIFIC COMMITTEE ON THE EFFECTS 01" ATOMIC RADIATION, 1982 Report to the General Assenbly, New York
100
0.24 0.07 0.12 0.02
0.45
The data in table 9 clearly show the effects of an aged population on the final PIF value of 0.45. This result means that the expected radiation detriment in the exposed CT population is only 45% of Chat which would oe expected if the same radiation doses (i.e. from the 8,200 CT scans in Manitoba in 1983) had been delivered to an "adult" (ICRP) population.
DISCUSSION AND CONCLUSIONS The per caput radiation dose (H ) in the province of Manitoba in IT/9 is estimated to be 1.0 mSv. It is important to note, however, that this value is based on a number of assumptions which require further Investigation before a definitive value can be obtained. The main area of investigation in likely to be the mean H per procedure In Manitoba. The overall accuracy of the per caput dose derived in this paper is probably within a factor of two t.o three of
(5) MACEWAN, D.W., GELSKEY, D.E., LOCK, J.R., POPOFF, J. and SOURKES, A.M., "1979 diagnostic radiology services in the province of Manitoba" Journal of the Canadian Association of Radiologists, 33:1982; 246-254 (6) HUDA, W. and GORDON, K., "Nuclear Medicine staff and patient doses in Manitoba (1981-1985)" Accepted for publication by Health Physics (7) HUDA, W,, SANDISON, C.A., and LEE, T.Y., "Patient doses from Computed Tomography in Manitoba (1977 to 1987)" Submitted to the British Journal of Radiology. (8) SHRIMPTON, P.C., WALL, U.K., JONES, D . C , FISHER, E.S., HILLIEK, M.C. and KENDALL, C M . , "A national survey of doses to patients undergoing a selection of routine X-ray examinations in English hospitals" National Radiological Protection Board report NRPB-R200, September 1986, Chilton, Dldcot, Oxon (9) HUDA, W., "Nuclear Medicine dose equivalent: a method for determination of radiation risk" Journal of Nuclear Medicine Technology,14:1986;199-201 (10) NATIONAL COUNCIL ON RADIATION PROTECTION AND MEASUREMENTS (NCRP), REPORT NO. 85, "Mammonraphya user's guide", 1986,Bethesda, MD (11) HUDA, W. jnd SANDISON, C.A., "CT dosimetry and risk estimates" Radiation Protection Dosimttry, .'2: 1985;241-249
C N S 9th A N N U A L C O N F E R E N C E , 1988 469
PRODUCTION AND RADIOCHEMICAL SEPARATION OF ARSENIC RADIONUCLIDES FROM GERM NIUM
JJ.G. DUROCHER*, D.N. ABRAMS + , M.W. BILLINGHURST + , t S.L. CANTOR +# , D.M. GALLOP*. J.S.C. McKEE* and G.R. SMITH*
•University of Manitoba Accelerator Centre Winnipeg, Manitoba, R3T 2N2 and Health Sciences Centre Winnipeg, Manitoba, R3E 0Z3
ABSTRACT Little effort has been made to evaluate arsenic cont'aining compounds as potential radionuclidic labels for new radiopharmaceuticals. Chemically, arsenic may be substituted for either phosphorous or nitrogen, both of which are found in abundance in biologically active molecules. Furthermore, several As radionuclides are readily produced via psoton irradiation of germanium. A joint research programme involving the Radiopharmacy group of the Department of Nuclear Medicine, Health Sciences Centre and the University of Manitoba Accelerator Centre is underway investigating methods for production and recovery of arsenic radionuclides.
administered to patients undergoing scans. These physical properties suggest that As may have potential in nuclear medicine.
Table 1 Q VALUES FOR (p,xn) REACTIONS
As Ge 70 72 73 74 76
69
70 71 72 73 74 75 76
16.2 7.0 34.4 25.2 13.5 5.1 41.6 32.0 20.3 11.9 1.1 51.4 42.2 30.5 22.1 11.3 3.4 46.5 37.4 27.3 19.3 9.0 1.8
INTRODUCTION Since arsenic is readily substituted for either nitrogen or phosphorus, the synthesis of analogs of biochemically active natural molecules and synthetic compounds with little or no alteration of biological activity is feasible. Furthermore, several As radionuclides are readily produced via proton irradiation of germanium. Table 1. lists a number of reactions of the (p,xn) type resulting in the production of radioarsenic. The radionuclide of choice for in vivo work would be As. The low energy positron (0.81 MeV) and gamma ray (0.1749 MeV) emitted are well suited for either PET or standard gamma imaging techniques. Its 64 hour half-life is convenient for performing the necessary Tadiochemistry while not requiring that large amounts of the isotope be
470
CNS 9th ANNUAL CONFERENCE, 1988
Q values (-ve MeV) for (p.xn) reactions on naturally occurring isotopes of germanium. The table shows only those reactions possible using the 20 to 50 MeV beams from the University of Manitoba Cyclotron and which resulted in radioisotopes of arsenic of interest to these studies.
Arsenic-71 is not commercially available, but can be produced by proton irradiation of germanium with a cyclotron. As can be seen from Table 1, it is easily produced using the 20 to 50 MeV variable energy proton beams from the University of Manitoba cyclotron. 71 Unfortunately, it is only possible to produce As without other As radionuclides being present by using a target of the separated isotope Ge (27% natural
abundance) irradiated with a proton beam of less than 25 MeV energy. The high costs of separated isotopes precludes their use in preliminary studies iuch as those described here. Therefore, throughout this work germanium of natural isotopic abundance was used. Table 2. lists the natural abundance of a number of Ge isotopes relevant to this work. As illustrated in Table 1, irradiation of natural Ge results in the production of radioisotopes of arsenic ranging from As to As.
Table 2 Ge AND As ISOTOPES OF INTEREST IN TfflS STUDY
69 70 71 72 73 74 75 76
Ge (%)
AsT
20.5
52.5m 64h 26h 80.3d 18d stable 26h
1/2
937 C without loss of target material, the latter technique seemed to offer the most elegant separation methodology, analogous to separation of I from Te [1]. However, the high solid phase solubility of arsenic in germanium [2] precluded this method. We did not recover any radioarsenic or observe any loss of radioarsenic from the germanium target material at temperatures up to 1200 C under a variety of conditions. Distillation of germanium from arsenic worked reasonably well but proved to be cumbersome in our hands due to our need to recover the enriched (future applications) germanium target material for further irradiations. However the chemical conversions inherent in the distillation of germanium from arsenic suggested that similar techniques might
15m 27.4 7.8 36.5 7.8
The natural abundance of each isotope of germanium and the half life of each isotope of arsenic of interest in this study.
Of the isotopes produced, the half-lives of As (80.3 days) and As (17.8 days) make them particularly useful isotopes in developmental work. For the initial studies. As/ As isotopic mixtures were generated by irradiating natural Ge targets and allowing short-lived isotopes to decay before the material was used. We have investigated various methods of separating the product from the target mxterial. The chemical and physical properties of arsenic md germanium indicate that they may be separated chemically by either selective distillation or selective extraction oi physically by sublimation. Since arsenic sublimes at 613 C and germanium melts at
MATERIALS AND METHODS
PRODUCTION OF RADIOARSENIC
Target Material Preparation. Germanium metal of natural isotopic abundance was ground to a fine powder (1.07 g, 14.7 mmol) and added to a three necked flask. Concentrated ammonium hydroxide and 35% hydrogen peroxide were added dropwise to an aqueous suspension of the germanium powder and the resultant suspension was stirred vigorously until no sign of the black germanium metal was visible. Ultrasonic agitatior also proved effective, but resulted in caking of the powder which decreased the rate of conversion. The solvent volume of the resultant while suspension was decreased in vacuo then dried at 200 C until a constant weight was obtained (1.51 g, 14.4 mmol, 98.2%). The oven dried anhydrous germanium oxide proved difficult to rehydrate and dissolve after irradiation, therefore the latter irradiations utilized powder which was dried in vacuo at 50 C.
Target Preparation. The germanium dioxide was prepared for irradiation by compressing the powder into a disc approximately 3 mm thick by 15 mm diameter.
CNS 9th ANNUAL CONFERENCE, 1988
471
3mm of compressed GeO- corresponds to a proton energy loss of approximately 20 MeV in the target. For an initial energy of 35 MeV, this energy loss covers the range of maximum cross section for the reactions of interest. The powder was compressed by applying up to 10 Pa pressure to a stainless steel piston/cylinder assembly containing the GeO_. To contain any material which might flake off, the disc was wrapped in thin aluminum after being compressed. The pellet was then fitted in a water cooled target holder which was in turn mounted in the irradiation chamber on the cyclotron's zero degree beam line.
20:
A
400
600
600
1000
UNTREATED IRRADIATED Ge0 2 TARGET
Irradiation. The germanium dioxide discs were irradiated with 35 MeV protons incident on target. The mean beam current during the irradiations was approximately 4 microamps on target. The approximately 80 watt poweT dissipation in the target resulted in no visible damage to the GeO_; the heat generated being easily removed via the water cooled target holder. B GE«MAMUr.' EXTRACTION PRODUCT
Irradiation time varied in accordance with the total activity required in each phase of the study. The integrated charge on target ranged between 1 and 50 microamp-hours. Following the irradiation, the target was removed from the beam line and stored in a hot area for two or three days until the activity of short lived As, Ge and Ga isotopes died out. When the activity had been reduced to a level which allowed for safe handling, the disc was packaged in a lead lined container for transport. The target was then transferred from the University of Manitoba Fort Garry Campus to the Health Sciences Centre via commercial courier service.
"As
J.
t
AS
1
..
g
0
02
04
06
0B
10
MeV - * • C GeO2 TARGET BBfORE ARSENIC EXTRACTION
ARSENIC/GERMANIUM SEPARATION
Sample Analysis. All samples were analyzed on a hyperpure germanium detector coupled to a Canberra Scries 35 MCA. Recoveries and crossover of arsenic and germanium were determined from the ratios of activity associated with the 595 keV gamma line of As and the 573 keV gamma line of Ge. Tjpical spectra obtained after an irradiation of natural Ge arc shown in Figures 1.
472
CNS 9th ANNUAL CONFERENCE, 1988
0
02
04
06
08
10
MeV —•• D ARSENIC EXTRACTION PRODUCT
Figures 1 A,B,C,D. SPECTRA OF GeO 2 IRRADIATION PRODUCTS AT VARIOUS STAGES OF As EXTRACTION
Germanium Recovery. The irradiated pellet was crushed and suspended in 50% hydrochloric acid and dilute hydrogen peroxide and either stirred or agitated in an ultrasonic bath until dissolution was complete. The resultant germanium tetrachloride in the aqueous phase was extracted into isopropyl or n-butylether (equilibrated with hydrochloric acid) three times and the washes were combined. The germanium tetiachloride in the ether phase was converted into germanium dioxide by treatment with dilute ammonium hydroxide and hydrogen peroxide. The white precipitate was filtered and dried in vacuo at 50 C.
Radioarsenic Recovery. The residual aqueous phase containing the radioarsenic was treated with sufficient concentrated hydrochloric acid to bring the overall hydrochloric acid concentration above 80%. Sufficient potassium iodide or sodium metabisulfite was added to consume the remaining hydrogen peroxide and convert the As (V) to As (ID). The As (HI) was extracted, as arsenic trichloride, into isopropylether or n-butylether.
OPTIMIZATION OF RECOVERY CONDITIONS
GERMANIUM RECOVERY
Hydrochloric Adi'- and Hydrogen Peroxide Concentration. Aliquots of the irradiated germanium dioxide were dissolved in aqueous concentrations of hydrochloric acid varying from 0 to 100% concentrated acid. Duplicate samples at each acid concentration were also treated with increasing concentrations of hydrogen peroxide. Each aliquot was extracted one time with n-ibutylether. The aqueous and organic phase of each aliquot was analyzed for As and Ge and recovery and crossover were calculated.
ARSENIC RECOVERY Evaluation of Reducing Agents. Aliquots of the aqueous phase from which the germanium had been extracted were treated (2 fold molar excess with Tespect to hydrogen peroxide concentration) with a scries of reducing agents after the hydrochloric acid concentration had been increased to at least 80%. Each
aliquot was extracted once with n-butylether and each phase was analyzed for As and recovery was calculated.
Evaluation of Hydrochloric Acid Concentration. Aliquots of the aqueous phase as above were treated with concentrated hydrochloric acid such that the final acid concentrations varied from 50 to 90%. Each sample was then treated with potassium iodide (as above) and extracted once with n-butylether. Each phase was analyzed for As and recovery was calculated.
RESULTS AND DISCUSSION
Severa'i methods for separating arsenic from germanium have been investigated. Since the ultimate goal of this piocedure was to provide As from an isotopically enriched Ge target several evaluation criteria were necessary. Ideally, the As should be separated in good yield and high radiochemical purity. The enriched germanium must be recovered quantitatively, in a chemical form suitable for re-irradiation and reprocessing. The relatively short half-life of As requires short processing times and the arsenic should be recovered in a chemical form readily amenable to further chemical manipulation. The physical properties of arsenic, which sublimes at 613 C, and germanium which melts at 937 C suggested that arsenic should sublime from a molten germanium target at approximately 950 C. We attempted to separate the arsenic radionuclides from an irradiated germanium 123 target in a manner similar to the sublimation of I from Te [1]. However, no sublimation of arsenic was observed at temperatures up to 1200 C, in an inert argon atmosphere. This failure to separate the two elements by sublimation was attributed to the high solubility [2] of arsenic in germanium. Indeed, although arsenic is routinely drifted into germanium to produce semi-conductors, it appears that the reverse process is not readily achieved with the small essentially carrier free amounts of arsenic encountered in this procedure. The Amersham corporation uses a distillation technique [3] for the separation of As from the germanium target material. This involved the dissolution of the germanium metal target in concentrated hydrochloric
CNS 9th ANNUAL CONFERENCE, 1988 473
acid and hydrogen peroxide. Under these conditions, the germanium is converted into germanium tetrachloride while the arsenic is mainiained as As (V). The germanium is removed by distillation, followed by reduction of the As (V) to As (HI) with hypophosphoric acid. The resultant arsenic trichloride is removed by distillation. Although this method was successful in our hands, the recovery of both germanium and arsenic was not efficient enough to warrant the use of expensive enriched isotopic material. However, the chemical conversions utilized in the distillation procedure were readily adapted to a liquidliquid extraction technique (Figure 2). The presence of Ge in the target material provided a convenient method for following and quantifying the recovery and separation efficiencies of arsenic and germanium. The 0.573 MeV gamma line of 6 9 G e and the 0.595 MeV gamma line of As have similar counting efficiencies and were used to quantilate the efficiency of each separation step.
Table 3 gives ihe optimum conditions for each step in the extraction procedure outline in Figure 2. The hydrochloric acid concentration necessary tn efficiently extract germanium as the tetrachloride was found to be equal to or greater than 50% concentrated hydrochloric acid on a volume basis. Concencations above 50% coextracted too much arsenic without the addition of hydrogen peroxide to prevent the conversion of As (V) to As (HI). As little as 1% concentrated (31%) hydrogen peroxide was sufficient to decrease co-extraction of arsenic to < 1%. Recovery of germanium was >97%.
Table 3. OPTIMUM EXTRACTION CONDITIONS Target Material
Target Dissolution Germanium Recovery
As (V) Reducing Agent Arsenic Recovery
basic Germanium Dioxide (converted in ammonium hydroxide) 50% hydrochloric acid 1% hydrogen peroxide dilute ammonium hydroxide, hydrogen peroxide Potassium Iodide or Sodium Metabisulphite > 80% HCI
The residual aqueous phase contained the arsenic radionuclides, as pentavalent arsenic. In order to extract the arsenic, the As (V) and the excess hydrogen peroxide were reduced. Various reducing agents were evaluated and potassium iodide and sodium metabisulfite were found to be the most efficient. The hydrochloric acid concentration of the residual aqueous phase was increased to greater than 80% (v/v) in order to convert the arsenic to the trichloride for extraction into the organic pha>e.
Figure 2. ARSENIC GERMANIUM SEPARATION SCHEME
474
CNS 9th ANNUAL CONFERENCE, 1988
The gamma spectral analysis of the natural isotopic germanium target, following each stage of the extraction process is given in Figures 1A through ID , Figure 1A represents the untreated germanium target. The major gamma lines have been attributed to mainly arsenic and germanium radionuclides. The main impurity was identified as Ga. Figure IB is the germanium
extraction product containing >97% of the original germanium target material and <3% of the original arsenic radioactivity. Figures 1C and ID show the spectral analysis of the aqueous phase before extraction and the organic phase after extraction of the arsenic. Recovery of the arsenic was >99% of the arsenic remaining after germanium extraction. Germanium crossover was < 1 % of the original germanium target material.
ACKNOWLEDGEMENTS This work is supported through grants from Children's Hospital Research Foundation, Winnipeg and the Natural Sciences and Engineering Research Council. *Deceased.
REFERENCES [1] ROBINSON, G.D. Jr., HELUS, R, Editor, "Radionuclides Production", CRC Press, Inc., USA, 1983, Chapter 4, page 130. [2] DANAEFFER, S., Private Communication, University of Winnipeg, November 1986. [3] Amersham Corporation, Private Communication
CNS 9th ANNUAL CONFERENCE, 198B 47S
THE APPLICATION O F ENGINEERING T O RADIO ISOTOPE PHODUCT1ON IV. IK ri-.TJ'IC.VS \loiiiic
l-nor^y
of
Canada
K;K[ i o c h e m i c a !
hnnata,
ABSTRACT MM- li'.-uliorhi'iiiir.-il • 'tni!|uiti> o f \ B "I h a s b e e n in • ijt'i-al i o n I'oi•H.MIn \ivws si if •['!> im< r a d i n o h e i n i c a i s a n d e n d p r o d u c t s n> i . o r l d markets. 17K- t-tWH|virit u i Hi ' U i T ^ n t a n n u a I i-evetiues in I ho SJOil M rari'-if -sol I s Mil.' \arie|\ of a i rad ioeheiniva Is , cmi pt o d u c i s. ami eu 4 i n e i - r o d products to manufacturers and i n s t i t u t i o n s around the world. f.'Cr a l s i . s t . M s ; U - c h n o l o - y i ^ c K a ^ e s IV.r t h e p r o d u c t i o n o f j s o l d | > e s o f c o m m e r c i a l i nt o r e s ! .
INTRODUCTION Ifie JV. td i ochem i oa I ('uj]]( I H< V ) u f \ k c l . h a s b e e n in njx-ivii i< in Cue o v e r |0 y o a t ' s , m a n u f a c t urin-j; a m i si-1 / iiiL-; rvuf i o r h e m i c a [ s a n d f in i s h e d product s (o i n d u ^ l r y and i n s t i t u t i o n s around tin1 w o r l d . The i i .inpan> w i f fi c u r r e n t a n n u a I r e v e n u e s i it I h o 100 >]$ ran^e "\ports about \li)% o f all p r o d u c l ion to i v o r j i l - w i d r m a r k e t s For1 w h i c h it lias m o r e t h a n a ~u% s h a r e ot c c r ' a i t i m a j o r i s o t o p e s . |>'i "<"' s hi is m o s s i nl i1 r e s t s a r t 1 d i v i d o d irtt o I wa hiMail i-;ilr-u-r)fi^s: / t r s t l > , tlie s a l e o f To-tin JUKJ I r m d j a t i o n oijui JIIKMI! a n r i , Si-c-oml I \ , I he- s a l o o f a w i (k' \ .i[' i f t \ of f" u l u r atnl e y e I ot [•(.'ti |>rorlui '^d I s o t i>|n'K, a s vvc-] I a s ( i r o i - o s s i n ^ frju i pim-nt . I'fit' ar'-a <1 Lsciissc-iJ in (his |iapi'r is t In* ^vin-tn I a p p l i c a ( ioti o f fni$itt(~**fniii t o t Jir p r o i j u c t j on o f b u l k f a d i or-hojiiit-fi] s a n d e n d j i r o d u r l s w i l h sjxviric r.-f<-rrm't's t o i n - t a i t i Uiannfafl u r i ML, u r o c c s s i ' s and t'ac i 1 ! t i <."-: i n riirr-i-nt us-.i'. In I \)H'i, U('i' moved iIs . i p c r a l i o n s from Tunru\v*s i ' a s d i r ' f in uttnwu I »> n e u J y i -of is I rijf'l eri (Viri J i t i'vs in K'inat a . These new i'fLf i I i 1 i ' \ s u i ' i ' c (Ji'sis;"iH'(| hy R('C, w i t h m a n u f a c t u r i n g of conijxjtK'iii s l>> ol h e r s and i nsl a I lal ion IA i-iititir/ttt fvic((ifK, Thru are sf nff - o f - f / i e - a r f f a c i I i t i e s f o r i s o t <>]»• jji-odnci i on w i l h fvip-tc i (.* s i - r v c tltr- w o r l d f o r t h e m;ijor JJJ d u r l s ,
Limit-''I
i '< m i p a n y
Ontario
Sina ! 1 s i inpl e . ric 1 u s u n •:•-. ar-c o f t CM u s e d when* I tie primar'y ret)\ii rem«Mit;-: a r c jV.tr con 1 a i nnv -ti 1 . }n t h o s e s i t ual i<»ns pr..).\ final i • s h i e l i| i nu; a l o n e will perini I s a f e oj«-rat i o n . fhese e n c l o s u r e s c a n U- d e s i q i n - d to a \ a r i e t y of - - t a n d a r d s iucludirrj; " c l e a n r'oom" s l a n d / i r d s which a r e r e q u i r e d f o r t h e mam t fact u r e o f phrirmaceiif i c a i s . . I'he d e s i iiu o f t tieso ;;iiiip]c « c^rxt ainment lx>\es usiia I I y i-'inploy-i r leaf, colourless, p l a s t ic c o n s t r-uct i on to ma>: inii /x\ i s i hi 1 j I y. Access I hroui*h s ^ J o v |xir( s for e a s e and s a f e ) y of o p e r a I irtn y ie-lfis H /VIC J l i l y wi t h e x c e l lent J ' l e \ J hi 1 i \ y. When r e u u i r e d , mat o r ia I s c a n he chauui'd t o meet must demands \'nf roiil n i nment and l o c a l s h i e l d i n g c a n he a d d e d t o i n c r e a s e t h e amount o f act i v i l y IKMMI^ h a n d l e d . J«ij-tft*r• f a c i l i t i e s s u c h a s hoi c e l l s c a n come in a v a r i el y o f s i /.es from t hi* smn II "Mini -cc 1 I " 1o the lar^'M("tlKxlt t y\x* cv\ I s , These units 1 y p i e a I Iy i-equi r e (he ed r a n ^ e s p e c i a l l y u n i t s iiirvinf f o r a s p e c i f i c p r o c e s s ; as t he (jenerat o r Landing bnci J i t y (IJI-K) » /Vicjjjlies for \enon, (n,T)Mo-99, e l c .
from such t he
In add i t ion to t he product / p r o c e s s S[XH; i f i e e e l I s t h e r e arc- c e l l s of a more1 g e n e r a l design which c a n he u s e d f o r a w i r t e i y of |>ur|xxse.s. Tliese t yjx\s o f c e J J s are typically used fnr coUilt-uO, i r i d ium-1'1^ , c a r b o n - I I , et c . The d e s n ; n s i n c I ude heavy shieldiim", l;tri»"e v i e w i n g wtncJ(}ws, t op end iiiati i p u l a l or-s and a v a r i e t y of' o t h e r s e r v i c e s ami opt i o n s .
to P R O C E S S EQUIPMENT AND TECHNOLOGY
W<'C h a s , o v e r i t s h i s t o r y , dr>xi^riod a \.'ar'iel\ ol' i wot u[X' i>rtKt*ssirm s y s t ems which d a i I y -j;enorat o s i g n i f i c a n t f r a c t i o u s o f f ho w o r l d ' s demarid. Thcsi:-;> '-•.( cms irK-fjCfKJ/vil o \n>t h ncu .-i/id e \ J s t i ni>' l(v|ir«)]oi?ii»s l o . s u c c e s s f u l l y y i e l d t h e h i ^ h - ' i " ' 1 ' ' ' > p r o d u c t s demanded by t h e market p l r i c c .
UCC has designed a u idr v a r i e t y oi' isot u\x' p r o c e s s i n g systems wtiieh ^eneraLe d a i l y s i ^ n i f ' i e a n l fract ions of the v^orld's demand. These systems rift on i ncori>orato txjt h new and e.\ ist in^ lee hnt d o r i e s to s u n essfu ( I y yi
W'C h a s , frtjin i t s loni* I ' S I X T i o r u - e , (Jrvc^cjjxvi enu; iiii^f T-j ni^ e.\fn'r( i HCV in amity **r*vi.s r-eJal \mX I" r v i d i o j s o t o/^e.s, i n c l u d i n g d<*si Ljn of I'm. i I i t ies-., p r o c e s s (•<|uii»meri( , s h i p p i n u ; c o n i a i i u M - s , i r r a d i a l ion f a c i I it i o s f o r hot.h r - o a c t o r s and c y c j o l r o n s nivi inanni'aci n r i n i ^ e>.)K-rl i s e for- v a r i o u s end itmduct.s u s i UK f a d i o i s o l o[x-s .
Most oi' t h e s e process syst.oms ar'e located in some form of e i t her con Iainment a n d / o r sh folding thus re<|uirin'4 design for remote o|torat i o n .
FACILITIES DESIOJ \{cc d e s i g n s uruJ w\\x w t r i o l y of hoi cell lt f/r'tfi'ft t. tit H KiiJr JiL-trket . The equ i j^nr-nt c a n vary in ' O i i i | i | e \ i t y fr-fjin a s i m p l e unsh i e l d e d \»>\ t o a l.-ii -A' i ln'fivi |> y . h i e l d o d hoi <<•! I .
476
CNS 9th ANNUAL CONFERENCE, 1986
liTC has rer-ent 1 v completed t h e del ivory, lo .lapan ol' a sysiem i'or t h e product ion • Jf lii-^h jnii-ity, ii)dino-l2H on a eye jot von. This syslem was r e c o c t i zed for its t eel in iea I excel 'Mice hy reei> i vim? t he bronze award in t he jo\al ion cal eLfor.v i'ram t ho Canada Awards \'o\- Busi ness K\coJ J e u i c 19H7 'um}H^ j J n>n.
Hit"' system desi *4i it -i I MIHJ liuill e n t i r e l y by Ift'C staff pmdueos nit in hi i^d pui-j t\ irjcJini-12.'* ,-iut iMII;it i i-.--i I I y on I ho i '[•- 12 |*)Ki t i vc ion niaoh ine located ;it (ho TNll'MI-' s i t e at t h e . I ht- process sysf i-in i n< hides (IK- t a r ^ r t whore 1iio I* van import s , the processing equ i piiir-ni , i hi' 1 'I .<' con I ro I l e r and I ho n'raphic t|>sj'la> modu l o . ()l h e r prm-i-KS
s\ st ems
IIMVC IH'I'H
[mill by («•(• t o IIIOL-I similar- c r i t o r i a s:t fol > , pfol'i 1 ;ihj 1 il > ;\mi i./ai»u.'i t y .
dos i gnod
for
and
increased
Iho lil I' is; .••mother ev-imple i>f such a r a d i o a c t i v e p r o c e s s i n g faci I i t ,\ . Work I > , i I produces sLeri l o , npyrogen i c Tc-99m generat oi's in a fu 1 1 y aitt oniat ic mode for (he Canadian market. The
Some of RCC's products a r c q u i t e small and t h i s prosenl y iles i t^n problems in t ho i r manuf a c t \ i r o . \-\2v sourc-os for oXruripJe a r e 3 jjun in diiimoLer, 9.fi rmn lon^' and c o n t a i n a bead of material 0.7 mm in diameter loaded with up to 2G GBq of 1-125. This product i s loaded and seal welded by a t o t a l l y remote p r o c e s s . Anolhcr jn-oduct with s i m i l a r production problems is l r - 1 9 2 s o u r c e s . Those sources c o n t a i n pellets of' I r - 192 3 . 0 nun in rI iaxiiotor, 0 . 3 mm Ihlck contained in a welded c a p s u l e . Again t h i s process is t o t a l l y re-mote from t he o p e r a t o r . RCC markets t e c h n o l o g i e s developed in-house a s a product . KvunpleB nf t.hese would bo t h e GI-F for Jc-99;» g e n e r a t o r s and t h e process system for t h e
prndttct ioti of high purify f-123 on a cyclotron. Those processes are- of commercial value because oC \ hei r i?ood design, rei iabi 1 i t y aiul proven ix'rformarico. I he Gl.K inonttonod earlir-r1 has l^een sold in both its manual and automatic version. Other designs of I he lil.h' have \jeon dovo loped to meet varying rust oiner demands. PRODUCT DESIGN
Application engineering is used in the design of ncu, arid maintenance of existing, products. Those act ivi t ies h|end various disci piinos of eni^ini•(• r iiii?, a 1 on^ wi t Vi Iht* dri ve t.o ]>r'oduce profitability for the product in question. Most new product s will include considerations ranging from isuto|x' |»roduetion, either in a reactor or a e\c lot ron t o \ ho f i rial packn^in^ . and shi pturnt of
I'rotJuct mainl onanco is an important as|wcl uf jjroduct en^i iioori n^. The- iromj.iet i t i \o ]>osil ion of any product in t he market is delormi nod by iIs design, ease of use, sui tabi I i ( y for i I s in I onderi use, qua I i ty, s.ift-t >- and, of real irn|>ortan, i I s price. At i of (he fore^oin^ need fo 6o corisffiercd in t he design of any product . I-'or radioact i \ e products, some of tliese desirables are difficull to achieve. Hy way of e\;vniple, let ' s oxajiune n product vvh ich HCC current 1\ markets across Canada, i he 1c—f*9m (ienerafor-. 'fh i s produc t. i s used by Sue I car Mod i c i ne dc'i'art ments in Canadian hospi tals for t he preparation of .iharmaceuticaJ s used in 7G% of a i l procedures using radjoacl i\'ity which amount to "'ixf'/o of all hospital procedures. The active component of this pharmaceutical is 'Ic—9(Jm, the daughter of Mo-y.9. Tc-99m has a 6-hour half-life and a decay energy of 41-10 kc-V. The product, design features ease of use, no required customer assembly and compactness. Dos i %x\ changes over t he past few years have resulted in reduced costs, a visually improved package and improved profitability. Another example of product improvementy which promises to increase performance and profitability is the current changes being considered with our lodine-125 point sources, RCC s e l l s 1-125 sources for bone densitomeIry and for a portable X-ray device. These sources can contain as much as 700 mCi ('*26 GBq) and -ire approx imateJ y \\ .0 mm in diameter by 9.5 mm long. Currently, this product is being redesigned to have a much thinner window and to be able to IK1 made by automat.ic- machinery. These chan^os will yield a product which has n higher, moro focused, more radially consistent output and with lower production costs. TRANSPORTATION OF RADIOACTIVE
MATERIAL
Annually, KCC ships in excess of 25,000 packages to dvst.inations around (he worid. These prcc/ca^os can vary in wei ght from a few kilogrnms to thousands of kilograms with contents which can vary 6 hy factors as high as 2 x 10 in a c l i v i t \ . The shipping containers used for these shipments wore substantially all designed and licensed hy Uc(\ The design, manufacture and maintenance of these packages meet al1 intermit ionally recognized standards. Package design includes thermal and st ress analysis often coupled with deslruct ive test ing. In many cases the shipping package must meet not only regulations hut. also customer demand for low transportation costs. This requirement can !«•• very chal longing tor sViijwicut.s where the price of I he order F.O.H. HCC and the cos! of transportation are similar-. Failure to respond to this challenge can rosult i n lost sales.
CNS9thANNUAL CONFERENCE, 1988 477
-" 1« « t Mm • ii" iu.it <•(• 1.1 K i s " I " e \ i [-I'MIimportance | -.•!•. k . i \ i ii'j. i l l ••-. i •-•.ii. M.-il i >[*i.-i I s rin is I Inaide to
in
nir-i •! •ifi
ill
( c i i i i i t i> ' t i s
J • l< •lit .
l-f.n1ii.-f-; mi •- -I
i f
in
i-;m
m ' u r r v i l i Hs l f i nU< 'til
adil i t i on
and
S-.<:".
-H:-»r
|<. j.i-.-\f-ut .-hiiicf-
ouile i •-. a
i i i ' I i'. i>[' -d
.-i Js(» o n
and
acids
some
e\Minple,
fits
^ticl'hn'j;
on
insl ly, therefore to
often
i m o l w
design,
is no|
of r.jn I y
real cost s
pr.jducts
aiilomat i o n ,
in
the
will areas
shielding
llici -| if in j cv , (-i HI I I'O|:- , c! « . | In- pro|>cr di "_;refof app! lea I loll a I I lie i licept i on st H-J,i • can make 1 he 11 i t i e r •' -nee hi't we< -n prof i i and 1 oss for* I he prf.diuM i o n p r o c e s s . Ki ' i ' i s c u r r e n l I y cmUtrked i\\*>n a \>nt.'n'<-t In i n s t a l l a roU.)t i c p r o d u c t i o n work u n i t i n i o one- o f M s p r o d u c t i on I i t i e s . In i s syst.eifr will f'tint *"• I he l a U m r c n n i e n l hy 2/'S and a l l o w r a p a c i ! \ incrtvisos o f (he o r d e r o f Fi tt, }i> uith no f u r t h e r investment requ i i'ed. i h i s cur-rent work i s d i r e d inu; i I s o l f t owards learning thtlundanicntals and ap]>lyin-^ those fundajiien! a 1 s t o s o l v e a p r o d u c t i o n p r o h l e m . K(.f' ant i i - i pat es t hat i t wi I I h a \ e r o b o t i c - s y s l o i n i n [ d a c e d u r - i n ^ 1HH9 u i t l i on I > a f'-u mrjiii hs In-h i n d .
its fi rsi I.he s«Tond
'I he d< 'velu|ifin""it o f now a m i irwiovat. i \pe prttduct s am 1 ] iroi.esse • i s c r 11 i f a 1 t o t he I'M 11 i r e KIH c e s s of t h« • h u s i r i e s s ami ' o n s i d e r a b l o e f f o r t s arTe *-\[x-no!ed on l hese act i v i I ie.s. \ew product s a r e cont inua I 1 y hcintj; soim'hf out t o tuid t o t h e p r o d u c t J i n o . MosI p r o d u i - t s e i t h e r - use radioact i\ i t y ilirec-lly or are used in support of produe)s I hat do. Much de\ e I o| OIK-nt
el f o r I
L^OI^S i n t o
l he do\ e l ojjfiient
of
end
f T'odiict s hecau.se- o f t he- h i ^)ier p r o f i t rnari^i n . These p r o d u c t s arc- a l s o more d i f f i c u l l t o design U-cause o f t he d j r-ect r-orit a c l wi I h 1 he t u s l omer , t h e demand f o r s»-r\ i c e and the- meiMini? o f customer r j e f j j,ed n«-eds. Some o f t hose product s a r e of r e s e a r c h i ut i-resI i n i t i a l l y hut have t he j>ot out ia 1 t o Ixv-ome e i l h e r s i u u i f i c a n t revenue i^ener-al o r s or com]•] et e f 1 o p s . Ijesii*ri i n I hese cases must ho eoi^ni/.nnt o f t h e market ami d e s i g n lo limit I hel i n a r i c ia I r i s k wh i I e erisur- ini? H I P o[>t i o n t o t ake a d \ a i i t aL,re o f t hp proriuct s ' [xit r-nt i n l i f i' does develop.
R7IURE DIRECTIONS What t h e future holds \'or t h e appl n a l i u i i u\' ermMMcerirm' l o isn\t}\M^ i>rr»hu-\.\nt) n v s u l t s from t he ( otil i tiua I need f o r p r o f i I a l i i I i I y . Th i s t r a n s l a t es i n f o i i i r r i v w i ) product j i i n , deer-eased M I K I S ami
47$
CNS 9th ANNUAL CONFERENCE, 1988
in
s:pi 11 •
u i I I «•»('
pluv
and t h< •
<,\'\ c i i
inakf-
11 i -_* J j r -1'
I Ii. 1 -'
CX3NCLUSIC*!
e i m i n c ^ - r i II-J; a p p l i c a l i o n s
process
11 >r\
handling,
cost:-;,
»jwi 1 i-
radial ion
at
In
cone I us i o n ,
appl i cal i on is to
place
can
U-
c-ti'J;' i n e f f i ivJ, of
PUSH r e and
the'
oiisincss
successful
pfodurr-
the-
seen Io
mill I i d i s i i p i i riai'y
t^oals
fuI I
view
i 1
o|"
f'iriHnfiul
in\'est.inenl .
of
remote
of
be
t f a d i t i onal I y ; i c a "I alilf • < ost:-;.
mu-ki-t ii'.-u
tar
\>rn< « - s s e s ,
can
I onicf.s
DEVELOPWFNT I'm
|JI<-sent•••
Miari|xjVM-f
mmtki
\ u l oinal i o n o f
|/f'O( •(•sSf'S,
Un-
need
nvjiir-r-
product ion
a c t i v. l i i c s
isot npe
I he
a n I oiii'il i o n
I ji-i -nsahi J i 1 > .
|)e\'-loiiuenl
1 e c h n i'p
howo\er
u h rr-|i
a r v - shi|>)M-rl
oiaf'-rial i i,,| >ac\
a u t OIILJI I i o n
| i . u ( i < 11 I a 1" I V
need
s d i r-1 <{ i ML*
I hf
h i g h e r (jiml i t y. I h e ; i c h i c w - m e n l rjf" | [ , e s e L-.oa I s i •-. s e e n a s r e ^ n i r1 iriL-;, m o r e ari'i imti'i-, I he * t p p l if -at i o n ol
wh ic*h
d . •-Tad.-, t i u n .
i in pi m l at i' "i. 1 a n d u i I I < >\'l e n out
for
n-i |i i i r-cnuMit ••; i nl fi
aminu
pt-Liim-t
l-or
f n n c l i< >t) i ML; x> s i em
c t u i l a i niiK'nt
|.-IIK>I l " i l
of
ami
pl'ol ecf
deUradal i o n .
Cum I i i HI; 11 tu
uh' 1 ] i
f i <' [i 11 [•* 'iiici it '-> liaiuls
Jin-
I »•
bol h norma 1
t ll< • in; 11 e r i a I s lill IS I
rfniri damage and
Ihr-;
! c-'.Wni
uti f- t
! r;itih]Hirl
\>lc| I t I' ill, i I I \
lli«' c D i i t t - n l s I he
of
in
! hat
I )H-
r v i d i o i s o l
must designs
mwssary
ho
held
reach return
I h< • in 1hion
Tlir CANADIAN TRRADIATION CliNTRl: Y. Doyle Al-XI, - Radiochemical Co.
(Mot A v a i l a b l e a t Time of
Printing)
CNS 9th ANNUAL CONFERENCE, 1986
479
PHYSICAL AND MECHANICAL CHARACTERIZATION OF RADIATION-CURABLE CARBON FIBRE COMPOSITES
C.B. SAUNDERS, A.A. CARMICHAEL, V.J. LOPATA AND A. SINGH
Radiation Applications Research Branch Vhiteshell Nuclear Research Establishment Atomic Nuclear Research Establishment Atomic Energy of Canada Limited Research Company Pinawa, Manitoba ROE 1L0
ABSTRACT Radiation processing is the utilization of ionizing radiat ion, usually photons or electron beams, to produce useful physical and chemical changes in a material. A potential application for electron beam processing for composi te manufacturing is for curing carbon fibre prepregs. These prepregs, carbon fibres or fabrics preimpregnated with liquid polymer resin, are commonly used in many industr ies, including aircraft and aerospace, automotive, electronics, construe t ion and vaiious commercial products. The objective of this experimental program is to design and manufacture a radiation-curable polymer-carbon fibre prepreg that meets the typical mechanical and physical property specifications set by the aircraft industry. This paper describes our current work in the design, manufacture and characterization of radiation-curable prepregs. INTRODUCTION c ibre-reinforced plastic composites are being used fo- rrnny industrial app]icat ions, both nonstructural and load bearing. The industries that use these composi tes extensively include the aircraft and aerospace industry, the electrical and electronics industry, various process industries for corrosionresistant tanks and pressure vessels, the construction industry and the commeic1 a 1 products and sporting equipment industries (1).
Basic and applied reseat ch and development in the field of radiat ion treatment of polymers and composites have continued since the first report of radiation polymerization by Hopvood and Phillips in 1938 (2). The interaction of high-energy radiation with organic compounds produces a variety of potential initiators, including an ions, cations and free radicals (3). Radiation-induced polymerization generally proceeds by the single mechanism that is most favoured by the conditions of the system (monomer type, temperature, impur i ties present, e t c ) The dose range required for most polymerization processes is usually 10 to 50 kCy, depend ing on the process objective and the material being treated (4). Highenergy irradiat ion of polymers also causes crosslinking and scission (2,3,4). Controlled crosslinking leads to an improvement of: the mechanical properties of the irradiated polymei, whereas scission usually leads to a deterioration uf mechanical properties. Crosslinking and scission occur simultaneously when polymeric materials are i rradiated. The i11adiation condi t ions can often be control lt_>d to favour crosslink 1 ng over sc i ss i on. Cross I inking is I a vmired dur ing Irradiation in vacuum HIM! i net l atmosphei e.s, and at high dose rales. The dose range? tequiied tor most crosslinking applications is bo to 200 k(Jy (4),
Radiation processing of plastics Is an established industry, vi th about 500 electron beam (E5) accelerators in operation worldwide, producing various
480 CNS 9th ANNUAL CONFERENCE. 198B
crosslinked products such as wire insulation and heatshrinkable films and tubes (5). Ultraviolet radiation sources are also extensively used for thin coating applications. The penetration limit for industrial high-energy EB accelerators (10-MeV beam voltage) is about 9 cm for tvo-sided treatment of unit-density material. The dose is distributed throughout the thickness of the material with approximately 252 variation (6), The penetration limit is inversely proportional to the density of the material: if the densi t y i s 2g/cm3, the penetration limit for twosided 10-MeV electron treatment is 4.5 cm. At present, radiation processing is not used commercially to manufacture fibre-reinforced composi tes. Nevertheless, EB processing does seem to be suitable for various potential applications for several reasons: 1. The continuing search for new matrix-fibre systems that can be easily processed and have the required mechanical and chemical properties. Radiation processing can convert a thermoplastic resin to a thermoset polymer. The processing ease of thermoplastics can be exploited, while still obtaining the high-temperature properties of thermosets. 2. The importance of automation of processes using fibre-reinforced composites to increase their use in high-volume consumer-oriented industries such as the automotive industry. Production speed is of primary importance, and extensive cure cycles limit production. Electron beam processing should be able to eliminate long cure cycles for some free radical-initiated resin systems, which would significantly increase production speeds. 3. Radiation processing can be easily retrofitted to accommodate many types of process lines and can also be an off-line process to treat selected products, as required, providing production flexibility. 4. The advantages that have been identified over the past 25 years in the radiation processing industry for the product ion of plast ics, including improved rate control, reduced curing limes, and curing at ambient temperature wi thout the need for chemical initiators and solvents. A specific application for election beam processing of composi tes is for curing carbon fibre prepiegs. These prepregs, carbon-ribres or fabrics preimpregnated with liquid resin, are used in the aircraft industry for hand lay-up applications, producing a high-strength, low-weight const ruet ion material. Epoxy resins are commonly used to manufacture these prepregs, but the curing cycle is both time-consuming and expensive. Radiation polymerization of most epoxy compounds proceeds by a cationic mechanism (7). These catlonic reactions are inhibited by water, even at
trace levels, and thus, most epoxies are not currently good candidates for Industrial-scale radiation polymerization. Good radiation curing properties can be imparted to an epoxy resin, however, by acrylating the terminal epoxy groups of an epoxy oligomer (?)•
3. X-ray diffraction (XRD) analysis using a Rigaku Rotoflex D/Max system to examine the crystallinity of the cured polymer (dose = 30 to 50 kGy). The minimum cooling rate was l°C/min from a 300°C maximum sample temperature.
The objective of this experimental program is to design and manufacture a radiation-curable polymercarbon fibre composite that meets typical mechanical and physical property specifications set by the aircraft industry. The program was divided into two phases: designing the composites and characterizing the selected resin system; and mechanical and physical testing of the radiation-cured materials. This paper describes our current work on the physical and mechanical characterization of the resins and composites.
4. Examining the effects of irradiation (both gamma and electrons) (dose < 60 kGy), atmosphere (air or nitrogen) and pressure (100 to 1000 kPa) on the gel fraction of the cured polymer. Gelation is the formation of an insoluble three-dimensional polymer network as a result of crosslinking reactions. Standard soxhlet extraction techniques, using chloroform as the solvent, were used for these experiments. A gammacell-220 (dose rate = 16 kGy/h) and a 1-kW, 10-HeV linear accelerator (average dose rate < 1300 kGy/h) were used to irradiate the samples.
RESIN SELECTION AND CHARACTERIZATION A resin system combining an epoxy diacrylate oligomer (502 by mass), a polybutadiene diacrylate oligomer (30£) and the multifunctional monomer dipentaerythritol monohydroxypentaacrylate (20%), vas selected for preparing our prepregs. Table 1 lists the relative product characteristics for these compounds provided by the supplier. The polybutadiene diacrylate vas included in the resin formulation to improve the flexibility and inpact resistance of the epoxy diacrylate polymer. Dipentaerythritol monohydroxy-pentaacrylate was added to improve the weatherability and the flexibility of the formulation. Various resin characterization completed, including
studies
have been
1. Fourier-transform infrared (FT-IR) spectroscopy to study the resin structure and the radiation-curing mechanism. A hydrogen-purged photoacoustic cell was used with a Bomen IMH4700L IR spectrometer to analyze 3-mm-diameter, 2-mm-long samples that had been gamma-irradiated in air from 1 to 300 kGy. 2. Linear thermal expansion of cured samples (dose = 30 to 50 kGy) using a Theta apparatus and a Lindberg furnace.
TABLE 1:
5. Determination of the viscosity of the selected resin formulation as a function of temperatuie. The resin viscosity is an important variable for prepreg manufacturing, affecting the flow-out properties and the resin wetting characteristics in the prepreging process. A Haake RV3 rotary viscometer was used for the measurements. PREPREG PREPARATION 1ND CHARACTERIZATION A plain weave carbon fabric, produced from polyacrylonitrile (PAN) precursor using a thermal decomposition process, was selected from an aircraft manufacturer's qualified products list. Table 2 lists the product specifications for the carbon fabric. The mechanical properties of carbon fibres are not deleteriously affected by irradiation when irradiated in a vacuum to a dose of as much as 50 MGy (8). This excellent radiation stability makes these fibres well suited for radiation-curable prepregs. The prepregs were prepared by a leading industrial prepreg manufacturer using a standard bead hot melt process ( 9 ) . The selected resin content in the prepregs was 35 t 2%.
RELATIVE PRODUCT CHARACTERISTICS FOR SELECTED COMPOUNDS
Property
Abrasion Resistance Adhesion Chemical Resistance Flexibility Hardness Heat Resistance Impact Resistance Low Shrinkage Water Resistance Weatherability Molecular Height Reactive Esters (%) Density (kg/m> • 25"C) Viscosity (Pa«s » 25°C) Resin Composition (wt.%)
Dipentaerythritol monohydroxypentaacrylate
X X X X X X
Polybutadiene diacrylate
X X A
X X
X X X X X
X S2S 97.5 1190 15.25 20
Epoxy diacrylate
X X
3000
99+ 940 0.01-0.02
30
428 98+ 1200 M000
50
X means that, vhen the compound is included in a resin formulation, the specified property is imparted to the cured polymer.
CNS 9th ANrJUAL CONFERENCE, 1988 481
RESULTS AND DISCUSSION TABLE 2: PLAIN' WEAVE CAR33N FABRIC SPECIFICATIONS AND FIBRE PROPERTIES
Specification/Property
Gelation
Value
Filarar.ts/Tov Fabric Areal Weight (g/ir.!) Cured Fabric Thickness (re) Fibre Ter.sile Strength (.MPa) Fibre Tar.sile Modulus tGPa) Fibre Tensile Strain {%) Fibre Density (kg/rc1) Prepreg Sssir. Content ( « . %)
3000 192 0.2 3790 234 1.5 183° 35 + 2.0
The prepreg physical tests include measuring the resin content, gel fraction, volatile content, moisture content, resin flow and ply thickness. Table 3 lists the mechanical tests and specifications for a 14-ply laminate made of carbon fabric-epoxy prepregs (10). Tests are conducted over the temperature range of 20 to 130°C (Table 3 ) . The effects of changing the irradiation conditions (dose rate, atmosphere, pressure) on the physical and mechanical properties of the laminate are also being examined. All tests are being done by the accepted procedures of the American Standard of Testing and Materials (ASTM) and the aircraft industry.
TABLE 3:
Temperature
CO Ultimate Ten3ile Strength (MPa) Tensile Modulus (GPa) Tensile Strain (m/m) x 10« Ultimate Compression Strength (MPa)
Compression Modulus (GPa) Compression strain (m/m) x 10'
Wet Compression Strength (MPa) 5
3 4 5
482
The gel point, the dose to initiate gelation, was approximately 2 kGy for both gamma (dose rate = 16 kGy/h) an electron (dose rate =1300 kGy/h) irradiation.
INDUSTRIAL CARBON FABRIC - EPOXY PREPREG LAMINATE SPECIFICATIONS 1
Property 2
1 2
Figures 1 to 3 illustrate the three important effects of irradiation on the gel fraction of the prepreg resin. Figure 1 shows that the gel fraction at a specified dose (2 to 60 kGy) depends on the dose rate and/or the type of radiation (gamma or electrons). The gamma-irradiated samples had a higher average gel fraction than the electron-irradiated samples. The maximum gel fraction was reached at a gamma or an electron dose of about 50 kGy. A maximum gel fraction of 972 was obtained with gamma irradiation and 87.5% was obtained with electron irradiation. This dose rate dependence is as yet unexplained. The high-energy electrons either reduce the overall crosslinking efficiency (G-value; G(X)) or increase the scission G-value (G(S)) as compared with gamma irradiation or. The G-value of a reaction is defined as the number of events (crosslinks or scissions) per 100 eV oE absorbed energy. Experiments are presently being completed to identify the mechanism causing this electron dose rate effect.
Minimum Average
Minimum Value
20-130 20-130
1275
145
117
20 130
9500 9500
7500 7500
20 70
1140 1070
1030
130
930
760
20-130
131
97
20 70 130
8500 7700 7000
20 70 130
483 414 241
1100
900
345 310 172
The laminate contains 14 plys, all in the same fibre orientation (± 1°). All properties will be measured using the American Standard of Testing and Materials approved methods. The time specified at the test temperature prior to the test is 10 ± 3 min for all dry tests and 2 + 1 min for all wet tests. At least five specimens are to be tested. Specimen conditioning: soaked in water at 71°C for 14 d prior to the test.
C N S 9th A N N U A L C O N F E R E N C E , 1988
maximum gel fraction in air at 100 kPa pressure was about 942. Increasing the pressure to 1000 kPa reduced the gel fraction over the entire dose range (2 to 50 kGy), with the maximum level being about 91X. This gel fraction reduction is a surface effect, probably the result of atmospheric oxygen diffusing further into the samples at an elevated pressure and inhibiting the crosslinking of the polymer by reacting with the free radicals produced during irradiation.
100 r
SO -
Figure 3 is a plot of gel fraction versus electron dose for resin samples irradiated in either an air or a nitrogen atmosphere. The atmosphere has little effect on the polymer gel fraction, confirming a theoretical conclusion made by Dickson and McKeown (11) that the high dose rates expected from a 10-Mev, 1-kU accelerator allow polymers to crosslink in the presence of oxygen. At these high dose rates the oxygen can not diffuse into the polymer fast enough to significantly inhibit crosslinking.
- 60 L
100
20
60
40 Dose, kGy
FIGURE 1:
EFFECT OF DOSE RATE FRACTION (GAHMA VS. ATMOSPHERE)
ON THE POLYMER GEL ELECTRONS; NITROGEN
nitrogen atmosphere. The radiation dose to initiate gelation in air, the gel point, was 2 to 3 kGy. The gel point of the samples irradiated in a nitrogen atmosphere was slightly lower than 2 kGy. The gel fraction increased rapidly with dose, reaching its constant maximum level at a dose of 50 kGy in both air and nitrogen. The maximum gel fraction in a nitrogen atmosphere was about 972 and was independent of the pressure during irradiation (100 to 1000 kPa). The
40 Dose, kGy
FIGURE 3:
EFFECT OF AThOSPHERE ON THE POLYMER GEL FRACTION (ELECTRON DOSE RATE = 1300 kGy/h)
Heating During Irradiation
100 r
80
. 60
-e- 100 kPo, air -B- 1000 kPa, air
40
- A - 1 0 0 kPo, nitrogen - $ - 1000 kPa, nitrogen
20
40
20 Dose, kGy FIGURE 2:
20
EFFECT OF ATMOSPHERE POLYMER GEL FRACTION 16 kGy/h)
AND PRESSURE ON THE (GAMMA DOSE RATE «
Thermal energy is added to material during radiation processing by radiation-induced exothermic reactions and by the net energy of the absorbed radiation (12). The final temperature of an irradiated object depends on its shape, specific heat and its thermal conductivity (12). Thick composites are of particular concern because the energy absorption properties of the reinforcements are different than those of the matrices which can result in extreme local heating problems, leading to composite damage (12). The exothermic reaction rates also increase as the temperature of the composite increases, providing even more energy to the system. Excessive heating in composites can be avoided by proper selection of the resin formulation to minimize the thermal energy released from exothermic reactions, reducing the required curing dose by using crosslinking promoters and fractional irradiation techniques. Figure 4 illustrates the effect of dose rate on the internal and surface temperatures of the 14-ply prepreg laminates. The electron dose rate was 290 kGy/h with a four pass irradiation and the gamma dose rate was 16 kGy/h. The maximum internal and surface temperature rise for the electron-irradiated samples, was 46°C and 39°C, respectively, at a dose of 42 kGy. The gamma-irradiated samples rose by \9°C both internally and at the surface at a dose of 30 kr,y and then remained at that temperature. C N S 9th A N N U A L C O N F E R E N C E . 19B8 483
Mechanical Testing
Initial experiments have shown that at 20°C, brittle fracture occurs in the uncoupled 14-ply laminates, the load rising to the breaking point linearly with increasing elongation (13). The evacuation time and the external pressure used in the vacuum-bag laminating procedure have affected the ultimate tensile strength or stress of these laminates by as much as 30%. The ultimate tensile strength of the 14-ply prepreg laminate is about 475 MPa. Experiments are being continued to optimize the laminating procedure and to complete the mechanical characterization of the uncoupled prepreg composite.
20 40 Average Dose, kGy
FIGURE 4:
TEMPERATURE RISE VERSUS AVERAGE IRRADIATION DOSE FOR 14-PLY LAMINATES (DOSERATE = 290 kGy/h FOR ELECTRONS AND 16 kGy/h FOR GAMMA; Ta = AMBIENT TEMPERATURE)
Figure 5 shows the effect of fractional irradiation on th= internal temperature of the 14-ply laminates. The maximum temperature rise for a single-pass irradiation was 65°C at a 42 kGy dose. A four-pass " r a d iation lowers this maximum temperature rise to 46 C. The properties of the composite matrix, the epoxy diacrylate polymer and the carbon fibres are not significantly affected by heating to 65°C above ambient temperature.
Miscellaneous Properties Table 4 lists some of the miscellaneous properties of the resin and the prepregs that have been examined in this project. X-ray diffraction analysis of the radiation-cured polymer (30 to 50 kGy) heated to 300°C and cooled at l°C/min confirmed that the polymer was amorphous and unchanged compared to the control samples. The polymer was not expected to crystallize because, as with most acrylated polymer formulations, the bulky side groups in the polymer structure make molecular ordering, and hence crystallization, difficult (15). The viscosity of the unirradiated resin was approximately 1 Pa-s at B0°C and 0.1 Pa-s at 200°C. The viscojity limit for a hot-melt prepreging process is 0.1 Pa*s to ensure proper iabric wetting and rp.sin flow-out,
70 60
The density of the unirradiated resin and irradiated polymer (30 to 50 kGy) was about 1110 kg/m 3 . Varying the curing conditions (dose, atmosphere and pressure) did not significantly affect the polymer density.
50 P
It must be noted that the mechanical properties of the epoxy diacrylate resin-carbon fibre composite, which is being tested at the present time, are not expected to meet the industrial specifications because no coupling agents were included in the resin formulation and the curing process requires optimization. Coupling agents are compounds that react with both the polymer and the fibre surface to enhance adhesion and improve the final properties of the composite (9). Coupling agents are used in composites to improve their properties (14). The uncoupled properties and the curing mechanism or a composite are being determined tD help select an appropriate coupling agent.
40
CL
a. in 20 40 Average Dose, kGy FIGURE 5:
EFFECT OF FRACTIONAL IRRADIATION ON THE INTERNAL TEMPERATURE RISE OF 14-PLY LAMINATES (ELECTRON DOSE RATE = 290 kGy/h)
484 CNS 9th ANNUAL CONFERENCE, 1988
Differential thermal analysis (DTA) of the radiation-cured polymer (dose = 30 to 50 kGy) in an argon atmosphere shoved a characteristic endotherm at about 23O°C. Linear thermal expansion analysis confirmed that the DTA peak was due lo polymer softening. The polymer softening point, about 235°C, is unaffected by varying the irradiation dose between 30 and 50 kGy. The linear thermal expansion coefficient is also unaffected by irradiation up to 50 kGy. The expansion coefficient was approximately 1.3 x 10"1" m/m°C for all three polymer samples (30, 40 and 50 kGy) below 150°C. The elucidation of the effect of irradiation on the thermal expansion of the polymer ahove 150°C requires further investigation. The expansion analysis also showed that the cured polymer (50 kGy) shrinks slightly, 1.3*, as a result of heating at l°C/min up to 250°C and cooling at l°C/min. Reheating the polymer to 25O°C did not shrink it further.
TABLE 4:
MISCELLANEOUS RESIN AND PREPREG PROPERTIES
Property
Value
Analytical Method/Equipment
Crystallinity of Cured Polymer (30-50 kGy) (%)•
<1
X-Ray Diffraction
Viscosity:
1 0.1
Rotary Viscometer
Pa's at 80°C at 200°C
Resin Density (kg/m J )
1110
Polymer Thermal Properties: Softening Point (°C) 2
235
Linear Thermal Expansion Coefficient (m/m/<>C;<150°C)2 Shrinkage (55)3 Cured Prepreg Thickness (mm): 1-ply 14-ply Actual Prepreg Resin Content (wt.%)
1.3x10"
Linear Thermal Expansion/ Differentia] Thermal Analyse Linear Thermal Expansion
1.3
Linear Thermal Expansion
0.4 3.5 36
Soxhlet Extraction
1 Samples were heated to 300°C and cooled at l°C/min. 2 Samples vere radiation cured with a 30 to 50 kGy dose. 3 Shrinkage occurred during the initial heating (l°C/min up to 25O°C) and cooling (l°C/min.) cycle. Fourier-transform infrared spectrpscopy was used to identify the qualitative effects of gamma irradiation (100 kGy) in air on each of the three resin components: epoxy diacrylate, polybutadiene diacrylate and dipentaerythritol monohydroxypentaacrylate. The following general conclusions were drawn from the FTIR spectroscopy study. 1. The amount of unsaturation, corresponding to peaks below 1800 cm" 1 , was reduced in each of the components upon irradiation. The carbon-carbon double bond is a common reaction site for radiationproduced free radicals during both polymerization and crosslinking. Acrylated peaks (1165 cm" 1 , 1250 cm" 1 and 1730 cm" 1 ) are reduced significantly, confirming that this moiety of each component takes an active part in the curing re?- 'ions. 2. Carbon dioxide (C02) (peak at 2300 to 2350 cm" 1 ) has been identified as a radiolysis product in both the polybutadiene diacrylate and the dipentaerythritol monohydroxypentaacrylate. The origin of the CO;, is as yet unknown. It may be produced during irradiation of the various acrylates. Atmospheric oxygen may also be involved in its production. Further experiments are required to identify tile CO;-producing reactions and determine the appropriate irradiation conditions to minimize its production. 3. Irradiation affects the -CH 2 and -CH 3 groups of each compound differently. The epoxy diacrylate peaks remain unchanged. The polybutadiene diacrylate peaks are all reduced, suggesting that the -CH 2 and -CH 3 groups may be involved in reactions to produce methane and hydrogen gases, which have not yet been identified. Only the CH 2 peak is reduced upon irradiation of the dipentaerythritol monohydroxypentaacrylate. h. The polyether iation.
structure
is
CONCLUSIONS A radiation-curable polymer-carbon fibre prepreg composite is being characterized as a first step in designing a laminated fibre-reinforced material to meet the typical physical and mechanical property specifications of the aircraft industry. The selected resin formulation is a mixture of epoxy diacrylate (502), polybutadiene diacrylate (30%) and dipentaerythritol monohydroxypentaacrylate (20%) resins. A plain weave carbon fabric was selected for the prepreg material. A hot-melt process was used to impregnate the fabric with about 36% (by mass) of ihe resin.
The selected prepreg resin radiation cures through crosslinking reactions involving the acrylated groups in the resin components, producing an amorphous polymer. The epoxy group takes little or no part in the curing reactions. Carbon dioxide has been identified as a radiolysis product in both the polybutadiene diacrylate and the dipentaerythritol monohydroxypentaacrylate. The gel content in the cured prepreg polymer depended on the total dose below 50 kGy, the dose rate, and the atmosphere and pressure during irradiation at a low dose rate (16 kGy/h). The maximum gel fraction was obtained with a gamma or an electron beam dose of 50 kGy. A difference in maximum gel fraction was observed between gamma-irradiated polymer (97%), irradiated under nitrogen, and electron-irradiated polymer (87.5%). This dose-rate dependence is not yet explained. The maximum gel fraction in air varied from 9b% at 100 kPa pressure to 912 at 1000 kPa pressure during gamma irradiation (16 kGy/h). The maximum gel traction in nitrogen was 97% (100 to 1000 kPa). The atmosphere has little effect on the polymer gel fraction during high dose rate (290 kGy/h) electron beam irradiation.
unchanged with irrad-
CNS 9th ANNUAL CONFERENCE, 1988
48S
During ejection beam iiradiation (dose rate = 290 kGy/h), the maximum internal temperature rise in the 14-ply pLepreg laminate was 65°C above ambient temperature at a dose of 42 kGy, for a single-pass treatment. Fractional irradiation (4 passes) lowered this temperature list' to 46°C. The properties of the composite are not expected tn be atfecled by heating to 100cC. X-ray diffraction confirmed that the polymer was amorphous, even at a cooling Late of l°C/min. The polymer softening point vas about 235DC and the linear thermal expansion coefficient was about 1.3xlO~4 m/m°C (30 to 50 kGy) up to 150°C. Heating the cured polymer to 250°C and cooling if at l°C/min caused the polymer to shrink 1.1%. REFERENCES (1)
HARGOLIS, J.H., "Advanced Thermoset Composites", Van Nostrand Reinhold Company, New York, 1986, Chapters 1.
(2)
CHAPIRO, A., "Radiation Systems", Interscience 1962, Chapter 1.
O)
DOLE, M., "The Radiation Chemistry of Macromolecules", Academic Press, New York, 1973, Vol. 1, Chapter 2.
Chemistry of Polymeric Publishers, New York,
(4) WILSON, J.E., "Radial ion Chemistry of Monomers, Polymers and Plastics", Marcel Dekker, New York, 1974, Chapters 6 and 7. (5)
PETERS, R.A., "The Quiet Revolution: Electron Beam, Ultra-Violet and Other High Energy Radiation Processing", Omega Research Associates Report, Pittsburg, 1984.
(6) McKEOWN, J., "Radiation Processing Using Electron Linacs", IEEE Transactions in Nuclear Science, Vol. NS-32, pp 3292-96, 1985. (7)
DICKSON, L.U., and SINGH, A., "Radiation Curing of Epoxies", Radiation, Physics and Chemistry, Vol. 31, No. 4-6, pp 587-593, 1988.
(8)
FORNES, R.E., MEMORY J.D., and NARANONG, N., "Effecl oi 1.33 Key Gamma Radiation and 0.5 KeV Electrons on the Mechanical Properties of Graphite Fiber Composites", Journal of Applied Polymers Science, Vol. 26, No. 6, pp 2061-6, 1981.
(9) HANSON, N., private communication, Incorporated, Magna, Utah, 1987, July.
Hercules
(10) WEETON, J.W., PETERS, D.M. and THOMAS, K.L., "Engineers' Guide to Composite Materials", American Society of Metals, Metals Park, Ohio, 1987, Chapter 6. (11) DICKSON, L.W., and McKEOWN, J., "Radiation Interactions with Linac Beams", Proceedings of a Working Meeting on Radiation Interactions, Leipzig, E. Germany, 1989. (12) ZAGORSKT, Z.P., "Thermal Effects in Radiation Processing", Radiation, Physics and Chemistry, Vol. 25, No. 1-3, pp 291-304, 1985. (13) WARD, I.M., "Mechanical Properties of Solid Polymers", Wiley-Interscience, Toronto, 1971, Chapter 2. (14) LEE, H., and NEVILLE, K., "Handbook of Epoxy Resins", McGraw-Hill, Toronto, 1967, Chapter 21. 486 C N S 9th ANNUAL CONFERENCE. 1988
(15) SHARPLES, A., "Crystallinity", in Polymer Science, Jenkins, A.D., (ed.), North-Holland Publishing, London, 1972, Vol. 1, Chapter 4.
Author Index NAME Abrams, D.N. Alikhan, S. Allen, P.J. Anderson, J.W.D. Anderson, J.W.D. Atkinson, J.C. Aydogriu, K.M. Badie, N. Bagshaw, J.L Banks, F.B. Barclay, F.W. Baxter, D.K. Beamer, N.V. Beaudoin, R.R. Billinghurst, M.W. Bonin, H.W. Boss, C.R. Bowman, J.M. Bromley, C.J. Brown, W.P. Buckley, LP. Calabrese, C.R. Candlish, J.R. Cantor, S.L. Carlson, P.A. Carlson, P.A. Carmichael, A.A. Chada, J. Chan.A.M.C. Chan, C.K. Chaplin, R.A. Cheadle, B.A. Coleman, C.E. Comeau, J.G. Cornblum.E. Currie, T.C. Dalton, K.C. De Paz, A. De Santis, A. DeVaal, J.W. Degregorio, R. Dinnie, K.S. Dinnie, K.S. Donnelly, J.V. Doyle, Y. Drew, D.J. Durocher, J.J.G. Durocher, J.J.G.
PAGE 470 447 284 8 347 325 144
391 32 245 284 332 157 135 470 297 144 360
32 157 157 25 277
470 305 320
480 20 252 238 300 396 401
13 353 325 277 370 20 110 379
423 426 62 479 173 453 470
El-Nesr, S. Ellis, P.J. Ellis, R.J. Ellis, R.J. Elphick, I.G.
360 213 305 320 252
Fowles, G.A. Frost, C.R.
210 129
Gallop, D.M.
470 164 74
Gil Gerbino, J.J. dirarri %niicirui RHi Grant, 1. Gray, R.G. Grondin, D.M. Gulshani, P.
OfV\ oUU
379 325 173
89
Hadaller, G.I. Hanna, B.N. Hayamizu, Y. Helbrecht, R.A. Hembroff, R.L. Hewitt, J.S. Hey, I.S. Holt, R.A. Hopwood, J.M. Huda.W. Huda, W. Hussein, E.M.A.
103 229 83 157 332 312
3 391
259 459 466 300
Inglis, 1. Irish, J.D.
379
Jenkins, D.A. Jin, C.D. Johnson, A.R.
25 20 417
Kenchington, J.M. Kirkpatrick, R.A. Koroll, G.W. Kowalski.J.E. Kowalski, J.E. Kozier K S Kozier K S Kroeger, V.D. Kumar, A.N. Kumar, R.K. Kundurpi, P.S.
213
Lafortune, J.-F. Lam, A.H.T. Lam, K.F. Lawrence, B.E. Le, V.T. Lee, A.G. Lemire, R.J. Lewis, R.E. Lichtenberger, P.C. Lie, S.G. Liederman, Joel Lindsay, R.P. Litke, D.G. Lopata, V.J. Lopez, A.M. Manzer, A.M. Marshall, B. McKee, J.S.C. McLellan, J. McQuade, D.R. Meneley, D.A. Meneley, D.A. Meranda, D.G. Moan.G.D. Mochizuki, H. Moffett, D. Moreno, C. Mosey, D. Moyer, R.G. Mulpuru.S.R. Murphy, E.V. Muzumdar, A.P. Muzumdar, A.P. Muzumdar, A.P. Natesan, M. Neuman, R.
67
360 238 229 332 41 62 110 411 238 94
48 237 341 151 157 320 164
194 384 428 202 365
120 480
20 347 277 470 459 183
48 300 213 401 83 151 370 437
120 245 379 94 103 110 379 20
Oliva, A.F. Oliva, A.F. Ordonez, J.P. Pang, S. Papiez, L. Paquette, J. Pattantyus, P. Patterson, B.K. Patterson, D.W. Pellow, M.D. Penner, G.R. Pettipas, W.H. Price, E. Price, E.G. Razzaghi.M. Reeves, D.B. Rolfe, B.A. Rosinger, H.E. Rouben, B. Sagert, N.H. Saluja, P.P.S. Sandison, G.A. Sandison, G.A. Saunders, C.B. Sauve, R.G. Savoie, D. Schankula, M.H. Schulte, C.W. Shalaby, B.A. Sheridan, S.D. Shim, S.Y. Sills, H.E. Singh, A. Singh, N.P. Smith, G.R. Smith, H.J. Smith, H.J. Smith, J.E. So, C.B. So, C.W. Spinks, N.J. Stafford, D.L. Stewart, W.B. Stone, R.E. Storey,H. Sunder, S. Symmons, W.R.
55 312 74 259 459 164 197 417
135 245 62 476 379 401 223 110
365 41
25 164 164 459 466 480
391 459 110 391 401 110
332 8 480 270 470
305 320 291 120 347 284 417 379
55 417 164 360
Taylor, P. Tennankore, K.N. Thompson, P.D. Townes, B.M. Tseng, CM.
447 67 67
164 237
Ungurian, M.
237
Vandergaast. G. Vandergraaf, T.T. Vikis, A.C. Vinez, J.C.
173 164
370
Wasywich, K.M. Weaver, K. Weeks, D.F. Whittall.W.R.
129 437 447 265
Yee, F. Yu,V. Yuen, P.S.
284 353
151
120
CNS 9th ANNUAL CONFERENCE, 1988 487
