Text

Transactions of the Thirteenth Water Reactor Safety Research Information Meeting
	ML20205E218
Person / Time
Issue date:	10/31/1985
From:	Weiss A; NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:	;
References
	CON-FIN-A-3283 NUREG-CP-0071, NUREG-CP-71, NUDOCS 8510170230
	Download: ML20205E218 (276)
	v • d • e

-

1 .

NUREG/CP-0071 f

Transactions of the Thirteenth Water Reactor Safety Research Information Meeting To Be Held at National Bureau of Standards Gaithersburg, Maryland October 22-25,1985 U.S. Nuclear Regulatory Commission Office of Nuclear Regulatory Research j%,o$a *' 9 Ogot g 851031 CP-OO71 PDR

t NOTICE These proceedings have been authored by a contractor 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, 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 views expressed in these proceedings are not necessarily those of the U.S. Nuclear Regulatory Commission.

Availab!e 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-0071 Transactions of the Thirteenth Water Reactor Safety Research Information Meeting To Be Held at National Bureau of Standards Grithersburg, Maryland October 22-25,1985 Date Published: October 1985 Compiled by: Alien J. Weiss, Meeting Coordinator Office of Nuclear Regulatory Research l U.S. Nuclear Regulatory Commission W:shington, D.C. 20666 l

l l

i PREFACE This report contains sumaries of papers on reactor safety research to be presented at the 13th Water Reactor Safety Research Information Meeting held at the National Bureau of Standards in Gaithersburg, Maryland, October 22-25, 1985. The summaries briefly describe the programs and results of nuclear safety research sponsored by the Office of Nuclear Regulatory Research, USNRC.

Sumaries of invited papers are also included, which cover the highlights of reactor safety research conducted by the electric utilities through the Electric Power Research Institute, 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. I A number of speakers did not submit summaries for inclusion in this report (indicated by an asterisk [*] in place of a page number in the Table of Contents).

l l

A sunwary of the agenda is printed on the inside of the back cover.

til

t TABLE OF CONTENTS 13th WATER REACTOR SAFETY RESEARCH INFORMATION MEETING October 22-25, 1985 Pagg PREFACE , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii Tuesday, October 23, 1985 PLENARY SESSION Chai rman: G. Marcus (NRC)

INTRODUCTION: G. Marcus (NRC)

To be announced: R. B. Minogue (NRC)

To be announced: E. G. Gomez (Spain)

SESSION 1 Integral Systems Tests Chai rman: W. D. Beckner (NRC)

FIST ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-1 W. A. Sutherland (GE)

OT I S TE ST R ESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3 J. Gloudemans (B&W)

UNIVERSITY OF MARYLAND 2x4 LOOP TEST RESULTS. . . . . . . . . . . . . . 1-5 Y. Y. Hsu, G. A. Pertmer, M. DiMarzo and D. Sallet (J. of Md.)

MIST FACILITY STATUS. . . . . . . . . . . . . . . . . . . . . . . . . . 1-7 H. R. Carter (B&W)

HIGHLIGHTS OF THE OECD LOFT EXPERIMENT PROGRAM. . . . . . . . . . . . . 1-9 J. Birchley and P. North (EG8G)

SESSION 2 Risk Analysis /PRA Applications Chairman: G. Burdick (NRC)

OPERATIONAL PHASE OF INSPECTION PRIORITIZATION . . . . . . . . . . . . 2-1 D. J. Campbell and V. H. Guthrie (JBF Associates)

THE USE OF RISK ANALYSIS IN EVALUATING TECHNICAL SPECIFICATIONS . . . . 2-3 J. L. Boccio and P. K. Samanta (BNL) v

I SESSION 2 (Cont'd)

Page PRA APPLICATION IN OPERATION. . . . . . . . . . . . . . . . . . . . . .

J. H. Bickel (Northeast Utilities)

USE OF PRA IN REGULATORY CONSIDERATION OF SEVERE ACCIDENTS. . . . . . .

Z. Rosztoczy and T. Speis (NRC)

RISK ANALYSIS OF DECAY HEAT REMOVAL SEQUENCES DURING SHUTDOWN . . . . . 2-5 J. Gaertner and W. Reuland (EPRI)

SESSION 3 Process Control Chai rman: F. P. Cardile (NRC)

IDENTIFICATION AND EVALUATION OF FACILITATION TECHNIQUES FOR DECOMMISSIONING LIGHT WATER REACTORS. . . . . . . . . . . . . . . . . 3-1 T. LaGuardia (TLG Engineering)

IMPACTS OF DECONTAMINATION ON SOLIDIFICATION AND WASTE DISPOSAL . . . . 3-3 P. L. Piciulo and J. W. Adams (BNL)

EVALUATION OF NUCLEAR FACILITY DECOMMISSIONING PROJECTS PROGRAM - STATUS. . . . . . . . . . . . . . . . . . . . . . . . . . . 3-5 R. L. Miller (UNC Nuclear Industries)

LWR SPENT FUEL R00 BEHAVIOR DURING LONG-TERM DRY FUEL STORAGE CONDITIONS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-7 C. S. Olsen (EG8G)

LESSONS LEARNED FROM A NUREG-0737 REVIEW 0F HIGH RANGE EFFLUENT MONITORS AND SAMPLERS . . . . . . . . . . . . . . . . . . . . . . . . 3-9 A. P. Hull (BNL) and J. R. White (NRC)

SESSION 4 Integral Systems Tests Chai rman: D. E. Solberg (NRC) I SEMISCALE SFCONDARY TRANSIENT INVESTIGATIONS: RESULTS FROM SEMISCALC MOD-2C FEEDWATER AND STEAM LINE BREAK TESTS . . . . . . . . 4-1 T. J. Boucher (EG8G)

SEMISCALE LIQUID HOLDUP INVESTIGATIONS: A COMPARIS0N OF RESULTS FROM SMALL BREAK LOCA TESTS PERFORMED IN THE SEMISCALE N00-2A AND MOD-2C FACILITIES . . . . . . . . . . . . . . . . . . . . . . . . 4-3 G. G. Loomis and T. J. Boucher (EG8G)

THE RESULTS OF THE ROSA-IV LSTF SMALL-BREAK LOCA EXPERIMENTS. . . . . . 4-5 K. Tasaka, M. Kawaji, M. Osakabe and Y. Koizumi (JAERI) vi

SESSION 4 (Cont'd)

Page PKL REFLOOD TESTS INCLUDING END-0F-BLOWDOWN . . . . . . . . . . . . . . 4-7 R. M. Mandl, B. Brand and H. Watzinger (KWU)

TEST REPORTING IN SUPPORT OF COMPUTER CODE ASSESSMENT AND UNCERTAINTY QUANTIFICATION. . . . . . . . . . . . . . . . . . . . . .

R. Shaw (EG8G)

INTEGRAL SYSTEM TEST (IST) PROGRAM FACILITY SCALING STUDY . . . . . . . 4-9 T. K. Larson (EGAG)

CONTINUING INTEGRAL TESTING CAPABILITY - APPROACH AND SCALING STUDY . . 4-13 T. K. Larson, J. S. Martinell and K. G. Condie (EGAG)

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

NRC OVERVIEW 0F MECHANICAL / STRUCTURAL RESEARCH. . . . . . . . . . . . .

J. E. Richardson (NRC)

STRUCTURAL LOA 0 COMBINATIONS. . . . . . . . . . . . . . . . . . . . . . 5-1 H. Hwang and M. Reich (BNL), B. Ellingwood (NBS), and i M. Shinozuka (Columbia Univ.)

STANDARD PROBLEMS FOR COMPUTER CODES. . . . . . . . . . . . . . . . . . 5-3 M. Reich et al. (BNL)

STEEL CONTAINMENT BUCKLING. . . . . . . . . . . . . . . . . . . . . . . 5-5 T. A. Butler and W. E. Baker (LANL)

PIPING RESEARCH OVERVIEW. . . . . . . . . . . . . . . . . . . . . . . . 5-7 D. Guzy (NRC)

PIPE DAMPING. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-9 A. G. Ware (EG8G)

PIPE CAPACITY TESTS . . . . . . . . . . . . . . . . . . . . . . . . . . 5-11 A. T. Onesto (ETEC)

PIPE RUPTURES IN BWR PLANTS . . . . . . . . . . . . . . . . . . . . . . 5-13 G. S. Holman (LLNL)

VALVE PERFORMANCE TESTING . . . . . . . . . . . . . . . . . . . . . . . 5-15 N. M. Jeanmougin (ETEC) vil

SESSION 6 Nuclear Plant Aging a Chai rman: J. P. Vora (NRC)

Pgge AN APPROACH TO UNDERSTAND AGING - SYSTEM INTERACTIONS AND EVALUATION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

J. Cleveland (SEA, Inc.)

~

AN UPDATE ON SQl'G ACTIVITIES. . . . . . . . . . . . . . . . . . . . . .

A. Marion (BGaE)

MATERIALS ASPECTS OF BWR PLANT LIFE EXTENSION . . . . . . . . . . . . . 6-1

8. M. Gerdon and G. M. Gordon (GE)

MEASUREMENT OF EQUIPMENT DETERIORATION. . . . . . . . . . . . . . . . . 6-3 G. J. Toman (FRC)

AN APPROACH TO EVALUATE THE SAFETY AND RISK IMPLICATIONS OF AGING AND SERVICE WEAR. . . . . . . . . . . . . . . . . . . . . . . . 6-5 W. Vesely (BCL)

Wednesday, October 24, 1985 SESSION 7 Severe Accident Sequence Analysis Chai rman: R. T. Curtis (NRC)

APPLICATION OF RAMONA-3B TO BWR ATWS. . . . . . . . . . . . . . . . . . 7-1 G. C. Slovik, L. Y. Neymotin and P. Saha (BNL)

TRAC /MELPROG Analyses of TMLB' TRANSIENTS IN OCONEE-1. . . . . . . . . 7-3 B. E. Boyack and R. J. Henninger (LANL)

ATWS ANALYSIS FOR BROWNS FERRY NUCLEAR PLANT UNIT 1 . . . . . . . . . . 7-5 R. J. Dallman and W. C. Jouse (EGaG)

ANALYSIS OF FEEDWATER TRANSIENT INITIATED SEQUENCES FOR THE BELLEFONTE NUCLEAR PLANT. . . . . . . . . . . . . . . . . . . . . . . 7-7 C. A. Dobbe (EGAG)

THE UKAEA PWR SEVERE ACCIDENT CONTAINMENT STUDY . . . . . . . . . . . . 7-9 A.T.D. Butland, B. D. Turland and R.L.D. Young (UKAEA)

STATION BLACK 0UT CALCULATIONS FOR BROWNS RIVER. . . . . . . . . . . . . 7-11 L. J. Ott, C. F. Weber and C. R. Hyman (0RNL)

STATION BLACK 0UT CALCULATIONS FOR PEACH BOTTOM. . . . . . . . . . . . . 7-13 S. A. Hodge, S. D. Clinton, C. R. Hyman and C. F. Weber (ORNL) viii

SESSION 7 (Cont'd)

Page MELRPI - DEVELOPMENT AND USE. . . . . . . . . . . . . . . . . . . . . . 7-15 ,

A. Sozer (0RNL)

HYDROGEN TRANSPORT IN A LARGE, DRY CONTAINMENT FOR SELECTED ARRESTED SEQUENCES. . . . . . . . . . . . . . . . . . . . . . . . . . 7-17 D. B. King and A. C. Peterson (SNL)

PRESSURE-TEMPERATURE RESPONSE IN AN ICE-CONDENSER CONTAINMENT FOR SELECTED ACCIDENTS. . . . . . . . . . . . . . . . . . . . . . . . 7-19 S. E. Dingman and A. L. Camp (SNL)

SESSION 8 Separate Effects /Exaeriments and Analyses Chai rman: 1. Zuber (NRC)

HEAT TRANSFER, CARRY-0VER AND FALL-BACK IN 0-TUBE AND STEAM GENERATORS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-1 L. Y. Liao, A. Parlos and P. Griffith (MIT)

MB-2. STEAM GENERATOR TRANSIENT TEST PROGRAM. . . . . . . . . . . . . . 8-3 M. Y. Young, O. J. Mendler and K. Takeuchi (W)

STEAM GENERATOR MODELING DURING TRANSIENTS. . . . . . . . . . . . . . . 8-5 C-Y Paik and P. Griffith (MIT)

SIMULATOR STUDY OF HOT LEG U-BEND TWO-PHASE FLOW UNDER NATURAL CIRCULATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

M. Ishii and S. B. Kim (ANL)

CRITICAL FLOW THROUGH A SMALL BREAK ON A LARGE PIPE WITH STRATIFIED FLOW . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-7 V. E. Schrock, S. T. Revankar, R. Mannheimer and C-H Wang (UCB)

CRITICAL FLOW THROUGH IGSCC IN PIPES. . . . . . . . . . . . . . . . . . 8-11 V. E. Schrock, S. T. Revankar and S. Y. Lee (UCB)

A FINAL REPORT ON THERMAL MIXING. . . . . . . . . . . . . . . . . . . . 8-13 T. G. Theofanus (UCSB)

STEAM EXPLOSIONS: ENERGY CONVERSION EFFICIENCIES OF STEAM EXPL0SIONS FROM TWO MAJOR ACCIDENTS IN THE PULP AND PAPER INDUSTRY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-15 T. M. Grace (Inst, of Paper Chem), R. R. Robinson (IIT Res. Inst.)

and J. Hopenfeld (NRC) ix

4 SESSION 9 Seismic Research Chai rmar.: J. J. Burns, Jr. (NRC) !

Page l NRC OVERVIEW 0F SEISMIC RESEARCH. . . . . . . . . . . . . . . . . . . .

J. E. Richardson (NRC)

EASTERN UNITED STATES SEISMIC HAZARD RESEARCH . . . . . . . . . . . . . 9-1 A. J. Murphy and L. L. Beratan (NRC)

NRC'S SEISMIC MARGINS PROGRAM . . . . . . . . . . . . . . . . 1

.... 9-3 R. J. Budnitz (FRA)

SEISMIC CATEGORY I STRUCTURES PROGRAM . . . . . . . . . . . . . . . . . 9-5 J. G. Bennett, W. E. Dunwoody and C. R. Farrar (LANL)

COMPONENT FRAGILITIES - DATA COLLECTION, ANALYSIS AND INTERPRETATION. . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-7 K. K. Bandyopadhyay and C. H. Hofmayer (BNL)

COMPONENT FRAGILITY RESEARCH PROGRAM - COMP 0NENT PRIORITIZATION AND TE S T I N G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-9 G. S. Holman (LLNL)

COOPERATIVE EFFORTS FOR VALIDATION OF SEISMIC CALCULATION . . . . . . .

J. Costello (NRC)

HEISSDAMPFREAKTOR PHASE II VIBRATION TESTS. . . . . . . . . . . . . . .. 9-11 L. Malcher (KfK) and H. Steinhilber (LBF)

BWR RISK ASSESSMENT . . . . . . . . . . . . . . . . . . . . . . . .;. . 9-13 T. Y. Chuang et al. (LLNL) '

SESSION 10 Fission Product Release & Trcas) ort in Containment Chai rman: T. J. Walcer (NRC)

STATUS OF THE ORNL AEROSOL RELEASE AND TRANSPORT PROJECT. . . . . . . . 10-1 R. E. Adams (0RNL)

STATUS OF THE DEMONA EXPERIMENTS - A COMPARIS0N WITH NAUA CALCULATIONS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-3 W. Sch6ck, H. Bunz and J. P. Hosemann (KfK)

STATUS OF THE LWR AEROSOL CONTAINMENT EXPERIMENTS (LACE) PROGRAM. . . . 10-5 G. R. Bloom et al. (HEDL) and F. J. Rahn (EPRI)

CONTAINMENT INTEGRITY UNDER SEVERE ACCIDENT CONDITIONS. . . . .>... . . 10-7 W. A. von Riesemann et al. (SNL)

X

SESSION 10 (Cont'd)

Page EXPERIMENTAL VALIDATION AND IMPROVEMENT OF CORE DEBRIS / CONCRETE INTERACTION MODELS. . . . . . . . . . . . . . . . . . . . . . . . . . 10-9 A. J. Suo-Anttila (SAIC), J. E. Gronager and D. R. Bradley (SNL), and R. E. Blose (Ktech)

REVIEW 0F THE LARGE-SCALE CORE-CONCRETE INTERACTION EXPERIMENTS AND ANALYSIS AT THE KfK BETA FACILITY . . . . . . . . . . . . . . . . 10-13 R. K. Cole, Jr. (SNL) and M. Reimann (KfK)

THE INFLUENCE OF REACTOR GE0 METRY ON THE BEHAVIOR OF DISPERSED DEBRIS. . . . . . . . . . . . . . . . . . . . . . . . . . . 10-15 W. W. Tarbell, M. Pilch and J. E. Brockmann (SNL)

ANALYSIS OF M0LTEN FUEL-CONCRETE INTERACTIONS AND FISSION PRODUCT RELEASE FROM EX-VESSEL CCRE DEBRIS. . . . . . . . . . . . . . 10-19 D. R. Bradley (SNL)

BNL SEVERE ACCIDENT SEQUENCE EXPERIMENTS AND ANALYSIS PROGRAM . . . . . 10-21 G. A. Greene, T. Ginsberg and N. K. Tutu (BNL)

SESSION 11 International Code Assessment Program Chai rman: F. Odar (NRC)

METHODOLOGY FOR CODE ACCURACY QUANTIFICATION. . . . . . . . . . . . . . 11-1 L. N. Kmetyk et al . (SNL)-

UNCERTAINTY DEVELOPMENT AND APPLICATION . . . . . . . . . . . . . . . . 11-3 G. E. Wilson and G. S. Case (EG8G)

IMPROVEMENTS OF BWR LOCA/ECCS ANALYSIS' IN JAPAN . . . . . . . . . . . . 11-5 K. Yahagi (TEPCO)

NEW JAPANESE CORRELATIONS ON CORE COOLING AND CCFL CHARACTERISTICS DURING BWR LOCA . . . . . . . . . . . . . . . . . . 11-7 H. Nagasaka (Toshiba)

APPLICATION OF TBL RESULTS TO BWR PLANTS. . . . . . . . . . . . . . . . 11-9 M. Naitoh, M. Murase and H. Suzuki (Hitachi) and J. A. Findlay and F. D. Shum (GE)

SAFER QUALIFICATION BY TBL TEST ANALYSIS. . . . . . . . . . . . . . . . 11-11 S. Miura, K. Moriya and T. Sugisaki (Hitachi)

SAFER QUALIFICATION AGAINST ROSA-III RECIRCULATION LINE BREAK SPECTRUM TESTS. . . . . . . . . . . . . . . . . . . . . . . . . 11-13 ,

S. Itoya and J. Otonari (Toshiba) and K. Tasaka-(JAERI)

Xi

SESSION 11 l (Cont'd)

Page ASSESSMENT OF THE SAFER CODE FOR LOCA WITH DATA FROM THE ADVANCED BOILING WATER REACTOR TEST FACILITY. . . . . . . . . . . . . ~11-15 F. D. Shum, A. B. Burgess and B. S. Shiralkar (GE), and K. Yahagi (TEPC0)

SAFETY PERFORMANCE OF THE ADVANCED B0ILING WATER REACTOR. . . . . . . . 11-17 F. M. Paradiso, J.G.M. Andersen and C. D. Sawyer (GE),

and A. Omoto (TEPC0)

ON THE PROBABILISTIC ESTIMATION OF CONTAINMENT FAILURE BY

, STEAM EXPLOSIONS. . . . . . . . . . . . . . . . . . . . . . . . . . . 11-19 T. G. Theofanous (UCSB), M. Abolfudl and H. Amarasooriya (Purdue Univ.), and B. Najafi and E. Ruuble (SAIC)

SESSION 12 Equipment Qualification Chai rman: W. S. Farmer (NRC)

PROGRESS ON QUALIFICATION TESTING METHODOLOGY STUDY OF ELECTRIC CABLES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12-1 S. Okada et al. (JAERI)

EQUIPMENT QUALIFICATION AND SURVIVABILITY RESEARCH AT SANDIA NATIONAL LABORATORIES . . . . . . . . . . . . . . . . . . . . . . . . 12-3 L. L. Bonzon (SNL)

FIRE PROTECTION AND HYDR 0 GEN BURN EQUIPMENT SURVIVAL RESEARCH . . . . . 12-7 D. L. Berry (SNL)

ENVIRONMENTAL AND DYNAMIC QUALIFICATION OF EQUIPMENT RESEARCH AT THE INEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12-9 1

J. A. Hunter (EG&G) 1 Thursday, October 24, 1985 SESSION 13 Severe Accident Source Term Chai rman: M. Silberberg (NRC)

REVIEW 0F SOURCE TERM REASSESSMENT. . . . . . . . . . . . . . . . . . .

M. Silberberg (NRC)

THE SOURCE TERM CODE PACKAGE. . . . . . . . . . . . . . . . . . . . . . 13-1 J. A. Gieseke, P. Cybulskis, H. Jordan and K. W. Lee (BCL)

SEVERE ACCIDENT CODE DEVELOPMENT PROGRAM. . . . . . . . . . . . . . . .

J. Han (NRC)

Xii l

l . _ _ - - -- - -

l l

SESSION 13 (Cont'd)

Page RELAPS/SCDAP: AN INTEGRATED COMPUTER CODE FOR SEVERE ;

, ACCIDENT ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . 13-3 ;

T. C. Cheng (EG8G) '

MELPROG: AN INTEGRATED MODEL FOR IN-VESSEL MELT PROGRESSION ANALYSIS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13-5 J. E. Kelly (SNL)

THE BEHAVIOR OF REACTOR CORE-SIMULANT AEROSOLS DURING HYDROGEN / AIR COMBUSTION . . . . . . . . . . . . . . . . . . . . . . . 13-7 L. S. Nelson et al. (SNL), and J. H. Lee et al. (McGill Univ.)

TELLURIUM CHEMISTRY, TELLURIUM RELEASE AND DEPOSITION DURING THE TMI-2 ACCIDENT . . . . . . . . . . . . . . . . . . . . . . 13-9 K. Vinjamuri et al. (EG8G) and R. A. Sallach (SNL)

ICE CONDENSER EXPERIMENTAL PLAN . . . . . . . . . . . . . . . . . . . . 13-11 L. D. Kannberg, P. C. Owczarski and A. M. Liebetrau (PNL)

SESSION 14 International Code Assessment Program Chai rman: D. E. Bessette (NRC)

APPLICATION OF RELAPS TO ANALYSIS OF THE DOEL STEAM GENERATOR TUBE RUPTURE AND STUDIES OF THE LOSS OF FEEDWATER AND FEEDWATER LINE BREAK TRANSIENTS. . . . . . . . . . . . . . . . . . . . . . . . . . . 14-1 E. J. Stubbe and L. Vanhoenacker (Belgium)

FRG ASSESSMENT OF TRAC-PF1/M001 AND RELAP5/M002 . . . . . . . . . . . . 14-3 F. Winkler (KWU)

TECHNICAL RESEARCH CENTER OF FINLAND USE OF RELAP5/M002 . . . . . . . .

H. Holmstrom (Finland)

ASSESSMENT AND APPLICATION OF RELAP5/ MOD 2 AT STUDSVIK; USE OF TRAC-PF1/ MODI TO ANALYZE LOSS OF GRID TRANSIENT . . . . . . . . . . .

0. Sandervag (Sweden)

ANALYSIS OF BWR6 LOCA WITH REDUCED ECCS USING TRAC-BD1. . . . . . . . . 14-5 S. N. Aksan and G. Th. Analytis (EIR, Switzerland)

UK EXPERIENCE WITH TRAC-PF1/M001 AND RELAP5/M002. . . . . . . . . . . . 14-7 I. Brittain (UKAEA), S. Board (CEGB), and K. Routledge (NNC)

SIMILARITY ANALYSIS OF LARGE STEAM LINE BREAK LOCAs IN ROSA III, FIST, AND BWR/6 . . . . . . . . . . . . . . . . . . . . . . . . . . .

M. Suzuki et al. (JAERI), J. A. Findlay and W. A. Sutherland (GE)

Xiii

SESSION 14 (Cont'd)

Page RELAP5/M002 ASSESSMENT AT BABC0CK & WILC0X. .............. 14-9 C. K. Nithianandan, N. H. Shah and R. J. Schomaker (B&W),

and C. Turk (AP&L)

SESSION 15 Surry Steam Generator / Examination and Evaluation Chai rman: J. Muscara (NRC)

DECONTAMINATION AND CLEANING. . . . . . . . . . . . . . . . . . . . . . 15-1 R. A. Clark (PNL)

GENERATOR DEGRADATION CHARACTERIZATION. . . . . . . . . . . . . . . . . 15-1 R. A. Clark (PNL)

EDDY CURRENT INSPECTION ROUND ROBIN . . . . . . . . . . . . . . . . . . 15-1 P. G. Doctor and Re Ferris (PNL)

STEAM GENERATOR TUBE VIBRATION STUDY. . . . . . . . . . . . . . . . . . 15-3 W. I. Enderlin (PNL)

STRESS CORROSION CRACKING 0F PWR STEAM GENERATOR TUBING . . . . . . . . 15-5 D. van Rooyen (BNL)

SESSION 16 Risk Analysis / Dependent Failure Analysis Chai rman: D. Rasmuson (NRC)

AN OVERVIEW 0F THE RISK METHODS, INTEGRATION, AND EVALUATION PROGRAM ...............................

J. C. Shepherd (NRC)

ANALYSIS OF DEPENDENT FAILURES AND EXTERNAL EVENTS. . . . . . . . . . .

M. P. Bohn (SNL)

AN NRC APPROACH TO DEPENDENT FAILURE ANALYSIS . . . . . . . . . . . . . 16-1 D. J. Campbell, J. R. Kirchner and H. M. Paula (JBF Associates)

PRA PROCEDURES FOR DEPENDENT EVENTS ANALYSIS - AN INDUSTRY PERSPECTIVE K. N. Fleming (PLG)

............................. 16-3 INTEGRATING ROOT CAUSES INTO PRAs . . . . . . . . . . . . . . . . . . . 16-5 L. C. Cadwallader et al. (EG8G), and W. E. Vesely (BCL)

ROOT CAUSE ANALYSIS OF COMP 0NENT DATA . . . . . . . . . . . . . . . . .

G. Crellin and A. Smith (LATA) e xiv l

l

SESSION 17 Materials Engineering Research/Non-Destructive Evaluation Chairman: J. Muscara (NRC)

Page INTEGRATION OF NONDESTRUCTIVE EXAMINATION RELIABILITY AND FRACTURE MECHANICS. . . . . . . . . . . . . . . . . . . . . . . . . . 17-1 S. R. Doctor et al. (PNL)

~

DEVELOPMENT AND VALIDATION OF A REAL-TIME SAFT-UT SYSTEM FOR INSERVICE INSPECTION OF LWRs. . . . . . . . . . . . . . . . . . . . . 17-3 S. R. Doctor et al. (PNL)

NDE OF STAINLESS STEEL AND ON-LINE LEAK MONITORING 0F LWRs. . . . . . . 17-5 D. S. Kupperman and T. N. Claytor (ANL)

PROGRESS FOR ON-LINE AC0USTIC EMISSION MONITORING 0F CRACKS IN REACTOR SYSTEMS. . . . . . . . . . . . . . . . . . . . . . . . . . 17-7 P. H. Hutton and R. J. Kurtz (PNL)

IMPROVED EDDY-CURRENT TESTING 0F STEAM GENERATOR TUBING . . . . . . . 17-9 C. V. Dodd (0RNL)

SESSION 18 Environmental Effects in Piping Chairman: A. Taboada (NRC)

MEASUREMENT AND PREDICTION OF MICR0 STRUCTURAL DEVELOPMENT IN 18-1 I STAINLESS STEEL PIPE WELDMENTS. . . . . . . . . . . . . . . . . . . .

S. M. Bruemmer and D. G. Atteridge (PNL)

BWR PIPE CRACKING AND WELD CLAD OVERLAY STUDIES . . . . . . . . . . . . 18-3 W. J. Shack et al. (ANL)

AGING DEGRADATION OF CAST STAINLESS STEEL . . . . . . . . . . . . . . . 18-5

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

SESSION 19 Containment Systems Research/ Containment Leads Analysis 3

Chai rman: P. M. Wood (NRC)

MODELS FOR BWR AND PWR ENGINEERED SAFETY FEATURES IN THE CONTAIN CODE. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19-1 F. J. Schelling et al. (SNL)

VALIDATION, ASSESSMENT AND APPLICATION OF THE CONTAIN COMPUTER CODE. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19-3 K. D. Bergeron et al. (SNL)

AN INVESTIGATION OF STEAM EXPLOSION LOADINGS USING SIMMER-II. . . . . . 19-5 W. R. Bohl (LANL)

XV 1

F-SESSION 19 (Cont'd)

Page SHORT-TERM AND LONG-TERM ASPECTS OF RECENT HDR CONTAINMENT TESTS ................................

L. Wolf, L. Valencia and K.-H.

19-7 Scholl (KfK)

HMS-BURN: A MODEL FOR HYDROGEN DISTRIBUTION AND COMBUSTION IN NUCLEAR REACTOR CONTAINMENTS . . . . . . . . . . . . . . . . . . . 19-9 J. R. Travis (LANL)

HECTR DEVELOPMENT AND ASSESSMENT. . . . . . . . . . . . . . . . . . . - 19-11 C. C. Wong (SNL)

CONCHAS-SPRAY MODELING OF FLAME ACCELERATION AND AIR FLOWS.19-13 ......

K. D. Marx (SNL)

FLAME ACCELERATION AND DETONATION RESEARCH. . . . . . . . . . . . . . 1. 9-15 M. P. Sherman, S. R. Tieszen and W. B. Benedick (SNL)

PLATINUM CATALYTIC IGNITERS FOR LEAN HYDROGEN / AIR MIXTURES. 19-17 ......

L. R. Thorne, J. V. Volponi and W. J. McLean (SNL)

INFLUENCE OF FLOW CHANNEL GEOMETRY AND DROPLET SIZE ON CONTAINMENT SUBCOMPARTMENT ANALYSIS . . . . . . . . . . . .19-19 ..... l K. Almenas (U. Md.) and R.Y.F. Lee (NRC)

Friday, October 25, 1985 SESSION 20 Materials Engineering Research/ Pressure Vessel Research Chairman: M. Vagins (NRC)

HSST CRACK C. E. ARREST STUDIES OVERVIEW. . . . . . . . . .20-1 Pugh (0RNL) .........

WIDE PLATE CRACK ARREST TESTING . . . . . . . . . . . .20-3 R. deWit and R. J. Fields (NBS)

ELAST0 DYNAMIC FRACTURE ANALYSES OF LARGE CRACK-ARREST EXPERIMENTS B. R. Bass, C. E. Pugh. . and

. . .J.. K.

. .Walker

. . . . (0RNL)

................. 20-5 DEVELOPMENT M. F. Kanninen et al.OF VISC0 PLASTIC FRACTURE MECHANICS.

(SwRI) . 20-7 .

PRESSURIZED-THERMAL-SH0CK EXPERIMENTS:

PTSE-2 PLANS. . . . . . . . PTSE-1 RESULTS AND R. H. Bryan et al. (ORNL) ................ 20-9 REACTOR W. R. CorwinVESSEL (0RNL) CLADDING SEPARATE EFFECTS ...... 20-11 STUDIES. . .

xvi

SESSION 20 (Cont'd)

Page IRRADIATION EFFECTS IN LOW-ALLOY REACTOR PRESSURE VESSEL STEELS (HEAVY-SECTION STEEL TECHNOLOGY PROGRAM SERIES 4 AND 5) . . . . . . . 20-13 J. J. McGowan et al. (0RNL), and B. H. Menke (MEA)

RADIATION SENSITIVITY AND ANNEALING PARAMETER STUDIES . . . . . . . . . 20-15 J. R. Hawthorne (MEA)

LWR DOSIMETRY PROGRAM OVERVIEW. . . . . . . . . . . . . . . . . . . . . 20-17 E. D. McGarry (NBS) i REACTOR VESSEL CAVITY 00SIMETRY . . . . . . . . . . . . . . . . . . . .

i J. Butler (UKAEA)

SESSION 21 Code Assessment and Improvement Chairman: F. Odar (NRC)

TRAC-PF1/ MODI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21-1 D. R. Liles and S. B. Woodruff (LANL)

TRAC-BWR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21-3 S. Z. Rouhani et al . (EG8G)

RELAP5/M002 DEVELOPMENT . . . . . . . . . . . . . . . . . . . . . . . . 21-5 G. W. Johnsen (EG8G)

COBRA / TRAC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

D. Trent (PNL) 0VERVIEW 0F RAMONA-3B CODE. . . . . . . . . . . . . . . . . . . . . . . 21-7 L. Y. Neymotin et al. (BNL) 0VERVIEW 0F TRAC-BWR ASSESSMENT: TRAC-BD1/M001 & TRAC-BF1. . . . . . . 21-9 G. E. Wilson et al. (EG8G)

LOS ALAMOS TRAC-PF1/ MOD 1 CODE ASSESSMENT PROGRAM USING LOFT AND OTIS DATA . . . . . . . . . . . . . . . . . . . . . . . . . 21-11 T. D. Knight (LANL)

ASSESSMENT OF TRAC-BD1/ MOD 1 USING FIST DATA . . . . . . . . . . . . . . 21-13 J. H. Jo and H. R. Connell (BNL)

RELAP5/M002 ASSESSMENT AT INEL. . . . . . . . . . . . . . . . . . . . . 21-15

. P. D. Wheatley et al . (EG8G)

STATUS OF TRAC-PF1/M001 INDEPENDENT ASSESSMENT AT SANDIA. . . . . . . . 21-17 L. D. Buxton and L. N. Kmetyk (SANL) xvii

l SESSION 22 Industry Safety Research Chairman: W. B. Loewenstein (EPRI)

Page SAFETY RESEARCH IN TRANSITION , . . . . . . . . . . . . . . . . . . . 22-1 W. B. Loewenstein and R. B. Duf fy (EPRI) l l

RESOLUTION OF STEAM GENERATOR INTEGRITY QUESTIONS . . . . . . . . . . . 22-3 J. F. Lang and S. P. Kaira (EPRI)

SIGNAL VALIDATION: A NEW INDUSTRY TOOL . . . . . . . . . . . . . . . . 22-5 S. M. Divakaruni and B.K.H. Sun (EPRI)

ON-LINE CORROSION CRACKING MONITOR DEVELOPMENT. . . . . . . . . . . . . 22-7 J. D. Gilman and R. L. Jones (EPRI)

SEVERE ACCIDENT CONTAINMENT INTEGRITY . . . . . . . . . . . . . . . . . 22-9 H. T. Tang (EPRI)

EFFECTS OF THERMAL CONVECTION IN POSTULATED HIGH PRESSURE ACC IDE NTS IN PWRs . . . . . . . . . . . . . . . . . . . . . . . . . . 22-11 B. R. Sehgal (EPRI)

FISSION PRODUCT RETENTION IN BWR COMPONENTS . . . . . . . . . . . . . . 22-17 M. Merilo (EPRI)

SESSION 23 Materials Engineering Research/ Piping Research & Fracture Mechanics Chairman: M. Mayfield (NRC)

THE UK MATERIALS AND NON-DESTRUCTIVE TESTING RESEARCH PROGRAMME FOR THE PWR . . . . . . . . . . . . . . . . . . . . . . . . 23-1 G. J. Lloyd and R. A. Murgatroyd (UKAEA)

NRC PIPING REVIEW COMMITTEE;

SUMMARY

ON PIPE BREAKS . . . . . . . . . .

R. Klecker (NRC)

DEGRADED PIPING PROGRAM - PHASE II PROGRESS . . . . . . . . . . . . . . 23-3 G. M. Wilkowski et al. (BCL)

PIPING RESEARCH IN THE FEDERAL REPUBLIC 0F GERMANY. . . . . . . . . . . 23-5 K. Kussmaul (U. of Stuttgart)

OUTLINE OF NUCLEAR PIPING RESEARCH CONDUCTED IN ITALY . . . . . . . . 23-7 P. P. Milella (ENEA-DISP)

RESEARCH ACTIVITIES ON PIPING FRACTURE IN JAPAN . . . . . . . . . . . . 2.$ -

G. Yagawa (V. Tokyo), and Y. Takahashi and K. Kuwabara (CRI)

XViii 6 _.

SESSION 20 (Cont'd)

Page IRRADIATION EFFECTS IN LOW-ALLOY REACTOR PRESSURE VESSEL STEELS (HEAVY-SECTION STEEL TECHNOLOGY PROGRAM SERIES 4 AND 5) . . . . . . . 20-13 J. J. McGowan et al. (0RNL), and B. H. Menke (MEA)

RADIATION SENSITIVITY AND ANNEALING PARAMETER STUDIES . . . . . . . . . 20-15 J. R. Hawthorne (MEA)

LWR D0SIMETRY PROGRAM OVERVIEW. . . . . . . . . . . . . . . . . . . . . 20-17 E. D. McGarry (NBS)

REACTOR VESSEL CAVITY 00SIMETRY . . . . . . . . . . . . . . . . . . . .

J. Butler (UKAEA)

SESSION 21 Code Assessment and Improvement

~~

Chai rman: F. Odar (NRC)

TRAC-PF1/M001 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21-1 D. R. Liles and S. B. Woodruff (LANL)

TRAC-BWR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21-3 S. Z. Rouhani et al. (EG&G)

RELAPS/M002 DEVELOPhENT . . . . . . . . . . . . . . . . . . . . . . . . 21-5 G. W. Johnsen (EG&G)

COBRA / TRAC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

D. Trent (PNL)

OVERVIEW 0F RAMONA-3B CODE. . . . . . . . . . . . . . . . . . . . . . . 21-7 L. Y. Neymotin et al. (BNL)

OVERVIEW 0F TRAC-BWR ASSESSMENT: TRAC-BD1/ MODI & TRAC-BF1. . . . . . . 21-9 G. E. Wilson et al. (EG&G)

LOS ALAMOS TRAC-PF1/M001 CODE ASSESSMENT PROGRAM USING LOFT AND OTIS DATA. . . . . . . . . . . . . . . . . . . . . . . . . . 21-11 T. D. Knight (LANL)

ASSESSMENT OF TRAC-BD1/M001 USING FIST DATA . . . . . . . . . . . . . . 21-13 J. H. Jo and H. R. Connell (BNL)

RELAPS/ MOD 2 ASSESSMENT AT INEL. . . . . . . . . . . . . . . . . . . . . 21-15 P. D. Wheatley et al. (EG&G)

STATUS OF TRAC-PF1/M001 INDEPENDENT ASSESSMENT AT SANDIA. . . . . . . . 21-17 L. D. Buxton and L. N. Kmetyk (SANL) xvii 1

l l

SESSION 22 !

Industry Safety Research '

Chairman: W. B. Loewenstein (EPRI)

Page SAFETY RESEARCH IN TRANSITION . . . . . . . . . . . . . . . . . . . . . 22-1 W. B. Loewenstein. and R. B. Duffy (EPRI)

RESOLUTION OF STEAM GENERATOR INTEGRITY QUESTIONS . . . . . . . . . . . 22-3 J. F. Lang and S. P. Kalra (EPRI)

SIGNAL VALIDATION: A NEW INDUSTRY TOOL . . . . . . . . . . . . . . . . 22-5 S. M. Divakaruni and B.K.H. Sun (EPRI)

ON-LINE CORROSION CRACKING MONITOR DEVELOPMENT. . . . . . . . . . . . . 22-7 J. D. Gilman and R. L. Jones (EPRI)

SEVERE ACCIDENT CONTAINMENT INTEGRITY . . . . . . . . . . . . . . . . . 22-9 H. T. Tang (EPRI)

EFFECTS OF THERMAL CONVECTION IN POSTULATED HIGH PRESSURE ACCIDENTS IN PWRs . . . . . . . . . . . . . . . . . . . . . . . . . . 22-11 B. R. Sehgal (EPRI)

FISSION PRODUCT RETENTION IN BWR COMPONENTS . . . . . . . . . . . . . . 22-17 M. Merilo (EPRI)

SESSION 23 Materials Engineering Research/ Piping Research & Fracture Mechanics Chairman: M. Mayfield (NRC)

THE UK MATERIALS AND NON-DESTRUCTIVE TESTING RESEARCH PROGRAMME FOR THE PWR . . . . ; . . . . . . . . . . . . . . . . . . . 23-1 G. J. Lloyd and R. A. Murgatroyd (UKAEA)

NRC PIPIf:G REVIEW COMMITTEE;

SUMMARY

ON PIPE BREAKS . . . . . . . . . .

R. Klecker (NRC)

DEGRADED PIPING PROGRAM - PHASE II PROGRESS . . . . . . . . . . . . . . 23-3 G. M. Wilkowski et al. (BCL)

PIPING RESEARCH IN THE FEDERAL REPUBLIC 0F GERMANY. . . . . . . . . . . 23-5 K. Kussmaul (V. of Stuttgart)

OUTLINE OF NUCLEAR PIPING RESEARC.i CONDUCTED IN ITALY . . . . . . . . . 23-7 P. P. Milella (ENEA-DISP)

RESEARCH ACTIVITIES ON PIPING FRACTURE IN JAPAN . . . . . . . . . . . 2?-

G. Yagawa (U. Tokyo), and Y. Takahashi and K. Kuwabara (CRI) xviii I____._______________ _

4

SESSION 23 (Cont'd)

Page FRACTURE ANALYSIS OF WELDED TYPE 304 STAINLESS STEEL PIPE USING LIMIT LOAD AND J-INTEGRAL TECHNIQUES. . . . . . . . . . . . . . 23-11 R. A. Hays, M. G. Vassilaros and J. P. Gudas (DTNSRDC)

PIPING FRACTURE MECHANICS DATA BASE . . . . . . . . . . . . . . . . . . 23-13 A. L. Hiser (MEA)

SESSION 24 2D/3D Research Chairman: G. 5. Rhee (NRC)

RESULTS OF CCTF TESTS . . . . . . . . . . . . . . . . . . . . . . . . . 24-1 Y. Murao et al. (JAERI)

RESULTS OF SCTF REFLOOD TESTS . . . . . . . . . . . . . . . . . . . . . 24-3 T. Iwamura et al. (JAERI)

TRAC ANALYSIS FOR CCTF AND SCTF TESTS AND UPTF DESIGN /0PERATION . . . . 24-5 J. W. Spore (LANL)

STATUS OF THE GERMAN UPTF PROGRAM . . . . . . . . . . . . . . . . . . . 24-7 K. R. Hofmann (GRS)

SESSION 25 Nuclear Plant Analyzer Chairman: C. R. Troutman (NRC)

NUCLEAR PLANT ANALYZER DEVELOPMENT AT THE IDAHO NATIONAL ENGINEERING LABORATORY. . . . . . . . . . . . . . . . . . . . . . . . 25-1

, E. T. Laats et al. (EG&G)

THE LOS ALAMOS NUCLEAR PLANT ANALYZER . . . . . . . . . . . . . . . . . 25-5 T. D. Knight and D. R. Liles (LANL)

BWR PLANT ANALYZER DEVELOPMENT AT BNL . . . . . . . . . . . . . . . . . 25-7 E. Cazzoli, H. S. Cheng, A. N. Mallen and W. Wulff (BNL) l l

I KiX

_ -. . _ _ . m. _ _ _ _ _ _ ._ _ _ . _ . _ .

f 1

l FIST ANALYSIS i Wm. A. Sutherland Nuclear Energy Business Operations General Electric. Company

~

San Jose, California The Full Integral Simulation Test (FIST) Program is a three pronged approach to the development of best-estimate analysis capability for BWR systems. An experimental program in a single bundle BWR system simulator ,

facility extends the LOCA data and adds operational transients data. An

analytical method development program extends the BWR-TRAC computer code to model BWR specific components (e.g. jet pump, separator, etc.) and all f

i major interfacing systems to improve application modeling flexibility, and to reduce computer running time. A method qualification program tests TRAC l against experiments run in the FlST facility and extends the program results to reactor system applications. With the recent completion and integration of these three activities, the best-estimate analysis capability objective has been achieved.

The FlST facility is an integral system capable of full power steady state, operation as well as real time LOCA and operational transients. It is a full height simulation of the reactor vessel and internals, with scaled regional volumes, and includes all major interfacing systems and automatic control system trip signals. This provides full scale values for I fluid conditions, heat transfer performance, and thermodynamic state. The l design of the facility incorporates BWR system design features important to ,

} ' thermal-hydraulic performance. Since each BWR fuel bundle is individually channeled (i.e., there is no cross flow in the core region), the thermal-hydraulic conditions within the core are accurately represented by a single bundle.

l l-1 e

l Large and small break LOCA, steam line break LOCA, and power transient ;

simulations have been completed, as well as measurement of the natural circulation flow characteristics. The results from these tests have been compiled and comparisons made to show trends in different types of events.

Tests of greatest interest have been evaluated in detail, and the analysis i of these data has led to the identification and understanding of, and modeling of, the controlling physical phenomena. These analyses provide guidance for development of BWR system features and components in TRAC.

A version of BWR-TRAC which includes the component and phenomena models developed under the Program was used to analyze the FIST facility response in three LOCA tests and an operational transient test. Particular attention to system definition and application modeling was given to the lower plenum region, two-phase level tracking within all regions, vessel stored heat, flow path loss coefficients, and break geometry. The pre-test analyses of the large break and small treak LOCA tests are found to represent the observed controlling thermal-hydraulic phenomena very well.

The analyses of a break from inside the shroud (i.e. LPLI line) and a turbine trip transient with delayed shutdown compare equally well with the system performance measured in the tests. Careful system definition leads to the TRAC analyses correctly handling lower plenum flow split performance, and the resulting prediction of core flow and liquid inventory lead to representative thermal performance in the bundle. Detail in the vessel stored heat model and break geometry contribute to good agreement with results from the small break and operational transient tests as well.

In summary, the data base has been extended to a wide range of system response situations, and analysis of these results has provided guidance for many model improvements. The method development program has provided improved TRAC analysis capability as well as system definition and modeling i'

flexibility and improved running time. Qualification against LOCA and operational transient tests has demonstrated BWR-TRAC capability to handle these transients and application for reactor system performance analyses.

1-2 I

OTIS TEST RESULTS J. R. Gloudemans Nuclear Power Division The Babcock & Wilcox Company Lynchburg, VA OTIS (Once Through Integral System) was a 1 x 1 full-elevation model of a domestic, raised-loop B&W plant. Testing was performed to obtain integral system data for code benchmarking. OTIS was sponsored by the Nuclear Regulatory Commissio n, the B&W Owners Group, the Electric Power Research Institute, and Babcock & Wilcox. Five types of tests were performed.

The results of the OTIS-GERDA benchmark test indicated that the facilities performed similarly, demonstrated the effectiveness of the BCM (boiler-condenser mode), and illustrated the stabilizing effect of the high-elevation injection of auxiliary feedwater. Bro types of cooldown tests were performed, HPI-PORY (Feed and Bleed) cooling and natural circulation cooldown. Increasingly variable system conditions were observed during HPI-PORY cooling, but the core-region fluid was cooled continuously. Head voiding was encountered during the natural circulation cooldown transients, as the primary system pressure was reduced.

This voiding had little effect on loop flow and the continuing cooldown.

Head venting eliminated the head voids.

Two o perator-controlled tests with differing HPI characteristics were performed in OTIS. The operator actuated the PORY (relief valve) to control system pressure until auxiliary feedwater became available; refill and post-refill cooldown were achieved relatively rapidly. Guard heating effects were examined in two added tests. The initial transients (depressurization to saturation, flow interruptions, BCM, and the start of refill) were similar with and witLout guard heating. Guard heating increased the primary system repressuriz ation during refill and delayed the completion of refill.

Seven of the fifteen OTIS tests singly varied the nominal SBLOCA boundary conditions (leak and HPI characteristics, SG level control, and break isolation).

The steady state initial conditions were self-consistent and virtually identical among the tests. Following the initial depressurization to saturation, the transients progressively diverged in response to the imposed boundary system variations. But the primary system repressurizations during interruptions of flow and primary-to-secondary heat transfer were similar in magnitude --

to some extent, the integral system interactions countered the imposed inter-test variations. These pressure trends underscored the significance of the BCA and the importance of event timing.

Several trends were discernible during the refill of the primary loop:

In most tests, the primary system gradually repressurized, the system conditions tended toward equilibrium, and the rate of refill was consequently supressed.

The refill of the HL was usually completed only after the hot leg vent had been opened. Natural circulation and the post-refill cooldown commenced within a few minutes after the completion of HL refill. Core cooling was maintained throughout each test transient.

l l-3

.o.

.e - x se s. 2 -

z.a-- --

w-+.-.=1.- , -- - --

i

'l s

1 1

4f I

l t

4 i l

i 1

s f

y T

1 p 5 l i

t I

4 1

1 t

I i

. I l

1 4

t 4

i l i,

3

+

i

f I l i

\

l 1

5 i i t

i 4

I i i i

i

]

l l

4 f

I r.

r f

l - -.. _. _ . _ . _ - _- - . - m-.- ---.--.0,--.. -.,- -. ------ --- -- - . --,.- - .- -- - - . _ - - - - . - - -, - --,-.. --- --,-- --

UNIVERSITY OF MARYLAND 2X4 LOOP TEST RESULTS Y.Y. Hsu, Dept. of Chemical & Nuclear Engineering G.A. Pertmer, Dept. of Chemical & Nuclear Engineering M. DiMarzo, Dept. of Mechanical Engineering D. Sallet, Deot. of Mechanical Engineering University of Maryland, College Park, MD 20742 he LNCP 2x4 Icop mnstruction was completed in this summer. We test facility is a 1/500-scale by volume of the B&W Ibwer Plant (lowered loop),

with scaled pressure vessel, OTSG-type steam generators (2), pressurizer.

We pressure vessel is provided with annulus downcomer and flapper-type vent valves. We power is provided by 15 electrical heater rods.

Since late spring, the following tests were performed:

Shake-down tests Hydraulic Characterization tests Heat Transfer Characterization tests Preliminary Natural Circulation tests Preliminary Blow-down tests

) Preliminary High-point Venting & Flow Besumption tests Preliminary SB-LOCA Simulating Nominal Base Case run h e results for various tests can be summarized as follow:

i' Hydraulic Characterization tests - Wrough many runs of various inlet i and exit combinations, the resistances of each component have been identified to enable synthesis of overall flow resistance for various flow combinations.

Heat Transfer Characterization Tests - We tests were run to check heat balance and to determine heat loss. Heat loss measurements were also performed at various locations using AT across cork -

patches.

Preliminary Natural Circulation tests - Contrary to the initial concerns voiced in various review meetings, natural-circulation were readily initiated & increased with power setting.

Preliminary Blow-down tests - Blow down of scaled small break and large breaks were performed. Natural circulation interruption, boiling-condensation nodes were all observed. Laser Doppler Anmenometry was used to measure flow rate.

1 Preliminary High-point Vent and Flow Resumption - After flow

interruption, the /1-bend void was vented and natural circulation j resumption achieved.

Preliminary SBLOCA Simulating Norminal Base-case Run - Based upon the scaling analysis, we have developed a pressure-scaling methodolo-4 gy. Wis series of tests showed that we can simulate a norminal i base case (TRAC Run 300,000) with scaled-down pressure but retaining the phenominalogical sequence of a SBLTA scenerio.

i 1-5 1

Data of tests and som video tape record will be presented.

p 'Ibe tests arri analyses were conducted by the following students:

M. Massoix1, M. Popp, W.K Lin, Z.Y. Wang and C.J. Minno.

l 1-6

_ _ ._. =. . _ _ _ ..

l I

MIST Facility Status H. R. Carter Babcock & Wilcox Company I

The Multi-Loop Integral System Test (MIST) is part of the Integral System Test Program (IST) being sponsored by the Nuclear Regulatory 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 the MIST experimental facility will be used to benchmark system computer codes, such as RELAP 5 and TRAC against simulated abnormal plant transients. The subject paper describes the design of the MIST facility, the instrumenta tion, and the planned test program.

Design and construction of the MIST facility was initiated in October, 1983, and construction will be completed in September,1985. 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 general, followed in order of priority; elevation, small break loss-of-cooling accident (LOCA) phenomena, 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 generator 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 locations.

)

q The MIST core and steam generators are full-length subsections of their plant counterparts. The core is simulated by 45 full-length 0.430 inch

, diameter heater rods and 4 in-core guide tubes with plant typical fuel pin

pitch. The simulated rods will be capable of full power output but, because of available power, are limited to approximately 10% scaled power (330 KW) for the planned program. A fixed axial heat flux profile with a peak to average flux ratio of 1.25 and a flat radial heat flux profile are used. Two 19-tube once-through steam generators (OTSG's) are used in MIST. The OTSG's are constructed with prototypical tubing and support plate and tube arrangements.

The hot and cold leg piping were sized to preserve two-phase phenomena. '

This resulted in hot leg and cold leg nominal pipe diameters of 2-1/2 inches and 2 inches, respectively. Scaled reactor coolant pumps are installed in each of the 4 cold legs. The reactor coolant pumps are designed to provide single-phase scaled flows at plant typical heads, simulated plant pump coastdown and spinup characteristics, and allow for pump operation under two-phase and single-phase conditions.

l .

1-7

l l

Other components in MIST include simulated reactor vessel vent valves, l pilot-operated relief valve (PORV), hot leg and reactor vessel vents, high pressure injection (HPI), core flood system, and a closed secondary system.

Critical flow orifices are used to provide control primary system leaks and vent. Passive and active (electrical guard heaters) insulation is used on all the reactor coolant system components to minimize system heat losses.

Over 850 instruments are present in MIST and are interfaced to a computer controlled high speed data acquisition system. MIST instruments consist of measurements of temperature, pressure, and differential pressure. Fluid level and phase indications are provided by optical viewports, conductivity probes, and differential pressures. Fluid density and flow regime information are provided by gamma densitometers and optical viewports. Mass flow measurements at the system boundaries are made using special low flow meters.

Testing in the MIST facility is to be performed in three major phases; debug, characteriza tion, and transients. The debug tests are to demonstrate the operability of each MIST component and will provide for corrective action when necessary. The characterization tests generally follow the debug tes to to examine the behavior of individual key systems and explore limited integral system interactions. The transient phase of MIST testing, and most extensive phase, includes 41 tests. These tests will explore small break loss-of-coolnnt accidents, feed and bleed cooling operation, steam generator tube ruptures, and the effect of the reactor coolant pump operation on system cooldown.

i l

l l-8

Highlights of the OECD LOFT Experiment Program J. Birchley, UKAEA/EG&G Idaho, Inc. i P. North, EG&G Idaho, Inc.

A group of countries,(a) each of which is a member of the Organization for Economic Cooperation and Development (0 ECD), formed and sponsored the OECD LOFT Project under the auspices of the OECD Nuclear Energy Agency (NEA). The purpose of the Project was to undertake research intended to improve the understanding and predictability of transient behavior and to enhance the reliability, availability, economics and safety of pressurized water reactors.

The specific objectives of the OECD LOFT experiment program were formulated to meet the needs of the participating countries. The resultant experiment program was defined on the basis of consensus among the participants and com-prised a total of six thermal-hydraulic experiments and two fission product release and transport experiments.

, The Loss of Fluid Test (LOFT) facility, in which the experiment program was conducted, is a nuclear integral test facility at the Idaho National Engineer-ing Laboratory, operated by EG&G Idaho, Inc. for the U.S. Department of Energy.

The LOFT pressurized water reactor (PWR) incorporated the major functional components of a commercial PWR, was capable of operation under nominal PWR

, operating conditions and a wide range of off-normal and accident conditions, and was unique in providing nuclear heat generation within an integral test facility.

The objectives and conduct of the six the-mal-hydraulic experiments addressed 5

a number of issues of current concern to the various member nations. The ob-jectives of the two fission product experiments addressed the magnitude and physical and chemical character of the radionuclide source term for accidents resulting in fuel damage.

The International OECD LOFT Experiment Program has benefited greatly from the active and expert participation by the Project members, and has been success-fully completed with the achievement of the major imediate objectives. The details of the experiment specifications and the resultant data are proprietary I to the sponsors. The bulk of the experiment documentation and data has been j~ issued to the participating countries, and is auilable to organizations within those countries through the respective OECD LOFT Project Program Review Group

member. A list of the various Review Group members and their respective organizations is presented in Appendix I.

4 (a) Current Project membership comprises: Austria, Finland, Italy, Japan, Spain, Sweden, Switzerland, United Kingdom, U.S.A. (EPRI, USDOE and USNRC as individual participants), and West Germany.

4 l l-9

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

1 APPENDIX I OECD LOFT PROGRAM REVIEW GROUP Mr. Jesse Fell Dr. Giorgio Palazzi UKAEA Ente Nazionale Energie Alternative Winfrith CRE Cassaccia l Dorchester, Dorset C.P. No. 2400 United Kingdom I-00100 Rome Italy Mr. D. F. Giessing U. S. Department of Energy Mr. Jose Puga MS B-107 ENUSA Washington, D. C. 20545 c/ Santiago Rusinol, 12 E-28040 Madrid Prof. E. F. Hicken Spain Gesellschaft fur Reaktor-sicherheit mbH Mr. O. Sandervaag D-8046 Garching Studsvik Energiteknik AB Federal Republic of Germany S-61182 Nykoping j Sweden Mr. Heikke Holmstrom Technical Research Center of Dr. Brian Sheror.

Finland (VTT) Office of Nuclear Reactor Regulation P0B 169, SF-00181 ". S. Nuclear Regulatory Commission

Helsinki 18, Finland Washington, D. C. 20555 Mr. P. D. Jenkins Dr. Robert Van Houten Central Electricity Generating U. S. Nuclear Regulatory Commission Board Washington, D. C. 20555 Barnwood, Gloucester GL4 7RS Dr. Kanji Tasaka United Kingdom Japan Atomic Energy Research Institute Tokai-mura, Naka-gun Mr. O. M. Mercier Ibaraki-ken 219-11 Eidgenossisches Institut fur Japan Reaktorforschung CH-5303 Wurenlingen Japanese Industries Group

Switzerland c/o Mitsubishi Corporation Power and Electrical Systems i Mr. M. Merilo Attention: Mr. T. Hada, Manager EPRI Nuclear Power Systems, Team A P.O. Box 10412 6-3, 2-Chome Marunouchi Palo Alto, California 94304 Chiyoba-ku, Tokyo 100 Japan S. M. Modro Osterreichisches Forschungszentrum Seibersdorf Gesellschaft mbH 1 c/o EG8G Idaho, Inc.

P.O. Box 1625

! Idaho Falls, Idaho 83415 i

i 1-10 i

. , - . _ , . - - - , . ,,. , . - ,..c.- .--------.-e.- - - , . . . _ , , . . - - - - ,_ ,

l OPERATIONAL PRASE OF INSPECTION PRIORITIZATION l 1

D. J. Campbell V. H. Guthrie JBF Associates, Inc.

Knoxville, Tennessee G. F. Flanagan Oak Ridge National Laboratory l Oak Ridge, Tennessee l

l The objective of the Risk Assessment Application to NRC Inspection Program being performed by the Oak Ridge National Laboratory and JBF Associates, Inc. is to develop methods for applying the results of probabi-listic risk assessmenta (PRAs) to manpower allocation decisions made by NRC inspectors. Accomplishing this objective will help inspection personnel decide which of the activities that demand their time have the greatest risk-reduction or safety-assurance potential.

Two key observations made early in the first phase of this program have had a major influence on the program's direction. First, PRAs are limited to quantifying a plant's bottom-line risk and showing how important various component and system failures are to this risk. While inspection personnel do inspect individual components and systems, they are validly more concerned with ensuring that nuclear power plant owners have adequate reliability assurance programs in place. Equipment reliability performance is a useful barometer for evaluating a licensee's programs; however, when equipment performance suffers, inspection personnel are more ef fective if they focus on the root causes of failures and their associated corrective actions rather than responding to individual failure events.

With the above observation in mind, Phase I of the program continued until a four-step procedure was developed to relate PRA results to inspection decisions. These steps are:

1. Relate system and component failure probabilities to plant risk.

2. Relate root causes of failure to system and component failures.

3. Relate reliability assurance programs to root causes of failure.

4. Relate inspection actions to reliability assurance programs.

The first step is accomplished using the results of a PRA. The second step, relating root causes of failure to system and component failures, is 2-1

the key step in this procedure becausa if the various root causes of failures can be ranked according to their importance to plant risk, then the door is opened for inspection personnel to carry out the last two l steps. Thus, the gap between PRA results and the needs of inspection I personnel can be bridged by identifying the relationships between root causes of failure and system and component failures. The NRC has programs in progress to evaluate root causes of failure.

The second observation that influenced the direction of the Risk !

Assessment Application to NRC Inspection Program is that PRA reports are written in the language used by PRA practitioners- a language that is not readily understood by others. Phase II of the program is focusing on developing a program for a microcomputer to present PRA-based information that can be interpreted by inspectors for use in making decisions. The Plant Risk Status Information Management System (PRISIM) is a decision-oriented, user-friendly, menu-driven program that contains data base management and interactive routines that will aid inspectors in allocating their efforts toward those areas where they will have the greatest impact on safety.

A c'omputer program was chosen to catalog and present the PRA information because the total amount of information is large, but the amount needed for any particular decision is relatively small. PRISIM allows the user to quickly and logically access the desired information without being overwhelmed by enormous quantities of data. PRISIM's data base consists largely of screen images that present PRA information in both textual and graphic formats. Each screen also acts as a menu, giving the user options to see more-detailed information in the area of his interest.

As one option in the program, PRISIM screens list h1C inspection modules and procedures, identify decisions inspectors must make to implement the procedures effectively, and direct the user to more screens of PRA-based information that influence specific decisions of interest.

The user indicates to PRISIM the areas of interest he wishes to pursue by using a cursor. The position of the cursor on the screen determines what information the data base management routine will present. The user does not need to have a background in computer operation or PRA to use the program er understand and employ the information it presents.

Because some decisions made by inspectors depend on the current status of the plant, PRISIM contains an interactive routine that allows the user to specify components or subsystems that are out of service. The user is

then apprised of the impact the specified condition (s) places on instantaneous risk and the components that are most critical to maintaining plant safety under the specified condition (s). Thus inspectors can plan their actions using PRA-based information integrated with plant status information.

The ris't-based information being selected for presentation by PRISIM is complete within itself and will be as compatible as possible with the four-step approach developed in the first phase of this program. However, all the root cause information needed to implement this approach is not currently available. As this information is documented, PRISIM will be updated and will become an even more useful tool for inspection planning and decision-making.

2-2 l

- - -. - - _- - - - _. ~ = - _ - . . . - _ . - - . - - -

4 THE USE OF RISK ANALYSIS IN EVALUATING TECHNICAL SPECIFICATIONS l John L. Boccio and Pranab K. Samanta Department of Nuclear Energy Brookhaven National Laboratory Upton , NY 119 73 i

Technical Specifications (TS) are design and procedural limits that entail ex-plicit restrictions on the operation of a nuclear power plant and the mainte-

! nance of safety s>atens in a pre accident condition. There is general agreement ,

, that TS are complex and difficult to implement. Surveillance Requirements (SRs)

! within TS sometimes introduce system lineups that make operations and compliance j rather difficult. In addition to SRs, another element of TS, namely Limiting a

Conditions for Operations (LCOs), do not have valid technical bases. NUREG 1 1024, " Technical Specifications - Enhancing the Safety Impact" has documented ;

past experiences which indicate that lack of guidance on TS can affect both 7 licensing and operations.

}.

Industry and the NRC have embarked upon parallel and coordinated eff orts to fur-ther identify problems with TS, come up with alternatives, and develop revised technical specification inducive to help short- and long-term safety of nuclear power plants. This paper discusses how risk and reliability-based methods can be used or have the potential to provide a sounder basis for addressing the j safety issues associated with present technical specifications. The paper will ,

j focus on the current and future work of the Procedures for Evaluating Technical Specifications (PETS) program conducted within the Division of Risk Analysis and

Operations. The PETS program has essentially the following objectives: (1) ,

{ develop and demonstrate approaches for addressing specific TS safety issues, (2) j develop and demonstrate an overall procedure for determining Allowed Outage j Times ( A0Ts) and Surveillance Test Intervals (STIs), (3) propose A0Ts and STIs ,

! for select specific cases, and (4) provide a report or a " procedures guide" for i implementing TS analysis procedures. It will show how the products generated by the PETS program can assist the Technical Specification Improvement Program (TSIP) in meeting its stated objectives. The TSIP was initiated within the Division of Safety Technology.

j The paper discusses various aspects being addressed within the PETS program in a revising TS using risk and reliability insights. The evaluation of a number of i issues impacting on the determination of ADT and STI requirements were performed j using the emergency coolant system of the Limerick plant. In addition, using j the PRA for Arkansas Nuclear One - Unit I as the basis for measuring the risk j impact of testing and maintenance activities, the paper summarizes PETS evalua-l tion of a complete set of A0T and STI requirements of a TS. Through example l applications, specific ways risk assessments can be used to identify with TS are important f rom a risk standpoint and which are not outlined.

J Related concepts dealing with cumulative outage time requirements, the impact of i ' demand-stress versus time-stress related failure mechanisms in testing require-j ments, the utility and necessity of " time dependent" evaluations in revising TS are also being investigated and will be discussed. Finally, a decision f rame-i work for utilizing the regulatory decision making process for granting exten-sions and exemptions to present TS is soggested with the objective of achieving j short-and long-term safety through problem recognition and correction.

2-3

o Risk Analysis of Decay Heat Removal i Sequences During Shutdown l'

l J. Gaertner & W. Reuland Electric Power Research Institute j

Background:

, To help the nuclear utilities avoid accidents during shutdown and also to 1

provide the technical basis for resolving a major generic safety issue, the i

NSAC Generic Safety Analysis Program has an effort in the area of shutdown decay heat removal (DHR). Included in this effort are four studies that

together assess the operating experience of PWR and BWR decay heat removal j systems, determine the adequacy of decay heat removal capability at existing 1 plants, and identify potential improvements in design and operation that might ;

i be desirable.

The four investigations are reviews of both PWR and BWR industry wide operating experience and risk assessments of a PWR DHR system and a BWR DHR system. The operating experience reviews reflect the observed problems with i

DHR systems while the risk assessments extend our knowledge to include the j

4 likelihood of postulated problems with OHR systems. This paper describes the approach and conclusions of the two risk assessments, NSAC-83, Brunswick Decay Heat Removal Probabilistic Safety Study and NSAC-84, ZION Nuclear Plant Residual Heat Removal PRA.

The BWR Study:

This probabilistic study was performed on Brunswick Unit 1. Specifically, it identifies accident sequences involving failures of the Residual Heat Removal l (RHR) System, its support systems, and backup systems in their role of decay heat removal. It further identifies dominant failures which contribute to these sequences and considers the recovery options available. It also

! provides a means for ranking the importance of events by calculating an

-j expected frequency of reaching an undesirable plant state for each sequence.

The study considers sequences which originate at cold shutdown as well as at power operation.

The results of this study provide excellent insights which relate to the DHR issue. For this BWR, improvements in decay heat removal ccoability can be i achieved with minor hardware modifications, improvements in 7rocedures and

! maintenance practices, and attention to the prompt recovery c.f available

! equipment. Major hardware additions do not have apr eciably greater safety value. Specific contributors to loss of DHR Sequances are identified, and j sensitivity studies are performed to investigate uncertainty and sensitivity I to possible hardware or procedural changes.

I l

The PWR Study

i l Care was used to model the specific design, procedures and experience of the l Zion plant. Extensive use also was made of data from the Zion Probabilistic '

Safety Study (ZPSS). In order to obtain generic insight from the results of l l the model, sensitivity studies were included that assumed variations from Zion '

i l j 2-5 I i

i

--.n..m., . , , - , - , , ----n -, -..g.- , . v--,, ---,.,.. .,----~..,-.n. -,r, - - ,. . , - - - --. .

i design and procedures. Risk changes as plant conditions change during an outage. To account for the changes in plant conditions as an outage progresses, an innovativa procedure event tree approach was tailored to the~

PRA model. The procedure event trees also provide a method for modeling errors in operator actions that necessitate recovery. These actions supplement the standard PRA initiating events such as loss of off-site power which are also modeled for the shutdown plant.

The end states modeled are fuel damage, defined as inventory loss to the mid-plane of the reactor core, and three categories of cold over-pressure. The study concludes by comparison with ZPSS results that for Zion the annual risk of fuel damage from events initiating during shutedwn is less, by a factor of 5 to 20, than the risk of core damage from transients initiated at power.

However, shutdown risk is highly dependent on operator error, and wider -

uncertainty exists for the shutdown model result than for the results at power. The insight to be gained from this study is that proper procedures and modest hardware improvements appear to be effective in improving DHR safety.

This conclusion supports the results of the NSAC review of DHR operating experience.

Modeling and analysis scope differences make inappropriate a numerical comparison of Zion and Brunswick risk based on the difference in calculated end state frequencies.

i 4

i l

i s

t 9

2-6 f

Identification and Evaluation of Facilitation Techniques for Decommissioning Light Water Reactors

SUMMARY

This paper describes a study sponsored by the U.S. Nuclear Commission Regulatory to identify practical techniques to facilitate the decommissioning of nuclear power generating facilities. The objectives of these " facilitation techniques" are to reduce public/ occupational exposure and/or reduce volumes of radioactive waste generated during the decommissioning process.

The paper presents the possible facilitation techniques identified during the study and discusses the corresponding facilitation of the decommissioning process. Techniques are categorized by their applicability of being implemented during the three stages of power reactor life: design / construction, operation, or decommissioning. Detailed cost-benefit analyses were performed for each technique to determine the anticipated exposure and/or radioactive waste reduction; the estimated cost for implementing each technique was then calculated. Finally, these techniques were ranked by their effectiveness to facilitate

/ the decommissioning process.

The approach of the study was to focus on those dismantling activities that generated the greatest percentage of occupational exposure and waste generation. Based on these activities, techniques were identified that reduce exposure and/or waste volume when applied to those major activities and/or subactivities.

Each of the identified techniques were evaluated with regard to safety, feasibility and assurance that the objective of each technique was consistent with the goals of the study.

Implementing scenarios were developed to maximize the benefits of each technique in reducing exposure and waste generation. This process clarified each technique and provided a baseline for a safety / feasibility review.

Review criteria were formulated to standardize the evaluation of all identified techniques. Techniques failing identified safety criteria were not immediately rejected. An evaluation was first performed .to determine if corrective action would alleviate or correct the concern, and if so the technique was revised and the evaluation process repeated.

The safety / feasibility review of the technique eliminated the nonviable techniques. An engineering assessment of the remaining techniques was then performed to objectively evaluate the practicality of their implementation.

3-1

Thsso facilitation techniques were grouped into two categories:

those being " Good Practice" techniques and those techniques supported by detailed cost benefit analyses.

Good Practice Facilitation Techniques:

These techniques provided benefits that could not be l quantified because of the generality of the technique.

However, engineering judgment indicated that qualitative benefit could be obtained should the technique be implemented.

Facilitation Techniques Requiring Cost Benefit Analysis:

These techniques required a detailed analysis to determine their viability. A detailed cost benefit analysis provided quantitative results, i.e., manRem and waste volume reductions.

Some of the key techniques included in the study are:

1. Segregation of wastes 6. Enclosed cable trays

2. Waste compaction 7. Non-embedment of pipe

3. Incineration 8. Tracks for remote cutters

4. Intact removal of items 9. Preplaced blast holes

5. Prepolishing of components 10. Modular bioshield The techniques were rated by their potential savings and costs associated with those savings. The initial investigation produced a list of 104 potential techniques. Of these, 68 were determined to facilitate the decommissioning process. Forty nine were then identified as Good Practice techniques, those not lending themselves to rigorous cost benefit evaluation but providing some benefit in reducing exposure or waste volume. The remaining 19 techniques required that detailed cost benefit analyses be performed to determine and quantify their potential ~

benefits.

The techniques evaluated in this study were specific to the scope of the facilitation technique and the site characteristics of the reference station used in the analysis. These results will vary on a plant specific basis and with any change in the scope of technique application. Each of the techniques has the potential of providing dose or waste reduction benefits but should be reviewed on a site-specific basis to make that determination.

l 3-2

e IMPACTS OF DECONTAMINATION 01 SOLIDIFICATION AND WASTE DISPOSAL

P. L. Piciulo and J. W. Adams Department of Nuclear Energy Brookhaven National Laboratory Upton, New York 11973

SUMMARY

The Nuclear Regulatory Commission (NRC) is concerned with evaluating the effectiveness and safety of chemical decontamination processes which are being proposed for light water reactor (LWR) primary systems as a means of reducing occupational exposure and' ensuring continued safe operation. The areas of concern being addressed are: the type, volume and toxicity level of the rad-waste streams generated by decontamination and their subsequent management at the plant and at the disposal site.

Because of the large amounts of chelates or complexing agents required for a full system decontamination, it is desirable to determine if there are methods which would convert these reagents to more innocuous forms. The ob-jectives of the program at Brookhaven National Laboratory (BNL) are to iden-tify the information and conduct the tests necessary to aid the NRC in making regulatory decisions on the solidification and waste disposal aspects of chemical decontamination processes. In particular, this program focuses on in plant methods for converting decontamination wastes to more acceptable forms prior to disposal.

The BNL program has provided an evaluation of potential decontamination processes, the wastes generated and potential ' waste management practices.

Subsequent laboratory studies have confirmed that incineration and acid diges-tion are capable of destroying selected decontamination reagents. Further, this work has assessed the solidification of simulated decontamination resin wastes in cement and vinyl ester styrene.

The laboratory evaluation of processes for destroying decontamination reagents have continued and included wet air oxidation. A discussion is given in this paper of data which indicate that wet air oxidation can ef fectively degrade the organic complexing agents typically used for chemical decontamina-tions. However, the resulting waste stream may need to be evaluated before it can be effectively managed prior to disposal. The solidification studies have also been extended to include treatment of a liquid decontamination waste stream. In a limited study, the NS-1 reagent was solidified in cement. A free standing monolith was obtained having no associated free liquid and good mechanical strength both prior to and af ter immersion in water.

Work carried out under the auspices of the U.S. Nuclear Regulatory Commisison.

3-3

i-t--

}i The scope of future work in this program will be outlined. A laboratory evaluation of bitumen as a solidification medium for decontamination wastes

! .will be performed. Additionally, tests will be carried out to assess the com-patability of_ container materials with decontamination wastes. The objective of the work is to generate data that will aid NRC in assessing problems that may be encountered with the long-term storage or disposal of such wastes.

1, i

i i

l l

1 l

?

4 f'

u f

f i-r 3-4

l EVALUATION OF NUCLEAR FACILITY DECOMISSIONING PROJECTS PROGRAM - STATUS R. L. Miller UNC Nuclear Industries, Decommissioning Programs Department StM ARY The Evaluation of Nuclear Facility Decommissioning Projects (ENFDP) Program is being undertaken by the NRC to compile and evaluate the activities of ongoing reactor decommissioning projects. Assessment and evaluation of the methods, impacts, and costs will provide bases for evaluating licensees decommissioning proposals and for future decommissioning direction and regulation.

The computerized Decommissioning ~ Data System (DDS) developed for the program to collect, analyze, and report actual decommissioning data currently has information from 12 reactor projects and 12 reference reactor studies performed by Pacific Northwest Laboratory (PNL). Of the actual reactor decommissioning projects, seven are currently in progress; with data collection and analysis being the major ongoing ef fort. The ongoing projects included in the ENFDP Program are Humboldt Bay, Shippingport, Three Mile Island-Unit 2, Northrop TRIGA, Lingen, Niederaichbach, and Gundrammingen.

Some initial comparison studies were made on the decommissioning of research reactors. Parameters considered included total curies, total power generated, cost, exposure, waste volumes, and burial costs. Where possible, actual versus estinated data were obtained to allow licensees and other planning organizations to normalize their decommissioning planning estimates. The research reactor decommissioning projects included in the comparison studies were the North Carolina State University R-3 reactor, the Ames research reactor, the Northrop TRIGA reactor, and the PNL reference research reactor.

3-5

LBR Spent Fuel Rod Behavior During Long-Term Ory Fuel Storage Conditions C. S. Olsen Idaho National Engineering Laboratory EG&G Idaho Inc.

P. O. Box 1625 Idaho Falls. Idaho 83401

SUMMARY

/,

As wet storage pools at reactor sites become filled, dry storage in metal 4

casks, cement silos, or dry wells is considered to be a viable l alternative, since dry storage is not expected to produce seconuary waste nor require maintenance. Because perforated rods that occur in-reactor are not routinely isolated, some rods with cladding perforations that are stored in dry environments may develop cladding failures from the

)

continued growth and propagation of cladding perforations, and i contamination may result from release of fuel particulate, spallation of the crud coating from stresses imposed on the cladding, or fission gas i release.

A testing program was conducted to investigate their long-term stability l characteristics of nine commercial PWR and BWR spent fuel rods under a variety of possible dry storage conditions. The objective of this project is to provide the NRC with experimental information with which to evaluate the results of short-term, high-temperature tests, to establish licensing positions with regard to long-term, low-temperature (<250 C) spent fuel rod behavior during dry storage, and to establish radioactive 4 contamination limits arising from spallation of cladding crud.

Four intact and five intentionally defected spent fuel rods were heated d in a furnace to simulate temperatures that would be expected to occur during dry storage conditions. During testing, temperatures ranged between 217 C and 230 C for different periods of time up to a total of

, 13,200 hours0.00231 days 0.0556 hours 3.306878e-4 weeks 7.61e-5 months , and the atmospheres surrounding each test rod was either l air or an inert mixture of argon and helium. Selected test rods were j nondestructively examined at interim times of 2,235, 5962, 13,168 hours0.00194 days 0.0467 hours 2.777778e-4 weeks 6.3924e-5 months .

i One BWR fuel rod which ruptured was destructively examined and replaced l with another defected BWR fuel rod after the second interim examination.

1 During the tnird interim examination, the fifth intentionally defected l rod, the three other defected rods, and the remaining intact fuel rods j were examined. Six of the eight remaining fuel. rods were also destructively examined to determine the extent of fuel rod degradation under long-term, low-temperature, dry fuel storage conditions. In defected fuel rods, the penetration of the defects through the cladding wall, the location of the defects with respect to pellet / pellet interfaces, the extent of interaction between the cladding and the fuel, l

J the extent of UO2 oxidation (i. e., the oxidation front and oxide form), and the extent of crud or oxide deposits on the cladding surface were determined.

As would be expected, a minute amount of fuel was released from the artificially defected fuel rods. However, isotopic gamma scanning I

l 3-7 i

- .. _.. ._ _ _. m. - - , . , . . -

. ,-r_.. - . _ --. . , . _ _ - _ - , , - ,..,-...._y--._,,-

i indicated fuel release from an undefected BWR fuel rod heated in argon which was also confirmed by delayed neutron measurements. This release is j indicative of fuel rod failure but the defect may nave been too small to !

be detected visually since it was not discovered during the visual l

examination.

4 During heating in an unlimited air atmosphere at temperatures between 217 l

C and 229 C. BWR fuel rods have developed cladding cracks at some, but '

not all, artificial defects. Although some fuel oxidation was observed at the defects in the center of a fuel rod, the cladding failures occurred l at either the top or bottom portions of the fuel rods as a result of the j volume expansion associated with significant fuel oxidation.

In BWR fuel, the oxidation was localized and limited to within 7 to 8 cm of the defect whereas in PWR fuel the oxidation extended 30 cm from the defect. No cause could be identified for the localized oxidation in 8WR I fuel.

I The gas release from from a defected BWR fuel rod heated in argon indicated some oxidized fuel wnich was confirmed by the destructive :

examination. Limited fuel oxidation also occurred with one of the j intentionally defected PWR rods heated in an inert atmosphere. Although j sealed, the capsules still had a finite leak rate from the mechanical j seals which allowed air to penetrate the sealed capsules. The oxidation i

rate is limited by the partial pressure of oxygen in the capsule. After the 13,000-hour heating cycle, the extent of oxidation was not

! sufficient to form large enough quantities of U38 0 to cause fuel rod j

expansion. The fuel was oxidized to U 04 9 or perhaps U 03 7 with a j small amount of U38 0 i

The storage of PWR fuel rods in unlimited air at 229 C did not cause any j deformation of the fuel rod cladding. The fuel rod heated in an unlimited i

I air atmosphere was oxidized, but not to the extent that BWR rods were; and the oxidation extended up to 65 cm from the defect. The fuel was 1

oxidized to U 03 7 with a small amount of U38 0-I Similarly to the defected BWR fuel rod heated in argon , the PWR fuel rod i

heated in argon was also oxidized; and like the BWR rod, tne extent of i

I oxidation was limited by the partial pressure of oxygen. The oxidation rates of PWR fuel is slower than that of BWR fuel.

I With both PWR and BWR fuel, the oxidation paths are along the gas gap and cracks in the fuel so that the pellets are oxidized by oxidation of the i individual pellet fragments. This process enhances the oxidation of an individual pellet over what would be expected for the application of rates of individual fragments determined from out-of-pile tests on intact j fuel pellets.

i i The results from this study have been used to establish general criteria l regarding dry spent fuel storage licensing, but additional work is j required to establish specific design criteria for implementing dry spent

, fuel rod storage.

l \

3-8

" Lessons Learned From a NUREG-0737 Review of High Range Effluent Monitors and Samplers" Andrew P. Hull

and John R. White **

Safety and Environmental Protection Division Brookhaven National Laboratory, Upton, NY 11973

- US NRC Region 1, 631 Park Ave, King of Prussia, PA 19406 Among the lessons learned f rom the TM1-2 accident were 1) that conven-tional gaseous effluent monitors lacked sufficient range to measure possible pos t -a ccident concentrations, 2) that they were typically situated where they would be inaccessible for sample retrieval under post accident conditions and,

3) that licensees were inadequately prepared to handle and analyze post-accident level samples of particulates and/or radioiodines. Subsequently, the NRC issued requirements for the installation of high range gas monitors and for the collection, transport and analysis of samples of particulates and radioiodines in post accident concentrations. They are set forth in NUREG-0578, NUREG-0660 and NUREG-0737, items II.F.1. Attachments I and 2, with a completion date of January 1, 1982. l l

In 1983, the NRC's Region I contracted with Brookhaven National Laboratory to supply technical assistance in the performance of on site reviews of the 1,nstalled monitors and sampling arrangements at the twenty operating nuclear power stations within the Region.

To date, eighteen stations, which include twenty four reactors, have been reviewed. Thirteen of the stations have installed vendor f abricated, micro-processor controlled units which include both monitoring and sampling capa-bilities. Six of these employ GM, five employ Cd-Te, and two Ge-Li noble gas detectors. Of the remaining, four employ ion chambers mounted external to ducts. One employs an organic scintillator. All of the samplers have been located at distances of from 50-250' from their associated effluent release points (stack or vent), thereby occasioning long horizontal sample lines.

Most of the installed gas monitors have been found acceptable in terms of the requirements of NUREG-07g7, Iteg II.F.1. Attachment 1, which includes an upper range capability of 10 uC/cm for undiluted release paths. Three stations have installed normal range monitors to which a high-range channel has been added. However, their associated micro computers appear to be vulnerable to radiation damage at an integrated dose that could be reached under post accident conditions, making them unsuitable for the intended purpose of meeting NUKEG-0737 requirements. In only a few instances, have the vendors and/or licensees developed sufficient calibration information to enable unambiguous conversion of monitor readings (in terms of cpm or. mR/hr) to congentrations of a changing post-accident mix of radiogases (in terms of uCi/cm or uCi/sec).

3-9

The requirements of NUREG-0737, II.F.1, Attachment 2 include "representa-tive sampling per ANSI N. 13.1.-1969. Subsequent to issuance of this require-ment, it became apparent that its guidance was not fully adequate to charac-terize the behavior of particulates and radioiodines in long horizontal sample lines. An approach of making actual system tests to determine line losses was endorsed by the NRC Staff in 1982.

At the time of our reviews, no licensee had performed sufficient testing to document the extent of these losses under accident conditions. It has become apparent that further generic research is needed to adequately under-stand the behavior of radiciodines in such lines under a range of post-accident conditions (particulate loadings, temperatures and relative humidi-ties).

There is also an inherent conflict between the desirability of a high flow rate to minimize deposition in these long sample lines and that of establishing a very low flow rate, so as to minimize the activity of the collected particulates and/or radioiodines in order to facilitate their transport and analysis. With the exception of the adapted normal range monitors, all of the " packaged" monitor-samplers address this problem by the provision of a separate high range sample path at a reduced flow rate. One model utilized ,a separate small diameter nampling line which appears subject to large deposition losses. The remaining two incorporate a second isokinetic probe in the main sampling line with a very short reduced flow sampling line to the particulate and iodine sampler itself.

These latter packaged units also provide for the by pass and purge of their low-range detectors when mid-to-high range concentrations of radiogasses are present in their incoming flow. However, none of the reviewed in-house engineered systems contained a similar feature, thereby putting their detec-tors at risk of being stressed beyond subsequent recovery as post accident concentrations eventually decrease.

1 I

i l

3-10 l

SEMISCALE SECONDARY TRANSIENT INVESTIGATIONS:

RESULTS FROM SEMISCALE M00-2C FEE 0 WATER AND STEAM LINE BREAK TESTS T. J. Boucher Idaho National Engineering Laboratory This paper presents a discussion of the results from steam generator main steam line and bottom main feedwater line break experiments performed in the Semiscale Mod-2C facility. Tests S-FS-1 and S-FS-2 simulated double-ended offset shears of the main steam line upstream and downstream, respectively, of the flow restrictor with compounding factors. Tests S-FS-6, S-FS-7 and S-FS-11 (not yet performed at this writing) simulated 100, 14.3 and 50%, respectively, breaks downstream of the main feedwater line check valve with compounding factors. The Semiscale Mod-2C system consists of the Mod-2B system with a new Type III single loop steam generator which allows increased instrumentation while providing a more prototypical simulation of a full scale steam generator.

The background for these tests is discussed in brief below. A large number of assumptions and simplificaticns are employed when performing secondary side transient calculations. Main steam line break calculations performed for pressurized water reactor (PWR) Final Safety Analysis Reports (FSARs) have predicted primary fluid overcooling raising concerns regarding possible pressurized thermal shock (PTS) occurrences. Bottom feedwater line break calculations performed for the Combustion Engineering (C-E)

System B0 FSAR have predicted peak primary system pressures in excess of 110% of the system design pressure. These calculations employed a large number of assumptions and simplifications. Foremost for the former were assumptions regarding transient separator performance which were considered to be conservative. Foremost for the latter was an assumption regarding the degradation of primary-to-secondary heat transfer with secondary inventory which was considered by C-E to be highly conservative. Although the FSAR calculations are believed to be conservative, the degree of conservatism remains unanswered. Quantification of the degree of conservatism requires performing best estimate calculations utilizing a computer code which has been assessed for these type of events. To provide 4-1

data to help answer these concerns, the steam line and feedwater line break tests were performed with conditions consistent with, or scaled from, those used for Westinghouse and C-E System 80, respectively, FSAR calculations.

Test S-FS-1 and S-FS-2 results show the same basic phenomena as predicted by Westinghouse FSAR calculations. The Westinghouse FSAR calculations assumed that the steam generator separator would provide perfect separation during the transient. The S-FS-1 and S-FS-2 test results indicated almost perfect separation during the transient. Thus, the FSAR calculation assumption regarding separator performance is !

marginally conservative for the Semiscale Type III steam generator. During the overcooling of the primary fluid system, the Semiscale Mod-2C system indicated significant cooling for both tests; however, the minimum primary fluid temperatures reached during the tests do not indicate temperatures sufficiently low to allow PTS phenomena to occur. Test S-FS-6 and S-FS-7 results show, and Test S-FS-Il results are expected to show, the same basic phenomena as predicted by the C-E FSAR calculations. The C-E FSAR calculations assumed that the affected loop steam generator primary-to-secondary heat transfer remained at 100% of the initial value until the secondary liquid inventory was depleted. The heat transfer was assumed to then reduce to zero instantaneously. This same trend was indicated by the S-FS-6 and S-FS-7, and is expected to be indicated by the S-FS-ll, test results.

Thus, the C-E FSAR calculation assumption regarding the degradation of heat transfer with inventory is not conservative for the Semiscale Type III steam generator. During, and following, the loss of the affected loop heat sink, the Semiscale Mod-2C primary system indicated, or is expected to indicate, a significant pressurization for all three tests; however, the peak primary system pressures reached, or expected to be reached, during the tests do not indicate (even when scaled up to adjust for the fact that the Semiscale system loses only one-half of the sink that the C-E system loses) pressures in excess of the 110% of design value l pressure limit. The Semiscale tests have, or will, provide valuable insight into the phenomena occurring during secondary transients. The real value of the tests, however, will be in providing valuable data for code assessment and/or development which will then allow for higher-confidence best estimate calculations and final quantification of the degree of conservatism inherent in FSAR calculations.

4-2

l SEMISCALE LIQUID H0 LOUP INVESTIGATIONS:

A COMPARIS0N OF RESULTS FROM SMALL BREAK LOCA TESTS l PERFORMED IN THE SEMISCALE M00-2A AND M00-2C FACILITIES G. G. Loomis T. J. Boucher Idaho National Engineering Laboratory This paper presents a comparison of results from small break-loss-of-coolant accident (SBLOCA) experiments performed in two different versions of the Semiscale facility. Tests S-LH-1 and S-LH-2 were performed in the Mod-2C facility while S-UT-6 and S-UT-8 were performed in the older Mod-2A facility. All four experiments were 5% SBLOCA simulations with various downcomer to upper head core bypass flow resistances. Both Semiscale Mods were volume-scaled representations of a four loop pressurized water reactor (PWR) that simulate most of the major features of a PWR including U-tube steam generators, pressure vessel with electrically-heated core, pumps, pressurizer, and loop piping. For both Mods the volume was approximately 1/1705 that of a PWR, while elevation was generally scaled on a 1:1 basis.

By way of historical background, Tests S-UT-6 and S-UT-8, performed in the Semiscale Mod-2A facility, exhibited a possible strong relationship between downcomer to upper head core bypass flow resistance and accident severity. Test S-UT-6 had a 4.0% bypass flow (defined as the percentage of the total full power operating condition core flow which flows from the downcomer to the upper head, bypassing the core) while S-UT-8 had a 1.1%

bypass flow. During S-UT-8 the vessel collapsed liquid level was depressed to the bottom of the core resulting in core heat-up while during S-UT-6 the vessel collapsed liquid level was only depressed to 220 cm above the core bottom with no core heat-up. Both experiments had essentially identical initial and boundary conditions; however, other system hardware changes in addition to the core bypass flow resistance were made between the two experiments that could have affected the difference in core level 4-3 n

depression. (These changes in hardware were made between S-UT-6 and S-UT-8 to provide a test bed in S-UT-8 for a vendor vessel liquid level probe.)

Thus, the comparison between S-UT-6 and S-UT-8 to assess the effect of core bypass flow resistance on SBLOCA severity was not clear. In addition since S-UT-8 was designed as only a test bed for the vendor liquid level system, emphasis was not placed on measurement of certain boundary conditions and other parameters. Therefore; to provide a clean comparison to examine the effect of downcomer to upper head core bypass flow resistance and to provide a state-of-the-art simulation with the instrumentation focused on SBLOCA phenomena, S-UT-6 and S-UT-8 were approximately duplicated in the ncw Mod-2C system (Tests S-LH-1 and S-LH-2).

Tests S-LH-1 and S-LH-2 were essentially identical with the exception of the core bypass flow resistance (Test S-LH-1 had a 0.9% bypass flow and Test S-LH-2 had a 3.0% bypass flow). The Mod-2C facility is a state-of-the-art facility designed specifically for SBLOCA experiments (the Mod-2A facility was an intermediate step in the transition from large break facilities to small break facilities). Special improvements in the Mod-2C facility include comprehensive heat loss make-up techniques and better scaled steam generators.

Test S-LH-1 and S-LH-2 results show the same basic SBLOCA phenomena as S-UT-6 and S-UT-8 including a rapid depressurization to fluid saturation conditions, pump seal formation and core liquid level depressions, seal clearing, and core liquid boiloff mitigation by accumulator injection.

Most importantly, however, both sets of experiments exhibit a net hold-up of fluid above the cold leg which is affected by the hydraulic resistance i of the core bypass flow path. Comparison of S-LH-1 and S-LH-2 test results showed a more severe level depression for the higher core bypass flow resistance. This was the same trend indicated by the S-UT-6 and S-UT-8 test results; therefore, the system differences that might have affected the comparison between S-UT-6 and S-UT-8 were of second order. For both S-UT-8 (1.1% bypass) and S-LH-1 (0.9% bypass) the core collapsed liquid level depression below the level corresponding to the bottom of the pump suctions was possible only because of the net positive head of fluid held-upinpositionsabovethecoldleg(postnotablythesteamgenerator U-tubes).

4-4 2

The Results of the ROSA-IV LSTF Small-Break LDCA Experiments l

K. Tasaka, M. Kavaji, M. Osakabe and Y. Koizumi l Japan Atomic Energy Research Institute Tokai-mura Ibaraki, Japan 1

i The test results are reported on three small-break I.OCA experiments recently conducted at the Large Scale Test Facility (LSTF) of the ROSA-IV program. In all of these tests, a break was located horizontally at the side of the cold leg piping (207 mm in diameter). The break sizes tested were equal to 10.0*4, 5.0*4 and 2.5f4 of the scaled (1/48) flow area of the reference PhE's cold leg.

All tests were conducted from essentially the same set of initial conditions: primary pressure of 15.5 MPa, secondary pressure of 7.4 MPa, and core power of 10.0 MW. For the present tests, the loss-of-offsite power concurrent with reactor scram was assumed and the EI:CS pump flow rates, ie. HPIS and 1.PIS, simulated one of two pump's capacity for each system in the reference PWR. Additionally in the 2.5f4 break test, HPIS was assumed to be inoperative for 1200 s after break. The core power decay curve used in the 10% break test was that of ANS standard curve, but a new decay curve including the contribution of fissions due to delayed neutrons was used in the 5f4 and 2.514 break tests.

In all of these tests, the primary system depressurized rapidly immediately after the break occurred. However, the depressurization rate in the remainder of the test depended strongly on the break size. At a primary pressure of 12.Sf7 M?a, reactor scram occurred and the steam generator (SG) feedvater and main steam lines were isolated. At the same time, the core power was tripped and SG auxiliary feedvater was turned on. As the primary pressure further dropped to 12.27 Moa, an SI signal was sent with subsequent activation of HoIS, accumulator and 1.PIS at respective pressure setpoints.

In both the 10*4 and 5 14 break tests, HPIS was activated early in the transient (23 s and 30 s after break, respectively) and the rest of ECCS operated satisfactorily to prevent the major uncovery of the core. However, partial and temporary core dryout was observed early in the 5*4 break test, because of the core . liquid level depression due to liquid holdup in the SG U-tubes like' in Semiscale S-lTT-8 test. This manometric effect was terminated and the core liquid level recovered as soon as the loop seals cleared in both loops.

Though a similar depression in the core collapsed liquid level was noted in the 10*4 break test, the fuel rod te=perature excursion was not observed because the core was kept entirely covered with the two-phase mixture.

In the 2.5f4 break test, a slight core liquid level depression with insignificant fuel rod temperature excursion was observed prior to the major core uncovery due to boiloff. Because of the assumption that HPIS remains inoperative for 1200 s after break, the top half of the core was completely uncovered and the fuel rod temperatures rose rapidly at about 580 s after break.

Although accumulator injection occurred at 835 s after break, the fuel rod temperatures continued to rise. When the maximum fuel rod temperature reached Er73 K, the core power was tripped off manually in order to prevent damage to the core instrumentation.

The effectiveness of ECCS in preventing the major core uncovery was demonstrated in this series of tests for a cold leg small-break LOCA. At the same time, the manometric effect due to liquid holdup in the SG U-tubes was observed to cause core liquid level depression early in the transient, and even lead to temporary and partial core dryout in the 514 break test.

4-5 a

PKL Reflood Tests Including End-of-Blowdown R.M. Mandl, B. Brand, 11. Watzinger Most experimental programs of the '70s and early '80s concerned with simulating large break LOCAs were divided into two stages.

The first stage, so called blowdown, was usually simulated in full-pressure power / volume scaled test facilities with nine to one hundred rods capable of producing the heat flux corresponding to full power operation. The second stage of a large break LOCA, the so called refill and reflood, was also simulated in power / volume scaled facilities with the number of rods reaching from 300 to 2000 simulating only the decay heat. The initial conditions for the latter type of tests were pressure of about 4 bar (pressure equalization with containment) and heater rod temperatures stipulated by licensing procedures which, in turn, were based on conservative assumptions - e.g. no credit was taken for either blowdown residual water or for ECC water already injected in the latter phase of blowdown (p c 40 bar).

To simulate refill and reflood more realistically the Prim 5r-kreisinufe (PKL) test facility, designed for 40 bar, was adapted to simulate refill /reflood preceded by End-of-Blowdown (E0B) thereby creating more realistic boundary and initial conditions.

The test were started at 40 bar with steam-filled system by opening the break valve and injecting hot water into the pipes to the left and right of the break as well as into the upper plenum for cold leg break and into downcomer for hot leg break.

The objective was to achieve, on readiing 26 bar (start of accumulator injection),

- pressure gradient

- core mass flow rate

- fluid density distribution t - clad temperature

- containment backpressure similar to what these would be in a PWR after experiencing a complete blowdown starting at 160 bar. The prevailing conditions at 26 bar were precalculated by system codes TRAC and DRUFAN.

The conditioning procedure consisting of injecting hot water to the left and right of the break to control the pressure gradient and simultaneously injecting water into the upper plenum to effect the correct mass flow direction and distribution proved practicable and was succesfully used.

.-7 A

The E0B test series (PKL IIB) was completed in June 1985.

It included tests with breaks located in cold as well as hot legs, best-estimate and licensing conditions. '

Apart from tests with combined injection there were one test i each with cold leg break and cold leg injection only as well as hot leg break and hot leg injection only.

The first analysis shows the following trends:

A number of heater rod thermocouples quench in the E08-phase confirming the fact that presence of water in the core ( x < 1. 0 ) ,

particularly at elevated pressures, produces a considerable cooling effect in the unwetted region leading to faster quenching.

Even more important is the fact that for those rods which do not quench during E0B the temperature rise is considerably lower (5 - 50 K) in contrast with comparable refill /reflood tests starting with x = 1.0 where the temperature rise amounted up to 200 K. The results from tests with cold leg injection only show that during the refill phase the temperature rise is less than adiabatic.

For cold leg break and combined injection, hot leg oscillations take place during E0B in the broken loop only. The hot leg of the single intact loop experiences high negative flow due to surge line depressurization; during this phase the flow in the double loop is almost negligible. During reflood all three hot legs show periodic oscillations. The highest amplitude is recorded in the broken loop where subcooled water penetrates into the steam generator tubes and partly evaporates thereby reinforcing the oscillations. The single and double loops show similar oscillations in horizontal sections of their respective hot legs, however hardly any water enters the steam generators.

Oscillations in the hot legs are transmitted through the upper plenum and influence, among others, a periodic breakthrough of subcooled water into the core. In certain regions the subcooled water penetrates the whole length of the core.

The tests with a break in the hot leg using hot leg injection only or combined injection showed no increase beyond the initial cladding temperature. In spite of countercurrent flow in the upper plenum the hot-log-injected water penetrated into the core.

4-8 a

INTEGRAL SYSTEM TEST (IST) PROGRAM FACILITY SCALING STUDY T. K. Larson l

l Idaho National Engineering Laboratory The Integral Systems Test (IST) Program is a jointly funded research program initiated in 1982 by government and industry to help provide information needed to address issues raised by the accident at the Three Mile b land (TMI) Nuclear Power Station. Joint funding is provided to the program by the United States Nuclear Regulatory Comission (NRC), the Babcock and Wilcox (B&W) company, the Babcock and Wilcox Plant Owners Group Organization, and the Electric Power Research Institute (EPRI). Research in this program is a combired experimental / analytical effort geared toward the study of B&W type Nuclear Steam Supply System (NSSS) behavior during off-normal operating circumstances. Major emphasis is placed on the experimental approach through use of small-scale, nonnuclear, integral thermal-hydraulic test facilities to simulate postulated NSSS off-normal operating conditions. The major program objective is to provide experimental data for use in assessing analytical techniques used in calculating full size B&W type NSSS response.

The Multiloop Integral Systems Test (MIST) Facility (a full height, full pressure, volume scaled (approximately 1/817) 2 x 4 loop system) is the central integral facility supported under the auspices of the IST Program. This facility is being designed and constructed by B&W at the Alliance Research Center in Alliance, Ohio. Two other integral systems are being designed and constructed under funoing sources separate from the IST Program. These includa SRI-2, funded by EPRI and being built by SRI International at their Menlo Park, Ca. facilities, and the University of Maryland Facility, funded by NRC and being built by the University of Maryland at the Collage Park, MD. campus. Both of these systems are reduced height, reduced pressure, 2 x 4 loop facilities with the major 4 t)

differences being size and maximum operating pressure. SRI-2 and UMCP are approximately 1/298 and 1/500 the fluid volume of a B&W 177-FA lowered loop NSSS, respectively. The maximum operating pressures in the two facilities are 0.689 and 2,07 MPA respectively. All three facilities are considered to be integral in the sense that they contain most of the components germane to the primary side of a B&W reactor system. While it is generally agreed that the SRI-2 and UMCP systems do not have the same degree of simulation potential as the MIST system, both will produce data that should compliment MIST results and support the IST program in general.

Each of the three integral facilities mentioned above was conceived, designed, and is being constructed under a different set of constraints and assumptions. Such constraints and assumptions include funding limitations, design basis assumptions such as thermal-hydraulic scaling criteria and overall desires with respect to facility capability, facility testing methods, and schedules, etc. As a result, each facility is different in terms of hardware geometric parameters as dictated by the scaling criteria utilized in its designs and physical constraints introduced by conscious choice, construction material limits, support system limitations, or personnel safety codes.

As a result of differences in design and scaling approach and facility operational limitations in the systems described above, there is a desire and need to investigate what the inter-relationships between the facilities are and how facility results will be complimentary. Such an investigation is necessary so that a unified global approach to resolution of the issues forming the basis for the IST program can be ef fected. Examples of questions posed regarding this issue include:

1. What are scaling and design rationales used to construct each facility?

2. How does each facility differ from an ideally scaled facility (not subject to physical construction constraints) designed according to the critoria used for that facility?

4-10 L ^

3. What are the potential distortions relative to expected NSSS behavior introduced by facility limitations and atypicalities?

4. In light of each facilities scaling approach, limitations, and potential distortions:

a. Is there a method or rationale for best relating scaled facility behavior to plant behavior?

b. Is there a potential for any of the facilities to produce data that would (because of facility distortion induced behavior) unnecessarily complicate computer code calculational requirements?

5. What is the best method for conducting a " counterpart" experiment among the three facilities to facilitate data comparison and assess the influence of the different system scaling techniques?

The purpose of the work presented in this paper is to provide as complete as possible answers to the questions posed above. Ultimate work objectives are:

1. To provide a reasonably complete and centrally located description of the scaling philosophy, geometric hardware parameters, and limitations of each of the integral facilities contributing to the IST program.

2. Provide insight on rationale for relating the three facilities and data they produce to each other and to a plant.

3. Recomend an experiment (s) and associated boundary and initial conditions to be conducted in each facility 50 that the influence I of system scaling differences on phenomena can be assessed.

4-11

l CONTINUING INTEGRAL TESTING CAPABILITY -

APPROACH AND SCALING STUDY ;

l l T. K. Larson J. S. Martinell K. G. Condie l 4

IDAHO NATIONAL ENGINEERING LABORAT0iu

! Water Reactor Safety Research activities conducted over the past two decades l have demonstrated the usefulness and viability of a combined analytical experimental l approach for the resolution of regulatory and technical issues involving nuclear a

power plant operation. Both separate effects experiments and integral systems j such as LOFT, FIST, Semiscale, and FLECHT have been used in the development j and assessment of analytical tools (PWR-TRAC, BWR-TRAC, RELAP5, etc.) used in the calculation of power plant response.to abnormal operating circumstances.

! Since all of these facilities are scheduled to be shut down by FY-87, it is an obvious concern that within the next few years there will not be any readily j available thermal-hydraulic test facilities in the United States to support j regulatory needs or other technical issues. In the interest of maintaining ,

l a capability for conducting integral system experiments in the years beyond ;

FY-87, the Nuclear Regulatory Comission has requested the Idaho National i

! Engineering Laboratory to evaluate from both cost and technical bases the

! options available for maintaining said capability.

I i

j Several possibilities exist for maintaining a continuing integral system testing l capability. In general, desirable criteria for maintaining capability includes:

i

Minimizing initial costs and operating cost.

j

Any new facility should be designed to the minimum scale size necessary I while providing testing capability adequate for a wide range of transient ,

?

) simulations, j

The scaling basis for any new facility must be defendable. !

Test turn around must be rapid. !

] !

l i i 4-13

i

Facility versatility should be maintained so that existing vendor geometries and proposed new geometries can be accommodated.

The facility should provide for study of operational issues.

The approach used in this study is designed to identify the major needs for future testing, the various options available to NRC, provide a technical evaluation of each option, and provide a cost estimate for each concept whether it be use of existing facility hardware, or design and construction of new l hardware. The approach used includes a thorough review of existing plant calculations to establish a data base of phenomena important from a simulation viewpoint and to establish a base plant configuration for each vendor design.

In parallel with this activity, methods of scaling used in the design of integral thermal-hydraulic exptriment systems will be reviewed to identify the techniques

and the advantages and disadvantages of these techniques.

After base plant information for each vendor design is established, geometric parameters for facilities scaled according to one or more of the scaling methods mentioned above will be calculated so that a scaling evaluation of that method for a particular vendor design can be made. Full height, reduced height, full pressure, and reduced pressure facility designs will be considered as well as local scaling effects such as flooding, critical flow, flow regimes,

, and heat transfer, etc. Working fluids other than water will also be considered in light of reduced pressure facility design. Cost estimates and cost-benefit analysis will be performed for the concepts most promising from the scaling evaluation analysis. Computer code calculations will aisc be performed for the most' promising concepts to support the scaling analysis and for comparison to selected transients defined in the plant transient data base.

l The work described above will be conducted for each of the four U. S. Vandor designs. Efforts relating to the Westinghouse and Babcock & Wilcox design will be completed in late FY-85. The General Electric and combustion engineering plant designs will be considered in early FY-86. The final results of the l study will be formally documented and recommendations provided to NRC so that j a timely decision on a continuing experimental capability concept can be made.

1 i

4-14 i

Structural Load Combinations H. Hwang and M. Reich Brookhaven National Laboratory B. Ellingwood National Bureau of Standards M. Shinozuka Columbia University Summa ry This paper presents the latest results of the on-going program entitled,

" Probability based Load Combinations For Design of Category I Structures",

currently being worked on at Brookhaven National Laboratory (BNL) for the Of fice of Nuclear Regulatory Research, U.S. Nuclear Regulatory Commission.

The objective of this program is to develop a probabilistic approach for safety evaluations of reactor contairinents and other seismic category I struc-tures subjected to multiple static and dynamic loadings. Furthernure, based on this probabilistic approach, load combinations for the design of seismic category I structures will also be established.

A probability-based reliability analysis nethod has been developed for the safety evaluation of shear wall structures. The shear walls are modeled by stick rodels with beam elenents, and may be subjected to dead load, live load and eartnquake during their lifetimes. In addition to the appropriate prob-abilistic nodels for dead load and live loads, the earthquake load is assuned to be a segnent of a stationary Gaussian process with zero-nean and a Kanai-TaJimi power spectral density f unction. The seismic hazard at a site is also included in the reliability analysis. Both shear and flexure limit states are defined analytically. The flexure limit state is defined according to ACI ultimate strength analysis principles, while the shear limit state is established from test data. Illustrative examples are given to denonstrate the method and the applications.

Utilizing the reliability anahiis method described above, load combina-tion criteria for the design ci shear wall structures have been established.

The proposed design crit:Lia are in the load and resistance f actor design (UlFU) format. In order to test whether the proposed criteria acet the reliability-based perfonnance objectives, four representative structures are selected using a Latin hypercube sampling technique. These representative structures are designed using trial load and resistance factors. Then, the reliability analysis nethod is employed to assess their reliabilities. An ob-Jective function is defined and a minimization technique is used to find the optimum oad factors. In this study, the resistance f actors for shear and flexure vid load factors for dead and live loads are preassigned to simplify the minimization work. The load factor for SSE is determined for target limit state probabilities of 1.0 x 10-6 or 1.0 x 10-5 during a lifetime of 40 years.

5-1

Standard Prob,lems for Computer Codes l M. Reich, A.J. Philippacopoulos, C. Miller, C. Costantino and L. Qingwu l

Brookhaven National Laboratory Summa ry I

Various numerical approaches with diff erent degrees of approximations have been developed and utilized for the evaluation of the structural response of nuclear containments and other Class I nuclear structures. These approaches however, inherently rely on various degrees of approximations in order to simplify the mathematical equations associated with the analysis l

methods. Thus, they may not necessarily represent the actual response be-l havior of the structure in question. This is especially true for operating or accident conditions that involve seismic and dynamic loads. Under this pro-4 gram BNL is investigating the ranges of validity of the anatlytical methods used to predict the bahavior of nuclear safety related structures under acci-

dental and extrene environmental loadings.

Durng FY 84, the investigations were concerned with comparisons of pre-dictions made using the Standard Soil Structure Interaction (SSI) methods and j those obtained from measured experimental data at the fUKUSHIMA site and the SIMQUAKE experiment. This work is detailed in a report entitled, "Verifica-tion of Soil-Structure Interaction Methods", NUREG/CR-4182, published in the spring of 1985.

! During the current year, the investigations were concentrateo on special problems that can significantly influence the outcome of the soll structure interaction evaluation process. Specifically, limitations and appItcability of the standard interaction methods when dealing with lift-off, layering and water table effects, were studied in detail. it was found that under certain i

conditions, the above effect can highly influence the analysis results j obtained by standard methods. The nature of these conditions as well as pro-posed modifications for the standard methods to account for these effects, are discussed in the text of the report.

i i

1 e 5

4 5-3 l

l

. ~ . . . . , . . _ _ . . _ . _ .__,-._.-,_..-__--_,m . . _ , _ _ _ _ _ _ , , . - , _ _ _ . . _ . . _ , ,_ _.._... , _._.

MELRPI - DEVELOPMENT AND USE A. Sozer Severe Accident Sequence Analysis (SASA) Program Oak Ridge National Laboratory l

l l

The MELRPI code, developed especially for BWR severe accident analyses I

und3r the sponsorship of Oak Ridge National Laboratory (ORNL) and more recently the Empire State Electrical Energy Research Corporation (ESEERCO) is used in accident analysis at ORNL. The current version of the code, MELRPI. MOD 2, includes simplified mechanistic core degradation, heatup, oxidation, melting, relocation, clad failure and rubble bed formation, and Emergency Core Cooling System (ECCS) models for intact or rubblized reactor core thermal hydraulics.

For the core uncovery period (approx. 30 hours3.472222e-4 days 0.00833 hours 4.960317e-5 weeks 1.1415e-5 months ), a station blackout accident analysis has been performed in order to estimate the level of l

BWR core degradation. Core degradation results are presented in terms i oft

fuel, clad, control rod and channel wall temperatures,

' rubblized nodes,

' reduced radial zone heights (effect of porosity and melting),

' molten clad + channel wall (Zr) fraction and, molten fuel fraction.

l 4

4 7-15

i Hydrogen Transport in a Large, Dry l Containment for Selected Arrested Sequences

D. B. King A. C. Peterson Sandia National Laboratories l

Summary The hydrogen burn in the containment during the accident at Three Mile Island instigated renewed attention to the possible consequences of hydrogen generation during a nuclear core uncovery accident. This accident contributed to Hydrogen Control Rules for Pressurized Water Reactors (PWR) with ice condenser containments and Boiling Water Reactors with Mark III containments. The Hydrogen Control Rule for PWRs with large, dry containments is presently deferred.

To assist in providing a basis for a recommendation on rulemaking for PWRs with large, dry containments, the i

containment environment created by arrested sequences having a 75% metal-water reaction was calculated. The HECTR computer i

code was used to calculate the containment response. The primary focus of this analysis was the investigation of the potential for formation of detonable mixtures of hydrogen, steam and oxygen.

HECTR is a lumped paramrcer containment analysis code developed to analyze nuclear reactor accidents involving the transport and combustion of hydrogen. However, this study only analyzed the transport of hydrogen and did not consider the effects of combustion. The thermal response of structural i materials and offects of reactor building fan coolers and containment sprays cre also calculated by HECTR.

Since the primary interest in this analysis was the potential for formation of detonable mixtures, a " highly" compartmentalized model (up to 42 compartments, 130 flow junctions and 125 structural surfaces) was used.

Steam and hydrogen mass flows calculated by the MARCH computer code for small break loss-of-coolant accidents and loss of offsite power transients were used as sources into the HECTR code. The peak hydrogen mass flow rate calculated for these sequences was 114 lbs/ min. The steam, hydrogen, oxygen

This work is supported by the United States Nuclear Regulatory Commission and performed at Sandia National Laboratories which is operated for the United 6'ates Department of Energy under Contract Number DE-AC04-76DP00789.

t 7-17

-- . - -~ -__ - _ . - - - .-. --

i and nitrogen transport in the containment were calculated.

Calculations determining the effects of the method of compart-mentalization, effects of fan coolers, and the effects of i containment sprays were performed. Overall, fifteen calculations were run and analyzed.

The results of this study indicated that for the specific conditions selected for these calculations, the highest 4

concentrations of hydrogen were always in the source

compartment. The size and location of the source compartment l affected the calculated peak hydrogen concentration. For one source location, which was relatively small, a detonable mixture j was calculated. However, the potential for the steam and

hydrogen sources to exist at this location and the effects of a j detonation at this location are still being analyzed. The

! water / steam and hydrogen source flows were dominant factors in establishing the circulation flows in the containment. In the

long term, these circulation flows resulted in relatively good j m2xing of the gases throughout the containment.

] Overall, the calculations with water / steam and hydrogen

! source flows which are representative of small break loss-

) of-coolant accidents for degraded core scenarios showed that i relative to the limits for deflagrations and detonations, the

, source compartment was not steam inerted.For most cases j concentrations were calculated that were near yet outside the

currently assumed detonation limits; however, one case was i

within the detonation limit. The calculations with water / steam and hydrogen source flows representative of loss-of-offsite

, power transients showed that at the peak hydrogen concentration

) the steam concentration was greater than 40%. At this steam concentration, the compartment would be steam inerted because it I

l was above the detonation limit, but could be within the i

flammability limit.

I i

I I

I t

l 7-18

(

Pressure-Temperature Response in an Ice-Condenser Containment for Selected Accidents

S. E. Dingman l A. L. Camp Sandia National Laboratories Albuquerque, NM 87185 l Igniters, designed to allow controlled burning of hydrogen at relatively low concentrations, are present in most regions of ice-condenser containments.

However, igniters are not present in the ice bed region of the ice condenser.

Thus, controlled burning of hydrogen at low concentrations can not be guaran-teed in this region. In previous analyses that examined the containment pressure and temperature loadings in ice-condenser containments [1,21, detonable mixtures of hydrogen were predicted to occur in the ice bed during scenarios in which the containment air return f ans had f ailed. In these scenarios, burning was prevented in the lower compartment by steam inerting (fan failure prevented return air flow f rom the upper compartment) . Without the large driving force provided by the f ans, detonable mixtures were predicted to occur in the ice bed before a combustible level was reached in the upper plenum (which contained ignitors).

The previous calculations were performed with the MARCH, MARCOM, and HECTR computer codes. A model was used for the Lee bed region that consisted of a single vertical stack of 4 compartments. Since the ice bed was not subdivided l circumferential1y, natural convection loops could not form in the ice bed.

l Thus, the hydrogen concentration was much higher at the top of the ice bed e l than at the bottom. It has been suggested that recirculation loops driven by ,

natural convection could result in suf ficient mixing within the Lee bed (as well as mixing of the ice bed and upper plenum regions) to prevent detonable i mixtures from forming in the ice bed region. To address these concerns, we 1 have constructed a more-finely nodalized HECTR model (40 Volumes) for an ice-condenser containment. The ice bed region is divided circumferentially i into 4 columns of compartments with 4 Vertically stacked compartments in each

column (16 compartments in all). This allows recirculation loops to form in the ice bed region.

i

This work is supported by the U. S. Nuclear Regulatory Commission and 2

performed at Sandia National Laboratories which is operated for the US Department of Energy under contract number DE-AC04-76DP00789.

i d

7-19

l i

Three calculations were performed using this 40 volume deck to examine the potential for detonable mixtures and asymmetric melting when recirculation loops were possible in the ice bed region. The calculations were performed for two accident scenarios: a transient-initiated accident in which there is total f ailure of both AC power and steam generator feedwater (TMLB'), and a small break loss-of-coolant accident with failure of emergency core coolant and containment sprays in the recirculation mode (S tHF). Fan operation is precluded by AC power failure in the TMLB' scenario. For the S tHF scenario, we performed calculations both with and without fans operating to examine the sensitivity of the results to fan operation.

The HECTR calculations predicted that recirculation loops would form in the ice bed region for all three cases (even the S tHF scenario with fans operating). For the fans-off cases, these loops caused more mixing within the ice bed region than in the HECTR calculations reported in references 1 and 2; however, detonable mixtures were still predicted. Although the igniters were operating in the S1 HF fans-off case, they were ineffective in preventing the detonable ice bed mixtures because the ice bed region (which does not contain igniters) became detonable before the upper plenum (which contains igniters) reached an ignitable level. Detonable mixtures did not occur in the S t HF scenario with fans operating due to the strong mixing provided by the f ans, For all three calculations, the recirculation loops resulted in greater ice melting in two of the columns than in the remaining two. This would lead to earlier melt-through of a portion of the ice bed than if the melting were uniform in all four columns.

REFEREUCES

1. A. L. Camp, V. L. Behr, and F. E. Haskin, MARCH-HECTR Analysis of Selected Accidents in an Ice-condenser Containment, NUREG/CR-3912, Sandia National Laboratories, December 1984.

2. F. E. Haskin, V. L. Behr, and A. L. Camp, HECTR Results for Ice-Condenser Containment Standard Problem, Proceedings Second Workshop on Containment Integrity, Sandia National Laboratories, June 1984.

l l 7-20 l

Heat Transfer, Carry-over and Fall-back in U-tube and Steam Generators L.Y. Liao - Former Student, MIT A. Parlos - Research Assistant, MIT P. Griffith - Prof. of ME, MIT, Cambridge, 02139, MA Experiments have been run and models developed describing the i heat transfer, carry-over and fall-back which occurs on the secondary side of an inverted U-tube steam generator following steam line breaks, feed line breaks and steam line breaks accompanied by tube rup ture ( s) . The de tails are soon to be generally available in a NUREG report. They are available immediately by request from Prof. Griffith (above) on microfische f*om the respective thesis. The experiments, analytical results and findings are as follows.

The experiments were run in vessel 10' high, 4" ID at full tempera ture and pressure. Simulated rods and downommer were included. The rods are 05." OD copper and are unheated. The only heat provided to the system is from electrical heaters in the bottom, which are turned off when a run is initiated. That is, the rods dry out due to their own heat capacity. Pressure, several differential pressures, many temperatures and the carry-over are measured. A range of break sizes including all those of interest have been run.

A code which includes only the energy and continuity equations has been developed. Water levels are calculated using a drif t flux model. About ten control volumes suffice to describe the essentials of the process.

The principal findings are as follows:

1) Only two water levels need to be considered when modeling the secondary, one on top of the uppermost tube support and one on the bottom. Though holdup sometimes occured briefly on intermediate tube support plates, in general all the plates below the top are crained.

2) Existing correlations for the drift flux model constants were examined and the best values selected.

3) Even for a passively heated system, the wall and internals hea t capaci cy play a significant role in the amount of liquid left behind.

4) Measurable liquid carry-over only occurs when the leve'l swell early in blowdown brings liquid to the top of the vessel.

8-1

H3-2: STEAM GENERATOR TRANSIENT TEST PROGRAM by M. Y. Young, O. J. Mendler, and K. Takeuchi Westinghouse Electric Corporation

SUMMARY

This paper describes results from a test and analysis program desiEned to investigate the thermo-hydraulic behavior of a U tube steam generator mder l

i accident conditions. The primary objective of this program was to provide detailed data from simulated loss of feedwater (LCF), steam generator tube rupture (SGTR) and steam line break (R.B) transients which can be used to validate computer codes and improve safety analysis calculations. This program is jointly sponsored by Westinghouse, the Nuclear Regulatory Comrrission, and the Electric Pwer Research Institute. The follwing paragraphs sunmarize the subject areas covered in the paper.

The test facility which was used in this program is a nearly full height, 52 tube model boiler which simulates a typical U tube, feedring design steam generator. The model is scaled to approximately one percent in volune and steam generation rate, and contains a riser wirl vane, and dryer assembly which is typical of current designs. The boiler operates at 1000 psia, and has been usea in extensive steady state, full power performance testing.

For the purpses of this transient test program the tube bundle was re-built and extensively re-instrunented with thermocouples to allw determination of local heat flux within the bundle. In addition, differential pressure measurerents at many points within the test section are available.

The objectives of the various test series and the background behird the transients investigated are discussed. A test matrix is then presented which addresses the major questions raised by analysis of these transients.

Data from the three test types (LCF, SGTR, SLB) are presented. Heat transfer and thermchydraulic data for each test is presented and discussed. In particular, the heat transfer on the secondary side determined from the LCF experiments is presented. The implications of the SGTR tests relative to dose release calculations is discussed. Finally, the measured break mass f1w for the SLB tests is presented, and conclusions drawn on the effect of the liquid carryover which was observed during these tests.

8-3

Steam Generator Modeling During Transients Chan-Young Paik, RA, MIT Peter Griffith, Prof. of ME, MIT Cambridge, MA 02139 Separators are part of every steam generator. As long as the steam generator is operating within the design envelope virtually all the moisture is removed. However, during a steam line break with simultaneous tube rupture the amount of liquid (and thus iodine) released to the atmosphere depends very strongly on how the separator performs. This is a project which has as its goal developing a generic separator model which can be applied to a variety of steam generators.

Experiments run in a transparent model of the W MB-2 separator system show us how these separators perform.

The model has three stages of separation, a cyclone, followed by a gravity separator followed by a chevron separator. The separator appears to perform at flow rates well above the design values as long as the liquid does not accumulate on the deck plate. High liquid levels rather than a high flow rates appear to be the primary cause of the failure.

When the liquid level is high enough, so much liquid is delivered to the chevron separator that the drain lines are unable to remove it and the excess is carried out with the steam. A scenario leading to steam separator f ailure, a mathematical model which incorporates the principal geometric variables and

suggested values for all constants will be provided within a year when the project is completed.

8-5 s

- -- - - - - ,~

CRITICAL FLOW THROUGH A SMALL BREAK ON A LARGE PIPE WITH STRATIFIED FLOW V. E. Schorck, S. T. Revankar, R. Mannheimer, C-H. Wang University of California, Berkeley An experimental investigation of critical steam-water two-phase flow through a small break on a horizontal pipe carring a stratified two-phase flow has been carried out. This flow configuration may occur in small break LWR accidenth.

The critical dischargo depends upon the effective stagnation state of the fluid entering the break channel and this state is, in turn, dependent upon the orientation of the break (up, down, side) and the proximity of the free surface in the upstream stratified region. The upstream hydrodynamics have been discussed by Zuber (1) .

Air-water experiments have been reported by Crowley and Rothe (2) for side break, by Reimann and Khan (3) for down break, and Reimann and Smoglie (4) for up break orientation. These experiments provide data on the interface level for incipient pull through of the second phase and the flow quality of the mixture entering the break channel. The incipience depends upon the break flow-rate, and therefore, upon the upstream pressure. For the case of steam-water, flashing in the break channel is an additional complicating factor influencing the break flow. Thus, stream-water tests were essential to determine whether the air-water data are applicable to the LWR problem.

In the present study, a 4-inch IPS schedule 40 stainless steel pipe was used for the upstream stratified flow region and provision was made to independently j l

controlthe supply of water and either air or steam. Break flow channels were !

straight sections of tubing, 3.9 and 6.2 mm I.D. and 12.3 cm in length, 1 l

8-7

2 mounted perpendicular to the main pipe. Windows were provided for viewing 1

the flow pattern at the entrance of the break channel. The discharged fluid was ducted to a quench / weigh tank used to measure the mass flow-rate.

Incipients of vapor pull through tests were carried out for down-oriented and side-oriented break channels in the pressure ranges of 110 to 1100 kPa, and 230 to 910 kPa respectively, for steam-water system and pressure range of 110 to 550 kPa for air-water system. Incipient of liquid entrainment tests were carried for up-oriented and side-oriented break channels in the pressure

! i ranges of 110 to 170 kPa for air-water system.

The vapor pull through the incipient data were correlated as interface level

, vs liquid Froude (Fr=Nj j7) number. For both the cases of data obtained with down and side oriented break channels, for the same geometry, the air-water data differed from the steam-water data by approximately a factor of two.

1 The difference is thought to be caused mainly by difference in the surface tension. The data-could be brought together by multiplying the Froude number ,

by the ratio of Bond numbers (Bo=d h) raised to the fourth power. The e

quality of the mixture entering t.'.e break depends upon the level h divided by the incipient level h .bIn these coordinates, the present. steam-water data are in good agreement with the air-water data of Reimann and Khan, while the present air-water results are somewhat different. This difference may be due l to the greater slope of the upstream interface near the break in Reimann and Khan's experiment. In the case of liquid entrainment (up and side flow),

the incipient data were correlated as interface level vs gas Froude number (Frg =

hj }[ ) and the air-water data agree with the steam-water data.

In the cast of up-oriented break channel, the entrance quality data for i

air-water and steam-water system agree with each other when quality is plotted i

8-8 r

I c

3 as function of h/h . The present data set covers the range of x from 10~

b to 9.5 x 10~ , and Smoglie's data covers x from 0.95 to 1. The data are in general agreement, showing a cor sistent variation.

The pressure profiles along the length of break channel were measured. In i t

case of two-phase flow, the pressure drop in the break channel increased with ,

i increase in the entrance quality. In case of vapor pull through, the quality increased with decrease in the interface level, whereas in case of liquid i

entrainment, the entrance quality increased with increase in the interface

level. Data for flow qualities, mass flow rates and pressure drops will be presented as functions of mainline stratified liquid level.

I i

4 References

1. N. Zuber, " Problems in Modeling of Small Break LOCA" t

NUREG-0724, 1981

2. C. J. Crowley and P. H. Rothe, " Flow visualization and Break Mass Flow Measurements in a Small Break Separate Effects Experiment", Small Break LOCA Analysis in LWR's, EPRI WS-81-201, August 1981

3. J. Reimann and M. Khan, " Flow Through a small Break at the Bottom of a Large Pipe with Stratified Flow", 2nd International Topical Meeting on Nuclear Reactor Thermal Hydraulics, Santa Barbara, CA, January 1983.

4. J. Reimann and C. Smoglie, " Flow Through a Small Pipe at the Top of a e Large Pipe with Stratified Flow", European Two-Phase Flow Group, Zurich, Switzerland, June 1983.

5. C. Smoglie, "Two-Phase Flow Through Small Branches in a Horizontal Pipe with St. ratified Flow", Disertation, Kernforschungszentrum Karlsruhe GmbH,

Karlsruhe, KfK, 3861, December 1984, pp. 130 8-9

1 CRITICAL FLOW THROUGH IGSCC IN PIPES V. E. Schrock, S. T. Revankar & S. Y. Lee l l University of California, Berkeley l The presence of intergranular stress corrosion cracks (IGSCC) in thermal stressed zones in steel piping and associated components of commercial reactors is of much concern in reactor safety. The prediction of leak rate through the cracks is important in assessing the plant reliability. An analytical model has been developed to predict the flow rates of initially subcooled or saturated liquid through these cracks.

The model assumes the flow in the crack to be homogeneous and in thermal

]

equilibrium. The crack geometry was assumed to be a convergent slit with smooth entrance. The slit gap was taken as uniform in transverse and in flow directions. The gravitational effects were neglected and the control surface of the geometry taken as adiabatic. One dimensional steady state model accounts for the effects of gross area change, expansions / contractions, tortuosity and friction. A numerical scheme has been deeeloped for the model which is coded into a Fortran computer program called SOURCE. The SOURCE program has been provided with subroutine STEAM for the calculation of saturated fluid properties.

l Imputs to the SOURCE are stagnation fluid properties and the crack geometry specifications. The program automatically assumes the initial guess value of mass flow rate depending on the stagnation fluid conditions through Hall's Isoentropic Homogeneous Equilibrium Model (IHEM) Table.

8-11 l

n

2 This table is provided as supporting data to the SOURCE program. The program calculates the pressure and the fluid properties, at each grid point along the channel. The grid spacing is such that the downstream grid size is half the size of the immediate upstream grid. At each grid point, the acoustic speed and fluid velocity are compared to see whether the flow is choked at that location. An iteration scheme is performed until the acoustic and fluid velocities match with one another at the crack exit plane. The tolerances used in the program are $ 0.1%.

The SOURCE program has been run to reproduce the measured mass flow rates of Battelle Clumbus Laboratories (BCL), phase II experiments. A parametric investigation was carried out using different values of friction factors in the calculation of critical mass flow rates. It was foynd that, for each set of data (total of five crack geometries), a single value of friction factor yielded good agreement (error 6 20%) with BCL data. And for each set, it was found that the correction factor to be applied to the predicted mass flow rate, depends on the stagnation subcooling of the liquid. The ,

correction factor decreas?d with increasing subcooling. This trend was consistent with all five sets of crack geometry studied.

SOURCE program predictions with a subcooling connection factor are recommended for use in estimating the leak rate through IGSCC cracks.

i i

8-12 n

A FINAL REPORT ON THERMAL MIXING by T.C. Theofanous Department of Chemical and Nuclear Enineering University of California Santa Barbara, CA 93106 l

t i

In response to the Pressurized Thermal Shock (PTS) issue a multifaced, comprehensive, research program on HPI mixing has evolved over the past several years with the support of the USNRC, EPRI, and US Industry. International cooperative efforts, US FGR and US Finland, came also into existance to provide l additional support.

In order to address the important problem of scaling these programs included integral simulations at several scales, both thermal and/or salt induced buoyancy, and a variety of measurement techniques. Analytical work encompassed three dimensional finite difference approaches, mechanistic modelling as well as empirical correlation schemes.

The purpose of this presentation is to provide an updated summary of the experimental data base, and it's unified interpretation with the help of the Regional Mixing Model (and the associated computer code REMIX). In particular the most recent full scale tests at the HDR reactor in Germany and the multiloop tests at Imatran Voino 01 in Finland will be discussed in detail.

We conclude that cooldown transients due to HPI in all conditions relevant to US designs can be predicted with confidence.

1 j

8-13 n

Steam Explosions - Energy Conversion Efficiencies of Steam Explosions for Two Major Accidents in the Pulp and Paper Industry T. M. Grace, The Institute of Paper Chemistry, Appleton, WI R. R. Robinson, IIT Research Institute, Chicago, Il and J. Hopenfeld, U.S. Nuclear Regulatory Commission, Washington, DC The object of the study was to determine the overall energy conversion effi-ciency (thermal to mechanical) of steam explosions, by detailed analysis of smelt-water explosions that have occurred in kraft paper pulp mill recovery boilers. Recovery boilers are large steam generators which use spent liquors

, from pulping wood as fuel. Smelt is a mixture of molten salts which are produced when the liquor is burned. These are the recovered pulping chemicals and are drained from the furnace and processed for reuse.

Analyses were carried out for two major accidents; one in each of the two types of recovery boilers currently in use. In one case water entered the furnace from a ruptured boiler roof tube. In the other, wash water entered through a spray gun used for firing black liquor.

The energy conversion efficiency was calculated as the ratio of the deformation energy in the furnace structure to the energy content of the molten smelt in the unit. The mechanical deformation energy was determined by developing models of the furnace structure and calculating the deformation energy consistent with the observed damage. Estimates were made of the amount of smelt present within the unit, smelt temperature, and the amount of water which entered the unit. Only the sensible heat in the molten smelt above the freezing point and the heat of fusion were included in calculating the energy available for the explosion.

This is considered to be conservative. If the energy release was limited by the avialability of water, complete vaporization of all of the water present was used as the upper limit.

The best estimate of the energy conversion efficiency was 0.25% for one case and 0.6% in the other. Although there is a wide range of uncertainty in the effi-ciency values, it is considered extremely unlikely that the energy conversion efficiency exceeds 1% in either case.

8-15 l

a

Eastern United States Seismic Hazard Research A.J. Murphy & L.L. Beratan Earth Sciences Branch Office of Nuclear Regulatory Research United States Nuclear Regulatory Commission l

l Abstract The United States Nuclear Regulatory Commission (NRC) supports a very productive research program that addresses the issue of seismic hazards to nuclear power plants (NPP's). Estimation of these hazards is a very i significant factor in siting of NPP's and in evaluating the safety of existing plants. There is considerable uncertainty in estimating the seismic hazard particularly in the Eastern United States. The objectives of the NRC seismotectonic research program are to quantify and reduce the uncertainty in seismic hazard assessment and to develop methods of dealing with uncertainties.

Three of the principal contributors to the uncertainty in quantifying the seismic hazard at a site are: 1) the characteristics of the seismic source zone, 2) the propagation of seismic energy between the source and the site, and

3) the site response, including , soil response. The relative levels of contribution of these three to uncertainty are region dependent. (For programmatic ease, the Eastern United States have been divided into four regions: Northeast, Southeast, New Madrid / Anna, Ohio, and Nemaha Ridge.)

Currently, there is a reasonable level of confidence in a working hypothesis for the source of seismicity in the New Madrid area and a moderate understanding of the regional propagation characteristics. There are a number of hypotheses for the source of Southeastern seismicity including the Charleston, South Carolina, area. No generally accepted hypotheses are available for the Northeast or the Nemaha Ridge. There is a low level of knowledge about the propagation and site response characteristics in the East exept as noted for the New Madrid region.

The programs concerning the quantification and reduction of uncertainty in seismic hazard assessment due to these three contributors are grouped into the regional program and the topical program. The regional program consists of the operating of seismographic networks and geophysical / geologic investigations of key areas, i.e., the collection and interpretation of a basic data set. The topical program consists of studies related to generic issues rather than source zone problems, such as the frequency dependence of the propagation.

A brief description of these two programs will compose the first third of this presentation.

The program related to dealing with the uncertainty generally involves probabilistic analysis of existing data sets. The principal projects in the program has been the " Seismic Hazard Characterization of the Eastern United States" conducted by Lawrence Livermore National Laboratory. The seismic hazard curves calculated were based upon expert opinion generated in the areas of seismicity and source zones and of seismic strong ground motion. Prelimi-nary results from this study will be presented and discussed.

9-1

-= _ ._ _

NRC'S SEISMIC MARGINS PROGRAM Robert J. Budnitz Future Resources Associates, Inc.

Berkeley, California 94704 In order to provide technical guidance to the NRC on the subject of seismic mar-gins at nuclear power plants, the NRC in mid-1984 formed an " Expert Panel on Seismic Margins", consisting of seven members, supported by key technical person-i nel at Lawrence Livermore National L:boratory. Then Panel is charged to work l closely with an in-house NRC staff working group on margins, to address key regu-latory needs. Panel members include R.J. Budnitz (chair), P. Amico, C.A. Cornell, t W.J. Hall, R.P. Kennedy, J.W. Reed, and M. Shinozuka.

The Panel has recently issued a report entitled "An Approach to the Quantification The pur-of Seismic Margins at Nuclear Power Plants", NUREG/CR-4334, July 1985.

i pose of this paper is to provide a summary of that report, whose objective is to discuss progress accomplished to date in studying the issue of quantifying seismic margins.

In particular, the report covers progress accomplished toward the establishment of review guidelines that would be useful in assessing whether a nuclear power plant has a high confidence of a low probability of failure when subjected to a postula-ted earthquake selected for margin review. Although these interim review guide-lines are not fully developed in this report, the basis is laid for such develop-ment. Guidelines will be developed in interim form as the next task of the Panel.

The plan is for these interim review guidelines to be used on a trial basis for one or more plants, after which their true value can be assessed. After this trial use, they will be modified, and them made available for general use.

The Panel's definition of seismic marain emphasizes that the margin is to be ex-pressed in terms of earthquake ground motion level, often referred to as ' earth-quake size'. The Panel has agreed to use as an indicator of ' earthquake size' the

' peak ground acceleration (pga)', defined as the average of the two horizontal peak components of free-field ground-surface acceleration coupled with additional des-criptors discussed in the body of the report.

The measure of margin adopted by the Panel is a high-confidence, low-probability-l of-failure (HCLPF) capacity. This is a conservative representation of capacity and in simple terms corresponds to the earthquake level at which it is extremely

unlikely that failure of a component or structure will occur. From the mathemati-cal perspective of a probability distribution on capacity developed in seismic PRA calculations, the HCLPF capacity values are approximately equal to 95 % confidence (probability) of not exceeding about a 5 % probability of failure. It is possible to derive this point from a full set of fragility curves; however,this is not neces-sary and to use this procedure in every case would defeat the Panel's purpose. The HCLPF values for specific types of components and structures are derived from a combination of engineering data, either test data or data from real earthquakes, and engineering analyses. The Panel feels that it is much easier to identify this point directly, and we have much more confidence in the identification of this point than in the identification of the median, randomness, and uncertainty para-meters which would be required to derive full fragility curves.

I 9-3

- _ _. _ -.7, v.

The Panel expects that a group of engineers would agree on a ground motion level where a component or structure has a high confidence of a low probability of failure; although they are not likely to agree that the confidence can be expressed exactly as 95% - 5%. This contrasts with the total lack of agreement which the Panel believes is likely to occur if the median capacity level and associated variabilities were selected. This is why the Panel believes that the HCLPF con-cept is a more useful way to deal with the question of ' seismic margins' than an approach using the median fragility values.

The Panel's margins review approach consists of eight steps, the first of which is the selection of an earthquake level for which it is desirable to demonstrate margin. This would be of the form, "Show that the plant has seismic capacity (mar-gin) up to an earthquake of 0.x g". The approach attempts to show the existence of that margin.by showing that the plant core-melt HCLPF is at least u.x g.

The steps which make up the review guidelines constitute three major processes:

screening, HCLPF determination, and systems analysis. The screening process uses the functional / systemic insights developed from the review of available seismic PRAs and the fragility insights developed from the available fragility information to reduce the size of the margin review. By applying these insights, it is possi- I ble to ' screen out' (remove from further consideration) large numbers of components and structures. The product of this process is a list of components and struc-tures which potentially have HCLPF values lower than the review level (i.e., cannot be said to have a high confidence of a low probability of failure at the review

, level, without performing an engineering analysis) and are involved in the perfor-mance of the important plant functions (i.e., the failure of which can lead to an initiating event or to the failure to shut down the nuclear reaction or provide cooling to the reactor core in the time period immediately following the earthquake.)

This list of components and structures is carried forward to the other two parts of the review. In the HCLPF determination process, detailed analyses are done to derive a HCLPF value for each of the components and structures. In the systems analysis process, a list is developed which represents the combinations of failures 4

of these components and structures which can lead to a core melt. Combining these two products will result in one of two outcomes. First, it is possible that there will be no combinations of failures that have an overall HCLPF lower than the re-view level. This specifically means that the overall core melt HCLPF is at least equal to, and probably greater than, the review level. The second possible outcome is that there may be combinations of failures that have an overall HCLPF lower than the review level. The product in this instance will be a numerical value of the overall plant core-melt HCLPF, along with an identification of which component and structural failures in what specific combinations are responsible for these HCLPF values.

l The Panel plans to develop detailed interim review guidelines in the next phase of its work. These guidelines will then be applied on a test basis to a few plants, and from the insights gained during the testing phase the guidelines will be re-i vised and finalized.

I I

9-4

r SEISMIC CATEGORY I STRUCTURES PROGRAM l

by l

J. G. Bennett W. E. Dunwoody C. R. Farrar Los Alamos National Laboratory The Seismic Category I Structures Program is part of the seismic r'esearch being carried out by the Office of Nuclear Regulatory Research as described in NUREG-1147 " Seismic Safety Research Program Plan." The overall objective of the l Category I Structures Program is to conduct experiments and perform analyses I

that address the USNRC licensing issue: Can existing facilities continue to operate in light of more demanding criteria than the criteria considered in the initial design? Specific objectives include developing the experimental data a for determining the sensitivity of structural behavior of noncontainment Category I Structures to variations in configuration and seismic loading; identifying the sensitivity of floor response spectra changes used in the design of piping and equipment to these variations; developing a method of representing damping in the inelastic range; developing experimental data to verify ductility factors used in conjunction with other analyses; and developing data that can be used to validate computer codes used to predict the response of these structures in the elastic and inelastic range.

During FY 85 this program has focused on carrying out a plan to resolve two major issues that have emerged from the static and seismic testing of 23 different models representing two types of structures, a diesel generator building and an auxiliary building. Two different sets of scales of these buildings were used corresponding to one-inch and three-inch-thick walls for the models. In addition, the number of floors varied from one to three.

Although a number of results on items such as aging (cure time), effect of increasing seismic magnitude, etc., have been reported, two general and consistent conclusions came out of the data from these tests. First, the scalability of the results, i.e., the ability of models to predict prototypical structural behavior, was illustrated both in the elastic and inelastic range.

However, all models were fabricated using microconcrete.

9-5

.- - ~ . - _

The significance of using microconcrete instead of actual concrete needs evalu-ation. Second, the " working load" secant stiffness of the models was lower than the computed uncracked cross-sectional values of stif fness by a f actor of about 4. This result has come to be called the " stiffness difference issue."

The Technical Review Group (TRG) for this program pointed out that if such a low value of stiffness was scaled to prototypical structures, the implications for piping and equipment could be very significant in terms of both design and safety. In addition, the TRG pointed out that credibility for the scale model data must be established, and during two meetings held early in the fiscal year the TRG helped to lay out a plan and to design a model (called the TRG test structure) for resolving these issues, i j Two different scales (1/4 and full) of the TRG test structure have been

prepared and are in some phase of construction or testing. The full-size i structures are constructed of prototypical concrete and rebar while the 1/4 scale structures are made of microconcrete and simulated rebar as in our I

previous models. Testing of the first 1/4 scale structure is complete and the 1

. results verify conclusions from the previous models on reduced stiffness, with t

the additional result that the very low load level stif fness was indeed u nin 70-80% of the theoretical value. Testing of the second 1/4 scale structure is

) in progress and thus far verifies results from the first model. Construction i

of the first large structure is underway. Four specific scenarios have been

] identified as the likely outcome of the TRG experiments and have been addressed l accordingly in the program planning.

i Following the resolution of the stiffness difference issue, a limited number of ,

I tests will be carried out to meet program objectives and aid in benchmarking

for the analytical model development. If settlement of the scalability and ;

stiffness difference issues allows, these tests will be carried out on one-inch- ,

-thick-wall concrete models. A statistician, knowledgeable in experimental ,

design will be used to comment on the test configurations recommended and i assure that the controlled variables (i.e., number of floors, wall arrangement, ^

etc.) and uncontrolled variables (i.e., concrete strength) are incorporated into a cott-effective test matrix to meet program objectives.

9-6 l

l - - - - . . -n-~ . . . . - - - . . - - . . - - , , - - . , , - - - - - , - , - . - - . ,a-- ,.- - - - - -, - - - - . . - , ,

.= - _ - - = - . _ - _ . -_ _ _ - _ - _ _ _ - - - - . _ _

Component Fragilities - Data Collection, Analysis and Interpretation K.K. Bandyopadhyay and C.H. Hofmayer Brookhaven National Laboratory Upton, New York 11973 Summary As part of the component fragility research program sponsored by the U.S. Nuclear Regulatory Commission, BNL is involved in establishing seismic l fragility levels for various nuclear power plant equipment with emphasis on 1

electrical equipment, by identifying, collecting and analyzing existing test

! data from various sources. With cooperation from utilities, major reactor l suppliers, testing laboratories, A/E firms and equipment manufacturers, to date, BNL has reviewed approximately seventy test reports to collect fragility or high level test data for switchgears, motor control centers and similar electrical cabinets, valve actuators and numerous electrical and control de-

' vices, e.g., switches, transmitters, potentiometers, indicators, relays, etc.,

of various manufacturers and models. Through a cooperative agreement, BNL has also obtained test data from EPRI/ANCO.

An analysis of the collected data reveals that fragility levels can best be described by a group of curves corresponging to various failure modes. The

lower bound curve indicates the initiation of malfunctioning or structural

damage, whereas the upper bound curve corresponds to overall failure of the equipment based on known failure modes occurring separately or interactively.

For some components, the upper and lower bound fragility levels are observed to vary appreciably depending upon the manufacturers and models. For some de-vices, testing even at the shake table vibration limit does not exhibit any failure. Failure of a relay is observed to be a frequent cause of failure of an electrical panel or a system.

An extensive amount of additonal fragility or high level test data exists. If completely collected and properly analyzed, the entire data bank

is expected to greatly reduce the need for additional testing to establish i

fragility levels for most equipment.

t I

4 J

i

]

) 9-7

COMPONENT FRAGILITY RESEARCH PROGRAM

Component Prioritization and Testing l Carry S. Holman f

l Lawrence Livermore National Laboratory l P. O. Box 808, Livermore, California 94550 4

j

SUMMARY

Current probabilistic risk assessment (PRA) techniques for nuclear power plants utilize component fragilities which are for the most part based on a limited dato base and on engineering judgement. The seismic design of components is based on code limits and NRC requirements that do not reflect the actual capacity of a component to resist failure. At the request of the U.S. Nuclear Regulatory Commission, the j Lawrence Livermore National Laboratory (LLNL) has begun a Component Fragility Research Program (CFRP), whose general objectives are to (1) expand the present component fragilities data base; (2) establish component seismic margins; and (3) improve our general understanding of component failure modes due to seismic events.

The CFRP, currently projected as a three-year program, is being donducted in two phases. During FY85, Phase I has consisted of parallel activities to:

1. Compile and evaluate existing component fragility dato base from both a foreign and domestic sources.

2.

Identify, prioritize, and group components relevant to (i.e., having a potential influence on) plant safety to determine those for which odditional fragility j data must be generated.

3. Develop and demonstrate procedures for performing tests to obtain these new

] dato.

l The results of these activities will be consolidated into a program plan for compre-hensive component testing that will take place in Phase II (FY86-87). In Phase I, LLNL has been performing component testing and prioritization, and is coordinating its efforts with those of the Brookhaven National Laboratory (BNL), which is compiling the existing dato base. LLNL will incorporate fragility dato collected by BNL in the development of the Phase ll test plan.

Component identification and Prioritization The purpose of this task is to identify important plant components,in other words, those that would become condidates for Phase il testing. Recognizing that post efforts to develop generic importance lists through the use of PRA techniques have met with limited success, we elected to take on alternate approach relying primarily on expert judgement backed up by either analytic or testing experience. This approach proceeds os follows:

This work was supported by the United States Nuclear Regulatory Commission under a Memorandum of Understanding with the United States Department of Energy.

. 9-9

l. Identify plant systems that would be required to perform under normal and abnormal conditions, or that have o potential influence on plant safety. For each, list the major component along with those peripheral systems (e.g., lobe oil systems) that could offect the ability of the component to perform its intended safety function.

2. Af ter components have been identified for o porticular system, categorize them ;mo three importance groups: "very important"(i.e., component failure implies system failure), "important" (component failure significantly reduces j system performance), and "less important" (component failure does not significantly of feet system performance.)

3. After components have been assigned to their respective importance groups, subdivide each according to whether the components have o "high", "interme-dicte", or " low" expected seismic capacity. Here "high capacity" implies that the component would remain functional under local response greater than 29 zero period acceleration (ZPA), " low capacity" that functional failure is anticipated for ZPA levels less than Ig.

The results of this task will be used to prioritize components for Phase ll testing. For example, we would recommend that first priority be given to components identified as )

"very important" but having " low" seismic capacity.

We have completed on initial importance list that will be presented in the full conference paper.

Phase i Testing The test procedures developed for specific components recommended for Phase ll testing will depend on those factors most influential on component functionobility. In i

order to demonstrate that we con chorocterize fragility for o selected piece of nuclear power plant hardware and then investigate its dependence on a particular technical issue, we are also performing actual component tests in Phase 1. For this purpose, we selected a motor control center with " representative" electrical components (e.g.,

relays, local starters) installed in th :abinet. The functional fragility of the electrical components will be observed, choroderized by their electrical output (e.g., chatter,1-V chorocteristics), while the cabinet will serve os a load transmission device. The base flexibility and mounting configuration of the MCC will be varied to investigate the effect of "hord" vs "sof t" mounting on component behavior. Testing is currently in progress at Wyle Laboratories in Norco, California.

It is important to note that the selection of this porticular test configuration was

not mode with the intent to "torget" this component out of any special concern, but rather because we feel it has certain chorocteristics that will allow us to demonstrate our general approoch to fragilities testing. While we recognize that our component identification octivity will likely include MCCs os on important component, our selection for demonstration testing was mode without regard for its importance relative to other components.

The conference presentation will include first results from our Phase I test program.

1 9-10

HeiGdompfreaktor Phase ll Vibration Tests Lothar Molcher, Kernforschungszentrum Karlsruhe, Project HDR, FRG Helmut Steinhilber, Fraunhofer-Institut for Betriebs-festigkeit (LBF), Darmstadt, FRG in the second phase of the earthquake investigations at the HeiGdompfreaktor (HDR),

FRG, high level shaker tests will be performed in June 1986. The purpose of these tests, supported by the German and U.S, Governments, is to investigate full scale structural response involving significant concrete and soil strains os well as strong indirect excitation of vessels, pipes and other mechanical equipment.

The vibrator, designed by ANCO Engineers, is a " coast-down" shaker, whose two eccentric masses of 40 tons each on a common shaft are brought up to speed in bo-lanced condition. Unbalancing will take place ofter decoupling from drive system and the shaker will then coast down through the buildings resonances. Accelerotions 2

of 4 - 5 m/s and corresponding displacements of + 7 cm are expected in the fun-domental rocking mode of the HDR building at 1 - 1.4 Hz.

During design of the shaker, extensive computer simulations of the dynamic behaviour of the coupled system shaker /HDR building were corried out as well as safety calcu-lotions to evoluote the food carrying capacity of the HDR building. Compared to the status of work reported during 12 th WRSIM, October 1984, the following progress was achieved:

The modified shaker design with an eccentric bearing of one shaker arm proved to guarantee unboloncing of the shaker at any desired frequency, but computer simulations indicated rather high shock loads during contact of the two shaker arms.

9-11

.. __ _ _ _- __ - - _. - __- _= _ _ . - . -

To prevent the system from overloading by the impact of the two shaker arms nonlinear dampers were designed. Dynamic calculations showed that the im-pact loads con be reduced to a tolerable level with these dampers.

The weights of the shaker were redesigned using steel plates instead of I

baskets filled with lead.

1 -

To prevent the anchorage from local moments the shaker frame was completely redesigned with increased stiffness.

Construction of the shaker core and the hydraulic drive system is completed in the U.S., the frame is under construction in Gennony.

Functionality check-out of the shaker will take place in December 1985 at the HDR und will include 1

one balanced test run with maximum speed 4 --

one balanced test run with maximum weights one unbalanced test run with low eccentricity at 30 % of maximum force one unbalanced testrun with medium eccentricity at 30 % maximum force.

i The check-out tests serve also to gather dato on air-resistance and bearing friction l

4 os well as on HDR building's and neighbour building's responses to provide o first

calibration for the safety calculations.

The safety concept for the shaker tests is based on calculations and measurements.

The aim is to reach during the tests the highest possible loading without global struc-tural failure. Pre-calculations using linear and nonlinear methods are performed not I

only for the maximum lood case but for all test load cases. Selected occelerofion and strain measurements serve to compare predictions and reality offer each test to con-finuously update the safety margins. The test plan has been updated for the modifico-tion of the shaker weights and design. Instrumentation planning is well in progress, j also US-NRC plans for the accompanying equipment qualification program. Hence, out of today's viewpoint, all test preporations are on schedule.

F 9-12 l

BWR RISK ASSESSMENT

by T. Y. Chuang, D. L. Bernreuter, J. C. Chen G. E. Cummings, J. J. Johnson, D. A. Loppo, J. E. Wells Lawrence Livermore National Laboratory Summary The simplified seismic risk methodology developed in the U.S. NRC Seismic Safety Margins Research Program (SSMRP) was demonstrated by its application to the Zion nuclear power plant, o pressurized water reactor (PWR). A detailed model of Zion, including systems analysis models (initiating events, event trees, and fault trees),

structure and SSI models, and piping models, was developed and used in assessing the seismic risk of the Zion nuclear power plant, o PWR. The SSMRP simplified methodology is now being opplied to o boiling water reactor (BWR). To demonstrate its applicability and to provide a basis of comparison of seismic risk between a PWR and a BWR, a seismic risk analysis is being performed on the LaSalle nuclear power plant.

Key elements of the analysis are:

Development of the systems models -- event and fault trees Benchmarking best estimate seismic response of structures, components, and piping systems with design values for the purposes of specifying median responses for use in the seismic risk calculations Definition of the seismic hazard at the LaSalle site including the effect of local site conditions Development of building and component fragilities for important structures and components Investigation of the effects of hydrodynamic loads on seismic risk Estimation of seismically induced core melt frequency.

Selected probabilistic response analyses of the LoSolle County Station structures were performed for two ranges of earthquakes--a lower level earthquake in the approximate range of the SSE and a higher level earthquake in the opproximate range of three times the SSE. Results of the analyses were probability distributions on two types of response--in-structure forces and moments to be used in the fragility evoluotion of the structures themselves; and in-structure response spectro at equipment and component locations for their fragility evoluotion. Two occeleration levels were considered to permit interpolation for other earthquakes of different peak occelerations.

This work was supported by the United Sotes Nuclear Regulatory Commission under a Memorandum of Understanding with the United Sotes Department of Energy.

9-13

A review of the seismic design analysis rmults and development of a preliminary set of structure element capacities initiated the task. Simultaneously, o preliminary SMACS onalysis was performed for a single earthquake simulation at near the SSE level to provide o basis of comparison with the design results. Having reviewed the design analysis results and structure model, changes in the structure model to better capture the expected behavior of the structure were recommended and incorporated into the SMACS onalysis. Additional preliminary SMACS onalyses were performed, loads generated, and on assessment of the model modifications mode. The initial model changes led to limited load redistribution and motivated a second set of model changes which were incorporated into the SMACS model and again evoluoted. The result was the best estimate structure model. One hundred-and-forty-five structure forces and moments at two excitation levels were obtained from the SMACS onalyses and used in the fragility development. Component fragilities were developed for major LaSalle components identified as important in terms of systems behavior and risk. LaSalle specific design reports and equipment qualification data were used as the principal basis for fragility assessment. Median level responses were used in the fragility assessment as generated from the SMACS onalyses.

A limited investigation of the effects of internally generated hydrodynamic loads on the seismic risk was performed. This effort concentrates on SRV discharge os the most likely loading condition and considers its effect on components and equipment.

Realistic best estimate treatment of the phenomenon was emphasized along with on approximate load combination methodology to be applied to hydrodynamic and seismic loads.

The LoSolle seismic risk analysis is identifying the core melt accident sequences, developing the initiating event scenarios, coordinating all the fragility and response information, and performing the probabilistic calculations using the SEISIM computer code. The occident sequences in the LaSalle analysis were identified using event trees. Over 500 occident sequences were identified. The event trees used were developed in the Risk Methodology Integration Evaluation Program (RMIEP). Fault tree analysis is used to define the failure paths in the occident sequences.

For LaSalle, two sets of fault trees were generated. One is a simple set while the other is more detailed. The detailed set was generated by the RMIEP. The two sets were generated so that a comparison between them con be mode. This comparison would indicate whether or not highly detailed fault trees are worronted for o simplified seismic analysis. Once cut sets have been generated from the event and fault trees they are put in a form that is acceptable by the SSMRP risk onalysis computer code SEISIM. SEISIM will then process these cut sets and provide core melt frequency estimates.

9-14 I .-. -

STATUS OF THE ORNL AEROSOL RELEASE AND TRANSPORT PROJECT R. E. Adams 1

Oak Ridge National Laboratory Oak Ridge, Tennessee 37831 The behavior of aerosols assumed to be characteristic of those generated during light water reactor (LWR) accident sequences and released into contain-ment is being studied in the Nuclear Safety Pilot Plant (NSPP) which is located at the Oak Ridge National Laboratory (ORNL). This project is sponsored by the Division of Accident Evaluation, Nuclear Regulatory Commission, and the purpose is to provide experimental qualification for LWR aerosol behavior codes under development.

During the past year, several significant changes in the scope of this project have occurred. During a joint NRC aerosol modelers/ experimenters meet-ing in 1984, several areas were identified in which additional experimental information was needed. New activities to develop this information include the study of (1) the thermohydraulic conditions existing during NSPP aerosol tests in steam-air environments, (2) the thermal output and aerosol mass generation rates for plasma torch aerosol generators, and (3) the influence of humidity on the shape of agglomerated aerosols of various materials. To accommodate this change in scope and to accelerate the development of this needed information, future multicomponent aerosol tests in the NSPP are to be conducted at a much lower frequency. The status of each of these study areas is presented.

A new facility was prepared at the NSPP site to accommodate the study of I

the effect of various levels of humidity on the physical characteristics (shape factors) of aerosols of interest in LWR accident sequences. The test vessel l (0.56 m3 ) features a humidity measurement and control system as well as provisions for the measurement of aerosol number concentration, mass concentration, and particle size distribution, and for the acquisition of aerosol samples for elec-tron microscopy. The aerosol-moisture interaction test (AMlT) facility is now operational and several tests with Fe2O3 aerosol have been conducted.

As part of the NRC support of the LACE Program at Hanford, a priority pro-ject was conducted to determine the suitability of the technique of acrosol

, production by plasma torch aerosol generators under the operating conditions of future LACE tests. Demonstrations of the production of manganese (Mn) aerosol at elevated temperatures and pressure were carried out. The final test was a study of the behavior of manganese oxide (Mn20 3) aerosol in the NSPP vessel in a steam-air environment.

I t

10-1 l

Status of the DEMONA Experiments-A Comparison with NAUA Calculations W. Schuck, H. Bunz, J. P. Hosemann Kernforschungszentrum Karlsruhe, FRG The NAVA computer code has been developed to calculate the removal of particulate fission products in the containment of an LWR under core melt accident conditions. The code development is considered to be completed with the code version NAUA Mod 5 and a series of application calculations has been performed. In order to demonstrate that the computed results are reliable, a

, large scale simulation experiment is necessary which simulates a core melt ac-j cident scenario as closely as possible. Therefore, the DEMONA experiment /1/

1 is being conducted to demonsrate natural aerosol removal in a closed contain-ment under realistic thermodynamic and aerosol specific conditions.

DEMONA is a large scale experimental program, in which aerosol behavior is measured under the conditions of a core melt accident in a German PWR con-tainment. The facility is a quarter scale model of the Biblis A power

] station, capable of being operated in the correct temperature and pressure range. Aerosols are produced by vaporizing and oxidizing powders of iron, silver and tin, concentrations of well over 10 g/m 3 can be achieved. The aerosol measurement instrumentation measures all relevant data of the solid and liquid fraction of the aerosol as a function of space and time.

i The DEMONA test matrix consists of a basic experiment, which is a simula-tion of a late overpressure failure sequence in a German large dry PWR con-

. tainment, and of some variations of that experiment including changes in the aerosol source, the thermodynamic conditions and the geometry of the contain-ment. The present status is that six experiments have been conducted suc-cessfully: a thermodynamic test run, a " dry" aerosol test, the basic test 1 (twice), a low concentration test, and one test with transient thermodynamics.

The results of the " dry" aerosol tests have been reported earlier /2,3/,

especially the influence of the aerosol particle size distribution on the overall behavior and on the comparison with NAUA computations has been discussed in detail.

Quick look evaluations of the wet test results are available, except for the latest one. The results can be summarized as follows:

The condensation of steam on particles leads to a much faster removal of 3 the aerosol from the airborne state, whereas in the dry tests the airborne concentration decreased below the detection limit of the measurement after about 48 hrs. In a wet test this condition was reached after 8.. 12 hrs.

The onset of steam condensation on aerosol particles could be determined i

clearly by two independent measurements. The time of this event was in good agreement with predictions from the ther.nodynamics code C0CMEL.

The comparison of measured and computed concentrations was good. The condensation rates, however, were crucial input data for the calculation and a need for improved data from the thermodynamic code still exists. In the wet 4

10-3

i i

tests also the aerosol measurement errors are larger than in the dry tests l because of the more difficult operating conditions for the instrumentation.

The agreement between measurement and calculation is especially notice-able when considering that the codes C0CMEL and NAVA are single compartment I codes and the experiment is conducted in a structured building. Although in l

. the experiments up to now all movable structures had been removed, still a

! compartment structure can be identified. It is one aim of DEMONA to investi-

, gate whether the single volume codes can calculate aerosol behavior correctly i in such a geometry. The final check will be a multicompartment experiment in ;

which at least three physical compartments will be created in the building. !

e The remaining tests are a repetition of the basic test with a mixed multicomponent aerosol to show the independence of the removal rates on the !

material, the multicompartment test, and a final demonstration experiment l 4

involving the media.

I Up to now, the following conclusions can be drawn. Within the scope of its development, no deficiencies of the NAUA code have been found. Especially j the applicability of the single volume concept for both the NAVA and the i C0CMEL codes was demonstrated; the multicompartment test will be the final l check. The more sensitive parameters were found to lie outside the responsi-

. bility of the code, e.g., aerosol source parameters, and especially thermo-dynamic and containment conditions (condensation rates, etc.). No need for a

! further refinement of the NAUA code is visible. Therefore, presently the KfK l concept is not toJched, which regards the DEMONA and BETA experiments as a j final step of KfK's efforts in LWR safety research. This is of course to be

seen in the context of the application to German LWR power plants which do not I have some of the ECCSs known elsewhere which may influence aerosol behavior in

! the containment.

i

References:

/1/ W. O. Schikarski et al. , DEMONA Forschungsprogramm zur Demonstration nuklearen Aeroso1verhaltens, KfK 3636, EIR 502 (1983).

/2/ W. Sch5ck et al., The DEMONA Project, Objectives, Results and Signifi-cance to LWR Safety, 5th International Meeting on Thermal Nuclear Reactor Safety, Karlsruhe, September 9-13, 1984. ;

J

/3/ W. Sch6ck et al., DEMONA Jahresbericht 1984, KfK 3942, EIR 553, BF-R 65.523-30-3, GRS-A-1078, KWU-R-917/85/103 (1985).

I V

1 a

10-4 l

STATUS OF THE LWR AEROSOL CONTAINMENT EXPERIMENTS (LACE) PROGRAM G. R. Bloom D. R. Dickinson R. K. Hilliard J. D. McCorma:k L. D. Muhlestein Westinghouse Hanford Company Richland, WA 99352

! F. J. Rahn Electric Power Research Institute Palo Alto, CA 94303

! The LWR Aerosol Containment Experiments (LACE) program will investigate inherent aerosol behavior for postulated accident situations where high consequences are presently calculated because either the containment is bypassed altogether, the containment function is impaired early in the accident, or delayed containment f ailure occurs simultaneously with a large fission product release. Since the LACE program is directed towards i resolving LWR source term issues, it is of special interest in relation to l safety analysis and emergency planning. Consequently, the program is sponsored by an international consortium, consisting of eleven sponsors, organized by EPRI.[1]

Six large-scale tests will be completed as part of the LACE program in the Containment Systems Test Facility (CSTF) at the Hanford Engineering Development Laboratory (HEDL) which is operated for the U. S. Department of Energy by Westinghouse Hanford Company. Test objectives and major test parameters are described for all six tests, but more detailed information is presented for the first three LACE tests; LAl, LA2, and LA3. The aerosol generation systems used to generate soluble and insoluble aerosols are l described and data relating to the generation of these two aerosols are presented. Computer code comparisons planned in association with the LACE program are also described.

This report focuses on those tests dealing with the containment bypass accident sequence in which aerosols can pass directly from the reactor vessel to an auxiliary building through a long pipe. The tests include three containment bypass scoping tests, completed under EPRI sponsorship while the LACE program was being defined which are in addition to the six LACE tests, and LA1 and LA3 of the LACE test sequence. These containment bypass tests are characterized in Table 1. Computer code comparison efforts are still in i progress for test LAl. Therefore, test results presented will focus on the containment bypass scoping tests.

1. Lead organizations: Comission of European Communities; VTT Finland; CEA France; ENEA Italy; JAERI Japan; SKI Sweden; EIR Switzerland; AEA United Kingdom; and EPRI, 00E, and NRC, United States.

10-5

TABLE 1

CONTAINMENT BYPASS TESTS Carrier Gas

! Inlet Velocity at Auxiliary l Test Temp Pipe 0atlet Building

] Number Aerosol Composition (oC) (m/s) Conditions CB1 Soluble Na0H 50% 186 200 Saguratedsteam/ air, steam / air 85 C CB2 Soluble Na0H 50% 111 160 Saguratedsteam/ air, Insoluble Al(OH)3 steam / air 81 C f

CB3 Insoluble Al(0H)3 50% 160 10 Saguratedsteam/ air, steam / air 84 C LAl Soluble Cs0H 50% 280 240 Superheated Insoluble MnD steam /N 2 steam / air, 115 C LA3 Soluble Cs0H 50% 280 Variable No auxiliary Insoluble Mn0 steam /N 2 building The containment bypass scoping tests used a test pipe 63 m in diameter, 27 m long with five 90-degree bends, 4 horizontal sections, and 2 vertical i

sections. The pressure drop across the pipe was approximately g.1 MPa and i

the Reynolds number of the carryr gas was approximately 5 x 10 . ' The test pipe terminated within an 850-m auxiliary building which was vented to a scrubber.

i Observations and conclusions from the pre-project containment bypass scoping tests are:

1. . Aerosol retention in the 63 m test pipe was less than 5% of the i entering mass.

2. Aerosol vented from the auxiliary building was less than 2% for tests j CB1 and CB2, and much greater for test CB3. The hygroscopic nature of the aerosol in tests CB1 and CB2 appears to have been responsible for the order of_ magnitude higher aerosol retention compared to that of dry, solid aerosol in test CB3.

3. The characteristics of the aerosol material passing through the test pipe at high velocity were changed significantly, probably due to 4

deposition onto and resuspension from the wall of the pipe. The large particle size of the resuspended material resulted in greatly enhanced

! removal in the auxiliary building.

j The implication of these conclusions in relation to consequence assessment for the containment bypass accident sequence is discussed.

10-6 L -- - .

CONTAINMENT INTEGRITY UNDER SEVERE ACCIDENT CONDITIONS

Walter A. von Riesemann, Daniel S. Horschel, Larry N. Koenig, David B. Clauss Sandia National Laboratories Albuquerque, NM 87185 The U.S. Nuclear Regulatory Commission (NRC) is fundirg several programs that have the shared objective of developing a methodology for determining the performance of containment systems for light water reactors (LWR) during covere accidents. Thcu occidents result in internal pressurization and elevated temperatures within the containment that are more severe than those used in the design of the containment system (structure, penetrations, i isolation valves , etc.). The containment system is the last engineered barrier to the release of radioactive material that may be present in the building. Release could occur due to leakage through the items constituting the containment system. In order to assess the consequences ofa a severe accident, the timing, mode, and location of the release must be known.

U.S. containment buildings and penetrations are designed according to codes established by the American Society of Mechanical Engineers and the American Concrete Institu's. An elastic method of analysis is used in the design procedure. The pressure and temperature within the containment during a severe accident may significantly exceed the design basis loadings.

Although the containment has a capacity for loads beyond the design conditions, estimation of this capacity is not straightforward, since nonlinear (inelastic) behavior is likely to occur. For this reason and I because knowledge of containment behavior is a key safety issue, the NRC has funded programs that are using a combined experimental / analytical approach.

Four of these programs: " Experiments on Containment Models Under Extreme Loading Conditions," " Integrity of Containment Penetrations Under Severe Accident Loads," " Containment Integrity Under Extreme Loads," and

" Electrical Penetration Assemblies" will be briefly described in the paper.

Containment Model Testing Scale models of nuclear power containments are tested to f ailure under the program titled "ExperlaAnts on Containment Models Under Extreme Loading Conditions." The scope of the program is to test free standing steel and reinforced concrete containments. Separate effects tests and testing by others are anticipated to address prestressed concrete containment behavior.

Testing of the free standing steel containments (those typical of a PWR ice condenser or BWR MK-III) has been completed. The testing of reinforced concrete containments will consist of a single test on a 1:6-scale model.

This work is 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.

10-7

This model has been des.igned , and construction is planned to begin in the near future.

All containment models have and will be tested pneumatically using nitrogen gas. The models are loaded in discrete pressure steps using a computerized pressure control system. Tests were conducted on four 1:32 scale steel models and a 1:8-scale steel containment. Extensive instrumentation consisting of strain gages, displacement transducers, pressure sages, and thermocouples were utilized during testing. During the testing of the 1:8-scale containment model, an acoustic detection system was used .to determine the presence and location of a leak in the model. Results and !

conclusions from the testing of the steel containment models will be I presented in the full paper. I l

The concrete model has been designed and will be 22 feet in diameter, 35 I feet high, and have 9 3/4 inch thick walls. It will have two equipment hatches, two personnel locks, and several smaller piping penetrations.

Major reinforcing will be #4 (1/2" diameter) size bars, with 8 layers through the thickness: 2 layers of meridional, 4 layers of circumferential, and 2 layers of diagonal steel. Testing is scheduled to occur in FY87.

Penetration Testing Both mechanical and electrical penetrations are being tested under calculated severe accident conditions. A background study and scoping analysis of mechanical penetrations identified penetrations with a potential for release and the release paths. From this study, certain penetrations, such as equipment hatches and personnel airlocks, were selected for full scale or scaled testing. Also, different combinations of seals and gasket materials and geometries were selected for testing. Two equipment hatches were a part of the 1:8-scale steel model, and the 1:6-scale reinforced concrete model will contain two different hatches. A test of a full-size personnel airlock under accident conditions is also planned. Electrical penetrations are also being tested. One full-size electrical penetration assembly has already been tested, and two others will be tested. Both leakage and electrical continuity are ucaltored during the test.

Analystes and Methodology Development Structural analyses are performed before the scale models and penetrations are tested. Release criteria for leakage and material f ailures are proposed so that the performance of the containment model or penetrations can be related to the predicted strains and displacements. The predictions are used to help plan the instrumentation and conduct of the tests. After the test, the adequacy of the analytical method and release criteria are evaluated by making comparisons with experimental data. The effects of thermal and radiation aging and aerosols on performance will also be considered. If necessary, refinements and modifications to the method or the criteria, or both, are implemented.

l l

10-8 l

EXPERIMENTAL VALIDATION AND IMPROVEMENT OF CORE DEBRIS / CONCRETE INTERACTION MODELS Ahti J. Suo-Anttila, J. E. Gronager, Richard E. Blose, David R. Bradley Molten / core concrete interactions have been recognized as impor-tant aspects of severe nuclear reactor accidents. Concrete erosion, large quantities of non-condensable gas, aerosols, and fission products are known to result from such interactions. A number of experiments have been conducted over the years to simu-late these interactions. A product of these experiments is the development of several computer codes that model such behavior.

Among those computer codes are CORCON:(1) a model of core debris concrete interactions, and VANESA:(2) a model of radionuclide i

release from molten core debris.

Six large-scale (up to 200 kg) melt-concrete interaction experi-ments have been conducted at SNL. The six experiments are four TURC series (transient tests) and two TheSWISS series (sustained four TURC experiments heating with an overlying water pool).

had the following melt compositions:

TURCIT - 200 kg alumina iron thermite TURCISS - 200 kg stainless steel 304 TURC2 - 200 kg UO 2 /Zr02 TURC3 - 200 kg UO2 /Zr0 2 /Zr The two SWISS experiments (1 and 2) were of somewhat smaller scale (45 kg) and utilized stainless steel type 304 for the melt composition. All of the experiments were conducted in a one-dimensional concrete crucible that utilized non-ablative MgG sidewalls; consequently, the ablation was one-dimensional.

The experimental results provided by these tests were unexpected when compared to CORCON predictions. The CORCON prediction of both the transient stainless steel (TURCISS) and thermite (TURCIT) tests indicates that their behavior should be quite similar. However, the test data indicate significantly different behavior. The Lilk pool temperature and concrete erosion rate of the thermite test wwre consistent with the presence of a rela-tively constant bulk pool to concrete surface heat transfer coefficient of 1500 W/mk. In contrast, the transient stainless steel test eroded at a much faster rate which could be charac-terized by a variable heat transfer coefficient of 5,000,to 10,000 W/mk. The difference in the t--ts is most likely due to the presence of alumina in the therm 4 c test. Alumina comprises in excess of 50% by volume of +% ermite mixture. Given a sufficient gas velocity through t ,

ermite mixture, a mixed pool of frozen oxide and liquid me al ..!1 result. The effective viscosity of a mixed slurry pool will be greater than that of a pure metal pool. In addition, the mixture thermal conductivity 10-9

will be lower. The net effect of these property variations will be a decrease in the heat transfer rate.

The CORCON model of melt concrete interactions does not consider well mixed pools, but rather separates the oxides and metallic phases into superimposed layers. This separat ion assumes the metallic part of the thermite to be adjacent to the concrete, thus the calculated thermite behavior turns out to be quite simi-lar to the calculated stainless steel behavior.

Transient' crust growth occurred in the TURC2 and TURC3 and SWISS 1 and SWISS 2 tests. In the TURC2 and TURC3 tests, no internal heating was present, so the crusts which formed were permanent and no ablation of the concrete was found. In the SWISS 1 and SWISS 2 tests, internal heating was present so that the crusts which were formed were later remelted. The net effect in the SWISS tests was to delay the onset of ablation by 5 to 10 minutes.

The CORCON model of crusting successfully predicted the behavior in the SWISS tests. However, in the TURC2 and TURC3 tests, CORCON predicted that significant ablation would occur. The reason for this discrepancy was due to the predicted melt point depression of the UO 2 /Zr02 -melted-concrete mixture. The pre-dicted rate of melt point depression was rapid enough to maintain a significant rate of erosion before freezing of the melt pool occurred, thus resulting in a substantial overprediction of the amount of ablation.

A new model has been developed at Sandia National Laboratories in order to improve upon the deficiencies encountered in the CORCON simulations of these experiments. The new model includes a mech-anistic approach to pool mixing and segregation. The model consists of a balance between turbulent mixing effects and buoy-ancy effects. The turbulent mixing is caused by the stirring and agitation by the bubbles which. are released from the concrete.

Buoyancy effects arise because dhe oxides and metallic phases are immiscible and of differing densities. The turbulent mixing is affected by pool viscosity. The presence of a slurry, such as frpzen oxide-liquid metal, will decrease _the level of mixing due to the increased viscosity. When applied'to this set of experi-ments, the model is able to predict both the thermite and the stainless tests. It also predicts rapid crusting for the TURC2 and 3 tests. This model thus gives quantitative support for the hypothesis that, in some cases, melt-concrete ablation can be a well-mixed slurry with ablation properties significantly differ-ent from,those currently assumed in the CORCON model.

In summary, experimental data, together with a new numerical model,-imply the need for new model development in the CORCON code. The new models should include possible mixing of the oxide and c.etallic layers as well as further' refinement in the area of crust growth and decay.

l l

10-10 l

l REFERENCES

1. R. K. Cole, Jr.; D. P. Kelly; M. A. Ellis; "CORCON MOD 2:

A' Computer . Program for Analysis of Molten-core Concrete i.

. Interactions," NUREG/CR-3920, SAND 84-1246, August 1984, t

2. D. A. Powers, J. E. Brockmann, A. Shiver, "VANESA:

A Mechanistic Model of Radionuclide Relase and Aerosol Generation During Core Debris Interactions with Concrete,"

to be published.

t i-l l

i i

10-11

FEVIN OF TiiE LARCE-SCALE CORE-CONCRETE INTERACTION EXPERIMENTS AND ANALYSIS AT THE KFK HETA FACILITY R. K. Cole, J r. , Sandia National Laboratories M. Rei mann , Kernforschungszentrum Karlsruhe In the event of an uncontrolled core-melt accident in a nuclear power plant, molten core materials would contact the concrete basenat of the reactor con-tainnent buil d i ng following failure of the reactor pressure vessel. Steam and non-condensable. gases produced by the interaction between these core materials and the concrete could cause the containment to f ail by overpressurization af ter a period of several days. This interaction may also lead to the genera-tion of aerosols, and to further release of fission products f rom the fuel in the debris pool. Consequently, an understanding of the processes is essential to estimation of the potential release of radioactivity. This point was empha-sized in the source term study by the American Physical Society.

Because no data are available (or expected) for the reactor core-melt case, risk assessments must be based on results of experiments and on computer model-ling of the physical processes. Melt / concrete interaction work carried out for many years, both at KfK and at Sandia, has resulted in development of computer codes to model the interactions. These codes, CORCON at Sandia and WECHSL at KfK, and also the KAVERN code of KWU, were initially based on the results of separate-ef fects experiments with simulants and of transient tests using more prototypic materials. The codes have been used for calculation of hypothetical core-melt accidents.

The BETA f acility was constructed at KfK to provide data for code development and validation. In this facility, large melts (up to 300 kg of steel and 300 kg of oxide) are produced by a thermite reaction in an external vessel, and poured into a concrete test crucible. The melt is then inductively heated so that interactions may be observed during an extended period of time under quasi-steady conditions. Additives in the initial thermite are used to produce a metallic phase with a composition corresponding to stainless steel, and to modify the properties of the oxidic phase. Power inputs are chosen so that heat flux, temperature, melt front propagation rate, and most other relevant pro-cesses are simulated in a 1:1 scale.

Approximately 200 channels of data are recorded on magnetic tape during each experiment. In some cases, post-test data reduction is required to obtain the actual parameters of interest. These on-line data include the power input to the melt, temperatures at 110 points within the concrete, temperatures of the melt phases, release rate and composition of evolved gases, and arcosol concen-tration in the of f-gas line. The propagation o'f the melt front is inferred f rom the time of f ailure of the thermocouples in the concrete. Samples are taken of the melt, of the evolved gases, and of aerosols, for of f-line analysis. Af ter each experiment, the crucible is sectioned and the post-test cavity shape ob-served. In addition, TV and film cameras provide a visual record of melt be-havior including crust formation and gas flow, and of aerosol production.

10-13 i

l I

b l

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

i l

j in the first four heated experiments, conc rete erosion was observed to be pre-

! dominantly downward. This was in cont rast to transient experiment s, where ero-I sion was relatively symmetric, and some modifications to the heat-transf er

models in the codes were necessary to reproduce the observed behavior. The most important of these was the introduction of a partial film collapse model, ana-logous to transition boiling, into both WECllSL and CORCON.

,i Following these modifications, the next group of experiments was planned and run, and the codes were used to make " blind post-test" calculations. (Only the

initial conditions and the measured power history were availabic to the calcu-lators.) Agreement with measurements, including erosion rates, melt tempera-tures, and gas generation, was reasonably good. Data f rom these experiments ,

j were than used to make further adjustments to the models, and the process was repeated with a new set of experiments and " blind" calculations. Agreement was j again satisfactory in general.

The dominance of downward ablation noted above has also been observed in later

experiments, although the degree is slightly less for those with lower power inputs. Initial cooling of the melt has been very rapid in all experiments, l with the melt quickly attaining a temperature which is only slightly above the j solidification point of steel in most cases, and less than 300 K higher for

] experiments at maximum power input. Dispersion of metal droplets into the oxi-dic phase is observed in some experiments. Because the droplets are too small i ;

' to couple to the induction field, this causes a reduction in coupled mass and therefore in power input to the melt. The largest fraction of the evolved gas j is hydrogen; on- and of f-line analysis are in good agreement. Only very modest ;

} aerosol generation is observed visually; this result is confirmed by laser t

! scattering data and by filter samples. Concentrations at system conditions are

{ a f raction of a gram per cubic meter.

1 All experiments performed to this time have used silicate concrete of the type

common in Germany. Two crucibles with high carbonate content have now been

fabricated using material imported f rom the USA. One is full carbonate concrete

} (the so-called "CRBR" concrete), while the other, of limestone / common sand

} concrete, has an intermediate carbonate content. Both of these concretes pro-j duce substantially greater amounts of gas, with a much greater CO2 content, i compared to the silicate concrete. Experiments using these crucibles will be

conducted in late 1985, and should provide a significant test of the heat-i transfer modelling. They should also help to answer some remaining questions f concerning areosol generation. >

I i

10-14

_ _ . _ _ . _ _ _ __ *_ _ _ _ ~~ _ _ . _ _ - - - - --

l l

SAND 85-1725A l

THE INFLUENCE OF REACTOR GEOMETRY ON '

l THE BEHAVIOR OF DISPERSED DEBRIS

William W. Tarbell Marty Pilch John E. Brockmann Sandia National Laboratories Albuquerque, New Mexico

SUMMARY

Assessing the consequences of severe nuclear power plant accidents has until recently been based on the results of the Reactor Safety Study (RSS). Subsequent to the accident at Three Mile Island, there has been increased emphasis on the use of newer Probabilistic Risk Assessments (PRAs) to quantify the risk associated with operating a nuclear power plant. Recent studies have used and expanded upon the RSS methodology by incorporating improved analytical techniques and the more extensive data base now available. A significant conclusion from these studies is that in a large number of accident sequences the reactor pressure vessel (RPV) may fail while the primary system is pressurized. Failure of the RPV while pressurized may result in the jet-like ejection of molten core debris into the reactor cavity.

In some PRA analyses, the ejected core material is assumed to flow out of the cavity and spread evenly over the floor of the containment building, forming a shallow debris bed that can be easily quenched by existing spray systems. Because a protracted core debris-concrete interaction is avoided, it is predict ~ed that the accident promptly terminates without significant gas or aerosol generation or added pressurization of the containment building.

Recent experiments performed at Sandia National Laboratories using one-tenth scale models of reactor geometry have confirmed that debris is dispersed from the cavity. The melt expulsion is not a benign film-like flow of material, but it is an energetic discharge of melt particles and high velocity gas. The ejected material is accompanied by an intense aerosol generation. Typically, the debris recovered from the tests has a size distribution that follows a broad log-normal pattern with a median size on the order of a millimeter or less. The small

= This work is 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 No. DE-AC04-76DPOO789.

10-15

.- .=.

1 size of the particles enhances their ability to impart energy to the atmosphere and to act as a global ignition source for hydrogen that may be present.

A number of researchers believe that the fraction of the core mass discharged from a real cavity will not be as great as that expelled in the one-tenth scale HIPS (High Pressure Melt Ejection) tests (over 95 percent), principally due to the natural obstructions and restricted flow paths that exist in )

most plants. They argue that only the very small particles, representing an insignificant portion of the total available mass, will be carried by the gas stream into the containment.

Furthermore, in the few plants where it is predicted that a significant fraction of mass exits the cavity, it is assumed that the debris will be unable to propagate extensively because of the obstruction caused by numerous containment structures.

The disparate opinions of the various researchers lead to j several major unresolved issues associated with melt expulsion and direct containment heating: (1) the extent of debris dispersal (based on plant type and system pressure), (2) the interaction of the debris with containment structures, and (3) the debris to atmosphere energy transfer processes. The HIPS program is designed to provide appropriate experiments and associated modeling to understand these uncertainties. This 3

paper discusses the results of two experiments that investigate i the second of these issues--the interaction of the debris with j containment structural features.

The experiments are designed to investigate the potential for in-containment structures to trap and retain debris dispersed from the cavity exit. The simulated structures are constructed of concrete or steel to determine if the debris form a stable crust layer. If the ejected material freezes on i

structures, it cannot be swept into the containment and heat the atmosphere. In one test a simple wall is placed in the path of the debris stream to study the interaction of the ejected material with an obstruction. Crust formation on a steel pipe l is also examined. In the second experiment, the debris is directed into a simulated instrument shaft room to create a highly restricted path for the flow of material. Debris can only escape the structure through a single opening facing away from the normal direction of propagation. The material must necessarily undergo numerous interactions within the structure before escaping.

The results of the two tests show that as the debris stream j reaches the obstruction, the gases divert while the bulk of the debris impacts onto the surface. Unlike the analytical predictions; however, only a small fraction of the impinging debris is retained by the obstruction. For surfaces of steel or concrete, the impact apparently causes fragmenting or splashing of the particles and subsequent reentrainment into the gas 10-16 l

stream. Only a minor fraction of the particle's energy is lost to the surface, so that the overall velocity of the debris

! remains relatively unchanged. For a plant geometry, this may mean that the debris can undergo a number of interactions with structures and still retain sufficient velocity to be transported into the containment dome.

The experiments do not support the assumption that containment features greatly mitigate the potential for direct heating. -The debris is not significantly slowed nor deposited by the interactions with obstructions. Work is continuing in

order to uhderstand the debris dispersal mechanisms, interactions with containment structures, and energy transfer

processes. A methodology is being developed for applying the i results to all plants. Future testing will use detailed I representatives of containment structures placed in a large interaction chamber to determine debris behavior and the energy

) imparted to the atmosphere. This testing is needed to provide i

the basis for accurate modeling of the phenomena occurring during this type of accident sequence. ,

i l

1

?

i a

)

i i

i l

1 i

i i

10-17 i

.. _. _ _ _ _ _ _ . _ _ . - _ _ _ _ . _ . _ _ _ . _ . . _ . _ _ __ . . . ~ _ _ _ _ . _ _ _ _ _ _

i Analysis of Molten Fuel-Concrete Interactions and Fission Product Release from Ex-Vessel Core Debris-l D. R. Bradley Sandia National Laboratories Albuquerque, NM 87185 Introduction l The CORCONI and VANESA2 computer models have been developed l at Sandia National Laboratories (SNL) to model ex-vessel core debris-concrete interactions and the resulting aerosol and

! fission product release. Both models are integral components of l the NRC suite of computer codes used for severe accident source term analysis.

l I As demonstrated in the QUEST study 3 at SNL and in the American Physical Society review of NRC source term research4, i significant uncertainties remain in this area of source term

{ assessment. The ongoing experimental and-analytical efforts at j SNL are devoted to a better understanding of governing ex-vessel j phenomena and a reduction in ex-vessel source term uncertainty.

1

Background

I' CORCON models core debris-concrete interactions and feeds ,

pertinent information to VANESA. VANESA then calculates the aerosol and fission product release resulting from these interactions. There are four primary factors that determine the 4

aerosol and fission product release calculated by VANESA: the temperature of the melt, the gas flow rate through the melt, the composition of the gas stream, and the availability of a water pool to scrub the aerosols. Each of these is calculated by l CORCON.

! Melt temperature is a function of melt-concrete heat i transfer, upward heat loss from the melt surface, and the j ablation temperature of the concrete. Gas flow rate is governed by melt-concrete heat transfer, gas content of the concrete, and the rate of gas reaction with the melt. Gas composition is

!

This work supported by the United States Nuclear J Regulatory Commission and. performed at Sandia National l Laboratories which is operated for'the U. S. Department of l

Energy under Contract number DE-AC04-76DP00789.

i 1

4 i

i 4

t 10-19 l

, - , - - , , - _ _ . - , - , - , . - _ . . . . - _ . . , . - - , . , . - - . . _ . - 0,__--,_,_.__,,,,_..,,_-,.__

j

$ determined by these chemical reactions with the melt, and is

?-

therefore a function of melt composition. The scrubbing j efficiency of a water pool depends on its depth and the aerosol particle size distribution.

Recent Observations

! Four significant conclusions can be drawn based on recent i experiment observations and accident analyses. First, an 3 accurate melt composition is important in both accident analysis j and experiment simulation. Most important of the severe melt j constituents is zirconium, which can have a major influence on i gas. flow rate, gas composition, and melt temperature. This is i especially true if the reactor cavity concrete contains a significant amount of carbon dioxide. Second, while it is present, a water pool is very efficient at reducing aerosol and fission product release. For example, decontamination factors on i the order of 10 or more have been experimentally obs.4rved.

i Conversely, a water pool is not very effective at reducing melt-I concrete heat transfer and therefore has little effect on i concrete ablation rates. This is true because the melt surface

! remains sufficiently hot that a stable boiling film exists. Ilent j transfer is therefore dominated by thermal radiation regardless i of whetherra coolant is present or not. Finally, the tellurium '

j release model in VANESA has been validated by comparison to l experiment results. This is significant because VANESA j calculates that tellurium is released continuously over many hours, and it is therefore an important long-term radionuclide

source.

t

References 1

4 j 1. R. K. Cole et al., "CORCON-Mod 2: A Computer Program for '

Analysis of Molten-Core Concrete Interactions," NUREG/CR- ,

j 3920, SAND 84-1246, Sandia National Laboratories, August 1984.

l 2. D. A. Powers et al., VANESA: A Mechanistic Model of i Rad'ionuclide Release and Aerosol Cencration During Core i Debris Interactions with Concrete, SAND 85-1370,

] NUREG/CR-4308, Sandia National Laboratories, to be published.

3 R. J. Lipinski et al., " Uncertainty in Radionuclide Release

, Under Specific LWR Accident Conditions," Vols. I-IV, SAND 84-

0410, Sandia National Laboratories, 1985.

4. R. Wilson et al., "Radionuclide Release from Severe Accidents l at Nuclear Power Plants," to be published in Reviews of i Modern Physics.

, i l

l 10-20

.-.___.o-._- - .. . - - _ - , - , , - - - . . , - - , . - - -, - _. - -

(

BNL SEVERE ACCIDENT SEQUENCE EXPERIMENTS AND ANALYSIS PROGRAM

G. A. Greene, T. Ginsberg and N. K. Tutu Brookhaven National Laboratory Department of Nuclear Energy Upton, New York 11973 ,

r Analysis of containment response to severe accidents in LWRs requires mathematical characterization of several sources of pressure and temperature loading on the reactor containment buildings. Experiments are in progres's at BNL in support of analytical model development related to: (1) core debris-l water thermal interactions and (ii) molten core-concrete interactions. The

! work supports development and evaluation of the CORCON, VANESA, MARCH, CONTAIN l and MEDICI computer codes, under development at other NRC-contractor labora-tori es.

l Progress is described below in the two areas of: (i) core debris thermal-hydraulic phenomenology and (ii) heat transfer in core-concrete interactions.

Core Debris Thermal-Hydraulic Phenomenology A pour stream of molten corium could encounter a pool of water under both in-vessel and ex-vessel circumstances. The mixing of the melt with the avail-able water would involve breakup of the melt stream, followed by mixing of the smaller melt elements with the available pool of water. Both processes are l being considered in the area of core-debris phenomenology. A series of exper-iments are being carried out to study the mixing of pre-fragmented debris with saturated water. This work is directed towards providing a methed for calcu-lating the distribution of melt, water and steam in the thermal interaction region. Experiments are designed to evaluate one- and two-dimensional models of core melt-water mixing in a drop-mode configuration. They are performed with spherical steel particles, preheated to a prescribed temperature and dropped into a pool of saturated water of depth up to one meter. Experiments have been carried out in a one-dimensional system where the particle " stream" was of the same diameter as the water pool and in a two-dimensional system where the pool cross-sectional area is significantly greater than the particle

" stream" diameter. Experiments have been carried out with 3 , 6 , and 12-mm

diameter particles, heated to 1000K. Results from the one-dimensional experi-i nent'; suggest that the particles tend to become fluidized as they begin to

penetrate the water as a result of steam generation from the leading edge of particles as they interact with the water. The 2-D experiments display sig-nificantly lower steam generation rates during the early interaction period,

, and do not display the strong fluidization characteristics as do the 1-D experiments. The 2-D experiments are in progress at present, and analysis of them is incomplete. Analytical modeling is also in progress.

A program is under way to provide an understanding of the transient quenching of in-vessel debris beds and to develop transient debris bed quench j models that can be incorporated in the SCDAP code. The experimental results

Work performed under the auspices of the U.S. Nuclear Regulatory Commission.

l l

l

. I 10-21 i

i

,.,.,,---,n- - - - .. , , , , , - - , - . , .- ,- ,,-.-----.-w,.., , , , --n ,r , ,----n , - , -,-----.n.---,- -+ n- .----m,,----w-- , , - - ,

would, in addition, generate a data base that can be used to verify other transient quench models. Work performed to date deals with the case of a one-dimensional debris bed being quenched by the injection of liquid coolant at a constant rate from the bottom of the debris bed. Previous results have shown that for small liquid supply rate and low initial particle temperature, the bed quench process is a one-dimensional frontal phenomenon. For large liquid I supply rate and high initial bed temperature, the bed quench process is a com-plex multi-dimensional phenomenon. New experimental results, performed with J i

double the previous bed height, suggest that for large liquid supply rate and high initial bed temperature, the instantaneous peak heat flux increases with increasing bed height. This instantaneous peak heat flux is observed to exceed the heat flux limit based upon assuming liquid evaporation rate within the debris bed to exactly equal the rate at which liquid is injected into the debris bed. Temperature traces from thermocouples installed within the parti-cles indicate that as the particles quench, they go through a " droplet flow regime" and a " slug type" two-phase flow regime.

Heat Transfer in Core-Concrete Interactions: Coolant Layer Behavior Analyses of severe core damage accidents in LWRs have demonstrated the possibility of the ex-vessel attack of molten core debris upon the structural concrete. The C0RCON-M002 and VANESA computer models have been developed to analyze the ex-vessel core-concrete interaction and fission product release during this stage of the reactor accident. Under some circumstances, it may be possible for a pool of water to exist over the molten core-concrete inter-action. The presence of this water ,;ool will have a significant effect upon the molten corium heat balance, fission product decontamination, steam genera-tion rate, and possible steam explosions in the reactor cavity. Experiments have been carried out to measure the film boiling heat flux of a boiling fluid over pools of liquid metal with a non-condensable gas flux through the liquid metal to simulate the concrete decomposition gases. The results indicate that for zero gas flux, Ril boils stably over liquid metal pools and agrees with the Berenson model of film boiling within +3%. The non-condensable gas flux tends to disturb the interface, increasing the interfacial surface area with the Ril, resulting in an increase in the film boiling heat flux to the Ril.

The relation between the film boiling heat flux and gas injection velocity appears to be linear.

l 10-22

METHODOLOGY FOR CODE ACCURACY QUANTIFICATION L. N. Kmetyk, R. K. Byers, M. C. Elrickt and L. D. Buxton Sandla National Laboratocles Albuquerque, New Mexico Code assessment is performed to evaluate the accuracy of a code and its range of applicability. To achieve this, the code is exercised against a variety of experiments. Conclusions on thermal / hydraulic code accuracy drawn from previ-l ous studies have been mostly phenomenological (i.e., "the code predicted all j modes of natural circulation") and/or qualitative (i.e. , "the code predicted break flow reasonably well"). [1] However, there is an increasing emphasis within the URC code assessment ef fort to formulate more quantitative conclu-slons. [2]

Some limited quantitative assessment has been done (3], but only of important single-valued key parameters (e.g., peak clad temperature). These are neces-sary but not sufficient measures of a code's accuracy throughout an accident or transient. The purpose of this study is to develop a method of quantif ying code accuracy for continuously-varying parameters (e.g. , Primary system pres-sure, break flows) over the course of a transient. Using a common method to quantify the differences between data and code predictions will allow combin~

ing results from a number of independent assessors, will provide broad-based information on code accuracy for app 1Leations to regulatory needs and will help define further code development needs.

Using the concept of multiple safety barriers (4), a ilmited number of key assessment parameters have been selected as a function of the class of tran-sient (e.g., LOCAs, operational transients) and different plant design characteristics (e.g., B4Rs, B&W OTSCs). The selection of these parameters considered the normal availability of corresponding data.

Quantifying differences between code predictions and measured results for key parameters presupposes both are available at common times. Consequently, the code results and/or the data may have to be " massaged" before such a compacL-son is possible. This includes smoothing very noisy data or collapsing a number of thermocouple traces to a cell-averaged temperature, while retalning a measure of the variation in the original data for later use. Procedures for performing this task have been developed.

A quantitative accuracy statement for a key pIrameter can be derived for the entire time range covered by both measurement and calculation. Alternatively, individual accuracy statements can be obtained for time intervals chosen to Leolate different governing phenomena during the accident or transient being studied.

This work was supported by the US Nuclear Regulatory Commleslon, Office of Nuclear Reactor Regulation, and performed at Sandla National Laboratories which is operated by the US Department of Energy under Contract Number DE-AC04-76DP00789 tDLkewood - Division of Kaman Sciences Corporation 11-1

The current methodology we are using for the quantification of the code-data differences for a key parameter is based on simple statistical definitions.

The results include thn mean code accuracy and its accuracy variation in each major time interval for the given assessment analysis, and the corresponding absolute values of the accuracy mean and variation.

This initial code accuency estimate is relative to the " base" experimental i

data, and does not take'into account possible errors or variations in the data due to instrumentation uncertaintles or to the presence of several independent measurements in a region represented by a single code variable. Methods of factoring such different data uncertainty sources into the code accuracy conclusion are currently being addressed; the first approach belns evaluated i is simply to assume normal distelbutions for both the code-data dif ference and

! the data variability.

Several other problems remain to be addressed in developing a generalized accuracy quantification methodology for individual continuous key parameters.

These generally involve defining meaningful major time regions and code-data j differences when key events occur at very different times in the test and the calculation.

i In order ultimately to apply the results of code accuracy quantification to plant studies and regulatory needs, a further problem being studied is how to combine individual " localized" accuracy statements into more global code j accuracy estimates (e.g. , covering dif ferent major regions, analyses, plant i designs and users). The key parameters for different transients and plant designs and the major time regions were specifleally chosen to allow such combination into overall accuracy statements; this also requires a common methodology for quantifying the accuracles for any given analysis.

t

Using our prellmlnary methodology, we have calculated some quantitative code accuracles from test data and either ouc previous RELAPS/ MOD 1 or our current TRAC-PFl/ MODI assessment calculations. These examples help show how the various generic accuracy quantLfication issues relate to real problems.

REFERENCES

1. L. N. Kmetyk, "RELAPS Assessment: Conclusions and User Guidelines,"

NUREG/CR-3936, SAND 84-1122, Sandla National Laboratories October 1984.

1 2. F. Odar and D. E. Bessette, " Guidelines and Procedures for the International Thermal-Hydraulic Code Assessment and Applications Program,"

Nuclear Regulatory Commission, April 1985.

j 3. L. W. Kmatyk, L. D. Buxton and S. L. Thompson, "RELAPS Assessment:

1 Quantitative Key Parameters and Run Time Statistics," NUREG/CR-3802, 4

SAND 84-1013, Sandla National Laboratories, October 1984.

! 4. F. Odar, "Results of Independent Assessment," at Tenth Water Reactor Safety Research Information Meeting, October 12-15, 1982, published in i Proceedings, NUREC/CP-0041, Vol. 3 of 6, January 1983.

11-2

SUMMARY

i UNCERTAINTY DEVELOPMENT AND APPLICATION Gary E. Wilson and Glenn S. Case i

! Idaho National Engineering Laboratory j EG&G Idaho, Inc.

P. O. Box 1625 l Idaho Falls, Idaho 83415 ,

i

! The Nuclear Regulatory Commission (NRC) has organized an i International Code Assessment and Applications Program (ICAP) to

! supplement the domestic code assessment which it sponsors. The !

NRC intends to use the international and domestic assessment l studies to form a data base which will be used to quantify code ,

uncertainties in the simulation of plant transients associated with nuclear safety analyses. The NRC desires that the quantification of code uncertainty be an engineering measure

based on an adequate statistical methodology. The uncertainty

measurements will, in the short term, guide the continued ,

{ improvement of safety analysis codes, and in the long term, I support regulation of commercial reactors. The development of s an adequate, statistically based, uncertainty methodology is underway and the current status of that development is the subject of this paper.

A primary element in the successful characterization of code uncertainty is the selection of key assessment parameters.

These are the minimum set of parameters whose uncertainty

! determination will adquately quantify the code's quality of j performance in its application to safety analysis. The selection of the key parameters are based on the controlling phenomena expected in different types of transients and within

, the different phases of each transient. The necessary minimum ;

) set has been established by the NRC and its contractors and is :

given in the paper.

[

J Any number of statistically based methodologies can be posed j to give a measure of code uncertainty. Each method has its advantages and disadvantages associated with complexity,

{ statistical rigor, demands on the experimental data base, i sufficiency, applicability to the desired objective, etc. Two i

of the " front runner" methodologies have been tested with. !

1 1

11-3

code / data comparison data from a TRAC-BD1/ MOD 1 simulation of a subscale boiling water reactor small break experiment. The results from these studies, including their comparison, are given in the paper. Recommendations relative to the selection of a " standard" methodology are also provided.

I 11-4

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

l l

Improvements of BWR LOCA / ECCS analysis in Japan.

K.Yahagi The Tokyo Electric Power Company, Inc.

l :

1. Introduction l

i

} 3apanese BWR utilities and venders ( Toshiba & Hitachl ) have been jointly [

studying LOCA / ECCS phenomena since 1977 and reached the stage to ,

conclude that existing BWR have enough safety margin for LOCA Based on I

the results of such studies, Japanese BWR venders and GE developed

improved LOCA / ECCS evaluation code SAFER-03. In this paper, the l

features of SAFER-03 code as well as the joint research results and code

! qualification against these experiments are summerized. ;

l 2. Development of SAFER - 03 i

.; SAFER-03 code has been developed jointly by Japanese BWR venders and GE j by modifying SAFER-02 code so as to establish the LOCA / ECCS evaluation .

j model ( EM ) for all types of BWR including the non jet pamp type plants such j as BWR-2 and ABWR. The main modifications in SAFER-03 are:

1 i j (1) Mist cooling model is added to steam cooling model. ,

4 (2) Extended new spray core cooling correlation.

- (3) New CCFL correlation for the top / bottom of bypass region.

(4) Modified multi bundle model. t 1 These model modifications are based on the following joint research !

j experimental results.

i i

{ o Bottom of bypass correlation was obtained from the experiments using ;

1 GTCC test facility which mocks up bypass bottom region ( consisting of ;

four fuel bundles, one control rod and one fuel support ). >

a

o Spray heat transfer coefficient which is a function of pressure, flow, and i rod superheat was obtained from the SHTF experiments. The test 4 j conditions of SHTF experiment covered low flow and high pressure condition.

1 i j o Mist cooling model is based on the experiment using RRTF test facility 3

which was also used to evaluate BWR steam cooling effect and side entry l

oriffice CCFL correlation.

i^ o Top of bypass CCFL correlation was obtained from the ESTA experiments which were conducted by Toshiba.

j Among the above experiments, top / bottom of bypass CCFL correlations and mlst cooling correlation were recently obtained. ,

t f

}

l 11-5

i i

k i 3. Qualification of SAFER - 03 code i

)

The SAFER - 03 has been qualified by LOCA integral system tests such as TBL ( Japz.nese BWR utilities and venders ), ROSA - Ill( J AERI ), FIST-ABWR (GE) and FIX - 2 ( Sweden ) tests as shown in the table. These qualifications

! cover not only large break LOCA but also small and intermediate break i j LOCA and other pipe breaks such as steam line break. The qualifications by i TBL and ROSA - III are for the jet pump type plants, while those by FIST - '

! ABWR and FIX -2 are for the non jet pump type plants. The comparisons j between the test results and calculation by SAFER - 03 show good agreement j and the appropriate safety margin required for EM is maintained. ,

L i 4. Conclusion SAFER - 03, which is applicable to all types of BWR plants, was developed j based on the latest LOCA / ECCS research results. This code is now being ;

reviewed by the Japanese regulatory body.

j The fo!!owing advantages are expected from the use of SAFER - 03 code in i the future. ,

t

) o More economical fuel design which might have provided some impacts by i the previous LOCA analysis will be easily accepted.

o Operational flexibility in core mancavability and surveillance frequency j of ECCS pump.

i j o Optimization of ECCS design.

l 1

! o Establishment of the more adequate emergency operational procedures in

case of LOCA.

1 I Table : SAFER - 03 qualification 1

i' Large Break LOCA Small Break LOCA others j TBL 0 0 0 f ROSA .III O O 1

FIX -2 0 0 ;

FIST - ABWR 0 0 0 I

i i

I i

i l 11-6 t

l

NEW JAPANESE CGRRELATIONS ON CORE COOLING AND CCFL CHARACTERISTICS DURING BWR LOCA L B.Nagasaka Toshiba Corporation l

l

.IMTRODOCTION Revised version of SAFER, a realistic analysis code for LOCA in BWR, has been developed under the cooperative research among Toshiba, Hitachi and General Electric. In this revised SAFER, new Japanese expreimental correlations on core cooling and CCFL

< characteristics are incorporated. This paper summerizes the new correlations.

CORE COOLING CHARACTERISTICS TESTS (a) Mist Flow Heat Transfer l The tests were performed by establishing, first, the steady state single phase steam cooling conditions and then injecting saturated water into the bypass, which provided the dispersed droplets through the leakage at the bottom of the heated bundle.

It shows that the heat transfer coefficient is larger than Saha's mist cooling model(l) based on tube data due to trap of the droplets by spacers.

(b) Spacer Effect In both steam cooling and mist cooling tests, heat transfer coefficients were observed to increase right after the spacer due to turbulence promoting effect. Same multiplier was obtained for both steam and mist cooling heat transfer coefficients.

(c) Spray Heat Transfer d

The tests were conducted by spraying water within the extended channel box in the upper plenum, after the prescribed intial conditions of power, pressure and rod surface temperature were achieved. The heat transfer coefficient was evaluated from the rod surface temperature transients assuming the bulk temperature to be saturated. Correlations dependent on spray flow rate, pressure and wall superheat were obtained for both average rod and PCT rod. New correlation agrees with test data within 115% as shown in Fig.l. These correlations also agrees well with CORECOOL.

_CCFL CHARACTERISTIC _S_TESE i

(a) Top of Bypass CCFL Using Eighteen degree Section Test Apparatus (ESTA), which mocks up a BWR plant with full befght from jet pump bottom to stgndpipe top and main components within shroud are cut into an 18 sector, top of bypass CCFL characteristics were obtained.

11-7

The tests were conducted by injecting steam into bypass and supplying saturated water into the upper plenum with the core filled with saturated water. New correlation gives less restrictive CCFL drainage compared with the current correlation obtained by annulus geometry tests (2), as shown in Fig.2.

(b) Bottom of Bypass CCFL Thus far there were no data on bottom of bypass CCFL. Test facility, composed of one guide tube, one control rod, one fuel support, four bundles and simulating exactly the bypass bottom and core inlet regions, has been constructed. The tests were conducted by injecting steam into the guide tube and supplying saturated water into the top of bypass. New correlation gives more restrictive CCFL drainage than that of top of bypass CCFL.

CONCLUSION New correlations on CCFL and core cooling characteristics were obtained and incorporated into SAFER, thereby more realistic core cooling analysis during BWR LOCA became possible.

REFERENCE (1) P.Saha, "A post-dryout heat transfer model based on actual vapor generation rate in dispersed droplet regime",

NEDE-13443, May. 1976.

(2) D.T.Hanson, et al., "ECCS performance in the SEMISCALE geometry", ANCR-ll61, June 19 74.

'a ,-

, , n tst ,'.

c4 w ,'

,/ . ,' '

C.5

% i2 M l ,

4 15: N NEW CORRELATION k

,/ .,,' A 0.4 /

? ,','/*, , y

} c,3 0* *'

\.

,' ' w 4 s' ,-

t En .,'

',' C L>

< ,* C...

c p;s a ,

y a -

lW t{ b.

d ,,'

,' CURRENT 6, t,i E* ' '

CORRELATION .

l

' ' .\

0 4 12 16 C.c c.) 0.2 0.3 C.4 0.5 0.6

! O 8 j CORRELATION [ Btu /h ft .F] (jE)

D i Fig.1 COMPARISON OF AVERAGE ROD Fig.2 TOP OF BYPASS CCFL CORRELATION WITH TEST DATA CHARACTERISTICS l

11-8

l APPLICATION OF TBL RESULTS TO BWR PLANTS M. Naitoh, M. Murase, H. Suzuki Energy Research Laboratory, Hitachi Ltd.

J.A. Findlay, F. D. Shum General Electric Company; San Jose, California

SUMMARY

The TBL (Two Bundle Loop) large break tests '

showed good cooling in the high power bundle due to parallel channel effects, which were experimen-tally observed under the co g ons of countercurrent, gaa-liquid flow in parallel vertical channels Parallel channel models have been developed and used to predict hydraulic behavior in the small-scaled air-water and steam-water systems simulating a BWR core. The models have been applied to a transient analysis and confirmed by the TBL results. Finally, the hydraulic behavior in a BWR plant has been evaluated using the models.

The TBL-1 tests '

for the large breaks and TBL-2 tests for the small and intermediate breaks were conducted. The tests covered the effects for the break location, break size and ECCS failure mode. The tests showed the following results:

o which was The highest consistent peak cladding temperature was 650*C (92g ,and TLTA (Two Loop with Test Apparatus). g -III (Rig of Safety Assessment) o Typical differences of thermal-hydraulic responses between bundles were not observed in the small and intermediate breaks.

o In the large breaks, liquid was maintained in the bundle even after mixture level formation in the lower plenum due to CCFL (counter-current flow lim y gg) at the g ry orifices, which was also observed in ROSA-III and TLTA . Thermal-hydraulic responses differed between the bundles due to the CCFL (i.e., parallel channel effects) and the period of core uncovery was shorter in the high power bundle than in the low power bundle.

Experiments g gyntercurrent gas- y gyid flow was performed using small-scaled air-water and steam-water systems simulating a BWR core. The test parameters were the spray water flow rate, bypass leak flow rate, gas generation rates in the channels and the lower plenum, and entry orifice size.

The typical results were as follows:

o Three different flow patterns of liquid down-flow, countercurrent gas-liquid flow and cocurrent up-flow could appear simultan also observed in SSTF (30' Sector Steam Test Facility) ygysly, which were o Liquid down-flow, countercurrent flow and coeurrent up-flow could easily appear in the channels with small entry orifice, large entry orifice and low power or high bypass leak flow rate, and large entry orifice and high power or low bypass leak flow rate, respectively.

11-9

On increasing the gas generation rate in the lower plenum, flow pattern transitions from liquid down-flow or countercurrent flow to cocurrent up-flow occurred at the peak pressure drop point. On decreasing the gas generation rate, flow pattern transitions from cocurrent up-flow to countercurrent flow or liquid down-flow occurred at the local minimum pressure drop point. The observedbyWallis,etal.gvpatterntransitiog)werethesameas and Piggot, et al i

Experiments on countercurrent gas-liquid flow in parallel vertical channels and quasi-steady state calculations on hydraulic behavior in a BWR core showed that flow patterns were liquid down-flow in peripheral bundles with small entry orifice and countercurrent flow or cocurrent up-flow in central bundles with a large entry orifice. Therefore, the core was divided into three I

regions of liquid down-flow, countercurrent flow and coeurrent up-flow. The i

parallel channel models were composed of the flow split and flow pattern transition models, which calculated the steam flow rates from the lower plenum into each region and pressure drop in the core, and the numbers of the countercurrent flow and co-current up-flo g ndles. These models were applied to the transient analysis program, SAFER i

In order to verify the flow split model, a transient analysis for a TBL large break test was performed and close agreements of thermal-hydraulic l responses between the calet' lated and measured were obtained. Then, transient responses for a BWR plant were analyzed and the capability to calculate the number of the countercurrent flow and co-current up-flow bundles was confirmed.

! REFERENCES

1. M. Murase and M. Naitoh, J. Nucl. Sci. Technol., 22(3), p. 213 (1985).

2. M. Murase and M. Naitoh, J. Nucl. Sci. Technol., 22(4), p. 301 (1985). '

3. D.M. Speyer and L. Kmetyk, Nuclear Reactor Safety Heat Transfer, p.

55, ASME, New York (1977).

i

4. G.B. Wallis, et al., Int. J. Multiphase Flow, 7, p. 1 (1981).

5. B.D.G Piggot and M.C. Ackerman, Heat Transfer in Nuclear Reactor Safety, p. 361, Hemisphere Publishing Corporation, Washington, D.C.

(1982).

I

6. J.A. Findlay, NUREG/CR-2566 (1982).

7. M.R. Fakory and R.T. Lahey, Jr., Nucl. Technol., 65, p. 250 (1984)

8. H. Nagasaka, M. Katch and S. Yokobori, NUREG/CP-0058, 3, p. 709 (1985).

9. T. Ikeda, A. Yamanouchi and M. Naitoh, J. Nucl. Sci. Technol.,

22(4), P. 249 (1985).

10. M. Murase, T. Ikeda and M. Naitoh, NUREG/CP-0058, 3, p. 632 (1985).

11. K. Tasaka, et al. , NUREG/CP-0027, 2, p. 910 (1982) .

12. L.S. Lee, G.L. Sozzi, and S.A. Allison, NUPEG/CR-2229, 1, (1982).

13. K. Soda, et al., J. Nucl. Sci. Technol., 20(7), p. 537 (1983).

I

14. M. Murase and H. Suzuki, Nucl. Technol. , 68, p. 408 (1985) .

15. M. Murase and H. Suzuki, Proc. the Japan-US Seminar on Two-Phase Flow Dynamics, No. C-3, Lake Placid, New York (1984).

[ 16. M. Murase and H. Suzuki: Countercurrent Gas-Liquid Flow in Boiling j Channels:, J. Nucl. Sci. Technol. (to be published).

l 17. B.S. Shiralkar, et al., Proc. Int. Nucl. Power Plant Thermal Hydraulics and Operational Topical Meeting No. E4, Taipei, Taiwan (1984).

l I

l

! 11-10

SAFER Qualification by TBL Test Analysis S.Miura, K.Moriya, T. Sugisaki 4 Hitachi Works, Hitachi Ltd.

l

SUMMARY

j Significant core cooling effects at the loss of coolant conditions in BWR were confirmed with the TBL (Two Bundle Loop) tests (1-4) which were conducted

to obtain the integral system responses at piping breaks by Japanese BWR utilities joint study with venders.

i SAFER, which was developed to the LOCA/ECCS analysis program for BWR,

) has been applied to the TBL analysis and obtained good predictions with l l adequate conservatism.

}

The TBL is scaled to a reference BWR/5-251 plant with 764 fuel bundles. The l TBL consists of a pressure vessel with two bundles and core internals, a feed water line, a main steam line, two recirculation lines, and an emergency core j cooling system. A power supply system has the maximum capacity of 10 MW

) for the two bundles. The height from the bottom of two jet pumps to an upper

{ plenum in the TBL is the same as the actual plant height. The scale ratio of J

regional volumes, masses, and flow rates in a pressure vessel is 2/764 which is l the same as the bundle ratio of theTBL to the BWR plant.

1 The TBL tests covered the wide range conditions at the loss of coolant accident in BWR. The TBL-1 tests (1,2) were conducted for the recirculation j large breaks with the effects of the bundle powers and ECCS failure mode.

j The T.BL-2 tests (3,4) were conducted for the intermediate and small breaks with the effects of the break location, break size, and ECCS failure mode.

i' SAFER code (5) was developed for the LOCA/ECCS analysis program of jet-i pump BWR, and have been extended its capabilities to non jet-pump BWR and internal-pump BWR. Also, the results of Japanese joint study which were i performed by BWR utilities and venders contributed to improvements of I accuracy in SAFER. SAFER has capability to simulate the typical core cooling I

phenomena in BWR, such as

' 11-11 1

(i) Multi-bundle effect in the core, which are caused by three different flow patterns of liquid down-flow, counter-current-flow and co-current up-flow, observed in TBL (1, ~) and SSTF(30-deg. Steam Sector Test Facility)(6).

! (ii) Liquid maintenance in each regions through transient by the counter-current flow-Ilmitation which occures at the small flow area such as upper tie plates, core inlet orifices, top of bypass, and bottom of bypass.

(iii) Up and down flow cooling effects in bundles, which include flooding ,

sterm , mist , droplet down flow- and falling film quenching-cooling.

SAFEF. has been applied to the TBL analysis, because the typical hydraulic behaviores in the TBL correspond to the hydraulic behaviores in the actual BWR plants. Thermal hydraulic transients predicted by the SAFER were compared with the TBL data, for the cases not only the large recirculation line breaks but also any other line breaks, with the effects of the bundle powers and ECCS failure mode.

l In order to confirm the conservatism of SAFER code, the predictions and test data of the highest peak cladding temperatures are compared. Then, the l LOCA/ECCS analytical results were close agreement with thermal-hydraulle responses of TBL test with adequate conservatism.

i 4

f REFERENCE, 1

l (1) M. Murase and M. Naltoh,3. Nucl. Sci. Techno!.,22(3), p.213 (1985).

(2) ti. Murase and M. Naltoh, 3. Nucl. Sci. Techno!., 22(4), p.301 (1985). ,

(3) T. Leda, A. Yamanouchi and M. Naitoh, 3. Nucl. Sci. Technol., 22(4),

p.2%' (1985)

(4) M. Murase, T. Ikeda and M. Naltoh, NUREG/CP-0058, 3, p. 632 (1985).

(5) B. S. Shiralkar, et al., Proc. Int. Nucl.

Hydraulics and Operations Topical Meeting, No. E4, Taipel, Taiwan (1984).

I (6) 3. A. Findlay, NUREG/CR-2566 (1982).

(7) K. Soda, et al., 3. Nucl. Sci. Techno!., 20(7), p.537 (1983).

(8) L. S. Lee, G. L. Sozzi and S.A. 'Allison, NUREG/CR-2229,1 (1982).

l

! Il-12 l

l I

l l

SAFER QUALIFICATION AGAINST ROSA-III RECIRCULATION LINE BREAK SPECTRUM TESTS S. Itoya, J. Otonari Toshiba Corporation K. Tasaka i Japan Atomi'c Energy Research Institute l INTRODUCTION l Revised version of SAFER computer code is a newly developed evalution model for the analysis of various BWR loss-of-coolant accidents (LOCAs). After developed the revised SAFER code under

, the cooperative efforts of Toshiba, Hitach and General Electric, l many assessments have been done using a lot of test data of the different test facilities. This paper will provide the qualification results against the ROSA-III recirculation line break tests with the different break area.

ROSA-III TEST CONDITIONS The ROSA-III program was conducted at JAERI and initiated in 1978 to simulate LOCAs for BWR plants. The ROSA-III test facility is a volumetrically scaled (1/424) BWR system with an electrically heated core, which consists of four fuel assemblies of half length and a control rod simulator. The test data selected for qualifications are two large break tests and one small break test at a recirculation pump inlet. These test conditions are shown in Table 1.

Table 1 ROSA-III Test Conditions Run No. Break Diameter Break Type Initial Power ECCS Mode 1 26.2/26.2mm(200%) Nozzle 3.97Mwt HPCS failure 2 26.2mm (100%) Orifice 3.97Mwt HPCS failure 3 5.9mm ( 5%) Orifice 3.97Mwt HPCS failure Run No. 1 is the large break test (double ended break) in

! order to simulate DBA condition. Run No. 2 and No. 3 are respectively a large and a small split break tests. These three i tests assumed a HPCS failure.

QUALIFICATION RESULTS I Figure 1 shows the system pressure histories of three recirculation, line break tests compared with test data. The depressuizations in two large break tests are very faster than one in a small break test due to the loss of coolant through the

, large break area. It can be seen from those comparisons that the pressure transients are well predicted.

11-13

i

)

i Figure 2 shows the comparisons of peak cladding temperatures (PCTs). These comparisons show good agreement with l heatup initiation, because the core uncovery caluclated by SAFER 1 is in agreement with the measured one. After heatup initiation, I the calculated heater surface temperature gradients agree well with test, due to a new core cooling model in a revised SAFER.

] As results, slightly conservative PCTs are calculated by SAFER.

4 I

CONCLUSIONS I

j In Summary, the revised SAFER has been assessed against the i ROSA-III recirculation line break tests with the different break ;

! area. It has been demonstrated that SAFER predicts the system

! response and key phenomena well in a BWR LOCA situation. Good I agreement with the system pressure is obtained throughout the transient. It has been confirmed that SAFER calculates a slightly conservative PCT.

REFERENCES

1. S. Itoya, et al, "An Improved Model for BWR LOCA Analysis" I ANS Meeting, New Orleans, a

June, 1984. _ goo ,

y pg, y j 2. Y. Anoda, et al, " Experiment 5 '

/ ,,,' ,

l Data of ROSA-III Integral , 7gg , ,= g ,'

, s s,

Test Run 912 (5% Split g ', ,

Break Test without HPCS

V 's i

'\

Actuation)," JAERI-M82-010, f500 TEST DATA L March 1982. 3 ..... sArta

3. M. Irioka, et al, " Assessment " 300

{'

of the '1 HYDE-BI/NODO Code y 900 -

M NO. 2 with Data from ROSA-III e /,\, ,

,\

i Loss-of-Coolant Experiment," @ 700 - ,'

i JAERI-M84-188, October, 1984. . t, ,

,e' j 500 -

TEST DATA 10 -

..... sArEn a

TEST DATA 300

{ g

) 38 -

..... $AFER 0 100 200 300 1 5

, # 900 . TEST DATA

..~ WER ,

5 -p to, 3 h -

h4 ' .- M NO. 2

~

o. m . ,

5 2

. I*

M ,,,,'

" 500 u

M NO. 3

(

f 0

- 300 0 200 400 600 l 0 200 400 600 TIME (s) TIME (s) i i Figure 1 Systen Pressure Figure 2 Peak Cladjing Teperature I

! 11-14

---,-.-,-,,--n,,. ..,,-,,,-,-,,.n-,..-,,.,,,--..,,,.,,,---.n-.-_--,- , ,-_,,n--,-s,nn.- - _e+ n,.,, -

-,,-,-v-.-n - - , . - -_,c..-_.v~ --,n.c,

Assessment Of The SAFER ( } Code For LOCA With Data From The Advanced Boiling Water Reactor Test Facility by F. D. Shum, A. B. Burgess, B. S. Shiralkar (GE) and K. Yahagi (TEPCO)

SUMMARY

The" development of an Advanced Boiling Water Reactor (ABWR) for Japan is being performed by General Electric in partnership with Toshiba and Hitachi under the funding of the Joint Japanese BWR _ Utilities Group (JJUG). The ABWR incorporates numerous improvements in the design of the nuclear system, balance of plant, control systems, and fuel. From the point of view of safety analysis the most significant improvement is the replacement of the recirculation system consisting of external pumps, the recirculation lines and jet pumps, by internal pumps at the bottom of the downcomer. This eliminates the poss:bility of large breaks below the level of the top of the core.

The response of the ABWR to loss of coolant accidents was simulated intheAg/FISTtestfacility. For this purpose, the BWR/6-FIST was modified to simulate the performance of the internal Facility pumps. The facility volumetrically scales the ABWR to one full size electrically heated bundle. Breaks in various regions were simulated by opening quick-opening valves and discharging flow to a suppression tank.

This paper presents results from simulated breaks in the steam line, a high pressure core spray (HPCS) line and in the bottom head of the vessel. The results have been analyzed with the SAFER code. SAFER is a realistic design model for LOCA analysis which can also be used for licensingcalculationswigconservativeinputsasrequiredby 10CFR50.46 and Appendix K . The steam line break and HPCS line break resulted in very mild transients with the rod temperatures following the saturation temperature and essentially no heatup. Figure I shows the system pressure transient for the HPCS line break. The pres:ture is maintained steady by the pressure control system prior to isolation, following which there is a small increase. As the unbroken HPCS line begins to inject in the upper plenum, the mixture level in the upper plenum drops as shown in Figure 1 and the break is uncovered. The subsequent steam discharge through the break, aided by the Automatic D_epressurization System (ADS) depressurizes the system, leading to a level swell. The core does not uncover and consequently the temperatures follow the saturation temperature transient. Figure 1 also shows SAFER predictions which track the system pressure and PCT very closely. The phenomena of level collapse and swell in the upper plenum are also captured in the calculations.

11-15

The only transient which leads to any significant heatup (PCT N 900*F) is the bottom break of 450 cm2 . This is a much larger break than the design basis but the experiment serves to assess the capabilities of the code for more severe transients.

SAFER predictions agreed well with the experiments in all cases. l (More detailed results and comparisons will be presented in the final I paper.) Consequently, SAFER has been shown to be applicable for ABVR '

LOCA analysis.

,=, e s

== - *" eaua = = m ,

, _ o . m.6 -.

b riassenz

" I e o ***="ma**

} ,, Mi - ---- **ea

. , g s [ !a -

1 T

em -

l f \

~ll- j'ene nu.am ,

h"-l ~~~,N j I l \

\ b T' *"I """fUf?' - - - --

~ ~

l mn.nz .

g m, -

i -

4 t _

111 , , , '- -

= - - - . ., ,,,

m- ""

Figure 1: Transient Response for Core Spray Line Break References

1. F.D. Shum, S. Itoya, K. Tominaga, K. Tasaka, W.S. Hwang, H. Aoki, and B.S. Shiralkar, "An Improved Model for BWR LOCA Analysis",

Second Proceedings of Nuclear Thermal Hydraulics, 1984 ANS Annual Meeting, New Orleans, June, 1984.

2. A.G. Stephens, "BWR Full Integral Simulation Test (FIST) Program -

Facility Description Report", NUREC/CR-2576, December, 1982

3. B.S. Shiralkar, J.G.M. Andersen, A.B. Burgess, and S.A. Wilson,

" Evolution of LOCA Analysis at General Electric", Proceedings of International Nuclear Power Plant Thermal Hydraulics and Operations Yopical Meeting, Taipei, October, 1984.

11-16

Safety Performance of the Advanced Boiling Water Reactor by F. M. Paradiso, J. G. M. Andersen, C. D. Sawyer (GE) and j

A. Omoto (TEPCO)

SUMMARY

A comparison of the ABWR ECCS network to the BWR/5/6 ECCS network (refer to Figure 1), shows that the ABWR has several features which are a distinct improvement over BWR/5/6. The ABWR has an additional high pressure makeup system which gives the ABWR ECCS network another system, independent of the ~

automatic depressurization system (ADS), to mitigate the consequences of transients and small LOCAs. Also, more effective use of the three emergency ,

diesel generators is realized in the ABWR by having a low pressure flooder .

with a heat exchanger on each diesel. Compared to BWR/5/6, this arrangement ,

provides an additional means of establishing long-term cooling. Furthermore, the RCIC has been upgraded and incorporated into the ECCS network, improving i its ability to mitigate the consequences of transients and small LOCAs.

Finally, fine motion control rod drives have been incorporated into the ABWR design. This design provides an alternate means of rod insertion via an electrical motor on each drive.

Another ABWR design improvement over BWR/5/6 is the replacement of the recirculation lines and jet pumps with internal recirculation pumps. This design eliminates the possibility of any significant size break below the top of the core. For the ABWR, the lowest possible large break location on the vessel is the high pressure core spray line. Coupled with the single failure of the emergency diesel generator which supplies electrical power to the other HPCS and ene low pressure flooder, the core spray line break is clearly the most limiting case for the ABWR. Compared to the BWR/5/6 design basis acci-dent (i.e., the maximum recirculation line suction break approx. 3 ft'), the HPCS line break (approx. .15 f t ) is much smaller. Therefore, given the 8

improvements in the ABWR ECCS network and the significant reduction in the break size for the limiting case, it is obvious by inspection that there would be a substantial improvement in the safety performance of the ABWR over BWR/5/6. This has been confirmed by a preliminary probabilistic risk assess-ment (PRA) for the ABWR which shows that the ABWR is a factor of 5-10 better than BWR/5/6 in avoiding possible core damage from degraded events.

The SAFER transient thermal hydraulic computer model is used to predict the ABWR system response to a loss of coolant accident (LOCA). Currently, the SAFER code has been assessed against results from the BWR/6-FIST Facility which has been modified for ABWR application. These r g its demonstrated that there is no fuel uncovery for any design basis break . SAFER analysis for the actual plant have also been performed for several break locations and possible single failures. As expected, the most limiting case is the HPCS line break discussed earlier which has only the RCIC, two low pressure flooders and the ADS available. However, even for this extremely unlikely event, no fuel uncovery is predicted.

11-17 l

l

The successful SAFER predictions of the FIST results have greatly enhanced the qualification of the SAFER model for ABWR LOCA application. To address effects which are not covered in the FIST Facility, such as parallel channel effects, SAFER results have also been compared to the more detailed TRAC computer model. The three-dimensional thermal-hydraulic TRAC computer model has been assessed against the results from several different test facilities and has been highly successful in predicting the results of a variety of simulated transient and accident events. TRAC is a nonhomogenous, nonequilibrium two-fluid thermal-hydraulic model which includes a three-dimensional thermal-hydraulic analysis of the vessel. The results of test cases analyzed with both codes show that the SAFER simplified model predicts all the significant system responses exhibited in the corresponding TRAC calculation. Therefore, it is concluded that the simplifications of the SAFER model do not diminish its ability to adequately predict the ABWR LOCA response.

The SAFER comparisons with TRAC further c6nfirm that SAFER is applicable

for ABWR LOCA analyses. Furthermore, the results of the LOCA analyses and preliminary FRAs show that the safety performance of the ABWR is much improved over BWR/5/6.

AM BWR/5/6 t

LP LP

.P .,

Y LP W e-e W* FIGURE 1: TCCS NEWORK COMPARISON

\

References

1. Shum, F. D.; Burgess, A. B.; Shiralkar, B. S.; and Yahagi, K. " Assess-ment of the SAFER Code for LOCA with Data from the Advanced Boiling Water Reactor Test Facility", Light Water Reactor Information Meeting, Gaitnersburg, October 1985.

l 11-18 l' _. _

1 1

l i

ON THE PROBABILISTIC ESTIMATION OF j 4 CONTAINMENT FAILURE BY STEAM EXPLOSIONS l i by i

M. Abolfudl,* B. Najafi,** E. Ruuble,**

T.C. Theofanous, i and H. Amarasooriya* j 4

Department of Chemical and Nuclear Engineering

! University of California Santa Barbara, CA D3106

}

i The challenge of probabilistic modelling is to optimize the partition (herein called probabilistic framework or structure)the of the overall process with respect to both the reliability of quantification of each constituent part (stage) a r.d the recognition and handling of any dependencies among them during When the quantification of each stage is carried out i synthesis.

) by means of single numerical estimates we speak of Point Estimate l Mgdels. When distributions are used an appreciation of confidence

~

levels (uncertainties) may also be traced in the calculation and we speak of Distributed Parameter Models. If the overall structure is correct the finer partition will direct that the processes to be judged are more elementary, thus decreasing the degree-of subjectivity. On the other hand, when a sufficiently

definitive technical argument on a single but crucial link in the ,

overall, sequence can be identified, the most elementary one stage i

model may by far outweigh any further elaborations in both significance and reliability. In this case we speak of t Eleeggtgry Siggle_Stgge Mgdels. In all three cases the '

mechanistic reasoning employed in quantification is an integrgl part of the probabilistic model. Thus in judging validity of predictions, both the technical details of this reasoning and probabilistic framework must be considered.

Past attempts of quantifying the a-f a i l u r e probability encompass all three catagories of models mentioned above. In all cases the likelyhood of a-f ailure estimated is so low (less than

10

per core melt) that its contribution to risk seems to be negligible. Still, as the APS report on the source terms states this conclusion remains highly judgemental at this time. Our this presentation is three fold: _(a) Provide a

purpose in i

critical evaluation of past probabilistic models which have been i proposed in conjuction with original mechanistic considerations.

(b) present a new probilistic framework and input (c) introduce- some distributions

new -analytical evidence to support the proposed for the processes of premixing and lower plenum failure / venting. Quantitative results, obtained using the

! Discreet Probabilits Distribution Method, will also be discussed.

School of Nuclear Engineering, Purdue University

- Science Application Internation1 Corporation, Los Angeles, CA 1

11-19 4

- , - * -n,..#--. , . - , , , , . . , , _ . , , , , ..

i Progress on Qualification Testing Methodology Study of Electric Cables S.0kada, Y.Kusama, M.Ito, T.Yagi, M.Yoshikawa, K.Yoshida, N.Tamura, W.Kawakami Japan Atomic Energy Research Inst t3 ute (JAERI), Takasaki Radiation Chemistry Research Establishment

! Safety-related electric cables in nuclear power plants are required to function even if they should be subjected to a postulated design basis event such as a loss-of-coolant accident (LOCA) at the end of their intended service life. In a LOCA the cables are assumed to be exposed to combined stresses of radiation, high temperature steam, spray and, under certain circumstances, air. The condition of the stresses, such as the exposure time or dose rate, is expected to depend upon the plant type and the acci-dent scale. In almost all cases of the electric cable qualification in Japanese plants, however, the cables are tested by such a method as irradi-ation followed by steam / spray exposure i.e. secuential method in short term (about one week for each of the irradiation and the steam / spray exposure) l tests. Therefore, it has been one of the most important subjects in the testing methodology study to assure the validity of the short term sequen-

tial test by finding adequate conditions to simulate the long or short term combined stresses.

i We have performed short term and long term tests of electric cable ma-

'1 terials such as ethylene-propylene rubber (EPR) and chloro-sulfonated poly-ethylene (Hypalon) in combined environments of the irradiation and the steam / spray exposure i.e. simultaneous method according to a PWR LOCA tem-perature profile (max. 150 C; 120 C in the major part of the exposure peri-od) for both cases of containing and not containing air in the steam , as listed below.

(Case Simultaneous-A)

Short term (about 1 week) and high dose rate (about 10kGy/h); saturated steam (not containing air)

(Case Simultaneous-B)

Long term (3 months) and low dose rate (0.6kGy/h); saturated steam (Case Simultaneous-C)

Short term (about 1 week) and high dose rate (about 10kGy/h); containing

air (partial pressure 0.05MPa) in steam

(Case Simultaneous-D)

Long term (3 months) and low dose rate (0.6kGy/h); containing air (partial pressure 0.05MPa) in steam They are assumed to correspond to various types of LOCA.

We have also investigated the mechanical and electrical degradation of the materials by the sequential tests where -the stress environments are systematically altered in order to find the adequate conditions to cause the same degree of degradation as the simultaneous tests. In the experiments the effects of the steam temperature (120 to 160 C) and the air partial pressure (0 to 0.5MPa) in the steam / spray exposure after the irradiation (what is called pre-irradiation) have been studied. The condition of the pre-irradiation has been investigated as well. The irradiations have 12-1

4 i

been performed under such various conditions as at a high dose rate (about 10kGy/h) and a low dose rate (about 0.5kGy/h) in open air; at a rather high i

dose rate in pressurized oxygen; at a rather high dose rate in air at ele-vated temperature (70 C).

1 In comparison between these sequential test results and those by the '

simultaneous tests, the following degradation behaviors should be noted.

j 1) Simulation of (Case Simultaneous-A) by a sequential method !

Tensile strength degradation of EPR's and Hypalons in the sequential test of the high dose rate irradiation in open air followed by the saturated steam exposure is greater than in Case Simultaneous-A. Electrical resist-ance of the EPR's, however, does not decrease in this type of sequential test so much as in Case Simultaneous-A. On the other hand, when the steam contains air (0.05MPa) in the sequential test, the electrical property as well as mechanical ones such as elongation and strength degrade to the same

, extent or much more than in Case Simultaneous-A.

2) Simulation of (Case Simultaneous-B) by a sequential method Mechanical and electrical properties of the EPR's are not affected by dose rate in the simultaneous environments of saturated _ steam (Case-A and

-B). However, mechanical property degradation of the Hypalons is greater in the long term test (Case-B) than in the short term test (Case-A), which is assumed to be due to additional thermal degradation. Such_ degradation increase can be obtained in the sequential test when the steam temperature is elevated (within 120 to 160 C) in both cases of saturated and air-containing steam / spray exposure following the high dose rate pre-irradiation. ,

3) Simulation of (Case Simultaneous-C and -D) by a sequential method In the simultaneous environment, the mechanical and electrical degrada-tions are on the whole greater in case of containing air. It is much more remarkable in the long term test (Case D), which is supposed to be caused by oxidation degradation. Such remarkable degradations are observed as well in the sequential test in case where the steam contains surplus air (up to 0.5MPa) or the pre-irradiation is performed in such sufficient oxidation environments as at a low dose rate in open air, in pressurized oxygen or at

elevated temperature in air.

These results suggest that various types of LOCA will be well simulated by the sequential tests in which the pre-irradiation and the steam conditions are suitably selected, from the viewpoint of the material property degrada-tion.

j.

12-2

i 1

l SAND 85-1712A EQUIPMENT QUALIFICATION AND SURVIVAB!LITY RESEARCH AT SANDIA NATIONAL LABORATORIES" Lloyd L. Bonzon Sandia National Laboratories, Albuquerque Since its inception in 1975, the Qualification Testing Evaluation (QTE) Programl has been concerned with several broad issues in safety-related equipment qualification. These concerns encompass both aging simulation methods as well as accident simulation methods. Much of the effort is concerned with combined environments especially radiation in combination with other

, environments including oxygen, temperature, mechanical stress, and accident thermodynamic environments like pressure / temperature / chemical spray.

The Electrical Penetrations Assemblies (EPA) Program2 is specifically concerned with the survival (i.e., leak-rate integrity) of such assemblies under severe accident conditions.

A brief discussion of several current and planned projects will illustrate the scope of these NRC-sponsored efforts.

a QTE Procram This overall Program effort has three major thrusts: aging simulation methods, adequacy of radiation simulators, and accident testing methods.

Sindia researchers have published numerous reports 3-7 on material degradation as a function of dose, dose-rate, combined (radiation plus thermal) environments, sequential and simultaneous aging.

A principal benefit of this work has been the development of several, simple techniques to determine the uniformity of material degradation.7 These techniques include the use of density gradient columns, microhardness profiling, and metallographic polishing.4.8.9 More recent activities have concentrated on modeling techniques for dose-rate effects so that the extrapolation from accelerated-aging can be made to actual use conditions.10 This is an excellent l

example of applications-oriented NRC-sponsored research.

Some new work has been initiated in the area of combined dose (dose-rate), temperature, and mechanical stress on seals and gaskets. Also, based on an earlier study of electronics aging,Il electronic pieceparts are being studied by varying several parameters: component type, dose, dose-rate, temperature, and voltage bias.

Saismic fragility tests have been conducted on Class lE naturally-aged battery cells to determine failure modes and thresholds.12-15 Artificially-aged battery cells of the same type will be subjected to the same seismic test levels. The results from these two aging methods will be evaluated in order to establish the relevance and applicability of accelerated aging methods.

In evaluating the adequacy of radiation simulators, the concern is for the accident situation: high dose rates, large total doses, gamma and beta radiations, and combined accident environments.

Particularly, the presence of beta radiation complicates the testing of material and equipment "This paper was supported by the U.S. Nuclear Regulatory Commission, Office of Reactor Safety Research, as part of the Qualification Testing Evaluation (QTE) Program (FIN OA-1051) and the Electrical Penetration Assemblies (EPA) Program (FIN OA-1364) being conducted by Sandia National Laboratories, under Interagency Agreement DOE-40-550-75.

12-3 l

specimens. Electron-irradiation charge-breakdown experiments were previously done on rubbar insulation materlats:16 charge breakdown was not apparent using "real" conditions (but was observed during experiments in vacuum). Currently the simulator-adequacy evaluation activity i involves comparison of beta and gamma effects on material degradations to determine a gamma-equivalent test approach; this work is being done jointly with French researchers.

Combined-environments accident testing has been done, and continues. The importance of oxygen during accident simulations has been observed and reported.17 Enhanced degradation of materials in simultaneous (radiation plus accident thermodynamic) profiles has also been observed and I reported.18 The " sensitivity" of material degradation to the choice of aging and accident simulation methods has just been completed and reported.19 Two new test programs are currently under way. Tests of cables in superheated steam conditions will be compared with the results from saturated-steam conditions (as reported in Reference 18). I Postaccident radiation monitors will be subjected to combined radiation-thermal environments expected in postulated accidents to evaluate for combined-environments effects.

i Planning is also under way for (1) tests of coaxial / triaxial configured cable, (2) acceleration methods for the expected (about 1-year) "postaccident" period, and (3) tests of sealing systems under accident conditions.

EPA Proaram Tha goal of this Program is to evaluate the leak-rate integrity of electrical penetration assemblies, under severe accident conditions (i.e., beyond design basis accidents); the program supports the ovsrall containment integrity evaluation efforts sponsored by the NRC. Based on a previous study,20 EPAs representing the three remaining U.S. manufacturers have been, or will be, i subjected to three representative severe accident profiles, i

The first such test has recently been completed.21 A D. G. O'Brien EPA was subjected to a simulated (large-PWR) severe accident with peak pressure of 155 psia at 3610F (saturated steam);

radiation and thermal aging was a part of the test program.

Although evaluations continue, there were no detectable leaks through the EPA during the steam pressurized portions, including none through the header plate O-ring seals. (As a significant secondary issue, the electrical properties of the EPA degraded over the first 2 days to the point that all modules measured less than 106 ohms to ground at 50 volts, and 5 out of the 8 circuits allowed 1/2 amp leakage currents to ground after 10 days.)

Tests of a Westinghouse EPA are currently under way. It will be subjected to a simulated (Mark !!!,

BWR) severe accident with peak pressure of 75 psia at 4000F (superheated steam). And planning is under way for the third test in this series; that will involve a Conax EPA subjected to a simulated (Mark I, BWR) severe accident with the most extreme environments, 135 psia and 7000F (superheated steam).

Summary i The USNRC concerns for data-based equipment qualification methods and equipment integrity 4

information are being addressed in the QTE and EPA Programs. The results presented illustrate the

. most recent, and planned, activities.

Referrnces

l. Stndia Laboratories Staff, Status Report on Equipment Qualification Issues Research and Resolution, SAND 85-1309, NUREG/CR-4301. To be published.

2. F. V. Thome, Results of Leak-Rate Testina of Electrical Penetration Assemblies under Severe Accident Conditions, SAND 84-1496A. Presented at the 8th International SMIRT Conference, Brussels, August 19-23, 1985.

12-4

3. L. L. Bonzon, K. T. Gill n end E. A. Salazar, Qu*lification Testina Evelu* tion Proaram Licht Wrt r R"ctor Sefaty R* search Quarterly Report. October-Decnmb*r 1978, SAND 79-0761, NUREG/CR-0813, Juns 1979.

4. R. L. Clough, K. T. Gillen, and C. A. Quintana, Heteroceneous Oxidative Dearadation in Irradiated Polymers, SAND 83-2493, NUREG/CR-3643, April 1984.

5. R. L. Clough and K. T. Gillen, investiaation of Cable Deterioration inside Reactor Containment, Nuclear Technology, Vol. 59, November 1982, pp. 344-354.

6. R. L. Clough and K. T. Gillen, Complex Radiation Dearadation Behavior of PVC Materials Under Accelerated Aoina Conditions, SAND 82-1414, NUREG/CR-3151, July 1983.

7. R. L. Clough, et al, Accelerated-Aaina Tests for Predictina Radiation Dearadation of Oraanic Materials, S AND83-0798], Nuclear Safety, Vol. 25, No. 2, March- April 1984, pp. 238-254.

8. R. L. Clough and K. T. Gillen, Techniques for Studyina Heteroceneous

Dearadation in Polymers,

SAND 83-2171. Presented at the American Chemical Society Symposium on Polymer Degradation, St. Louis, Missouri, April 8-13,1984.

9. K. T. Gillen, R. L. Clough, and N. J. Dhooge, Density Profilina of Polymers, SAND 85-0795J, March 1985. To be published in Polymer.

10. K. T. Gillen and R. L. Clough, General Extrapolation Model for an Important Chemical Dose-Rate Effect, SAND 84-1948, NUREG/CR-4008 December 1984.

11. R. T. Johnson, F. V. Thome, and C. M. Craft, A Survey of the State-of-the-Art in Aoina of Electronics with Application to Nuclear Power Plant Instrumentation, SAND 82-2559, NUREG/CR-3156, April 1983.

12. L. L. Bonzon and D. B. Hente, Test Series 1: Seismic-Fraallity Tests of Naturally-Aaed Class IE Could NCX-2250 Battery Celts, SAND 84-1737 NUREG/CR-3923, September 1984.

13. L. L. Bonzon and D. B. Hente, T_ est Series 2: Seismic-Fraallity Tests of Naturally-Aced Class IE Exide FHC-19 Battery Cells, S AND84-2628, NUREG/CR-4095, March 1985.

14. L. L. Bonzon and D. B. Hente, Test Series 3: Seismic-Fracility Tests of Naturally-Aaed Class IE C&D LCU-13 Battery Celts, SAND 84-2629, NUREG/CR-4096, March 1985.

15. L. L. Bonzon and D. B. Hente, Test Series 4: Seismic-Fraallity Tests of Naturally-Aced Exide EMP-13 Battery Cells, SAND 84-2630, NUREG/CR-4097, March 1985.

16. W. H. Buckalew, F. J. Wyant, and G. J. Lockwood, Response of Rubber Insulation Materials to Monoeneraetic Electron Irradiations, S AND83-2098, NUREG/CR-3532, November 1983.

17. K. T. Gillen, et. al., Loss of Coolant Accident (LOCA) Simulation Tests on Polymers: The importance of includina Oxyaen, SAND 82-1071, NUREG/CR-2763, July 1982.

18. L. D. Bustard, The Effect of LOCA Simulation Procedures on Ethylene Propylene Rubber's Mechanical and Electrical Properties, SAND 83-1258, NUREG/CR-3538, November 1983.

19. L. D. Bustard, et. al., The Effect of Alternative Aaina and Accident Simulations on Polymer Properties, S AND84-2291, NUREC/CR-4091, May 1985.

20. W. A. Sebrell, The Potential for Containment Leak Paths Throuah Electrical Penetration Assemblies Under Severe Accident Conditions, SAND 83-0538, NUREG/CR-3234, July 1983.

21. Quick-Look Report on Test for Evaluatina Leak Behavior of a D. G. O'Brien EPA under Severe i Accident Conditions, July 12,1985. Letter F. V. Thome (SNL) to W. S. Farmer (NRC).

l 12-5 )

l 1

I Fire Protection and Hydrogen Burn Equipment Survival Research*

D. L. Berry Sandia National Laboratories Summary The fire protection and hydrogen burn survival research programs at Sandia involve both testing and analysis. Each program has followed separate, yet parallel, paths for assessing the likelihood that equipment will survive either a fire or a hydrogen burn accident environment. To do this, both programs have strived to characterize their respective accident conditions and to test the survivability of equipment to these conditions.

Fire Protection Research The fire protection research program has involved testing and analysis to:

- characterize source fire heat and mass release rates

- establish failure thresholds of cabling and equipment j - measure full-scale room fire environments i This work has been aimed at supporting licensing and probabilistic risk analysis (PRA) needs for data relevant to nuclear power plant situations. During FY85, accomplishments in the program included:

- Twenty-four heat release rate tests on various types and sizes of fuels I - Over 15 full-scale cabinet fire burn tests, using the IEEE-383

qualified or unqualified cabling

- Room fire environment testing in a 60' x 40' x 18' room, including a control room mockup

- Numerous damage threshold tests of IEEE-383 qualified and unqualified cable under transient temperature conditions j - Three thermal damage threshold tests of control relays

- Electrical component fire environment exposure tests Based on these accomplishments, it has been found that:

- Accurate and reproducible measurements of source fire heat

!^ release rates can be made for use in fire analyses and PRA models.

This work is supported by the United States Nuclear Regulatory Commission and' performed at Sandia National Laboratories which is 1

operated for the United States Department of Energy under Contract Number DE-AC04-76DP00789.

12-7

- Cabinet fires involving unqualified cabling are much more severe than qualified cabling.

1

- Cabinet fires can produce significant smoke which may obscure the vision of operators or fire teams.

- The failure threshold of cabling is affected by cable configuration and convective heat transfer.

- Control relays and other electronic components appear less susceptible to thermal degradation than manufacturers predict.

Hydrogen Burn Survival Research The hydrogen burn survival research program has involved testing and analysis to:

- characterize hydrogen burn thermal environments (in cooperation with Hydrogen Behavior Program)

- establish failure thresholds of cabling and pressure transmitters This work has been aimed at performing experiments and analyzing available data to develop an understanding of equipment thermal response and survival in a hydrogen burn. During FY85, accomplishments in the program included:

- Local heat transfer and global thermal calculations of 21 Nevada Test Site (NTS) pre-mixed hydrogen burn tests sponsored by the Electric Power Research Institute (EPRI)

- Post-accident environment analysis of the TMI-2 containment using the " Hydrogen Event: Containment Transient Response" (HECTR) Code

- Ten NTS environment simulation tests (six with cable and four with pressure transmitters)

- Twenty tests (ten cable and ten pressure transmitters) at heat flux conditions beyond NTS tests Based on these accomplishments, it has been found that:

- Global hydrogen burn environments measured during the NTS tests should be useful for benchmarking global computer models.

- Local environments cannot be resolved from the (NTS) data made available to Sandia, given problems with instrumentation performance during the NTS tests.

- HECTR analyses performed to date agree within 10% with the TMI-2 l accident pressure conditions.

- With few exceptions, the cabling and transmitters tested exhibited no failures when exposed to simulated heat fluxes, l even as high as three times those experienced in an NTS burn I test using 13 volume percent hydrogen.

12-8

ENVIRONENTAL AND DYNAMIC QUALIFICATION OF EQUIPENT RESEARCH AT THE INEL

, J. A. Hunter

! IDAH0 NATIONAL ENGINEERING LABORATORY l EG&G IDAHO, INC.

Equipment qualification research is being conducted by the Idaho National l Engineering Laboratory to provide the technical basis for improving standards used for qualifying electrical equipment for dynamic conditions and mechanical equipment for dynamic and environmental conditions. The following summarizes the most important results and accomplishments to date for the program.

Main Coolant Pump Shaft Seal Failure of main coolant pump shaft seals under station blackout conditions could lead to a significant loss of primary coolant. Of particular concern is the influence of increasing primary coolant temperature on non-metallic elements of the seals during station blackout conditions when the cooling to the seals is interrupted. To assess the possible degradation effects of prolonged high temperature (approximately 550 F) on the non-metallic elements, 0-ring extrusion and hydrostatic seal blowdown tests on specific 0-ring and seal simulations were performed. The results indicated that some 0-ring materials will not withstand pressurized water reactor station blackout conditions. 0-ring extrusion was found to be a function of material, channel seal presence, and 0-ring clearances. Seal stability assessed during the seal blowdown tests was found to be primarily dependent on the seal balance ratio and the extent of water subcooling.

Additional research has been initiated to identify critical seal stability parameters and to assess friction effects on elastomer seal elements under station blackout conditions. Additional seal designs will also be evaluated during this phase of the research.

Valve Research Containment integrity is most likely to be compromised during an accident by failure of one of the many penetrations rather than by failure of the structure. A technical basis is required to assess the affect of accident conditions on the operability and leak integrity of containment isolation system (CIS) valves which include containment purge and vent valves. The purpose of the valve research is to identify requirements and acceptable methods for qualifying specific types of CIS valves to withstand design basis loads and to characterize the behavior of selected CIS valves under accident conditions.

Experiments with three containment purge butterfly valves were conducted under accident flow conditions with valve inlet and outlet ducts installed to assess system effects on purge valve performance. Two 8-inch valves and one 24-inch

. valve were tested in two duct configurations while varying valve position ;

and inlet pressure from 5 to 60 psig. i 1

i 12-9 !

The purge valve experimental data analysis concluded that the torque measurements derived from a scale-model valve at a given inlet pressure can be extrapolated to bound the torque requirements for a larger valve when several conditions are satisfied. Some of the conditional parameters include disc shape, disc aspect ratio, disc thickness to nominal bore ratio, disc orientation relative to the flow and valve inlet pressure.

The purge valve experimental data will be augmented by the data from the additional CIS experiments to be conducted in the near future. The CIS I experiments will assess the behavior of two sizes of containment valve penetration assemblies (two-inch globe and eight- inch gate and butterfly valves) under accident conditions. The results will be used to assess system effects on the valve dynamic response, operability and leak integrity. The results will also be used to verify the capability of pipe system analytical procedures to predict CIS valve and piping system response to dynamic design loads.

Additional testing will assess the effects of severe accident phenomena (containment wall displacement) on the ability of CIS valves to operate and maintain leak ir.tegrity.

Valve experiments will also be performed in the Federal Republic of Germany Heissdampfreaktor Facility to assess and predict valve and pipe system response to dynamic loads imposed on a complex piping system.

Dynamic Research The equipment qualification research at EG&G Idaho /INEL is t.lso investigating several topics directly related to dynamic qualification criteria, requirements, and methodologies. All topics have been initiated but they have not progressed to the development of final conclusions.

Research is also being condu-ted to develop methods that can be used to establish dynamic load qualification margins for mechanical and electrical equipment. The methods will assist in the development of test equivalency guidelines and data transfer methods such that existing data obtained by different techniques can be used to qualify various categories of equipment to ensure that the resulting margin is unique. The methods will be used to quantify margins in new and operating equipment.

Research is being conducted to evaluate current qualification methods for safety injection pumps which intermittently operate. Specifically, the influence of normal operating loads, including flow induced vibration, on pump performance is to be investigated. The objective of the research is to determine whether or not normal operating loads are properly being simulated in the current pump qualification procedures.

Research is also in progress to evalute the need to include frequency ranges above the seismic range in current qualification procedures. Research is in progress to define conditions under which equipment can be excited by frequencies greater than the seismic frequency range, define sensitive components, and develop and validate guidelines for qualifying equipment for the higher frequencies.

The program is also initiating flow induced vibration research which will identify equipment susceptible to flow induced vibration, characterize the vibration most detrimental to the equipment, and develop and validate necessary procedures to account for the flow induced vibration loading.

12-10

THE SOURCE TERM CODE PACKAGE by JA Gieseke, P Cybulskis, H Jordan, and KW Lee BATTELLE Columbus Laboratories 505 King Avenue Columbus, Ohio 43201 The Source Term Code Package (STCP) is a set of computer codes which allows analyses of nuclear reactor accidents to produce predictions of fission product release to the environment as a function of reactor design and specifi-cations for the assumed accident. The codes are basically those used in the analyses performed for the BMI-2104 report but have been combined, improved, and streamlined for easier use. The objective in preparing this code package was to make the calculations more direct, traceable, and user-independent with documentation for release to the public. It is important to note that the STCP is not intended to be a research tool but a code for general use in making source term predictions that has a sound and definable basis and produces reasonable accurate results in comparison with more detailed codes.

The Source Term Code Package has four major elements which are shown in Figure 1. The overall thermal hydraulics is provided by the MARCH-3 code which combines the previously separate codes MARCH-2, CORCON-Mod 2, and CORSOR-M.

Release of fission products and aerosols during core-concrete interactions is predicted with the VANESA rode. Detailed thermal hydraulics and fission product transport in the reactor coolant system is provided by the TRAP-MERGE code formed by combining the previously separate TRAP-MELT and MERGE codes. Finally, fission product transport in the containment is predicted by the NAUA-4 code as modified to include fission product removal by pressure suppression pools (SPARC code) and within ice compartments (ICEDF code).

Fission product and aerosols groups being tracked in the code package are as follow:

e Noble gases e Iodine (I, Br) i e Cesium (Cs, Rb) )

e Tellurium (Te, Sb, Se) e Barium l l

13-1 l

e Strontium o Ruthenium (Ru, Rh, Pd, Tc, Mo) e Lanthanum (La, Eu, Pr, Nd, Pm, Sm, Y, Nb, Zr[f.p.])

e Cerium (Ce, Np, Pu) e In-vessel produced aerosols (structures, cladding, control rods, etc.) .

e Ex-vessel produced aerosols (structures, concrete, etc. l The species or materials forming each group have also been identified in the above list. The code package produces time-dependent locational distributions, physical forms, and transport rates for these groups throughout the course of the accident.

INPUT SELECTION FROM MANUALS If MARCH 3 (PEROi. CORQN. CORSOR)

If i f TRAP-MERGE VANCsA (TPAP-MELT 2. MERGE)

If NAUA/

m- sPARC/ s ICEDF 1

I f 50tRCE TEM CODE PACUlGE 13-2

l i

RELAPS/SCDAP: An Integrated Computer Code for Severe Accident Analysis By T. C. Cheng Idaho National Engineering Laboratory EG&G Idaho, Inc.

I

SUMMARY

RELAP5/SCDAP is a computer code under development at the Idaho National Engineering Laboratory for the Office of Nuclear Regulatory Research that is designed to mechanistically model postulated severe a:cidents in light water reactors. Specifically, the code is intended for use in studying the consequences of light water reactor core disruptive events for terminated accidents (such as occurred in Three tiile Island, Unit

2) or unterminated accidents up to the loss of internal vessel structural integrity. The current version of RELAPS/SCDAP resulted from the integration of the RELAP5, SCDAP, and TRAP-MELT codes. Consequently, RELAPS/SCDAP

] is capable of modeling the coupled behavior of a severe accident of the entire reactor coolant system (RCS), including the thermal-hydraulics, fuel damage progression, and fission product release, transport, and retention.

Treating the coupled effects is important in modeling a severe accident.

A typical accident progression in severe accident would involve: loss i

of coolant, core uncovery, core heatup, cladding ballooning and oxidation, cladding breach, cladding liquefaction, fuel dissolution, liquefied fuel / cladding relocation, and fission product release and transport in the RCS. During a severe accident, direct coupling exists between reactor system thermal-hydraulics, fuel damage progression, and fission product behavior. For example, the progression of fuel damage is strongly influenced by the rate of the metal-water reaction, which is in turn influenced by steam availability.

Obstructed regions in the core as well as loop hydraulic behavior and hydrogen release will influence core steam flow and thus metal-water reaction. The energy release from fission products transported within and outside the vessel will also affect loop and vessel flow patterns.

Based on these and other interactions, it is necessary that a coupled effects be modeled to obtain credible results.

In addition to the existing capabilities of the individual codes, several modifications and code restructuring were implemented in RELAP5/SCDAP to provide direct coupling between models. These modifications include:

(1) Multi-specie noncondensible and solute field equations. The source term for the noncondensibles (Xe, Kr, H, N, He) and solutes (C3 0H, CsI) are included in the conservation equations.

13-3

(2) State properties for coolant mixtures at temperatures up to 3500K.

(3) Time-varying geometry based on rod ballooning and frozen core materials for fluid volumes.

(4) Multi-component coolant properties for the fission product behavior model. A liquid state is also included.

(5) Input processing, initialization of TRAP-MELT and part of the SCDAP model, together with unified output, restart, and plot capabilities.

User conveniences and efficiency were also addressed. For example, although RELAPS/SCDAP is an integral code, it is structured such that the user can run RELAP5 alone, RELAPS with the SCDAP models, RELAP5 with the TRAP-MELT models, and of course, the entire set of RELAPS/SCDAP/ TRAP-MELT models.

Detailed input checking identifies all errors in one pass through the input deck, thereby reducing user time spent on problem initialization.

Dynamic storage allocation based on the number of hydraulic volumes, fission product species, aerosol size bins, etc., reduces memory requirements.

Two calculations have thus far been used to demonstrate overall RELAP5/SCDAP capability. The first is a simple bundle calculation similar to the Power Burst Factility (PBF) SFD 1-1 test scenario. Nearly all the fuel behavior and fission product models and part of system thermal-hydraulic model were exercised using this problem. The second is a pretest calculation for the OECD Loss of Fluid Test (LOFT) FP-2 test. Post-test analysis showed that the system thermal-hydraulic response and fuel temperature histories predicted by RELAPS/SCDAP were in good agreement with experimental data.

Future development activities will focus on assessment and code refinement.

Assessment is planned using data from experiments conducted in the Annular Core Research Reactor, NRU reactor, PBF, Marviken, and OECD-LOFT facilities.

In addition to code improvements identified through assessment, new ca'pabilities will be added in response to requirements of the Severe Accident Sequence Analysis (SASA) program.

13-4

l l MELPROG: AN INTEGRATED MODEL FOR

( IN-YESSEL MELT PROGRESSION ANALYSIS J. E. Kelly Sandia National Laboratories Albuquerque, New Mexico 87185 t

i l Since the TMI-2 accident, renewed emphasis has been placed on improving i our understanding of severe accidents in light water reactors (LWRs) . One l avenue of research has been to create detailed systems models for the i response of the reactor core and vessel, the reactor coolant system (RCS)

} and the reactor containment. In this paper, we describe one such core and

vessel response code, MELPROG, which is being developed at Sandia and Los Alamos National Laboratories as part of the USNRC and Foreign Partners' Severe Fuel Damage program.

i The purpose of MELPROG is to provide at any time during a core meltdown i accident sequence (i.e., from the initial core-damaging stage to the stage

! at which the vessel fails and core materials are discharged to the containment) a description of (1) the state of the reactor core and sur-j rounding in-vessel environment, and (2) the disposition of core materials

(in particular, fission products) contained within the reactor coolant sys-i tem boundary. The principal processes which are to be modeled by MELPROG include:

i 1. Vessel thermal-hydraulics and core heat-up;

2. Liquefaction, melting, and mechanical failure of reactor core 4 materials;

~

3. Solidification and freezing of fuel, cladding, control, and structural materials;

4. Fragmentation and relocation of liquefied and solidified materials;

5. Stressing and ablation of major structural features including core supports and the pressure vessel;

6. Energy-generation processes, including decay heating and chemical j oxidation; 4 7. Associated thermochemical and physical processes which affect the

release, transport, and deposition of fission products.

The first version of MELPROG, MELPROG-PWR/ MODO, has been completed and

. is currently being tested and applied to a variety of accident sequences.

This version consists of five (5) explicitly linked modules each of which is comprised of models that treat the physical processes that occur during a severe core-damaging accident. A one-dimensional fluid dynamics model (FLUIDS module) and PWR core structure models (STRUCTURES module) are used I in this version. Modules for debris behavior (DEBRIS module), radiation heat transfer (RADIATION module), and fuel and control rod behavior (PINS module) are also included. Additionally, this version is linked to TRAC-PF1 to provide a complete RCS analysis capability.

I k

l i

13-5 i

M

The second version, MELPROG-PWR/ MODI, is currently under development.

This version includes all features of the original code plus additional enhancements. In particular, this version includes a two-dimensional fluid dynamics model (FLUIDS-2D module), a fission product analysis model (VICTORIA module), an improved core structure model (CORE module), a melt-water interaction model (IFCI module), and a melt-ejection model (EJECT module). With the two-dimensional fluid dynamics model, such important effects as in-vessel natural circulation can be calculated. The VICTORIA module allows the code to analyze the important aspects of fission product behavior in an integrated manner. The CORE module will allow the treatment of both BWR and PWR core structures in a detailed and consistent manner.

The MODO version of the code is currently being exercised on three separate calculations. The first is an SID sequence which involves a medium-size break in the cold leg followed by failure of the emergency core cooling system. This calculation has been run from initiation through vessel failure both with and without accumulator dump. Although the modeling of this accident is relatively simple, the calculation does cover the range of anticipated phenomena. The second is a TMLB' sequence for the Surry plant. This sequence leads to core meltdown at high pressure. The modeling of this case is relatively complex in order to obtain a best-estimate within the one-dimensional fluid dynamics approximation for this sequence. This calculation is being compared to MARCH code results as well as those obtained with the 2-D version of MELPROC (MODI). The third calculation is actually an analysis of the DF-1 experiment. The MODO version of the code has been modified in order to model the specific features (e.g., view factors) of this experiment. The results of this calculation can be compared directly to the experimental results.

In the full paper, both the models, methods, and status of the two versions of MELPROC will be discussed. In addition, a description of the calculations together with the results will be presented.

i 13-6 i

i

__ - = . = _ - . - - __.

THE. BEHAVIOR OF REACTOR CORE-SIMULANT AEROSOLS DURING HYDROGEN / AIR COMBUSTIONa L. S. Nelson, W. Einfeld, K. P. Guayb and G. D. Valdeze l Sandia National Laboratories Albuquerque, New Mexico and l J. H. Lee, R. Knystautas, M. Fresko and M. Gaug l McGill University Montreal, P.Q., Canada i

! During a nuclear power plant accident in which the reactor core is hypothesized to overheat and melt, aerosols could be generated in large con-1 centrations along with a combustible hydrogen / air / steam mixture in the i

containment building. If this mixture were to ignite, it is not known how the aerosols would affect the combustion and the accompanying heat transfer and pressure rises, nor how the burn would affect the aerosols chemically.

4 In order to learn more about the effects of aerosols on gas phase combustion, t we have ignited hydrogen / air mixtures (steam was omitted to simplify the experiments) throughout which we dispersed simulated condensates and mists from molten or vaporized (a) fuel and structural materials and (b) fission products.

i Dur simulants of (a) were Al23 0 , Fe2 03 and metallic Fe. The combustible

mixture was 6.5 v/o hydrogen in air used because of its relationship to the deliberate ignition systems installed in existing plant containments. Our simulant of (b) was a mixture of 10 w/o Csl and 90 w/o Al 02 3 powders simi-larly dispersed throughout the chamber gases. With this aerosol, we used combustible mixtures from 6.5% v/o up to the stoichiometric composition of 29.6% v/o hydrogen in air in order to assess the range of chemical effects produced by the burn.

I The aerosol-hydrogen combustion experiments were performed in the 5.1 m3

! VGES chamber at Sandia National Laboratories and in a second chamber with a volume of 0.18 m3 , located in the Department of Mechanical Engineering at

McGill University, Montreal, Canada. The aerosols were produced by gas-burst dispersal of specially prepared powders into the chambers that contained previously mixed hydrogen-air compositions. Powders from the same batches were used in both chambers. Several seconds after dispersal, combustion aThis work was supported by the U.S. Nuclear Regulatory Commission and l performed at Sandia National Laboratories which is operated for the U.S.

Department of Energy under Contract Number DE-AC04-76DP00789.

j b Assigned to Sandia National Laboratories by Ktech Corporation,

Albuquerque, NW.

cAssigned to Sandia National Laboratories by Technadyne Engineering Consultants, Albuquerque, NM.

);

13-7

I was initiated with a spark or bridgewire. (Control experiments were l performed by omitting the powders but otherwise following identical gas-burst and combustion procedures.)

The oxidic aerosols in (a) only slightly reduced the peak pressures and temperatures produced during the hydrogen-air combustions. Moreover, there '

was little difference detected between the thermally stable and unstable oxides, A1203 and Fe2 0 3, respectively. Both slightly lengthened the decay of pressure and temperature after the peak values roughly proportional to the amounts of aerosol dispersed. The behavior appeared to be independent of the chemical nature of the oxides.

On the other hand, the presence of a metallic iron aerosol substantially

] increased the energetic effects of the burn. In experiments performed with i

iron aerosols dispersed in 6.5 v/o hydrogen / air mixtures, the peak pressure rises produced with metallic aerosol present were 20.3 MPa, compared to 20.1 MPa in control experiments performed identically without the aerosol. Ther-i mocouples exposed to the burning mixtures registered maximum absolute temperatures 50 to 90% higher than in the control experiments. Moreover, pre- and post-combustion gas analyses indicated that 95% of the hydrogen burned with the aerosol present compared to only about 60% without the aerosol. Reddish brown and black deposits were found in the chamber after the burns, suggesting the formation of Fe203 and Fe30 4, respectively.

Clearly, the iron aerosol had burned rapidly along with the hydrogen, adding its enthalpy of combustion to that of the gas mixture.

We also studied the ignition limits of iron aerosols dispersed in air

] alone and in lean hydrogen-air mixtures. We performed experiments in both chambers. We found that the lean limits for ignition of the iron aerosol dispersed in air alone were 2400 g/m3 and 275 g/m3 for the smaller and larger chambers, respectively, suggesting a possible scale effect. As small amounts of hydrogen were added to the air, the minimum concentration of iron needed to sustain combustion decreased roughly as predicted by the Le Chatelier rule for mixtures of fuels. Deviations from the rule were greater for the larger chamber than for the smaller chamber, again suggesting a scale elfect.

l j Large amounts of elemental iodine were produced when the CsI-Al 023 l

aerosol was exposed to hydrogen / air combustion. We performed these burns in the VCES chamber with compositions between 6.5 and 29.6 v/o hydrogen in air;

each burn was performed with I kg of 10 w/o CsI-90 w/o Al 023 dispersed

. throughout the combustion volume. Chemical analyses indicate that as much as

75% of the iodide ion present in the CsI was oxidized to molecular iodine during the burn at 29.6 v/o, the stoichiometric composition.

We are currently applying the MARCH and CONTAIN codes to a hypothetical Surry-type accident sequence. We are investigating off-site consequences of releasing radioactive iodine as particulate Cs1 compared to gaseous elemental l iodine.

Although these experiments are highly simplified simulations of the j accident situations that might prevail in a nuclear plant, we feel that

significant trends have been identified for further study with more proto-typical aerosols in larger volumes.

13-8

TELLURIUM CHEMISTRY, TELLURIUM RELEASE AND DEPOSITION DURING THE TMI-2 ACCIDENT Krishna Vinjamuri Robert A. Sallach*

Daniel J. Osetek Richard R. Hobbins Douglas W. Akers Idaho National Engineering Laboratory EG&G Idaho, Inc.

P.O. Box 1625 l Idaho Falls, Idaho 83415 l

I *Sandia National Laboratory l

P.O. Box 5800 Albuquerque, New Mexico 87185 Summary This paper presents the estimated chemistry and transport behavior of tellurium during and after the THI-2 accident. Thermodynamic calculations indicate that H 2Te is the predominant vapor species which results from the presence of excess hydrogen and high total pressure (8.2 to 15.2 MPa) in the upper plenum of TMI Unit-2. Increasing the system temperature will tend to disso-ciate H2Te. However, temperatures 21200 K are needed for this to occur.

The tellurium behavior oresented in this paper is based on all available measurement data for 129mTe, 132 T e, stable tellurium (126T e, 128Te, and 130T e), and best estimate calculations of tellurium release and transport.

The predicted release was calculated using current techniques that relate release rate to fuel temperature and holdup of tellurium in zircaloy until significant oxidation occurs. The calculated release fraction was low (10%),

but the total measured release for samples analyzed to date is about 5.8%.

Of the measured tellurium in the containment sump water, upper plenum assembly surfaces, containment solids in the sump water, makeup and purification demineralizer, containment inside surface, and the reactor primary coolant there was about 2.4,1.8, 0.88, 0.42, 0.17 and 0.86% of core inventory, respectively. A significant fraction (54%) of the tellurium predicted to be retained on the upper plenum surfaces (5.4% of the core inventory) was deposited during the high pressure injection of coolant at about 200 min after the reactor scram. Comparison of tellurium behavior with in-pile and out-of-pile tests suggests that zircaloy holds tellurium until significant cladding oxidation occurs. Analyses of samples from the core region of TMI-2 indicate that about 49% of core inventory is retained in the surface of the debris bed. Core samples taken from 0.28 to 0.94 m into the debris bed contained lower amounts of tellurium, suggesting that a highly volatile tellurium species was released form the hot debris bed and deposited in the cooler surface debris bed. No correlation was found between the atoms of tellurium and those of tin, zirconium, iron, chromium, and nickel.

13-9 I

ICE CONDENSER EXPERIMENTAL PLAN L. D. Kannberg, P. C. Owczarski, and A. M. Liebetrau Pacific Northwest Laboratory *

SUMMARY

An experimental plan is being developed to validate the computer code ICEDF.

ICEDF was developed to estimate the extent of aerosol retention in the ice compartments of PWR ice condenser containment systems during severe accidents. The development of the experimental plan has involved a somewhat unique ite.ative procedure whereby candidate statistical test designs are examined with ICEDF prior to experimentation. The merit of the method is that it permits the use of efficient testing methods, as derived from statistical test design procedures, with insight gained from application of the computer code. Since the data developed from the testing will be used for validation of the code, the development process cannot rely solely on ICEDF calculations lest significant phenomena overlooked in the code are similarly overlooked in the experiments.

The development of the experimental plan began with review of available information on the conditions under which the code will be applied.

Typically, the code will be used to estimate aerosol retention during severe accidents. Available computer generated estimates of thermohydraulic and aerosol conditions entering the ice condenser were evaluated and representative ranges were identified for each variable. These ranges of conditions (henceforth called design criteria) were used in conceptual test assembly design and for generation of the first round of statistical test designs. Consideration of the phenomena to be evaluated in the testing program as well as equipment and measurement 1 imitations has led to changes in the design criteria.

The overall strategy in development of the experimental plan includes the following sequence of events: 1) development of design criteria, 2) development of candidate statistical test designs, 3) evaluation of candidat,e test designs (using ICEDF), 4) revision of design criteria and/or candidate test designs (if necessa ry), 5) statistical comparison of the selected design (s) against competitive designs, 6) reevaluation of the selected design (s) (if necessary) and, finally, 7) utilization of the final experimental plan in the conduct of tests. Estimates of experimental variability made prior to actual testing will be verified by replicate testing at preselected design points.

This work was supported by the U. S. Nuclear Regulatory Commission under Contract DE-AC06-76RLO 1830, NRC FIN B2444.

Pacific Northwest Laboratory is operated by Battelle Memorial Institute for the U. S. Department of Energy.

13-11

_. .~ -- _ _ _ __ .__- .- . . - . - . - .-. . - .

APPLICATION OF RELAP 5 TO ANALYSIS OF THE DOEL STEAM GENERATOR TUEE RUPTURE AND STUDIES OF THE LOSS OF FEEDWATER AND FEEDWATER LINE BREAK 6

I TRANSIENTS By E.J. STUSBE and L. VANHOENACKER TRACTIONEL -

BRUSSELS -

BELGIUM l

The RELA 1 5 Code has been used extensively since 1991 as the main I

thermal-hydraulic system code at TRACTIONEL. The principal analyses

concerned the DOEL 1 and 2 power plants- (392 MWe, 2 Loop PWR) which are not strictly covered by the existing generic vendor studies, with the purpose

1

- to improve the understanding of various phenomena occurring during postulated plant transients

- to update the various emergency procedures and improve operator interventions

- to reanalyse some accident analyses in the framework of the 10-yearly revision of those plants

- to define and analyse the PTS limiting transients.

The DOEL-2 steam generator tube accident of June 1979, yielded a rare opportunity to perform some code assessment studies on a full scale a

plant, with clean initial conditions but some unknown boundary conditions (i.e. operator interventions).

l This complex incident has been analysed by means of the PELAP-5 code l versions MOD 1-CYCLE 14 and MOD 2-CYCLE 36. This paper highlights some im-portant improvements over the MOD 1 code version in terms of local mass errors, pressurizer spray modeling, break flow behaviour and code !

performance (CPU times) , and illustrates the overall agreement between calculated and recorded plant data, i

I .

14-1 4

l

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

The loss of feedwater, and the feedwaterlinebreak accidents were analysed with very conservative initial and boundary conditions (e.g. extreme setpoints, minimal safeguard systems) and the analyses

! manifest some modalisation sensitivity of the secondary side steam generator while break flow behaviour and code performance were quite 1

i good, i

i For the different transients, several safety issues are discussed such I as

- importance of pressurizer spray and spray efficiency

- interpretation of water level measurements in pressurizer and steam generator j - operator control over the HPSI systems in a SGTR accident i

- the efficiency of the auxiliary feedwater to remove decay heat.

i a

l This paper concludes by some specific remarks about the need to freeze

!- the code in its actual state.

l i

i i

1 I

l l

l t i l

l l

14-2

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

FRG-Assessment of TRAC-PF1/ MOD 1 and RELAP 5/ MOD 2 BY F. WINKLER KRAFTWERK UNION AG I

The BMFT of the FRG has entered into an agreement with the USNRC to per-l l form 50 code assessment calculations in return for receiving the 4 advanced computer codes TRAC-PF1/ MOD 1, RELAP 5/ MOD 2, TRAC-BD and TRAC 43F. These assessment analysis will be performed entirely by KWU and GRS over a five year period, starting during 1985.

The amlysis will include 8 KWU-PWR and 7 KWU-BWR postealculations of NPP-commissioning transient tests, pre and post calculations fer UPTF, Karl-stein calibration test facility and PKL, and a limited number of calculations for other mutually agreable facilities. The PWR-NPP commissioning test analysis will contain post test calculations of reactor scram, turbine trip, trips and starts of main coolant pumps and loss of on-site and off-site electric power. The BWR-NPP analysis is related to load rejection, re-circulation pump trip, trip of feedwater heater, closing of main steam line isolation valves and reactor power and level control. The analysis of UPTF consists of both seperate effect and integral test calculations of the full size 3 dimensional upper plenum test facility. Seperate effect tests will be primarely used for assessment of the special models contained in the codes. The integral tests will be utilized for assessing the codes capability to calculate full system behaviour and the interaction between the various special models. The Karlstein calibration test data will be used to study the special TRAC co-current and counter-current flow models in the end box of a single fuel bundle under saturated and subcooled conditions. PKL-I experimental resalts will be applied to small leak code assessment. The major part of the code assessment using PKL is devoted to PKL II large LOCA tests including end of blowdown, refill and reflood. Emphasis for PKL will be to validation of RELAP 5/ MOD 2 also taking into account multi channel simulation. Eleven calculations have not yet been selected and will be chosen as needs become better prioritized in the future through allready performed studies, to include additional facilities, or to address future issues. The specific analysis to be perfbrmed are described in the following table.

14-3

FRG PROPOSAL FOR 50 TRAC AND RELAP5 CODE ASSFrbENT STUDIES TEST FACILITY STUDIES RELAPS/M002 IRAC-Pf1/M001 IRAC-BD1/M001 CODE per TEST TRAC-BF1 SELECIION OPEN f 180 TOTAL F IBD I0IAL F IBD 10iAL -

UPif 12 - - -

1 - 12 - - -

4 7 KARLSTEIN 2 - - - 2 -

2 - - - -

CALIBRATION TEST PKL-I 1 - - -

PKt 10 6 4 - ~ ~ ~

pg 77 3 4 2 2 PWR-COMISSIONING 8 3 -

3 1 - 1 - - - 4 IESTS BWR-ComISSIONING 7 - - - - - -

7 -

TESTS EXPERIMENT AND TEST- 11 - - - - - - - - -

11 FACILITY OPEN TOTAL 50 5 4 9 10 9 19 7 0 7 15 (f=22)

F = FIXED; 180 = CALCULATION 10 BE DETERMINED i

All four codes have been delivered to FRG and are being installed on the CYBER 176 at KWU, and partly on the AMDAHL at GRS and the CRAY-1M at the UNIVERSITY OF STUTTGART. The PWR-NPP and BWR-NPP commissioning, PKL and l

, Karlstein tests have been completed, and the UPTF tests will start early l

l 1986.

l By participating in this international code assessment program, FRG expects to make a useful contribution to the development and validation of advanced thermohydraulic computer codes which can be used reliable for safety analysis of nuclear power plants.

14-4

ANALYSIS OF BWR6 - LOCA WITH REDUCED ECCS USING TRAC-BD1 S.N. Aksan and G.Th. Analytis Swiss Federal Institute for Reactor Research (EIR),

! 5303 WUrenlingen, Switzerland 1

SUMMARY

A very important problem in the analysis of Loss-of-Coolant Accident (LOCA) and severe reactor accidents is related to the efficiency of the emergency core cooling systems (ECCSs) in preventing reactor fuel from reaching excessive heat-up and degradation. The evaluation of core-coolability becomes particularly important in the case of delayed action of the aforementioned systems and/or if the actual water injec-tion rate they provide is limited to a fraction of the full desigti capacity.

The transient reactor analysis computer code TRAC-BD1 version 12 was employed to analyze the BWR-6 core behaviour under LOCA conditions with limited ECCS. In this paper, we shall be presenting the results of a (generic)BWR-6 recirculation line break for the case that only one low pressure coolant injection (LPCI) is operational, and we shall compare these results to the case that all ECCSs are normally functioning. The details of the model as well as the nodalization used will also be presented.

In the case that all ECCSs are operational, the transient lasted appro-ximately 150 sec between pipe rupture and core quench. Fuel rod clad-0 ding temperatures in the high power bundles reached a peak of 700 K .

In the case of limited ECCS (only one LPCI operational), the transient was calculated to the point where the gradient of the temperature in-crease of the high and average power channels had decreased to a very low value indicating that steam cooling was becoming effective. A 14-5

y 2-temperature of 1000 K was reached 320 sec. after the pipe rupture, while the low power channel had already quenched. The continuous in-jection of water from the one LPCI, its subsequent evaporation and the steam cooling which take place in the average and high power chan-nels brings the system into a kind of " pseudo-steady state", while )

the low pressure persisting in the one-dimensional components of the system (pipes) leads to some stability problems which can be resol-ved by employing a very small time-step at the later stages of the transient.

The main conclusion of our calculations is that even with one LPCI operational, the operators have at least 6 minutes to bring additional systems into operation before any severe damage occurs in the core.

Clearly, the steam cooling taking place delays any sharp temperature increase in the high power bundle.

a-i I

14-6

UK EXPERIENCE WITH TRAC-PFl/ MODI and RELAP5/ MOD 2 I Brittain*

S Board +

K Routledge*

SUMMARY

The UK has been using versions of TRAC and RELAP5 for best estimate PWR LOCA analysis for a number of years. In the pre-construction phase of the Sizewell 'B' plant the codes were used to provide an independent assessment which could be compared with the EM-based safety case. It is generally agreed that advanced code calculations will play a more direct role in the pre-operation phase of the project, though the precise mode of use has not yet been determined.

Our experience with TRAC-PFl/ MODI covers both large and small break LOCA.

The general approach has been to carry out plant studies to identify the main phencuena of interest, to see to what extent these phenomena arr present in integral tests, and to use separate effects tests to improve and/or validate specific models which have been shown to be inadequate or not tested in the integral tests. For the double-ended cold leg break a coarse 300 cell model of a typical W 4-loop plant was set up and run.

The results have been compared with those from a detailed 950-cell model which had previously been developed in a joint exercise with LANL.

Although there are some differences, it is concluded that the coarser model is adequate for most purposes. Integral test comparisons have concentrated on LOFT, the most recent being test LP-LB-1. LOFT confirms most of the phenomena seen in the plant calculation, in particular core liquid flows during blowdown, condensation-driven oscillations, and the importance of nitrogen injected from the accumulators in the initial rapid core reflood. Because of scaling problems, LOFT does not exhibit the post- accumulator phase of reflood under pumped ECCS conoitions which is seen in the plant calculations. Important models requiring further study include post-dryout heat transfer, condensation and entrainment during reflood. Comparisons have been made with condensation experiments in a glass rig and with the THETIS bundle reflood experiments. For small breaks the integral analysis has concentrated on the LOBI International Standard Problem, a 1% cold leg break. This has highlighted the importance of modelling stratification in the horizontal pipes, and its effect on the flow through a small diameter off take. Separate effects studies have been carried out on horizontal flow modelling and also on level swell in partially-uncovered cores.

United Kingdom Atomic Energy Authority

+ Central Electricity Generating Board

National Nuclear Corporation 14-7

[

For RELAP5/ MOD 2 our experience is -more - limited, and consists of collaboration in some - of the code development (reported at the 1984 Information Meeting) preliminary work in analysing LOFT small break tests and carrying out small break sensitivity studies for the Sizewell 'B '

plant.

Our overall conclusions are:

1. For large breaks RELAP5 'is not a serious contender because, unlike TRAC, it does not have a 3-D capability. TRAC contains models of all the important large break phenomena, though modest improvements to some of the models are desirable.

2. For small breaks both RELAP5 ' and TRAC appear to be potentially capable of useful BE predictions. However, both codes-require some further development, particularly in the modelling of flow separation in large horizontal pipes.

3. The running times are such that the codes can only be used for limited investigations at present. Both codes require further attention to the numerics of model implementation and the better use of parallel processing in order to reduce computer running time to an acceptable level for more general use.

4

4. For both codes the available documentation is incomplete. It is necessary to examine the FORTRAN source in order to understand some of the models. Further work-on User Guidelines is also needed.

a Y

a 14-8 l

l_ , . .

nur.aps;/M)D2 ARM 6T AT RARCOCK & WIL(X)I C. K. Eithianandan, N. H. Shah, R. J. hhamaker Babcock & Wiloor C. Turk Arkansas Power and Light Company Babcock & Wilcox (B&W) has been working with the code developers at B3&G and the NRC in assessing the REA5/ MOD 2 computer code for the past year by simulating selected separate effects tests. The purpose of this B&W Owners Group-sponsored assessment was to evaluate RELA 5/ MOD 2 for use in design calculations for the MIST and OTIS integral system tests and in the prediction of pressurized water reactor transients. B&W evaluated various versions of the code and made recommendations to improve code performance. As a result, the currently released version (Cycle 36.1) has been improved considerably over earlier versions. However, further refinements to some of the constitutive models may still be needed to further improve the predictive capability of RELA P5/ MOD 2, Method of Evaluation A selected number of separate effects tests were simulated using the first released version of RELAP5/ MOD 2 (Cycle 22). The results were compared with the test data and with REDBL5 (B&W Version of RELA 5/ MOD 1) and/or REAP 4/ MOD 6 predictions. Observed problems or limitations and the probable causes were reported to EG &G and NRC. ED &G made improvements to the code to eliminate these deficiencies. B&W subsequently reevaluated the updated versions of the code and further recommendations were made as needed.

RELAP8i/lGD2 Yernions Evaluated The following versions of RELAPS/ MOD 2 were evaluated by B&W o REA5/ MOD 2/ Cycle 22 - First released version o YELA5/ Cycle 32 - B3&G test version of REA5/ MOD 2/ Cycle 32 o RELA 5/ MOD 2/ Cycle 36 - Frozen cycle for international code assessment o Updates to Cycle 36 based on recommendations developed by B&W during the simulation of an MIT pressurizer test o Cycle 36.1 Updates received from EG&G.

Senarata Effecta Si=ninted Most of the separate effects calculations selected for this study were based on experience with RELAPS/MODl. In most cases, the RELAP5/ MOD 2 input models were generated from REDBL5 (B&W version of RELA 5/ MOD 1) input files (withminimum changes) so that direct comparison between the two code-calculated re7ults could be made. The separate effects and the corresponding tests simulated are given below:

14-9

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

I

1. Transient critical flow GE Vessel Blowdown Test Edward's Pipe Blowdown Test j -

Batelle-Northwest CSE Blowdown Test B-9 i -

Marviken Critical Flow Test 24 l 2. Level swell during depressurization - '

GE Level Swell Test 1004-3 3 Steam Generator s

.B&W 19 Tube Steam Generator (IE0TSG) Steady State and Transient Test 13

4. Pressurizer MIT Pressurizer Test r Phenomenological Test Problem; Results Compared with E* LAM / MOD 6

, and REDBL5

' 5. Subcooled Boilina j Christensen's Subcooled Boiling Tests 10 and 15

j. . 6. Wall condensation MIT Pressurizer Tests

' B&W Single Tube Tests

7. Voiding of hot leg U-Bend

' Phenomenological Test Problem. Results compared with REDBL5 'and RELAPS/ MOD 1 l, 8. Reflood.

FLECHT-SEASET Forced Reflood Test-31504 i

Emanita and Dimanmaion

' All' the tests (except the MIT pressurizer test, the B&W single tube ~ tests, and'the reflood test) were simulated using Cycle 22. The following observations 4 were made from this study:

, o The pressure was highly underpredicted in the MIT pressurizer test.

} o For the IE0TSG test, Cycle 22 undercalculated primary. to secondary heat transfer and predicted non-equilibritaa mist flow (about 975 quality) l instead of dry superheated steam flow at the secondary side exit during steady state. During refill, following the loss of feedwater, large positive spikes were observed in steam; flow and heat transfer.

o For the Edward's pipe test case, Cycle 22 calculated oscillatory vapor generation rates and void fraction along the pipe axis.

o . Void oscillations were also observed in the hot leg U-bend sample

~ problem 4

o The void fraction was over-predicted throughout the test section for both Christensen's testa 10 and 15. Near the bottom of the test section

Cycle 22 predicted slug flow even ' though the void fraction was about ,

0.001.

1 l While the assessment was . ongoing at B&W, ED&G was actively revising the

! co.de. When the Cycle ~22 assessment was completed at B&W, ED40 reported that i the code' had be en completely revised and provide d the test version,

YELAP5/ Cycle-32, to B&W for further assessment.

t 14-10

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

V Using YELAPS, B&W simulated the MIT pressurizer , IE0TSG, Edward's pipe, the following conclusions were made.

l and U-bend tests. From this evaluation,

! o For the MIT pressuriz er test, the pressure response was reasonably well predicted.

{

4 o For the IE0TSG test, YELAPS and Cycle 22 results were almost the same.

o YELAPS did not predict non-physical oscillations in Edward's pipe and bot-leg U-bend teets.

Hearathile, fur ther code modifications were made by E &G, and the revised code was officially released as REAPS / MOD 2 Cycle 36.

l B&W subsequently simulated the YELAP5 test cases using Cycle 36. The resulta j were similar to ' the YEAPS results except for the MIT pressurizer test. In j the pressurizer test case , Cycle 36 results showed oscillatory behavior in pressure when a maxista time step of 0.1 see was used. Temperature spikes <

on the order of 2000F were observed in the steam node immediately above the volume containing a two-paase level. E&G had found that the results became worse when a maxista time step of 0.05 seconds was used. However, the oscillations l disappeared when the maximum time step used was 0.025 sec. EM ooncluded that a deficiency in the vertical stratification model was the cause.

The B&W version of Cycle 36 was updated to include the E&G code improvements and the MIT pressurizer test was simulated using the three maxista time steps.

The pressure oscillations and temperature spikes in steam voltaes were not observed in any of the three cases. However, in the 0.1 sec maxista time step case, large pressure spikes (on the order of 2000 paia) in liquid filled

voltaes were observed due to water packing. E&G modified the water packing j code logic which was included in the officially released Cycle 36.1 updates.

4 These updates were incorporated in the B&W version, and the MIT pressurizer case was rerun with satisfactory results.

RELAPS/ MOD 2/ Cycle 36.1 was further evaluated by simulating the GE level t swell test, the two Christensen's tests, and two B&W tests to calculate I condensation inside an IEOTSG tube at high pressure. In all cases, the calculated l results agreed reasonably well with the data.

The reflood model in Cycle 36 was evaluated by simulating the FLECHT-SEASET forced reflood test 31504. From this study, it was concluded that further refinement to the wall heat transfer model and some of the constitutive models would be needed to correctly predict the reflooding phencoena, i

! ele &EI 1

In stamary, the RELA 15/ MOD 2 code assessment at B&W has led to improvements by E M in subsequently released versions. As a result, the performance of j the latest released version, REAP 5/ MOD 2/ Cycle-36.1, is much better than earlier versions of the co de. However, further improvements of some of the models

' may be required to improve simulation of 19-tube IE0TSG test case and to more accurately predict local reflooding phenomena, i I 14-11 t

__ _ , . - _ . ,___ _ , _ . _ _ _ _ _ _ _ _ ~

l 1 l l

SURRY STEAM GENERATOR - EXAMINATION AND EVALUATION Robert A. Clark, Pamela G. Doctor, and Robert H. Ferris ;

Pacific Northwest Laboratory l Richland, Washington The Steam Generator Group Project, an NRC program with joint sponsorship by EPRI and consortia from France, Italy and Japan, is completing the fourth year of a five year research program. Goals are to utilize the retired from service Surry steam generator to better understand the reliability of detec-tion and accuracy of defect sizing provided by current practice eddy current in-service inspection (ISI). In addition to defining the performance of current field practice, the program is also exploring advanced nondestructive examination (NDE) techniques to determine performance. Validation of NDE is provided through the removal of specimens from the generator with subsequent destructive and out-of-generator nondestructive examination. The NDE valida-tion information and data on remaining integrity of service defected tubes (obtained from burst testing) will be incorporated into an ISI model which will address the period between the extent of inspection for safe generator operation.

In previous years, the generator was prepared for extensive primary side investigation by decontamination of the channel head and removal of the plugs inserted into the defected tubes during service. Two approximately 100%

multifrequency eddy current inspections were conducted to determine a post-service baseline of generator conditions. From these examinations, extensive secondary side characterization and historical records, a subset of tubes was determined for repeated NDE investigation. Round robin multifrequency eddy current examinations / analysis by five teams using identical equipment were then conducted.

During the past year, analysis of the data acquisition / analysis round robin was completed. A domestic eight team data analysis round robin, with all teams analyzing the same set of Zetec MIZ-12 generated eddy current signals, was conducted. A similar analysis round robin using Intercontrole IC3FA generated signals is just being completed in France. Multiple inspections of the Surry generator round robin tubes were conducted by alternate and/or advanced NDE techniques. This included investigation by teams from ORNL (using the NRC funded multiparameter eddy current system), Japan (using the Mitsubishi Heavy Industries eddy current system and special probes), Germany (using Kraftwerk Union eddy current systems, ultrasonics and rotating eddy current probe), EPRI-J.A. Jones NDE Center (using Zetec MIZ-12 and MIZ-18 systems), and Babcock and Wilcox (using the PROFIL 360 profilometry system).

Results from the advanced / alternate NDE inspections in general agreed with the detection of defect indications from multifrequency eddy current inspection.

Special probes did in cases discover additional defect indications not found in other inspections. There were several instances where alternate NDE 15-1

l techniques resulted in different defect sizing for common detections. Pre-liminary analysis of the domestic data analysis round robin showed a signifi-cant variation in common defect detections and sizing agreement. In all of the round robins, the level of agreement in sizing and detection varied .at different locations throughout the generator. The best agreement was generally at the hot leg top of the tube sheet, where most defects appear to be located

' in this unit. There is -poor agreement in most other locations. The subject generator is not easy to nondestructively characterize. Heavy conductive deposits, dented tubing and permeability variations in the tubing all con- l tribute to difficulties in analysis of eddy current data. '

During this last year, the first specimen removals were made allowing initial validation of the NDE results. Tubes were pulled from both the hot and cold leg sides of the tube sheet to above the sludge pile. In addition, three tube sheet sections (2 hot leg and I cold leg), each including 9 tubes, were removed by overbore drilling of surrounding tubes. The cold leg tube sheet tube specimens have a heavy. continuous layer of copper metal deposit. The hot leg tube sheet tube specimens exhibited a narrow band of deep pitting, slightly above the top of the tube sheet. Two pulled hot leg specimens, which had remained in-service at the end of generator life, had s85% throughwall pits. All eddy current examinations detected this defect but undersized the maximum defect depth. Ultrasonic examination appears to have closely deter-mined the maximum pit depth. Post-removal eddy current examination has also i,

revealed what might be intergranular attack (IGA) near the top of the tube sheet and 3" below the top of the tube sheet. At this writing, metallurgical validation is being conducted on these specimens. However, no in-situ NDE identified IGA in the tube sheet crevice.

Efforts are currently concentrated in specimen removals and metallurgical validation of NDE detections. Preliminary results from the removed tube sheet sections are expected by mid-October. These investigations are addressing damage to the tube sheet as well as the tube, examining the corrosive environ-ment which exists in the tube to tube sheet crevice.

l 1

4 i

l l

l 15-2

STEAM GENERATOR TUBE VIBRATION STUDY Walter I. Enderlin Pacific Northwest Laboratory Richland, Washington 99352 Dr. Joseph Muscara, NRC Project Manager The buildup of magnetite in the steam generators of some presscrlced water reactors (PWRs) has led operators to propose chemical cleaning to remove this product. In some cases, the volume of magnetite formed by the corrosion of the carbon steel has been sufficient to cause " denting" or reduction of the outer diameter of the tubes where they pass through the support plates. Moreover, when the magnetite is removed by a chemical cleaning process, the diameter of the hole in the tube support plate is increased. The U.S. Nuclear Regulatory Commission (NRC) has expressed concern that the resulting increased clearances may allow an increased level of flow-induced vibrations in chemically cleaned steam generators that may in turn lead to high tube wear rates and unacceptable levels of tube failure.

The Pacific Northwcst Laboratory (PNL) is supporting the NRC staff in address-ing the effects of increased tube / tube-support clearances. The objective on PNL's work is to provide NRC with criteria with which to evaluate licensees' specific proposals for chemical cleaning of steam generators.

Research has been performed in both the U.S. and Canada on vibration and wear of PWR steam generator tubes. However, most of the work performed in the U.S.

is currently classified as proprietary by its sponsors and is unavailable for general use.

In Canada. P. L. Ko and co-workers have been investigating wear in CANDU steam generators for over 10 years. Their work has included experimentation to determine the fretting wear mechanisms involved, development of analytical techniques to predict impact forces at the tube supports, and the correlation of tube wear and tube motion parameters. Ko et al. developed a bench-scale tube fretting apparatus in which they extensively studied the effects of various tube / support plate parameters on tube wear. These researchers found that high wear was caused not by the high-force components such as impact motion that have low probability of occurrence, but instead by the high probability intermediate-range force components (usually combined impact and rubbing motions). The amount of wear was observed to increase exponentially with excitation frequency, and to increase approximately linearly with diametral clearance and excitation amplitude, up to a point. However, the data to substantiate the latter were extremely limited. Moreover, this work was not predicated on tube motions and tube support impact forces that were o perienced within an actual reactor operating environment. Rather, a computer code (VIBIC) was developed to simulate the motion of a cantilevered beam impacting against supports with clearance, and to predict the midspan displacements and the support impact forces. The difference between the experimental and analytical results using this code was found to be greater for those tubes tested in the presence of water than for those tested in air. The VIBIC code 15-3

considers only tube motion in air. The fluid damping effect in the tube / support clearance space and the film effect of the water at the contact surface had not been included in the VIBIC simulation used in Ko's work.

In the U.S., EPRI is currently sponsoring a research project to quantify tube /

tube-support plate interaction characteristics of multispan steam generator tubes and to gain a better understanding of their dynamic characteristics.

This project employs a single-tube test fixture with 3 instrumented supports (1 l eggcrate and 2 cylindrical-hole type). Measurements were recorded in air and !

in cold stagnent water environments for different tube alignment conditions and i for different support clearances. The data were recorded over a range of sinu-soidal and random excitations and then digitally analyzed to compute statisti-cal force and sliding distance parameters. These data provide the basis for wear tests being conducted in an autoclave by Kraftwerk Union AG in Germany.

The investigation undertaken by PNL extends the previous research in that flow tests at prototypic Reynolds number, with tube clearances at the flow entrance region representative of as-built and of post-cleaned conditions, were performed at elevated tceparature and pressure, using a full length scale model of a steam generator tube bundle. The tests were performed to 1) establish the forcing boundary conditions and 2) determine if an environment conducive to fretting wear actually exists under both tube clearance conditions at the instrumented support location.

From the flow test results. PNL researchers concluded that a tube clearance of 10 mils or greater in excess of design clearance would not tend to increase tube wear at the instrumented tube support location. The tube support plate was shown to be considerably less active at this clearance than at the design clearance when fluid flows at 400 F ranged from 50% to 150% of the flow required for prototypic Reynolds number. Tube motion was elliptic under these conditions, and vibration amplitudes were greater than for the design clearance case, suggesting that the tube in this case was restrained by its stiffness rather than by the support plate. It was evident that frequent tube contact did occur at design clearance conditions. Based on these results, a forcing function for accelerated wear tests cannot be defined. It was further concluded that the data did not justify further testing at tube clearance conditions in excess of 10 mils over design clearance, and that accelerated wear testing was not justified.

The tube excitation force expected under field conditions would be even less than what was imposed by the test conditions. Moreover, based on the results of chemical cleaning tests performed by other researchers, the post-cleaning tube / tube-support plate clearance is expected to remain at no less than 10 mils over design clearance, and will likely be even larger. Hence, PNL concluded that under normal operation conditions there is little potential for increased tube wear rate as a result of chemical cleaning at tube support locations with-in a steam generator where conditions are similar to those at the instrumented tube support location. However, the data obtained to date are not sufficient to predict the potential for increased tube wear rate at other tube support locations within a steam generator tube bundle, where conditions differ from those at the instrumented tube support location. To date, no follow-on work has been planned for this project.

15-4

Stress Corrosion Cracking of PWR Steam Generator Tubing Daniel van Rooyen Brookhaven National Laboratory Department of Nuclear Energy Upton, New York 11973

SUMMARY

In a long-term program, sponsored by the NRC, Brookhaven National Laboratory is investigating the quantitative aspects of factors involved in the intergranular stress corrosion cracking (IGSCC) of nuclear grade Inconel 600 steam generator tubing. Such IGSCC has occurred in many plants world-wide, at sites of high stress / strain such as roller expanded transition zones, U-bends or " dented"* regions.

Testing is done in high temperature water (290*C-365'C), either pure or simulated AVT or primary water. Relationships are being formulated between failure times, crack growth rates and factors that influence them, such as temperature, stress, strain, strain rate, alloy processing and microstructure, and electrolyte composition. Equations are set up for predicting expected lifetimes of the tubes. Quantitative estimates of crack initiation times and growth rates are believed useful in predicting service behavior based on accelerated test data. Together with other techniques, such calculations would be useful in determining criteria, for instance, for tube plugging or other remedies, before leaks occur, thus eliminating unnecessary release of primary water and accompanying plant shut-downs.

This paper adds to information given in earlier reports, and includes final plots of failure time vs. temperature for U-bends from which activation energy values are obtained. Also in final state is the relationship between stress and IGSCC initiation. Slow strain rate data are nearly complete, and several sets of new results are included for one heat of commercially produced Inconel 600 tubing. These findings include strain rate and temperature effects.

For validating the quantitative models, arrangements are being made to obtain tubing from the Surry steam generators at PNL.

Future work should include the examination of materials other than Alloy 600, since future plants will be likely to choose a material such as Alloy 690 for which there is no independent NRC data base.

Denting refers to a corrosion-related deformation of tubes during service in afflicted steam generators.

i I

i 15-5

AN NRC APPROACH TO DEPENDENT FAILURE ANALYSIS D. J. Campbell J. R. Kirchner H. M. Paula JBF Associates, Inc.

l Knoxville, Tennessee I

The NRC is currently supporting research to develop an approach to dependent failure analysis that will (1) enable analysts to determine dependent failure contributions to nuclear power plant risk and (2) enable decision-makers to see what factors influence plant risk. The results of this project will be useful to those who must focus their efforts on risk reduction and/or safety assurance.

As part of the Risk Methods Integration and Evaluation Program (RMIEP),

JBF Associates and Sandia National Laboratories personnel have developed I qualitative procedures for identifying and screening dependent failure scenarios that are potentially significant contributors to nuclear power plant risk. Ongoing work is directed toward evaluating and developing methods that can be used to quantify these dependent failure scenarios.

Dependent failures represent a broad class of event types. They include all types of external events; human-error-related events associated with the design, fabrication, installation, and operation of plant components and systems; and events caused by plant environmental conditions. Many external events (e.g., seismic, fire, and flood) are analyzed separately in PRAs and are therefore not being considered in this project. This project is most concerned with events caused by human error and events caused by environmental extremes, although some consideration is being given to the analysis of events caused by fires and floods.

Performing a dependent failure analysis as part of a nuclear power plant PRA involves the collection and evaluation of enormous quantities of data. It can also involve expensive solutions of large fault trees since the quantitative truncation typically used in an analysis of independent failures is generally inappropriate when considering dependent failure scenarios. With these concerns in mind, we are structuring the dependent f ailure analysis procedures to efficiently focus an analysis on only those dependent failures that are important to plant risk.

Human-error-related events can result in multiple component failures regardless of the relative locations of the components. Environmentally caused events only affect components in the same location with respect to a particular environment. Therefore, we have developed two sets of procedures for identifying and screening dependent failure events: one for the analysis of human-error-related events and one for the analysis of events caused by harsh environmental conditions.

16-1

The procedures involve screening to quickly exclude unimportant dependent failures from further analysis. Different types of screening apply to the different types of dependent failures. For the human error-related events, we can screen out multiple component failure combinations that involve dissimilar components. This screening step is based on an extensive review of licensee event reports and other studies of dependent failure events that show multiple component failures due to human-error-related causes almost always involve similar components. (One exception to this rule is the consideration of multiple component failure combinations if all of the components share a common emergency operating procedure.)

Screening steps that apply to the analysis of environmentally caused dependent failures include (1) screening based on component failure susceptibilities to environmental failure causes, (2) screening based on component locations, and (3) screening based on locations of sources of harsh environments.

Initial demonstration applications indicate that the use of these screening steps makes the qualitative portion of a dependent failure analysis tractable. Personnel at Sandia National I.aboratories have successfully used the SETS computer program to solve several example PRA accident sequences for dependent failure scenarios, and JBF Associates personnel have done the same using COMCAN 3.

Research is now focusing on evaluating and developing quantitative modeling and screening techniques for dependent failure analysis.

Techniques being evaluated include parametric models, such as the 8-f actor Model and the BFR Model, and specialized models, such as the Stress-strength Model and the Maintenance Personnel Performance Simulation Model.

The final product of this project will be a procedures guide for performing dependent failure analyses. The guide will be published early in 1986.

i l

1 I

l 1

16-2

l PRA PROC"DURES FOR DEPENDENT EVENTS ANALYSIS - AN INDUSTRY PERSPECTIVE Karl N. Fleming Pickard, Lowe and Garrick, Inc.

I 2260 University Drive Newport Beach, California 92660

SUMMARY

I !

{

The purpose of this paper is to report on recent progress that has been made !

in the development of systematic procedures for the analysis of dependent events I 1

in PRAs. This work is sponsored by the Electric Power Research Institute and will l l produce a two-volume guidebook; one dealing with plant level, and the other with system level dependent events analysis. This guidebook, which is intended to l update the material on this subject in the IEEE/ANS PRA Procedures Guide (NUREG/

i CR-2300), is based on the procedures that were followed in a number of recent and l ongoing PRAs, particularly those on the Seaorook Station, Oconee, and LaSalle.

I There are several motivations for developing a guidebook of this kind. One is that dependent events analysis has been a major source of confusion, inconsistency, i and incompleteness in earlier PRA applications. Another is that much progress has

been made recently in the development of systematic procedures for the definition,

classification, and analysis of dependent events - at both the plant and system levels.

4 The most important motivation is that operating experiences and the results of completed PRAs consistently point to various types of dependent events as major contributors to risk.

1 Between the two volumes of the guidebook, relatively more progress has been made thus far on the systems level volume. The systematic procedure described in this volume consists of the following basic steps: system familiarization, logic model i development, Boolean analysis, algebraic model development, parameter estimation, j system quantification, and interpretation of results. This procedure is generally

{ applicable to a number of specialized CCF models such as the basic parameter (BP),

> multiple Greek letter (MGL), binomial failure rate (BFR), and other parameter models. 7 i Case studies are included to illustrate practical problems of application, j The plant level volume deals with PRA analyses of dependent events that i transcend system boundaries. A simplified nuclear power plant example is used to i

demonstrate how such plant level events as intersystem dependencies and common cause initiating events are identified and incorporated into a PRA model. Emphasis is placed on the use of plant models that facilitate an integrated analysis of internal j events, external events, and internal plant hazards (e.g. , fires and floods).

I l The final topic of the paper is a plan to integrate industry, NRC, and NRC j contractor efforts to achieve a consensus on how these problems should be tackled i in future PRAs.

i i,

l 16-3 l

i i

l l

l Integrating Root Causes Into PRAs L. C. Cadwallader, S. Z. Bruske and P. L. Stepina EG8G Idaho, Inc.

Idaho Falls, ID 83415 Dr. W. E. Vesely Battelle Columbus Laboratory Columbus, Ohio 43201 1

[ The performance of probabilistic risk assessments (PRAs) of nuclear i power plants requires the use of data on the failure of components that comprise safety and safety-support systems. Certain component i failures are part of dominant accident sequences. Determining the root causes of these important component failures is a necessary step in the process of evaluating and reducing nlant risk. An understanding

of the root causes of component failure can provide valuable insights in such areas as (a) PRAs, (b) reliability assurance, and (c) the application of PRAs to inspection activities.

i Root causes cover a wide range of activities and situations.

Components may fail due to errors in design,' manufacturing, installation, i test, maintenance, or repair. Operator errors due to inadequate procedures or training are also possible root causes as well as harsh environments that exceed design specifications. Root causes for certain component failures have been identified from existing failure data bases. All root causes are tied to component failures modes. By determining root cause importances, cut set basic events and root causes can be combined.

As a result the identification of root causes makes it possible to understand why the important component failures identified from PRAs are occurring. To demonstrate this process, the root cause categorization is discussed and an example of integrating root causes into PRAs is presented.

l 4

ll l

16-5

INTEGRATION OF NONDESTRUCTIVE EXAMINATION RELIABILITY AND FRACTURE MECHANICSa 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 Integration of Nondestructive Examination (NDE) Reliability and Fracture Mechanics (FM) Program at Pacific Northwest Laboratory (PNL) was established to determine the reliability of current ISI techniques and to develop recom-mendations to ASME Section XI that will ensure a suitably high inspection reliability. The objectives of this NRC program are to:

e determine the reliability of ultrasonic ISI performed on commercial light-water reactor (LWR) primary systems e using probabilistic FM analysis, determine the impact of NDE unreliability on system safety and determine the level of inspection reliability required to ensure a suitably low failure probability e evaluate the degree of reliability improvement that could be achieved using improved and advanced NDE techniques e 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 crogram consists of three basic tasks: a piping Task, a Pressure Vessel Task, and a Fracture Mechanics Task. Because of the problems 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 Piping Task and the Fracture Mechanics Task.

aFIN: B2289; NRC

Contact:

J. Muscara 17-1

i The Piping Task had a large number of activities in progress during the past year. A document containing the criteria and requirements for the qualification of personnel, equipment, and procedures for inservice inspection (ASME Se *. ion i XI) was developed and presented at a workshop to a large cross-section of

people from the nuclear industry. The recommendation of the workshop was to form an Ad Hoc Task Group to ASME Section XI. Thus, a major effort involved participation in an Ad Hoc Task Group commissioned to develop revised qualifi-cation and training requirements for inservice inspection. A pipe insoection

" mini round robin" was conducted to help quantify the effects of extensive training efforts that have evolved since 1982, to look at effects of crack i'

length on detection performance, and to quantify the performance of individuals versus teams. An international round robin on inspection of cast stainless steel was conducted as part of PISC II. This is the first phase of an. inter-national round robin study on pipe inspection capability and reliability that will be continued under PISC III. A preliminary report based on PNL and other reported work which evaluated the inspectability of weld joints repaired by the weld overlay process was completed. Other activities included work to evaluate advanced ultrasonic inspection systems, studies and modelling of ultrasonic

equipment interactions, and participation with international groups (CSNI Task Group on NDE Reliability, PISC II, and PISC III) to coordinate technical activities to complement rather than duplicate work.

The Fracture Mechanics Task focused efforts on developing probability risk assessment and value impact models for inservice inspection. These analyses j evaluated engineering traceoffs of effectiveness of NDE and inspection intervals

versus system safety.

A 4

4 i

i l

17-2

l l DEVELOPMENT AND VALIDATION OF A REAL-TIME SAFT-UT SYSTEM FOR INSERVICE INSPECTION OF LWRsa S. R. Doctor, Project Manager H. D. Collins, S. L. Crawford, T. i. Hall, L. D. Relo Pacific Northwest Laboratory Operated by Battelle

SUMMARY

A program for the development of the synthetic aperture focusing techn.ique for ultrasonic testing (SAFT-UT) has completed the third year. The program is designed to provide the engineering required to transfer the SAFT-UT tech-nology from the laboratory into the field. Specific program objectives are:

e Design, fabricate, and evaluate a real-time defect detection and charac-terization system based on SAFT-UT technology for preservice and inservice inspection of LWR components.

e Establish calibration and field test procedures.

e Demonstrate and validate the SAFT-UT system through actual field inspec-tions.

e Generate an engineering data base to support Code acceptance of the real-time SAFT-UT technique.

e Facilitate technology transfer to the commercial ISI community.

The program scope is defined as follows:

e Conduct laboratory tests to provide engineering data for defining SAFT-UT performance.

e Complete the development of a special-purpose SAFT processor to make SAFT-UT a real-time process for ISI applications.

e Fabricate and field test a fieldable real-time SAFT-UT system on nuclear reactor piping, nozzles, and pressure vessels.

e Encouraging ISI community interest in implementation of SAFT-UT technology in commercial applications.

The activities over the past year have dealt with implementing the SAFT-UT data acquisition system and integrating the complete field system. This includes modifying the pipe scanner to accommodate multiple SAFT-UT configura-aFIN: B2467; NRC

Contact:

J. Muscara 17-3

l 1

tions and reducing the amount of operator interaction required to position

! the scanner in a potentially harsh environment. Also work has been performed on reducing the amplifier noise level to improve signal characteristics. The modular SAFT-UT field software on the field host computer has been developed i so that simultaneous data acquisition, processing and display has been achieved. An extremely versatile graphics package has been completed that provides extensive and expedient image analysis under operator control. A l major effort of this past year has been to generate a field usable fully imple-mented real-time SAFT-UT system.

Development of a real-time SAFT processor has been undertaken within PNL that will enable the field system to display the final processed UT image while the specimen is being scanned on-site. This is a peripheral device to

the SAFT field host computer that performs the intensive computations involved

in the SAFT algorithm. Laboratory testing of this device is to be completed j by the end of this fiscal year.

j Other activities this year included initiating the technology transfer effort by locating companies interested in implementing SAFT-UT in commercial ISI applications. In light of this, a cooperative demonstration with Combustion Engineering equipment and the SAFT-VT equipment was held at the Commonwealth Edison facilities. This effort proved to be very successful in demonstrating SAFT-UT to an ISI end user. Also this was another milestone in the SAFT-UT ,

j program in that the host field computer was transported to the field thus j allowing SAFT processing and image display to be performed on-site. This l factor came to be very effective in presenting the SAFT-UT system as a viable j tool for the industry.

i 2

l i

i l

1 17-4

NDE OF STAINLESS STEEL AND ON-LINE LEAK MONITORING OF LWRs*

D. S. Kupperman (MST) and T. N. Claytor (CT)

Argonne National Laboratory Argonne, Illinois 60439 Introduction The objectives of the ultrasonic NDE program are to (a) assess methods for characterizing the microstructure of cast stainless steel (CSS) and determine ISI reliability, and (b) evaluate ultrasonic inspection problems associated with weld overlays, including problems in trying to distinguish intergranular cracks from geometrical reflectors.

The objectives of the leak detection program are to (a) develop a facility to evaluate acoustic leak detection (ALD) systems quantitatively and (b) assess the effectiveness and reliability of field-implementable ALD systems.

Technical Progress A. Cast Stainless Steel Four 60-mm-thick CSS plates with microstructures ranging from equiaxed to primarily colu=nar grains have been examined with ultrasonic waves. We have found that the longitudinal velocity of sound and longitudinal-to-shear velocity ratio as a function of position can be used to characterize the crystallographic texture. We have also found that the beam-skewing phenomenon present in columnar (but not equiaxed) structures is strong enough so that measurements of probe separation at maximum received signal intensity for 45* shear-wave pitch-catch transducers can be correlated with micro-structure.

Two 45' shear-wave transducers used in a pulse-echo mode and one 45* longitudinal-wave transducer used in a pitch-catch mode were compared for their effectiveness in detecting notches in CSS. The shear probes were manufactured by Magnaflux (300 kHz) and Panametrics (500 kHz). The longi-tudinal probe was an RTD 45*, 1-MHz, side-by-side pitch-catch transducer.

41 The two shear-wave transducers gave virtually identical signal-to-noise (S/N) i ratios for detection of the corner of a 60-mm-thick wrought stainless steel block. The response from a 50%-deep notch in a 67-mm-thick cast block with equiaxed grains, 1-5 mm across, was 6 dB greater for the 300-kHz shear-wave probe (when compared to 500 kHz) in both amplitude and S/N. The 45* longi-tudinal pitch-catch probe could also be used to detect the 50% notch, but the S/N and signal amplitude were considerably lower than for the shear-wave probes. Notches 50% deep in material with larger grains (5-15 mm across) could not be detected with any of the probes.

A laser interferometer has been used successfully to map beam dis-tortion in a piece of CSS 19 mm thick. The profile of the beam af ter propa-gating parallel and at 45' to the columnar grain axes is qualitatively con-sistent with model predictions. A quantitative analysis is being carried out

)

to assess the validity of the transverse isotropic model in predicting wave distortion.

Work supported by the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research.

1 4

I 17-5

B. Pipes with Weld overlays An NDE workshop on inspection of 12-in. Schedule 80 pipe-to-elbow weldments with weld overlays, from the Georgia Power Co. Hatch-2 reactor, was held at ANL in May 1984. A second workshop on NDE of endcaps with weld over-lays, also from Hatch, was held at ANL in January 1985. Because of the limited amount of cracking, the emphasis was on trying to understand the nature of crack overcalling. We concluded that it is extremely difficult I to inspect overlayed pipes by traditional shear-wave methods because of the unpredictable beam distortion due to the overlay and the absence of effective reference pipes. Improvements may be possible through the use of 1-MHz longitudinal angle-beam probes propagating through the weld.

C. Acoustic Leak Detection Acoustic leak data have been acquired from a third field-induced IGSCC. The data in the 300-400 kHz region are comparable to the data acquired

~

from two other ICSCCs. This result supports the argument that, at least for low flow rates, flow-rate information can be obtained from acoustic signal amplitude data despite variations in crack geometry. Acoustic leak data have also been acquired from a 2-in. valve with a leak in the stem. As expected, the magnitude of the acoustic signal is lower than for IGSCCs with comparable flow rates. Furthermore, the frequency spectrum for the valve is significantly different than for IGSCCs but comparable to that of fatigue cracks. This result suggests that frequency analysis of acoustic leak signals may be useful in discriminating IGSCC leaks from other types of leaks.

A laboratory test has been carried out to help evaluate the capa-bility of the GARD /ANL digital continuous acoustic monitoring system to locate a leaking field-induced IGSCC by averaging cross-correlation functions. Two AET 375-kHz receivera were placed on waveguides; one 61 cm, the other 103 cm from ICSCC #1. A 0.003-gal / min leak was generated from IGSCC #1 at 504*F and 1000 psi. With the flow off and electronic filters passing 150- to 500-kHz signals, the electronic background noise levels were 31 and 42 mV. With the flow on, the signal amplitudes increased to 51 and 68 taV. The sampling rate for these tests was 500 kHz (2 ps between data points). Nine correlograms were averaged. In generating these correlograms, one of the two waveguides was moved circumferentially (about 60*) before the next waveform was captured.

This averaging technique permitted a leaking field-induced IGSCC to be lo-cated, for the first time, by cross-correlation techniques. The location accuracy of the system, however, has yet to be determined. Similar tests

carried out with an electronic leak signal indicated that location accuracy improves with signal amplitude. This result suggests that larger leaks would be located with greater reliability. Only three dif ferent positions at each

, waveguide location were required to acquire the nine correlograms. Thus, the l averaging procedure could be carried out with three transducer /waveguide

systems at each monitoring site. It may be possible to carry out cross-correlation analysis with one transducer at each monitoring site, if larger leaks are prasent. Future tests will address this point.

Three waveguides with AET 375-kHz acoustic emission transducers were attached to check valves at the ETEC valve leak facility, Canoga Park, CA. Initial tests allowed acquisition of background (electrical and acoustic) noise data and provided an opportunity to check the compatibility of ETEC tapes and ANL signal analyzing equipment.

17-6

PROGRESS FOR ON-LINE AC0USTIC EMISSION MONITORING OF CRACKS IN REACTOR SYSTEMS P. H. Hutton R. J. Kurtz Pacific Northwest Laboratory I Operated by Battelle Memorial Institute l

SUMMARY

The objective of the acoustic emission (AE) monitoring program is to develop ,

and validate the use of AE methods for continuous surveillance of reactor pressure boundaries to detect flaw growth. Benefits expected from the program include:

]. e Increased assurance of physical integrity of the pressure system under both normal and abnormal operating conditions by early detection of crack-ing.

e Facilitate longer life allowance for weld overlay pipe repairs by providing an effective method of detecting crack growth under the weld overlay, o Provide the capability for early detection and characterization of coolant leaks. This is in conjunction with a companion program at Argonne National Laboratory.

To achieve the objective and associated benefits, the program follows the format of developing basic AE/ crack growth relationships in the laboratory, evaluating and improving these through testing an intermediate scale pressure vessel, and finally, demonstrating the technology by performing on-reactor monitoring. Initial laboratory development and evaluation of an intermediate scale vessel have been completed. Preservice testing on a new reactor has been monitored and preparations are completed for in-service monitoring of the same reactor.

Specific accomplishments in FY85 in support of achieving the program objectives include:

e Preparation for on-line monitoring at Watts Bar Unit I have been completed.

e An improved AE signal identification method has been developed and imple-mented as part of the prototype AE instrument being installed at Watts Bar Unit 1.

e Arrangements are in progress to demonstrate AE monitoring to assure con-tinued integrity of weld overlay pipe repairs.

e A laboratory test is being performed to investigate the effect of very 1 low crack growth rate on detec Crack growth ratesdowntoatleast1x10-yionofthecrackingbyAE. inches /second are being considered, 17-7

e An ASTM Standard for continuous AE monitoring of metal pressure systems has been submitted for ballot. Action has been started to work with ASME Code Section XI to achieve Code acceptance of continuous AE monitor-ing.

The end product of the program is expected to be demonstrated technology for l continuous AE monitoring of reactor pressure systems with supporting engineering I field data and the Ccde and regulatory documentation required for administration of technology application.

l l

17-8

l i

! IMPROVED EDDY-CURRENT TESTING OF STEAM GENERATOR TUBING

C. V. Dodd Metals and Ceramics Division OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee 37831

SUMMARY

The Oak Ridge National Laboratory multiple-f requency eddy-current system was prepared for field testing at the steam generator test facility at the Pacific Northwest Laboratory (PNL). The field test was originally planned to emphasize array coils, but emphasis was changed to absolute circumferential coils due to the greater speed and larger number of tubes to be tested. As a result, both types of coils were evaluated.

The circuitry driving the two types of coils was changed to reduce the effects of cable capacitance and resistance. By placing a voltage dropping resistor in the probe head, we have improved the signal-to-noise ratio, the sensitivity, and the frequency response of the probes.

The field test on the Surry steam generator at PNL was performed, and good data were obtained. The multiplexing circuitry in the probe head of the array probe worked well in the radiation field. Tube denting in the Surry generator was greater than anticipated, and the magnetite deposits on the Surry tubes had a larger permeability than the material used in our training standards. This gave errors in the on-line data reduction, which will be corrected when the system is recalibrated using a new standard. The different probes gave similar readings, demonstrating that the calibration readings performed on one probe can be used for another probe of the same design.

Research sponsored by the Of fice 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 Martin Marietta Energy Systems, Inc.

17-9

Heasurement and Prediction of Microstructural Development in Stainless Steel Pipe Weldments c

S. H. Bruemer and D. G. Atteridge Pacific Northwest Laboratory Microstructural development including material degree of sensitization has been quantitatively measured and modeled as a function of various thermal and thermomechanical treatments. The effects of material bulk composition and initial condition have also been examined. Comparisons will be made between measurement and prediction af ter both simple and complex thermo-mechanical treatments.

Detailed thermomechanical history measurements have been completed on an instrumented 24-in.-dia. , schedule 80, pipe weld. These measurements of temperature and strain were conducted dynamically during each pass at inner diameter surface locations to map the weld heat-affected zone. Degree of sensitization measurements were also obtained across the heat-affected

zones using the electrochemical potentiokinetic reactivation (EPR) test technique.

Pass-by-pass thermomechanical ' istories as a function of heat-affected zone location were then input into the microstructural development model and sensitization predicted.

Simultaneous deformation was found to have a significant effect on sensiti-zation development in laboratory tensile tests and in the weld heat-affected zone. Deformations of about 3-5% plastic strain promoted sensitization at shorter times and to higher values during an isothermal exposure at 600 C. These levels of plastic strain were similar to those neasured in the heat-affected zone during the initial passes of the instrumented 24-in.-dia. pipe weld. Cumulative measurements indicated that the heat-affected zone underwent compressive and tensile deformations totalling more than 75% plastic strain. The potential effects of this i hot / cold work on microstructural development and SCC susceptibility of weldments will be discussed.

j

-1 18-1

= _ _ _ - __- -- -

BWR Pipe Cracking and Weld Clad Overlay Studies W. J. Shack, T. F. Kassner, P. S. Maiya, J. Y. Park, and W. Ruther Materials Science and Technology Division Argonne National Laboratory Argonne, Illinois 60439 l

l Introduction I

' Leaks and cracks in the heat-affected zones (RAZs) of weldments in austenitic stainless steel piping and associated components in boiling water reactors (BWRs) have been observed since the mid-1960s. Proposed remedies include (1) procedures to produce a more favorable residual-stress state in the l HAZ adjacent to welds, (2) replacement of the piping with materials that are more resistant to stress corrosion cracking, and (3) modification of the reactor coolant environment to decrease the susceptibility to cracking. During this i year, experimental and analytical studies of residual stresses, the behavior of

Type 316NG stainless steel in impurity environments, and the effect of the reactor coolant on the cracking of sensitized Type 304 stainless steel have been carried out. The results have important implications for all three classes of remedies.

Technical Progress Finite-element calculations were performed to validate the elastic super-position approach used in most fracture-mechanics analyses of weld overlays with cracks. Compared with complete elastic-plastic solutions, elastic superposition gives unconservative estimates of stress intensity f actors. However, for typical applied loads the stress intensity factors are negative even for large cracks, and positive stress intensity factors are obtained only for large, deep cracks under high axial loads.

Surface residual-stress measurements have been made on an pipe-to-elbow weld overlay and a recirculation header-endcap overlay from the Hatch-2 reactor.

The stresses on the inner surfaces of these weldments were compressive, although not as compressive as those measured on a mockup pipe-to-pipe weldment prepared with nominally identical procedures. The stress distributions for the pipe-to-elbow overlay are strongly nonaxisymmetric, in contrast to those of the mockup weldments. The distributions in the header-endcap weldment were nearly axi-symmetric, and the stress distributions and magnitudes approximated those ob-l tained from mockup weldments and axisymmetric finite-element analyses.

A detailed metallographic examination of the recirculation header-endcap weld overlay from Hatch-2 was completed. The conclusions are similar to those from the previous analysis of the pipe-to-elbow overlay. Crack blunting was observed, and there was no evidence of tearing or throughwall extension of the crack beyond the blunted region. However, only relatively few, short cracks were present in either weldment, although several of the cracks were quite deep (460% throughwall). In both weldments cracking occurred more extensively in the forged component (elbow or endcap) than in the pipe.

Our previous work has shown that Type 316NG stainless steel is susceptible to TGSCC in water with sulfate impurities at levels within Reg. Guide 1.56 specifications. Additional tests revealed some TGSCC at impurity levels corresponding to water conductivities of <0.5 US/cm. Observations from tests "RSR FIN Budget No. A2212; RSR

Contact:

A. Taboada.

18-3

interrupted at strains corresponding to <20% of life indicate that cracks initiate early in the tests and are not an artifact associated with the high strains reached during the final stcges of slow-strain-rate experiments. Pipe tests are under way to investigate susceptibility under more prototypic loading conditions.

A cooperative effort with the EPRI NDE Center is in progress to study the suitability of German " Nuclear Grade" Type 347 stainless steels for BWR primary system piping. The NDE Center will focus on the development and verification of welding practices, and ANL will evaluate the SCC resistance of the material.

Preliminary SCC tests have been carried out on one heat, which appears to be very resistant to cracking even in impurity environments. Additional heats have been obtained and are being welded at the NDE Center.

Additional data have been obtained which confirm the dependence of various SCC susceptibility parameters, such as the average crack growth rate i on the applied strain rate predicted by a phenomenological model based on estima,tes of crack-tip strain rates. The model has been extended to obtain a relationship between A and the average crack-tip strain rate, which is in good agreement with expe*lmental results for intergranular and transgranular cracking of Types 304 and 316 SS.

The effect of impurity elements (sulfur and phosphorus) on depletion of grain boundary chromium in Type 304 SS was investigated. Scanning-transmission electron microscopy (STEM) showed that phosphorus strongly promotes chromium depletion at low temperatures whereas sulfur does not.

Long-term fracture-mechanics crack growth tests were performed to quantify the influence of water chemistry on the rate and mode of crack growth in Type 304 SS. Crack growth in the sensitized materials ceased at low dissolved oxygen levels even in the presence of 0.1 and 1.0 ppm sulfate in the 289'C water. The data from the fracture-mechanics specimens are consistent with the more extensive results from slow-strain-rate experiments concerning the effects of dissolved oxygen and sulfate on the SCC behavior of the material in high-temperature water.

Irradiation-induced segregation of alloying or impurity elements can result in irradiation-assisted stress corrosion cracking (IASCC) of solution-annealed material. The major environmental parameter that controls IASCC is the open-circuit corrosion potential. The extent to which hydrogen-water chemistry suppresses r.adiolysis and alters the open-circuit corrosion potential of core materials and hence the SCC behavior of the irradiated material has not been determined. Since in-core experiments are complex, laboratory experiments are being performed to determine the effect of intense gamma radiation on the open-circuit corrosion potential. Preliminary results obtained with a 90-kCi cobalt source indicate that the gamma field degrades the Teflon tubing and seals used in our standard 0.lM KC1/AgC1/Ag reference electrode. Hence a modified reference electrode is being developed.

Additional experiments have been performed to examine the SCC susceptibility of Type 304 SS under torsion (Mode III) and tensile (Mode I) loading. SCC did not occur in either sharply notched or fatigue precracked specimens under Mode III loading in aggressive environments at 289'c for times much longer (by a factor of 10 to 100) than those required for failure under Mode I. This is consistent with a hydrogen embrittlement mechanism for crack advance. However, other evidence pertaining to the open-circuit corrosion potential of the steel and the influence of dissolved oxygen and several oxyanion species on the crack growth behavior suggests that the predominant cathodic reaction involves the reduction of these species and not hydrogen ion reduction and hydrogen absorption at the crack tip. At present, it is difficult to reconcile this information with the behavior in the Mode I/ Mode III tests.

l 18-4

i AGING DEGRADATION OF CAST STAINLESS STEEL *

0. K. Chopra and H. M. Chung Materials Science and Technology Division Argonne National Laboratory Argonne, Illinois 60439

SUMMARY

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 compo-nents 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 after 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 under-standing of the ef fects of key compositional and metallurgical variables on the kinetics and degree of embrittlement.

The relationship between aging time and temperature for onset of embrittlement will be determined by microstri:ctural examination and measure-ments of hardness, Charpy-impact and tensile strengths, and Jyc fracture toughness. Estimates of the degree of embrittlement obtained from laboratory-aged material will be compared with data for material from actual reactor service. The kinetics and fracture toughness data generated in this program and from other sources will provide the technical basis to define the aging histories, chemical compositions, and metallurgical structures that lead to significant embrittlement of cast stainless steels under LWR operating conditions.

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-duplex stainless steel. Measurements of the bulk chemical composition, hardness, ferrite content, and composition of the ferrite and austenite phases as well as a metallographic evaluation of the grain structure of the various unaged materials have 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 specimens have accumulated 14,000 h of thermal aging. Charpy-impact and tensile tests were conducted at room temperature on reactor-aged material and materials that were aged in the laboratory for up to 3000 h at 320, 350, 400, and 450'C. Initial data indicate that interstitial content, e.g., C and N, and cast structure influence the embrittlement

Work supported by the Office of Nuclear Regulatory Research, U. S. Nuclear Regulatory Commission.

18-5

behavior. Mechanical property data generated in this program, as well as from other sources, will be used to estimate the degree of embrittlement expected I at reactor operating temperatures. l

! Microstructural characterization was performed on three heats of laboratory-aged material obtained from Georg Fischer Co. in Switzerland and on reactor-aged material from the KRB reactor pump cover plate. The transmission i

electron microscopy (TEM) study revealed that all low-temperature aged mate-l rials contain two types of precipitates, designated Types M and X. Intensity spectra from small-angle neutron scattering (SANS) were also obtained from several batches of the Swiss material to determine the size and distribution of precipitates in the ferrite phase of specimens aged at 400'C for ~70,000 h.

The results are consistent with the TEM observations. Thus, the SANS technique ca'n be used to obtain quantitative data on the size and distribution

- of precipitates in aged duplex stainless steel. However, it does not discrim-1 -inate.between the different types of precipitates. The KRB material, in addition, contains o' precipitates and a grain boundary phase. The ferrite

phase exhibited cleavage fracture in Charpy test specimens from materials that 2

were aged for -6 yr at 300*C or ~1.2 yr at 400*C. The amount of cleavage fracture is greater than the ferrite content in the steel, indicating that the I

crack propagates preferentially through the ferrite phase.

j 1

f 4

r i

f f

i

+

3 h 5

\, s i

18-6 l

Models for BWR and PWR Engineered Safety Features in the CONTAIN Code

l F. J. Schelling, K. K. Murata, D. C. Williams, and P. E. Rexroth l Sandia National Laboratories Albuquerque, NM 87185 Summary Configured in a number of ways in both PWRs and BWRs, engineered safety features (ESFs) form a major part of the overall design of reactor containments. Reducing containment pressure and temperature loads and efficiently reducing concentrations of airborne aerosols are the chief functions of these systems during severe accidents. Mechanistic models of several types of complete ESF systems are available in the CONTAIN computer code, which is designed to be the NRC's principal best-estimate calculational tool for severe accident containment analysis. A unique feature of CONTAIN is its integrated treatment of the interactions between thermal-hydraulic, aerosol, and fission-product behavior. Because of their strong influence on containment conditions, these effects are particularly important for modeling ESF performance.

The effects of sprays, ice condensers, pressure suppression pools, and fan coolers on the containment environment can be modeled for a variety of accident sequences. The models developed for CONTAIN have been designed with considerable flexibility to enable the user to tailor the code for treating specific features of individual plant designs. In addition, arbi tra ry combinations of component parts, including tanks, pumps, pipes, valves, orifices, heat exchangers, mass and energy sources, and pool overflow can be utilized for modeling the distribution of liquid throughout containment and for more detailed modeling of specific systems.

An overview of available ESF models in CONTAIN is presented through a discussion of recent modeling results. These models are unique in that they are highly mechanistic in their treatment of physical processes.

Modeling of spray behavior, for example, includes a mechanistic treatment of the condensation process on a single droplet falling through an atmo-sphere, the size of the drop continually changing as mass and energy are transferred between the liquid drop and the atmosphere; the effects are then scaled to the total number of drops present. A similar treatment of stemn condensation on surfaces is also utilized in the mechanistic fan cooler and ice condenser models. Aerosol removal processes, including impaction, inte rce ption, sedimentation, diffusion, thermophoresis, and diffusiophoresis, are also modeled mechanistically for these systems.

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.

19-1

The containment spray model has recently been used in a number of detailed calculations of severe accident sequences. As part of a series of sensitivity studies using CONTAIN, processes occurring during TMLB' spray recovery sequences were analyzed in detail. In particular, the possibility of hydrogen burns occurring as a result of the spraya inducing de-inerting conditions, the magnitude of the aerosol decontamination factors produced by the sprays, and the interactions between sprays, aerosols, and burn behavior were studied. The results indicated, at least for default burn conditions, that while spray recovery can de-Inert an atmosphere and promote hydrogen burns, it also significantly reduces the temperatures and pressures which the burns produce. Decontamination factors resultiag from the capture of aerosol particles by falling spray droplets were found to be quite large, even for a period of time following a burn. Perhaps most significant, the analysis showed the luportance of the Integrated treatment of the complex and interrelated physical pro-cesses occurring. Aerosol removal rates considerably larger than might be expected from a consideration of simpler model) were observed as a result of this synergy.

A new feature of CONTAIN is a pressure suppression pool model, one of the major elements of the containment design for BWRs. Suppression pools are designed to greatly reduce temperatures and pressures in the event of a severe accident by efficiently condensing steam released from the primary system. They are also expected to provide a substantial barrier to aerosol release to the-upper containment by scrubbing aerosols passing through the wetwell pool. A discussion of some recent calculational results is presented which illustrates some of the features and limita-tions of the model. Used in conjunction with other E3F systems such as sprays, fan coolers, heat exchangers, and tanks, CONTAIN will have the new capability of reasonably complete modeling of BWR containment systems.

In the suppression pool model, vents open between the drywell and wetwell if a sufficient hydrostatic head develops; gas flows between the cells, condensing steam into the wetwell pool until the pressure equilibrates, after which the vents reclose. The wetwell pool itself utilizes.CONTAIN's pool model for adjusting thermal-hydraulic conditions between the pool and atmosphere, and for treating physical processes such as boiling. Flow can occur in either direction; for example, in a Mark III, hydrogen burns may cause vents to be opened from the wetwell side.

Overflow back into the drywell is then possible, as is unrestricted gas flow through vents left completely open by a depleted wetwell pool. Under 1 l such conditions, steam can pass directly between the wetwell and drywell without significant condensation occurring, i l An aerosol decontamination model, based on Fuch's treatment of aerosol

) deposition in rising bubbles as developed for the VANESA code, is included in the suppression pcol model. An initial rapid decontamination occurs during the turbulent condensation occurring at the pool entrance, followed by a mechanistic treatment of impaction, sedimentation, and diffusion in i growing bubbles. (Phoretic mechanisms during bubble rise are assumed to be inactive.)

19-2

Validation, Assessment, and Applications of the CONTAIN Computer Code

K. D. Bergeron, K. K. Murata, D. C. Williams, and J. L. Tills **

Sandia National Laboratories Albuquerque, NH 87185 l

l l

Summary The CONTAIN cocputer code is a versatile system of phenomenological models for analyzing the physical, chemical, and radiological conditions inside the containment building during severe reactor accidents. The code has been released to more than twenty research institutions in five countries, and is being used in a variety of ongoing research studies.

In this paper, we will describe recent accomplishments in three areas:

(1) Code Validation, which we define as comparison of code predictions with experimental results; (2) Code Assessment, which involves comparisons among computer codes and tests of the code's efficiency and speed for a variety of multi-cell problems (with and without the new implicit solution technique); and (3) Applications, which consist of a number of full-scale accident sequence calculations devised to study one or more aspects of

" integrated" containment analysis (i.e., the coupled treatment of thermal-hydraulic, aerosol, and fission product phenomena).

Validation CONTAIN has been used in two different code validation exercises. The first to be discussed is the ABCOVE aerosol experiment series at Hanford Engineering Development Laboratories (HEDL). Results for tests AB-6 and AB-7 were recently released by HEDL, along with the blind predictions of a number of aerosol codes. Both experiments were good tests of multi-component aerosol models, since they involved two aerosol components, NaI and NaOH, released at different times. The AB-7 test was run in order to resolve a number of problems in the AB-6 test which were identified after the tests were run. In general, agreement between CONTAIN and experi-mental data was excellent, except in the case of those aspects of the AB-t data which were suspect. Results of both experieents and the corresponding code predictions will be presented.

Another code validation exercise which was recently completed involved debris-concrete interaction experiments at Sandia. For this program, which was sponsored and organized by the USNRC, the latest version of CONTAIN with the CORCON-Mod 2 option was used. Results for test CC-2 showed that the feedback between above-pool containment phenomena and the

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.

- J. L. Tills and Associates, Inc., Albuquerque, hH.

1 19-3 i

f\

actual debris-concrete interactions could be significant. Being able to treat feedback effers with the combined codes is thus potentially important.

Assessment 1

CONTAIN has been used in an international aerosol code comparison exercise organized by Committee for the Safety of Nuclear Installations ,

(CSNI) of the OECD. The purpose is to illuminate the current state of I the art of aerosol modeling. CONTAIN results from sensitivity studies will be presented illustrating the importance of multi-component analysis, compared with single component models (e.g., NAUA). Comparisons between CONTAIN predictions for thermal-hydraulic conditions and those used for

, the CSNI code comparison exercise (based on the JERICII0 code) will also be discussed.

One of the most significant new developments for the CONTAIN code is the implementation of a new implicit numerical solution technf se for the intercell gas flow equations. A series of test calculations based on realistic simulations of actual plants will be presented. These multi-cell calculations indicate an increase in running speed by factors of 200-300, compared with the released version of the code, which utilizes an explicit solution technique, 1

Applications The unique capability of CONTAIN to simultaneously rodel thermal-hydraulics, aerosol physics, and fission product decay has been exploited in a series of sensitivity studies intended to explore the importance of feedback effects among these different areas of phenomenology. Two cal-culations will be presented as examples from this series. The first study is based on the TMLB' sequence at the Bellefonte plant; it illustrates the 1

importance of modeling mobile heat sources for accurate prediction of thermodynamic conditions throughout the containment. Since decay heat can be carried by aerosols as well as gases, and since the associated radioisotopes can be deposited on surfaces or collected by engineered safety features, an integrated analysis of aerosols and decay heating, as well as the mal-hydraulics, is needed.

i The second study considers the condensation on aerosols which occurs as a result of depressurization at the time of containment failure. It is well-known that rapid dapressurization of a nearly-saturated atmosphere l

will cause condensation of water onto nucleation sites (as in a Wilson l cloud chamber); the question investigated in these sensitivity studies is whether this process results in a significant reduction of the source term to the environment. The accident sequence considered is based on the TMLB' sequence at the -Zion plant. The most important conclusion of the analysis is that significant decontamination can take place due to condensation followed by agglomeration and settling, but that it takes l time for these processes to occur l reduced for a small hole (0.02 m2); hence

, but not the for source a largeterm holeis(0.8 significantly m2 ),

l 19-4

. n , -- - -- -

l AN INVESTIGATION OF STEAM EXPLOSION LOADINGS USING SIMMER-II by W. R. Bohl Los Alamos National Laboratory

SUMMARY

The purpose of this work was to provide a reasonable estimate of the maxi-mum loads that might be expected at the upper head of a pressurized water reactor (PWR) following an in-vessel steam explosion. These loads were deter-mined by parametric cases using a specially modified and calibrated version of the SIMMER-II computer code. Using the determined range of loads, the poten-tial for containment failure by missile production resulting from a steam ex-plosion, commonly called alpha-mode failure, was estimated assuming core melt.

The SIMMER-II code first was upgraded for this assessment. An improved equation of state was implemented to treat the corium-water system,and appro-priate nonequilibrium heat transfer models were formulated to treat the low thermal conductivity of liquid water. Model parameters then were calibrated to Sandia National Laboratories (SNL) experimental data. Both pre-explosion coarse-mixing calculations and explosion analyses were done. These analyses suggested parameters that would envelop the real behavior, although a unique fit to the data was not obtainable. To investigate a basis for judgments on the ability of SIMMER-II to calculate post-explosion slug breakup, a series of shallow pool acceleration experiments was conducted and then analyzed using SIMMER-II. Correct global behavicr was calculated, and the exercise suggested code parameters that should be conservative for calculating the reactor case.

Finally, a lower vessel head failure and motion model was developed and added to SIMMER-II. Only energetic steam explosions are a concern in evaluating con-tainment failure. These energetic explosions could cause a rupture in the lower vessel head area that could be expected to mitigate upwardly directed kinetic energy significantly.

Five major types of parametric steam explosion cases were analyzed.

First, several cases were calculated similarly to those in a previous SIMMER-II study 2 in which 20'/, of the total corium was mixed with water. These indi-cated a substantial reduction in the likelihood of energetic missile production with the code modifications, particularly with the lower head failure model.

Second, more mechanistic (but arguably conservative) cases were calculated starting from initially separated corium and water. The maximum upwardly di-rected fluid kinetic energy achieved in these cases was ~500 MJ about the 19-5

time of head impact, with a peak upper head force of 0.81 GN. Third, worst-case assumptions from a recent SNL study 2 were used; in the study, 94 000 kg of corium was mixed with 20 000 kg of water. Heat transfer from solid partic-ulate corium to steam was limited, and a diffuse spray led to a peak upper head force of 0.78 GN. Fourth, the calculated coarse-mixing situation from the second configuration was homogenized and used with degraded heat transfer to simulate a multiple explosion environment. In this situation, lower head fail- ,

ure was not calculated to occur, and the peak upper head force was 1.52 GN.

Fifth, the premixture from the third configuration was used with corium and water thermally equilibrated before the expansion. The peak upper head force was now 12.4 GN.

A rough analysis suggests that for the vessel head to become a large missile, a loading of approximately 1 GN is required under two-phase impact situations. After dismissing the fifth situation as impossible, only the multiple explosion simulation exceeded this threshold. However, true multiple explosions were not calculated. The initial conditions were idealized (for 4

example, with uniform material distributions radially), dissipation of energy in the upper core structure was neglected (with only 765 MJ of upwardly-directed fluid kinetic energy this could be significant), and the time-dependence of the nonuniform upper head loading suggested formation of multiple missiles from the head apex that could be stopped by the missile shields.

To estimate a limit on the conditional probability of containment failure given core melt, an accident progression diagram was prepared and evaluated using an approach from Theofanous and Bell.a The present technology base, the SIMMER-II cases, and the known SIMMER-II biases influenced the judgment made. Because of the edge-of-spectrum character of both the calculative as-sumptions leading to a large-scale stream explosion and those required to obtain significant head loading, this study suggests an upper limit for the containment failure probability given core melt of 10-2 if the vessel upper head and bolts are near nominal operating temperatures. Suggestions on future research to obtain increased confidence on steam explosion issues were formu-lated. However, it was noted that, until a judgment is made on what level of residual uncertainty is tolerable, the steam explosion issue will tend to remain open-ended.

REFERENCE

1. M. G. Stevenson, " Report of the Zion / Indian Point Study, Volume II,"

Los Alamos National Laboratory report LA-8306-MS, NUREG/CR-1411 (April 1980).

2. M. Berman, D. V. Swenson, and A. J. Wickett, "An Uncertainty Study of PWR Steam Explosions," Sandia National Laboratories report SAND-83-1438, NUREG/CR-3369 (May 1984).

3. T. G. Theofanous and C. R. Bell, "An Assessment of CRBR Core Dis-ruptive Accident Energetics," Los Alamos National Laboratory report l LA-9716-MS, NUREG/CR-3224 (March 1984).

1 1

19-6

SHORT TERM AND LONGTERM ASPECTS OF RECENT HDR CONTAINHENT TESTS f

L. WOLF *, L. VALENCIA, K.-H. SCHOLL Project HDR, Kernforschungszentrum Karlsruhe GmbH, FRG t

Battelle-Institut e.V., Frankfurt, FRG This presentation summarizes new experimental and analytical

~

results of a series of additional HDR-containment experiments performed during Phase II of the HDR Safety Program (1984 - 1987) thus far.

The new test series of simulated steam blowdowns, termed T31.1 through T31.1, results from the still open questions raised from the integral HDR-containment experiments V42 - V45 (steam blow-downs) and V21.1 and V21.3 (water blowdowns) which were performed during Phase I (1982 - 1983) of the HDR Program, results of which have been reported in detail at the 11th WRSIN.

In the test T31.-3 the flow field was varied between the break location and the overflow vent by different inclinations of the jet impingement plate in order to study the influence of flow around existing equipment and room contours, the influence of the flow directed towards the overflow vent, and the influence of the fraction of suspended water droplets on the differential pressure between neighboring subcompartments. All other tests boundary con-ditons, especia21y the break mass and energy flows, are identical.

In order to relate the news tests to previous ones. T31.1 with the jet impingement plate perpendicular to the flow was designed such as to be about si0ilar to V45 with the exception of the overall vent cross-section out of the break subcompartment which was only one-half that of V45.

Worthmentioning is that besides the extensive instrumentation of compartments and vents to follow major changes of the containment atmosphere along the flow path, first use of a laser velocity mea-suring equipment for two-dimensional recording of local velocity components was sucessfully employed. Furthermore, the containment spray system was intermittently operated. The resultant data are to be compared with those obtained from continuous spray operation during V45.

The first assessment of the data indicates that the flows around 19-7

internal equipment and following the specific break subcompartment contours respectively have obviously greater effects upon the differential pressure than the flows directed towards the overflow j vent. Substantial differences in the local pressures within the j break subcompartment develop during the short-term transient (0-3 sec). These data set can be used for verifying 3-D code features of sophisticated best-estimate codes such as COBRA-NC and the re-sults of preliminary multidimensional predictions will be presen-ted.

Surprisingly, during 31.3, where the flow was directed into the corner of the subcompartment, a brief period of superheating was recorded shortly after blowdown initiation. Moreover, the atmo-spheres in all other compartments of the containment substantially changed even at later times (100 - 200 sec) depending upon the flow details in the break subcompartment.

! Results of the comparisons between the data and blind pre-test predictions by the codes: COFLOW, CONDRU (GRS); DDIF, COCO (KWU),

GRUYER 2 (CEA) and COBRA-NC (PNL) for the short and long-term containment responses will be presented.

The results of intermittent on-off operation of the spray system upon the dynamic compartment responses at different axial planes 1

in the facility will be compared to the results obtained by e

continuous spray operation.

By recognizing the present need for long-term (hours and days)

containment data, an additional small-break blowdown (T30.4) into i

the HDR-containment was performed in the aftermath of the RPV-I Main Test Series. This steam blowdown lasted for approximately 1 hour1.157407e-5 days 2.777778e-4 hours 1.653439e-6 weeks 3.805e-7 months and resulted in substantial temperature gradient over the 60m height of the HDR-containment, measureo results of which will be reported for a total of 20 hours2.314815e-4 days 0.00556 hours 3.306878e-5 weeks 7.61e-6 months in detail.

A similar test planned for November 1985 will be briefly outli-ned.

I 19-8

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

HMS-BURN: A MODEL FOR HYDROGEN DISTRIBUTION AND COMBUSTION IN NUCLEAR REACTOR CONTAINMENTS J. R. Travis Theoretical Division, Croup T-3 University of California Los Alamos National Laboratory Los Alamos, New Mexico 87545

SUMMARY

During and after a loss-of-coolant accident in a light-water reactor, water may be decomposed by chemical reactions and radiolysis to release gase-ous hydrogen. Should hydrogen be released, two deleterious effects could occur. The noncondensable gas can increase the containment pressure, and in sufficient amounts, the hydrogen could burn in the presence of air, causing considerable heat flux loads on the containment walls or crucial control de-vices. Each effect represents an additional safety risk. To better assess the problem, we have developed a multi-dimensional fluid dynamics finite-difference code to calculate the details of hydrogen transport and possibly combustion through containment structures.

This detailed model of the full three-dimensional time-dependent Navier-Stokes equations with multiple species transport coupled to the global chemi-cal kinetics of hydrogen combustion is solved by a variant of the Implicit Continuous-fluid Eulerian (ICE)1 technique. We make use of the idea that for low Mach number flows, such as dif fusion flames and low-speed deflagrations, pressure wave propagation need not be resolved in detail, and therefore, the local fluid density maybe expressed as a function of the average fluid pres-sure, local temperature, and relative concentrations of available species.

This formulation further allows accurate representation of flows driven by large density variations due to temperature and concentration gradients for which the classical Boussinesq approximation may not provide sufficient accur-acy.2 In addition, a transport equation for the subgrid scale turbulent kinetic energy density is solved to produce the time and space dependent tur-bulent transport coefficients. The heat and mass transfer coef ficients gov-erning the exchange of heat and condensation of steam between fluid computa-tional cells adjacent to wall cells is calculated by a modified Reynolds anal-ogy formulation.

Analyses have been performed on a MARK-III type containment where the formation of dif fusion flames above the release areas in the suppression pool are established and on a large dry PWR type containment to determine, in the absence of combustion, hydrogen concentrations as functions of space and time.

REFERENCES

1. F. H. Harlow and A. A. Amsden, " Numerical Fluid Dynamics Calculation Method for All Flow Speeds," J. Comput. Phys. 8, 197 (1971).

2. C. K. Forester and A. F. Emery, "A Computational Method for Low Mach Number Unsteady Compressible Free Convective Flows," J. Comput. Phys. 10, 487 (1972).

19-9

i l

j llECTR Development and Assessment i

C. Channy Wong Sandia National Laboratories, Albuquerque, New Mexico 87185

! Since the accident at Three Mile Island (TMI), there has been a great deal of interest regarding the problem of hydrogen production and combustion in light water reactors (LnRs). Sandia National Laboratories is engaged in an i NRC-sponsored program to study hydrogen phenomenology in LWRs. As part of

that effort, we have developed a computer program, HECTR (Hydrogen Event

Containment Transient Response) to predict the transient pressure and temperature Tesponses within reactor containments for hypothetical accidents involving the transport and combustion of hydrogen.

! Capabilities of HECTR Version 1.5 HECTR is a relatively fast-running, lumped-volume containment analysis code developed for calculating the containment atmosphere pressure-temperature response with respect to hydrogen combustion. Six gases are modeled: steam, nitrogen, oxygen, hydrogen, carbon monoxide and carbon dioxide. The compartment conditions and junction flow rates during the transient are determined by solving a set of ordinary differential equations for conservation of mass, momentum and energy. The thermal response of surfaces and equipment in the containment is calculated, using either one-dimensional finite difference slabs or lumped masses. Most of the important phenomena occurring in reactor containments, as well as most engineered safety features (ESFs), are modeled. The phenomena included are hydrogen combustion, radiative heat transfer, convective heat transfer and steam condensation or evaporation.

The ESFs modeled are containment sprays, fans, ice condensers, sumps, suppression pools, fan coolers and heat exchangers.

Model Development - Diffusion Flame Modeling During a degraded core accident, hydrogen generated from metal-water reaction or radiolysis can combust as a standing flame near the point of release of the hydrogen-steam mixture into the containment. The primary-threats from this diffusion flame are the high thermal loads imposed by the flame on safety-related equipment and containment penetration seals, rather than the mechanical loading of the containment. A diffusion flame can be characterized by its basic flow field as either a momentum dominated jet-like flame or a buoyancy dominated plume-like fire. For example, hydrogen release during a degraded core accident through a stuck-open valve (2-10 cm) of a PWR will result in a momentum-dominated jet of hydrogen and steam, driven initially by very high pressure (1000-2000 psia) and expanding into a containment at or near atmosphere pressure. On the other hand, hydrogen release into a BWR containment will come from spargers in the suppression pool. These flows will be at low pressure, spread over a relatively large area (3 m. diameter) and will have a small steam content relative to the PWR cases.

The basic flow is buoyancy dominated.

19-11

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

I Work on the development of a diffusion flame model for the HECTR code is

under way. Two different approaches are being evaluated which will be assessed using data from the NTS continuous injection experiments. The first

model utilizes empirical correlations to characterize the diffusion flame and also the flow induced by the flame. These empirical correlations are based

, upon intermediate-scale experiments. The second model is a lumped parameter 2 approach where a built-in correlation generates " flame," " plume,' and "down-chimney" compartments. Combustion takes place in the flame compartment. HECTR calculates the gas pressure and temperature with respect to hydrogen combustion. A separate correlation computes the flow rate between the flame l and plume compartment. The major dif ferences between these two approaches are that in the lumped parameter approach the standing flame is modeled in more detail and the characteristics of the flame are calculated by HECTR rather than predicted by empirical formulas.

Code Assessment - HTS Calculations Although HECTR was designed primarily to investigate hydrogen combustion phenomena in LWRs, it may also be used to analyze hydrogen transport and

- combustion experiments. It.is in this manner that new models are developed and assessed. In particular, the results from the large-scale test program sponsored by the Electric Power Research Institute (EPRI) at the Nevada Test Site (NTS) have been used in this effort.

The premixed hydrogen combustion experiments at HTS were performed to study combustion processes in a large-scale vessel and to evaluate associated safety-related equipment response to the resulting thermal environment. The experiments were conducted in a 2048 cubic meter spherical vessel (hydrogen dewar) with mixtures of hydrogen, steam, and air ignited by glow plugs or heated resistance coils. Hydrogen concentrations ranged from.5 to 13% (by

+

volume) and steam concentrations from 4 to 40%. Several tests also i

incorporated spray systems and/or fans which enhanced the combustion rate and significantly altered the post-combustion gas cooling rate.

HECTR calculations of twenty-one NTS premixed tests were performed. In these calculations, the NTS dewar was modeled as one compartment with two major surfaces: the dewar wall and the water pool at the bottom of the vessel.

For the cases with lean hydrogen combustion (i.e., hydrogen concentration less than 8%), HECTR generally overpredicts both the peak gas pressures and temperatures. This is due to an overprediction of the combustion completeness together with an underprediction of the burn time. The faster combustion rate limits the heat transfer from the gas to the vessel, thus retaining more i energy in the gas which results in a higher than measured gas pressure and temperature. For those tests with high concentrations of steam, HECTR underpredicts the flame speed, which leads to a longer burn time and lower i peak' pressure. A modified HECTR combustion completeness correlation based on an expanded data set including NTS, FITS and VGES data has been developed.and incorporated in HECTR version 1.5. Recalculations of the same NTS tests with the new correlations show much better agreement with the experimental data.

19-12 i

( _ ~ _ _ _ __ _ . . _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ . _ , _ _ __ __ _ _ . _

[

l l

CONCHAS-SPRAY MODELING OF FLAME ACCELERATION AND AIR FLOWS

K. D. Marx Sandia National Laboratories Livermore, CA 94550 Sununary The Conchas-Spray computer code has been used to simulate two very different pro-cesses which are ofinterest in the field of hydrogen controlin nuclear reactors. The first of these is the acceleration of flames in confined geometries. The second is the generation of air flow due to activation of water spray systems in reactor containments. In this paper, recent results from both types of calculation will be presented.

Flame acceleration is the process in which a deflagration, or slowly burning flame, increases the rate at which it consumes fuel and oxidizer, and thereby increases its prop-agation velocity. The present work deals with the simulation of flame acceleration in small-scale experiments performed at McGill University. The experiments consist of the propagation of premixed hydrogen-air flames in tubes with obstacles which disrupt the gas flow. The major emphasis in the computations is placed on the comparison of two combustion models and a discussion of their ability to simulate the formation of a quasi-steady propagation mode. The flow is highly turbulent; a transport model is included to describe turbulence effects. Experimental data at one concentration are used to fix certain constants in the combustion models. The resulting mode!s are then employed in computations at different concentrations. The results are compared with the experimental parametric behavior.

The flow of air in reactor containments due to the introduction of water sprays is of interest because of potential effects on hydrogen igniter systems. The particular contain-ment geometry studied here is that of Sequoyah and Catawba, pressurized-water reactors equipped with ice condenser containments. The calculations employ the spray option avail-able in the code, which accounts for the coupling of the Navier-Stokes equations for fluid flow with the equations of motion of water droplets in two dimensions. An approximate theoretical description of some aspects of the flow field is derived. The dependence of the flow velocities on such parameters as spray flux, spray ring location, droplet size, spray in-jection characteristics, air fans, and the turbulence model is described. For typical values of water spray flux, the calculations indicate that peak flow velocities of approximately 12-14 m/s are possible in an empty containment building shell. This is consistent with o')servations made in tests of a spray system. Simulations which are more representative of complete reactors have been carried out by including approximations to the ice condenser and steam generator enclosures in the containment model. In that case, it is found that the peak flow velocities are quite sensitive to the relative locations of the steam generator wall and the spray headers. However, the effect of the internal structures is generally to lower the flow velocities over large regions of the building. Implications for igniter performance are briefly discussed.

This work was performed at the Combustion Research Facility and supported by the U. S. Nuclear Regulatory Commission under Memorandum of Understanding DOE 40-550-75, NRC FIN No. A1246.

19-13

SAND 85-1793A I

Flame Acceleration and Detonation Research' W. P. Sherman, S. R. Tieszen, and W. B. Benedick I i

! Sandia National Laboratories,

Albuquerque, N. M.

Two large experimental hydrogen combustion facilities, FLAME and the Heated Detonation Tube (HDT), built at Sandia National Laboratories in Albuquerque, have been operational for two years. Data obtained from the two facilities are used to address the possibility of detonation in reactor ,

containment. The HDT addresses the question of prm 2;auion (independent of i ignition source); the FLAME facility addresses the q2estion of transition to detonation. This paper summarizes.the work done, with emphasis on the j results obtained since the last information meeting.

The HDT is a 13.1-m long by 0.43-m diameter heated tube with associated piping and heating hardware in which H -Air-Steam 2 mixtures are detonated.

Tests have been completed for mixtures with an air density corresponding to i

1 atm and 20*C (i.e., the air in the plant prior to an accident). This mixture is heated to 100*C, and H2 and steam are added to simulate an accident. Mixtures with up to 30% steam have been detonated.

j The results show that steam dilution greatly desensitizes the mixture.

For 10%, 20% and 30% steam dilution of stoichiometric mixtures, the energy

required to detonate from a point source increases by factors of 220, 27000, and 216000, respectively. The desensitizing effect of steam is partly offset by the sensitizing effect of the increase in temperature and pressure from steam and H2 addition. The H2 -Air-Steam results can be applied directly to the propagation question through existing correlations between detonation cell width (which is measured in the HDT) and geometric propagation criteria.

i FLAME is a large horizontal "U" shaped channel, 30.5 m long, 2.4 m high and 1.8 m wide, made of heavily reinforced concrete. It was designed to study hydrogen combustion problems relevant to nuclear reactor safety, for

, example, flame acceleration, transition to detonation, simulation of com-bustion in reactor containment geometries, etc. Various amounts of transverse venting are achieved by moving steel plates on the top. Obstacles can be attached to the walls and floor. Ignition is made at the closed end by a single point bridgewire igniter (weak ignition source). In each test, approximately one hundred data channels measure flame position and pressure.

1

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 No. DE-AC04-76DP00789.

4 19-15 l 4

Three series of tests without obstacles in the channel were completed, tests with no top venting, 50% top venting and 13% top venting. The first l tests with obstacles are now under way. Plywood sheets are attached symmetrically on two walls, obstructing one third of the cross section.

Future tests are planned to model the geometry of the ice condenser upper plenum more closely.

> In tests with 50% top venting and no obstacles, the overpressures were i

low. There was no flame acceleration, and no transition to detonation. In tests with 0% top venting and no obstacles, the overpressures were much higher. There was flame acceleration, and transition to detonation.

Transition was observed near the channel exit in tests with 24.7% and 30.0%

l hydrogen. In tests with 13% top venting, no obstacles, and lean H2 -Air mixtures, the results were intermediate between those of 50% and 0% top venting tests. For mixtures above about 18% hydrogen, the flame acceleration and overpressure with 13% top venting were larger than with 0% top venting.

A transition to detonation was observed one third the distance down the channel in a test with 24.8% hydrogen.

Results with obstacles to date show much higher flame speeds and overpressures than in comparable tests without obstacles. The tests with I

obstacles are expected to give considerably higher flame speeds, overpressures, and possibly wider compositional range in which transition to detonation is observed.

t I

)

4 l

19-16 t

l PLATINUM CATALYTIC IGNITERS FOR LEAN HYDROGEN / AIR HIXTURES*,**

L. R. Thorne and J. V. Volponi, V. J. McLean Sandia National Laboratories Livermore, California 94550 A loss-of-coclant accident (LOCA) may produce large quantities of hydrogen gas. The hazards associated with the resulting hydrogen-air mixture may be reduced by intentionally burning it at sufficiently lov hydrogen concentrations that little if any damage to the containment building vill occur. Current implementations of this hydrogen mitigation strategy make use of glovplugs which require an uninterrupted supply of electrical power. However, in the event of a serious accident, electrical power may not be available. Thus there is a need for a non-powered igniter.

Ve have developed a catalytic igniter that can operate under conditions that may prevail during an LOCA and that does not require an external source of power of any kind. The igniter is composed of a catalytic substrate and several vires which project into the unreacted gas.

The substrate is an alumina honeycomb (4.4-cm diameter, 3.0-cm height, with 0.2-cm diameter cells) that is coated with high-surface-area platinum particles to about 1.7 veight % platinum. The wires are also platinum (0.0123-cm diameter, 4.0-cm long). The igniter operates by catalyzing the surface reaction between hydrogen and oxygen. This reaction produces heat that in turn accelerates the reaction rate of hydrogen and oxygen impinging on the surface. When the igniter substrate, initially at room temperature, is exposed to a hydrogen / air mixture, its temperature begins to rise. This in turn heats the wires so that the reaction on the wire surface accelerates and they become hot enough to ignite the gas phase mixture.

Ignition occurs after an induction time of 20 - 400 s that depends on the hydrogen concentration, gas flow velocity, gas temperature and the relative humidity. The induction times were measured for hydrogen concentrations in the range of 5.5 - 11.0%, gas flow velocities between 1.7 and 19.5 cm/s, gas temperatures between 20 and 65 C and relative humidities between 5 and 98%. Induction times are shorter for mixtures with higher hydrogen concentrations, higher flow velocities, higher gas temperatures and lover relative humidity.

This research 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.

- The design of the optimized catalytic igniter described herein is currently the subject of a Sandia patent disclosure.

19-17

The igniter operates repeatedly. Some of the igniters used in this study were cycled tens of times without any sign of reduced performance.

This is a desirable characteristic since the hydrogen produced during an LOCA may require several separate ignition events to eliminate hazardous concentrations of hydrogen. Liquid water blocks the catalytic sites and defeats the igniter. When a vet igniter is dried it operates normally, indicating that water does not poison the catalytic sites. Preliminary tests indicate that liquid water may not defeat igniters made with the " vet proofed" catalytic substrates developed at Atomic Energy of Canada's Chalk River Laboratories. However, higher hydrogen concentrations are required for ignition to occur.

i

)

l i

19-18 l

l INFLUENCE OF FLOW CHANNEL GEOMETRY AND DROPLET SIZE ON CONTAINMENT SUBCOMPARTMENT ANALYSIS l

by K. Almenas Department of Chemical and Nuclear Engineering University of Maryland i College Park, Maryland 20742 and R.Y.F. Lee Office of Nuclear Regulatory Research U.S. Nuclear Regulatory Commission Washington, D.C. 20555 4

Subcompartment analysis requires the determination of mass transfer of a two-phase (liquid-vapor), two component fluid (air-water) between large complex geometry volumes. Comparisons of model results with newly available data have revealed substantial divergences between calculated and measured intercompart-ment pressures differences. The modeling of short term pressure distribution in containment subcompartment after a D.B. LOCA thus does not meet the accuracy criteria expected for thermal-hydraulic calculations.

There are several reasons for this situation. A successful calculation of pressure gradients within a network of subcompartments requires a specification of:

a) flow resistances, b) the evaluation of an effective local fluid density, c) consideration of energy transfer to structures and d) the determination of relative droplet to vapor phase velocities.

Available experimental data is not sufficiently extensive to allow an empirical evaluation of the effects of the listed parameters. Therefore, models had to be simplified by approximating the influence of some independent variables or neglecting such influences altogether.

This study analyses the effect of droplet size and the nature of the aperture i

flow field on the buildup of inter-compartment pressures. These are modeling aspects which up to now have not been accessible to direct experimental investi-l gation. The only remaining alternative then is to choose a sufficiently idealized flow situation for which a mechanistic, non-lumped parameter analytical approach

- becomes feasible. The analysis has been performed for circular openings and channels j of various diameters and channellengths. The effect of drop size is investigated i

19-19 i

i

by assuming that the drop population can be represented by a representative drop diameter. Neither mechanical nor thermal equilibrium is assumed. The effect of drop size on pressure buildup is analyzed for the entire feasible drop spectrum extending from Ip to 1500p diameter drops. During transportation between compartments drops are accelerated by interphase drag. Larger size drops having large inertias lag increasingly behind the vapor flow field. Since they are transported at lower velocities the inter-compartment pressure required for acceleration is correspondingly reduced. This results is expected and can be modeled at least approximately in lumped parameter codes by using appropriate slip relationships. A less familiar aspect of the study quantifies the effect on pressure buildup of the diamter and length of the flow channel. It is shown that for increasing diameter flow channels the flow field becomes less steep. Drops spend more time in an accelerating field, reach higher velocities and thus contribute more to pressure buildup. The same relationship exi,sts between channel length and pressure buildup.

1 i

l i

19-20 l

_ _ _ _ . . . . _ ._. . _ . _ - __ . . _ _ __ . __ __ ~.

HSST CRACK ARREST STUDIES OVERVIEW

C. E. Pugh l

Cak Ridge National Laboratory Oak Ridge, Tennessee 37831 The purpose of this presentation is to summarize the crack-arrest studies that are underway in the Heavy-Section Steel Technology (HSST) program, and to assist in understanding the integration of the following presentations. In general, the HSST program addresses light-water reactor (LWR) pressure vessel integrity under accident scenarios, including pressurized-the: mal-shock (PTS) events. Central to these studies is understanding vessel conditions that would initiate growth of an existing flaw and conditions that would lead to strest of I

a moving crack. Low temperatures and secumulated neutron exposure increase the tendency for flaws to propagate under abnormal loadings. In the case of PTS conditions, the most severely irradiated (inner surface) meterial in a vessel is exposed to the most severe stresses (combined thermal and pressure) at rela- '

l tively low temperatures that result from the injection of low-temperature cool-i ing water. Progress is being made in understanding phenomena associated with the behavior of cracks that might exist in a reactor pressure vessel under these conditions. The studies involve several laboratories and are integrated into an overall program plan.

Prior studies of crack arrest have utilized small specimens and focused on re-ducing dynamic effects of the running crack. The appropriateness of current ASTM recommendations on procedures ~ for testing small crack-arrest specimens is '

being examined through a round-robin test program. Small specimens, however, j

provide limited constraint of deformation in the crack plane region and permit only the generation of data at temperatures below those where arrest is likely i to occur in some PTS scenarios. The HSST Program has been and is continuing to provide crack-arrest data over an expanded temperature range through tests of thermally shocked cylinders, PTS vessels, and wide plate specimens. The wide-

, plate tests allow a significant number of data points to be generated at

, affordable costs, while the thermal-shock and PTS tests provide validstion data under full-constraint, transient, multiaxial loading conditions.

While the HSST thermal-shock experiments (TSEs) have produced a significant number of data points, the driving force in those experi:nents is thermal stress

*Research sponsored by the Office of Nuclear Regulatory Research, U.S.

Nuclear Regulatory Commission under Interagency Agreements 40-551-75 and 40-552-75 with the U.S. Department of Energy under Contract DE-AC05-840R21400

with Martin Marietta Energy Systems, Inc.

4 By acceptance of this article, the publisher or recipient acknowledges the

i U.S. Government's right to retain a nonexclusive, royalty-f ree license in and to any copyright covering the article.

a I a

l i 20-1

, . - v - - - , . , , , - --

-. , , , , - -, -- ,, -, . n - - . . , , - . .-m-

only and, consequently, crack-arrest data have been limited to below about 150 MPa+6. An important conclusion from the TSEs is that the Ky , data from these highly-restrained propagations fall well within the range of Ky , data from lab-

! oratory specimens and slightly above the K IR C"TV8' The pressurized-thermal-shock experiments (PTSEs) have the capability to pro-vide higher K7 , values under similar highly restrained conditions. To date, i one three part experiment (PTSE-1) has been performed and is discussed in a j later contribution to this session. PTSE-1 has provided K 7 data as high as

~

300 MPa*6 and at temperatures up to 30*C above the onset of the Charpy upper shelf.

j Most recently the HSST program has initiated a program to investigate the crack i

run-arrest behavior in large plates with steep toughness gradients. These I tests use wide plate specimens that possess a single-edge notch (crack) that initiates at low temperature and arrests in a region of increased fracture j toughness. The toughness gradient is achieved through a linear transverse tem-j perature profile across the plate. The experiments require the application of large tensile loads and are being conducted by the National Bureau of Standards in Gaithersburg, Md. The first objective is to provide KIa data above the ASME i K IR Curve upper-limit criterion for prototypical pressure vessel steels. The

. steels include typical base metals and low-upper shelf materials representative of degraded weld materials. Other objectives include providing data from which dynamic fracture analyses can be performed.

i Efforts are underway for analysis methods to take into account the dynamic na-ture of crack propagation-arrest events. These include consideration of iner-4 tial and strain-rate effects. The Southwest Research Institute (SwRI), the l University of Maryland and ORNL are working on integrated ef forts to develop l' elastodynamic fracture analysis finite-element programs and viscoplastic analy-sis methods. Elastodynamic fracture analysis procedures have been applied to the analyses of the wide plate crack-arrest experiments and the PTS tests. The results obtained to date show that the essence of the dynamic behavior is being

modeled and will be reported later in this session. Further refinements in l quantitative representation of material parameters and the inclusion of rate

dependence through viscoplastic modeling are expected to give a more accurate

] basis for assessing the f racture behavior of LWR pressure vessels under PTS and j other off-normal loading conditions. SwRI will report on experimental and analytical studies of a viscoplastic material model and associated fracture criterion.

I

! l l

( 20-2

l WIDE PLATE CRACK ARREST TESTING R. deWit, and R.J. Fields Fracture and Deformation Division National Bureau of Standards Gaithersburg, MD 20899 To predict the behavior of a nuclear pressure vessel undergoing pressurized thermal shock, certain information on dynamic crack propaga-tion and arrest is required. The purpose of the work described here is to provide such data on wide plates fracturing at temperatures up to the upper shelf region. This presentation discusses the results of the first tests.

Twelve single edge cracked tensile specimens are to be tested in a thermal gradient. Four tests have been completed on the 26 MN Universal Testing Machine at NBS. The edge-notched plates in the middle of the specimen were supplied by ORNL. The pull plates and tabs were designed and constructed by NBS. These are welded together into a pull plate assembly.

The specimens are all to be fractured in a thermal gradient that, in the most extreme case, might extend from -150*F to 500*F across the cold 1 m specimen width. This is done so that the crack will initiate in a cold brittle region and arrest in a hot tough region. To establish this gradient, a heating and cooling system was constructed. The temperature gradient will cause a slight in-plane bending of the specimen. When a load is applied to this bent specimen, a moment results that will affect the stress state of the specimen.

An important part of this study is data acquisition from the numerous strain gages, thermocouples, timing wires, crack mouth opening displace-ment gages, and acoustic emission transducers that are mounted on the specimen. Low reactance bridges, wide range dynamic amplifiers, high speed digital oscilloscopes, and wide band FM magnetic tape recorder are used to record the response to crack propagation and arrest. This equip-ment is controlled remotely by a microcomputer. In addition a lead-wire junction box and data logger with digital voltmeter are used to monitor the temperature and load. The data are recorded on a second microcom-puter. Details of the data acquisition system will be given.

Each test has been different with respect to conditions of testing, specimen configuration, and instrumentation used. The progressive changes in test procedure represent attempts to obtain the desired crack run and arrest behavior and to improve upon the quality of the data collected. In particular, efforts were made to initiate crack propagation at lower stress intensity factors. Also, strain gage combinations and locations were optimized to better deduce the crack position as a function of time.

Each test yields a wealth of information. Since using the tape recorder on the second test (WP-1.2) it has been possible to record 28 signals. Some 20 of these were strain gages. Others were load, crack mouth opening displacements, timing wires, and acoustic emission. These 20-3

signals are recorded at 30 inch per second over the half hour period of loading till the specimen breaks into two pieces.

To analyze the results the data are played back from the recorder into the digital oscilloscope and from there into a microcomputer. Hence, a record of the complete loading history can be obtained. However, the capability of the tape recorder is so good that details of the run arrest events on the order of 1 ms can also be resolved.

The strain gages near the crack plane show a peak as the cleavage crack runs past each gage. This helps determine crack position as a function of time. Other clues are used to determine crack position when the crack growth becomes ductile. By differentiation these results the crack velocity is obtained as a function of time. Another result of great interest that can be deduced from these tests is the initiation fracture toughness and the arrest toughness.

1 i

20-4 I

ELAST0 DYNAMIC FRACTURE ANALYSES OF 1ARGE CRACK-ARREST EXPERIMENTS

B. R. Bass, C. E. Pugh, J. K. Walker Oak Ridge National Laboratory Oak Ridge, Tennessee 37831 I The WP-1 series of HSST wide-plate crack-arrest tests are being performed at the National Bureau of Standards (NBS), Caithersburg, MD, using specimens from HSST Plate 13A of A533, Gr B, C1 1 steel. The six tests in the WP-1 series are aimed at providing crack-arrest data at temperatures up to and above that cor-responding to the onset of the Charpy upper-shelf, as well as providing infor-mation on dynamic fracture (run and arrest) processes for use in evaluating improved fracture analysis methods. The tests use single-edge-notched (SEN) plate specimens that are cooled on the notched edge and heated on the other edge to give a linear temperature gradient along the plane of crack propaga-tion. Upon initiating propagation of the crack in cleavage, arrest is intended to occur in the higher-temperature ductile region of the specimen. The speci-

, mens are instrumented with strain gages and thermocouples to provida strain and i temperature data as functions of position and time.

Crack propagation-arrest behavior has historically been interpreted and ana-lyzed in terms of static fracture mechanics concepts. However, the events are inherently dynamic, so that the analysis methods should consider inertial and strain-rate ef fects that are known to be present. Accordingly, the HSST pro-gram is developing, in concert with subcontracting groups, dynamic fracture analysis procedures and applying them to the analyses of the wide-plate crack-arrest experiment. The elastodynamic analyses of the wide plate tests de-scribed in this paper were carried out with the SWIDAC and ADINA/EDF dynamic crack analysis codes. These codes utilize a displacement-based finite element formulation and the implicit Newmark-Beta scheme for time integration of the equations of motion. The dynamic stress-intensity factor Kyis determined in each time step from the dynamic J-integral containing the appropriate inertia and thermal terms. The crack-growth modeling technique of these codes utilizes a scheme in which crack-plane nodes, initially restrained normal to the crack plane by stiff springs, are released incrementally according to the selected analysis mode (application or generation). In an application-mode analysis,

Research sponsored by the Office of Nuclear Regulatory Research, U.S.

Nuclear Regulatory Commission under Interagency Agreements 40-551-75 and 40-552-75 with the U.S. Department of Energy under Contract DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc.

i By acceptance of this article, the publisher or recipient acknowledges the U.S. Government's right to retain a nonexclusive, royalty-free license in and to any copyright covering the article.

f I 20-5

I the crack is propagated incrementally according to a prescribed dynamic frac-ture toughness relation, and the crack-front position history is determined f rom the finite element solution. In a generation-mode analysis, the crack tip is propagated according to a prescribed position-time history estimated f rom ,

crack-line strain-gage data recorded during the test. l i

Elastodynamic analyses have been completed for the actual test conditions of the four tests WP-1.1 through WP-1.4 conducted thus far in the WP-1 series. In this paper, the computed results are compared with data for crack-line strain-time response, crack-propagation speed, crack opening displacement, arrest location and post arrest tearing. Results f rom both application-mode and gen-eration-mode dynamic analyses are presented. In addition, the paper includes a

! summary of the arrest toughness calculations compiled in the four tests at tem-peratures that range from transition to upper-shelf values for the wide plate material.

These same elastodynamic fracture analysis techniques have been applied to the analysis of the first pressurized-thermal-shock experiment (PTSE-1) performed j at ORNL. The calculated results for crack depths and crack opening displace-i ments compared very favorably with measured data.

) The results obtained to date show that the ' essence of the run-arrest events, including dynamic behavior, is being modeled. Refined meshes and optimum solu-tion algorithms are important parameters in elastodynamic analysis programs to give sufficient resolution to the geometric and time-dependent aspects of frac-ture analyses. Further refinements in quantative representation cf material parameters and the inclusion of rate dependence through viscoplastic. modeling is expected to give an even more accurate basis for assessing the fracture be-havior of reactor pressure vessels under PTS and other off-normal loading con-

ditions.

i a

I 20-6

DEVELOPMENT OF VISCOPLASTIC FRACTURE MECHANICS l M.F. Kanninen, S.J. Hudak, Jr., K.W. Reed, l R.J. Dexter, E.Z. Polch, J.W. Cardinal, J.D. Achenbach" and C.H. Popelar**

Engineering and Materials Sciences Division Southwest Research Institute San Antonio, Texas The objective of this research is to provide a fundamentally correct methodology for the prediction of crack arrest at the high upper shelf conditions occurring in a postulated pressurized thermal shock (PTS) event.

Current predictive methods assume linear elastic material behavior and quasi-static conditions for analyzing crack propagation / arrest events in both nuclear pressure vessels and in the small-scale test specimens that are used to determine the material fracture properties. While these key assumptions are not necessarily inappropriate, because of the crucial nature of crack arrest predictions in PTS safety assessments, they must be critically examined. This can only be accomplished by analyses that take dynamic (inertia) effects and rate-dependent yielding directly into account.

Accordingly, a dynamic-viscoplastic analysis procedure is being pursued for this purpose.

This research centers on the development of a finite-element method for the solution of time-dependent boundary value problems that admit inertia effects, a prescribed spatial temperature distribution, and viscoplastic constitutive behavior. Supporting this development are (1) material characterization testing, (2) detailed mathematical analyses of the near-tip region, and (3) small-scale fracture experimentation. The material characterization work has provided the constitutive behavior for A533B steel in the form introduced by Bodner. The mathematical analyses are designed to determine the effect of local heat generation on the temperature-dependent yielding accompanying a rapidly propagating crack. The experiments employ compliant-loaded compact tension specimens with refined instrumentation to determine the crack length and load-line displacement histories accurately.

The results of " generation-phase" analyses will then provide an effective temperature and crack speed-dependent fracture criterion for use in analyses of PTS events.

Because of the difficulty in obtaining crack arrest toughness values at high temperatures using conventional compact specimens, wide plate tests are being performed by the National Bureau of Standards (NBS) in a companion HSST program. As a first step, generation-phase dynamic-viscoplastic analyses are currently being made of the NBS tests in this program. The preliminary results indicate that a substantial amount of inelastic deformation occurs along the crack line prior to the arrival of the crack tip. This offers a possible explanation for the observed intermediate crack arrests; observations that are inconsistent with linear elastic analyses.

l l

Northwestern University, Evanston, Illinois '

- The Ohio State University, Columbus, Ohio 1

20-7 I

PRESSURIZED-THERMAL-SHOCK EXPERIMENTS:

PTSE-1 RESULTS AND PTSE-2 PLANS

R. H. Bryan J. G. Merkle l

R. K. Nanstad G. C. Robinson R. Wanner G. D. Whitman Oak Ridge National Laboratory Oak Ridge, Tennessee 37831 Results of the first pressurized-thermal-shock experiment PTSE-1 have now been evaluated, and the second experiment PTSE-2 is planned for mid-1986. Prelimi-nary conclusions f rom PTSE-1 were reported at last year's information meeting.

Now all the data recorded during the test and fractographic evidence of the be-havior of the flaw have been thoroughly evaluated. These experimental results are important because fracture phenomena were observed under conditions that were as representative of plane strain as are likely to be studied experimen-tally at temperatures and stresses representative of real reactor pressure ves-f sels.

The series of pressurized-thermal-shock experiments was motivated by a concern for the behavior of flaws in reactor pressure vessels having welds or shells exhibiting low upper-shelf Charpy impact energies, ~68 J or less. Evaluations j of overcooling accidents, however, involved consideration of other complexities that had not been explored under particularly realistic conditions. Issues

(

that have an important impact on accident evaluation and are also amenable to investigation in pressurized-thermal-shock experiments are: effects of se-quences of warm prestressing and anti-warm prestressing episodes on crack ini-tiation; behavior of cleavage fracture propagating into ductile regions; transient crack stabilization in ductile regions; and crack shape changes in bimetallic zones of clad vessels.

The first experiment, PTSE-1, was performed with an unclad vessel with a 1-m-long surf ace crack in a welded-in plug of specially tempered steel. The plug was made of a forging that, with normal heat treatment, would meet the

specifications for SA 508 class 2 steel, a material with extensively studied l properties. P,TSE-1 demonstrated the strongly inhibiting ef fect of simple warm j prestressing (K7 < 0) on crack initiation. In at least one of the anti-warm-prestressing phases of the experiment (Kg > 0) Ky became much greater than KIc j *Research sponsored by the Office of Nuclear Regulatory Research, U.S.

Nuclear Regulatory Commission under Interagency Agreements 40-551-75 and

, 40-552-75 with the U.S. Department of Energy under Contract DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc.

i By acceptance of this article, the publisher or recipient acknowledges the U.S. Covernment's right to retain a nonexclusive, royalty-free license in any and to any copyright covering the article.

20-9

1 I

without causing the crack to run. In this instance, however, K during anti-7 warm prestressing was much smaller than its previously attained maximum value.

Subsequently , the crack propagated by cleavage and arrested twice. Nearly pure

= 300 MPa =[m cleavage and persisted a temperature of to the point

-179'C, which of is final arrest well at the above a value onsetofof K the

, Charpy upper shelf.

In the second experiment, PTSE-2, the warm prestressing phenomenon will be ob-served, if possible, in . a transient in which yK will, during anti-warm pre-stressing, exceed previous maxima. The flaw in the PTSE-2 vessel will reside in material designed to have - a low upper-shelf impact energy, namely 54 to 68 J. In the final phase of the experiment, the crack will propagate initially in cleavage, run unstably in a ductile tearing mode, and be arrested by unloading the vessel.

i

]

l l

i g

l -> '

i l

20-10

_m. _ - - ,.

i REACTOR VESSEL CLADDING SEPARATE EFFECTS STUDIESO W. R. Corwin i\'

i e Metals and Ceramics Division OAK RIDGE NATIONAL LABORATORY i' Oak Ridge, Tennessee 37831

SUMMARY

f The existence of a layer of tough weld overlay cladding on the interior of a light-water reactor pressure vessel could mitigate damage caused during certain overcooling transients. The potential benefit of the cladding is that it could keep a short surface flaw, which would otherwise become long, from growing either by impeding crack initiation or by arresting a running crack. Two aspects critical to cladding behavior being investigated by the Heavy-Section Steel Technology Program

! will be reported: irradiation effects on cladding toughness and the l response of mechanically loaded, flawed structures in the presence of cladding.

In evaluating the potential for radiation damage of weld overlay cladding, a two-phase irradiation experiment is being conducted. In the first phase, Charpy impact and tensile specimens from a three-layer stainless steel weld overlay fabricated using the oscillating single-wire submerged-arc technique were irradiated to 2 x 1023 neutrons /m2 (>i gey) at 288'C. Cladding from the upper weldment layers, typical of good quality pressure vessel cladding, exhibited very little irradiation-induced degradation. However, ductile-to-brittle transition behavior, caused by temperature-dependent failure of the residual 6-ferrite, was a observed during impact testing. In contrast, specimens from the first weldment layer, which also exhibited transition type behavior, were markedly embrittled. The cause of the embrittlement was determined to be high-radiation sensitivity of the atypical microstructure resulting f rom excessive base metal dilution of the first weldment layer. In the

! second phase of irradiations, now in progress, a commercially produced three-wire series-arc weldment will be evaluated under identical irradiation and testing conditions as the first series. In addition, 0.5T compact specimens of both weldments will be examined to confirm that the fracture toughness behavior follows the trends already determined. Moreover, the maximum fluence examined will be increased to 5 x 1023 neutrons /m2 ()1 gey),

Research sponsored by the Office of Nuclear Regulatory Research,

U.S. Nuclear Regulatory Commission, under Interagency Agreements DOE 40-551-75 and 40-552-75 with the U.S. Department of Energy under contract DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc.

I 20-11 r

aI d- .,. , _ _ _ . . , , _ c_ , _ , , _ _

To examine the structural ef fect of cladding in limiting crack extension, a two-phase program is being conducted utilizing relatively large (914- x 406- x SI-mm) four-point bend specimens that have been clad and flawed on the tension surface. The testing rationale in both series is that if a surface flaw is pinned by the cladding and cannot grow '

longer, it will also not grow beyond a certain depth, theraby arresting the entire flaw in a stress field in which it would otherwise propagate through the specimen and cause total failure. In the first phase, which used cladding identical to that of the first irradiation phase, testing at a temperature at which the base plate was frangible resulted in the cladding being relatively low in its Charpy transition. The results showed that cladding with low-to-moderate toughness appeared to have a limited ability to mitigate crack propagation. For the second phase, three-wire cladding like that being utilized in phase 2 of the irradiation experiment has been deposited on a base plate with a very high ductile-to-brittle transition temperature. This will allow testing on or near the upper shelf of the cladding to ascertain the crack-inhibiting capability of tough cladding.

2

=

1 A

20-12 i p;

IRRADIATION EFFECTS IN LOW-ALLOY REACTOR PRESSURE VESSEL STEELS (HEAVY-SECTION STEEL TECHNOLOGY PROG (GUI SERIES 4 AND 5)*

J. J. McGowan, R. G. Berggren, R. K. Nansted ,

K. R. Thoms ,t and B. H. Menkei Metals and Ceramics Division OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee 37831

'+

SUMMARY

The Fourth and Fif th Irradiation Series are a part of the Heavy-Section Steel Technology (HSST) Program. The Fourth HSST Irradiation.

Series, almost completed, applies statistical analyses to fracture toughness results for four irradiated " current practice" (circa 1979) submerged-arc welds and an A533 grade B class 1 plate (HSST-02). The Fif th HSST Irradiation Series (irradiations now in prcgress) is aimed at obtaining a statistically significant fracture toughness data base on two weldments with high-copper content to determine the shift-and shape of the KIc curve as a consequence of irradiation.

In the Fourth HSST Irradiation Series, Charpy V-notch (CVN), tensile, and 25-mm-thick compact specimens (1TCS) were irradiated at 288*C to

'~

neutron fluences of 0.7 to 2.0 x 1023 neutrons /m2 (>l MeV) . The materials were a 305-mm-thick plate of A533 grade B class 1 steel and four submerged-arc welds. The plate material contained 0.14% Cu and 0.67% N1.

The four submerged-arc welds contained 0.04, 0.12, 0.06, and 0.05% Cu and 0.13, 0.10, 0.63, and 0.63% Ni, respectively. For a fluence of about

~

2x 1023 neutrons /m2, the plate material showed a yield strength increase of 30%, ultimate tensile strength increase of 20%, a CVN impact transition temperature increase of 68*C, and a CVN upper-shelf energy drop of 46%.

The four submerged-arc welds showed smaller changes than did the plate material, 12 to 19% and 3 to 9% increases in yield and ultimate tensile strengths, respectively, 6 to 34*C lhereases in transition temperature, and upper-shelf energy changes were within data scatter. The fracture

~

toughness results from the ITCS specimens showed approximately the same temperature shift as did the CVN results. All results were statistically

Research sponsored by the Office of Nuclear Regulatory Research, U.S. Nuclear Regulatory Commission, under Interagency Agreements DOE 40-551-75 and 40-552-75 with the U.S. Department of Energy under contract DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc.

tEngineering Technology Division.

TMaterials Engineering Associates, Lanham, Maryland.

20-13

e analyzed by nonlinear regression methods. Charpy and fracture toughness tests were both conducted by two laboratories and there was excellent agreement between results from the two laboratories. The results imply that submerged-arc welds with both low copper and low nickel contents can exhibit essentially zero radiation embrittlement. In addition to the l effects of copper content, the results imply that nickel (0.63%) can contribute to radiation embrittlement even when the copper content is low

, (0.05%).

The Fif th HSST Irradiation Series is designed to obtain statistically j significant data on the shift of fracture toughness (KIc) and to correlate 4

these results with CVN and drop-weight test results. Tensile, CVN, drop-weight, and IT, 2T, and 4T compact specimens have been irradiated and are i l being tested. Additionally, 6T and 8T unirradiated compact specimens are being tested to attain the same K Ie measuring capacity as the irradiated specimens. The materials for this irradiation series are two wcldments f abricated from special heats of weld wire with copper added to the melt.

g Copper levels for the two welds are 0.25 and 0.34% _and the nickel level is

0.60%. Irradiations have been conducted at the Oak Ridge Research Reactor

at an irradiation temperature of 288'C to a fast neutron fluence of 1.7 x 1023 neutrons /m2 (>l MeV). Testing of fracture toughness specimens will be conducted _ primarily to establish valid KIe versus test temperature curves . Tests on IT and 2T compact specimens will be used to establish test temperatures for the larger' specimen tests, and to provide a comparison of JK e results with valid K Ie results from the larger s pecimens.

I e

l l

t J

l 20-14

RADIATION SENSITIVITY AND ANNEALING PARAMETER STUDIES J. R. Hawthorne Materials Engineering Associates, Inc.

Thrusts made in exploration of steel radiation embrittlement sensitivity and properties recovery by postirradiation heat treatment have produced significant new information in both areas. Information developed on radiation sensitivity conc erns the influence of neutron flux levelgn thg magnitude of embrittlement The produced studies by a fluence of notch of ductility

~ 0.6 x 10 n/cm , E > 1 MeV.

recovery tested for the ef fects of steel composition on 399'C (750*F) annealing response, and additionally, weld metal re-emb rittlement trends following 399*C annealing as functions of preanneal fluence level, postanneal fluence level and the flux type used in weld fabrication. Highlights of findings by Materials Engineering Associates (MEA) in each of the research investigations are presented.

The significance of neutron flux level to reactor vessel resistance to radiation-induced embrittlement is a long standing question. Closely associated with this uncertainty is the comparability of power vs.

test reactor environments in producing notch ductility, fracture toughness and strength changes. To help resolve this issue, a series of irradiations under closely-controlled conditions were undertaken with reference plate and weld deposit materials. Charpy-V (Cy ),

fatigue precracked Charpy-V (PCCy ), compact tension and tension test specimens of each are beigg irradiatg in the UBR regtor gt thrqe 7-8 x 10 ,g 4-6 x 10 and and 27-9xf0 flux levels: nfcm-sec Target fluences are 0.5 x 10 , 1.0 x 10 x 10 n/cm .

The first specimen assembly irradiated at the intermediate flux level has been discharged from the reactor and C y specimens tested. The specimens, fabricated from the ASTM A 302-B reference plate (0.21% Cu, 0.20% Ni) and a gde 8g submerged-arc weld (0.36% Cu, 0.65% Ni),

received 0.64 x 10 n/cm at 288'c (550*F). The results, when compa' red against projections based on data for the materials irradia-ted at the highest flux, indicate a material-dependent dose rate effect. The magnitude of weld metal embrittlement was mucgmore ghan expec ted _g eat as that measured after 1.4 x 10 n/cm at 8.5 x 10 being asn/cm -secThis

[. indication was not observed for the A 302-B plate which was in the same assembly. Moreover, weld perform-ance is the reverse of that trend reported for welds irradiated in power reactor surveillance vs. test reactor facilities to higher The geutron exposure to a second intermediate flux capsule fluences.19 n/cm ) will be completed in late 1985 and will permit an

(~ 1 x 10 adc1tional test of the material dependency.

Material composition is also expected to be a major factor in post-irradiation annealing recovery, both for welds and plate materials.

One study of composition effects on annealing recovery initiated by MEA, is using materials initially acquired for investigating radiation sensitivity variables and mechanisms. One of two planned experiments 20-15

have ~been completed. The experiment contained five composition variations about A 533-B steel and focused on t he role of copper,

nickel and phosphorus in recovery by 399*C annealing. The data do not indicate a major influence of phosphorus content on residual embrit-tiement after annealing when copper content is high. Nearly full recovery was observed when copper content was low. Jointly, ,the results signify that residual embrittlement is a func tion of copper, and not . phosphorus content. A second experiment, now in reactor, will evaluate the effect of variable copper and variable nickel content for the . c ase of low phosphorus. A parallel effort involving welds with high/ low copper and high/ low nickel contents and multiple flux types is also underway. Objectives are the evaluation of 399'C vs. 454*C 1 (850*F) ' annealing and reirradiation-embrittlement behaviors for the

! several material types.

Results from the Irradiation-Anneal-Reirradiation (IAR) Phase 2

, program investigations on Linde 80 and Linde 0091 submerged-arc welds 1

are presentg. Ngtch ductility data for a high, first cycle fluence

(~ 2.4 x 10 n/cm ) confirm the relatively rapid rate of re-emb rittlement after annealing that was firsp9 obsegved with the

, welds with a low first cycle fluence (~ 1.2 x 10 n/cm ). A benefit of IAR treatment is also clearly evident in the data. Weld dissimi-larities in radiation embrittlement trends and in their reirradiation susceptibility are evident and are believed to be a manifestation of a composition factor (or factors). A pronounced trend toward radiation embrittlement saturation was found for the Linde 80 weld.

l J

t f

l l

i 20-16

LWR-DOSIMETRY PROGRAM OVERVIEW by E. D. McGarry National Bureau of Standards, Gaithersburg, Maryland ABSTRACT (to be presented at NBS at the 13th NRC WRSR Meeting)

The Division of Engineering Technology, Office of Nuclear Regulatory Re.3earch supports research to improve neutron surveillance dosimetry of Light Water Reactor (LWR) pressure vessels. The research program is carried out by three contractorat the Westinghouse-Hanford Engineering and Development Laboratory (HEDL), the Oak Ridge National Laboratory (ORNL), and the National Bureau of Standards (NBS). This LWR-PV Surveillance Dosimetry Improvement Program (SDIP) employs neutron dosimetry measurements, transport calculations and metallurgical damage analyses to more accurately assess the neutron exposures and embrittlement damage rates in LWR pressure vessels.

Accuracy goals for dosimetry are in the range of 5% to 15%.

Currently being incorporated in new and updated ASTM standards are methods by which dosimetry measurements and transport calculations can be consolidated to reduce uncertainties on neutron exposure parameters.

These will be reviewed. Emphasis is on prediction of exposures at locations different from those where surveillance measurements are made, particularly at critical welds and within the pressure vessel.

To validate measurement and calculational procedures, the SDIP has built and neutronically enaracterized several pressure-vessel and surveillance-capsule simulators at ORNL. A summary of results will be given for the various benchmarking experiments performed in these controlled neutron-field environments. In addition to verification of fluence perturbation effects of various types of surveillance capsules, and investigations of lead factors, the simulator facilities have been used to benchmark reference and provide quality assurance for conventional and new types of neutron sensors used for PV 1rradiation surveillance. The status of state-of-the-art dosimetry and commercial interests will be discussed.

A.world-wide research effort to solve PV embrittlement problems exists and there is a strong cooperative link between the NRC-supported research and research at SCK/CEN (Mol, Belgium), KFA (Julich, Germany, EPRI (Palo Alto, USA) and several UK laboratories. In particular, the Belgian VENUS Experiment has investigated the accuracy of transport calculations in the vicinity of the step-shaped periphery of the fuel in commercial reactors. The British Winfrith-NESDIP Experiments provide for simulations of pressure-vessel cavity environments. The significance of results from these experiments to those from the SDIP will be considered.

20-17 n

The damage analysis phase of the LWR-PV-SDIP serves the NRC objective of advancing the knowledge of irradiation embrittlement of steel. It combines both neutron dosimetry and transport calculations with metallurgical results. It necessarily draws upon the results of ,

irradiation testing, damage correlation studies and trend curve analyses. HEDL and other Program participants have studied possible correlations between chemical impurities, exposure parimeters, fluence rates and temperature effects. This supports the development of trend curve multiplicative correction-factor terms that account for neutron fluence-rate effects and nickel (and copper) impurity effects on PWR plant-specific trend curves. Conclusions will be summarized.

In support of the damage analysis studies, HEDL least-square adjustment analyses were used to re-evaluate neutron exposure fluences and displacements per atom (dpa) for 47 PWR and BWR surveillance capsules from W, B&W, CE and General Electric (GE) (for BWR) power plants. The revised fluence values for energies greater than 1 MeV average 25% higher than originally reported values, with a (new/old) range from 0.8 to 2.4.

The LWR-PV-SDIP studies conclude that consistent with dosimetry accuracy goals and realistic allowances for competing uncertainties, damage exposure parameters, such as fluence greater than 1 MeV and dpa, may be specified with accuracles in the 10% to 30% range. Furthermore, it is concluded that displacements per atom (dpa) is a damage parameter that is superior to fluence greater than 1 MeV because dpa better accounts for displacement damage caused by neutrons scattered below 1.0 MeV. The dpa concept and specific cross section data are documented in an ASTM Standard. The general use of the ASTM standards to guide in the selection, deployment and interpretation of advanced dosimetry from reactors will be discussed.

20-18 i

n

l TRAC-PF1/ MOD 1

Dennis R. Liles and Susan B. Woodruff Safety Code Development Group Energy Division Los Alamos National Laboratory Los Alamos. NM 87545

SUMMARY

The Transient Reactor Analysis Code (TRAC), an advanced best-estimate systems code for the analysis of light-water reactor accidents, is being developed at Los Alamos under the sponsorship of the Office of Nuclear Regulatory Research of the US Nuclear Regulatory Commission. TRAC is used around the world for reactor-safety applications.

TRAC-PF1/ MOD 1 was designed to provide full balance-of-plant capabilities. The code thus incorporates a general capability for modeling plant control systems as well as a special turbine component. The characteristics of TRAC-PF1/ MOD 1 include variable-dimensional fleid dynamics. a full two-fluid (six-equation) hydrodynamics model, a flow-regime-dependent constitutive equation package, and a comprehensive heat-transfer capability. The code incorpo-rates the stability-enhancing two-step numerical method in the one-dimensional components.

This numerical method a!lows large time steps to be used for slow transients. An updated version of TRAC-PF1/ MOD 1 (version 12.1) was " frozen" in January 1985. This version is maintained by Los Alamos. Periodic updates addressing code errors and upward-compatible user enhancements are issued to external users through dial-up procedures. The primary forthcoming user enhancement is a flexible heat-structure capability. Users will be informed when this capability has been implemented and tested.

Improvements recently incorporated in version 12.1 include a plenum component. high-pressure thermodynamic properties, and improved posteritical heat-flux heat-transfer mod-eling. The plenum component allows an arbitrary number of one-dimensional components to be connected to a single hydrodynamic cell. The high-pressure properties module allows calculations above the critical point: however, this feature has not been thoroughly assessed.

One heat-transfer improvement considerably reduces the computational costs of the model.

This work was funded by the US Nuclear Regulatory Commission. Office of Nuclear Regulatory Research. Division of Accident Evaluation.

21-1

.k.-

. -a 4 3 4 _ ,..4.m .m A.2 - a a e .ua a..ma s s.., ,e 2 m .- _. . _ __

4 d I i

1, d

h.

1 r i

l f I

4 1

i 1

6

,3 d

t e

l i,

f i

I 4

i l 9

/

i

't 4

4 7

4, 0

9 1

b 4

i r

4 1

l' ... . , . _. - _ . . , ,. . v .. .. . _ _ _ . - - - - _ . . . , _ . ,,_--- .._m. , . . . - . . . , .. , _. - . _ , . . . . . ,. . _ _ . _ _,_

l TRAC-BWR By S. Z. Rouhani, W. L. Weaver III, R. W. Shumway and G. L. Singer Idaho National Engineering Laboratory EG&G Idaho, Inc.

I Summary The Transient Reactor Analysis Code for Boiling Water Reactors (TRAC-BWR),

which is being developed at the Idaho National Engineering Laboratory (INEL) for the NRC, is an advanced best-estimate computer code for analyzing transient behavior and safety aspects of boiling water reactors. From its start in 1979, this project has been coordinated with TRAC-PWR development at Los Alamos National Laboratory and has been conducted in close technical collaboration with a similar development at General Electric. The latest publicly released version of this code is TRAC-BD1/M001, which was released in 1984. A new version of TRAC-BWR (TRAC-BF1) will very shortly be released.

The focus of this paper is on the progress of the TRAC-BWR development program during 1985. In this year efforts concentrated on creating a faster running and more versatile version of TRAC-BWR. This goal was achieved in two steps, by first creating TRAC-BF0 and then TRAC-BF1.

TRAC-BF0 added to the capabilities of TRAC-BD1/ MOD 1 a one-dimensional neutron kinetic model and a material-Courant-limit-violating (fast running) numerical solution capability for one-dimensional hydraulic components (except the steam separator and turbine components.)

After preliminary independent assessment of TRAC-BF0 a number of new capabilities were integrated into the code, which resulted in the creation of TRAC-BF1. The salient additions in TRAC-BF1 are: an implicit solution scheme for steam separator and turbine components, improved control system logic and solution method, a consistent set of flow regime dependent constitutive relations for interfacial mass and momentum exchange (developed at GE), a condensation model for vertical ' stratified flow, and other improvements for treating the transition from single-phase to two-phase flow and vice-versa. Another addition in TRAC-BF1 is a preload processor which automatically checks the input for each problem and loads only those subroutines needed for its solution, thereby reducing computer 4

memeory requirements. Over 95% of the TRAC-BF1 code has been converted i to ANSI Standard FORTRAN 77, which makes the code more transportable to different computer systems. The graphics file has been modified to make TRAC-BF1 compatible with the Nuclear Plant Analyzer (NPA) being i

21-3 i

1 developed at the INEL and has also been made compatible with the CRAY i computer at Kirtland Air Force Base in Albuquerque, New Mexico.

In September of 1985 a TRAC-BWR workshop was held for the first time

to discuss the TRAC-BWR code capabilities in detail with users and potential j users of TRAC-BWR. User guidelines and the experience of the independent i assessment group at INEL were also discussed.

i A frozen version of TRAC-BWR is earmarked for international assessment under the auspices of the NRC's International Code Assessment Program (ICAP). Under this program a group of nations will assess code performance over a period of several years utilizing a wide variety of experimental

data. Results of the assessment will identify areas where further code j improvements are needed.

4

Plans for TRAC-BWR development in FY-1986 include the addition of a non-homogeneous boron transport model and a central differencing approach j in the solution of the momentum equation. Furthermore, work will begin j to extend the fast numerical solution scheme .to three-dimensional components.

j Also, efforts will be made to restructure the code for increased computational

efficiency, and in particular, for execution on vector and parallel processing computers. .These efforts are aimed at adaptation of TRAC-BWR to the Nuclear Plant Analyzer (NPA) system.

)

i i

i l

l l

i 21-4 ,

l

RELAPS/ MOD 2 DEVELOPMENT :

1 G. W. Johnsen Idaho National Engineering Laboratory EG&G Idaho, Inc.

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

A major development task completed in 1985 was the nearly-implicit solution scheme option. This optional scheme evaluates the convective terms implicitly, thereby removing the material Courant limit as a stability criteria. For quasi-steady processes, time step sizes of between five and ten times the Courant limit may be used without

! sacrificing accuracy. The optional scheme is particularly useful

for computing steady-state conditions (i.e., null transients) where

intermediate states are of no interest. In such cases, even larger j time steps may be used.

In connection with the conversion of RELAP5/M002 to execute on CRAY computers, work has continued at a modest level in 1985 toward the exploitation of vector processing on these machines. Since RELAPS

~ is structured primarily on the basis of looping over hydrodynamic volumes, junctions, and heat structure mesh points, it is ostensibly amenable to vector processing. However, the generality afforded by arbitrary volume connections leads to recursive calculations, which are a hindrance to vector processing. Conditional testing, necessitated by the two-phase model (e.g, different equations of state are required for each of several. possible coolant compositions also represents a challenge to vectorization. Compiler advances are now being made to overcome these obstacles, but are not now widely available. Several RELAP5/ MOD 2 subroutines that do not require the new compiler options have been rewritten for vector processing, with modest speed gains achieved.

With the basic development of RELAP5/ MOD 2 complete, emphasis has shifted toward maintenance and user support. To provide a mechanism for serving the many domestic organizations using RELAP5, a newsletter service was inaugurated in July. 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 up-to-date information on development

and application activities. Each' user may also contribute to the

! newsletter concerning their usage and experience.

I 21-5

^

Another avenue of user support is being provided through workshops.

This year, three workshops were held at domestic utility facilities and a fourth at EIR (Eidgenossiches Institute fur Reaktorforschung) in Switzerland. These workshops provide users with lectures on code fundamentals and application as well as hands-on practice with real problems. The workshops also facilitate feedback from users on desired code improvements.

This past year saw the formation of the International Code Assessment Program (ICAP). Sponsored by NRC and member countries, this program will undertake a rigorous plan of assessment of current light water reactor safety codes over the next three years. The plan calls for the usage of " frozen" code versions during this period. This strategy ensures each member utilizes the same code versior.. 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/ MOD 2(36), released in January 1985 was designated as the frozen version of RELAP5, and has been distributed to all member countries.

The documentation for RELAP5/ MOD 2 was upgraded and brcught up to date to be consistent with M0D2(36). Volumes 1 and 2 of the users manual (issued in draft form in April 1984) were expanded to include new code features (e.g. , the nearly-explicit numerical scheme) as well as material from the MOD 1 manual, which was previously referenced.

Errors detected by users in the draft manuals were also corrected.

Plans for FY-86 call for continued maintenance and user support for RELAP5. In addition, a new self-initialization option will be added to reduce the time and cost presently needed to initialize large plant models. This option will utilize the existing steady-state and nearly-implicit solution scheme features, coupled with a generic, built-in controller package. The latter will permit the user to specify any one of several parameter sets to be kept fixed, while other variables may " float". The steady-state option and new solution i scheme will cause the model to relax to a steady-state condition l many times faster than would be physically possible.

t l

21-6

~

0VERVIEW 0F RAMONA-3B CODE

i L.Y. Neymotin, G.C. Slovik, H.R. Connell, and P. Saha Department of Nuclear Energy i Brookhaven National Laboratory Upton, New York 11973 RAMONA-3B [1] has been developed at Brookhaven National Laboratory (BNL) under the sponsorship of the U.S. Nuclear Regulatory Commission (NRC) for ana-

. lyzing BWR systems transients. RAMONA-3B is the only currently available BWR systems transient code designed to predict three-dimensional power in the core, i fuel and cladding temperature and vessal thermal hydraulics phenomena. The code l capabilities cover a variety of BWR normal and abnormal system and accident transients including those with full, partial or no scram, control rod drop i transients, main steam isolation valve closure, turbine or recirculation pump i trip, transients induced by change in feedwater conditions, failure of pressure .'

J regulator, etc.

RAMONA-3B employs models for typical BWR components, namely, jet pump, re-circulation pump, steam separator, and steam line. The code has a comprehensive set of models for plant control and protection systems such as pressure regula-tor, feedwater control, standby liquid control system (boron injection), safety

and relief valves, High Pressure Coolant Injection (HPCI) and Reactor Core Iso-j lation Cooling (RCIC) systems. In addition, a model for the recirculation flow j control system is being implemented in the code. A brief description of the RAMONA-3B models is given below.

3 The neutron kinetics model of RAMONA-3B starts from the following set of j two-group, three-dimensional, time-dependent diffusion equations:

i 84 -

1 1 4

Fast Neutrons: T 1 at = v.D iv4 1- E21 +1- Eal*1 (1) 4 I

E 4 E 3 ij

+(1-8)[v1 fi t+v2 f2+2 +9k A c i

1 3*2 "

Thermal Neutrons: T2 at " Y'0 Y'2 2 + E21 *1 - Ea2 42 (2)

BC 1

Delayed Precursors: at "di E"1E f1*1 + "2E f2+23-Ac ii (3)

{

i = 1 to I where I = 6 i

Work performed under the auspices of the U. S. Nuclear Regulatory Commission.

i.

1 21-7 l

Postulating that the thermal neutron leakage term, i.e., v.D2 Vt2, can be either neglected or assumed to be constant, the model is further reduced to the well-known 1-1/2 group, coarse mesh diffusion model. The boundary conditions at the core periphery are specified using parameters related to the extrapolation length for the fast flux and albedo for the thermal flux.

The three-dimensional power distribution is calculated as a sum of the prompt and delayed energy generation rates. The prompt component is proportional to the instantaneous fission rate, whereas the delayed energy generation rate is calculated using the 1979 ANS Standard 5.1 for decay heat from fission products.

The cross section dependencies on fuel and moderator temperatures, core void fraction, boron. concentration and control rod pattern are taken into account in the neutron kinetics calculation. Energy generated in the fuel is transferred into the core coolant in two ways: through heat conduction in the fuel rods (fuel pellet, gas gap and cladding) and as heat deposited directly into the coolant.

To be compatible with 3-D neutron kinetics, the code uses parallel coolant channels in the core. The reactor vessel thermal hydraulics is based on a one-dimensional, four-equation slip model including vapor and mixture mass equa-tions, mixture momentum and energy equations. Use of slip correlations and non-equilibrium vapor generation models, including subcooled boiling and post-CHF correlations, constitutes a nonhomogeneous nonequilibrium approach in the treat-ment of the vessel thermal hydraulic phenomena. Condensation on the cold safety injection water in the downcomer region is accounted for by using a separate jet and film condensation model. The code also includes a boron transport model.

To reduce computational burden, the code uses a momentum integral approach and an average reactor vessel pressure which can, of course, vary with time. How-ever, in the steam line, pressure is a function of both time and space. Thus, the acoustic effects in the steam line due to valve closures or openings are taken into account.

RAMONA-3B has been and is being used for the analysis of a number of MSIV closure ATWS scenarios for the Browns Ferry plant under the NRC Severe Accident i Sequence Analysis (SASA) Program [2]. In the past, RAMONA-3B has been used to analyze center and off-center control rod drop accidents, MSIV closure full- and partial-ATWS, and Peach Bottom Turbine Trip Tests. The code has been assessed l using the FRIGG data and is being assessed using the FIST data.

RAMONA-3B is available to any U.S. organization, on a royalty-free basis, for the analysis of U.S. reactors. So far, the code has been distributed to ten U.S. organizations including a national laboratory, a reactor vendor, a univer-sity, several utilities and a number of consulting and service organizations.

REFERENCES

1. Wulff, W., et al., "A Description and Assessment of RAMONA-3B MOD 0 Cycle 4

- a Computer Code with Three-Dimensional Neutron Kinetics for BWR System Transients," NUREG/CR-3664, BNL-NUREG-51746, January 1984.

2. Slovik, G.C. , et al ., " Application of RAMONA-3B to BWR ATWS," to be pre-sented at the 13th Water Reactor Safety Research Information Meeting, Octo-ber 1985.

21-8

SUMMARY

OVERVIEW OF TRAC-BWR ASSESSMENT:

TRAC-BDl/ MOD 1 & TRAC-BF1 i

4 i Gary E. Wilson' Briant L. Charboneau K. C. Wagner John D. Burtt Idaho National Engineering Laboratory EG&G Idaho, Inc.

i P. O. Box 1625 Idaho Falls, Idaho 83415 1

i

! The assessments of the Boiling Water Reactor, Transient l Reactor Analysis Code (TRAC)-BDl/ MOD 1 performed during FY-1984

and FY-1985 have been analyzed as a whole. The objective of 2 this effort was twofold;.to evaluate the overall predictive quality of the code and to determine user guidelines relative to good modeling practice. In addition, an assessment of the i latest released version, TRAC-BF1, of this series of codes was

! performed with experimental data from a small break experiment I in the Rig of Safety Analysis (ROSA)-III facility.

The evaluation of the TRAC-BD1/ MOD 1 assessment studies indicate that most of the deviations between the~ code j simulations and experimental data were more related to modeling i practice than to code technology deficiencies. User guidelines l which provide better modeling practice were identified in the areas related to break flow, level tracking, CCFL, fine mesh

. nodalization, vapor condensation, separators, containments, l '

safety relief valves and flow restrictions. These guidelines are given in the paper.

An evaluation of the interplay between fast numerics and the

, quality of code prediction was made with the'results from the TRAC-BF1 assessment. This code appears to be approximately 60%

faster in execution than was TRAC-BD1/ MOD 1 while maintaining the same general quality of simulation. These results are discussed in the paper along with recommendations for the. continued assessment of TRAC-BF1, particularly with respect to efficient j simulations of~ full scale reactor systems. Other

recommendations, with respect to the further refinement of the j code technology, are also given.

1 i

i 21-9 i-i

l LOS ALAMOS TRAC-PF1/ MOD 1 CODE ASSESSMENT PROGRAM USING LOFT AND OTIS DATA

Thad D. Knight Safety Code Development Group Energy Division Los Alamos National Laboratory Los Alamos. NM 87545

SUMMARY

los Alamos National Laboratory is developing the Transient Reactor Analysis Code (TRAC) to provide an advanced. best-estimate, predictive capability for the analysis of pos-tulated accidents in light-water reactors. The TRAC-P series of codes provides this analysis capability for pressurized water reactors (PWRs) and for many thermal-hydraulic test facilities.

The code features a three-dimensional treatment of the pressure vessel and its associated in-ternals two-phase nonequilibrium hydrodynamic models. flow-regime-dependent constitutive relations. optional reflood-tracking capability for both bottom-reflood and falling-film quench fronts, and consistent treatment of entire accident sequences, including the generation of consistent steady-state conditions.

Two currently available code versions are TRAC-PD2/ MOD 1 and TRAC-PF1/ MOD 1.

TRAC-PD2/ MOD 1. which was developed primarily to analyze large-break loss-of coolant accidents (LOCAs), is older than T RAC-P F1/ MOD 1. Relative to TR AC-PD2/ MOD 1.

TRAC-PF1/ MOD 1 includes changes in the hydraulic modeling. the numerics. the constitutive relations. the trips and controls. and the user-convenience features to enhance applications of

! the code to transients other than large-break LOCAs. The newer code version also adds the capability to track a noncondensable gas and a solute, such as boron in the liquid field.

I will describe two separate assessment efforts at Los Alamos to investigate the large-and small-break LOCA capability of TRAC-PF1/ MODI. The large-break-LOCA assessment focuses on a test in the loss-of-Fluid Test (LOFT) facility. which simulates the geome-try of a PWR with U-tube steam generators: we previously had analyzed this test with

! TRAC-PD2/ MOD 1. The small-break-LOCA assessment relies on tests in the Once-Through i Integral System (OTIS) facility, which simulates the geometry and behavior of a Babcock and Wilcox (B&W) raised-loop PWR with once-through steam generators (OTSGs).

We performed the bulk of the code assessment for large-break LOCAs with TRAC-PD2/ MOD 1. In this current case we analyzed the LOFT LP-02-6 transient with both code versions in an effort to benchmark TRAC-PF1/ MOD 1 against both data and TRAC-PD2/ MOD 1. Both versions performed very well in calculating the hydraulic behavior of the transient, but could not calculate the early core quenching in the higher powered re-gions during the blowdown phase of the transient. Results from the two code versions are comparable: TRAC PF1/ MOD 1 is the preferred version because of its increased generality and improved robustness.

This work was funded by the US Nuclear Regulatory Commission. Office of Nuclear o';egulatory Research. Division of Accident Evaluation.

21-11 l

The two OTIS tests that we used investigate small break LOCA behavior under conditions that are dominated by gravitational heads, natural circulation and flow interruption. and steam-generator heat transfer. During this assessment. we have developed new noding techniques for use in TRAC to represent OTSG heat transfer better during high-clevation auxiliary-feedwater injection. For these tests, this improved representation of OTSG is important because the heat-transfer distribution in the OTSG affects the calculated natural-circulation flow. We have demonstrated that TRAC-PF1/ MODI can calculate the boiler-condenser mode of heat transfer which is the OTSG equivalent of reflux cooling in the U-tube steam generators. We also have shown that the code correctly calculates the system-pressure response and inventory distribution into the primary-system refill-phase of the transient. To calculate inventory distri-bution and system pressure, the code had to calculate a very complex mode of energy removal from the core simulator. This energy-removal mechanism relies on the mixing of highly sub-cooled emergency-core coolant injected into the cold leg with the high-enthalpy fluid flowing through the reactor-vessel vent valves.

Results of the assessment activities described here support the application of TRAC-PF1/ MODI to both large- and small-break LOCA analyses. We also have demon-strated that the code can calculate phenomena that are important to the B&W plants. Results indicate that TRAC-PF1/ MOD 1 is a useful best-estimate calculational tool to support any revisions contemplated in the reactor licensing process for PWRs and to resolve regulatory questions.

i i

l l

P 1

21-12

ASSESSMENT OF TRAC-BD1/M001 USING FIST DATA

i Jae H. Jo and H. R. Connell Department of Nuclear Energy Brookhaven National Laboratory Upton, New York 11973 The TRAC-BD1/riODI code [1], developed at Idaho National Engineering Labora-tory (INEL) fcc BWR accident and transient analysis, has been assessed at Brook-haven Nathnal Laboratory (BNL) using some of the FIST Phase I experiments. The code version used in this study was A22 plus the correction set E1Z1.

The FIST (Full Integral Simulation Test) facility [2] is a BWR safety test facility which was built to investigate small break LOCA and operational tran-sients in BWRs and to complement earlier large break LOCA test results from TI.TA (Two-Loop Test Apparatus). The FIST program is sponsored jointly by the NRC, Electric Power Research Institute (EPRI) and General Electric Company (GE).

The facility is a full BWR height, integral test facility with volume scal-ing of 1/624 to the BWR/6 vessel and contains a single full-size BWR fuel bundle (electrically heated). It has all the prototypical components of a BWR/6. The flow areas and the fluid volumes in all regions are also closely scaled to 1/624. However, because of scaling difficulty, the test facility has a cylin-drical external downcomer connected to the main vessel.

The FIST tests consist of two phases. The Phase I tests [3] were completed ~

in 1983, and the Phase II in early 1985. In this study the TRAC-BD1/M001 code has been assessed with the Phase I tests. The FIST Phase I test matrix consists of eight tests. Five of these tests were selected to be simulated. These were:

a BWR/4 MSIV closure ATWS (Test 4PMC1), a BWR/6 small break LOCA without HPCS (6SB2C), a BWR/6 large break LOCA (6DBA1B), a BWR/6 small break LOCA with stuck open SRV (6SB1) and a BWR/6 main steam line break test (6MSB1).

The " VESSEL" component of TRAC-BD1/ MOD 1 was used to represent the FIST facility. The VESSEL was nodalized with 12 axial levels, 2 radial rings, and 2 azimuthal sectors. Three different steady states were cbtained because Tests 6SB2C and 6DBA1B, and Tests 6SBl and 6MSB1 started from the same initial condi-tions, respectively.

Test 4PMC1 was a power transient simulation test for a BWR/4 with MSIV clo-sure and without power scram. The transient calculation of this test was termi-nated at 400 seconds since all the significant events occurred during this peri-od and the rest of the transient was predictable. The calculated results were generally in good agreement with the test data, particularly the pressure, steam line mass flow rate, downcomer collapsed water level, and the frequency of SRV opening and closing. The coru was always covered and no rod heat-up was ob-served in the calculation as well as in the test.

Work performed under the auspices of the U. S. Nuclear Regulatory Commission.

21-13

Test 6SB2C was a small break test, simulating a BWR/6 recirculation line 2

break of 0.05 ft with High Pressure Core Spray (HPCS) assumed to be unavaila .

ble. The MSIV was tripped when the downcomer water level reached " Level 1" and the Automatic Depressurization System (ADS) was activated with a 120 second de-lay. The transient was calculated up to 450 seconds. The results showed gener-ally good agreement with the test data. However, in the calculation, the Level l

1 was reached about 10 seconds later and the depressurization was slightly slow-er after ADS activation than in the test. This resulted in about 30 seconds de-lay in Low Pressure Core Spray (LPCS) and Low Pressure Core Injection (LPCI) initiations.

Test 6DBA1B was a large break test with a 200% recirculation line break for l a BWR/6. All the ECCs were assu.aA to be available. The calculation was run up to 120 seconds. The preliminary results indicated somewhat slower depressuriza-tion in the calculation than in the test.

Two more transient calculations for Tests 6SBl and 6MSB1 are in progress.

It appeared that the TRAC-BD1/ MODI code adequately predicted some of the FIST tests. However, the code did not appear to be completely robust numerical-ly as manifested by occasional failures and need for restarting with small time steps. The code also needed some manipulation for geometric data such as cell length, area and/or hydraulic diameter around the " VALVE" components, which were used to simulate breaks and SRVs, to avoid taking excessively small time steps due to the material Courant limit. This difficulty was caused by the semiimpli-cit numerical scheme used in the code and is expected to be eliminated in the new code version with an implicit numerical method.

REFERENCES

1. Taylor, D.D. , et al . , " TRAC-BD1/M001: An Advanced Computer Program for Boil-

! ing Water Reactor Transient Analysis," NUREG/CR-3633, April 1984.

2. Stephens, A.G., "BWR Full Integral Simulation Test Program," Contract No.

NRC-4-76-215, NUREG/CR-2576, December 1982.

! 3. Hwang, W.S., et al., "BWR Full Integral Simulation Test (FIST) Phase I Test i

Results," Contract No. NRC-4-76-215, NUREG/CR-3711, November 1983.

l i

l 21-14 l

RELAP5/ MOD 2 ASSESSMENT AT INEL P. D. Wheatley M. A. Bolander

R. Chambers J. M. Cozzuol C. B. Davis J. L. Steiner i

Idaho National Engineering Laboratory EG&G Idaho, Inc.

s P. O. Box 1625 Idaho Falls, Idaho 83415 l

Independent assessment of the RELAPS code continued with the assessment of RELAP5/ MOD 2 during the past year. RELAPS was l

! assessed using a range of integral and separate effect data.

Semiscale tests S-UT-8, S-UT-6 and S-PL-4 simulating small break transients were used for assessment. GERDA test 1605AA and a Model Boiler (MB)-2 test were also used. The crossflow junction capability in RELAPS was assessed using the Electric Power

Research Institute's (EPRI) single-phase liquid subchannel l blockage test data. International Standard Problem (ISP) 18 was

! also reviewed as part of the assessment of RELAPS/ MOD 2 though .

! the actual calculation was performed in support of an ISP-18 submittal.

Results of the independent assessments indicated that l

RELAPS/ MOD 2 was capable of simulating the range of transients

selected. RELAP5 was able to simulate crossflow or two j dimensional behavior with the addition of the crossflow junction

. model. Code simulations of the EPRI single-phase blockage data i showed very good comparisons. Nine of the blockage tests were simulated, including both smooth and plate type blockages.

Results from the Semiscale tests indicated that RELAP5 was

{

able to calculate the system behavior in general. The 4

1 21-15

_. _ . _ _ _ _ . _ _.- . _. .- _.__m . _ _ . _ _ .

4 l

l 4 calculated behavior for tests S-UT-6 and S-PL-4 was very similar j to the experimental results. Liquid hold-up during the early pcrtions of S-UT-8 was underpredicted. However, the remainder f of the S-UT-8 calculation was similar to the experiment.

! RELAP5/ MOD 2 did not calculate the core heat up during the latter I

I portion of the Semiscale transients. Accumulator injection l started before sufficient liquid had been removed during the-l core boil-off period to cause core heat-up.

i l The physical phenomena observed in the GERDA test 1605AA ,

also occurred in the calculation. The magnitudes of the l ,

3 phenomena were generally well calculated by RELAP5/ MOD 2; in ,

particular, the magnitude of the primary and secondary pressure ;

changes were calculated accurately, as was the minimum liquid level in the vessel. ISP-18 and the MB-2 loss-of-feedwater test i i-1 also showed good agreement with the experimental data. '

i 4

l t

f

i

1 i

r i

r f

i i

I l I l

21-16

STATUS OF TRAC-PF1/ MOD 1 INDEPENDENT ASSESSMENT AT SANDIA L. D. Buxton and L. N. Kmetyk Sandia National Laboratories Albuquerque, New Mexico Sandia's TRAC-PF1/ MOD 1 independent assessment program is part of a multi-faceted effort sponsored by the NRC to determine the ability of various systems codes to predict the detailed thermal / hydraulic response of LWRs during accident and off-normal conditions. This program is a successor to the RELAPS/ MOD 1 independent assessment project previously conducted at Sandia.

TRAC-PFl/ MOD.1 is being assessed against test data f rom various integral and separate effects experimental facilities. The calculated results will also be compared with results from our previous RELAPS/ MOD 1 independent assessment analyses whenever possible. Our TRAC-PF1/ MOD 1 matrix includes:

-- LOFT large break test L2-5,

-- LOFT loss-of-feedwater test LP-FW-1,

-- Semiscale Mod-2A intermediate break test S-IB-3,

-- Semiscale Mod-2A feedwater line break test S-SF-3,

-- Semiscale Mod-2A steam line break test S-SF-5,

-- Semiscale Mod-2B loss-of-power test S-PL-3,

-- two Semiscale Mod-2B steam generator tube rupture tests,

-- PKL natural circulation test series ID1,

-- LOBI large break test Al-04R,

-- LOBI intermediate break test B-R1M,

-- FLECHT SEASET natural circulation test 8,

-- B&W OTSG steady state test 28 and loss-of-feedwater test 29

-- NEPTUNUS pressurizer test YOS,

-- FLECHT SEASET reflood tests 31504 and 31701,

-- a Dartmouth University 3-tube CCFL test,

-- Northwestern University horizontal stratified cocurrent condensing flow tests, and

-- Northwestern University perforated plate CCFL tests.

Analyses for the PKL natural circulation, NEPTUNUS pressurizer, B&W OTSG and Northwestern University condensing flow tests were reported at last year's WRSRIM [1]; this paper will present the results of more recent work.

For LOBI Al-04R, we saw good agreement with measured system pressure, break flow and accumulator injection; however, the calculated blowdown PCT was low, probably because TRAC cannot model the LOBI rod geometry exactly. The TRAC results were generally comparable to our previous RELAPS/ MOD 1 results for this

This work was supported by the US Nuclear Regulatory Commission, Office of Nuclear Reactor Regulation, and performed at Sandia National Laboratories which is operated by the US Department of Energy under Contract Number DE-AC04-76DP00789 21-17

same test, with RELAPS doing slightly better during blowdown and TRAC slightly better during refill /reflood. Similar primary pressures and break flows were calculated by both codes, with RELAP5's PCT and TRAC's accumulator injection in better agreement with data, respectively.

In the LOBI B-RlM transient, we also saw generally good agreement with measured system depressurization, break flow and core temperatures, but some details of early-time broken loop flow behavior were not calculated correctly.

Qualitative dif ferences between B-R1M and its counterpart Semiscale test, S-IB-3, were correctly predicted. We did, however, see too little core heat-up in the S-IB-3 calculation due to problems predicting the pump head degradation and the loop seal clearing phenomena. Another important result was that significantly worse core and vessel refill response was calculated for S-IB-3 when a purely 1-D model was used instead of a 3-D VESSEL component.

The qualitative behavior of both the Semiscale S-SF-3 and S-SF-5 transients was calculated reasonably well. The predicted steam generator heat transfer degradation began too early for S-SF-3, possibly due to overprediction of liquid entrainment in the boiler and a resulting incorrect secondary side fluid distribution. The calculated primary-side fluid temperature response for S-SF-5 was in reasonable agreement with data, but the calculated pressure response was not. Data uncertainties prevent us from determining the exact cause of this S-SF-5 discrepancy, but there is indirect evidence that the calculated rate of phase change in the pressurizer was incorrect.

Calculations for the Northwestern University perforated-plate CCFL tests show that, for high water flow rates, TRAC overpredicts the steam flow rate needed for complete CCFL; however, for flow conditions typical of PWR transients, TRAC provides a reasonable prediction of CCFL. These analyses also showed nodalization and time step sensitivities.

The results of the Dartmouth CCFL calculations indicate that TRAC was able to predict the qualitative aspects of the four different annular flow regimes observed, in which different pressure-loss mechanisms dominate. However, some i discrepancies were discovered involving the transitions between the different flow regimes. The CCFL predictions were sensitive to the number of cells and the types of compt. nts used, and to the choice of friction factor options.

l Several coding, documentation and modelling inadequacies have been identified in the course of our TRAC-PFl/ MODI assessment calculations; they are being or have already been addressed by the code developers at LANL.

i REFERENCES

1. L. D. Buxton et al., " TRAC-PF1/ MODI Independent Assessment at Sandia l National Laboratories," at Twelfth Water Reactor Safety Research Inforaation Meeting, October 22-26, 1984, published in Proceedings, NUREC/CP-0058, Vol. 2 of 6, January 1985.

l 21-18

i l

SAFETY RESEARCH IN TRANSITION:

FROM ACCIDENT DESCRIPTION TO ACCIDENT PREVENTION l W. B. Loewenstein R. B . Du f fy l

l Safety Technology Department l Electric Power Research Institute A review of research underway in EPRI programs is given, with particular j emphasis on recent results.

l There have been clear changes in safety research and direction, and this

~

is becoming more apparent as the technology and results are focused on power plant applications. Traditionally, safety research has emphasized detailed analysis and description of reactor accidents, from defined transients to design basis accidents (LOCA), and events beyond the design basis. The tools to perform such analyses are now approaching a mature state, and the technology is in wide use in the utility industry.

Significant safety and operation margins are perceived for design basis events.

The extension to beyond design basis events, including degraded cores, shows that the dominant interest is in the potential radioactive source term. Siting and emergency planning aspects are the dominant implications.

What the analyses show is that there is great opportunity and potential to avert significant accidents and consequences by preventing accidents. This also reduces the economic exposure of the industry, and reduces significant plant events.

The ways this can be achieved are many, and in marked contrast to traditional safety analysis:

o reduction in risk due to seismic events, by realistic definition of the margins inherent in plant design o recognition and removal of common mode events as potential j accident initiators j o emphasis on reliability centered maintenance with a thorough knowledge of plant and equipment status o reduction in safety challenges and trips, by upgrading control and safety systems, where demonstrated to be needed I

i, 22-1

. - - - . _ . , - .. . ~ . . , . -_ _ - . - . . _ . . _ , - - _ , _ . . . - . ..____._.-__._.__z._

i i

o improved accident and transient diagnosis by operator information systems and signal validation, including the potential of utilizing validated expert systems o enhanced transient control by integrating emergency guidelines, and by quantifying the influence of operator actions o examining plant recovery strategies for all stages of accidents

! and transients o utilizing source term studies to quantify those transients and situations of major risk, and the areas of potential risk reduction.

The safety research underway in EPRI is examined in the context of this l transitional phase and revised objectives, and in how the application to i U.S. plants is being explored.

I j

{

l t

4 i

i i

l 4

}

22-2

Resolution of Steam Generator Integrity Questions J. F. Lang S. P . Kai ra Electric Power Research Institute 1

Background

The principal safety question regarding steam generator; is the effect on public health and safety of steam generator tube rupture events. Most of the research sponsored by EPRI and the Steam Generator Owners Group addresses steam generator reliability and thus the detection, correction, and long term 1

prevention of conditions that could lead to tube rupture. While EPRI results t

in the reliability area decrease the probability of steam generator tube ruptures, this paper discusses EPRI work to define and minimize the impact of steam generator tube ruptures should they occur. The principal topics are leak-before-break, effect on core cooling, and radiation release.

Leak-Before-Break While there are types of defects that h. > ruptured before they leaked (e.g.,

large volume defects), they are types ai.a able to detection and sizing with nondestructive inspection. EPRI's leak-before-break analysis has therefore focussed on cracks which are more difficult to detect and size by nondestructive inspection. Generalized methods have been developed to calculate critical crack sizes and primary-to-secondary leak rates for steam generator tubes. Considering typical tube sizes, crack locations, loadings and boundary conditions for once-through and recirculating steam generators, leak-before-break margins have been calculated for both types of steam generators.

Core Cooling Steam generator tube ruptures that have occurred in the Prairie Island-1 and 1 Ginna plants and steam generator tube rupture experiments at the SEMISCALE and Model Boiler (MB-2) facilities have provided data for verification of the systems thermal and hydraulic codes RETRAN and RELAP. These codes, in turn, have been applied to calculate the responses of plants to postulated steam generator tube rupture events. RELAP5 analysis for a range of single and multiple tube rupture transients in a B&W design plant demonstrated that current procedures ensured core cooling for all conditions analyzed and

. further demonstrated the value of main coolant pump operation during steam j generator tube rupture transients. Similar RETRAN analysis currently underway for a plant with recirculating steam generators is yielding similar results.

j Experiments in MB-2 have also provided thermal hydraulic data on other conditions beyond the design basis, conditions such as a stuck open relief valve in conjunction with a steam generator tube rupture.

22-3 1

. . _ . . - _ . - - _ , m, _ _ . _ . _ - . , _ , - , __ , - . . ..-, m . ~ ..7,, _

Radiation Release EPRI work has proceeded on the parallel paths of using existing information to develop analytical techniques for calculating radiation release during a steam generator tube rupture event and determining experimentally the processes by which radionuclides are transported through steam generators from primary-to-secondary leaks. Conservative analysis applied to once-through steam generators calculated average DFs several times greater than the values of 1 or 2 used in other analyses. The analysis techniques are being extended to recirculating steam generators. Experiments underway at Northwestern University should provide more accurate data on water scrubbing. Experiments at Oak Ridge National Laboratory should provide better data on the Iodine partition factor in the steam generator. Further, integral tests in the MB-2 model steam generator indicate that steam generator partition factors may be one to several orders of magnitude higher than assumed in current licensing analysis. The analyses and experiments both are pointing toward lower radiation release rates than previously calculated.

\

s fn

~

(, 22-4

SIGNAL VALIDATION: A NEW INDUSTRY TOOL S. M. Divakaruni and B.K.H. Sun Electric Power Research Institute I

l The signal validation approach used in EPRI project utilizes parity-space and analytic redundancy methods, and it emphasizes reliable fault isolation resolution to a few physically proximate elements in the shortest reasonable time. The philosophy underlying this approach is that if the operator is quickly given a reasonably short but exhaustive list of the possible causes of detected measurement inconsistencies, he is capable of making the final j

decision as to the ultimate cause. Of paramount importance is that the >

operator be reliably alerted to a problem area within the plant in the shortest practicable time.

In the current plants, the plant operating crew usually relies on judgment-4 based criteria for detecting signal failures, beyond the hardware redundancy implemented for selected safety-related signals. The operating crews at power plants have always performed some measure of manual signal validation traditionally. Redundant gauges are intercompared, and functionally diverse indications are correlated with operator's mental model of plant behavior. Today's powerful computers in the plants promise relief to the operators in this area by providing on-line signal validation automatically. -

The system-like SPDS can be made robust to yield a low false alarm rate.

High information reliability must be and can be assured at the front eno by the on-line signal validation methods.

Most common signal validation practices in the nuclear industry today are limited to a few rather simple techniques that rely, for the most part, on process symmetry characteristics and physical redundancy of observables.

These techniques include methods such as sensor comparisons, limit checking, auctioneering, instrument-loop integrity checking, and calibration checking.

In this paper, the general concepts used in EPRI signal validation projects are outlined and a brief report on each of the key projects is given.

EPRI is currently pursuing to develop the signal validation effort for key parameters supporting the SPDS functions using GPU's Oyster Creek as the reference plant. The major part of this effort is to validate the BWR sup-pression pool bulk temperature and the vessel level both inside and outside the core shroud, using the extended Kalman filtering approach. The Oyster i Creek suppression pool bulk temperature, vessel level and pressure signal validation procedures are described in the paper, as examples. The vari-ables used in the suppression pool and BWR vessel validation are combined into a single library of software modules that would provide the signal validation for the Oyster Creek plant and other operating BWRs.

22-5

l .C Working with nine utilities, EPRI developed a list of seventeen key variables for which the signal validation is necessary and that would support the critical safety functions used in the SPDS in the operating PWRs. The paper (

describes the validation routines for pressurizer and steam generator pressures, steam generator level and reactor coolant system hot leg temper-atures, and the results from failure tests performed on the routines. A library of seventeen modules developed for PWR variables, will be tested in Millstone-3 and M111 stone-2 plants.

It is expected that not only will a precise and robust sensor validation

., technique enhance plant safety, but it will also be cost beneficial as a i result of increased plant availability. The potential of the methods used in EPRI signal- validation project to identify failures of sensor and non-sensor plant components promises additional gains in safety and availability

! by permitting timely and accurate operator action to alleviate plant disturbances.

i

^

i i

i 22-6 l

On-Line Carrosion Cracking Monitor Development by J. D. Gilman and R. L. Jones Electric Power Research Institute

SUMMARY

Stress corrosion cracking of LWR components has caused significant forced outages with associated high costs for repairs or replacement of affected components. When cracks are discovered, plant owners must choose between repair or replacement alternatives, or they may elect to defer this decision on the basis of evidence that the plant is safe to operate. A reliable assessment of future crack enlargement is often necessary to support a deci-sion to defer repair or replacement. Similarly, if a component is repaired

! without removing the crack, a crack growth assessment is usually required.

EPRI is developing techniques capable of continuous on-line monitoring of known cracks in these situations, to augment current crack evaluation methods based on analysis and periodic nondestructive examination.

A crack-following technology developed at General Electric is being adapted to two related in-plant monitoring applications. In one case, a known crack in a piping component is instrumented and monitored as discussed above. In the second application, a sensor incorporating an instrumented, preloaded crack is exposed to the reactor operating environment. Data from the instrumented crack are used to predict the behavior of component cracks in like material and exposed to the same environment.

EPRI supported research beginning in 1981 to adapt a GE " reversing de poten-tial monitoring technique" to surface " thumbnail" cracks in laboratory test specimens. Electrical probes in the vicinity of the crack sense small voltage changes resulting from distortion of a uniform potential field by the crack.

A calibration model converts voltage differentials between various pairs of probes to an estimate of the crack size and shape. Several innovations and refinements make the system unusually stable and sensitive. Crack growth rates of the order of a few thousandths of an inch per year can be resolved by some operational laboratory systems.

To adapt the laboratory system to pipe crack monitoring, other developments are required. Electrical probes must be attached to the outside surface rather than the cracked surface, resulting in a substantial reduction of sensitivity. However, data from laboratory specimens and from an instrumented 4-inch pipe indicate that adequate accuracy and sensitivity are possible with external probes, particularly when the crack is deep enough to be of signifi-cance to structural integrity.

Another development is needed for applcations to large pipe. It is rat practical to pass a uniform current down the pipe, as is done for 4-inch pipes, to create the potential field around the crack. A local current source 22-7

and sink must be used. An objective of EPRI research now is progress is to define the optimum number and locations of current probes and voltage probes, and to develop the logic and software for conversion of data to crack size and shape, in applications of the pipe crack monitor to large pipes in field service.

The calibration model for pipe crack monitoring characterizes the crack as the semi-ellipse that best fits the voltage readings. Cracks having well-defined length and depth should be adequately represented by this idealization. If the crack shape is very irregular, as it may be for some IGSCC pipe cracks, a source of error is introduced.

Some of the limitations in monitoring a piping component crack are overcome by introducing a stress corrosion cracking sensor into the reactur environment.

A sensor consisting of a standard fracture mechanics specimen of sensitized Type 304 stainless steel, instrumented with electrical probes, has been opera-tional for some time in the Dresden 2 plant. A separate loop circulates reactor water to an autoclave housing the sensor. Measured crack growth rates reflect changes in plant water chemistry due to oxygen suppression by hydrogen water chemistry (HWC).

A more recently developed sensor of the double cantilever beam (DCB) type could be used in an autoclave, but it is also suitable for direct insertion at a location of interest in the reactor system. The DCB specimen is very sensi-tive to stress corrosion crack growth. It is designed to control the crack configaration and stress as the crack progresses, so that crack growth data can be interpreted quantitatively. Two useful applications of data from an in-plant corrosion cracking sensor are: (a) Plant operators may obtain an immediate indication of the effect of reactor water chemistry changes or transients on crack extension in components of like material. Transients of interest include loss of HWC and intrusion of impurities; and, (b) Data can be interpreted witn the aid of predictive models of SCC to estimate crack growth rates and service lifetime of specific components. Laboratory development and demonstration of an integrated monitoring and life prediction system are in progress.

22-8

Severe Accident Containment Integrity H.T. Tang Electric Power Research Institute 3412 Hillview Avenue Palo Alto, CA 94303 One aspect of source term consideration is the integrity of reactor contain-ment. The key question to be answered is will the containment fail given pressure and temperature histories associated with hypothetical, low orobabil-ity, degraded core accident scenarios. And if so, what is the ultimate fail-ure mode. This has significant ramifications. For if the failure mode were a sudden gross one, such as a sudden rupture of containment wall, the fission product would be released to the environment immediately. However, if the f ailure mode were some localized leakage which would lead to containment de-pressurization, the release of fission product would be gradual and limited.

To address this issue, a research program has been undertaken by EPRI at Construction Technologies Laboratories (CTL) and Anatech International, Inc.

The former has the testing responsibility, and the latter has the scope of de-veloping an analytical model for failure mode prediction.

In the first phase of the testing program, concrete slabs representing seg-ments from reinforced and prestressed containment walls were tested under uni-axial and biaxial tensile loading to understand prestressed and reinforced concrete deformation behavior. The slabs developed discrete cracks and stretched as much as 2%, which is equivalent to an increase in the containment building diameter of about one meter. Also tested in the first phase were segments of steel liner which, in a typical concrete containment, is anchored to the containment inner surface to serve as a leak tight membrane. The liner plates, containing butt welds or pipe penetrations, withstood up to 6% elonga-tion without rupturing. Data from selected tests in this phase of testing have been used to benchmark the nonlinear, structural ABAQUS-EPGEN computer code.

The second phase of testing addresses specifically the failure mode charac-terization. A large reaction rig was fabricated at CTL. Within this rig, deformation and leak tests of full-scale containment wall segments of various configurations were planned. At the present, four tests have been completed.

All specimens tested had the steel liner anchored.

Out of the four tests, two are of particular significance. The first is a full-scale wall segment representing prestressed secondary containment build-ing design. A manufactured crack in the liner plate was included along the weld seam to act as a " calibrated" leak path. The specimen was loaded to 80%

of reinforcement ultimate. At the end of the test, it was found that the original six-inch-long liner plate crack extended to a length of nine inches and the width of the crack changed from zero to three-eighths inch. This liner tearing occurred in a very controlled manner indicating the significant 22-9

~

interaction between the linear and concrete. The controlled tearing contra-dicts the hypothesis of uncontrolled liner rupture. Leak rate informction was also obtained up to liner plate crack width of 0.1054 inch beyond which the air supply system was exhausted to give any meaningful data.

The second is a full-scale prestressed wall segment with a large pipe penetra-tion sleeve located at the center of the specimen. Both biaxial tensile load and outward punching shear were applied to the specimen. The final applica-tfon of punching shear was to a total force of 897 kips at the time when the maximum average strain in the hoop reinforcing was 2.5%. Upon unloading of the shear load, the liner developed a crack along one side of the penetration that completely depressurized the air system for leakage monitoring. The crack was 16 inches long. The exact sequence of crack initiation and propaga-tion is being studied.

The results of these two specimens show that considerable interaction between liner and concrete exists. This interaction has strong bearing to define

" failure" mode of lined reinforce concrete containments. It demonstrates that

" leak-before-break" is possible for a lined concrete containment. Although the liner plate is significantly more ductile than the reinforcing, liner crack can occur prior to rupture of the reinforcing.

22-10 ,

EFFECTS OF THERMAL CONVECTION IN POSTULATED HIGH PRESSURE ACCIDENTS IN PWRs B. R. Sehgal Electric Power Research Institute The consequences of the postulated PWR high pressure scenarios, e.g.,

TMLB' and 2S D (small LOCA with failure of ECC injection system), have been evaluated with the NRC code MARCH (1) and the IDCOR code MAAP (2) for the I core heat-up and degradation calculation. These codes mainly employ the "once-through" forced convection flow modeling, in which the steam and hydrogen generated flow through the core, PWR vessel, piping and compo-nents, in a uni-directional fashion. The MAAP code transports the fission products generated during the core heat-up and deposits them on the piping and the components, which increase in temperature due to the fission product self-heating. Some natural convection flow is modeled in the MAAP code but does not include convection through the core prior to vessel breach.

The "once-through" convective flows may not be valid during the core heat-up and degradation phase when there is a radial variation of the steam heat-up and hydrogen generation in the core. Buoyancy-driven natural convective flows develop between the core and the upper plenum, the upper plenum and the dome, and the upper plenum and the rest of the primary system loops.

These flows distribute the heat generated in the core to the upper parts of the PWR vessel and the primary coolant system. These flows, therefore, af'fect (a) the magnitude and the rate of hydrogen generation, (b) the elapsed products intime before the BWR startcoolant primary of coresystem melting, ((c) the transport of fissionPCS), (d) th the PCS piping and components, (e) the revaporization of the fission products from the PCS surfaces, and lastly (f) the course of high pressure accidents.

Inclusion of natural convection flow modeling, and determination of its effect on the parameters mentioned above, is the objective of the recent analytical work with the CORMLT code (3, 4) and of the recent experimental work at Westinghouse. Of particular iiiiportance is the temperature history of the PWR PCS to assess if, for the risk-dominant high pressure scenarios, a local failure in the PCS may occur before the postulated vessel melt-through and radically alter the consequences of such accidents.

Some initial results for the temperature histories of PCS piping and The pressurizer gas obtained from the CORMLT code are shown in Figure 1.

lowermost curve set suggests that the mechanical integrity of the vessel discharge nozzles may be threatened by thermal stresses in this (rather thick) heat structure. Also shown in the figure are temperatures of surge-line piping and of exit gas from the pressurizer. If concern is restricted to times prior to core degradation (say, %3000s) for which geometrical 22-11 m

i complications in flow modeling have yet to occur, it is seen that pipe-wall temperatures are still so high [ approaching 1260K (18000F)] as to question mechanical survivability. (At 18000F, the yield stress of steel is less than 20% of its value at normal operating temperatures.) Also of interest are the estimates of gas-discharge temperatures from the pressurizer; at these temperature levels, it is doubtful that the (spring-loaded) safety valves will operate as designed.

The work at Westinghouse R&D Laboratories involves simulated accident condi-tions in a 1/7th linear-scale-replica of a Westinghouse PWR core, upper plenum, piping and steam generators. Experiments using low pressure water and SF6 gas have been performed. Visual observations of flow patterns and motion picture records were obtained and core heater watt-meter data were digitally recorded and stored at selected intervals. A two-color, laser-doppler anemometer system with traversing optics was used to obtain velocities in the front row of fuel assemblies, and in line with or in the lane behind the front row of guide tubes, and support tubes in the upper plenum (Fig-ure 2).

Argonne National Laboratory uses the COMMIX-1A code (5), a three-dimensional fluid flow code, to analyze and pre-predict the data measured in the West-inghouse testc. The ultimate objective is to obtain benchmark predictions of the temperature histories, in a few prototypic postulated accidents, and to use the results to normalize the predictions made with the CORMLT code for relevant part of the in-vessel accident progression. This code is de-scribed later in this paper.

The experimental work described above is continuing; however, it has pro-gressed to the point that some conclusions can be drawn. There are ample data to show that natural convection flows will play a dominant role in the early part of the postulated high pressure severe accidents in PWRs.

The analytical developments in the CORMLT code, and their application to the postulated TMLB' and S 2 D scenarios, have shown that a substantial part of the heat generated in the core will be transferred to the primary system during the early part of these accidents. Oxidation of zircaloy cladding, and initiation of melting in the core, may be delayed substantially to provide the operator with more time to correct the faulted conditions, which lead to the postulated accident. ;

The primary system calculated temperatures for the postulated TMLB' acci-dent achieved high values quite early in the accident. It is probable that local failures in the primary circuit to the pressurizer will occur. As pointed out earlier, this will alter the course of a high pressure accident and reduce the consequences, particularly with respect to containment challenge from direct heating loads.

References

1. R. O. Wooton and H. I. Avci, " MARCH (Meltdown Accident Response Char-acteristics) Code Description and User's Manual," NUREG/CR-1711 (1980).

22-12 o

2. Fauske and Associates, "MAAP, Modular Accident Analysis Program, User's Manual," Volumes 1 and 2, Draft Technical Report 16.2-3, Atomic Industrial i Forum, Washington, D. C. (1983).

3. V. E. Denny and B. R. Sehgal, " Analytical Predictions of Core Heat Up/

Liquifaction/ Slumping," Proc. Int. Mtg., LWR Severe Accident Evaluation (1983).

l 4. V. E. Denny, "The CORMLT Code for the Analysis of Degraded Core Accidents,"

EPRI NP-3767 CCM (1984).

t

5. W. T. Sha et al . , " COMMIX-1 A: A Three-Dimensional Transient Single Phase i Single Component Computer Program for Thermal Hydraulic- Analysis,"

i NUREG/CR-0785, ANL-77-96, Argonne National Laboratories (1978).

i I

a c

i i

j-

'l t

4 1

f 9

22-13

- y . ,- _ _ . . . . , _ _ . -.__,,f , , _ _,_._.c,, _

(f930)

(b'9 1) dW31 SV9 B3ZlBASS38d o o o o o o o o o M n - o

. . . . . . . i. g j l o g

- n' 1 s o 1n x

- LD a r -

u o u n .

N x -

o j m i _

o u - o r w>

9 g -

e 2-s m

s c> => . - -

w 4 * > EA

-u g

l -

a - o Ee "g

oz =o u s ox _ a3 9 3.mm mm-te e u

{ =

99 oW 13 EE x - O$ ."

t:

m- or

== -

31

= -- x x o Mo --

w

-a H a

a gw rs oo ce o_

W u

"E 5 E OS au W ., 2 G r0 ## .

r ur E= wr== == M Se Wo "E"t GG -

b to 3R WOtr == -

_ 4~

MOH 8 "d dS 3 m "S "

S _

o

'e 89e= a c-5 if -

_= c < s a b = s" Es l - 5:wr8.."* -

$5!!5EEE!**

-m a x

b2='m==8m== oo =E i -

l

_s-m.~~~m7"!,,,,I,,,,l,,o 8- , ,

o o o o o o o o o o o o n n -

e (d'930) (f'M )1dW31 3 Nil 3930S

! (I'M 1) 30lS dO1 dW31 3lZZON 22-14

_ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ - ._.5 , .-. _ _ . . . _ _

o g . u rew.s Upper Plenum J

Left N /. Right Hot leg I I\l \ l\ Hot Leg l p I' p d

a d'

J %,

A i \

w axt -<

3/F t ' ,,'*'

, , ' a, d a i, '

i'1' g

d' "

,, 1 oa o a, d, J Row f 4 1'

Core o

1' Vector Magnitude L0 cm/s Lower Plenum Fig. 2 : -Velocity vectors with 28 KW input into water and with steam generator connected 22-15

FISSION PRODUCT RETENTION IN BWR COMPONENTS M. Merilo Electric Power Research Institute To describe fission product transport and deposition in BWRs under conditions postulated to occur during severe accidents, it is important to understand the aerosol retention characteristics of the components where significant depo-sition is expected to take place. A set of experiments has consequently been initiated at Battelle Columbus Laboratories to measure the retention of aero-sols in the separator / dryer system, a system which was in fact designed to remove water droplets from the flowing steam prior to entering the steam turbines. To perform these experiments, G.E. made available to EPRI a full-scale separator and segment of a dryer, which had previously been used to obtain steam / water separation data under reactor operating conditions.

The current tests are being conducted under low flow conditions typical of those calculated to occur during severe accident sequences, but flows which are far removed from normal operation. To simplify the performance of the experiments most of the tests have utilized ambient air as the carrier gas, though several experiments with steam are also planned. Since these conditions are not totally prototypical of those expected, a scaling analysis was per-formed to try and achieve phenomenological similitude, a task which was simplified greatly by the exact geometrical representation.

To date, only the steam dryer has been tested using ambient air as the carrier gas with an inlet plenum velocity of about 6 cm/s. This corresponds to the condition where maximum retention was expected to occur.

Two types of aerosols covering three different sizes were used. To produce aerosols with a mass median diameter of 0.75 pm, liquid droplets of dioc-tylphthalate (D0P) tagged with sodium fluorescein were generated by atomiza-tion in a two-fluid nozzle followed by evaporation ar.d subsequent recondensa-tion. Aerosol particles with mass mean diameters of 5 and 10 pm were obtained by dispersion of precut AC test dust with a commercial dust feeder.

The experiments which have been performed to date indicate that the measured retention efficiencies are substantially higher than those predicted by methods currently used in severe accident analysis. The theoretical retention efficiencies range from 0.1% for 0.75 pm particles to 20% for 10 pm particles, whereas the measurements indicate retention efficiencies of approximately 50%,

independent of the particle size. The reasons for this discrepancy have not yet been determined; however, further analyses and testing at varying flow rates with different aerosols are being performed to elucidate this situation and to generate more applicable data.

22-17

THE UK MATERIALS AND NON-DESTRUCTIVE TESTING RESEARCH PROGRAMME FOR THE PWR by G J LLOYD AND R A MURGATROYD United Kingdom Atomic Energy Authority Risley Nuclear Power Development Laboratories Cheshire WA3 6AT England l

In the UK, a substantial research programme is being conducted in support of the PWR project in which CEGB, the UKAEA and the National Nuclear Corporation are collaborating closely. The purposes of the programme are to produce safety information for licensing assessments and to provide information which will, in the event, enable the PWR system to be introduced into the UK in a safe and economical manner. As part of the programme, collaboration with overseas organisations on PWR issues also occurs.

The research programme addressing these objectives is broadly divided into technical areas, one of the largent of which deals with Materials and Non-Destructive testing.

In the Materials area, studies have encompassed the ' validation' of materials properties, the long-term degradation of RPV materials, large-scale structural validation, environmentally assisted cracking and fault-related problems.

The purpose of the NDT programme is to ensure that techniques are available with the capability to reliably inspect the pressure containing components during both fabrication and operation. These include regions of complex geometry in the RPV, such as the nozzle-vessel weld, the inspection of underclad region in the RPV. Inspection techniques are also being developed for other components in the pressure circuit, including the pressuriser, steam generator and austenitic materials, such as the pipework and primary coolant pump casing.

The NDT programme incorporates theoretical modelling of the scattering of ultrasound by flaws and the propagation of ultrasound through anisotropic media. A substantial programme is proceeding aimed at developing methods of assessing, analysing and assuring the reliability of all stages of the ispection process. The techniques developed have been applied successfully international inspection exercises including the Defect Detection Trials and PISC II.

The paper to be presented will consider briefly the scope of each part of the materials and NDT programme and illustrate the contribution being made by reference to areas where research has led either to an improved understanding of PWR pressure vessel materials behaviour or to the development of effective flaw detection and sizing techniques.

Finally, a brief discussion of links with the overseas organisations and the future direction of the programme will be given.

23-1

"Degrad:d Piping Program - Phasa II Prcgress" l

by l G. M. Wilkowski, J. Ahmad, C. R. Barnes, D. Broek, F. Brust, D. Guerrieri, G. Kramer, M. Landow, C. W. Marschall, W. Maxey, M. Nakagaki, P. Scott, V. Papaspyropoulos, V. Pasupathi, and C. Popelar Battelle's Columbus Division Sumary The Degraded Piping Program began in March 1984. Its main objective is to verify elastic-plastic fracture mechanics and limit-load analyses for cracked nuclear piping. This program integrates advanced elastic-plastic fracture

, mechanics, detailed material property testing, and full-scale pipe fracture experiments. These efforts are combined to develop and assess relatively simple analyses methods.

Current related safety issues involve two specific concerns. The first is the i removal of pipe whip restraints, if leak-before-break can be shown. Current applications are specifically for systems not susceptible to waterhammer, IGSCC, or significant fatigue. The second safety issue is the assessment of cracks found in service. The verification of flaw acceptance criteria, such as the ASME Section XI IWB 3640 analysis, is an important aspect of this program.

Some important developments from the work to date are listed below.

(1) For through-wall cracked pipe in pure bending:

-A simple screening criteria was developed to show when limit-load analysis can be used and when elastic-plastic fracture mechanics is needed. The data showed that even high toughness pipes could fail at stresses below the limit load if the pipe diameter was sufficiently large. Conversely even low toughness materials may fail at limit load if the pipe diameter is small.

- The GE/EPRI estimation scheme was found to predict conservatively the loads at crack initiation and the maximum loads.

(2) For internal surface cracked pipe in pure bending:

- The pipe fracture data show that the limit-load is affected by the pipe R/t. Thin-wall pipe will ovalize reducing the load-carrying capacity.

An ovalization correction factor for the limit load is needed.

23-3

l l

- Experimental data have shown that several surface flawed carbon steel pipes have failed at stresses below limit load. A finite length surface cracked pipe J-estimation scheme is being developed.

l -A laboratory specimen was developed to determine the crack growth resistance in the through-the-thickness direction.

(3) Three experiments were conducted on stainless steel pipes with i prototypical weld overlay repairs. The experimental data showed that the l ASME section XI IWB 3640 analysis was non-conservative by 9 percent.

This assumed that the total thickness of the pipe and the weld overlay was accounted for in the nominal stress analysis for sizing the overlay.

Frequently only the pipe thickness is used to size the overlay. In this case, the ASME analysis method was conservative by 36 percent. The

deformations that occurred in these experiments were extremely large.

(4) Flux welds in carbon steel and stainless steel welds had initiation toughness values of less than one-tenth of the base metal toughness. The carbon steel welds failed in a " brittle" manner at 550 F (288 C).

(5) J-R curves for large amounts of crack growth were calculated from IT, 3T and 10T CT specimens for Type 304 stainless steel and A516 Grade 70 plates at 550 F (288 C). The IT CT specimen Jg-R curves could be extrapolated to 1 inch (25.4 mm) of crack growth without significant error.- Since only small specimens can be machined from pipes, this is important when evaluating the maximum load-carrying capacity of through-

, wall cracked pipe.

(6) Unlike JM , an incremental plasticity modification of J, called T*, tended toreachasteady-statevalueafterasmallamountofcrackgrow$h. This was consistent with crack tip opening angle data. If the T p fracture parameter proves fruitful, small specimen data could be extrapolated to very large amounts of crack growth.

Numerous other related technical issues were evaluated and are discussed in ,

3 greater detail in NRC NUREG/CR-4082 Volume 2. )

i l

I 23-4

PIPING RESEARCH IN THE FEDERAL REPUBLIC 0F GERMANY Karl Kussmaul Staatliche Materialpruefungsanstalt (MPA)

University of Stuttgart In 1977, a national R&D program (BV) was launched focussed on the generic issue of leak before break (LBB) of piping and pressure vessels. The goal was to investigate the fracture and crack opening process and to develop an unim-peachable deterministic safety concept. This concept allows the possibility of catastrophic failure to be completely excluded, without invoking prchabilistic arguments. Therefore, validation work of existing fracture mechanics approaches had to be performed on the basis of full scale testing under real operational conditions. Post analysis was done by maans of finite element modeling of fluid and structure interaction during crack propagation up to crack arrest. The BV activity has been divided into two phases. Phase one finalized in 1984 concen-trated on primary piping 790 mn 0.D., 45 mm wall thickness as used for German 1300 ME PWRs and on longitudinal defects. Phase two deals with circumferential defects. The full scale tests again under operational conditions of pressure and temperature include such tests with superimposed static and dynamic bending load.

A high capacity bending test rig has been constructed to provide a bending moment up to 10 MNm. The bending load will also be applied cyclicly for evaluation of pipe whip. In general, surface defects, through wall defects partly produced through fatigue and corrosion cracking will be investigated. Parallel to these efforts, German nuclear industry prepared their specific LBB approach. Within the scope of that R&D work, smaller diameters than for BV were used and also other materials including austenitic stainless steel. Within that framework a ,

new cooperative industry -- MPA program -- was started. The main features are to i generate a gross plastic strain concept for circumferentially welded safety-relevant nuclear piping made of ferritic, martensitic and austenitic stainless s teel . High speed impact tensile tests on flat specimens with defects will support the full scale tests on precracked or notched components partly subjected to impact bending loading.

It could be demonstrated that for adequate system design, the postulate of longitudinal circumferential large breaks (spontaneous catastrophic failure) is not justified. Failure probabilities are so low as to be meaningless. This conclusion is based on the use of high toughness materials not susceptible to degradation during manufacture, processing and operation, pre-service and in-service quality assurance programs. Load and leakage as well as water chemistry monitoring systems provide the necessary redundancies for the LBB concept.

23-5

y OUTLINE OF NUCLEAR PIPING RESEARCH CONDUCTED IN ITALY Pietro Paolo Milella ENEA-DISP Paper to be presented at the NRC 13th Water Reactor Research Safety Information Meeting Gaithersburg, Maryland October 22-25, 1985 ABSTRACT Following an early phase of limited activity at the University of PISA on small stainless steel pipes containing axial cracks, in 1981 ENEA, the Italian Committee for Research and Development of Nuclear Energy and Alternative Energies, has started a massive research campaign on fracture of carbon and stainless steel piping containing through and part-through cracks loaded either under pressure or in bending.

The purpose of the program was to develop a better understanding on pipes fracture behaviour in order to set new design criteria more realistic, jet conservative, than the guillottine break and prepare acceptance criteria for in-service flaws particularly under the growing pressure of IGSCC that has merciless affected worldwide practically any BWR piping system.

The analysis of more than 100 tests carried out at CISE research centre; in Milano, on 4 in, 6 in, 8 in and 10 in pipes has indicated that unstable fracture requires at least 150' through wall crack under ASME maximum design stresses.

The leak area, before instability tekes place, is always less than 10%

of the net cross section area of the pipe.

This has led ENEA to consider a 10% break area as a reascnable value to calculate jet forces.

Further, it was found that the net section collapse load criterion by far underestimates the actual collapse load and that 360* part-through cracks tend to switch from ductile to brittle the failure mode of a pipe loaded in bending.

Further work is planned for the next 3 years including high temperature tests, stainless steel weldments and HAZ tests, high compliance tests and eventually burst tests.

Besides the ENEA's research program, Ansaldo AMN, the Italian Nuclear Architect Engineer, is developing theoretical studies and codes ' to treat the problem of pipe fracture.

23-7

l l

Research Activities on Piping Fracture in Japan i

G. Yagawa*, Y. Takahashi"* and K. Kuwabara**

University of Tokyo

- Central Research Institute of Electric Power Industry Research on the pipe fracture started several years ago in Japan for the purpose of assessing the safety of the nuclear piping under the presence of the intergranular stress corrosion cracking. For this purpose, Crack Behavior Verification Test Program was managed by Nuclear Power Engineering Test Center (NUPEC) under the sponsorship of Ministry of International Trade and Industry from 1978 to 1984 to show the enough safety margin against unstable Tracture in type 304 stainless steel piping.

First, ductile fracture tests for 20 center cracked plates and 20 6-inch diameter pipes with a circumferential crack were performed using a tension-type compliant test apparatus, where dynamic as well as quasi-static loadings were applied to the specimens through a disk spring whose compliance is 10 mm/N. Unstable ductile fracture mode was observed in these tests. Secondly, leak-before-break tests were conducted for five 4-inch diameter pipes, where cyclic tension loading was applied through high compliant test apparatus with the constant internal pressure by hot water of BWR condition. The boundary between tne leak-before-break and the break-before-leak was clarified by this test. Thirdly, the thermal shock tests were carried out to demonstrate the integrity of the piping under the thermal transient due to tne operation of ECCS, where no crack growth was observed in all the specimens tested.

Based on the results of these tests, applicability of some fracture criteria was studied. The effectiveness of the net-section stress criterion was shown for the estimations of maximum load and leak-before-break condition. Detailed numerical calculations were performed in University of Tokyo and Central Research Institute of Electric Power Industry to make clear the applicability of nonlinear fracture mechanics parameter to pipe fractures. Three-dimensional finite element codes were developed for the purpose of crack growth analysis in piping system.

Geometry dependence of the fracture resistance curves of J-integral and crack tip opening angle were studied using these codes. Crack growth behavior from a part-through crack was also studied by the finite element method.

The Crack Behavior Verification Test Program for carbon steel piping started in 1985 and will finish in 1988, also managed by NUPEC. This program consists of (a) crack growth analysis, (b) material property and fracture behavior test, and (c) leak-before-break verification test. Crack growth analysis is based on fatigue crack growth and ductile crack growth analyses. It is planned that the former will be made with Paris' law while the elastic plastic fracture mechanics will be employed for the latter. In the material property test, tensile test, Charpy test, compact-tension specimen test and center-cracked panel tension test will be performed to obtain fundamental data of STS 42, STS 49 and weld metals. Fracture behavior test will be made for 6 and 16-inch diameter pipes containing a circumferential crack by using the compliant test apparatus. Bending load 23-9

l test will be performed by the modification of the test apparatus. Leak-before-break verification test consists of various component tests.

Specimens to be employed are straight pipe, elbow and tee joint pipes with a part-through crack. Based on these test results, evaluation will be made for the leak-before-break in the actual plant condition.

In Japan Atomic Energy Research Institute, pipe fracture tests under four point bending load are in progress. This program started in 1983 and will finish in 1987. The stainless steel and carbon steel pipes of 6-inch diameter with a through-wall or part-through circumferential crack were tested at ambient temperature or 285 C. Effectiveness of the net-section stress criterion and the J-integral / tearing instability criterion are being studied based on these test results. The J-integral resistance curves were obtained using experimental records and compared with others for type 304 stainless steel taken from the literature. It was made clear from the research that the net-section stress criterion gives the unconservative prediction for the stainless steel pipes with a deep surface crack.

Conducted by Hitachi Group are fracture tests using flat plates and 2-inch diameter pipes with a surface crack of type 304 stainless steel.

Based on the test results, study was made on the loads at the wall penetration and the final break. For the former, they proposed a new formula modifying that given by Battelle Columbus Laboratories and developed the leak-before-break assessment diagram based on their new formula.

Several analytical works are also being conducted in Japan. The development of elastic-plastic line spring model for the surface crack is under r,rogress in University of Tokyo. This method was applied to the problem of surface crack extension obtaining reasonable crack growth behavior. The three-dimensional fully plastic solutions are also obtained in University of Tokyo. To assess the accuracy of the numerical methods, a round-robin test for crack growth problem is being conducted in Japan Society of Mechanical Engineers.

23-10

FRACTURE ANALYSIS OF WELDED TYPE 304 STAINLESS STEEL PIPE USING LIMIT LOAD AND J-INTEGRAL TECHNIQUES R.A. Hays, M.G. Vassilaros, and J.P. Gudas David Taylor Naval Ship Research and Development Center Bethesds, MD 20084 13th WRSR Meeting, October 22-25, 1985 An experimental investigation was performed to evaluate the fracture performance of circumferentially welded type 304 stainless steel pipe. Tests were performed at 288 C (550 F) on 1219 mm (48 in.)

long, 4 in. NPS schedule 80 pipe specimens and compact tension specimens ranging in plan size from IT to 3T with thicknesses equal to the nominal pipe wall thickness. Flaws were machined into the weld centerlines and sharpened in fatigue prior to testing. The pipes were loaded in four point bending in a 136,000 kg (300,000 lb) capacity screw type testing machine under Jisplacement control in a manner similar to that of Vassilaros et a1 [1]. Five channels of data were taken during the tests including load, cross-head displacement, load line displacement, crack mouth opening displacement, and electrical potential across the crack mouth. Two crack geometries were investigated. These include a through-wall crack growing circumferentially (simple) and a through-wall crack growing circumferentially superimposed on a 360 degree radial flaw (complex).

The limit load expression used in evaluating these tests was that of Tada et a1 [2]. In this evaluation it is assumed that the cross-section containing the crack is fully yielded and that the material stress strain behavior is elastic perfectly plastic.

J-integral values were calculated using an expression due to Zahoor and Kanninen [3]. The expression requires actual load line displacement and bending moment data and contains a crack growth component. J-integral values for the compact specimens were calculated using the crack growth corrected deformation J expression published by Ernst et al [4].

Limit loads were calculated using both the ASME (3Sm) flow stress for type 304 stainless steel at 288 C which is 352 MPa (51 Ksi) and the average of the yield and ultimate strengths from tensile tests conducted at 288 C which was 373 MPa (54.1 Ksi). Good agreement between the limit load calculated using the ASME code flow stress was found for three of the five tests conducted on pipes with simple crack geometries. On average, the limit load predicted using the ASME code flow stress was 8%

higher than that actually attained by the pipe specimens during testing.

4 Results from tests conducted on pipes with complex crack geometries assuming no crack closure on the compressive side of the neutral axis and the ASME code flow stress were conservative in both cases.

Results from the J-integral resistance curve tests indicate that crack initiation occurs at a J level of approximately 1120 kJ/m2 (6400 in-lb/in2 ) for pipes containing simple flaws. A reduction in the J level at initiation of about a factor of four and a reduction in 23-11

apparent tearing resistance in pipes containing the complex flaw was observed due to the presence of the internal flaw. Good agreement in theJlevelatcrackinitiationbetweenthepipescontainingthesimple flaw and the compact specimens (J-initiation approximately 1050 kJ/m )

indicates that initiation toughness measurements on laboratory sized compact specimens may be applicable to pipe geometries.

REFERENCES

l. Vassilaros, M.G., R.A. Hays, J.P. Gudas, and J.A. Joyce, "J-Integral Tearing Instability Analyses for 8-Inch Diameter ASTM A106 Steel Pipe", U. S. Nuclear Regulatory Commission Report NUREG/CR-3740, April 1984.

2. Tada, H., P. Paris, and R. Gamble, "A Stability Analysis of Circumferential Cracks in Reactor Piping Systems", U.S. Nuclear Regulatory Commission Report NUREG/CR-0838, June 1979.

3. Zahoor, A., and M.F. Kanninen, "A Plastic Fracture Mechanics Prediction of Fracture Instability in a Circumferentially Cracked Pipe in Bending- Part I.: J-Integral Analysis", ASME Journal of Pressure L Vessel Technology, Vol.103, No. 4, November 1984. ,

4. Ernst, H.A., P. Paris, and J.D. Landes, " Estimations on J-Integral and Tearing Modulus T from a Single Specimen Test Record", ASTM 13th National Symposium on Fracture Mechanics, STP 743, Richard Roberts ed.,

1981.

l d

4 23-12

Piping Fracture Mechanics Data Base by A. L. Hiser '

Materials Engineering Associates, Inc.

9700-B George Palmer Highway Lanham, Md 20706 A comprehensive, computerized data base of fracture toughness (J-R curve) !

and other support data from nuclear piping steels is being established. This data base will be used for the materials input to assessments of piping integrity with known or assumed flaw and loading conditions.

An important consideration in assessing nuclear power plant safety is a determination of the structural integrity of the coolant water piping systems. Besides system-dependent concerns such as loading and flaw distributions, properties of the constituent materials such as tensile strength, fatigue crack growth rates and fracture toughness values are also required. The fracture toughness takes on special significance since it is used in determining whether or not a crack will propagate through thickness under specified loading scenarios, and additionally, whether or not a through-thickness crack will be stable or will result in a large break.

The development of fracture analysis procedures for LWR piping systems has passed through several stages. Currently, J-integral concepts are receiving considerable attention as a suitable tec hnique, with the J-R (or resistance) curve as the fracture toughness parameter required to perform such analyses. To facilitate such analyses, a computerized data base of J-R curves and other pertinent information on steels used in nuclear piping is being developed.

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 survey documented the ASME specification, as well as the pipe size and wall thickness.

Concurrent with this survey ac tivity , a format for the data base was established, with data input forms and any 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 suc h that future modifications to J-integral analysis methods can be accommodated in a relatively easy manner.

23-13 1

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 ha s centered on procurement of piping materials for additional testing, as required to fulfill the data matrix requirements.

This data base will be accessible to NRC personnel, contractors and other outside users, as required. Regular updates will be made as more data are generated at HEA and by others with close coordination of testing plans tied to other NRC-sponsored research programs.

e 23-14

Results of CCTF Tests Yoshio MURA0, Tadashi IGUCHI, Jun SUGIMOTO, Hajime AKIM0TO, Tsutomu OKUB0, Tsunetuki H0J0 Japan Atomic Energy Research Institute I

1. Introduction j A reflood test programIII for a large-break loss-of-coolant acci- I dent (LOCA) of pressuri:ed water reactor (PWR) has been conducted at Japan Atomic Energy Research Institute (JAERI), by using large scag Core Test Facility (CCTF) test facilities, which are the Cylindri and the Slab Core Test Facility (SCTF)g< 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 1000 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 2000 electrically-heated rods and four loops with passive and active components.

The main purpose of the CCTF tests is to investigate the integral system behavior as well as the core thermal hydraulic behavior during the refill and reflood phases of a PWR-LOCA. ,

Since 1979, JAERI has performed 56 CCTF tests. They can be classi- l fied into 5 categories. l (1) Cold leg injection simulation tests under evaluation model(EM) condition .

(2) Cold leg injection simulation tests for parametric effect 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 initia-tion The experimental work of JAERI for CCTF has completed at March, 1985. Currently, the analytical work is in progress.

The major findings upto the last year for the cold leg injection simulation tests are:

(1) The thermal-hydraulics in the primary system are nearly the same as the current EM models assumed in the safety evaluation analysis.

(2) The core cooling is much better than that predicted with the cur-rent EM model.

The work was performed under contract with the Atomic Energy Bureau of Science and Technology Agency of Japan.

24-1

It can be concluded that the current EM model is reasonable and it predicts the clad temperature conservatively during the reflood phase.

The JAERI's activity is mainly focused on alternative ECC simula-tion tests and refill simulation tests for this year.

In this presentation, the following topics are explained; (1) Refill test, (2) Upper plenum injection test and (3) Combined injection test.

2. Refill tests The pressure vessel was pressurjzed upto 0.6 MPa and the lower plenum was filled with saturated water to the specified level for the pre-conditioning of the test. By opening the cold-leg-break-simulation va; ves, the depressurization of the vessel was initiated and subsequen-tly ECC water was injected into cold legs. At the same time, the steam injection into upper plenum was initiated to simulate the steam genera-tion due to flushing in the upper head.

The test results indicated that (a) A oart of the steam injected into upper plenum flowed downward in the core and delayed the end of bypass.

(b) During the period of the depressurization, the downcomer CCFL conti-nued and the lower plenum filling was restrained. During this period, the core cooling due to steam flow was observed.

(c) At the termination of the depressurization, the lower plenumum was rapidly filled with water due to the fall back of the water held in the downcomer, and eventually the reflood initiated.

3. Upper plenum injection test A test simulating upper-plenum-injection-type PWR was performed with single failure assumption of LPCI pumps. The ECC water was injec-ted into the upper plenum asymmetrically through one injection nozzle located on the side wall of the upper plenum.

The test results showed that (a) The effective core cooling was observed even under the conditions of the single failure assumption of LPCI pumps.

(b) The asymmetric upper plenum injection gave rather better core coo-ling than the symmetric upper plenum injection.

(c) Asymmetric upper plenum injection resulted in extention of top-quench region and showed weak effect on quench velocity of bottom-quench region.

4. Combined injection tests Three tests were performed to simulate a combined-injection-type PWR in which the ECC water was injected into both cold and hot legs.

The test results showed that (a) The following three charachteristic thermal hydraulics were obs.er-ved:

(i) localized CCFL break-through at end box (ii) siginificant horizontal ununiformity of core cooling and upper plenum thermal hydraulics (iii) fluid oscillation in broken loop (b) The effective core cooling was observed either under EM condition or best estimate condition.

(c) Most ECC water injected into hot legs flew into upper plenum and then into core.

(d) Steam generated in core was completely condensed in upper plenum due to subcooled ECC water injected into hot legs.

24-2

RESULTS OF SCTF REFLOOD TESTS Takamichi IWAMURA, Makoto SOBAJIMA, Hiromichi ADACHI, 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 (Cylindrical 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.

The two-dimensional effects observed in the SCTF tests are t-1L_

classified into the following two individual ef fects :

(1) Effect of radial core power / temperature distribution a The heat transf er above the quench f ront is enhanced in the highar powe r/ tem pe ra tu re bundles and degraded in the lower powe r/ t empe r a tu re bundles and resultantly the turnaround temperature is reduced in the higher power / temperature bundles.

(2) Effect of non uniform water accumulation in the upper plenum As the collapsed water level in the upper plenum becomes higher in the hot leg side on the periphery than in the radial center side, the quench in the upper half of the core is delayed in the peripheral bundles.

In the SCTF tests, the radial core power / temperature distribution has more significant effect on the reduction of peak cladding temperature

~

than the non uniform water accumulation in the upper plenum. When the radial core power distribution existed, the radial rod temperature distribution was also accompanied in the previous tests and therefore the ef fects of core power and rod temperature distributions could not be The work was performed under contract with the Atomic Energy Bureau of g Science and Technology Agency of Japan.

24-3 '

distinguished from each other. In order to separately evaluate these two effects, four tests were performed with various combinations of radial core power and rod temperature distributions as follows :

Test number S2-12 S2-14 S2-15 S2-21 Core heating power distribution Steep Flat Steep Flat Initial rod temperature distribution Steep Flat Flat Steep

- Normalized power ratio : .l.0 (Bundles 1 & 2),1.2 (Bundles 3 & 4),

1.0 (Bundles 5 & 6) and 0.8 (Bundles 7 & 8)

- Nearly flat. That is, the temperature in Bundles 3 & 4 was slightly higher and the temperature in Bundles 7 & 8 was slightly lower than the average temperature.

These tests were performed under the forced flooding condition to make the core inlet flow rate the same. The water in the upper plenum was extracted in these tests so as to avoid the effect of non uniform water accumulation in the upper plenum.

Two-dimensional thermal-hydraulic behavior was clearly observed in Test S2-12, whereas the thermal-hydraulic behavior in Test S2-14 was almost one-dimensional over all bundles. The thermal-hydraulic behavior in Test S2-15 was similar to that in Test S2-14 during the initial l period and then approached to that in Test S2-12 as the radial temperature distribution was developed due to the steep radial power distribution. On the other hand, the thermal-hydraulic behavior in Test S2-21 was initially similar to that in Test S2-12 and then approached to that in Test S2-14 as the radial temperature distribution became flat due to the flat radial power distribution. Therefore, it was concluded that the radial temperature distribution which was induced by the radial power distribution was the important f actor of the two-dimensional thermal-hydraulic behavior in the core.

s l

l 24-4

TRAC ANALYSES FOR CCTF AND SCTF TESTS AND UPTF DESIGN /0PERATION*

by Jay W. Spore Code Development Group Energy Division Los Alamos National Laboratory Los Alamos, New Mexico 87545 The 2D/3D Program is a multinational (Germany, Japan, and the United States) experimental and analytical nuclear reactor safety research program.

Its main purpose is the investigation of multidimensional thermal-hydraulic behavior in large-scale experimental test facilities having hardware proto-typical of pressurized water reactors (PWRs). The Japanese are currently operating two large-scale test facilities as part of this program: the Cylindrical Core Test Facility (CCTF) and the Slab Core Test Facility (SCTF).

The CCTF is a 2000-electrically-heated-rod, cylindrical-core, four-loop facility with active steam generators primarily used for investigating inte-gral system reflood behavior. The SCTF is a 2000-electrically-heated-rod, slab-core (one fuel assembly wide, eight across, and full height), separate-effects reflood facility. Both facilities have prototypic power-to-volume ratios preserving full-scale elevations, and both are much larger than any existing facilities in the United States (including LOFT). The German con-tribution to the program is the Upper Plenum Test Facility (UPTF), a full-scale facility with vessel, four loops, and a steam-water core simulator which has been constructed in Mannheim, Germany. All of these facilities have more instruments than do any existing facilities: each has more than 1500 con-ventional instrumentation data channels, alone. As its contribution to the program, the United States is providing advanced two-phase flow instrumenta-tion and analytical support.

The los Alamos National Laboratory is the prime contractor to the United States Nuclear Regulatory Commission (NRC) in the latter activity. The main analytical tool in this program is the Transient Reactor Analysis Ccde ITRAF),

a best-estimate, multidimensional, nonequilibrium, thermal-hydraulics computer code developed for the US NRC at Los Alamos. Through code predictions of experimental results and calculations of PWR transients, TRAC provides the ,

analytical coupling among the facilities and extends the results to pre-dict actual PWR behavior.

During fiscal year 1985, TRAC-PFl/ MODI analyses were provided for seven CCTF-II experiments. Predictions of upper plenum injection (UPI) tests 57,

Work performed under that auspices of the US Nuclear Regulatory Comission.

24-5

72, 76, and 78 demonstrated that TRAC can predict correctly when UPI flows enhance core cooling and when they contribute to steam binding and degraded core cooling. Both the TRAC calculations and the experimental data indicated channeling of Emergency Core Cooling (ECC) water from the upper plenum into the core. The experimental data indicated asynmetric behavior in the core and upper plenum, therefore multidimensional analysis capability was required to accurately simulate the test behavior. In addition, TRAC was used to analyze nine SCTF experiments: base case for Core-II (Run 604), flat power and initial rod temperature profile (Run 605), steep power and initial rod tenperature profile (Run 611), FLECHT-SET coupling test (Run 613), best estimate base case (Run 614), separate effects countercurrent flow-limiting (CCFL) tests (Runs 608 and 610), and others. The analyses of these tests demonstrated that in general TRAC-PFl/M001 accurately simulates the reflood thermal-hydraulic behavior of the SCTF tests.

In support of the UPTF, three pre-test predictions were performed with TRAC-PFl/MODl: downcomer separate effects analyses, a German PWR base case analysis, and a small-break test analysis. From these analyses, initial and boundary conditions for the tests can be determined to ensure proper opera-tion of the test facility.

Finally, a fine-node 200% Loss-Of-Coolant Accident (LOCA1 calculation of a B&W 2772 MWt PWR, assuming licensing-type boundary and initial conditions, was completed. This calculation predicted a peak cladding temperature IPCT) of 995 K to occur in the average rod during b!owdown.

The Los Alamos analysis effort is functioning as a vital part of the PD/3D P rogram. Results from this program have already addressed, and will con-tinue to address, key licensing issues including scaling, multidimensional ef-fects, downcomer bypass and refill, reflood, steam binding, core blockages, al ternate emergency core-cooling systems, and code assessnent. The CCTF and SCTF analyses have demonstrated that TRAC-PFl/ MODI can correctly predict multi-dimensional, nonequilibrium behavior in large-scale facilities prototypical of actual PWRs. Through these and future TRAC analyses, the experimental findings can be related from facility to facility; and more important, the results of this multinational research program can be related directly to licensing concerns affecting actual PWRs.

l l

l 1

24-6 i

Status of the German UPTF Program K. R. Hofmann Gesellschaft fuer Reaktorsicherheit The Upper Plenum Test Facility (UPTF) sponsored by the Ministry for Research and Technology (BMFT) is the German contribution to the trilateral 20/30 Pro-ject. The objective of the 2D/3D Project, performed within international co-operation between Japan (JAERI), USA (USNRC) and the Federal Republic of Germany (BMFT) is the experimental and analytical investigation of the multi-dimensional flow behaviour in the primary cooling system of a PWR during a loss of coolant accident (LOCA). The effectiveness of emergency core cooling systems (ECCS) will be studied considering multidimensional flow effects in the reactor core, in the upper plenum and in the downcomer during the end of blowdown, refill and reflood phases of the LOCA with large and intermediate breaks. Data from large scale experimental facilities as CCTF (Cylindrical Core Test Fac11ity) and SCTF (Slab Core Test Facility) in Japan and from UPTF in Germany are being used to assess the TRAC computer code developed by USNRC.

' Emphasis is given to improve the computer code capability for best estimate analyses and to quantify the existing margins in safety analyses for LOCAs.

The UPTF simulates the primary cooling system of a KWU 1300MW 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. TRAC analyses and data of SCTF, where the core behaviour is studied, are used to specify the boundary conditions in order to create the required flow conditions at the core / upper plenum interface. The cross section of the core simulator, in the upper part consisting of 193 fuel element dummies, is subdivided into 17 injection zones where the injection and mixing of steam and water can be controlled indepen-dently. A total of 1500 kg/s water and 360 kg/s saturated steam is available for injection into the core simulator.

The three intact loops are equiped 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 separators and break valvcs into the containment simulator. Breaks of variable sizes can be simulated in the hot and in the cold leg respectively.

The containment simulator, with a volume of 1500 m3, is designed as pressure suppression system with steam injection capability in order to keep the back-pressure level according to realistic containment conditions. To maintain the mass balance in the UPTF system, appropriate drainage devices are installed.

The ECC system simulating accumulator and low pressure injection by using four pressurized storage tanks, is designed to inject into the cold and hot legs of 24-7

the loops and into the downcomer in any configuration according to the various reactor designs. Vent valves for the B&W/BBR reactor simulation can be acti-vated. The capability for nitrogen injection is also available.

Large amounts of steam and water needed to operate the UPTF are provided by a power plant and stored in supply tanks before the experiment.

Nearly 1200 measurement channels are being used to record the data from various kinds of instruments during the test. An extensive number of advanced irlstru-ments to measure two phase flow phenomena have been developed and provided by USNRC including the data acquisition system.

The construction of the UPTF which began in 1981 was recently completed. The first test is scheduled for the end of this year. The status of the current commissioning activities will be presented. A total of 30 experiments is planned including integral and separate effects tests. Approximately one half of the experiments will be specified to investigate the phenomena occuring during cold leg injection and the resulting downcomer behaviour, the other half will focus on the effects during combined hot and cold leg injection and the resulting upper plenum behaviour.

Within the German 2D/3D program the calibration of the UPTF tie plate instru-ments provided by USNRC has been performed. These measurements basically con-sisting of tie plate dragbodies, flow turbines and break through detectors play an important role to determine and control the boundary conditions at the core /

upper plenum interface. A single bundle test loop was used to calibrate the instruments for the various single and two-phase flow conditions. Appropriate algorithms for data evaluation and interpretation have been developed. The results of the calibration program will be presented.

To define the initial and boundary conditions for the experiments, evaluation of TRAC analyses for the reference reactor and the UPTF system are required. These analyses as well as the later test calculations are partly performed by LANL and the German contractors of BMFT.

24-8 i

l

i NUCLEAR PLANT ANALYZER DEVELOPMENT AT THE IDAHO NATIONAL ENGINEERING LABORATORY E. T. Laats R. J. Beelman T. R. Charlton l

N. L. Hampton i J. D. Burtt EG&G Idaho, Inc.

1 5 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-developedcomputercolorgraphicsprogramsand{agge An repositories of reactor design and experimental data * .

important feature of the NPA is the capability to allow an analyst to redirect a RELAP5 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 j actual plant. The NPA resides on the dual Control Data corporation Cyber 176 mainframe computers at the Idaho National Engineering Labortory and a Cray-lS computer at the Los Alamos National Laboratory (LANL).

During the past year, the NPA program at the INEL has addressed three areas: software development, advanced graphics development,'and user support.

The primary emphasis in software development has been the development of workstations capability for performing local data storage and replay functions. -The_same capabilities available on the-mainframe computer to control the replay of data (i.e.,

start,-stop, skip, pause).are now available ?.t the' workstation.

A second software development activity was the implementation of

Work supported by-the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research, under Department of Energy ~ Contract No. DE-AC07-76IDO1570.

25-l'

the Computer Visual System 3 to generate graphical display

" masks", thereby replacing the MAPPER syctcm and reducing the time and skill level required to build the masks.

The activities in the graphics development area have been somewhat experimental in nature. A study was conducted to investigate rudimentary animation capabilities that are possible through the workstation working with the mainframe computer. In particular, real-time animation using bubble movement and vectors were two techniques pursued. Graphics to represent 3-dimensional flow were also investigated.

The user support area included assisting various users (NRC and contractor) in establishing and maintaining workstations, and providing user training sessions with support materials 4.

Also, the RELAPS/ MOD 2 model of the H. B. Robinson Plant was l

modified to enable interactive execution and control through the NPA.

One last noteworthy activity was the publishing of a plan to develop the next generation " Type-3" workstation that could be completely standalone.3 That is, the NPA and the simulation codes would reside and operate on the workstation without requiring access to the mainframe computer.

The NPA has been used in FY-1985 to support key NRC tasks.

One particular application was the analysis of the June 9, 1985 incident at the Davis-Besse Plant.

Within two weeks after the event, a detailed RELAP5 model was used to replicate the event, and then extend it beyond the actual time when the event was mitigated. The NPA's capability to interactively " control the plant" enabled the NRC staff to overcome a problem of minimal measured plant data and still be able to duplicate plant response by perturbing plant control systems that produced the observed response. This simulation l capability enabled the NRC to evaluate the timing and consequences of other possible operator actions.

Other noteworthy uses of the NPA have been the evaluation of abnormal transient operator guidelines by the NRC's Division of i

Human Factor Safety, and the simulation of a stricken power j plant whose real-time data were provided the NRC Operations j Center during emergency preparedness exercises.

l The upcoming NPA program at the INEL will focus on production operation. The major emphasis will be to modify existing input decks to enable their' interactive execution through the NPA. Some software enhancements will be developed 25-2

to provide more flexibility and variety in graphically viewing data. User assistance will continue, with the goal to meet the

< needs of the expanding user community.

l l 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.

REFERENCES

1. K. D. Russell, et. al., Nuclear Plant Analyzer and Data Bank Common User Interface Functional Requirements, Conceptial Design, and Hardware Considerations, EGG-SAAM-6419, September 1983.

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

3. D. M. Snider and K. C. Wagner, Computer Visual System (CVS)

Reference Manual, EGG-IS-6478, January 1985.

4. E. T. Laats et. al., User's Manual for The U.S. Nuclear Regulatory Commission's Nuclear Plant Analyzer, EGG-NTP-6657, Rev 1, September 1985.

5. E. T. Laats et. al., Minicomputer-Based (" Type-3")

Workstation For The Nuclear Plant Analyzer, EGG-RST-6803, February 1985.

6. R. J. Beelman et. al., Nuclear Plant Simul & tion Using The Nuclear Plant Analyzer, Transaction of The International Conference on Power Plant Simulation, Cuernavaca, Mexico, November 1984.

25-3

.n

= ._.

THE LOS ALAMOS NUCLEAR PLANT ANALYZER

Thad D. Knight and Dennis R. Liles Safety Code Development Group Energy Division Los Alamos National Laboratory Los Alamos. NM 87545

SUMMARY

los Alamos National Laboratory, with Idaho National Engineering Laboratory (INEL). is developing the Nuclear Plant Analyzer (NPA). The Los Alamos NPA effort is divided into two l parts. The first involves conversion of INEL NPA executive programs to Cray computers, and the second involves development of the NPA hydrodynamic code to function more efficiently l .

with the NPA. We have continued to use the earlier, unreleased. Los Alamos-developed NPA i in our analyses of the Multiple-Loop Integral-System Test (MIST) facility, of the R. E. Ginna

~

steam-generator tube-rupture transient, and of the Davis Besse loss-of-feedwater transient.

The goal ot' NPA development is to provide an interactive-analysis capability for the avail-able thermal-hydraulic codes The executive programs, among other things. should incorporate similar commands to execute all available codes. Conversion of INEL executive programs is an ongoing effort to produce an NPA version on the Cray computers [Cray Time Sharing Sys-tem (CTSS) operating system] with the same interface to the user as the Cyber-1/6 version developed at INEL. This process should produce the first functioning version at Los Alamos j in early fiscal year 1986. We are removing much of the machine-dependent programming from this version, and the code should be reasonably easy to convert to other computers in the l

j future, we will modify the NPA to improve and enhance execution and user interface.

We are continuing development of the NPA hydrodynamic code for use with the NPA l

~

executive. This code, which currently is in the check-out stage of development, will provide several features to enhance its computing efficiency. A major advance will be application of the stability-enhancing two-step numerics (SETS) to the multidimensional VESSEL compo-nent. Multidimensional SETS. together with the one-dimensional SETS previously available in the Transient Reactor Analysis Code (TRAC-PF1/ MOD 1), will permit ths NPA code to take time steps larger than the material Courant limit, and should decrease significantly the 2 computer time required for long, slow transients. To promote calculational efficiency on vector-processing computers, we have incorporated significant vectorization in the programming. We have vectorized most of the coding specific to the VESSEL component and significant portions specific to the one-dimensional components. We also have added generalized heat structures based on fuel-rod modeling. This new generalized heat structure can connect thermally cells within the same component or cells in two different components. and includes cartesian or cylindrical two-dimensional conduction solutions. Because of the magnitude of the changes

This work.was funded by the US Nuclear Regulatory Commission. Office of Nuclear

Regulatory Research. Division of Accident Evaluation.

i 25-5 i

made to the code, we anticipate an extended period of checking and debugging will be re-quired: this phase will be followed by developmental assessment. Future work will focus on modifications necessary to gain the benefits of parallel processing and to utilize minicomputers.

We have continued to use our own, in-house. NPA prototype to aid the analysis and visu-alization of transients in the MIST facility and to support plant analyses under the calculational assistance program, in addition to the specific contribution made to these various analyses, this work has had a beneficial effect on the development of graphic displays of calculational results. and has generated experience working in an interactive mode. We are factoring this overall experience into our NPA-development activities.

i l

25-6

BWR PLANT ANALYZER DEVEL'PMENT AT BNL*

E. Cazzoli, H. S. Cheng, A. N. Mallen and W. Wulff Department of Nuclear Energy Brookhaven National Laboratory An engineering plant analyzer has been developed earlier [1] for realis-tically and accurately simulating transients and severe abnormal events in BWR power plants interactively at high simulation speeds, low costs and outstand-ing user conveniences. This plant analyzer has now been shown to be fully operational in interactive mode, from anywhere in the United States and Europe, using a readily available IBM Personal Computer and a commercial tele-phone line.

I The plant analyzer executes safety analyses ef ficiently. It could also be an outstanding tool for parametric studies, for optimizations of emergency procedures and for control system designs. It can also serve as the basis for developing ex?ert systems to support normal, abnormal and emergency opera-tions. Its modeling features and simulation capabilities are found in Refer-ence (2]. Figure 1 shows schematically the simulated system.

During the last year, the plant analyzer has been extended with remote access capabilities and the ability to simulate indefinitely long transients.

The remote access capability has been achieved by programming an IBM PC to control the PDP-il/34 host computer via standard MODEM and commercial tele-phone lines, and to display tabulated and graphical data transmitted from the AD10 peripheral processor via the PDP-ll/34 host computer. All operator actions and malfunctions can be entered from the keyboard without interrupting a simulation. Geometric parameters, operating conditions, control systems parameters and safety system trip setpoints can be changed from the keyboard.

Two selected parameters can be displayed in labeled, two-color graphs during the simulation, while 100 additional parameters are stored for later remote replay and display. The simulation speed is controlled by data transmission rates. At 1200 baud transmission rate, the simulations are executed in one-fourth the time of the actual transient.

In order to meet the needs of USNRC-IE, the plant analyzer has now been programmed to simulate indefinitely long transients for emergency drills in the Emergency Operations Center of USNRC. As shown in Figure 2, the plant analyzer can maintain steady-state conditions for 20 hours2.314815e-4 days 0.00556 hours 3.306878e-5 weeks 7.61e-6 months . Figures 3 and 4 show typical results obtained during the first two hours of the drill. In addition to the graphical display shown in Figures 2 through 4, the plant ana-lyzer also transmits tabulated parametets as specified by the NRC. These tables are updated in specified time intervals and also rhow results f rom the previous two updates for comparison and trend determinations.

The BNL Plant Analyzer is now the only operating facility that simulates two orders-of-magnitude faster than the CDC-7600 computer, that is accessible and fully operational in on-line interactive mode from anywhere in the U.S.

and Europe, and which is affordable and meets the needs of the USNRC, the utilities and other institutions with limited access to super computers. It constitutes a new technology based on an integrated concept for cost-efficient simulation.

Work performed under the auspices of the U.S. Nuclear Regulatory Commission.

25-7

References

1. NUREG/CR-0058, Vol. 2, pp. 431-451 (1984).

l

2. NUREG/CR-3943 (1984).

s .* I

>g, - < g, s';.

} sisa FE 522E cc:rH p?:v: rEsn grivE

'* li ? -

U wr. 8a -

Q ""

i ube .~

__ c..cu Y ';E 1

}l' _.

t h"' 0-O 0

~

M__qgkan - 1 :

,, r. w *

~

~ d')l 9u !

" U"* 9 r.mos i '

0.0 4 8 12 15 20 [

.a r u.a o.xm.c o-.== cwn T M (n) i Fig. 1 Flow Schematic and Control Fig. 2 System Pressure and Its Blocks for BWR Simulation; FW-Feed- Derivative During a 20-Hour water Controller, P-Pressure Control- Steady-State Simulation ler, RF-Recirculation Flow Controller K.1Ta1EETR ^

0 cssE==os 771 inst . Eisa Fi'E232E a 0 .O ~

c:: Err P:uf: LUE SLC*I:4TE rg l g 0 0

.a g- r Uh _ h E O

'O -

0 3

4 0 1

3 y g i ni "O Ig ) 30 \

~

f3 j h

~

~I O E I 0 IG 3E 5 00 75

! [0* !? k -

03 J

7;;a' 't 'e" j 0 1E ED Sul 7E t Tid (!)

Fig. 3 Suppression Pool Temperature Fig. 4 Vessel Pressure and Valve and Liquid Level During Tsc-Hour Discharge Rate During Two-Hour ATWS Transient Carried Out for IE ATWS Transient Carried Out for Emergency Drill IE Emergency Drill 25-8

, 4 e T ~uw t ,.s.,- r,oc .- .. . ., ,

uix oc o u m u m o ,co . .o=

g,,on= u.

E'"28 BIBLIOGRAPHIC DATA SHEET NUREG/CP-0071 SEE imsTRUCTt04$ ON Tat mivim56 3Lt.VtSL.%.

3 TeTLE .N0 5v8 TITLE Transactions of ths Thirteenth Water Reactor Safety Research Information Meeting 4 0.Tt REpoMT CoMPLETf D Wo%TH Tt.R i auf oam September 1985 Conference papers by various authors; compiled by

. o n at om issuto l

o' ' - '<.a Allen J. Weiss October l 1985

. ,aostcTa.s. .ca. u=4 ~-na

. . ts oa-,~a o.a.~,1 7,os ...t . o o. .L,~o .oc E ss ., ., < c ,

l

! Office of Nuclear Regulatory Research

' ""a "a ^ * ' * *" a U. S. Nuclear Regulatory Commission Washington, D. C. 20555 A-3283 10 SPopesomsseG 08tG.%i2. Tion N.vt .ND W. LihG .DoRESS fiw'ar8'l* C88 '" "'O#"*PURF Transactions of conference on safety research Same as Item 9 above. , ,,,,,oo Co,4,, t o ,, . ,

October 22-25, 1985 12 SUPPLEME8eT.Rv NOTE 5 13 s_S5f m.CT (200 *erds er iess, This report contains summaries of papers on reactor safety research to be presented at the 13th Water Reactor Safety Research Information Meeting held at the National Bureau of Standards in Gaithersburg, Maryland, October 22-25, 1985.

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 are also included, which cover the highlights of reactor safety research conducted by the electric utilities through the Electric Power Research Institute, 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 mean-ingful discussion and information exchange during the course of the meeting, and are given in the order of their presentation in each session.

u ooCo.E.T . .Lvs.. . . . voo os of sc . To . i s .,v,.g,,y -

reactor safety research nuclear safety research Unlimited

't SECum TV CL.85lflC.flom

e. sotNTsf stal Ortm ENotD Ttmws ifne reeerto Unclassified

,,~ - . , .. 04.

is paict

l i mi row T Nuves a . A-c., e, r,oc ve, .e . .,,,,

I u s. =ucLam e e tuuTon y co==essiom gc,,*on= sm

%'," 'E BIBLIOGRAPHIC DATA SHEET NUREG/CP-0071 SEE IN57muCT60NS ON TMt stytast 2 f tTLE ANO 5ueTITLE 3 LE AVE SL A'A Transactions of the Thirteenth Water Reactor Safety Research Information Meeting 4 oaf t atPORT CowPLETED MONin

EAR

. aufenomist September 1985 Conference papers by various authors; compiled by . o ATi miPomr iisvio

"^"

"o"

Allen J. Weiss l October 1985 8 PROJECT Tass woms unit NuwetR 7 b E AFOmwiNG OmGANi2 AT+0N Newt ANO MaittNG AcontSS traefweele Cess #

Office of Nuclear Regulatory Research

'"'"'"^"*""

V. S. Nuclear Regulatory Commission Washington, D. C. 20555 A-3283 10 SPONsOmiNG OmGANilaTioN Nawt AND u A.Lineo Aoomiss ##are w erle Ceeer tiaTvetOFmteomT Transactions of conference on safety research Same as Item 9 above. ,, ,,,,oo co, , , , o ,,,,,,,. e.,,,,

October 22-25, 1985 12 SUPPLEMENT Amv NOTES 13 A $f m ACT 420 7 eerse er ressi This report contains-summaries of papers on reactor safety research to be presented at the 13th Water Reactor Safety Research Information Meeting held at the National Bureau of Standards in Gaithersburg, Maryland, October 22-25, 1985.

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 are also included, which cover the highlights of reactor safety research corducted by the electric utilities through the Electric Power Research Institute, 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 mean-ingful discussion and information exchange during the course of the meeting, and are given in the order of their presentation in each session.

,. ooCu..u AN.L.m . . n .omos onCm,, Toms i. A.,Ag.gt ,

reactor safety research nuclear safety research Unlimited it SECuntTV CLAlla8sCATION rTs.e eneer

.so NTie m or.N No.cTim ,,

Unclassified i Vne reeertl Unclassified

'7 NuwetmC*PAGES is Pasca

u s. huCLELA ELELULATO%v COMMession i 8ttruMT mvMIEa rAss,ew ey Froc ese ver We, ># eard hmC 2 0mM 335

)f, "*,'- BIBUOGRAPHIC DATA SHEET NUREG/CP-0071 sEE ,=sravCr.o~s o~ T-E .ENE 3 LE AVE 4Lamu

/

2 TITLE AN0 5L87tTLE Transactions of t Thirteenth Water Reactor Safety Research Inform ion Meeting efaTE REP 0mf COMPLETED Mo n a, l

s aumoam Sep nber 1985 Conference papers by rious authors; compiled by f . oari EPoai .ssuto

" Aa Allen J. Weiss /Wo~ ~ l Opober 1985 7 PE AFORMING 0m",A%42 AY EON N AME ANO M A'Le%- 00at$5 riacbee to codes a Paftcf.Y Ass Wo** u%it NWSER Office of Nuclear Regulat Research / '*"a U. S. Nuclear Regulatory Co iission 'f oa o"^*'

Washington, D. C. 20555 [ A-3283 Y

10 SPONsomi%G ORGANi2 ATION NAvt ANO MaitthG AODRE53 0 ver le Cose8 Ita TYPE OF REPORT Transactions of conference n safety resea d Same as Item 9 above. . ,E . oo co, E to ,,,,, ..e ee ,

October 22-25, 1985 12 SUPPLEMENT AR v NOTES 13 Ae$f n ACT r200 spores or 'essJ This report contains summaries of pape n reactor safety research to be presented at the 13th Water Reactor Safety R aarch Information Meeting held at the National Bureau of Standards in Gaither u , Maryland, October 22-25, 1985.

The summaries briefly describe the program and esults of nuclear safety research sponsored by the Office of Nuclear Regula ry Res rch, USNRC. Summaries of invited papers are also included, which cover th highligh of reactor safety research conducted by the electric utilities thr gh the Ele ric Power Research Institute, the nuclear industry, and the research f government nd industry in Europe and Japan. The summaries have been compi d in one repor to provide a basis for mean-irgful discussion and information ex ange during the urse of the meeting, and are given in the order of their pre ntation in each ses ion.

i. ooc,E ~T A~ AL n.. . e u ..omos otsca. Pro i is A,v,Agagp reactor safety research nuclear safety research Unlimi ted 14 SECURITY CLAll.FICATION

.,otst,,,E. oPi~ENotavt. ,

lYEcTassified a n. r.e-,

Unclassified

,, ..E.o,,AoEs t4 PRICE

13th WATER REACTOR SAFETY RESEARCH INFORMATION MEETING October 22-25, 1985 Session Schedule RED AUDITORIUM GREEN AUDITORIUM LECTURE RDOM B Tues Plenary Session AM Integral Systems Tests Risk Analysis / Process Control PRA Applications Session 1 Session 2 Session 3 Tues Integral Systems Tests Mechanical and Nuclear Plant Aging PM Structural Research Session 4 Session 5 Session 6 Wed Severe Accident Separate Effects /Ex- Seismic Research AM Sequence Analysis periments & Analyses Sessico 7 Session 8 Session 9 Fission Product Release International Code Equipment Wed and Transport in Assessment Program Quali fication PM Containment Session 10 Session 11 Session 12 Severe Accident International Code Surry Steam Gener-Thurs Source Tern Assessment Program ator/ Examination AM and Evaluation Session 13 Session 14 Session 15 Risk Analysis / Dependent Paterials Engineering Containment Systems Failure Analysis Research/Non-Destruc- Research/Contain-Thurs tive Evaluation ment Loads Analysis PM Session 17 Environmental Ef fects in Piping Session 16 Session 18 Session 19 Materials Engineering Code Assessment Industry Safety Fri Research/ Pressure and Improvement Research AM Vessel Research Session 20 Session 21 Session 22 Materials Engineering 20/30 Research Fri Research/ Piping Research Session 24 PM and Fracture Mechanics Nuclear Plant Analyzer Session 23 Session 25 l

l l

l

r ,

1 l #' I'1 1 S 1 U

- 12055507EF77 UNITED STATES US t,R C V 0 " ' ' '

NUCLEAR REGULATORY COMMISS ADF I(1 bP-PDR NUF E6 O tmo" n'*e'is"r'a'io WASHINGTON, D.C. 20666 (,9h01 wf0"!c CC 20555 bASblNGT0f; Peauer n. o er OFFICIAL SUSINESS PENALTY FOR PRIVATE USE,8300 1

i i

i 1

i.

I 1

l.

l 4

- i

1 I

I 1

+

e I

r l

l

ML20205E218

Text

SUMMARY

SUMMARY

Background:

SUMMARY

SUMMARY

SUMMARY

SUMMARY

SUMMARY

References:

SUMMARY

Background

SUMMARY

SUMMARY

SUMMARY

SUMMARY

SUMMARY

Dearadation in Polymers,

SUMMARY

SUMMARY

SUMMARY

SUMMARY

SUMMARY

SUMMARY

SUMMARY

Contact:

SUMMARY

Contact:

SUMMARY

SUMMARY

Contact:

SUMMARY

SUMMARY

SUMMARY

SUMMARY

SUMMARY

SUMMARY

SUMMARY

SUMMARY

Background

SUMMARY

SUMMARY

Navigation menu

Search