ML20215K677

Transaction of the Fourteenth Water Reactor Safety Information Meeting

ML20215K677
Person / Time
Issue date:	10/31/1986
From:	Weiss A NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:
References
CON-FIN-A-3283 NUREG-CP-0081, NUREG-CP-81, NUDOCS 8610280255
Download: ML20215K677 (295)

Text

- -

NUREG/CP-0081 Transactions of the Fourteenth Water Reactor Safety Information Meeting To Be Held at National Bureau of Standards Gaithersburg, Maryland October 27-31, 1986 U.S. Nuclear Regulatory Commission Office of Nuclear Regulatory Research p= %,,

I;f i ,'

• 3*

eggo2gg,862 pua CP-OOB1 R 1

NOTICE These proceedings have been authored by a contractor l of the United States Government. Neither the United l States Government nor any agency thereof, or any of their employees, makes any warranty, expressed or implied, or assumes any legal liability or responsibility for any third party's use, or the results of such use, of any information, apparatus, product or process disclosed in these proceedings, or represents that its use by such third party would not infringe privately owned rights. The i views expressed in these proceedings are not necessarily '

those of the U.S. Nuclear Regulatory Commission.

Available from Superintendent of Documents U.S. Government Printing Office P.O. Box 37082 Washington D.C. 20013-7082 and National Technical Information Service Springfield , VA 22161

NUREG/CP-0081 z - _ :; -

- _ _ _ n:::- - -_ _ _ . _ _.

Transactions of the Fourteenth Water Reactor Safety Information Meeting To Be Held at National Bureau of Standards Gaithersburg, Maryland October 27-31,1986 Date Published: October 1986 Compiled by: Allen J. Weiss, Meeting Coordinator Office of Nuclear Regulatory Research U.S. Nuclear Regulatory Commission Washington, DC 20555 s.,

(kh/

s...../!

PREFACE This report contains sunmaries of papers on reactor safety research to be presented at the 14th Water Reactor Safety Information Meeting held at the National Bureau of Standards in Gaithersburg, Maryland, October 27-31, 1986.

The summaries briefly describe the programs and results of nuclear safer /

research sponsored by the Office of Nuclear Regulatory Research, USNRC.

Summaries of invited papers concerning nuclear safety issues are included from the Of fice of Nuclear Reactor Regulation, USNRC, in addition to summaries of invited papers that cover the highlights of reactor safety research conducted by the Department of Energy (00E), the electric utilities thrcugh the Electric Power Research Institute (EPRI), the nuclear industry, and the research of goveranent and industry in Europe and Japan. The summaries have been compiled in one report to provide E basis for meaningful discussion and information exchange during the course of the meeting, and are given in the order of their presentation S. each session.

A nunber of speakers did not submit sunmaries for inclusion in this report (indicated by an asterisk [*] in place of a page number in the Table of Contents).

A summary of the agenda is printed on the inside of the back cover, iii L_____. _ __ , __ _..._.

TABLE OF CONTENTS 14th WATER REACTOR SAFETY INFORMATION MEETING October 27-31, 1986 Page PREFACE. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii Monday, October 27, 1986 PLENARY SESSION Chairperson: G. H. Marcus (NRC)

INTRODUCTION: G. H. Marcus (NRC)

KEYNOTE ADDRESS: V. Stello (NRC)

To be announced : L. W. Zech (Chairman, NRC)

To be announced: J. Bussac (CEA)

SESSION 1 ECCS Rule Revision Chairperson: W. Beckner (NRC)

SUMMARY

OF NRC PROPOSED REVISION TO THE ECCS RULE CONTAINED IN 10 CFR 50.46 AND APPENDIX K. .....................

• W. Beckner (NRC)

DRAFT REGULATORY GUIDE, "BEST ESTIMATE CALCULATIONS OF EMERGENCY CORE COOLING SYSTEMS PERFORMANCE" . . . . . . . . . . . . . . . . . . .

• H. Tovmassian (NRC)

NRC EXPERIENCE USING REALISTIC CALCULATIONS IN THE LICENSING PROCESS . .

• B. Sheron (NRC)

Integral Systems Testing Chairperson: D. Solberg (NRC)

STATUS OF INTEGRAL FACILITIES AND FUTURE PLANS FOR INTEGRAL TESTING. . .

• D. Solberg (NRC)

EVALUATION OF SCALING CONCEPTS FOR INTEGRAL SYSTEM TEST FACILITIES . . . 1-1 K. G. Condie -and T. K. Larson (INEL)

' PWR RECOVERY PROCEDURES INVESTIGATED IN THE L0BI-M002 TEST FACILITY , . . . . . . ........................ 1-3 C. Addabbo and L. Piplies (ISPRA) v

SESSION 2 Nuclear Plant Analyzer and Code Development Chairperson: H. Tovmassian (NRC)

Page THE STATUS OF TRAC-BWR PROGRAM . . . . . . . . . . . . . . . . . . . . . 2-1 W. L. Weaver III and G. W. Johnsen (INEL)

RELAPS/ MOD 2 DEVELOPMENT. . . . . . . . . . . . . . . . . . . . . . . . . 2-3 C. S. Miller (INEL)

TRAC CODE DEVELOPMENT STATUS AND PLANS . . . . . . . . . . . . . . . . . 2-5 J. W. Spore et al . (LANL)

NUCLEAR PLANT ANALYZER DEVELOPMENT AT THE IDAHO NATIONAL EilGINEERING LABORATORY . . . . . . . . . . . . . . . . . . . . . . . . 2-7 E. T. Laats (INEL)

BWR PLANT ANALYZER AT BNL. . . . . . . . . . . . . . . . . . . . . . . . 2-11 W. Wulff, H. S. Cheng and A. N. Mallen (BNL)

THE NUCLEAR PLANT DATA BANK. . . . . . . . . . . . . . . . . . . . . . . 2-13 C. P. Booker, M. R. Turner and J. W. Spore (LANL)

SESSION 3 Equipment Qualification Chairperson: G. H. Weidenhaner (NRC)

EFFECTS OF SEISMIC LOADS ON THE OPERABILITY AND LEAK INTEGRITY OF CONTAINMENT ISOLATION VALVES. . . . . . . . . . . . . . . . . . . . 3-1 R. Steele, Jr. and R. C. Hill (INEL)

PARAMETERS IMPORTANT TO REACTOR COOLANT PUMP SEAL STABILITY DURING STATION BLACK 0UT. . . . . . . . . . . . . . . . . . . . . . . . 3-3 R. C. Hill (INEL) and C. A. Kitbner ( AECL)

DYNAMIC LOAD EFFECTS ON GATE VALVE OPERABILITY: RESULTS OF THE HDR-VKL TEST PROGRAM . . . . . . . . . . . . . . . . . . . . . . . 3-5 R. Steele, Jr. (INEL)

PROGRESS ON QUALIFICATION TESTING METHODOLOGY STUDY OF ELECTRIC CABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-7 Y. Kusama et al . (JAERI)

GENERAL TECHNICAL REQUIREMENTS FOR SPECIAL VALVES USED IN NUCLEAR POWER PLANTS . . . . . . . . . . . . . . . . . . . . . . . . . 3-9 J. Zdarek and F. Jukl (SRI, Czechoslovakia)

EFFECTIVENESS AND SAFETY ASPECTS OF SELECTED DECONTAMINATION METHODS FOR LWRs . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11 S. W. Duce and F. B. Simpson (INEL) vi

SESSION 4 Integral Systems Testing Chairperson: D. Solberg (NRC)

Page PKL III SMALL BREAKS AND TRANSIENTS EXPERIMENTAL PROGRAMME . . . . . . . 4-1 R. M. Mandl, B. Brand and H. Watzinger (KWU)

MIST TEST RESULTS. . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3 H. R. Carter and J. R. Gloudemans (B&W)

TRAC-PF1/ MODI PRETEST PREDICTIONS OF MIST EXPERIMENTS. . . . . . . . . . 4-5 B. E. Boyack and J. L. Steiner (LANL)

UMCP 2x4 LOOP TEST RESULTS . . . . . . . . . . . . . . . . . . . . . . . 4-7 Y. Y. Hsu et al . (U. of Md.)

SEMISCALE REC 0VERY INVESTIGATIONS: A COMPARIS0N OF RESULTS FROM SEMISCALE MOD-2C SMALL BREAK LOCA WITHOUT HPI TESTS. . . . . . . . . . 4-9 J. E. Streit and T. J. Boucher (INEL)

THE RESULTS OF 5% SMALL BREAK LOCA TESTS AND NATURAL CIRCllLATION TE S TS A T R OS A-I V L S TF . . . . . . . . . . . . . . . . . . . . . . . . . 4-11 K. Tasaka et al . (JAERI)

COMPARIS0N OF THE TRAC CALCtlLATION TO THE DATA FROM LSTF RUN SB-CL-05. . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13 F. Motley (LANL)

MANAGEMENT OF EMERGENCY FEEDWATER DURING NATURAL CIRCULATION C00LDOWN FOLLOWING A LOSS OF 0FF-SITE POWER SCENARIO . . . . . . . . . 4-15 E. H. Davidson and M. A. Stutzke ( Fla. Power)

SESSION 5 Mechanical and Structural Research Chairperson: J. J. Burns (NRC)

PROBABILITY OF FAILURE IN BWR REACTOR COOLANT PIPING . . . . . . . . . . 5-1 G. S. Holman, T. Lo and C. K. Chou (LLNL)

SS I AND STR UCTURAL BE NC HMAR KS. . . . . . . . . . . . . . . . . . . . . . 5-3 A. J. Philippacopoulos, C. A. Miller and C. J. Costantino (BNL),

and H. Graves (NRC)

VALVE PERFORMANCE TESTING - CHECK VALVE RESULTS. . . . . . . . . . . . . 5-5 N. M. Jeanmougin (ETEC)

PIPING AND FITTING DYNAMIC RELIABILITY PROGRAM . . ........ .. 5-7 D. Guzy (NRC) 6-INCH DI AMETER PIPE SEISMIC FRAGILITY TEST. . . . . . . . . . . . . . . 5-9 W. P. Chen, V. teVita and A. T.,Dnesto (ETEC) vii a _ - _ . - - - . -

SESSION 5

( Con t ' d)

Page PIPING DAMPING STUDIES . . . . . . . . . . . . . . . . . . . . . . . . . 5-11 A. G. Ware (INEL)

NONLINEAR PIPING DAMPING AND RESPONSE PREDICTIONS. . . . . . . . . . . . 5-13 L. K. Severud et al . (HEDL)

EXAMPLES OF NRC RESEARCH PRODUCTS USED IN REGULATION . . . . . . . . . . 5-15 N. Anderson (NRC)

SESSION 6 TMI-2 Analyses Chairperson: R. B. Foulds (NRC)

INTRODUCTION TO TMI-2 RESEARCH PROGRAM . . ...............

D. McPherson (00E)

O V ER V I EW 0F TM I -2 PRO GR AM . . . . . . . . . . . . . . . . . . . . . . . .

P. Grant (INEL)

TMI-2 LOWER VESSEL DERRIS EXAMINATIONS . . . . . . . . . . . . . . . . . 6-1 D. W. Akers, C. S. Ol sen and R. V. Strain (INEL)

PRELIMINARY RESULTS OF THE TMI-2 CORE BORES. . . . . . . . . . . . . . . 6-3 R. K. McCardell et al . (INEL)

TMI-2 ACCIDE NT SC ENAR IO UPDATE . . . . . . . . . . . . . . . . . . . . . 6-7 B. Tolman, P. Kuan and J. Broughton (INEL)

UPDATE ON STANDARD PROBLEM, DATA BASE AND UNCERTAINTIES. . . . . . . . . 6-13 D. W. Golden (INEL)

Tue"1ay, October 28, 1986 SESSION 7 Severe Accident Sequence Analysis Chairperson: T. J. Walker (NRC)

EFFECTIVENESS OF THE BWR MARK I SECONDARY CONTAINMENT IN SEVERE ACCIDENT MITIGATION. ......................... 7-1 S. R. Greene and S. A. Hodge (ORNL)

CONTAINMENT VENTING AS A BWR ATWS MITIGATION TECHNIOUE . . . . . . . . . 7-3 R. M. Harrington (0RNL)

EFFECTS OF LATERAL SEPARATION OF OXIDE AND METAL COMPONENTS OF CORE DEBRIS ON THE BWR MK I CONTAINMENT DRYWELL FLOOR. . ....... 7-5 C. R. Hyman and C. F. Weber (ORNL) viii

SESSION 7

( Con t ' d )

Page SMALL BREAK LOCA MITIGATION FOR BELLEFONTE . . . . . . . . . . . . . . . 7-7 P. D. Bayless and C. A. Dobbe (INEL)

INTEGRATED SCDAP/RELAPS ANALYSIS OF A BWR HIGH PRESSURE BOILOFF. . . . . 7-9 R. Chambers (INEL)

CALCULATIONS OF CONTAINMENT LOADING DUE TO HIGH PRESSURE EJECTION OF CORE DEBRIS. . . . . . . . . . . . . . . . . . . . . . . .

R. Gasser (SNL)

PROBLEMS WITH INDUSTRY'S IMPLEMENTATION OF REQUIREMENTS FOR UPGRADING EMERGENCY OPERATING PROCEDURES . . . . . . . . . . . . . . . 7-11 J. P. Bongarra, Jr. (NRC)

SESSION 8 Separate Ef fects/ Experiments and Analyses Chairperson: N. Zuber (NRC)

TWO-PHASE FLOW MEASUREMENT IN UPPER PLENUM 0F A PWR DURING REFLOOD. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-1 P. Gri f fith and K. M. Gawlik (MIT) l DRYOUT FRONT MODELING FOR R0D BUNDLES. . . . . . . . . . . . . . . . . . 8-3 P. Grif fith, J. A. Mohamed and D. Brown (MIT)

STEAM SEPARATOR MODELLING FOR VARIOUS NUCLEAR REACTOR ACCIDENTS. . . . . 8-5 C. Y. Paik et al . (MIT)

MEASUREMENT AND MODELING 0F VOID AND VELOCITY PROFILES IN CURVED CHANNELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-7 Y. Y. Hsu and W. K. Lin (U. of Md .)

SIMULATION EXPERIMENTS FOR HOT LEG U-BEND TWO-PHASE FLOW PHENOMEba. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-9 M. Ishii, J. T. Hsu and G. Lambert ( ANL)

THERMAL MIXING TESTS IN A SEMIANNULAR 00WNCOMER WITH INTERACTING F LOW S F R OM COLD LE GS . . . . . . . . . . . . . . . . . . . . . . . . . 8-11 H. Tuanisto (IV0, Finland)

SCALING OF THERMAL MIXING PHENOMENA FROM 1/5 TO FULL SCALE TE S T F AC I L I T I E S . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-13 T. G. Theofanous, K. Iyer and E. Shabana (UCSB) 1x

SESSION 9 Seismic Research Chairperson: A. Murphy (NRC) '

Page OVERVIEW . ...............................

J. E. Richardson (NRC)

COMP 0NENT FRAGILITY RESEARCH PROGRAM: PRIORITIZATION AND 1 DEMONSTRATION TESTING. . . . . . . . . . . . . . . . . . . . . . . . . 9-1 G. S. Holman and C. K. Chou (LLNL)

SYNTHESIZING SEISMIC FRAGILITY OF COMP 0NENTS BY USE OF EXISTING DATA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-3 K. K. Bandyopadhyay and C. H. Hofmayer (BNL)

NRC'S SEISMIC MARGINS REVIEW 0F NUCLEAR POWER PLANTS . . . . . . . . . . 9-7 R. C. Murray and P. G. Prassinos (LLNL)

VALIDATION OF SEISMIC S0Il-STRUCTURE INTERACTION ANALYSIS METHODS. . . . 9-9 2PRI/NRC Cooperation in Lotung, Taiwan Experiments C. A. Kot, M. G. Srinivasan and B. J. Hsieh ( ANL) and Y. K. Tang and R. P. Kassawara (EPRI)

HDR PHASE II VIBRATIONAL EXPERIMENTS . . . . . . . . . . . . . .... 9-11 L. Malcher (KfK) and C. A. Kot ( ANL)

THE SEISMIC CATEGORY I STRUCTURES PROGRAM. . . . . . . . . . . . . . . . 9-13 J. G. Bennett, C. R. Farrar and W. E. Dunwoody (LANL)

DEVELOPMENT OF SITE SPECIFIC RESPONSE SPECTRA. . . . . . . . . . . . . . 9-15 J. B. Savy, D. L. Bernreuter and J. C. Chen (LLNL) 1 NATIONAL SEISM 0 GRAPHIC NETWORK FOR THE EASTERN UNITED STATES . . . . . . 9-17 A. J. Murphy (NRC)

SESSION 10 International Code Assessment Program Chairperson: D. Bessette (NRC)

SR D OF U K A EA V I EWS O N C H ER N0B Y L . . . . . . . . . . . . . . . . . . . . .

• J. Gittus (UKAEA)

PHYSICAL MODEL FOR REACTOR COOLANT PUMPS . . . . . . . . . . . . . . . 10-1 K. Schneider and F. Winkler (KWU)

RESULTS FROM ASSESSMENT OF RELAP5/M002 AND TRAC-PF1/M001. . . . . . . 10-3 F. Winkler (KWU)

CSNI VALIDATION MATRIX FOR PWR AND BWR CODES . . . . . . . . . . . . . 10-5 K. Wol fert (GRS) and I. Brittain (UKAEA)

X

SESSION 10

( Con t ' d )

Page SWEDISH EXPERIENCE WITH RELAPS/ MOD 2 ASSESSMENT . . . . . . . . . . . . 10-7

0. Sandervag (Studsvik)

FINNISH ASSESSMENT OF RELAP5/ M002. . . . . . . . . . . . . . . . . . . 10-9 H. Holmstrom (VTT)

TRAC DEVELOPMENT AT GENERAL ELECTRIC . . . . . . . . . . . . . . . . . 10-11 J.G.M. Andersen, J. C. Shaug and B. S. Shiralkar (GE) 3-D NEUTRONICS MODEL IMPLEMENTATION INTO TRAC-B01. . . . . . . . . . . 10-13 S. Tsunoyama and H. Uematsu (NAIG), J. C. Shaug and B. S. Shiralkar (GE), and H. Namba (Toshiba)

VALIDATION OF TRAC-PD2 AGAINST EXPERIMENT LP-02-6 0F THE LOF T-0 ECD E XP ER IM E NT . . . . . . . . . . . . . . . . . . . . . . . . 10-17 A. Alonso et al . (Madrid Poly. Inst.) and J. L. Mora (Vandellos Nuclear Assn.)

SESSION 11 Fission Product Release and Transport in Containment Chairperson: S. B. Burson (NRC)

INFLUENCE OF MOISTURE ON THE BEHAVIOR OF AEROSOLS. . . . . . . . . . . 11-1 R. E. Adams, A. W. Longest and M. L. Tobias (ORNL)

L AC E PR OGR AM R E SULT S . . . . . . . . . . . . . . . . . . . . . . . . . .

• L. Muhlestin (HEDL)

SUMMARY

OF AEROSOL CODE - COMPARIS0N RESULTS FOR LWR AEROSOL CONTAINMENT TESTS LA1, LA2 AND LA3 . . . . . . . . . . . . . . . . . 11-3 A. L. Wright, J. H. Wil son and P. C. Arwood (ORNL)

INTEGRATED EX-VESSEL SOURCE TERM ANALYSIS WITH THE CONTAIN 1.1 C0hPUTER CODE. . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-5 K. K. Murata et al . (SNL) and G. D. Valdez (Technadyne)

SUSTAINED URANI A/ CONCRETE INTERACTIONS (SURC): EXPERIMENTS A ND A NAL YS I S . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-7

0. R. Bradley and E. R. Copus (SNL)

RESULTS FROM THE DEMONA AEROSOL EXPERIMENTS, . . . . . . . . . . . . . .

• T. Kanzleiter (Battelle-Frankfurt) and W. Shoeck (KfK)

THERMAL-HYDRAULICS STUDIES ON M0LTEN CORE-CONCRETE INTERACTIONS. . . . 11-11 G. A. Greene (BNL) l l

xi

SESSION 11

( Cont' d )

Page ICE CONDENSER TESTING F ACILITY AND PLANS . . . . . . . . . . . . . . . 11-13 L. D. Kannberg et al . (PNL)

SUMMARY

OF KfK BETA EXPERIMENTS AND ANALYSIS WITH THE WECHSL A ND CORCON C ODE S . . . . . . . . . . . . . . . . . . . . . . . . . . 11-15 H. Al smeyer and M. Reiman (KfK) and R. K. Cole, Jr. (SNL)

SESSION 12 Nuclear Plant Aging Chairperson: J. Vora (NRC)

RISK EVALUATIONS OF AGING PHENOMENA. . . . . . . . . . .... . . . . 12-1 W. E. Vesely (SAIC) and D. G. Satterwhite (INEL)

REACTOR PROTECTION SYSTEl1 AGING: RESULTS OF A PILOT COMMERCIAL PLANT S TUDY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12-3 P. T. Jacobs and L. C. Meyer (INEL) and O. Larsen (Duke Power)

OVERVIEW 0F RECE NT OPERATING EXPERIENCE. . . . . . . . . . . . . . . . 12-5 G. M. Holahan and M. A. Caruso (NRC)

APPLICATION OF DI AGNOSTICS TO DETERMINE MOTOR-0PERATED VALVE OPERATIONAL READINESS. . . . . . . . . . . . . . . . . . . . . . . . 12-7 D. M. Eissenberg (ORNL)

THE EFFECTS OF RELAY AND CIRCUIT BREAKER AGING IN A SAFETY-RELATED SYSTEM . . . . . . . . . . . . . . . . . . . . . . . . . . . 12-9 G. J. Toman et al . (Franklin Res. Center) and W. Gunther (BNL)

NUCLEAR SERVICE EMERGENCY DIESEL GENERATOR AGING RESEARCH. . . . . . . 12-11 J. W. Vause III and D. A. Dingee (PNL)

Wednesday, October 29, 1986 SESSION 13 International Code Assessment Program Chairperson: F. Odar (NRC)

APPLICATION AND TESTING 0F A METHOD TO QUANTIFY CODE UNCERTAINTY . . . 13-1 R. G. Hinson (INFL)

SANDIA CODE ACCURACY QUANTIFICATION AND ITS APPLICATION TO TRAC-PF1/ M001 ASSESSMENT , . . . . . . . . . . . . . . . . . . . . . 13-3 L. N. Kmetyk , R. K. Byers and L. D. Buxton (SNL) ar

• M. G. Elrick (Dikewood) x11

SESSION 13

( Con t ' d )

Page TEST OF THE TRAC CODE AGAINST KNOWN ANALYTICAL SOLUTIONS FOR S TRA T IF IE D F L OW. . . . . . . . . . . . . . . . . . . . . . . . . . . 11- 5 P. S. Black and D. C. Leslie (Queen Mary College), and G. F. Hewitt (UKAEA)

COLD LEG CONDENSATION IN A LARGE BREAK LOCA USING TRAC-PF1/M001. . . . 13-7 J. T. Dawson (CEGB, UK)

SOME PRELIMINARY RESULTS OF POST-DRYOUT HEAT TRANSFER MEASURE-MENTS AT LOW QUALITIES AMD PRESSURES UP TO 20 BAR. . . . . . . . . . 13-9 K. G. Pearson, D. Swinnerton and R. O'Mahoney, UKAEA JRC ISPRA EXPERIENCE WITH THE IBM VERSION OF RELAPS/M002 . . . . . . . 13-11 W. Kolar and H. Stadtke (Ispra)

ASSESSMENT AND UNCERTAINTY IDENTIFICATION FOR RELAP5/M002 AND TRAC-BD1/M001 CODES UNDER UNC0VERY AND REFLOODING CONDITIONS . . . . 13-13 S. N. Aksan, M. Richner and G.Th. Analytis (EIR), and M. Andreani (ETH), Switzerland STATUS OF MEXIC AN NUCLEAR POWER PROGR AM. . . . . . . . . . . . . . . . .

M. Medina (CNSNS)

STATUS OF J-TRACK CODE DEVELOPMENT . . . . . . . . . . . . . . . . . . 13-15 H. Akimoto et al . (JAERI)

APPLICATION 0F ENGINEERING AND BEST-ESTIMATE CODES TO HDR THERMAL M I X I NG E X PER IME NTS . . . . . . . . . . . . . . . . . . . . . . . . . 13-17 L. Wol f et al . (Kf K) and T. G. Theofanous (UCSB)

SESSION 14 Severe Accident Source Term Chairperson: J. Mitchell (NRC)

OVA S AR U NC ER TA I N T Y STUD Y . . . . . . . . . . . . . . . . . . . . . . . 14 - 1 M. Khatib-Rahbar et al . (BNL)

MECHANISTIC CODE DEVELOPMENT AND APPLICATIONS. . . . . . . . . . . . . .

• J. Han (NRC)

MELPROG/ TRAC UPDATE AND APPLICATIONS . . . . . . . . . . . . . . . . 14-5 R. J. Henninger (LANL) and J. E. Kelly (SNL)

SCDAP/RELAPS UPDATE AND APPLICATIONS . . . . . . . . . . . . . . . . . 14-7 C. M. Allison et al . (INEL) xiii

SESSION 14

( Con t ' d )

Page IMPLEMENTATION OF SEVERE ACCIDENT POLICY . . . . . . . . . . . . . . . .

Z. Rosztoczy (NRC)

STCP VALIDATION. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

J. Gieseke (BCD) and M. Khatib-Rahbar (BNL)

SESSION 15 Research in Nondestructive Evaluation Chairperson: J. Muscara (NRC)

EVALUATION AND IMPROVEMENT IN NDE RELI ABILITY FOR INSERVICE INSPECTION . . . . . . . . . . . . . . . . ............. . 15-1 S. R. Doctor et al . (PNL)

DEVELOPMENT AND VALIDATION OF A REAL-TIME SAFT-UT SYSTEM FOR INSERVICE INSPECTION OF LWRs . . . . . . . . . . . . . . . . . . . . 15-3 S. R. Doctor et al . (PNL)

PROGRESS FOR ON-LINE AC0USTIC EMISSION MONITORING 0F CRACKS IN REACTOR SYSTEMS. . . . . . . . . . . . . . . . . . . . . . . . . . . 15-5 P. H. Hutton, R. J. Kurtz and M. A. Friesel (PNL)

RELI ABILITY OF LEAK DETECTION SYSTEMS IN LWRs. . . . . . . . . . . . . 15-7

0. S. Kupperman ( ANL)

PROGRAMME FOR THE INSPECTION OF STEEL COMPONENTS, PISC 11 RESULTS AND P I SC 111 PL A NS . . . . . . . . . . . . . . . . . . . . . . . . . 15-9 S. J. Crutzen (CEC /JRC-Ispra)

IMPROVED EDDY CURRENT TESTING 0F STEAM GENERATOR TUBING. . . . . . . . .

C. V. Dodd (ORNL)

SESSION IS A Poster Session AN EXPERT SYSTEM FOR USNRC EMERGENCY RESPONSE. . . . . . . . . . . . . 15A-1 D. E. Sebo, M. A. Bray and M. A. King (INEL)

NUCLEAR REGULATORY COMMISSION PROGRAMS ARE SUPPORTED BY THE TECH-N0 LOGY RESOURCES OF THE ENGINEERING PHYSICS INFORMATION CENTERS. . . 15A-3 N. A. Hatmaker , R. W. Roussin and J. E. White (0RNL)

NRC MOBILE NDE LABORATORY. . . . .....................

H. Kerch and J. Wiggins (NRC) xiv 9 _ _ _ _ _ _ _ _ _ _ _ -

SESSION 16 Containment Systems Research/ Containment Loads Analysis Chairperson: T. Lee (NRC)

Page DCH-1: THE FIRST DIRECT CONTAINMENT HEATING EXPERIMENT IN THE SUr7SEY TEST FACILITY. . . . . . . . . . . . . . . . . . . . . . . 16-1 W. W. Tarbell, J. E. Brockmann and M. Pilch (SNL)

RECENT EXPERIMENTAL AND ANALYTICAL RESULTS OF BNL DIRECT CONTAINMENT HE ATING PROGR AMS . . . . . . . . . . . . . . . . . . . . 16-3 T. Ginsberg and N. K. Tutu (BNL)

DEVELOPMENT AND APPLICATIONS OF THE INTERIM DIRECT HEATING MODEL FOR THE CONTAIN COMPUTER CODE. . . . . . . . . . . . . . . . . . . . 16-5 K. D. Bergeron et al . (SNL) and J. L. Tills (Jack Tills and Associates)

RECENT RESULTS IN HYDROGEN RESEARCH. . . . . . . . . . . . . . . . . . 16-7 M. Berman, M. P. Sherman and C. C. Wong (SNL)

STE AM EXPLOS ION ENERGET ICS . . . . . . . . . . . . . . . . . . . . . . 16-9 T. G. Theofanous et al . (UCSB) and B. Najafi and E. Rumble (SAIC)

RECENT RESULTS IN FCI RESEARCH , . . . . . . . . . . . . . . . . . . . 16-11 M. Berman et al . (SNL)

PRELIMINARY RESULTS OF AN EXPERIMENTAL STUDY OF CORE MELT-WATER MIXING IN MFTF . . . . . . . . . . . . . . . . . . . . . . . . . . . .

• D. F. . Fletcher and M. Bird (UKAEA)

LARGE MODEL TESTS FOR CONTAINMENT PERFORMANCE AND SEPARATE EFFECTS FOR PENETRATION. .......................

• W. A. von Riesemann (SNL)

EXPERIMENTS AND BEST ESTIMATE CODE DEVELOPMENT ON SMALL BREAK LOCA THERMAL-HYDRAULICS AND HYDR 0 GEN MIXING WITHIN SUBCOMPARTMENTAL C ONTA I NME NT VE SSEL . . . . . . . . . . . . . . . . . . . . . . . . . 16-13 A. Tsugue (Mitsubishi)

SESSION 17 Environmental Effects in Primary System Components Chairpersons: J. Muscara and A. Taboada (NRC)

BWR PIPE CRACK REMEDIES EVALUATION . . . . . . . . . . . . . . . . . . 17-1 W. J. Shack et al . ( ANL)

AGING DEGRADATION OF CAST STAINLESS STEEL. . . . . . . . . . . . . . . 17-3

0. K. Chopra and H. M. Chung ( ANL)

EFFECTS OF WELDING AND WELD REPAIR ON STRESS CORR 0SION CRACKING RESISTANCE OF STAINLESS STEEL PIPING . . . . . . . . . . . . . . . . 17-5 S. M. Bruemmer and D. G. Atteridge (PNL) xv

SESSION 17

( Con t ' d )

Page SURRY STEAM GENERATOR EXAMINATION AND EVALUATION . . . . . . . . . . . 17-7 R. J. Kurtz et al. (PNL)

THE EFFECT OF CRACK SHAPE AND VARIABLE AMPLITUDE LOADING ON FATIGUE CRACK GROWTH IN PWR ENVIRONMENTS . . . . . . . . . . . . . . 17-9 W. H. Cullen (MEA)

SESSION 18 Risk Analysis /PRA Applications Chairperson: R. C. Robinson (NRC)

USE OF PRAs IN REGULATION. . . ... . . ................

F. Congel (NRC)

OPERATIONAL SAFETY RELI ABILITY RESEARCH. . . . . . . . . . . . . . . . 18-1 R. E. Hall and J. L. Boccio (BNL)

R ISK-B ASED PERFORM ANCE INDIC A TORS. . . . . . . . . . . . . . . . . . . 18-3 M. A. Azann and J. L. Boccio (BNL) and W. E. Vesely and E. Lofgren (SAIC)

PROCEDURES FOR EVALUATING TECHNICAL SPECIFICATIONS (PETS) . . . . . . . 18-5 P. K. Samanta and J. L. Boccio (BNL)

AN OVERVIEW 0F THE PLANT RISK STATUS INFORMATICd MANAGEMENT SYSTEM . . 18-7 J. R. Kirchner and D. J. Campbell (JBF Associates)

SYSTEM ANALYSIS AND RISK ASSESSMENT SYSTEM ( SARA) . . . . . . . . . . . 18-9 E. A. Krantz, K. D. Russell and H. D. Stewart (INEL)

MANAGEMENT OF GENER IC SAFETY ISSUES. . . . . . . . . . . . . . . . . . 18-11 W. Minners (NRC)

STATUS OF SAFETY G0AL IMPLEMENTATION . . . . . . . . . . . . . . . . . .

G. Sege (NRC)

Thursday, October 30, 1986 S_ESSION 19 Pressi're Vessel Research Chairperson: M. Vagins (NRC)

INTEGRATION OF WIDE-PLATE CRACK-ARREST TEST RESULTS. . . . . . . . . . .

D. J. Naus and C. E. Pugh (ORNL)

VISC0 PLASTIC ANALYSES OF LARGE CRACK-ARREST SPECIMENS WITH VARIOUS CONSTITUTIVE EQUATIONS AND FRACTllRE CRITERI A . . . . . . . . . . . . .

B. R. Bass (0RNL) and C. W. Schwartz (U. of Md.)

xvi

SESSION 19

( Con t' d)

Page DEVELOPMENT AND APPLICATION OF A DYNAMIC VISC0 PLASTIC FRACTURE MODEL. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19-1 M. F. Kanninen et al . (SWRI)

BRITTLE-TO-DUCTILE TRANSITION BEHAVIORS IN NUCLEAR REACTOR VESSEL STEELS. . . . . . . . . . . . . . . . . . . . . . . . . . . . 19-3 G. R. Irwin (U. of Md .)

FLAW DENSITY EXAMINATIONS OF A CLAD BOILING WATER REACTUR PRESS URE VESSEL SE GME NT. . . . . . . . . . . . . . . . . . . . . . . 19-5 K. V. Cook and R. W. McClung (0RNL)

PTSE-2: AN HSST PRESSURIZED THERMAL SH0CK EXPERIMENT WITH LOW UPPER-SHELF ENERGY TOUGHNESS MATER I AL. . . . . . . . . . . . . . . . .

• R. H. Bryan et al. (0RNL)

EVALUATION OF SURFACE CRACKS EMBEDDED IN REACTOR VESSEL CLADDING . . . .

D. E. McCabe (MEA)

SESSION 20 Reference Plant Risk Analysis - NUREG-1150 Chairperson: J. Murphy (NRC)

N UR EG-1150 - A N OVE RV I EW . . . . . . . . . . . . . . . . . . . . . . . .

• J. Murphy (NRC)

ACCIDENT CHARACTERIZATION METHODOLOGY. . . . . . . . . . . . . . . . . .

• A. Camp (SNL)

RESULTS OF CORE DAMAGE FRE00ENCY ANALYSIS FOR THE REFERENCE PLANTS . . .

F. Harper (SNL)

CONTAINMENT VENTING ANALYSIS FOR PEACH BOTTOM. . . . . . . . . . . . . 20-1

0. L. Batt, H. S. Blackman and W. R. Nelson (INEL) and M. T. Leonard (BCD)

UNC ER TA I NTY ANAL YS IS METHODS . . . . . . . . . . . . . . . . . . . . . .

• A. Benjamin (SNL)

SESSION 21 Industry Safety Research Chairperson: W. B. Loewenstein (EPRI)

NUCLEAR SAFETY - FORWARD ON TECHNOLOGY AND BACKWARD ON PERCEPTION. . . 21-1 W. B. Loewenstein and G. R. Thomas (EPRI)

CONTAINMENT AND PIPING RESEARCH. . . . . . . . . . . . . . . . . . . . 21-3 H. T. Tang and S. W. Tagart (EPRI) xvii

SESSION 21

( Con t' d )

Page USE OF PROBABILISTIC SYSTEM ANALYSIS FOR ENHANCING PLANT OPERATION SAFETY AND PRODUCTIVITY. . . . . . . . . . . . . . . . . . . . . . . 21-5 B. B. Chu (EPRI)

A UTILITY PERSPECTIVE ON PLANT MODIFICATIONS RESULTING FROM S AF E T Y R E S EA RC H. . . . . . . . . . . . . . . . . . . . . . . . . . . 21 -7 R. E. Deem (NYPA) and P. P. Bieniarz (Risk Mgmt. Assoc.)

A SEISMIC HAZARD METHODOLOGY FOR THE CENTRAL AND EASTERN UNITED STATES. . . . . . . . . . . . . . . . . . . . . . . . . . . . 21-9 J. C. Stepp and J. L. King (EPRI)

THE ROLE OF OVERSIGHT WITHIN NRR AND ITS EFFECT ON INDUSTRY. . . . . . 21-11 R. J. Bosnak (NRC)

SESSION 22 Irradiation Effects on RPV Steels Chairperson: A. Taboada (NRC)

EVALUATION OF SURFACE CRACKS EMBEDDED IN REACTOR VESSEL CLA0 DING . . . 22-1 D. E. McCabe (MEA)

COMPOSITION AND TEMPERATURE EFFECTS ON ANNEALING /REIRRADIATION AND DOSE-RATE EFFECTS ON IRRADI ATION EMBRITTLEMENT . . . . . . . . . . . 22-3 J. R. Hawthorne (MEA)

RADIATION EMBRITTLEMENT CORRELATION BETWEEN POWER AND TEST

• REACTOR DATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

G. L. Guthrie (HEDL)

SESSION 23 Degraded Piping Research Chairperson: M. Mayfieli (NRC)

PROGRESS AND RESULTS FROM THE DEGRADED PIPING PROGRAM, PHASE II. . . . .

G. Wilkowski et al. (BCD)

STATUS OF RESOLUTION OF INTERGRANULAR STRESS CORROSION CRACKING

• AT BWRs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

W. Hazel ton (NRC)

PIPING FRACTURE MECHANICS DATA BASE. . . . . . . . . . . . . . . . . . 23-1 A. L. Hi ser ( MEA) xviii f

SESSION 24 Reference Plant Risk Analysis - NUREG-1150 Chairperson: J. Murphy ( NRC)

Page R ISK AND RISK REDUCTION RESULTS FOR SURRY. . . . . . . . . . . . . . . .

• G. Boyd (SAROS)

CONTAINMENT PERFORMANCE ANALYSIS METHODS . . . . . . . . . . . . . . . .

• A. Benj anin (SNL)

SOURCE TERM ANALYSIS METJ0DS FOR NUREG-1150. . . . . . . . . . . . . . .

• R. Denning (BCD)

DEVELOPMENT AND STATUS OF MELCOR CODE. . . . . . . . . . . . . . . . . .

24-1 F. E. Haskin and S. f. Webb (SNL)

ANALYSIS OF PEACH BOTTOM STATION BLACK 0UT WITH MELCOR. . . . . . . . . 24-3 S. E. Dingman et al . (SNL)

MELCOR VALIDATION RESULTS. . . . . . . . . . . . . . . . . . . . . . . 24-5 C. D. Leigh and R. K. Byers (SNL) and C. J. Shaffer (SEA)

SESSION 25 2D/3D Research Chairperson: G. S. Rhee (NRC)

SUMMARY

OF CCTF TEST R ESULTS . . . . . . . . . . . . . . . . . . . . . 25-1 T. Iguchi et al . (JAERI)

SCTF-!!! TEST PLAN AND RECENT SCTF-III TEST RESULTS. . . . . . . . . . 25-3 T. Iguchi et al . (JAERI)

HEAT TRANSFER ENHANCEMENT IN SCTF TESTS. . . . . . . . . . . . . . . . 25-5 T. Iwamura et al . (JAERI)

DEVELOPMENT OF REFLOOD MODEL AT JAERI. . . . . . . . . . . . . . . . . 25-7 Y. Murao et al . (JAERI)

UPTF TEST RESULTS, FIRST 3 SEPARATE EFFECT TESTS . . . . . . . . . . . 25-9 P. A. Weiss and R. J. Hertlein (KWU)

ANALYSIS RESULTS FROM THE LOS ALAMOS 2D/3D PROGRAM . . . . . . . . . . 25-13 B. Boyack and M. Cappiello (LANL)

MULTIDIMENSIONAL REPRESENTATION OF GPWR PRIMARY SYSTEM IN 200%

BR EAK LOCA CALC ULAT ION . . . . . . . . . . . . . . . . . . . . . . . 25-15 B. Riegel and K. Liesch (GRS) and H. Plank (KWU)

XIX

Friday, October 31, 1986 SESSION 26 Panel Discussion of Regulatory Issues Panelists to be announced in the final agenda.

SESSION 27 Innovative Concepts for Increased Safety of Advanced Power Reactors Co-Chairpersons: T. King and C. N. Kelber (NRC)

Page A SAFE ENVIRONMENT FOR DEMONSTRATING ADVANCED REACTORS . . . . . . . 27-1 J. L. Dooley and R. P. Hammond (Telephos)

AN EXAMINATION OF THE BASES FOR PROPOSED INNOVATIONS IN REACTOR TECHNOLOGY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-3

0. L. Moses (ORNL)

A RECONFIGURED PWR WITl ULTRA-SAFE CHARACTERISTICS . . . . . . . . . . 27-5 M. A. Schultz (Penna. State Univ.)

AUTOMATION AND ARTIFICIAL INTELLIGENCE FOR INCREASED SAFETY. . . . . . 27-7 W. G. Kennedy ( NRC)

ADVANCED LIGHT WATER REACTORS FOR THE NINETIES . . . . . . . . . . . . 27-9 F. A. Ross (DOE) and W. R. Sugnet (EPRI)

THE IMPACT OF PASSIVE SAFETY REQUIREMENTS ON PLANT DESIGN. . . . . . . 27-11 F. X. Gavigan, J. R. Fbmphreys and A. C. Millunzi (DOE)

THE PASSIVE CONTAINMENT SYSTEM (PCS-2) . . . . . . . . . . . . . . . . 27-13

0. B. Falls, Jr. and F. W. Kleinola (NucleDyne)

PLENARY SESSION Chairperson: D. F. Ross (NRC)

Concluding Renarks XX

/

EVALUATION OF SCALING CONCEPTS FOR INTEGRAL SYSTEM TEST FACILITIES

• K. G. Condie T. K. Larson Idaho National Engineering Laboratory EG&G Idaho, Inc.

The United States Nuclear Regulatory Commission (NRC) is currently assessing their future research needs for integral thermal-hydraulic reactor safety experiments. To assist NRC in this task, the Idaho National Engineering Laboratory (INEL) has developed a technical basis on which to evaluate scaling concepts for proposed facilities. '

Several scaling concepts were evaluated based on their ability to produe.e important thermal-hydraulic phenomena. The important theimal-5ydraulic phenomena were determined in four steps as follows.

First, the important thermal-hydraulic transients were identified based on potential significance to reactor safety and the production of a wide range of thermal-hydraulic phenomena. Five important transient classes were identified including 1) increase in heat removal, 2) decrease in heat rem; val, 3) anticipated transients without scram 4) small break loss-of-coolant accident, and 5) large break loss-of-coolant accident.

Second, the calculational data base for each of the transients was reviewed. The data base consisted of thermal-hydraulic calculations made with the RELAPS and TRAC advanced computer codes. Third, a base plant was identified for Westinghouse and B&W type plants. The base plants were the reference for developing paper models of potential facilities. The base plants were selected using the criteria of plant typicality and the completeness of the thermal-hydraulic calculational data base. The Westinghouse base plant was Seabrook Unit 1 while the B&W base plant was Oconee Unit 1. Finally the important thermal-hydraulic phenomena were identified for both vendor types for each of the important transients. The important phenomena were identified based on a review of the calculational data base and engineering judgment.

General scaling relationships for single-phase and two-phase flow were developed. From the general scaling relationships four potential scaling concepts were identified and important parameter ratios were developed in order to generate a scaled paper model representing each of the four scaling concepts for both Westinghouse and B&W plant designs. The four scaling concepts considered were:

Full Height Full Pressure Water (FHFPW)

Reduced Height Reduced Pressure Water (RHRPW)

Reduced Height Full Pressure Water (RHFPW)

Reduced Height Full Pressure Freon (RHFPF)

For the Full Height Full Pressure Water concept, two potential facilities were evaluated, the " Ideal" and the " Actual." The ideal concept

• Work supported by the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research, Under Department of Energy Contract No.

DE-AC07-76ID01570.

1-1

refers to a facility which follows the theoretical scaling laws ideally or without compromise. The actual concept refers to a facility which is built in compliance with most of the theoretical scaling laws but includes some compromises to meet practical or physical limitations. Semiscale and MIST are examples of Actual FHFPW facilities where the primary piping diameter is significantly larger than the ideal or theoretical value in order to provide a more realistic frictional pressure drop uistribution.

An additional facility concept was also developed as a special case of the Reduced Height Full Pressure concept in which the length and diameter ratios were both equal to 0.1 resulting in a 10 to 1 " Xerox" reduction of the plant. This concept is referred to as the 0.1 scaled Linear Full Pressure Water (LFPW) concept.

The phenomena were evaluated on a local or separate effects basis. The general procedure was to determine how the phenomena actually scaled and how the phenomena should ideally scale. The distortion in the scaling was then determined by a comparison of the actual and ideal scaling. The ability of each concept to scale a wide variety of thermal-hydraulic phenomena, such as break flow, flooding, heat transfer, multi-dimensional effects, etc. was evaluated and a numerical ranking was assigned to each concept to qualitatively indicate its ability to scale each phenomenon. The relative differences between the scaling concepts were then evaluated.

It should be noted here that the local phenomena investigated in this study have been addressed only on an individual basis. Ultimately the synergistic effects will need to be determined using a thermal-hydraulic code to model each facility and then address those transients as specific needs for a proposed facility become more well defined.

The results of this study have shown that the actual FHFPW model, represented by Semiscale and MIST facilities for a PWR, provide the best representation of the largest number of phenomena which were investigated with the RHFPW a close second. Conversely the RHRPW concept represented the least number of phenomena investigated. The RHFPW and LFPW concept ranked closely together in between the actual FHFPW and RHRPW concepts. For each concept, except the RHRPW, there were shown to be phenomena which were best represented by that particular concept.

The scaling concept to be used for the design basis for any new facility should be based on the results of this study combined with the transients and phenomena to be simulated. Considerations must also be given to the amount and type of data already available, the areas of greatest uncertainties in the codes and the availability of existing experimental facilities.

1-2

- _ _ _ _ - . - - - ~ - - _ _ , _ - _ i

PWR Recovery Procedures Investigated in the LOBI-MOD 2 Test Facility C. Adiabbo, L. Piplies Commission of the European Communities Joint Research Centre, Ispra Establishment Thermodynamics Division, LOBI Project I-21020 Ispra (I)

The results from a series of tests simulating PWR recovery procedures in the LOBI-MOD 2 test facility / 1 / are presented. The reported experimental re-sults are specific to the scaled facility and are not directly extrapolable to full size plants; they provide, however, a basis for the phenomenological understanding and for the development and/or assessment of analytical tools used in PWR safety analysis.

In the event of a small break LOCA, the secondary system may be used as an additional heat sink to provide an adequate primary system energy removal and cooldown. In the LOBI-MODE test facility the primary system followed closely the secondary system cooldown for the range of break sizes 0.01A and smaller; the break flow and the emergency-core-cooling systems were seen to be sufficient to ensure an effective primary system depressurization and cooldown for break sizes 0.02A and larger.

Test A2-90-3 represented the recovery of steam generators heat sink from highly degraded conditions via secondary side refill and cooldown. At the start of the test the primary system was at high pressure and temperature, and the secondary side of each steam generator was nearly dry. Throughout the test, core heat removal relied mainly on single-phase and two-phase natural circulation within the primary cooling system. The reference scena-rio may be related to the progression of a fault initiated by the loss of off-site and normal on-site electrical power, with emergency Diesel power being availa'le.

u Test BT-00-3 investigated the main features of potential long term cooldown via primary system bleed (manual opening of the pressurizer relief valve) and feed (automatic actuation of the high pressure emergency-core-cooling in-jection system). The initiating fault was a loss of all feedwater (main and, in time, auxiliary) to the steam generators. At the start of the bleed and feed heat removal process, the primary system was at high pressure and tempe-rature and the steam generators secondary sides were completely dry.

Test BT-01-3 simulated a steam line break accident mitigation phase based on the optimized recovery guidelines. Primary system depressurization using the pressurizer cooling system and unaffected steam generator cooldown constitu-ted the preferred procedures for the recovery from high system pressure and low cold leg temperature.

1-3

k f

i The LOBI-MOD 2 experimental programme / 2 / is, at present, mainly focused on the events characterizing the occurrence of small break LOCA and Special i Transients in PWRs. There are plans to include in such a programme the syste-matic investigation of prospected emergency operating procedures and the verification / validation of transient management strategies.

References:

1. C. Addabbo, L. Piplies, W. L. Riebold: l

} " LOBI-MOD 2: Geometrical Configuration of the Test Facility 'I f l Communication tio. 4010, CEC-JRC, July 1983 i

i 2. W. L. Riebold:

! " LOBI-MOD 2 Programme Status and Plans" Proceedings of the Specialists' Meeting on Small Break LOCA Analyses in l

LWRs", Pisa (I), June 23 - 27, 1985 I

i i

4 4

i a

> 1-4 l

l

, - , , , , - - , - - , , - - - ,-- , - - - - - - - , - - - - ---e- , - - - , - . - - -n-..,.-- , - , - - - - - - ., -n------ ,---- --

THE STATUS OF TRAC-BWR pROGRAMa Walter L. Weaver, III Gary W. Johnsen Idaho National Engineering Laboratory EGSG Idaho, Inc.

The Transient Reactor Analysis Code for Boiling Water Reactors (TRAC-BWR) is being developed at the Idaho National Engineering Laboratory (INEL) for the Division of Accident Evaluation, Office of Nuclear Regulatory Research of the United States Nuclear Regulatory Commission (USNRC).

The objective of this development is to provide the USNRC and the  !

public with a detailed, best estimate, and efficient computer code for the  ;

analysis of postulated accidents and transients in boiling water reactor l (BWR) systems. This program is unique among advanced code development projects in that it focuses on the hardware, thermal-hydraulics, and heat transfer phenomena that distinguish BWR systems and their response in transients. In addition to providing a best estimate analysis capability for BWR systems, the code can also be used to address current licensing concerns such as anticipated transients without scram (ATWS) or the small break loss-of-coolant accident (SBLOCA). It also provides analytical support to the USNRC experimental safety programs. The success of this development is attributed in part to the continuing participation of the General Electric Company as a part of the Full Integral System Test (FIST)

Experimental Program composed by General Electric, the USNRC, and the Electric Power Research Institute (EPRI). l Work on the TRAC-BWR series of codes began in 1979, starting with a developmental version of TRAC-PD2 received from the Los Alamos National l Laboratory. Several versions of TRAC-BWR have been released by the INEL, the latest one being TRAC-BF1. The major new features of this code include a one-dimensional (1-D) two-group neutron kinetics model, a Courant limit violating numerical solution procedure for one-dimensional components, and l an improved solution procedure for the control system. Other improvements I to models in previous versions of the code include improvements to: the water packing logic, momentum coupling between the VESSEL component and the one-dimensional components, subcooled critical flow, the jet pump model and the addition of a stratified interfacial heat transfer model. Modifications were also made to the separator / dryer and turbine component models to make them compatible with the Courant limit violating numerics in the one-dimensional hydrodynamics solution procedure.

a. Work supported by the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research, under DOE Contract No. DE-AC07-761001570.

2-1

In addition to the modeling improvements, extensive changes to the FORTRAN coding were made to make the TRAC-BWR code more portable and machine independent. This work focused on compatibility with the CRAY and IBM computer systems. Finally, the code was modified to make it compatible with the Nuclear Plant Analyzer (NPA) code executive and display system.

TRAC-BF1 was tested against twelve developmental assessment test cases, which included both separate effects and integral system cases. This developmental assessment was completed in June 1986 and TRAC-8F1 is now i available upon request and with the concurrence of the USNRC. The separate effects test cases included both hydrodynamic and heat transfer test cases while the integral system test cases included: a large break in the Two Loop Test Apparatus (TLTA), a small break and a power transient ( ATWS like test) in the Full Integral System Test (FIST) facility and both a large break and a small break in a BWR/6. Recent work has focused on inclusion of the results of these test cases into the draft code manual which was published in June 1986.

l The work planned for FY-1987 includes the extension of the Courant limit violating numerics to the three dimensional VESSEL component, further modification of the FORTRAN coding for improved portability and support for users of the TRAC-BWR codes.

2-2

RELAP5/ MOD 2 DEVELOPMENT

• C. S. Miller Idaho National Engineering Laboratory EG&G Idaho, Inc.

RELAP5/M002 is a pressurized water reactor (PWR) system transient analysis computer code developed for the U.S. Nuclear Regulatory Commission (USNRC) Safety Research and Regulatory Programs. MOD 2 is the latest in the RELAPS series, having been officially released in April 1984. Since that time, development has focused on refinements designed to increase code speed, usability, and reliability. ..

The International Code Assessment Program (ICAP), sponsored by NRC ar.d member countries, has undertaken a rigorous plan of assessment of current light water reactor safety codes to span some three years. The plan calls for the usage of " frozen" code versions during this period. This strategy ensures that each member utilizes the same code version. Moreover, the preclusion of code improvements during the assessment period (i.e., only errors may be corrected) provides a uniform basis for drawing conclusions on code capability. RELAP5/M002 Cycle 36, released in 1985, was designated as the frozen version of RELAP5. Cycles 36.01 through 36.05, reflecting error corrections and user convenience changes only, have been transmitted to all participants.

With the basic development of RELAP5/M002 complete, emphasis has shifted toward maintenance and user support. The RELAPS newsletter service provides a mechanism for serving the many domestic organizations using RELAPS. This service, supported by the users themselves, utilizes a menu-based electronic newsletter stored on an IBM PC with an auto-answer modem at the Idaho National Engineering Laboratory. By accessing the newsletter through their own local terminal, users are able to obtain code updates and up-to-date information on development and application activities. Each user may also contribute to the newsletter concerning their usage and experience. A quarterly report summarizing all reported user problems, resolutions and other code modifications is also sent to newsletter service members.

A major development task completed in 1986 was the PWR self-initialization option using specially constructed controllers in conjuction with the existing steady-state and nearly-implicit solution scheme options. Provisions are included for pump, steam and feed controllers. The controllers are flexible in that, like standard RELAPS controllers, they can take input signals from a wide variety of sources and have outputs that can be used by many components. They are specific to self-initialization in that they have default controller time constants found to be suitable for a range of cases but not necessarily optimum for every situation.

• Work supported by the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research, Under Department of Energy Contract No.

DE-AC07-761001570.

2-3

The self-initialization input requires the type of control, the controller device number, signal sources and the final destination of the controller ensure thatsignal,,. the c~ot h,This information rollers are is checked against correctly characterized. the regular The method is input to capable of controEdng individual components from either individual or multiple signals or controlling multiple components from a single signal or from an array of signals. The existing steady-state scheme is used to determine when the initialization is completed.

The documentation for RELAPS/M002 was upgraded and brought up to date

- to be cons stent with Cycle 36.05.

i Volumes 1 and 2 of the users manual were expanded to include current code features and to correct errors detected by users.

Plans for FY-87 call for continued maintenance and user support for RELAP5. In addition, an improved version of RELAPS to augment and/or supplant the frozen version of RELAPS will be released. It will embody the improvements coming from the comments, suggestions and corrections of the combined ICAP, domestic, INEL and newsletter service users. Improvements will be in the areas of interfacial drag, noncondensible gases, flow regime i maps and CCFL. .Also a quality assurance document for RELAP5 will be assembled delineating the technical basis and limitations of the physical correlations used in the code.

b i

1 i

r I

2-4

TRAC CODE DEVELOPMENT STATUS AND PLANS

• by Jay W Spore. Dennis R Liles. Ralph A Nelson. Paul J Dotson Victor Martinez. Richard P. Jenks. Michael W. Cappiello, James F. Dearing. Robert G. Steinke. and Paul T. Giguere Safcty Code Development Group Energy Division Los Alamos National Laboratory Los Alamos. New Mexico 87545 TRAC CURRENT STATUS Version 12.1 of the TRAC PF1/ MODI code was released as the frozen version of MODI in January of 1985 for the purposes of independent code assessment. The principal features of this code are.

a. Variable dimensional fluid dynamics model that can address 3-D flow in the vessel component, while both the primary and secondary loop components are treated as 1-D flow components. " should be noted that a user can specify a 1 D vessel component, which will result in reduced computer costs.

b. Nonhomogeneous nonequilibrium full two-fluid six equation hydrodynamics model that describes the steam-water flow. A horizontal stratified flow model has been added to the one dimensiona! hydro.

dynamics. A seventh field equation (mass balance) that describes a noncondensable gas field and an eighth field equation that tracks the solutes in the liquid phase have also been added to the TRAC hydrodynamics model.

c. Flow regime dependent constitutive equation packagc that describes both the transfer of mass energy, and momentum between the steam / water phases and the interaction of these phases with the heat flow from the system structures.

d. Flow regime dependent wall to fluid heat transfer correlations that are obtained from a generalized boiling curve based on socal conditions

c. Two-dimensional fuel rod conduction model that includes a dynamic fine-mesh rezoning capability that can resolve both bottom flood and falling-film quench fronts.

~

f. Consistent analysis of entire accident sequences including initial conditions, blowdown, refill. and reflood phases of a loss-of coolant accident. In addition. TRAC can be used to simulate a complete spectrum of break sizes as well as operational transients

g. TRAC has component and functional modularity which allows the user to model virtually any pressurized water reactor design or experimental configuration. TRAC has component models for accumulators, breaks. fills, cores. pipes, pressurizers, pumps, steam generators. turbines. valves, and vessel with associated internals

h. TRAC trip and control models gives the user the flexibility to model virtually any PWR control system

~\ or protection system or any experimental control system.

Since version 121. eight additional error-correction / user convenience update sets have been generated.

These correction sets contain modifications to the TRAC code that can be grouped into three categories. The first group encompasses error corrections to logic and modcls infrequently activated by the code or users: the second group addresses user convenience changes that deal with input or output, the third group are new model options that require user input to activate.

Of the updates in the first group. a major error correction improves the interfacial condensation model by limitingits rate of change to those observed experimentally This replaces the old method of logarithmic-averaging the old and the new time values which yields time step. size-dependent results A major user convenience added to TRAC is the multiple component connections to a single cellin a vessel component This allows user to reduce the vessel noding for transients in which multi-dimensional flow in the vessel is not signihcant. thereby saving computer computation costs Another significant user convenience is the self initialization capability added to

• Work performed under the auspices of the US Nuclear Regulatory Commission.

2-5

l TRAC. This capability allows the user to input uscr desired initial conditions (ie.. cold leg fluid temperature.

loop mass flow rate. pressurizer pressure. reactor power. etc.). Built in controls will force the TRAC steady state to thosc user-specified conditions.

A major inodel improvement added to TRAC that must be activated by input is the counter current flow limiting (CCFL) model The user is given the option of specifying a flooding curve for any location within the vessci component. for which he/she has experimental data Since the effective interfacial shear for flooding or CCFL is strongly dependent upon geometry. TRAC will use this user input flooding curve to infer an effective interf acial shear for those locations where the user has flooding data. If no flooding data are available. then the i user can use the TRAC default interfacial shcar packagt which will predict a typical flooding curve for a straight l pip e.

PLANNED CODE IMPROVEMENTS Currently Los Alamos is following two ccde development paths. Both paths are essentially the same. except l

one has the 3 D stability enhancing two-step (SETS) method and some partial vectorization of TRAC while the other does not. A decision will made late this year or early next year as to which path will yield the MOD 2 version of TRAC-PF1. For both paths, a new separator component is being developed. This new separator component will allow the user to input experimentally determined carryover and carryunder curves as a function of any TRAC signal variable or control block For both paths. a new core void fraction distribution model is being developed it is intended that this new core void distribution model will accurately simulate a wide variety of transient conditions in the reactor core from reflood to boiloff transients. Since the core voids and heat transfer models are very coupled. a new core void fraction model implies improvements in the TRAC post critical heat flux heat transfer package.

A generalized heat structure capability has been developed for both paths: however. testing has not been completed on this model This capability will allow the user to model arbitrary conduction paths between any two fluid cells as wcll as allowing any number of structures to exist in a fluid cell. Heat-structure geometry is I

two dimensional and can be specified as either Cartesian or cylindrical. Axial conduction and fine-mesh rezoning capability can be utilized in any of the generalized heat structures. Generalized heat structures will allow the user more flexibility in mode ling steam generators and willimprove the modeling of conduction pathways in the vessel component The 3 D SETS method has already undergone some developmental testing. One of the tests was a compar-ison between the current TRAC PF1/ MOD 1 code and the 3-D SETS code for a steam generator tube rupture analysis of the H. B Robinson plant. The calculated results were the same as expected The 3-D SETS code ran the steady state and most of the transient at ~1/4 the CPU time required by TRAC-PF1/ MODI. Before g

' pump trip. the 3-D SETS code was running the transient ~2 CPUs to 1 transient second and it should be noted i that this was a very detailed model of both the primary and secondary systems containing 106 components and I 621 hydraulic cells with a complete 3 D representation of the vessel.

I 2-6

l NUCLEAR PLANT ANALYZER DEVELOPMENT AT THE IDAHO NATIONAL ENGINEERING LABORATORY E. T. Laats Idaho National Engineering Laboratory EG&G Idaho, Inc.

P. 0. Box 1625 Idaho Falls, Idaho 83415 The Nuclear Plant Analyzer (NPA) is a state-of-the-art safety analysis and engineering tool being used to address key nuclear power plant safety issues. Under the sponsorship of the U.S. Nuclear Regulatory Commission (NRC), the NPA has been developed to integrate the NRC's computerized reactor behavior simulation codes such as RELAP5, TRAC-BWR and TRAC-PWR, with well-developed computer color graphics programs and large repositories of reactor design and experimental datal ,2,3 An important feature of the NPA is the capability to allow an analyst to redirect a RELAPS or TRAC calculation as it progresses through its simulated scenario. The analyst can have the same power plant control capabilities as the operator of an actual plant. The NPA resides on the dual Control I Data Corporation Cyber 176 mainframe computers at the Idaho National i '

Engineering Laboratory and Cray-1S computers at the Los Alamos National Laboratory (LANL) and Kirtland Air Force Weapons Laboratory (KAFWL).

During the past year, the NPA program at the INEL has addressed two primary areas: software development and user support.

The primary emphasis in software development has been the implementation of the NPA on the Cray-1S computers, first at the KAFWL, which was completed in early January 1986, and then at the LANL. Thc KAFWL implementation utilized the RELAP5/ MOD 2 code, and the LANL implementation utilized the TRAC-PF1/M001 code to provide the reactor simulation capabilities.

A second software development activity was the addition of an X-Y plotting capability. The user of the NPA system on the INEL computers has the option to pause an interactive simulation being viewed on the color graphics display, switch to the X-Y plotting function and review selected data channels, and then return to the color graphics display and resume the calculation.

The user support area included assisting various users (NRC and contractor) in utilizing the NPA in various applications. A major activity was the modification of RELAP5/ MOD 2 models of the H. B. Robinson (HBR) and Arkansas Nuclear One-Unit 2 (ANO-2) plants, to enable

• Work supported by the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Regulatory Research, Under Department of Energy Contract No. DE-AC07-76IIIID01570.

2-7

interactive execution and control through the NPA. Also, a second set of RELAP5/M002 models of the HBR and Oconee plants were developed significantly reducing the number of computational cells in each model. '

'. Real wall-clock computational times were realized for selected of operational transients where the primary loop remained liquid-full. The fidelity of the simulation was apparently not compromised when the smaller 1

sized models were used for this type of transients.

Two related activities at the INEL are noteworthy. First, a separate NPA program was begun in September that is co-sponsored between the NRC and the Netherlands. A training program has started to familiarize a Dutch representative assigned to the INEL with the operation of the NPA l with RELAP5/M002. Second, using Department of Energy program development I

f funds, a minicomputer version of the RELAP5/ MOD 2 is being l

" proof-of-concept" tested on a MASSCOMP 5700 computer. Thisprojecgisa first big step toward developing a stand-alone " Type-3" workstation The upcoming NPA program at the INEL will focus on production operation. The major emphasis will continue to be modifying existing input decks to enable their interactive execution through the NPA. The NPA software will be maintained on the INEL and LANL computer systems.

Efforts toward closely integrating the Nuclear plant Data Bank with the NPA will receive increased emphasis. Finally, the Dutch /USNRC co-sponsored program will continue.

ACKNOWLEDGEMENTS 1

The NPA program at the INEL was highly successful during 1986 due to the diligent efforts of several people. They are (in alphabetical order) t R. J. Beelman, J. D. Burtt, T. R. Charlton, J. N. Curtis, C. D. Fletcher,

! W. H. Grush, R. N. Hagen, N. L. Hampton, H. A. Hardy, M. A. Lintner, F. S. Miyasaki, K. D. Russell, D. H. Schwieder, H. D. Stewart, and 1

J. E. Tolli.

NOTICE This paper was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, or any of their employees, makes any warranty, expressed or implied, or assumes any legal liability or responsibility for any third party's use, or the results of such use, or any information, apparatus, product or process disclosed in this report, or represents that its use by such third party would not infringe privately owned rights.

' The views expressed in this paper are not necessarily those of the U. S.

Nuclear Regulatory Commission.

i REFERENCES

1. K. D. Russell, et. at., Nuclear plant Analyzer and Data Bank Commnn ,

User Interface Functional Requirements, Conceptual Design, and i Hardware considerations, EGG-SAAM-6419, September 1983.

2. H. D. Stewart, et. al., NECTAR (NpA) Program and Reference Manual, EGG-CMD-6825, May 1985.

l i

I 2-8 l

. ~ . .. _-_-_.- , _ - _ - - . . -

3. E. T. Laats and R. J. Beelman, "U.S. Nuclear Regulatory Commission's Nuclear Plant Analyzer," Proceedings of the International ANS/ ENS Topical Meeting on Thermal Reactor Safety, San Diego, USA, February 2-6, 1986.

4. E. T. Laats and R. N. Hagen, " Nuclear Power Plant Simulation using Advanced Simulation Codes Through a State-of-the-Art Workstation,"

Proceedings of the 1985 Summer Computer Simulation Conference, Chicago, USA, July 22-24, 1985.

2-9

BWR PLANT ANALYZER AT BNL W. Wulff, H. S. Cheng and A. N. Mallen Department of Nuclear Energy Brookhaven National Laboratory Upton, New York 11973 An engineering plant analyzer has been developed at BNL [1] for realisti-cally and accurately simulating transients and severe abnormal events in BWR power plants. Simulations are being carried out routinely with high fidelity, i

' high simulation speed, at low cost and with unsurpassed user convenience. The BNL Plant Analyzer is the only operating facility which (a) simulates more than two orders-of-magnitude faster than the CDC-7600 mainframe computer, (b) is accessible and fully operational in on-line interactive mode from anywhere in the U.S. or Europe, via widely available IBM-PC, standard modem and tele-phone lines, (c) simulates both slow and rapid transients seven times faster than real-time speed in direct access, and four times faster in remote access modes, (d) achieves high simulation speed without compromising fidelity, and (e) is available to remote access users at the low cost of $160 per hour.

L Detailed and accurate simulations are achieved in a minicomputer at high simulation speeds, because five distinct modeling principles [2] and the special-purpose parallel processor AD10 for time-critical simulations of com-plex systems [3] have been used. Modeling formulation, simulation language and computer architecture have been optimized as an integrated system to achieve cost-effective BWR power plant simulations in the shortest time possible.

The analyst using the BNL Plant Analyzer controls the simulation as the plant operator controls the plant and has two arbitrarily selected parameters displayed on-line. Over 150 additional parameters are being stored for con-venient graphical or tabulated display af ter the transient. The history of operator actions introduced during the transient is cataloged. More than 10 transients can be analyzed in detail and documented in a single day.

During the past year, the containment simulation has been completed.

Detailed models have been implemented to describe the suppression pool dynam-ics, the pressures, temperatures and vapor mass concentrations in drywell and wetwell atmospheres, the liner wall temperatures, the condensation of vapor at the walls and in the drywell coolers, as well as in the atmospheres of dry and wet wells and the evaporation and condensation at the pool surface.

Simulated are the High-Pressure Core Injection and Reactor Core Isolation Cooling Sy stems , the low-Pressure Injection System, the Control Rod Drive Hydraulics System, the Condensate Storage Tank , the Residual Heat Rejection System and small breaks in steam line, recirculation loop and feedwater loop inside the containnent, as well as the intercompartment flows and leakages.

• Work perf6rmed under the auspices of the U.S. Nuclear Regulatory Commission.

2-11

1 The BNL Plant Analyzer has been employed for NRC/NRR-sponsored analyses of the effects from Tech. Spec. changes, fuel burnup and boron stratification.

The Plant Analyzer was programmed to display the minimum critical power ratio for this program.

! The BNL Plant Analyzer was also employed to simulate a four-hour-long station blackout transient at the Fermi-2 power plant. Plant parameters were

computed by the plant analyzer and transmitted by voice phone to the emergency

[ center.

i The BNL Plant Analyzer was called upon by the NRC to simulate the transi-i ent following the feedwater pump speed excursion at the LaSalle-2 power

plant. A program change had to be implemented, yet the BNL Plant Analyzer j succeeded in simulating the transient and four variations on the postulated scenario on the same day. Results were transmitted to the NRC by facsimile the next morning.

BNL has continued to provide remote access to six off-site users of the

! BNL Plant Analyzer. LILCO has completed the plant-specific input file for the Shoreham power plant and uses the Plant Analyzer routinely. NYPA is preparing to use the Plant Analyzer for the FitzPatrick power plant. Niagara Mohawk is

, preparing to modify the Plant Analyzer for Nine Mile Point-1, a plant without

jet pumps.

BNL continues to promote the development of PWR plant simulation, using again the five newly developed modeling principles and the newest special-purpose peripheral processor, the AD100. The intent is to form a consortium j of interested parties who support this development. Proposals have been sub-mitted to the Empire State Electrical Energy Research Corporation, to

Taipower, Taiwan, and to the Consejo de Seguridad Nuclear of Spain. Another proposal has been submitted jointly with Grumman Corporation to develop a PWR plant analyzer as part of a control and instrumentation test facility.

l The simulation technology developed at BNL is being recognized as a i viable alternative to widely used but expensive and time-consuming methods which are ' based on standard finite dif ference techniques, standard FORTRAN l programming and program execution on generai-purpose mainframe or supercomput-

ers. The advantages of the Plant Analyzer arise from the integrated optimiza-

tion of modeling techniques, numerical integration methods, programming lan-I guage and special-purpose minicomputer architecture, all aimed at the single purpose of cost-ef fective simulation with high fidelity.

References i 1. NUREG/CR-3943 (1984).

i

2. W. Wulff et al., " Plant Analyzer Development for High-Speed Interactive l

Simulation of BWR Plant Transients," 2nd Int. Topical Meeting on Nuclear

! Plant inermal Hydraulics and Operations, Tokyo, Japan (1986). ,

, 3. E. O. Gilbert and R. M. Howe, " Design Considerations in a Multiprocessor l Computer for Continuous System Simulation," Proc. AFIPS Conf. , Vol . 4 7 l

(1973), AFIPS Press, Montvale, NJ 07645, 2-12

THE NUCLEAR PLANT DATA BANK

• by Clay P. Booker, Michael R. Turner, and Jay W. Spore Safety Code Development Group Energy Division Los Alamos National Laboratory Los Alamos, New Mexico 87545 i

The Nuclear Plant Data Bank (NPDB)is being developed at the Los Alamos National Laboratory to assist

! analysts in the rapid and accurate creation ofinput decks for reactor transient analysis codes. We envision that l this attribute will reduce the time and cost of the creation of a typical input deck Further, we expect that the NPDB will be an invaluable toolin the timely investigation of recent or ongoing nuclear reactor accidents and other transients.

The NPDB will give the analyst the interactive capability to build. store, retrieve, edit, and display nuclear l l power plant models from a quality assured data base of plant data. With the NPDB, the analyst will be able to ,

generate interactively a complete Transient Reactor Analysis Code (TRAC) input model of an operating nuclear '

plant or to renode an existing TRAC model in a significantly reduced time. The NPDB development is divided into three phases.

The NPDB will be available in the summer of 1987. That version is specified to produce a complete TRAC PF1/ MOD 1 input file for a Westinghouse nuclear power station with basic trips and controls and a rudimentary model of the secondary side. In phase 1. the fundamental modules for the complete NPDB are designed The relational data base (RDB). which handles both global plant characteristics and specific plant features, provides the text menus for the user to select a specific plant and, for each plant. provides the key to the general data in addition to plant data, the RDB manager administers the user.model file library where the user has plant nodings in various stages of completion. The indexed data base which keeps plant data in a data dictionary (index), stores all the available data for each plant. For portability, the NPDB is restricted to ANSI FORTRAN TT coding and the ASCll character set as much as possible: any machine dependent coding is isolated and clearly marked as such To promote transportability. the graphics section uses the ANSI Graphical Kernel System (GKS) module. at any facility where the GKS system is available. the NPDB GKS module may l be replaced by the system resident GKS. Where the GKS is unavailable the NPDB GKS module may be left I

in place, and the user may supply his own terminal drivers or may modify the NPDB Tektronix 41xx terminal driver module The modeling module uses geometric information from the data bases to produce a display of any subsystem with sufficient detail for the user to node the plant accurately. With this module in control of the terminal. the user specifies noding choices: the selections and the relevant plant data are recorded in a model file that may be stored and modified later. The input-file generator module retrieves allinformation from the main module through the model file; it does not access the data bases. lhus. a user will be able to access his own model file or a model file from another user, modify it. and produce an input file with a minimum of intensive data base processing by the NPDB modeling module.

During NPDB phase 1. we have completed the implementation of the GKS routines: the development of the terminal driver routines. menus. data base managers. and the TR ACIN generator for most of the one dimensional components, and the design of the data bases (both the RDB and the indexed data base). The major tasks to be completed are the TRACIN generators for the steam generator and the vessel.

During phase 11, which will begin immediately af ter completion of phase 1. we will add teveral useful features.

One will be the capability to treat all commercial US pressurized water reactors (PWRs) Another feature will incorporate a more sophisticated inp and control system interactive design with a complete secondary side model.

In phase lit, we will extend the NPDB to boiling water reactors (BWRs).

This work was funded by the US Nuclear Regulatory Commission (NRC), Othee of Nuclear Regulatory Research. Division of Accident baluation I

2-t3  !

EFFECTS OF SEISMIC LOADS ON THE OPERABILITY AND LEAK INTEGRITY OF CONTAINMENT ISOLATION VALVES R. Steele, Jr.

R. C. Hill Idaho National Engineering Laboratory l

This paper presents the results of seismic loading experiments in the ,

Containment Isolation System (CIS) Valve Integrity Program being performed by l the Idaho National Engineering Laboratory. The program is sponsored by the '

United States Nuclear Regulatory Commission, Office of Nuclear Regulatory Research.

i The purpose of this program is to provide a technical basis for the support and development of equipment qualification procedures in the area of design i base dynamic loads. Specifically, the test program addresses the operability and leak integrity of representative CIS valves when subjected to typical l operating basis earthquake (0BE) and safe shutdown earthquake (SSE) loads.

By testing full size, complete systems (valves, containment penetration, piping, and supports) the uncertainties concerning modeling, interfaces, and other analytical assumptions were avoided. The two main criteria used to identify the systems to be tested in the program were: 1) Systems with a relatively high potential for leaking the containment environment to the outside atmosphere; and 2) systems that would be required for mitigating an l advanced severe accident. Three systems were tested: An eight-inch l

butterfly valve system modeling a containment purge and vent system; an

{ eight-inch gate valve system modeling a containment spray system; and a j two-inch globe valve system modeling the numerous small bore piping systems t

that penetrate containment.

Each piping system was configured to be typical of a wide range of piping designs so that results are applicable to the highest percentage of plants.

Numerous CIS piping configurations were reviewed to determine typical piping lengths from containment wall to valves, lengths to first elbow, direction of first bend, and type and location of supports.

The test apparatus consisted of a test frame, measuring 23' x 13' x 8' and constructed from 14" square tubing, mounted on eight pressurized air bags.

Each piping system was individually installed in the test frame using nuclear grade supports including rigid struts, spring hangers, and box beam l supports. The total test system with installed piping weighs approximately 23,000 pounds. Simultaneous independent triaxial motion was input into the frame with the use of large hydraulic actuators mounted to the frame in three orthogonal directions, piping systems were pressurized with air, nitrogen, or water (depending on system) for evaluation of valve operability and leakage.

Each piping system was subjected to a series of dynamic loads which included:

(1) Low level dynamic loadings for instrumentation and operational check-out; (2) a simulated operating basis earthquake; and finally (3) a simulated safe shutdown earthquake. Load levels were developed from a review of other NRC sponsored projects and information obtained in Final Safety Analysis Reports.

3-1

4 l Leakage was measured after each dynamic excitation and during the OBE; operability (stroke time and valve operator current) was recorded after each i dynamic excitation and during the SSE. There were essentially no detrimental effects on leakage or operability due to loading at typical OBE & SSE

, levels. Stroke times and operator currents ' vere not affected by the dynamic loads; during the OBE excitation on the eight-inch gate valve leakage

increased, however, it returned to zero after the excitation stopped.

The program has emphasized the typicality of valves, operators, piping, l penetrations, supports, configurations, and dynamic loads. No unanticipated system effects were observed, thereby providing credibility to the single effects testing techniques typically followed in valve qualification. The conclusion from this research is that typical CIS penetrations and valves are 1 not threatened by typical SSE loads.

i i

1 1

-l l

i i

1 l

3-2 l

l PARAMETERS IMPORTANT TO REACTOR COOLANT l PUMP SEAL STABILITY DURING STATION BLACK 0UT l

R. C. Hill Idaho National Engineering Laboratory C. A. Kittmer

! Atomic Energy of Canada, Ltd.

l For the past three years, the Nuclear Regulatory Commission has sponsored research, directed by the Idaho National Engineering Laboratory, into the behavior of reactor coolant pump shaft seals in predicted station blackout conditions. The objective of this research has been to determine if typical shaft seals can function adequately to prevent serious loss of primary coolant, and subsequent core uncovery, if all AC power is lost.

Although there are design variations among the several pump and seal vendors, most seal assemblies consist of three or four similar stages.

Each stage is made up of a rotating ring attached to the pump shaft and a stationary ring attached to the pump housing. The rings of each stage are maintained in close proximity to each other by a balance of forces, the most significant of which are the hydrostatic loads generated by system pressure and the pressure distribution radially across each stage of seal rings. The equilibrium gap between the rings establishes a small controlled leakage out of the primary system. One of the rings can accommodate limited axial motion, either hydrostatically or thermally generated, by sliding over a secondary elastomer seal that prevents leakage around the seals.

Normally, the seals operate at primary system pressure and at a maximum temperature of about 1500F, If all AC power were lost, cooling to the water passing through the seal assembly would also be lost and the seals would be exposed to primary coolant conditions. The subject research program has been performed to investigate seal behavior under these conditions with emphasis on the behavior with two phase flow between the seal rings. As the hot water flashes into the lower pressure region of the flow path, the pressure distribu? ion radially across the seal rings will change'from the normal, single phase distribution. The distribution may change enough that no equilibrium gap can be maintained between the seal rings, causing the seal rings to separate to the limits of axial travel. The large gap and resulting high leakage could lead to core uncovery if the blackout continued for more than a few hours.

An analysis of a model seal identified three parameters with the most influence on stability - inlet coolant conditions, gap convergence (the slight taper or out-of-parallel relationship between the two seal rings),

and the seal backpressure. In general, inlet conditions approaching saturation, larger convergence (more taper), and a lower backpressure all contribute to unstable operation.

3-3

Testing on a small scale model was performed at representative inlet conditions and cor. vergence; the backpressure was atmospheric pressure.

The test results were consistent with analytical predictions.

There are variables to be considered in addition to the seal itself including downstream flow restrictions, seal ring surface conditions, and staging f ows that 2.rc .;:d in scme .31 assembly J.:igns to control the pressure at each stage. The plant specific configuration and system operation in a postulated station blackout must be considered in determining seal stability. However, based on the subject research it is  !

concluded that stability margins would be small during a station blackout and there are credible scenarios in which seals could become unstable with subsequent high leakage of primary coolant.

3-4

l DYNAMIC LOAD EFFECTS ON GATE VALVE OPERABILITY:

RESULTS OF THE HDR-VKL TEST PROGRAM R. Steele, Jr.

Idaho National Engineering Laboratory The United States Nuclear Regulatory Commission (NRC) and Kernforschungszentrum Karlsruhe (KfK) have undertaken a joint seismic research program performed at the Heissdempfreaktor (HDR) decommissioned experimental reactor facility in the Federal Republic of Germany. The program provides a unique opportunity to study the response of an installed piping system to dynamic excitation from the building. As part of the larger seismic research program, the Idaho National Engineering Laboratory developed a task to obtain in-situ experimental data to aid in assessing the adequacy of existing qualification standards with regard to load combinations and application of nonnuclear experience to the qualification of equipment used in nuclear power plants.

An existing piping system, the Versuchskreislauf (VKL), was modified to include an eight-inch gate valve and five different pipe support configurations. The five support systems (two flexible systems of German design; one typical of existing United States nuclear installation; and two experimental energy absorbing designs) were installed individually on the same piping system.

The test program consisted of mounting a large mechanical coastdown shaker in the HDR reactor containment building. This shaker transmitted mechanical motion into the building at various initial frequencies, ranging from 1.6 to 8.0 hz (before coastdown) with a maximum eccentric force of 10,600 KN (2.4 x 10 lbf). This motion, in turn, excited piping and components internal to the building.

The VKL system, selected supports and the gate valve were instrumented to monitor the response to flow related and seismic loads while the system was subjected to more than 20 individual simulated seismic excitations.

The complex structural filtering of the shaker input resulted in piping excitation very similar to that which would be expected from earthquake induced excitation. Hydrodynamic loads were applied to the valve with flows in the VKL up to 400 gpm and differential pressures, across a nearly closed valve, of about 350 psi.

Overall seismic excitations were typical of the SSE response spectrum for a Standard huclear Unit Power Plant System (SNUPPS) with 3% damping.

Hydrodynamic loads, while not equal to the maximum that would be experienced by typical safety related gate valves, were'large enough to permit observation of effects on the valve operator power demands.

3-5

The slow related loads resulted in repeatable demands on the valve operator j and at no time was the operator capacity challenged; nor, on the basis of projected loads, would it have been expected to be unable to cycle the valve at design conditions.

There were no inherent differences in valve operating characteristics when l dynamically excited in a flexible rather than rigid piping system. There i were no observed responses that would invalidate consideration of nonnuclear piping system experience in the qualification of valves installed in nuclear j power plants.

In general, the valve behavior was consistent from one test to the next -

I there were no apparent system effects that would question the viability of

] the separate effects testing applied in current qualification practices. j 1

s i

l i

l l

I l

I i

1 1

2 i

! I

)

.I i i

)

3*6 i

u. - - - - . - - . . . - _ . _ - - _ - - _ . .

Progress on Qualification Testing Methodology Study of Electric Cables Y.Kusama, T.Yagi, S.Okada , M.Ito, M.Yoshikawa, K Yoshida, N. Tam ura, W.Ka wakami Japan Atomic Energy Research Institute (JAERI), Takasaki Radiation Chemistry Research Estabitshment Safety-related electrical cables (class 13) in nuclear power plants are suppose i to be expose:I to radiation, high temperature steam, spray, and in certain cases air simultaneously when a LOCA (loss of coolant accident) occurs.

These cables are required to ba functional even Lf they are subjected to a LOCA at the end of their intended service life. To confirm the integrity of the cables under LOCA, qualifiestion tests have been carried out under simulated environments which were specified by individual countries according to their types of reactors.

In Japan, cable makers qualify their cables by the sequential method based on the recommendation issued by IEEJ in 1982. In the 13th meeting, we have reported the results about qualifLcatton testing methodology in which degradation of Ilypalons and CPRs under four kinds of combined environments including an extreme case can be well simulated by selecting sultable sequential conditions. Present paper deals with the results about the effects of environmental conditions in the sequential methods on the degradation of mechanical and electrical properties and also on the water sorption of the insulating materials during LOCA.

Sequential test procedure is divided into two parts, that is irradiation (I) and steam / spray exposure (II). In order to find adequate conditions of the sequential method which can bring about the degradation to the same degree as each simultaneous environment, the irradiation ani the steam / spray exposure conditions were systematically changed and the effects on the degradation were investigated.

Five kinds of insulating and jacketing materials such as ethylene propylene rubber (EPR), crosslinked polyethylene (XLPE), silicone rubber (S L R),

chlorosulfonated polyethylene (llypalon), and chloroprene(CR) were tested under various conditions of the sequential test.

I. Irradiation a) Effect of dose Samples were irradiated with gamma rays up to 2 MGy with a dose rate from 0.45 to 10 kGy/h in air at room temperature. Irradiation under elevated temperature f rom 50 C to 90 C and under pressurized oxygen (0.5MPa) at room temperature were examined.

Mechanical properties of the materials, especially elongation at break, were markedly decreased with increasing dose. Decrease in volume resistivity of EPR, XLPE and StR were within two orders of magnitude by 1.5 MGy irradiation under oxidntive conditions such as at low dose rates, at elevated temperature or in oxygen.

b) Effect of dose rate The samples irradiated under various conditions described above were exposed to constant temperature steam (mostly 120 C)/ spray environments.

Irradiation at low dose rate caused greater degradation in nochanical properties of the materials except SLR and C't than at high dose rate when they were subjected to the following stean/ spray exposure. In the case of EPR, rato M

i i

i of degradation in the following steam exposure depended upon about -1st power of i dose rate. Decrease in volume resistivity was remarkable when the samples were irradiated at lower dose rate than in the case at high dose rate irradiation, j

c) Ef fect of irradiation temperature and atmosphere i Samples irradiated at elevated temperature (50-90 C) at 5 kGy/h and in i pressurized oxygen (0.5 MPa) at 4 kGy/h were exposed to steam / spray environment.

Similar degradation behavior in mechanical properties between high

] temperature irraitation at 70 C and low dose rate irradiation at 0.45 kGy/h were j observed except for XLPE. For XLPE, remarkable elongation recovery by steam  !

j exposure following the irradiation in air at roo n temperature was observed but I not in the case of the irradiation at elevated temperature. Both mechanical j and electrical properties were greatly degraded by irraitation under pressurized j oxygen except for sir.

In many cases, higher water sorption were observed for the samples irra-i diated at high temperature than irradiated at low dose rate at room temperature.

J

H. Steam Exposure '

! i) Effect of temperature in steam / spray exposure j Irra11ated samples were exposel to saturated steam and air-containing steam environments at 120 C, 140 C and 160 C.

j Both mechanical and electrical properties were degraded significantly with

! increasing temperature. Volume resistivity and water sorption of the sampics

! except sir were greatly af fected in air-contalning steam as compared with in 1 saturated steam, i

j 11) Ef fect of air over pressure in steam Irradiated samples were exposed to steam / spray environment where air added to suturated steam by 0.05, 0.13, 0.25 and 0.5 MPa.

4 Slight changes in mechanical properties of the materials except for XLPE were caused by air up to 0.25 MPa, but significant damage were observed at 0.5

! MPa.

l On the other hand, volame resistivity of some EPR's and XLPE decreased when l

the ad litional air partial pressure was 0.05 MPo, although the additional air

did af fect the properties.

I 1

tit) Effect of spray in steam exposure Irradiated samples were exposed to steam environments with and without spray in which air over pressure was 0.05 MPa at 120 C.

4 Most samples irradiated at high dose rate showel larger damage in

! mechanical and electrical properties when exposed to steam environments with j spr ty. For certain materials, larder damage were observed when the samples l were irrallated at 70 C an-1 then exposed to steam environments without spray.

j Similar tendency was foun I la water sorption of the 9amples. These phenomena j were assumei to depend upon chemical change in the materials such as i accumulation of the oxidized substances which were produced durIng irradiatlon.

i 3

l From thene experimental facts, it can be concluded that mechanicn1

. propertlos were maialy affected by dose, dose rate and steam temperature in the j sequential method. Electrical properties were nainly af fected by oxidation causet by irradiation at low lose rate, under pressartzel oxygen and at elevated temperature followed by exposure to high temperature steam with alr.

3-8

General Technical Requirements for Special Valves Used in Nuclear Power Plants i J. Zdarck F. Jukl I

Sigma Research Institute, Prague Czechoslovakia With the aim to reach a high standard and to unify the quality of supplied components the " General Requirements for Special Valves, Used in Nuclear Power Plants" /OTT-82/ had been worked out.

Within the scope of OTT-82 fall all types and sizes of special valves, used in pipelines and systems of the nucler power plants /NPP/ with PWR and RBMK / boiling water cooled, graphite moderated, channel type /

reactors.

OTT-82 involve comprehensive design data and requirements for production, inspections, testing, packaging, transportation, storage of the valves and manufacturer guarantees.

3-9

T Effectiveness and Safety Aspects of Selected Decontamination Methods for LWRs S. W. Duce, F. B. Simpson EG&G Idaho Inc., Idaho National Engineering Laboratory In October of 1983 the Nuclear Regulatory Commission (NRC) Office of Nuclear Regulatory Research funded this program with the objective to obtain information on chemical decontamination processes that the NRC might be expected to review. Information was to focus on methods, effectiveness, and safety aspects of the chemical decontamination processes as they were applied in primary coolant recirculation systems (PCRSs) at commercial nuclear power reactors. This information was to be obtained through observation of the process application in light water reactors (LWRs).

During the years 1984 and 1985 in-situ chemical decontamination processes were applied extensively in boiling water reactor (BWR) PCRSs to help facilitate system replacement or inservice inspections, and more limited in pressurized water reactor (PWR) steam generators to facilitate tube plugging or sleeving operations. During these two years four chemical decontamination processes were observed: CAN-DECON, LOMI, DOW NS-1 and PNS CITROX A at a total of nine different facilities, seven BWRs and two PWRs.

In general there are two major areas of safety concern: system hardware and personnel, whether they be facility or member of the public. During l

the course of the study information was obtained by direct measurement, i

observation, and queries as to the effect on safety of the chemical decontamination processes. Concerning the first area of system hardware there were three main areas of focus: reduction in system integrity due to pipe wall thinning, chemical attack on other critical system components, and plant radwaste systems. Prior to application at the nuclear facilities all.of the chemical decontamination processes had been tested in laboratory experiments to determine if the process enhanced intergranular stress corrosion. None of the processes used showed any enhancement of intergranular stress corrosion in the initial testing programs. In addition, artifacts from the system to be decontaminated were sent to the vendor for testing to quantify the attack on the base metals by the chemical process. During the application of the chemical process, artifact coupons were installed in the vendor process system to measure the attack on the base metal for the actual application parameters. In all cases, regardless of chemical process, the attack on the piping system base metal was within acceptable limits.

Concern for chemical attack on other critical system components was taken seriously at all facilities. Several different methods were used to ensure the decontamination chemicals did not move into systems where their -

affect was unknown or not desirable. In PWR steam generators plugs were put into the hot and cold leg coolant lines to prevent chemicals from leaking into the reactor system. These plugs were pressure tested at pressures that would exceed the expected operational pressures. In BWRs the PCRS was physically cut loose from the reactor vessel in one instance to ensure the decontamination chemicals would not enter the reactor vessel. As the PCRS was to be removed following the decontamination this 3-11

was an acceptable option. However, most facilities used plugs and j solution level administrative controls to prevent the decontamination solutions from entering the reactor vessel. There were some cases where

! tha decontamination solutions were deliberately sent through tha annulus

! area in an effort to remove the oxide film that had built up in the i annulus. In general these methods were effective in protecting the

] critical components. In the few rare cases where small amounts of

' chemicals did get into the reactor vessel or piping the vendors were able to remove the chemicals in a minimum time, using their ion-exchange systems. There were few instances where systems leaks occurred. In those 4

few situations that did occur the chemical solutions were normally contained and cleaned up.before they could enter into the plants radwaste treatment system. In the one instance where the leaked solution did enter the radwaste treatment system the system was able to handle the decontamination chemicals.

l Personnel safety, be it facility personnel or member of the public, was j probably the most significant driving factor for performing a chemical

decontamination, albeit the greatest concern was for facility personnel.

! Chemical decontamination processes were selected because of the large dose j (i.e., man-rem) savings that could be realized by removing the highly radioactive oxide films that had built up on the internal pipe surfaces.

l Average curies of gamma emitting activity removed from all facilities were 51.4 Cl. Removal of this activity resulted in an average man-rem savings ,

! of 1482 man-rem for all facilities. The range of man-rem savings for BWRs  !

I was >450 to 2193 and for PWRs 1400 to 3660 man-rem. Several utilities I chose chemical decontamination methods solely for the dose reduction to I personnel and not because they felt that there would be dollar savings.

l Radioactive low level wastes generated from the chemical decontamination

! process were generally solidified ion-exchange resins. There was one

} instance where liquids were solidified and one instance where the resins

were simply dewatered. Ave j decontaminationwass150ftgagewastevolumesfromthechemical

, with the range being 80 to 1020 ft 3, In many instances the facilities were able to significantly reduce the i volume of radioactive wastes going to the burial grounds by taking advantage of the Quadrex Recycle Center in Oak Ridge, TN. The recycle i center would receive the stainless steel pipe, that had been removed from j the PCRSs, and; using other chemical and electro-chemical techniques, decontaminate the pipe so that it could be sold for scrap metal to the j public. Handling of the waste material in the manner used by the

utilities showed a concern for the public dose by minimizing the volume

! and lowering the exposure rate of the wastes going to the low level waste r burial grounds, which resulted in lower population dose now and should result in lower doses to future generations, t

3-12

PKL III Small Breaks and Transients Experimental Programme R. M. Mandl, B. Brand, II . Watzinger KRAFTWERK UNION AG During the last decade experimental research concentrated its ef forts on loss-of-coolant accidents (LOCA) caused by double-ended breaks of a pr' ary coolant pipe.

In Germany the analysis of PKL IIB and UPTF test results shows that the containment and consequences of double-ended breaks are largely understood.

Operational experience as well as risk studies show the probability of a small break or stuck open valve being larger by several orders of magnitude than a double-ended break of a primary coolant pipe. For this reason it was decided that the next series to be carried out in the PKL test facility should cover the area of small breaks and operational transients.

The planned experiments include among others base line tests with pumps on/off ( forced / natural circulation) shutdown with 1, 2 or 3 isolated steam generators and loss of off-site power breaks in the main steam line loss of feedwater small breaks in loops, pressurizer PORV stuck open rapture of steam generator tube (s)

The phenomena to be investigated differ, sometimes considerably, from the ones occuring during large break LOCAs. As the PKL test facility was originally designed mainly for large break LOCA experiments it was considered necessary to carry out a scaling study to identify those components to which changes muat be made.

4-1

4d--- - 6m6 -

4 s A- -~4-'A+A . w &- b A s-4-L +ec -M<-n4,'- t r-< L= =- A-The resulta cf t!.a scaling study show that most of the relevant two-phase phenomena occuring at 80 bar (heat transfer j in core and steam generators, onset of counter current flow l limitation etc.) can be successfully simulated at 40 bar (limiting pressure of PKL primary side). However, it was found necessary to make the following major changes to hardware:

II

- replacing the existing double loop by two identical single loops with independent steam generators I - introducing a downcomer top section " wrapped" around the

! upper plenum thus creating symmetrical layout of all four i loops and avoiding different elevations of hot and cold i legs i - installing active reactor coolant pumps in all four loops

- adding all necessary interfacing systems on both the primary and secondary.

1 The modifications to the test facility are to be completed by March 1987. The shake down period will be followed by j an experimental programme lasting 12 months. The project is 4

l Jcintly funded by KWU, BMFT (German Ministry of Research and 1

i Technology) and EVU (Consortium of German utilities).

i I

I r

4-2

f MIST Test Results H. R. Carter and J. R. Cloudemans The Babcock & Wilcox Company l

The Multi-Loop Integral Sys tem Test (MIST) is part of the Integral System Test (IST) Program being sponsored by the Nuclear hegula tory Commission, Electric Power Research Institute, Babcock & Wilcox (B&W) Owners Group, and B&W. The IST Program is being performed to obtain experimental integral system data for the B&W designed nuclear steam supply system (NSSS).

The da ta acquired from MIST will be used to benchmark system computer codes, such as RELAP 5 and TRAC, against simula ted small break loss of coolant l accident (SBLOCA) plant transients. The subject paper describes the design i

of MIST, the tes t program, and tes t results obtained.

MIST is a 2 x 4 arrangement (2 hot legs and 4 cold legs) of the B&W lowered loop NSSS. Scaling of MIST, in gene ral, followed in order of priori ty; eleva tion, SBLOCA phenomenon, component and piping volumes, and loop irrecoverable pressure losses. Full plant elevations are maintained at key interfaces such as the hot leg U-bend spillover, steam genera tor upper and lower tubesheets (secondary faces), cold leg low point, pump discharge, cold leg and hot leg nozzles, core (throughout), and emergency core cooling system (ECCS) injection loca tions.

The plant to MIST power scale factor is 817, which is set by the core and steam ge ne ra tor components. The core and s team gene ra tors are f ull-leng th subsec tions of the plant counterparts. The core is simula ted by 45 f ull-leng th 0.430 inch diame ter heater rods and 4 in-core guide tubas with plant typical fuel pin pitch. The simulated rods are capable of full scaled power ou tput but are limited to approxima tely 107. scaled power (330 kw) for the planned program. Two 19 tube once-through steam genera tors (OTSG's) are used in MIST. The OTSG's are approxima tely 52 fee t in length (between secondary faces of the upper and lower tubesheets) and are constructed with prototypical tubing and tube support plate arrangements.

Testing in MIST was performed in three major phases; debug, cha rac te ri za tion , and transients. The debug tests were performed to demons tra te the ope rabili ty of each HIST component. The cha ra c te ri za tion tests generally followed the debug tes ts to examine the behavior of the individual key systems and to explore limited integral system in te rac t ions .

The transient phase of MIST testing, and most extensive phase, includes 41 tests. These tests explore SBLOCA's, feed and bleed cooling opera tion, steam generator tube ruptures, the influence of non-condensable gases during a SBLOCA, and the ef feet of reactor coolant pump opera tion on system cooldown.

Test results presented in the s ubj ec t paper are from the tra ns i en t phase of testing, which started in June, 1986. A summary of the results from one te s t in the SBLOCA, feed and bleed, and s team gene ra tor tube rupture groupings are described.

4-3

i TRAC-PF1/ MOD 1 PRETEST PREDICTIONS OF MIST EXPERIMENTS

• by B. E. Boyack and J. L. Steiner Safety Code Development Group

! Energy Division Los Alamos National Laboratory Los Alamos, New Mexico 87545 i

l l Los Alamos National Laboratory is a participant in the integral System Tcst (IST) program initiated in l June 1983 for thc purpose of providing integral system test data on spuihc issues / phenomena relevant to post-small break loss-of coolant transients (SBLOCAs)in Babcock & Wilcox (B&W) plant designs. The Multi-Loop Integra! System Test (MIST) facility is the largest single component in the IST program MIST is a 2v4 [2 hot legs and steam generators (SGs). 4 cold legs (CLs) and reactor coolant pumps (RCPs)] representation of lowered loop reactor systems of the B&W design. The objectivc of the MIST tests is to generate high quality experimental data to be used fc assessing thermal-hydraulic <.afety computer codes.

Los Alamos is currently providing analytical support to the IST program; the largest part of our analytical efforts involve the use of TRAC-PF1/ MOD 1 There are three applications in which TRAC is. or will be. used as a complement to the MIST experimental program. The first app!ication is related to test specihcation or design.

This assumes that one has sufhcient conhdence that TRAC will correctly predict the dominant test phenomena and, therefore, that the predicted results can be used to ensure that facility limits will not be exceeded and that the resultant test will satisfy the test objectives. For exampic TRAC can be used to determine whether the design limits of the facihty or facility instrumentation could be exceeded during a test The second application is related to test evaluation. It is impossible to include all the desired instrumentation in a facility. Constraints of cost. complexity, space. etc. are rapidly reached. Again. if one has sufhcient conhdence that TRAC will correctly predict the dominant test phenomena, calculations can be used to fill in gaps about quantities that are not measured in the facihty The third application is related to TRAC assessment. The ability of a thermal hydraulic code to accurately calculate experimental behavior in scaled facilities is an important link in demonstrating that the code can be used to predict how an operating PWR would perform under accident conditions.

During Fiscal Year 1986 Los Alamos performed hve MIST pretest analyses The hve experiments were chosen on the basis of their potential either to approach the facility limits or to challenge the predictive capability of the TRAC codt Thc five tests are identihed and briefly discussed in the following paragraphs. The summary hndings of cach pretest analysis are also provided.

Test 310000 This was the MIST nominal test. The nominal conditions included a scaled 10-cm; CL-discharge leak. full high pressure injection (HPI) and auxiliary fcedv ater (AFW) unavailabic RCPs. available, no noncondensible gas automatic reactor vessel vent valve actuation on differential pressure, automatic guard heater (ontrol. constant SG level control after SG rehil. and symmetric SG cooldown Test 310503 This test focused on the effect of throttled AFW and asymmetric SG cooldown. The longer-duration SG rehil obtained by reduc.ing the SG refIl rate was intended to obtain AFW boiler condenser mode (BCM) cooling Following AFW BCM or SG rehlt the SGs were depr(ssurized at unequal rates. to generate asymmetric loop conditions. The purpose was to accentuate interloop asymmetries toinvestigate their long term impact.  !

Test 320604. The tests of group 32 were controlled in the same manner as the Nominal Test (310000) but with altered leak and HPI characteristics This test used the Evaluation Modcl (EM) HPl capacity, which is roughly one half that of full HPl Relative to the nominal case extensive system voiding and perhaps an early occurrence of the BCM were expected

• Work perfermed under the auspices of the US Nucicar Regulatory Commission 4-5

Test 330201 The tests of group 33 cxamined system interactions that occur while using feed and-bleed (HPl PORV) cooling Thest. tcsts simulated a totalloss of secondary side fecdwater. Tcst 330201 used the EM rather than thc full HPl hcad flow charactcristics. EM HPl flow is initiated upon fifting of the Power-Operated Relkf Valvc (PORV)

Test 330302 This test imposed dclaved HPl activation Thc PORV discharge was the only available energy removal mechanism for th( fnst 2i minutes following PORV lif t Thc full HPl is then providcd.

Although therc are many d(tailed ph(nom (na of int (rest in cach pr(test calculation. we will summarize only thc gcncral conchtsions of out analyses hcrc The initialloop bchavior was co.nmon to all hve of the calculations.

T his txhavior inc.luded thc inturuption of natural circulation in the intact loop. followed later by a sharp reduction in broken loop natural circulation flow and th( start of primary system rcpressuritation, followed by a period of spillover circulation in thc brok(n loop suthcient to tcrminate the repressurization. Aftcr this sequence of events, natural circulation was compktcly tuminated in both loops and thc cakulated syst(m response dcpended on th( control procedures specihed for the test Strong systcm interactions and coupling wuc observec in each of th( pr(t(st c alculations For exampic. during liquid iontinuous primary natural circulation, the primary loops exptri(nc(d sevcral long-pcriod flow oscillations T h( prcscnc( of the pressurizer in the intact loop was the trinyt that initiattd the diffoing thermal hydraulic bchaviors of the intact and brok(n loops The pr(test cakulations wcre pcrformed through thc blowdown phas( of the transient and each calculation was tcrminated n(at th( beginning of rchil whcn the HPl flow exceeded th( kak flow During blowdown the calculated r(sults genvally wuc not affcct(d by thc MIST facility scaling atypicalitics. Thc only atypicality that could hate afluted the tcsults is the excess mctal mass awl surfan area of the stcam genuator shcIls. This atypicahty would tend to extcod thc time r(quir(d to rchlt the stcam gencrator secondaric5 and increase the secondary pressurt attcr th( rehil As a result, thc timing of major cvents is shifted slightly: howevcr. the ovcrall tr(nds of the (akulations arc not affectcd Results of th( pr(test calculations indicated that the MIST facility safety limits would not be approached in the hvc t(sts considered fast, core uncovery will not occur during any of the tests cxamined T he lowest vesscl liquid Icvcis wcre prednted for T(st 330302. fced-and biccd cooling with HPl dclayed 20 minutes af ter PORV actuation Suond. PORV capacity is sufhcicnt to permit control of the primary system pressure throughout the te(d and bleed tcsts We cncountered onc situation in whic.h either the code or the model was predicting unexpccted physical b(havior We bclievc the problem is associated with the input model and not with code We concluded that the spht-channel SG model is d(hcient in modeling boiloff transients We bc!ievc that the impict on the calculated results is not major but that SG mod (I enhanccments for this type of transient should be studied in summary. w( bcliese that Los Alamos is f unctioning as a vital participant within the IST program The tcsults of our analytical efforts ar( b(ing used to support test specihtation and dcsign and to identify and improve om understandmg of ph(nomena that may occur in the MIST tests Wt look forward to using MIST data to assess the predntivc capabihty of IRAC and to assist us in identifying nceded arcas of code and modcling tt chnique unprov(ment i

4-6

UMCP 2x4 Loop Test Results Y.Y. Hsu, M. Dimarzo K. Almenas, F. Munno, D. Sallet M. Massoud, M. Popp Z.Y. Wang, J. Munno, H. Haper S.L. Shieh, K.Liaou University of Maryland Tests were performed in the following sequences of small break LOCA Rapid depressurization Natural circulation, single -phase Natural circulation, two -phase Boiling-condensation , including flow interruption and flow resumption

• Flow injection For each sequence, steady-etate or quasi steady state, runs were performed to understand the phenomena of each mode of flow and also to characterize the flow parameters. The loop was first run with low resistance (without pump resistance) and then with high resistance (with pump resistance). l The tests verified that during depressurization the pressure scaling parameters are the p/pi and n/ai where i stands for initial condition. For single-phase natural circulation mode, the test results verified the classical natural circulation equation which balances hydrostatic head versus the flow resis-  !

tance. The primary side flow distributions are relatively insensitive to the in-balance of coolant flow in the secondary 4-7

i side. For boiling-condensation mode, strong intermittent flows were observed. The flow stability was strongly affected by the inventory level; the relative height of flow in primary and secondary sides; the flow resistances in the secondary side.

Throughout the year the test loop went through several extensive remodeling in the periods between test sequences. The remodelling were performed to improve the control and instrumen-tation, as well as to make the test loop more responsive to the demands imposed by each new sequences of test.

i i

f i

i i

4-8

SEMISCALE RECOVERY INVESTIGATIONS:

A COMPARISON OF RESULTS FROM SEMISCALE MOD-2C SMALL BREAK LOCA WITHOUT HPI TESTS

• J. E. Streit T. J. Boucher Idaho National Engineering Laboratory EG&G Idaho, Inc.

This paper presents a comparison of the results from the S-NH test series performed in the Semiscale Mod-2C facility. The S-NH test series simulated small break loss-of-coolant accidents (SBLOCAs) without high pressure injection (HPI) and was designed to gain information on various recovery procedures that are outlined in the emergency operating procedures (ECPs) of commercial '

Pressurized Water Reactor (PWR) plants. Tests S-NH-1, S-NH-3, and S-NH-5 simulated a shear of a 2-inch line penetrating the cold leg of a commercial PWR plant, while test S-NH-2 simulated the shear of a 4-inch line. Three different recovery scenarios were simulated to assess their effectiveness in mitigating the resulting core heater rod temperature excursions. In tests S-NH-1 and S-NH-2 a steam generator steam-and-feed recovery was initiated when the core peak cladding temperature (PCT) reached 811K (10000F). In test S-NH-3 a primary coolant pump was restarted when the PCT reached 811K (10000F) and the steam generator steam-and-feed operations were begun when the PCT reached 950K (12500F). In test S-NH-5 the steam generator steam-and-feed recovery was initiated when the vessel liquid level reached the elevation of the top of the core, before a heater rod temperature excursion began.

SBLOCAs in the form of steam generator tube ruptures, pump seal leaks, and stuck open pressurizer power operated relief valves (PORVs) have already occurred in commercial PWR plants. Additional anticipated small breaks include instrumentation lines and small pipe cracks associated with normal or abnormal operation. The safety issue associated with SBLOCAs without HPI is the possibility of inadequate core cooling resulting from the loss of coolant before the primary pressure decreases to the accumulator and low pressure injection system pressure setpoints. If a core heater rod dry out should occur fuel damage may result before fluid from the low pressure safety injection system can reflood the core. The emergency operating procedure for the Zion nuclear plant (EOP-11) specifies that the operator initiate steam-and-feed or pump restart recovery operations should a condition of inadequate core cooling occur. The E0P defines inadequate core cooling to exist when the average of the ten highest indicating core exit thermocouples is greater than 12000F (922K). It may be too late to prevent fuel damage if recovery procedures are not initiated until this time.

E0P-11 for the Zion nuclear plant specifies that following the unavailability or failure of both a steam generator steam-and-feed operation and a primary steam-and-feed operation to cool down and depressurize the primary system, the reactor coolant pumps (RCPs) are to be restarted (one at a time) in an attempt to reestablish core cooling through either steam flow or two-phase flow. The restart of a RCP under inadequate core cooling conditions is a final effort to prevent severe core damage during a SBLOCA due to loss of other engineered safety features (HPI and auxiliary feedwater).

4-9

Concern has been expressed that the restart of a RCP in a highly voided system may not reestablish adequate core cooling. This concern is partially based on the results of a pump restart experiment (L8-2) performed in the LOFT facility, in which the RCPs were restarted with a voided core and failed to reverse the fuel rod thermal excursion. Another concern is whether the steam generator steam-and-feed operation applied late in the transient (after a condition of inadequate core cooling already exists) can reduce the primary system pressure to the accumulator pressure setpoint.

The results of this series of experiments show that the operator actions were effective in reducing the primary pressure to the accumulator pressure setpoint in order to mitigate the core thermal excursions by accumulator i

injection. In tests S-NH-1 and S-NH-2 the steam generator secondary steam-and-feed operation proved adequate to reduce the primary pressure to the accumulator pressure setpoint and mitigate the core thermal excursions. The larger break in S-NH-2 reduced the primary system inventory faster and, as a result, the core PCT was higher during the thermal excursion. In test S-NH-5 the early steam generator secondary steam-and-feed operation prevented the core j thermal excursion from occurring. The restart of the primary coolant r. ump in j test S-NH-3 redistributed the primary coolant and mitigated the core thermal excursions. Continued depletion of the primary coolant by the break, however, caused the core to heat up again. This heat-up was slower than the first due to continued pump operation. The steam generator secondary steam-and-feed operation reduced the prime.ry pressure to the accumulator pressure setpoint, which slowed the core thermal excursion further.

In all four tests the accumulators emptied before the LPIS pressure setpoint was reached. As a result, the core experienced an additional thermal excursion and was heating up when.the primary system pressure reached the LPIS pressure setpoint and the tests were terminated. It was anticipated that the accumulator injection would decrease the primary pressure to the LPIS pressure setpoint. However, in a recent 5% break experiment (SB-CL-05) run in the ROSA IV facility in Japan the accumulators also emptied before the LPIS pressure setpoint was reached and an additional thermal excursion resulted.

I

• Work supported by the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research, Under Department of Energy Contract No.

DE-AC07-76ID01570.

i 4-10

l The Results of 5% Small Break LOCA Tests and Natural Circulation Tests at ROSA-lV LSTF K. Tasaka, Y. Kukita, Y. Koizumi, M. Osakabe and H. Nakamura l

Japan Atomic Energy Research Institute Integral sirrulation experiments on PWR smal1-break loss-of-coolant accidents (SBLOCAs) are being conducted using the Large Scale Test Facility (LSTF) of the ROSA-lV Program. The LSTF is a volumetrically-scaled (1/48 scale) simulator of a W-type 4-Loop (3423 MWt) PWR.

Table 1 summarizes experiments conducted so far with a break area of 5% of the sca led cold leg flow area. Figure 1 compares core collapsed liquid levels obtained in the cold-leg break experiments. The core liquid level is dependent on two factors:

the vessel mass inventory, and the differential pressure between the upper plenum and the downcomer. SB-CL-05 and -08 indicated core level depression below the loop seal bottom elevation during the clearing of loop seal. This resulted f rom the dif ferential pressure (in 1

addition to the loop seal water head) due to liquid holdup in the steam generator (SG) U-tube upflow sides. SB-CL-06, with a smaller core power and a larger bypass area, indicated a smaller core level depression since the U-tubes were almost empty of liquid at the time of loop seal clearing (LSC).

The opening of a core barrel vent valve in SB-CL-07, af ter 60 s into the transient, reduced the di f ferent ial pressure between the upper plenum and downcomer, so that the loop seal cleaing was delayed. The total failure of ECCS in SB-CL-08 had no significant influence until the loop seal clearing, however, it caused a continued decrease of the primary mass inventory and thus a partial core uncovery before the accumulators came on.

Figure 2 shows the break location ef fects on the core liquid level. The loop-scal break test, SB-LS-01, indicated qualitatively the same response as the cold-leg break experiment. In the hot leg break experiment, SB-HL-01, the condensation depressurization in the cold leg following the accumulator injection caused a steep but temporary decrease of the core level.

The two natural circulation tests (Table 2) consisted of many steady-state stages obtained for different primary mass inventories. The change of primary loop flow rate wi th the decrease of mass inventory (Fig. 3) was qualitatively the same as observed on other facilities. Two phase natural circulation ini tiated when the mass inventory decreased below about 90%.

This resulted in an increase of the loop flow rate until the mass inventory decreased to about 70%. The two phase natural circulation ceased at 56% and 64% inventories for 5% and 2% core power, respectively, as stationary vapor bubbles were formed at the top of U-tubes. Then, the reflux condensation cooling mode followed. The core remained submerged and cooled until the mass inventory decreased to 30% for the 2% core power.

Af ter the completion of the primary-side natural circulation experiments, the SG heat exchange rate was measured for reduced secondary side mass inventories, while the primary side was kept in the reflux condensation mode.

The measured minimum heat excha e rates (recorded for the maximum i secondary respectively.

le vel) were 2.6 a nd 1.7 kW/m /K f or the 5% and 2% core power '

4-11

Table 1 52 Bruk Tstt Conditiens SB-CL-06 SB-CL-07 SB-CL-08 SB-ilt-01 SB-LS-01 Test ID SB-CL-05 Cold tag Cold Ing Cold Ing Hot Leg Loop Seal Break location Cold tag (Bottom)

Bypass (% of core flow)

Downcomer-Hot Leg 0.2 0.6 0.0 0.2 0.2 0.2 Onwncomer-Upper Head 2.1 0.9 0.9 0.3 0.3 0.9 8.3 0.0 0.0 0.0 Internal Vent Valve 0.0 0.0 Single Train Single Train Single Train None None Single Train High-Pressure ECCS 4 Core

] l [ [

JP Table 2 Natural Circulation Test Conditions a

k. I e

( 7

- h ,se ~

• 3 jI ST-NC-01 ST-NC-02

'3 .%--,. Test ID

. I

/ST-SG-02

.[ N.

f, N- /ST-SO-01 3 d .-

s ly\ ~ Q A

q. . . y .. . (. . -- ;

8 -

Nj

• Loop Sea! 3.56/* 1.42/2

.% 2

~ Core Power (MW/ )

• J A {.

,, Bottom

+--

)  !LSC k SB-CL-08 6 8 - 15.5 y - . - SB -CL- 0 7

~

Primary System (MPa) 7.2 - 15.5 I -

4 -- SB -CL- 0 6

'# '#I I" ** * *

• d LSC --.. . . .. S D - CI.- 0 3 8 Core Bottom if L,SC , , ,

0 -

12- '

i 0 200 400 600 800 1000 "

20- .

LS.. sr i\

11 t Time (s) - . Semis:::e lg.

Fig.) Core Collapsed 1.1guid Level Transients to- (NURE3/CR.2325)/ k (51 Cold Leg Break Tests) _

/ l g

,- _ / i.

O t5- /,s s l [ l l Core ._ g I

• O% b"" t' iib \

.......SB-LS-01 t$

SB-CL-08

- --- SB -IIL- 01 j7-2 s'l.I a \ -

, ,\

~

WR.m s

,r, " ii . f. (% '- a.

~

e.

8b " 10-i l 1 t

\'N e'

+

• T I.. . '

i/1(K f J '"

~

, < _ .g

$ j

. .i . . . ,. . n 4 # ,

y . . ,., ' I.oop -- Seal u~ -

.a. *  ;:  ; i i v ...f g2 -

{ l -1 gog g o,, g c-l ,8 .,

.: .:- }4; i +--- ga .8 ,i  !

\ 2

- . , a

, y

! LSC 5 -

[ , p 3

2P / I LS *.' 6er d Core l ,a, l LSC 1 Bottom l /\l l l l l '

4 1',/ / 'l, O

0 200 400 600 800 1000 /., O-

/ ,__f. i p y 4 50 60 70 80 90 @

Time (s)

Fig.2 Core Collapsed Liquid 1.evel Transients Fig 3 L CP Flow Race during Natural Circulation Tests (51 Break Tests) 4-12

COMPARISON OF THE TRAC CALCULATION TO THE DATA FROM LSTF RUN SB-CL-05

• by Frank Motley Safety Code Development Group Energy Division Los Alamos National Laboratory Los Alamos New Mexico 87545 The results of the hrst Large Scale Test Facility (LSTF) transient calculated by the Transient Reactor Analysis Code (TRAC) show moderate agreement with the experimental data. The LSTF is a large-scale (1/48) integral test facility for the study of overall pressurized water reactor (PWR) benavior during a small-break loss-of-coolant accident and anticipated transients. The LSTF models all the major components of the primary and secondary systems of a PWR. The test facility was built by the Japan Atomic Energy Research Institute (JAERI) as part of the ROSA IV program. The test results from the LSTF are shared by JAERI with the United States Nuclear Regulatory Commission according to terms of a bilateral agreement. TRAC analysis of selected tests will be conducted at Las Alamos and Idaho National Engineering Laboratory.

Run SB-C1-05 is a 5% break in a cold leg. The test resulted in a brief core uncovery and minor overheating of the rods at certain core locations Liquid holdup on the upflow side of the steam generators was observed. The clearing of the loop seat refilled the core and cooled the rods The TRAC input model of the LSTF is very detailed and describes the primary system and the steam generator secondary system to the steam valves.

The first calculation of a ROSA IV test showed moderate agreement between the data and the TRAC calculated results. The calculated system pressure response was in moderate agreement with the dat.. We define moderate agreement to mean that TRAC correctly pre-dicted the major tiends and phenomena. The hquid holdup on the upflow side of the steam generator was predicted accurately by the TRAC code. The core liquid depression was also accura (ely predicted. The refilling of the core when the loop seals cleared occurred in the calculation just as it had in the data. When the core uncovered, the rods began overheating.

There was moderate agreement between the data and the calculation for the rod temperatures.

. k F

e

[ Work performed under the auspices of the US Nuclear Regulatory Commission.

4-13

MANAGEMENT OF EMERGENCY FEEDWATER DURING NATURAL CIRCULATION C00LDOWN FOLLOWING A LOSS OF 0FF-SITE POWER SCENARIO Edwin H. Davidson Martin A. Stutzke FLORIDA POWER CORPORATION Recent analyses indicate that rapid cooldown is not necessary to conserve emergency feedwater (EFW) following loss of off-site power at the Crystal River Unit 3 (CR-3) nuclear generating station. Rapid cooldown may not be necessary to conserve EFW for other PWR's which have similar relationships between atmospheric dump valve (ADV) areas and decay heat removal (DHR) system cut-in temperatures.

During the loss of of f- si te power scenario for PWR's, the large steam turbine demand on the nuclear steam supply system is removed by the turbine trip. Reactor trip follows immediately, and the heat removal process continues by diverting steam from the turbine to the atmosphere. For prolonged periods of off-site power loss, with natural circulation cooling, reserves of EFW to the steam generators are utilized to provide removal of heat from core nuclear decay, system metal, and primary coolant. Since the EFW is once-through, assurance is needed that the EFW inventory (condensate-grade) will be sufficient to effect cooldown to the DHR system conditions. The amouitt of time required to cooldown has often been assumed to be a strong function of operator actions and procedural requirements for meeting constraints of 1) pressurized thermal shock and 2) subcooling to preclude forming a steam bubble in the reactor vessel head. These considerations are a part of the NRC's Generic Letter 81-21 on Natural Ci rcul ation Cooldown. Our calculations suggest that wide latitudes may exist at some PWR's for operator cooldown response without affecting EFW requirements.

The analyses performed at Florida Power Corporation for the CR-3 plant conditions gave results, for cooling rates >4F/hr, that were initially unexpected:

1) A minimum time (91 hours0.00105 days <br />0.0253 hours <br />1.50463e-4 weeks <br />3.46255e-5 months <br />) was required to cooldown to DHR system conditions, regardless of initial cooldown rate.

2) Virtually the same amount of EFW was consumed during cooldown to DHR conditions, regardless of cooldown rate.

3) The operator could delay cooldown for up to 77 hours8.912037e-4 days <br />0.0214 hours <br />1.273148e-4 weeks <br />2.92985e-5 months <br /> without exceeding the minimum time for cooldown, without increasing the EFW consumption, and without exceeding a 50F/hr initia! cooldown rate.

The results appear to be determined by the DHR system cut-in conditions and by reductions in the steaming capacity of the ADV's as pressure and temperature are reduced in the steam generators.

4-15

.= __ -

We suggest that the following conclusions are generic to all PWR's using natural circulation cooldown with EFW steam released through ADV's to the atmosphere:

A) A minimum time can be calculated to reach DHR system conditions using i the maximum permissible initial cooldown rate. The cooldown rate may l or may not be near-linear, depending on the DHR system cut-in (

temperature. l l

B) If the cooldown rate is not approximately linear, a linear cooldown i rate may be estimated graphically or calculated consistent with the  !

i minimum time calculation.

C) A minimum EFW requirement exists consistent with the minimum time to reach DHR system initiation temperature.

D) If the maximum permissible plant initial cooldown rate far exceeds the linear rate calculated in B) above, large delays in initiating cooldown

, are possible without exceeding the minimum time to reach DHR system initiation temperature and without exceeding the EFW minimum requirement.

l E) Plant specific values for the items above are strongly influenced by the DHR system cut-in temperature and the total steam release area of the ADV's.

The calculated CR-3 results and the suggested generic applicability were formulated from a computer program which models decay heat generation and heat storage in the metal and water contained in the CR-3 reactor primary cooling system. Energy transfer was modeled via the EFW steaming through the ADV's by accounting for appropriate steam-table values of pressure, temperature and enthal py and for diminished steam rel ease rates via the ADV's as the pressure decreased. Rcdiation heat losses and the cooling effect of makeup water to the primary system were neglected. Simplicity of the modeling and the programming permitted use of the IBM-PC desk top

computer as an in-house and part-time exercise. No outside consultants or vendor organizations were utilized for modeling, programming, or calculating.

Since startup of the CR-3 plant in 1977, only one loss of off-site power has occurred. The duration was only 6 seconds. The probability of losing off-site power for a period in excess of 70 hours8.101852e-4 days <br />0.0194 hours <br />1.157407e-4 weeks <br />2.6635e-5 months <br /> is estimated to be less than 10-5 Since cooldown at CR-3 may be delayed for as long as 77 hours8.912037e-4 days <br />0.0214 hours <br />1.273148e-4 weeks <br />2.92985e-5 months <br /> without increasing EFW consumption, it may be prudent to remain at hot standby conditions, rather than initiating an early cooldown, until off-site power is restored. Operator / management decisions on how long to remain at hot standby do not appear to be constrained by EFW availability but would more likely be influenced by the cause of the AC power loss, the state of the interconnecting grids, and the anticipated time for restoration of off-site power.

1 4-16

PROBABILITY OF FAILURE IN BWH REACTOH COOLANT PIPING

• G.S. Holman, T. Lo, and C.K. Chou Lawrence Livermore National Laboratory University of California Livermore, California / U.S.A.

SUMMARY

As part of a research program for the Nuclear Hegulatory Commis-sion, the Lawrence Livermore National Laboratory (LLNL) has estimated the probability of double-ended guillotine break (DEGB) in the main steam, feedwater, and recirculation loop piping of Mark I boiling water reactor (BWH) plants.

Two causes of pipe break are considered: crack growth at welded joints and seismically-induced failure of component supports. For the former we use a probabilistic fracture mechanics model implemented in our PRAISE computer code, for the latter a probabilistic support relia-bility model. Past LLNL evaluations of pressurized water reactor (PWR) piping indicated that the probability of reactor coolant loop DEGB is low under all plant conditions, including earthquakes, and that the probability of DEGB due to crack growth (less than IE-10 events per reactor year) is much lower than that due to heavy component (i.e.,

reactor pressure vessel, steam generator, reactor coolant pump) support failure (less than IE-6 events per reactor year). Thermal stresses dominated the probability of crack-induced DEGB, while earthquakes contributed only negligibly. The crack growth evaluations also indi-cated that DEGB probabilities are several orders of magnitude lower than leak probabilities.

Although the BWR and PWR reactor coolant system evaluations followed the same general approach, two BWR-specific factors required consideration: potential failure of intermediate pipe supports and supports for light loop components (PWR reactor coolant loop piping being supported solely by loop components), and intergranular stress corrosion cracking (IGSCC).

Pipe support (e.g., hanger, snubber) failure would redistribute the stresses at weld joints and thereby change the predicted leak and break probabilities. We developed a method of incorporating support

" fragility" -- the probability of support failure conditioned upon the occurrence of various levels of seismic intensity -- into the DEGB evaluation. and then investigated how support failure affected the probability of recirculation loop break for a reference BWH plant.

Seismic stresses at the piping weld joints were calculated for each of "This work was supported by the United States Nuclear Regulatory Commission under a Memorandum of Understanding with the United States Department of Energy."

5-1

16 support failure scenarios. Sensitivity analyses were then performed which indicated that the probability of DEGB was very sensitive to the characteristics of the particular seismic hazard curve used in the eval-uation.

To account for'lGSCC effects, we developed a probabilistic model for the PRAISE code based on experimental and field data compiled from several sources. In this model, times to crack initiation as well as crack growth rates for Types 304 and 316NG stainless steel are corre-lated to material-specific " damage paramet'ers" which consolidate the reparate effects of coolant environment (e.g., temperature, dissolved oxygen content, level of impurities), applied loads, and degree of sen-sitization. The model also considers the effect of residual stress on crack initiation and growth. l The results of our evaluations to date have indicated that IGSCC clearly dominates the probability of DEGB in recirculation piping fabricated from Type 304 stainless steel. Failure results mainly from cracks that initiate within a few years after plant operation has begun. Replacing Type 304 piping with Type 316NG not only decreases the number of initiations, but those cracks that d'o initiate do so much later in plant life. The resultant failure probabilities are therefore several orders of magnitude lower than in Type 304 piping.

A N

5-2

SSI and Structural Benchmarks A.J. Philippacopoulos, C. A. Miller, C.J. Costantino Brookhaven National Laboratory Upton, NY 11973 H. Graves Office of Nuclear Reactor Research U.S. Nuclear Regulatory Commission Washington, DC 20555 Summary This paper presents the latest results of the ongoing program entitled,

" Standard Problems for Structural Computer Codes", currently being worked on at BNL for the USNRC, Office of Nuclear Regulatory Research. During this year, efforts were focussed on three tasks, namely, (1) an investigation of ground water effects on the response of Category I structures, (2) the Soil-Structure Interaction Workshop and (3) studies on structural benchmarks associated with Category I structures.

The objective of the studies on ground water effects is to verify the applicability and the limitations of the SSI methods currently used by the industry in performing seismic evaluations of nuclear plants which are located at sites with high water tables. In a previous study by BNL (NUREG/CR-4588),

it has been concluded that the pore water can influence significantly the soil-structure interaction process. This result, however, is based on the assumption of fully saturated soil profiles. Consequently, the work was further extended to include cases associated with variable water table depths. In this paper, criteria related to " cut-off" depths beyond which the pore water effects can be ignored in seismic calculations, are addressed.

Comprehensive numerical data are given for soil configurations typical to those encountered in nuclear plant sites. These data were generated by using a modified version of the SLAft code which is capable of handling problems related to the dynamic response of saturated soils.

Further, the paper presents some key aspects of the Soil-Structure Interaction Workshop (NUREG/CP-0054) which was held in Bethesda, MD on June 16-18, 1986. This workshop was set up with the following objectives:

(1) to examine the SSI related licensing concerns and various procedures and alternatives jointly by the regulators, practitioners, researchers, utilities and other interested groups in the light of the recent analytical and

• This work was performed under the auspices of the U.S. Nuclear Regulatory Commission.

5-3

experimental development; (2) to examine the areas of greater uncertainties and means to address them; (3) to review the licensing criteria in the SSI area with an overall view and discuss suggestions to improve the licensing process; and (4) to present results of recent USNRC research. During the workshop the emphasis was placed on particular aspects related to the Unresolved Safety Issue (USI) A-40, " Seismic Design Criteria". Consensus opinions and recommendations given by the panelists on these issues are presented in this paper.

Finally, recent efforts related to the task on the structural benchmarks are described. The objective of this task is to develop benchmark problems which can be used to validate methods and computer codes used by the industry i in making predictions of the behavior of Category I structures. Each benchmark consists of an experimental observation which is compared with code predictions. The structural loads, the response characteristics as well as i the uncertainties in strength are discussed in detail for one benchmark problem.

I i

l l

d f

s i

5-4 i

VALVE PERFORMANCE TESTING - CHECK VALVE RESULTS N. M. JEANM00 GIN ENERGY TECHNOLOGY ENGINEERING CENTER The Valve Performance Test Program addresses current requirements for testing of pressure isolation valves (PIVs) in light water reactors. The objectives of the program are to evaluate (1) the use of valve leakage as an indicator of valve condition, (2) the Section XI correlation for extrapolating leak rates to higher differential pressure conditions, (3) the use of. motor signature monitoring as a measure of gate valve operability, and (4) the use of acoustic emission techniques for quantifying valve leakage and as an indicator of valve condition. Six valves were procured as test articles, three check valves and three gate valves. Check valve testing was completed in fiscal year 1986 and is the subject of this pape .

Swing check valves with nominal diameters of four, six and twelve were used for the l Valve Performance Program. Life-cycle testing of the check valves, interspersed with leak rate measurements, was done to assess the correlation between valve wear and leak rate. The valves were subjected to long-term flashing to evaluate the effects of erosion on valve seating surfaces. Ir, addition, loosened internals testing of the check valves, which examined the capability of check valves to perform their pressure isolation function with damaged internals, was completed. ,

Acoustic emission traces were recorded throughout the check valve leak rate measurements and loosened internals testing.

Wear of the check valve seating surfaces due to life cycling and long-term flashing did not result in significantly increased leak rates. An important result from the l check valve testing is that leak rate testing was ineffective in detecting check valve loosened internals damage or deformation of the disk hinge pin / stud. Thus, the standard method for determining valve condition may not alert the plant operators to advanced degradation or possible impending valve failure.

Acoustic emission traces were recorded throughout the life cycle tests and the loosened internals tests. These data have not been fully analyzed; however, acoustic emission monitoring does show promise as a method for detecting loosened internals in check valves. The check valve leak rates were so small during the program, never exceeding 3 x 10-3 lb/sec, that the acoustic emission monitors would not be expected to detect the leak rates either qualitatively or quantitatively.

5-5

Piping and Fitting Dynamic Reliability Program Dan Guzy USNRC This cooperative EPRI/NRC program was initiated in mid-1985, with the following objectives:

o To identify failure mechanisms and failure levels of piping components and systems under dynamic loadings.

o To provide a data base that will improve our prediction of piping system response and failure due to high level dynamic loads.

o To develop an improved, realistic and defensible set of piping design rules for inclusion in the ASME Code.

By October 1986, ANC0 Engineers will have completed approximately one-half of the 40 planned component tests in this program. Also the planned piping systems tests shnuld be underway by then. Although this 3-year program will entail much future testing, analysis and standards development activities, the results to date provide important evidence that the assumptions used in establishing the current ASME Code criteria for piping dynamic loads were inaccurate,and conservatisms are greater than previously believed. Measured failure levels have been typically 20 times greater than the level D stress limits, and the failure mechanism has been shown to be attributable to fatigue, not collapse.

These iritial findings are being used to support the development of an ASME Code Case that will provide immediate interim relief for level B dynamic stress criteria for piping (e.g., OBE stress limits). They may also serve in changing current NRC reouirements involving the " loss of function" concept for piping.

The tests have shown that (put simply) pressurized piping fails by localized bulging and cracking rather than by cross sectional collapse that leads to flow reduction.

The comprehensive piping design rule changes that will come eventually from this program have the potential for being the greatest improvement in nuclear piping design in recent years, outweighing changes that effect only response calculation (e.g. Code Case N-411 damping). By quantitatively establishing failure mechanisms and margins-to-failure, more realistic and balanced piping design rules will be made that should lead to better overall reliability and safety of nuclear piping systems.

5-7

6-INCH DIAETER PIPE SEISMIC FRAGILITY TEST W. P. Chen V. DeVita A. T. Onest..

ENERGY TECHNOLOGY ENGINEERING CEN1ER P. O. Box 1449 Canoga Park, California 91304 This paper contains the results and assessments of seismic fragility tests performed on a 6-inch diameter piping system. The objective of thewithstand to test was high to investigate the ability level dynamic seismicofand representative piping other loadings by: (systems

1) testing a representative 6-inch diameter nuclear piping system to failure under dynamic loads; (2) characterizing the high level dynamic response; (3) identifying the failure mode; and (4) providing a benchmark test for a HEDL analysis (Ref) to quantify the analytical conservatism in: (a) current ASME B&PV Code (ASME Code) design criteria, (b) failure predictions based on several emerging nonlinear piping response analysis methods, and (c) probabilistic risk assessment (PRA) methods for piping systems.

Testing was performed in the ETEC Seismic Fragility Test Facility. This facility was designed for low frequency, high acceleration seismic testing of piping systems and components and is capable of applying high level, dynamic seismic base motions in moderate sized piping systems. Levels of loadings achieved during seismic testing were of the order of 20 to 30 times larger than usually specified for Safe Shutdown Earthquakes (SSE's) of contemporary nuclear power plants. Failure of the piping system occurred during dynamic testing and provided data directly applicable to attain the test objectives.

The piping system utilized in the test (test article) consisted of 48 feet of 6-in. diameter and 17 feet of 3-in. diameter carbon steel piping and components and included a simulated valve assembly. Materials of construction, fabrication, and proof pressure testing of the test article were in accordance with ASME Code,Section III, Class I requirements.

The test article was internally pressurized at 1000 psi and was subjected to the following three levels of dynamic seismic loads:

Low level seismic load: 5 g nominal ZPA*

Intermediate level seismic load: 14 g nominal ZPA High level seismic load: 25 g nominal ZPA Provisions were also made to conduct the following sequence of three sine burst tests following seismic testing if failure of the test article did not occur during the seismic tests:

Sine burst - 4 Hz: 8 cycles of 1 7 in. maximum displacements Sine burst - 5 Hz: 11 cycles of f 7 in, maximum displacements Sine burst - 6 Hz: 7 cycles of 1 7 in. maximum displacements

• ZPA (Zero Period Acceleration) - Peak g level 5-9 l

" Failure" (i.e., rupture) of the test article did not occur during seismic testing. However, a 2-in. wide circumferential bulge indicative of ratcheting was observed following the high level (30 g actual ZPA) seismic test in a vertical leg of the test article. Rupture occurred in the circumferential  !

bulge during the sixth of the planned eleven cycles of the 5 Hz harmonic input. Failure resulted from a 300 degree circumferential break in the bulge; a classic double-ended guillotine break was avoided with prompt tarmination of testing.

The tests demonstrated the increasing resistance of the piping system to respond to increasing levels of seismic loadings. This characteristic l was exhibited by the peak acceleration or amplification observed during testing.

Strain gage data indicated that inelastic straining occurred in the highly stressed elbows and straight pipe near the critical support during high level seismic testing. The circumferential, longitudinal, and average radial residual strains at the failure location were 9.2%, 0.7%, and 12%, respectively. These strains were in good agreement with the results of a simplified ratcheting analysis performed by ETEC.

Based on test results, estimated damping for the seismic tests were between 1-6%, 3-12% and 13-22% for the low, intermediate and high level seismic tests, respectively, and 19% for the 4 Hz sine burst test.

t Failure of the test article was attributed to incremental ratcheting due to the internal pressure in the piping system resulting in wall thinning and bulging and subsequent fracture.due to tensile overloading. Local wall thinning of up to 25% was found in the bulge during post-test examinations

This failure mode was different from the collapse failure mode on which current nonlinear failure analyses are based. Furthermore, failure did not occur in any of the locations of high stresses considered critical by the ASME Code.

Based on the maximum input acceleration of 30 g applied during the high level seismic test, it was concluded that the tests demonstrated that piping systems are inherently strenger to resist dynamic loading than currently permitted by design criteria in existing codes and standards or would be predicted by several nonlinear methods and analysis.

REFERENCE i

M. J. Anderson, M. R. Lindquist, et al., " Pretest Failure Predictions.

Post Test Analyses, and Comparisons to Test Results of the NRC/ETEC Seismic Fragility Demonstration Piping Test," Draft Report to NRC, HEDL-TC-2779, June 1986 i

L l LL/K/24 5-10

PIPING DAMPING STUDIES A.G. WARE IDAHO NATIONAL ENGINEERING LABORATORY The Idaho National Engineering Laboratory (INEL) has been conducting a research program to assist the United States Nuclear Regulatory Commission (USNRC) in determining best-estimate damping values for use in the dynamic analyses of nuclear power plant piping systems. The results of this program are being used to assess whether the current damping guidelines in Regulatory Guide (R.G.) 1.61 and the American Society of Mechanical Engineers Boiler and Pressure Vessel Code (ASME Code) can be revised to allow higher damping values for dynamic piping analyses. Higher damping values would allow more flexible piping systems with fewer supports, making the piping systems better able to expand and contract during thermal loadings, less susceptible to fatigue, and will reduce inspection, maintenance and worker radiation exposure. This paper will briefly describe four tasks performed as part of this program at the INEL in FY-86.

In the first task damping data in the seismic frequency range (0 to 33 Hz) from three laboratory test piping systems were evaluated. A 5-in.

system was tested at the INEL and two 6-in. systems were tested by ANCO Engineers. The primary sources of energy dissipation observed during these tests were through the supports, the insulation, and material damping. The effect of pressure on damping at seismic design levels was considered to be minimal. The supports influence damping at all response amplitudes of vibration. In the lower level tests the damping was higher with a mechanical snubber in the system than when this snubber was replaced by a rigid strut or a hydraulic snubber. The support effects observed can be attributed to the interaction of the vibrating modes with the supports. Damping values for the ANCO systems were relatively low, in the 1 to 3% of critical range, for most of the tests. The addition of insulation in the INEL tests produced a large increase in damping during the shaker tests with earthquake-like motion, but there was no comparable increase during the snapback (pull and release) tests. This is attributed to the fact that during the random input motion shaker tests the pipe was continually vibrated against the insulation, while during the snapback tests the pipe and insulation motions decayed freely together. Once a certain response amplitude threshold was reached (approximately one-half yield stress), overall system damping appears to increase to high levels.

This effect is most likely due to a combination of material hysteresis in the pipe material, and greater impacting and energy losses in the supports.

The second task involved gathering and evaluating damping data in the 33 to 100 Hz frequency range. Both R.G. 1.61 and the ASME Code prescribe damping values for seismic analyses, but there are no formal guidelines to be used for higher frequency loads such as water hammer, pressure relief valve discharge, and loss-of-coolant-accident (LOCA) related hydrodynamic loads. The research plan included collecting data from the published literature, conducting tests at high frequencies on laboratory 3- and 5-in. piping systems at the INEL, and evaluating the data to describe damping values representative of'the test data for piping vibrations at these higher frequencies. The published high-frequency data was rather 5-11

sparse, but pertinent data was obtained from General Electric (GE) and Kraftwerk Union (KWU) sources. The data chosen as the basis for estimating high-frequency damping values were heavily weighted by the results of the INEL 3- and 5-in tests. One significant characteristic of the data is that damping appears to be higher when insulation is present.

Another feature is that damping levels were greater at the higher excitation levels (when stress levels were near yield stress), and for systems with intermediate supports. It was concluded that suitable representations of the test data could be achieved by increasing the Pressure Vessel Research Committee (PVRC) recommended uninsulated damping curve of 2% of critical in the 20 to 33 Hz range to 3%, and extending both this curve and the PVRC insulated curves at the recommended levels to 100 Hz.

The third. task involved selecting pipe damping data representative of at least moderate severity seismic or hydrodynamic transients, and performing a statistical study to determine the probability distributions for probabilistic risk assessment (PRA) analyses and to present the data graphically to assist in preparing regulatory guidelines. Relevant i damping data was chosen from 27 different light water reactor (LWR) piping systems, most of which were from actual nuclear power plants. In the first part of this study, damping was treated as independent of frequency (or mode number). The statistical analysis showed that a lognormal probability distribution provided a suitable approximation of the raw data. For analyses in which results from more that one test on a single mode were eliminated (one damping value per mode per piping system), the average damping (arithmetic mean) ranged from 4.4 to 4.8% of critical, while the averages of the natural logarithms produced values (geometric means) ranging from 3.1 to 3.7% of critical. The scatter in the data produced relatively large standard deviations.

The final task has been to collect and evaluate damping data at high strains, at levels considerably beyond the design basis. Results have shown that damping increases to significant levels in the high strain region, causing response motions to be limited so that a large design margin exists above the elastically computed allowable stresses. Data are available from tests conducted at the INEL, the University of Akron, Hanford Engineering Development Laboratory, KWU, and the British Central Electric Generating Board. Further data is being developed in USNRC and Electric Power Research Institute (EPRI) programs in which testing is being conducted at the ANCO Engineers and Energy Technology Engineering Center (ETEC) laboratories. The GE Company is also an active participant in the latter programs.

5-12

NONLINEAR PIPING DAMPING AND RESPONSE PREDICTIONS

• L. K. Severud, E. G. Weiner, M. R. Lindquist, M. J. Anderson, S. E. Wagner Hanford Engineering Development Laboratory Westinghouse Hanford Company, Richland, WA The objective of the research being carried out in this program is to provide the bases for recommending pseudo-linear-elastic methods and procedures to be used in predicting the high-level seismic response and failure of piping systems. These methods can then be used to assess the margins provided by the present ASME Code rules for piping seismic design, to reduce the number of needed piping restraints, to increase the piping reliability and contribute to Code improvements.

During 1986, the NRC/NRR sponsored fragility testing of a 6-inch diameter piping system at the Energy Technology Engineering Center (ETEC), Reference 1.

Previously, a 1-inch pipe system was tested at HEDL, Reference 2, a 3-inch system at ETEC, and a branch 2 and 4-inch system at KWU, Reference 3.

These test results provided correlation data for comparison to pre and post-test predictions using various cimplified methods. We are preparing a NUREG report on this work.

The following simplified elastic, elastic-plastic, and nonlinear inelastic methods were employed:

a. A standard ASME Class 2 design analyses

b. The Newmark modified spectra method (Reference 4) c: The Campbell / Kennedy / Thrasher dynamic / static margin ratio method (Reference 5)

d. A progressive hinge simplified limit analysis method by Jaquay (Reference 6)

e. A nonlinear transient dynamic inelastic analysis using the NONPIPE computer code (Reference 7)

f. Seismic PRA-type piping fragility estimations per the SSMRP and industry procedures.

The test results revealed significant, very high (over 30%) equivalent system damping when the pipe system experienced inelastic response. This corresponded to ductility ratios in the 5 to 10 range.

• Work sponsored by the U.S. Nuclear Reguletory Commission, Office of Nuclear Regulatory Research, under FIN D1611 with the U. S. Department of Energy.

5-13

1 The ETEC 6-inch pipe withstood 25g's ZPA testing without collapse but eventually failed after repeated tests. The failure mode was described as a local ratchet-fatigue rupture subsequent to local diametral bulging at one pipe leg anchor location. The predicted collapse capacity levels were only

about half that withstood by the piping. Moreover, ASME code allowable design levels were less than 10% of the test-withstood level.

A ratchet-fatigue failure mode analysis was accomplished and compared well l with the test findings. However, further development of procedures and limits are recommended.

Simplified methods for calculating modal equivalent viscous damping for plastic hinges and snubbers were also developed. These methods were used to assess damping energy distribution in four pipe systems in high-level response.

l It appears that the simplified inelastic methods and acceptance criteria have

very good potential for becoming reliable, conservative design tools. Before

~

simple design-oriented methods are endorsed by NRC, however, additional test i data and analytical correlations are needed to provide adequate confidence in l the procedures and code limits. l REFERENCES

1. A. T. Onesto, W. P. Chan, and V. Devita, Six-Inch Pipe Seismic Fragility Test, Draft NUREG Report to NRC, ETEC, July, 1986.

f l 2. M. R. Lindquist, M. J. Anderson, L. K. Severud, E. O. Weiner," High Level i Dynamic Testing and Analytical Correlations for a Small Bore Piping i System," ASME PVP Conference Special Publication PVP-Vol. 108, July, 1986.

3. E. Haas, et al, " Experimental and Analytical Investigation of the Dynamic Response of a Branched Piping System Due to Elasto-Plastic Behavior",

Trans. of the 8th SMIRT, paper K 15/3, Brussels, Belgium, August 1985.

! 4. N. M. Newmark, "A Response Spectrum Approach for Inelastic Seismic Design of Nuclear Reactor Facilities," Transactions Third International i Conference on Structural Mechanics in Reactor Technology (London), Paper

! K5/1, Vol. 4, Part K, 1975.

) 5. R. D. Campbell, R. P. Kennedy, R. D. Thrasher, " Inelastic Response of l Piping Systems Subjected to In-Structure Seismic Excitation," ASME Paper 83-PVP-50.

l 6. K. Jaquay, "A Simplified Limit Analysis Approach for Seismic Design of

! Ductile Piping Systems," Presentation at the ASME Winter Annual Meeting, New Orleans, December 9-14, 1984.

l f 7. NONPIPE A Computer Program for Nonlinear Analysis of Piping Systems-User l Information Manual, Engineering Decision Analysis Co., Inc., Palo Alto, CA, September 30, 1983.

5-14

Eadmples of NRC Research Products Used in Regulation N. Anderson, USNRC The key to effective research is a close relationship between information needs and research results. This can only be achieved by close cooperation between the " researchers" and the " regulators." At the NRC, this relationship has matured over the years until now the researchers participate in definition of the information needs and the regulators help define the research programs.

The more formal means of ensuring a close match between needs and results include joint research review groups, oversight working groups, and a system of Research Information Letters (RILs). On an informdl basis there are many day to day discussions and meetings on the various programs which ensure effective program guidance and early identification of significant findings.

This paper describes both the formal and informal researcher / regulator interface ano discusses some examples of how specific research programs are utilized in the reguldtory process.

Specific progroms described are the pressurized thermal shock program, the seismic margins progrom and the Category 1 structures program. Other exomples cited are the aging and life extension and piping programs.

5-15

TMI-2 LOWER VESSEL DEBRIS EXAMINATIONS

0. W. Akers, C. S. Olsen, and R. V. Strain Examination of the TMI-2 reactor core began in 1982 when a closed circuit television camera revealed extensive damage to the top of the fuel assemblies. From this initial observation, a broad examination program has evolved to determine the extent of core damage. This examination program includes documentation of the post-accident core condition, and sampling of the loose debris, fuel rod stubs, once-molten slag and other damage features. One of the more important examinations being performed is of prior molten material that was relocated from the reactor core proper to the lower vessel head of the reactor.

Twelve particles of debris from the lower reactor head were received at the INEL. Four of the particles had apparently broken off from larger pieces since only eight particles were originally shipped from TMI. The particles ranged in weight from 0.4 g to 553.9 g with corresponding radiation levels to 42 R/hr beta / gamma at 10 inches for the largest particle.

The matrix density which excludes the open porosity was determined by immersion density for six particles that weighed more than 15 g. The density varied from 6.57 to 8.25 g/cc with an average of 7.07 g/cc. The porosity was extensive in these particles varying between 2 and 51% with an average of 25%.

Particles were selected and sectioned for mechanical testing, metallographic, and radiochemical analyses. A compression sample removed from the largest particle failed at 111.0 MPa compared with 95% dense UO2 pellets failing at 166.2 MPa. The higher porosity in the TMI samples account for the lower strength. The TMI compression strength is calculated to be slightly higher than that of pure UO 2 based on porosity-free material.

The microstructural data indicate that the grains are a solid solution of 76 w/% UO2 and 21 w/% Zr0 oxides of nickel, chrome,2iron, with and smaller amounts silicon. (lessboundaries The grain than 1%) ofwere much more diverse and consist princip tily of a dendritic structure of UO 2 or Zr02/U0g platelets in a mixture of chrome and iron oxides with the occasional presence of nickel and aluminum oxides.

In the largest particle, two small areas of pure U02 were found.

Although the stoichiometery was not measured, the stoichiometric ratio for (U,Zr)02 in adjoining areas would oggest that this material was nearly stoichiometric, and that the I:% was near the melting point.

} This observation places the minimum .aTculated peak temperature near 3000

{ K rather than 2800 K for the ceramic mixture of uranium and zirconium

) oxides.

i i

h 6-1

The lower plenum samples contain little metallic material. However, two of the ceramic particles examined contain many metallic inclusions which are generally associated with the dendritic structure observed in the grain boundaries. These inclusions are frequently rich in Ni and often contain Sn. In a few cases, the fission product Ru was also present, and in one area, Te was detected. The origin of the Te could be either impurity or fission product. Less frequently the metallic element silver was found along with molybdenum and indium.

Several chemical separation and measurement methods are being used in the radiochemical analysis of the lower vessel debris. Tracer techniques are used during sample dissolution for the Sr-90, I-129, and elemental tellurium analyses. Principal measurement methods used are gamma spectroscopy for gamma ray emitters, neutron activation analysis for I-129, U-235, and U-238, liquid scintillation counti'ng for Sr-90, and inductively coupled plasma spectroscopy (ICP) for elemental composition.

An experimental analytical technique (microgamma spectroscopy with tomographic reconstruction of the location of gamma ray emitters) will also be used in the lower vessel debris analysis to allow comparison of the distribution of gamma ray emitters in small samples with elemental distribution data obtained from SEM/WDX and microprobe examinations.

Elemental analysis results indicate that the lower vessel debris bed is generally similar in composition to what would be expected if all components of the original core were mixed and a fraction transported to the lower head of the reactor.

Calculation of the retention of fission products in the fuel debris has been performed. This analysis is done by comparing the measured fission product concentrations with the average, predicted quantities of fission products in a gram of fuel as calculated using the code ORIGEN-2.

Although actual and predicted concentrations may vary by a factor of two, this analysis provides an indication of relative, fission product retention in the debris. Using this method, calculated retention of radioiodine in the lower debris bed ranges from 0-10 percent, whereas in the upper debris bed retention ranges from 10-28 percent. The difference in retentions may be due the greater fraction of prior molten material

(~100 percent) in the lower vessel debris. For Cs-137, retentions at the two sample locations are similar with the lower vessel debris retaining 6-22 percent of the expected inventory and the upper vessel debris retaining 6-32 percent. These data suggest that Cs-137 may be retained to a greater exent than expected in fuel material that has completely melted. Anomalous behavior has also been observed for the radionuclides, Sb-125 and Ru-106.

I 1

6-2

Preliminary Results of the TMf-2 Core Bores R. K. McCardell E. L. Tolman M. R. Martin R. P. Smith j Idaho National Engineering Laboratory, EG&G Idaho, Inc. Idaho Falls, ID 83415 The TMI-2 Accident Evaluation Program (1) is being conducted for the United States Department of Energy (DOE) as part of the overall TMI-2 recovery effort. 'he major objectives of the Accident Evaluation Program are to complete cur understanding of the core damage progression, including the end-state distribution of fission products and to apply the major TMI-2 research findings towards resolution of severe accident and source term technical issues.

The principle information necessary to complete our understanding of the mechanism controlling the progression of core damage includes 1) the end-state visual characterization of the damage to the lower half of the core, the core support structures below the core and the reactor vessel lower head, and 2) samples of the degraded core materials for studying the important physical and chemical interactions that took place during the accident and that allow closure to be made relative to the distribution of fission products.

A comprehensive Sample Acquisition and Examination Project (2) is underway to visually characterize the core damage via closed circuit television and to obtain the necessary physical samples of the core materials as the core is being defueled. To provide samples of core materials from known locations of the core and lower plenum regions, a drilling unit was designed to drill through the degraded core material. The drill unit was modified using available mining / geology equipment and technology to provide (a) precise positioning over the reactor vessel; (b) a microprocessor for operational control and safety interlocks; and (c) recorded drilling parameters (torque, load,etc.) The machine drilled " core" samples approximately 2.5 inches in diameter which were enclosed in slightly larger casing tubes. The encapsulated samples were then removed from the core region. The resulting hole provided access for video inspection. The drill samples will be gamma-scanned, sectioned, and radiologically cnd metallurgically examined to determine fission product retention, material composition, and prior peak temperatures.

The core bore machine was used to acquire ten samples from those fuel assembly positions shown in Figure 1. In four of the locations, drilling extended to below the core region into the core support regions. In two of these locations, the drilling and sample acquisition extended through the core support structures into the fuel debris resting on the lower reactor vessel head.

6-3

Preliminary Results of the TMI-2 Core Bore Acquisition and Analysis - Page 2 Video data characterizing each of ten drill locations, along with the data characterizing the "drillability" of each location, has provided important data to interpret the damage conditions of the lower core and core support structures. The data have also been helpful in identifying the core damage locations and evaluating the mechanisms leading to the core degradation and fuel migration into the lower plenum. Major findings including the following:

1. The central two thirds of the lower part of the core generally consists of two distinct regions:

an upper region of previously-liquified ceramic and structural materials, starting approximately five feet above the lower end fittings, forming a convex lens-shaped layer roughly three feet thick near the core center and tapering to a foot near the outer edge a region of intact rod stubs (in some cases, oxidized) below the previously liquified materials

2. The migration path of the previously-liquified material to the lower plenum is located near the core periphery

3. There appears to be no major damage to the lower core support assembly based on the limited video data

4. The fuel debris resting on the bottom vessel head near the center of the reactor vessel appears to be loose and relatively fine, compared with the larger agglomerated debris existing near the edge of the reactor vessel Specificimplicationsofthecoreboreacquisitiondatarelativeto9gt understanding of the core damage progression is discussed in a papert 1 to be presented later in this session. Examination of the core bore samples will further confirm our understanding of the mechanisms controlling the core damage progression and fission product behavior during the accident.

6-4

References

1. E. L. Tolman et al, TMI-2 Accident Evaluation Program, EGG-TMI-7048, February 1986

2. M. Russell et al, TMI-2 Accident Evaluation Program Sample Acquisition and Examination Plan, EGG-TMI-7132, January 1986

3. E. L. Tolman, P. Kuan, " Implications of the Core Bore Acquisition Data on our Understanding of the Core Damage Progression, 1986 Light Water Reactor Safety Symposium (October 1986) 6-5

g Access. w N ] Locations mte Lower Plenum Burnable Poison Fuel Assembh i5 Axial Power u h Shaping Fuel us.msn

9 33 . . ii.m .

Control Rod 12 M- - M# -

M M -

A r i usembw ii  !!si;  ::si:  !:sc , _

s s . hstrumented i0

'iiiii;

• A iiiii '

h * *

• Fuel Assembh g  !!si  !!?!! lifl .

M  !!?! =

M w too, 3 +  !!#l $  !!$i $ li!$i.  !!Wi! C Outlet g . iis .  !!*o W Q .  !!!!! $  !!!E! = !!!#i:

5 A Loop

" EN6 "

!!si IM !!!E!

g Outlet M e h iiiii h  !!$

- - i;#l s _

9;ii , .

iign g ^'

. i; . . .:i 1 =

A B C D E F G lH K L M N O P R

$ Core Bore Limited to Core Region

@ Core Bore Extended hto Core Support Assembb Q Core Bore Extended into Lower Plenum Region Figure 1. Ten Core Bore Locations 6-6

TMI-2 ACCIDENT SCENARIO UPDATE B. Tolman P. Kuan J. Broughton Idaho National Engineering Laboratory The TMI-2 offers the opportunity of increasing our understanding of many currently unresolved severe accident and source term technical issues.

The major objectives of the TMI-2 Accident Evaluation Program (l) are to complete our understanding of the core damage progression and the end-state fission product distribution, and to apply the major TMI-2 research findings towards resolving reactor safety issues. The basic research approach to complete our understanding of the accident will require integration of the end-state core and reactor system characterization data, the reactor system thermal hydraulic response data and major research findings from independent severe accident research.

An accident scenario using this approach has been developed and is summarizedinReference1ggdgasbeenpresentedanddiscussedat several technical meetings 4 The basic features of the scenario include:

1. Rapid zircaloy oxidation and heatup of the upper half of the core occurred during the time interval between 150-160 minutes. The rapid heating resulted in the downward floy of molten zircaloy cladding and some liquified UO' fuel which then froze at the coolant interface (0.5 to 1.0 m from the bottom of the fuel rods). This relocated core material formed a solid, non-coolable configuration estimated to be several cubic meters in volume. An approximation of this configuration is shown in Figure 1(a).

2. The pump transient at 174 minutes shattered the upper fuel rod remnants and oxidized cladding, forming a rubble bed on top of the already non-coolable region near the center of the core as shown in Figure 1(b).

3. The core material in the interior of the non-coolable region continued to heat up and remelted but was contained by a crust of solified prior molten core 6-7

material at the periphery of the non-coolable region.

Previous discussions have suggested that at approximately 224 minutes the bottom crust failed allowing the molten core material to migrate to the lower plenum.

The core bore acquisition data (5) have provided added insights in developing details of the above core failure mechanisms. The implications of the core bore acquisition data on the core damage scenario are summarized below:

1. The basic core relocation scenario, i.e. formation of a non-coolable core configuration from the relocation of molten zircaloy and liauified fuel appears to be sound. Subsequent heatup of the degraded core leading to failure of the supporting crust also appears reasonable, although the failure mechanisms need to be more clearly identified and substantiated by the core bore acquisition and examination data.

2. Intact fuel rod stubs in the lower core regions confirm that Th cooling in the lower 0.5 m of the core was maintained.isconsistentwithprevio for the minimum core liquid levels.

3. The data suggest that the crust failure location and molten material flow paths to the lower plenum are in the east quadrant of the reactor vessel, near the core periphery as shown in Figure 2.

4. The crust failure mechanisms appear to be mechanical in i nature. The mechanical stress in the crust at the time of failure needs to be further evaluated.

5. The lack of evidence of damage to the lower core support structure suggest rapid flow of the molten core material to the lower plenum regions.

I Detailed examination of the core bore samples together with additional characterization of the lower regions of the core, support assembly and lower plenum region will provide the basis for evaluatingongoingengineeringcalculationssimulatingdegradedcore formation ag heatup( ) and the failure mechanisms of the degraded corecruststg) .

1 6-8

REFERENCES

1. E. Tolman et.al, TMI-2 Accident Evaluation Program, EGG Idaho Report EGG-TMI-7048, February 1986.

2. E. Tolman et.al, " Thermal Hydraulic Features of the TMI-2 Accident", ACS Symposium Series 293, 189th Meeting of the American Chemical Society, Miami Beach, Florida, April 28-May 3,1985.

3. J. Broughton, " Core Condition and Accident Scenario",

Proceedings of the First International Information Meeting on the TMI-2 Accident, Conf-8510166, October 1985.

4. J. Broughton ,"TMI-2: Core Conditions and Postulated Accident Scenario", ANS Transactions, Volume 50, November 1985.

5. R. K. McCardell, et. al., " Preliminary Results of the TMI-2 Core Bore Acquisition and Analysis " 1986 Light Water Reactor Safety Symposium (These Proceedings).

6. K. Ardron, D. Cain, TMI-2 Accident Core Heatup Analysis, l NSAC-24, January 1981.

7. P. Kuan, TMI-2 Core Debris Bed Coolability, EGG Idaho Report EGG-TMI-7150, March 1986.

8. P. Kuan, Core Relocation in the TMI-2 Accident, EGG Idaho Report To Be Published.

6-9

l'I I PI'I' (a) Core condition prior l r-- to pump transient jf 9 I' !I l -

i d

%M c![! j t

"* q Highly oxidized fU i

}j y ,- - rod like geometry Approximate l $

Relocated and partisfly coolant -

liquid level solidified prior molten N., core material Intact fuel - .

rod stubs

,i Solidilled crust Instrumer.t penetration 1 tubes l;l g s195I?I!I:inen (b) Core condition after pump

5 5 si i2Q FC

^

transient at 175 minutes fif0 -

N _ s

(

l i i c 1,f Upper volded regicn I

Ob caused by shattering upper oxidized fuel rods j '. ,

Debris bed of oxidized f[ ,

fuel rod materials Molten core materials--

• Solldified, prior molten materlaf surrounding standing fcel rod remnants Prior mottog solidilled' ' ' "

material r

~ "e::.. -

4 em

( [j]; u i

Figure 1. Core Damage Progression Scenario.

6-10

i 1

A Accesst>k N  % tocatwwinto

- Lower Phnum Burnable Poison Fuel Assembh 15 j gA %L Axia1 Power g[ O Shaping Fuel 14 .

' -  : % L, Assembh is I. A . .

q 12 l lJ% [ .

{ }* .i j =

Contro1 Rod Fuel Assembh 11 I [s  :!

y i;...!!

!!n!!

!!M!! =

l io gg . g = . g . s  ;

. - tu ra'i ^==*

, lpm . . _

w; y 3 -> fs' Estimated l Outer r.m

• p B Loop outht E - l Edge of Molten Core 7

Materials in Degraded E

• _fL Core Region

= l-m.~ ...

5 =

limi!

=

pWi pg. (

1 4%lYN Outki . g $g g '\ stimated E Locotton L ( gq of Molten Material 3 . I 2 1 . g liF r Breakout t 9%K A B C D E F G H K L M N O P R Figure 2. Estimated Core Breakout Location.

6-11

Update on Standard Problem, Data Base and Uncertainties D. W. Golden, Idaho National Engineering Laboratory, EG&G Idaho, Inc.

Idaho Falls, ID Summary The TMI-2 accident provides a unique opportunity to develop an understanding of the behavior of power reactors under severe accident conditions and to benchmark the severe accident computer codes. This requires that the TMI-2 plant be viewed as a full scale, integrated test facility.

In this regard three separate, interrelated tasks have been undertaken.

These tasks are:

(a) development of the international TMI-2 standard problem, (b) development of a set of comprehensive data bases, and (c) analysis and qualification of the accident data.

The standard problem requires that a data base exist providing sufficient data (e. g. plant configuration, event sequence, initial and boundary conditions and etc.) to perform a benchmarking calculation. This in turn requires the data in the data bases be qualified with appropriate r

uncertainties.

6-13

The standard problem package has been released to those organizations participating in the standard problem. There will be a work shop in January 1987 to initiate the calculations. As part of the package a demonstration calculation, to 174 minutes after turbine trip, was included to show that a calculation is possible.

l The Sequence of Events, Initial and Boundary Conditions, and Plant Configuration data bases have been developed during the past year to support the standard problem, and are an integral part of the standard problem package. The Time Series data base has been developed to contain all other time dependent thermal hydraulic data. All except the Plant Configuration data base are available as interactive IBM compatible

' microcomputer data bases in the SAGE data base management system.

The data, in the data bases, has been provided as best estimate values with the uncertainties determined by the data qualification and analysis effort. However, there are a number of significant data gaps and data which have been questioned. First the make-up and letdown flow and the auxiliary feedwater flow were not measured, these data are significant to i

primary mass inventory and primary to. secondary heat transfer. Efforts to-compute auxiliary feed flow are in progress as part of a steam generator thermal hydraulic analysis.

J During FY-87 the data bases will be completed with the development of the fission product data base and addition of a manipulative capability in the data bases. The data for the Time Series data base will be qualified and data analyses completed.

l 6-14

n. -- . _ _ - .

EFFECTIVENESS OF THE BWR MARK I SECONDARY CONTAINMENT IN SEVERE ACCIDENT MITIGATION S. R. Greene S. A. Hodge Severe Accident Sequence Analysis (SASA) Program Oak Ridge National Laboratory This paper presents the results of calculations performed with the CONTAIN code to assess the effectiveness of the BWR Mark I secondary containment in mitigating fission product transport under severe accident conditions in which the primary containment pressure boundary has f ailed. Many features of the secondary con-tainment are plant-specific; the Browns Ferry and Peach Bottom de-signs are discussed in this study. It is demonstrated that the retention of aerosols by the secondary containment structures would be significant, particularly in the Browns Ferry design which has the two important advantages of 1) additional internal flow paths to promote mixing, and 2) fire protection system atmo-sphere sprays that would be actuated under severe accident condi-tions.

7-1

CONTAINMENT VENTING AS A BWR ATWS MITIGATION TECHNIQUE R. M. Harrington

. Severe Accident Sequence Analysis (SASA) Program Oak Ridge National Laboratory This paper describes the results of analyses performed to assess the effectiveness of wetwell venting as a severe accident mitigation technique for potential Anticipated Transient Without Scram (ATWS) events at BWR plants with the Mark I containment design. ATWS has been identified by the Accident Sequences Evaluation Program (ASEP) to be one of the two accident sequences that dominate the BWR risk of core melt.

The model plants used for this study are Browns Ferry and Peach Bottom. Dif ferences in plant design that must be considered in severe  !

accident studies are identified.

In the ATWS accident sequence, a great deal of energy is deposited in the pressure suppression pool during the period before permanent core uncovery or core damage of any kind. As the suppression pool temperature increases, so does the associated saturation pressure; ' the primary containment is pressurized by the resultant suppression pool evaporation and steaming. Should the primary containment fail by over-pressure, the consequent blowdown into the secondary containment might incapacitate the reactor vessel injection systems therein; without con-tinued injection, the core would be uncovered.

Previous work has demonstrated that no additional operator action, including wetwell venting, is necessary if the injection of sodium pentaborate solution is initiated within 15 minutes. If, however, an ATWS event should be compounded by loss of the standby liquid control (SLC) system, then plant survival would depend on measures taken to delay core damage until the SLC system could be repaired. In this study, the criteria for success of wetwell venting are that without l venting, severe core damage would occur during the first three hours of the accident sequence and that with venting, the onset of core damage is delayed beyond the three-hour point. Three hours are believed sufficient to permit SLC system repair.

Wetwell venting to atmosphere under ATWS conditions has very undesirable side effects. The pressure suppression pool would become saturated and all systems that pump from the pool would be threatened by loss of their necessary net positive suction head. Personnel access to the secondary containment would be sacrificed because the vented steam would be released into the lower levels of the reactor building.

Examination of the MSIV-closure ATWS accident sequence scenarios from the standpoint of the effect of venting indicates that wetwell venting does not significantly affect the outcome in most cases. In one case, venting is counterproductive, causing core melt within the three-hour time frame. In one case, venting would be beneficial in delaying core melt.

7-3

_ _ _ _ _ _ _ _ O

It is a conclusion of this study that wetwell venting should not be automatically required by symptom-oriented emergency procedures for BWRs. The integrity of the containment should not be intentionally violated unless there is a clear understanding of the accident sequence i in progress and the effect that venting would have on the operating pumping systems.

l 1

.h i

1 i

7-4

EFFECTS OF LATERAL SEPARATION OF OXIDE AND METAL COMPONENTS OF CORE DEBRIS ON TELE BWR MK I CONTAINMENT DRYWELL FLOOR C. R. liyman C. F. Weber Severe Accident Sequence Analysis Program Oak Ridge National Laboratory In evaluating core debris / concrete interactions for the BWR MK I containment design, it has been common practice to assume that the core debris released at reactor vessel breach is homogeneous and of low viscosity, so that it flows through the pedestal doorway and spreads in a radially uniform fashion over the entire drywell floor. Recent calculations with improved codes indicate, however, that the initial temperature of the released debris would be such that the metal components (Zr, Fe, Ni, Cr) are completely molten while the oxides (UO2 '

Zr02 , Fe0) are completely frozen. Thus, most of the frozen oxide would remain within the reactor pedestal while the molten metals flowed through the pedestal doorway and spread over the annular region of the drywell floor between the pedestal and the drywell shell. It now becomes necessary to assess the effect of the lateral separation of the debris components.

Calculations of core debris / concrete reaction processes have been examined for two dif ferent configurations of BWR core debris. Results for the standard model involving radially uniform, axially separated oxide and metal layers are compared with results for a model that assumes significant radial separation of oxides and metals.

Results indicate that larger releases of combustible gases (C0 and 11 2 ) are produced in the uniformly distributed case while a greater CO2 and 11 2 0 release and a greater total gas release are predicted for the radially separated case. The combustible gas release is important in determining the response of the secondary containment, where hydrogen or carbon monoxide burns might occur.

Of significance for acrosol and fission product release are debris temperatures and the total gas flow from the debris surface. Oxide layer releases occur only during the first few hours of the transient, while debris temperatures exceed 1900K. Although the total cumulative gas release for the uniform case is lower, the flows during the initial high-temperature period are greater, thus releasing more oxidic fission products relative to the separated case. Regardless of the case, all cesium and iodine is released completely from the core debris.

The results of this work demonstrate that the effects of lateral spreading of the core debris over the drywell floor do significantly affect the timing, composition, and magnitude of the release of gases, aerosols, and fission products. Lateral spreading changes the location of the metallic debris relative to the oxidic debris and its effect on temperatures, gas flows, and chemistry.

7-5

Small Break LOCA Mitigation for Bellefonte Paul D. Bayless and Charles A. Dobbe Idaho National Engineering Laboratory A series of small break loss-of-coolant accidents (LOCAs) was analyzed for the Bellefonte Nuclear Plant as part of the Severe Accident Sequence Analysis Program. The transients were all initiated by a 2-in.

l diameter break in a reactor coolant pump discharge line. Variations on

! available systems and on operator actions were investigated.

l Bellefonte is a pressurized water reactor designed by Babcock and l Wilcox and being built by the Tennessee Valley Authority. It has a rated core thermal power of 3600 MW, raised coolant loops, and two once-through steam generators. The transients were calculated using the RELAP5/M002 and SCDAP/RELAP5 computer codes.

The first transient calculated assumed that there were no emergency core cooling (ECC) systems available and that there was no auxiliary l feedwater available to the steam generators. This sequence led to core damage (the onset of calculated cladding oxidation) at about 2760 s.

The base transient was the SpD sequence, which is a small break LOCA with no high pressure injectTon (HPI). Auxiliary feedwater, low pressure injection, and core flood tanks are available, but no operator actions are considered. Using RELAP5, this transient was found to lead to core damage at about 3600 s after transient initiation. The SCDAP/RELAPS computer code was then used to calculate the transient from initiation through core damage. This calculation stopped at 11,900 s, with the cladding temperatures within 20*F of the melting temperature of I zirconium oxide (4892*F). The top 70% of the fuel rod cladding was I

completely oxidized; because of the slow heatup rate, no relocation of molten zircaloy was observed. The system pressure remained above the initiation pressure of the available ECC systems throughout the transient, so that there was no flow into the reactor coolant system (RCS).

The next transient calculated assumed that one high pressure injection pump was available, but that no other ECC systems were and there was no auxiliary feedwater. This transient resulted in a stable long-term cooling mode being established by 10,680 s with the core covered and the liquid level in the reactor vessel increasing. The HPI flow was feeding the RCS while the break was bleeding it. The system pressure was about 650 psia and decreasing at the end of the calculation (15,000 s). There was sufficient water in the borated water storage tank to maintain the HPI flow for another 15 to 20 hours2.314815e-4 days <br />0.00556 hours <br />3.306878e-5 weeks <br />7.61e-6 months <br />.

These RELAPS calculations showed that the availability of a single HPI pump was sufficient to prevent core damage for the small breaks. To determine if core damage could be prevented without high pressure injection, operator actions were taken into account.

Two operator action cases for the S D2 sequence were evaluated using the SCDAP/RELAPS code. The first case followed the Anticipated Transient Operating Guidelines (ATOG) for the Bellefonte plant. The first part of the transient was nearly identical to the base case, except that the i

7-7  !

pressurizer spray line block valve was closed at 500 s. At about 800 s, a controlled cooldown of the RCS (50*F/hr) was begun. The cooldown was controlled by depressurizing the steam generators. When superheat was detected at the core outlet, the Inadequate Core Cooling procedure was followed. This involved a more rapid depressurization of the steam generators. This action was effective in returning the steam temperature at the core outlet to the saturation temperature, so that the controlled cooldown procedure was re-entered. The operational control of the plant cooldown alternated between these two procedures until the pressure was low enough (215 psia) that the low pressure injection pumps began delivering liquid to the RCS (at 7886 s). The core was quickly covered with liquid, and the plant was in a stable, long-term decay heat removal configuration. During the transient, the peak fuel rod cladding temperature was lower than the steady state temperature.

The second operator action case was the same as the first except that the core flood tanks were assumed to be unavailable. The transient response was similar to that of the first operator action case, with the plant alternating between the two cooldown rates. The absence of the core flood tanks resulted in more extended temperature excursions and a faster depressurization of the RCS, such that the low pressure injection began at about 7420 s. Although the fuel rod cladding temperatures were higher than in the first operator action case, the peak temperature was only about 65aF above the steady state temperature.

The small break sequences investigated showed that without operator actions, the flow from one HPI pump is necessary and sufficient to prevent core damage. Even without HPI, simple operator actions can effectively mitigate the transient and bring the plant to a stable condition removing the decay heat.

Work supported by the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research, under DOE Contract No. DE-AC07-761001570.

7-8

Integrated SCDAP/RELAPS Analysis of a BWR High Pressure Boiloff Rosanna Chambers Idaho National Engineering Laboratory During an anticipated transient without scram (ATWS) in a boiling water reactor (BWR), high pressure emergency core cooling (ECC) can be lost.

Without mitigating operator actions, the situation could degrade to a high pressure boiloff. The only mass entering the reactor pressure vessel (RPV) would be the control rod drive (CRD) hydraulic fluid; not enough to maintain adequate core cooling. Steam generated in the core would be relieved through the safety relief valves (SRVs). A simulation of such a postulated high pressure boiloff was performed using the SCDAP/RELAPS computer code. The purpose for using SCDAP/RELAP5 was to study the effects of coupling the reactor core behavior with the system thermal-hydraulic behavior. A model of the Browns Ferry Nuclear Plant Unit 1 was used for the analysis. The work was performed for the Severe Accident Sequence Analysis program.

The SCDAP/RELAP5 simulation reported here was initialized from previous RELAPS calculations and begun just prior to heatup of the fuel rod cladding. Since emergency injection systems and pumps were assumed to be no longer operational, only the RPV was modeled. The only flow into the RPV was 112 gpm of CRD fluid.

To study multi-channel effects on hydrogen production and flow distribution, four parallel core regions were modeled. In each region, fuel rods, water rods, and fuel channels were modeled. The four regions were chosen to best match axial power distributions and radial power peaking factors from the Browns Ferry core at end of cycle. Each region contained eight calculational cells in the axial direction. In the center region, 116 fuel bundles were modeled with a radial power peaking factor of 1.137. In regions 2, 3, and 4; 168, 292, and 188 bundles were modeled with radial power peaking factors of 1.175, 1.092, and 0.616, respectively. The maximum axial peaking factor of 1.407 occurred in the second region.

Calculated damage to the core consisted of cladding failure, heavy oxidation, and relocation of molten core material. No cladding ballooning or change in flow distribution was calculated. The cladding failed, but not until the Zr02 melting temperature (4892*F) was reached 2126 s after transient initiation. The initial calculated heatup rates in the top half of the core ranged from 1.0 to 1.4*F/s. That slow heatup rate allowed a thick zirconium oxide shell to develop on the rod and channel surfaces, slowing oxidation and preventing earlier cladding failure and relocation of molten core material.

l 7-9

The importance of multi-channel, integrated thermal-hydraulic and core behavior calculations was demonstrated by this boiloff simulation. When the SRVs opened, steam was pulled through the core region and oxidation increased. When the SRVs closed, the flow decreased and the zircaloy oxidation reaction became steam limited, causing the oxidation to slow.

These changes in flow caused spikes in the core hydrogen generation rate l and variations in the heatup rate. The center of the core was heavily damaged, but the periphery remained intact. Oxidation, melting, and relocation of molten material were calculated for the inner three-fourths of the core, but very little oxidation and no melting was calculated on the outside of the core.

Work supported by the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research, under DOE Contract No. DE-AC07-76ID01570.

i l

7-10

PROBLEMS WITH INDUSTRY'S IMPLEMENTATION OF REQUIREMENTS FOR UPGRADING EMERGENCY OPERATING PROCEDURES Jame: P. Bongarra, Jr.

United States Nuclear Regulatory Commission Office Of Nuclear Reactor Regulation Division Of ;luman Factors Technology Washington, DC 20555 Following the Three Mile Island (TMI) accident, the NRC issued two documents in 1980, the "TMI Action Plan" and the " Clarification of TMI Action Plan Requirements" which required the industry to reanalyze transients and accidents and upgrade emergency operating procedures (E0Ps). The NRC staff also developed a long-term plan that integrates and expands the efforts at writing, reviewing, and monitoring plant procedures. As an initial step in developing a plan to upgrade procedures, the NRC published " Guidelines for the Preparation of Emergency Operating Procedures" in 1982 which provides guidance for preparing and implementing E0Ps and describes the use of a " Procedures Generation Package" (PGP) to prepare E0Ps. In the PGP, a licensee describes its program for developing and implementing upgraded E0Ps by addressing four areas: the technical basis of the procedures; preparing the text and visual aids for E0Ps (referred to as the writer's guide); verifying / validating the E0Ps; and training operators to use the E0Ps. The NRC staff has emphasized evaluation of PGPs rather than E0Ps because it is more important to ensure that the process for producing procedures and their technical basis is sound and well documented than to perform a one-time review of E0Ps with no assurance that future revisions of E0Ps will be technically adequate and consistent with existing E0Ps.

The NRC staff has collected information on the quality of the upgraded E0Ps throughout the industry. This information has come from several sources including: (1) comprehensive audits of E0P upgrade programs at four plants, (2) review of PGPs for 48 of 89 plants submitting PGPs to the NRC staff, (3) the review of procedures as a part of operating event analyses, (4) reviews of detailed control room design review program plans and summary reports, (5) examiners' experiences during operator license examinations, (6) industry comments on procedure-related research efforts, and (7) other interactions with licensees. The assessment of this information indicates that significant deficiencies in implementing the E0P upgrade programs exist. Specifically:

A. Licensees have not followed commitments they made in their PGPs.

That is, the E0Ps, the Verification and Validation program (V&V),

and/or the operator training program was not found to adhere adequately to the PGP at any of the four plants audited.

B. Undocumented deviations from the generic technical guidelines have been founa at all four plants audited. Deviations are expected, but they should be identified, and a technical justification provided in 7-11

the PGP. In addition, programs have been found deficient where deviations were identified after the PGP had been submitted, reviewed, and approved by the NRC.

C. Failures to appropriately adapt or deviate from the Owners Groups' generic procedure guidelines have been found at all of the plants audited. Some licensees have literally adopted the generic guidelines as E0Ps. The NRC staff review and approval of the generic technical guidelines has been limited to assessing their technical suitability as guidelines not as E0Ps for typical plants. The guidelines have not received human factors review to assure that, as procedures, they are likely to minimize confusion or error in use, nor have they been reviewed for their suitability to each individual plant.

D. Failures to adhere to the PGP writer's guide have been found at all plants audited. The purpose of following an approved writer's guide is to assure that potentially confusing or error-prone writing styles, nonmenclature, formats, conventions, or inconsistencies are avoided.

E. Failures to adhere to PGP commitments on the V&V of E0Ps have been found at all plants audited. The implementation of V&V programs has not been successful in identifying and correcting the types of deficiencies cited at any of the plants audited.

F. Training programs at some plants audited have been found deficient in areas such as (a) providing adequate instruction in the philosophy of the function orientation of upgraded E0Ps, (b) training all operators on all E0Ps before they are implemented, and (c) evaluating operator knowledge and performance after they have received training on the upgraded E0Ps.

i While the above findings do not necessarily represent the situation at any single plant, their occurrence across the plants contacted suggests that the overall quality of E0Ps may not meet current requirements. To address these potentially safety significant deficiencies, the NRC staff is (1) issuing an Information Notice to alert licensees to the deficiencies, (2) continuing the audit program to determine the scope, safety significance, and probable cause of these deficiencies, and (3) conducting inspections at all plants to evaluate the implementation of the licensees' commitmer.ts to develop and implement upgraded E0Ps.

The paper will elaborate on the historical background of the NRC's requirements to upgrade E0Ps and the industry's efforts to meet these requirements. The paper will also discuss the NRC's future direction for promoting the development and implementation of effective E0Ps.

l 7-12

TWO-PHASE FLOW MEASUREMENT IN THE UPPER PLENUM 0F A PWR DURING REFLOOD i

l P. Griffith, Prof of Mechanical Engineering K. Gawlik, Research Assistant Massachusetts Institute of Technology i

The 20-3D reflood hydraulics experiment and analytical program being performed in Germany, Japan and the United States includes flow measurement at the upper tie plate. In order to determine the flow through the upper tie plate, either up or down and for either phase, a set of instruments has been developed including a turbine meter, a collapsed liquid level measure-ment, a drag body, a pressure difference across the tie plate, and several temperature readings. These measurements are calibrated and used to infer the flow rate through the tie plate.

This paper describes an air-water calibration and flow rate calculation algorithm suitable for calculating the flow of each phase through the tie plate. The extrapolation to steam water conditions is discussed. The accuracy in the different regimes is determined.

8-1

DRYOUT FRONT MODELING FOR R0D BUNDLES Peter Griffith, Prof. of Mechanical Engineering Jorge A. Mohamed Former Student Daniel Brown, Former Research Assistant Massachusetts Institute of Technology The dryout zone is defined as the region between the calculated liquid level and the place where the heater rod or tube dries out. That place is called the dryout front. Depending on the void fraction and superficial vapor velocity, the dryout zone can contribute appreciably to the overall heat transfer because heat transfer coefficient in that region it very high.

The best values for the drift flux model constants are chosen for three different tube bundle geometries. These are used to calculate the water level so the extent of the dryout zone can be determined. Steam-water data from three different experiments, a blowdown experiment, essentially steady state experiment, and a high pressure steady state are used to develop a model to predict the extent of the dryout zone.

It is found that the dryout front and liquid level are essentially identical as long as the void fraction at the front is less than 45%. Above that the dryout zone extent is proportional to the superficial vapor velocity rising to about 1 foot when the vapor velocity is six feet per second.

8-3

. I

Summary Steam Separator Modelling for Various Nuclear Reactor Accidents Chan Young Paik, Research Assistant Greg Mullen, Resarch Assistant Christoph Knoess, Visiting Engineer Peter Griffith, Professor of Mechanical Engineering Massachusetts Institute of Technology This is a progress report on the MIT program on separator modelling.

In a pressurized water reactor steam generator, a moisture separator is used to separate steam and liquid to ensure that essentially dry steam is supplied to the turbine. During a steam line break or combined steam line break plus tube rupture, a number of phenomena can occur in the separator which have no counterparts during steady-state operation. How the separator will perform under these circumstances is very important because it can substantially change water inventory in the system. A steam line break combined with a tube rupture provides a leadage path through and out of the secondary system for radioactive materials contained in the primary. As the radioactivity is primarily in the form of iodine, the amount of liquid (and thus iodine) released to the atmosphere depends very strongly on how the separator performs. This project has as its goal the development of a simple separator model which can be applied to a variety of steam generators for of f-design conditions.

Experiments were performed using air and water on three different types of centrifugal separators (a cyclone as a generic separator, a Westinghouse type stationary swirl vane separator, and a Combustion Engineering type stationary swirl vane separator). Experiments were also performed in the MIT blowdown rig with and without a cyclone separator using steam and water. All air-water experiments were performed in transparent models of the separator system under steady-state conditions to show how these separators perform. The blowdown experiments were performed ander transient conditions. Thus, these experiments cover most of the geometric differences in present PWR separators and an extreme range of flow conditions from slow quasi-steady-state transients to fast transients.

The air-water cyclone separator system has three stages of separation:

first a cyclone, then a gravity separator, and finally a chevron pilte separator. The other air-water separator systems have only a centrifugal separator to isolate the effect of the centrifugal separation while the blowdown experiment has a cyclone and a gravity separator.

All three centrifugal separators appear to perform well at flow rates well above the design values as long as the downcomer water level is not high. High downcomer water levels rather than high flow rates appear to be the primary cause of degraded performance. Appreciable carry-over from the separator section of a steam generator occurs when the drain lines from the three stages of separation (the centrifugal, gravity, and secondary) are unable to carry of f the liquid flow. A scenario leading to the steam separator failure and simple models which incoporate the principal geometric and flow variables are provided. The models developed vary in degree of complexity. The first is a single switch such that the separator is on or off depending on a critical downcomer water level which is a function of inlet conditions. The second is a more refined model that includes all three stages of separation and can calculate each separator's l

8-5 l

performance for given boundary conditions.

Tests in the blowdown rig showed that for rapid transients the separators did not make a lot of difference. Oddly enough, when the separators did make a difference, they tended to cause more water to be expelled. What happened is, early in the transient, the separator was overwhelmed by the surge in the water level immediately following the break. Subsequent high steam flow then carried water trapped above the separator out for a considerable time after the water level below the separator had receeded below the separator inlet.

Without a separator the rapid fall of the water level af ter the initial surge, results in negligible carry over for most of the transient.

Such carry-over as is observed occurs during the initial surge.

The work on this project will be completed in thesis form by December 1986 and will appear as a report in May 1987.

8-6

l l Measurement and Modeling of Void and Velocity Profiles in Curved Channels l

l l Y.Y. Hsu & W.K. Lin l

l University of Maryland, College Park, MD This project is a part of UMCP 2x4 Loop program, with the objective of obtaining more fundamental data on two-phase flow characteristics in a hot-leg near U-bend during two-phase flow natural circulation.

Two-phase (air / water) measurements of the phase distribution phenomena were made in a transparent model of a PWR hot leg by pitot tubes and conductivity probes at room temperature.

Test conditions were selected to cover a range of gas and liquid superficial velocities (0<Jo<10 cm/s, 0<J <40 L cm/s) expected to occur in a prototypical reactor geometry during a

, small break (SB) loss of coolant accident (LOCA). For compari-l son, tests at high gas superficial velocities (10 cm/s<JG <150 cm/s) were also performed in which the void fraction were obtained with the three DP-units. Test results include average void fraction, local pressure profile, local velocity profile and the local void fraction profile. Data obtained are found to be in general good agreement with the drift-flux model of Zuber for vertical pipes. For the U-bend section, the local velocity as well as the local void profile can be measured within the bubbly flow regime, the Co value which is affected by the curvature can 8-7

i

then be calculated by semi-empirical method. The Co obtained f appears to be in the range of 1.0 to 1.2, which is lower than the

, value of 1.2 to 1 5 as derived by Zuber for the vertical pipe.

f-The results also showed that both the profiles are affected by i

i the curvature channel. In addition, new expressions are proposed i

for velocity and void profiles. Integration of these profiles yield the global parameters such as Co which agree with those ,

4 globally measured quantities. )

1 1

i A " Weir Model" was developed to relate the liquid flow rate with the level of water in the n-bend. To test this weir model, data were taken from three different test loops, one with a 3 inch (7 62 cm) diameter plastic pipe, and the other with a 3 5 t

inch (8.89 cm) hot leg pipe on the 2x4 B&W Simulation Loop in UMCP. The third data source is Dr. Ishii's experiment movie film with 4 inch (10.16 cm) glass pipe loop. Resulta appear to be in good agreement with the theoretical calculation as long as it ,

j is in the single liquid phase-or the bubbly flow regime. This j model can be used as a means of determining liquid flow rate from liquid height in the U-bend.

This study also presents a new technique developed to

) estimate the local average void fraction and the liquid mean superfical velocity profile by using the combined conductivity-pitot tube probes. Since this is the new attempt to measure the  ;

j velocity and the void at same time and the same location, some limitation can't be avoided, and reliable measurements are limited to a certain range of conditions.

i 8-8 l

Simulation Experiments for Hot-Leg U-Bend Two-phase Flow Phenomena M. Ishii, J. T. Hsu and G. Lambert Reactor Analys,is and Safety Division Argonne National Laboratory The major issues addressed under this study are the interruption and reestablishment of the two-phase flow natural circulation in a loop. The research is motivated by the need to understand some important aspects of a small break loss of coolant accident in a Babcock & Wilcox (B&W) light water reactor. The two-phase natural circulation, termination and reestablishment have been studied using a nitrogen-water simulation loop with complete flow visualization capability. The loop design is based on the two-phase flow scaling criteria developed under the program. Furthermore, Freon 113 boiling  !

and condensation icop with similar flow visualization capability has been built to study the effect of phase change and property effects on these phenomena.

The main focus is the simulation of a prototype system as well as the evaluation of various scale distortions which arise in typical scale models or in integral test facilities such as Multi-loop Integral System Test (MIST) and University of Maryland facilities. For example, it is noted that in the )

prototype system, the diameter of the hot legs is 90 cm and the length to diameter ratio relatively small at about 20. This indicates that the two-phase flow in the hot leg can be quite different from the standard small scale experiments.

The phenomena were studied experimentally by focusing on the hot leg U- l bend phase separation, two-phase flow regimes, void fraction distribution, natural circulation rate, flow termination and various geometrical ef fects.

Special attention has been paid to the simulation and scale distortion effects. The effects of parameters such as the gas volumetric flux, thermal center, loop frictional resistance, inlet section geometry and U-bend curvature have been studied by various experimental measurements and flow visualization. The visual observation presented clear understanding of the hydrodynamic phenomena in addition to detailed data on the flow, pressure drop and void fractions. The major results from the air-water simulation experiments are summarized below:

The induced liquid circulation rate increased with increased gas flux.

There is a certain minimum gas flux below which the two-phase natural circulation is terminated.

The flow termination is determined by the head balance between the hot leg and downcomer. Thus the thermal center (separator water level) and the hot leg void distribution are the key parameters for the flow termination. A simple criterion for predicting this phenomena has been developed.

8-9

. Temporal voiding and intermittent U-bend liquid flow have been encoun-tered in slug and churn flow. However, these do not lead to flow termination. Even under this condition, the natural circulation flow rate stays very steady due to the liquid inertia in the loop.

. Near the flow termination point, large amplitude flow oscillations have been encountered. The amplitude can be as high as the average flow. The instability appears to be related to the kinematic wave propagation and has a period of 10 to 25 sec. This instability is important in terms of safety analyses, because it may lead to loop to loop oscillations or flow excursions.

. The inlet geometry to the hot leg has significant effects on the hot leg two-phase flow regimes and void distribution. With a horizontal section and 90* elbow, only cap bubble and slug flow are possible due to slugging in a small diameter tube (less than 10 cm diameter). On the other hand, in prototype system the regime should be mostly bubbly flow.

A preliminary experiment in Freon 113 boiling and condensation loop has indicated that the phase change can be quite important for two-phase natural circulations. The most significant effect seems to be the intermittent flashing of liquid in the upper part of the hot leg. It was observed that the j

~

hydrostatic head decrease was sufficient to superheat the liquid. The flashing of ten lead to a relatively high void fraction and restarted a two-phase natural circulation for a short period of time. In general, the hydrodynamic behaviors were much more unstable in the Freon 113 loop than the adiabatic loop.

I i

4 i

1 l

8-10 I

THERMAL MIXING TESTS IN A SEMIANNULAR DOWNCOMER WITH INTERACTING FLOWS FROM COLD LEGS by Harri Tuomisto Imatran Voima Oy Helsinki, Finland An experimental facility for thermal mixing tests in a semiannu-lar downcomer and cold legs was constructed at the Hydraulic Laboratory of Imatran Voima Oy (IVO) in 1983. The facility was designed to model the Loviisa VVER-440 reactor, and an experi-mental program of about 50 tests was performed /1/.

The test facility was redesigned and modified for new test series carried out by mutual agreement of the pressurized thermal shock (PTS) information exchange between USNRC and IVO. Three cold legs and high-pressure injection (HPI) geometry were changed to model typical features of Western type PWRs.

The facility operates at atmospheric pressure with loop and HPI flows from different cold legs in the area of interest to PTS.

Transparent materials were used to allow flow visualization during the tests. The choice of transparent materials limit the upper tenperature to 750C. The full buoyancy effect was induced by salt addition and the IIPI temperature was used as a tracer.

The test matrix consists of 20 tests. The varied parameters are flow rates, and the number and configuration of cold legs with HPI and loop flows. The test conditions can be divided into five different groups:

Group 1. Interaction of plumes caused by IIPI into two cold legs. The loop flow is introduced to the cold leg without IIPI.

Group 2.  !!PI into one cold leg with the loop flow in another.

Group 3. Tests with zero loop flow in all cold legs.

Group 4. Tests with the loop flow temperature de-creasing stepwise from 750C to 250C.

I Group 5. Tests in which the plume interaction was studied with both loop and IIPI flow intro-duced to two cold legs.

8-11

l The performed tests confirm good mixing in a semiannular down-comer with interacting buoyant plumes. Interacting plumes often show a tendency to join, especially if loop flow is maintained in another loop. When loop flows are very low or zero, inter-acting plumes are unstable, thus increasing mixing. In case of unstable behaviour, plumes are not necessarily vertical all the time, but cold plumes can even turn to the side or upwards depending on the recirculation flows in the downcomer and on the fact, how strong the plumes are.

These tests supplement the thermal mixing data base with multiloop operation experiments. The test results have been utilized to assess REMIX-code that has been developed for thermal mixing calculations /2/. This assessment work will be presented by Prof. Theofanous in this Session.

References:

/1/ H.Tuomisto, Y.Hyt6nen and P.Mustonen, "Two-fifths Scale Thermal Mixing Tests in a Semiannular Downcomer and Cold Legs", Proceedings of the International ANS/ ENS Topical Meeting on Thermal Reactor Safety, San Diego, California, 2-6 Feb 1986, Paper XVIII.5.

/2/ K.Iyer, H.P.Nourbakhsh and T.G.Theofanous, " REMIX: A Computer Program for Temperature Transients Due to HPI after Interruption of Natural Circulation", Purdue University, USNRC Report NUREG/CR-3701, February 1986.

8-12

Scaling of Thermal Mixing Phenomena from 1/5 to Full Scale Test Facilities by T.G. Theofanous, K. Iyer, and E. Shabana University of California, Santa Barbara Thermal mixing in relation to Pressurized Thermal Shock has been examined experimentally at various scales. In order of size the test facilities include: CREARE (1/5), IVO (2/5), PURDUE (1/2),

CREARE (1/2), and HDR.

The HDR experiments were run at fully prototypic pressure and temperature conditions. This is a full-scale reactor, however, its cold leg diameter and downcomer gap dimensions are nearly one-half of those in full-scale commercial PWRs. This experimental effort has just culminated with the UPTF tests. The UPTF is a full geometric replica of a PWR.

However, 200 C. As because such it of its pressure rating, it can operate only up to composing a full-scale represents a useful complement to the HDR in behavior.

The UPTF tests were run at 5, 10, 20, 40, and 70 kg/s injection rates.

Because of a peculiar injection nozzle which was introduced to eliminate excessive stratification in the cold leg due to the horizontal injection port available the tests were primarily aimed for downcomer behavior. However, pretest calculations indicated that at the higher flow rates the flooding-limited mixing regime of the code NEWMIX paper

[1] should be approached. The purpose of this is to compare these pre-test calculations with the experimental data and to discuss the downcomer behavior. With this discussion (taken together with that presented in last year's conference) the assessment of REMIX (2] and NEWMIX as scaling tools for this problem is complete.

Peferences

1. Iyer, K.,

and T.G. Theofanous, " Flooding Limited Thermal Mixing

- The Case of High-FR Injection," Proceedings of the Third International Topical Meeting on Reactor Thermal Hydraulics, Newport, RI, October 15-18, 1985.

2. Iyer, K.,

and T.G. Theofanous, " Decay of Buoyancy Driven Stratified Layers with Applications to PTS: Reactor Predictions," ANS Proceedings of the 1985 National Heat Transfer Conference, Denver, CO, August 4-7, 1985, 358-371.

8-13

COMPONENT FRAGlLITY RESEARCH PROGHAM:

PH10RITIZAT10N AND DEMONSTHATION TESTING #

G.S. Holman and C.K. Chou Lawrence Livermore National Laboratory University of California Livermore, California / U.S.A.

SUMMARY

Current probabilistic risk assessment (PRA) methods for nuclear power plants utilize seismic fragilities which are mostly based on limited test data and on engineering judgement. As part of the Nuclear Regulatory Commission Component Fragility Research Program (CFRP), the Lawrence Livermore National Laboratory (LLNL) developed procedures for fragilities testing. We applied these procedures to evaluate actual seismic capacities of motor control center (MCC) electrical devices, to develop seismic fragilities suitable for PRA application, and to suggest various methods of improving MCC seismic performance.

When used for probabilistic risk assessment, component " fragility" relates the probability of failure conditioned upon occurrence of some level of forcing or response function (e.g. , seismic acceleration).

Failure probability is described by " fragility curves" plotted at vari-ous levels of statistical confidence. These curves include a " median" function, representing the "best estimate" fragility, bounded by curves which characterize uncertainty in the median function. For certain practical applications, the lower bound curve can be used to estimate a "HCLPF"j(ligh Confidence Low Probability of Failure) capacity for the component.

In principle, empirically developing a statistically meanirgful seismic fragility for any component would require subjecting a large population (e.g. , hundreds or thousands) of identical samplan ta successively higher levels of acceleration while recording the dis-tribution of failures acceleration level. (however " failure" is defined) as a function of Practical constraints on time and resources clearly make this infeasible for a single component under well-defined load conditions, let alone for the effectively infinite combinations and permutations of component type, manufacturer, mounting, and loading conditions that could be identified for actual nuclear power plants.

However, useful yet cost-effective fragilities testing can be conducted within these constraints the objective is not to explicitly develop

" generic" fragilities broadly applicable to wide ranges of components, "This work was supported by the United States Nuclear Regulatory Commission under a Memorandum of Understanding with the United States Department of Energy."

9-1

. - - - _ .. _. - . . . _ . == _

J but rather to enhance understanding of how certain components fail

(" failure modes"), what important factors affect component performance, j and what their relative influence is. Parametric testing offers one j method of providing the fragility analyst with a firm technical basis i for making the inevitable judgements that must be made when developing specific fragility descriptions for specific components, particularly when considering seismically " rugged" components for which a limited or i " lower bound" fragility description may suffice for PRA purposes.

Through such sensitivity tests, for example, LLNL systematically investigated how varying the cabinet mounting configuration affected MCC functional and structural behavior. For each of the internal elec-trical devices (atarters and relays) the tests yielded actual failure data as well as a basis for developing fragility curves (referenced to in-cabinet zero period accelerations) with confidence limits from which lower bound (i.e., HCLPF) seismic capacities could be estimated. The tests also suggested possible hardware modifications to enhance seismic capacity, such as cabinet top bracing or the use of reed- rather than armature-type relays, and indicated that spectral acceleration might be a more appropriate response parameter for characterizing device behavior.

Such " sensitivity tests" can also improve the technical basis for interpreting and applying data from other sources, such as when using qualification data to assess actual component capacity. As another CFRP activity, LLNL developed fragility curves for five components based on high-level seismic qualification data. Although not true

" fragility" data, these qualification data provided useful information

~

on component behavior under conditions exceeding any anticipated change in design basis earthquakes for U.S. plant sites. Here we assumed that qualification levels represented the HCLPF capacity of each component.

As opposed to fragility tests, where the median seismic capacity is determined explicitly, in this case the median capacity can only be inferred through judgement regarding the uncertainty parameters that define the HCLPF capacity.

As another part of its CFRP activities, LLNL has identified other candidate components for " Phase II" CFRP testing. The presentation and paper will discuss how we propose to similarly apply test data to inves-tigate the fragility of these components.

9-2

Synthesizing Seismic Fragility of Components by use of Existing Data K.K. Bandyopadhyay and C.H. Hofmayer Brookhaven National Laboratory Upton, NY 11973 -

Summary i

Brookhaven National Laboratory has completed Phase I of the Component Fragility Program and is now perigtming Phase II. i been published in NUREG/CR-4659.1J In both Phases, The results of Phase I have existing test data for various models are utilized to determine the seismic fragilities of different equipment categories.

This represents the first large scale attempt to assemble, compile and interpret the very large heretofore fragmented data base.

In Phase I, a methodology has been established to compile the test data for variations of testing methods, vibration inputs, damping values, etc.

Test response spectra have been used as a measure of the test vibration inputs.

Fragility data have been collected and stored in a computurized data '

bank at BNL for many electrical and control equipment pieces. The data collected to-date for Switchgears and Motor Control Centers (MCCs) have been considered adequate to generically represent a majority of the respective equipment family. As such, a detailed fragility analysis of these two types of equipment has been performed. The fragility levels with the associated malfunctions have been included in the analysis. The front-to-back fundamental natural frequency of these equipment families varies from 5 Hz to 15 Hz primarily depending on the model, manufacturer, mounting configurations and modifications of the test specimen.

The Phase I data indicate ZPA values of 1.1 g, 1.5 g and 1.8 g as the respective lower-bound fragility levels of MCCs, low Voltage Swchgears and Medium Voltage Switchgears. These values suggest that the resistance of these items might have been understated in PRAs and margin studies performed in the past for some specific applications.

Products from four major manufacturers have been included in this study. The following observations have been made in the MCC and Switchgear analyses:

o The overall fragility of an equipment assembly is limited by malfunction of devices. Some of these devices and their failure modes are as follows:

a) contactors - chattering b) motor starters - chattering, dropping out load, changing state, erratic behavior c) time delay relays - time setting d) other relays - chattering, non-operability e) switches - chattering

• This work was performed under the auspices of the U.S. Nuclear Regulatory Commission.

9-3 <

o Among the sensitive devices, certain types may withstand a substantially higher vibration level than others, o Many electrical devices are frequency-sensitive.

o Self-tapping screws strip at a much lower vibration level than what the equipment frame can otherwise withstand, and their use should be avoided.

o Equipment design has evolved with time. The later products have been observed to withstand higher vibration inputs. i l

In current Phase 11 efforts of the Component Fragility Program, BNL is l concentrating on collecting and analyzing the test data for the following equipment categories:

o Switchgear (additional data) o Electrical distribution equipment a) panel boards b) other panels o DC distribution switchboards o Instrumentation and control panels and racks a) nuclear instrumentation system b) proceis control equipment o Local Instruments a) transmitters b) switches c) indicators o Motor Control Centers (additional data) o Relays o Power Supplies The above components have been selected primarily by considering their relatively low seismic capacity levels and importance in performing safety functions as indicated in a prioritization study performed by Lawrence Livermore National Laboratory. Other equipment categories identified in the prioritization study will also be included as part of the Phase II effort during FY 1987 and 1988.

Since PRA and the Margin Studies in their current forms cannot accept more than one fragility indicator, e.g., a g-value, in Phase II each fragility TRS data set is being reduced to a single g-value. To this end, three different approaches are being considered at this time: first, the ZPA of the TRS; second, an equivalent ZPA of a standard response spectrum comparable to 9-4

the TRS; third, the spectral g-value at the natural frequency of the equipment. These g-values are then statistically used to determine a single-valued fragility descriptor for the equipment. The fragility descriptor used in this program is the HCLPF (High Confidence of g (ow Probability of Failure) value at the equipment mounting location.L2J A study of the damping characteristics of selected equipment assemblies is being performed in Phase Il by use of low level resonance search test data collected as part of the fragility information. For MCCs, the average damping values are observed from a limited data base to vary between 1.7% and 3.4% in the front-back direction at a level of 0.2 g.

REFERENCES 1.

Seismic Fragility of Nuclear Power Plant Components (Phase I), NUREG/CR-4659, June 1986.

2. An Approach to the Quantification of Seismic Margins in Nuclear Power Plants, NUREG/CR-4334, August 1985.

9-5 i

NRC'S SEISMIC MARGINS REVIEW 0F NUCLEAR POWER PLANTS R. C. Murray P. G. Prassinos Lawrence Livermore National Laboratory Livermore, California 94550 Recent studies have indicated that nuclear power plants are capable of withstanding earthquake motion substantially greater than the safe shutdown earthquake (SSE) acceleration. However, there has not been any systematic program to quantify the actual seismic capacity of these plants. For several reasons, especially in light of the changing perception of the seismic hazard, there is a continuing need to determine this capacity for licensing purposes. The NRC and industry are developing soundly based, ef ficient, and effective methods for identifying how much margin actually exists in important safety-related components, structures, and the plant as a whole. These methods are designed to be applied when questions arise about the seismic capacity of a plant.

An approach and trial guidelines for performing seismic margin reviews have been developed and issued as NUREG/CR-4334 and NUREG/CR-4482. The level of effort and cost associated with performing a seismic margin review are dependent on its scope and end use. The objective is to determine whether a plant can resist with high confidence a specified earthquake level greater than the SSE. To accomplish this objecti ve, analyses a re performed on components, systems, and the plant as a whole to determine if there is a high confidence of a low probability of f ailure (HCLPF) at this earthquake. The HCLPF represents a lower bound estimate of seismic capacity and is expressed in terms of ground acceleration.

The margins review methodology involves both the screening of components based on their importance and seismic capacity, and the quantification of HCLPF values for components, systems, accident sequences, and the plant. Systems analysis is used to determi ne those plant systems and components that are important contributors to plant seismic sa fety and thus allow focusing of ef fort on components requiring a margin review. From previous Probabilistic Risk Assessment (PRA) studies, it was found that, for PWR 's there are primarily two plant-safety functions that were identified as being major contributors to plant seismic sa fety : reactor su bc ri ticality and ea rly emergency core-coolant injection. For PWR's, the initial candidates for margin review will include those components that make up the systems that perform and support these two safety functions. At the present ti me , for BWR's, the margins review will. include those components and st ructu res relevant to all the plant safety functions. A separate study is currently underway to examine available BWR PRA's to determine which specific functions are major contributors to plant seismic safety. When this study is complete, it is hoped that specific BWR functions will be identified to focus margin reviews.

9-7

Concurrently, knowledge of the inherent capacity of components (structures and equipnent) obtained from previous PRA's and earthquake experience data, and the present seismic margin approach, is used to screen out high capacity items. However, all important components are initially reviewed and inspected to assure that thei r seismic capaci ty can be represented by the generic i nf ormati on used for this screening before they are removed from futher consideration. In addition, any potential systems interactions and impcrtant plant unique global features that are discovered during plant walkdowns are added to the list of review items.

The components remaining after the systems and fragility screenings plus the systems interactions and plant-unique global features are then subjected to a margins quantification. Each of these remaining components are thoroughly inspected and studied, and systems models are developed to describe the seismic-initiated accident behavior of the plant. The quantification is accomplished by calculating the HCLPF values for each of these components using structural / mechanical analyses based on the results of the detailed studies and inspection, and then analyzing the minimum cut sets derived from the systems analysis. The results of the quantification are the HCLPF values for each important low capacity component, important system, accident sequence, and the plant as a whole, which provide information that can be used to make decisions about the seismic capacity of the plant in relation to a selected earthquake level.

While the approach and the seismic margin review guidelines represent our best attempt at this time, they are being applied to trial plant reviews before being finalized. A trial seismic margin review of the Maine Yankee Plant is currently in progress. Fragility and systems analysis teams have been formed and are currently obtaining information on the characteristic and configura-tion of the plant. In addition, a' peer review group has been selected to review the technical results of this study.

The trial seismic margins review of Maine Yankee is more than 50% complete. A first plant walkdown has been completed by both the Analysis Team and the Peer Review Group. Event tree / fault tree development is underway, and HCLPF values are being developed for important components. A second plant walkdown is being planned for November 1986. The PWR trial review is scheduled for completion in February of 1987, and a trial BWR review is planned to start in 1987.

This trial review is intended to demonstrate whether this approach, the guide-lines, and the quantification techniques are adequate and provide sufficient information concerning the seismic capacity of the plant. The purpose of this paper is to report the status of this trial seismic margin review. The results will include important components, systems and accident sequences to seismic-induced core melt along with HCLPF capacities of these components, systems, sequences, and the plant as well as the ef fort needed to conduct a seismic margins review.

9-8

VALIDATION OF SEISMIC SOIL-STRUCrDRE INTERACTION ANALYSIS METHODS EPRI/NRC Cooperation in lotung, Taiwan Experiments C. A. Kot, M. G. Srinivasan, and B. J. Hsieh Argonne National Laboratory Argonne, Illinois 60439 Y. K. Tang and R. P. Kassawara Electric Power Research Institute Palo Alto, California 94303 While much effort has been devoted in the past to the development of seismic soil-structure interaction (SSI) analysis methods, few experimental data exist to validate these techniques. Consequently, the uncertainties and the assumptions / approximations made in SSI analyses (used in the design, licensing, and risk assessment of nuclear power plants) often introduce unnecessary conservatisms. Mindful of this, both the Electric Power Research Institute (EPRI) and the U.S. Nuclear Regulatory Commission (NRC) have undertaken programs for the validation of SSI analysis methods using experimental and earthquake data. A cornerstone of these efforts is the EPRI/NRC cooperation in the Lotung, Taiwan, seismic experiment that is described in this paper.

EPR1, in cooperation with Taiwan Power Company (T PC) , has built two concrete containment models (1/4 and 1/12 scales) in Lotung, Taiwan, where f requent earthquakes are known to occur and where a UC-Berkeley strong ground motion array (SMART-1), sponsored by the National Science Foundation, is in operation.

More than 130 channels of seismic instrumentation have been installed. These include accelerometers on the containment models and in the free field, the latter consisting of both surface and downhole arrays.

Pressure gages are located adjacent to the basemat of both scale models. The instrument records provide data to verify predictions of spacial variation of ground motion, foundation input motion, and structural response.

In addition to the monitoring of earthquake response, forced vibration tests were performed (primarily under NRC sponsorship) on the containment models. The purpose of the tests was to define the dynamic characteristics of the soil-structure systems and to obtain additional data for analysis validation. The tests for the 1/4 scale model were performed in two stages.

In the first stage, with only the basemat construction completed, the basemat was vibrated. Second, tests were performed after the entire structure was completed, with the force applied on the roof. S teady-s ta te (sinusoidal in time) excitation with a single eccentric-mass shaker was used to apply a uniaxial force in the frequency range from 1 to 40 Hz. In each run, the eccentricity of the rotating masses was held constant as the shaker was operated successively at a number of discrete frequencies. In the basemat tests, the directions of shaking were radial, tangential, and vertical, while for the completed structure, shaking was either tangential or radial. Up to 20 channels of acceleration were recorded in these experiments.

9-9 l

To date, a number of earthquake events ranging in Richter Scale magnitude from 5.3 to 6.5 have been recorded by the Lotung facility since completion of the models. Data reduction indicates that a good database has been obtained. With this database in hand, EPRI, NRC, and TPC have initiated a cooperative research effort, the obj ective of which is to evaluate and validate current SSI analysis techniques. This evaluation will be performed by using the methods to make blind predictions of the response of the soil-structure system to forced vibrations and to the actual earthquakes. The calculational results will then be compared with the actual measurements.

This process offers the unique opportunity to evaluate the SSI methods in a completely blind, independent, and unbiased fashion, in that the actual responses will not be known to those making the predictions. The ultimate goal of the program is to provide the basis for a more realistic SSI analysis practice for industry that is also acceptable to regulatory agencies. A round-robin approach is used in the program, with EPRI sponsoring U.S.

industry firms emphasizing current industry practice, NRC sponsoring universities and national laboratories involved in SSI methodology, and TPC sponsoring local Taiwaa organizations. Independence of all participants will be maintained during the course of the study.

Both direct mett.ods and substructure methods are candidates for verification. At Itast two models are to be constructed for each m'ethodology in successive phases. The first models will be based only on information typically available in nuclear power plant design, i.e., soil data and geological data. The final models will reflect additional information gained f rom the low-level forced-vibration testing of the as-built structure. All models will be used to predict response to at least one strong motion event recorded by EPRI. Where feasible, the models based solely on design type information will be used to predict forced-vibration test response.

Conservatisms and sensitivities in the SSI parameters and procedures are to be quantified. Assumptions, approaches, and their justification are to be documented to permit the comparison / evaluation of methods and measurements. A workshop is planned to discuss research findings, i

9-10

HDR PHASE II VIBRATIONAL EXPERIMENTS L. Malcher, EDR Project Kernforschungszentrust Karlsruhe, FRG C. A. Kot, Argonne National Laboratory Argonne, Illinois, U.S.A.

As part of the second phase of vibrational / earthquake investigations at the HDR (Heissdampfreaktor) Test Facility in Kahl/ Main, FRG, high-level shaker tests were performed during June and July 1986. The purpose of these experiments is to investigate full-scale structural response, soil-structure interaction, and piping and equipment response under strong excitation conditions. Data obtained in the tests also serve to verify analysis methods. This effort is supported by the government of the Federal Republic of Ge rmany (FRG) and the U.S. Nuclear Regulatory Commission (NRC), with additional participation by the Electric Power Research Institute (EPRI) and German industry.

The very large eccentric-mass coastdown shaker used in this series of experiments (SHAC) was designed by ANCO Engineers of Culver City, CA, while all the design and functional calcul,ations of its dynamic behavior were performed by the Fraunhofer Institut fur Betriebsfestigkeit (LBF), Darmstadt, FRG. The shaker is capable of developing 1000 tons of force. It is mounted on the operating floor of the HDR building and provides strong excitation to the entire HDR structure / soil / equipment system. The shaker is brought up to the desired starting speed (1.2-8.0 Hz) with its two arms in a balanced condition. Unbalancing takes place after the arms are decoupled from the drive system. Firing an explosive bolt releases the movable arm, which swings a round and couples with the fixed arm, forming a large eccentric mass that provides a variable (both in magnitude and direction) force during coastdown through the building resonances. Details of the shaker design were discussed at the previous Water Reactor Safety Research (WRSR) Information ' Meeting.

Preliminary tests conducted in February 1986 proved the functionality of the shaker system, except that it was not possible to reach speeds higher than 5.2 Hz due to air resistance that was much higher than expected. This problem was readily overcome by the provision of an enclosure for the shaker.

The SHAG test sequence was planned to provide the maximum possible loading for the HDR building and piping without global structurs/ soil failure. A combination of linear and nonlinear precalculations and comparison after each test with selected acceleration and strain measurements served to i establish the capacity limits of the HDR system. All major tests in the frequency range 1.2-8.0 Hz were originally intended to be run with approximately 10 kN of force with expected maximum building accelerations of 5 m/s2 and peak displacements of t7 cm. However, safety considerations both for the HDR building and adjacent installations necessitated curtailment of these plans. In particular, test runs that strongly excited the main structural mode of the HDR building (out-of phase bending at nominally 2.5 Hz) had to be reduced in load or abandoned. The major reason for this is the weakness of the outer concrete HDR shield building, whose cylindrical walls 9-11

have very little reinforcement in the embedded region and which could thus fail in tension during bending.

The final SHAG test series consisted of 25 runs with starting frequencies at 8.0 Hz (9 runs), 6.0 Hz (5 runs), 5.6 Hz (1 run), 4.5 Hz (6 runs), 2.1 Hz Only the 8.0 and 5.6 Hz runs were at full load (1 ryn), and 1.6 Hz (3 runs).

(>10 kN); all others were performed with reduced eccentricities. In particular, the highest load for the 1.6 Hz case, which mainly excites the building rocking mode, had a maximum force of 6,800 kN. Nevertheless, peak accelerations and displacements in the building were quite substantial and nonlinear behavior of the soil-structure system was clearly observed. Much local damage, such as concrete cracking and interior masonry wall collapse, i occurred. During the rocking experiments (1.6 Hz), much energy was i transferred to the soil, as evidenced by high accelerations, cracking, and subsidence. Also, impact occurred between the HDR building and the equipment tower and the connecting bridge to the office building. Strains in the HDR shield building walls approached or exceeded their estimated limit values. l Of major interest to all participants in the SHAG tests was the response of the VKL piping system, which was tested with up to seven multiple support (hanger) configurations. The hangers ranged from very rigid systems using snubber / strut supports typical of U.S. nuclear power plants, to very soft configurations (mainly spring and constant force supports) typical of conventional piping systems. Also included were energy absorbers, seismic stops, and viscous damper supports. The 8.0, 6.0, and 4.5 Hz experimental runs provided the data for the evaluation and comparison of the performance of the various hanger configurations. The 4.5 Hz tests were run under hot

(>200*C) and high-pressure (70 bars) conditions, while the remainder of tests were at room temperature. These experiments also provided qualification data for a typical U.S. gate valve that was specially installed in the VKL piping system.

Pretest and blind post-test calculational predictions are performed by a number of organizations for many aspects of the SHAG tests, including the soil-structure interaction and the building and piping response. All experimental data are held secure and will not be released until these calculations are completed (December 1986). Comparison with these data will serve to validate the computational methods. Ambient response measurements before, during, and after the shaker tests together with the SHAG data will be used to study nonlinear effects of the soil-structure system due to high-level loading.

9-12 J

THE SEISMIC CATEGORY I STRUCTURES PROGRAM by J. G. Bennett, C. R. Farrar, and H. E. Dunwoody Los Alamos National Laboratory

SUMMARY

At the end of FY84, the Seismic Category I Structures program entered a new phase. During the prior fiscal years, tests on microconcrete scale model shear deformation dominated structures were completed. The results indicated that these structures responded to seismic excitations with frequencies that were reduced by factors of two or more over those calculated based on an un-cracked cross section strength-of-materials approach. This further implies that stiffness associated with seismic working loads (operating basis earth-quake) are down by a factor of four or more. These reductions were also con-sistent with those measured during quasistatic tests to an equivalent level of loading. It was further pointed out that although the structures themselves have sufficient reserve margin, the equipment and piping are designed to re-sponse spectra that are based on uncracked cross sectional member properties, and that these spectra may not pertain to actual building responses.

The new phase of the program was aimed at verification of these conclu-sions using real concrete structures. These test structures were designed based on guidance from the program's Technical Review Group (TRG), a group of nationally recognized nuclear civil structure experts. Results of 1/4 scale model tests on the TRG-suggested geometry (but still using microconcrete) were also consistent with all prior results and were reported at the Thirteenth Light Water Reactor Safety Information Meeting.

During FY86, the first large TRG structure (4-inch walls of real concrete, No. 3 rebar, and with about 15 tons of added mass) was tested seismically at the Construction Engineering Research Laboratory (CERL) in Champaign, IL.

When measured property values were used to predict the first mode fre-quency as opposed to using assumed design values, results indicated stiffness reductions on the order of 4, consistent with previous results.

An analytical model must be developed to predict this reduction as a func-tion of shear wall aspect ratio and percentage steel reinforcing. A matrix of experiments has been planned using statistical methods that will allow this analytical model to be developed. Loading for these tests will be quasistatic and cyclic in nature, with the magnitudes increasing in 100 psi (average base shear stress) multiples after every 3 cycles of full load reversal. The first cycle is planned to be a single 50 psi cycle, followed by 3 cycles at 100 psi and then the 100 psi increments. After full load carrying capacity has been reached, the final 3 cycles will again be 50 psi cycles to look at hysteretic energy loss for the degraded structure. This sequence tests the assumption of l

9-13 l i

some computer programs that the same energy losses during low level cyclic loading of the structure occur both before and after the maximum earthquake shock has passed. These tests will also supply data on structural degradation as a function of load level. This set of statistically planned experiments is now underway on a geometry similar to that of the TRG series, except wall thickness has been increased to 6 inches. In addition, several assumptions that are used in industry design practices related to the contribution of transverse walls to the bending stiffness of the structures will be assessed.

No uniform procedure for treating these walls exists among the architecture engineering community. In order to supply data to develop guidelines for this procedure, separation of bending deformation from shear deformation is deemed important and instrumentation to obtain this data has been carefully planned and sized based on computed elastic deformation fields from finite element codes.

Sensitivity experiments are planned for this program that will examine torsional responses due to structural asymmetries, effects of large openings, and effects of different earthquakes. The TRG has recommended that these ex-periments can be carried out dynamically on smaller models. Planning for these experments is to be completed during the current fiscal year.

9-14

DEVELOPMENT OF SITE SPECIFIC RESP 001SE SPECTRA

• J.B. Savy, D.L. Bernreuter, and J.C. Chen Lawrence Livermore National Laboratory

SUMMARY

The United States Nuclear Regulatory Commiesion (NRC) has employed site specific epectra (SSSP) in their evaluation of the adequacy of the Safe Shutdown Earthquake (SSE) for a number of years. Since these spectra were first developed, a number of additional records from both the Eastern and Western U.S. and other areas have become available and should be incorporated into the existing sets of spectra. In the original applications of SSSP, only horizontal spectra were used. However, recent data indicate that it may be appropriate to compare the vertical component of the SSE design spectra to an appropriate vertical SSSP.

One basic concept behind the development of SSSP is that earthquakes in Eastern North America (ENA) have the same average range of source parameters as those in other parts of the world, most notably the Western United States (WUS). This implies that on the average, the only differences in ground motion records from earthquakes in different regions are due to differences in regional attenuation. Based on this assumption eets of SSSP are developed for ENA sites by selecting sets of earthquakes recorded on similar site conditions within distances where the differences in regional attenuation are thought to be emall. The available data used in these etudies are primarily from the WUS. The development and use of the SSSP is complicated by two factors:

first, in actual practice, it is difficult to match eite conditions, and still find sufficient number of records from earthquakes with the proper distance and magnitude ranges to develop etatistically significant sete of SSSP; and secondly, data from several recent ENA earthquakes suggest that ENA earthquakes may be richer in high frequency energy than is typical of earthquakes in the WUS.

It is observed that the short period spectral content of several ENA records is much higher than for typical WUS records. Several factors may lead to the observed differences:

1. Earthquake source mechanism differences.

2. Effect of local eite conditions, e.g., very shallow soil over an unweathered bedrock with a high shear wave velocity (high when compared with WUS).

3. The instruments for collecting the strong motion data and the methods used to correct for instrument response introduces errors in the digitized strong motion data.

• This work was supported by the United States Nuclear Regulatory Commission under a Memorandum of Understanding with the United States Department of Energy.

9-15

It is of some interest to note that all ENA recorde do not appear to exhibit the same enhanced high frequency content.

The purpose of the study reported here is principally to develop new sets of SSSP applicable to several categories of site conditions, thus updating previous work. In addition, the methodology le applied to the vertical component of the strong motion.

The approach adopted consists of five parte. In the first part (Section

2) we review some of the work that has been performed to mathematically model source, travel path and the local site conditions that sheds some light on the issues at hand. The results from these studies, along with the modeling studies of Section 4 are used to aeslet in understanding the comparisons that are made between the spectra from ENA earthquakes and other world wide data.

1 In the second part, we point out some of the physical limitations present I in the ground motion data due to the data collection technique, including the

! Instraments used in the ground motion data equipment.

The third part, reported in Section 4, consists of a set of analyses performed using eimple modeling to better understand the importance/effect of shallow soil layers, weathered rock and deep sof ter rock columns typical of WUS sites.

In the fourth part, reported in Section 5, we examine the contribution to the uncertainty associated with our SSSP and attempt to determine which factore are the most significant as a guide to the development of criteria to select suites of records to represent a given category.

The fifth part, reported in Section 6, we present the results of calculations for the updated SSSP for the various site categories considered and examine the eensitivity of the SSSP to the makeup of the sets.

The major conclusion reached in this study is that the ground motion at a site will exhibit large variability between earthquakes due to eource and travel path variabilitiee. Thus, any approach used to estimate the ground motion at a site will have large uncertainties and the only way to reduce these uncertainties is to include the important source and travel path parameters in the model. We also found that site parameters are very important and should be modeled. These are not new conclusions, but we believe that in Section 5 we have better defined the relative contribution of each of the major sources of uncertainty in the method used to estimate the ground motion.

We concluded that the SSSP were not very eensitive to the dietance distribution. That is, the variability (uncertainty) introduced by the range of distances used was relatively small compared to the variability introduced by other factore. We also concluded that the SSSP are somewhat eensitive to the distribution of magnitudes used, particularly at rock and, by inference, at shallow eoil sites. We found that one important criteria in eelecting records to generate SSSP is the depth of soil.

9-16

NATIONAL SEISM 0 GRAPHIC NETWORK FOR THE EASTERN UNITED STATES BY ANDREW J. MURPHY The U.S. Nuclear Regulatory Commission (NRC) currently operates four regional seismographic networks in the Central and Eastern United States through 15 individual contracts. Some of the stations of the regional networks have been in operation for over ten years, are expensive to maintain and should be upgraded to take advavtage of improved technology. The data are needed to address the seismic hazard issues currently confronting the NRC, including the surveillance of background seismicity as it would affect operating NPPs.

The NRC has entered into negotiations with the U.S. Geological Survey (USGS) to establish a National Seismographic Network. The NRC would fund the purchase of the capital equipment while the USGS would support the design, testing, and installation of the network. The USGS would also have responsibility for the operational expenses of the network as a national facility.

The two principal technical design criteria are:

a. detection and location capability for any earthquake in the Central and Eastern U.S. greater than magnitude 2.5.

b. the ability to report any such earthquake within thrity minutes of its occurrence.

The NRC and the USGS are actively soliciting additional partners in the de-velopment of the National Seismographic Network both from the Federal and industrial community.

9-17

Physical Modal for Reactor Coolant Purps by K. Schneider, F. Winkler Kraftwerk Union AG 8520 Erlangen (FRG)

In LOCA-analysis of large and small leaks in PWR's the two phase flow performance of the main coolant pump has an influence on the flow distribution in the primary system and on core cooling.

Within the world-wide efforts to investigate two-phase pump behaviour, two main research projects have been initiated in Germany by the Federal Ministry of Research and Technology (BMFT). Beginning in 1976 extensive steady state and transient tests with both 1:4 and 1:5 scaled models of an axial-type main coolant pump have been performed; and in the following analytical program an empirical two-phase flow pump model was developed.

Using single-phase pump-theory two-phase flow pump performance is described in the model on the basis of the Eulerian pump equations. For homogenous and incompressible medium they read AP = p ( u_2 c 3 - u2 .

c 2)

T= (u 3 c 3 - u 2 .

c 2) and allow the calculation of pump pressure rise and pump torque to the determination of the velocity triangles at the inlet and outlet of the pump impeller. In investigating two-phase flow behaviour, consequently one has to consider phenomena which modify those triangles. In our model the effects taken into account are

- a phase separation effect in the impeller perpendicular to one-phase streamlines

- the extreme compressibility of two-phase fluid

- the phase slippage between void and liquid When neglecting the mu,tual perturbations, on additive approach for pump pressure rise and torque becomes practicable ap2ph

• P 1ph + P separ

+ P comp

+ P slip 2ph

• 1ph separ

+

camp

+

slip' and for each process a characteristic number can be derived when idealizing the process concerned. As for the phase separation effect, this number can be calculated from the maximum reduction of circumferential speed occuring with total phasic segregation at the discharge of the impeller. The characteristic number of 10-1

compressibility depends on two-phase sonic velocity and on pump head wereas the slip number can be derived from the simultaneous validity of the Eulerian equation for both phases.

Exploiting those characteristic numoers, K in a dimensional analysis treatment it becomes apparent that 1,ll a terms possess a common representation O pi = p (n2 q ) g i f i

(E) 0 T

i

=

.g (n2 @ ) K g g1 @)

0 Thus with this analytic treatment the evaluation of the two-phase tests is reduced to the determination of the functions f. and g.

which are calculated in the model by application of the leasd squares method.

Evaluations have been performed for an axial-type pump (Andritz),

a pump of mixed-flow-type (Byron-Jackson) and a radial flow-type pump (LOBI). The enclosed figure shows the main results and some peculiarities of the pump model presented.

0.6

"'2

- There is of course a 0A: =.5 degradation of pump _

head and torque de- u m E2 pending on void . -

fraction and pressure "a=

% 0- }"

- The single-phase similarity laws are 1 2 h test] /"

125 not fulfilled - model

- The degradation maxi- -01.2 mum depends on flow- 1 y speed-ratio, pressure n : ,7

-Q6 and magnitude of flow and speed. g ' ' ' g'2 ' ' ' d.I. ' '06' ' '0'.8' ' ' 10 void fraction The pump model was used with a calculation of a blowdown from a double-ended brbak in the cold leg of a 4 loop-GPWR. As the pump model predicts a pronounced degradation of head and torque at high mass-flow, the increase of pump speed in the broken leg is smaller than calculated with previous models.

10-2

Results from Assessment of RELAP5/M002 and TRAC-PF1/ MOD 1 BY F. WINKLER KRAFTWERK UNION AG The BMFT of FRG is participating in the INTERNATIONAL CODE ASSESSMENT PROGRAM (ICAP) of the US-NRC in which KWU and GRS as its agents are performing the required code assess-ment studies. The studies include analysis of experiments in FRG test facilities as well as analysis for commissioning transient tests in modern KWU-1300 MW PWR and BWR nuclear el power plants.

The six studies which have already been performed by KWU and GRS include 2 post-test and 1 pre-test calculation of PKL-I and PKL-II experiments, one post-test analysis of a safety valve ATWS separate effect experiment, one post-test calculation of a GPWR commissioning test and a multi-dimensional analysis of a LARGE LOCA in a GPWR.

The first study was devoted to the post-test analysis of the PKL-IDG- Small Leak Experiment with hot leg HPIS injection using RELAPS/ MOD 2.

The study performed by T.NGUYEN and G.SEEBERGER of KWU indicated that a parallel channel noding in the core and upper plenum is necessary to simulate hot leg ECCS in-jection. Pressure, temperatures and collapsea water level in the core and upper plenum prior to the be-ginning of refluxing are in good agreement with the test data.

The second study was devoted to the post-test analysis of PKL-IIB 2 large cold leg break experiment with simul-taneous hot and cold leg ECCS-injection using RELAPS/ POD 2.

The calculation performed by H. DANG-VIET, F.FOTIADIS and P.FRIEDMANN of KWU shows that parallel channel noding in upper plenum and core is necessary. Pressure, 10-3

collapsed level in the test vessel, CCFL behaviour in the downcomer, loop oscillations and heater rod tempera-tures are in good agreement with test data, steam pro-duction in steam generator appears to be too high.

- The third study involves a pre-test calculation of the PKL-IIB 3 "Large HOT LEG Break Experiment" with hot leg ECC-injection using TRAC-PF1/M001 (Version 11.1). The calculation performed by K.TRAMBAUER of GRS indicates, that the code predicted the overall behaviour reasonably well especially the mass flows rates and water distrib-ution. Quenching was calculated earlier than observed.

Water down flow was calculated and measured to occur mainly in regions adjacent to the not legs which had ECC injection.

- The fourth study included post-test calculations of "ATWS SAFETY VALVE Experiments" with two phase mixture discharge using both RELAP5/ MOD 2 and TRAC-PF1/M001. The study per-formed by U.NEUMANN, P.PUZALOWSKI and E.SCHELLER of KWU was a somewhat unusual application of the codes in that it was a pressure wave experiment with a total duration of only a few seconds.

In general the codes were able to calculate the experiment well, but some problems were observed in the propagation and tracing of a steam-water interface and recommendations were made for certain desirable code improvements.

~

- The fifth study was a post-test calculation of a 1300 MW-GPWR commissioning test " Reactor Trip at Full Load" using RELAP5/M002. The calculation was performed by G.GERTH of KWU and showed good agreement with measured data. It was difficult to establish correct initial conditions in the steam generator and it was apparent that more effort is needed for simulating the control and limitation systems.

- The sixth study is a GPWR large LOCA calculation using TRAC-PF1/M001, was performed as part of the 2D/3D-program and will be presented in the 20/3D-session.

Code assessment activities are continuing within the FRG with current emphasis being placed on UPTF pre- and post-test an-alysis.

10-4

CSNI VALIDATION MATRIX FOR PWR_ AND BWR_ CODES K. Wolfert Gesellschaft fur Reaktorsicherheit (G RS )

I. Brittain United Kingdom Atomic Energy Authority (UKAEA)

To provide an international common basis for the independent assessment of large thermal-hydraulic system codes, the task was given by the CSNI Working Group 2 to the " Task Group on Status and Assessment of Codes for Transients and ECC" to prepare an internationally agreed validation matrix for PWR and BWR codes.

For PWR codes four individual matrices have been prepared, differentiating between transients small and intermediate breaks (one separate matrix for PWRs with U-tube steam generators, one separate matrix for PWRs with once-through steam generators) large breaks For BWR Codes two individual matrices have been prepared, differentiating between transients loss-of-coolant accidents Each individual matrix consists of a Cross Reference Matrix and a table of selected tests. In the Cross Reference Matrix the physical phenomena which are assumed to occur during the transient or LOCA, the experimental facilities suitable for reproducing these effects, and the test types of interest are l

10-5 l

listed. Based on the knowledge and experience of the Task Group members, the relationships phenomena versus test types, phenomena versus test facilities and

- test facilities versus test types have been assessed for relevance and suitability.

Based on the Cross Reference Matrices, well-balanced sets of tests were selected according to the criteria each phenomenon should be addressed in all relevant test types and in test facilities of different scale.

all test types should be included.

The matrices are focussed mainly on integral system test data and operating data from power reactors, assuming that the developmental assessment of individual models has been satisfactorily completed by the code developer. Separate effects tests have been selected only in cases where suitable integral system tests could not be found to address a particular phenomenon. Where counterpart tests or near-counterpart tests have been identified between two or more test facilities, they have been considered favourably in order to address questions relating to scaling and facility design compromises.

At the end of 1986 the matrices will be published as a single CSNI report covering LWR's. The matrices will be a guide for independent code assessment, will be a basis for the comparisons of code predictions performed with different system codes, and may contribute to the quantification of the uncertainty range of code predictions.

References

/1/ CSNI Validation Matrix for the Assessment of Thermal-Hydraulic Codes for PWR LOCA and Transients, Task Group on Status and Assessment of Codes for Transients and ECCS, SINDOC (86) 12, December 1985

/2/ CSNI Validation Matrix for the Assessment of Thermal-Hydraulic Codes for BWR LOCA and Transients, Task Group on l

Status and Assessment of Codes for Transients and ECCS, May 1986 10-6

SWEDISH EXPERIENCE WITH RELAP5/M002 ASSESSMENT 0 Sandervag Studsvik Energiteknik AB, Sweden ABSTRACT The Swedish assessment of RELAP5/M002 is a part of the International Code Assessment program which is organized by the US NRC. The major part of the experimental data used for assessment is of Swedish origin. The data encompass critical flow and level swell data from the Marviken facility; axial void profiles, dryout and post-dryout data from the FRIGG and Royal Institute of Technology-facilities; and integral system thermal-hydraulics from the FIX-II facility.

A part of the agreed assessment matrix has been completed. Assessment against separate and integral experiments shows that the dominant uncer-tainty in prediction of clad temperatures is due to a poor calculation of dryout. Predicted post dryout wall temperatures, given the experimental dryout location as input parameter, generally agree well with data.

Simulations of level swell following depressurization of the large diameter Marviken vessel showed that RELAP5/M002 was able to calculate overall axial void profiles in fair agreement with data. The assesment showed that increasing the modeling detail could give rise to numerical instabilities.

Assessment against large scale critical flow data shows that the agreement with data is somewhat dependent on upstream fluid conditions and modeling.

Low quality two phase flow is,in general, accurately predicted while subcooled liquid flow and saturated steam flow are generally overpredicted if no discharge coefficient is applied. It is recommended not to model short length nozzles explicitly.

I 10-7

Finnish Assessment of RELAP5/ MOD 2 H. Holmstr6m Technical Research Centre of Finland (VTT)

The first version of the frozen version of RELAP5/ MOD 2 (cycle

36) was received in February 1985. Since then four sets of updates have been received to create cycle 36.03, the last ones in summer 1986.

Four assessment cases have been calculated before August 1986:

1) LOBI-Mod 2 test A2-81 (ISP18, 1 % cold leg break)

2) LOFT test SB-3 (0,5 % cold leg break, sec. feed and bleed)

3) LOFT test LB-1 (200 % cold leg break)

4) Plant transient (turbine bypass valve stuck open)

In addition, the code has been used in several calculations of hypothetical LPWR cases, mainly steam line breaks and SBLOCA transients.

The code has generally produced good results and run well.

Some problems, however, have been encountered:

Sometimes severe flow oscillations Discontinuity in flow when flow pattern changes Horizontal stratification model unsatisfactory Pressurizer insurge not calculated well Accumulator behaves incorrectly (pressure stays high)

Flow distribution in LBLOCA incorrect Temperature distribution in core unsatisfactory (LBLOCA)

Horizontal steam generator difficult to model The next test cases will probably include one reflood and one natural circulation test in the Finnish REWET facility, loop seal behavior tests in a Finnish large scale air-water loop and international standard problems.

10-9

- . - _ . . - . - . . , - ~. - - . . . .-

l i

TRAC DEVELOPMENT AT GENERAL ELECTRIC J. G. M. Ande*sen J. C. Shaug B. S. Shiralkar General Electric Company 175 Curtner Avenue San Jose, California 95125 TRAC is a computer code for transient analysis of light water reactors. The BWR version of TRAC has been developed as a result of a close cooperation between the General Electric Company and the Idaho National Engineering Laboratory under the GE/NRC/EPRI funded BWR Refill /Reflood and FIST programs. At INEL which has the main responsibility for tb- NRC version of TRAC this work has led to the development of TRACBD1 and TRACBF, while at GE, TRACB04 was the final product of the Refill /Reflood and FIST programs.

TRAC development has continued at General Electric after the completion of the Refill /Reflood and FIST programs in 1985. The purpose of this paper is to describe the ongoing TRAC development at General Electric.

The TRAC development at General Electric can be divided into two main categories: improved benchmark capability and user convenience.

In the area of improved benchmark capability the major development has been the addition of a hot rod model for upper bound peak cladding temperature (PCT) prediction, and a 1 or 3 dimensional neutron kinetics model.

The fuel channel model in TRAC is 1-dimensional and the capability of the TRAC channel component to predict PCT's is limited by this assumption. The inability of TRAC to model local variations i the cross sectional fluid conditions results in a smaller spread in the predicted temperatures than often observed in the data. The purpose of the hot rod model is to simulate this local variation in the hydraulic conditions leading to an estimate of the upper bound PCT.

10-11

Two transient neutron kinetics models have been implemented into TRAC, a 1-dimensional and a 3-dimensional model, providing the capability to simulate a wide range of transients such.as pressuriza-tion events, partial scram, rod drop accidents with void feedback, time domain oscillations and ATWS scenarios. The 3-dimensional model is based on a separation of the flux and precursor concentrations into shape and amplitude functions. This approach reduces the execution time, e.g., for a simulation of a turbine trip event the execution time for the 3-dimensional model is only twice that of the 1-dimensional model.

Both the 1-dimensional and 3-dimensional models are consistent with the GE 3-dimensional core simulator, PANACEA, making input preparation a simple automated task. Comparisons with data vill be shown in the paper to illustrate the capability of these models.

The other main effort has been aimed at making TRAC more user convenient. To accomplish this, two approaches have been taken. One is to reduce the execution time of TRAC, and the other is to implement TRAC on a distributed system of mini and micro computers.

In order to reduce the execution time an implicit two-step method was implemented for both the 1-dimensional and 3-dimensional components under the FIST program. The implicit integration method has recently been upgraded to include the channel leakage flow and the separators making the hydraulic model 100% consistent with the implicit schemes.

In order to further improve the user convenience of TRAC, it has been implemented on to a distributed system of 32 bit mini and micro computers. This task was completed early this year and since then the code has been extensively tested. The testing has demonstrated that previous results obtained a CDC main frame computer can reliably be reproduced on a 32 bit micro computer. The CPU time on the micro computer has increased; however, by using a distributed system of microcomputers, wherc one or more computers could be dedicated to TRAC, the elapsed turn around time as well as availability has been greatly improved. TRAC is currently implemented on a MicroVAX-II computer.

In summary, the development at General Electric has significantly increased the benchmark capability of TRAC and reduced the cost of executing TRAC making TRAC a more useful and usable engineering tool.

10-12

3-D Neutronics Model Implementation Into TRAC-BD1 (TRAC /TOSDYN)

S. Tsunoyama*

H. Uematsu*

J. C. Shaug*

H. Nambat B. S. Shiralkar+

NAIG Nuclear Research Laboratory t Toshiba Corporation

+ Ceneral Electric Company

SUMMARY

A three-dimensional neutronics model has been incorporated into TRAC-BD1. The resulting code (TRAC /TOSDYN) thus provides detailed thermal hydraulics and neutronics capability for the analysis of a wide spectrum of BWR transients and accidents.

The purpose of this paper is to provide (1) a brief description of the TOSDYN neutronics model and solution method, and (2) simulation of a sample transient with statistics of computer running time for TRAC /

TOSDYN.

Neutronics Model The modified one-group time dependent diffusion equations, assuming N delayed neutron groups, are solved in the TOSDYN module.

To obtain the transient solution, TOSDYN uses the so-called

" Improved Quasistatic Method". This method has proved successful for several multidimensional space-time kinetics programs ( .

1. Ott, K. O. and Meneley, D. A., " Accuracy of the Quasistatic Treatment of Spacial Reactor Kinetics", Nuclear Science Engineering, 36 402 (1969). ~

10-13

The procedure used in the quasistatic method is based on the factorization of the space-time dependent neutron flux $, (r,t) into a scalar amplitude function T (t) which determines primary time dependent and a shape function S (r,t) which determines primary spatial dependence of the solution.

The amplitude function T (t) represents the magnitude of neutron flux over whole core and may change rapidly in a transient. On the other hand, the shape function S (r,t) represents the spatial dis-tribution of neutron flux normalized by T (t) so that the temporal change of S (r,t) is expected to be much less than T (t). So, T (t) and S (r,t) are solved over different time steps in the quasistatic method.

T (t) is zolved over the so-called reactivity step which is usually much smaller than the so-called shape step for S (r,t). This procedure reduces computational time considerably without reducing accuracy.

In order to demonstrate the capability of TRAC /TOSDYN in simulating transients, a control rod drop event from power operation for a BWR/5 was analyzed. The initial plant condition is about 50% rated power and 100% rated flow. The neutronics module has the capability to evaluate a full core model or half, quarter and octant symmetry core models. For this case, octant symmetry was assumed during the transient to reduce computer time.

The transient starts by the movement of rod, with velocity of 95 cm/second. Figure 1 shows changes in the power in the adjacent and average bundles calculated by the neutronics modules of TOSDYN. The power increases during the first 1.2 seconds, and then turns to decrease. Effects of the various reactivity components (control, void and Doppler) were correctly calculated.

The percentage of computational cost for this case spent in the three dimensional neutronics model was of the order of 11%.

i 10-14

(MWt) e 20 I

i i

Adjacent Channel

____ Average Channel 15 _

10 i

/ \

1 p 4

/ 5 r

a

.' \

i \

/

,i 5 _ '

\

t g

1 j / \

/. /  %

/

e

/ \

\

s ,

/./.",/

s s

. .. _ _ s #j:.

s s

N i

i t

i I

0 l l I I I I I

(( 0.0 0.3 0.6 0.8 1.2 1.5 1.8 (Sec.)

Figure 1. Histories of Channel Powers

\

t l

1

VALIDATION OF TRAC-PD2 AGAINST EXPERIMENT LP-02-6 0F THE LOFT-0 ECD EXPERIMENT A. Alonso; J. Blanco; J. C. Martinez: A. Querol, and J. Rivero, De-partment of Nuclear Technology, Madrid Polytechr.ical University; J. L. Mora, Vandell6s Nuclear Association.

SUMMARY

In early 1985, Spain decided to join the LOFT-0 ECD Project. To secure transfer of knowledge and technology into the country, five responsible Or-ganizations decided to establish a formal Agreement. One of the parties in-volved is the Polytechnical University of Madrid, through its Department of Nuclear Technology.

The Department was assigned with the responsibility of analyzing the results of experiment LP-02-6, using a NEA supplied version of TRAC-PD2. Such experiment was selected to determine the system behaviour and fuel response for conditions close to the assumptions specified in Apendix K to 10CFR50. The computer code had been exercised previously in a CDC-Cyber-180 machine.

Previous work with the code was mainly devoted to solve the graphics problem, as the version received was not compatible with our software. Two codes were witten and tested. TRISAD is able to extract the information from l TRCGRF and to condition the information for our code ISADORA, developed in the 1

Department of Nuclear Physics. On its side, computer code ISAISA is able to join two or more outputs from TRISAD, corresponding to successive time inter-vals in the transient.

l The analys'is of the experiment consists in running pre-test and post-test calculations. The first was conducted usino the parameters supplied to us by INEL, with small modifications suggested by the experiment proper. The results were similar to the ones by INEL, with the exception of pump oscilla-tions, startilo after 30 seconds into the transient, instead of after 40 se-conds, at in the reference calculations. No explanation has been ventured.

10-17

Previous to the post-test calculation, the steady-state was corrected to take into account experimental conditions; the corrections affected the secondary side of the steam generator, pressure losses in the loop, hot leg pressure and reactor power. All variables maintained their values within the margins specified in the experimental measurements.

Calculations revealed that pump behaviour was not well predicted by TRAC. Corrections were introduced in the inertia, which was made dependent on angular velocity, and on torque, by introducing a degenerated component.

With these corrections, experimental results were well predicted. Other co-rrectionsaffected the accumulators, friction factors and injection systems, which showed satisfactorilli into the results.

Our calculations, as well as those by other Groups working with TRAC and RELAP, did not predict well the rod temperature profile and time to re-wet. This is due to an early rewetting taking place in the experiment, which is not well predicted by the codes. By using the fine mesh possibilities of TRAC some improvement was obtained, but at a heavy penalty in calculation time. A decision was taken in modifying corelations in the code.

The first was a modification into Forslund factor to estimate heat transfer to liquid. The second affected the minimum boiling temperature, as it is known corelations normally used do not predict it well at high pres-sures. Three corelations were tested, the one used in RELAP and those by Siegel and Sakurai. It was determined that Sakurai's gives acceptable results for pressures above 5.5 bars. For pressures below 5.0 bars the homogeneous nucleation corelation is superior. For intermediate values, an interpolation b etween the two.

The thermal part of the analysis was finally performed with the cor-rections above obtaining a temperature profile for the rod very close to the experimental. A considerable improvement was also obtained in the time to rewet.

10-18

INFUJENCE OF MOISIURE ON THE BEHAVIOR OF AEROSOIS R. E. Adams, A. W. Inngest, M. L. Tobias Oak Ridge National laboratory Oak Ridge, Tennessee 37831 he behavior of aerosols assumed to be characteristic of those generated during light water reactor (UG) accident sequences and released into containment has been studied in the Nuclear Safety Pilot Plant (NSPP) located at the Oak Ridge National Iaboratory (ORNL) . We purp of this project, sponsored by the Division of Reactor System Safety, NRC, was to provide experimental qualification for U m aerosol behavior codes 'm current use. Due to budget limitations all experimental activities were susparled in February 1986.

We research plan for the NSPP aerosol project provided for the study of the behavior, within containment, of simulated UVR accident aerosols emanating from fuel, reactor core structural materials, and from concrete-molten core materials interactions. W e aerodynamic behavior of each of these aerosols was studied individually to establish its characteristics; a limited number of experiments involved a mixture of these aerosols to establish interaction and collective behavior within containment. Tests have been conducted with U 03 8 aerosols, Fe2O3 aerosols, and concrete aerosols in an environment or' either dry air (relative humidity (RH) less than 20%] or steam-air [ relative humidity (RH) approximately 100%] with aerosol mass concentration being the primary experimental variable. Experiments with a test aerosol mixture of Fe2O 3 + U 03 8 were conducted with the primary experimental variables being aerosol mass concentration and aerosol cor: position (mass ratio) .

l We NSPP facility is corposed of a test containment vessel, aerosol I generating equipment, aerosol sampling equipment, and environmental measuring equipment. W e vessel is a stainless steel cylinder with dished ep having a diameter of 3 m and a height of 5.5 m and a volume of 38.3

m. We test aerosols were produced with a generator which consists of a commercial plasma metalizing torch asserbly and a special high-temperature reaction chamber into which metal or concrete powder is injected together with argon aml oxygen gases. For the dry aerosol tests the vessel atmosphere was dry air (RH 20%) and the temperature and pressure were slightly above ambient. W e steam-air aerosol tests were conducted under quasi-steady-state steam conditions. The test atmosphere was prepared by injecting steam into the vessel (initially at subatmospheric pressure) to form a steam-air mixture at elevated temperature (around 380 K) and pressure (about 0.2 MPa absolute); upon achievement of this condition the aerosol was generated and introduced.

A number of aerosol tests were conducted under both dry air and th steam-air test environments;3.

ranged between 0.3 and 5 g/m e aerosol mass concentration generally W e presence of steam in the test Il-1

environment causes a change in both the aerodynamic behavior and the physical shape of U 38 0 , Fe2 O3, or U O3 g + Fe2 O 3 aerosols. 'Ihe most I obvious effect of steam is an enhanced rate of removal of aerosol from the vessel atoesphere emmred to that under dry conditions. 'Ihe aerosols are changed frcxn loosely-emmeted webs of chain-agg1cxnerates (noted under dry conditions) to almst spherical " balls" assumed to be emmeted from these webs. 'Ihe aerodynamic mass median diameter (AMMD) for these spherical

" ball" aerosols ranged from about 2 to 4 um; under dry conditions the AMMD ranged between 1.5 and 3 pm.[1]

Ooncrete aerosol is not affected by the presence of steam in the same 3g O or Udry 03 and O 3 aerosol. 'Ihe rates of removal 8 + Fe2 manner as U O , Fe2 of concrete aerosol un 3, der under steam-air conditions are essentially the same. This aerosol was generated by passing powdered limestone-aggregate concrete through the plasma torch aerosol generator.

Concrete aerosol is not a simple, single-component aerosol (such as UO3g or Fe2 O3 aerosol) but actually a complex mixture of particles of Al 23 O , SiO 2, Cao, MgO, Fe2 O 3, and various silicates with Al, Ca, Mg, and Fe as the cations. Steam affects the physical shape of concrete aerosols (to a lesser degree than for U 038 r Fe 03 aerosols) producing some <mmeted spherical balls. 'Ihe of the concrete aerosol was about 1 pm or less.[2]

Sirue the aerodynamic behavior of some, but not all, of the test aerocols was observed to be influenced by the presence of steam (water vapor) and the influence is probably the result of a change in shape of the aerosol, it was of interest to determine at what level of humidity this change in shape occurs for the aerosols under test. A project was initiated to systenntically study the effect of moisture level in an air atmosphere on the shape of agglomerated aerosols composed of U 380 '

Fe2O 3, and concrete, singly and in various combinations. 'Ihe aerosol-moisture interaction test (AMIT) facility, incorporating a test vessel of 0.52 m3 volume, was prepared for this study, Six tests were conducted with Fe2 O 3 aerosol before the study was terminated. 'Ihe Fe293 aerosol was at chain-agglomerates noted to be in RH values upthe fann at to 92%; of 100%

webs RH of the agglomerated aerosol was in the fonn of compacted spherical balls.[3]

References

1. Adams, R. E., "Recent Activities in the Aerosol Generation and Transport Program," proceedings of the USNRC 'IVelfth Water Reactor Research Infonnation Meeting, Gaithersburg, MD, October 22-26, 1984, NUREG/CP-0058, Vol. 3 (January 1985) .

2. Adams, R. E. and 'Ibbias, M. L. , " Aerosol Release and Transport Frugicua

Semiannual Progress Report for April 1984-September 1984,"

I NUREG/CR-3830, Vol. 2, (ORNI/'IM-9217/V2) December 1984.

3. Adams, R. E. and Tobias, M. L., " Aerosol Release and Transport Program Semiannual Prugr==ss Report for October 1985-March 1986,"

NUREG/CR-4255, Vol. 3, No. 1, (ORNI/TM-9632/V3&N1) June 1986.

11-2

SUMMARY

OF AEROSOL CODE-COMPARISON RESULTS FOR LWR AEROSOL CONTAINMENT TESTS LAl, LA2, AND LA3*

A. L. Wright, J. H. Wilson, and P. C. Arwood Chemical Technology Division Oak Ridge National Laboratory Oak Ridge, Tennessee 37831

SUMMARY

The Light-Water-Reactor (LWR) Aerosol Containment Experiments (LACE) are being performed in Richland, Washington, at the Hanford Engineering Development Laboratory (HEDL) under the leadership of an international pro-ject board and the Electric Power Research Institute. These tests have two objectives: (1) to investigate, at large scale, the inherent aerosol retention behavior in LWR containments under simulated severe accident con-ditions, and (2) to provide an experimental data base for validating aero-sol behavior and thermal-hydraulic computer codes.

Aerosol computer-code comparison activities are being coordinated at the Oak Ridge National Laboratory (ORNL). For each of the six LACE tests,

" pretest" calculations (for code-to-code comparisons) and " blind posttest" calculations (for code-to-test-data comparisons) are being performed. The overall goals of the comparison effort are (1) to provide code users with experience in applying their codes to LWR accident sequence conditions, and (2) to evaluate and improve the code models.

The LACE experiments use the 852-m3 -volume Containment Systems Test Facility (CSTF) vessel at HEDL. A two-component aerosol source, consisting of Mno and Cs0H aerosols, is used in the tests. Test LAl was designed to simulate LWR " containment bypass" accident sequence conditions. In this test, aerosols were injected into a 0.63-m-diam, ~30-m-long test pipe (the pipe had six 90* bends). The pipe inlet flow velocity was roughly 100 m/s, and the outlet flow velocity was roughly 200 m/s. Aerosols transported through the pipe were then made airborne in the CSTF vessel.

Test LA2 was designed to simulate LWR "f ailure to isolate contain-ment" accident conditions. In LA2, aerosols were injected directly into the CSTP vessel under condensing steam conditions; a high leakage rate (similar to the expected aerosol deposition rate) was maintained during the test.

Test LA3 actually consisted of three te s t s ( LA3 A, LA3 B , and LA3C),

again designed to simulate containment bypass accident conditions.

Variables in these tests were the Mn0/Cs0H mass ratio (values of 8:1 and 2:1) and the pipe flow velocity (values of 20 and 80 m/s).

• Research sponsored by the Electric Power Research Institute under Interagency Agreement DOE-40-551-75 with the U.S. Department of Energy under contract DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc.

11-3

1 For pretest calculations, code users were provided specific values of code input parameters, such as system geometry, aerosol source rates and sizes, pipe and vessel temperatures, and vessel gas leakage rates. We also specified both the code output parameters (total aerosol deposited in the test pipe, airborne aerosol concentrations, etc.) and the times for code outputs. Aerosol computers codes used in these studies include the CONTAIN, NAUA, AEROSIM, MAAP, RETAIN-2C, RETAIN-S, MCT, REMOVAL, and TRAP-MELT 2 codes; in some cases different investigators used the same codes.

Detailed results from the LAl, LA2, and LA3 tests have been documented -in a series of proprietary LACE technical reports; brief sunmaries of the results are provided below:

1. LA1: The major result was that errors in code inputs, partic-ularly the input acrosol source sizes, produced most of the observed dif ferences in code results. Four of the five pipe calculations produced similar results, even though some calcula-tions were performed with incorrect aerosol source sizes. Large differences in calculated aerosol concentrations, depositions, and sizes were observed for the CSTF vessel calculat!3ns, mostly because of the incorrect source-size inputs. Coding errors in the TRAP-MELT 2, MCT, and REMOVAL codes were identified and corrected.

2. LA2 : All calculations were performed with the requested aerosol code inputs, and the overall resulting code agreement was much improved compared to results for test LA1. It was discovered that a number of codes did not calculate the aerosol aerodynamic mass-median diameter (AMMD) correctly (this is an important size para-meter that is typically measured with cascade impactors in containment tests); this error has been corrected in some of the codes used. Differences found in the calculated diffusiophoretic deposition can be attributed to the dif ferent models used in the various codes; this issue is presently being resolved.

3. LA3: Preliminary evaluation of the LA3 pretest results shows that code results for calculated pipe deposition dif fered by factors of 2 to 6 for the three LA3 tests. Some of the investigators feel that these dif ferences are due to dif ferent assumptions related to the " control volumes" used to model the pipe. To resolve this question, a " benchmark" calculation is presently being performed in which all investigators use the same control volumes to model the pipe; these results will be reported in the full paper.

The LACE pretest code comparisons for tests LA1, LA2, and LA3 are pro-viding useful information to the project participants. More information will be obtained from future pretest and posttest LACE code comparisons.

11-4

8 Integrated Ex-vessel Source Term Analysis with the CONTAIN 1.1 Computer Code K. K. Murata, D. E. Carroll, F. J . Schelling, K. E. Washington, and G. D. Valdez Sandia National Laboratories Albuquerque, NM 87185 l

l ABSTRACT 1

The CONTAIN code is a system-level code developed for the USNRC and is intended to provide best-estimate predictions simultaneously for the thermal hydraulic conditions and fission product distributions within the con-l tainment building of a nuclear power plant during a severe accident. The l first version of the code, CONTAIN 1.0, in addition to the basic thermal l hydraulic and fission product behavior modules, provided models for en-gineered safety features appropriate to PWR containments.

In a forthcoming new version, CONTAIN 1.1, a number of new models and improvements in old models will become available. The ability of the code to handle calcula-tions involving BWR containments will be significantly improved with this

new version.

l l The most significant of the new features are:

1) The CORCON-VANESA models which treat the thermal hydraulic and fission product sources from the core-concrete interaction will be available as modules within CONTAIN and will replace the regular CONTAIN lower-cell modules. Radiative transfer from the top of the debris and the boiling of an overlying coolant pool, if any, may be treated through improved models available for any CONTAIN cell. The pool scrubbing of aerosols in conjunction with CORCON will be handled exclusively through the VANESA SCRUB model, i

Technadyne Engineering Consultants, Inc., Albuquerque, NM This work supported by the United States Nuclear Regulatory Commission and performed at Sandia National Laboratories which is operated for the U.S.

Department of Energy under contract number DE-AC04-76DP00789.

11-5 l

1

r

2) A new suppression pool vent model will describe the vent flow between the drywell and the wetwell of a BWR. In order to give the user flexibility in the aerosol scrubbing by the pool, the SPARC code has also been incorporated into CONTAIN. The user will be able to select either the VANESA SCRUB model or the SPARC module to describe this scrubbing. I l

3) The user wil?. be able to specify primary coolant system sources J directed to the suppression pool through the Safety Relief l Valves (SRV) of a BWR. These sources are scrubbed prior to introduction to the wetwell cell atmosphere, much like the drywell-wetwell vent flow.

Significant improvements in existing models will also become available:

1) A new type of radiative transfer model which handles multiple reflections (a net enclosure model) will be available.

2) An improvement has been made in the specification of fission product hosts, allowing more realistic modeling of fission product ;cansport.

3) A new pool boiling algorithm will be available. The algorithm is fully implicit with respect to pressure and coupled to gas flows. The potential for numerical chugging is therefore almost completely eliminated.

4) A new type of CONTAIN flow path will allow the specification of an unrestricted number of independent parallel flow paths between any tvo CONTAIN cells.

In order to demonstrate the operation of the maj or new features, a calculation of a hypothetical TB-sequence for the Peach Bottom BWR has been carried out. This calculation is strictly for demonstration purposes and not necessarily a best estimate. However, the conditions of the calculation are selected to be reasonable ones for this sequence. A core melt, which results in fission product release in aerosol form to the suppression pool, is followed by vessel failure and direct attack of the drywell basemat concrete by the core debris. Primary containment fails shortly after vessel failure, and the depressurization causes the suppression pool to flash. The subsequent pressurization of the secondary containment causes it to fail.

The fission products evolving subsequent to failure of the primary containment bypass the suppression pool. However, natural depletion of these fission products occurs within secondary containment prior to escape to the environment. The SPARC module is used to calculate the suppression pool scrubbing of aerosolized fission products. The CORCON-VANESA module is used to calculate the evolution of gases and fission products during the core-concrete interaction. A detailed 9-cell nodalization of the secondary containment developed at Oak Ridge National Laboratories is used to calculate the retention in the secondary containment.

11-6

SAND 86-2293A SUSTAINED URANIA-CONCRETE INTERACTIONS (SURC):

EXPERIMENTS AND ANALYSIS D. R. Bradley*

E. R. Copus*

Sandia National Laboratories Albuquerque, NM 87185 ABSTRACT Introduction In the event of a severe reactor accident in which molten core debris penetrates the reactor vessel, the interaction of the molten debris with structural concrete in the reactor cavity is an important factor in the reactor containment loading and aerosol source terms associated with the accident. Since the time of the Reactor Safety Study [1], this aspect of reactor safety analysis has been poorly understood, with little substantive experimental data available. Out of necessity then, computer models were initially developed based on data from simulant experiments and on observations from the few existing melt-concrete experiments. The CORCON(2] and VANESA(3) computer models were developed using this limited data. base.

In the last two years, experiment programs at SNL and at Kernforshungszentrum Karlsruhe (KfK) have investigated prototypic melts interacting with concrete. Experiments at both facilities have been well-instrumented and have yielded an abundance of useful data. The availability of these data has allowed validation of CORCON and VANESA and further model development. However, there are still important gaps in the experimental data base for molten core-concrete interactions.

Specifically, little or no data is available for sustained prototypic oxide melts interacting with concrete or for the effects of Zirconium (Zr) oxidation on melt-concrete interactions. Zirconium has been found in previous CORCON/VANESA calculations [4] to have a major influence on almost every aspect of the interaction from melt-concrete heat transfer to aerosol and fission product release. The SURC

• This work supported by the U.S. Nuclear Regulatory Commission and performed at Sandia National Laboratories, which is operated for the U. S. Department of Energy under Contract number DE-AC04-76DP00789.

l 11-7

oxpsrimants at Scndia National Laboratories cro dasignsd to extend the data base in these areas and consequently, to provide validation of the models in CORCON and VANESA.

This paper reports recent results obtained from the first experiment in the SURC test series, SURC3.

SURC3 Experiment Description and Results The SURC3 experiment was designed to examine the interaction of a sustained stainless steel melt with concrete both before and after the addition of zirconium. The interaction crucible used in the experiment had a limestone concrete test bed approximately 20 cm in diameter and 40 cm deep. The test bed was surrounded by an Mg0 annulus that was 10 cm thick and 90 cm high. This crucible design limits concrete erosion to the axial direction. A .3 m3 top hat was sealed to the Mg0 annulus to direct aerosol and gas effluents through the instrumentation stream. The charge in SURC3 was composed of 50 kg of stainless steel and 2 kg of fission product mocks. An additional 5 kg of Zr metal was added to the charge during the tect. The charge was inductively melted and sustained using a 55 cm induction coil and a 250 kW-1000 Hz power supply.

Instrumentation for the test consisted of 50 thermocouples embedded in the concrete to measure pool temperatures and axial erosion; 40 thermocouples in the Mg0 and top hat to measure sideward and upward heat losses; five independent flow measurement devices; gas composition measurement via an infrared C0/C02 monitor; and aerosol source term measurements using cyclone impactors, filters and a photometer.

The experimental procedure for SURC3 was to first melt the steel and allow it to erode 6 to 8 cm of concrete. Zirconium metal was then added to the melt to determine the impact of Zr oxidation on concrete erosion, gas flow, gas composition, and aerosol and fission product release. Post-test inspection of the crucible and its contents revealed that only 1.5 kg of Zr had actually entered the melt during this phase. After Zr oxidation had apparently been completed and the interaction had returned to pre-Zr steady state behavior, the experiment was terminated by turning off the induction power supply.

Preliminary results of the test indicate that during the 30 minutes prior to Zr addition the concrete erosion rate was .14 cm/ min, the flowrate of effluent gases was 50 1pm, aerosols created an opacity of 25%, the gas composition was 85% Co - 15%

CO2, and pressure in the containment top hat was .75 psig. Two minutes after Zr was added to the melt, the flowrate jumped to 135 1pm, the aerosol opacity increased to 50%, pressure increased to 2 psig, and the gas composition shifted to 96% CO -

4% CO2 This excursion continued for approximately ten minutes, during which time the erosion rate averaged .4 cm/ min. After ten minutes of interaction, the pressure, flowrate, erosion 11-8

rato, and gas cc position all returnsd to thair origincl valuas for tha rocaindar of tha tost. Forty-fiva sinutes after the first Zr addition, an attempt was made to add an additional 5 kg of Zr. This attempt failed due to a thick buildup of overlying crust material in the MgO annulus. The test was terminated one hour later.

SURC3 Analysis Detailed analysis of the SURC3 experiment using CORCON and VANESA has not yet been performed. Results from CORCON and VANESA calculations vill be included in the final paper and will be presented at the neeting.

References

1. USNRC, Reactor Safety Study: An Assessment of Accident Risks in US Commercial Nuclear Power Plants, WASH 1400, NUREG 75/04, October 1975.

2. R. K. Cole, D. P. Kelly, and.M. A. Ellis, CORCON-Mod 2: A Computer Program for Analysis of Molten-Core Concrete Interactions, NUREG/CR-3920, SAND 84-1246, Sandia National Laboratories, Albuquerque, NM, 1984.

3. D. A. Powers, J. E. Brockmann, and A..W. Shiver, VANESA: A Mechanistic Model of Radionuclide Release and Aerosol Generation During Core Debris Interactions with Concrete, NUREG/CR-4308, SAND 85-1370, Sandia National Laboratories, Albuquerque, NM, 1986.

4. D. R. Bradley and A. W. Shiver, " Uncertainty in the Ex-Vessel Source Term Caused by Uncertainty in.In-Vessel Models," ANS/ ENS Topical Meeting on Thermal Reactor Safety, San Diego, CA, February 1986.

11-9

THERMAL-HYDRAULIC STUDIES ON MOLTEN CORE-CONCRETE INTERACTIONS

• G. A. Greene Brookhaven National Laboratory Department of Nuclear Energy Upton, New York 11973 Recent assessments of risk due to unterminated severe accidents in light water power reactors have indicated that the consequences of molten core-concrete interactions (MCCI) dominate the consideration of containment loads and performance, as well as the release of non-volatile fission product aero-sols to the containment building. The issues of containment pressurization, combustible gas generation, structural erosion, fission product release, and fission product decontamination are critical in the evaluation of risk as cur-rently being quantified by the Severe Accident Risk Reevaluation Program (SARRP). The program at BNL has focused on investigating several important aspects of MCCIs in an effort to support the integral melt-concrete programs as well as the CORCON and VANESA computer code development and verification programs at SNL. Experimental, analytical, and computational programs are currently under way to investigate and support validation of a variety of physical processes that occur during a MCCI, including interlayer heat and mass transfer, liquid-liquid boiling and pseudo-boiling processes, and aerosol formation and decontamination.

The issue of interlayer heat and mass transfer addresses the phenomena that occur primarily interior to the molten core debris itself. These include such phemonena as pool boilup and void fraction, onset and rate of entrainment between layers, rate of interlayer heat transfer, distribution of fission products, and mode of pool solidification and porosity. Results have indicat-ed that pool boilup and void fraction as calculated by CORCON are in excellent agreement with experimental data. Models have been developed to predict the onset of entrainment between liquid layers and, subsequently, - the rate of entrainment and the rate of heat transfer, both with and without entrainment.

Application to the reactor case indicates that entrainment and mixing may or may not occur, depending upon the prevailing thermal hydraulic conditions in the melt, and that provisions for such modeling should be incorporated into the CORCON code.

The issues of liquid-liquid film boiling and pseudo-film boiling address the phenomena that occur primarily on the periphery of the molten core debris, either at the melt-concrete interfaces or at the melt-coolant interface.

These include such phenomena as melt-coolant film boiling heat transfer with non-condensable gas flux, f requency and magnitude of melt-coolant energetic thermal interactions from a stratified, sparged state, gas film stability, and melt-concrete heat transfer. Results from melt-water film boiling studies have suggested that the coolant heat flux increases above the flat plate limit

• Work performed under the auspices of the U.S. Nuclear Regulatory Commission.

11-11

with non-condensable gas inj ection from below. Steam explosions occur fre-quently; the frequency and magnitude vary with gas inj ection rate. Results f rom melt-R11 film boiling studies support the contention that R11 boils sta-bly on liquid metal melts, even under the influence of non-condensable gas flux from below. The gas injection tends to increase the liquid-liquid boil-ing heat flux over the flat plate limit by two mechanisms; area enhancement and generation of turbulence in the vapor film.

The issues of aerosol formation and decontamination address the phenomena that govern the mechanical and vaporization mechanisms of aerosol formation as well as the decontamination processes in water pools. Simple, scoping studies of mechanical aerosol formation by bubbles bursting at a free interface have been performed. The mass rate of aerosol production has been found to exceed available models and data by a wide margin, possibly due to use of a high vis-cosity fluid instead of water. Additionally, studies of aerosol decontamina-tion by overlying water pools have begun to measure decontamination factors appropriate to MCCI aerosols. Although it was intended to generate a monodis-persed aerosol of 0.3 pa particles in 2 centimeter bubbles, it was observed that the bubbles always fragmented and the Al 02 3 Particles agglomerated. The resulting pool DFs were in the range 25-75. When bubble size and aerosol size were taken into account, these DFs were consistent with VANESA predictions.

11-12

ICE CONDENSER TESTING FACILITY AND PLANS L. D. Kannberg, B. A. Ross, E. J. Eschbach, M. W. Ligotke, and G. F. Piepel Pacific Northwest LatioratoryN '

Richland, Washington 99352

SUMMARY

As part of an investigation of engineered safety feature system fission product retention effectiveness, a facility is being constructed to experimentally evaluate aerosol deposition in the ice compartment of PWR ice condenser containment systems. The testing conducted in this facility will provide data for validation of the computer code ICEDF. The ICEDF code was developed as part of the high priority effort to reassess and update analytical procedures l for estimating accident source terms for nuclear power plants.

Careful consideration of validation requirements has led to the development of a test plan that spans the ranges of experimental conditions while optimizing the potential value of the experimental data for code validation. Accident conditions were reviewed in 1985 and a preliminary set of test design criteria were developed and reported at the Thirteenth Water Reactor Safety Meeting.

A methodology for development of the detailed test plan was included in the paper provided at that meeting.

Duringthelastyearthemethodologyhasbeenapp)liedinthedevelopmentof the detailed test plan (matrix of test conditions . The test plan is comprised of 15 cases. The test conditions are varied among the cases in such a fashion to ensure that the cases are widely dispersed from one another, making delineation of effects easier and reducing the number of tests required.

Various statistical test design codes were utilized in assembling candidate test plans. The computer code, ICEDF, was used to simulate the various cases comprising each of 10 candidate test plans. The test plan having the best technical and statistical features was selected for use in the planned testing.

Design of a testing facility was conducted during and following development of the test plan. The testing facility has unique requirements that placed a special burden on site preparation and necessitated fabrication of a special test assembly. A building having a 50-ft deep pit and over 70 ft of overhead enclosed conditioned space was identified for use. Various facility modifications have been undertaken to prepare the facility for the planned testing. A test section over 50-ft high, with a cross section accommodating the equivalent of four ice baskets has been fabricated and installed. Necessary thermohydraulic and aerosol generation and mixing equipment is being installed, as well as interconnecting piping. Other features of the facility include extensive data acquisition and instrumentation and various ice handling, test assembly, and safety equipment.

(a)0perated for the U.S. Department of Energy by Battelle Memorial Institute under contract no. DE-AC06-76RLO 1830.

11-13

The testing will require careful measurement of thermohydraulic and aerosol parameters to gain information not only on bulk aerosol deposition, but on various thermohydraulic and aerosol deposition processes and phenomena. Aerosol generation has received considerable effort, concentrating on identification of suitable aerosol materials and generation methods. Because code validation is the primary emphasis of the project, identification of aerosols materials and generation methods has focused on aerosols having a wide range of properties as represented in the test plan.

The facility is scheduled to be completed by March 1987, at which time testing will be initiated. It is anticipated that testing will be completed in 1988.

11-14

SUMMARY

OF KFK BETA EXPERIH2NTS AND ANALYSIS WITH THE WECHSL AND CORCON CODES H. Alsmeyer, M. Reimann, Kernforschungszentrum Karlsruhe R.K. Cole, J r. , Sandia National Laboratories The BETA experiments investigate the melt / concrete interaction in a large scale melt facility using simulated core melts with sustained internal in-duction heating. The experimental data were used to verify the corresponding computer codes, such as the KfK-code WECHSL and the Sandia-code CORCON by detailed analysis. Extrapolation of the results to a core melt down accident in a LWR is by application of the verified codes.

The BETA experimental test series was completed in early 1986 af ter 2 years of experiments.19 experiments were carried out, most of them with silicate concrete of German standard specification. The melt consisted of typical 300 kg steel with Cr and Ni and 150 kg of aluminium oxide with additions of SiO2 and Cao, produced by a thermite reaction with initially some 2000 C.

Sustained power input to the melt was by induction heating from up to 1800 kW to some hundred kW only, as postulated by the phenomena to be investigated.

The main experimental results for silicate concrete are:

- For melts with high power input and small influence of crusts at the melt /-

concrete interface, propagation of the melt is predominantly downward.

- Downward heat transfer is considerably higher than predicted by earlier modelling. The high heat transfer results in fast cooling of the melt, and in steady melt temperature close to the solidification temperature of the metal, even for power input to the melt being an order of magnitude higher than according to decay heat.

- Dispersion of the steel into the oxidic melt may occur for high gas fluxes, mainly dependent on density difference of the melt phases and on viscosity of the oxide. For low temperature melts with reduced gas production, the segregated state is the stable configuration.

- Experiments with low power input show the role of solidification processes.

Sidevard and downward propagation are found to be more balanced. Crusts are always permeable for the gases released f rom the concrete, and the gases continue to agitate the melt, thus establishing an effective heat transfer mode which keeps the long term temperatures of the melt phases low, that is close to the solidification temperature of the metal.

- The aerosol release is observed to be very low, with the exception of an initial aerosol peak during and shortly af ter pouring of the melt. Aero-sols are condensation aerosols f rom the hot melt.

- Material investigation shows the f ast oxidation of chromium, followed by the iron oxidation. The admixture of molten concrete to the oxide reduces the f reezing temperature below 1300 O C, as is also expected for the reactor material af ter some interaction time.

- Cas analysis gives the release rates of H 2, H 0, 2 CO, and CO2 with a small amount of CH4 . Hydrogen release, dominant in the high temperature experi-ments, is released also af ter crust formation at lower melt temperatures.

Cas composition is in thermodynamic equilibrium at 800 to 1000 K.

11-15

The BETA experimental series includes three exparicsnts with linnotcna con-crete, fabricated according to US-specifications. The most striking differen-ce from the silicate concrete experiments is the release of dense white aero-sols from a pure limestone crucible (CRBR concrete). This occured during l

heating and persisted even af ter f reezing of the melt, although reduced, for

more than 30 minutes af ter power off. The crucible was sectioned and shows a considerable amount of grey, very fine powder at the melt / concrete interface and in the bulk of the melt, while the upper surface of the crucible is covered by the white aerosols.

! These observations may be understood from the process of lime-burning oc-curing at temperatures up to 1000 *C. The resulting product Ca0 is a very sof t material, which is easily powderized and transfered from the wall into l

the gas stream, containing a certain amount of Na and K. The . process of line burning must also be considered with respect to the decomposition of the j concrete, for which current modelling assumes melting at 1500 *C. The loss of mechanical strength of the burned limestone and its subsequent removal by spalling or the agitating melt may reduce the decomposition temperature of

, the concrete as low as 1100 *C.

The experiments with lime / common sand concrete show that the presence of a i relatively small amount of SiO2 in the concrete reduces the aerosol release i considerably, probably by dissolving the Ca in the silica.

l Temperatures of the melt in limestone concrete experiments are found to be comparably low as in the silica experiments. Gas release is largely enhanced j by the decarboxylation of the calciumcarbonate. Radial erosion of the crucib-le is more pronounced than in the silicate experiments, and may be increased by spallation, at least in the upper crucible.

f Modifications of the computer codes were necessary to account for the observ-l ed f ast downward heat transf er and erosion. Both computer codes introduced a

! partial film collapse model, analogous to transition boiling, in addition to i file. model heat transfer at very high gas rates and nucleate boiling type of heat transfer at lower gas rates. Additional modifications were made for void behavior and oxidation kinetics of the gas-metal reaction.

j _

The codes give good agreement with measurements of cavity shapes, erosion j

rates, melt temperatures, and gas generation and composition for the silicate concrete experiments. However, for the limestone concrete experiments, pre-diction of cavity shapes and tas generation rates is less satisfactory. This may indicate that modelling of the limestone concrete decomposition needs further improvement.

l l 11-16

.- _. . . - - _ - _ - . - _ . - _ - - - - _ . . . _ - _ . - - - - - . . . . - . . _ - . - . . _ . ..--.~ -

RISK EVALUATIONS OF AGING PHENOMENA W. E. Vesely, Science Applications International D. G. Satterwhite, Idaho National Engineering Laboratory Work Supported by the U. S. Nuclear Regulatory Commission Office of nuclear Regulatory Research Division of Reactor Analysis Operations under DOE Contract No: DE-AC07-76ID01570 In the first phase of this work, approaches are being applied to evaluate the relative impacts of aging on risk. Also, models are being developed and are being demonstrated to quantify the time dependent, absolute impacts of aging on risk.

The evaluations of relative age impacts have as their objective, the quantification of the relative importances of aging causes to component and system failures and unavailabilities. Various systems have been evaluated for the relative contributions from aging. Operating systems in general show more contributions from aging related failures than do standby systems.

However standby systems also show important contributions from aging. Pumps, valves, and heat exchangers are among the components showing the greatest contributions from aging. Erosion and foreign material intrusion are among the most dominant causes of aging failures.

The objective of the time dependent analysis is to develop mechanistic models to quantify the absolute changes in component and system unavailability from aging. The models incorporate the effects from specific aging mechanisms and cover specializations which can utilize existing data.

Linear and non-linear time dependent aging models have been developed and have been applied to evaluate the effects of stress corrosion and foreign material buildup. Stress corrosion if not arrested is shown to cause an order of magnitude change in the unavailability after approximately 25 years. Foreign material buildup if not controlled is shown to result in unavailabilities of .01 to .1 after approximately 20 years. The effects of these aging mechanisms are compounded if multiple components are simultaneously aging.

In the second phase of the work, plans are to evaluate the relative importances of aging causes to accident sequences contributing to core melt.

With regard to the time dependent evaluations, plans are to extend the models to evaluate the changes in core melt frequency with plant age.

l l

12-1

REACTOR PROTECTION SYSTEM AGING:

RESULTS OF A PILOT COMMERCIAL PLANT STUDY P. T. Jacobs L. C. Meyer Idaho National Engineering Laboratory D. Larsen Duke Power Company Summary This paper summarized the status of studies to assess the aging impact on nuclear plant safety systems. The work is being performed by the Nuclear Power Aging Research Branch of the Idaho National Engineering Laboratory (INEL) and Duke Power Company under the sponsorship of the United States Nuclear Regulatory Commission (NRC), Office of Nuclear Regulatory Research.

The purpose of this facet of the research program is to conduct system engineering studies to satisfy goals of the NRC Nuclear Plant Aging Research (NPAR) program. Those goals are: (1) to identify and characterize aging and service wear effects which could cause degradation impairing plant safety; (2) to identify methods of inspection, surveillance, and monitoring which will detect aging degradation prior to loss of safety function; and (3) to evaluate methods to control aging or mitigate the effects of aging degradation.

The approach taken in the plant system studies was to establish a cooperative research program with a utility so that insights into the aging process could draw from an experience base larger than that I available in the public domain. A pilot research program was conducted for a single system, the Reactor Protection System (RPS), at one of the

! utility's plants to assess the feasibility of this approach and to l determine which utility information sources best met program needs.

The RPS was studies because of its central importance in initiating a reactor trip and all frontline and support system functions in the plant safety hierachy. The understanding of RPS and any aging related degradation of that function is a prerequisite to understanding system interactions within the safety hierachy.

• Work sponsored by, Division of Engineering Technology, Office of Nuclear Regulatory Research, United States Nuclear Regulatory Commission, Washington, DC 20555, Under DOE Contract No.

DE-AC07-761001570.

12-3

Data sources for this study involved several data bases and selected operating plant records. The data bases included Licensee Event Reports (LERs), Nuclear Power Experience (NPE), and Nuclear Plant Reliability Data System (NPRDS). Specific plant experience on the operation, testing and maintenance of these systems was supplied by the Duke Power Company.

The RPS is a four-channel system that receives redundant inputs from both nuclear and non-nuclear instrumentation. It initiates a reactor trip whenever any two of the four channels indicates a safety limit has been reached. The system protects the reactor core form fuel cladding damage and the reactor coolant system from high-pressure damage.

A functional description of given of a typical RPS trip string and representative instrumentation channels along with various component data related to aging. In addition, the stressors due to environment, operations and maintenance activities are shown on the channel one-line diagrams along with indicators of degradation.

Specific plant experience with aging system components and improved maintenance practices to mitigate the effects of system degradation are discussed. The role of equipment qualification, obsolescence, spare parts and operating schedules must be factored into the maintenance program along with the surveillance testing, repairs and allowable downtime. A good maintenance program reduces the aging impact on plant safety by periodic rejuvenation which returns systems to as good-as new condition.

This coupled with a good surveillance program to periodically monitor equipment condition are necessary to assure the safety design objectives are achieved.

12-4

OVERVIEW 0F RECENT OPERATING EXPERIENCE by Gary M. Holahan and Mark A. Caruso Office of Nuclear Reactor Regulation Nuclear Regulatory Commission

SUMMARY

This paper presents a review of significant events which have occurred at operating plants in 1985 and 1986 and examines the extent to which plant age has played a role. The paper also discusses the results of recent studies of experience at newly licensed plants. The events which are reviewed have been identified through the normal systematic event analysis program conducted by NRR and IE. Information regarding these events was obtained through followup by reviews conducted by NRC Resident Inspectors and special inspection teams as well as event reports submitted by licensees. The review focuses primarily on events which involved infrequent or complicated plant transients, significant failures in key safety systems or both in combination. The key safety systems include reactor protection system (RPS), emergency core cooling system, emergency electric power system, containment isolation system, and overpressure protection systems, i

Recent, as well as past, studies of reactor trip frequencies and other types of operating experience have shown that relatively high frequencies are likely in new plants with little accumulated operating time. In order to better understand all the factors which contribute to high frequencies in new plants, the authors have made a comparison of reactor trip frequencies between plants which went into operation in the 1960's and early 1970's and those which have gone into operation more recently. Trip frequency versus accumulated operating time for the two plant groups are compared to see the extent to which design differences (e.g. capacity, thermal margin) affect trip frequency.

12-5

) ac optanc3 if this atticl3, the

$4ther or rec 6p;ent acknowledges p U.S. Covernenent's right to Zfnonon2clusive, eovalty f ree tense in and to any copyright 2 .,6ng th. utics APPLICATION OF DIAGNOSTICS TO DETERM1NE MOTOR-OPERATED VALVE OPERATIONAL READINESS D. M. Eissenberg Oak Ridge National Laboratory As part of the NRC Nuclear Plant Aging Research (NPAR) Program, ORNL is carrying out a study of motor-operated valves (MOVs) used in safety systems of nuclear power plants. The primary objective of this study is to identify and recommend diagnostic methods for detecting and trending time dependent degradations and service wear (aging) of the MOV such that timely maintenance can be performed prior to loss of safety function.

The motor of an MOV transmits power to the valve through a complex drive train within the operator which serves the purpose of providing the needed

! torque to raise or lower the valve stem while protecting both the motor and the valve from damage or excessive wear. As a result of experimental investigations at ORNL utilizing two large MOVs, it was discovered that the electric current supplied to the motor during a valve cycle was a very useful diagnostic parameter for detecting and trending many d rive train load variations, within both the operator and the valve.

Utilizing a simple remote, nonintrusive current probe as a detector, the motor current signal is obtained during a valve actuation, conditioned to minimize 60 Hz interference, and plotted as both time domain and frequency domain plots. The resulting signatures are analyzed at four levels - mean value for a cycle, gross trends during a cycle, transients, and noise frequency spectra.

The results have been used to identify many drive train mechanical load features previously not detectable or requiring intrusive monitoring methods. Several types of degradation have been identified. The use of electric motor current signature monitoring has also been shown to detect mechanical load features of other devices, including a vacuum pump and a fan.

Future work includes developing a data base of diagnostics versus degradations obtained from both laboratory and field tests, and developing a generic logic for analysis of the MOV signatures, including criteria for continued operation.

Research sponsored by the office of Nuclear Regulatory Research, U.S., Nuclear Regulatory Commission under Interagency Agreement DOE 40-551-75 with the U.S.

Department of Energy under contract DE-AC05-840R21400 with the Martin Marietta Energy Systems, Inc.

12-7

l The Effects of Relay and Circuit Breaker Aging in a Saf ety-Related System G. J. Toman, V. P. Bacanskas, T. A. Shook, and C. C. Ladlow (Franklin Research Center)

W. Gunther (Brookhaven National Laboratory)

In conjunction with the U.S. NRC's Nuclear Plant Aging Research Program, Franklin Research Center analyzed the aging of circuit breakers and relays, and the effects of this aging on a safety-related system. The work was performed under contract to Brookhaven National Laboratory. The age-related deterioration of relays and circuit breakers was evaluated through analysis of the designs and applications of relays and circuit breakers in plants, evaluation of manufacturers' instructions, and analysis of failure reports.

The primary failure causes were determined. The safety injection system for a pressurized water reactor was evaluated to see if operation and testing of the system added to the rate of aging of the relays and circuit breakers.

Failures that could result from aging were evaluated to determine if failures that prevent system operation would occur under hypothesized accident conditions.

The basic types of relays used in safety-related systems are protective, control, and timing relays. Protective relays detect abnormal conditions on the plant's power system and initiate opening of circuit breakers to prevent damage to the protected equipment such as motors, buses, and transformers.

Control relays are used in the logic and safety system actuation circuits of nuclear plants. Control relays are basically two-position relays with contacts that transfer position when their coils are energized. Timing relays are used to delay or sustain a signal for a specific period in accordance with system operating requirements.

The predominant failure mechanisms identified through Licensee Event Report (LER), Nuclear Plant Reliability System (NPRDS), and In-Plant Reliability Data System (IPRDS) failure data evaluation relate to setpoint drift for protective and timing relays, and cell burnout, binding, and contact problems for control and timing relays. The evaluation of NPRDS data indicated that normally energized relays fail approximately twice as often as normally deenergized relays. Thermally induced damage of organic coil and contact carrying components causes the energized relays to f ail faster than their normally deenergized counterparts.

12-9

The two types of safety-related circuit breakers (CBs) are molded-case and metal-clad switchgear drawout CBs. Molded-case CBs are used in low-voltage applications up to 480 V for lower level distribution systems and low power loads. They are manually closed and may be manually opened; however, they open automatically during overload conditions. The metal-clad switchgear CBs are used in applications where large loads and high fault currents can occur. They can be closed by electrical signals and are tripped by signals from protective relays and plant control equipment. They are mechanically and electrically more sophisticated than molded-case CBs and are used on safety systems with voltages ranging from 480 V to 6900 v. The evaluation of failure data for the metal-clad switchgear indicates that j in-service failures are most often caused by electrical and mechanical control components rather than by the main contact or arc extinguishing components.

The failure data also indicate that the failure rate increases for most types of failure after the sixth year of service.

1 The aging interaction study of the safety injection system showed that each relay is connected to at least one CB for a power source and that each metal-clad CB is affected by at least one control and one protective relay and its control power flows through a molded-case CB. The most prevalent failure causes for the relays and CBs were evaluated for their effect on safety-injection system operation for five types of accidents. The study I

concluded that inadequate maintenance and testing of CBs and relays would have a significant effect on the probability of failure of a safety injection chain. Failure of redundant chains is not expected from failure of an individual CB or relay. However, there are multiple ways in which aging can affect relays and circuit breakers such that failures of more than one CB or relay could occur under accident conditions if aging were left unchecked.

Prevention of aging effects warrants the need for a strong maintenance and test program to prevent multiple age-related failures of CBs and relays from disabling the safety function through concurrent loss of redundant safety 1

i chains.

l 12-10

Nuclear Service Emergency Diesel Generator Aging Research J. W. Vause III D. A. Dingee Summarv This paper reports the progress of diesel-generator aging research currently underway at PNL.

Because equipment aging can potentially reduce the capability of a nuclear plant to operate safely, the Nuclear Regulatory Commission's Office of Research, Division of Engineering Technology, Electrical Engineering Branch, has established a comprehensive research program intended to evaluate the effects of ' aging on selected safety related components. As part of this program, entitled Nuclear Plant Aging Research (NPAR), research is being conducted to identify, evaluate, and recatmend methodologies to mitigate the aging of nuclear service diesel generators.

Diesel generators occupy a unique position in the NPAR Program. Even though they are on the list of components to be evaluated, diesel-generators are complex systems: acutally, small power plants within the larger nuclear f a c il i ty. As a complex of systems, the diesel-generator comprises the engine, aux 111ary systems, and interfacing equipment. This includes the fuel, lubrication, air start, cooling, and electrical systems as well as.special components such as the governor, turbocharger, and generator. An additional complication is that more than ten vendors and several diesel power systems contractors supply diesel-generators to the nuclear industry, each with several models, modifications, and applications in service.

Because of the complexity of this task, the identification of aging degradation was separated into two major phases: 1) an evaluation of current expert and operational experience; and 2) detailed component aging assessments.

Under Phase I of the research, an evaluation was performed of a representative sample of four data bases--Licensee Event Reports (LERs), Nuclear Plant Reliability Data System (NPRDS), Nuclear Power Experience (NPE), and Emergency Diesel Generator Component Tracking System (EDGCTS)--to formulate one largo database of f ailures evaluated for aging degradation. The purpose of the evaluation was to determine which components f ailed; why the components f ailed; if the f ailure was due to an aging mechanism; and what were the preliminary corrective actions that should be suggested. The methods to resolve these issues included the participation of expert diesel consultants, who reviewed the available data with knowledgeable objectivity and considerable past diesel experience. Phase II of the study is directing research toward selected components identified as problematic under Phase I.

Each f ailure, categorized by a predetermined and coded list of components and subcomponents, was evaluated for the primary and secondary causes of failure; primary and secondary recanmended corrective actions for the listed failure; and whether aging may have contributed to the f ailure. The types of f ailures and recommended corrective actions were also coded to represent a predetermined selection of evaluation categories for each component and subcomponent.

12-11

Accompanying each aging evaluation was a commsnts s:ction of up to 250 cords that amplified details of the evaluation.

The data was entered on an IBM-PC using the dBase III relational database management ', system. Using the dBase III programming language, programs were developed to f ormulate component / analysis and component / recommended-corrective-action matrices to total the number of f ailures for each component in each f ailure category and each associated recommended corrective action category.

Because the comments field was word searchable, many relevant topical areas could be segregated and reviewed.

Due to the diversity of information in the database, an interesting and valuable selection of computer sorts were possible. The resulting computer output provided an indication of the aging process and a basis for the aging analysis.

As a result, 50 matrices plus selected word searchable topical sorts were produced that eventually were categorized into four major groups of nuclear service diesel-generator aging indicators.

The results gained f rom analysis of the matrices indicate that the aging of diesel-generators: is observable, follows recognizable patterns, experiences changes in the type of degradation mechanism with time, is confined to a relatively f ew, major components and subsystems, and increases as a percentage of all f ailures with time.

Additionally, the follewing components were identified as most susceptible to aging degradation:

Governor Overspeed trip system Injection pumps Piping on engine Controls Starting motors Lubricating system pumps Cooling system pumps Heat exchangers Turbocharger The primary causes of diesel-generator aging degradation were, in order of significance: vibration, inferior quality component, adverse environment, and human error.

General trends were revealed and methodologies developed that provide realistic insights into diesel-generator aging. The results of this study can therefore di rect the course of future research and in-depth component aging evaluations.

12-12

Application and Testing of a Method to Quantify Code Uncertainty 0 Robert G. Hanson Idaho National Engineering Laboratory EG&G Idaho, Inc.

P.O. Box 1625 Idaho Falls, Idaho 83415 An effort to develop a statistically rigorous methodology for the quantification of computer code uncertainty sponsored by the United States Nuclear Regulatory Commission (USNRC) has been supported at the Idaho National Engineering Laboratory (INEL). The objective of the methodology is to quantify code uncertainty relative to experimentally measured data for general transient types.

The application of the methodology for statistically based code uncertainty quantification consists of four major subtasks. First the transient type is identified and the transient is subdivided into time intervals. The time intervals are characterized by a dominant thermal-hydraulic phenomena which remains constant throughout the time interval. Secondly, the transient is characterized by selection of key single valued and continuous valued parameters. Data consisting of the difference between the calculated and measured results for each key parameter are determined for each of the time intervals. The third step in the analysis is to statistically reduce the multiple measurement continuous valued key parameters (such as clad temperature) to characterize the code capability relative to the specific transient. The single measurement continuous valued parameters (such as system pressure) and the single valued parameters (such as peak clad temperature) are retained for analysis when data from several similar test from different facilities are available. Finally the composite data from a statistically significant number of simulations are analyzed to formulate code uncertainty statements for the key parameters for the type of transient and controlling thermal-hydraulic phenomena. The code uncertainty statement will contribute to the overall uncertainty determination for a full-sized plant application of the code.

To test the methodology a limited uncertainty study has been performed using the RELAPS/ MOD 2 computer code. The code was applied to small break type transients in the Semiscale and L0BI facilities. The small break transients were subdivided into three time intervals. The system response during the first interval was dominated by the subcooled break-flow phenomena; the second interval was dominated by the saturated break flow related phenomena; the third time interval was dominated by the refill related phenomena in the core region. Key parameters selected for the analysis were the system pressure, break flow-rate, minimum core inventory, and the clad temperature response. Typical results of the study are shown on the accompanying figure. The figure characterizes the overall code uncertainty statement for I

• Work supported by the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Regulatory Research, Under Department of Energy Contract No.

DE-AC07-761111001570.

13-1

the break flow-rate during the saturated break-flow time interval. The figure presents the mean of the calculated minus measured break flow for the small break tests. Also presented are the confidence and tolerance intervals. The confidence interval contains the mean response 95% of the time. The tolerance interval contains 95% of the data on the average. Both intervals are time dependent measures of the composite code uncertainty. The figure shows that the confidence interval contains the value zero throughout the time interval suggesting the code predicted the saturated break flow-rate with no appreciable bias. The confidence and tolerance intervals at the beginning of the time segment are wide due to the calculated transition response from sub-cooled to saturated blowdown.

13-2 ,

e

SANDIA CODE ACCURACY QUANTIFICATION i

AND ITS APPLICATION TO TRAC-PFl/ MODI ASSESSMENT

• L. N. Kmetyk, M. G. Elrickt. R. K. Byers and L. D. Buxton Sandla National Laboratories,6444 Albuquerque, NM 87185 The product of code assessment is a body of information containing comparisons l

between code predictions and measured data. Thus far, conclusions on code accuracy i

l from previous assessment studies have been mostly phenomenological and/or qualitative. Such conclusions are somewhat ambiguous and difficult to utilize i

directly by those engaged in plant safety-related subjects. There is increasing emphasis within the NRC-sponsored code assessment effort to formulate more coherent, quantitative conclusions on the capabilities and accuracles of the codes.

l Quantifying assessment analysis accuracy using a common method and a common set i of key parameters will allow results from a number of independent code assessors to be combined, will provide broad-based information on code accuracy for application in regulatory needs and power plant studies, and will define further code i development needs and priorities.

l The output of the accuracy quantification methodology being developed at Sandla is a set of time-independent accuracy statements for a selected continuous-valued key parameter in a given test or plant analysis, giving:

-- the average code accuracy or bias within a specified time region, and two bounds centered on this code blas, giving measures of

-- the time-dependent variation of the average code-data difference within that time region (e.g., the accuracy in predicting an average temperature), and

! -- any within-instrument uncertalntles and between-instrument variabilities in the original code and/or data results, in addition to the time dependence of the i

average behavior (e.g., the accuracy in predicting any particular thermocouple response).

The methodology includes a way of comparing the code result to data that allows for the possibility that the code result, while not exactly on the reported data, is still within the data range (which for a given transient evaluation usually would be defined as accurate enough). The approach taken is to compute the fraction of time that the code result falls within the data band. This fraction, or " accuracy percentage", provides an additional quantitative code accuracy evaluation tool, allowing easy objective defl..ition of such common assessment terms as "very good agreement", " reasonable agreement", etc., regardless of the magnitudes and units of the actual code and data values.

• This work was supported by the U. S. Nuclear Regulatory Commission and performed at Sandia National Laboratories which is operated for the U. S.

Department of Energy under Contract Number DE-AC04-76DP00789.

t Dikewcod - Division of Kaman Sciences Corporation 13-3

The potential applications of code accuracy quantification can be grouped into three distinct areas: j

-- individual analysis evaluation concentrates on objectively quantifying the code accuracy in a given assessment calculation, replacing such subjective results as the statement "the code did a reasonably good job predicting the test behavior";

-- code-to-code (or calculation-to-calculation) comparison allows judgements on f the relative merits of different codes and/or code versions, different l nodalizations, etc., measuring the relative ability of each to predict the required phenomena of interest in given transients; and

-- overall code accuracy conclusions provide a subsequent plant analysis accuracy estimate for regulatory personnel, by combining individual analysis accuracles from numerous analyses and experiments.

Sample applications of the described code accuracy quantification methodology to some individual continuous-valued key parameters have been done. Comparison of the final accuracy statements to the original measured vs calculated parameter plots provided an idea of what significant information from the original analysis is retained and/or lost after the accuracy quantification process. Applications included quantifying the code accuracy for single well-behaved parameters, for noisy variables, and for multiple code and/or data results, all from one of our completed TRAC-PF1/ MODI assessment analyses.

In another application, the described code accuracy quantification methodology has been used to compare the relative accuracles of different nodalizations (for the same test analyzed with the same code) and of different codes (for the same test analyzed with similar nodalizations). In each case studied, the quantitative code accuracy was evaluated using both single-valued key assessment parameters and l continuous-valued key assessment parameters; the quantitative assessment analysis results were then compared to our previous qualitative conclusions to see if the same conclusions were reached.

l The ultimate licensing application of quantitative accuracy statements requires combining individual accuracy statements into more global code accuracy estimates covering different major regions, different analyses, different plant designs, and different users. Although the limited data base of completed calculations available to us preclude any final conclusions, we have studied the problem of combining individual-transient accuracy statements from our available RELAP5/ MODI and TRAC-PFl/ MODI assessment programs, to help identify some of the problems to be considered in deriving overall code accuracy statements.

The process of development and implementation of an accuracy quantification methodology is a large problem with many facets. The methodology we have been developing is basically complete, but a few questions remain. Some of these questions have been partially addressed; all are important and will require eventual resolution.

l l

13-4

Tmt of th7 TRAC Coda Agninct Known Annlytical Solutions for Stratified Flow Mr- P S Black Prof D C Leslie Queen Mary College, London Prof G F Hewi.tt UK Atomic Energy Authority, Harwell Accurate predictions of the safety of Light Water Reactors under accident conditions are dependent on a detailed knowledge of in the dynamic arbitrary behaviour of complex mixtures of steam and water geometries.

such problems, even in the The absence of general solutions to simplest geometries, leads to a reliance on the results obtained from large two phase flow codes such Los as TRACNational Alamos (Transient Reactor Analysis Code), developed by the Laboratory.

the subject of criticism, However, these codes have been ranging from the accuracy of their finite-differencing schemes to the validity of their physical models; from their use of flow-regime maps to the very nature of the differential equations solved. In this paper we employ certain stratified known analytical two phase solutions to problems in horizontally flow as a means of addressing these criticisms, using detailed comparisons with code predictions to assess the range of applicability of results, and to highlight code deficiencies. The effect of varying spatial and temporal nodalisation also examined.is recognised as being of major importance, and is It is first shown that the time- and space-averaged equations of horizontal two-phase interfacial shear, identical with the flow are, in the absence of equations of standard non-linear shallow water theory. It is noted that setting the wall and interfacial statement is true for drag all to speeds.

flow zero in the code implies that this

( may also recognise Having recognised this we the solutions to these equations basedexistence of a number of analytical around the method of characteristics, the United Kingdom and in this we build on the work of Gardner of Central Electricity Generating Board.

Adopting his solution for drainage of a horizontal pipe (the dam break problem), and bore motion (the internal dam extending solutions fo r internal flow and break) to circular geometry provides the equations us with used sufficient in TRAC. information to confirm an error in When this is corrected it is seen that the results are in excellent agreement with the analytical solution, though the coarse mesh results, particularly for the internal dam break, do exhibit a degree of numerical diffusion.

13-5

Additionally, we have also examined the work of Ardron (also of the UK CEGB), who demonstrates that, under reasonable assumptions, a perturbation analysis of the one-dimensional TRAC equations gives results for the Kelvin-He lmholtz instability which are in per fect agreement with those from an exact two-dimensional solution. The prediction of this instability clearly represents a stringent test of any code, in terms of both the numerical scheme used and the accuracy of its coding. We have produced a solution valid in pipes of circular cross-section, for a range of mesh and have obtained code results a sufficiently fine spacings. These demonstrate that for nodalisation the analytical growth rate is approached, though a fully converged solution is unobtainable dueAdditionalto problems with problems ill-posedness of the basic equation set.

were encountered due to the non-conservative form of the finite-differencing, though these were circumvented through careful treatment of the initial conditions.

In general, TRAC results are in good agreement with theory, and any discrepancies that did exist have led to the detection and correction of errors in the coding. Having established these firm foundations we now feel confident enough to address the problems of flooding in horizontal pipes.

I l

l i

13-6

COLD LEG CONDENSATION IN A LARGE BREAK LOCA USING TRAC-PFl/ MOD 1 J.T.Dawson.

Central Electricity Generating Board, Berkely Nuclear Laboratories Direct contact condensation of steam on emergency core cooling (ECC) water is potentially an important phenomenon in determining the course of a Large Break Loss-of-Coolant Accident (LBLOCA). A key area where condensation might occur in a PWR is at the ECC injection point in the cold leg, where the incoming jet enhances local turbulence in the water giving high condensation heat transfer. This can reduce the pressure sufficiently to suck back the liquid away from the vessel towards the pump. When this liquid passes the injection point, there is far less turbulence at the steam-water interface and the condensation rate falls.

The pressure can therefore build up again and force the water back towards the vessel. These condensation driven oscillations have been detected in several large scale experimental LOCA simulations.

Work is being carried out at Berkeley Nuclear Laboratories to assess the condensation models used in the best estimate thermal hydraulics code TRAC-PFl/Modl. The 1-D condensation modelling has been tested against a small (1/30th) scale flow visualisation experiment. The use of bulk liquid parameters in the simplified condensation heat transfer correlations cannot take account of local turbulence enhancement at a plug interface from impinging jets or wall films, resulting in an under-prediction of condensation rate. The code also has no model to predict the observed interface shattering which can result in order of magnitude increases in condensation rate.

A simple algorithm is employed by TRAC-PFl/Modl to detect the existence of a liquid plug. This algorithm is not well based physically, and in particular it gives rise to two liquid plugs in adjacent cold leg cells where only one plug exists. This can result in reduced heat transfer coefficients compared with the normal flow regimes, and in plant calculations the occasional cold leg flow reversal is predicted when the algorithm flips from two plugged cells to one.

TRAC-PFl/Modl does not predict the sustained cold leg flow oscillations observed in the LOFT large break tests. This may in part be due to the plugging algorithm, as well as inadequate enhancement of turbulence in the ECC injection region. Ironically, the grossly over-simplified condensation treatment in the earlier code TRAC-PD2 predicts the LOFT oscillations reasonably well due to the use of inappropriate liquid velocities in the heat transfer correlation.

A sensitivity study was carried out on a Sizewell 'B.' plant calculation to investigate the effect of increasing the TRAC-PFl/Modl condensation coefficients in the cold leg due to the impinging ECC jet.

Condensation was also switched off when the plug moved upstream past the injection point. Although this gave some short term changes in the predicted course of the transient, and in particular a 2 second delay in the end of bypass, the two cases converged towards the end of the refill period. The plug in the cold leg simply moved to a new position (and hence had a different interfacial area) to give a similar condensation rate dictated by the steam supply.

I 13-7

Only one complete oscillation past the injection point was predicted in the sensitivity study, but it should be noted that the pressure boundary conditions were different to many of the large break experiments. However, a further TRAC calculation was carried out in which the onset of condensation was delayed until the plug had travelled about half a scetre downstream of the injection point. A very high condensation rate was then assumed until the plug had reversed back to the injection point. This assumption, which was intended to represent interface shattering as observed in some of our experiments, introduced a phase lag which resulted in sustained oscillations.

i c

13-8

l Some Preliminary Results of Post-Dryout Heat Transfer Measurements at Low Qualities and Pressures up to 20 bar K G Pearson D Swinnerton R O'Mahoney UK Atomic Energy Authority, Winfrith Steady state data have been obtained on pos t-dryout heat transfer for upwards flow in a tube of inside diameter 10 mm and length 920 mm. The experiments covered mass velocities up to 200 kg/m2 s at 10 and 20 bar and up to 1000 kg/m2 s at 2 and 5 bar.

Inlet qualities were close to zero and the equilibrium quality at exit ranged up to 60%. The tube was prevented from rewetting by massive copper hot patches, brazed to it at each end. Surface temperature measurements were made along the length of the tube by precision spot-welded bare wire thermocouples. An in-stream thermocouple inserted into the flow at exit from the tube provided a measure of vapour temperature at this location.

Typical sets of data are presented and the trends discussed.

Comparisons are made with predictions obtained using the TRAC-PFl/ MODI computer code. This code is used to carry out pressurised water reactor safety calculations such as postulated los s-of-coolant accidents. The prediction of post-dryout heat transfer is important in calculating the blowdown and reflood phases of such accidents. These new data extend the range of the available database against which the combined effects of the heat transfer and hydraulic models in the code can be assessed.

Selected tests have been simulatd by using a TRAC ' rod' to represent the tube heat transfer surface. The experimental fluid inlet conditions and nett heat input were specified as boundary conditions. The calculation was then run as a transient i.,ntil a steady state was reached. The results are presented in the form of total heat transfer coefficient against egilibrium quality.

This permits a direct comparison with the experimental results.

Comparisons are also made with predictions obtained using the reflood code BERTHA.

13-9

JRC Ispra Experience with the IBM Version of RELAPS/ MOD 2 W. Kolar, H. Stsdtke Joint Research Centre, Ispra Establishment I-21020 Ispra, Italy The RELAP5 code has been used extensively within the LOBI project at the Joint Rescrach Centre (JRC) in Ispra. The code is used mainly for test design calculations, pre-test predictions and post-test analysis. The results obtained represent an important contribution to the multi-national effort for the RELAPS code assessment and verification. )

The use of RELAP5 within the LOBI project started in 1983 with the code {

version RELAP5/ MOD 1 which was converted from CDC to IBM in order to allow the code to run on the JRC AMDAHL computer. Since than several updates have been implemented up to cycle 25. Severe deficiencies obtained with RELAP5/ MOD 1 were the main reason for a number of model improvements which have been incorporated, leading to the JRC Ispra version of RELAP5/ MOD 1 denoted RELAP5/ MOD 1-EUR. This code which shows substantial improvements with regard to reliability, accuracy and code running times compared with the original code version, has been the main analytical tool within the project over the last few years.

In 1985/86 the RELAPS/ MOD 2 code was converted from CDC to IBM and a number of test cases have been calculated using the CDC and the IBM version of the code in order to demonstrate the correct implementation of the code on the AMDAHL computer at Ispra. A copy of the RELAPS/ MOD 2 IBM version was sent to the Idaho National Engineering Laboratory in June 1986.

In order to examine the prediction capabilities of RELAP5/ MOD 2, post-test predictions have been performed for three different LOBI tests which cover a wide spectrum of LOCA conditions:

Test A2-81: 1 % small break LOCA experiment (ISP 18 test)

Test Al-83: 10 % intermediate break LOCA test Test Al-04R: 2 x 100 % large break LOCA test.

13-11

The RELAPS/ MOD 2 results show considerable improvements compared with the RELAP5/ MOD 1 data. The unphysical (numerical) oscillations typical of RELAP5/ MOD 1 results are largely reduced and a generally better agreenient is obtained with experimental data. In addition, the code run times are significantly reduced for slow transients. Frequent code problems due to steam table failures for small break LOCA calculations could be avoided by minor code modifications.

In the paper, RELAPS/ MOD 1 and RELAPS/ MOD 2 results will be compared with measured key parameters for the three LOBI test cases and the deviations between measured and predicted data will be analysed. Recommendations will be given on the extent to which the model improvements in the JRC Ispra version of RELAPS/ MOD 1 may be incorporated also into RELAP5/ MOD 2.

j 13-12

ASSESSMENT AND UNCERTAINTY IDENTIFICATION FOR RELAP5/M002 AND TRAC-BDl/ MODI CODES UNDER UNC0VERY AND REFLOODING CONDITIONS S.N. Aksan, M. Richner, G.Th. Analytis, M. Andreani*

Swiss Federal Institute for Reactor Research (EIR)

CH-5303 WUrenlingen, Switzerland

• Swiss Federal Institute of Technology (ETH)

Nuclear Engineering Laboratory, j CH-8092 Zurich, Switzerland i

Summary Assessment calculations for the thermal hydraulic transient computer codes RELAP5/ Mod 2 (frozen version 36.02) and TRAC-BDl/Modl (frozen version 22) were perfonned, at Swiss Federal Institute for Reactor Research (EIR), under both core uncovery (boil-off) and reflooding conditions. The aim of the work being to assess the predicting capabilities of the frozen versions of the best esti-mate computer codes.

4 Some of the reflooding and boil-off experimental data obtained from NEPTUN test facility at EIR are used for the assessment work. Model optimization cal-culations on nodalization and effect of available options (e.g. heat slab sizes) are performed with a selected base case and the same model is applied to the rest of the other experimental cases covering wide range of parameters.

In this paper, we shall be reporting the results of these assessment calcula-tions and we shall also identify and point out the existing uncertainty areas in boil-off and reflooding phenomena. On the basis of the further calculations by changing some of the models such as bubbly / slug flow interfacial friction and film boiling heat transfer correlations in the frozen versions of the codes, improvements which eliminate some of the problems will be recommended.

13-13

Status of J-TRAC Code development Hajime AKIM0TO, Takamichi IWAMURA, Akira OHNUKI, Yutaka ABE , Yoshio MURA0 Japan Atomic Energy Research Institute

1. Introduction J-TRAC program is a code developmental program which started in 1984 at Japan Atomic Energy Research Institute to develop a standard code for reactor s fety assessment. J-TRAC will be developed by using TRAC-PFl/ MODI code ) as the framework of the code and by improving the physical models for the refill and reflood phases of a PWR LOCA based on the physical understanding of the phenomena. J-TRAC code is expected to be used for the calibration of the simplified best-estimate code or the licencing code and the simulation of the transient in PWRs.

In Fy 1984, TRAC-PFl/ MOD 1 listing has been investigated on the code structure, the numerics and the constitutive relations to understand details of TRAC-PF1 code. In Fy 1985, the ef forts in J-TRAC program were focussed on the assessment of the predictive capability of TRAC-PF1 code for typical large break LOCA experiments, the improvement of the reflood model and vectorization of TRAC-PFl/ MOD 1 for 3D calculation. In this presentation, the followings are explained; (1)

Assessment of TRAC-PFl/ MOD 1 code for typical large break LOCA experiments (2) Improvement of reflood model

2. Assessment of TRAC-PFl/ MOD 1 for large break LOCA experiments In order to understand the predictive capability of TRAC-PFl/MODl, systematic assessment calculations were perf ormed for typical large l break LOCA experiments including IDFT L2-3 and L2-5, Semiscale S-05-3, l Marviken Test 4 and Test 24, CREARE 1/15-scale downcomer test, a ECC mixing test, CCTF base case test, FLECHT SET F2714B, FLECHT SEASET Run 31805B, SCTF radial power profile ef fect tests, SCTF tie plate CCF test. As the results of these assessment calculations, the following areas are identified where the future improvements of the physical models are required for the reasonable agreement with test results; (a)

Interfacial friction model in core during reflood (b) Core heat transfer model during reflood (c) CCFL model at downcomer and tie plate (d) Direct contact condensation model (e) Physical properties in high pressure range. It is also found the necessity of the speed-up of the 3D calculation for practical use of the code in reactor safety assessment anal yses.

The work was performed under contract with the Atomic Energy Bureau of Science and Technology Agency of Japan.

13-15

- =_ _. ., , .

3. IEprovem2nt of reflood modal

! TRAC-PF1 has model deficiencies to predict core thermal hydraulic behavior during the reflood in a PWR LOCA.ge The main items need to be improved are (a) the hydraulic model related to the core water accumulation behavior and (b) the wall heat transfer model.

The interf acial friction model is relaced with thegrrelation equivalent to the Murao-Iguchi void fraction correlation. The heat transfer correlation for t film boiling regime is also replaced with 1 Murao-Sugimoto correlation in J-TRAC. Both models are used in the REFLA code, whick5h"* ""

reflood analysis.

Figures 1 and 2 show the caculated results for CCTF flat power test with J-TRAC and original TRAC-PF1. J-TRAC result shows excellent agreement with CCTF results for the core water accumulation and the core cooling behaviors except the period right after the reflood initiation. With the modification of the interfacial friction and film i boiling models, the prediction capability for the reflood phenomena is j i highly improved. The model improvement also resulted in the less CPU time due to the stabilization of the transient which allows the

! calculation with the big _ time step size. In case of the analysis of the CCTF flat power test with ID CORE model including ten cells, J-TRAC was about 40 times faster than original TRAC-PF1.

References (1)Liles, D.,et al.: NUREG/CR-3567,LA-9944MS, February 1984.

(2)Akimoto,H.: Nuclear Engineering and Design 88, 215 (1985).

(3 )Mure o ,Y. ,Iguchi,T. : J. Nucl. Sci. Tecnol. 19(8), 613 (1982).

(4)Murao,Y.,Sugimoto.J.: J. Nucl. Sci. Technol. 18(4), 275 (1981).

(5)Murao,Y.,et al.: JEARI-M 84-243, February 1985.

I i

U

] 'l I lj il l 1300 g N

(

TR10 (ID CORC) k TRAC 03 -

Q t10 CORib7 "*

~

g y J. TRAC E \ J.TA10

y y '(ID CORE)

(ID CORE! j ,

heeomred h E50 heemred :b /

2  ?

/

TRAC (vtsstL) \N\

y,%

1. ' ' O '

t Ll s

/

l Devoten t 22- 1.E3 m 0 4 l

D 10 0 200 300 400 500 500 Time is) Time is1 Fig. 1 Void fraction in core Fig. 2 Clad temperature at midplane 13-16

APPLICATICN OF EN3INEEDING AND BEST-ESTIMATE CODES TO HOR THERMAL MIXIN2 EXPERIMENTS TEMB L. Wolf; U.Schygulla; Project HDR KfK, FRG W.Hufner, K.Fischer: Battelle-Frankfurt, FRG W.Baumann, INR/KfK, FRG T.G.Theofanous. UCSB, USA During Phase II of the HDR Safety Program, a total of 26 thermal mixing experiments have been performed at the HDR-facility. These large-scale experiments were performed with three different HPI-nozzle arrangements (with different distances from the downcomer) over a broad spectrum of Froude-numbers (0.00625 < Fr CL 10.27) for both finite loop flow (natural circulation in the sy-stem) and stagnant loop conditions. Only one cold leg loop was simulated at the HDR, thereby resulting in an asymmetric cooldown behavior of the downcomer. The local and global transient response of the three-dimensional HDR-vessel with built-in core barrel was measured both with respect to fluid and wall temperatures.

Parallel to the international experimental research (CREARE, SAI,IVO) to provide a broad data base for resolving the PTS issue, substantial analytical efforts were put into the development, verification and application of best-estimate, multi-dimensional codes, such as COMMIX-1B and SOLA-PTS. These efforts were supple-mented by the development of fast, inexpensive engineering-type of codes concentrating either upon special phenomena such as HPI-jet mixing under cross-flow (JETHIX by Battelle-Frankfurt based on Kim's model EPRI), or using a zonal approach like VOLMIX (Battel-le-Frankfurt based on Sun, Oh model of EPRI) together with empi-rical correlations for entrainment primarily deduced from small-scaled facilities thus far. A different approach has been taken in j

the REMIX-Code by Theofanous which uses correlations developed on the basis of numerical solutions of the basic set of conservation equations for planar plumes accounting for a three equation turbu-lence model.

Prior to the HDR-experiments.all of the computer codes had been more or less validated against the results of small-scale faci-lities by means of post-test computations only.Due to its large scale and close to realistic fluid conditions, the results of the HDR-TEMR test series constitute the only data set worldwide at the high end of the. injection pressure spectrum. Therefore, the.HDR-data deemed ideal for code verification purposes.

In a truely multi-national analytical effort (USA-FRG) the fol-lowing codes: COMMIX-1B (INR/KfK); SOLA-PTS (BF); JETMIX (BF);

REMIX (Purdue Univ., UCSR, USA); VOLMIX (BF) were applied to the HDR-experiments under strictly blind pre-test conditions. First, the four experiments constituting the Preliminary Test Phase were analyzed and after their data became available, additional post-test calculations were performed with the engineering-type codes.

With some noticeable improvements, primarily for VOLMIX, the engi-13-17

neering-type codes were applied for pre-test predictions of the Main Test Phase. Due to expenditures and timing the best-estimate codes COMMIX-1B and SOLA-PTS have been used only for blind pre-test predictions of a few selected HDR-experiments, thus far, with some post-test analyses expected to be completed in the near future.

Nevertheless, the selection of tests for these exercises covers a rather broad spectrum of overall tested conditions and thereby allows a rather fair judgement of the codes predictive capabili-ties for a variety of injectiori modes under different flow condi-tions.

Under the rather stringent conditions imposed the following con-clusions can be drawn from the comparisons between experimental data and calculated results:

1. The so-called learning effect from the Preliminary Test Phase was larger for engineering-type codes than for best-estimate codes. Among the engineering-type codes those using purely empirical informations, such as VOLMIX were more affected than for instance REMIX with a more basic approach.

2. Once adjusted, all engineering-type codes show about the same type of pre-test predictive capabilities for the experiments of the Main Test Phase.

3. Certain ambiguities still remain by comparing local data with zonal results.

4. For the specific HDR-asymmetric cooldown, best-estimate multi-dimensional codes are needed to derive the thermal loa-ding function for structural analysis.

5. Best-estimate code results are surprisingly close to the mea-sured data; reasons for deviations have been deduced from the comparison with data and will be factored into the post-test analyses.

6. Both computational approaches ideal]y supplement each other for conditions examined at the HDR.

7. For both types of codes, verifications on behalf of HDR-data means one step further in extrapolation to real plant condi-tions for the range of conditions examined.

Advantages and disadvantages of the various codes and models will be identified on the basis of the consistent comparisons. Over-all, it will be demonstrated that the applications of both best-estimate and engineering-type methodologies provide the potential user with the opportunity to choose from a variety of tools depen-dend upon his resources and needs for insights into physical phe-nomena.

13-18

QUASAR UNCERTAINTY STUDY M. Khatib-Rahbar, C. Park, R. Davis, H. Nourbakhsh M. Lee, E. Cazzoli, and E. Schmidt Department of Nuclear Energy Brookhaven National Laboratory Upton, New York 11973 Over the last decade, substantial development and progress has been made in the understanding of the nature of severe accidents and associated fission product release and transport. As part of this continuing effort, the United States Nuclear Regulatory Commission (USNRC) has sponsored the development of the Source Term Code Package (STCP), which models core degradation, fission product release from the damaged fuel, and the subsequent migration of the fission products from the primary system to the containment and finally to the environment.

In order to better establish the validity and potential applications of source term predictions from these phenomenological models, quantification of the uncertainties associated with the STCP calculated source terms is essential.

An initial attempt at quantifying the uncertainties in the source term estimates based upon the BMI-2104 1 methodology was completed at Sandia as part of the QUEST 2 program. However, this study was preliminary and limited in scope. Comparable studies have also been performed by the Nuclear Industry as part of the Industry Degraded Core Rulemaking (IDCOR) prograin using the MAPP code.3

~ The objectives of the QUASAR (Quantification and Uncertainty Analysis of Source Terms for Severe Accidents in Light Water R_eacEors) prograiii are: (1) to address the uncertainties associated with input parameters and phenomeno-logical models used in the STCP, and (2) to define reasonable and technically defensible ranges and assumptions for the use in the STCP.

The QUASAR study consists of:

1. Screening Sensitivity Analysis: This stage is necessary to reduce the number of input variables to a manageable size. This has been accomplished by parametric sensitivity studies on the various codes in the STCP. The results of the QUEST program were a useful starting point for this exercise. The most sensitive input variable are then selected for detailed uncertainty analysis.

2. Uncertainty Analysis: This stage consists of: (a) Identification and Classification; (b) Quantification; and (c) Propagation. Identi-fication and Classification of uncertainties entails a detailed exam-ination of the various models and their associated _ computer codes in the STCP. In general, QUASAR addresses uncertainties in input param-eters, modeling options, and unmodeled phenomena. The quantification 14-1 l

process in QUASAR uses the available experimental data-base followed by an extensive expert review process to establish reasonable upper and lower bound estimates together with Probability Density Functions (PDFs) for the sensitive input parameters to the STCP. The propaga-tion of input uncertainties through the STCP is accomplished through a stratified Monte Carlo simulation.

i It must be noted that, in assigning the PDFs, various dependencies between the parameters are included. Furthermore, uncertainties due to significant unmodeled physical phenomena (models not included in j

the STCP) will also be addressed in the second phase of the study.

3. Output Sensitivity Analysis: Following the completion of the uncer-tainty analysis stage, a sizeable number of STCP generated samples become available which enables the use of a regression type tech-nique, such as the Response Surface Method (RSM) for sensitivity  ;

analysis. In this stage, the sensitivity of the output PDFs to the '

i input PDFs are established.

4. Importance Analysis: In this stage of the study, an importance rank-ing of the sensitive input parameters /models are established by defining an appropriate unit of importance.

The resulting PDFs for the radionuclide releases are then used to calcu-late corresponding statistical parameters such as the mean, median, and upper i and lower percentiles.

This comprehensive methodology is initially being applied to the three

accident sequences summarized in Table 1. Other accident sequences will also be studied, with the objective of establishing the overall uncertainties asso-

! ciated with severe accident source terms in LWRs. Application of the QUASAR l approach to the MELCOR severe accident computer code is also envisioned.

I i

Table 1 Plants and Accident Sequences to be Studied in QUASAR

! Plant Reactor Type Containment Type Accident Sequence Peach Bottom BWR Mark I Anticipated Transient Without Scram (TC)

Sequoyah PWR Ice Condenser Small Break LOCA with Recirculation Failure (S 2HF)

Sequoyah PWR Ice Condenser Interfacing Systems LOCA.(V) 14-2

References l

1. J. A. Gieske et al., "Radionuclide Release Under Specific LWR Accident '

Conditions," BMI-2104 (1985).

2. R. J. Lipinski et al., " Uncertainty in Radiological Release Under Specific LWR Accident Conditions," SAND 84-0410 (1985).

l

3. "MAAP Uncertainty Analyses," Fauske & Associates, Inc. (April 1985).

)

4. M. Khatib-Rahbar, "Quantification and Uncertainty Analysis of Source Terms for Severe Accidents in LWRs (QUASAR), Part I: Methodology and Program Plan," NUREG/CR-4688, BNL/NUREG-52008, Vol. I (1986).

5. J. A. Gieske et al . , " Source Term Code Package: A User's Guide,"

NUREG/CR-4587, BMI-2138 (1986).

14-3

MELPROG/ TRAC UPDATE AND APPLICATIONS

• by R. J. Henninger Safety Code Development Group Energy Division Los Alamos National Laboratory Los Alamos. New Mexico 87545 and

.l. E. Kelly Sandia National Laboratories P.O. Box 5800 Albuquerque. New Mexico 87175 In a joint project. Los Alamos and Sandia National Laboratories are developing the MELPROG PWR/ MODI code to predict within-vessel behavior under degraded core conditions and the linked MELPROG/ TRAC to provide

, predictions of the entire system for light-water reactor (LWR) accidents. We used these codes to perform the l first complete, coupled. and mechanistic analysis of a TMLB' (station blackout) core meltdown accident for the Surry plant. The calculation began when boiling started in the core region and ended when the reactor vessel failed. Most of the important phenomena that occurred during the accident sequence were modeled.

l The important exceptions were a treatment of motion before major disruption of the fuel rods (candling) and a l treatment of the fission-product release. transport, and deposition (as implemented by the VICTORIA module that is being incorporated in MELPROG). Although this calculation is preliminary. it does demonstrate the advanced capabilities of this version of MELPROG.

The " base-case' calculation provided the timing of the major events that occurred during the accident.

the amount and timing of hydrogen production created by oxidation of Zircaloy cladding, and the condition and composition of the disrupted material when the vessel failed. Because this is the first preliminary calculation, we performed a limited number of auxiliary MELPROG and linked MELPROG/ TRAC calculations. The base-case calculation. the auxiliary calculations. and a comparison of their results with previous calculations have~provided further insights into this accident.

In particular, the calculations have shown that natural convection reduces the rate of core heating but increases the rate at which the upper-plenum structures heat. This result implies that a significant amount of core energy is deposited in the plenum and primary piping. This increased heating can inhibit fission-product tieposition and can increase the amount of molten structural steel in the debris at vessel failure. We also have shown that coupling between the vessel and primary system flows may lead to early heating and failure of the p:; mary system. Natural-convection cooling from the top of a debris region. such as in the lower head, also lengthens the time to vessel failure. Hence, natural circulation within the vessel with coupling to the primary system can change completely the course and timing of a meltdown sequence. This conclusion emphasizes the importance of a multidimensional vessel-flow capability. as provided by MELPROG.

In addition, we tested the modeling effects of the initial fuel-rod melting and relocation. Variations in the assumptions were found to affect strongly hydrogen production and the subsequent course and timing of the accident (total hydrogen production was doubled. and vessel failure occurred 20 min earlier for an increased failure and relocation temperature). Thus. we have shown the need for the more accurate models provided by the MELPROG CORE module that currently is being implemented.

• Th5 work was funded by the US Nuclear Regulatory Commission (NRC). Office of Nuclear Regulatory Research. Division of Accident Evaluation.

14-5

An important assumption in this, and all other. TMLB' sequence calculations is that the primary system remains at. or near the pressure corresponding to the set point of one of the relief valves. The possibility of a failure somewhere in the primary system, followed by depressurization, needs to be examined. With an auxiliary calculation. we estimated the thermal condition of one of the primary-system weak points. the connection between the vessel outlet nozzle and the hot leg The temperature. thus determined. then can be used in a structural analysis to determine if and wh(n a failure may occur. The temperature was above 1000 K for more than 1 h before core slump occurred. Initial indications are that the connection fails rapidly above 1000 K. If this is the case. then system depressurization by this means is possible before core slump and probable before vessci lower-head f ailurc.

We ran a linked MELPROG/ TRAC calculation from accident initiation through fuel disruption. The model includes the primary loops with relief valves and a simplified secondary system. MELPROG provides the vessel component in the linked calculation. Modeling of the entire primary system produces important differences in the results. In particular, all of the major events, such as uncovering of the core, onset of cladding oxidation.

and disruption of the control and fuel rods. are delayed from 10 to 25 min when additional heat sinks provided by the primary system are included. The primary-system heating. which is calculated directly, has shown similar (although delayed) hot-leg nozzle heating when compared to that discussed previously. The calculation also indicated that the relicf capacity may be insufficient to prevent pressurization during core slump. The resulting pressure increase probably will cause some component to fail long before vessel f ailure.

When the vessel failed in the base case. the vessel debris was 30% moltcn on average. The lower-plenum debris consisted of all the core material and some structural steel. Table I summarizes the state of the debris when the vessel failed included is the amount of steelin the debris most of which is added to the debris after core slump. Also included here is 9600 kg of unoxidized zirconium that represents 58% of the originalinventory.

The average temperature of the debris was 2460 K.

Finally. the calculation has shown the need for additional experimental and analytical information on such phenomena as debris-crust failure related to core slump and vessel-penetration failure before gross vessel-wall collapse. We recommend that such information be obtained and incorporated in MELPROG along with the implementation of the CORE and VICTORIA modules. With these changes in the code. we further recommend that the calculation be rerun to assess the improved predictive capability of the code and to identify additional coding and modeling modifications nceded.

TABLEI STATE OF DEBRIS AT VESSEL FAILURE Mass Molten / Liquid Material (kg) (%)

UO; 96000 14 Zr 9600 100 ZrO 2 9250 0 Steel 19300 78 Controi rod 2850 100 Total 137000 30 14-6

SCDAP/RELAPS Update and Applications C. M. Allison, G. A. Berna, T. C. Cheng, L. J. Siefken A. S. L. Shieh, R. J. Wagner Idaho National Engineering Laboratory The objective of the SCDAP/RELAPS code development effort is to develop a best estimate computer code to predict system thermal-hydraulics, core damage progression, and fission product transport within the reactor coolant system. The code uses fully coupled models with system thermal-hydraulics and material transport, structural and debris behavior, and radionuclide deposition behavior described by the RELAP5, SCDAP, and TRAP-MELT components of the code, respectively. The individual components of the code have undergone significant assessment, with limited assessment of the integrated code coming from comparisons with early data from LOFT FP-2, idealized test cases, and RELAPS calculations. The basic model development effort has been completed with models in place for PWRs, BWRs, and experimental facilities. Additional model development is being supported with non-NRC funding to support the TMI-2 standard problem analyses efforts and to support other potential users of the integrated code. The additional models being incorporated in an experimental version of the code include models that treat debris formation, debris behavior, melt relocation, vessel failure, and melt ejection.

During the past year, there have been several accomplishments that are worthy of special note. The fission product behavior models have been expanded or improved through the addition of the CORSOR-M correlations in medium to low volatility fission products, a tin release model, and upgrades to the PARAGRASS model. In addition, a model to predict the decay heat associated with fission product deposition has been completed and incorporated into the code. The criteria to select appropriate ballooning and rupture models has been modified to better predict the extended balloons that occur under slow heatup conditions. This improvement, along with experimental modifications to the oxidation models to account for double sided oxidation, appears to be a significant factor in properly predicting the oxidation-driven heatup shown in the severe fuel damage experiments. The BWR capability of the code has been completed with the modification of the code to allow the definition of multiple shrouds and with the completion of the B4C control rod model.

This allows the code user to analyze a BWR core with multiple assemblies.

The increased speed of the integrated code, as a result of improved model numerics in the TRAP-MELT models and vectorization of selected models, is also worthy of particular note. This increased speed has been of particular importance in the application of the integrated code to full plant analysis.

Recent applications of the integrated code to TMLB', S D, 2 and BWR High

! Pressure Boiloff sequences mark an important milestone in the development 14-7

of the code, although this type of analysis is not performed as part of the code development effort, but as part of the NRC's SASA effort. These calculations have demonstrated important effects that can be seen only in a full integrated plant analysis and provide an important justification of the integration effort. These calculations also have demonstrated for the first time that a fully coupled plant analysis using detailed phenomenological models can be performed at near or faster than real time i on a CRAY computer.

Other activities occurring in FY-86 include the refinement of existing models with an increasing emphasis on code assessment and user and NRC technical support. These activities include the completion of the OECD LOFT FP-2 posttest analysis, support of the CORA experiments being performed in West Germany, and a formal technical review of the i SCDAP/RELAP5 models. In FY-87, most of the activities will be directed to developmental assessment of the codes using the data becoming available at that time.

! I Work supported by the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research, under DOE Contract No. DE-AC07-76ID01570.

t i

14-8

EVALUATIONANDIMPROVEMENTINNDERELIABILITY l FOR INSERVICE INSPECTI0fr I S. R. Doctor, Program Manager D.J. Bates, M.S. Good, P.G. Heasler, G.A. Mart F.A. Simonen, J.C. Spanner, T.T. Taylor Pacific Northwest Laboratory Operated by Battelle Memorial Institute

SUMMARY

The primary pressure boundaries (pressure vessels and piping) of nuclear power plants are inspected periodically during the service life of the power plant. The rules and requirements for such inservice inspections are specified in Section XI of the ASME Boiler and Pressure Vessel Code (Rules for In-Service Inspection of Nuclear Power Plant Components).

The Evaluation and Improvement in NDE Reliability for Inservice Inspection program at Pacific Northwest Laboratory (PNL) was established to determine the reliability of current ISI techniques and to develop recommendations to ASME Section XI that will ensure a suitably high inspection reliability. The objec-tives of this NRC program are to:

a determine the effectiveness and reliability of ultrasonic ISI performed on commercial light-water reactor (LWR) primary systems recommend Code changes to the procedures to improve the reliability of ISI using fracture mechanics (FM) analysis, determine the impact of NDE unre-liability on system safety and determine the level of inspection reliabil-ity required to ensure a suitably low failure probability evaluate the degree of reliability improvement that could be achieved using improved and advanced NDE techniques based on material properties, service conditions, and NDE uncertainties, formulate recommended revisions to ASME Code,Section XI, and Regulatory requirements needed to ensure suitably low failure probabilities.

The scope of this program is limited to ISI of primary systems, but the results and recommendations are also applicable to Class II piping systems.

The program consists of three basis tasks: a Piping Task, a Pressure Vessel Task, and an Improvement in NDE Reliability Task. Because of the prob-lems associated with the reliable detection of intergranular stress corrosion cracks (IGSCC) and the accurate characterization of IGSCC, the past year's major programmatic efforts were concentrated in the first and the last tasks.

a FIN: B2289; NRC

Contact:

J. Muscara 15-1

There were a number of activities that occurred in the Piping Task during the past year. A significant effort was expended in support of ASME Section XI Subgroup on NDE to develop requirements for the qualification of UT/ISI systems (personnel, equipment and procedures). Although, there are no new Code rules, the documents are almost completed and are proceeding through the acceptance process. In parallel, input has been gathered for forming a Regulatory Guide for UT/ISI qualification if the Code work does not meet NRC safety require-ments.

A round robin was conducted to evaluate performance by personnel who had passed IEB 83-02 in terms of their ability to detect IGSCC, correctly classify geometry and metallurgical conditions, detect short versus long (up to 360')

IGSCC, automated versus manual system performance, crack depth sizing accuracy and any team effects, single inspector versus teams. Destructive testing of all IGSCC was performed and an analysis conducted.

Pipe from the Vermont Yankee reactor was cut to remove the pipe weldments  ;

and these weldments were decontaminated. The pipe material from the Monticello l reactor was cut to isolate the weldments from the large pipe sections, but it I was not decontaminated. These specimens will be used for round-robin studies such as PISC III, other technology evaluations, and to compare field ISI per-formance versus true conditions. ,

1 Work was performed looking at the inspectability of weld overlaid pipe joints and in studies to develop a requirement for surface irregularities.

Both surface waviness and smoothness will adversely effect UT performance but no specification exists for acceptable limits.

The PISC-II round robin data bank was received, and it is undergoing ana-lysis to ensure that all necessary data is present. A test plan for analyzing the data was also developed to interpret the result in terms of U.S. needs.

A test plan was developed as part of the PISC-III program for the austen-

itic steel tests that are being planned for reliability, capability, and para-metric studies.

Calculations were performed to look at the effect of NDE inspection inter-val and NDE effectiveness versus probability of leakage for IGSCC and flaws of interest to the PTS issue.

I i

L 15-2

DEVELOPMENT AND VALIDATION OF A REAL-TIME SAFT-UT SYSTEM FOR INSERVICE INSPECTION OF LWRs" S. R. Doctor, Program Manager T. E. Hall, L. D. Reid, and G. A. Mart

SUMMARY

A program for the development of the synthetic aperture focusing technique for ultrasonic testing (SAFT-UT) has completed the third year. The program is designed to provide the engineering required to transfer the SAFT-UT technology from the laboratory into the field. Specific program objectives are:

Design, fabricate, and evaluate a real-time defect detection and charac-terization system based on SAFT-UT technology for preservice and inservice inspe.ction of LWR components.

Establish calibration and field test procedures.

Demonstrate and validate the SAFT-UT system through actual field inspec-tions.

Generate an engineering data base to support Code acceptance of the real-time SAFT-UT technique.

Facilitate technology transfer to the commercial ISI community.

The program scope is defined as follows:

Conduct laboratory tests to provide engineering data for defining SAFT-UT performance.

l Complete the development of a special-purpose SAFT processor to make pulse-echo and tandem SAFT-UT a real-time process for ISI applications.

Fabricate and field test a fieldable real-time SAFT system on nuclear reactor piping, nozzles, and pressure vessels.

Encouraging ISI community interest in implementation of SAFT-UT technology in commercial applications.

The work this year has focused on the implementation of a different and more effective scanning procedure for tandem SAFT. This new procedure involves the scanning of both transducers simultaneously in opposite directions. This results in better illumination of deep defects and, hence, better depth sizing information. The tandem SAFT modes were also programmed on the real-time processor, since the algorithm is different for pulse-echo and tandem SAFT.

a FIN: B2467; NRC

Contact:

J. Muscara 15-3

{

_ . _ _ l

A considerable effort was spent in the evaluation of new 68020 micropro-cessor boards for upgrading the real-time processor. These were first genera-tion boards, and they had numerous hardware problems that had to be solved.

Extensive effort was also spent in achieving other software changes that produced further improvements in the processing times for pulse-echo SAFT.

Extensive tests were conducted on a large number of IGSCC and thermal fatigue cracked specimen. These tests were used to determine the best algo-rithm to use for length and depth sizing of defects. In addition a large number of laboratory tests were conducted to look at the performance of the SAFT system for flaw parameters such as orientation, or system sensitivity to operational parameters such as transducer center frequency. These tests are helping to show which operating mode is best for different SAFT applications and what are the critical parameters that could cause system errors (lowering of system resolution).

Exchange meetings were held with combustion Engineering staff to start j training some of them on the operation of the SAFT system. Since Combustion I Engineering staff are to aid PNL in validating the SAFT technology under field conditions, it is important that they learn the procedures for properly using SAFT in anticipation of this testing.

15-4

)

PROGRESS FOR ON-LINE AC0USTIC EMISSION MONITORING OF CRACKS IN REACTOR SYSTEMS P. H. Hutton, R. J. Kurtz, and M. A. Friesel Pacific Northwest Laboratory Operated by Battelle Memorial Institute

SUMMARY

The acoustic emission (AE) monitoring program has the objective of devel-oping and validating the use of AE technology for continuous surveillance of reactor pressure boundaries to detect flaw growth. The benefits expected include:

Early detection of cracking in primary pressure boundaries.

Support of the leak-before-break concept through crack growth detection.

Support ALARA by reduced manual inspection of service sensitive piping and nozzles.

Facilitate toring for crack safe extension growth. of the service life of pipe weld repairs by moni-Increase operational safety through pressure boundary and/or valve leak detection.

Accomplishments in FY86 include:

Continued readiness to monitor Watts Bar Unit i reactor during initial operation.

Arranged to monitor piping joints at Peach Bottom 3 reactor during opera-tion and have installed a monitor instrument at the plant.

Gained direct support from utilities for completion of validation testing.

Further verified the AE signal pattern recognition method by determining that it is capable of separating crack growth AE signals from crack inter-face noise.

Test evidence indicated that AE from IGSCC is identifiable using the AE signal pattern recognition method.

An ASTM standard for continuous AE monitoring of pressure boundaries has been approved.

A draft AE application procedure has been developed under ASME Section V.

15-5

The program continues to follow the original plan with current emphasis on validation of the technology through monitoring on operating reactors and tech-nology transfer through seeking acceptance of the technology by code and regu-latory functions.

l 1

l r

I l

d l

15-6

RELIABILITY OF LEAK DETECTION SYSTEMS IN LWRs*

D. S. Kupperman Argonne National Laboratory Materials and Components Technology Division Argonne, Illinois 60439 U.S. Nuclear Regulatory Commission Guide 1.45 recommends the use of at least three different detection methods in reactors to detect leakage. Monitoring of both sump-flow and airborne particulate radioactivity is mandatory. A third method can involve either monitoring of condensate flow rate from air coolers or monitoring of airborne gaseous radioactivity. Although the methods currently used for leak detection reflect the state of the art, other techniques may be developed and used. Since the recommendations of Regulatory Guide 1.45 are not mandatory, the technical specifications for 74 operating plants have been reviewed to determine the types ofleak detection methods employed. In addition, Licensee Event Report Compilations from June 1985 to June 1986 have been reviewed to help establish actual capabilities for leak detection.

Generally speaking, reactor opcrators rely on sump pump monitoring to establish the presence of leaks, although for most reactors, the surveillance periods are too long to detect a 1-gal / min leak in I h, as suggested by Regulatory Guide 1.45. Also, the review of event reports indicates that in a number of cases, leaks were not detected until flow rates were well above those allowed in reactor technical specifications. These leaks were primarily from valves, pumps, and I to 2-in.-diameter lines. Recent exceptions are leaks from a 1-in.-long through-wall IGSCC in a riser weld and a break in a 6-in. steam line elbow. It further appears that radiation monitors are unreliable, primarily for two reasons:

l (1) The high background radiation level in some reactors forces the alarm trip point so high as to be potentially insensitive to the rise in radiation level from unidentified leakage.

In one reported case, the radiation level did not increase with a leak rate of 25 gal / min. (2)

Spurious electrical signals cause false alarms to occur at a high rate. In summary, leak detection procedures appear to be deficient in meeting NRC goals in some cases.

Although current leak detection systems nevertheless appear to be adequate to ensure a leak-before-break scenario in the great majority of situations, one must also consider the possibility that large cracks may initially produce low leak rates. This situation may arise because of corrosion plugging or fouling of relatively slowly growing cracks or the relatively uniform growth of a long crack before penetration. The possibility of a relatively low leak rate from a large crack is not simply a matter of conjecture. The Duane-Arnold safe-end cracking incident indicates that the current approach to leak detection is clearly inadequate in some cases. In the Duane-Amold case, the plant was shut down on the basis of the operator's judgment when a leak rate of 3 gal / min (below that required for shutdown in BWRs) was detected. Examination of the leaking safe-end showed a crack that extended essentially completely around the e umference.

Simply tightening the current leakage limits to improve >cer .ivi - s not adequate, however, since this might produce an unacceptably high number fof spnri a d 2 owns owing to the inability of current leak detection systems to identify leak sources. None or ttle systems currently used provides any information on leak location, and leaks must be located by visual examination after CRSR FIN Budget No. A2250; RSR

Contact:

J. Muscara 15-7

shutdown. Since cracks may close when the reactor is shut down and thus reduce leak rates considerably,it would be desirable to locate cracks during plant operation.

In order to improve detection of IGSCC leaks, some U.S. utilities have installed either acoustic emission monitors or moisture-sensitive tape at specific welds. With moisture-sensitive tape, quantitative information is limited and false alarms are a potential problem. Acoustic systems have great potential for providing quant;tative information regarding leak detection, location, and leak rate estimates. Although the utilities' current experience with acoustic monitoring mainly involves valves, experiments at Argonne National Laboratory have shown that acoustic leak signals can provide detection, location, and leak rate information for the entire primary coolant system. Studies carried out with field-induced IGSCC specimens and acquisition of acoustic background data from reactors have led to estimates of acoustic leak detection sensitivity as a function of background noise, leak rate, and acoustic receiver location. Analysis of experiments with a digital continuous acoustic monitoring system (developed with GARD Inc.) under laboratory and field conditions has shown that enhanced capabilities for locating and sizing leaks in reactor primary systems can be realized. Recent efforts with computer-simulated leak signals have been encouraging and should lead the way to improved system design.

15-8

Programme for the Inspection of Steel Components PISC II Results and PISC III Plans S.J. CRUTZEN, CEC, JRC Ispra Establishment, NDE Laboratories

SUMMARY

l For several years the question has been asked : are NDE techniques used l i

on heavy sections sufficiently effective and reliable to satisfy the requirements for integrity assessment on important structures such as pressure vessels? Early answers were often optimistic opinions and the only known facts arose from inconclusive, limited trieals. Occasionally it was discovered that a large defect had not been detected, but there was little public discussion or explanation of scuh failures. On the other hand, there have been claims that unreliable NDE has led to unnecessary and costly repairs. The series of major cooperative program-mes known collectively as PISC (the Programme for the Inspection of Steel Components) has been developed progressively to provide as comple-te an answer as possible to these questions on the capability of NDE.

PISC involves three stages :

PISC I (1975-1980) assessed the capability of a manual ultrasonic procedure based upon the relevant section of the 1976 edition of the ASME boiler and pressure vessel code, particularly Section XI. The results first published in 1979, showed several shortcomings of the PISC I (ASME XI) procedure, but higher effectiveness was found (and published in 1980) for European Institutions. some alternative procedures demonstrated by 10 PISC II was set up in 1980 by OECD and CEC to examine in more detail which techniques could provide the desired level of capability for detecting and sizing defects. The experimental work of this phase has been completed successfully to the planned programme and evaluation of the results is completed.

i 15-9

PISC III is the most recent phase initiated by OECD and CEC, and started in June 1985, to amplify, verify and validate aspects revealed in the PISC II work by considering : real defects (PISC I and II considered mainly artificial or deliberately induced flaws); real geome-tries in real surroundings (PISC I and PISC II had only one case of a real geometry); complementary laboratory exercises and validation of mathematical models.

The PISC 11 exercise reached its first objective : the identification of high-performance procedures and ways of complementing or optimizing present procedures to improve their capability.

Detection and sizing is conditioned by defect characteristics. Techni-ques have thus to be optimised as a function of the defect types they are supposed to detect and size. Validation samples as well as calibra-tion blocks have to be conceived in compliance with this important parametre.

The second objective of PISC II was to draw important conclusions to the attention of the appropriate licensing, code writing and specifica-tion bodies.

This last objective appears to be reached too.

15-10

An Expert System for USNRC Emergency Response D.E.Sebo, M.A. Bray, and M.A. King Idaho National Engineering Laboratory, EG&G Idaho, Inc.

Summary The Reactor Safety Assessment System (RSAS) is an expert system under development for the United States Nuclear Regulatory Commission (NRC).

RSAS will be used at the NRC Operations Center during serious incidents at licensed nuclear power plants and during emergency respcnse drills.

RSAS is an heuristic classification system which converts parametric data from a power plant into plant status information which is useful to NRC emergency response personnel. The key technical challange in RSAS is to produce useful conclusions and cover all licensed nuclear power plants without writing 100 seperate expert systems.

The NRC Operations Center in Bethesda, Maryland, serves as the operational communications center for the NRC. The center is the focus for NRC response to an incident until the team from the region office arrives at the licensee site. A key NRC function during incidents is to provide independent technical assessment of existing and expected plant status, offsite consequences and licensee actions. The Reactor Safety Team is one of the several teams in the center and is responsible for dssessing the Current and anticipated status of the nuclear power plant, particularly those aspects of plant status that could lead to offsite radioactivity releases. The RSAS expert system is a tool to be used by the Reactor Safety Team.

RSAS uses, as input, parametric data which the phone talker obtains from the plant. Typically 20 to 30 values of temperatures, pressures, radiation readings, etc. are provided at 15 minute intervals. RSAS produces three types of conclusions based on this data. First, diagnoses are produced when appropriate. An example diagnosis is that a steam generator is dry if its pressure is significantly below the saturation pressure corresponding to primary system temperatures (i.e. the steam generator pressure is decoupled from the primary system). A second type of conclusion points out important relationships between parameters or between parameters and plant specific setpoints. For instance, the system notes relationships between reactor system pressure and ECCS setpoints or between pressure and reactor temperatures. The third type of conclusions are about significant behavior of individual parameters such as high heatup or cooldown rates.

In addition to producing conclusions from data, RSAS provides several other functions. The system has ranges for plant instruments and is able to do limited error checking. It deals selectively with out of range readings. For instance, reactor coolant temperature may have a range of 50-650 degrees F. An out of range value of 30 degrees is clearly invalid but a value of 750 degrees may be from an alternative instrument. It is possible to create rules on line to monitor specific param>;er values or rates. Input data can be modified to correct bad data or to what-if the available. Also a graphic display of pressure temperature readi1g is situation.

15A-1

A challange of RSAS has been to produce useful conclusions for all licensed reactors without having to develop 100 expert systems. This is being handled by producing modular rulesets for a short list of plant type. Rulesets exist now for Babcock and Wilcox, Westinghouse 2 loop, 3 loop and 4 loop plants. For each ruleset, seperate setpoint files are produced for each reactor covered by the ruleset. A rule might compare pressure to a relief setpoint. The setpoint would be obtained from the setpoint file rather than be written into the rule itself.

The RSAS expert system has been developed on a stand-alone computer, a Xerox 1186 AI workstation. During July the system was installed at the Operations Center for user testing. Because the system is intended to support a particular group, the Reactor Safety Team, it is critical that a very interactive relationship be maintained between the developers and the users. Installation of the system in the center and use during drills will help to ensure that the tool meets the team's needs.

Work supported by the U.S. Nuclear Regulatory Commission, Office of Inspection and Enforcement, under DOE Contract No. DE-AC07-76ID01570 15A-2

NUCLEAR REGULATORY COMMISSION PROGRAMS ARE SUPPORTED BY THE TECHNOLOGY RESOURCES OF THE ENGINEERING PHYSICS INFORMATION CENTERS Nancy A. Hatmaker Robert W. Roussin John E. White Engineering Physics Information Centers Oak Ridge National Laboratory Oak Ridge, Tennessee The Engineering Physics Information Centers (EPIC) at Oak Ridge National Laboratory have two activities which support NRC Programs, the Radiation Shielding Information Center (RSIC) and the Technical Data Management Center (TDMC). The older of the two, RSIC, was i

established in 1963 as an Information Analysis Center in the general (

I field of radiation shielding, transport, and protection. It has multiple agency funding to acquire, evaluate, organize, and distri-bute information (including computing technology and numerical data) relevant to its field. The TDMC was established in 1978 to perform work for NRC in fields not related to .RSIC's subject coverage and to perform tasks beyond the normal level of activity for RSIC. The two centers share administrative functions, ' physical premises, special-ists skills, and computing resources. The NRC/ADM/TIDC monitors the cosponsorship of RSIC for coverage of agency-wide interests and plans. directs, and coordinates the work of the TDMC, including the establishment of TDMC as the agency-wide repository for packaged computer-related technical information products. The contract monitor is Myrna L. Steele.

I RADIATION SHIELDING INFORMATION CENTER l

The disciplinary areas of RSIC's coverage include radiation 1 l

production, transport, shielding, and protection. The analysis of '

penetrating radiation in an external exposure context is one of the main subjects of interest, but the computation of internal dose and other radiation protection analyses are of equal importance to the NRC staff and contractors. The technology base in RSIC offers the types of resources required for such analyses.

As a national technology resource, RSIC's purposes are to avoid duplication of effort, advance the state of the art, increase value and content, and make widely available information, computing techno-logy, and numeric data. The program is integrated with other programs in the Department of Energy and the Defense Nuclear Agency to provide technology necessary for the solution of a variety of radiation transport problems. There are currently more than 700 computer code packages and 120 data libraries in the RSIC collection. Many were developed under NRC sponsorship, but much technology developed in other fields has provided a foundation for application to problems of interest to the agency. Computing technology is treated by RSIC in the "open code / data package" concept that fosters a continual RSIC-User collaboration which moves the technology ahead at a faster pace.

15A-3

Communication with the user community is provided by a monthly newsletter that includes information about new acquisitions and details changes to existing packages. It is an efficient means of binding the RSIC community together in an effective collaboration in the solution of radiation transport problems.

TECHNICAL DATA MANAGEMENT CENTER In order to effectively serve NRC programs, TDMC was established with goals and priorities determined by an advisory committee of NRC technical staff members chaired by the NRC/ADM/TIDC contract monitor.

Its primary purpose is to support NRC programs by assisting in verification, validation, documentation, and standardization of computing technology (including codes and other technical data) used in the licensing and regulatory processes.

Current work includes the following subtasks:

(A) technical data management in "open code / data package" concept in "nonshielding" areas; (B) computer program enhancement, development of new options and extensions, standardization, and documentation of selected

^

computing technology; (C) technical data generation, validation, and documentation:

(D) publications in support of TDMC tasks and other NRC Technical Data Management programs; and (E) dissemination of computer codes and data and technical informa-tion exchange.

Application of the open code / data package concept to atmospheric transport, environmental transport, accident evaluation and radiolog-ical assessment methods has added in excess of 75 packages to the collection. Computing technology developed or evaluated by TDMC is made operational on the NRC MV/8000 or other specified computer to allow maximum utilization by NRC staff.

Inquiries to either Center should be mailed to:

RSIC or TDMC or EPIC Oak Ridge National Laboratory P. O. Box X Oak Ridge, Tennessee 37831 Telephone: 615-574-6176 FTS 624-6176 15A-4

SAND 86-22200 DCH-1: The First Direct Containment Heating Experiment in the SURTSEY Test Facility.

William W. Tarbell John E. Brockmann Marty Pilch Sandia National Laboratories Albuquerque, NM 87185 Direct heating of the atmosphere by debris expelled from the reactor cavity may potentially threaten containment integrity. The SURTSEY Direct Heating Test Facility has been constructed at Sandia National Laboratories to perform experiments where molten debris are ejected from a scale model of a reactor cavity into a well-defined and contained atmosphere. The size of the facility (3 m diameter by 11 m tall) permits experiments on the order of 1:10 of full scale. The SURTSEY test chamber permits direct measurement of the pressure and temperature caused by the dispersed debris and facilitates acquisition of aerosol generation and behavior data.

The objective of the SURTSEY/DCH test matrix is to provide the experimental data for understanding of the phenomena associated with pressurized melt ejection and direct containment heating. This understanding is used to develop physical models of the processes that are incorporated into larger containment response codes. The coupling of these and other phenomena within the code allows predicting the potential consequences of severe accident sequences that involve failure of the reactor vessel while at elevated pressure. Subsequent experiments will consider other cavity geometries, in-containment structures, water in the cavity and atmosphere, and containment atmosphere composition to investigate the mechanisms of energy transfer and debris transport. The DCH-1 test did not include any potential mitigative aspects other than a relatively small mass of melt and the addition of a non-prototypic structure at the cavity exit. The results of these tests cannot be extrapolated directly to reactor scale because of the complexity of the scaling issues.

The DCH-1 test was the first experiment performed in the SURTSEY

} facility. It involved 20 kg of molten material ejected into a 1:10 linear l scale model of the Zion reactor cavity. The model was placed within the test chamber so that the cavity exit was along the vertical centerline of the vessel. To prevent damage to the vessel by direct impact of the debris, an open-ended concrete-lined steel box structure (0.9 m tall) was added to the cavity exit. The structure was designed to direct the debris vertically upward. The molten material was produced by a metallothermitic reaction in a

• This work was supported by the U.S. Nuclear Regulatory Commission and performed at Sandia National Laboratories which is operated for the U.S.

Department of Energy under Contract Number DE-AC04-76DP00789. melt generator 16-1

placed at the scaled height of the reactor pressure vessel. As the reaction front reached the bottom of the melt generator, it failed a fusible brass plug and formed the exit aperture. Bottled nitrogen gas was used to pressurize the melt to 2.55 MPa prior to ejection. The SURTSEY chamber was instrumented primarily to measure the atmosphere pressure and temperature increase and aerosol generation and behavior. Other diagnostics obtained the pressure and temperature history of the pressurizing gas in the melt generator. Debris and aerosol samples were collected from the chamber after the test for laboratory analyses.

Results from the experiment showed that the dispersed debris caused a rapid pressurization of the chamber atmosphere. Peak pressures from the six transducers in the chamber ranged from 0.09 MPa to 0.13 MPa (13.4 to 19.4 psig). The large spread in the recorded values was from debris heating the sensing element. The time from the start of debris dispersal to the pressure peak averaged less than one second. Camera records indicated that the debris from the cavity emerged and propagated upward at a velocity on the order of 40 m/s. The debris rapidly expanded laterally as it moved upward, filling the entire cross-section within a few meters of the cavity exit. The aerosol material accompanying the debris discharge was extensive, effectively causing the atmosphere to become opaque within about five seconds.

Post-test collection of the debris showed that approximately 11.6 kg was dispersed from the cavity. The measured mass included an estimated 1.6 kg due to uptake of oxygen by the metallic constituent of the reaction (55 w/o Fe). Gas samples from the chamber following debris dispersal showed nominally a 0.05 v/o reduction in the oxygen concentration. The melt material retained within the cavity and chute was in the form of a thin crust layer on the concrete and steel surf aces, and a large mass (~1.2 kg) found at the base of the inclined tunnel. This material was presumably a film layer in the inclined section that was not entrained because of an inadequate supply of gas from the melt generator. Mechanical sieving of the debris collected from the chamber showed a log-normal size distribution with a mass mean size of 0.55 mm and a geometric standard deviation of 4.2.

The aerosol measurements indicated a substantial amount of material of size less than 10 micrometers. Most of the collection devices were overloaded by the greater than expected amount of material. Because of this, the calculated range in the aerosolized fraction of the dispersed debris was large; from 5% to 25%. The aerosol concentration was initially high (~10

! g/m3 ), that fell off rapidly at first and then slowed later in time. This behavior was consistent with the bi-modal size distribution determined from cascade impactors. The distribution data suggested modes at I and >10 micrometer with a possible third mode at 5 micrometer.

Work is continuing on the analysis and interpretation of the data from the DCH-1 experiment. The next SURTSEY experiment (DCH-2) will be performed at similar conditions except the initial melt mass will be 80 kg.

16-2

l RECENT EXPERIMENTAL AND ANALYTICAL RESULTS OF BNL DIRECT CONTAINMENT HEATING PROGRAMS

• T. Ginsberg and N. K. Tutu Brookhaven National Laboratory Department of Nuclear Energy Upton, New York 11973

1. INTRODUCTION The direct containment heating scenario involves high-pressure ejection of molten core material from the reactor vessel into the region beneath the vessel and into various subcompartments of the containment building. The stored energy in the melt consists of the sensible energy of melt and the chemical reaction energy of the various components assuming that they can react with either the oxygen or the steam within containment. The metallic phase may first react with steam, if local conditions permit, and thereby pro-duce hydrogen. The hydrogen may then burn at some later time at a different location. In order to predict the containment response, one must follow the melt through the various subcompartments of the containment building, while computing the integrated _ release of energy from the melt to the containment atmosphere and the quantity of hydrogen produced during the time period that the melt is suspended. The BNL research program in the area of direct con-tainment heating is directed towards the development of a methodology to pre-dict the hydrodynamic, chemical and thermal interactions which could take place in three regions of PWR containment buildings: the reactor cavity, the intermediate compartments (e.g., steam generator room) and the containment dome. Separate effects, scaled expe'riments are performed related to selected aspects of the DCH problem, and analytical models are developed to character-ize the relevant phenomena. A summary of recent progress is presented below.

2. DEBRIS DISPERSAL FROM REACTOR CAVITIES The objective of this experimental program is to determine the potential for debris dispersal from PWR reactor cavities of various designe during the high-pressure melt ejection accident. Small-scale (1/42) transparent models of the cavities for the Zion, the Surry, and the Watts Bar nuclear plants have been built.

Molten Wood's metal and water have been used as the melt simu-lants. To begin an experiment, a known mass of melt simulant is injected into the cavity model from an orifice above the cavity floor. This is followed by the blowdown of N2 gas from a scaled pressure vessel. During the experiment, high-speed movies of the melt simulant-gas interaction in the cavity are taken, and at the end of the experiment, the mass of melt simulant trapped in the cavity is measured.

A dimensional analysis of the melt dispersal phenomena has been per-formed. This analysis is used to select the experimental parameters and to serve as the basis for interpreting the experimental data and for application

• Work performed under the auspices of the U.S. Nuclear Regulatory Commission.

16-3

of the results to full scale. The analysis shows the fraction of debris dio-persed from the cavity to be dependent upon six dimensionless groups. Based upon physical reasoning it is believed, however, that only three of the dimen-stonless groups are dominant, namely: (i) the Kutateladze number, (ii) the dimensionless blowdown time, and ('iii) the ratio of debris (particle) time constant to the fluid transit time through the cavity. The experiments are designed to have the values of these groups as close as possible to full-scale cavity. An interesting result of this scaling analysis is the finding that water is a better melt simulant than Wood's metal for small-scale experi-ments. Detailed experimental results will be presented in the paper; here we just mention that, with water as the melt simulant, the Zion and Surry cavity models showed virtually complete dispersal.

3. INTEGRATED CONTAINMENT PHENOMENOLOGY The debris dispersal experiments will provide guidance as to the extent of melt dispersal f rom PWR reactor cavities of various designs. If, for a given design, significant melt dispersal is observed in the experiments and is predicted for accident conditions, then a model of the containment building will be constructed and experiments will be conducted directed towards predic-tion of the extent of transport of the melt to the various subcompartments of the containment building. In experiments at BNL and elsewhere, simulations of melt ejection into Zion reactor cavities have led to the observation of com-plete melt ejection from the experimental cavities. A model of the Zion con-tainment is being constructed with which melt transport experiments will be performed. A description of the experimental model is presented in the paper.

DHCVIC, a direct heating containment vessel interactions code, is being developed to predict the thermal, chemical and hydrodynamic interactions which could occur during high-pressure melt ejection sequences. DHCVIC is a three-region, integrated model, consisting of descriptions of a melt vessel blow-down, interactions within a reactor cavity and wit.hin a containment volume.

The model is described and results are presented of a DHCVIC analysis of the SNL "SURTSEY" Test DCH-1.

16-4

SAND 86-1989A Development and Applications of the Interim Direct Heating Model for the CONTAIN Computer Code

• K. D. Bergeron, D. E. Carroll, J. L. Tills 1, K. E. Washington, and D. C. Williams Containment Modeling Division 6449 Sandia National Laboratories Albuquerque, New Mexico 87185 Summary The phenomena of melt ejection, debris dispersal, chemical reactions and heat transfer between debris, water, and gases which collectively contribute to Direct Containment Heating (DCH) are extremely difficult to predict with con-fidence using existing calculational tools. An ongoing NRC research program is under way to study these processes ex-perimentally, and it is expected that a significantly improved understanding of DCH will result from these experiments.

However, in the interim, it is useful to analyze the problem with the best calculational tools available in order to assess the important uncertainties and to be able to interpret the results of the experiments as efficiently as possible when the data become available. A model that treats the highly uncertain class of phenomena parametrically, and the better understood phenomena with best-estimate models is therefore a reasonable goal for an interim calculational tool. This paper describes an Interim Direct Heating Model (IDHM) that has been developed as an adjunct to the CONTAIN computer code and which is intended to provide an improved understanding of the uncer-tainties in the analysis of accident sequences involving direct heating, and which can be of use in interpreting and guiding the DCH experiments. Along with the standard models available in the released version of the CONTAIN code, the new model treats transport of the finely dispersed debris with the blowdown gas, heat transfer among the gas, wall surfaces and debris, removal or trapping of the debris as it is trans-ported, and chemical reactions between the unoxidized metals and oxygen or steam.

1 Jack Tills and Associates, Albuquerque, NM 87123

!

• This work supported by the United States Nuclear Regulatory Commission and performed at Sandia National Laboratories which is operated for the U.S. Department of Energy under contract number DE-AC04-76DP00789.

16-5

We will present results from a number of accident sequence calculations (primarily Station Blackout) at the Surry and Sequoyah plants. One of the most important conclusions from these multi-cell calculations is that early in time the debris exists in high-steam / low-oxygen environments, so that the steam-metal reactions are the dominant oxidation process (and not oxygen-metal, as has often been assumed in single volume calculations). Consequently, nominal direct heating pressure rises are less than would be predicted by single volume calculations, but the situation is complicated by the generation of very large quantities of hydrogen in a short time. For the Sequoyah plant it is found that the hydrogen is efficiently transported to the dome even when the debris and steam are trapped in the ice chest, and that hydrogen burns can then occur, causing containment failure in some cases. For Surry, the containment atmosphere is steam-inerted for hydrogen combustion by conventional standards, but there are a number of factors which might invalidate conventional inerting criteria. These include (a) very high temperatures in the bulk volume; (b) even higher temperatures of the entering hydrogen / steam jet; and (c) the possible presence of debris particles whose surfaces could act as catalysis sites for hydrogen / oxygen recombination.

16-6

RECENT RESULTS IN HYDROCEN RESEARCH.

o M. Berman, M. P. Sherman, C. C. Wong Sandia National Laboratories Albuquerque, New Mexico HECTR-MAAP Code Comoarison Sandia developed the HECTR code to analyze the transport and combustion of hydrogen during reactor accidents. IDCOR uses the MAAP code to perform similar analyses. These codes differ in the way that various phenomena are modelled, especially in the areas of (1) ignition and flame propagation criteria, (2) burn time and combustion completeness, (3) continuous recombination of hydrogen and oxygen in the reactor cavity, and (4) natural circulation. We have performed some calculations to compare.the two. codes to estimate the impact of the modelling differences. -

Modelling differences are particulcMy pronounced in multicompartment systems such as the Ice Condenser (IC) and Mark III containments. HECTR calculations tend to allow higher concentrations of hydrogen to develop, which leads to the prediction of higher containment pressures and temperatures. HECTR also permits flames to propagate into the IC upper plenum region, where potentially detonable mixtures can develop for some accident scenarios (e.g. TMLB ' ) . Flame propagation into the IC upper compartment is also possible in the HECTR model, and the global burns which ensue generate much higher pressures than burns restricted to the lower compartment. MAAP code calculations generally do not predict these effects.

A standard problem, an S2HF accident sequence in a PWR Ice-Condenser containment, was defined to compare the HECTR and MAAP code predictions.

Preliminary results show that the two models yield very different pictures of the burning process. MAAP calculations which implicitly employ a 5% hydrogen ignition criterion yield a burn time on the order of two' hours; i.e., the burning process resembles a standing diffusion flame, rather than a flame propagating through a homogeneous mixture. Calculated combustion pressures and temperatures are very low.

FIAME ACCELERATION EXPERIMENTS Ten tests were conducted last year in the FIAME facility. Eirbt involved periodic " simple" obstacles, rectangular baffles on opposite vertical walls, extending from floor to ceiling, and blocking one-third of the channel cross section. Two tests were conducted using a more accurate simulation of the geometry of the IC upper plenum region, including the air handling. units; only 11% of the channel cross section was bloc.ked.

To assess scale e f fec ts , we conducted a set of MINIFLAME experiments, which used 8% scale models of FIAME.

16-7

Analysis of the results of the FLAME and MINIFLAME tests, together with the results of other researchers, has produced a qualitative picture of flame acceleration and deflagration-to.-detonation transition (DDT). The most important four parameters are, in approximate order of importance: mixture sensitivity (how close the hydrogen: air ratio is to stoichiometry); the presence and characteristics of obstacles; scale (larger gas clouds more readily undergo DDT); and degree of transverse venting.

When periodic obstacles were present, DDT was observed in a mixture containing 15% hydrogen. Without obstacles, DDT occurred for 24% hdyrogen.

LOCAL DETONATIONS An important question in reactor safety is how to utilize experimental results to make decisions concerning plant operation and licensing. In some cases, models can be developed to accomplish the necessary extrapolation. In the case of flame acceleration and DDT, the phenomena are far too complex to model with today's knowledge and computing capacity. Furthermore, the applicable data base is always much too sparse to make decisions with high confidence. Nevertheless, the experimental data do provide useful information that should be factored into rational decision making.

We have attempted to develop a framework which converts our sparse knowledge into a qualitative decision guide. We have acsigned two subjective categories to mixture sensitivity, and to a category which incorporates consideration of geometry, scale and obstacle configuration. Based on calculations of a particular accident scenario, numbers from 1 to 5 are assigned to these categories, with 1 being the most favorable to flame acceleration and DDT. These numbers can then be combined to qualitatively estimate the likelihood of DDT under the given conditions. It is important to distinguish between qualitative guidance as to the potential for DDT, and numerical subjective probabilities. Since there are no statistical data whatsoever, numerical probabilites assigned to this " guidance" table would be purely fictitious guesses.

s 16-8

Steam Explosion Energetics by T.G. Theofanous, M. Abolfadl, H. Amarasooriya, B. Najafi*, G. Lucas, and E. Rumble

• University of California, Santa Barbara

• Science Applications International Corporation j The purpose of this presentation is to summarize the results of a study aimed to quantify the likelihood of the a-mode containment failury . Key elements of the study include:

1. A new probabilistic framework that is based on the quantification of a series of causal relationships between parameters that characterize the essential events in the sequence arising from an energetic steam explosion. This allows consideration of all applicable previous results (i.e., it is not limited to propagation of uncertainties in the parameters of a given mechanistic model) and is convenient for a continuing enrichment and refinement as new results become available.

2. New numerical results on the magnitude of premixtures attainable within the lower plenum of a PWR. The analyses include transient and two-dimensional effects as well as, in a special idealized case, water / steam slip.

3. A detailed structural analysis of lower plenum failure and associated venting (energy dissipation) mechanisms.

4. An evaluation of energy dissipation due to uppcr internal structures, including consideration of a new mechanism due to core barrel straining.

5. A new vessel head loading mechanism due to the presence of the massive upper core support plate and upper internal structures.

That is, solid / solid rather than the previously used liquid / solid impact phenomena.

6. Estimation of the fraction of the head bolts that would be expected to undergo brittle fracture and of the loading necessary to produce massive upper head failure.

The results lend additional support to the conclusions of the Steam Explosion Review Group that the likelihood of the a-mode containment failure can be neglected. However, in order to gain fully of the advantages of the present approach as a means leading to convergence of technical judgement it is expected that additional efforts by others to enrich the basis for, and refine the uncertainty bounds of, the causal relations introduced here would be required.

16-9

RECENT RESULTS IN FCI RESEARCH M. Berman, B. W. Marshall, Jr. , M. F. Young, L. T. Pong Sandia National Laboratories Albuquerque, New Mexico Over the last few years, intermediate-scale (10 - 50 kg of melt) fuel-coolant-interaction (FCI) research has addressed the following phenomena: 1.

The rate of steam and hydrogen production from explosive and non-explosive FCIs resulting from " coherent" drops of fuel into water and 2. The behavior of molten " jets" of fuel pouring through water.

COHERENT POURS Presious FITS and EXO-FITS experiments have shown that the qualitative and quantitative aspects of FCIs depend rather strongly on a large number of initial and boundary conditions including fuel composition, mass, and temperature, fuel-to-coolant mass ratio, ambient pressure, water subcooling, drop height (or entry velocity), water chamber depth and cross-sectional area, and degree of confinement. Since this parameter space is far too large to investigate in a statistically reliable fashion, models are being developed to correlate and understand key aspects of FCI phenomena. The two most recent in-vessel test series include the completed FITSC series, and the partially completed FITSD series. These tests are specifically aimed at measuring the rate of steam and hydrogen generation, and the characteristics

, of the post-test debris from FCIs, and Hol the efficiency of steam explosions.

1 A model was developed (1) to predict hydrogen generation from three types or phases of FCIs: coarse mixing, steam explosions, and stratified melt quenching. The model appeared to agree quite well with the results of the FITSC tests, and indicated that the rate of hydrogen generation depended strongly on the explosivity of the FCI. In contrast, subsequent FITSD tests seemed to show that explosivity was not very important to hydrogen generation, and that water subcooling was more important. This model was recently reevaluated against some of the FITSD tests. After making some corrections, the model now appears to agree quite well with most of the FITSC and FITSD tests. The experimental obse rvation of the insensitivity of hydrogen generation to FCI explosivity appears now to also be confirmed by the model. The earlier misinterpretation was probably influenced by the statistically inadequate number of FITSC tests, and the excessive weight assigned to the very low hydrogen generation in test FITS 3C.

We are developing another model to distinguish between the three phases of the iCI, and to investigate maximum possible rates and two limiting mechanisms, steam availability and hydrogen blanketing. The model predictions are strongly influenced by the assumed particle size, relative vclocity, heat transfer mode, water subcooling, and chemical reaction rates.

Nes3rtheless, reasonable assumptions on input conditions produce quantities of hydrogen that are within the envelope of experimental observations.

16-11

The success of both of these separate-effects models implies that the incorporation of a hydrogen / steam generation model into an integrated code (e.g., TEXAS or IFCI) may be a relatively straightforward procedure, at least for comparison to tests involving 50 kg of fuel or less, the limit of current testing capability.

FUEL JETS A startling discovery during the investigation of the TMI accident has altered some previously held assumptions concerning the mode of fuel-coolant contact for in-vessel steam explosions. In direct contradiction to the almost unanimous expert opinicn that no fuel melted during TMI, it is now believed that much of the core melted. Furthermore, the molten fuel appears to have poured through a lower plenum region in which the structures remained intact. Many arguments have been proposed to dispose of the alpha-mode failure issue by arguing that large-diameter fuel pours are self-limiting, and can only result in relatively small steam explosions. The fact that the fuel may be constrained to flow in the form of many small-diameter jets leads one to question the validity or even relevance of these mixing-limit arguments to in-vessel steam explosions. Sandia, under NRC funding, has begun a program to investigate the behavior of single and multiple jets pouring through water. Some scoping experiments have been conducted in the new 50-kg EXO-FITS facility. Based on some very preliminary analysis of the data, it appears that some cautious qualitative conclusions can be drawn for initial jet diameters of 4, 8 and 16 cm.

The jet behavior may be dominated by transient phenomena throughout the duration of the pours. Apparent jet. diameter, or equivalently local void fraction or degree of fragmentation, does not increase linearly with increasing depth. Rather, the jet falls for a period or depth, and then discontinuous 1y changes dimension, sometimes simultaneously throughout the entire length of the jet in the water. It may be that the jet at its deepest penetration point " suddenly" undergoes a change in flow regime, and that this change is propagated rapidly up the jet mixture by rising steam.

The ultimate interpretation of these single -j e t tests will involve additional experiments with improved instrumentation and melt-delivery techniques, coupled to a multi-dimensional fragmentation analysis tool, which is under development, but has not been completed. Additional experiments are also planned using multiple jets. Extrapolation from these complex three-dimensional experiments to reactor conditions will require a great deal of careful analysis.

REFERENCES

1. M. L. Corradini, D. E. Mitchell, N. A. Evans, " Hydrogen Generation During a Core Melt-Coolant Interaction," Proc. International Meeting On IRR Severe Accident Evaluation, pp. 16.5-1 to 16.5-8, Cambridge, MA, August 1983.

l 16-12

EXPEltIALENTS AND 13EST ESTIA1 ATE CODE DEVELOPA!ENT ON SAIALL liltEAK LOCA TillillAIAL-IIYDilAULICS AND IIYDItOGEN A11XING WITIIIN SUI 3COMPAltTAIENTAL CONTAINALENT VESSEL A . T S U O ls Takasago Technical Institute, Technical Ileadquarters, Alitsubishi IIeavy Industries, LTD. 2-1-1, Shinhama, Arai-cho, Takasago, Ilyogo Pref. 676, J APAN

1. Ilackground Af ter the incident of TA1132, extensive safety researches have been carried l out by Japanese industry group together with regulatory group. l Special technical issues addressed in the program are the realistic evaluation of the transient phenomena of the small break LOCA and the hydrogen mixing within subcompartmental containment vessel.

With the funding of the five PWIt utilities in Japan, comprehensive experimental in vestigations together with development of the best estimate computer program have been conducted by Takasago Technical Institute of Atitsubishi lleavy Industries, LTD.

2. Small 11reak LOC A Thermal-Ilydraulics EOS Test Loop, 2 AtW thermal hydraulic simulation loop, was used to simulate various pattern of small break LOCA transient together with the post accident operator action. lireak points were selected both at the pressurzer and the cold leg with the break size range 1.0 inch and 4.0 inch. Outstanding item concluded

are, (1) Core uncoverage mechanism due to loop seal formation in the ItCP suction pipe and mass inventory reduction was fully examined.

(2) The importance of operator action to s.ool down ItCS by the steam generator feed and bleed was demonstrated.

(3) The steam generator is capable of cooling down ItCS both with aru without the natural circulation flow under the existence of noncondensable gas.

(4) Newly developed best estimate computer code ; CANAC- 1 was demonstrated to have the capability of predicting the thermal hydraulic behavior during small 1.OC A transient with highly economical comuputer running time.

3. Ilydrogen Alixing within Subcompartmental Containment Vessel A computer program, AtAl'IlY was developed for the prediction of the transient behavior of hydrogen mixing in each subcompartment of a contaimnent vessel af ter 1.OC A .

AtAl'lIY has the capability of solving the t ransport of oxygen, nitrogen, hydrogen and steam with the assumption of equilibrium state vapor and liquid phase in each subcompartment. Afore than 50 subcompartments and 100 paths can be modeled.

Alixing experiments of simulated hydrogen in the model C/V having 26 subcompartments simulating a broken lo;p and an intact loop was conducted for 16-13 A

the verification of A1APilY. The test results were satisfactory showing the capability of AIAl'IIY code in predicting the highest concentration level and its location together with the concentration distribution anong 26 subcompartments.

Ilydrogen mixing in an actual containment vessel of PWR was analyzed using MAPIlY codes and it was confirmed that hydrogen mixing is good enough to eliminate local high concentration area without the containment recirculation f an.

This results provide the technical base for not requiring post accident recirculat-ion f an within the Japanese contaiment vessel.

25 00 12 00 20 00 10 00 -

2 - Esperispent

-t""'""'

is 00 -

700 . . . . .

--- - w _

Ei  ;;:

$ ,L/ otH les elevataost g" o 00 C no,,,,, s .- - - - - .

e' .p.,..

1

_a a== r _ __ .

1. 7

, ., o e 00 h

}' y oe a 2w -

, p CoHa***d 8've' ACC Inw.sa s'ert

-6 00 e iro0 a.* 3.= .soo .0= rio0 ,.co ioa0o ,2o00 00, , , , gg , ,g , , , g Il488 femel TIME (.ec)

Fig.1 Primary flow rate Fig.3 core water level Test No.SL-2 PRZ 2.61nch break Test No.SL-5 cold leg 41nch break am, --_ 500 00 _

)

2 50

• From te 450 00- _

'. Too G 2 00' +- ~ E* ~~ ~ ~ ~'M f '.5. I e.evee on ' 430 00- "" J O

-~--*-.., f ,

'"f'

- t see,inmi

' w- t- g >%

I 300 00-p "

t

&f

% ,/ .

250 00 -

[ i -h O ' 5200 ~2 73eh Tso'o e000 7200 s.Y 960010b2000 200 00' Tiut usen 0 150 300 450 600 750 900 1050 1200 1350 1500 TIME teocl Fig.2 Pressurizer water level rig.4 Heater surface temperature Test No.SL-2 PRZ 2.61nch break Test No.SL-5 cold log 41nch break C5 .dtf CH2. d t f CC2. dt f Gs2.d t f Cs,a XXXV sxxxxxx Modei C/V Emperiment + - - + Bench-Scale Emperiment (convecuen. dif f sten) (N

> (hear trans fer) f f i I

j hatt xgn T MH 2. MC2. NN 2. P5 ses phone N Development of Analyti- Development of Analyti- O PN2. P02. PN}. PS cal Approach (11 cal Approach til) Cs.s(Cu2.G02.Cu2.CS )

{ transient nodal analy- Ithree-dimensional analy- (Q y

(Gu2.CO2) t+g t

u (c y0, s+g Pt y Pi+2 it e ses MAPHYi sis-COMT AC 3D)

< 6 6 (Cu2. CO2) r**

l l ( _ ( Cu 20 ) t *

• 4 1 S Mi Q2. M8M2 Verification of Analytical n20.

. . Min 2' .M L Evaluation Approach l tieute phase l _xxv , Axxxs Cr .e C tM 20. g (GH2O)t*g flow rate by CIM2 .8 evaporation Prediction of Actual Plant cm3 .s ( 2)g*1 0 i flow rate by and Investigation of its 1N2 8 condensation improvement ( Z .GO 2 ) g

• 1 t absorption rate 1+ g i desorption rate Fig.5 Logic diagram of research pecgram Fig,6 MAPHY's model in a subcompartment 16-14

BWR PIPE CRACK REMEDIES EVALUATION W. J. Shack, T. F. Kassner, P. S. Maiya, J. Y. Park, and W. Ruther Materials and Components Technology Division Argonne National Laboratory Argonne, Illinois 60439 Introduction Cracking in sensitized austenitic stainless steel (SS) piping and associ-ated components in boiling water reactors (BWRs) has been observed since the mid-1960s. Proposed remedies include (1) procedures that produce a favorable residual stress state in the weld regions, (2) replacement of the piping with materials that are more resistant to stress corrosion cracking (SCC), and (3) modification of the reactor coolant environment. During this year, studies that have important implications for all three classes of proposed remedies have been carried out. These studies include fracture-mechanics crack-growth-rate tests on Type 316NG SS and Type 304-Type 308 SS weld overlay specimens in im-purity and high purity environments, metallographic and residual stress studies on weldments treated by the Mechanical Stress Improvement Process (MSIP) devel-oped by O'Donnell and Associates, slow-strain-rate tests on the German Nuclear Grade Type 347 SS, and slow-strain-rate tests for the characterization of reac-reactor coolant impurities that permit alternate cathodic reduction reactions.

In addition, studies of the effect of high gamma irradiation on reactor coolant environments and on the corrosion potential of irradiated stainless steel have been initiated. Together with related studies being performed under another program to characterize stainless steels irradiated under light water reactor conditions being performed under another program, these results will permit a better assessment of the potential for irradiation-assisted SCC in the core region.

Technical Progress Fracture-mechanics crack-growth-rate tests have been performed on a Type 316NG SS to confirm that the susceptibility to transgranular SCC observed in slow-strain-rate tests can also occur under less sevege mechanical loading.

The tests included a lightly sensitized (EPR = 2 C/cm ) Type 304 SS control specimen.gjgost of the tests were carried out with a maximum stress intensity of

%30 MPa.m , a load ratio R of 0.95, and a frequency of 0.08 Hz, i.e., a nearly constant load with a small superposed cyclic ripple. Under these loading condi-tions, both materials cracked in water with 0.25 ppm dissolved oxygen and 0.1 ppm H2 SO 4 , and the crack growth rates were somewhat higher in the sensi-tized specimen. In a high purity environment with 0.25 ppm dissolved oxygen, crack growth stopped in the Type 304.SS specimen, but in the Type 316NG speci-menitcont{guedatalmostthesamerateasintheimpurityenvironment (N1.7 x 10 m/s). In a simulated hydrogen water chemistry environment, crack growth was halted in both specimens, although sulfate was present. In a high-purity environment with 0.25 ppm dissolved oxygen under constant-load (R = 1) conditions, the crack growth rate in the Type 316NG SS specimen decreased by an order of magnitude compared with that observed under the R '= 0.95 load, while the crack growth rate in the sensitized specimen increased to a level close to

• RSR FIN Budget No. A2212; RSR

Contact:

A. Taboada.

17-1

I that observed in the Nuclear Grade specimen under the R = 0.95 load. Since the test is still in progress, no fractographic information on the mode of cracking is available. However, previous tests have shown that the sensitized specimen cracks intergranularly under these conditions, as expected.

Fracture-mechanics crack-growth-rate tests are also being performed on specimens fabricated from a pipe with a weld overlay. One specimen is being tested in a high-purity environment, the other in an environment with 0.1 ppm H S0 . Compliance measurements indicate that in the impurity environment the 2 3 is growing through the overlay, although it is possible that it is in a crack diluted portion of the overlay and may slow down or arrest in the higher ferrite region. In the test in the high-purity environment, the crack is at present just at the interface between the pipe and the overlay, and additional testing is required to determine whether or not the crack will arrest.

Two pipe-to-pipe weldments treated by the MSIP process were examined. The process did produce favorable residual stress states in both 12-in.-dia and 28-in.-dia weldments. Unlike most residual stress improvement processes, the effectiveness of MSIP does not appear to be a strong function of pipe diameter (or thickness), although the peak compressive axial stresses in the smaller pipe weldment are somewhat higher. The 12-in.-dia weldment had been tested in boiling MgCl2at the EPRI NDE Center before being sent to Argonne. Although no cracking was reported by the NDE Center after the MgC1 7 test, metallographic studies at Argonne did reveal some cracking that was attributed to highly localized tensile stress regions which were susceptible to chloride-induced SCC.

No cracks were observed in the 28-in.-dia weldment that was subjected to higher plastic strains during MSIP but not tested in MgCl2' Slow-strain-rate tests have been performed on additional heats of German Nuclear Grade Type 347 SS. No evidence of knife-line attack has been observed in weldment specimens, but transgranular stress corrosion cracking (TGSCC) occurs in impurity environments in tests at very low strain rates 7

(<5 x 10 /s). However, Type 347 SS appears at least as resistant to TGSCC as Type 316NG SS. The deleterious effect of nitrogen concentrations greater than 0.1 wt.% on TGSCC susceptibility has been confirmed in additional tests. A preliminary examination of six Type 347-Type 347 and Type 347-Type 304 SS weldments received from the New York Power Authority has been completed. The weldments were prepared by either the German narrow-gap welding process without an insert or the more conventional welding procedure developed at the EPRI NDE Center, in which an insert is used. Before shipment to Argonne, the weldments were radiographed and examined ultrasonically; welds produced by either pro-cedure were found to be of high quality with relatively few potential defects.

At Argonne the pieces were checked by dye penetrant and ultrasonic tests, and no defects were observed.

Experiments to characterize the effect of high gamma rgdiation on the cor-rosion potential of SS have been carried out using a 9 x 10 C1 cobalt source.

A standard reference electrode was modified to operate satisfactorily in the gamma environment. The ef fect of the gamma dose rate and total dose on the open-circuit potential of Type 304 SS and the redox potential of a platinum electrode was measured in conventional BWR-like environments with 0.2-0.3 ppm dissolved oxygen and in deoxygenated water (<0.005 ppm oxygen). The gamma radiation causes a significant increase (N200 mV) in the corrosion potential of the Type 304 SS, but no significant changes were noted in the potential of the platinum electrode. No influence of gamma dose on the potentials was found over the range from 0.08 to 10 Mrad.

17-2 a

AGING DEGRADATION OF CAST STAINLESS STEEL

• O. K. Chopra and H. M. Chung Materials and Components Technology Division Argonne National Laboratory Argonne, Illinois 60439 Introduction A program is being conducted to investigate the significance of in-service embrittlement of cast duplex stainless steels under light-water-reactor (LWR) operating conditions and to evaluate possible remedies to the embrittlement problem for existing and future plants. The scope of the program includes studies to (1) characterize the microstructure of in-service reactor components and laboratory-aged material, correlate microstructure with loss of fracture toughness, and identify the mechanism of embrittlement; (2) determine the validity of laboratory-induced embrittlement data for predicting the toughness of component materials af ter long-term aging at reactor operating temperatures; (3) characterize the loss of fracture toughness in terms of fracture mechanics parameters in order to provide the data needed to assess the safety significance of embrittlement; and (4) provide additional understanding of the effects of key compositional and metallurgical variables on the kinetics and degree of embrittlement.

Microstructural and mechanical properties data are being obtained on 19 experimental heats and 6 commercial heats as well as reactor-aged material of CF-3, -8, and -8M cast stainless steels. Metallurgical characterization of the various materials has been completed. Specimen blanks for Charpy-impact, tensile, and J R-curve tests are being aged at 290, 320, 350, 400, and 450*C.

The aging times range f rom 100 to 50,000 h. The main areas of effort during the past year have been (a) mechanical properties of medium-term aged material and (b) characterization of the microstructure and fracture morphology of reactor-aged and long-term laboratory-aged material.

Technical Progress Microstructures of several heats of laboratory-aged material and material from the KRB pump cover plate were characterized by transmission electron microscopy, small-angle neutron scattering, and atom probe field-ion micros-copy techniques. The results indicate that low-temperature embrittlement of cast stainless steel is primarily due to the precipitation of chromium-rich o' phase in the ferrite matrix by spinodal decomposition. APFIM profiles for chromium show a very fine segregation (4-5 nm) in the KRB pump cover plate (reactor-aged ~6 yr at 280*C) as well as materials aged at 400*C for >3000 h.

Two other phases, viz., the nickel- and silicon-enriched G phase and an unknown Type X phase, are also observed in all low-temperature aged materials.

However, the significance of these phases in low-temperature embrittlement is not known. Microstructural data indicate that precipitation of carbides at

• FIN No. A2243; contact: J. Muscara.

17-3 l

the ferrite /austenite phase boundaries also contributes to embrittlement of the high-carbon CF-8 and -8M grades of cast stainless steel. The precipita-tion of carbides is significant for materials aged at 400 or 450*C.

Charpy-impact tests were completed on 17 experimental and commercial heats of cast materials that were aged up to 10,000 h at 450, 400, 350, 320, and 290*C. The results indicate that the ferrite content, concentration of C and N, and distribution of ferrite are important parameters in controlling aging degradation of cast duplex stainless steel. The extent of embrittlement increases with an increase .in ferrite content. The low-carbon CF-3 grades of steels exhibit greater resistance to embrittlement than do the CF-8 and -8M grades. The greater reduction in impact energy for the high-carbon grades is due to the weakening of ferrite /austenite phase boundaries by carbide precipi-tation, particularly for materials aged at temperatures above 400*C. Conse-quently, extrapolation of high-temperature laboratory data to reactor temper-atures may not be valid for the high-carbon grades of cast steels.

Impact tests conducted to obtain ductile-to-brittle transition tempera-ture indicate that thermal aging decreases the upper-shelf as well as the lower-shelf energy. The reduction in lower-shelf energy is substantial for CF-8 and -8M grades of steels because of carbide precipitation. The fracture surfaces of these specimens show cleavage of the ferrite phase and separation of ferrite /austenite phase boundaries.

Tensile and J -curve tests on three commercial heats of 10,000-h aged R

material have been completed by Materials Engineering Associates (MEA). The l results indicate that the yield and ultimate strength increase while the JIC and tearing modulus decrease with aging. The fracture toughness results show reasonable agreement with the Charpy-impact data, i.e., the relative decrease in J IC values for the aged materials is similar to the relative decrease in Charpy-impact energy. The mechanical property data are used to examine the existing correlations to predict the extent of embrittlement expected at reactor operating temperatures.

17-4

EFFECTS OF WELDING AND WELD REPAIR ON STRESS CORROSION CRACKING RESISTANCE OF STAINLESS STEEL PIPING

• S. M. Bruemmer and D. G. Atteridge Pacific Northwest Laboratory Richland, WA 99352

SUMMARY

An integrated modeling approach is being developed to enable evaluation of fabrication / welding procedures and ensure adequate cracking resistance of current and replacement piping. The modeling approach and its associated experimental program centers on three separate efforts: 1) thermomechanical (TM) history prediction from welding parameters, 2) microstructural and micro-chemical development prediction from material and TM history information, and

3) stress corrosion cracking (SCC) resistance prediction from resultant mate-rial condition and service environment. Significant progress has been made in each of these areas enabling quantitative evaluation of sensitization develop-ment in current materials and identifying a potential concern in replacement materials due to welding induced deformation in the heat-affected zone (HAZ).

Material and fabrication effects on microstructural development in austenitic stainless steels have been extensively studied. A computer model for the quantitative prediction of sensitization as a function of simple ther-mal to complex TM histories has been completed. The model is theoretically based but empirically optimized, written for use on personal computers and set up to be operator-interactive for ease of use. Pass-by-pass HAZ sensitization predictions have been made and verified by comparison to measurements on pipe weldments. Preliminary capability has also been developed to predict grain boundary impurity segregation (e.g., phosphorus) within the same model frame-work.

• Research sponsored by the Office of Nuclear Regulatory Research, U. S.

Nuclear Regulatory Commission, NRC

Contact:

J. Muscara.

17-5

Detailed measurements of pass-by-pass TM history for 4 to 24 in, diameter pipe weldments have been obtained. Weld-induced HAZ temperatures decrease, but HAZ deformation increases with pipe diameter (for same pipe schedule).

Cumulative cyclic strains of more than 75% were measured on the ID surface of a 24 in, pipe compared to about 15% for 4 in pipe. This large amount of warm / cold work in the HAZ of a large diameter pipe weldment may influence SCC resistance of both current and replacement materials. TM history measurements have also been used to calibrate a computer model to predict HAZ temperatures and cooling rates as a function of weld parameters. This model has been used to define acceptable repair welding parameters and avoid increases in HAZ sen-sitization.

Experiments were also initiated to link microstructural and microchemical conditions to SCC resistance. The effect of grain boundary chromium depletion and EPR-measured degree of sensitization (DOS) on the IGSCC of Type 304 stain-less steel has been investigated. A series of specimens were produced with various minimum grain boundary chromium concentrations and EPR-DOS values, but with comparable carbide distributions and chromium depletion widths. Ductility and fracture mode after slow-strain-rate tests in oxygenated, high-temperature water were directly dependent on minimum chromium contents. Intergranular cracking was observed in specimens with grain boundary chromium minimums up to about 14% and which showed no attack during EPR tests.

17-6

/

1 1

SURRY STEAM GENERATOR EXAMINATION AND EVALUATION R. J. Kurtz, Program Manager R. L. Bickford, E. R. Bradley, R. A. Clark, P. G. Doctor, R. H. Ferris, L. K. Fetrow and M. Lewis Pacific Northwest Laboratory Richland, Washington The Steam Generator Group Project is a multi-task effort for the NRC with additional sponsorship by EPRI and consortia from France, Italy and Japan.

The principle objectives of this program are to determine the adequacy of the current regulatory guides to; 1) define the frequency, extent and procedure for conducting nondestructive inservice inspections (ISIS) of steam generator tubing, and 2) determine tube plugging limits of service-degraded tubing under normal operating and accident loading conditions. To reach these objectives the program utilizes a retired-from-service steam generator that was removed from the Surry 2 Nuclear Station. Nondestructive and destructive examinations of the generator are providing information on the accuracy of eddy current (EC) inspection techniques to detect, characterize and size flaws in steam generator tubing. In addition, burst testing of service defected tubing shall be performed to validate models of remaining integrity developed previously using laboratory defected tubing. With this information an ISI model will be developed to serve as the technical basis for updating the regulations. The model's goal is to determine the optimal frequency, extent of inspection and tube repair and plugging criteria for specific types of ISI procedures utiliz-ing presently available field-use EC nondestructive examination (NDE) equip-ment and procedures. The program is also evaluating alternate or advanced NDE techniques for the ISI of steam generator tubes.

In prior years, the generator was prepared for extensive primary side inspec-tion by decontamination of the channel head and removal of the plugs inserted into tubes during service. Two multifrequency EC inspections were conducted to determine a post-service baseline condition of the generator. From these examinations, historical records and secondary side examinations, a subset of tubes was selected for additional detailed NDE examinations. Four round robins were conducted to estimate the reliability of multifrequency EC inspec-tions and developmental NDE techniques. Results from these exaininations displayed a wide range in numbers and types of indications with substantial agreement in defect detection only at the hot leg top of tube sheet. Reported defect sizes varied significantly between teams for all round robins. The variability in detection and sizing appears to be due to analyst interpreta-tion of complex EC signals rather than inspection equipment.

During the past year, removal of tube specimens for validation of the in-situ NDE results has been the main activity. Approximately 270 specimens were removed from the tube sheet by pulling with a single point tube puller. All specimens from the U-bend level of the generator have been removed by abrasive saw cutting flush with the 7th support plate. Remaining specimens between the 17-7

(

top of tube sheet and 7th support plate (about 200) have been removed by pulling the tube from above with an overhead crane. Severe service-induced support plate corrosion and cracking on the hot leg side of the generator permitted break-up of these plates and pulling of tubes full length with minimal damage.

Examination of removed specimens began with out-of-generator NDE for defect location and characterization using bobbin-coil EC equipment. Selected specimens were additionally inspected with rotating EC and ultrasonic probes to provide enhanced detection. Following initial NDE examination, specimens were chemically cleaned to remove a pervasive surface deposit consisting of a copper and magnetite mixture. By comparing EC inspection results before and after cle:ning it was established that the surface deposit created false defect indications in some cases and masked tube degradation in others.

All specimens have been visually inspected after out-of-generator NDE to determine locations for metallographic examination. Metallography on selected specimens has been perfonned for determination of defect extent. Most defects have been observed in the sludge pile at the hot leg top of tube sheet with the dominant defect type consisting of wastage and/or pitting. Shallow intergranular attack (s60 pm) has been observed on both the OD and ID surfaces of tubes from this region. Deep, primary side initiated, intergranular cracking of the roll expanded region of tubes plugged during service has been found, an example of continued tube degradation following removal from ser-vice. Defects identified in the U-bend region have included grinder marks, anti-vibration bar wear coupled with corrosion and cracking in Row I and 2 tubes. At intermediate levels mostly denting has been observed that, in some cases, has lead to OD initiated cracking. Minor pitting at tube support plate intersections has also been found.

At this juncture definitive conclusions regarding EC inspection reliability cannot be made since validation activities are in progress. Never-the-less it is evident from validation results to date that copper containing deposits on tube OD surfaces affect EC inspection and may lead to unnecessary tube plug-ging and/or masking of potentially serious defects.

17-8

The Effect of Crack Shape and Variable Amplitude Loading on Fatigue Crack Crowth in PWR Environments W. H. Cullen Materials Engineering Associates, Inc.

After several years (1970-1983) of fatigue crack growth rate testing of a variety of pressure vessel steels in light water reactors, with the principal aim of defining material properties effects, water chemistry effects and loading variable effects, attention is now turning towards applications-oriented studies using specimen configurations, crack geometries and loading variables which af ford a greater degree of similitude than earlier (pre-1980 or so) test practice.

Compact fracture specimens have been used to develop an extensive data base through which the effect of several critical variables have been identified, However, laboratories such as MEA are now developing research techniques for the use of part-through cracked specimens of both clad and unclad pressure vessel steels in air and reactor-grade water environments.

Until recently, most of the fatigue crack growth rate data base has been developed using constant amplitude loading. The effects of variable amplitude loading, which have been shown to be considerable in some other structural materials, have not been defined to any significant degree. To address this concern, MEA is currently c.onducting an extensive series of tests using both simple variable amplitude waveforms, and multi-component waveforms which are more reactor-typical in their construction.

Plate specimens (25 mm x 100 mm) have been prepared with 10 mm x 20 mm (a x 2c) part-through surf ace cracks located on the centerline. The specimens are rigidly gripped in tension and are instrumented with several pairs of d.c.

electric potential drop probes used to monitor the crack extension and symmetry. Specimens have been f abricated f rom both clad and unclad pressure vessel and piping steels. The matrix of tests in an air environment has been completed and the companion matrix in PWR environment is about to begin.

Tests in air environments indicate that crack shape is highly dependent on test temperature. At high temperatures, the width-to-depth ratio is decreased significantly. For clad steels, the tensile stresses in the clad at low (room) temperature result in increased crack opening forces and a large width-to-dcoth ratio.

The effect of variable amplitude loading have been studied through a program of test matrices of increasing complexity. After completion of a series of constant amplitude tests at various load ratios, the initial variable amplitude testing was in a dry helium gas evironment in order to ascertain the effects of single and multiple overload and underload schemes witnout the complex ef fects of a PWR environment. Later tests employed more complex spectra, containing blocks of spectral components with load ratios 17-9

oituleting start-ups/ shut-downs, norrsi pewsr lordings and unlordings, and simple reactor or turbine trips. Af ter completion of testing in helium gas environments, the testing in PWR environments was initiated, with the result that crack extension in a block was essentially dependent on the test frequency of the spectral component.

l f

i l

t

{

17-10 l

I

SUlelARY OPERATIONAL SAFETY RELIABILITY RESEARCH R. E. Hall and J. L. Boccio Engineering Technology Division Brookhaven National Laboratory Operat ing reactor events such as the TMI accident and the Salem automat-ic-trip failures raised the concern that during a plant's operating lifetime the reliability of systems important to safety might degrade from the de-signed-in reliability that was considered in the licensing process. This con-cern became Generic Issue II-C-4.

To address this concern, NRC, DOE, EPRI, and other organizations spon-sored surveys of reliability techniques that have proven successful in aero-space and commercial-aviation industries. NRC-sponsored studies included a Rome Air Development Center survey of Air Force reliability techniques, an ap-plication by NASA Kennedy Space Center of a reliability activity called system assurance analysis.

In May, 1986, Brookhaven National Laboratory (BNL) began the evaluation which is the subject of this report. The objective of this project are to identify the essential tasks of a reliability program and to evaluate the ef-fectiveness and attributes of a reliability program applicable to maintaining an acceptable level of safety during the operating lifetime at the plant.

Achieving high availability of safety systems, involves both controlling the configuration of safety systems so that sufficient safety equipment is al-ways available, and providing safety systems that function reliably when chal-lenged.

The objective of a reliability program is to assure that an acceptably low likelihood of core-melt frequency or risk is maintained for the plant lifetime. This is accomplished by assuring that reliability is maintained for equipment that is performing within reliability goals; and by improving the reliability of equipment that is not performing within goals, or is experienc-ing failures that can be prevented.

Some if not most reliability program tasks are presently performed by many utilities without formal reliability programs. The differences are that reliability technology allows prediction of potential problems before they re-sult in deteriorated reliability; a reliability program identifies reliability technology that can be used to perform these tasks so that pre-established re-liability targets are met; a reliability program provides techniques for moni-toring equipment performance, and alerting when pre-defined reliability tar-gets, consistent with safety goals, are not met by actual equipment perfor-mance; and a reliability program involves a discipline by which each task is performed in a way that is consistent with the risk from the problem, thus focusing resources on risk important problems.

There are definable characteristics that a reliability program must have if it is to accomplish the objective stated above. Obviously, the program v

18-1

should provide means for recognizing when a reliability problem exists. The program should also be capable of anticipating, or predicting, potential re-liability problems that could occur in the future, based on evaluations that show design or testing inadequacies; vulnerability to common cause; observed deterioration in equipment that has not yet resulted in failure; and problems that have occurred elsewhere in the industry. If a problem is predicted or diagnosed, the program should provide means of indicating the priority of the detected problem, correcting it in a time frame commensurate with its priori-ty, provide assurance that the corrective actions have been applied, and close the problem out when assurance has been obtained that the corrective action has been effective.

This research program consists of: (1) case studies of five operating plants to better understand techniques that have worked for utilities; (2) cooperative trial application to evaluate the effectiveness of reliability techniques; (3) assessment of the potential usefulness of reliability tech-niques to help resolve generic safety issues, abnormal occurrences and acci-dent precursor events; and (4) the development and application of responsive performance indicators for monitoring the risk and reliability performance at plants and for determining when degraded performance alert levels are achieved.

A closed-loop reliability program process is described which can help re-duce the frequency of transients faults that challenge safety systems and help assure that safety systems will function reliably when called upon to mitigate abnormal occurrences. A reliability program (tasks, activities, techniques, work requiremente) is described which includes both diagnosis of recurring problems and prognosis of impending problems, prioritization, and corrective actions to reduce the likelihood of recurring problems and potential problems before they take effect.

For LWR application, eight reliability tasks are identified along with a variety of techniques, methods and activities for performing each task. The basic features of reliability technology, applicable to LWR operational safe-ty, are that it allows prediction of potential problems before deteriorated reliability ensures, it identifies reliability technology that can be used to perform reliability tasks so that pre-established reliability targets are met; it provides techniques for monitoring equipment performance and alerting when pre-defined reliability targets are not met; and, it involves a discipline to focus resources on risk-important problems and problem closeout.

A ma jer conclusion from this study is that all of these features are con-sistent uith the eventual implementation of a safety goal policy and perceived ways to assure policy conformance. However, at present, implementation of current reliability programs, both within and without the nuclear industry, do not contain specific criteria which can be used by regulators to define what constitutes an acceptable program. This study has identified what it con-siders as primary ingredients for an acceptable reliability program. It has addressed how specific reliability criteria can be established for not only assessing top-level performance indicators, which a regulator may audit, but also how consistent low-level performance indicators, which the industry can use to evaluate present reliability, can be tied to top-level indicators.

18-2

RISK-BASED PERFORMANCE INDICATORS M.A. Azarm aad J.L. Boccio Reliability & Physical Analysis Group Department of Nuclear Energy Brookhaven National Laboratory Upton, New York 11973 and W.E. Vesely and E. Lofgren Science Application International Corporation

SUMMARY

i The NRC is presently focusing on ways to develop better indicators of plant performance. The goals of the NRC effort are to develop indicators that faith-fully track different aspects of plant performance and to utilize these indica-l tors to assess plant safety. However, since plant performance covers a wide spectrum of areas, the characterization of plant performance will necessarily involve multiple descriptors, where each descriptor focuses on a different as-pect of plant performance, such as operations, management, or test 'and I maintenance.

For eventual utilization, the indicators may be kept separate, whereby they constitute a multi-attribute description of plant performance or, the indicators can also be combined into an overall performance indicators, where the weighting of each indicator is determined according to some policy or utility criteria.

l For any set of descriptors of plant performance, one descriptor that is critical in directly assessing the plant safety is the descriptor of the risk level of the plant. Probabilistic risk level indicators, or simply risk level indicators, provide an updated estimate of the probability that an accident will I

occur at the plant or that a safety system will fail at the plant if called upon. The information provided by risk level indicators is similar to the prob-abilistic risk type of information that is provided by a probabilistic risk analysis (PRA). However, what makes the risk level indicators different is that contrary to PRAs the risk levels of the plant are continuously updated to

{ reflect events which occur at the plant, including not only failures but also design and operating chcnges and plant status changes.

The concept, involved in risk-based indicators is to directly utilize values or models of the risk impacts of each event and combine these with frequency estimates (including recurrence and duration considerations) from event occur-rences to determine the risk implications of the events. The risk implications of the different events which occur at a plant, and those which do not occur, are combined to give the overall risk level, or risk performance, of the plant.

18-3 i

I

As part of the Operational Safety Reliability Rssearch (OSRR) Project, this concept was initially explored using selected systems at the Davis-Besse plant (namely, auxiliary feedwater system (AFWS), and steam and feedwater rupture con-trol system (SFRCS). The results, presented in NUREG/CR-4618,1 indicate that the degraded performance of AFWS, in terms of an increase in plant core-melt f requency, could have been determined long before the June 9,1985 incident.

Furthermore, the susceptibility of the system to common-mode failures of redun-dant trains was identified as a major contributor to degraded performance. A preliminary application of counted Cusum statistical techniques,2 3 under the assumption of sampling from a Poisson process, was performed in order to track AFWS reliability vs a prespecified risk / reliability target. The results tenta-tively show that there is a potential for developing an optimized statistical technique having a fast response time and low false-alarm rate. However, fur-ther work beyond this exploratory study is considered warranted.

Several technical aspects and issues related to risk-based performance in-dicators were not addressed by the OSRR Project. NRC research is now addressing these various issues in its program on risk-based indicators. These include the following:

1. The various types of risk-level indicators that can address past, pres-ent, and future risk; and their utilization in terms of their ability to identify the specific contributors to a degraded performance.

2. The identification of various statistical indicators of risk-level in-dicators and the determination of their operational characteristics to assure early and reliable detection of degraded performance.

3. The specific information required by the indicators must be defined and missing information identified.

4. The regulatory role of risk-based indicators needs to be developed along with its integration with other indicators presently being used (SALP) or under investigation.

In conclusion, risk-based indicators have the potential of providing up-to-date readings of plant risk levels. Risk-based indicators can utilize past and present records of failure occurrences and plant activities and combine this in-formation with plant configuration and status information to give readings on public risk, core-melt frequency, and safety system unavailabilities. Projec-tions of future risk levels are also obtainable.

References

1. Azarm, M.A., Lofgren, E.V., et al., " Evaluation of Reliability Technology Applicable to LWR Operational Safety," Draf t NUREG/CR-4618, May 30,1986.

2. Van Dobben DeBruyn, C.A., Cumulative Sum Tests: Theory and Practice, Griffin, London, 1968.

3. Vesely, W.E., "Cusum Indicators of Plant Risks," Draft Report, OSRR Pro-gram, Science Applications International Corporation, February 1986.

18-4

l PROCEDURES FOR EVALUATING TECHNICAL SPECIFICATIONS (PETS) i P.K. Samanta and J.L. Boccio Reliability & Physical Analysis Group Department of Nuclear Energy Brookhaven National Laboratory Upton, New York 11973 and l

' W.E. Vesely Science Application International Corporation l

l

SUMMARY

The Procedures for Evaluating Technical Specification (PETS) Program has the objective of developing and demonstrating methodologies to utilize risk and reliability techniques in evaluating the scope, detailed requirements, and l safety impact of plant technical specifications. Many facets of the problems in l establishing risk analyses methods as a tool for establishing or revising tech-nical specifications are being studied. The program was originally scoped to focus on two of the TS elements, viz., allowed outage times (A0Ts) and Surveil-lance Test Intervals (STIs), and to establish procedures that employ risk-based insights for ascribing valid downtimes and testing frequencies of safety system components. In addition, the program is evaluating risk-based approaches for redefining action statements and is addressing additional technical specifica-tion issues.

In this paper, two aspects of technical specifications relating to Generic Issues B-56 and B-61 are discussed. The PETS Program was restructured to ad-dress these issues to assist NRC's Office of Nuclear Reactor Regulation (NRR).

During FY86, the PETS Program completed the development of methodology and ap-plication to example system to support the resolution of the issues on methodo-logical aspects.

In evaluating Generic Issue B-56 relating to diesel surveillance test in-tervals and adaptive diesel test requirements approaches have been developed which allow risk acceptable diesel test intervals to be determined for any diesel parameters. From a risk standpoint, the objective of diesel surveillance tests is to control the risk arising from diesel failures which can occur while the diesel is on standby. At the same time, risk caused by the test from test-caused failures and degradation also need to be controlled. Risk acceptable test intervals balance these risks to give a low acceptable overall risk. A PC software was developed for efficient use by regulator or plant personnel to simply determine risk acceptable test intervals for individual diesels. The FRANTIC computer code was applied to evaluate issues impacting diesel unavaila-bility as a function of diesel test intervals. The study distinguishes between the diesel accident unavailability (the probability that the diesel will not 18-5

perform its function in an accident) and the diesel test unavailability (the probability that the diesel will fail in a surveillance test). The results ob-tained show that the diesel accident unavailability is generally higher than the diesel test unavailability. The accident unavailability is higher than the test unavailability because the test does not measure the maintenance and downtime contributions which can cause the diesel to be unavailable in an accident. The behavior of the test unavailability is consistent with the requirements in Reg.

Guide 1.108, i.e., to test more frequently whenever problems arise in order to decrease unavailability. Whereas, the accident unavailability can be unaccept-ably high from a risk standpoint if test intervals are too short demonstrating that Reg. Guide 1.108 is not an effective approach. This paper will also demon-strate the effect of testing strategy, the effect of test-caused degradation, the relative effect of demand and time related failures, and the effect of in-creased number of failures on diesel test intervals. Overall, an approach to effectively detc.rmine test intervals to maintain the accident unavailability at acceptably low levels will be presented.

In response to Generic Issue B-61, the PETS program is analytically evalu-ating the effectiveness of cumulative outage time requirements for safety system components. The allowed cumulative outage time (ACOT) is an alternative to present allowed outage time (A0T) requirement where a component may undergo as many corrective maintenances and repairs as deemed necessary by plant personnel and can stay down as long as necessary to complete a particular repair if the sum of downtimes for the component in a defined time period is less than the assigned ACOT for the component. The paper will discuss analytical models for describing the component up and downtimes. Due to the random nature of the up-times and downtimes, the successive periods for which a component is up or down vary. The probability distribution of the uptime durations and the probability distribution of the downtime durations are assumed to form an alternative re-newal process and standard renewal equations are used to develop these distribu-tions. The cumulative downtime distribution developed in this manner gives the likelihood that a given ACOT will be violated in the defined time period. An associated distribution for the cumulative downtime risk is developed from the distribution of cumulative downtime and the downtime risk associated with a given downtime. The cumulative downtime risk distribution show the degree to which the cumulative downtime risk is controlled for a given allowed cumulative downtime. An application to a group of components in system logic configuration will also be presented. The risk control capability of the ACOT for different repair times and repair time distributions and the methods for selecting ACOT values will be discussed. The paper will als) present the implementation characteristics of an ACOT comparing its advanteges and disadvantages over an A0T.

i 18-6

An Overview of the Plant Risk Status Information Management System J. R. Kirchner D. J. Campbell JBF Associates, Inc.

Knoxville, Tennessee The Plant Risk Status Information Management System (PRISIM) is a personal computer program that presents PRA results and related information in an easy-to-understand format that is free of PRA jargon. It provides users with easy access to specific information by making extensive use of menus. PRISIM was developed to provide NRC inspectors with useful information in a format that can help them decide where to focus their efforts in limiting or reducing plant risk.

PRISIM contains two basic types of information: (1) pre processed information that is useful for long-term inspection planning and (2) a plant risk model that allows the user to specify equipment that is out of service and determine the risk implications of the plant status. Specific uses of both types of information will be described in the full paper.

To assist the inspector in using the information in PRISIM, we have provided a procedure-related access path. The inspector first specifies the inspection procedure he needs to perform. The program then presents a list of decisions required by the procedure. After selecting a decision, the inspector is provided with information relevant to the decision and, where appropriate, guidance on applying the information. In addition, the user may access all information contained in PRISIM directly without following the procedure-related path.

In developing PRISIM, we obtained input from inspectors and the inspectors have been enthusiastic in their support of the program. We have completed a prototype version of PRISIM for Arkansas Nuclear One - Unit 1 (ANO-1). This prototype is currently being field tested at the plant and at Region IV headquarters. The interim results of this field testing will be documented in the full paper.

We are currently developing -a second PRISIM prototype for the Peach Bottom - Unit 2 Plant. This prototype is scheduled for completion by the end of CY86, at which time it will be field tested. We are concurrently developing for both ANO-1 and Peach Bottom - Unit 2 modified versions of PRISIM that are designed for a wider range of regulatory applications.

These programs will allow the user to change the PRA input data and assess the impact of these changes on the core melt frequency.

18-7

In conjunction with the Idaho National Engineering Laboratory we are developing a PRISIM production facility that will make possible (1) the development of PRISIM versions for all plants that have documented PRAs and (2) the timely updating of these programs as plant designs and operating procedures change.

i 18-8 i

L_____ _ __ _ _ __ . _ _ _

l i

SYSTEM ANALYSIS AND RISK A_SSESSMENT SYSTEM (SARA)

E.A. Krantz, K.D. Russell, H.D. Stewart E.G.&G. Idaho, Inc.

Idaho National Engineering Laboratory P.O. Box 1625 Idaho Falls, Idaho 83415 Summary Utilization of Probabilistic Risk Assessment (PRA) related information in the day-to-day operation of plant systems has, in the past been impracticable due to the size of the computers needed to run PRA codes.

This paper discusses a microcomputer-based database system which can greatly enhance the capability of operators or regulators to incorporate PRA methodologies into their routine decision making. This system is called the System Analysis and Risk Assessment (SARA) system.

SARA was developed to facilitate study of frequency and consequence analyses of accident sequences from a large number of light water reactors (LWRs) in this country and to provide a means of tracking the expected change in risk from nuclear power plants resulting from the implementing modifications resulting from generic issues. This information is being amassed by several studies being sponsored by the United States Nuclear Regulatory Commission (USNRC). The amount of information gathered via these studies is massive, and a means to manage it was necessary. It was also found necessary that many people have access to this information.

To meet the need of portability and accessibility, it was felt that a microcomputer-based system would be most suitable. The database was developed by E.G.&G. Idaho, Inc. at the Idaho National Engineering Laboratory to optimize the performance and speed of the microcomputer while maintaining the capability to significantly expand its capacity in the future. The relational database was developed in the Modula-2 structural language, is highly interactive and has a variety of quality control and security features. The SARA database has a full complement of search, sort and reporting routines. Reports can be generated containing all the information that the system can display to the user on screen.

The SARA system also has the capability to perform a variety of calculations. SARA users can make changes to basic event unavailabilities and sequence initiating event frequencies to simulate changes in component reliability / availability. This calculational capability permits the users to conduct sensitivity or risk / benefit studies on component perfomance and directly compare the before-and-after results. Provisions have been made in the SARA system to examine the impacts of the user imposed model changes on basic event importances, sequence frequencies, severe core damage frequencies, and offsite consequences. The consequence 18-9

calculations include tha capability to make modifications to the consequ2nco modals to conduct sensitivity studies similar to those for sequence analysis.

The SARA system was designed to provide the user with the capability to examine the PRA results from several analyses and perform direct cross plant comparisons. These results can then be examined comparatively to determine similarities / differences in the factors driving risk at those plants. Such comparisons can be made, not only of PRA results, but also of plant system configuration. SARA has the capability to store and present plant system diagrams, component pictorials, graphs and charts and to present results comparisons graphically. Examples of these graphics are presented in the body of this paper.

All the capabilities of SARA can be used by interactirg with the system through a hierarchical menu scheme. The menus are highly user-friendly, fill-in-the-blank type screens, and there are many on-line user " helps" to further simplify operation of the system. The screens developed for SARA have been human-engineered to optimize user performance. The screens utilize display techniques which aid the user in understanding the indicated results.

l l

f d

18-10

Management of Generic Safety Issues Warren Minners Division of Safety Review and Oversight Office of Nuclear Reactor Regulation The NRC program for the management of generic safety issues consists of six l stages: identification, prioritization, resolution, imposition, implementation and verification. The purpose of the program is twofold. One purpose is to assure that appropriate measures to control the risk to the public from significant safety issues that apply to several or a class of nuclear power plants are implemented. An agency-wide Safety Information Management System is being developed to track issues through the entire process and assure that this purpose is fulfilled. The second purpose of the program is to assure that the resources of the NRC and the industry are applied to the resolution and implementation of these issues as efficiently as possible.

The identification of safety issues is the primary function of the Office of Analysis and Evaluation of Operational Data (AE0D); however, other offices in the NRC, the Advisory Committee on Reactor Safeguards, the industry and the public are also sources of safety issues. A catalogue of about 140 generic activities was first issued in 1978. In 1980 the NRC Action Plan Developed As the Result of the TMI-2 Accident added another set of about 350 issues. About 180 new generic safety issues have been identified since then.

The prioritization stage is performed by the Division of Safety Review and Oversitht in NRR, which first screens the issues for duplication or overlap with existing generic issues. When accepted as a new generic issue, issues are succinctly defined, assigned a number and title, classified and catalogued with all other generic issues in NUREG-0933. Not all suggested issues are safety issues, that is, related to possible deficiencies in the design, construction or operation of a nuclear power plant. Some are classified as possibly unnecessary or overly restrictive requirements while others are 18-11

related to protecting the environment or to improving the licensing process.

The program focuses on generic safety issues, but does not exclude these other types of issues. Currently there are about 50 issues in the prioritization stage.

As a basis for efficiently allocating resources, a high, mediun, low or drop priority is assigned. High and medium priority issues are being scheduled for resolution, while low and drop are not. Priority is based on rough quantitative estimates of the risk reduction and the net cost to the NRC, the licensees and the public if a resolution were implemented. Relaxations of requirements are also ranked on this same quantitative basis, but a qualitative basis is used to rank issues related to environmental protection j or the licensing process.

During the resolution stage, engineering analyses, quantitative risk assessments and cost estimates are performed. Based on this information, the possible alternatives, which always include no action, are identified, a cost /Denefit analysis is prepared and a solution recommended. If new requirements or regulatory guidance are proposed, they are reviewed by the regulatory staff, the Advisory Comittee for Reactor Safeguards, the Comittee for Review of Generic Requirements, the public and, in some cases, the Commission. There are about 100 generic safety issues that are now in the resolution stage.

If new requirements or guidance are approved, they are issued to licensees by the licensing divisions in NRR to begin the imposition stage. Plans and

schedules for implementing the new requirements or guidance are submitted by the licensees and then reviewed and approved by the NRR licensing divisions.

Licensees then implement these commitments by making the necessary changes in equipment, structures, procedures or staffing. If warranted, the Office of Inspection and Enforcement may perform inspection to verify implementation by licensees.

I 18-12

Development and Application of a Dynamic Viscoplastic Fracture Model M.F. Kanninen, S.J. Hudak, Jr., R.J. Dexter,  !

K.W. Reed, E.Z. Polch and J.W. Cardinal Engineering and Materials Sciences Division Southwest Research Institute San Antonio, Texas This research is being pursued to assist the HSST Program in obtaining reliable procedures for the prediction of crack arrest at the high upper shelf toughness conditions occurring in postulated pressurized thermal shock (PTS) events. The ultimate objective of the research is to quantify an effective crack propagation / arrest criterion for PTS conditions through viscoplastic-dynamic finite-element analyses of fracture experiments; e.g., the NBS wide plate tests. Supporting this are companion activities involving (1) viscoplastic material characterization of A5338 steel, (2) the development of test methods to obtain dynamic crack propagation / arrest data on A533B steel using small-scale laboratory specimens, (3) asymptotic crack tip and small-scale yielding crack propagation analyses of viscoplastic-dynamic crack propagation, and (4) elastic-plastic pipe fracture analyses. These activities are aimed at providing the basis for a more effective analysis procedure, and to set the stage for extending the analyses to include re-initiation by ductile tearing following arrest.

The focal point of the work is the development of a versatile viscoplastic-dynamic finite-element analysis model. There are two reasons for requiring this level of complexity. First, unless the crack jump length is small, dynamic effects can significantly influence a run-arrest event.

Second, at upper shelf conditions, rapid crack propagation induces rate-sensitive inelastic deformation that requires strain rate effects to be taken into account. The accompanying experimental work consists of (1) tensile split Hopkinson bar testing to obtain stress-strain data at higher strain rates, and (2) compact tension fracture testing instrumented to obtain detailed crack length and load-line displacement histories. In the former area, data obtained at strain rates in the order to 550 sec-g have been incorporated into a Bodner-Partom viscoplastic constitutive model. This representation is input for the finite element fracture computations.

Duplex compact , tension fracture experiments on A5338 steel at room temperature were performed in which crack length and load-line displacement data were successfully captured. These data, which were significantly improved by instituting new crack length measurement procedures developed in this program, are needed to set the time-dependent boundary conditions for the finite element analysis. In these experimer.ts, rapid fracture is initiated in 4340 steel and propagates into a A533B steel test section at a high speed.

However, it has not vet been possible to analyze these experiments.

19-1 l

In the computations that have so far been performed, viscoplastic-dynamic finite element simulations of dynamic crack propagation and arrest in the generation-phase sense were made of NBS wide plate experiments WPl.2 through WPl.6. The dynamic stress intensity factors, determined with a rscently implemented energy release rate algorithm, were consistent with values developed from the CT00. These gave values that were nearly the same as the elastodynamic values reported by ORNL and others. It was further found that the zones of significant inelastic deformation were much smaller than would be expected from quasi-static analyses. Because these zones cannat be reliably determined with the mesh sizes that are possible on the VAX 780, conversion of the code for a CRAY supercomputer was made. Preliminary re-analyses of the wideplate tests on the CRAY will be provided in the paper.

19-2

BRITTLE-TO-DUCTILE TRANSITION BEHAVIORS IN NUCLEAR REACTOR VESSEL STEELS i

George R. Irwin University of Maryland College Park, Maryland Brittle-to-ductile (cleavage-fibrous) transition behaviors of reactor vessel steels are important to fracture-safe operation of nuclear plants.

These behaviors determine the RT-NDT temperature rating, add complexity to selection of conservative toughness values, and control the maximum temper-ature below which rapid cleavage can develop. Test specimen sizes have ranged from Charpy and small compact specimens to the ORNL intermediate vessels and thermal-shock cylinders. Wide-plate tests at NBS are currently of special interest. The temperature region of main practical interest is the upper portion of the transition range, adjacent to the loss-of-cleavage temperature. This temperature region is elevated substantially by increase of constraint and of loading rate. In order to make efficient use of frac-ture toughness information obtained with a variety of tests, we need a sound understanding of cleavage-fibrous transition behaviors.

The local inhomogeneities natural to steels employed in reactor vessel fabrication provide a wide variation in local fracture strength. Using tests in which the measurement point is determined by onset of cleavage, previous studies have shown that the scatter of toughness values for tests of different specimen sizes can be approximately represented using a flaw probability analysis method. Other methods, mainly fractography and metal-lography, have been used to determine the physical mechanisms which control initiation and spreading of cleavage at fine scale. Results from investiga-tions of this kind are discussed in this paper.

Several inhomogeneity features are easily seen and are closely related to cleavage initiations. These are the carbide density banding in A5338 and A508 steels and variations of inclusion density in weld metals. The high stress conditions necessary to initiate cleavage in a ferrite grain are pro-vided by abrupt local-region separations. After initiation, spreading of the cleavage crack is resisted by lattice disorder and by discontinuities at grain boundaries. At larger scale, connections between non-coplanar cleavage cracks form and subsequently break by ductile hole-joining. Increase of temperature favors separation by hole-joining and enhances the resistance to spreading of cleavage. Small islands of cleavage can be found by careful inspection of fractures at temperatures moderately above the loss-of-cleavage temperature. These represent the most persistent conditions for cleavage initiation and their study has been of special value.

The brittle-to-ductile (cleavage-fibrous) transition behaviors are com-pl ex . However, enough understanding is now available to support development efforts toward quantitative models which are clearly related to the essential features. Progress toward this goal can provide improved input for risk calculations and improved efficiency in the use of fracture testing.

I 19-3 l

l

FLAW DENSITY EXAMINATIONS OF A CLAD BOILING WATER l REACTOR PRESSURE VESSEL SEGMENT

• K. V. Cook and R. W. McClung Metals and Ceramics Division OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee 37831

SUMMARY

As part of the Heavy-Section Steel Technology Program, studies have been conducted to determine flaw density in a section of reactor pressure vessel cut from the Hope Creek Unit 2 vessel [ nominally 0.7 by 3 m (2 by 10 ft)]. This section (removed from the scrapped vessel that was never in service) was evaluated nondestructively to determine the as-fabricated status. We had four primary obj ectives: (1) evaluate longitudinal and girth welds for flaws with manual ultrasonics, (2) evaluate the zone under the nominal 6.3-mm (0.25-in.) clad for cracking (again with manual ultra-sonics), (3) evaluate the cladding for cracks with a high-sensitivity l

fluorescent penetrant method, and (4) determine the source of indications l detected.

Three pieces were used for the study; approximately 0.67 m (2 ft) of girth weld and 2.5 m (8 f t) of longitudinal weld were included. The i cladding surface in the three samples is approximately 1.8 m2 (18 ft2 ), i All ultrasonic examinations were performed manually using contact techniques. Different search units were applied, depending on the specific ultrasonic method being employed. We used a nuclear-grade commercial water-soluble couplant with a stable gel viscosity. This nontoxic, rust-inhibitive couplant performed satisfactorily for the contact methods employed. A commercial flaw detector was used for all ultrasonic testing.

Volumetric ultrasonic, pulse-echo, angle-heam examinations of the length of seam welds in the three samples, were, as nearly as possible, performed using ASME Code techniques. The small size of the samples limited the application of 60* shear waves; however, all weld sections were interrogated with 45* shear-wave beams from two orientations (perpendicular and parallel to the weld centerline) and, when possible, from two directions. Additionally, these tests were performed from both

• Research sponsored by the Office of Nuclear Regulatory Research, Division of Engineering Safety, U.S. Nuclear Regulatory Commission under Interagency Agreement DOE 40-551-75 with the U.S. Department of Energy under contract DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc.

By acceptance of this article, the pubh,her or recipient ack nowledget the U S. Government's nght to retain a nononclus.ve, royalty-free he.n.. .n and to .ny coovneht covering the article.

19-5

the outer and inner surfaces of the samples. No recordable indications were detected from the outer surface. Five ASME Code-recordable indica-tions were detected from the clad side [1.e., 50% distance amplitude correction (DAC) or more in amplitude response]. Seven others were detected when 20% DAC recording was used, for a total of 12 indications.

A dual-element search unit that generates nominally 70* longitudinal beams with a second angle to produce a pitch-catch maximum sensitivity to underclad cracking at a depth of about 9.5 mm (0.375 in.) was used to examine the approximate 1.8 m2 clad surface for underclad cracking. No significant indications were detected near the clad interface.

Approximately 1.8 m2 of clad surface was inspected for surface-breaking flaws with a fluorescent penetrant technique. A black residue was present on all three samples that required scraping to remove. In addition, a solvent cleaner was selected to remove the residual residue.

After the dye penetrant had remained on the cladding surfaces for a minimum of 30 min, a soft cloth / mpened with solvent was used to remove the penetrant from the surface. A developer was applied and the surfaces inspected with a 100-W fluorescent black light in the darkened laboratory a rea. All indications were obviously caused by rough surface areas and, in most instances, were produced by the weld perturbations that exist between weld passes [at about 25.4 mm (1 in.) along the longitudinal axis of the pressure vessel section]. Thus, no significant indications of surface flaws were detected.

A brief comparison was made between the apparent flaw density detected in this examination and the predictions made in the Marshall report. Data for the Marshall report were primarily obtained from unclad nonnuclear vessels. Our evaluation implies a greater flaw density than p redicted in the examined sections. Evaluation of these indications is continuing using nondestructive and destructive methods. However, to date, results indicate that many indications may have been associaten with the cladding.

l 19-6

ContainmGnt Venting Analysis for Peach Bottom D. L. Batt EG&G Idaho, Inc.

M. T. Leonard Battelle Columbus Division H. S. Blackman EG&G Idaho, Inc.

W. R. Nelson EG&G Idaho, Inc.

Work Supported by the U. S. Nuclear Regulatory Commission Office of Nuclear Regulatory Research Division Of Reactor Analysis Operations under DOE Contract No: DE-AC07-76ID01570 There is a widely perceived potential for reducing risk from severe accidents in BWRs with Mark I containments by venting the primary containment during accident sequences that would otherwise challenge containment integrity by gradual overpressurization. Initial studies by the Industry Degraded Core Rulemaking (IDCOR) Program concluded that a substantial reduction in risk could be realized through containment venting. Because of this perceived potential, the use of venting as a means of controlling containment pressure, was included in the BWR Owners' Group Emergency Procedure Guidelines (EPGs) and is being incorporated into some plant specific emergency operating procedures (EOPs). However, the IDCOR studies were not of sufficient detail to confidently conclude that existing E0Ps, vent path systems, and appropriate operator actions could be effectively used during risk significant accident sequences to lessen an overpressure threat to containment integrity.

A plant specific evaluation of the effectiveness of containment venting l was, therefore, performed for the NRC to support its reevaluation of nuclear power plant risk (NUREG-1150). The Peach Bottom Atomic Power Station Unit 2 was selected for analysis because a draft venting procedure has been developed by the licensee and it is the BWR Mark I reference plant being analyzed for NUREG-1150.

Technical Approach and Results Three severe accident sequences were selected for analysis: (a) an l

anticipated transient without scram (ATWS) with early containment failure (labeled sequence TC1); (b) an ATWS with core degradation preceding the conditions of containment failure (sequence TC2); and (c) a station blackout l with core degradation preceding the conditions of containment failure (sequence TB1). These sequences were selected from those identified by the Accident Sequence Evaluation Program (ASEP) as being significant contributors

( to severe accident risk at Peach Bottom. For each of these sequences, i

operator performance, equipment performance, and physical phenomenology as they relate to containment venting were examined in detail. The effect of containment venting on several additional accident sequences was evaluated in a qualitative manner in an attempt to consider a broader spectrum of severe accident scenarios.

20-1

Our evaluation indicates that the effectiveness of containment venting at Peach Bottom (i.e., the capability of existing plant systems and E0Ps to lessen an overpressure threat to containment integrity) is highly sequence specific. For accident sequences such as TW (transient with a failure of containment heat removal) in which the rate of energy addition to the containment is limited to decay heat of an intact core, containment venting is reliably effective in preventing containment overpressurization and, as a result, prevents core damage. The effect of this conclusion is evident by the elimination of this sequence from the set of significant risk contributors in ASEP's analyses (whereas TW headed the core melt frequency l

list in WASH-1400). For the accident sequences that remain as significant contributors to risk at Peach Bottom (station blackout and ATWS) containment venting was not found to be effective. In general, the estimated probabilities of operators being able to successfully control containment pressure by venting in these sequences are moderate to low, with the extreme case being station blackout, where the success probability is zero. The results of our estimation of the success probability of venting for the specific accident sequences analyzed in detail (mentioned above) are summarized in Table 1.

Other accident sequences were considered for which containment venting was judged likely to be effective e.g., loss-of-coolant-accidents with breaks inside containment. However, the risk reduction potential for venting in these sequences is inherently small, due to their relatively low frequency of occurrence. The net effect is that containment venting, as currently l envisioned at Peach Bottom, has a limited potential for further reducing risk from severe accidents.

Results to date suggest that changes in the Peach Bottom venting procedure could improve the operator performance aspect of containment venting. For station blackout and ATWS sequences, however, some changes in j plant hardware are also likely to be necessary to significantly improve the likelihood of successfully implementing the venting procedure. Unlike

{

changes in procedures, hardware changes are usually quite expensive and would require a value/ impact assessment, such as that presented in the NRC's j backfit rule, to ascertain the cost effectiveness of such changes.

I i

20-2

TABLE 1. VENTING SUCCESS PROBABILITIES Success Sequence probability Basis Uncertainty TC1 0 Operator opens four of 95% confidence bound (power >12%) four 18-in. vent paths for operator success:

0 .93 Insufficient flow to Assumes vent valves reduce pressure do not open fully.

Four 18-in. paths will control pressure if valves open fully TC1 0.6 Operator opens four of 95% confidence bound (power <12%) four 18-in. vent paths for operator success:

0 .93 Pressure controlled Assumes vent valves do not open fully. Two of four 18-in. paths will control pressure if valves open fully TC2 0.7 Operator opens one of 95% confidence bound four 18-in. vent paths for operator success:

0 .94 Pressure controlled Assumes containment survives earl.i pressure spike TB1 0 Hazardous environment Assumes that personnel at valve locations will not be exposed during or following to lethal environments valve operation 20-3 l

l 1

NUCLEAR SAFETY Forward on Technology and Backward on Perception W.B. Loewenstein G.R. Thomas Safety Technology Department Electric Power Research Institute The nuclear power industry is completing a momentous year.

This year saw the United States pass 1000 cumulative reactor-years of commercial operation

- with the number of licensed reactors exceeding 100 (Hope Creek 1 received the 101st license). Nuclear power should supply ~16% of all electrical power in the U.S. for this year.

Major technical advances in nuclear safety have occurred in the past year with improvements in:

• application of advanced technologies for improving both plant control, operational efficiency and safety;

• knowledge and understanding of seismic event consequences;

• analytical ability to predict plant operation ranging from normal conditions to hypothetical severe accident source terms;

• the experimental data base for supporting the above activities.

However, this year has also seen the unfortunate Chernobyl accident. For the first time in commercial nuclear power history, a reactor accident has claimed lives. Chernobyl has dominated the year in:

• the public's perception of nuclear power safety;

• the industry's focus away from the technical accomplishments in nuclear safety towards a response mode of placing Chernobyl in the proper perspective to the vastly different light water reactor plants that dominate outside of the Soviet Union.

There is a positive side to this tragic accident. Chernobyl has already and undoubtedly will continue to foster a much greater international exchange in the arena of nuclear safety.

This past year has been a year of major accomplishments in the maturation of EPRI's nuclear safety research program. Some of these accomplishments are noted below:

• Use of advanced control technology became a reality in the United States with operation of the first digital controller for a major plant system - the Digital Feedwater Control System at the Monticello BWR.

21-1 l

. Initial versions of several advanced artificial intelligence tools were produced for enhancing plant on-line monitoring and analysis. These tools improve both plant output and safety.

• Understanding seismic ground responses improved with programs that:

- consolidate and qualify existing central and eastern U.S. seismic data;

- initiated monitoring of actual seismic events in both soft soil and bedrock in the U.S. and in Taiwan.

The ability to analyze normal conditions through hypothetical severe accident source terms was enchanced with the completion or improvement of several codes ranging from system thermal hydraulics to detailed fission product behavior in primary system piping.

Over the last four years, EPRI has managed large-scale efforts directed at improving knowledge in the areas of hydrogen-air mixing and combustion. A generic program of research ranging up to large-scale tests (in a 15.9-m diameter sphere) has been completed. A major owner group experimental effort is nearing completion characterizing BWR6 Mark III containment responses to hydrogen burns.

Within the United States, EPRI is involved in many other collaborative efforts with owner groups, host utilities, vendors, DOE and NRC. The TREAT STEP tests and the MIST B&W safety programs are current examples.

In the international arena, EPRI is involved in several areas which foster R&D interchanges.

• The OECD LOFT program completed the experimental phase last year. EPRI has continued major support resulting in extension of the program to include detailed analysis and examination of the LOFT FP-2 experiment.

. Both the LACE and Marviken aerosol-behavior experimental programs will be completed by the end of this year, with EPRI continuing to support post-experiment evaluations.

As a significant and new area of international participation, since the day of the Chernobyl accident, EPRI has been involved in technical evaluations of the accident and active participation in resultin, international meetings. This participation will continue as long as the need exists.

This paper examines EPRI's nuclear safety research program in the context of this new transitional phase and how it may be used to meet the everyday challenges of commercial nuclear power.

21-2

- - _____________ _ m

I Containment and Piping Research H. T. Tang and S. W. Tagart Electric Power Research Institute 3412 Hillview Avenue Palo Alto, California 94303 Containment The concrete containment integrity research sponsored by EPRI at Construction Technology Laboratories and Anatech International, Inc. is aimed at establishing the true failure modes and load-carrrying capebility of reinforced and prestressed containments under internal pressures beyond those for which they were designed. The goal is to provide industry with a test-verified analytic tool for evaluation of containment integrity under various postulated severe accident scenarios.

In the first phase of the testing program, carried out in a preexisting NRC facility, concrete slabs representing segments from reinforced and prestressed containment walls were tested under uniaxial and biaxial tensile loading. The objective of the first phase testing was to check out the testing procedure and to provide a database for checking out concrete and steel liner behavior for actual failure characterization to be investigated in Phase 2.

The second phase of testing used a 50-million pound multiaxial reaction rig, the largest of its kind, and addressed specifically the prestressed concrete containment failure mode characterization under internal overpressurization.

The results of the second phase testing demonstrated that:

1. the liner will develop a crack in the discontinuity region where high load strain concentration is induced,

2. a preexisting crack or a crack initiated at the discontinuity region will arrest due to concrete-liner interaction,

3. the liner anchorage appears to play a critical role in concrete-liner interaction.

The current phase experimental effort will be tests of similar specimens with reinforced configurations where reinforcement in the containment wall is much more dense, and the liner anchorage design is slightly different from the prestressed one.

The analytic effort has centered on the verification and application of the nonlinear finite-element code ABAQUS-EPGEN for predicting the complex concrete cracking and concrete-steel interaction behavior that could produce local failure of the liner plate. The challenge here is to develop a material model, a liner crack criterion, and a finite element idealization that can 21-3

account for the overall response of the containment and the local deformations leading to liner rupture or warping of penetrations.

Piping The nuclear power industry has a need for more realistic piping design rules than are present in the current ASME code together with supplementary NRC regulations. This need is made clear when we observe the large number of dynamic snubbers which appear in current nuclear plants. We see 2000 or more pipe snubbers in typical new plants and these snubbers have become a greater problem than the pipes they are intended to protect.

The ASME Code rules for piping were developed more than 15 years ago where dynamic loading of metallic components was not as well understood as it is today. The assumption made was that from an allowable stress viewpoint, pipes may fail if dynamic stresses reach levels which would cause static collapse.

This was recognized as a conservative assumption. However, since the degree of dynamic margin was not quantitatively established, it was not practical to justify a less conservative rule.

The EPRI/NRC research project 1543-15 was designed to provide data on pipe failure behavior which will clearly establish the true dynamic margin available and permit the development of realistic and defensible design rules for dynamic type loading. These types of experiments (component, system, and material) are planned in this 3 year effort.

Approximately 40 pressurized pipe component tests will be dynamically loaded to failure to form the principal database in the program. These tests will include pipe elbows, tees, reducers, nozzles and, lug attachments. Three kinds of dynamic loads are included, low frequency seismic, mid frequency hydrodynamic, and high frequency waterhammer loads. To date 12 component tests have been completed and all tests confirm that static collapse does not occur. Fatigue retcheting is the consistent failure mode.

The pipe system tests are intended to verify that the component test results allow us to sufficiently understand dynamic pipe system failure behavior, so that the design rules can be significantly changed. Only three pipe system tests are planned since the failure behavior of the components in the system test is expected to be identical to the test results more thoroughly explored in the component tests. The system tests will also show how apparent damping is affected by significant plastic behavior, and how redundancy causes a shedding of dynamic load from one location in the system to another.

The material tests are intended to establish how ratcheting affects fatigue life for nuclear piping materials and to quantify how much ratcheting is expected with various degrees of new stress. This information is needed if we are to take full advantage of the new knowledge about dynamic pipe failure.

That is, once we have confirmed that static collapse failure does not occur before fatigue ratchet failure occurs, then current code fatigue rules need to be modified to include the influence of ratcheting.

21-4

_______ ___ _ __ a

l l

"USE OF PROBABILISTIC SYSTEM ANALYSIS FOR ENHANCING PLANT OPERATION SAFETY AND PRODUCTIVITY" Boyer B. Chu Safety Technology Department Nuclear Power Division Electric Power Research Institute Many nuclear power generation stations are installing upgraded on-site computers and developing computer-based plant information management systems. An on-site computerized information management system could support the data needs of many engineering and opera-tional applications, for instance, compiling equipment failure data used in system reliability analysis. Probabilistic system analyses have increasingly been gaining industry-wide and regulatory accept-ance. They can effectively and systematically assess system design reliability, identify critical equipment list, and rank the relative importance of each equipment to the overall plant performance.

These analytical techniques, together with the plant specific equip-ment availability information data can be useful for determining system operability status, monitoring technical specifications compliance and planning equipment testing and maintenance activities.

The study, "Use of Probabilistic System Analyses for Enhancing Plant Operational Safety and Productivity" is a major on-going R&D endeavor within the Risk Assessment Program of Nuclear Power Division at Electric Power Research Institute (EPRI). The study intends to develop a practical tool which combines key features of plant information management systems with system reliability analysis techniques to assist plant personnel and engineering staff in performing their functions more effectively and accurately. To achieve this objective, the study has two major tasks: (1) to develop a user-friendly, integrated computer software, and (2) to demonstrate the applications and value of this software on-site.

The software named Reliability Assessment Program with In plant Data, RAPID, is an aggregate of many stand-alone computer codes. It consists of three interrelated elements:

~

(1) an Executive Controller which provides the users with interface to and control of the other two elements, (2) a Data Base Manager which administers the data file management, and (3) Applications Modules which perform specific engineering functions. A broad range of these functions has been developed, for instance:

o Equipment status and system operability monitoring, o Equipment technical specifications compliance monitoring and time action tracking.

o Reliability-based system engineering analysis, o Technical specification evaluation and optimization, o Dynamic plant productivity / reliability assessment, and o Reliability and root cause data compilations and report gen-eration.

The immediate emphasis focuses on the development of four applica-tion modules: a Plant Status Module, a Technical Specification 21-5

Evaluation Module, a Reliability Assessment Module, and a Utility Module.

The Plant Status Module, RAPID /PSM, has been developed and its functionalities are being demonstrated on site. Forty-six (46) plant-specific GO system models consisting of five thousand four hundred (5,400) elements have been developed and integrated to perform six primary functions:

o Maintain equipment (component) status information o Calculate operational status o Determine technical specification compliance o Track timing of technical specification action statements o Assess plant health o Prepare event record and shift record logs The input to PSM is only a list of current out-of-service compo-nents. By using this list, PSM can evaluate the plant system logic models and determine the operational status of the plant, its systems, and their trains. It also accepts from a postulated list of out-of-service components for "what if" inquiries. The resulting plant operational status information is then used to monitor technical specification compliance and track time action statements. The " Plant health" is a new dimension. It assesses

" Plant health" both in terms of the probability of continued power operation at various established power levels of the current plant configuration and the availability of safety systems to function when it is required. The RAPID /PSM is in its final software testing and data verification.

The Reliability Assessment and Utility modules, RAPID / RAM-UM, are being demonstrated at another member utility company. The RAPID / RAM software provides a computer environment for performing reliability tasks. The emphasis is to demonstrate, in a cost-effective manner, that a "living" Level I PRA study can be a reality. The RAM soft-ware creates and maintains a series of model files which contain information regarding the model logic, the data for model quantifi-cation, model cut sets and other information necessary for maintenance of the models. It verifies the model input data, edits the model, and prepares a new input deck to execute. The RAM provides updated results and models to the database manager for automatic documentation, update and quality assurance.

The Utility Module (UM) acquires data from plant information system and converts it, when necessary, for use in RAPID. The interface with the plant system is flexible and is adaptable to most plants.

The RAPID /UM is being used to extract the data from the on-site plant information management system in support of RAPID / RAM demonstration.

This paper summarizes RAPID software design and functionalities, details of two ongoing demonstration studies and their preliminary results.

21-6 r

A UTILITY PERSPECTIVE ON PLANT MODIFICATIONS RESULTING FROM SAFETY RESEARCH Richard E. Deem Peter P. Bieniarz New York Power Authority Risk Management Associates This paper will present how a practical application for an operating nuclear power plant can result from plant specific source term analyses. Source term results for a BWR MK I plant will be presented and how conclusions from that analysis determined the design of the modification. The overall criterion the modification design had to adhere to for implementation in the plant with its overall benefits will also be discussed, l

1 21-7

A Seismic Hazard Methodology for the Central and Eastern United States J. Carl Stepp and Jerry L. King Electric Power Research Institute i

Palo Alto, CA 94303 I l

The purpose of a probabilistic seismic hazard-analysis is to provide a documented basis for making informed decisions about ground-motions appropriate for seismic design of a specific facility at a given site.

Seismic hazard is usually depicted as probabilities that given levels of earthquake ground shaking will be exceeded annually. Given current understanding of earthquake causes and processes, uncertainty on the hazard at any site may be large; therefore, a hazard methodology should include procedures to quantify uncertainty that properly reflect uncertain ihput parameters and computational models.

Recently, development of probabilistic seismic hazard metho characterizingseismichazardatlowprobabilitiesof(<10gology per year) at '

locations in the central and eastern United States has been based on input interpretations of seismic sources and their associated seismicity parameters l by multiple experts. To express uncertainty, rrultiple alternative interpretations are elicited.

An ongoing seismic hazard methodology development program at the Electric Power Research Institute (EPRI) follows.this general approach. The goal of the program is to develop procedures that are consistent with earth science practice, that facilitate expressions of uncertainty in input interpretations, and that are generally applicable to assess seismic hazard at the low probability levels required for nuclear plant safety assurance.

New developments and approaches, incorporated into the EPRI methodology, include formation of Earth Science Teams to perform all input interpretations; development and utilization of a common, uniform data base for interpretations; a structured approach to develeg interpretations and to characterize uncertainty on them that specifically includes separate considerations of scientific and data uncertainty; extensive analysis and upgrading of the historic earthquake catalog; new procedures to estimate seismicity parameters; and a number of computational efficiencies. Input interpretations have been developed by six Earth Science Teams rather than by individual scientists. By taking the team approach it is believed that bias in the input interpretations due to uneven discipline knowledge is minimized.

A strong emphasis has been the development of a uniform data base for the entire region of the central and eastern United States. By developing and using a uniform data base it is believed that bias due to uneven knowledge of data among teams has been minimized. The procedure to interpret seismic sources consists of two parts; definition of source zone geometry and derivation of probability that the source is active. The starting bases for defining source geometries are tectonic features contained in a tectonic framework interpretation. The marginal probabilities of activity assessed for 21-9

elements of the tectonic framework form the basis for deriving the probability that each source is active. Alternative interpretations result in multiple combinations of active sources that might affect a site, depending on interpretations of dependencies among them. Depiction of alternative interpretations is facilitated by use of logic tree structures.

A number of innovations in the assessment of temporal and spatial rates of earthquake occurrence have been made to maximize use of the available earthquake data set. The earthquake catalog was subjected to extensive analysis to: 1) establish a single magnitude measure mb with error estimate,

2) analyze clustering (classify earthquakes as main er dependent), and

3) estimate incompleteness. The procedure used to obtain uniform magnitude estimates uses all measures of size (maximum intensity, felt area, magnitude) for each earthquake in the catalog and yields an estimate of error. The procedure to identify clusters and aftershocks is general and accommodates spatially non-homogeneous catalogs with incompleteness-induced non-stationarity. By this procedure main earthquakes are distinguished from dependent events in the master catalog for recurrence estimation.

Incompleteness is assessed using a probability-of-detection model which is a function of time, magnitude, and geographic position. Together these analyses permit maximum use cf the earthquake catalog to estimate source zone seismicity parameters.

Assessment of uncertainty in hazard estimates is facilitated by use of a logic tree structure to weight alternative input interpretations. Each end branch of the logic tree structure represents a weighted alternative interpretation of all input parameters and can be used for a single seismic hazard computation. The computation has a weight which is the product of the assessed probability values of all intermediate branches. The uncertainty in seismic hazard at a site due to a Teams' alternative interpretations of all input parameters is then the range of hazard results obtained from the end branches for all alternative source zone combinations relevant to that site.

Total uncertainty can be obtained through a weighted combination of Teams' interpretations.

21-10

The Role of Oversight Within NRR and Its Effect on Industry Robert J. Bosnak, Chief Engineering Issues Branch Division of Safety Review and Oversight Industry performing research, developing new technology, developing new analytical methods and procedures not previously used, intended to be applied to operating nuclear power plants or plants still under construction, should be eware of the fact that the last reorganization of the Office of Nuclear Reactor Regulation established a Division of Safety Review and Oversight (DSR0) and three separate licensing divisions, each with its own separate technical review staff. The need for " Oversight" was anticipated as being necessary to achieve consistency in the licensing reviews of the three divisions which are identified organizationally by NSSS manufacturer type.

In its broadest sense, the oversight process must cope with two distinct situations.

The first involves consistency in the day to day application of regulatory requirements dealing with conventional normal practice. This situation presents no great problem provided experienced staff are available. The second situation is much more difficult, in that the results of research or the development of new methods or procedures which have not been previously used in the licensing process, must now be coordinated, communicated and evaluated by each of the technical discipline groups responsible for its application, in order that such epplication be cora,istent, irrespective of the group performing the review.

There have been several specific instances in the past few months of " Oversight" situations of the second type from which lessons may be drawn on improving the process.

Industry should be aware of and plan for necessary presentations and discussions with the NRR staff on the results of their work and how it is intended to be cpplied to plant specific licensing situations. DSR0, in its Oversight function, stands ready to assist the industry in the acceptance of new analytical methods and procedures, but unless such considerations are planned for and factored into specific plant schedules, there is a risk that such methods and procedures j may not be accepted for plant specific application in the time frame desired.

1 l

21-11

EVALUATION OF SURFACE CRACKS EMBEDDED IN REACTOR VESSEL CLADDING D. E. McCabe Materials Engineering Associates, Inc.

An extensive review by NRC and industry specialists on the pressurized thermal shock scenario has resulted in the recognition that the effect of a surface flaw embedded in the clad layer of the pressure vessel wall should be j better understood. The condition of the local region containing such a flaw j

is an extremely complex mix of microstructures and stress gradients. Metal-lurgical complexity is from stainless steel weld metal, heat-af fected zone, base metal; and for stress complexity there are residual stresses, thermal stresses, and the primary membrane stress. Efforts to analytically determine the f racture properties of this region have yielded mixed results, suggesting that the final analysis must reside in experimental evaluations where the composite effects are integrated into one result.

An experimental program has been undertaken to investigate the fracture behavior of clad plate containing a flaw. Two 102-mm (4-in.) thick plates of A 533-B steel were clad layered on top and bottom surf aces with approximately 9.5 mm (0.37" in.) of 304, 308, and stainless steel welds using a commercial three-wire process. They were subsequently stress relief annealed at 620*C (1150*F) for 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />. Residual stresses in the region of the clad layer were determined. Baseline mechanical properties have been determined for the three metallurgical zones. This includes tensile values, Charpy energy, and f rac-ture toughness in terms of J -cunes.

R The focus of the study is on crack initiation from surface cracks in irradiated and unirradiated clad RPV steels. The cracks have an aspect ratio, a/c, of 0.6. The two crack depths chosen were 5 mm (0.2 in.) and 10 mm (0.4 in.). The small flaw is completely embedded within the clad layer, and the larger one generally penetrated into the heat-affected zone. All were EDM-machined notches subtended with f atigue precracks. Specimens were four-point loaded bend bars that are known to exhibit some of the features of a flaw in a thermal shock scenario. The tests were made over a transition range so that the overall interactions of the composite would, be evident in the form of a transition temperature shift. Base metal vs. ' clad-layered specimen comparisons (with comparable flaw sizes) will demonstrate either a detrimental effect (transition temperature increase) or a beneficial effect (transition temperature decrease) due to the clad layer.

Quantitative evaluation in the form of elastic plastic stress intensity factors was developed using empirical techniques that have been successfully applied to specimens with large through-thickness flaws. Three models were tried: Equivalent Energy, a Simplified Tangent Modulus method, and Effective Crack via ASTM Standard E 561. All were found to be reasonably consistent in that they agreed substantially with each other. The E 561 method, however, did not perform well when specimen deformation approached plastic collapse.

Tests made on base metal yielded a K 3e transition curve that was about the same as that obtained with IT and 1/2T compact specimens.

22-1

c--

At this point, only the unirradiated materials have been tested, and the following characteristics were shown:

e The clad specimen transition temperature was lower than that of base metal by 20*C to 30'C.

e Clad material has low JR-curve slope that results in easy slow-stable crack growth. Ductile cracking in the clad layer is a significant consideration in the safety analysis. Growing cracku~in cladding material tend to enhance onset of cracking growth in the normally highly growth resistant base metal.

e The Jpcurve characteristics of the materials tend to be similar in compact specimens and in surface-cracked specimens.

e. ,

s 22-2

COMP 03ITION AND TEMPERATURE EFFECTS ON ANNEALING /REIRRADIATION AND DOSE-RATE EFFECTS ON IRRADIATION EMBRITTLEMENT J. R. Hawthorne Materials Engineering Associates, Inc.

l MEA investigations in 1986 have probed further the contributions of metallurgical variables to postirradiation annealing behavior and have extended the assessment of neutron flux level effects on radiation-induced embrittlement accrual. Studies of dose-rate effects have evolved new data from the UBR test reactor experiment series and separately, have generated first experimental comparisons of radiation sensitivity for the Gundremmingen reactor vessel (KRB-A) material exposed in dissimilar reactor environments.

Material composition is expected to play a major role in postirradiation annealing recovery, both for welds and plate materials. MEA has underway two tasks to probe the significance of this variable to annealing response. One task, coded Composition Effects on Annealing (CEA), is studying the annealing response of materials initially acquired for investigations of variable material sensitivity to 288'C radiation embrittlement and underlying damage mechanisms. Initial findings from one of two planned reactor experiments were presented to the 1985 WRSRIM. A primary determination was that residual embrittlement (after 399'C annealing) is a function of copper but not phosphorus content. The second CEA experiment has now been completed. Its results illustrate the effects of variable copper and variable nickel contents on recovery for the materials having low phosphorus levels. The other task, identified as the High Temperature Annealing (HTA) study, is investigating welds with high/ low copper and high/ low nickel contents and dif ferent flux types. Here, two postirradiation annealing temperatures: 454*C vs 399'C are being employed for insight both into composition effects and annealing temperature effectiveness. Results of the first postirradiation annealing comparison for the materials and preliminary data from re-irradiation exposures following annealing at 454*C and 399"C are presented.

New qualifications of potential dose-rate effects, using the UBR, involve assessements of the ASTM A 302-B reference plate and a reference submerged arc (S/A) weld af ter irradiation in positions adjacent tobutoptsidetheUBRfuel lattice. The neutron flux level was about 5x 10 n/cm -sec~I, E > 1 !!eV.

The S/A weld was made with Linde 80 flux and MnMoNi filler and has a high

, (0.35%) copper content. Postirradiation mechanical properties determinations were notch ductility via C tests, quasi-static fracture toughness via 0.5T-CT specimen tests (J-R curve)y dynamic f racture toughness via f atigue precracked Cy testsandtensilestreyth. pesultsarecomparedtoarecentevaluationof the materials at 5x 10 n/cm from an i n-co re (high flux) irradiation and higher fluence data trends.

22-3

In parallel with the UBR exposures, the significance of irradiation exposure rates is being tested using archive material from the Gundremmingen reactor (KRB-A) vessel. The effort represents a joint USA-FRG undertaking initiated by the NRC where MEA is reponsible for the USA portion of the research investigations dealing with archive material. MPA is the MEA counterpart in the FRG. In 1986, Harwell joined in the effort through an agreement between the USA and the UK. At this time, MEA has largely completed the experimental phase of the USA portion of the program. This presentation describes the effects of test reactor irradiation vs power reactor service for the KRB-A material and the relative sensitivity of the KRB-A material to light water vs heavy water environments.

l l

\

22-4 l

PIPING FRACTURE MECHANICS DATA BASE A. L. Hiser Materials Er.gineering Associates, Inc.

A comprehensive, computerized data base of fracture toughness (J-R curve) I and other support data from nuclear piping steels has been established. This data base is for use to provide the materials input required for assessments of piping integrity with known or assumed flaw and loading conditions.

This computerized data base is the culmination of a multi-step process.

First, a survey of the FSAR's (Final Safety Analysis Reports) from operating ,

plants revealed the type of steels used most commonly in U.S. plants. This I survey documented the ASME specification, as well as the pipe size and wall thickness.

Concurrent with this survey activity, a format for the data base was established, with data input forms and an instruction manual for the forms written. This format provides for all possible characterizations of the material (including chemical composition, tensile, impact and K yc data, if available) and sufficient data from the J-R curve test such that future modifications to J-integral analysis methods can be accommodated in a relatively easy manner.

Once a data matrix and a data base format had been established, a survey of the technical community was made to ascertain the availability of pertinent test data and future plans for testing. A parallel path has ~ centered on procurement of piping materials for additional testing, as required to fulfill the data matrix requirements.

This data base is accessible to NRC personnel, contractors, and other outside users, as required. Regular updates are made as more data are generated at MEA and by others with close coordination of testing plans tied to other NRC-sponsored research programs.

23-1

SAND 86-2115A Development and Status of MELCOR Code

• by F. Eric Haskin and S. W. Webb Sandia National Laboratories

SUMMARY

MELCOR is a fully integrated, relatively fast-running code that models the progression of severe accidents in light-water-reactor plants. An entire spectrum of severe-accident phenomena is modeled in MELCOR. Characteristics of severe accident progres-sion that can be treated with MELCOR include the thermal-hydraulic response in the reactor coolant system, reactor cavity, and containment / confinement buildings; core heatup and degrada.

tion; core-concrete attack; heat-structure response; radio-nuclide release and transport; and the impact of engineered safety features on thermal-hydraulic and radionuclide behavior.

MELCOR has been developed for the US Nuclear Regulatory Commission to succeed the Source Term Code Package. MELCOR is intended for use in conjunction with probabilistic risk assess-ments (PRAs). It has been designed to support sensitivity and uncertainty analyses. MELCOR will be used to evaluate the radio- j logical source terms for three NUREG-1150 type PRAs in 1987. j MELCOR is written in ANSI Standard FORTRAN 77 and has a struc-tured, modular architecture that facilitates the incorporation of new or alternative phenomenological models. MELCOR includes a number of features that enhance user flexibility and permit sensitivity and uncertainty analyses. These include input pre-processing and checking, control functions, tabular functions, sensitivity coefficients, a restart capability, and a general plotting routine.

MELCOR uses a control volume and junction approach to model the thermal hydraulics of both the reactor coolant system and containment. The nodalization is user specified. Mass and energy are explicitly conserved. Natural circulation flow patterns are calculated by MELCOR based on the control volume and flow path input data and the driving forces predicted by the physical models. A new treatment of melt progression has been implemented that semimechanistically models the candling of molten core materials and the gravitational settling of rubblized

• This work was supported by the U. S. Nuclear Regulatory Commission and performed at Sandia National Laboratories which is operated for the U. S. Department of Energy under Contract Number DE-AC04-76DP00789.

24-1

debris in the core and lower plenum. The CORCON cod 2 has been integrated into MELCOR to model core-concrete attack. Multiple corium pools and time-dependent additions of core materials and water can be modeled in MELCOR. MELCOR includes models for many engineered safety features including suppression pools, contain-ment sprays, fan coolers, and filtration units. Incorporation of an ice condenser model is not complete.

In-vessel radionuclide releases are based on models extracted from the CORSOR code. Ex-vessel releases from core-concrete interactions are based on the VANESA code, which has been incorporated into MELCOR. Aerosol agglomeration and deposition is modeled using the MAEROS sectional aerosol code. Fission product vapor physics is modeled in a manner similar to that used in TRAPMELT. Simple chemical reactions can be modeled in MELCOR as specified by user input. Models for removal of radionuclides by suppression pools, containment sprays, and ice condensers are patterned after models in SPARC, CONTAIN, and ICEDF respectively.

Removal of radionuclides by fan coolers is also modeled.

24-2

SAND 86-2129A Analysis of Peach Bottom Station Blackout with MELCOR S. E. Dingman F. E. Haskin R. K. Cole R. M. Summers Sandia National Laboratories Albuquerque, NM 87185 An analysis of the Station Blackout sequence at Peach Bottom has been performed using the MELCOR code. The calculated core and containment thermal-hydraulic responses and the radionuclide releases are presented in this paper. In a number of areas, the MELCOR results are significantly different from results that have been obtained with other codes (e.g., MARCON 2.lB and the Source Term Code Package (STCP)). These differences are mainly due to phenomena which are modeled in MELCOR but not in the other codes, such as in-vessel natural circulation and continued oxidation and heat transfer from core debris follcwing bottom head dryout. In addition, a more detailed nodalization has 1 en used for MELCOR (6 reactor coolant system volumes, 2 containment volumes, and 9 secondary containment volumes) than in the other codes. The results of calculations that were performed to determine the sensitivity of the results to modeling assumptions regarding core-concrete attack are also presented.

• This work is supported by the United States Nuclear Regulatory Commission and performed at Sandia National Laboratories which is operated for the US Department of Energy under contract number DE-AC04-76DP00789.

24-3

SAND 86-2128A MELCOR Validation Results*

by C.D. Leigh and R.K. Byers Sandia National Laboratories C.J. Shaffer Science and Engineering Associates

SUMMARY

The term " code validation" as used herein refers to the comparison of code predictions with experimental results. As stated in NUREG-0956, "The NRC agrees with the strong recommendation of the American Physical Society's study scoup cn radionuclide release from severe accidents that additional validation of the NRC's severe accident computer codes be undertaken."

To provide for improved code validation, a coordinated validation effort is being conducted by NRC. Past and near-term future experimental results that are suitable for code validation are first identified. Schedules are then established for comparing calculations with detailed mechanistic codes such as SCDAP, TRAP-MELT, MELPROG, and CONTAIN to the selected experimental results. Finally, calculations performed with the integrated codes, MELCOR and the Source Term Code Package, are compared to (bench- )

marked against) a subset of the cases selected for validation of the detailed mechanistic codes. This provides comparisons of MELCOR results to detailed mechanistic code results as well as to the experimental data.

MELCOR comparisons to three sets of experiments, the ABCOVE Aerosol Experiments, the HDR V44 Steam Blowdown Experiment, and the Battelle-Frankfurt Gas Mixing Experiments are complete. Comparisons of MELCOR calculations to other experiments, analytic solutions, standard problem results, and other calculations performed with the Source Tem Code Package and ORNL SASA codes are also underway.

The Aerosol Behavior Code Validation and Evaluation (ABCOVE) tests are of interest for LWR applications because the spherical cluster structure of the aerosols is similar to that expected of particulate aerosols in a steam environment. MELCOR was run for ABCOVE Tests AB5, AB6, and AB7, and the results were compared to the experimental data and to the CONTAIN results for the same tests. MELCOR and CONTAIN results for these tests agree. Both codes predict the suspended mass of aerosol in each test very well during the aerosol sources and within a factor of two at later times when the suspended concentration has been significantly reduced.

Predictions of the mass deposited due to settling are all within an 11%

error. Predictions for the mass deposited on vertical surfaces are less

• This work was supported by the U. S. Nuclear Regulatory Commission and performed at Sandia National Laboratories which is operated for the U. S. Department of Energy under Contract Number DE-AC04-76DP00789.

24-5

l accurate. In ABS , where thermophoresis is the primary deposition mecha-nism, MELCOR predictions are within 12% of the experimental results.

However, when plating occurs by inertial impaction or by diffusion, neither MELCOR nor any of the other codes involved in the ABCOVE comparisons gives accurate results for the amount of material deposited on the walls at the end of the tests. This is primarily because impaction is not currently included in any of the codes, and the diffusional boundary layer thickness, a parameter involved in current models for diffusional deposition, is highly uncertain for the given experimental conditions.

The HDR-V44 experiment was a reactor-scale steam experiment conducted in 1982 by Kernforschungszenrum Karlsruhe (KfK) at the decommissioned HDR reactor facility near Frankfurt, West Germany. Experiment V44 was one of a series of water and steam blowdown experiments conducted to simulate full-scale loss-of-coolant accidents, was initiated from saturated steam conditions, and had the highest RpV liquid level. HELCOR calculations for HDR V44 were performed and compared to the experimental results and the CONTAIN calculations. In the MELCOR calculations, the predicted peak system pressure is approximately 30% higher than measured and the predicted peak temperature is about 20 K higher. This level of agreement is relatively good, since the peak pressure during blowdown requires accurate modeling of phenomena which are dif ficult to characterize such as the water carryover and the forced convection heat transfer rates.

The longer term pressure and temperature (af ter about 1000 sec) are in good agreement with the measured values. During these later times (from 50 - 1500 seconds) natural convection predominates. The agreement between the experimental and the calculated results in this regime validates the MELCOR natural convection heat transfer modeling and heat conduction in structures.

The Battelle-Frankfurt Mixing Tests comprised a series of experiments in which hydrogen-nitrogen mixtures were injected into a model containment at the Battelle Institut e.V. Frankfurt. Reported data were variations in hydrogen concentrations with time and location. MELCOR calculations were performed for tests with simple and relatively complex containment configurations, each of which had a uniform and a nonuniform initial temperature distribution, preliminary calculations for a steady state (i.e., no injection) in the more complex geometry showed that elevation discontinuities and numerical roundof f can cause circulating flows which might override the flow patterns after injection was initiated. With injection active, MELCOR results and previously obtained RALOC and HECTR results for the two isothermal cases agreed well with experimental data; any quantitative discrepancies could be attributed to differences in the accuracy with which the calculations modeled injection rates and compartment volumes. In contrast, none of the codes produced uniformly good agreement with experiment for the cases with nonuniform initial temperatures. Extreme sensitivity to initial temperature distributions and surface heat transfer is probably responsible for the observed discrepancies.

24-6

Summary of CCTF test results

- Assesment of current safety evaluation analysis on reflood behavior during a LOCA in a PWR with cold-leg-injection-type ECCS -

Tadashi IGUCHI , Yoshio MURAO, Jun SUGIMOT0 Hajime AKIMOTO, Tsutomu OKUBO, Tsuneyuki H0JO Japan Atomic Energy Research Institute Science and Technology Agency of Japan

1. Introduction A reflood test program (I) for a large-break loss-of-coolant acci-dent (LOCA) of pressurized water reactor (PWR) has been conducted at Japan Atomic Energy Research Institute (JAERI), by using large sea test f acilities, which are the Cylindri{g1 Core and the Slab Core Test Facility (SCTF) .

Test Facility (CCTF)

This program has been in a part of 2D/3D project which is performed by USNRC, BMTF and JAERI.

The CCTF is an experimental facility designed to model a four-loop 1100 MWe class PWR with the flow scaling ratio of 1/21.4 and to simulate the thermo-hydraulic behavior in the primary system during the refill and the reflood phases of a PWR-LOCA. The CCTF has a full-height scaled pressure vessel with a cylindrical core of 1824 electrically-heated rods and four loops with passive and active components (for example, active steam generators, primary pumps).

Since 1979, JAERI has performed 56 CCTF tests. They can be classi-fled into 5 categories.

(1) Cold-J eg-injection simulation test under evaluation model (EM) condition.

(2) Cold-leg-injection simulation tests for parametric ef fect study.

(3) Cold-leg-injection tests to verify that the CCTF simulates a PWR properly.

(4) Alternative ECC simulation tests, such as upper plenum injection, downcomer injection and combined injection (cold legs and hot legs).

(5) Refill simulation tests to investigate the thermal hydraulics in the primary system during the end-of-blowdown to reflood initiation.

The test conduction of JAERI for CCTF has completed at March,1985.

The analytical works have been performed in the following fields.

The work was performed under contract with the Atomic Energy Bureau of Science and Technology Agency of Japan.

25-1

1 (1) Physical un erstan in on refill and reflood phenomena for PWRs with cold-leg-injection-type ECCS (2) Investigation on parameter ef fect on reflood phenomena (3) Evaluation on adequacy of similarity of CCTF to PWRs (4) Evaluation on conservatism of current saf ty analysis (5) Evalugiononpredictabilityofbest-estimatecodes, i.e. TRAC and REF LA (6) Physiscal understanding on reflood phenomena for PWRs with other injection types of ECCS (Downcomer injection, Upper plenum injection and Combined injection)

This presentation focuses on (4).

2. Evaluation on conservatism of current safety analysis We selected WREM code (5) as a representative safety analysis code. The calculated result of the code was compared with CCTF test.

This calculated clad temperature was higher than CCTF data, indicating the overall conservatism of the code against the CCTF data.

The code calculates the core inlet boundary conditions for the first step, and then calculates the peak clad temperature for the second step by using the calculated core inlet boundary conditions. The calculation on the core inlet boundary conditions showed the reasonable agreement with data, indicating that the conservatism of the code comes mainly from the second step calculation (peak clad temperature calculation).

The conservatism of the code against CCTF data in the second step calculation (peak clad temperature calculation ) was attributed to the following three points.

(1) No modeling in the code on heat transfer enhancement caused by the ununiform radial core power profile (2) No horizontal cross flow assumption between subchannels at each elevation in the code (3) Conservative heat transfer correlations in the code References (1) Hirano, K. and Murao, Y. : J. Nucl. Sci. Technol,22(10), 681(1980)

(2) Murao , Y., et al.  : J. Nucl. Sci. Technol, 19(9) , 705(1982)

(3) Adachi, H., et al.  : NUREG/CP-0048, 2, 277(1984)

(4) Murao , Y., et al.  : JAERI-M 84-243(1984)

(5) WREM  : NUREG-75/056(1975) 25-2

SCTF III test plan and recent SCTF III test result Tadashi IGUCHI, Takamichi IWAMURA, Hajime AKIMOTO, Akira ONUKI , Yutaka ABE , Tsuneyuki H0JO, Isao SAKAKI, Akihiko MINATO , Hiromichi ADACHI, Yoshio MURA0 Japan Atomic Energy Research Institute

1. Introduction A reflood test program (1) for a large-break-loss-of-coolant accident (LOCA) of pressurized water reaccor (PWR) has been conducted at Japan Atomic Energy Research Institute (JAERI), by using large scale test facilities, which are the Cylindrical Core Test Facility (CCTF) and the Slab Core Test Facility (SCTF). This program has been in a part of 2D/3D project performed by USNRC, BMTF and JAERI.

The main objective of SCTF tests is to study two-dimensional ef fect on thermal hydraulics during reflood phase in the core of full radius.

Therefore, the SCTF was designed to have a full-length, full-radius and one-bundle width core and to simulate the multi-dimensional thermal-hydraulics in the core. The core flow area scaling ratio against typical 1100 MWe class PWRs is 1/21.

The SCTF tests have been parformed by using Core-I and Core-II(2) ,

In these tests, the reflood phenomena in PWRs with cold-leg-injection-type,ECCS are focussed. These tests have been completed by July, 1985 and the analytical activity is underway. Since January 1986, the SCTF test with Core-III (SCTF III test) have been performed. In this presentation, the followings are expained.

(1) SCTF III test plan (2) Quick review of SCTF III test results

2. SCTF III test plan Special objectives of SCTF III are as follows.

(1) For PWRs with cold-leg-injection-type ECCS The objective is to investigate further two-dimensional thermal-hydraulics.

(2) For PWRs with combined-injection-type ECCS The objective is (i) to attain physical understanding on reflood phenomena, (ii) to establish a quantitative prediction method on core cooling, (iii) to provide data base for analytical codes.

According to the previous tests (3)*(4) for PWRs with combined-injection-type ECCS, the following features have been observed.

The work was performed under contract with the Atomic Energy Bureau of Science and Technology Agency of Japan 25-3 a

(1) CCFL behavior with two-region separation (water downflow region and two phase upflow region)

(2) Significant multidimensionality on core cooling behavior (Rapid quenching in water downflow region and slow quenching in two-phase upflow region)

(3) Net negative flow of coolant in core We think it is necessary to confirm above (1)d3) in full scale facility and quantify them, in order to attain the objectives of (2)

(1) and (ii). Hence, we perform three groups of SCTF III tests (core cooling test, CCFL test and integral test) with full-radius core. The accoglishment code .

by data analysis will be incorporated into REFLA

3. Quick review of SCTF III test results By core cooling tests with ununiform ECC injection into upper plenum, it was confirmed that the two-region separation and the multidimensional thermal hydraulics were realized in the full-radius core. In the water downflow region, even before bottom reflood initiation a significant core cooling (50~100 W/m K)2 was observed and the upper half of the core was quenched, in spite of no (or slight) water accumulation in the region.

The water downflow infuenced the core cooling in the region adjacent to the water downflow region. However, littJe core cooling (practically no core cooling ) was observed in the f ar region before bottom reflood initiation.

Af ter bottom reflood initiation, good core cooling was observed even in two phase upflow region, although the multidimensional core cooling remained. The entire quenching was much earlier in comparison of the estimated one for PWRs with cold-leg-injection-type ECCS.

By CCF tests with steam injection of forced f eed, the limited water downflow (by CCFL) and the massive water downflow (i.e. break through) were observed, depending on steam injection rate.

References (1) Hirano, K. and Murao, Y.: J. Nucl. Sci. Technol, 22(10),681(1980)

(2) Adachi, H. , et al.  : NUREG/CP-0048, 2, 277(1984)

(3) Murao, Y. , et al.  : Thirteenth Water Reactor Safety Research Information Meeting, Gaithersburg (1985)

(4) Watzinger, H. , et al.  : Secord International Topical Meeting on Nuclear Power Plant Thermal Hydraulics and Operation, Tokyo (1986)

(5) Murao, Y. , et al.  : JAERI-M 84-243 (1984) 25-4

HEAT TRANSFER ENHANCEMENT IN SCTF TESTS Takamichi IWAMURA, Hiromichi ADACHI, Makoto SOBAJIMA, Akira OHNUKI, Tsutomu OKUBO, Yutaka ABE and Yoshio MURA0 Japan Atomic Energy Research Institute The SCTF (Slab Core Test Facility) Program is a part of the Large Scale Reflood Test Program together with the CCTF (Cyl in-drical Core Test Facility) Program. The major objective of the SCTF Program is to investigate two-dimensional thermal-hydraulic behavior in the core during the reflood phase of a loss-of-coolant accident (LOCA) of a pressurized water reactor (PWR). In order to meet this objective, SCTF simulates a full radius slab section of a PWR with 8 bundles arranged in a row and the heating power for each bundle can be independently controlled.

It was revealed from the SCTF tests that heat transfer was enhanced for the high power bundles and degraded for the perip-heral low power bundles due to the ef fect of radial power distri-bution.

The radial distribution of heat transfer coef ficient was strongly related to the f act that the vertical pressure drop above the quench f ront was higher for the high power bundles and lower for the low power bundles. The higher vertical pressure drop in the high power bundles is explained to be caused mainly by the higher liquid f raction in these bundles due to the follo-wing three ef fects;

1) The entrainment generation rate around the quench f ront is larger due to the higher cladding temperature,

2) A natural convection below the quench front is induced by the radial distribution of steam generation rate, and The work was performed under contract with the Atomic Energy Bureau of Science and Technology Agency of Japan.

25-5

3) The concave distribution of the quench front enhances the water concentration into the high power bundles.

The higher steam flow rate in the high power bundles also contri-butes to the higher vertical pressure drop in these bundles.

l The Bromley type film boiling heat transfer correlation was not able to predict the large dif ference of heat transfer coef fi-cients between bundles.

Based on a dispersed flow model, the heat transfer coeffi-cient was expressed as the sum of three terms ; radiation, drop-let impingement, and forced convection to steam. The heat tran-sfer coef ficients calculated with this model agreed well with the data from the SCTF, the CCTF and the FLECHT-SEASET when the liquid fraction was less than 0.1. The steam temperatute tended to be underestimated in the present calculation, resulting in the overestimation of heat transfer coef ficient especially during the initial period.

The heat transfer enhancement for the high power bundles due to the radial power distribution was predicted well with the present dispersed flow model as shown in Fig.1.

5 _ Test S2-12 (steep power)

"E Elevation 2.33 m x

3C 5

Bundle 4 (high power)

.2 u

k100 o - , I"" ~~

~

.d -

v ,N y j_ .' g Y -'

g .' I/

e / y Experiment Z Bundle 8 ---- Present method j

9 (low power)

-- Murco-Sugimoto 0

O 100 200 Time after beginning of reflood (s)

Tig. 1 Prediction of two-dimensional heat transfer behavior with dispersed flow model and Murao-Sugimoto correlation 25-6

DEVELOPMENT OF REFID0D MODEL AT JAERI Yoshio MURA0, Tadashi IGUCHI, Hiromichi ADACHI Hajime AKIMOTO, Tsutomu OKUBO, Tsuneyuki H0JO Takamichi IWAMURA, Akira OHNUKI, Yutaka ABE Japan Atomic Energy Research Institute ,

1. Introduction In the safety analysis on the Loss-of-Coolant Accident (LOCA) of pressu-rized light water reactors (PWRs), it is very important to evaluate the tempera-ture history of the fuel rod claddings during the reflood phase, which is governing the integrity of the first enclosure of the fission product. As tools for realistic evo2 uation of the fuel claddings, we are developing best estimate models for core enoling during reflood phase based on the physical understanding of the phenomena. In order to investigate the reflooding phenomena, data f rom Small scale reflood test facility, Cylindrical Core Test Facility (CCTF), Slab Core Test Facility (SCTF) and Upper Plenum Test Facility (UPTF) in FRG has been used.

2. Model development for PWRs with a cold leg injection type ECCS.

For PWRs with a cold leg injection type ECCS, a reflood analysis code REFLA has been developed.

The latest version of the code is REFLA/ MOD 4. This version includes some new improvements in the core thermo-hydrodynamic models, that is (1)1ocal power ef fect model for evaluating radial power distribution ef fect, (2) pellet-cladding gap conductance effect, (3) cladding material ef fect on quench propagation. This code successfully shows the capability of realistic simulation of the CCTF tests and the SEFLEX test.

Grid spacer ef fect and flow housing ef fect are being incorporated into REF LA/ MOD 5. And the mixing vane ef fect and mechanisms of core water accumula-tion are studied in small scale reflood tests.

The work was performed under contract with the Atomic Energy Bureau of Science and Technology Agency of Japan.

25-7

The SCTF and CCTF test results showed the core cooling enhancement due to the radial power ef fect. The ef fect on the peak clad temperature, however, seems to be weak, since the turn-around of clad temperature occurs in a short time af ter reflood initiation. In lower flooding rate cases which expect to cause delay of turn-around, the effect on the peak clad temperature becomes more important. Therefore qualitative explanation on the core cooling enhancement are being derived as shown in the previous presentaton.

The system model of the REFLA code includes the pressurization model of the pressure vessel due to the pressure loss in the broken cold leg nozzle, and the water accumulation model in the upper plenum, the downcomer bypass model and the pressure loss model in the broken cold leg nozzle.

The core model of the REFLA code was installed in the TOODEE 2 code in the WREM code system. The impro e WREM code provides more flexibility, since the original WREM code uses empirical FLECHT heat transfer correlation, while the improved uses the physical model. The core model is also being installed in J-TRAC code explained in the next presentation.

3. Model development for PWRs with a combined injection type ECCS For PWRs with a combined injection type ECCS like German PWRs, modification of the REFLA code and the J-TRAC code are considered. In the modeling, the following items should be studied :

(1) Condensation of steam in the hot legs, the upper plenum, the lower plenum, and the core, (2) Counter current flow at the end box of the core, and (3) Core cooling by fall-back water and bottom flooding water.

For these purpose, CCTF, SCTF and UPTF test results are being analysed. In the case of UPTF, Once the water fall-back occurs, the steam injected into core seems to be condensed by f alling-back subcooled water in the core and upward steam flow at the end box is decreased, if there is no feed-back system for steam injection. While in the facility with simulated fuel rod the f all-back water feed the source of steam in the heated core and the response of the upward steam flow against f all-back flow expected to be dif ferent. SCTF in expected to provide the information on the feed back system for steam injection into the core of UPTF.

25-8

UPTF TEST RESULTS FIRST 3 SEPARATE EFFECT TESTS BY P.A. Weiss, R.J. Hertlein Kraftwerk Union AG 8520 Erlangen (FRG)

Summary The Upper Plenum Test Facility (UPTF) Experimental Program, sponsored by the Ministry for Research and Technology (BMFT) is the German contribution to the trilateral 2D/3D Project, and is performed within international cooperation among Japan ( JAERI), USA (USNRC) and the Federal Republic of Germany (BMFT).

The UPTF simulates the primary cooling system of a KWU 1300 MW PWR. The upper plenum, including internals, the downcomer and the four connected loops are represented in 1:1 scale. The core is simulated with controlled injection of steam and water supplied from external sources.

The three intact loops are equipped with flow restrictors to simulate the reactor coolant pumps, and with steam / water separators representing the steam generators. The hot and cold legs of the broken loop lead through steam / water separa-tors and break valves into the containment simulators. Breaks of variable sizes can be simulated in the hot and in the cold leg respectively.

According to the internationally agreed test matrix, the following three Separate Effect Tests have been performed during March to June 1986.

- Test No. 12: Simultaneous steam up-and water downflow at the tie plate

- Test No. 1: Fluid-fluid mixing in cold leg arid downcomer

- Test No. 11: Countercurrent flow of steam and saturated water in a PWR-hot leg.

In the following a brief summary of the test objectives, initial and boundary conditions, test results and major findings is given.

25-9

Test No. 12:

The objectives of this test were to study the phenomena at the tie plate and in the upper plenum with hot leg ECC injection by:

- Detecting water break through events and steam up flow regions at the tie plate for steam only core simulator injection.

- Determining the ef fective water delivery through the tie plate into the core region.

- Tracing the water pool formation above the tie plate.

The initial and boundary conditions for the test were as follows:

- Initial test vessel pressure 4.5 bar

- All loops blocked at pump simulators, hot leg break valve opened to the containment.

- An initial core simulator (= CS) steam mass flow rate of 284 kg/s was injected (which is much higher than expected for the refill- and reflood phase of a PWR) and ramped down to 156 kg/s during the test.

- About 57 s after start of CS steam injection ECC injection (T = 35 C) was initiated into the three intact loops wikb 400 kg/s/ leg.

The main findings are:

- Immediately af ter start of ECC injection subcooled water reached the tie plate and broke through into the core region.

- The break through areas established in front of the injection loops did not change their location.

- Around the down-flow channels there are locally restricted transition zones showing intermediate up - or downward flows.

- Dependent on the ECC mass flow rates water pools with different water inventories were established on the tie plate during the test runs; a lateral void distribution in the pool could be observed.

Test No. 1:

The objectives of this test were:

i

- To investigate phenomena which occur in the downcomer of a pressurized water reactor as a results of high pressure coolant injection into the cold leg at a time when the reactor coolant system is at elevated temperature.

This mixing relates to the overall reactor safety problem of pressurized thermal shock (PTS).

- Special attention has been given to the mixing in the cold leg, the mixing and propagation of a density front in the downcomer and the overall cooldown process in the test vessel.

25-10

. _ _ _ _ _ _ _ _ _ .a

The initial and boundary conditions for this test.were as follows:

- Test vessel and loop nozzles as well as pump seals completely filled with water.

- Test vessel pressure during the test 18 bar.

- Initial fluid temperature 190 C (kept subcooled).

- Break valves closed and pump simulators blocked.

- ECC water mass flow rates injected into cold leg of loop no. 2 were 40, 20, 10 and 5 kg/s (TECC = 30ac).

The main findings are:

For the cold leg region:

- In the isolated volume between injection port and blocked pump simulator a homogeneous cool down could be observed.

- On the path between port and downcomer the stratification of fluid temperature and mixing was measured.

For the downcomer region:

- In the test with the highest ECC-water injection (40 kg/s) the degradation of the fluid temperature difference between the minimum cold water plume temperature and the temperature of the ambient fluid at the entrance and near the bottom of the downcomer was fromeT = 70 K to tLT = 9 K.

- This fast degradation, confirmed by measurements, indicates a good circumferential mixing.

Test No. 11:

The objectives of this test were to study countercurrent flow limitation phenomena in a PWR hot leg for steam and saturated water under hydraulic conditions related to Small Break LOCA and Reflood.

The initial and boundary conditions for this test were as follows:

- Test vessel pressure 15 bar and 3 bar.

- Saturated water mass flow into steam generator simulator inlet chamber 30 kg/s

- Steam injection into the hot leg via the core simulator (stepwise).

The main findings are:

- There is a fairly good agreement between data gained in the test and a correlation proposed by Richter and Wallis based on a small scale hot leg experiment.

- The data of this experiment indicate that a pressure scaling based on the Wallis Parameter J' (modified Froud number) is appropriate.

25-11

ANALYSIS RESULTS FROM THE LOS ALAMOS 2D/3D PROGRAM

• by B. E. Boyack and M. W. Cappiello Safety Code Development Group Energy Division Los Alamos National Laboratory Los Alamos, New Mexico 87545 The 2D/3D program is sponsored jointly by Japan. the Federal Republic of Germany, and the United States (US) The safety-related objectives of the 2D/?D program are as follows: first. to provide an improved understanding of the effectiveness of various emergency core-cooling (ECC) systems in limiting peak fuel-rod cladding temperatures during vessel refill and corc reflood for medium- to large-break loss-of-coolant accidents (LOCAs) in pressurized water reactors (PWRs); second, to reveal core coolant inventory and system flow characteristics during the ref ll and reflood phases of a medium- to large-break LOCA: third, to study convective flow and temperature distributions inside a heated core during reflood for a medium- to large-break LOCA: fourth, to assess the predictive capability of best estimate computer codes and the conservatisms of evaluation model computer codes; and fifth, to obtain information that may be used to improve thermo-hydrodynamic models in best estimate, evaluation-model and other computer codes.

Activities conducted at Los Alamos National Laboratory in support of 2D/3D program goalsinclude analysis support of facility design, construction, and operation: provision of boundary and initial conditions for test facility operations based on analysis of PWRs: performance of pretest and posttest predictions and analysis: and use of experimental results to validate and assess the single and multidimensional nonequilibrium features in the Transient Reactor Analysis Code (TRAC).

Three experimental facilities provide data to 2D/3D pogram participants. The Cylindrical Core Test Facility (CCTF) is an approxHately 1/21-scale facility located in Japan: this facility has completed its test program and the Los Alamos counterpart analysis program is nearing completion The Slab Core Test Facility or SCTF is a separate-effects reflood facility, also located in Japan. A full-height 1/21-scale section of the core one fuel element wide from core centerline to outer periphery. is modeled. This facility began testing its third electrically heated core during 1986. Los Alamos will complete analysis of Core Il tests in 1986 and will begin analysis of Core Ill tests in FY-1987. The Upper Plenum Test Facility (UPTF). located in the Federal Republic of Germany.

is a 1/1-scale integral test facility focusing on phenomena in the downcomer, lower plenum, upper plenum, and primary system loops of a PWR. Los Alamos analytical efforts to date have largely supported test design and specification, posttest analyses of UPTF experiments will be started as soon as data beccme available. In the following paragraphs we provide a brief synopsis of the significant efforts and accomplishments during FY-1986.

CCTF Analysis. We issued our summary report to document results from TRAC PD2 analyses of the Core 1 test series. TRAC was able to consistently predict peak cladding temperatures that were in moderate agreement with the measured values. We def ne mcderate agreement to mean that TRAC correctly predicted the major trends and phenomena that are dominant in determining the peak cladding temperature. Although TRAC did not predict all measured quantities usiformly well,it was able to properly characterize the parametric impact on peak cladding temperature and quench-front propagation displayed by the experimental test series.

We completed our analysis of hve Core ll upper. plenum injection (UPI) experiments and prepared a draft report summarizing our findings. With respect to UPI test phenomena. we found channeling of ECC fluid into the core from the upper plenum which causes a nonuniform cooling of the core. With respect to our analysis

, of the TRAC predictions of the UPI tests we found that for conditions expected in PWRs. TRAC was able to predict asymmetric downflow of liquid into the core. The TRAC predictions were in modcrate agreement with measured peak cladding temperatures for one UPI test conducted with off-normal (high power. Iow UPI t;ow)

• Work performed under the auspices of the US Nuclear Regulatory Commission.

25-13

conditions, insufficient agreement between test and predictions was obtained Additional analytical studies of this test are in progress. We also completed our analysis of Run 71. the single best-estimate test run in the CCTF. Significant phenomena observed in this test included long period hydraulic oscillations in the core that caused overcooling in the lower core and dryout and reheat of the upper Icvels of the core. TRAC also predicted core oscillations but with a shorter period and smaller amplitude TRAC predictions of average peak cladding temperatures, heater rod behavior in the lower levels of the core, and overallloop pressure drop were in moderate agreement with measurements SCTF Analysis. We prepared our summary rcport to document results from TRAC-PD2 analyses of the Core-1 test series This test scries showed that the impacts of radial power peaking and ~ hot channel" factors are mitigated by multidimensional hydrodynamics. resulting in a

• chimney" cooling effect in the high-power bundles.

For this series. TRAC PD2 was able to accurately predict both the hydrodynar., e and heat-transfer phenomena during the cort reflood phase in particular. there was moderate agreement betwecn the predicted and measured peak cladding temperatures under blind. posttest conditions for a spectrum of parametric experimental tests.

We have completed two operational studies using TRAC-PF1/ MOD 1 in support of the initial phase of SCTF Core til testing Finally. we have developed and applied stand-alone models of (a) the SCTF upper plenum and (b) the SCTF hot leg and separator. These models have allowed us to isolate and focus on both input model and code features that are obscured in a full facility model.

UPTF Analysis. During this period, we continued to provide support analysis for the planning of UPTF tests. Although actual tests have been completed in the f acility. the test data are not yet available and therefore.

posttest analyses have not been attempted. Work also continued on improving the input decks for the facility.

Improvements include the modeling of the subcooled injection system recently added to the facility and the addition of the nitrogen injection systems to the intact loop ECC-injection nozzles.

Analysis of the possible loads on the instruments in the facility is receiving special attention. In light of the recent damage to the drag disks in the loops and the degradation of many of the instrument signals, increased efforts are being made to avoid further potential damage. In working closely with consultants from MPR. we have identified the potential for damage to the downcomer level detectors in the upcoming Downcomer Separate Effects Test (DC-SET) tests. Possible alternatives are being assessed to mitigate this potential.

TRAC Improvement. We have either completed or are currently working on three areas of modelimprove-ment. First. we have added a CCFL model to the code. Thus far, wt are seeing significant improvements in the comparison to tie-plate flooding data for the saturated injection case. We are in the process of assessing performance for the subcooled injection case Second. we have added the capability to connect more than one PIPE or TEE to one VESSEL cell. This multiple source connection capability provides a marked improvement in modeling capability. Finally. we are in the process of adding a separator model that willinclude liquid carryover and vapor carryunder as a function of mixture flow rate and quality. The user will provide these data to the code

,na input.

Assessment of TRAC predictive capabilities in the CCTF and UPTF analysis series has resulted in the definition of areas in which code improvements are needed. The most significant model deficiency identified was that TRAC did not correctly predict liquid distribution in the core. This dehtiency is particularly important because liquid entrained in the core first impacts heat transfer above the quench front. and subsequently affects upper plenum liquid accumulation. steam binding in the steam generators, and loop component pressure losses.

Work is currently undcrway to improve the TRAC core void fraction package. Work is also needed to improve multidimensional hydrodynamics in the upper plenum including de-entrainment and carryover to the hot legs.

A third area of needed improvement is related to condensation effects resulting from the injection of highly subcooled ECC watcr into the primary In summary. we believe that Los Alamos is functioning as a vital participant within the 2D/3D Coordination Program The results of our analytical efforts are being used to support test specification and design, improve understanding of phenomena occurring in tests, assess the predictive capability of TRAC, and identify needed areas of code improsement.

25-14

_ _ _ _ _ _ _ _ _ _ _ _ _ u

MULTIDIMENSIONAL REPRESENTATION OF GPWR PRIMARY SYSTEM IN 200 % BREAK LOCA CALCULATION B. Riegel Gesellschaft fur Reaktorsicherheit (GRS)

H. Plank Kraftwerk Union (KWU)

K. Liesch Gesellschaft fur Reaktorsicherheit (GRS)

The 3D-Vessel capability of TRAC-PF1/ Mod 1 (12.5) was used to study the influence of combined cold and hot leg ECC injection on flow and tem-perature distributions in the upper plenum as well as on the water break thru from the upper plenum to the core region and finally on the core cooling.

2 LOCA calculations have been performed. Based on a German Pressurized Water Reactor (GPWR) system a TRAC-PF1 input deck was designed to simulate the pressure vessel by a three-dimensional component of 15 axial levels, 4 radial rings and 8 azimuthal segments (see flg.1) the 4 loops seperately by one dimensional components including a fine noding of the steam generators and the main coolant pumps

- the complete ECC-injection system (cold and hot leg injection) consisting of high and low pressure pumps (HP/LP) and accumulators (ACCU).

The same nodalization (ab ut 1160 computational cells) was used in both calculations.

I Paper prepared for the l 14th Water Reactor Safety information Meeting October 27-31, 1986, Gaithersburg, Maryland 25-15

The only differences are: (see fig.1)

1. Calculation HP/LP/ ACCU injection into hot legs of loop 1 and pressurizer loop 2.

100 % power.

2. Calculation HP/LP/ ACCU injection into hot legs of loop 1 and loop 3.106 % power.

All other initial and boundary conditions are identical.

A time sequence of major events is tabled below:

1. calc 2. cale pressurizer empty 16.5 s 18.0 s accu injection started 22.5 22.0 LP injection started 36.0 38.0 first av. rod quenched 17.5 (cell 15) 38.2 (cell 16) begin of reflood 50.0 55.0 all av. rods quenched 88.0 107.0 It can be also seen in figure 2 which shows the envelope of the maximum rod temperature in cells 1-16 is quite similar in both cases up to 55 s after break initiation neglecting the small difference in the temperature values of the 2. calculation which may come from the initial 6 % over-power. A major difference after 55 s indicates that the thermohydraulic behaviour in the core region changed significantly.

The discharge flow from the pressurizer into cell 15 causes a substantial precooling within this area. In the 1. calculation a quenching of the average rod in cell 15 at 17.5 s is the consequence. When accumulator injection starts a cooled channel is already prepared to drain down the ECC-water to the lower plenum. The average rod in the adjacent cell 16 quenches down shortly after accumulator injection started (22.5 s) at about 28 s. Average rods in cells 7 and C are the next which are 25-16 w

_ _ _ _ _ _ _ _ _ _ _ _ m

quenched. The code calculates a major continuous break thru of ECC water in the vicinity of the main coolant pipe nozzles (hot tog loop 1 and 2); this is in excellent agreement with experimental results from UPTF test 12, a subcooled tie plate CCF test, which showed that the break thru areas are near by the hot leg nozzles.

In the calculation the steam which is generated by removal of the stored energy and decay heat, flows upwards (cells 3, 4, 11, 19) to the hot leg of the broken loop.

At about 55 seconds of the transient the reflood phase begins. The substantial precooling (i.e. partially quenched rods, reduction o'f steam superheat) of about 30 % of the core region leads to a further cool down of the remainder part of the core and finally to a complete quenching of all average rods at 88.0 seconds.

In the 2. calculation the pressurizer effect is not further supported because the HP/LP/ ACCU-injection is connected to loop 3. This results in a later quenching of the average rod in cell 16. At about 50 s the average rods in cells 11 and 12 are quenched. Also in this case the break thru areas stay stable in the vicinity of the hot leg nozzle in which ECC water is injected. This is also in excellent agreement with test resdits (UPTF-test 12). The ECC injection into loop 3 hot leg next to loop 4 tends to restrict the outflow of steam via the broken hot leg. More stored energy (106 % power) and less precooling at this time causes the clad temperatures not to decrease as fast as in the previous case. A steam upflow in cells 5, 6, 7 is established (chimney effect) which results in a delayed quetiching of the rods in this cells.

A first confirmation of establishing " flow channels" between core and upper plenum for water down- and steam upflow is given by experimental results of UPTF-tests.

25-17

1. CALCULATION 2. CALCUL ATION HP/LP/ ACCU HP/LP/ ACCU HP/LP/ ACCU h Pressurtzer b'.

?

.y-  ;

..c.

i ./. Pr.

/ -.

or52.r f.-

..<,, .. .43,

,.**"**~~. (pjg(( 2 ~~,* . 23 25 's , 25 LP/ACCUg,,j . ." 16 ,.i ,6 .

g LP/ ACCU ,

7 *j' [,,,'.( \=\g "'I

. . . ..., .s 22 7 e f. .,/ 'l ',,..

7 , ' N.s. \22 m V

. i., .. .

5 -( '\ 'I [ . / '*/ 21 k .

), e .* \ .- 3 i'3 f2 ,(

s '.25 'x.'. '; .3 ;f,,X 2'/'@ tP/ ACCU gk.2. ..-]...,,..g.,,.y

' 2,/% t,yn

.a.. .2 . .. .

. . /

o. ! m.2,y2s

, .,j ,.. i.

m.. g. .

.,2s,

..a, ,

.? *. ,v.

el g

i 6 -

HP/LP HP/LP/ ACCU HP/LP FIG. 1: ECC-INJECTION MODES GPWR 200*/. CLB, 5/8 ECC (4/86)

LEGE'ND a - 1. CALCULATION q

8- o = 2. CALCULATION 9

2-Un x

o -

6--.

e x

E9 ug t

g m o

Oo x -

es 8-C x9 y c.

g g_

N 9

0' 9 -

a L

6b.0 7b.0 85.0 95.0 1d0.0 11'0.0 1d0.0 130.0 0.0 lb.0 2d.0 3b.0 ib.0 Sb.0 TIME (S)

FIG 2: VESSEL CLOBRL DRTA 25-18

ll

,+

A SAFE ENVIRONMENT FOR DEMONSTRATING ADVANCED REACTORS by James L. Dooley & R. Philip Hammond (Consulting Engineers. TELEPHOS, Santa Monica, CA)

Advanced reactor concepts cenerally have improved safety as one of their objectives, for if the public is to accept any nuclear power at all, the perceived level of sirety must be improved. However, another coal is equally important--improved economics and reliability. Unless this goal is also achieved, an ultra-safe reactor will remain an " academic" reactor--

still on paper, for no one will pay for developing and commercializing it.

Unfortunately, achieving inherent safety often requires sacrifices in power density, fuel burnup, ease of construction, load following capability, or other characteristics needed for low cost. Thus the two goals tend to be mutually exclusive.

If 2t comes to a choice between these conflictino objectives, which should covern--inherent safety or economics?

The fact is that there as really no choice at all. Any new, untried reactor will automatically be classified as a hazardous experiment until the safety features claimed for it have been satisfactorily demonstrated in a working installation. The public has heard promises before that all contingencies have been thought out and provided for, only to find that a fatal flaw was still lurking undiscovered. Even the nuclear community has become skeptical of untested new desians, and tends to favor the older ones that have been tried out and worked over repeatedly. One expects that these old deslans have fewer surprises left in them, and it is human nature to prefer the demon one knows than an unknown.

Even the safest reactor in the world will not be able to ao beyond a paper design unless it can be built and tested in an environment where unexpected failures will have no way of affecting public safety and health. This paper describes such an environment--one that gives any reactor, new or old, a secure operating containment from which even a severe core-melt accident or hydrogen fire cannot threaten or injure the public.

This special environment is provided by s. rather ordinary containment shell equipped with a new addition--an immense condensing filter. The containment is vented continuously into the filter, insuring that the internal pressure stays close to atmospheric. The filter is capable of condensing, adsorbing and retaining all the heat, particulates, vapors, and radioactive gases (including krypton, xenon and iodine) that might be released in the worst unforseen event.

Such a containment makes it physically impossible for the containment to f ail structurally or leak significantly, because it can have no sustained internal pressure. Once it is in place no radioactive materials can be I discharged into the atmosphere. Moreover its performance is simple, direct, easily demonstrable, and thus highly credible.

27-1

With a leakproof containment the unproven nature of the reactor is not a barrier to public acceptance, because the reactor system is completely decoupled from the external environment, even though a severe internal accident should destroy the reactor. Having the means for assuring absolute public safety, especially for the developmental and prototype versions, the designer of an advanced reactor concept can concentrate on the essential operating properties of the system--economics, reliability, inherent safety, or whatever, without having to prove in ablance that every contingency of risk to the public has been removed--a requirement that is impossible to fulfill. Therefore, the prototype reactor would not have to be encumbered with layers of safety devices to protect the public against hypothetical accidents. Such devices are not only costly, but they can interfere with the normal operation of the plant, or even cause abnormal behavior by their own failures.

The concept of a filtered vent to protect a containment structure is not new, and the low-temperature trapping technology of the filter system extends back nearly 50 years. Although not widely known in reactor safety circles, the technology is highly developed, well proven, and thoroughly documented. Every U. S. reactor makes use of it in the off-gas recovery system, but it has never been applied at such a large scale as we propose now, where it must have sufficient flow capacity and condensing power to accept a full discharge of primary system contents over a short time.

The work presented on this concept formed part of an intensive study by NRC of severe accidents and means for mitigating their effects. The specific contract included categorizing the modes of failure of U. S. containment types in severe accidents, and the design of specific systems for preventing such failures. The chill-vent filter system was the most versatile, least costly, and most successful of the systems studied.

27-2

An Examination of the Bases for Proposed innovations in Reactor Safety Technology Day Id L. Mases Oak Ridge National Laboratory During fiscal years 1984 and 1985, a group at the Oak Ridge National Laboratory performed a program of work entitled the Nuclear Power Options Viabli Ity Study (NPOVS). The NPOVS sought to characterize the research and devel opment needs essential for the successf ul deployment of advanced reactor concepts early in the next century, by which time the need for new baseload generating capacity is likely to be evident. To limit the study'to a tractable scope, the NPOYS examined only those concepts claimed by their proponents to emphasize passive safety features in the design. For the purpose of the NPOVS, " passive safety" was defined as the reliance on natural physical l aws and properties of materials to ef fect shutdown and radioactive decay heat removal without fuel damage and without relying on mechanically or el ectrical ly actuated and driven dev ices such as those employed in active (engineered) systems. To assess and evaluate each concept's v iabi l Ity with respect to marketability and i Icensabil Ity, the NR)VS derived and employed seven quantifiable criteria and a nunber of less easily quantifled essential and desirable characteristics.

Although the NR)VS was purposely limited to advanced reactor concepts which onph asized passive saf ety, it was f ut ly understood that passive saf ety is not en end in itsel f but rather en approach that some proponents belleve can be used to salvage the nuclear power option. Currently, there appear to be at least four distinct bases upon which different proponents approach i nnovation in reactor saf ety technologies. These bases for Innovation can be characterized as foilows:

1. The fIrst approach, which has been espoused by Weinberg, Crane, Hannerz, and others, is that a major, if not always radical or revolutionary, change is needed in reactor design. This change invol ves almost excl usive rellance on passive safety features in order to preclude or signIf Icantiy deIay core damage. WIthout core damage, the publIc health and safety is always assured. With signif icant increases in the time available before core damage, public health and saf ety is more easily and conv incingly accmmodated. " Licensing by test" is of ten espoused by the proponents of this approach as a means to convince the publIc and the regul ator s of inherent pl ant saf ety, in general, this approach is reflected in the Swedish Process inherent Ultimate Safety (PlVS) light water-cooled reactor and to a lesser extent in the General Electric small boil ing water reactor. Both concepts sacrif ice the econanics of very high power rati ngs for the abil ity to prevent or del ay core damage and for the potential cost savings fran reduced requirements for engineered saf ety devices.

l Research sponsored by U.S. DOE Office of the Assistant Secretary for Nuclear Energy under Contract #DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc.

27-3

2. The second approach is that adopted by the major U.S. reactor vendors in an alliance with those of Japan. This approach to designing advancea light water reactors attempts to focus Innovation within the context of accumul ated experience. The objectives are to simplify the design and demonstrate safety probabilistically by caref ully redesigning the reactor tc reduce the total core mel t f requency to about 10-6 per reactor year.

Elements of passive safe'/y features are incl uded but are Ilmited since desi gn evol ution rather than radical change is believed to be most cost effective. This approach also seeks to maintain the perceived economy of scale of high power ratings.

3. The third approach is one based on the contention that current, reactors are adeq uately . designed f or both economy and saf ety but that the publ ic concern about perceived safety hazards can be overcome best by innovative desi gn of the contai nment structures to ach ieve an ultresafe conf igurati on. Ultrasaf ety is to be ach ieved by the use of core retention dev ices below ground level and of a canbination of passive or active containment coolers. In some ca ses, proponents h ave proposed making exi sti ng pl ants ul trasaf e with backfitted enhancements to the contai nment.

4. Finally, there is an aspect of innovation that is conmon to some extent to al l of the above approaches to enhanced reactor saf ety. This approach seeks to focus the attention of licensing to one or at most only a few sy stems and com ponents which assure the public health and saf ety. The idea is to danonstrate that the safety of the design is not contingent upon mul ti pl e, independent and redundant def ense-in-depth features but rather depends only upon one or a few features which can be rel ied upon exclusively or with a very high probabilistic confidence. In particular, this approach seeks to simplify licensing by separating out the safety features from those which are argued to represent "i nvestment protection."

This paper anploys the NPOVS criteria to examine the current bases for proposed Innovations in l ight water reactor saf ety technology. The bases f or such innovations are also reviewed against the licensing history of the Fort St. Vrain HTG R, which was the first reactor to attempt the use of " Inherent saf ety" arguments in l icensi ng documentation. Special attention is given to the potential for problems which may be incurred in " licensing by test" and in separating out "investnent protection" features f rom the traditional defense-In-depth approach to licensing.

27-4

A RECONFIGURED PWR WITH ULTRA-SAFE CHARACTERISTICS by M.A. Schultz The Pennsylvania State University An ultra - safe reactor is defined as one that con sit in piece safely and indefinitely with a complete loss of power. A PWR using conventional components con be reconfigured to have this property. In addition the normal method of shutdown con be simply turning off the power to key components.

The primorg loop of the reconfigured system is conventional except for the pressurizer. Any existing design reactor con be used. Added safety con be provided by using a BWR core in a PWR vessel with some loss in specific power. The method of pressurization is via a high reverse leakage charging pump that uses a Francis type impellor, (similar to those used in pumped storage facilities ) thus enebling the device to run in the forward direction as a pump and in the reverse direction as a turbine. The pressurizer pump inlet water supply is from en atmospheric pressure tank located inside containment and above the reactor in elevation. There are no power operated relief valves (PORVs)in the system.

Normel operation of the primary loop is conventional and essentially at constant pressure. If the system pressure drops, the pump quickly recharges to operating pressure. If the system pressure rises, this pressure is relieved via the reverse leakege of the pump impellor. The high pressure injection system (HPI) pumps are fed from the atmospheric pressure tank and can be turned on and lef t on indefinitely without any danger of the system overflowing.

The primary system decay heat at shutdown is transferred to the secondary system by natural convection and water levels and piping are possively set so as not to block full natural circulation. The secondary system contains a dedicated illi tank and possive heat exchanger to enable full removal of decay heat without power. A suggested possive heat exchanger is the " earth condenser", a buried pipe network located approximately 1 f t. underground spread over roughly 30 acres of the exclusion area. All conventional seconderg components are present including roughly full steem dumping copobility. The dedicated pessive shutdown heat removal system is connected into the normal secondary system by loss of power to two diverse volve clusters.

In normal shutdown or failure of power to the pressurizer pump, high r

' pressure high temperature water is ejected from the primary system to the etmospheric tank via en orfice and the Francis impellor, which now becomes o l

27-5

turbine. The pump motor becomes a generator, and with possive diode connections, this generator now supplies power to en HPl pump ogcin using the atmospheric tank as a water source. As the enthalpy of the exiting hot water is much greater than the cold water of the atmospheric tank,it becomes possible for a while to pump back into the primary system as much or more water then was lost in the blow-off without external power.

Preliminary calculations indicate that the amount of water ejected through a 1 in. diameter orfice in blow-off for I hour would be roughly the some as that lost in the TMI accident through the PORY in 3 minutes. And if the HPl turbine generator pump described above were present, the some amount of water would not be lost in 7 hours8.101852e-5 days <br />0.00194 hours <br />1.157407e-5 weeks <br />2.6635e-6 months <br /> of ter shutdown. During this 7 hours8.101852e-5 days <br />0.00194 hours <br />1.157407e-5 weeks <br />2.6635e-6 months <br /> the primary system is cooling off by water mixing with the atmospheric tank water. Pressure is dropping because of specific density change end because the HPl pump involved cannot get full power input from the cooler primary blow-off water. At the end of roughly 7.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> ef ter shutdown, the pressure has dropped to the gravity induced pressure of the atmospheric tank. The primary system now floods completely by gravity and remains in this state indefinitely.

On shutdown, the power to the secondary system is turned off; the system depressurizes; the volve clusters open; natural circulation transfers the decoy heat to the seconderg fill tank and possive heat exchanger, and this device transfers the heat to the ground or atmosphere for as long as necessary.

The atmospheric tank provides anothr benefit in that it enables full on-line clean-up of radioactive gases and liquids at atmospheric pressure rather then primary system pressure. An overflow tenk, located at the lowest elevation in the working plant,is provided with pumps and filters to take water back to the atmospheric tank. Any spilt water from pipe leeks or breaks or even a pressure vessel leck con be returned to the primary system via gravity to the overflow tank and then pumped back to the atmosphedc tonk.

The safety of the plant is improved two ways. First, anotherlayerin meltdown protection is provided by the ability of the plant to sit in place without power. Secondig, preliminary failure mode and effects onelysis (FMEA) indicates that for f ailures of most primary system components or pipe breaks, the operator response is always the same--simply turn on the HPl pumps (electric or diesel) and leave them on. In this way the primary system water always goes around in a circle from the primary system to the atmospheric tank and back. Operator trotning and intuitive response is thus simplified.

Such on ultro-sofe plant con be built out of conventional components that are service proven. Safety could be dramatically demonstrated by turning off the power, setting up the possibility of greater public acceptance.

27-6

AUTOMATION AND ARTIFICIAL INTELLIGENCE FOR INCREASED SAFETY William G. Kennedy United States Nuclear Regulatory Commission Office of Nuclear Reactor Regulation Division of Human Factors Technology Washington, DC 20555 This is not a discussion of requirements for safety but one of possibilities for safety. Because computers are becoming commonplace and artificial intelligence (A1) is in the news, the nuclear power industry may be ready to seriously apply these new technologies to reduce operating costs, improve performance, and increase safety.

Automation is the transition, from humans to machines, of the sources of power, control, and decision-making to perform certain function. It is more of an evolutionary process than an overnight revolution. This evolutionary process can be described as 5 major levels. Starting at the low end of automation, the first level is no automation at all; all actions are powered and controlled directly by people. The next level is to replace human power with machine power but with the power still under direct local control of the operator. The third level is to replace the direct, local human control with indirect, somewhat remote control, with all decision-making still with the human operator. The next level is turning over some of the decision-making to the machine. This allows the machine to perform preplanned sequences without human intervention during the sequences. It is commonly called

" semi-automatic" operations. The last level is full automation, where the machine provides the power, control and decision-making.

The commercial nuclear power industry is generally at the third level of automation with some second level activities. We see complex control rooms with many operations being performed locally by auxillary operators. To compound the control problem of these complicated machines, the operators face a very large number of inputs describing the machine's current condition and a very large number of controls with which to manage the machine. This situation continually results in a relatively large number of minor errors.

These are tolerated because the corrections are apparently easy: increasing the number of procedures or providing more training. Although human control of nuclear power plants is considered safe enough, the title of this conference indicates we believe there is some room for improvement.

Other machines that have evolved, and the industries associated with their manufacture and use, have faced similar problems.

Consider the task of getting to the second or higher floor of a building. We could walk (1st level of automation, none). We could use an elavator that has a human operator who controls the movement and stopping at the desired floor (second to third level of automation). At the fourth level of automation is the common semi-automatic operation of current elevators. At the fully automatic fifth level, we use escallators that automatically move from one floor to another without human decision-making or controls.

27-7

l As another example, the first commercially available automobiles were seen as advanced technology in their day. They required " engineers" to operate them.

Controls included " spark advance", a hand brake, a " choke", and skill to change gears that were not synchronized. It took almost 50 years, but now manual chokes are things of the past. We have power brakes and power steering and automatic transmissions, and who knows what a spark advance was for?

The experience of NASA the evolution of automaton associated with launch control is of particular interest because of its similarity to the control function of the nuclear industry. It was about 30 years ago that our space program was controlled by engineers operating controls that remotely operate individual valves and breakers. This has evolved into engineers controlling the functions of systems by initiating sequences of what were individual operator actions and, in some cases, equipment failures of components are adjusted fer automatically. NASA is experimenting with applying AI to automatically detect equipment failures and find ways to work around those failures. Automation at NASA was initially justified as a technique to contol the manpower costs as the launch control process became more complicated. What happened was that there has been a significant improvement in performance and safety through reduction of the voluminous, low-level demands on the operators which allowed them to concentrate on the higher-level purpose of the controls.

The commercial nuclear power industry is also experimenting automation and AI. The Electric Power Research Institute (EPRI) has demonstration projects in the area automating recognition of anticipated transients without a scram (ATWS) for boiling water reactors (BWRs). The reactor vendors have efforts started in automation or AI and other firms, such as Technology Applications, Inc. and the Management Analysis Company, are offering services to provide automated aids to operations.

Although AI is receiving a lot of attention, the construction of intelligent machines requires that the decision-making process be well understood. Since there are many aspects of human decision-making that are not yet well understood, it is probably too early to fully automate significant portions of the decision-making in control of nuclear power plants although tools to aid operators are possible.

While automating the operators' decision-making (the fifth level) is some time off, progress from the other end of the evolutionary spectrum, i.e.,

automating the low-level operator actions may be a good place to start.

Nuclear operators are working at a very low le;el of automation when they must turn a series of switches to start routinely a system or switch the operating pump. NASA has demonstrated the cost benefit of moving up the evolutionary scale with better performance and therefore safety. In the nuclear power industry, this approach could significantly increase the safety associated with the current use of human operators.

27-8

Advanced Light Water Reactors for the Nineties Frank A. Ross ALWR Program Manager Office of LWR Safety and Technology U.S. Department of Energy William R. Sugnet ALWR Program Manager Electric Power Research Institute Cooperative Advanced Light Water Reactor (ALWR) programs of the Electric Power Research Institute (EPRI) and the U.S. Department of Energy (DOE) over the next three to four years arc directed toward providing a viable light water reactor (LWR) option in the mid-1990's to meet electrical load growth and replacement capacity. requirements. The principal objectives will be to achieve improvements and simplifications to enhance safety and reliability, facilitate licensing, and reduce construction duration and cost. To effectively meet the objectives, these programs will be directly responsive to the needs of the electric utility industry, who sponsor the EPRI program, and closely coordinated with the Nuclear Regulatory Commission (NRC), where appropriate, in areas involving significant safety issues.

The programs include the development of mid-size (about 600 MWe) PWR and BWR plant concepts that offer a favorable prospective for systems simplifications, significant improvements in construction, and advantageous economics for the utilities. These activities will investigate simplified systems configurations, including the application of more passive safety systems and design features. Plant arrangements and construction studies will be conducted to achieve improved constructability, such as through the increased use of prefabrication and modularization. Proof-of-principle or qualification tests will be performed on selected key systems features and components to verify the performance and practicality of selected concepts.

A fundamental basis of the EPRI program is the ALWR Utility Requirements Document, that will include an extensive compilation of the utility requirements for design, construction, and performance of ALWR plants for the 1990's and beyond. This document will emphasize the resolution of significant problems experienced at existing nuclear plants. The Requirements Document chapters are being reviewed by the NRC as they are developed to assure that the safety-related requirements and criteria are being appropriately addressed.

A most important part of the DOE program includes support for the design verification, leading to design approval and certification of two large (nominally 1300 MWe) reactor plant designs: a General Electric ABWR and a Combustion Engineering APWR. This activity will be closely coordinated

, with development of the EPRI Requirements Document to assure these requirements are appropriately reflected.

27-9

The DOE program also include substantive studies and industry workshops directed toward achieving significant improvements in LWR construction.

This work will consider past and current problems experienced by the industry, and will address all aspects of construction, including construction management, project control, construction technology, and plant design to facilitate construction.

Finally, the DOE program will include significant studies of advanced instrumentation and control, in the areas of integrated plant control, data communication, control system performance, and advanced sensor applications.

The EPRI and DOE believe that these programs will significantly contribute to a revitalization of the LWR to fulfill its role of providing a safe, reliable, and economic option to meet electric power needs in the nineties.

27-10

The Impact of Passive Safety Requirements on Plant Design Francis X. Gavigan J. Ross Humphreys Andrew C. Millunzi U.S. Department of Energy Over the past 2 years, the Department of Energy has been actively involved with contractor design organizations throughout the United States to develop modular Liquid Metal Reactors and High Temperature Gas Reactors that reflect increased passive safety, decreased costs, and high reliability. The emphasis on passive safety has led to a number of beneficial outcomes in the form of reduced costs, different safety philosophies, simpler plant designs, lower operating costs, and increased safety margins. Features such as air cooling for removal of decay heat, magnetic curie switches to cause rod drop, core analysis that produces inherently safe cores, and others have resulted in a reexamination of the need for safety-grade control rooms and a move away from reliance on redundancy and diversity in engineered safety systems toward heat removal and reactivity control through the action of passive devices and inherent characteristics. The result to date has been reactors that are reduced in risk and in cost, which are simpler to operate and simpler to maintain and that may well meet the Environmental Protection Agency's Protection Action Guides inside the site boundary thereby introducing the possibility of eliminating the need for evacuation planning.

I 27-11

THE PASSIVE CONTAINMENT SYSTEM (PCS-2) by O. B. Falls, Jr. and F. W. Kleimola NucleDyne Engineering Corporation PCS-2 enables a design for a zero source term; i.e. , nct release of radio-activity from containment under any accident conditions. PCS-2 provides the needed engineered safety improvements for advanced light-water reactors as described herein for a four-loop pressurized water reactor.

PCS-2 innovative engineered safety features consist of structures, sys-tems and components. The structures encompass three basic containment systems constructed as one integrated structure on the reactor building foundation:

the Primary Reactor Containment System (PRCS) , the Auxiliary Containment Sys-tem and the Secondary Centainment System.

The PRCS encloses the reactor ecolant system (RCS) and engineered safety system components. The PRCS prevents the uncontrolled release of radioactiv-ity to the environment from the postulated accidents (i.e. pipe breaks in the RCS and secondary system) and anticipated events, including anticipated trans-ients without scram and station blackout. The stored water within the PRCS exceeds one million gallons. The heat sink capacity of the 50F stored water is over 13 million Btu. This heat sink capacity is more than 3.5 times that required for the postulated LOCAs. The Auxiliary Containment System prevents the uncontrolled release of radioactivity to the environment from postulated accidents outside the PRCS. These include accidents postulated for the reactor auxiliaries, air handling, stored radioactive waste and fuel handling and stor-age systems. The Secondary Containment System (i.e. the reactor building) pro-vides a second barrier for all accidents postulated.

For the loss-of-coolant accident (LOCA) three sources of emergency core cooling (ECC) are housed entirely within the PRCS. These consist of depres-surizer vessels and refill and deluge tanks all physically elevated above the RCS piping. In the small break LOCA, the heat sink capacity of the depressur-izer coolant is augmented by steam blowdown from two select steam generators (SG) through DC-actuated relief valves into the adjoining deluge and quench tank heat sinks. The steam blowdown transfers residual and decay heat from the RCS into these tanks. A portion of the steam flow is routed to injectors that entrain water from the adjoining quench tanks to FUpply emergency feed-water for continued decay heat transfer. The depressurizer vessels (maintain-ed at RCS pressure) contain a sufficient inventory of water to prevent steam bubble formation in the reactor vessel during RCS depressurization. With the RCS depressurized to the pressure range in the other two SG secondaries, being retained at pressure, steam flow from the latter two SG through injectors en-trains ECC water from adjoining refill tanks. RCS blowdown to the containment backpressure, passively initiates continued ECC with gravity flow from the de-luge tanks. Flooding of the PRCS to an elevation above the pipe break initiates decay heat removal througn closed loops to a cooling pond for the term of an accident.

27-13

Functioning of the PCS in a major LOCA is not dependent on mechanically or electrically driven components or operator actions. For the small break LOCAs, secondary system pipe breaks, transients and adverse incidents, the only active safety features are select automatically-actuated valves activated by DC-power (storage batteries).

The PRCS consists of interconnected steel cells maintained at a vacuum.

RCS blowdown into the PRCS results in steam flow through variable orifice vents at the deluge and quench tanks. These tanks have sufficient heat sink capacity to reduce the PRCS pressure below atmospheric within 30 seconds after a major pipe break. Passive depressurization of the PRCS prevents the leak-age of radioactivity enabling a design for a zero source term.

The main steam isolation valves (MSIV) and feedwater check valves are po-sitioned immediately outside the PRCS on strong pipes. In the event of a steam or feedwater pipe break, the valves on this piping automatically close, pre-venting the blowdown of potentially radioactive steam to the environment. The closure of the MSIV results in the overpressure steam blowdown of the SG secon-daries into the deluge and quench tanks. The SG safety valves and relief valves are positioned on the steam headers within the PRCS immediately above the quench tanks.

With the SG safety valves and relief valves within the PRCS the concern re-lating to a SG tube rupture is negated. Any overpressure in a SG resulting from a tube rupture is relieved into the adjoining quench and deluge tanks with-out need to relieve potentially radioactive steam to the environment.

PCS-2 prevents the so-called Class 9 accidents; core melt, metal-water re-actions, hydrogen explosions and burning, and containment overpressure. The PRCS pressure is restored in a sub-atmospheric condition passively within a few minutes after an accident depending on the accident-type. There is no need for containment venting or core retention devices.

The design of the PCS-2 is in keeping with the criterion on single-failure, common cause failures, equipment separation, systems interactions and the en-vironmental and seismic qualification of equipment. Redundancy and diversity are even provided for the passive systems and components.

PCS-2 further advances the design of nuclear power plants for the next gen-eration by providing improvements relating to seismic events, wind and tornado loadings, flooding, missiles (internal, external, turbine and natural phenomena generated missiles} Also, improvements are provided relating to fire preven-tion and insider / outsider sabotage.

PCS-2 improves on the plant construction time, significantly reduces the cost of the plants engineered safety features, improves plant availability, reduces personnel exposure in keeping with "as low as reasonably achievable" (ALARA) considerations, improves replacement of RCS and other major components enabling plant recommissioning beyond the present day license period. PCS-2 enhances modular construction and standardization of the engineered safety features for light water reactors.

27-14

g,Po=M= u s Nucuan .Eaut4TO=, cOMMis .ON i R E POa T Nv Mu a ,w~,

• reoC. .d. v., N. . .,,,,

ETYE BIBLIOGRAPHIC DATA SHEET in .NiTRuctioN:oN T E Rivers. NUREG/CP-0081 TITLE AN0 5vtTITLE 3 LgAvgSLANK Yransactions of the Fourteenth Water Reactor Safety Information Meeting 4 DAf g REPORT COMPLETED MONT- WEAR i AuT-ORise September 1986 Conference papers by various authors; compiled by * ^ " " ' 'o" " 55

Allen J. Weiss " ' " "

l October 1986 7 PERFORMtNG ORGANI2ATION NAME AND MasLING ADDRESS tsac8edele Codes 8 PROJECTeT ASK* ORE UNei NUM8tR Office of Nuclear Regulatory Research , , , ,, ,,,

U. S. Nuclear Regulatory Commission Washington, D. C. 20555 A-3283 10 SPONSomeNG ORGAN 12 Af EON NAME ANO mailing ADDRE55 ##acs v erle Codes tie TYPE OF REPORT Transactions of conference Same as Item 7 above. on safety research D PERIOD COV ERED (inc#us.ve derest October 27-31, 1986 12 SUPPLEMtNT ARY NOTES This report contains summaries of papers on reactor safety research to be l

presented at the 14th Water Reactor Safety Information Meeting held at the National Bureau of Standards in Gaithersburg, Maryland, October 27-31, 1986.

The summaries briefly describe the programs and results of nuclear safety research sponsored by the Office of Nuclear Regulatory Research, USNRC. Summaries of invited papers concerning nuclear safety issues are included from the Office of Nuclear Reactor Regulation, USNRC, in addition to summaries of invited papers that cover the highlights of reactor safety research conducted by the Department of Energy (D0E), the electric utilities through the Electric Power Research Industry (EPRI), the nuclear. industry, and the research of government and industry in Europe and Japan. The summaries have been compiled in one report to provide a basis for meaningful discussion and information exchange during the course of the meeting, and are given in the order of their presentation in each session.

le DOCUMtNT AN ALv $i$ - a R E vvwORD5 DE SCR t* TOR 5 16 av A4L A8a Lif v STATEMENT Reactor safety research Nuclear safety research Unlimited 16 SECURITY CLA55aFeCAT:CN E Tro pe,ep

. loENTs.iERs O,E~ E~oiD TiaMs Unclassified iTme repartl Unclassified 17 NVM9tR OS P AGES t g Peqq([

.. _ n

14th WATER REACTOR SAFETY INFORMATION MEETING Octobar 27-31, 1986 Session Schedule

• ED AUDITT)RIUM CREEN AUDITORILH LECTURE ROOM B Mon Flenarr Session AM ECCS Rt.le Revision Nuclear Plant Analyzer Equipment Integral Systems Testing and Code Improvement Qualification Session 1 Session 2 Session 3 Mon Mechanical and PM Integral Systems Testing Structural Research TMI-2 Analyses Session 4 Session 5 Session 6 Tues Severe Accident Separate Effects /Ex-AM Sequence Analysis periments & Analyses Seismic Research Session 7 Session 8 Session 9 Fission Product Release Tues International Code and Transport in Nuclaar Plant Aging PM Assessment Program Containment Session 10 Session 11 Session 12 Materials Engineering Wed International Code Severe Accident Non-Destructive AM Assessment Program Source Term Evaluation Session 13 Session 14 Session 15 Materials Engineering Containment Systems Environmental Effects Risk Analysis /PRA Wed Research/ Containment on Primary System Applications PM Loads Analysis Components Session 16 Session 17 Session 18 Materials Engineering Thur Pressure Vessel Reference Plant Risk Industry Safety AM Research Analysis - NUF.EG-ll50 Research Session 19 Session 20 Session 21 Materials Enginaering Radiation Effects Reference Plant Risk 2D/3D Research Thur Session 22 Analysis - NUREO-1150 PM Degraded Piping Session 23 Session 24 Session 25 Panel Discussicn of Innovative Concepts for Regulatory Issues Increased Safety of Fri (NRR, IE, NMSS) Advanced Power Reactors AM Session 26 Session 27 Plenary Session Concluding Remarks g

POSTER SESSION - WEDNESDAY ( ALL DAY)

- Mobile NDE Van (Parking Lot)

[ - NRC Operations Center Info (Employees' Lounge)

- Technology Resources-ORNL RSIC and TDMC (Emp. Lounge) l

UMlVED STATES speciat rountw-ctass narr MUCLEAR REGULATORY COMMISSION 'oST^$3 l[ES PAto WASHINGTON, D.C. 20555 , was,w, o ej OFFICIAL BUSINESS PENALTY FOR PR:VATE USE,6300 1 1AN19A19819C1 120555078877 US NRC TIOC AOM-DIV 0FPUB MGT BR-PDR NUREG POLICY &

W-501 DC 20555 WASHINGTON C