ML20058H759
| ML20058H759 | |
| Person / Time | |
|---|---|
| Site: | Surry  |
| Issue date: | 10/31/1990 |
| From: | Breeding R, Helton J, Murfin W, Laura Smith SANDIA NATIONAL LABORATORIES |
| To: | NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| References | |
| CON-FIN-A-1322 NUREG-CR-4551, NUREG-CR-4551-V3R1P1, NUREG-CR-4551P1, SAND86-1309, NUDOCS 9011260018 | |
| Download: ML20058H759 (511) | |
Text
{{#Wiki_filter:- . .. . - .. . .. 4 NUREG/CR-4551 SAND 86-1309 Vol. 3, Rev.1, Part 1 Evaluation of Severe Accic.ent Risis: Surry Unit 1 Main Report
.J l e ng, J. C liciton, W. B. Murfin, L N. Smith Sandia National Laboratories -Operated by Sandia Corporation Prepared for -
U.S. Nuclear Regulatory Commission 4 f P
AVAllABILITY NOTICE Avn!!atWitty of Referenco Materials Cited h NRC Put2 cations Most documents cited in NRC pubucations will be svanable from one of the foDowing sources:
- 1. The NRC Public Document Room, 2120 L Street, NW, Lower Level, Washington, DC 20555
- 2. The Superintendent of Documents, U.S. Government Printing Office, P.O. Box 37082. Washington, DC 20013-7062
- 3. The National Technicalinformation Service, Springfield, VA 22161 Although the isting that foCows represents the majority of documents cited h NRC publications, it is not hunded to be exhaustive.
Referenced documents avanable for hspection and ot :ybg for a fee from the NRC Puble Doranent Room hclude NRC correspondence and intomal NRC memoranda; NRC Offloe of Inspection and Er,icreement bulletins, circulars, information notices, hspection and Investigation notices; Ucensee Event Reports; ven-dor reports and cor-espondence; Commission papers; and appBcant and Iconsee documents and corte-spondonce. The following documents in the NUREO series are available for purchase from the GPO Sales Program: formal NRC staff and contractor reports, NRC-sponsored conference proceedings, and NRC booidets and brochures, Also evalable are Regulatory Guides NRC regulations h the Code of Federal Regularlons, and Nuclear Regulatory Commission Issuances. Documents avaRable from the National Technical Informa'4n Service hclude NUREG series reports and technical repc' 3 prepared by other federal agencies and reports prepared by the Atomic Energy Commis-sion, forerunrw agency to the Nuclear Regulatory Commission. Documents avaRable fiom public and speclal technical Ebraries hebde.at open Iterature items, such as books, joumal and periodical articles, and transactions, Federal Reglster notices, federal and state legesta-tiort, and congressional reports can usualy be obtained from these Ibraries. Documents such as theses, dissertations, foreign reports and translations, and non-NRC conference pro-ceedings are avaAable for purchase from the organlzation sponsoring the pub 0 cation cited. Single copies of NRC dra9 reports are avalable f*ee, to the extent of supply, upon written request to the Office of Information Rotmes Management, Distribution Section, U.S. Nuclear Regulatory Commission, Washington, DC ' 20555. Copies of houstry codes and standards used h a substan*,!ve manner h the NRC regulatory process are maintained at the NRC Ubrary,7020 Norfolk Avenue, Bethesda, Maryland, and are avaRable there for refer. ence use by the pubBc Codes and standards are usuaRy copyrighted and may be purchased from the originathg organization or, if they are American National Standards, from the American Natiois ; Ot-'vlards institute,1430 Broadway, New York, NY 10018, DISCLAIMER NOTICE This report was prepared as an account of work sponsored by an agency of the United States Govemmont. NolthortheUnitodStatosGovemmentnoranyagencythoroof,oranyof theiremploycos,makesanywarTanty, exprosed or implied, or assumes any legal liability of responsibility for any third pany's uso, or the results of such uso, of any Information, apparatus, product or process disclosed in this roport, or represents that its use by such third party 'vould not infringo privately owned rights, s 3
)
l NUREG/CR -455. ! SAND 86-1309 Vol. 3, Rev.1, Part 1 l 1 Evaluation of Severe Accident Risks: Surry Unit 1 i Main Report j Manuscript Completed: July 1990 i Date Published: October 1990 i ! Prepared by R. J. Breeding. J. C. Helton', W.11. Murfin', L N. Smith 8 l Sandia National laboratories Albuquerque, NM 87185 Prepared for Division of Systems Research Omce of Nuclear Regulatory Research U.S. Nuclear Regulatory Commission Washington, DC 20555 NRC FIN A1322 l 1 Arizona State University Tempe, AZ Frechnadyne Engineering Consultants, Inc., Albuquerque, NM
- Science Appikations International Corporation. Albuquergoe, NM 1
l
ABSTRACT in support of the Nuclear Regulatory Commission's (NRC's) assessment of the risk from severe accidents at commercial nuclear power plants in the U.S. reported in NUREG 1150, the Severe Accident Risk Reduction program (SARRp) has completed a revised calculation of the risk to the general public from severe accidents at the Surry Power Station, Unit 1. This power plant, located in southeastern Virginia, is operated by the Virginia Electric Power Company. The emphasis in this risk analysis was not on determining a "so called" point estimate of risk. Rather, it was to determine the distribution of risk, and to discover the uncertainties that account for the breadth of this distribution. T1.c offsite risk from internal initiating events was found to be generally below the risk estimates reported about a decade ago in the Reactor Safety Study (RSS) (WASH 1400). The upper end of the current distributions for common risk measures is of the same order of magnitude as the point estimates obtained by the RSS, but the bulk of the distribution is much lower. The containment appears to be quite likely to successfully withstand the loads that might be placed upon it if the core melts and the reactor vessel fails. Most of the risk, in the current view, comes from initiating events which bypass the containment, such as interfacing system pipe breaks and steam generatcr tube ruptures. These events are estimated to have a relatively low frequency of occurrence, but their consequences are relatively large. The uncertainties in risk from internal initiators are largely due to uncertainties in the initiating frequency of these bypass events, and in the magnitude of the radioactive release that results. ' The offsite risk from external initiating events was found to be of the same order of magnitude or higher than the risk from internal initiating events. Only fires and earthquakes were found a be important enough external initiators to warrant a complete analysis. The risk from seismic initiators is an order of magnitude or more greater than that from fires. I The Surry containment is not expected to fail directly due to an earthquake. As there are no seismic bypass initiators, the offsite risk from a radioactive release due to an earthquake is low unless the ground motion indirectly a.ils - the containment. Failure of the supports of the I reactor coolant pumps or the steam generators was judged to have this potential. There is great uncertainty in the frequency with which large magnitude earthquakes may be expected in the eastern U.S. Risk was calculated for two hazard distributions, which differ markedly. Both the absolute value of the seismic risk distributions and the uncertainties in these distributions are driven by the uncertainties in the seismic hazard. l The RSS considered external initiating events only in a cursory manner. 1 1 tii/iv
1 l 1 i i l l 2 1 J CONTENTS i l i SUKKARY.... ........................................................ S 1 l
- 1. INTRODUCTION................................................... 11 1.1 Ba ckground and Obj e c t ive s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 j 1.2 Overview of Surry Power S tation, Unit 1. . . . . . . . . . . . . . . . . . . 1. 3 l 1.3 Changes Since the Draft Report............................. 1.5 i
1.4 Structure of the Analysis................................. 1.8 " i 1.5 Organization of this Report............................... 1.16 i L 1.6 References................................................ 1.18 l
- 2. THE ACCIDENT PROGRESSION ANALYSIS ............................. 2.1 2.1 Surry _ Features Important to Accident )
Progression ............................................. . 2.1-2.1.1 The Surry containment Structure................... 2.1 2.1.2 Subatmospheric Containment During Operation....... 2.2 2.1.3 The Containment Heat Removal System............... 2.2 2.1.4 Service Water Cana1............................... 2.3 2.1.5 Sump and Cavity Arrangement.......'................. 2.3 1 l 2.2 Interface with the core Damage ! Prequency Analysis.~.......'............................... 2.4 1 2.2.1. De finition of Plant Damage States . . . . . . . . . . . . . . . . . 2.4 2.2.2 PDS Frequencies.................................... 2.6 v 2.2.2.1 PDS Frequencies for Internal Initiators..................... 2.6 2.2.2.2 PDS Frequencies for-Fire Initiators......................... 2.14 L 2.2.2.3 PDS Frequencies for ! Seismic Initiators...................... 2.15 2.2.3 High Level Grouping of PDSs . . . . . . . . . . . . . . . . . . . . . . . 2. 20 1 L 2.2.4 Variables Sampled in the i Accident Frequency Analysis...-................... 2.20' 2.3 Description of the APET....-............................... 2.32 ; l 2.3.1 Ove rvi ew o f the APET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 . 3 2 ! 2.3.2 Ove rview of APET Quantification. . . . . . . . . . . . . . . . . . . 2. 35 2.3.3 Variables Sampled for the Accident Progression Analysis.................... 2.42 l l t-I
- l. V ,
e-w+
2.4 Description of the Accident Progression Bins.............. 2.53 2.4.1 Description of the Bin Characteristics............ 2.53 2.4.2 Rebinning....... ................................. 2.60 2.4.3 Summary Bins for Precantations.................... 2.64 2.5 Results of the Accident Progression Analysis.............. 2.68 2.5.1 Results for Internal Initiators................... 2.68 2.5.1.1 Results for PDS Group 1: Slow SB0........ 2.68 2.5.1.2 Results for PDS Group 2: LOCAs........... 2.71 2.5.1.3 Results for PDS Group 3: Fast SB0........ 2.73 2.5.1.4 Results for PDS Group 4: Event V......... 2.75 2.5.1.5 Results for PDS Group 5: Transients...... 2.77 2.5.1.6 Results for PDS Group 6: ATWS............ 2.79 2.5.1.7 Results for ,$ Group 7: SGTRs........... 2.81 2.5.1.8 Core Damage acrest and. Avo idanc e o f VB . . . . . . . . . . . . . . . . . . . . . . . . . 2 . 84 2.5.1.9 Early Containment Failure ............... 2.85 2.5.1.10 Summary ................................. 2.87 2.5.2 Sensitivity Analyses for Internal Initiators...... 2.90 2.5.2.1 Pressure at VB for S and S 3 Breaks...... 2.90 2.5.2.2 No T 1 SGTRs or llot leg Breaks. . . . . . . . . . . 2.92 2.5.2.3 Second S amp 1 e . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 .103
-2.5.3 Results for Fire Initiators...... ................ 2.107 2.5.4 Sensitivity Analyses for Fire Initiators.......... 2.109 2.5,5 Results for Seismic Initiators -
LLNL llazard Distribution......................... 2.109 2.5.5.1 Results for PDS Group EQ 1, LOSP: liigh Acceleration, LLNL llazard Distribution...................... 2.109 2.5.5.2 Results for PDS Group EQ 2, SB0:. liigh Acceleration, LLNL llazard Dis tribution. . . . . . . . . . . . . . . . . . . . . . 2.113 2.5.5.3 Results for PDS Group EQ 3. LOCAs: LLNL itazard Curve liigh Acceleration. . . . . . 2.116 2.5.5.4 Results for PDS Group EQ 1, LOSP: (No SBO): Low Acceleration, LLNL llazard Distribution................. 2.118 2.5.5.5 Results for PDS Group EQ 2, SB0: Low Acceleration, LLNL llazard - Distribution. ...........................,2.121 2.5.5.6 Results for PDS Group EQ 3, LOCAs: Low Acceleration, LLNL llazard Distribution............................. 2.123 2.5.6 Sensitivity Analyses for Seismic Initiators - LLNL llaz a rd D i s t r ibu t i on . . . . . . . . . . . . . . . . . . . . . . . . . 2 .12 6 I 1 l vi
2.5.7 Results for Seismic Initiators..EPRI L Hazard Distribution............................... 2.126 2.5.7.1 Results for PDS Group EQ 1. LDSP: l (No SBO): High Acceleration EPRI Hazard distribution................. 2.126 2.5.7.2 Results for PDS Group EQ 2, SB0: High Acceleration, EPRI Hazard Distribution...................... 2.130 1 2.5.7.3 Results for PDS Group EQ 3. LOCAs: , High Acceleration, EPRI i Hazard Distribution...................... 2.132 , 2.$.7.4 Results for PDS Group EQ 1, LOSP ; (No SBO) Low Acceleration EPRI Hazard Distribution ................ 2.134 2.5.7.5 Results for PDS Group EQ 2. SB0: ! Low Acceleration, EPRI Hazard Distribution............................. 2.137 ! 2.5.7.6 Results for PDS Group EQ 3. LOCAs: Low Acceleration, EPRI Hazard Distribution...................... 2.139 l 2.5.8 Sensitivity Analyses for Seismic Initiators: EPRI Hazard Distribution........................ 2.141 2.5.9 Comparison of Results for Seismic Initiators...... 2.141 2.6 Insights from the Accident Progression f Analysis................................................. 2.144 l
- 3. RADI OLOGI CAL SOURC E T ERM ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1- i I
3.1 Surry Features Important to the Source # l Term Analysis............................................. 3.1 3.2 Description of the SURSOR Code............................. 3.3 3.2.1 Overview of.the Parametric Mode 1.................... 3.3 ' 3.2.2 Description of SURS0R.........................,. .. 3.4 3.2.3 Variables Sampled in the Source Term Analysis.............................. 3.11 r 13 . 3 Resul ts o f the Source Te rm Analysis . . . . . . . . . . . . . . . . . . . . . . . . 3.15 [ l '3.3.1 Results for Inte rnal Initiators . . . . . . . . . . . . . . . . . . . 3.16 3.3.1.1 Results for PDS Group 1: Slow SB0........ 3.16 - 3.3.1.2 Results for PDS Group 2: LOCAs........... 3.18 3.3.1.3 Results for.PDS Group 3: Fast SB0........ 3.19 3.3.1,4 Results for PDS: Group 4: Event V......... 3.19 3.3.1.5 Results for PDS Group 5: Transients........ 3.20 3.3.1.6 Results for PDS Group 6: ATWS............ 3.21 3.3.1.7 Results for PDS Group 7: SCTRs........... 3.21 3.3.1.8 Results for Summary Accident Progression Bins......................... 3.22 vii L
. - , -%+, -
m- v + T
3.3.1.9 S am a ry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.23 i 3.3.1.10 -ntributors to Uncertainty.............. 3.23 3.3.2 Source Terms Results for the Second Sample for Internal Initiators . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. 28 3.3.3 Results for Fire Initiators....................... 3.67 1 1 3.3.4 Sensitivity Analyses for Fire Ini:iators.......... 3.67 I 1 3.3.5 Results for Seismic Initiators LLNL Hazard Dis tribution. . . . . . . . . . . . . . . . . . . . . . . . . . 3. 67 , 3.3.5.1 Results for PDS Group EQ 1, LOSP (No SBO); High Acceleration LLNL Hazard Distribution................. 3.68 3.3.5.2 Results for PDS Group EQ 2, SB0: High Acceleration, LLNL Hazard Distribution............................. 3.68 3.3.5.3 Results for PDS Group EQ 3, LOCAs: High Acceleration, LLNL Hazard Distribution...................... 3.69 3.3.5.4 Results for PDS Group EQ 1, LOSP: (No SBO): Low Acceleration LLNL Hazard Distribution. . . . . . 4 ......... 3.69 3.3.5.5 Results for PDS Group EQ 2: SB0: Low Acceleration, LLNL Hazard Distribution...................... 3.70 3.3.5.6 Results for PDS Group EQ 3: LOCAs: low Acceleration, LLNL Hazard Distribution............ ......... 3.70 3.3.6 Sensitivity Analyses for Seismic Initiators: LLNL Hazard Distribution. . . . . . . . . . . . . 3. 71 L 3.3.7 Results for Seismic Initiators: EPRI Haz a rd Cu rve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 . 71 e 3.3.8 Sensitivity Analyses for Seismic Initiators: EPRI Hazard Distribution............. 3.72 3.4 Partitioning of the Source Terms for the Consequence Analysis..................................... 3.95 3.4.1 Results for Internal Initiators.................... 3,95 i 3.4.2 Results for Fire Initiators....................... 3.108 i L 3.4.3 Results for Seismic Initiators: (- 1 LLNL H a z a rd Cu rve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 .115 3.4.4 Results for Seismic Initiators: EPRI Hazard Curve.................................. 3.128 1 3.5 Insights from the Source Te rm Analysis. . . . . . . . . . . . . . . . . . . . 3.141 3.6 R e f e r e nc e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 .14 2 viii
1 1 k l
\
1 4 CONSEQUENCE ANALYSIS............................................ 4.1 1 4.1 Description of the Consequence Analysis..................... 4.1 1
)
4.2 KACCS Input for Surry....................................... 4.2 r 4.3 Results of the KACCS Consequence Calculations............... 4.7 4.3.1 Results for Internal Initiators.................... 4.7 4.3.2 Results for Fire Initiators....................... 4.13 ' 4.3.3 P.esults for Seismic Initiators' 4 LLNL Hazard curve.............................'.... 4.13 4.3.4 Results for Seismic Initiators: EPRI H a z a r d Cu rve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 .13 P 4.4 References................................................. 4.29
- 5. SURRY RISK RESULTS ............................................. 5.1 .
5.1 Results for Internal Initiators............................ 5.1 5.1.1 Risk Results....................................... 5.1 5.1,2 Second Sample............... ...................... 5.8 I 5.1.3 Contributors to Risk.............................. 5.13 5.1.4 Contributors to Uncerte.inty....................... 5.23 5.2 Results for Fire Initiators................................ 5.38 5.3 Results for Seismic Initiaters:'LLNL Hazard Curve.............. .............................. 5.44
)
1 5.3.1 Risk Results....................................... 5.44 i f 5.3.2 Results of Sensitivity Analyses................... 5.49 ;
-5.4 Results for Seismic Initiators:
- E PR I - H a z a r d Cu rve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . '. . . . . . 5.53 5.4.1 Risk Resulta....................................... 5.53 5.4.2 'Results of Sensitivity Analyses................... 5.58 5.5 Comparison of Results for Seismic Initiators.............. 5.59 5.6. R e f e r e nc e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.64
- 6. INSICHTS AND CONCLUSIONS........................................ 6.1 ix
. . - . , . --- -, - . - - - , .. ~_ -
FIGUr.ES
- 1. Back End Documentation for NUREG-1150........................... viii S.1 Overview of Integrated Plant Analysis in NUREG 1150. . . . . . . . S.3 S.2 Mean Probability of APBs for PDSs Internal and Fire....... S.10 '
S.3 Probability of Core Damage Arrest Internal................ S.11 S.4 Probab!,11ty of Early Containment Failure Internal......... S.11 S.5 Mean Probabilities of APBs dor PDSs -Seismic............... S.13 ' S.6 Probability of Core Damage Arrest Seismic................. S.14 0.7 Prabability of Early Containment Failure- Seismic.......... S.1/ . S.8 Exceedance Frequencies for Release Fractions for Surry: All Internal Initiators..... .................. S.19 S.9 Exceedance Frequencier for Release Fractions Seismic Initiators for.LLNL Hazard Distributions.......... S.21 4 S.10 Exceedance Frequencies for Release Fractions Seismic Initiators for EPRI Hazard Distributions.......... S.22 S.11 Internal Initiators........................................ S.25 S.12 LLNL Hazard Distributions, Low PGA......................... S.26 f i S.12 LLNL Hazard Distributions, High PGA........................ S.27 S.13'EPRI Hazard Distributions, Low'PGA......................... S.28 S.13 EPRI Hazard Distributions, High PCA....................... S.29 S.14 Results of the Integrated Risk Analysis ; for. Internal Initiators................................... S.31
- S.15 Distributions-of Annual Risk Surry: Internal Initiators................................ S.35 l S.16 Results.0f the Integrated Risk Analysis ,-
for Leismic Initiators- LLNL Hazard Distributions......... S.36. ! S.17 P.esults'of the Integrated Risk Analysis. for Seismic Initiators- EPRI Hazard Distributions......... S.39 J. S.18 Distributions of Annual' Risk for Seismic Initiators . LLNL Hazard Dis tributions . . . . . . . . . . . . . S.42 1 x i1 I
4 S.19 Distributions of Annual Risk for Seismic Initiators. EPR1 Hazard Distributions............. S.43 , S.20 Statistical Measures for Both Samples...................... S.45 i S.21 Contributions of Summary PDS Groups to Mean Risk........... S.47 S.22 Contributions of Summary APBs to Mean Risk................. S.48 1-1 Section of Surry Containment............................... 1.4 12 Overview of Integrated Plant Analysis in NUREG 1150........ 1.9 i 1-3 Example Risk CCDF.......................................... 1.15 l t 2.5 1 Probability of Core Damage Arrest Internal......... 2.85 ; 2.5 2 Probability of Early Containment ~ Failure Internal.................................. 2.86 . 2.5 3 Mean Probability of APBs for PDSs ! Interr.a1 and Fire................................. 2.87 2.5 3A Range of Mean Core Damage Frequencies for - 200 Obse rvations for Seven PDS Groups . . . . . . . . . . . . . . 2.88 a 2.5 3B Distribution of Frequencies for APB Groups.......... 2.89 l 2.554 Mean Probability of APBs for PDSs-- ( Internal Sample 2.................................. 2.104 i 2.5 5 Mean Probabilities of APBs for PDSs -Seismic........ 2.142 . . 2.5 6 Probability of Core Damage Arrest Seismic........... 2.143-1 l 2.5-7 Probability of Early Containment E Failure- Seismic................................... 2.144 3.2 1 Blood Flow Diagram for SURS0R....................... b.7 l' l 3.3 1 Exceedance Frequencies for Release Fractions a for Surry Internal ~ Initiators
'PDS Group 1: Slow SB0.............................. 3.32 >
3.3-2 Exceedance Frequencies for Release Fractions for Surry Internal Initiators PDS Group 2: LOCAs................................ 3.34 L 3.3 3 Exceedance Frequencies for Release Fractions for Surry Internal Initiators PDS Group 3: Fast SB0............................. 3.36 Xi
3.3 4 Exceedance Frequencies for Release Fractions for Surry Internal Initiators PDS Group 4: Event V.............................. 3.38 3.3 5 Exceedance Frequencies for Release Fractions , for Surry Internal Initiators ' PDS Group 5: Transients........................... 3,40 3.3 6 Exceedance Frequencies for Release Fractions i 1 for Surry Internal Initiators PDS Group 6: ATWS................................. 3.42 ; 3.3 7 Exceedance Frequencies for Release Fractions for Surry Internal Initiators l PDS Group 7: SGTRs................................ 3.44 3.3 8 Exceedance Frequencies for Release Fractions for Surry Internal Initiators. Alpha Mode.......... 3.45 1 3.3 9 Exceedance Frequencies for Release Fractions for Surry Internal Initiators. Containment Failure at Vessel Breach with the RCS at High Pressure............................... 3.46 3.3 10 Exceedance Frequencies'for Release Fractions for Surry Internal Initiators. Containment Failure at Vessel Breach with the RCS at Low Pressure................................ 3.47 f 3.3 11 Exceedance Frequencies for Release Fractions for Surry Internal Initiators. Late Containment Failure........................... 3.48 3.3 12 Exceedance Frequencies for Release Fractions for Surry Internal Initiators . V Dry. . . . . . . . . . . . . . . 3.49 3.3 13 Exceedance Frequencies for Release Fractions for Surry Internal Initiators. V Wet............... 3.50 l 3.3 14 Exceedance Frequencies for Release Fractions 3 l for Surry Internal Initiators.
"G" SGTRs (Secondary SRVs Rec 1osing)............... 3.51 3.3 15 Exceedance Frequencies for Release Fractions for Surry Internal Initiators.
"H" SGTRs (Secondary SRVs Stuck.Open).............. 3.52 3.3 16 Exceedance Frequencies for Release Fractions for Surry Internal Initiators.
All Internal Initiators............................ 3.53 xii
t i t 3.3 17 Partial Rank Correlation Coefficients for Exceedance Frequencies for Release Fractions for Surry Internal Initiators. Iodine................................. 3.61 3.3-18 Partial Rank Correlation Coefficients for Exceedance Frequencies for Release - Fractions for Surry Internal Initiators. Cesium................................. 3.62 3.3 19 Partial Rank Correlation Coefficients for Exceedance Frequencies for Release Fractions for Surry Internal Initiators. Strontium.............................. 3.63 3.3 20 Partial Rank Correlation Coefficients for Exceedance Frequencies for Release Fractions for Surry Internal Initiators. Lanthanum.............................. 3.63 3.3 21 Comparison of Exceedance Frequencies for the Two Independent Latin Hypercube Samples Surry Internal Initiators.......................... 3.64 3.3 22 Comparison of Two PRCC Curves for Two ' Representative Variables.... ..................... 3,66 3.3 23 Release Fractions CCDFs for Fire Initiators......... 3.73 i 3.3 24 Release Fractions CCDFs for LLNL PDS EQ 1: LOSP Hi PCA....................................... 3.76 3.3 25 Release Fractions CCDFs for LLNL PDS EQ 2 SB0....... 3.78 L 3.3 26 Release Fractions CCDFs for LLNL PDS EQ 3 LOCAs. . . . . 3.80 3.3 27 Release Fractions CCDFs for All LLNL l 6 All PGA...............-........................... 3.82 , 3.3 28 Release Fractions CCDFs for EPRI PDS EQ 1 L0SP...................................... 3.84 3.3 29 Release Fractions CCDFs for EPRI PDS LQ 2 SB0....................................... 3.86 3.3 30 Release Fractions CCDFs for EPRI-PDS EQ 3 LOCAs..................................... 3.93 3.3 31 Release Fractions CCDFs for ALL PDS & All PGA........................ .......... .. 3.94 3.4 1 Distribution of Nonzero Early and Chronic Health Effee Weights for Internal Initiators....... 3.97 xiii
3.4 2 Partition Plot Fire Initiator...................... 3.114 3.4-3 Partition Piot Seismic LLNL Initiators. . . . . . . . . . . . . 3.126 3.4 4 Partition Plot- EPRI Initiators..................... 3.129 = 4.3 1 Consequences Conditional on Source Terms Surry: Internal Initiators......................... 4.11 4.3 2 Consequences Conditional on Source Terms Surry: Fire Initiators............................. 4.14 , 4.3 3 consequences conditional on Source Terms Surry: Seismic Initiators; LLNL llazard Dis tributions , low PCA. . . . . . . . . . . . . . . . . . . . . . 4.17 4.3-4 Consequences Conditional on Source Terms Surry: Seismic Initiators; LLNL Hazard Distributions High PCA..................... 4.18 4.3 5 Consequences. Conditional on Source Terms Surry: Seismic Initiators; EPRI _ Hazard Distributions, Low PCA...................... 4.23 4.3 6 Consequences Conditional on Source Terms Surry: Seismic Initiators; EPRI Hazard Distributions, High PCA..................... 4.24 5.1 1 Execedance Frequencies for Risk Surry--Internal Initiators......................... 5.2 5.1 2 Distributions of Annual Risk S.rry All Internal Initiators............................ 5.6 5.1-3 Comparison of the Exceedance Frequencies for Risk. Surry: Two Samples for Internal Initiators... .................. ......... 5.9 5.1 4 Comaprison of the Cumulative distributions for Annual Risk; Surry: Two Samples for Internal Initiators................................ 5.12 5.1 5 Fractional PDS Contributions to Annual Risk: Surry -Internal Initiators................... 5.16 5.1-6 Fractional APB Contributions to Annual Risk: Surry Internal Initiators................... 5.19 5.1 7 Distributions for Fractional PDS Contributions to Annual Risk. Surry--Internal Initiators......... 5.20 x
,, 5.1 8 Partial Rank Correlation Coefficients < for Exceedance Frequencies for Early Fatality Risk...................................... 5.30 xiv
5.1 9 Partial Rank Corre?ation Coefficients ! for Exceedance Frequencies for Latent Fatality Risk. Surry: Internal Initiators.......... 5,31 5.1 10 Partial Rank Correlation Coefficients ! for Exceedance Frequencies for Population Dose Within 50 Miles. Surry: Internal Initiators........ 5.32 i 5.1 11 Partial Rank Correlation Coefficients for Exceedance Frequencies for Population Dose for the Entire Region ! Surry: Internal Initiators......................... 5.33 5.1 12 Partial Rank Correlation Coefficients for Exceedance Frequencies for Individual Early Fatality Risk Within 1 Mile Surry: Internal Initiators......................... 5.35 5.1-13 Partial Rank Correlation Coefficients for Exceedance Frequencies for Population Dose for the Entire Region Surry: Internal Initiators......................... 5.36 5.1 14 Partial Rank Correlation Coefficients for Exceedance Frequencies for Individual Latent Cancer Fatality Risk Within 10 Miles Surry: Internal Initiators......................... 5.37 5.2 1 Excoedance Frequencies for Risk. Surry -Fire Initiators............................. 5.40 , 5.2 2 Distributions of Annual Risk ' Surry Fire Initiators............................. 5.43 5.3 1 Exceedance Frequencies for Risk. ' Surry: Seismic Initiators LLNL Hazard Distributions.......................... 5.45 5.3 2 Distributions of Annual Risk. Surry: Seismic Initiators LLNL Hazard Distributions.......................... 5.48 5.4 1 Exceedance Frequencies for Risk. Surry: Seismic. Initiators EPRI Hazard Distributions.......................... 5.54 s 5.4 2 Distributions of Annual Risk. Surry: Seismic Initiators EPRI Hazard Distributions.......................... 5.57 xv
TABl.ES S.1 Design Features Relevant to Severe Accidents Surry Unit 1.............................................. S.4 S.2 Surry Core Damage Frequencies Internal, Fire, and Seismic Initiators.................... S.7 ; 2.2 1 Pk'R Plant Damage State Characteris tics . . . . . . . . . . . . . . 2.7 ; 2.2 2 PDSs for Surry Internal Initiators.................. 2.9 ' 2.2 3 Comparison of PDS Core Damage Frequencies for Surry Internal Initiators...................... 2.10 ; 2.2 4 PDSs for Surry Fire Initiators . . . . . . . . . . . . . . . . . . . . . . 2.14 l 2.2 5 Comparison of PDS Core Damage Frequencies I for Surry Fire Initiators..................,....... 2.14 ; j 2.2-6 Plant Damage States for Surry Seismic Initiators LLNL Hazard Distribution........................... 2.16 2.2 6 Plant Damage States for Surry Seismic Initiators EPRI Hazard Distribution........................... 2.17 2.2-7 Comparison of PDS Core Damage Frequencies L Surry: Internal Initiators LLNL Hazard Distribution........................... 2.18 + 2.2 7 Comparison of PDS Core Damage Frequencies Surry: Seismic Initiators EPRI Hazard Distribution........................... 2.19 ' L 2.2-8 Relationship Between PDS Croups
- and Summary Groups................................. 2.20 1
2.2 9 Variables Sampled in the Accident Frequency Analysis for Internal Initiators................... 2.23 2.2 10 Variables Sampled in the Accident Frequency 2.29
=
Analysis for Fire Initiator e....................... 2.3 1 Questions in the S" .y APET......'................... 2,37-2.3-2 Variables Sampled in the Accident Progression Analysis for Internal Initiators.................., 2.45 2.5.1 Results of the Accident Progression Analysis for Surry Internal Initiators
- PDS Group 1: Slow SB0............................ 2.69 xvi
l l 2.5.2 Results of the Accident Progression Analysis for Surry Internal Initiators PDS Group 2: LOCAs ............................. 2.72 2.5.3 Results of the Accident Progression Analysis for Surry Internal Initiators ,
- PDS Group 3: Fast SB0........................... 2.74 '
I 2.5.4 Results of the Accident Progression , Analysis for Surry Internal Initiators l
-PDS Group 4: Event V............................ 2.76 J i
2.5.5 Results of the Accident Progression Analysis for Surry Internal Initiators
--PDS Group 5: Transients.......................... 2.78 ]
2.5.6 Results of the Accident _ Progression Analysis for Surry Internal Initiators j
- PDS Group 6: ATWS............................... 2.80 q
.1 2.5.7 Results of the Accident Progression Analysis for Surry Internal Initiators PDS Group 7: SGTRs.............................. 2.82 2.5.8 Comparison of APET Results With and Without T 1 Hot Leg Breaks and SCTRs PDS Group 1--Slow SB0.............................. 2.97 2.5.9 Comparison of APET Results With and Without T I Hot Leg Breaks and SGTRs i
PDS Group 3 -Fast SB0......................... .... 2.98 2.5.10 Comparison of APET Results With and Without l T I Hot Leg Breaks and SGTRs PDS Group 5- Transients............................ 2.99 2.5.11 Comparison of APET Results With and Without T.I Hot Leg Breaks and SGTRs l PDS Group 6 ATWS.................................. 2.100 e 2.5.12 Probability of CF and Probability of Case for CF at VB for Surry PDS Group 1- Slow SB0.............................. 2.101 2.5.13 Comparison of Selected Bins for - PDS Group 1. Slow SBO for Two Samples for Surry.......................... 2.106 2.1 14 Results of the Accident Progression Analysis
-for Surry Fire Initiators.
PDS Group Fire..................................... 2.108 xvii
2.5 15 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Croup EQ 1, LOSP (No SBO): High Acceleration, LLNL Hazard Distribution........................... 2.111 2.5.16 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Group EQ 2, SB0: High Acceleration....................... .... 2.114 2.5.17 Results of the Accident Progression Analysis ] for Surry Seismic Initiators. PDS Group EQ 3 - LOCAs: High Acceleration, i LLNL Hazard Distribution........................... 2.117 2.5.18 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Croup EQ 1, LOSP (No SBO): Low Acceleration, I LLNL Hazard Distribution........................... 2.120 j 2.5.19 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Croup EQ 2, SB0: Low Acceleration, , LLNL Haza rd Di s t ribu t ion . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.122 2.5.20 Results of the Accident Progression Analysis ; for Surry Seismic Initiators. PDS Group EQ 3, LOCAs: Low Acceleration. - LLNL Hazard Dis tribution. . . . . . . . . . . . . . . . . . . . . . . . . . . 2.124 2.5.21 Comparison of the Accident Progression Analysis Results for Surry Seismic Initiators; With and Without Containment Failures at the Start of the Accident; LLNL Hazard Distribution............. 2.127 2.5.22 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Group EQ 1, IDSP (No SBO): High Acceleration, EPRI Hazard Distribution........................... 2.129 2.5.23 Results of the Accident Progression Analysis + for Surry Seismic Initiators. PDS Group EQ 2, SB0: High Acceleration.
- EPRI Hazard Distribution........................... 2.131 2.5.24 Results of the Accident Progression Analysis i
for Surry Seismic Initiators. PDS Group EQ 3, . LOCAs: High Acceleration, EPRI Hazard Distribution........................... 2.133 l l 2.5.25 Results of the Accident Progression Analysis I for Surry Seismic Initiators. PDS Croup EQ 1 LOSP (No SBO): Low Acceleration, f EPRI Hazard Distribution........................... 2.136 xviii
I 2.5.26 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Group EQ 2, l SB0: Low Acceleration. l EPRI llazard Distribution........................... 2.138 2.5 27 Results of the Accident Progression Analysis i for Surry Seismic Initiators. PDS Group EQ 3, ' LOCAs: Low Acceleration, EPRI Hazard Distribution......................... . 2.140 3.2 1 Isotopes in Each Radionuclide Release Glasa....... 3.3 3.2 2 Variables Sampled in the Source Term Analysis............................... 3.11 3.3 1 Hean Source Terms for Surry Internal Initiators. PDS Group 1: Slow SB0......... 3.31 3.3 2 Hean Source Terms for Surry Internal Initiators. PDS Group 2: IDCAs............. 3.33 3.3 3 Mean Source Terms for Surry Internal Initiators. PDS Group 3: Fast SB0.......... 3.35 3.3 4 Mean Source Terms for Surry Internal Initiators. PDS Group 4: Event V........... 3.37 3.3-5 Mean Source Terms for Surry Internal Initiators. PDS Group 5: Transients........ 3.39 3.3 6 Mean Source Terms for Surry Internal Initiators. PDS Group 6: ATWS.............. 3.41 3.3-7 Mean Source Terms for Surry Internal Initiators. PDS Group 7: SGTRs............. 3.43 3.3 8 Summary of Variables Sampled Surry Internal Initiators........................... 3.55 3.3 9 Summary of Rank Regression Analyses for Anrual Release Rates .............................. 3.57 3.3-10 Summary of P.ank Regression Analyses for Annual Reltase Rates at Surry for Two Samples for Internal Initiators................................ 3.59 3.3 11 Mean Source Terms for Surry Fi r e I n i t i r.t o r s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.74 3.3 12 Mean Source Terms for burry Seismic Initiators. PDS Group EQ 1: IDSP liigh Acceleratien. LIRL Hazard Distribution. . . . . . . 3.75 xix
3.3 13 Mean Source Terms for Surry Seismic Initiators. PDS Group EQ 2: SB0 High Acceleration. LLNL Hazard Distribution....... 3.77 3.3 14 Mean Source Terms for Surry Seismic Initiators. PDS Group EQ 3: LOCAs High Acceleration. LLNL Hazard Distribution....... 3.79 3.315 Mean Source Terms for Surry Seismic Initiators. PDS -up EQ 1: LOSP Low Acceleration. LLNL r . ard Distribution. . . . . . . . 3.81 3.3 16 Mean Source Terms for Surry , Seismic Initiators. PDS Group EQ 2: SB0 Low Acceleration. LLNL Hazard Distribution........ 3.83 . 3.3 17 Mean Source Terms for Surry Seismic Initiators. PDS Group EQ 3: LOCAs , Low Acceleration. LLNL Hazard Dis tribution. . . . . . . . 3.85 ' 3.3 18 Mean Source Terms for Surry ) Seismic Initiators. PDS Group EQ 1: LOSP High Acceleration. EPRI Hazard Distribution........ 3.87 3.3 19 Mean Source Terms for Surry Seismic Initiators. PDS Group EQ 2: SB0 High Acceleration. EPRI Hazard Distribution........ 3.88 3.3 20 Mean Source Terms for Surry Seismic Initiators. PDS Group EQ 3: LOCAs High Acceleration. EPRI Hazard Distribution........ 3.89 3.3 21 Mean Source Terms for Surry Seismic Initiators. PDS Group EQ 1: LOSP Low Acceleration. EPRI Hazard Distribution......... 3.90 3.3 22 Mean Source Terms for Surry Seismic Initiators. PDS Group EQ 2: SB0 Low Acceleration. EPRI Hazard Distribution......... 3.91 3.3 23 Mean Source Terms for Surry Seismic Initiators. PDS Group EQ 3: LOCAs Low Acceleration. EPRI Hazard Dis tribution. . . . . . . . . 3.92 3.4 1 Summary of Early and Chronic Health Effect Weights for Internal Initiators................................ 3.96 3.4 2 Distribution of Source Terms with Nonzero Early Fatality and Chronic Fatality Weights for Internal Initiators................................ 3.100 t XX l l 1
{ i i 3.4 3 Distribution of Source Terms with Zero Early , Fatality and Nonzero Chronic Fatality Weights for l Internal Initiators................................ 3.103 3.4 4 Mean Source Terms Resulting from Partitioning for Internal Initiators. Surry..................... 3.104 3.4.5 Summary of Early and Chronic Health Effect Weights for Fire Initiators................. 3.108 3.4.6 Distribution of Source Terms with Nonzero Early i Fatality Weight and Nonzero Chronic Fatality Weight for Fire Initiators......................... 3.109 3.4 7 Distribution of Source Terms with Zero Early i Fatality Weight and Nonzero Chronic Fatality Weight for Fire Initiators . . . . . . . . . . . . . . . . . . . . . . . . . _3.110 3.4 8 Mean Source Terms Resulting from Partitioning for r ire Initiators................... 3.111 3.4 9 Summary of Early and Chronic Health Effect Weights for Seismic Initiators LLNL Hazard Distributions.......................... 3.115 l 3.4 10 Distribution of Source Terms with Nonzero Early a Fatality and Chronic Fatality Weight for Seismic
- Initiators: LLNL Hazard Distributions.............. 3.116 3.411 Distribution of Source Terms with' Zero Early.
Fatality Weight and Nonzero Chronic Fatality Weight for Seismic Initiators: .
- LLNL Hazard Distributions.......................... 3.118 3.4 12_Mean Source Terms Resulting from -
Partitioning for Seismic Initiators t LLN1 Hazard Distribution: High-PCA................. 3.120 l 3.4 13 Summary of Early and Chronic Health ( Effect Weights for Seismic Initiators i EPRI Hazard Distributions........................... 3.128' l 3.4 14 Distribution of Source Terms with Nonzero Early Fatality and Chronic ~ Fatality Weights for Seismic
- Initiators: EPRI Hazard Distributions.............. 3.130 3.415 Distribution of Source Terms with Zero Early Fatality Weight and Nonzero Chronic Fatality Weight for Seismic Initiators:
- EPRI Hazard Distributions.......................... 3.132 xxi
i-3.4 16 Mean Source Terms Resulting from Partitioning for Seismic Initiators l EPRI Hazard Distribution: High PCA................. 3.133 4.1 1 Definition of Consequence Analysis Results.......... 4.3 i 4.2 1 Site Specific Input Data for Surry MACCS Calculations................................. 4.5 4.2.2 Shielding Factors Used for Surry MACCS Calculations................................. 4.6 4.3 1 Mean Consequence Results fdr , Internal Initiators................................ 4.9 : 4.3 2 Mean Consequence Results for Fire Initiators............................. ...... 4.15 I' 4.3 3 Mean Consequence Results for Seismic Initiators. LLNL Hazard Distributions i Low PGA........ .............................. .... 4.19 4.3 4 Hean Consequenci sults for Seismic Initiators. Li atard Distributions High PGA..... ................................. 4.21 4'3 5 Mean Consequence Results for Seismic
)
Initiators. EPRI Hazard Distributions > Low PGA............................................ 4.25
.4.3 6- Mean Consequence Results-for Seismic '
Initiators. EPRI Hazard-Distributions ' High PGA........................................... 4.27 5.1 1 Distributions for Annual Risk'at Surry, Due to Internal Initiators......................... 5,11 5.1 2 Fractional PDS' Contributions to Annual Risk , at Surry Due to-Internal Initiators................ S.15 5.1 3 Fractional PDS Contributions to Annual Risk at Surry Due to Internal Initiators................ 5.18 5.1 4 Summary of Rank Regression Analyses fot Annual Risk at Surry for Internal Initiators... ............................ 5.25 5.1-5 Rank Regression Summaries for. Annual Risk et Surry for Internal Initiators Using Product Variables,........ .................. 5.28 xxii
i I r 5.1 6 Summary of Rank Regression Analyses i for Annual Risk at Surry for m Two Samples for Internal Initiators....... ........ 5.29 5.2 1 Distributions for Annual Risk Due to Fire Initiators......................................... 5.39 : 5.3 1 Distributions for Annual Risk Due to Fire Initiators. LLNL H9 ' stributions.............. 5.50 5.3-2 Fractional PDS Contric atons to Annual Risk at Surry Due to Seismic ' Initiators. LLNL Hazard Distributions. . . . . . . . . . . . . . 5.51 5.3 3 Distributions for Annual Risk Due to Seismic Initiators. LLNL Hazard Distributions.............. 5.52 5.3 4 Fractional PDS Contributions to Annual Risk at Surry Due to Seismic
. Initiators. LLNL Hazard Distributions Three PCA Subgroups................................ 5.52 ,
5.4 1 Distributions for Annual Risk Due to Seismic ! s. Initiators. EPRI Hazard Distributions.............. 5.58 5.4 2 Fractional PDS Contributions to Annual Risk at Surry Due to Seismic ' Initiators. EPRI Hazard Distributions.............. 5.59 4
- 5. 5 -l' Distributions for Annual Risk Due to Seismic Initiators......................................... 5.61 ,
5.5 2 Fractional PDS Contributions to L Annual Risk at Surry Due to Seismic ! Initiators......................................... 5.62 5.5 3 Fractional-APB Contributions to Annual Risk at Surry Due to Seismic . H- Initiators.... .................................... 5.63 s xxiii
FOREWORD This is one of numerous documents that support the preparation of the final NUREG 1150 document by the NRC Office of Nuclear Regulatory Research. Figure 1 illustrates the documentation of the accident progression, source term, consequence, and risk analyses. The direct supporting documents for the first draf t of NUREG 1150 and for the revised draft of NUREG 1150 are given in Table 1. They were produced by the three interfacing programs at Sandia National Laboratories that performed the work: the Accident Sequence Evaluation Program (ASEP), the Severe Accident Risk Reduction Program (SARRP), and the PRA Phenomenology and Risk Uncertainty Evaluation Program (PRUEP). The Zion volumes were written by Brookhaven National Laboratory and Idaho National Engineering Laboratory. The Accident Frequency Analysis, and its constituent analyses, such as the Systems Analysis and the Initiating Event Analysis, are reported in NUREG/CR 4550. Originally, NUREG/CR 4550 was published without the designation " Draft for Comment. " Thus, the current revision of NUREG/CR-4550 is designated Revision 1. The label Revision 1 is used consistently on all volumes, including Volume 2 which was not part of the original documentation. NUREG/CR 4551 was originally published as a " Draft for Comment". While the currenc version could have been issued without a revision indication, all volumes of NUREC/CR 4551 have been designated Revision 1 for consistency with NUREC/CR 4550. The material contained in NUREG/CR 4700 in the original documentation is now contained in NUREC/CR 4551; NUREG/CR 4700 is not being nvised. The contents of the volumes in both NUREG/CR 4550 and NUREC/CR 4551 have been altered. In both documents now, Volume 1 describes the methods utilized in the analyses, Volume 2 presents the elicitation of expert judgment, Volume 3 concerns the analyses for Surry, Volume 4 concerns the analyses for Peach Bottom, and so on. Note that the Surry volume of NUREG/CR 4551, now Volume 3, was Volume 1 in the original Draft for Comment version of NUREG/CR 4551, published in February 1987. - The Surry plant was also treated in Volume 1 of the original Draft for Comment version of NUREG/CR 4700. The topics covered in NUREG/CR 4700 are now included in NUREG/CR 4551. In addition to NUREG/CR 4550 and NUREG/CR 4551, there are several other reports published in association with NUREG 1150 that explain the methods used, document the computer codes that implement these methods, or present the results - of calculations performed to obtain information specifically for this project. These reports include: NUREG/CR-5032, SAND 87 2428, 'Modeling Time to Recovery and Initiating Event Frequency for Loss of Off-site Power Incidents at Nuclear Power Plants," R. L. Iman and S. C. Hora, Sandia National Laboratories, Albuquerqt.e, NM, January 1988. NUREG/CR 4840, SAND 88 3102, " Recommended Procedures for Simplified External Event Risk Analyses," M. P. Bohn and J. A. Lambright, Sandia National Laboratories, Albuquerque, NM, December 1988, i l Xxiv l l . , . _ . . .
l l 1 NUREG/CR 5174, SAND 88 1607, J. M. Criesneyer and L. N. Smith, "A Reference Manual for the Event Progression and Analysis Code (EVNTRE)," Sandia National Laboratories, Albuquer:tue, NM, 1989.
)
NUREG/CR 5380, SAND 88 2988, S. J. Higgins, "A User's Manual for the Post Processing Program PSTEVNT," Sandia National Laboratories, Albuquerque, NM, 1989. NUREG/CR 5360, SAND 89 0943 H. N. Jow, W. B. Murfin, and J. D. Johnson. *XSOR Codes User's Manual," Sandia National Laboratories, Albuquerque, NM, 1989. NUREG/CR 4624, BMI 2139, R. S. Denning et al., "Radionuclide Release 3 Calculations for Selected Severe Accident Scenarios," Volumes I V, Batte11e's Columbus Division, Columbus, 011, 1986. , NURE0/CR 5062, BMI 2160, M. T. Leonard et al., " Supplemental Radionuclide Release Calculations for Selected Severe Accident Scenarios," Battelle Columbus Division, Columbus, OH, 1988. NUREG/CR 5331, SAND 89 0072, S. E. Dingman et al., "MELCOR Analyses for Accident Progression Issues," Sandia National Laboratories, Albuquerque, NM, 1989. NUREG/CR 5253, SAND 88 2940, R. L. Iman, J. C. Helton, and J. D. - Johnson, "A Users Guide for PARTITION: A Program for Defining the Source Term / Consequence Analysis Interfaces in the NUREG-1150 Probabilistic Risk Assessments," Sandia National Laboratories, Albuquerque, NH, 1989. NUREG/CR 5382, SAND 88 2695, J. C. Helton et al., " Incorporation of Consequence Analysis Results into the NUREG-1150 Probabilistic Risk Assessments," Sandia National Laboratories, Albuquerque, NM, 1989. l l l XXV
i..
- SUPPORT DOCUMENTS TO NUREG-1150 j v -
j NUREG-1150
/(NRC Staff) l
' i i EVALUATION OF SEVERE ACCIDENT RISKS NUREG/CR-4551 i 20EMOOS MR N PARAREETERS SURRY PEACH 90TTOtt SEOUOYAH GRANO GUtf ZION NUREGICR-4550 Vol.1 Vol. 2 Vol. 3 Vol. 4 - Vol. 5 Vol. 6 Vol. 7 n x r e 5 - t 3 . i
> $ E o M o 5 . =- g 8 8 $ 5 E E E E E
e
- 8 [m SE 3* $ $ $ $ $
E O E E E O w o 1 _8, . w o E w x a
=
t z a 2
=
e e
=
w o
=
a e w o
=
a e w o
=
a e e w o
=
a e u I s: z
- 6 E E R = = E -
E 5 m 5 a 5 E 5 a 5 0 y 5 $ 5 8 E 2 5 0 5 5 5 5 5 6 5 5 U E 8 - R.
- 5 5 8 % 5 W I & 1 a 1 & 's & 1 o 1 <
vlvlslslaefols E u m m s x < ss p< < als < < alv < v l g v& Figure 1. Back-End Documentation for NUREG-1150.
Table 1. NUREG-ll50 Analysis De.:usentation Oririnal Documentation NUREG/CR-4550 NUREC/CR-4551 NUREC/CR-4700 Analysis of Core Damage Frequency Evaluation of Severe Accident Risks Containment Event Analysis From Internal Events and the Potential for Risk Reduction for Potential Severe Accidents Vol. 1 Methodology Vol. 1 Surry Unit 1 Vol. 1 Surry Unit 1 2 Summary (Not Published) 2 Sequoyah Unit 1 2 Sequoyah Unit 1 3 Surry Unit 1 3 Peach Bottom Unit 2 3 Peach Botton Unit 2 4 Peach Botton Unit 2 4 Grand Gulf Unit 1 4 Grand Gulf Unit 1 5 Sequoyah Unit 1 6 Crand Gulf Unit 1 7 Zion Unit 1 Revised Documentation y NUREG/CR-4550, Rev. 1. Analysis of Core Damage Frequency NUREG/CR-4551, Rev. 1, Eval. of Severe Accident Risks g Vol. 1 Methodology Vol. 1 Part 1, Methodology; Part 2. Appendices 2 Part 1 Expert Judgment 211 cit. Expert Panel 2 Part 1 In-Vessel Issues Part 2 Expert Judgment Elicit. Project Staff Part 2 Containment Ioads and MCCI Issues Part 3 Structural Issues Part 4 Source Tern Issues Part 5 Supporting Calculations Part 6 Other Issues Part 7 MACCS Input 3 Part 1 Surry Unit 1 Internal Events 3 Part 1 Surry Analysis and Results Part 2 Surry Unit 1 Internal Events App. Part 2 Surry Appendices Part 3 Surry External Events 4 Part 1 . Peach Botton Unit 2 Internal Events 4 Part 1 Peach Botton Analysis and Results Part 2 Peach Botton Unit 2 Int. Events App. Part 2 Peach Botton Appendices Part 3 Peach Botton Unit 2 External Events 5 Part 1 Sequoyah Unit 1 Internal Events 5 Part 1 Sequoyah Analysis and Results Part 2 Sequoyah Unit 1 Internal Events App. Part 2 Sequoyah Appendices 6' Part 1 Grand Gulf Unit 1 Internal Events 6 Part 1 Grand Gulf Analysis and Results Part 2 Grand Gulf Unit 1 Internal Events App. Part 2 Grand Gulf Appendices 7 Zion Unit 1 Internal Events 7 Part 1 Zion Analysis and Results Part 2 Appendices
ACRONYMS AND ANITIALISMS ADV atmospheric dump valves ATW auxiliary feedwater APWS auxiliary feedwater system APB accident progression bin APET accident progrecsion event tree ASEP accident sequence evaluation ATWS anticipated transient without scram BMT basemat melt through BNL Brookhaven National Laboratory BWR bciling water reactor CCF common cause failure CCDF complementary cumulative distribution function 001 core concrete interaction CCW component cooling water CDP cumulative distribution function CF containment failure CFW chronic fatality weight CilR ev.tainment heat removal CSS containment spray system C3T condensate storage tank C.iT condensate storage tank DCil direct containment heating DC diesel generator EACPS emergency ac power system ECCS emergency core cooling system (s) EP early fatalities EFW early fatality weight E0P emergency operating procedures EPRI Electric Power Research Institute ESW emergency service water EVSE ex vessel steam explosion FSAR final safety analysis report flPI high pressure injection llPIS high pressure injection system llPRS high pressure recirculation system llPME high pressure melt ejection llRA human reliability analysis ICIR in core instrumentation room INEL Idaho National Engineering Laboratory IVSE in vessel steam explosion LCF latent cancer fatalities 12tS Latin flypercube Sampling 11NL iawrence LLvermore National laboratory LOCA loss of coolant accident xxviii
i 1 i l LOSP loss of offsite power LPI low pressure injection LPIS low pressure injection system LPRS low pressure recirculation system LVR light water reactor MCDF mean core damage frequency MDP motor driven pump MFWS main feedwater system MOV motor operated valve MSIV main steam isolation valve MSL main steam line NRC Nuclear Regulatory Commission PCC partial correlation coefficient PDS plant damage state PGA peak ground acceleration PORV power operated relief valve PRA probabilistic risk analysis PRCC partial rank correlation coefficient PVR pressurized water reactor . PER pressurizer RCP reactor coolant pump RCS reactor coolant system RHR residual heat removal RPS reactor protection system RSS Reactor Safety Study RWST refueling water storage tank SB0 station blackout SERG steam' explosion review group 50 steam generator SGTR steam generator tube rupture SLC standby liquid control SNL Sandia National Laboratories SOV solenoid operated valve h SRV safety relief valve
. STSO source term subgroup TAF top of active fuel , TDP turbiot, driven pump TI tempertrure induced UTAF uncovering of TAF VB vessel breach i
XXIX
ACl'SOVLEDOMENTS Many people at Sandia National Laboratories and elsewhere contributed to the work reported in this volume. Julie Gregory provided many helpful comments during the development of the accident progression event tree. Dave Williams assisted in resolving numerous containment loading and source term problems. Ann Shiver performed the i.atin Hypercube Sampling runs and j produced many of the figures. Jay Johnson, Hong Nian Jow, Jerry Sprung, ! and Sarah Higgins made the MACCS runs and produced figures and tables for j this report. Tom Brown and Arthur Payne made many helpful suggestions. ' Bob Bertucio and Tim Wheeler worked with us to iron out interface problems j for the internal initiators. John 1.ambright provided interface information j for the fire initiators, and Mike Bohn performed the same function for the seismic initiators. Steve Hora assisted in deriving the method for determining the mode of containment failure for fast pressure rise. Ron Iman provided regression results and the distributions used for the recovery of offsite power. John Kelly and Ed Boucheron provided information and insights about the core melt progression. Reeta Garber was i ' of great help in preparing this document. Jan Frey provided invaluable general support in numerous ways. This project would not have been possible without the cooperation of all those who participated in the t expert panels, and those who organized and participated in the training, discussion and elicitation sessions. 1.as t , but certainly not least, we gratefully acknowledge the support of Elaine Gorham and Fred Harper in leading this project to its completion, B i XXX
l l
SUMMARY
S.1 Introduction The United States Nuclear Regulatory Commission (NRC) has recently i completed a major study to provide a current characterization of severe accident risks from light water reactors (LVRs). This characterization is derived from integrated risk analyses of five plants. The summary of this study, NUREG 1150,1 has been issued as a second draf t for comment. The risk assessments on which NUREG 1150 is based can generally be characterized as consisting of four analysis steps, an integration step, an1 an uncertainty analysis step:
- 1. Accident frequency analysis; the determination of the likelihood and nature of accidents that result in the onset of core damage,
- 2. Accident progression analysis: an investigation of the core damage process, both within the reactor vessel before it fails and in the containment afterwards, and the resultant impact on the ;
containment.
- 3. Source term analysis: an estimation of the radionuclide transport within the reacte ; coolant system (RCS) and the containment, and the magnitude of the subsequent releases to the environment, 4 Consequence analysis: the calculation of the offsite conseqaences, primarily in terms of health effects in the general population, 5, Risk integration: .the assembly of the outputs of the previous tasks i
into an overall expression of risk, ; 6, Uncertainty analysis: the propagation of the uncertainties in the initiating events, failure events, accident progression branching ratios , and parameters, source term , parameters threugh the first three analyses above, and the determination of which of these uncertainties contributes the most to the uncertainty in risk, This volume presents the details of the last five of the six steps listed above for the Surry Power Station, Unit 1, The first step is described in NUREC/CR-4550.2 S,2 Overview of Surry Power Station. Unitj The Surry Power Station, Unit 1 is operated by - the Virginia Electric Power Company and is located on the south bank of the James River in southeastern Virginia, about 10 miles south cf Williamsburg, VA, and approximately 35 miles northwest of Norfolk, VA. Two units are located on the site; Unit 2 is essentially identical to Unit 1. Surry Unit I nuclear raactor is a 2441 MWt pressurized water reactor (PWR) i designed and built by Wes tinghouse . The RCS has three U-tube steam generators (SGs) and three reactor coolant pumps (RCPs). The containment S.1
and the balance of the plant were designed and built by Stone and Webster. Unit 1 began commercial operation in December 1972. Table S.1 summarizes the design features of the plant that are relevant to severe accidents. Of particular interest is the large, dry containment that is kept at subatmospheric pressure during operation. The high pressure injection system (HPIS) and auxiliary feedwater system (APWS) crossties between the two units make the core damage frequency lower than it would be without the crosstics. S.3 Description of the Integrated Risk Analysis Risk is determined by combining the results of four constituent' analyses: the accident frequency, accident progression, source term, and consequence analyses. Uncertainty in risk is determined by assigning distributions to important variables, generating a sample from these variables, and propagating each observation of the sample through the entire analysis. The sample for Surry consisted of 200 observations involving variables from : the first three constituent analyses. The risk analysis synthesizes the results of the four constituent analyses to produce measures of offsite risk and the uncertainty in that risk. This process is depicted in Figure S.I. This figure shows, in the boxes, the computer codes utilized. The interfaces between constituent analyses are shown between the boxes. A mathematical summary of the process, using a matrix representation, is given in Section 1.4 of this volume. The accident frequency analysis uses event tree and fault tree techniques to investigate the manner in which various initiating events can lead to core damage and the frequency of various types of accidents. Experimental data, past observational data, and modeling results are combined to produce frequenc; 9timates for the minimal cut sets that lead to core damage. A minimal c sei is a unique combination of initiating event and individual hardware w operator _ failures. The minimal cut sets are grouped into plant damage states (PDSs), where all minimal cut sets in a PDS provide a similar set of initial conditions for the subsequent accident progression analysis. Thus, the PDSs form the interface between the accident frequency analysis and .the accident progression analysis. The outcome of the accident frequency analysis is a frequency for each PDS or group of PDSs for.each observation in the sample. The accident progression analysis uses large, complex event trees to determine the possible ways in which an accident might evolve from each
~
PDS. The definition of each PDS provides enough information to def'.ne the initial conditions for the accident progression event tree (APET) naalysis. Past observations , experimental data, mechanistic code calculations, and expert j udgment were used in the development of the model for accident progression that is embodied in the APET and in the selection of the branch-probabilities and parameter values used in the APET. Due to the large number of questions in the Surry APET and the fact that many of these questions have more than two outcomes, there are far too many paths through the APET to permit their individual consideration in subsequent source term and consequence analysis. Therefore, the paths through the trees are grouped into accident progression bins (APBs), where each bin is a group of paths through the event tree that define a sin.ilar set of conditions for S.2
ACCIDENT ACCIDENT SOURCE FREQUENCY- PROGRESSION TERM CONSEQUENCE ANALYSIS ANALYSIS ANALYSIS ANALYSIS a acroLOGCAs. Pt ANT SYS TEU EQUfDE4 7 23 ME47 PAR AMEmic WFCawAftog A80uf 6*E ALN EFFECTS DATA WODELS itm1Engal CPER ATOR FastuRE PROGRES904 r,55,og paODUCT TA A'vSPOR T wE TEOmCLOGCAL D ATA AND ENTERNAL EVENTS) DATA E VEN T TREE mNo nEvoval DEvocmAM OATA
~
, ,, F ACC m ~1 ,.
o, b SETS 7 UAC
- T o-Ac STM y ... E S,,
y Scu CE.,.*.
- : EWME -
XSOR - VACC.S PR Aut$ : RIS'( E XPRE S90NS FRE QUE **OE S N [ O ' U *C $ N rmE QUE ssC'ES
,, ,,,, m,,
I WFORUATION FROM *CCMTAmmsE*sTLCADmG E PE AFORwars0E rssuf S! E xPg n:MENTS ec ' F AONT ENO tSSUE S l ,4 WESSEL WELT PRWE SSON 6S'ES SOUaCETtRufSSUtS DETA# LED CDESS Lame HvPERCUBE SnuPLE Figure S.I. Overview of Integrated Plant Analysis in NUREG-1150
source term analysis. The properties of each accident progression bin define the initial conditions for the estimation of a source term. The result oifthe accident progression analysis is a probability for each APB, conditional on the occurrence of a PDS, for each observation in the sample. Table S.1 Design Features Relevant to Severe Accidents Surry Unit 1 Emergency Core llPIS Cooling (ECCS) Three motor-driven pumps (MDPs) (charging pumps) Suction from low pressure injection system (LPIS) discharge Dedicated two train cooling system Crosstie to Unit 2 IIPIS LPIS Two self-cooled MDPs Suction from the refueling water rcorage tank (RWST) or containment sump Accumulators Three accumulators containing borated water pressurized to 650 psig Emergency Core lleat Auxiliary Feedwater System (AFWS) Removal Two MDPs and one turbine driven pump (TDP) Crosstie to Unit 2 AFWS Feed and Bleed Utilizes llPIS and power-operated relief valves (PORVs) Reactivity Control -Reactor Protection System (RPS) (automatic scrau) Manual scram Emergency Electrical AC Electrical Power Power Three diesel generators (DGs) for both units Dedicated DG always aligned to Unit 1 Swing DG may be aligned to either Unit DCs are self-contained and=self-cooled DC Electrical Power Station batteries designed to last 2 h Batteries estimated to last 4 h due to design margin and load-shedding S.4
e
.i Table S.1 (continued) [
Containment-Structure Built of reinforced concrete with welded steel
-liner Volume is about 1,8 million ft3 Design pressure is 45 psig Maintained at 10 psia-during reactor operation containment Heat All emergency heat removal is by spray systems Removal- Two-spray injection trains Four spray recirculation trains ..
All spray trains are independent I Cooling of Reactor Two independent sources of RCP seal cooling Coolant Pump Seals Component cooling water system Charging (HPIS) pumps RCp seal failures likely only in SB0 accidents Service _ Water = Supplied from an elevated canal' Canal likely to drain during extended SB0s , l Delay in restoring ECCS after power recovery Service water for cooling the ECCS pumps , comes from the canal ,! The canal must be refilled before starting the ECCS pumps
] !
, Sump and. Reactor No connection between sump and cavity at a low Cavity level.in the containment- i z1 1 Sump ca; city 'is large , so there is . no ,)
possibility .of overflow' from the ~ sump filling i the cavity Cavity,can be filled with. water only by operating the containment sprays
-l a
h A ~ source term - is calculated for each APB with a i non zero conditional
~
j probability for each: cbservation in the sample by SURSOR,, a fast running 1 p~ parametric computer code, SURSOR is not a detailed mechanistic-model; it t is not. designed to be a realistic simulation of fission product transport, i physics,= and chemistry. Instead, SURSOR integrates the results of- many detailed codea and the conclusions of many experts. Most=of the parameters-usedJto calculate fission product release fractions in SURSOR are sampled from -distributions : provided by an expert panel, Because of the large number of APBs, it is necessary to use a fast-executing code like SURSOR, S,5
The number of APBs for which source terms are calculated is so large that it is not computationally practical to perform a consequence calculation for every source term. As a result, the source terms had to be combined into source term groups. Each source term group is a collee P n of source terms that result in similar consequences. The process of catermining which APBs go to which source term group is called partitioning. It involves considering the potential of each source term group to cause early fatalities and latent cancer fatalities. The result of the sou ce term calculation and subsequent partitioning is that each APB fot each observation is assigned to a source term group. A consequence analysis is performed for each source term group, generating both mean consequences and distributions of consequences. As each APB is assigned to a source term group, the consequences are known for each APB of each observation in the sample. The frequency of each PDS for each obser-vation is known from the accident frequency analysis, and the conditional probability of each APB is determined for each PDS group for each obser-vation in the accident progression analysis. Thus, for each APB of each observation in the sample, both frequency and consequences are determined. ! The risk analysis consists of assembling and analyzing all these separate estimates of offsite risk. S,4 Results of the Accident Freauency Analysis The accident frequency analysis for Surry is documented elsewhere.2 This section only summarizes the results of the internal, fire, and seismic accident frequency analyses since these form the starting point for the analyses that av. covered here. Table S.2 lists four summary measures of the core damage frequency distributions for Surry for the seven internally initiated PDS groups, the fire PDS group, and the six seismic PDS groups. The four summary measures are the mean, and the 5th, 50th (median) and 95th percentiles. The'25 internally initiated PDSs that had mean frequencies above 1.0E-7/R-yr are placed into the seven PDS groups listed in Table S.2. These 25 PDSs account for over 99% of the total mean core damage frequency (MCDF) of 4.1E-5/R yr. In both SB0 groups, offsite power is lost and the DCs fail to start and run. In the slow SB0 group, the steam turbine driven (STD) AFWS operates until the batteries are depleted; in the fast SBO group the STD AWS fails. In both SB0 groups, core melt may be arrested before the i vessel fails, if offsite power is recovered in time. The loss-of coolant accident (LOCA) PDS group consists of accidents initiated by breaks of all four. sizes (A, S, i 5, 2 and S3 ). In some of the PDSs in this group, the LPIS is operating at the onset of core damage, therefore, it is possible to arrest core degradation before the vessel lower head fails' for these PDSs. Event V is initiated by the failure of two check valves that isolate LPIS piping from the RCS. The check valve failures expose the low pressure l piping to full primary system pressure, causing the pipes to rupture. l Since the break is outside containment, the break fails both the RCS and the injection system, and bypasses the containment. The Transient group consists of two PDSs that have failure of both the AFWS and Feed and Bleed. Core damage arrest is possible for this PDS group if the RCS pressure can S.6
Table S.2 Surry Core Damage Frequencies Internal, Fire, and Seismic Initiators Core Damage Freauenev (1/R-vri % Mean TCP PDS Groun St Median Mean 95% rreauency 1 Slow SB0 1.6E 06 1.1E 05 2.2E 05 6.4E 05 56 2 LOCAs 1.2E 06 3.9E 06 6.1E 06 2.0E 05 15 3 Fast SB0 1.2E-07 1.5E-06 5.4E 06 2.1E 05 13 4 Event V. 3.6E-11 4.9E 08 1,6E 06 8.2E 06 4 5 Transients 1.1E 07 B.2E 07 1.8E-06 5.5E-06 4 6 ATWS 2.9E 08 4.2E 07 1.4E-06 6.5E 06 4 7 SGTRs 4.5E 07 1.4E 06 1.8E-06 4.7E 06 4 Internal. Initiators 9.8E 06 2.5E 05 4.1E 05 1.0E 04 FIRE. 2.3E 6 8.4E-6 1.1E 5 2.6E-5 Seismic Initiators LJJH Hazard Distribution -- Peak Ground Acceleration > 0.6 g EQ 1 LOSP 9.1E 9 5.8E 7 9.4E 3.4E-5 5 EQ 2'SBO 2.2E 8 9.2E 7 1.1E 5 5.3E-5 6 EQ 3 LOCAs= 9.5E-9 5.5E 7 7.5E-6 3,6E 5 4 ; High PGA 5.6E 8 2.4E-6 2.8E 5 1.3E-4 15 LLNL Hazard Distribution -- Peak Ground Acceleration < 0.6 g , 1 EQ 1-LOSP 1'.0E 7 - 6.2E-6 8.1E-5 3.5E-4 42 EQ 2 SB0 1.2E 7 5.8E 6 6.8E-5 2.9E-4 35 EQ 3'LOCAs 1.8E 8 1.1E 1.5E-5 7.3E-5 8 Low PGA 4.9E-7 1.5E-5 1.6E-4 6.4E-4 85 All PGA - LLNL 5.3E-7 1.8E-5 1.9E-4 7.6E-4 S.7
I Table S.2 (contirued)
, Seismic Initiators Core-Damace Freauency (1/R vri % Mean TCD PDS , Grouo- 5% Median Mean 95% Preauency EPRI 11azard Distribution Peak Ground Acceleration > 0.6 g EQ 1 14SP 8.8E 9 2.5E-7 1.1E 6 4.9E 6 4 EQ'2 SB0 1. 8E 8- .3.2E-7 1.1E 6 4.7E 6 4 .
EQ 3 IDCAs 2.1E 9 2.0E.7 9.8E 7 5.5E 6 3 l liigh.PGA 3.3E-8' 9.9E-7 3.2E 6 1.4E-5 11 EPRI llazard Distribution --' Peak Ground Acceleration < 0.6 g EQ-1 1 ASP ~ 9.6E 8 3.6E 6 1.4E-b 7.6E-5 50
- EO~2 SB0 1.3E 2.5E 6 8.4E 6 3.5E-5 30 !
EQ 3 LOCAs 5.3E 9 5.0E-7 2.5E-6 1.2E 5 9 ~! 1
~ Low PGA- 3dE-7 8.6E 6 2.5E 5 1.3E 4 89
'All PGA EPRI 3.7E-7 9.4E 6' 2.8E 1.4E 4
- be reduced since the LPIS is operable in c,ne PDS' and both LPIS and llPIS are
- operable -in the other. The ATWS group contains three PDSs in which the nuclear reaction is not brought under control at the start of. the accident.
The four . PDSs ' that comprise .the steam generator tube rupture (SGTR) group-include- two PDSs, in which the safety relief valves (SRVs) in the secondary system a tick open - ("11" SGTRs)', and two PDSs in which these SRVs - reclose-
- after opening (_"G" SGTRs), l
~ There are only four fire PDSs, all placed together in a 'singlea fire PDS
. group. - Significant fire , locations for Surry are: emergency switchgear room,< auxiliary build'ng, control - room, and cable vault and tunnel.- Coro damage arrest is not possible for the fire PDSs. since the fire destroys.
either the control.or motive power cables. .i The seismic PDS frequencies are calculated-for two different sets of hazard ' distributions: one generated by Lawrence - Livermore ; National Laboratory 1(LLNL): and one generated 'by the Electric Power Research Institute (EPRI), Table S . 2 shows that use of the EPRI- hazard distributions results in'a
-total core damdge frequency that is almost an order of magnitude lower than the t total . core damage frequency obtained using the LLNL hazard distribu-tions. The seismic PDSs are divided into three groups. Ilowever, as the evacuation response differs for earthquakes with a peak ground acceleration i (PGA) over.0.6 g:and earthquakes with a PGA below 0.6 g, it is necessary to subdivide those groups on this-basis. The loss of offsite power (LOSP) PDS 'i Lgroup consists of accidents criggered by the LOSP, but in which the DGs l'
S.8
start and:run so there is no SBO. In the SB0 accidents, the LOSP is , followed - by failure of the DGs. Some of the PDSs in the SBO group have large A size pipe breaks. When the earthquake caused both SB0 and LOCA, the PDS was placed in the SB0 group. Large LOCA PDSs appear in both the SB0 and the LOCA PDS groups. These A-size pipe breaks are due to failure , of the ' steam ' generator or reactor coolant pump supports. In addition to J causing a break in.the primary recirculation lines, these support failures are judged to place enough strain on the main steam line penetrations to fail the containment pressure boundary. Thus, these "A" PDS have initial containment failure.-
-S.5 ' Accident Procression Analysic S.S.1~-Description of the Accident Procression Analysis ;
g The accident = progression analysis is performed by means of a large and ' detailed event tree called the APET. This event tree forms a high-level model of the accident progression, including the response of the i containment to the loads placed upon it. The APET is not meant to be a substitute for detailed, mechanistic computer simulation codes; rather,-it ( is a framework for integrating the results of these codes together with
- experimental results and expert judgment. The detailed, mechanistic codes require too much computer time to be run for all the possible accident
-progression paths. .Furthermore, no single available code treats all the important phenomena in a complete and thorough manner that is acceptable to all those : knowledgeable in the field. Therefore, the results from these codes, as , interpreted by experts, are summarized in an event tree. The resulting APET can be evaluated quickly by computer so that the full- j '
diversity of _ possible accident progressions can be considered and the uncertainty in the many phenomena involved. ;
' The APET treats the . progression of the accident from the onset of core damage _ to the core-concrete interaction (CCI) . The APET accounts for all- .
the events 1 that may lead . to the release of fission products due to the
-accident, even though -some of the events may not occur until several days after the accident._ The Surry APET consists of 71 questions, most of which ' have more than two branches. There are seven time periods considered in $
the tree. The recovery of offsite power- is considered both before vessel j failure as well as after vessel failure. The possibility of arresting the ; core degradation process before failure of the vessel is explicitly ! considered. ' Core damage arrest may occur following the recovery of offsite power or when depressurization of the RCS ' allows injection by an operating system (llPIS or LPIS) ' that - previously could not function. Containment . failure is considered at vessel breach (due to vessel blowdown, hydrogen l combustion,; direct containment heating, and steam explosions), af ter vessel i failure (due J to hydrogen combustion), and af ter several days (due to E .basemat meltthrough or eventual overpressure if containment cooling is not 1 restored). Five mechanisms , four of them inadvertent, for depressurizing the vessel before failure are included in the APET. The APET is so large and complex that it cannot be presented graphically _ l- and must.be evaluated by computer. A computer code, EVNTRE, has been l S.9 i
written for this purpose. In addition to evaluating the APET, EVNTRE sorts the myriad possible paths through the tree into a manageable number of outcomes called the APBs. S.S.2 Results of the Accident- Procression Analvsis Results of the accident progression analysis for internal initiators at Surry are summarized in Figures S.2, S . 3, and S .4. Figure S.2 shows the mean distribution among the summary accident progression- bins for the summary PDS groups. Technically, this figure displays the mean probability , of a summary hPB conditional on the occurrence of a PDS group. Since only mean values are shown, Figure S.2 gives no indication of the range of values - encountered. Figure S.3 shows the distributions of the expected conditional probability for core damage arrest given a PDS group. Simi-larly, the distributions of the expected conditional probability for early containment. failure (CF) given a PDS group are displayed in Figure S.4. Early CF means CF at or before vessel breach (VB). *
SUMMARY
SUMMARY
PDS GROUP (ween Con Darnese nequmy) ACCIDENT
. . . . . . . .. . . . . .. . . . . . - la t e rn et t a tu a te ro - . . --- . - . . ..- . - ..
PROGRESSION LOSP ATWS Transients LoCAs Dypass All Fire DIN GROUP ( 2.8E-05) ( t.4E-06) ( tee-06) ( 6 IE.06) ( 3 4E-06) ( 41E-05) ( 1.lE-05) VB. alpha. 0.003 0.003 0.005 0.003 0.005 early CF VD > 200 pol. 0.005 0.001 0 001 0.004 0.013 early CF VB. < 200 psl. early CF VD. DwT or late CL 0.079 0.046 0.013 0.055 0.069 0292
~
( Dypass 0.003 0.078 0.007 1.000 0.122 VB. No Cr 0,310 0.$2 1 0.217 0.660 0.346 0.690 l l l No VD 0.599 0.350 0.762 0 352 0.4 6f Figure S.2. Mean Probability of APBs for PDSs--Internal and Fire S.10
l I 100 W m mean 85t h, m = median m I th = percentile i i n ! ' I 0.75-esth. .
- g. kl J
l
- esth.)
by esth. ge _ . sis. . L
~
m W.-e
}L _ !
a8k 1 - .b - I e
,,3 is "
l l
.O - 8 j
y - fO m-8-. d: 87 _l 0.25- ,, l 3, } ot u _1 0 00 PDS Group 1.0S P . ATWS Transients LOCA Dypass All Core Damage Freq 28E-05 14 E- 00 1.IIE- 00 61E-06 3 4E-06 4.lE-05 Figure S.3 Probability of Core Damage Arrest -Internal 1.In l SURRY 1.E-1. 'Stu 1 ..iu . .. i 6. . E new h u yj1E-2. u.
"'b 3p 06th
$& u u.
6
. j .h 1.E-3.
OY s6
$ *+ l **
l I I c t.E-4. ,
, i m.
64 O J sit. J o set.
** stk.
Stk 1E-5. set. I W = ms 7 m = metsien l th a percentile I . E- 6
---- - -- - -- - --- --- - inte r n al in ttia to rs -- - - - -- ----- -- - - - - -
PDS Group 500 ATWS Tra nsients toCAs Bypass All Fire . Core Damage heq 2 8E-OS 14E-06 ISE+06 61E-06 3 4E-06 41E-05 11E-05 ( Figure S.4. Probability of Early Containment Failure--Internal l S ll l l
d
, m e
i Figure S.2-indicates the mean probability of the possible outcomes of the accident ' progression analysis. The- width of each box in the figure indicates how likely each accident progression outcome is for each - type of accident. -Except for the Bypass initiators, either no failure of the ; vessel (safe stable state) or no failure of the containment are by far the most likely outcomes for internal initiators and fire initiators. If core damage is not arrested and the accident proceeds to failure of the vessel, Figure S.2 shows that no failure of the containment is the most likely outcome for all types of accidents. If CF does occur, late failure is nore -likely than failure at or before VB. Late failure may be due .to hydrogen ignition some hours af ter VB, basemat meltthrough, or eventual overpressure after several days if CHR is not restored. Of these three : late failure modes, basemat meltthrough is the most likely for internal F initiators. Results of the accident progression analysis for fire initiators at Surry are summarized in Figures S.2 and S.4. Figure S.2 shows that early CF is very unlikely for core damage accidents started by fires, but that' the probability of late CF is about 0.30. This-is because the. initiating fires destroy the~ ability to operate or restore CHR (sprays) within a few days of the, accident. Arresting the core damage process before vessel failure is not possible .for the fire initiators because the fires render all the
-injection systems inoperable. Figure S.4 shows that early- CF is quite-unlikely for fire initiators. All four fire PDSs have a break that
- partially depressurizes the RCS before VB.
i Figure S.5 summarizes the results of.the accident progression analysis for ! seismic initiators. As the core damage frequency of a PDS group has no
'effect on the evaluation of the APET, the accident progression analysis results ' for the two hazard distributions are very similar. The differences are due-to= differences in the frequencies of ' the individual PDSs - relative to other PDSs in'tho' group. The majority of seismic. core damage accidents .t
-result in either= no vessel failure or no containment failure. There.is no
- possibility of- avoiding vessel failure - for the SB0 accidents. The mean E
probability of .early containment failure -is -on the order of 0.01 - if the L initial failures'due-to SG or RCP support failures' arc excluded. Initial CF p occurs only~for.large breaks, so essentially all the low pressure early CFs g are attributable. to SG~or RCP failures. The probability- of late CF is [ relatively high for the ~ SB0 accidents because there is no -long-term ( L recovery of CHR~as there is in the seismic LOSP and internal SB0 accidents, l The : probability.. of late CF in the' scismic accidents is higher than it : is for internal initiators because many of . the PDSs (including the most I frequent ones) in ' the LOSP and IECA groups are ones in 'which the sprays have failed.
' Core Damace Arrest. It is possible to arrest the core damage process, avoid VB, and: achieve a safe, stable state (as at Three Mile Island-(TMI-2)) if coolant injection is restored before the core degradation process
'has gone too far. Recovery of injection is due to one of two events. In the LOSP accidents, recovery of injection follows the restoration of offsite . power. In other types of accidents, an inj ection system is ;
operating ' when core degradation starts, but no inj ec tion is taking place S.12 !
- _ = _ - __ _
PDS GROUP ACCIDENT ("'* " C " D* *
- 8' h"I"'acr)
...............LLut............... ...............tPR1--+---~~-----.--
, . PROGRESSION losp sD0 LoCAs Total 1.Ose suo LoCAs Total l BIN GROUP ( 9 0E-Os) ( 7.9c-Os) ( 2 3E-Os) ( l 9t-04) ( tsc-Os) ( 9 st-06) ( s os-06) ( 2 et-Os) VD. alpha. 0 005 0.006 0 006 0 006 0 006 0.006 0.007 0.006 early CF VD > 200 psi. 0 006 0 012 0 000 0 006 0 014 0 000 early CF VB < 200 pst. 0.072 0.350 0 082 0 050 0 322 0.058 early CF VD. DWT or late CL 0.214 0 377 0.124 0.200 0.232 0.002 0.137 0.293 Dypass 0001 0 002 0 001 0 001 0.001 0 001 VD. No CF 0.391 0.S31 0.216 0.435 0.400 0.54e 0.227 0 457 i No VD 0.383 0.305 0 189 0,346 0.307 0.176 Key. DMT . nasemat Well-Through SURRY Cf' = Containment Ihtlure
'li 'e s e en h Figure S.S. Mean Probabilities of APBs for PDSs- Seismic l
because- the RCS pressure is too high. If a break in the RCS pressure l boundary allows the RCS pressure - to decrease to the point where the
-operating system can inject, then there is some chance of arresting the-core degradation. process. The probability of arresting core degradation depends on the time the injection starts relative to the state of the core.
The RCS failure that allows inj ection to commence may be an initiating break or a temperature-induced failure that occurs after the onset of core damage (such as a break in the hot leg or surge line, the failure of an RCP seal failure, or the sticking-open'of a PORV.) For the internally initiated PDS groups, core damage arrest ir possible for all groups except Event V. Offsite power may be recovered for the two SB0 groups. Some PDSs in the Transients, LOCAs, ATWS, and SGTR groups have LPIS, or LPIS and HPIS operating. The initiating break in the interfacing LOCA fails the LPIS by diverting the flow out the break. Figure S.3 contains no plot for the bypass accidents since the fission products may escape to the environment whether or not the vessel and containment fail, so vessel failure is not of particular interest. Figure S.3 indicates that core damage arrest be fore VB is especicily likely for the Transients PDS S.13 l
i l t L i group. One rf 'he PDSs in this group has the LPIS operating, and the other
- 1. has both L1 IE and llP7S operating, at the onset of core damage. The l probability of core damage arrest for this group reflects the probability
!. that one of the five means of depressurizing the ECS reduces the RCS to a sufficiently low pressure to allow injection. l_ Core damage arrest (no VB) is not possible for the fire initiators because each fire initiator destroys the ability to supply either motive or control power to the ECCS. Figure S.6 displays distribution of the probability of arresting the core-melt process and avoiding failure of the reactor vessel for the seismic initiators. For seismic initiators, recovery of offsite power in the SB0 i accidents is not considered feasible dun to damage to the switchyard; no l histogram is shown for the SB0 PDS group as core damage arrest is not possible. Inclusion of the SBO accidents in the total accounts for the fact that the total distribution shows core damage arrest to be 1er likely than it is for either the LOSP or LOCA groups. The differences between the l' distributions for the hazard distributions are due to differences in the j' frequencies of the PDSs in which an injection system is operable relative to other PDSs in the group. The probability of core damage arrest for the seismic initiators is lower than it is for internal initiators because the 100-W = mean m = median th = percentile 95th. l 95th. l 0.75-l 95th ,. DY 95th. .
=!c <t b ]
E
- a. C - ]
- 6 0.60- soth.
$$ esth. .
.a e dL v. -
% W_.
OO ** L , O l 0.25- i J : : y
) ]
O 00
- - - - - - - -- L L N L - - - - - - - - - - - - - - -- E P R l -- - - - - - - -
PDS Group LOSP LOCAs Total LOSP IACAs Total Core Damage ITeq 9 00-05 2 3E-05 1.9 E- 04 1SE-05 3 SE-06 2.0E-05 Figure S.6. Probability of Core Damage Arrest--Seismic S.14
i frequencies of the PDSs with one or more injection systems operable are lower relative to the total frequency than they are for internal initiators. RCS Depressurization. The reduction of the RCS pressure in the period between the onset of core damage and VB is important for two reasons. First, pressure reduction may allow the LPIS to function and thus avoid vessel failure in accidents where the LPIS is operable but not inj ecting due to high RCS pressure. Second, lower RCS pressures at VB reduce the loads placed on the containment structure at that time and reduce the probability of containment failure at VB. Four of the five means of depressurizing the RCS considered in the Surry accident progression analysis are temperature induced (T-I) and inadvertent. The five mechanisms are:
- 1. T-I hot leg or surge line failure;
- 2. PORVs or SRVs stuck open;
- 3. T-I RCP seal failure;
- 4. T-I SCTR; and
- 5. Deliberate opening of the PORVs by the operators.
T-I failures of the RCP seals and PORVs sticking open are also considered in the accident frequency analysis. Of these five mechanisms, only the first three are effective for most accidents. Expert panels provided distributions for the probability of hot leg failure, SGTR, and RCP seal failure, n.he effective means of RCS depressurization ensured that very few accidents proceeded from the onset of core damage to lower head failure at the PORV setpoint pressure (about 2500 psi). Early Containment Failure. For those accidents in which the containment is not bypassed, the offsite-risk depends strongly on the probability that the containment will fail early, i.e., before or at VB. There are four possibilities for early CF:
- 1. Pre-existing containment Icak;
- 2. Isolation failure;
- 3. CF before VB due to hydrogen combustion; and
- 4. CF at VB due to the events at VB.
j As the Surry containment is maintained about 5 psia below ambient atmospheric pressure during operation, an unsealed hatch or an open vent line would be quickly discovered since the vacuum pumps could not keep the l containment at the desired pressure. Thus, the probability of a pre-existing leak at Surry is negligible. Isolation failures at Surry are also negligible. Because the structural experts considering this issue found the Surry containment to be quite strong, CF due to hydrogen burns before l VB is not considered at Surry. It was estimated to be unlikely that L sufficient hydrogen would be generated in the vessel and escape to the containment before VB to cause a hydrogen deflagration large enough to l threaten the Surry containment. This failure mode was included in the APET l used in the previous analysis for the first draft of NUREG 1150, and CFs I before VB were negligible. S.15 1
Except for the initial CF in some seismic PDSs, the only significant cause -! of early CF at _ Surry is the pressure rise due to the events that occur at VB. Figure S.4 indicates that early CF is fairly unlikely for the internal and fire initiators. The probability distributions for early CF in this figure are conditional on VB, not on core damage. There is no histogram for Bypass accidents. When the containment function is bypassed by Event V or SGTR, early CF ceases to be important in determining the release of fission products and the offsite risk. Thus, the conditional probability of early CF is not plotted for the Bypass group. For internal and fire accidents other than Bypass, Figure S.4 shows that the mean probability of early CF is on the order of 0.01, and the median is about two orders of magnitude lower. This is largely due to the robust nature of the Surry containment relative to the loads expected at VB and the effectiveness of the RCS depressurization mechanisms. The pressure loads on the containment due to the failure of the reactor vessel and the
-escape of the molten core into the reactor cavity are strongly dependent on the pressure in the RCS at the time the lower head of the vessel fails.
Therefore, depressurization of the RCS before VB plays an important part in determining the probability of containment failure at VB. Even without RCS depressurization, the distribution for the containment failure pressure at Surry provided by the Structural Response Expert Panel falls generally above the distributions for the loads expected to be observed at VB from the expert panel on containment loads. i The conclusion that CF at VB is unlikely for Surry also holds for the fire initiators and the seismic initiators. The initial CF due to seismic failures of the SC or RCP pump supports constitutes a separate failure mechanism. Figure S.7 shows that the mean probability of early CF for all f PDS groups for the LIRL and EPRI seismic hazard distributions are on the l order of 0.10. Most of. these early failures of the containment- are initial failures due to SC and RCP support failures. These initial CFs account for L the fert that early CF is more likely for the IDCA group than for the SB0 group, and more likely for the SB0 group ' than for the LOSP group (which' contains no "A" PDSs). ' S.6 Source Term Analysis S.6.1 Description of the Source Term Analysis The source term for a given bin consists of the release fractions for the nine radionuclide classes for the early release and for the late release, and additional information about the timing of the releases, the energy associated with the releases, and the height of the releases. It includes the information required for the calculation of consequences in the 4 succeeding analysis. A source term is calculated for each APB for each observation in the sample. The nine radionuclide classes are: inert gases, iodine, cesium, tellurium, strontium, ruthenium, lanthanum, cerium, and barium. S.16
- . _ =
J_ lIIu i 1.El SURRY 1.E0- estk. esth. e6th. I I 1 esth, esth. , I esth. ' r y - m. Nh1.E-1, v. w -
= W. ,+
y _I eSth. -
- m. - -
o e
, I
- 95th.}- m.
2& m. _ A .5 1 .E-2. W. - g 3 _ .tu. i _ ec ' 6 t h. 88 1 q _ 6c @1.0-3. sih , Ob _ 3g g, o 6th. .
.l -
6t h
- m. g
, l.E-4, j
] W = mean oth, m a median oth. .l th = percentile
_ 1. E- 5.
----------LLNL---------.--- ------------EPRI---------~~
PDS Group LOSP Si10 LOCAs Total 1.0SP SitO LOCAs Total core Damage heq DOE-05 7 9E 2.3 E- 05 19E-04 15E-05 950-06 35E-06 28E-05 m Figure S.7. Mean Probability of Early Containment Failure--Seismic The source term analysis is performed by a relatively small computer code, SURSOR. The purpose of this code is ap_t; to calculate the behavior of the fission products from their chemical and physical properties and the flow and temperature conditions in the reactor and the containment. Instead, SURSOR provides a means of incorporating into the analysis the results of the more detailed codes that do consider these quantities. This approach is needed because the detailed codes require too many computer resources to be able to compute source terms for the numerous APBs and the 200 observa-tions that result from the sampling approach used in NUREG-1150. SURSOR is , a fast-running, parametric computer code used to calculate the source-terms for each AFB for each observation for Surry. Since there are normally about a hundred bins for each observation, and 200 observations in the' sample, the need for a-source term calculation method that requires'few computer resources for one evaluation is obvious. SURSOR provides a framework for . synthesizing the results of experiments and mechanistic codes, as interpreted by experts in the field. The reason for " filtering" the detailed code results through the experts is that no code available S.17
treats all the phenomena in a manner generally acceptable to those knowledgeable in the field. Thus, the experts are used to extend the code results in areas where the codes are deficient and to judge the applic-ability of the model predictions. They also factor in the latest experi-mental results and modify the code results in areas where the codes are known or suspected of oversimplifying. Since the majority of the param-eters used to compute the source term are dorived from distributions determined by an expert panel, the dependence of SURSOR on various detailed codes reflects the preferences of the experts on the panel. It is not possible to perform a separate consequence calculation for each -of the approximately 20,000 source terms computed for the Surry integrated risk analysis. Therefore, the interface between the source term analysis and the consequence analysis is formed by grouping the source terms into a much smaller number of source term groups. These groups are defined so that the source terms within them have similar properties, and a single consequence calculation is performed for the mean source term for each group. This grouping of the source terms is performed with the PARTITION l program, and the process is referred to as " partitioning." The partitioning process involves the following steps: definition of an early health effect weight (Ell) for each source term, definition of a chronic health effect weight (Cil) for each source term, subdivision (partitioning) of the source terms on the basis of Eli and Cil, a further subdivision on the basis of the time the evacuation starts relat?.ve to the start of the release, and calculation of frequency-weighted mean source terms. The result of the part'tioning process is that the source term for each accident progression bin is assigned to a source term group. In the risk computations, each accident progression bin is represented by the mean source term for the group to which it is assigned, and the consequences are calculated for that mean source term. S.6.2 Results of the Source Term Annivsis When all the internally initiated accidents at Surry are considered together, the plots shown in Figure S.8 are obtained. These plots show four statistical measures of the 200 curves (one for each observation in the sample) that give the frequencies at which release fractions are exceeded. Figure S.8 summarizes the complementary cumulative distribution functions (CCDFs) for four representative radionuclide groups (iodino, cesium, strontium, ' and lanthanum). The mean frequency of exceeding a release fraction of 0.10 for idoine and cesium is on the order of 10-6/yr. The mean exceedance frequency for release of 0.10 of the core strontium is somewhat lower. The mean frequency of exceeding a release fraction of 0.01 for lanthanum is on the order of 10-6/yr. The highest fractional releases are computed for bypass accidents (Event V and SGTRs) and early containment failures. The releases for late containment failures, most of which are basemat meltthroughs, are quite small. Regression-based sensitivity studies for the internally initiated accidents indicate that the largest contributors to the uncertainty in release S.18
C LE-3 ftE-3 o o ,, , , , ,, , - E : 1 Percentile g 0 u - t . . r.5_t.h. . . o tE-4 , stE-4 _ Meon u o o o - 50 h t e - e u u t.E-5 s. i 1.E-5 ,-------___'--_--------.5.th - - - - p r-- g Q
, i w : , ', _ w N s '.
v- . ., .- tr ',
- 1.E-6
. *tE-6 ;- -_. ;
Iu ~ Percentile , 95th - o .
'i c - - - - - - - - - c .
OLE-7 ' Weon '. o 1.E-7 7 i i* 50th : i* ., 5th
- 1 u u ~ -- ------ '
"- "- "- '- " ' w L, w tE-8 '- ' " " - " " " - " - ' ' - ' -
is tE- B " i LE-8 LE-6 LE-4 LE-2 LEO 1.E-8 LE-6 LE-4 LE-2 1EO Release Froction for i Release Froction for Cs
.M i w \
'8 ftE-3 o , , , . . . , ,,, , , . . , m. a ftE-3 o
E - Percentile: Percentile 8 9 u 8
. - .9.5.t.h. .
e on-, _ - o tE-4 , _ W_5_t_h SLE-4 7 Meon -
- 50C
~
l 8 i 50th g a u -. _ _ , __5th- - - - - - - - 5th utE-5 ------ ate-5 ,'---- _____ , ---- e a 7 e a _, Y . '. - M - n N.. tr - . - 0-fl.E-6 7 N., , ytE-6 7 '.,
, -,,,,_ -., '.= . .
o ~. * -
- u . .
c -
.
- c - .
OLE-7 . ., OLE-7 '. ; , m 7 . . _ o 7 e . . e . e u
'.. . e u
x
' x __.'
w tE-8 ' " = " " * " . '" - - w tE-8 " L' - " I ' "- ' LE-8 LE-6 LE-4 1.E-2 LEO LE-8 LE-6 LE-4 LE-2 LEO Release Froction For Sr Release Froction For lo s Figure S.8. .Exceedance Frequencies for Release Fractions for Surry: All Internal Initiators _ -. _ _ . - ~ . - - _ . _ . . _ _ . .
fractions and their frequencies are accident frequency and source term variables. Few accident progression variables are important in determhiing the uncertainty. The important initiating event variables are the fre-quencies for Event V and SGTR, Failures of the DC to start and *.un are also important. For the less volatile radionuclide classes (tollurium, strontium, ruthenium, barium, lanthanum, cerium, and barium), the variable that contributed-the most to uncertainty is the fractional release from the core to the vessel. Also important are the fractional release from the vessel to the environment for SGTRs, and the release fraction for CCIs. The frequency of any given size release due to fire initiators at Surry is so low relative to the frequency of 'a similar release due to internal initiators that no fire source term results are presented. The release frequencies for seismic initiators are much larger than those for fires. Figures S.9 and S.10 present statistical measures of the families of curves that.give the frequencies with which release fractions are exceeded for the LLNL and EPRI hazard distributions. They may be compared to Figure S.8 for. > internal initiators. It may be seen that the seismic releases based on the EPRI hazard distribution are roughly comparable to those due to the internal initiators, while the seismic releases based on the LLNL hazard distribution are greater than those due to the internal initiators. S.7 Consecuence Analysis S.7.1 Descrintion of the Consecuence Analysis MACCS is used to calculate offsite consequences for each of the source term groups defined in the partitioning process. MACCS tracks the dispersion of the radioactive material in the atmosphere from the plant and computes its deposition on the ground. MACCS then calculates the effects of this ra. dioactivity on the population and the environment. Doses and the ensuing health effect:: from 60 radionuclides are computed for the following path-ways: immersion or cloudshine, inhalation from the plume, groundshine, deposition 'on - the skin, inhalation of resuspended ground contamination, ingestion of contaminated water, and ingestion of contaminated food. MACCS treats atmospheric dispersion by- using multiple, straight-line Gaussian plumes. Each plume can have a different direction, duration, and initial radionuclide concentration. Cross-wind dispersion is treated by a multi-step function. Dry and wet deposition are treated as independent L processes. The weather variability is treated by _ means of a stratified sampling process. l For early exposure, the , following pathways are considered: immersion or , cloudshine, inhalation from the plume, groundshine, deposition on the skin, l- and inhalation of - resuspended ground contamination. For the long-term l exposure, MACCS considers the following four pathways: groundshine, l inhalation of resuspended ground contamination, ingestion of contaminated L water, and ingestion of contaminated food. The direct exposure pathways, l groundshine, and inhalation of resuspended ground contamination, produce doses in the population living in the area surrounding the plant. The indirect exposure pathways, ingestion of contaminated water and food, produce doses in those who ingest food or water emanating from the area around the accident site. The contamination of water bodies is estimated S.20
, :! l m
0 0 - E E . 0 0 _ , - 1 : ,: 7 - 1 7 - - ,: - , : - - ,: : 1 le - _ i - . . t - n h;, th - c5g0 et 55-h- t- a. w a.
..**.i.
r9
- 2 i 2 .
e - '
- s - o s P - c .
EC . EL n
- 0 0 o -
, - " 1r o
4 a. t r o i t s F F c n n .... . n o a o
- io , i r it d
a t c - t c F u _ ' 4 a 4 a ebi
' - r E F .-
- r E F s
.' 0 e 0 e a t r
s .' e t
- a. 1 s
". s -
s o - a l i
- .-- .' e e R e D m\ !e le
_ - - d R - R r d
=,
- a. o r f az 6 - .
6 s Ha E - . E
- 0 _= - ,: _ _ - ,: 0 p: - r i - _r 7 7- .
e
. ,. ~
1 1 i e L 2 3 4 5 6 7 8 3 4 E 5 6 E 7 E E
- cN nL E E- E E E E E E E 0
t 0 t 0 t 0 t 0 1 0 t 0-t 0 t 0 t 0 1 0 1 0 t 0 t eL u
" q e r
^h, > g's8 & ,gMe E euc0v.euxw~ To >lLo uOeu 'eawCy euCOveeua r fo F
-_ e s c r n to D 0 a E O E 0 da e i q:- 15 3,. f - 3- 1- L ,
- ~ , - ,:. : , - ,:- a 1 e it
.,.' c n
.='.,'.
6' .....'. - x E I
, , c
,' a-
.- a .im
- ,. 2 2 .
- r s 9.
S ei
,. E 0 !
,' E 0
S
- a. 1 ro
- s 1r s. -
.- f o .
F e S
. r n - . n u o
- o 3
- i g
- it ' t i
, - - a. c
- a. c F 4 a 4 o >
- Fr s
- r >
E F - E
. - 0. e - 0 e _
- N -
" 1 s
a
- a t s a
r
.e -
e N le - . le
- R ,
' R
- - a. - a.
- - - .. 6 . '
, . 6 g: - r
,._~_ 7 ~ .
r r-- r: E 0
," - 7- - - - - - ,
E 0 1 1 2 3 4 5 6 7 s 3 4 5 6 7 e
- - - - - - - - - - - - - ~
E E E E E E E E E E E E E 0 0 0 0 0 0 0 0 0 O 0 O o t 1 1 1 1 t t t 1 L 1 L t e5,>g'ag* I ayw E ,oc0Oeeuxw
- To >Iu0 oOeu beawC' [ eoCOmeeu6 m.NW
,, 1
~
=
T - A A gtcE-2 ,, , , ,, .., , _, . _, , 8t0E-2 ., , , , , , i Percentile . gtot-3 , ., gt0E-3 r _ _95t h_ _ _ ,
; E _
g ... MeoL o _ : o : 501h e t0E-4 , e t0E-4 , 5th g ,-------- ._...'-~~~~ ...._, [ g [---
~ ~ ,,, '
O t0E-5 . '. O t0E-5 ---'~-' '- - _' g 1 g * - e g : '. . 6 .'. .- e t0E-6 o r
*1 e t0E-6 o
r ',9 c p :- - .._, - , ', :' , _ c-o- _ o 10E-7 ' ---- - t t0E-7 ---- - e r 5, e r , E o u s e u x
- x ~
~.'
w LOE-8 u' = = " " *.-*~1 *n' = - w toe-8 * . ' ' = "sr<~' = "' ...u- - I 10E-6 * ' '.0E-4 1 1.0E-2 10E0 1.0E-6 1.OE-4 10E-2 t0E0 Release Froction For I tt) Release Fraction For Cs ' l 1 M ! E t0E-3 o , o,, . . , , , um c, . .u, , .om O 1.0E-3 ii,, 1
.om o, . ,
e . , . e g _ g u - u o 10E-4 , o 1.0E-4 ,. - u - o O : _ O e u -- -- .- e s 10E-5 e -, ,,,'---- ,, , u e ut.0E-5 , G. * ~ O. ,,* Mg _- ..'s ,- ,
- M ,' _ , .
y 6 E t0E-6 ,
. Le t0E-6 ;-
. ~ .
.e E , e ',
u -
~
u -
- C' - -
C - - O 10E-7 r 01.0E-7 *
- n. n r .
e e c : e : u X
._-
- u x -
w LOE-8 - 'c"*"' = * ' ' . >n' * - w LOC-8 . ~ - - ' - . " ' ' ' * * "' - h' - -- 10E-6 1.0E-4 10E-2 10E0 1.0E-6 1.0E-4 1.0E-2 10E0 Release Fraction For Sr'- Refeose Froction For lo l l Figure S.10. Exceedance Frequencies for Release Fractions. Seismic Initiators for EPRI. Hazard Distributions i l
~
! ~ . - , , , - . . -. _ _ _ _ _ _ _ _ _ _ _ . . _ _ , , - . . , -
for the washoff of land deposited material as well as direct deposition. The food pathway _model includes direct deposition onto the crop species and l uptake from the soil. Both short-term and long term mitigative measures are modeled in.MACCS, Short-term actions include evacuation, sheltering, and emergency relocation out of the emergency planning zone. Long term actions incluoe relocation , and restrictions on land use and crops. Relocation and land decontamina-tion, interdiction, and condemnation are based on projected long-term doses from groundshine and the inhalation of resuspended radioactivity. The , disposal of agricultural products and the removal of farmland from crop production are based on ground contamination criteria.
'The health effects models link the dose received by an organ to morbidity ,
or mortality. The models used ir. MACCS calculate both short-term and long-term effects to a number of organs. Although the variables thought to be the largest contributors to the uncertainty in risk are sampled from distributions in the accident frequency, accident progression, and source term analyses, there is no analogous treatment of uncertainties in the consequence analysis. Variability in the weather is fully accounted for, but the uncertainty in other parameters, such- as- the = dry deposition velocity or the evacuation rate, is not considered. The _ MACCS consequence model_ calculates a large number of different consequence measures, This . report gives results for the following six consequence measures: early fatalities, total latent cancer fatalities, ; _ population dose within 50 miles, population dose for the entire region, early fatality risk within 1 mile, and latent cancer fatality risk within 10 miles. For NUREG-1150, . 99. 5% of the populatio.. evacuates and 0.5% of the population continues normal activity. For internal initiators at Surry, the evacuation delay time between warning and the beginning of evacuation ~is 2'h, i For seismic initiators, the evacuation parameters were altered since earthquakes are judged to- affect the evacuation. There is no evacuation at all for those - earthquakes - in which the maximum peak ground acceleration (PGA) exceeds 0.6 g. These earthquakes form a relatively small portion of the seismic distribution at. Surry. The evacuation is degraded for earth-quakes in which the maximum PGA does not exceed 0.6 g. The delay. period (from the warning to the start of evacuation) is-increased to 1.5 times its normal value,. and the evacuation speed is decreased to half its normal
-value, S . 7. 2- Results of the Consecuence Annivsis The results presented in this section depend on the occurrence of a source .
term group. That is, if a release takes place wii.h release fractions and other characteristics as defined by one of the source term groups, then the tables and figures in this section give the consequences expected. This section contains no indication at all about the frequency with which these consequences may be expected. Implicit in the results given in this S.23
section_is that 0.5% of the population does not evacuate and that there is a 2 h delay between the warning to evacuate and the actual start of the evacuation. CCDFs display the results of the consequence calculation in a compact and complete form. The CCDFs in Figure S.11 for early fatalities and latent cancer fatalities - display the relationship between consequence size and consequence frequency due to variability in the weather for each source term group that has a non zero frequency. Depending on the occurrence of a release, each of these CCDFs gives the probability that individual consequence values will be exceeded due to the uncertainty in the weather conditions that will exist at the time of an accident. Figure S.11 shows that there - is considerable variability in the consequences that is solely due to the weather. There is, of. course, considerable variability among the consequences that is due to the size and timing of the release as well. The risk from fire initiators at Surry is low relative to that from internal initiators, so no fire consequence results are displayed in this summary. Figures S.12 and S.13 present CCDFs for the LLNL and EPRI hazard distributions for the non zero source term groups. As these results are conditional on the occurrence of'the release, and contain-no information about the expected frequency of the release, no conclusions concerning risk can be drawn from Figures S.11, S.12, and S.13. S.8 Intecrated Risk Analysis S.8.1 Determination of Risk Risk is determined by bringing together the results of the four constituent analyses; the accident frequency analysis, the accident progression analysis, the source term analysis, and the consequence analysis. This
. process is described in - general terms in Section S.2 of this summary, and in ' mathematical terms in Section 1.4 of this volume. Specifically, the accident frequency analysis - produces a frequency for each PDS group for each observation, and the accident progression analysis results in a probability for each APB, conditional on the occurrence of- the- PDS group.
The absolute frequency for each bin for each observation is obtained by summing the product of the PDS group frequency for that observation and the conditional probability for the APB for that observation over all the PDS groups. For each APB for each observation, a source term is calculated; this source term is-then assigned to a-source term group in the partitioning process. Then the consequences are computed for each source term group. The overall result of the source term calculation, the partitioning, and the conse-
-quence calculation is that a set of consequence values is identified with each APB for each observation. Because the absolute frequency of each APB =is known from the accident frequency and accident progression results, both frequency and consequences are known for each APB. The risk analysis consists of assembling and analyzing all these separate estimates of offsite risk.
S.24 i
i 4 i , e
' ' 1 4 .
=
= --
w~~ . 9
- o
,- M._ -- m g; Y(f ' -
a e-w ,- - <r L4 Z, m g 3: I oW
$ a 1' cr - '
- S r. z ,
w ; (w-r' <-- i i i. [b4 7
- g y
e co o w
? 7 7 ? 7 e M o o o o o o X ONIO333X3 30 A11119V80dd b e
- ~m. ,
r- , . r- , . g 7 o .
"' M
> > /- / N, .
F-
,, in - y W k y
~
<t .
p ;/ - 2 - - k z
./ ,Y/ i
'f 2 h !
a a-
, ., ~
m J ^ y ./ ?- ~--- ;- > cr f' F. ' 7o 'dg 2 w
/ --
j ,, ,
'. x n ,
y y - g 't
~
~
f
' ~
i - i: .,),
~
i I O N Y w D w o e
'o 'o 'a b- b b- b-X DN10333x3 30 All119v80dd S.25
.n
' c.
-wp- ,
,. . ._.-5W M' 1 e
~
#~ __ :: - -
F ~' p tn
/ a O
7 c
- S3u 7 U O
J o "l
~
r 3 I Le 7 e o-
& w D <
m -
- . a g
>! .3
- ~.
o .
- m a
- i. . . m 3
'o~
y i 7 o 7 o 7 o 7 o du
- o. -- - - - a m
X DN10333XP 30 A11718V808d y
, t
, r- r- r- ,
g g
- u (0
.T.
e g F e
~O , y
,/ e- M
- .J .
g i - a
& +
$ N i / -
a l - g i / /
- m .
l jl
~ 'e h.! l. .[
i i n. . y o I o o o e o o e.s e.e o.
- v. - -e - -
X ON10333x3 30'A11716V806d S.26
.-- i i i , ?
o fA O I l r
. u,
; a o
4 I- r J . Z r 1 o LAJ J N E I .
? U -,
>- o. H E
- g w
a a l
&=
O M _
'O J m
W X E
'o .
i h O
, i. :-. i. i 7 p o s Y
o I o 7 o 7 o I 7 du a .-e o- e
.-e - - 4J u)
X ON10333X3 30 All118v808d g o u i i i i i g g d e ~
'o O I' _ N' E '
- m e-4
^
. j v I / / _- 0 -
u j ffb *
,/
o4 i r
~
J '
. v J [ .j g I
/
-w
. m'o W4 y
p . .
?
t> m i a f j ;
; b , ;
a l I . 1 [ . . , , e i T $ $ $-- $- e o o e o e o e e m ,-e w e ~ X ON10333x3 30 AlllI8v80Bd S.27
. s, . , ,.n,-- , . ,
o
.r u,- ~
~~
, -gj W O'
O ," , 7 -4 o-l e. 4 k e * - a e .f -- o w W - s , y U 4
>- e e - *-
e = a, W m - - . 7 - i c o <
- J g
, J >
'a-
.3 m
s.. m h i a i i a w
~ ? 7 7 7 $ '7 ? N o o o, o o o a '
u X ONIO333X3 40 All118V808d + a -
, C i.e- , , , 7 'O o W M G3 N
ft)
~
w X _# +
" Cd A t y
d n
/ O g n' v
/ w
- ~.
/.j'
. y : - l - M 7 r
/
/ -
. ~
o . W g gy ' D - X W r
/ -
i
~
/
~
~
/ 'o-l
-/. ... . .
( o S ? 7 $ $ $ o o o a
.-. ee o.
X ONIO333X3 30 A11718V808d S.28
, , y a
~
- e f /. ,- ._
I 4 ! S[
$ ,// /
3
*;' ,' r
# 1
- c, 2
-e o u
^
~ " fr
/
'o w a f - h' e
cr [ 3 %. w . or -
" 7 <
8
~
E
/ ,
,, x 6
o
~
m
. t
. . . , . - m, b
~
3
? T 7 7 i ? $
e o o o e e .o g X ONIO33')x3 30 A11110VPCad A
.,... , , , -.w r.
. m g" - 1 'ca sf -
l I f^'I' /M
/ i, e
/,. -
- - T emw Q .
/ , ,,)
, 'I -
7
,/; # f v
/ - .
oa a (c D - 1 !, (e 1 8 x W / ' [ l .l
. ?
.I , o.
. ) i ; .. i i %-
- 7 o
? 7 7 7 I i T o o o o o C'
- o. - - - - -
K ON10333x3 30 All 118v800d S.29
I S.8.2 Ermilts of the Risk Annivsis Measures of Risk. Figure S.14 shows the basic results of the integrated risk analysis for internal initiators at Surry. This figure shows four statistical imeasures of the families of the CCDFs for early fatalities, latent cancer fatalities, individual risk of early fatality within 1 mile of the site boundary, and individual risk of latent cancer fatality within 10 miles of the plant. The CCDFs display the relationship between the frequency of the consequence and the magnitude of the consequence. Since there are 200 observations in the sample for Surry, the actual risk results at the most basic level are 200 CCDFs for each consequence measure. Figure S.14 displays the 5th percentile, imedian, mean, and 95th percentile for these 200 curves, and shows the relationship betwem the magnitude of the consequence and the frequency at which the consequence is exceeded, as well as the variation in that relationship. The 5th and 95th percentile curves provide an indication of the spread between observations, which is often large. This spread is due to uncertainty in the sampled variables, and not to differences in the weather at the time of the accident. As the magnitude of the consequence recasure increases, the mean curve typically approaches or exceeds the 95th percentile curve. This results when the mean is dominated by a few observations, which of ten happens for large values of the consequences. Only a few observations have nonzero exceedance frequencies for these large consequences. Taken as a whole, the results in Figure S.14 indicate that large consequences are relatively unlikely to occur. Although the CCDFs convey the most information about the offsite risk, summary measures are also useful. Such a summary value, denoted expected risk, may be determined for each observation in the sample by summing the product of the frequencies and consequences for all the points used to construct the CCDP. This has the effect of averaging over the different weather states as well as over the different types of accidents that can occur. Since the complete analysis consisted of a sample of 200 observa-tions, there are 200 values of expected risk for each co.aequence measure. These 200 values may be ranked and plotted as histograms, .hich is done in Figure S.15. The same four statistical measures used above ere shown on these plots as well. Note that considerable information has oeen lost in going from the CCDFs in Figure S.14 to the histograms of expected values in Figure S.15; the relationship between the size of the consequence and its frequency has been sacrificed to obtain a single value for risk for each observation. The plots in Figure S.15 show the variation in the expected risk for internal initiators for four consequence measures. Where the inean is close to the 95th peteentile, a relatively small number of observations dominate the mean value. This is more likely to occur for the early fatality consequence measures than for the la*ent cancer fatality or population dose consequence measures due to the threshold effect for early fatalit!cs, i S.30
.. ~ _ -_ _ _ _ - - . _ . . . . - -
1 1.0E-3 , , , , , _ 7 , Percentile
. . S.th, , ,
ft.0E-4 y 5 'b b ~ jiDE-5 ..pjp..., r...........' t 31.0t-s - 1.00-7 r ',
- I
,~. ]
d: 1.0t.g \. r.,***......, 1 g 7 1.00-9 r ,'*.. i
. 1 L -
.,r-d 1.0E-in
> - - a - >
1.0L-4 '.0L-1 3 '.1 0 L - 21.0L-1 --a 1.000 1.0Lt 1.0L2 i.0L3 1.0C4 [orly fotolilles . 9o 0E-3 1 , ,., , ..., . . , , .., ., 3 , i> 1.0E-4 [5 I........ 1 3,1.00-5 ,,'~.. ................... '**.
- i. p....,,,,, . ,, ]
t
,. 0e. ,
r .. ..
? . . 1
'J
'., ', r 1.00-7 '
r ' I > 1.0E-8 r Percentile
'.'. i i
\
I, 1
.. h.. ,
l 1.00-9 r '~ 5 h
\ 1 I. y .. PJb.... i 1.0E- F 1.0i-3 IOL-2 ~'.0l-iIOLO IOLi 1 1.002 i.003 1.Di4 1.0t$
Latent Concer rotolities Figure S.14. Results of the Integrated Risk Analysis for Internal Initiators ; S.31 1
- w - --m-- w-- -+w--
i 10E-3 ' ' . e, Percentile , (LOC-4 ;
. 95 l
- -f,.th . .f ~l tot-5 , . . > t ti. . . . ,
g va t0t-t, i. w 1 b tot-7 r . 1 i, . tot-8 .. - I 7
.......... - .... i Lot-9 ,. ...,,,,', ,
1 u '..., O 10[-F . s " r
. L-6 t0L-7 .0t - t, .0L-b 0L '4 t0L-3 1.0L-2 ".M- l W 1
[orly fotoilly Risk With,ni 1 Mile 10[-3 , , , , , , , , ,, , , , , ,,, ,, , . , , , w I t00-4 h . I g tot-5 .................". g, _ u t0[-6
" " ' ' - - - - ~ ~ . . . . . , ,',', ,. ,
i-g , H LOC-7 ,. ', , N, , tot-e Percentile '. \, 3 l t*-> r ,g5.1h ..2- i i , l' y ..>.ttt,.. ,
, i l' w LOC-In >
a ' l LCL-8 LOL-? t0E-6 100-5 1 00-4 tol-3 1.0l-2 LOL-1 1.0E0 Latent Concer Fotolity Risk within 10 Miles Figure S.14. (continued) S.32
l t l l t i l l tot-3 , , ., 7100-4 s o -
. )
t00-5 ......., t, '
,3 tot-6 TOC-7 '. ', - '.
1
'. I
,, tot-8 r Percentile \. \. ,
y ', I t00-9 ,. ~gth ..50kL ' i U ..P!ti.... 1 O tot-1n ' ' ' . a .. a a .a - ' 1.0LO LOli t0L2 LOL3 t0L4 LOLL 1.0L6 1.0L7 0008 r Population Dose (person-rem) Within !>0 Miles . ' tot-3 , , ,
,Et00-4 5 +
,3, tor-5 r- ....., .... ................- i u ,..... .,,. .,...,- ,
j v t00-6 r
.e '
j.LOL-7 p. i
., i.,
l '. i jLOC-B r Percentile \, '
; 3 ,
/ .. g5Jh, .. ', ;
jtot-9 r r,o$L '- . , g ,.E!tt... ',- w tog in . . , , ,,a a , . . ,i, . . . , ,, t0L1 t0L2 1.0L 3 10l4 LOLS tol6 t0L7 tol8 tot 9 Population Dose (person-rem) Within Region Figure S.14. (continued) S.33 l - -
The safety goals are written in terms of individual fatality risks. The i plots in Figure S.15 for individual early fatality risk and individual l latent cancer fatality risk show that essentially the entire risk distribution for Surry falls below the safety goals. A single measure of risk for the entire sample may be obtained by taking the mean value of the distribution for expected risk. This measure of risk ! is commonly called mean risk, althour.h it is actually the average of the l expected risk, or the mean value of the mean risk. Mean risk values for 1 internal initiators for four consequence measures are given in Figure S.15. The risk from fire initiators at Surry is well below that from internal initiators. The mean early fatality risk due to fires is 3.8E 8/R yr, and l the mean latent cancer fatality risk due to fires is 2.7E 4/R yr. Both these values are more than an order of magnitude lower than the comparable , values for internal initiators. The risk from seismic initiators at Surry is comparable to, or higher than, , that from internal initiators. Figures S.16 and S.17 present the statisti-cal summaries of the CCDFs for the LLNL and EPRI hazard distributions. They may be compared to Figure S.14 for internal initiators. Figures S.18
- and S.19 present the histograms of mean risk for the LLNL and EPRI hazard distributions. They may be compared to Figure S.15 for internal initia-tors. Offsite seismic risk based on the EPRI hazard distribution is roughly comparable to the risk due to internal initiators; the seismic risk based on the LLNL hazard distribution is greater than the risk due to the '
internal initiators. Comoarison with Previous Studies. The offsite risk at Surry from internal initiators is lower than that computed in the Reactor Safety Study 3 (RSS) of 1975. For early fatalities. Figure S.15 shows that almost the entire distribution is now below the lower end of the RSS distribution. The median value for this study is about two orders of magnitude below - the median for the RSS, (The RSS did not report mean values, nor did it consider risk from external events.) For' latent cancer fatalities, the 5th percentile of the RSS distribution falls between the mean and median of the current distribution. The median value for this study is about one order of magnitude below the median for the RSS. These decreases in risk are greater than the decrease in the core damage frequency.- The RSS reported a point ' estimate value of 4.6E 5/R yr. For the integra",ed risk analysis, the median and mean of . the core damage , frequency . distribution are 2.5E 5/R yr and 4.1E 5/R yr, respectively. Because the changes in the consequence calculation are relatively small, ! much of the decrease in the risk at Surry compared with those in the RSS ! comes from changes in our understanding of how reactor accidents progress, and how much of the fission product inventory may be expected to be
- released in the course of an accident.
l S.34
_ f>MM- b - ] a
.}
fa E C a E g C L 1 ih
*n, I
. e o
l E *Er E Y t~ t r r W ( r r b-b- b-b - b
- $$.v b-b b - -
5 X8/ son 11e14 Xpe2 tenpM1pui Xu/sentivig aeauva luairl tenpM;pui y 6 3 S D s m x x n !q ,-r1TIb b r ! y F 4 , q z C d { E g y g a E g g c Il f . fE1 w
'u n o 55 3
a S
't t ? % h t t t t h j
'AgAusg. nns tod) satin og - osou uonvinded (AHAaen uosJ8d) uo19eg oJnu3 esou uoneinded **
b ' i t ,a 8. s . E-H Bl l 9
- x g I j rrrrrfDThh cerrI'Th c e
a m E ,
.s i a e
s ze , 1 C h Eg ju s y Es w Ag/sel)tte)q Apv3 Ag/sanitsig .toouva quairi S.35
t , , r T.0E-3 Percentile 7t0E-4 . 95.th ,,
~
-- ib t0E-5 r ...,,, ,, .. Afb....
t0E-6 r '., 1 y ., 5 1.0E-7 p t t0E-6 r
., 1-10E-9 7 .....,,,
v '
~~.. .
O LOE-1n ~a . 'a . ' a . 'r ~a . .a - a , t0E-4 t0L-3 t0L-2 LOL-1 LOLO 1.001 1.0L2 t003, ,.0E4 ' Early Fotolities , i e t0E-3 ,, 1 l LOE-4 r \ ...........,-*-...n 1
..., SURR ft0E-5 r N .,
j . y t0E-6 r \. 1-w ' t0E-7
- ~ ~ - .. ,," " " "- ., Seismic
,,, '\,
\., ~
f LDE -8 r Percenille N,... l
,,' q g I .. 5,111 . , ..,
l t0E-9 - 5 th
..Rth,,,,
' '1 o .,
O LOE-In ' ' i> a >' >1a >' a .- ' 'a ' ' ' ' <- t00-3 t0L-2 t00-1 LOLO t001 t'oL2 t003 t004 t0E5 - Latent Concer Fotollfles l l 1 Figure S.16. Results of Integrated Risk Analysis for for Seismic Initiators. LLNL llazard Distributions S.36
1 I l l 1 l l l 1.0E-3 vn- '
>I 1.0t-4 r
, 1 1.0E-5 ". -
7 ., b '. S1.0[-6 g 7.............. t t.0E-7 r
., 1 ,1.00-8 r Percentile \.,' 'q -
g 1.0t-9
. . gth . .
gh ',N', e 7 , , o ..EJb_... \, d 1.0t-in - a ., a .a 1 <a >....o - . . 1, . - 1.000.0L1 1 1.0L 2 1.003 1.0L4 1.0L5 1.006 1.007 .0E8 Population Dose (person-rem) Within 50 Mdes 1 p . 00- 3 . . ~ . . , , , , ,,, , , , , , j p
, , ,,, s., , , , ,
1.0C-4 ~N ' .....,,,,
- [5 7
i 1
.,'+ sugg
! jg1.0E-5 7 X , S 1.0C-6 , ',
' " " " '., l Seismle l: 1.0E-7 , N-7 3 1.0E-8 \
r Pementile ,,,,,, 1~ g . 1.0t-e . - go$ gth.. ',
..?!b.... ', -
~
o Ut.0C-1n' < "a a ..."s - .-s - * . - '
- - - {~ 3 1.0Lt 1.0L2 1.003 1.004 1.0L5 1.0l6 t.007 1.008 1.0E9 ,
Populoilon Dose (person-rem) Within Region Figure S.16 (continued) 1 S.37 i I
tot -3 -.. 3 eenenw. "1 1 7 tot-4 7 I j ,...,
,............................................. :. m .... 3 !
t.0t-6 r 1 e '\. t0E-7 r ., i LOC .* r tog.g f .. .............................,,,,,,,,...... , 1.00-in
" "^,, ., , . .a .. -
"joh "kM-6 IOE 5 t0E-4 LOE-5 t0E-2 LOE-t 1.0E0 Eorly Fotolity Risk Within 1 Wile 1
9 00- 3 ,, ,, l .3 g .. tot-4 7
......,,' ......,, )-
, t0E-6 ....,,,'. -
SURR , I
- i. ',
]
1.0E-6 r \, h t00-7 .........*****.... Selsmic t r ....., 1
,LOE-B r Percenille '. . , g J
tot-9 r ',. p ..RD.... s, t Q LOE-In ' < ' ' ~' - m' ' - - LOE-8 t0E-7 'ot-6 t0E-5 t0E-4 t0E-3 t0E-2 1.0E-1 t0[0 t Lolent Concer Fotolity Risk Within 10 Miles Figure S.16 (continued) S 38
p .0E-3 1 j Portenlile
?1.0t-4 . . $5.'h . . ,
E ~5$ h 1.0[- 5 . . k! h . . . y r
.y ,a-s ,
1.0E-7 . I N .' - ,
' t.0[-8 7 i
3 1.00-9 F-............. , 1
,,,," .... ,'18-o ',
"a ' ' ' " ' '-' " " " '
O 1.0C-td ' "d *' 1.0L-4 1.0L-3 1.0L-2 1.0L-1 1.000 1.001' ' 1"'.0L2 1.0L5 ".0[4 1 Early Fotollfles 1.0C-3 ,,, , ,., .,, ., ...,, .,, , , ,
^1.00-4 g
r""".............., ........... ! l.1.0E-5 i
...., .. ., 1 vY1 .0E-6 ,
' \., , )
, SURR 1.0[-7 .
+
y
~
1m-s r Percentiie
'. i 3,3,mi,
, p[fh 1.0[- 9 g ",, -
, i, y u . . Plh . .. N.
"a o ' '
O 1.00-in > i 1 =' EPRI 1.0L-5 1.00-2 1.0L-1 1.Olo 1.0 01 1.0L2 1.0LJ> I "'.0041.005 1 Lotent Concer Fotollfles Figure S.17. Results of Integrated Risk Analysis for Seismic Init iators. EPRI llazard Distributions l l l S.39
i i i i 10E-3 , , , , _,, , Et0E-4 i. q) r" " " " ~m........ ...'.......... ,'.. . 1
; tot-5 r 1
i g N , , vca.0E-6 1 r ', 1 i '.,
)pt0E-7 r
- -..... ,,- \, q i
, ', l t0E-8 r Percenille - -
. . g5.th . .
',- \, ,! !
10E-9 *Mfh -~~ ' 7 ,
'k u ' . . P.t tt . . . ',
L5 LOE-In > "a - - "a - i - "a *' a - "d - ' "a - - " - ' LOLO t.001 1002 t0L3 1.0L4 1.005 LOL6 10E7 1.008 Population Dose (person-rem) Within 50 Miles , y y r y yr.g r v ,5 y g is yg y y1 105 a 1e ng 1 vvg vug rt0E., g r........' - " ......, 1 q
'f1 ;
30E-5 vn t0E-6 ,
, y y ', SURR
~ 10E-7 , .........,, ',
,h
's, . [
t0E-8 r Percenille ', i
, 1 Seismle
.. 's , i t0E-9 ,
gth..gff- ', u ..>.ftt... ,' .'. 65 t0E-in i > .a .""a - "; .<-a * 'a .' ma EPRI t0L1 1.0L2 t0L3 Lold LOLS LOLb 1.0L 7 t0LB t0E9 Population Dose (person-rem) Within Region Figure S.17 (continued) , S.40
m 9 t0E-3 ,, , , , , , .., g Percentile Etot-4 . . E5' 2
- I.
_M hr),,,_. . 1 i. l tot-5 , ..>.th. ., g1.00-6 i 7 ................................. " -- ...., ,, 1 tot-7 7
. . . . ",1 i, . . t00-8 ,
1
/
j t0E-9 , u d tot-FL{,8j0L,7 .....................................,,,,-.,.,ma "j,g[,[g,g[-b L0b-4 t0L-3 t.0L-2 LOL-t toto Early Falolity Risk Within 1 Mile 1.00-3 .,, ., ,,, ,, ,, , , , , , , , ,, tot-4 ,.....,,,, yl. ; 1.00-5 .........,'"'., , u tot-6 + x .,,, . SURR 8, tot 7 .."-......., ', s , tT 1
$ tot-a . Percentine ' ,,
- 1 setsmic g . gth.. '.,
7 tot-9 .. gfn~- ', r 8 ..>.tb.... '.- - d tot-6 a . - .o > ' - ' - ' - EPRI tol-8 1.00-7 t00-6 tot-b t0L-4 tol-3 t0L-2 t00-1 LOLO Latent Concer Fotoilty Risk Within 10 Miles Figure S. 17 (continued) S.41
h 16' R18 16', E YL g.. , gr g 95Lh , 16 1 h M Iety OOBI O ii 2 16' , W-- 6 gg . u &g
- g. }3g*q '
E E 5 8 2 Id - 8 16'1
$ IO. '
h h 8 k ' i
- y2C, su 5 26 f ie - ~ 2 c; c
16"
<3 E-10 C k
3o. 10
"' h '
gg" g 8.3g' Probability Probability Probability
." Key M = mean _ .",$;[o Ky M = meen d m = median A N h* tik m = Mien th = Wie cy,s 6 tli = mtak
$ : __ Safety Goal 7 yA
- 16* -
M E > g6- % 16 3 1 > c d 26', F 6 y { c - , 3 M_ 2 .3 16* , t c F rf d 16*- 2 Id - ' u E 8 ' S 3 cif 310~* , C r5 3 e I l e- E >
$ 16 * '
a s bI
- 5 StL
[J N
~ gga 'J h g gg' gg*
o. Probability 2 Probatnhty Probability Figure S.18. Distributions of Annual Risk for Seismic Initiators LUIL Hazard Distributions
y y t t h i l f . i b i b a _ ^ t m - - b h
# w ~ P o
r s a *
~ ~ " P o
r e l F nt li l w " e ne e W y ai ed mw p
.m.G o y
s r
= t e o
== t W Mmh rt f
% i t
a i n
, 1 , 1 1 1 61 gg 6I g g
61 1 ' g '61 6 1 G 6 1 I c Ed 23$]- 6m_
. t Nb[- :>$5e Ig0 E' 331gT6c~ .
i m s i e S r s o n f o i k t s i u _. t y y R ib h ti t l r a ts i i b b
- _ - ~ ui n
, A. h ,
A. P nD
, h P e ~ "
A dr
" nt li W " f ne u aid u o zaa e e s m mma nH o
==p= i I t R P Mmh buE y t e
K i
- r gg g 13 6 t 81 dI 81 3 dx d I 1 s i
% efu 2a 35 e$s.5& x .yS5e $r.E $ ch 38e9"$o2 2 D 9 1 S e r u g i F _ t y t y _ h i h i b - b a m
- - - b - __ - i _
o r _ v. h h
- h P + u " h P W" 5 t
nt l i a v. t S na. aid m e e i mmg
===
Mmh rt e i 1: 1 ~ 1 1 , 1 K , - .
'6 6 g 6 6 6 6 6
* * , ~
"6 6
'0 'g *0 '6 1 1 g 1 1 1 1 g 1 1 1 1 1 1 m _e , gis N?g5e tE6,.6]
e"ot n a.
A major change in this study relative to the RSS is the decreased probabi. lity of early failure of the containment for non bypass accidents. The estimates of the failure pressure of the Surry containment have increased, while the estimates of the containment loads at the time of lower head failure have decreased. Although the pressure rise at VB now contains the contributions from direct heating of the containment, this addition has been more than offset by the consideration of mechanisms that lead to depressurization of the RCS before failure of the vessel. Furthermore, the current analysis includes the possibility of arresting the core damage process before vessel failure and achieving a safe, stable state. This analysis includes accidents initiated by SGTRs that are not treated in the RSS. The normal (*G") SGTRs are not large contributors to risk, but , the "11" SGTRs, in which the SRVs on the secondary system stick open, are ! major contributors to latent cancer fatalities. Stability of the Analysis. To determine the stability of the integrated j risk analyses performed for NUREG 1150, a second sample was generated and the entire analysis for internal initiators at Surry was repeated. The , second sample is just as valid as the first sample, and differs from the !' first sample only in the use of a different random seed in the Latin ilypercube Sampling (LilS) program. Therefore, differences in the results : between the two samples are an indication of the robustness,of the analysis methods. i Figure S.20 displays four statistical measures of the families of CCDFs for both samples. Considering the range of the distributions, as indicated by '. the. distance between the 5th and the 95th percentile curves, the agreement i between the two sampics is remarkably good. Since the family of CCDFs is the most basic measure of risk, this agreement indicates that the methods , used for the propagation of this integrated risk analysis are sound, i l S.8.3 Imoortant Contributors to Risk l There are two' ways to calculate the contribution- to mean risk. The
. fractional contribution to - mean . risk :(FCMR) is ; found by dividing the average risk for.the subset of interest for the sample by the average total risk for. the sample. The mean' fractional contribution to risk (MFCR) is-
' found by determining the ratio of the '. risk for the ; subset of-interest to ,
- the total risk for each observation and then averaging _over the sample. ,
Results of computing the contributions to the mean risk for internal
. initiators, by the. two methods are .given below, percentages are shown for the- two samples for early fatalities and latent cancer fatalities for the four PDS groups making substantial . contributions to risk. LOCAs, %" Transients, and ATWS each contributed less than 3% of the mean risk for all.
risk measures. != S.44 i l
1 Y t0t -3 .-, ., , -, Y Percentile 7 tot-4 . . hbjh , . b t ort 1
+
$0th .
tot-5 g . . >.itt . . . , b 'ose........,**::::.# 3 t0E-6 , :.', ; ,. , , , tot-7 '. ' t0E-8 ' L 1 l...'a....u , t0E-9 \ k ,':' v '. ...- ,'r- . O tot-in - - " 'b "- "' 1'- - t0L-4 LOL-3 t0L-2 LOL-1 LOLO LOLI - " l'.0L2 - LOL3 - t0[4 Early Fatalities tot-3 , , , - Et0E-4 r sit::s... , M "' :: htot-5 7 ,....
- t00-6 7
,...%'***r:................ * ,
a
...,*.*e, g%
4 , o t00-7 r
, 's',,
- 3 "
5 , tt 10E- B Percentile
* \* '3
- 9 *'N
$ . . u.5th con . . ',
" t;om _
}1.00-9 . . >.t h . . .
1 O tot-1n , m, b
,- ,3 ;,gg g {pt.3 tot 0 .Oli toL2 t0L3 t0L4 tot 5 l Laient Concer rotolities i
Figure S.20. Statistical Measures for Both Sarnples S.45
- h. - i-..- n
Sontributors (%) to Mean Early Fatality Risk for Internal Initiators Samtse 1 Samole 2 PDS Groun ,J,QiB MFCR FCMR MFCR Slow SB0 8.6 7.7 15.6 7.3 ; Fast SB0 8.6 1.3 13.9 1.7 i Event V 77.3 $7.4 62.6 64.0 SGTR 4.1 29.0 6.9 22.3 i Contributors (%) to Mean Latent i Cancer Fatality Risk for Internal Initiators ! Samnie 1 Samole 2 PDS Croun ,J, Gig E _FCMR .MFCR Slow SB0 10.9 15.2 14.6 14.9 Fast SB0 4.6 3.6 8.6 3.9 , Event V 34.3 15.9 25.5 15.8 SGTR 46.5 57.0 47.0 56.8 Figure ' S .21 shows pie charts for the contributions of the summary PDS groups to mean risk for internal initiators for these two risk m00sures for both methods and both samples. Figure S.22 displays similar pie charts for the contributions of the summary APBs to mean risk. . Since the second sample is as valid as the first sample, and more basic measures of risk indicate that the analysis is robust and repeatable, it is clear that the fractional contributions to mean risk can only be interpreted in a broad sense. That is, it is valid to say that Event V is the major contributor to early fatality risk at Surry, or that Event V contributes on the order of 2/3 of the mean risk due to internally initiated accidents at Surry. It is not valid to state that Event V contributes 77.3% of the early fatality risk at Surry. The reason the contributors to mean risk appear to be unstable is that the : expected risk for each obse.vation is typically dominated by a few APBs ' which have both high frequent.v and high source terms, and that the mean , risk is dominated - by a few ob.ervations which have very large values of I expected risk, About 10 observa. ions contribute to most of the mean risk. While-the sample as a whole is rep oducible, the 10 or so observations that control mean risk are generally no reproducible. Since it is - the exact 4 nature of t.hese 10 (approximately) observations that determine the contributors to mean risk, it'is not surprising that this is not a robust measure of tha entire risk analysts. Even though the measures for determining the contributors to mean risk are only approximate, the types of accidents that are the largest contributors to the risk from internal initiators at Surry are clear. For early fatalities, which depend on a large early release, the risk is dominated by Event V. Event V not only proceeds quickly to VB, but it creates a bypass S.46
I " ; ; i;I L l b k, 3, z - 3c
'n r
t a r r/ n a e y r w ks ae i y n, e va r-R -r y o w- k s _ ks o t t t l i c , a. i i c R ae t a ae R y R SR n a t l i l a
/
0 rec "/ 0 M e B 1
- n1
- o y 6 1
Ca 84 - t . ii t a n s E e p n - t
- ,= u
. e 2 la 2 = o r e e ns e a G M l p s l p t n '3 2 e a S m a M me r m D F S S u y
. r
- t. a n
a 3N u n
#no - -
u S r _. " f b e w o
*, 3, m IT's g""'
n s n
, y I{* o
, i
- t 3ao u b
i _ r f t r n e r I m n a r n . o w k a e u C ks e y i s ey r w _ i - R r R -r . y t yt o c t i t o j=. 1 2 l c i l a ea t a ae S . t R S/ SR e r */ r f y *0 e0 c u g r 1 n1 a*
- Ca 2 i -
E O. F - n2 t 5 , e e n e h 3, M- t a - i 8 1 l 1 3, lep n a p le n e , m a u c M ma y c S P S e z
'ao -
s
".C
; ; l j:( ?l'!
- t 3S 3z ee t T *,
o , ' r a n e r s r e w e y r e s ks e i a y _ R - ks to i c yor l
?r-t R ae i t k l
aace = s tyR t i i l / a *- SR- / R - t 0 5 n S1 1' 0 = a - im'~1
- 8 - -
y6 1 e
' M _
Ca 8' l r 1
*s
= :, .
a 4 . o E n - t n- / F= t . a e 2 t e a-r. e 7 _ B s M ep l mn C F 6 l 2
- ap nel - Ca a cvo F
A
= en y a w t
1
-_ Mm e r S F _
S a _ k :7 =
=
u _ S - f o 3 % s
,%4 s 0 5 n i 4 =
0 o E . *c o .
- =' i
. . t
- u _
b i r _ t n o r a enr - . C e w S"" - r y r k a iy s e _ - 2 2 ks to i c R - yro ?r 4 S _ R a t i t 7 e e aace l r , tyR
/ t u aR i
l g a* re-*/ _ i t h10* c0 F y0 n1
- s. h a* "
l r 2 C 2 w E a t 5 'se n - e a e 1 M ep l a t L1 n-a n ne ap l m h w e r e Mma - s S _ =
'g
of the containment as well. The probability of CF at VB is low at Surry for the most frequent CD accidents, SB0 and IDCA, as discussed above. The SGTR accidents that lead to large releases, the "H" SCTRs with stuck open secondary SRVs, progress to VB only af ter many hours. Thus, Event V accounts for most of the large, early releases, and most of the early fatality risk. For latent cancer fatalities, and the other consequence measures that depend primarily on the total amount of radioactivity released, the risk is dominated by Event V and SGTRs. The SGTRs contribute tnore than Event V, and most of this contribution comes frorn the "H" SGTRs (stuck open second-ary SRVs) . Although this accident is unlikely (MCDF about 1.0E 6/R yr), there is a direct open path from the reactor vessel to the environment throughout the accident. The probability that the break point in Event V will be underwater when the releases start is estimated to be about 0.85. Results of computing the contributions to the mean risk for seismic initiators by the two methods are given below. Percentages are shown for the two hazard distributions for early fatalities and latent cancer fatalities for the three seismic PDS groups. Contributors (t) to Mean Early Fatality Risk for Seismic Initiators LLNL EPRI PDS Group FCMR MFCR FCMR .MFCR LOSP 4.7 4.0 1.4 6.5 SB0 65.2 37.2 44.1 38.1 LOCA 30.2 58.8 54.5 55.5 Contributors (t) to Mean Latent Cancer Fatality Risk for Seistnic Initiators LLNL EPRI PDS Grotm .JQEl MFCR FCMR MFCR LOSP 6.0 8.9 10.8 15.4 SB0 62.3 43.8 43.1 43.2 LOCA 31.7 47.3 46.0 41.4 There are no bypass initiators among the seismic accidents, and the proba-bility of CF at VB is low, so early fatality risk is largely due to the initial contaitunent failures that accompany failure of the SG or RCP sup-ports and CF at VB, S.49
S.8.4 1mnortant Contributors to the Uncertainty in Risk The important contributors to the uncertainty in internally initiated risk are determined by performing regression based sensitivity analyses for the mean values for risk and partial rank correlation analyses for the risk CCDFs. The largest contributors to the uncertainty in the risk at Surry are variables that determine the frequency of bypass accidents and parameters that determine the release fractions. The most important contributors to the uncertainty in mean risk are: the initiating event frequency for Event V, the initiating event frequency for SCTRo, the fractional release from the reactor core to the vessel, and the release fraction from the vessel to the environment for SGTRs. These same four variables are the largest contributors to the variability in the risk CCDFs as well. While not dominant for all risk measures for all values of the risk measure, they, or a subset of them, are important for a significant fraction of the range of the risk measure for all risk measures. Other variables that are important for some parts of the range, or for some risk measures, include: the release fractions from the contaitunent , the LDSP event, failure of the DCs to start, and failure of the DGs to run. Important contributors to risk are not determined for the fire initiators. Uncertainty in the seismic risk is dominated by the variability in the seismic hazard distribution. S.9 Insichts and Conclusions Core Damare Arrest. The inclusion of the possibility of arresting the core degradat; ion process before vessel failure is an important feature of this analysis. For internal initiators, there is a good chance that non bypass accidents will be arrested before vessel failure. This may be due to the recovery of offsite power or the reduction of RCS pressure to the point where an operable system can inject. The arrest of core damage before VB plays an important part in reducing the risk due to the most frequent types of internal accidents SB0s and LOCAs. For fires, there is no possibility of core damage arrest since the initiating fire destroys the ability to provide control or motive power to the coolant injection systems. For accidents initiated by earthquakes, core damage arrest is not possible in the SBO accidents because the switchyard is destroyed. The fraction of accidents that do not progress to vessel failure in the LOSP (No SBO) and LOCA seismic groups is significant, however. Dearessurization of the RCS. Depressurization of the RCS before the vessel fails is important in reducing the loads placed upon the containment at vessel breach and in arresting core damage before VB. For accidents in which the RCS is at the PORV setpoint pressure during core degradation, the effective mechanisms for pressure reduction are temperature-induced failure of the hot leg or surge line, temperature induced failure of the RCP seals, and the sticking open of the PORVs. All of these mechanisms are inadver-tent and beyond the control of the operators. The apparent beneficial effects of reducing the pressure in the RCS when lower head failure is S.50
t E imminent indicate that further investigation of depressurization may be warranted. It is somewhat unsettling that the probability of containment failure depends on RCS pressure boundary failures that occur at unpredict-able locations and times. Studies of the effects of increasing p0RV capacity, providing the means to open the p0RVs in blackout situations, and changing the procedures to remove restrictive conditions on deliberate RCS pressure reduction might prove rewarding in decreasing the probability of early containment failure at pVRs. Containment Failure. If a core damage accident proceeds to the point where the lower head of the reactor vessel fails, the containment is unlikely to fail at this time. This is partially due to the depressurization of the RCS before vessel failure and partially due to the strength of the Surry containment relative to the loads expected. No containment failure is more likely than containment failure for all types of initiators. If the containment does fail, it is more likely to fail many hours af ter VB than at VB. The mode and time of failure depends upon the availability of CIIR. If CilR is recovered within a day or so, basemat meltthrough is the most probable failure mode, If CilR is not recovered within days, an overpressure failure is the likely mode about a week after the stare of the accident. For seismic initiators, almost all early failures of the containment result from initial failures to failures of the steam generator or reactor coolant pump supports. Bvoams Accidents. Bypass accidents dominate the risks that depend on a large early release as well as those which are functions of the total release. Event V is the accident most likely to result in a large, early release for internal initiators. SGTRs are also important contributors to large releases, but most of the large releases due to SGTRs occur many hours after the start of the accident. The most important SCTRs are those in which the SRVs on the secondary system stick open. Although the bypass accidents are not the most frequent types of internal accidents, the low probability - of CF, especially early CF, for the non bypass accidents results in the largo contributions of the bypass accidents to risk. Fission Product Reles.31g. There is considerable uncertainty in the release fractions for all types of accidents. For most accidents, the central portions of the release fraction distributions are below most release fraction estimates made several years ago. While the upper portions of the release fraction distributions are comparable with the values of the RSS,3 many of these distributions now extend to release fractions several orders i of magnitude lower than those of the RSS. , Comoarison with the RSS. The distributions for annual risk resulting from the current analysis of the offsite risk from internally initiated acci-dents at the Surry nuclear power plant are lower than those found about 15 years ago in the RSS. For early fatalities, the 95th percentile of the current distribution lies below the 5th percentile of the RSS distribution. 1 i For latent cancer fatalities, the 95th percentile of the current distribu-tion is slightly greater than the median of the RSS distribution. The most l frequent accidents. SB0s and IDCAs, are unlikely to result in early con- l l tainment failure. This is due to a number of factors, including: consider-S.51
ation of core dama6e arrest, higher estimates of the containment failure pressure, reduced estimates of the pressure risa at VB, and lower release fractions for many accidents. Uncertainty in Rigl;. Considerable uncertainty is associated with the risk es:imates produc3d in this analysis. The largest contributors to this uncertainty are the frequencies of the initiating events, especially for the bypass and seismic initiators, and the uncertainty in some of the parameters that determine the magnitude of the fission product release to the environment. The distributions for annual risk resulting from this analysis are much wsder than those from the RSS. The additional ; uncertainty is all in the direction of lower risk. Propagation of the 1 uncertainties in the accident frequency, accident progression, and source i term analyses throu6h to risk allows the uncertainty to be quantitatively calculated and displayed. Risk from Fire. The risk due to fires at Surry is lower than that from internal initiators or carthquakes. While there is no chance of core damage arrest due to disruption of ECCS control or motive power, the fire core damage frequency is about one fourth that for internal initiators, there are no bypass initiators, and the probability of early containment failure is small. Risk from Earthauakes. The offsite risk at Surry due to earthquakes depends very strongly on the set of hazard distributtons used in the accident frequency analysis. If the LLNL hazard distributions are used, the upper portions of the annual risk distributions are about an order of magnitude higher than the risk distributions from internal initiators. If the EPRI hazard distributions are used, the annual risk distributions are roughly comparable with the risk distributions from internal initiators. Much of the offsite seismic risk is attributable to initial containment failures due to.t.he RCP or steam generator support failures that accompany the large "A" 1DCAs. Containment failute at vessel breach is relatively unlikely for the seismic initiators. Comoarison with the Safety Coals. For both distributions for individual fatality probability for internal and fire initiators, the 95th percentile value for annual risk falls more than an order of magnitude below the safety goal. For the seismic initiators, the 95th percentile values for individual latent cancer fatality risk are more than an order cf magnitude below the safety goal for both the LLNL and EPRI hazard distributions. For the probability of an individual early fatality from seismic initiators using the EPRI hazard distributions, the 95th percentile value is abo 9t a factor of four below the safety goal. If the LLNL hazard distributions are used, the upper 10% or 15% of the distribution for the probability of an individual early fatality exceeds the safety goal. l S.52
4 .t ( F, ;
"" - , , References
~h- .!
'i '
- 1. USNRC, " Severe Accident Risks: An Assessment. for Five U.S. Nuclear .
p. 4 Power Plants," Second Draf t for Peer Review, NUREC.1150, June 1989. [
, t ' c 2. R. C. Bertucio and J. A. Julius,
- Analysis of Core Damage Frequency: j
, Surry Unit 1," Sandia National Laboratories, NUREG/CR.4550. Vol. 3 *
, iW Rev. 1 SAND 86 2084, April 1990. .{
lL .
- 3. US!;RC , " Reactor Safety Study..An Assessment of Accident Risks in U.S.
- f. Commercial Nuclear Power Plants," WASH.1400 (NUREG 75/014), October v fi 1975.
m r;
!' /
l j .. ' I i ,
.i y .j !'
. . _ f, . ' I
.i : ' - .
/ '
, , l 4 q q
- l;l .1.. ' l
+
y l i 45 . t
@ :[!
6 !
, Eil'. l f.3: ,
e h i c t I
-r i
l f t ((
.t
.I q
l- S.53 J '
\
i: 4 l' , 1
- 1. INTRODUCTION The United States Nuclear Regulatory Commission (NRC) has recently completed a major study to provide a current characterization of severe accident risks from light water reactors (LVRs). The characterization was
' derived from the analysis of five plants. The report of that work, NUREG. 11501 has recently been issued as a second draft for comment. NUREG 1150 is based on extensive investigations by NRC contractors. Several series of reports document these analyses as discussed in the Foreword. These risk assessments can generally be characterized as consisting of four ! analysis steps, an integration step, and an uncertainty step.
- 1. Accident frequency analysis: the determination of the likelihood and nature of accidents that result in the onset of core damage.
- 2. Accident progression analysis: an investigation of the core damage process, both within the reactor vessel before it failm-and in the containment afterwards, and the resultant impact on the containment.
- 3. Source term analysis: an est'sation of the radionuclide transport within the reactor coolant system (RCS) and the containment, and the magnitude of the subsequent releases to the environment.
- 4. Consequence analysis: the calculation of the offsite consequences in terms of health effects and financial impact.
- 5. Risk integration: the combination of the outputs of the previous L tasks into an overall expression of risk.
6 .- Uncertainty analysis: the determination of which uncertainties in the preceding analysea contribute the most to the uncertainty in risk. This volume is one of 'seven that comprise NURE0/CR 4551. NUREG/CR 4551 presents the details of the last five of the six analyses listed above. The analyses reported here start with the onset of core damage and conclude with an integrated estimate of overall risk and uncertainty in risk. This volume, Volume 3, describes these analyses, _ the inputs utilized in them, and the results obtained, for Surry Power Station, Unit 1. The methods utilized in these analyses are described in detail in Volume 1 of this report and are only briefly discussed here, j 1.1 Rackground and Objectives of NUREC 1150
~
Assessment of risk from the operation of nuclear power plants, involves determination of the likelihood of various accident sequences and their , potential offsite consequences. In 1975, the NRC completed the first j comprehensive study of the probabilities and consequences of core meltdown accidents -the " Reactor Safety Study" (RSS).2 This report showed that-the probabilities of such accidents were higher than previously believed, but that the consequences were significantly lower. The product of probability 1.1
]
l l and consequence a measure of the risk of core melt accidents was estimated to be quite low when compared with natural events such as floods and earthquakes and with other societal risks such as automobile and airplane accidents. Since that time, many risk assessments of specific plants have been performed. In general, each of these has progressively reflected at least some of the advances that have been made in reactor safety and in the ability to predict the frequency of several accidents, the amount of radioactive material released as a result of such accidents, and the offsite consequences of such a release. In order to investigate the significance of more recent developments in a comprehensive fashion, it was concluded that the current efforts of research programs being sponsored by the NRC should be coalesced to produce an updated representation of risk for operating nuclear power plants.
" Severe Accident Risks: An Assessment for Five U.S. Nuclear Power Plants"1 is the result of this program. The five nuclear power plants are Surry, Peach Bottom, Sequoyah, Grand Gulf, and Zion. The analyses of the first four plants were performed by Sandia National Laboratories (SNL). The analysis of Zion was performed by Idaho National Engineering Laboratory (INEL) and Brookhaven National Laboratory (BNL).
The overall objectives of the NUREG 1150 program are given below.
- 1. Provide a current assessment of the severe accident risks to the public from five nuclear power plants, which will:
- a. Provide a " snapshot" of the riska reflecting plant design and operational characteristics, related failure data, and severe accident phenomenological information extant in 1988;
- b. Update the estimates of the NRC's 1975 risk assessment, the
" Reactor Safety Study";2
- c. Include quantitative estimates of risk uncertainty, in response to the principal criticism of the " Reactor Safety Study"; and
- d. Identify plant specific risk vulnerabilities, in the context of the NRC's individual plant examination process.
- 2. Summarize the perspectives gained periorming these risk analyses, with respect to:
- a. Issues significant to severe accident frequencies, consequences, and risk;
- b. Uncertainties for which the risk is significant and which may merit further research; and
- c. Potential for risk reduction.
- 3. Provide a set of nothods for the prioritization of potential safety l issues and related research. l l
1.2
i l 1 These objectives required special considerations in the selection and development of the analysis methods. This report describes those special considerations and the solutions implemented in the analyses supporting NUREG 1150, 1.2 Overview of Su gv Power Station. Unit 1 The subject of the analyses reported in this volume is the Surry Power Station, Unit 1. It is operated by the Virginia Electric Power Company and ' is located on the south bank of the James River in southeastern Virginia, about 10 miles south of Williamsburg, Virginia. The nearest large city i:: Norfolk, Virginia, approximately 35 airline miles to the southeaat of the plant. Two units are located on the site; Unit 2 is essentially identical to Unit 1. The nuclear reactor of Surry Unit 1 is a 2441 MWt pressurized water reactor (PWR) designed and 'ouilt by Westinghouse. The RCS has three U tube steam generators (SCs) and three reactor coolant pumps (RCPs). The containment-and the balance of the plant were designed and built by Stone and Webster. Unit 1 began commercial operation in December 1972. 4 There are three diesel generators (DGs) at the Surry site to supply emergency ac- power if offsite power from the grid is lost. One of these DGs is dedicated to Unit 1, one is dedicated to Unit 2 and the third DG may be aligned to supply either unit. Each unit has its own set of batteries to supply general emergency de power. Each DG obtains starting power from a separate set of batteries. The auxiliary feedwater system (A}VS) has three pumps; two are driven by electric motors; the third is driven by , steam turbine. The A}VS takes suction from the condensate storage tank (CST). There are three charging pumps; they also serve as high pressure injection (llPI) pumps. There are two low pressure inj ection (LPI) pumps; they are self cooled. Both the llPIS and the LPIS can function in a recirculation modo as well as in an injection mode. In the injecticn mode they take suction from the refueling water storage tank (RWST); in the recirculation mode they take suction from the sump. Surry also has three accumulators to provide immediate, high-flow. Iow pressure injection. Reactor coolant system (RCF) overpressure protection is provided by three code safety valves and two power operated , relief valves (PORVs). Service water for cooling condense. s, pumps, and heat exchangers is obtained by gravity flow from an elevated service water canal. This canal is continuously supplied with river water by electric pumps. If ac' power is lost, the service water canal will drain in approximately 30 minutes unless a number of large manual valves are closed. 1 The Surry containment is a reinforced concrete cylinder with a hemispherical dome. A welded steel liner forms the pres are boundary, rigure 1.1 shows a section through the Surry containment. The volume is 1,800,000 f t3 , and the design pressure is 45 psig. During operation, the interior of the containment is maintained about 5 psig below ambient atmospheric pressure. Normal containment cooling is by fan coolers. These are not safety grade and they will be partially submerged if the sump is l full of water. Emergency containment heat removal is by the spray systems. 1.3
Containment Sprays 200-ton Crane df A . g:' WRecirc. 8 pray ' 8 .- .
',.,.i I'
*. I I
)<
. ,j ..'
T ,, Manipulator '
'/.'.-
Steam Crane Pressurizer 4 . yenerator , . .,; O
.h !
, \ f
'*. t I
.p . . .
k
.. . .w., c. . . . .
3, a ,. . .., ..m ... ,, .
.g s
* ,. . s- )
i;
'a j.' .
e .
\\ ..
.I' D, , . ,
N., 6 .
.o
., '.'c,. J .. 5 -,
e,
- .< ~g- ,- .
, . . e. . _ ..
> ',1" l .'
, i i l. '.',
,7 ,
.. H. ,
g
' 'y
. '. Reactor R. C. '.
'.. [.'f
. Pump- .
- P.:.....r..
, ,. . ..z....
_ r,.. , ,
,e. '* . * .. '. : ...
,e..'. ,e*
- - e' a. ..
. Accumulator : - .
F
- p. , .
. : Reactor ..,
. D. (/ ; .' : Cavity ,P. l'.,
. ,.. , . .J. ,.. ... . . . . .. .. . -... __ .. .i D.. g..3.,:;.,.
.f m. ..,' '.e,..3.. i;s..;..:,.'f. ... .. . ;;i.; .)e..;
..9 .. ,'. ..
. . i.. .'..t. .q. .. . .. ..;. . -; y:.
t . . e.,:.v.'. .. .; ,;g. i.;,. ; '..s.').;..;
.'....e
.... ' . . ..'.'s,,.. : ' :/ '. *.,.. .: .t\'. *a :'.', .'.* ? ' * . i'.. ,s. e. ,. . ._.. ' fe.* **.; 'g.;/ , '.'..i..t P ,t .*..?.
. *..'.... *: .. .. . .i f f.'l.
.i'.','.*
Figure 1.1. Section of Surry Containment. 1.4
The containment spray injection system has two_ trains, each with one pump which takes suction from the RWST. There are two containment spray recir- I culation systems, each with two trains. Each of the six containment spray i trains is independent of the other spray systems; except for that, each train requires electrical power for the pumps. Each containment spray recirculation train includes a heat exchanger that is cooled by the service water system and a pump that takes suction directly from the containment sump. One system has its pumps located inside the containment and the other has its pumps located outside the containment. There is no connec-tion between the sump and the reactor cavity at a low elevation in the Surry containment. Water from a pipe break in containment will flow to the sump. The reactor cavity will remain dry unless the containment sprays operate. Section 2.1 of this volume contains more detail on the plant's features important to the progression of the accident ' and to the containment's performance. < 1.3 Ch mtes Since the Draf t Report g The Surry analyses for the February 1987 draft of NUREG 1150 were presented in Volumes 1 of the original "Draf t for Comment" versions of NUREC/CR 4551 and NUREC/CR-4700, also published in February 1987. The analyses pa:: formed for NUREG-1150, Second Draft for Peer Review, June 1989, and reported in this volume, are completely new. While they build on the previous analyses and the basic approach is the same, very little from the first analyses is used directly in these analyses. This section presents the major differ-ences between the two analyses. Essentially, the accident progression analysis and the source term analysis were completely redone to incorporate new information and to take advantage of expanded methods and analysis capabilities. Quantification. A major change since the previous analyses is the expert clicitation process used to quantify variables and parameters thought to be large contributors to the uncertainty in risk. This process was used both for the accident progression analysis and the source term analysis. The sizes of the panels were expanded, with each panel containing experts from industry and academia in addition to experts from the NRC contractors. The number of issues addressed was also increased to about thirty. Separate panels of experts were convened for In-Vessel Processes Containment Loads, Containment Structural Pesponse, Molten Core-Concrete Interactions, and Source Term Issuec. To ensure that expert opinion was obtained in a manner consistent with the state of the art in this area, specialists in the process of obtaining expert judgments in an unbiased fashion were involved in designing the clicitation process, explaining it to the experts, and training them in the methods used. The experts were given several months between the meeting at which the problem was defined and the meeting at which their opinions were elicited so that they could review the literature, discuss the problem with colleagues, and perform independent analyses. The results of the clicitation of each expert were carefully recorded, and the reasoning of each expert and the process by which their individual conclusions were aggrc6ated into the final distribution are thoroughly documented. 1.5
Accident Progression Annivsis. Not only was the Accident Progression Event Tree (APET) for Surry completely rewritten for this analysis, but the capabilities of EVNTRE, the code that evaluates the APET, were considerably ! expanded. The maj or improvements to EVNTRE were the ability to utilize user functions and the ability to treat continuous distributions. A user function is a FORTRAN subprogram which is linked with the EVNTRE code. When refe;enced in the APET, the user function is evaluated to perform calculations too complex to be handled directly in the APET. In the current Surry APET, the user function is colled to determine the mode of containment failure and to compute the pressure rise in containment due to hydrogen deflagrations. These problems were handled in a much simpler ! fashion in the previous analysis. The current method explicitly treats the failure modes due to pressure rises that are fast with respect to the depressurization rates from small failures of the containment. The event tree used for the analysis for the 1987 draft of NUREG 1150 could only treat discrete distributions. For example, for the containment failure pressure, only values of 67, 85, 119, 143, and 180 psig were possible in the previous analysis. In the analysis reported here, a continuous distribution is used for containment failure pressure, so the values are not constrained to these five values. Use of continuous distributions removes a significant constraint from the expert clicitations and eliminates any errors introduced by discrete levels in the previous analysis. Another major change in the accident progression analysis is in the binning or grouping of the results of evaluating the APET. In the first analysis, all results were placed in one of about 30 previously defined bins. There were many pathways through the tree that did not fit well into these previously defined bins. For the current analysis, a ficxible bin structure, defined by the characteristics important to the subsequent source term analysis _ was used. This eliminates a major problem in the original analysis process. The event tree that forms the basis of this analysis was completely rewritten. In addition to utilizing a user function for added flexibility, the APET now considers offsite electric power recovery in the period between the onset of core damage and vessel' failure. This led to a significant portion of the station blackout accidents terminating not with vessel breach but in an arrested core damage state similar to THI-2. Additional means of depressurizing the RCS are now in the event tree, These additional mechanisms, along with the higher probabilities for some of them that resulted from the expert clicitations, mean hat the likelihood is small that.an accident that is at full system pressure at the onset of core damage.will still be at that pressure when the vessel fails. Accidents in which core damage begins with LPIS, or both LPIS and HPIS operating are treated in the current APET whereas they were omitted in the previous version. If an event occurs to reduce the RCS pressure in these situations, core damage may be arrested before the vessel fails, leading, by another path, to an arrested core damage state similar to that of THI 2. i 1.6
Source Term Analysis. While the basic parametric approach used in the original version of SURSOR, the code used to compute source terms, has been retained in the present version of SURSOR, the code has been completely rewritten with a different orientation. The previous version was designed primarily to produce results that could be compared directly with the results of the Source Term Code Package (STCP). Discrete values for the parameters that differed from those that produced results close to STCP l results were then used in the sampling process, with the probabilities for i each value or level determined by a small panel of experts. Thus, the l first. version of SURSOR determined uncertainty in the amount of fission 1 products released for the limited number of predefined bins from the STCP as a base. The current version of SURSOR is quite different. First, it is not tied to the STCP in any way. It was recognized before the new version was devel-oped that most of the parameters would come from continuous distributions defined by an expert panel. Thus, the current version does not rely on results from the STCP or any other specific code. The experts utilized the results of one or more codes in deriving their distributions, but SURSOR itself merely combines the parameters defined by the expert panel. Second, SUP.SOR now treats any consistent accident progression state defined by the eleven characteristics that constitute an accident progression bin for Surry. It is not limited to a small number of pre-defined bins as it was-in the-original version. , Finally, a new method to group the source terms computed by SURSOR has been devised. A source term is calculated for each accident progression bin for each observation in the sample. As a result, there are too many source terms ' to perform a consequence calculation for each and the source terms have to be grouped before the consequence calculations are performed. The
" clustering" _ method utilized in the previous analysis was somewhat subjective and not as reproducible as desired. The new " partitioning" scheme developed for grouping the- source terms in this analysis- eliminates these problems.
Conseauence Analysis. The consequence analysis for the current NUREG 1150
~does not differ so markedly from that for the previous version of NUREG-1150 as does the accident- progression analysis and the ' source term .
analysis- . Version 1.4 of MACCS was used for the original analysis, while ! version 1.5 is used for this analysis. The major difference between the two versions is in the data-used in the lung.model. Version 1.4 used the i1 lung data contained in the original version of " Health Effects Models for Nuclear Power Plant Accident Consequence Analysis",3 whereas version 1.5 l of MACCS uses the lung data from Revision 1 (1989) of this report.4 Other changes were made to the structure of the code in the transition from 1.4 to 1.5, but the effects of these changes on the consequence values calculated are small. Another difference in the consequence calculation is that the NRC specified evacuation of 99.5% of the population in the evacuation area for this analysis, as compared with the previous analysis in which 95% of the population was evacuated. 1.7
Risk Analysis. The risk analysis combines the results of the accident frequency analysis, the accident progression analysis, the source term analysis, and the consequence analysis to obtain estimates of risk to the offsite population and the uncertainty in those estimates. This combination of the results of the constituent analyses was performed essentially the same way for both the previous and the current analyses. The only differences are in the number of variables sampled and the number of observations in the sample, 1.4 Structure of the Analysis The analysis of the Surry plant for NURFC 1150 is a level 3 probabilistic risk assessment composed of four constituent analyses:
- 1. Accident frequency analysis, which estimates the frequency of core damage for all significant initiating events;
- 2. Accident progression analysis, which determines the possible ways in which an accident could evolve given core damage;
- 3. Source term analysis, which estimates the source terms (i.e.,
environmental releases) for specific accident conditions; and
- 4. Consequence analysis, which estimates the health ano economic impacts of the individual source terms.
Each of these analyses is a substantial undertaking in itself. By taking care to carefully define the interfaces between these individual analyses, the transfer of information is facilitated. At the completion of each constituent analysis, intermediate results are generated for presentation and interpretation. An overview of the assembly of these components into an integrated analysis is shown-in Figure 1.2. The NUREG-1150 plant studies are fully integrated probabilistic risk assessments in the sense that calculations leading to both risk and uncertainty in risk are carried through all four components of the individual plant studies. The frequency of the initiating event, the conditional probability of the paths leading to the consequence, and the value of the consequence itself can then be combined to obtain a risk measure. Measures of uncertainty in risk are obtained by repeating the calculation just indicated many times with different values for important parameters. This provides a distribution of risk estimates that is a measure of the uncertainty in risk. It is important to recognize that a probabilistic ris* -sessment is a procedure for assembling and organizing information from wany sources; the models actually used in the computational framework of a probabilistic risk assessment serve to organize this informtion, and as a result, are rarely as detailed as most of the models that at, actually used in the original generation of this information. In order to capture the uncertainties, the first three of the four constituent analyses attempt to utilize all availabic sources of information for each analysis component, including 1.8
ACCIDENT ACCIDENT -SOURCE FREQUENCY PROGRESSION TERM CONSEQUENCE ANALYSIS ANALYSIS ANALYSIS ANM.YSIS
" ^
^ ^
- ,r ., ,, ,
m aOOLOOC% ACDDE4T PanaMETmcsesFomuantre AsOuY HE AL TH E FFE CTS DATA PL ANT SYSTEu EQusPMEmT maio PROGRESSlom FstOns pnCouCT Tut ApeSPOR Y ME TE OnoLCOCAL OnTa MODELS tmTEmasAL OPER4TUR FmLus*E EVENT TREE ANDREMQvAL DEuoGA APHIC DATA As0 E x1Eammt EVENTS) DATA 1r 1. ,, ACCIDEN T 1, CUT SE F PLAN T OAwaGE STATE - / O
, engangssion ems
[ s 3 scuaCE TEmu PAATITf04S S - umCCS COwSE QUEmCE S
- g RS y ..
SETS : TE MAC : E VNTRE - $ r MSOR % a : PAAnes N '" I
@ E NPRESSIONS F REQUE h0E S k FRE QUENCES N h
', < \ -n N ,, o 2, n
"'FORwa p FROM *CopeYMeuEm T LOAOs*G 8 PERFOmuAseCE fSSuf S.
E xPEMHhdENTS ANO r FRONT END r$$UES SOURCE TEnna SSUES mWSSEt MT PROORE S904 rSSUES
- CE TAsLE D CDE SS LATWW HYP(RCUS( SAMPLE Figure l.2. -- Overview of Integrated Plant Analysis in NUREG-1150.
i
.past observational data, experimental data, mechanistic modeling and, as appropriate or necessary, expert judgment. This requires the use of relatively quick - running models to assemble and manipulate the data developed for each analysis.
To facilitate both the conceptual description and the computational s implementation of the NUREG 1150 analyses, a matrix representation.e is used to show how the overall integrated analysis fits together and how the progression of an accident can be traced from initiating event to offsite consequences. Accident Frecuency Analysis. The accident frequency analysis uses event tree and fault tree techniques to investigate the manner in which various initiating events can lead to core damage. In initial detailed analyses, the SETS program? is used to combinc experimental data, past observational
- data and modeling results into estimates of core damage frequency. The ultimate' outcome of the initial accident frequency analysis for each plant is a group of minimal cut sets that lead to core damage. Detailed descriptions of the systems analyses for the individual plants are available els ewhe re . e,o,10,11.12 For the final integrated NUREG 1150 analysis for each plant, the group of risk-significant minimal cut sets is uued as the systems model. In the integrated analysis, the TEMAC programlS l' is used to evaluate the minimal cut sets. The minimal cut sets themselves are grouped into PDSs, where all minimal cut sets in a PDS provide a similar set of conditions for the subsequent accident progression analysis. Thus, the PDSs form the interface between the accident frequency analysis and the accident progression analysis.
With use of the transition matrix notation, the accident progression analysis may be represented by f?DS - fIE P(IE*PDS), (Eq. 1.1) where fPDS is the vector of frequencies for the PDSs, fIE is the vector of frequencies for - the initiating events, and P(IE*PDS) is - the matrix of transition probabilities from initiating events to the PDSs. Specifically: fIE - [ fIE t, ..., f1Enig), f1E t - frequency (yr-1) for initiating event i,
-nIE - number of initiating events, fPDS - [fPDS . ..., fPDSnros] .
t fPDS3 - frequency (yr-1) for plant damage state j ,
~
n'DS - number of PDSs, pFDS33 . . . pPDS t,nros P(IE*PDS) - . . PPDSnig,1 ... pFDSnts,nros and pPDSy - probability that initiating event I will lead to plant damage state J. 1.10
The. elements pFDSy of P(IE*PDS) are conditional probabilities: given that initiating event - i has occurred, pPDSg is the probability that plant l damage state j will also occur. The elements of P(IE*PDS) are determined by the analysis of the minimal cut sets wi'5 the TEMAC program. In turn, i both the cut sets and the data used in their analysis come from earlier studies that draw on many sources of information. Thus, although the l-elements pPDSy of P(IE-+PDS) are represented as though they are single numbers, in practice these elements are functions of the many sources of information that went into the accident frequency analysis. Accident Procression Analysis. The accident progression analysis uses event t ee techniques to determine the possible ways in which an accident might evolve from each PDS. Specifically, a single event tree is developed for each plant and evaluated with the EVNTRE computer program.15 The
- definition of each PDS provides enough information to define the initial conditions for the accident progression event tree (APET) analysis. Due to the large number of questions in the Surry APET and the fact that many of these questions have more than two outcomes, there are far too many paths through each tree to permit their individual consideration in subsequent source term and consequence analysis. Therefore, the paths through the trees are grouped into accident progression bins, where each bin is a group of paths through the event tree that defino a similar set of conditions for source term analysis. The proporties of each accident progression bin define the initial conditions for the estimation of the source term.
Past observations, experimental data, mechanistic code calculations, and expert judgment were used in the development and parameterization of the model . for accident progression that is embodied in the APET. The transition matrix representation for the accident progression analysis is fAPB - fPDS P(PDS*APB), (Eq. 1.2) where fPDS is the vector of frequencies for the PDSs defined in Eq. 1.1, fAPB is the vector of frequencies for the accident progression bins, and P(PDS*APB) is the matrix of transition probabilities from PDSs to accident progression bins. Specifically: fAPB - [fAPBt . ..., fAPBgps), n fAPB3 - frequency (yr-1) for accident progression i bin k, nAPB - number of accident progression bins, pAPBgt ... pAPB ,ngp3 3 pAPBnros,1 ... PAPBn ros ,Arn i 1.11
and pAPBx3
- probability that plant damage state j will lead to accident progression bin k.
The properties of ffDS are given in conjunction with Eq.1.1. The elements pAPBg3 of P(PDS*APB) are determined in the accident progression analysis by evaluating the APET with EVNTRE for each PDS group. Source Term Annivsis. The source terms are calculated for each APB with a non-zero condit ional probability by a fast-running parametric computer code entitled SURSOR. SURSOR is not a detailed mechanistic model and makes no ;
~
pretense of modeling the fission product transport, physics, and chemistry from first principles. Instead, SURSOR integrates the results of many detailed codes and the conclusions of many experts. The experts, in turn, . based many of their con::1usions on the results of calculations with codes ( such as the Source Term Code PackaSe,18.17 MELCOR, and MAAP. Most of the parameters utilized ' calculating the fission product release fractions in SURSOR are sampled from distributions provided by an expert panel. Because of the large number of APBs , use of fast executing code like SURSDR is absolutely necessary. The number of APBs for which source terms are calculated is so large that it was not practical to perform a'cor. sequence calculation for every source term. That is, the consequence code, M/.CCS ,18.18. 20 required so much computer time to calculate the consequer.ces of a source term that the source terms'had to be combined into scurce term groups. Each source term group is a collection of source t9rms that result in similar consequences. The frequency of the source terr group is the sum of the frequencies of all the APBs which make up the group. The process of determining which APBs go to ' which source term group is denoted partitioning. It involves considering the potential of each source _ term g 'ap to cause early fatalities and latent' cancer fatalities. ' Partitioning is a complex process; it is discussed in detail in Volume--I of this report and in the User's Guide for the' PARTITION Program.21 The transition matrix representation of the source term calculation and the grouping process is-fSTG = fAPB P(APB*STG) (Eq. 1.3) where fAPB is the vector of frequencies for the accident progression bins defined in Eq. 1.2, fSTG is the vector of ' frequencies for the source- term groups, and P(APB-*STG) is the matrix of transition probabilities from accident _ progression-bins to source term groups. Specifically, fSTG - (fSTG t, ..., fSTGeo), fSTGf - frequency (yr-1) for source term group f, nSTG = number of source term groups, 1.12 l l
1 pSTG11 ... pSTG ,ns7o 3 PSTGngn,3 ... pSTGnAn,nsto and pST0gf - probability that accident progression bin k will be assigned to source term group 1.
'l if accident progression bin k is assigned to source term group 1
,0 otherwise.
The properties of fAPB are given in conjunction with Eq. 1.2. Note that the source terms.themselves do not appear in Eq. 1.4. The source terms are used only to assign an APB to a source term group. The consequences for
. each APB are computed from the average source term for the group to which
-- the1APB has'been assigned.
Consecuence Analysis, Tho ' consequence analysis is performed for each source term group' by the MACCS program. The results for each source term group include estimates for both mean consequences and distributions of consequences. When these consequence results are combined with the frequencies for the source term groups, overall measures of . risk are obtained. The consequence analysis differs from . the preceding three constituent ' analyses in that uncertainties _ are not explicitly treated in the consequence analysis. That is, important values and parameters are determined . f r_om - distributions by a . sampling process = in the accident frequency analysis,.the accident progression analysis' and the source term. analysis.' This is not the case for the consequences in the analyses performed for NUREG 1150.- In the transition matrix notation, the risk may be expressed by rc - -fSTG cSTG (Eq. 1.4) where fSTG is the vector of frequencies for the source term groups defined in Eq. 1.3, rc is the vector of risk mea ures, and cSTG is the matrix of mean consequence measures conditional on the occurrence of individual source. term groups. Specifically, rC'- [ rC3 , ..., rCe), n rc, - risk (consequence /yr) for consequence measure m,. nC - number of consequence measures, 1.13
cSTGn . . . c S TG1.nc cSTO - . . cSTGnsto,2 . . . cSTGsto.nc and c STG f , - mean value (over weather) of consequence measure m conditional on the occurrence of source term group 1. The properties of fSTG are given in conjunction with Eq.1.3. The elements cSTG f , of cSTG are determined from consequence calculations wi*.h MACCS for individual source term groups. Computation of Risk. Equations 1.1 through 1.4 can be combined to obtain the following expression for risk: rc - f1E P(IE*PDS) P(FDS*APB) P(APB4STG) cSTG. Eq.(1.5) This equation shows how each of the constituent analyses enters into the calculation of risk, starting from the frequencies of the initiating events and ending with the calculation of consequences. Evaluation of the expression in Eq. 1.5 is performed with the PRAMISu and RISQUE codes. The description of the complete risk calculation so far has focused on the computation of mean risk (consequences / year) - because doing so maker the overall structure of the NUREG-1150 PRAs more easy to comprehend. The mean risk results 'are derived from- the frequency of the_ initiating events, the-conditional probabilities of the many ways that each accident may evolve and the_ probability of occurrence for each type of weather sequence at the time'of an accident. The mean risk, then, is a summary risk-measure. . More -information is conveyed when distributions for consequence values are displayed. The form typically used for this is the complementary cumulative distribution fuction (CCDF). CCDFs are defined by pairs of-values (c , f) ,- where c is a. consequence value and the f is the frequency with which c ~is exceeded. Figure 1.3 is an example of a CCDF. The construction- of CCDFs is described in Volume 1 of this report. Each mean risk result is the outcwe from reducing a curve of the form shown in Figure 1.3 to a single value. While the mean risk results are often'useful for summaries or high level comparisons, the CCDF is the more basic measure: of risk because it displays the relationship _ between the size of the consequence and frequency exceedance. The nature , of this relationship, 1.e., that high consequence events are much less likely than low consequence events is lost when mean risk results alone are reported. This report utilizes both mean risk and CCDFs to report the risk results. l 1.14
1.0E-3 .., ... ,,, .., , ., . . , , . . . , . T
.o1.0E-4 :
e . x - l
- u. .
2o1.0E-5 .
~
o i 2 - L m 8.1.0E-6 , _ x : ! o - C - I aj1.0E-7 . tr : LE
' ' .1.0E-8 _
[ 8 o w
$1.0E-9 o
.x ,
( ,. L
. 1.0E-10 .................. ...............a . . . . . . . . . . . . . . . . . .......... , . .
1.0E-3.1.0E-2 1.0E-1 1.0E0 1.0E1 1.0E2 1.0E3 1.0E4' 1.0E5 Latent Cancer Fatalities l Figure 1.3. Example Risk CCDP. E l i' l 1.15
'% _8 'T,, "- - _ . . - ' .. ' '. ' ' ' ' Pronaration of Uncertainty throuch the Analysis. The integrated NUREG 1150 analyses use Monte Carlo procedures as a basis for both uncertainty and the sensitivity analysis. This approach utilizes a sequence: X,X, i 2 . . . , X,y (Eq. 1.6) of potentially important variables, where nV is the number of variables selected for consideration. Most of these variables were considered by a panel of experts representing the NRC and its cont <:ctors, the academic world, and the nuclear industry. For each variable treated in this manner, two to six experts considered all the information at their disposal and provided a distribution for the variable. Formal decision analysis techniques 23 (also in Vol. 2 of this report) were used to obtain and record each expert's conclusions and to aggregate the assessments of the indivi-dual panel members into summary distribution for the variable. Thus, a sequence of distributions Dg, D,2 .... D n y, (Eq. 1.7) is obtained, where Di is the distribution assigned to variable X i. From these distributions, a stratified Monte Carlo technique , . Latin Hypercube Sampling,24,25 is used to obtain the variable values that - will
- actuelly be propagated through the integrated analysis. The result of generating a sample from the varfables in Eq. 1.6 with the distributions in Eq. 1.7 is'a sequence Si - (Xit, Xi a, ..., X ,ny], i - 1, 2, . . . , nulS ,
i (Eq. 1.8) of sample elements , - where X u is the value for variable X3 in sample element i and nulS is the number of elements in the sample. The expression in Eq. 1.5 is then determined for each element of the sample. This creates a sequence of results of the form
'rCi - fIE3 Pi (IE-*PDS) P (PDS*APB) i P (APB-*STG) i cSTG , (Eq. 1.9) where_the subscript i is used to denote the evaluation of the expression in Eq. 1. 5 - with the 'ith sample element in Eq. 1.8. The uncertainty and sensitivity analyses in NUREG 1150 are based on the calculations sumnarized in Eq. 1.9. Since P(IE*PDS), P(PDS*APB) and P(APB4STG) are based on re-sults obtained with TEMAC, EVNTRE and SURSOR, determination of the expres-sion in Eq. 1.9 requires ' a separate evaluation of the cut sets, the APET, and- the source term model for each elerc.ent or observation in the sample.
The matrix-cSTG in ' Eq. 1.9 is not subscripted because the NUREG-1150 analysee do not include consequence modeling uncertainty other than the stochastic variability due to weather conditions. 1.5 Ora nization of this Renort This report is published in seven volunes as described briefly in the Foreword. The first volume of NUREC/CR-4551 describes the methods used in the accident progression analysis, the source term analysis, and the consequence analysis, in addition to presenting the methods used to 1.16
assemble the results of these constituent analyses to determine risk and the uncertainty in risk. The second volume describes the results of convening expert panels to determine- distributions for the variables thought to be the most important contributors to uncertainty in risk. Panels were formed to consider in vessel processes, containment structural response, molten core containment interactions, and source term issues. In addition to documenting the results of those panels for about 30 important - parameters, Volume 2 includes supporting material used by these panels and presents the results of distributions that were determined by other means.
-Volumes 3 through 6 present the results of the accident progression analysis, the source' term. analysis , and the consequence analysis, and the ,
combined risk results for Surry, Peach Bottom, Sequoyah, and Grand Gulf, respectively. These analyses were performed by SNL, Volume 7 presents analogous results for Zion. The Zion analyses were performed by BNL, < i This volume of NUREG/CR. 4551, Volume 3, presents risk and constituent analysis results for Unit 1 of the Surry Power Station, operated by the Virginia Electric- Power Company in southeastern Virginia. Part 1 of this volume presents the - analysis and the results is some detail; Part 2 consista of appendices which contain further detail. Following a summary and an introduction,-Chapter 2 of this volume presents the results of the accident progression analysis for internal initiating event, fires, and-earthquakes. Chapter 3 - presents the . result of the source term analysis, and Chapter 4 gives ' the result of the consequence analysis. Chapter 5 summari::es the risk results, including the contributors to uncertainty in risk, for Surry, and Chapter 6 contains the insights and conclusions of the complete _ analysis, l. i 1.17
1.6 References
- 1. U.S. Nuclear Regulatory Commission, " Severe Accident Risks: An Assessment for Five U.S. Nuclear Power Plants," NUREG 1150, June 1989,
- 2. U.S. Nuclear Regulatory Commission, " Reactor Safety Study + An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants,"
WASil-1400 (NUREG 75/014), 1975.
- 3. J. S. Evans et al., "licalth Effects Models for Nuclear Power Plant Accident Consequence Analysis ," NUREG/CR 4214, SAND 85 7185, Sandia National Laboratories, August 1986.
4 J. S. Evans et al., " Health Effects Models for Nuclear Power Plant Accident Consequence Analysis," NUREG/CR-4214, Revision 1, SANDBS-7185, Sandia National Laboratories, and Harvard University, Cambridge, MA, (Part I published January 1990; Part II published May 1989).
- 5. S. Kaplan, " Matrix Theory Formalism for Event Tres Analysis:
Application to Nuclear Risk Analysis," Risk Analysis, Vol. 2, pp. 9-
'18, 1982.
- 6. D. C,_Bley, S. Kaplan, and B, J. Garrick, " Assembling and Decomposing PRA Results: - A Matrix Formalism," in Proceedines of the International Meetinc on Thermal Nuclear Reactor Safety, NUREG/CP 0027, Vol. 1, pp.
173-182, U. S. Nuclear Regulatory Commission,- Washington, D.C., 1982. ^
- 7. R. B. Worre .. " SETS Reference Manual," NUREG/CR-4213, SAND 83 2675, Sandia Nati aal Laboratories, May 1985.
- 8. R. C. Bertucio and J. A. Julius, " Analysis of Core Damage Frequency from Internal Events: Surry, Unit 1," NUREG/CR-4550, Vol. 3, Revision 1, SAND 86-2084, Sandia National Laboratories, April 1989,
- 9. R. C.'Bertucio and S.;R. Brown, " Analysis of Core Damage Frequency from Internal Events: Sequoyah, Unit 1, " NUREG/CR-4550, Vol. 5, Revision 1. SAND 86 2084,- Sandia National Laboratories, 1990.
- 10. A. M. Kolaczkowski et al., " Analysis of Core Damage : Frequency from Internal Events: Peach Bottom, Unit 2," NUREG/CR 4550, Vol. 4, Revision 1 SAND 86-2084, Sandia National Laboratories, August 1989,
- 11. M. T. Drouin et al., " Analysis of Core Damage Frequency from Internal Events: Grand Gulf, Unit 2," NUREG/CR 4550, Vol. 6, SAND 86 2084, Sandia National Laboratories, 1989.
- 12. M. B. Sattison and K. W. Hall, " Analysis of Core Damage Frequency from Internal Events. Zion, Unit 1," NUREG/CR-4550, Vol. 7, Revilson 1, EGG-2554, Idaho National Engineering Laboratory, May 1990.
- 13. R. L. Iman, "A Matrix-Based Approach to Uncertainty and Sensitivity Analysis for Fault Trees," Rink Analysis. 7, pp. 21 33, 1987.
1.18
I
- 14. R. L. Iman and M. J. Shortencarier, "A User's Guide for the Top Event Matrix Analysis Code (TEMAC)," NUREG/CR 4598, SAND 86 0960, Sandia National Laboratories, April 1986.
- 15. J. M. Griesmeyer, and L. N. Smith, "A Reference Manual for the Event Progression Analysis Code (EVNTRE)," NUREG/CR 5174, SAND 88 1607, Sandia National Laboratories, September 1989.
- 16. R. S. Denning, J. A. Gieseke , P. , Cybulskis , K. W. Lee , 11. Jordan, L.
A. Curtis, R. F. Kelly, V. Kogan, and P. M. Schumacher, "Radionuclide Calculations for Selected Severe Accident Scenarios," NUREG/CR-4624, BMI-2139, Vols. 1-5,-Batte11e's Columbus Division, 1986.
- 17. M. T. Leonard et al., " Supplemental Radionuclide Release Calculations for Selected Severe Accident Scenarios," NUREG/CR 5062, BMI-2160, Batte11e's Columbus Division, 1988.
- 18. D. I. Chanin, J. L' Sprung, L. T. Ritchie and H. -N Jow, "MELCOR
-Accident Consequence Code System (MACCS): User's Guide," NUREG/CR-4691, SAND 86-1562, Vol. 1, Sandia National Laboratories, February 1990, 19 . : li - N . _ J ow , . J . L. Sprung, J. A. Ro11stin, L. T. Ritchie and D. I.
Chanin,- "HELCOR Accident Consequence Code System (MACCS): Model ; Description,'" NUREG/CR-4691, SAND 86-1562, Vol. 2, Sandia National Laboratories, February 1990.
- 20. J. A. Ro11stin,- D. I. Chanin and ll . - N . Jow, "MELCOR Accident Consequence Code System (MACCS): Programmer's Reference Manual,"
NUREG/CR-1562,_Vol. 3, Sandia National Laboratories, February 1990.
- 21. R. L. Iman, J. C. llelton, and J. D. Johnson, "A User's ' Guide for PARTITION: A Program Defining the Source Term / Consequence Analysis Interface in the NUREG 1150 Probabilistic Risk Assessments," NUREG/CR-5253, SAND 88-2940, Sandia National Laboratories, May 1990.
- 22. R. L. Iman, J. D, Johnson, and J. C. Helton,'"A User's Guide.for the Probabillstic Risk Assessment Model Integration System (PRAMIS),"
NUREC/CR-5262, SAND 88 3093, Sandia National Laboratories, May 1990.
- 23. S. C. llora and R. L. Iman, " Expert Opinion in Risk Analysis -
The NUREG 1150 Methodology," Nuclear Science and Enciriee rine . 102: pp. 323-331 (1989).
- 24. M. J. McKay, W. J. Conover, and R. J. Beckman, "A Comparison of Three
' Methods for Selecting Values of Input . Variables in the Analysis of Output from a-Computer Code," Technometrics. 21, 239-245, 1979,
- 25. R. L. Iman and M. J. Shortencarier, "A FORTRAN 77 Program and User's s Guide > for the Generation of Latin flypercube and Random Samples for Use with Computer Models," NUREC/CR-3624, SAND 83-2365, Sandia National Laboratories, March 1984.
1 I 1.19 1
- 2. ANALYSIS OF THE ACCIDENT PROGRESSION This chapter describes the analysis of the progression of the accident, starting from the uncovering of the top of active fuel (UTAF) and continu-ing for about 24 h or until the bulk of the radioactive material going to be released has been released. As the last barrier to the release of the fission products to the environment, the responso of the containment to the stresses placed upon it by the degradation of the core and failure of the reactor vessel is an important part of this analysis. The main tool for performing the accident progression analysis is a large and complex event tree. The methods used in the accident progression analysis are presented in Volume 1, Part 1. The accident progression analysis starts with infor-mation received from the accident frequency analysis: frequencies and definitions of the plant damage states (PDSs). The results of the accident progression analysis are passed to the source term analysis and the risk analysis.
Section 2.1 reviews the plant features important to the accident progression analysis and the containment response. Section 2.2 summarizes the results of the accident frequency analysis, defines the PDSs, and presents their frequencies. Section 2.3 contains a brief description of the accident progression event tree (APET). A detailed description of the APET is contained in Appendix A. Section 2,4 describes the way in 'which the results of the evaluation of the APET are grouped together into bins. This grouping is necessary to reduce the information resulting from the APET evaluation to a manageable amount while still preserving the information required by the source term analysis. Section 2.5 presents the results of the accident progression analysis for internal initiators, fires, and earthquakes. 2,1 Surry Features Important to Accident Procression The entire Surry plant was briefly described in Section 1.2 of this volume. This section provides more detail on the features important to the progression of a core degradation accident and the response of the containment to the stresses placed upon it. These features are:
- The containment structure;
- The maintenance of the containment atmosphere below ambient pressure when the plant is operating; e The containment heat removal system;
- The service water canal; and
- The sump and cavity arrangement.
2.1.1 The Surry Containment Structure The Surry containment is constructed of reinforced concrete; it has the ahape of a cylinder topped by a hemispherical dome. The cylindrical portion of the containment sits on a basemat that is 10 ft thick. The wall of the cylinder is about 4.3 ft thick. The dome is about 2.6 ft thick. 2.1
The inner surface of the containment is a liner of welded steel plate, which forms the pressure boundary. Figure 1.1 shows a section through the Surry containment. The volume is 1,800,000 ft 3 , and the design pressure is 45 psig. Due to conservatisms in design and construction, most estimates of. the failure pressure are between two and three times the design pres-sure. The mean of the aggregate distribution for the failure pressure of the Surry containment provided by the Structural Response Expert Panel is 126 psig. The size and strength of the Surry containment mean that it can absorb a great deal of energy without failing. 2.1.2 Subatmosoberic Containment During Operation i When the reactor is operating, the pressure inside the containment is kept at about 10 psia,. about 5 psia below ambient atmospheric pressure. The implication of this is that it makes the probability of pre-existing leaks - -negligible. The vacuum pumps that keep t.he containment atmosphere below ambient pressure are limited in their capacity, so an open hatch or airlock would be quickly discovered even though the opening has an area of only a fraction of.a square foot. The vacuum-pumps would be unable to keep the pressure at 10 psia. The Technical Specifications prevent plant operation much above this pressure, so the rise in containment pressure would force ; -the plant to be shut down until 10 psia could be maintained in the containment. The size of hole which would go urinoticed is so small that it may be ignored. The fact that the containment is maintained below ambient pressure also means that very few lines are normally open into the -containment during normal- operction; thus, the probability of isolation failure is low. 2.1.3 The Containment Heat Removal System During normal operations, containment cooling is by fan coolers. These are not qualified for operation during severe accidents, and in any event,-if the. contents of the refueling water storage tank (RWST) are pumped into the containment, the fan coolers will be partially submerged. Further, the service water flow to the fan coolers is shut off when the containment is isolated. For these reasons, the fan coolers are not considered a viable means of containment heat removal in the Surry - accident progression analysis. Emergency containment heat removal at Surry is only by the spray systems. The containment spray injection system has two trains. The pumps take suction f rom the RWST, and it can function only in the injection mode. There are two containment spray recirculation systems. each- with two , trains. Each of the four recirculation spray trains is completely separate from the other recirculation - spray trains and from the containment spray injection system. Each containment spray recirculation train includes a heat exchanger that is cooled by the service water system. All four spray recirculation pumps take suction directly from the containment sump. One system has its pumps located inside the containment and the other has its pumps located outside the containment. The diversity and redundancy of the spray systems means that there are no significant accident sequences at Surry where the sprays are failed by harc; ware faults. If there is electric power available, and water in the sump, containment heat removal by the recirculation sprays is always available in the accident scenarios of interest. 2.2
_ _ _ _ _ _ _ _ . _ . . . . _ _ _ . . . . . . n . .. 2,1.4 Service Water Canal Service water for cooling condensers, pumps, and heat exchangers is obtained by gravity flow from a service water canal which is elevated above the levels in the plants on which the pumps and condenser are located. This canal is continuously supplied with river water by electric pumps. If ac power is lost, the service water canal will drain in approximately 30 minutes. While the service water flow to many pumps can be shut off from the concrol room, the large lines to the condensers can be isolated only by manually closing a number of large valves outside the plant. The accident frequency analysts estimated that during a station blackout (SBO), these -~ valves would not be closed in time to prevent the canal from draining. The implication of this unique service water arrangement is that when ac power is restored to the plant, emergency coolant injection cannot be restored to the core at once. The pumps are cooled by the service water system, and it takes on the order of 20 to 30 minutes to refill the canal and establish _ pump cooling. This means that to restore core cooling, power has to be restored about half an hour earlier at Surry than it would in a plant [ without a gravity-fed service water system. , 2.1.5 S.umpmand Cavity Arrancement There is no connection between the sump and the reactor cavity at a low elevation in the Surry containment. That is, the sump can be full when the reactor cavity is dry and the cavity can be full when the sump is dry. If the cavity is dry, the water in the sump is unavailable to mitigate the effects of vessel breach (VB) or to cool the core af ter VB. The only effective way to fill the reactor cavity is for the containment sprays to operate. The water falling inside the shield wall and in the refueling basin will drain into the reactor cavity. Overflow from the reactor cavity will drain to the sump. There is no overflow from the sump to the cavity. The sump is so large that, even_with the entire RCS and the contents of the RWST in the sump, there is no overflow into the cavity. Because the sump and the cavity are not connected, the cavity is dry at VB in a number of accident scenarios at Surry in which it would not be dry if the sump and the cavity were connected at a low elevation. Whether the cavity is dry or contains water at VB has implications for the magnitude of the containment pressure rise at VB and whether core concrete interaction (CCI) occurs. The design of the cavity and the adj acent in-core instrumentation room (ICIR) is . such that two containment failure modes important in some other _ plants are negligible at Surry. The seal table forms part of the ceiling of the ICIR. In some plants, the seal table is located between the crane wall and the containment wall. If high-pressure melt ejection (HPME) accompanies VB, it may fail the seal table and allow hot core debris to accumulate in the vicinity of the seal table. At Surry, the seal table is inside the crane wall, so this material cannot attack the containment pressure boundary. Were the seal table outside the crane wall, this material could attack and fail the steel liner. The other negligible failure mechanism at Surry is a direct impulse resulting from an ex-vessel _ steam explosions (EVSEs) at VB. In plants which have a direct water pathway from the reactor cavity to the containment wall, it is possible that the impulse from an EVSE could be transmitted in water to the contain-ment wall and fail it. There is no such pathway at Surry. 5 i 2.3 !
---...i..n ...i.n im
2.2 Interface with the Core Damace Frecuency Analysis 2.2.1 Definition of Plant Damare States Information about the many different accidents that lead to core damage is passed from the accident frequency analysis to the accident progression analysis by means of PDSs. Because most of the accident sequences identified in the core damage frequency analysis will have accident progressions similar to other sequences, these sequences have been grouped together into PDSs. All the sequences in one PDS should behave similarly in the period following the UTAF. For the pressurized water reactors (PWRs), the PDS is denoted by a sever-letter indicator that defines seven characteristics that largely determine the initial and boundary conditions of the accident progression. More information about the accident sequences may be found in NUREG/CR 4550, Vol. 3. Part 1. The methods used in the l accident frequency analysis are presented in NUREG/CR 4550, Volume 1. l Table 2.2 lists the seven characteristics used to define the PDSs for PWRs. Under each characteristic are given the possible values or attri-butes for that characteristic. For example, the first characteristic denotes the condition of the reactor cooling system (RCS) pressure boundary at the time core damage begins (assumed to be approximately when the top of active fuel [TAF) is uncovered) . Table 2.2-1 shows that there are eight
. possibilities for this characteristic: T for transient or no break; A, S, t Sai and S3 for the four size: of break which do not bypass the containment; G and H for steam generator tube ruptures (SGTRs), and V for the large ;
bypass pipe failure. The first charecteristic in the PDS is not necessarily an indication of the initiating event. -It is an indicator of the RCS integrity at the time the core uncovers. That is, if.the initiating event is a transient, say loss of offsite power (lDSP) , but a reactor coolant pump (RCP) seal failure occurs before the onset of core degradation, then there is a small hole in the RCS pressure boundary at the time that core damage begins, which is the time the accident progression analysis begins. The PDS ' for. this accident i would begin with S3 to reflect the fact that there is a small hole in the RCS when this analysis starts. It is the plant's condition at the onset of core - damage that is important- for the accident progression analysis, not what the original initiator may have been. Thus, the first character in the PDS indicates the condition of the RCS at the onset of core degradation. As a holdover from the use of this character to indicate the original initiator, "T" is used to indicate no , break (transient) . An S 2 break is a break equivalent to a double-ended guillotine breck of a pipe between 0.5 and 2 in. in diameter; an Sa break is less a break of a pipe than 0.5 in. in dieneter. An A break is a break of a pipe greater than 6 in. in diameter and an Si break is a break of a pipe between 2 and 6 in, in diameter. Both A and Si breaks are considered together in the accident progression analysis since both result in low pressure in the RCS. SGTRs are S3 size. Almost all pump seal failures result in a leak area equivalent to an S3 break. A stuck-open power-operated relief valve (PORV) is equivalent to an S2 break. Event V is such 2.4
l Table 2.2-1 PWR Plant Damage State Characteristics
- 1. Status of RCS at Onset of Core Damage T - no break (transient)
A - large break in the RCS pressure boundary j St - medium break in the RCS pressure boundary i S2 - small break in the RCS pressure boundary S3 - very small break in the RCS pressure boundary , G - steam generator tube rupture (SGTR) . H - SGTR with loss of secondary system integrity V - large break in an interfacing system
- 2. Status of ECCS B - operated in injection and now operating in recirculation I - operated in injection only R - not operating, but recoverable N - not operating, not recoverabla !
L - LPIS available in both injection and recirculation modes
- 3. Containment Heat Removal Y - oparating or operable if/when initiated
~
R - not operating, but recoverable N - never operated, not recoverable S - sprays operable, but no CHR (no SW to HXs) i
- 4. .AC Power Y - available I P - partially available j R - not available, but recoverable j N - notjavailable, not recoverable
- {
L..5 Contients of RWST l ! Y - injected 'into containment . R - not injected, but could be injected if power-recovered i N - not injected, cannot.be injected in the- future U - injected, but confined to upper compartment
- 6. Heat Removal from the Steam Generators X - at least one AWS operating, SGs' not depressurized Y - at least one AWS operating, SGs depressurized S --S AWS failed at beginning, E-AWS - recoverable C - S-AWS operated until battery depletion, E-AWS recoverable, SGs not depressurized D - S-AWS operated until battery depletion, E AWS recoverable, SGs depressurized N - no AWS -operating, no AWS recoverable
- 7. Cooling for Reactor Coolant Pump Seals Y - operating R - not operating, but recoverable N - not operating, not recoverable 2.5 i
a well known and unique type of accident that the subsequent six characteristics are usually not written out. The second characteristic concerns the status of the emergency core cooling system (ECCS). Recoverable means that the ECCS will operate if or when electric power is recovered. The value "L" for the second characteristic is used when tha low pressure injection system (LPIS) is available to inject when the core is uncovered but cannot because the RCS pressure is too high, "L" implies that HPIS is failed. The letter "L" is chosen for the second characteristic, for example, for the S 2D 3 sequence. This is a small break with failure of the high-pressure injection (HPI) and it is placed in PDS S 2 LW WN . The low pressure injection (LPI) pumps are operable, so if the operators recognize the situation and depressurize to allow injection by the LPIS there is no core damage. The only portion counted toward core damage is the small (about 2%) fraction where the operator does not recognize the situation and does not depressurize the primary system. The-use of the letter "B" for the second characteristic indicates that both the high-pressure injection system (HPIS) and the LPIS are operating but are unable to inj ec t because the RCS pressure is too high. In sequence TaLP, PDS TLW-YNY, for example, the operators cannot open the PORVs and the auxiliary feedwater system (AFW) has failed. Thus bleed and feed is noc possible using the HPIS, nor can the operators depressurize the system to use the LPIS, As in S2 LW-WN, a temperature induced failure of the RCS pressure boundary or the sticking open of the PORVs or the SRVs will allow injection when the RCS pressure falls to the appropriate level. The third characteristic concerns the status of containment heat removal (CHR). Recoverable means that the CHR systems will operate if or when electric power is recovered. The value "S" for the third characteristic is used when the sprays are available, but that there is no heat removal from the spray heat exchangers. Even if there is no heat removal, it is important to know if the sprays are operating because they reduce the acrosol concentrations in the containment atmo ghcre. The fourth characteristic concerns _ th e status of ac pawer. Recoverable means that power can be restored wJ chin One timeframe of the accident, roughly 24 h. Exc(pt for some seiscic events and fire in the emergency switchgear room, electric power in the plant in general ds always consid-cred to be recoverable in those PDSs where it is_not available. However,_ there are cases where an earthquake has failed motor control centers or switchgear although power is available at some levels in the plant, i.e., there is no SBO. The letter "P" is used to denote this situation. At the levels of_ interest (pumps and valves) this type of power loss is considered to be not recoverable. The fifth characteristic concerns the status of the water in the RWST. It is important for the accident progression to know if the water from the RWST is inside the containment where it fills the sumps and the reactor cavity. The value "N" for this characteristic is used when some failure prevents the injection of the RWST, such as a seismic failure of the tank itself, or when the water from the RWST has been injected into the RCS but 2.6
has ended up outside the containment. This occurs in event V when the water is injected into the RCS but flows out through the break into the auxiliary building, and thus is not available inside the containmenr. The attribute "U" applies to PWRs with ice condenser containments and is not applicable to Surry. The sixth characteristic concerns the heat removal from the steam generators (SGs). There are six possible values for this characteristic I since the AFWS may operate for some time in a blackout accident, and the [ secondary system may or may not be depressurized by the operators. The following abbreviations are used in describing the sixth characteristic in Table 2.2-1: E AWS - Electric-motor driven auxiliary feedwater system; and S-AFWS - Steam turbine-driven auxiliary feedwater system. The seventh characteristic concerns cooling for the reactor coolant pump (RCP) seals. Recoverable means that cooling will become available if or when electric power is recovered. 2.2.2 PDS Freauencies This subsection presents the core damage frequencies for the PDSs and PDS groups. The accident frequency analyses for internal initiators, fires, and earthquakes were performed with more observations per sample than were the accident progression analysis and the subsequent analyses. As the samples were different in the random seed as well as the number of observations, the core damage frequencies differ slightly as is to be expected. This subsection lists these differences to facilitate the transition from the accident frequency analysis to the accident progression analysis. The internal initiators, fire initiators, and seismic initiators are considered in turn. 2.2.2.1 PDS Frecuencies for Internal Initiators. Table 2,2-2 lists PDSs for Surry as placed into seven internally initiated PDS groups and gives their core damage frequencies from the sample of 200 observations used for the integrated risk analysis. The 25 internal initiated-PDSs are-all those which had mean frequencies above 1.0E-7/R yr in the sample of 1000 observations used for the stand alone accident frequency analysis. These 25 PDSs account for over 99% of the total mean core damage frequency (TMCDP) of 4.06E-5/yr. One PDS with a mean core damage frequency (MCDF) lest than 1.0E-7/R-yr, CLYY YXY in the ATWS PDS group, is listed in Table 2.2 2. When the list of cut sets and PDSs was finalized, this PDS had a MCDF slightly above 1.0E-7/R-yr. Last minute changes to the cut sets, the reduction in the sample. size from the 1000 observations used for the stand-alone accident frequency analysis to 200 observations used for the inte-grated risk analysis, and the selection of a new seed for the sample selection changed all of the PDS frequencies somewhat. The MCDF for GLYY-YXY dropped to slightly below 1.0E-7/R-yr, but it was retained in the set of PDSs used in the integrated risk analysis. 2.7
Table 2.2 2 PDSs for Surry Internal Initiators Mean CD Group % Mean CD Group Freq , (1) TMCD Freq . (1) %TMCD Number Group Name (1/R vr) Freq. PDSs (1/R-vr) Freq. 1 Slow Blackout 2.2E-5 55,4 TRRR RDY 1,0E 5 24.7 S3RRR-RDR 8,4E-6 20.7 S2RRR RCR 2.0E 6 4.8 TRRR RDR 1.1E 6 2.7 ? S2RRR RDR 7.0E-7 1,7 S3RRR-RCR 2.8E-7 0.7 2 1DCAs 6.1E-6 15.0 S1W 3 WN 1.7E 6 4.2 SLW WN 3 .9.3E-7 2.3 l AIW WN 8,5E-7 2,1 J ALYY YYY 6,7E 7 1.6 StNYY-YYN 6.1E-7 1.5 SLW-WN 3 6.0E 7 1.5 S LYY YYN 4.5E 7 1.1 ANW-YYN 2,7E-7 0,7 3 Fast Blackout 5.4E-6 13.4 TRRR RSR 5.4E-6 13.4 8 i 4 Event V 1.6E 6 4.1 V 1.6E-6 4.1 , i 5 Transients 1.8E-6 4.3' TBYY-YNY 1,0E-6 2.6 ! TLYY-YNY 7.1E-7 1.8 I' 6- ATWS 1.4E 6 3,5 S3NYY-YXN 7.5E-7 1,8 5.7E-7 1.4 TLYY-YXY GLYY-YXY 9.0E-8 0,2
- 7. SGTRs 1.8E-6 4.4 HINY-NXY 1.4E-6 3.4
.GLYY-YXY 1.8E-7 0.4 HINY-YXY 1.3E-7 0.3 GLYY-YNY 1.0E-7 0.3 Total 4.1E-5 Internal Initiators (1) Based on the sample of 200 observations used in the risk analysis.
i. l Note that while Table 2.2-2 reports frequencies for the 25 PDSs, the acci-dent frequencies actually used in the integrated risk analysis were those of the seven PDS groups. That is, the accident progression analysis was
~
performed for aach of the seven PDS groups individually. The 25 PDSs were used in determining the branching for some of the initialization questions in the APET, but the APET was not evaluated for each PDS separately. 2.8
The accident frequency analysis reports the PDS frequencies based on a sample size of 1000 (see Section 5 of NUREG/CR 4550, Vol. 3, Part 1). When considered as a separate entity, a great many variables could be sampled in the accident frequency analysis, and a sample size of 1000 was used. A sample this large was not feasible for the integrated risk analysis. Based on the results from the 1000 observation sample, those variables which were not important to the uncertainty in the core damage frequency were elimi-nated from the sampling, and the cut rets were re evaluated using 200 observations for the integrated risk analysis. As some variation from sample-to sample is observed even when the sample size and the variables sampled remain the same, there are variations between the 1000-observation sample used for the stand alone accident frequency analysis and the 200-observation sample used for the integrated risk analysis. These differences are summarized in Table 2.2-3. For each PDS group, the first line of Table 2.2-3 contains the 5th per-centile, median, mean, and 95th percentile core damage frequencies for the 1000 observation sample used in the stand alone accident frequency analysis. These values are taken from Table 5 5 of NUREG/CR 4550, Volume 3, Part 1. Samples containing 200 observations are used for the integratar' risk analysis at Surry. The 5th percentile, median, mean, and 90th percontile core damage frequencies for first sample are shown on the secuc line of Table 2.2-3 for each PDS group. For Surry only, a second sa sple was drawn and run all the way through to risk. The same statis:ical , measures are shown on the third line for this second sample for each FDs group. The differences between distributions for core damage frequency for the three samples are within the statistical variation to be expected. Note that the fractional contributions of each PDS group to the MCDF in Table-2.2-2 are slightly different from those In Table 2.2-3. This is due to the l fact that the group fractional contributions in Table 2.2 2 are based on ! the first sample of 200 observations, and the contributions in Table 2.2-3 l are based on the sample of 1000 observations. After all the risk calculations were completed and the results reported, it was determined that 'ie cut sets used in the stand-alone accident frequency analysis were not a,actly those used in the integrated risk analysis. A last minute change to the cut sets did not get made for the cut sets used for the integrated risk analysis. The result is that the frequency for PDS TRRR-RSR, the only PDS in internal initiators Group 3, Fast SBO, was about 10% too low. Following Table 2.2-3 are listed the 5th percentile, median, mean, and 95th percentile core damage frequencies for TRRR-RSR for the 1000-obse rvation- sample used in the stand alone accident frequency analysis, the 200 observation sample used in the risk analysis, and the 200 vation sample based on the revised cut sets used in the 1000-
- ion sample. The differences were not great enough to warrant 3 the entire analysis.
2.9 l
Table 2.2-3 Comparison of PDS Core Damage Frequencies for Surry Internal Initiators Core Damare Frecuency (1/R-vr) LHS Sample % Mean TCD PDSs Size (D 5% Median Mean 95% Frea.(U 1 1000 6.1E 07 8.2E-06 2.2E 05 9.5E 05 54.6 Slow SB0 200 1.6E 06 1.1E 05 2.2E 05 6.4E 05 200 S2 1.4E-06 1.0E 05 2.4E-05 7.0E 05 2 1000 1.2E-06 3.8E-06 6.0E 06 1,6E 05 14.7 LOCAs 200 1.2E 06 3.9E 06 6.1E 06 2.0E-05 200 S2- 1.2E 06 3.7E-06 5.9E 06 1.8E-05 3 1000- 1.1E 07 1.7E 06 5.4E 06 2.3E-05 13.3 Fast SB0 200 1.2E 07 1.5E-06 5.4E-06 2.1E 05 200 S2 1.4E-07 1.5E-06 5.7E-06 2.3E 05 4 1000 3.8E 11 4.9E 08 1.6E-06 5.3E-06 4.0 Event V 200 3.6E-11 4.9E 08 1.6E-06 8.2E-06 200 S2 3.6E 11 4.9E 08 1.6E-06 8.5E 06 5 1000 7.2E 08 6.9E-07 2.1E-06 6.0E-06 4.8 Transients 200 1.1E 07 8.2E-07 1.8E-06 5.5E-06 200 S2 7.8E 08 8.0E-07 1.7E 06 5.2E-06 6 1000 3.2E-08 4.2E 07' 1.6E 06 5.9E-06 3.8 ATWS 200 2.9E 08 4.2E-07 -1.4E 06 6.5E-06 200-S2 4.2E-08 4.0E 07- 1.5E-06 5.6E 06 7 1000 1.2E-07 7.4E-07 1.8E 06 6.0E 06 4.8 SGTR 200 4.5E-07 1.4E-06 1.8E 06 4.7E-06 200 S2 5.0E 07 1,4E-06 1.8E 06 4.4E 06 Total 1000 6.6E 06 2.3E-05 4.0E-05 1.3E-04 200. 9 8E 06 2.5E 05 4.1E 05 1.0E-04 200 S2 8.7E-06 2.6E-05 4.2E-05 1.2E 04 (1) The accident frequency analysis used a Latin Hypercube Sampling (LHS)- sample size of 1000. The accident progression analysis used an LHS sample size of 200; 200 S2 denotes the second sample of size 200, (2) Percentages based on the LHS sample size of 1000. I 2.10 ;
Core Damage Frequency Distributions for TRRR RSR Distribution 5% Median Mean 95% L 1000 Observation Sample 1.1E-7 1.7E-6 5.4E 6 2.3E 5 200-Observation Sample 1.2E 7 1.5E-6 5.4E 6 2.1E 5 Used in Risk Analysis / Revised 200 Observation Sample 1.7E 7 1.8E 6 6.0E-6 2.3E 5 _ The differences are even less when all seven internally initiated PDS groups are considered. Core Damage Frequency Distributions for All Seven Internally Initiated PDS Groups b Median Distribution 5% Mean 95% 1000-Observation. Sample 6.6E-6 2.3E-5 4,0E-5 1.3E-4 200-Observation Sample 9.8E-6 2.5E-5 4.1E 5 1.0E-4 Used in Risk Analysis Revised 200 Observation Sample 1.0E 5 2,6E-5 4.1E-5 1.0E 4 l The seven PDS groups for internal initiators-are discussed below. This is followed by presentation and discussion of the fire and seismic initiators. PDS Group 1 consists _of six slow blackout PDSs. . In these accidents,' off- _. site power is lost and the DGsfail to start or.run. The steam turbine-driven AFWS ' operates until. the L&;.i.eri : are depleted. Without power for instruments and controls, the STD AFWS eventually fails. Battery depletion is estimated to take about 4 h. During this time the RCP seals may fail or the PORVs may stick open. Thus the six PDSs in this-group have the RCS in
.different states at the onset of core damage. In two of the PDSs-in'this group, the RCS is intact when the TAF is uncovered. Another two of the PDSs have3 S size breaks (failures of the RCP seals), and the final two PDSs in this group have S 2-size breaks (stuck-open PORVs). The difference between the two "T" PDSs in Group 1 is whether there is cooling for the RCP seals. The difference between the two "S 3" PDSs and the two "S2" PDSs is whether the secondary system is- depressurized while the AFW is operating (before the core uncovers).
PDS . Group 2 consists of seven ~ 1oss-of coolant accident -(LOCA) PDSs. Four of the PDSs have an A size break, and two of. the PDSs have an S t-size break, - which are treated together and denoted "A" PDSs in this portion of the analysis. There- is one PDS with an S2 size break and one PDS with an S3 -size break. Four of the PDSs in this group have the LPIS operating. In PDS ' ALYY-YYY, the accumulators have failed and the LPIS is operating successfully (all trains) . For an A break, the success criteria require 2.11 '
both accumulator injection and LPIS operation. Thus, even though the RCS pressure is low and the LPIS is injecting water successfully, core damage has been assumed. In PDS S1 L W 'YYY, HPIS has failed and the LPIS is operating successfully (all trains). For an S t break, the success criteria require HPI early in the accident and LPIS operation later. In this PDS also, the RCS pressure is low and the LPIS is injecting water successfully, but core damage has been assumed since the success criteria have not been met. In PDS S 2LW-YYY and SaLW-YYY, the break does not depressurize the RCS enough to allow LPI. Thus the accident will progress to vessel failure at pressure too high to allow LPI unless a large temperature-induced break occurs or the primary system is deliberately depressurized. PDS Group 3 consists solely of TRRR-RSR fast blackout. This accident is similar to PDS Group 1, except that the STD AFW fails at the beginning. It proceeds to the onset of core damage before the RCP seals are likely to fail or the PORVs are likely to stick open. Group 4 consists solely of Event V. This is a large break in low-pressure piping following the failure of the two check valves that isolate the low-pressure piping from the RCS. The break is outside containment 'in the auxiliary building, so the break both fails the RCS pressure boundary and bypasses the containment. Internal PDS Group 5 consists of two PDSs that havo failure of both AFW and Bleed and Feed. This PDS group is denoted Transients. In PDS TBYY-YNY, both LPIS and HPIS are available but the PORVs cannot be opened. The operators have failed to depressurize before the onset of core damage. In PDS TLW-YNY, only LPIS is av/ilable. All AFW has failed and Bleed and Feed is not successful because the HPIS has failed. The operators have faf red to depressurize before the onset of core damage in this PDS also. l As the operators have already failed to follow procedures and depressurize the system, no credit may be given for their depressurizing the RCS after j the onset of core damage in PDS Group 5. Since:there is RCP seal cooling and SGTRs are not very likely, the only effective means of depressurizing the RCS are the PORVs/SRVs sticking open or the failure of the hot leg / surge line. (Even though the PORVs cannot be opened from the control room, they may still open as part of their safety function. If they do not I open at all, then the SRVs will open at a slightly higher pressure, e l frequency of SRVs sticking open is assumed to be the same as for P,: r l sticking open.) If the RCS pressure decreases to the high or intermediate range, the HPIS will, if not failed, inj ect. If the RCS pressure decrerms to the low range, the LPIS will inj ect. Group 6 contains the three ATWS PDSs. They differ in the status of the RCS at the time the core uncovers, whether the ECCS worked in the inj ection phase, and in whether cooling for the RCP seals is operating or failed. In this group also, the LPIS is available in some of the fDCs, and will inject if the RCS reaches low pressure. 2.12
PDS Group 7 consists of four PDSs that are initiate d by SGTRs and which do not have scram failures. Il1 NY - NXY i s an SCTR with stuck-open SRVs in the secondary system. IIINY YXY is similar t o FilNY NXY , but in addition has the l RCS PORVs stuck open. ClYY-YNY has the RCS PORVs open since the operators are attempting to keep the core cooled by feed and bleed. GNYY-YXY has no unusual features. HINY-NXY has no possibility of the water from the RWST being injected into the containment; the HPIS pumps the water through tie broken tube and out of the containment through the main steam line in the other three PDSs, the sprays operate while there is still water in the RWST or in the sump, so the cavity is full when the TAT uncovers, or shortly thereafter. t In grouping the PDSs into the seven internally initiated groups shown in Table 2.2 2, no information is lost, nor are inappropriate assumptions made to facilitate this grou ing. For example, all the breaks in PDS Group 2 are not treated as large (A) LOCAs simply beccusa the majority of the group frequency is in the large LOCA PDSs. The appropriate division between large, small ($2 ) . and very small (S3 ) LOCAs is made by using fractions for the branching ratios in Question 1 in the APET. By using fractional branch ratios in Question 1 and other places in the first twelve questions, plac-ing the 25 PDSs into the seven PDS groups causes no loss of information. The six PDSs left out of t. h e accident progression analysis because their MCDFs fell below the cutoft of 1.0E-7/R yr are PDS JCDf_ AINY YYN 2.5E 8
#NNY-NYN 1.4E-9 S 31 NY - YYN 5.0E-8 S 3NNY - NYN 2.7E-9 S 2NNY - NYN 2. 7E 9 S 3NNY - NYN 3.5E-8 Their total MCDF is 1.lE-7/R-yr. Had they been included, they would all have been in PDS Group 2, LOCAs. This group has a MCDF of 5.9E 6/R-vr The PDSs excluded do not have any features which would make the risk i r o .,
them any higher than other PDSs retained in PDS Group 2. Furthermoce, PDS Group 2 does not make a very large contribution to risk. 2.2.2.2 PDS Freauencies for Fire I n i t i a t. o r s . Table 2.2 4 lists fire PDSs for Surry. There were only four PDSs, and they were all placed into a single Fire PDS group. Fire in the emergency switchgear room, S3NNN-NDN , the dominant ftre accident. The S3 break in the RCS is an RCP seal failure due to the lack of seal cooling. The fire has failed the electrical power to all safety systems, so there is no way to replace the water lost through the RCP seals. The APWS train driven by the steam-turbine pump operates until the batteries deplete, but this has little effect on the accident due to the pump seal failure. The S 3 break in the PDS due to fire in the auxiliary building, S 3NYY NYN . is also an RCP seal f ailurt- due to the loss of coM ing. All ECCS is failed, so there is no way to replace the water los: t. cough the pump seals APWS and the containment sprays operate PDS S;NNY - NYi is due to a fire in the control room. Al1 ECCS and sprays are 2 13
failed due to loss of control from the control room. Although the operators could take control of these systems from auxiliary control ; locations, core damage results because they failed to do so. The Sa break is due to stuck open PORVs. The fourth fire PDS results from fire in the cable vault and tunnel. The S3 break is an RCP seal failure due to the loss of cooling. All ECCS and sprays are failed due to loss of control or uiotive power. 1 I Table 2.2 4 PDSs for Surry Fire Initiators Mean Mean Group CD Freq. CD Freq. 4 TMCD HAEA (1/R-vr) Fire Location __ PDSs (1/R vri Fieg. FIRE 1.1E.5 Emergercy Switchgear Room S3NNN.NDN 5.9E.6 $4.3 2.2E.6 Auxiliary Building S3NYY.YYN 20.0 Cable Vault & Tunnel S3NNY NYN 1.4E.6 13.0 Control Room S2NNY.NXY 1.4E.6 12,7 Table 2.2 5 compares the core damage frequency distribution for the 1000 , observation sample used in the fire core damage frequency analysis with the distribution for the 200 observation sample used in the integrated risk analysis.. The stand alone fire analysia included a fif th fire location, the room that houses the service water pumps: for the charging pumps. As the mean value of the core damage frequency for this fire location was less than 1.0E 7/R yr, it was not included in the integrated risk analysis. The differences between the means and medians of the two distributions are negligible. The differences at the extremes of the distribution are larger due to the elimination of the charging pump service water pump room loca-tion.- The 1000 observation values are taken from Table 5.1 of NUREG/CR-4550, Volume 3, Part 3. Table 2.2 5 Comparison of PDS Core Damage Frequencies for Surry Fire Initiators LHS Core Damage Frecuenev (1/R-vr) f Sample PDSn Sleed) 56' Median Mean 954 FIRE 1000 5.4E 7 8.3E.6 "1.1E 5 3.8E 5 200 2.3E 6 -8.4E 6 1.1E 5 2.6E.5 , L (1) The accident frequency analysis used an IJIS sample size of 1000, The accident progression analysis used an LHS sample size of 200. 2.14
i-.-,...-in. W'. Atter all the :isk calculations were complet ed and t he results toported, it was, determined that the input to the UlS program for the 200 - obse rva t ion sample used for the integrated risk analysis, was not the s aive as that used (- L for the final fire runs made for the stand-alone accident frequency analy-i n., t e a d , a preliminary version of the UlS fire input had been used sis for the integrated ris,k analysis. The main difference between the two was the use c' gamma rather than lognormal distributions for the fire initia-tors. The IJIS input for the 200 observation sample was revised to agree with that for the final 1000 observation sample, and TEMAC rerun with the 5 corrected sample Listed below are the 5th percent ile, median, mean, and 95th percentile core damage frequencies for the 1000 observation sample, [ the 200 observation sample u s, e d in the integrated risk analysis, and the revised 200 observation sample The differences were not great enough to warrant terunning the ftre risk analysis. Core Damage Frequency Distributions for Fire Distribution St Median Mean __?bt lOOO Observation Sample 5.4E 7 8.3E 6 1.10 5 3.8E-5 200 Observation Sample ?.30-6 8.4E-6 1.10 5 2.6E-5 Used in Risk Analys,is Revised 200 Observation Sample 1.80-6 8.60 6 1.10 5 2.80 5 2.2.2.3 PDS Freauencies for Seismic Initiators. Table 2.2-6 lists PDSs for Surry as placed into three n,eismically initiated PDS groups. The seismic PDS frequencies were calculated for two difforent sets of hazard distributions; one generated by Lawrence Livermore National Laboratory (LIEL) and one generated by the Electric Power Research I r.s t i t u t e (EPRI). More information on the differences between there two hazard distributions may be 'ound in NUREG/CR 455(), Vol. 3, Patt 3. Table 2.2 6 shows that the core damage frequency is almost an order of magnitude higher for the LIEL hazard distributions than it is fo- the EPRI hazard distributions. SB0 accounts for a larger fraction of the core damage frequency when the LIEL rather than the EPRI hazard distributions are u s, e d . Table 2.2 7 cuspares the core damage f requency dist ribut ion for the 4000-observation straight Monte Carla sample used in the seismic core damage frequency analysis with the distribution for the 200-observation UlS sample used in the accident progression analysis and the subsequent analyses. Each of the three "EQ" PDS groups is divided into two subgroups It was judged that the evacuation of the area around the reactor site would proceed differently for large earthquakes. Therefore, for each seismic group there is a high and a low acceleration subgroup. The dividing line between the two subgroups is a peak ground acceleration (PGA) of 0.6 g. _ The two subgroups must be kept separate in all the constituent analyses of the integrated risk analysis. Thus, Table 2.2-7 lists six seismic PDS groups. The differences between the 4000-observation sample distributions and the 200-observation sample distributions are of the magnitude to be expected considering the different sample sires and the different random seed used. and are not significent
?.15
i j Table 2.2 6 ! Plant Damage States for Sucry Seismic Initiators , LLNL Hazard Distribution 4 Mean CD Mean CD Group- Group Freq.(t) Group 4 Freq.(1) 4 TMCD I Number Hagg (1/R.vri IjiCD Prea. PDSm (1/R vr) Freo. . EQ 1 IDSP 9.1E 5 47.1 TNNY NNY 3.9E.5 20.2 (No SBO) l TLNP NNY 2.2E 5 11.4 l S3NNY.NYN 1.7E.5 8.8 TBYP.YNY- 7.5E.6 3.9 - TLYP YNY 5.3E 6 2.7 EQ 2 SB0 7.9E.5 41.1 TRRN.RNR 2.4E.5 12.4 TRRN.RDR 1.5E 5 7.8 S3RRN RDR 1.3E.5 6.7 ' TNNN.NDN 8.3E.6 4.3 ANNN.NNN 6.6E.6- 3.4 S RRN RDR 5.1E.6 2.6 ARRN RDR 3.9E.6 2.0 S3NNN NDN 3.4E.6 1.8 .; EQ 3 1hCAs 2.3E 5 11.9 S NNY.NYY 5.4E.6 2.8 SgLNP.NYY 5.0E.6 2.6 ANNY NYY 2.4E.6 1.2 AINP.NYN 2.4E.6 1.2 - t SaLYP YYY 1.7E.6 0.9 AIYP.YYN 1.5E 6. 0.8 AIYP.YYY 1.3E.6 0.- 7 S3NNY NYY 1.2E.6 0.6 S3 LNP NYY 9.5E 7 0.5 S3 LYP.YYY 5.0E.7 0.3 AIYY YYY 4.2E.7 O2 l- 1.9E 4 Total All Seismic Initiators based on the LLNL Hazard Distributions l- .(1) Based on the sample of 200 observations used in the risk analysis. 2.16
Table 2.2 6 (continued) Plant Damage States for Surry Seismic Initiators EPRI Hazard Distribution Mean CD Mean CD Croup Group Fre q . (1) Croup % Freq.(1) % THCD Nusher .Name (1/R-vri TMCD Freo. PDSs (1/R vr) Freo. EQ 1 LOSP 1.5E 5 $3.7 TNNY NNY 6.6E 6 23.6 (No SBO) TLNP NNY 2.6E 6 9.3 S3NNY NYN 3.2E 6 11.4 TBYP YNY 1.8E 6 6.4 TLYP YNY 8.4E 7 3.0 EQ 2 SB0 9.4E 6 33.7 TRRN RNR 2.5E 6 8.9 TRRN RDR 1.5E 6 5.4 S3RRN.RDR 2.4E 6 8.6 TNNN NDN 4.2E 7 1.5 ANNN NNN 3.8E 7 1.4 S RRN RDR 6.4E 7 2.3 ARRN RDR 5.0E 7 1.8 S3NNN NDN 1.1E 6 3.9 EQ 3 LOCAs 3.5E 6 12.5 S NNY NYY 8.1E 7 2.9 j S LNP NYY 6.9E 7 2.5 ANNY NYY 4.8E 7 1.7 AINP NYN 3.6E 7 1.3 S LYP YYY 2.3E-7 0.8 A1YP YYN 3.1E 7 1.1 A1YP YYY 1.5E 7 0.5 , SgNNY NYY 1.7E.7 0.6 ' SgLNP NYY 1.4E 7 0.5 SgLYP YYY 5.0E-8 0.2 5.8E 8 0.2 A1YY YYY Total' 2.BE 5 All Seismic Initiators based on , the EPRI Hazard Distributions ' (1) Based on the sample of 200 observations used in the risk analysis. 1 I 2.17
Table 2.2 7 Comparison of PDS Core Damage Frequencies Surry: Internal Initiators LLNL llazard Distribution Core Dnenge Frecuenev (1/R vri UlS Sample % sn TCD PDSs S i ze(1) 5% Median Menn 954 . Ire a . m Peak Ground Acceleration > 0.6 g EQ 1 4000 1.4E 8 7.3E 7 8.3E 6 3.5E 5 LOSP 200 9.1E 9 5.8E 7 9.4E-6 3.4E 5 5.0 EQ 2 4000 1.7E 8 7.7E-7 8.8E 6 3.9E 5 SB0 200 2.2E 8 9.2E 7 1.1E 5 5.3E 5 5.8 EQ 3 4000 5.5E 9 5,4E-7 7.9E 6 3.4E 5 LOCAs 200 9.5E 9 5.5E 7 7.5E 6 3.6E 5 4.0 Peak Ground Acceleration < 0.6 g EQ 1 4000 1.3E 7 7.2E 6 7.1E 5 3.5E-4 IDSP 200 1.0E 7 6.2E 6 8.1E.5 3.5E 4 42,0 EQ 2 4000 1.2E 7 5.1E 6 5.4E 5 2.1E 4 SB0 200 1.2E 7 5.8E 6 6.8E 5 2.9E 4 35.3 EQ 3 4000 1.1E 8 1.2E 6 1.5E 5 6.1E 5 lhCAs 200 1.8E.8 1.1E 6 1.5E 5 7.3E 5 7.9 TOTAL 4000 4.5E ? 1.8E 5 1.7E 4 7.7E 4 All PGA 200 5.3E 7 1.8E 5 1.9E 4 7.6E 4 (1) The seismic accident frequency analysis used a Monte Carlo sample size of 4000. The accident pro 8ression analysis used an Lits sample size of 200, (2) Percer.tages based on the Uls seaple of 200 observations used in the integrated risk analysis. 2.18
i t i Table 2.2 7 (continued) Comparison of PDS Core Damage Frequencies l Surry: Seismic Initiators EPRI Hazard Distribution
- l Core Damage Precuency (1/R vr) f IRS Sample 4 Mean TCD PDSs Sirem St .. Median Mean 954 Frea . W f
Peak Ground Acceleration > 0.6 g i
-EQ 1 4000 8.8E 9 2.0E.7 1.1E.6 4.4E.6 ;
LOSP 200 8.8E 9 2.5E.7 1.1E.6 4.9E.6 3.9 . t EQ 2 '4000 1.1E.8 2.7E.7 1.3E.6 4.6E.5 SB0 200 1.8E.8 3.2E.7 1.1E.6 4.7E.6 3.8 , EQ 3 4000 3.9E.9 1.5E.7 9.4E.7 4.0E.6 ! LOCAs 200 2.1E.9 2.0E.7 9.8E.7 5.5E.6 3.5 { Peak Ground Acceleration < 0.6 g EQ 1 '4000 9.3E.8 2.6E.7 1.3E.5 5.2E.5 [ LOSP 200 9.6E.8 3.6E.6 1.4E.5 7.6E.5 49.9 , EQ 2 4000 8.6E.8 2.0E.6 1.0E.5 3.9E 5 . SB0 200 1.3E.7 2.5E.6 8.4E.6 3.5E.5 29.9 EQ 3 4000 9.6E.9 3.8E 7 2.3E 6 9.6E.6 -{ LOCAs 200 5.3E.9 5.0E.7 -2.5E.6 1.2E.5 8.9 't TOTAL 4000 3.4E.7 7.0E.6 2.9E-5 1.3E.4 Low PGA 200 3.7E.7 9.4E.6 2.8E.5 1.4E 4 . (1) The seismic' accident frequency analysis used a Monte Carlo sample size ! of 4000. The - accident progression analysis. used an LHS sample sizec of 1 200, j
-I (2) Percentages -based on the IMS sample of 200 observations used' in the l integrated. risk analysis.
t F 2.19
i I { i 2.2.3 High Level Grouping of PDSs To provide simpler, more easily understood summaries for NUREG 1150, the seven internally initiated PDS groups described above were further con. densed into the following five groeps: j
- 1. Loss of offsite Power (LOSP) ,
- 2. IJDCAs
- 3. Transients
- 4. Bypass LOCAs
- 5. ATWS.. i These five groups are denoted Summary. Groups or collapsed PDS Groups. The mapping from the seven groups described in the previous section into the five Summary Groups used in the presentation of many of the results is i given in Table 2.2 8. In combining two groups to form one super group, frequency weighting - by observation is employed. The percentages of the t total MCDP given abo've provide only approximate weightings.
Table 2.2 8 Relationship Between PDS Groups and Summary Groups Su===ry Creuo 1 TMCDP PDS Croups t TMCDP
- 1. LOSP 66 1. Slow Blackout 55
- 3. Fast Blackout 12 1
- 2. 14CAs 15 2. LOCAs 15 ,
. 3, Bypass LOCAs 10 4. V _
6
- 7. SGTRs _4 4.iTransients 5 5, Transients 5
- 5. ATWS 4 6.- ATWS 4 .,
l 2.2.4 Variables Samoled~in the Accident Preauency Analysis In the stand alone accident frequency analysis'for internal events -a large. ., number of variables were sampled. (A list of these variables may be found ; in NUREG/CR.4550, Vol._ 3, , Part 1.) Only those variables found to be important to the uncertainty in the accident frequencies were selected for sampling in the integrated risk analysis. These variables are listed.and i defined in Table 2.2+9 (at the end of this subsection). For the regression analysis, identifiers of eight characters or less were required, and these are listed in the first column. Where these differ from the identifiers ,
~
used in the fault trees, these identifiers are listed in the description in. brackets. Generally, the eight character identifiers have been< selected to be as informative as possible to those not familiar with the conventions j used in systems analysis. For example, while Event K is commonly used to 1 t 2.20
- l indicate the failure of the Reactor Protection System (RPS) to insert enough control rods to make the reactor suberitical, the identifier AU.
SCRAM was chosen since it was felt that " auto scram" conveys more meaning to most readers than "K". The second column in Table 2.2 9 gives the range of the distribution for the variable and the third column indicates the type of distribution used and its mean value. The entry " Experts" for the distribution indicates that the distribution came from the accident frequency analysis expert panel. The fourth and fifth columns in Table 2.2 9 show whether the varia-ble is correlated with any other variable and the last column describes the variable. More complete descriptions and discussion of these variables may be found in the Surry accident frequency analysis report (WREG/CR 4550, Vol. 3). This report also gives the source or the derivation of the dis-tributions for all these variables. Host of the variable distributions come from the generic accident sequence evaluation (ASEP) data base. Others were derived specifically for the Surry equipment using plant data. The distribution for the frequency of the LOSP initiating event was derived by combining data from all nuclear , power plant sites with the historical experience at Surry, using the methods of NUREG/CR 5032. The distribution for the frequency of transient initiating events was darived froin Surry data as described in NUREG/CR-3862. The distribution for the probability of failure to scram (AU SCRAM, Event K) was derived from the information in NUREG 1000. The human error probability distributions were derived using the human reliability analysis (HRA) methodology as described in NUREG/CR 4772. Failure of the RCP seals due to lack of cooling was sampled in the follow-ing manner in the accident frequency analysis: eight states were defined, and one of these states had a probaaility of 1.0 in each observation while ' .the other seven states had a probability of 0.0. (When all the probability is assigned to one branch in every observation, the sampling is denoted zero one sampling.) The eight RCP seal states are: Total Start Fault Tree State Leak Rate Time Probability- Identifier 1 750 gpm 90 min 0.535 RCP MCA 750 90M 2 467 spm 150 min 0.120 RCP MCA 467 150 3 183 gpm 150 min 0.020 RCP LOCA 183 150 4 163 gpm 210 min 0.015 RCP WCA 183 210 90 min 5 1440 gpm 0.005 RCP MCA 1440 90 6 183 gpm 90 min 0.010 RCP LOCA 183 90 7 561 gpm 150 min 0.005 RCP LOCA 561 150 8 Normal N.A. 0.290 NO RCP SEAL LOCA The probability for each state was determined by a special expert panel as described in NUREG/CR 4550, Volume 2. The use of this information in the Surry accident frequency analysis is described in more detail in NUREG/CR. 4550, Volume 3. The last state represents success, i.e., no failure of the RCP seals. Design leakage through the seals is about 3 gpm/ pump during normal operation, but non failure leakage could be as high as 21 gpm/ pump when there is no flow of cooling water to the seals. Leakage following 2.21
seal failure could be as high as 480 gpm/ pump or 1440 gpm total. As there were 200 observations in the sample used to determine risk for Surry, state 1 (a total leak of 750 gpm from three pump seals starting at 90 minutes) had a probability of 1.0 for 107 observations and a probability of 0.0 for 93 observations. State 7 (561 spm starting at 150 minutes) had a probabi-lity of 1.0 for only one observation. A random number generator was used to determine which state had the unity probability for which observation. Only two accident frequency variables were correlated in the integrated analysis. As indicated in Table 2.2 9, DG FRUN1 and DG FRUN6 were corre-lated with each other since they represent failures to run for different timec for the same equipment. The failures to run for the steam turbine-driven APW pump (ATp FR6 and ATP FR24) should have been correlated for the same reason, but this correlation was omitted due to an oversight. Neither of the AFW pump failure-to run variables was important in determining the uncertainty in risk, so the effect of omitting the correlation between them is not significant. Table 2.2-10 (also at the end of this subsection) lists the variables sampled in the accident frequency analysis for fire initiators. Since the fire analysis considered random failures in addition to fire damage, these variables were sampled in addition to the variables sampled for the inter-nal initiators. Although only four fire locations are presented in Table 2.2 4, there was a fifth fire location which was not discussed because its mean frequency was below the cutoff value of 1.0E 7/R yr. However, two variables for this fire location, the room housing the service water pumps which cool the charging pumps, remained in the sample file and are listed in Table 2,2 10. The variables sampled in the accident frequency analysis for seismic initiators are listed i'.. NUREG/CR 4550, Vol. 3 Part 3. The variability in the hazard distributio:. is by far the major contributor to the variation in the seismic core damage frequency. Therefore, variables other than the hazard distribution were not considered in the regression analyses performed as part of the integrated risk analysis. 2.22
+
Table 2.2-9 Variables Sampled in the Accident Frequency Analysis for Internal Initiators Variable Range Distribution Correlation Correlation With Descrintion V-TRAIN 1.8E-13 Experts -None Initiating event: frequency (1/yr) 1.5E-5 Mean-5.5E-'7 of check valve failure in one of the LPIS trains. IE-IDSP 2.6E-5 'IDSP Data -None ' Initiating event: frequency (1/yr) 0.28 Mean-0.077 of IDSP. [IE-T1] IE-A 5.0E-5 Iognormal None Initiating event: frequency (1/yr) 0.0032 Mean-5E-4 of a large (dia. > 6 in.) break in the RCS (IDCA). y IE-S1 1.0E-4 Iognormal None Initiating event: frequency (1/yr) ; y 0.0063 Mean-0.001 of an intermediate size (6 in. > w dia. > 2 in.) IDCA. IE-S2 1.0E-4 fognormal None Initiating event: frequency (1/yr) 0.0063 Mean-0.001 of a small break (2 in. > dia. > , 0.5 in.) in the RCS. IE-S3 0.0013 Iognormal None Initiating event: frequency (1/yr) 0.082 Mean-0.013 of a very small (0.5 in. > dia. ) break in the RCS (LOCA). IE-T-ALL 0.67 Legnormal None Initiating event: frequency (1/yr) 41.6 Mean - 6.6 of all transients that require scram (Surry data). [IE-T] 4 IE-T-HIP 0.60 Iognormal None Initiating event: frequency (1/yr) , 37.2 Mean-5.9- of all transients from high (>25%) i power that require scram (Surry , data). [IE-TN] IE-11tFWS 0.096 Iognormal None Initiating event: frequency (1/yr) 5.9~ Mean-0.94 of transients due to Ioss of the main feedwater system (Surry data).
Table 2.2-9 (continued) n= nee Distribution Correlation Correlation With Descriotion Variable O.001 Iognormal None Initiating event: frequency (1/yr) IE-SGTR {1E-T7] 0.063 Mean-0.01 of SGTRs (PWR data). 2.5E-5 Iognormal None Initiating event: frequency (1/yr) IE-DCBUS for loss of a DC power buss. [IE-0.14 Mean-0.005 T5] 9.9E-6 Ingnormal Rank 1 DG-FRUN6 Probability that the diesel DG-EUN1 generator fails to run for 1 h, 0.057 Mean-0.002 given that it starts.[DGN- a-1HR] 6.0E-5 lognormal Rank 1 DG-EUN1 Probability that the diesel DG-FRUN6 0.34 Mean-0.012 generator fails to run for 6 h, m given that it starts. [DGN-E-6HR] 0.0022 Iognormal None Probability that the diesel DG-FSTRT 0.14 Mean-0.022 generator fails to start, given a demand to start. [DGN-FS) 1.8E-4 Lognormal None Fraction of the time that the UNFV-MOD reactor operates with an 0.27 Mean-0.014 unfavorable moderator temperature coefficient. [Z] 1.8E-6 Iognormal None Probability of failure of the AU-SCRAM RPS to automatically insert 7.6E-4 Mean-6E-5 sufficient control rods to terminate the reaction. [K] O.017 Max. Entropy None Probability of failure to effect MN-SCRAM manual scram due to operator error 1.0 Mean-0.17 and hardware faults. [RJ 4.8E-5 Iognormal None Probability of failure of one train AUTO-ACT of an automatic actuation system 0.020 Mean-0.0016 (generic). [ACT-FA] It
Table 2.2-9 (continued) Variable Ranze Distribution Correlation Correlation With Descriotion CCF-RWST 1.5E-6 Iognormal None Probability of common cause failure 0.0085 Mean-3E-4 of the recirculation mode transfer system due to miscalibration of the water level sensors in the RWST (human error). [RMT-CCF-FA-MSCAL] BETA 2MOV 0.0089 Lognormal None Beta factor for common cause 0.55 Mean-0.088 failure of two motor-operated valves (generic). [ BETA-2MOV) BETA-AW 0.0057 Lognormal None Beta factor for common cause 0.35 Mean-0.056 failure of the AWS motor-driven y - pumps (generic). U BETA-LPI O.015 Lognormal None Beta factor for common cause 0.94 Mean-0.15 failure of the LPIS pumps I (generic). AW-STMB 2.0E-8 lognormal None Probability of common cause failure 0.0070 Mean-1.0E-4 of all AWS due to steam binding (back leakage through check valves from MWS). [CCF-IX-STMBD] MDP-FSTR 1.5E-5 Iognormal None Probability of failure to start 0.085 Mean-0.003 (per demand) for motor-driven pumps for which specific plant data was not available (generic). [MDP-FS] AWMP-FS 6.4E-4 Lognormal None Probability of failure to start 0.040 Mean-0.0063 (per demand) for AW motor-driven pumps (from Surry data). [ AW-MDP-FS-W3B]
Table 2.2-9 (continued) Variable Range Distribution Correlation Correlation With Descrintion ANTP-FS 5.5E-5 Iognormal None Probability of failure to start 0.31 Mean-0.011 (per demand) for AW steam turbine-driven pump (from Surry data). [TDP-FS] ATP-FR6 1.5E-4 legnormal None Probability of failure to run for 6 0.85 Mean-0.030 h for the AW steam turbine-driven pump (generic). [TDP-FR-6HR] ATP-FR24 0.01 Max. Entropy None Probability of failure to run for 1.0 Mean-0.12 24 h for the AW steam turbine-drz<en pump (generic). [TDP-FR-24HR}
$ PORV-B1JC 0.0041 lognormal None Probability of failure to open (per 0.25 Mean-0.040 demand) for the PORV block valves (MOVs). [PPS-MOV-FT]
LPRS-MOV- 2.6E-5 Iognormal None Probability of failure (per demand) 0.15- Mean-0.0052 for the suction MOVs in the LPRS, due to hardware failures or plugging. [LPR-MOV-FT] i MOV-FT 1.5E-5 Ingnormal None Probability of failure to transfer 0.085 Mean-0.003- -(per demand) for motor-operated valves (generic). MNV-PG1 4.1E-6 Ingnormal None Probability of failure due to 2.5E-4 Mean-3.6E-5 plugging for manual valves that are flow-tested every month (generic). [XVM-PG-lMO} MOV-PG3 1.0E-5 . .Ingnormal None Probability of failure due to 6.3E-4 Mean-1.0E-4 MOVs that are flow-tested every 3 i months (generic). [MOV-PG-3MO]
Table 2.2-9 (continued) Variable n=n-e Distribution Correlation Correlation With Descrintion MOV-PG12 4.5E-5 Iognormal None Probability of failure due to 0.0028 Mean-4.4E-4 plugging forMOVs that are flow-tested every 12 months (generic). [MOV-PG-12MO} AW-OCC 1.5E-5 Iognormal None Probability of common cause failure 9.5E-4 Mean-1.5E-4 of AFUS due to an inadvertently open cross-connect to Unit 2 (flow diversion). [ AW-PSF-FC} PORV-REC 1.5E-4 Lognormal None Probability of failure of the m 0.85 Mean-0.030 pressurizer PORVs to reclose after - opening (generic). [SOV-00] SSRVO-SB 0.030 Max. Entropy None Probability of failure of an SG SRV 1.0 Mean-0.27' to reclose within 1 h during SB0 (faulted steam generator). [QS-SB0] SSRVO-U2 0.016 Max. Entropy None Probability of failure of a 1.0 Mean-0.16 secondary system SRV at Unit 2 to reclose within I h during 580 at both units. [QS-UNIT 2} SOV-FT 1.0E-4 Iognormal None Probability of failure to transfer 0.0063 .Mean-0.001 (per demand) for solenold-operated valves (generic). CKV-FT 1.0E-5 Iognormal None Probability of failure to open (per 6.3E-4 Mean-1E-4 demand) for check valves (generic).
Table 2.2-9 (continued) Variable Ranne . Distribution Correlation Correlation With Descrintion HE-FDBLD 0.0071 Max.. Entropy None Probability of failure of the 0.71 Mean-0.071 operator to initiate feed and bleed (human error - open PORVs, and
- start charging pump and align suction and discharge valves).
[HPI-XHE-FU-FDBID] HE-PORVS 0.0044 Max. Entropy None Probability of failure of the 0.44 Mean-0.044 operator to initiate feed end bleed (human error; diagnose situation and open FORVs). [PPS-XHE-FO- , PORvSI [ HE-CST 2 0.0065 Max. Entropy None Probability of failure of the
= 0.65 Mean-0.065 operator to align the AWS suction to the backup CST during an 530 with a faulted SG. [ AW-XHE- FG- ;
CST 21 HE-UNIT 2 0.0')36 Max. Entropy None Probability of failure of the 0.36 Mean-0.036 operator to provide AW from Unit 2 via the cross-connect. {XHE-FO-UNIT 2) HE-SKILI. 1.3E-5 legnormal None Probability of human error for 0.077 Mean-0.0026 skill-based human errore (rudimentary actions performed " r_ memory). [XHE-FO-SKILLBASE] RCP-SL-F Experts None Probability of RCP seal failure
'before the onset of core damage.
[See text] i
, , ,. ._ ._. ,_ .v_ ~ -__. . . - . - - <,- .. ,-.~, . - - .. - , -. -- -
Table 2.2-10 Variables Sampled in the Accident Frequency Analysis for Fire Initiators Variable Range Distribution Correlation Correlation With Description IE-AUXBL 0.027 Iognormal None Initiating event - fire in 0.16 Mean-0.064 auxiliary building. [ AUX 1LIARY Brac.] AR-AUXBL 2.4E-4 Max. Entropy None Area ratio in auxiliary building 0.0011 Mean-4.8E-4-' where critical damage occurred. [ Fall SR-AUXBL 0.19 Max. Entropy None- Severity ratio for a large fire 1.0 Mean-0.30 based on generic combustible fuel loading. [FS1]
=
y FX-AUXBL 0.60 Max. Entropy None Fraction of fires extinguished e 1.0 Mean-0.87 manually before critical damage occurred. [Q1TG] HE-AUXBL 0.19 Max. Entropy None Probability the operators fail to 1.0 Mean -0. 26 obtain HPI from Unit 2 to prevent RCP seal failure. [R10P] IE-CBLVT 3.0E-6 lognormal' None Initiating event - fire in cable 0.016 Mean-0.0027 vault and tunnel. [ CABLE VAULT AND TUNNEL] AR-CBLVT 0.012 Max. Entropy None Area ratio in cable spreading room 0.047 f:ean-0.018 where critical damage occurred. [FA2] SR-CBLVT 0.50- Max. Entropy None Severity ratio for a large fire
- 1. 0 -- Mean-0.90 based on generic combustible fuel loading. {FS2]
FX-CBLVT 0.60 hax. Entropy- None Fraction of fires extinguished 1.0 Mean-0.87 ' manually before critical damage occurred. [Q2TC) , . _ . . _ . . . . , . ._ .. _ . , _ . . . - ~-.__..;.. _. . _ -.. _ _ . . _ . . _
' Table 2.2-10 (continued)
Variable ~ Range Distribution Correlation Correlation With Descrintion XA-CBLVT 0.50 Max. Entropy None Fraction of fires extinguished 0.90 Mean-0.70 . automatically before critical damage occurred. [QAUTO] HE-CBLVT 0.0044 . Max. Entropy None Probability the operators fail to 0.44 Mean-0.044 obtain HPI from Unit 2 to prevent
'RCP seal failure. [R20P]
IE-CNIRM 1.2E-6 Iognormal None Initiating event - fire in the 0.0074- Mean-0.0011 control room. [ CONTROL ROOM] AR-CNTRM 2.4E-4 Max Entropy None Area ratio of benchboard 1-1 to u 0.0011 Mean-4.8E-4 total cabinet area in the control L o room. -[FA3] HE-CNTRM 0.0074. Max. Entropy None Probability the operators fail to 0.74 ' Mean-0.074 from the auxiliary shutdown panel. [R3OP] IE-ESWGR 0.027- Iognormal None Initiating event - fire in 0.16 Mean-0.064 emergency switchgear room. [ ELECTRICAL SWITCHGEAR ROOM] ARS-ESWG. 0.020 Max. Entropy None Area ratio in emergency switchgear 0.099 Mean - 0.039 room for a small fire where critical damage occurred. [FA4] SRS-ESVG 0.33 Max. Entropy .None Severity ratio for a small fire - 0.81 Mean-0.70 based on generic combustible fuel loading. [FS4] ARL-ESVG 10.051 Max. Entropy None Area ratio in emergency switchgear 0.24 Mean-0.10 room for a large fire where critical damage occurred. [FA5]
- . - . . ~ . , . . . , - . , , . , . . . ,. - - - . . . . . . - . - , , - - .- _ _ ~ - ~ . _ - . - _ - - - - - -
Table 2.2-10 (continued) Correlation Correlation With Description Variable Rance Distribution SRL-ESWG 0.19 Max. Entropy None Severity ratio for a large fire 0.67 Mean-0.30 generic combustible fuel loading. [FS5} Fraction of fires extinguished FX-ESWGR 0.60 Max. Entropy None 1.0 Mean-0.37 manually before critical damage occurred. [Q4TG] IE-CPSWP lognormal None Initiating event - fire in charging Mean-0.0037 pump service water pump room. [ CHARGING PUMP SERVICE WATER PUMP ROOM} None Fraction of fires extinguished FX-CPSVP 0.60 Max. Entropy
~ 1.0 Mean-0.87 manually before critical damage occurred. [Q5TG]
i 2.3 Descrietion of the Accident Progression Event Tree This section describes the APET that is used to perform the accident progression analysis for Surry. The APET itself forms .: hich le'.el model
-of~the accident progression. The APET is too large to be drawn out in a figure as cmaller event trees usually are. Instead, the APET exists only as a coroputer input file. The APET is evaluated by the code EVNTRE, which ,
is described elsewhere.1 ' The APET is not meant to be a substitute for detailed, mechanistic codes such as the STCP, CONTAIN, MELCOR, and MAAP. Rather, it is an integrating i framework for synthesizing the results of these codes together with expert judgment on the strengths and weaknesses of the ' codes. The detailed, mechanistic codes require too much computer time to be run for all the possible accident progression paths. Therefore, the results from these codes are represented in the = Surry APET, which can - be evaluated very ; quickly. In this way, the full diversity of possible accident progressions - can be considered and tb uncertainty in the many phenomena involved can be < included. The following seccion contains a brie f . overview of the Surry APET. Details, including a complete listing of the APET and a discussion of each question, may be found in Appendix A of this volume. Section 2.3.2 is a - summary of how the APET was quantified, that is, how the many numerical values for branching ratios and parameters were derived. Section 2.3.3 presents the variables that ' were sampled in the accident progression analysis for Surry. 2.3.1 Overview of the APET The APET' for Surry _ considers the progression of _the accident from the time ' the TAF in the core is uncovered, which is assumed to be the onset of core i damage through the cci. Although the CCI may progress at ever slower rates for days,-- the end of this analysis has been arbitrarily set at 24 h. Ex. cept in very unusual accidents, almost all of the fission products that are going to be released from the containment will have been released by 24 h
.after the initiator. =While every effort has been made.to make this a general event tree that can~
be. applied to any large, dry containment, the cavity and sump arrangement ; Lat each plant is. unique, and this tree was constructed for.the Surry sump and cavity arrangement. - Therefore, some revision of this tree - will be .
. required for plants with other sump and' cavity configurations. The reactor cavity at Surry is not connected to the sump at.or near the basemat eleva-tion. The . sump; at Surry has a very large . capacity, so no matter how much ' '
of the water in the RCS and RWST escapes into the'surry containment through , a break, there vill be no water in the cavity unless the sprays operate. Table 2.3.1 lists the'71 questionn in the Surry APET. There are seven time periods in this APET. . To facilitate understanding of the AVET and
. referencing between questions, each branch of every question is assigned a mnemonic abbreviation. The mnemonic branch abbreviations for most branches i 2.32 i
start with a character er characters which indicate the time period of the question. The time periods used in the Surry APET, and their abbreviations, are: B Initial Questions 1 through 14 determine the conditions at the beginning of the accident.. E -Early Questions 15 through 31 concern the progression of the accident from the UTAF to just before VB. ! Questions 14 through 18 concern events or actions which may depressurize the RCS before breach. The possibility that core degradation may be arrested and VB prevented is considered in Question 23. ; I Intermediate . Questions 32 through 43 determine the progression of , the accident immediately before, at, and immediately af ter VB, including the possibility of containment failure at, or immediately after, VB. : 12 Late . Intermediate Question 44 determines the status of the sprays shortly after VB, during the RCS release.
'L Late- Questions 45 through $5 determine - the progression of the accident during the CCI.
L2 Very Late Questions 56 through 64 determine the progression of the accident in the period following CCI, including the possibility of containment failure due to hydrogen combustion. F Final Questions 65 through 71 determine the final status of . the containment. [ The clock time for each period varies depending upon the type of accident being modeled. The Surry APET does . not contain any questions to resolve core vulnerable sequences, which are accidents that have failure of containment heat , l- removal.. only. The continual deposition of decay heat'in the containment by > operation of . the ECCS in the recirculation mode is predicted to lead to - eventual containment failure in many hours or a few days. Containment failure, in tu rn ,- may lead to ECCS failure. The Surry PDSs with
. frequencies exceeding 1.0E 7/ year . did not contain any accidents of this type.
l Although the fan coolers at Surry are not safety grade, and the . accident progression analysis does not give any credit for their operation (although the accident frequency analysis does), this APET contains fan cooler questions. This is done to make this APET as applicable as possible to PWRs with large, dry containments that do have safety grade fan coolers. ' 2.33 e -
Centainment failure due to hydrogen combustion in the period before vessel failure is not considered in the Surry APET. This possibility was included in ihe event tree used in the analysis performed for the previous draft of NUREA 1150, and no containment failures before vossel breach due to defla-grati ns were observed. Due to the size of the Surry containment, a con-siderable amount of hydrogen must be produced and mixed into the contain-ment atmosphere to reach the minimum concentt stion for deflagrations. To reach the lower limit for deflagration in tt e absence of steam, hydrogen from oxidizing about one third of the zircon!um in the Surry core must be released from the RCS, If the atmosphere is half steam, hydrogen from oxidizing about two thirds of the zirconium in the Surry core must be released from the RCS to reach the lower limit for deflagration. To cause a deflagration or a detonation capable of failing the Surry, the hydrogen must accumulate in the containment until concentrations well above the lower limit for deflagration are reached. If electric power is available during this period, the sprays will keep the steam concentration low and sparks from electrical equipment will cause ignition near the lower defla-grable limit. Thus, large hydrogen concentrations will not occur. If electric power is not available during this period, the sprays will not operate and the containment is likely to be inerted by the high steam concentration. Furthermore, in many of the accidents in which power is unavailabic (SBos), the RCS remains at high pressure and much of the hydrogen produced remains within the RCS until the vessel fails. Hydrogen deflagrations are considered at vessel breach and af ter vessel breach in this analysis. The pressure rise at vessel breach was determined by a group of experts, the Containment Loads panel. This pressure rise includes contributions from hydrogen combustion, RCS blowdown, direct containment heating, and ex vessel steam explosions. The panel did not provide details about how much hydrogen is produced before vessel breach and how much is burned at vessel breach, so the amount of hydrogen remaining af ter vessel breach can only be approximated. If the containment does not fail at vessel breach, the APET determines if it fails later due to hydrogen deflagrations during or after CCI. Concentrations of hydrogen, oxygen, steam, and carbon dioxide in the containment atmosphere ace computed by means of a " User Function" for the periods after vessel breach. The user function also calculates the pressure rise due to the late hydrogen burns. In several places in the evaluation of the ApET, a User Function is called. This is a FORTRAN function subprogram which is executed at that point in the evaluation of the APET. The user function allows computations to be carried out which are too complex to be treated directly in the event tree. The user function itself is listed in Appendix A.2, and the manipulations performed by the user function at each question that utilizes the user function are described in Appendix A.1. The user function is called to: Determine containment failure and the mode of failure (Questions 43, 52, and 64); e Compute the hydrogen concentration in the containment; determine if the containment atmosphere is flammable; and, if it is, determine the total pressure in the containment from a hydrogen burn (Questions 51 and 63). 2.34 s -
J 2.3.2 Overview of the APET Ouantification This section summarizes the ways in which the questions in the Surry APET were quantified and discusses these methods briefly. A detailed discussion of each question, which includes comments on its quantification, may be found in Appendix A.1.1. In addition to the number and name of the quest' .6, Table 2.3 1 indicates if the question is sampled, and how the questi i is evaluated or quanti-fled. In the sampling column, an entry of DS indicates that the sampling is from a distribution provided by one of the expert panels, or from the electric power recovery distribution. The item sampled may be either the branching ratios or the parameter defined at that question. For questions which are sampled and which were quantified internally, the entry Z0 in the sampling column indicates that the question was sampled zero one, and the entry SF means the questions were sampled with split fractions. The difference may be illustrated by a simple example. Consider a question that has two branches, and a uniform distribution from 0.0 to 1.0 for the probabi:ity for the first branch, if the sampling is zero one, in half the observations, the probability for the first branch will be 1.0, and in the other half of the observations it will he 0.0. If the sampling is split fraction, the probability for the first branch for each observation is a random fractional value between 0.0 and 1.0. The average over all the fractions in the sample is 0.50. The implications of 20 or SF sampling are discussed in the methodology volume (Volume 1). If the sampling column is blank, the branching ratios for that question, and the parameter values defined in that question, if any, are fixed. The [ branchin6 ratios of the PDS questions change to indicate which PDS is being considered. Some of the branching ratios depend on the relative frequency of the PDSs which make up the PDS group being considered. These branching ratios change for every sample observation, but may do so for some PDS groups and not for others. If the branching ratios change from observation-to observation for any one of the seven PDS groups. SF is placed in the sampling column for the PDS questions. The abbreviations in the quantification column of the Table 2.3 1 are given below, with the number of questions which have that type of quantification. Number of Tyne of Ouantification Ouestions Comments PDS 11 Determined by the PDS l AcFrqAn 1 Determined in the accident frequency I analysis Other 4 See notes 1 through 4 Internal 17 Quantified internally in this analysis 2.35 l
Summary 17 The branch taken at this question follows directly from the branches ; taken at previous questions ROSP 3 The probability of the recovery of offsite power is determined by distributions derived from the electric power recovery data for this plant i UFUN Str. 3 Calculated in the User Function. l using distributions from the , Structural Expert Panel J UFUN Int. 2 Calculated in the User Function, using an adiabatic pressure rise calculation determined internally In Vessel 5 Distributions from the In Vessel Expert Panel Loads 2 Distributions from the Containment Londo F.xpert Panel Struct. 1 Distribution from the Structural Expert Panel N.A. 5 Fan cooler.. questions not applicable to Surry In some cases, a question may have more - than one function, so the entry-' i under Quantification in Table 2.3.1 can be only indicative. 'For example, ; Questions 43, 52, and 64 are listed as .being quantified by the user func. ! tion. based on distributions generated by the Structural' Response Expert Panel.. The. actual situation is this: a portion of the user function is evaluated which determines whether the containment fails using the load pressure and the failure pressure. The load pressure is determined in Questions 39 and 40 based on aggregate distributions from the Containment Loads Expert Panel. The containment failure pressure is determined in I Question 42 from the aggregate - distribution from the Structural Response Expert Panel. If = the failure pressure is lower than the _ load pressure, then the containment fatis and the mode of failure is determined using the random number defined in Question 42 and a table of conditional failure mode probabilities contained in the user function. This table was also i generated 'by the Structural Response Expert ' Panel. The sampling is indicated to be zero one because one of the four branches of . these questions always has a probability of 1.0, and'the other three always have a probability of 0.0 L t 2.36
Table 2.3-1 Questions in the Surry APET Ouestion Number Ouestion Sampline Ouantification
- 1. Size & location of RCS break when the core uncovers? SF PDS
- 2. Has the reaction been brought under control? SF PDS
.3. For SGTR, are the secondary system SRVs stuck open? SF PDS 4 Status of ECCS7 SF PDS
- 5. RCS depressurization by the operators? SF' PDS
- 6. Status of sprays? SF PDS
- 7. Status of fan coolers? N.A.
- 8. Status of ac power? PDS
- 9. RUST injected into containment? SF PDS
~
w 10. ' Heat removal from the steam generators? SF PDS w 11. Did the operators depressurize the secondary SF PDS before the core uncovers?
- 12. Cooling for RCP seals? SF PDS
- 13. Initial containment condition? AcFrqAn ,
- 14. Event V - break location' under water? ' SF Note 1
- 15. RCS pressure'at the start of core degradation? Summary *
- 16. Do the PORVs stick open? SF Note 2
- 17. Temperature-induced RCP seal failure? Z0 Note 3
- 18. Is the RCS depressurized before breach by opening Internal the pressurizer PORVs?
- 19. Temperature-induced SGTR7 DS In-Vessel
- 20. Temperature-induced hot leg or surge line break? DS In-Vessel
- 21. Is ac power available early7 SF ROSP
- 22. Rate of blowdown to containment? Summary
- 23. Vessel pressure just before vessel breach? 20 Internal
- 24. Is core damage arrested? No vessel breach 7 SF Internal
- 25. Early sprays? Summary i
- 26. Early fan coolers? N.A.
- 27. Early containment heat removal? Summary
Table 2.3-1 (continued) Ouestion Number Ouestion Sampline Ouantification
- 28. Baseline containment pressure before VB? Internal
- 29. Time of accumulator discharge? Summary
~.,0 . Fraction of zirconium oxidized in-vessel during P In-Vessel core degradation?
- 31. Amount of zirconium oxidized in-vessel during Summary core degradation?
- 32. Amount of water in the reactor cavity Summary at vessel breach?
- 33. Fraction of core released from the vessel at breach? P In-Vessel
- 34. Amount of core released from the vessel at breach? Summary
- 35. Does an alpha event fail both vessel & containment? SF Note 4 m
L 36. Type of vessel breach? 20 In-Vessel
- 37. Does the vessel become a " Rocket" and fail the cont.? Internal
- 38. Size cf hole in vessel (after ablation)? Z0 Internal
- 39. Total pressure rise at vessel breach? Lar;;e hole cases P Loads
- 40. Total pressure rise at vessel breach? Small hole cases P Loads
- 41. Does a significant ex-vessel steam explosion occur? Internal
- 42. Containment failure pressure? P Struct.
- 43. Containment failure and type of failure? ZO UR'N- S tr.
- 44. Sprays after vessel breach? Internal
- 45. Is ac power available late? SF ROSP
- 46. Late sprays? Summary
- 47. Late fan coolers? N.A.
- 48. Late cuatainment heat removal? Summary
- 49. How much h drogen burns at vessel breach? SF Internal
- 50. Does late ignition occur? Internal
- 51. Resulting pressure in containment? U R*N-Int.
- 52. Containment failure and type of failure? ZO UR'N- S t r .
- 53. Amount of core available for CCI? Summary 54 Is the debris bed in a coolable configuration? Internal
Table 2.3-1 (continued) Ouestion Number Ouestion Samplinz Ouantification
- 55. Does prompt CCI occur? Summary
- 56. Is ac power available very late? SF ROSP
- 57. Very late sprays? Summary
- 58. Very late fan coolers? N.A.
- 59. Very late containment heat removal? Summary
- 60. Does delayed CCI occur? Summary
- 61. How much hydrogen is produced during CCI? Internal
- 62. Does very late ignition occur? P Internal
- 63. Resulting pressure in containment? UFUN-Int.
64 Containment failure and type of failure? ZO URIN-S tr. m 65. Sprays after very late CF7 Internal L
- 66. Fan coolers after very late CF7 N.A.
- 67. Containment heat removal after very late CF Surmary
- 68. Eventual basemat melt-through (BMT)? Internal
- 69. Eventual overpressure failure of containment? Internal
- 70. BMT before overpressure failure? Internal
- 71. Final containment condition? Summary Notes to Table 2.3-1 Note 1. Whether the location of the break in the low pressure piping would be under water in Event V at the time the core was uncovered was determined by a special panel which considered only this problem for the draft version of this analysis. As there was no new information available, there was no reason to change the conclusions reached by this group. See the discussion of Question 14 in Appendix A.l.l.
Note 2. There is little or no data on the failure rate of PORVs when passing gases at temperatures considerably in excess of their design temperature. The quantification was arrived at by discussions between the accident frequency analyst and the plant analyst. See the discussion of Question 16 in Appendix A.l.l.
Note 3. In the accident frequency analysis, a special panel was convened to consider the issue of the failure of RCP seals. The quantification of this question is not as detailed as that done in the accident frequency analysis, but relies on the information produced by this panel. See the discussion of Question 17 in Appendix A.l.l. Note 4 The Alpha mode of vessel and containment failure was considered by the Steam Explosion Review Group a few years ago. The distribution used in this analysis is based on information contained in the report of this group. See the discussion of Question 35 in Appendix A.1.1. Key to Abbreviations in Table 2.3-1 AcFrqAn The quantification was performed as part of the Accident Frequency Analysis DS The branch probabilities are taken from a distribution; depending on the distribution the sampling may be SF or ZO. m Internal The quantification was performed at Sadia National Laboratories by the plant analyst with the i o assistance of other members of the laboratory staff. In-Vessel This question was quantified by sampling from an aggregate distribution provided by the Expert Panel on In-Vessel Issues. Loads This question was quantified by sampling from an aggregate distribution provided by the Expert Panel on Containment Loads. N.A. Not Applicable. P A parameter is determined by sampling from a distribution, in most cases an aggregate distribution from an expert panel. PDS The quantification follows directly from the definition of the Plant Damage State. ROSP This question was quantified by sampling from a distribution derived from the offsite power recovery data for the plant. SF Split Fraction sampling - the branch probabilities are real numbers between zero and one.
E- .. l Struct.. . This question was quantified by sampling from an' aggregate distribution provided by the Structural Response Expert Panel. Summary The, quantification for this question follows directly from the branches taken at preceding
- questions, - or the values of parameters defined in preceding questions.
UFUN-Str. This. question is. quantified by the execution of a part of the User Function, using distributiens from the Structural-Responsc Expert Panel. UFUN-Int. This question is quantified by the execution of a part of the User Function, using an adiabatic calculation for the pressure rise due to hydrogen combustion. 20 Zero-One sampling - the branch probabilities are either 0.0 or 1.0. H
2.3.3 Variables Samoled for the Accident Pronression Analysis i About 50 variables were sampled for the accident progression analysis. That is, every time the APET was evaluated by EVNTRE, the original values of about 50 variables were replaced with values selected for the particular observation under consideration. These values were selected by the LilS program from distributions that were defined before the APET was evaluated. Most of these distributions were determined by expert panels. Table 2.3 2 lists the variables in the APET which were sampled for the accident progression analysis. Some of them are branch fractions; the others are parameter values for use in calculations performed while the APET is being evaluated. In Table 2.3 2, the first column gives the variable abbreviation or identifier, and the question (and case if appropriate) in which the variable is used. The identifiers are limited to eight characters for the statistical package used to perform sensitivity studies. Where several variables are correlated, they are treated as one variable in the regression analysis (see section 3.3.1.10), but are different variables as far as the accident progression analysis and sampling process are concerned. Some of these variables in Table 2.3 2 have a number to distinguish the cases in the ninth position, which is dropped in the sensitivity analysis. For example, RCP SL P2 and RCP SL-P3 are treated as one variable, RCP-SL P, in the sensitivity analyses. The second column gives the range of the distribution for the variable. An entry of "0.0/1.0" in this column indicates that the variable took on fractional values between 0.0 and 1.0, An entry of "Zero/One" in this column indicates that the variable was sampled Zero one, i.e., it took on only the values 0.0 and 1.0. In each observation, one of these two values would be assigned. The third column in Table 2.3-2 indicates the type of distribution used. For uniform distributions from 0.0 to 1.0, the mean is obvious and so is
-not listed. Otherwise, the mean is given, if appropriate. The entry " Experts" for the distribution indicates that the distribution came from an expert panel and the entry " Internal" distribution indicates that the distribution was determined internally by the proj ect staff or others.
(None of the distributions obtained by aggregating the conclusions of experts can be described succinctly in words. Plots of the aggregate expert distributions are contained in Volume 2 of this report, A listing of the - input to the LilS program that contains many of these distributions in tabular form is given in Appendix E.) For Zero one variables, an indication of the probability of each state is given in this column. The fourth and fif th columns in Table 2.3 2 show whether the variable is correlated with any other. " Rank 1" indicates a rank correlation of 1.0. An "n" is used to indicate any integer. In the entry for RCP SL-P2, RCP-E'.-Pn in the " Correl. with" column indicates that RCP-SL P2 is correlated with RCP-SL P3 and RCP-SL P4. For further information on each of the variables listed in the table, see the detailed discussion of the indicated APET question in Appendix A. 2.42
Most of the variables listed in Table 2.3 2 need no further comment. The RCS pressure at VB variables, RCSPR-VB2 and RCSPR VB3 (Question 23), are sampled Zero One. The distribution column gives the fraction of the time each of the pressure ranges is chosen. Low is below 200 psia, Im indicates the intermediate pressure range, from 200 to 600 psia. The high pressure
-range extends from 600 to 2000 psia, but is nominally about 1000 to 1500 psia. Setpoint press,ure refers to the PORV and SRV settings, about 2500 psia.
RCP seal failure is considered both in the accident frequency analysis and in the accident progression analysis. The eight character code is RCP-SL F for RCP seal failures in the accident frequency analysis and RCP SL P for RCP seal failures in the accident progression analysis. These two variables should have been correlated with each other, but the ways in which seal failur's were treated in the two constituent analyses were so different that this was not feasible. Note that both the temperature induced (T 1) hot leg failure variable r., (Question 20) and all the fraction of zirconium oxidized variables (Question 30) are correlated with each other as the experts concluded that the oxidation of a lot of zirconium before VB would result in high tempera-tures, which in turn, would make hot leg or surge line failure more likely. This reasoning included the T-I SGTR as well as the hot leg break, and it was intended that variable TI SGTR would be correlated with TI 110 TLC and FR-ZROX. Due to an oversight, this correlation was omitted. As T I SGTRs were very infrequent, the omission of this correlacion was not significant. The type of vessel failure variables (Question 36) are samp1;d Zero One and the entries under " Distribution" indicate the probability of each type of vessel breach, llPME indicates ejection of the melt at high pressure through a hole that is small relative to the cross-section of the vessel. Btmlid indicates a gross failure of the entire bottom head of the vessel, and Pour indicates a slow release of the melt driven primarily by gravity. For the hole size (Question 38), large means greater than 0.4 m2 (nominally 2.0 m2) and small means smaller than 0.4 m2 (nominally 0.1 m2), For.the numerous pressure rise at VB variables (Questions 39 and 40), wet cavity means the cavity contains at least the accumulator water (depth about 4 f t) or that the cavity is full (depth about 14 f t) . The fraction of the core ejected at VB (Question 33) was placed into three groups in Question 34, liigh fraction ejected means greater than 40%, medium fraction ejected means between 20 and 40%, and low fraction ejected means less than 20%. The failure mode, as a function of failure pressure, was determined by the structural expert panel. The containment failure modo variable, CF MODE (Question 42), is only a random variable used to determine the failure mode. The method used to select the failure mode for each observation is explained in Volume 1, and the results of the expert panel on the failure pressure and failure mode for Surry may be found in Volume 2. Additional information is contained in the discussion of Question 42 in Appendix A of this volume. 1 2.43
s
-The final L variable in Table 2.3 2 (Questions 21, 45, ana $6) is used to ,
-select the probability .that offsite power will be recovered in a specified 1-time-interval'given that it was not recovered in a previous time interval.-
Distributions were developed for 12 cases, each with different start and "
' end - times ; corresponding - to' dif ferent classes of accidents. See the-discussions in' Appendix A'for the questions listed above. More detail on the inethods for determining the probability of offsite power recovery may be found in the methodology volume of this report and NUREC/CR 4550. The variable POWERREC defines a quantile for these distributions and the 6 associated' recovery probabilities are used in the' analysis.
i t i e t l l H .. 2.44
! M-Table 2.3-2 Variables Sampled in the Accident Togression Analysis for Internal Initiators Varieble Question Descrintion
& Case Range Distribution Correlation Correlated With V-UWATER- 0.70 Uniform None Probability that the break location 1.0 Mean 0.85 will be underwater when radioactive Ql4 releases begin, given Event V.
PORV-OPN 0.0 Uniform None Probability that at least one PZR PORV or RCS SRV sticks open, given Q16 C1 1.0 that the RCS is intact and the PORVs or SRVs are cycling. RCP-SL-P2 Zero Fail 0.71 Rank 1 RCP-SL-Pn Probability of a T-I failure of the One RCP seals given core damage, RCS at
, Q17 C2 setpoint pressure, and no cooling w
for the RCP seals. RCP-SL-P3 Zero- Fail 0.65 Rank 1 RCP-SL-Pn Probability of a T-I failure of the One RCP seals given core damage, RCS at Q17 C3 high pressure, and no cooling for the RCP seals. RCP-SL-P4 Zero Fail 0.60 Rank 1 RCP-SL-Pn Probability of a T-I failure of the One RCP seals, given core damage, RCS Q17 C4 at intermediate or low pressure, and no cooling for the RCP seals. TI-SGTR 0.0 Experts None Probability of a T-I SGTR, given 0.12 Mean- 0.014 core damage, RCS at setpoint Q19 C1 pressure, and no cooling for the steam generators. TI-HOTLG1 0.0 Experts Rank 1 TI-HOTIS2 Probability of a T-I failure of the FR-ZROXn hot leg or surge line, given core Q20 C1 1.0 Mean-0.77 damage, AFVS failure, and the RCS intact at setpoint pressure. I __
Table 2.3-2 (continued) Variable Question Description
& Case- stange ' Distribution Correlation Correlated With TI-HOTJC2 0.0 Experts Rank 1 TI-HOTM1 Probability of a T-I failure of the 1.0 Mean-0.035 FR-ZROXn hot leg or surge line, given core .Q20 C2 damage, AFUS failure, and an S3 break in the RCS.
RCSPR-VB2 Zero 0.20 Low Rank 1- RCSPR-VB3 RCS pressure just before vessel One 0.80 Im breach, given an initiating or Q23 C2 induced S2 break. RCSPR-VB3 Zero 0.33 Iow Rank 1 RCSPR-VB2 RCS pressure just before vessel One 0.34 Im breach, given an initiating or , Q23 C3 o 0.33 High induced S3 break. CDARRESTS 0.80 Uniform Rank 1 CDARRESTn Probability that core damage can be arrested before VB, given the 1.0 Q24 C5 conditions of case 5. (Also used ; for case 8.) CDARREST6 0.0 Quadratic Rank 1 CDARRESTn Probability that core damage can be arrested before VB , given the Q24 C6 1.0 Mean-0.67 ' conditions of case 6. CDARREST7 0.0 Uniform Rank 1 CDARRESTn Probability that core damage can be 1.0 arrested before VB, given the Q24 C7 conditions of case 7. Experts Rank 1 TI-HOTMn Fraction of equivalent core FR-ZROX1 0.0 FR-ZROXn zirconium oxidized in-vessel given Q30 Cl 1.3 Mean-0.44 that the RCS is at setpoint pressure and the accumulators discharge before or after core melt.
Tabla 2.3-2 (continued)- Variable Question
& Case Range Distribution Correlation Correlated With Description FR-ZROX2 0.0 Experts Rank 1 TI-HOTIEn Fraction of equivalent core zircon-Q30 C2 1.3' Mean-0.50 FR-ZROXn nium oxidized in-vessel given that the RCS is at setpoint pressure and the accumulators discharge during core melt.
FR-ZROX3 0.0 Experts Rank 1 TI-HOTLEn Fraction of equivalent core zirco-Q30 C3 0.80 Mean-0.32 FR-ZROXn nium oxidized in-vessel given that the RCS is at high pressure and the accumulators discharge before or y after core melt. b - " FR-ZROX4 0.0 Experts Rank 1 TI-HOTIEn Fraction of equivalent core zirco-Q30 C4 0.85 Mean-0.38 FR-ZROXn nium oxidized in-vessel given that the RCS is at high pressure and the accumulators discharge during core-melt. FR-ZROX5 0.0 Experts Rant 1 TI-HOTIEn Fraction of equivalent core zirco-Q30 C5 1.2 Mean - 0.48 FR-ZROXn nium oxidized in-vessel given that the RCS is at intermediate pressure and the accumulators discharge before or after core melt. FR-ZROX6 0.0 Experts Rank 1 TI-HOTIEn Fraction of equivalent core zirco-Q30 C6 1.2 Mean-0.52 FR-ZROXn nium oxidized in-vessel given that the RCS is at intermediate pressure and the accumulators discharge during core melt.
Table 2.3-2 (continued) Vari 6 1e Question
& Case Range Distribution Correlation Correlated With Descrintion FR-ZROX7 0.0 Experts Rank 1 TI-HOTLGn Fraction of equivalent core Q30 C7 1.2 Mean-0.45 ER-ZROXn zirconium oxidized given that the RCS is at low pressure and the accumulators discharge before core melt.
FR-HPME~ 0.0 Experts None Fraction of cors which participates Q33 0.60 Mean-0.30 in HPME at VB. VB-ALPHA 0.0 Experts None Probability that an Alpha mode CF. m Q35 C1 1.0 Mean .0091 occurs,given that the RCS is at low b pressure. (One tenth this value is utilized for case 2.) TYPE-VB1 Zero Experts Rank 1 TYPE-VB2 Type of VB given that the RCS is at Q36 C2 One HPME 0.79 setpoint pressure. BtmHd 0.08 Pour O.13 TYPE-VB2 Zero Experts Rank 1 TYPE-VB1 Type of VB given that the RCS is at Q36 C3 One HPME O.60 high pressure. (Also used for case BtmHd 0.27 4.) Pour 0.13 VBHOLSIZ Zero .0.1 Large None Size of the hole in the vessel Q38 Cl One 0.9 Small after ablation given high pressure melt ejection. PRISE-ID 0.0 Experts None Pressure rise at VB given that the Q39 C3 80 psi Mean-19 RCS is at low pressure or the mode of VB is Pour. . - . ,= .,- , .c - . - , - . . _ _ _ _ _ . _ _ _ _ _ _
Table 2.3-2 (continued) Variable Question. .
&' Case Range Distribution . Correlation correlated With Descrintion PRISE-VB1 6.5 Experts Rank 1 PRISE-VBn Pressure rise at VB given Im RCS
.Q39 C5 165 psi Mean-67 pressure,high fraction ej ected, large hole, wet cavity.
PRISE-VB2 5.0 Experts Rank 1 PRISE-VBn Pressure rise at VB given Im RCS Q39 C6 145 psi Mean-58 pressure, medium fract. ejected, large hole, wet cavity. PRISE-VB3 3.0 . Experts ' Rank 1 PRISE-VBn Pressure rise at VB given Im RCS Q39 C7 112 psi Mean-38 pressure, low fraction ejected, large hole, wet cavity. m b
- PRISE-VB4 11 Experts Rank 1 PRISE-VBn Pressure rise at VB given high or Q39 C8 185 psi Mean-90 setpoint RCS pressure, high fraction ej ected, large hole, dry
. cavity. PRISE-VB5 9.5: Experts ' Rank 1 PRISE-7Bn Pressure rise at VB given high or Q39 C9 166 psi Mean - 75.. setpoint RCS pressure, medium fraction ejected, large hole, dry cavity. PRISE-VB6 5.5 Experts Rank 1 PRISE-VBn Pressure rise at VB given high or Q39 C10 123 psi Mean-47 pressure, low fraction ej ected, large hole, dry cavity. PRISE-VB7 9.0 Experts Rank 1 PRISE-VBn Pressure rise at VB given high or , Q39 C11 180 psi Mean setpoint RCS pressure and a wet cavity or In pressure and a dry cavity, high fraction ej ec te d , large hole.
Table 2.3-2 (continued) Variable Question Descrintion
& Case Range Distribution Correlation Correlated With PRISE-VB8- 8.0 Experts Rank 1 PRISE-VBn Pressure rise at VB given high or Q39 C12 160 psi Mean=65 setpoint RCS pressure and a wet cavity or Im pressure and a ' dry cavity, medium fraction ejected, large hole.
PRISE-VB9 5.0 Experts Rank 1 PRISE-VBn Pressure rise at VB given high or Q39 C13 120 psi Mean-42 setpoint RCS pressure and a wet cavity or Im pressure and a . dry cavity, low fraction ejected, large hole.
?
U PRISE-VB10 5.5 Experts Rank 1 PRISE-VBn_ Pressure rise at VB given Im RCS 157 psi Mean-60 pressure, high fraction ejected, Q40 C2 small hole, wet cavity. PRISE-VBil 4.5 Experts Rank 1 PRISE-VBn Pressure rise at VB given Im RCS Q40 C3 135 psi Mean-49 pressure, medium fraction ejected small hole, wet cavity. PRISE-VB12 3.0 Experts Rank 1 PRISE-VBn Pressure rise at VB given Im RCS 107 psi Mean=34 pressure, low fraction ejected, Q40 C4 small hole, wet cavity. PRISE-VB13 8.0 Experts Rank 1 PRISE-VBn Pressure rise at VB given high or Q40 C5 160 psi Mean-75 setpoint RCS pressure, high fraction ejected, small hole, dry cavity. PRISE-VB14 7.0 Experts Rank 1 PRISE-VBn Pressure rise at VB given high or Q40 C6 140 psi Mean-61 setpoint RCS pressure, medium fraction ejected, small hole, dry cavity.
M-
~
Table 2.3-2 (continued) Variable Question Descriotion l
& Case Range ' Distribution Correlation Correlated With
. . Experts Rank 1 PRISE-VBn Pressure rise at .VB give. high or PRISE-VB15 4.4 108 psi Mean-40 setpoint RCS pressure, low fraction Q40 C7 ejected, small hole, dry cavity.
Experts Rank 1 PRISE-VBn Pressure rise at VB given high or PRISE-VB16 6.5 160 psi Mean-66 setpoint RCS pressure and a wet Q40 C8 cavity or Im pressure and a dry cavity. high fraction ejected, small-hole. Experts Rank 1 PRISE-VBn Pressure rise at VB given high or PRISE-VB17 6.0 y 136 psi Mean-55 setpoint RCS pressure and a wet Q40 C9 cavity or Im pressure and a dry ~ cavity, medium fraction ejected, small hole. 3.6 Experts Rank 1 PRISE-VBn Pressure rise at VB given high or PRISE-VB18 110 psi Mean-37 setpoint RCS pressure and a wet Q40 C10 cavity or Im pressure and a dry cavity, low fraction ejected, small hole. Containment failure pressure. CF-PRES 70 Experts None Q42 190 Mean-126 Psig CF-MODE 0.0 Uniform None Random number used to select 1.0 containment failure mode. Q42 0.20 None Ratio of the expected pressure rise HB-SCAN Internal to the adiabatic pressure rise for a Q50,.Q62 1.2 Mean-0.72 hydrogen burn resulting from rapid steam condensation. bil ..
- - . . s- .. .. ~-
l
- +; ,
d . ., s
=.,., ,
Table 2.3-2 (continued) Variable Question
& Case Range Distribution Correlation Correlated With Descrintion HB-SCAL 2 0.10 Internal None Ratio of the expected pressure rise to the adiabatic pressure rise for a Q62 1.0 Mean-0.53 hydrogen burn resulting from slow steam condensation.
POWERREC None Variable used te select the Q21, Q45, Q56 probability thac offsite power will be recovered in a specified time interval given that it was not recovered in a previous time interval. M l t I
E = 2.4 Descriotion of the Accident Progression Bins As each path through the APET is evaluated, the result of that evaluction [ is stored by assigning it to an accident progression bin. cribes the evaluation in enough detail that a source term (release of This bin des-radionuclides) can be calculated for it. The accident progression bins are the means by which information is passed from the accident progression analysis to the source term analysis. A bin is defined by specifying the , attribute or value for each of eleven characteristics or quantities which define a certain feature of the evaluation of the APET. Section 2,4,1 describes the eleven characteristics, and the values that each character-istic can assume. A more detailed description of the binner, discussing _ sach case in turn, is contained in Appendix A. l . 3. The binner itself, which is expressed as a computer input file, is listed in Appendix A.1.4, Section 2.4.2 contains a discussion of rebinning, a process that takes place between evaluating the APET (in which binning takes place) and the source term analysis. Section 2.4,3 describes a reduced set of binning characteristics, which is used in presenting the results of evaluating the L[ APET, 2.4.1 Descrintion of the Bin Characteristics The binning scheme for Surry uses eleven characteristics. That is, there are eleven types of information required to define a path through the APET. A bin is defined by specifying a letter for each of the eleven character-istics, where each letter for each characteristic has a meaning defined below. For a characteristic, the possible states are termed attributes. The Surry binning characteristics are: Characteristic Abbreviation Descriotion _ 1 CF-Time Time of containment failure
'2 Sprays Periods in which sprays operate 3 CCI Occurrence of CCIs 4 RCS Pres RCS pressure before vessel breach 5 VB Mode Mode of VB 6 SCTR Steam generator t u rupture 7 Amt-CCI Amount of core available for CCI 8 Zr-Ox Fraction of Zirconium oxidized in-vessel 9 HPME Fraction of the core in thr high pressure melt injection (llPME) 10 CF-Size Size or type of containment failure 11 RCS-Ilole Number of large holes in the RCS after VB 2.53 E.-
m.------------.i-------
Most of this information, organized in this manner, is needed by SURSOR to calculate the fission product source terms. Characteristic 5, Mode of Vessel Breach, is not used by SURSOR, but has been retained because it provides interesting output information about the APET outcome, or the paths taken through the APET. SURSOR obtains the information it needs concerning IIPME from Characteristic 9, Fraction of the Core in }{PME. The remainder of this section consists of a listing of each attribute for each characteristic, followed by a brief description of each characteris-tic, and finally an explanation of an example bin. 1 Characteristic 1: Containment Failure Time l l A V Dry Event V, Break Location not Submerged B - V-Wet Event V, Break Location Submerged C Early CF Containment Failure before VB D - CF-at-VB Containment Failure at. VB E - Late-CF Late Containment Failure (during the initial part of CCI, nominally a few hours after VB) F - VLate CF Very Late Containment Failure (during the latter part of CCI, nominally 8 to 12 hours after VB) C - Final CP Containment Failure in the Final Period (nominally about 24 h af ter VB) 11 - No-CF No Containment Failure Characteristic 2 - Sprays A - Sp Early The sprays operate only in the Early period. B - Sp-E+I The sprays operate only in the Early and Intermediate periods. l l C - Sp-E+1+L The sprays operate only in the Early, Intermediate, and Late periods. D - SpAlways The sprays Always operate during the periods of interest for fission product removal. E - Sp-Late The sprays operate only in the Late period. F - Sp L+VL The sprays operate only in the Late and Very Late periods. 2.54
C Sp VL The sprays operate only in the Very Late period. H - Sp Never The sprays Never operate during the accident. 1 - Sp Pinal The sprays operate only during the Final period, which is not of interest for fission product removal. Characteristic 3 - Core-concrete Interactions A Prmpt-Dry CCI takes place promptly following VB. There is' no overlying water pool to scrub the releases. B - PrmptShlw CCI takes place promptly following VB. There is a shallow (about 4.5 ft) overlying water pool to scrub the releases. C - No CCI CCI does not take place. D - PrmptDeep CCI takes place promptly following VB. There is a deep (about 14 ft) overlying water pool to acrub the releases. E - SD1yd Dry CCI takes place after a short delay. The debris bed is coolable, but the water in the cavity is not replenished. 'The delay is the time needed to boil off the accumulator water. P - LD1yd Dry CCI takes place after a long delay.- The debris bed is coolable, but the water in the cavity is not replenished. The delay is the time needed to boil off the water in a full cavity. Characteristic 4 - RCS Pressure before Vessel Breach A ---SSPr . System Setpoint Pressure (2500 psia) B - IliPr liigh Pressure (1000 to 2000 psia) C - ImPr Intermediate Pressure (?.00 to 1000 psia) D - LoPr Low Pressure (less than 200 psia) 2.55
Characteristic 5 - Mode of Vessel. Breach l A - VB l!PME High Pressure Melt Ejection occurs - direct heating always occurs to some extent. < B - VB-Pour The molten core Pours out of the vessel, driven primarily by the effects' of gravity. C VB BtmHd Gross failure of a large portion of the Bottom llead of the vessel occurs, perhaps as a result of a l circumferential failure. D - Alpha An Alpha mode failure occurs - resulting . in containment failure as well as vessel failure.
'E - Rocket' A Rocket mode failure occurs -
resulting in -; containment failure as well as vessel failure. i F - No VB No VB occurs. Characteristic 6 - Steam Generator Tube Rupture A'--SCTR- -A SCTR occurs. The SRVs on the secondary system are not stuck open. B- SCTR-SRVO A SGTR occurs. The SRVs on the secondary system , are stuck open. C - No SGTR- A SCTR'does not occur. 1 Characteristic 7 - Amount of Core not'in HPME available for CCI I l A - Lrg-CCI A CCI occurs and involves a Large Amount of the ', I Core (70 1004). i B - Med CCI A CC' occurs ' and involves ' a - Medium amount of the ; l Core (30-70%). ; 1 l C - Sml-CCI A CCI occurs and involves a Small amount of the' Core (0-304). r D - No-CCI No CCI occurs.
.i I
2.56
4 h Characteristic 8 - Zirconium Oxidation A - Lo-Zr0x A Low amount of the core Zirconium was Oxidized in the vessel before VB. The implies a range from 0 to 40% oxidized, with a nominal value of-25%. B lli-Zrox A liigh amount of the core Zirconium was Oxidized in , the vessel prior to vessel breach. This implies t that more than 40% of the Zirconium was oxidized, with a nominal value of 65%. r Characteristic 9 - High Pressure Melt Ejection (HPME) ' A - Hi HPME A liigh fraction (> 40%) of the core was ejected under pressure from the vessel at failure. B - Md HPHE A Moderate fraction (20-40%) of the core was ejected under pressure from the vessel at failure. C Lo llPME A Low fraction (< 20%) of the core was ej ected under pressure from the vessel at failure. a. D No HPME There was no HPME at vessel failure. Characteristic 10 - Containment Failure Size , 4 -A-- Cat-Rupt The containment failed by . catastrophic rupture, resulting in a very large hole and gross structural failure, t B Rupture The containment failed by the development of a large hole or rupture; nominal hole ~ size is' 7 f ta, C Lt.ak The " containment failed - by the development of a .s - small hole or a leak; ' nominal hole size is 0.~1 f ta, 3 o D'- BMT The contaltunent failed by BMT. E - Bypass Tho' containment was bypassed by event V or an SGTR. F - No CP The containment did not fall. b s 2.57
Characteristic 11 - Holes in the RCS A,- 1 Hole There is only One large llole in the RCS following VB, so there is no effective natural circulation through the RCS after breach. B Holes There are Two large lloles in the RCS following VB, so there will be effective natural circulation through the RCS after breach. Characteristic 1 primarily concerns the time of containment failure. There are eight attributes. Five of these attributes concern the time of con-tainment failure, two concern Event V, and one is for no containment failure. SCTRs are considered separately in Characteristic 6 since an SCTR can occur in addition to one of the modes of containment failure. SURSOR-does not. distinguish between Late- and Very Late containment failure, so these two attributes are combined in the rebinner. BMT and eventual overpressure . failure due to the inability to restore containment heat . removal in the days following the accident are the failures that occur in the Final period. Characteristic 2 concerns the _ periods in which the sprays operate. The division into the nine attributes is a straightforward sorting out of the various combinations of time periods. The final time period is of little consequence for the -. fission product release, but it must be included because there are cases where the sprays operate only in this period and, , fort each characteristic, the binner must have a location in which to place ' every outcome. As . SURSOR does. not distinguish between ' Sprays Never Operate' ,- Attribute - H, _and ' Sprays Operate only in the Final Period', Attribute 1,.these-two are combined in the rebinner for SURSOR. Characteristic ~3 concerns the CCIs. .There are six possibilities which cover the meaningful combinations of prompt CCI, delayed CCI, and_no CCI, with the amount of water in the cavity. The amount of water in the cavity may be. divided into three cases. If the cavity was dry at VB and the - accumulators- discharge before breach, the cavity is dry at the start of ,
-CCI. If.the cavity was dry at VB and the accumulators discharge at breach, the cavity will- be about one- quarter full. If the sprays operate before- i breach,; then the cavity will be full (14 f t . of water) . (At Surry, the cavity can hold about 12,400 f t3 of water. ) - The neutron shield. tank has-been -ignored in this analysis. It contains 1600 ft3 of water, and the walls are steel- 1. 5' in, thick (EPRI ' NP-4096, p. D-16). Although the supports may fail at or after vessel failure, there seems to be a good chance that-the> cylindrical tank (it surrounds the -vessel) would wedge in place. The cavity and in-core instrumentation room together have a floor area of- about 620- f 2 t , so the neutron- shield water would cover it to a ;
depth of.about 2.5 ft. I L Characteristic 4 concerns the pressure in the reactor vessel before vessel breach; there are four levels. The pressures shown in parentheses below I are approximate pressures just before VB. The RCS pressure dur.ing most of 2.58
the core degradation period may be less than this value except for SSPr where the reclosing of the PORVs will keep the system pressure at the setpoint value. Characteristic 5 concerns the mode of vessel breach; there are six possibilities, including no VB. Direct heating of the containmeut always occurs to some extent if there is HPME, so there is no simple way to distinguish whether direct containment heating occurs. Characteristic 6 concerns SGTR. There are only three possibilities: no SCTR, SGTR, and SGTR with the SRVs on the secondary system stuck open. SCTR is considered separately from the other containment failure modes since it can occur in addition to the other failure modes. That is, occurrence of an SCTR before VB does not preclude containment failure at VB or late containment failure. The SGTR creates a bypass of the containment which may have no removal mechanisms operating in the escape path, so it is important to treat it separately. Characteristic 7 concerns how much of the core not in HPME that is available to participate in the CCI. The fractions 0.30 and 0.70 divide the range into three portions. The fourth attribute is no CCI. As SURSOR subtracts out the fraction of the core involved in HPME, when HPME occurs the fraction of the core available for CCI is always set to Large. Characteristic 8 concerns the amount of the core zirconium which is oxidized in-vessel before vessel breach. There are two possible values for this characteristic: low and high. The demarcation point between the two ranges is 40%. Characteristic 9 concerns the amount of the core involved in HPME; there are four attributes. The possible range is divided into three portions by 20% and 40%. No HPME is the fourth attribute. Characteristic 10 concerns the size of the hole that results from containment failure or the type of containment failure. There are six attributes. The first three attributes concern failure of'the containment wall above ground. BMT results in a release from the containment below ground. As SURSOR does not distinguish Final period leaks from BMT, they are combined in the rebinner. SURSOR determines whether the containment was bypassed from Characteristic 1 (Event V) and Characteristic 6 (SCTR), so the Bypass attribute is combined with No CF in the rebinner. Characteristic 11 concerns the number of large holes in the RCS after breach. The experts on the Source Isrm Expert Panel who provided distributions for revolatilization from the RCS - surfaces af ter VB gave different distributions depending on whether an effective natural circulation flow would be set up within the vessel. A significant flow could be expected only if there were two large, effective holes in the RCS; for example the hole in the bottom head resulting from vessel failure and a large temperature-induced hole in the hot leg. SGTRs, failure of the RCP seals, and Event V's would not count as large effective holes since effective natural circulation through the RCS would not result in these cases. S3 size holes are not considered large enough to result in effective natural circulation after vessel breach. l 2.59
A typical bin might be CFADBCABDDB, which, using the information presented above, is: G Final CF Containment Failure in the Final Period F - Sp L+VL Sprays only in the Late and Very Late periods. A - Prmpt-Dry Prompt CCI, Dry cavity
-D - LoPr Low Pressure in the RCS at vessel breach B - VB-Pour Core material Poured out of the vessel at breach C - No-SGTR No Steam Cenerator Tube Rupture A - Lrg CCI A Large fraction of the core was available for CCI l B lii-ZrOx A liigh fraction of the Zirconium was Oxidized in-vessel D e No llPME No llPME D - BMT BMT B - 2 Iloles Two lleles in the RCS.
2.4.2 Robinning The binning scheme used for the evaluation of the APET does not exactly match the input information required by SURSOR. The additional information
'in the initial binning is kept because it provides a better record of the-outcomes of the APET evaluation. Therefore, there is a step between the evaluation of the APET and the evaluation of SURSOR known as "rebinning".
In the robinning, a- few attributes in some characteristics are combined because . there are no'significant differences between them for calculating the fission product releases. Characteristic 5, Mode of VB, is not used by SURSOR, but is not eliminated in the rebinning. The~ information SURSOR ; requires about itPME is obtained~from Characteristic 9.
-In the rebinning for Surry, there are no changes for Characteristics 3, 4, 5, 6, 7, 8, 9, and.11. -That is, for these eight characteristics, the-in-formation produced by the APET . is exactly that used by SURSOR. For Characteristic 1, the Late-CF and VLate-CF. Attributes E and F, are com-bined into VLate CF, Attribute E. Final ^i becomes Attribute F, and No-CF
.becomes Attribute G. For-Characteristic 2, the two final ^ attributes (H -
Sp-Nover,.and_I - Sp-Final) are combined into Attribute 11, Sp-Nonop, 'since whether the sprays operate in' the Final period does not affect the amount of;fisslon products released. For Characteristic 10, the third and(fourth attributes (C ' - Leak, and=D - BMT) are combined into Attribute C (Leak)'
. since SURSOR considers the radionuclides released from BMT to be the same.
as those released from a leak in this period. Also for Characteristic 10, the fifth and sixth attributes (E - Bypass and_F - No CF) are combined into-a new Attribute D (No-CF) since the containment pressure boundary is not failed by a bypass and the releases from the bypass events (V and SGTR) are treated separately in SURSOR.
-As rebinned,. the listing of each attribute for each characteristic is as
+
follows: Characteristic 1 - Containment Failure Time (Rebinned)
~A - V-Dry Event V, Break Location not Submerged B - V-Wet Event V, Break Location Submerged m 2.60
l l l C Early CF Containment Failure before VB l D CF at VB Containment Failure at VB E - VLate CP Late or Very Late Containment Failure (during CCI, nominally several hours after VB) F Final-CF Containment Failure in the Final Period (nominally about 24 h after VB) G.- No CF No Containment Failure I Characteristic 2 - Sprays (Rebinned) A Sp Early The sprays operate only in the Early period. B Sp-E+1 The sprays operate only in the Early and Intermediate.- periods. C - Sp-E+I+L The sprays operate only in the Early, Intermediate, and Late periods. f
'D SpAlways The sprays Always operate during the periods of
-interest for fission product removal.
E' Sp-Late The sprays operate only in the Late period. ; F.- Sp L+VL- The sprays operate only in the Late and Very Late l periods, G Sp-VL The sprays operate only in the Very Late period. H , l H - Sp-Non0p -The sprays Never operate during the - accident, or operate only during the Final period, which is not of .4 interest for fission product. removal. _ Characteristic 3 - Core-Concrete Interactions. (Rebinned) _
-A - Prmpt Dry CCI takes place . promptly following VB. There is no overlying water pool to scrub the releases.
B - PrmptShlw CCI takes place promptly following VB, There is a shallow (about 4.5 ft) overlying water pool to scrub
.the releases.
- C No-CCI CCI does not take place. !
- l. D - PrmptDeep CCI takes place promptly following'VB, There is a deep.
1 (about 14 ft) overlying water pool to scrub the , releases. ! l-L 2.61
E -'SD1yd Dry CCI takes place after a short delay. The debris bed is coolable, but the water in the cavity is not replen-ished. The delay is the time needed to boil off the accumulator water. F - LD1yd-Dry CC1 takes place after a long delay. The debris bed is coolable, but the water in the cavity is not replen-ished. The delay is the time needed to boil off the water in a full cavity. Characteristic 4 - RCS Pressure before Vessel Breach (Rebinned) A SSPr System Setpoint Pressure (2500 psia) B HiPr High Pressure (1000 to 2000 psia) C ImPr Intermediate Pressure (200 to 1000 psia) D - LoPr Low Pressure (less than 200 psia) Characteristic 5 - Mode of Vessel Breach (Rebinned) A - VB HPME HPME occurs - direct heating always occurs to some
+
extent. B - VB-Pour The molten core Pours out of the vessel, driven primarily.by the effects of gravity. C - VB BtmHd Gross failure of a large portion of the Bottom Head of the vessel occurs, perhaps as a result of a { circumferential failure. D - Alpha An Alpha modo failure occurs resulting in containment failure as well as vessel failure. E - Rocket A Rocket mode failure occurs.resulting in containment failure as well as vessel failure.
. F - No VB No Vessel Breach occurs.
Characteristic 6 - Steam Generator Tube Rupture (Rebinned) A - SGTR A SGTR occurs. The SRVs on the secondary system are not stuck open,
, B - SGTR SRVO A'SGTR occurs. The _ SRVs on the secondary system are stuck open.
C No-SCTR- A SCTR does not occur.
)
1 2.62
Characteristic 7 - Amount of Core not in HPME available for CCI (Rebinned) A Lrg-CCI A CCI occurs and involves a large Amount of the Core (70 100%). B - Med CCI A CCI occurs and involves a Medium amount of the Core (30 70%). C - Sml CCI A CCI occurs and involves a Small amotnt of the Core (0 30%), D - No CCI No CCI occurs.
- Charseteristic 8 - Zirconium oxidation (Rebinned) ;
A - Lo Zr0x A Low amount of the core Zirconium was oxidized in the vessel prior to vessel breach, The implies a range from 0 to 40% oxidized, with a nominal value of 25%, B -~Hi-Zrox A High amount of the core Zirconium was Oxidized in the vessel before VB. This implies that more than 40% of the zirconium was oxidized, with a nominal value of 65%, , Characteristic.9 - High Pressure Melt Ejection (HPNE) (Rebinned)
'A - Hi HPME A High fraction (> 40%) of the core was ejected under pressure from the vessel at failure.
B - Md HPME A Moderate fraction (20 to 40%) of the core was ejected under pressure from the vessel at failure. o L C - Lo HPME A Low fraction (< 20%) of the core was ejected under pressure from the vessel at failure.
'D No HPME There was no HPME at vessel failure.
L Characteristic 10 - Containment Failure Size (Rebinned) A - Cat Rupt .The containment failed by catastrophic rupture, resulting in a very large - hole and gross structural failure. B - Rupture The containment failed by .the - development of a large hole or rupture; nominal hole size is 7 fta, ; C - Leak The containment failed by the development of a leak (nominal size 0.1 f t 2) or BMT. D - No-CF The containment did not fail. It may have been bypassed. 2.63
e a Characteristic 11 - Holes in the RCS (Rebinned) A 1 1101e There is only One large llole in the RCS following vessel breach, so there is no effective natural circulation through the RCS after breach. B - 2 Holes There are Two large Holes in the RCS following vessel , breach, so there will be effective natural circulation I through the RCS after breach. 1 In the robinning process, bin CFADBCABDDB used as an example above becomes FFADBCABDCB since the rebinning affects first and tenth characteristics: P - Final-CP Containment Failure in the Final Period F - Sp L+VL Sprays only in the Late and Very Late periods,
.A - Prmpt Dry Prompt CCI, Dry cavity D - LoPr- Low Pressure in the RCS at vessel breach B - VB Pour Core material Poured out of the vessel at breach C - No SCTR No Steam Generator Tube Rupture A - Lrg CCI A Large fraction of the core was available for CCI j B - Ili-Zr0x A High fraction of the Zirconium was Oxidized in-vessel ao D - No HPME No HPME C-- Leak Leak.(includes BMT)
B .2-Holes Two lloles in the RCS, 2.4.3 Summary Bins for Presentation
-For presentation purposes in NUREC-1150, a ~ set of " summary" bins has been adopted. Instead of the 11 characteristics and thousands of possible bins that describe the evaluation of the APET.in detail, the summary bins place
;the outcomes cf the evaluation of the APET into a few, very general groups.
'The seven summary bins'for Surry are: ,
VB,'Early CF, Alpha VB, Early CF, RCS Pressure > 200 psia VB,:Early CF,-RCS Pressure < 200 psia VB,-Late CF ;
- Bypass VB, No CF H No VB.
I This - order is that used -in displays, It has containtant failure first, then Bypass, then no containment failure, and finally, no vessel breach. , Containment failure is divided into four subsets, which are listed roughly. p, in decreasing order of the severity of-the resulting release. s In assigning bins to one of these summary bins, however, the summary bins must be considered in the following order: t I 2.64
Bypass Alpha No VB Early CF l Late CF No CF. That is, if Bypass and Early CF both occur, the resulting bin is assigned to Bypass since Bypass occurs first in this list. The reason that the reduced bins must have a definite assignment priority is that all possible i outcomes do not fit neatly into the seven reduced bias. There are certain i combinations of events which can be put in different places in the reduced l bins and there are other combinations of events which do not fit well in
- i. any of the reduced bins. None of these combinations are very frequent occurrences, but they have to be put in one of the seren reduced bins. The principle determining the reduced bin is that the release path which results in the highest offsite risk should determine the reduced bin. Thus the reduced bins reflect the logic used by SURSOR in calculating the source L terms.
l l As an example, consider Event V followed by an Alpha mode failure of. the i vessel and containment. This results in Bypass and Early CF, Should this ' go in the Alpha reduced bin, or the Bypass reduced bin? By the priority list above, it is placed in the Bypass reduced bin. The reason is - that almost all of the fission products released from the core before VB will have-escaped.to the auxiliary building-through the bypass before VB.. Thus this_ path determines most of the risk. Although' SURSOR treats the CCI y release as if all of it escapes through the ruptured containment, the early release is more important for determining offsite risk. The placement in reduced bins of six other ambiguous combinations of events is discussed below. Combination 1: Event V and B-Leak. l The ; fission product release from Event V with an isolation failure at the ! start of the accidents - (as calculated by SURSOR) -is very similar cto the release from Event V without an isolation failure, and quito dissimilar to the releases from accidents with an isolation failures but no initial bypass of the containment. Therefore, this combination is placed in the-Bypass reduced bin. The relative' frequency of this combination is . small since the probability of an isolation failure at Surry'is 2E-4. Combination 2: Event V and CF-at-VB, i This combination is analogous to the situation in which Event V is followed by an Alpha mode failure of the containment just discussed, except that the containment falls at VB for other reasons. It is also placed in the Bypass-reduced bin. l l 2.65 l
Combination 3: SGTR and noVB. In this scenario, vessel failure is avoided but there may be considerable core damage, and the fission products from the degradation of the core have an escape path to-the environment through the secondary system. It is not possible in this ' analysis to determine how much core damage occurs before the arrest of the degradation process. For this combination cf events, SURSOR calculates a SGTR release assuming that the degradation proceeds to the point of VB. If the core degradation is arrested very late, this is probably a reasonable assumption. Thus the SGTR and noVB combination is placed in the Bypass reduced bin. This combination is very infrequent; the i three PDSs in the SGTR PDS group all have no ECCS operable, so the only PDS l with an initiatin6 SGTR that may have no VB is CLYY-YXY in the ATWS PDS group, which is less than 1% of the total mean core damage frequency. PDSs in which temperature induced SGTRs occur may result in this combination of 1 events, but temperature induced SGTRs are very unlikely. Combination 4: B-Leak and noVB The combination of an isolation failure, no bypass, and the arrest of core damage before VB is a difficult combination to place in a reduced bin. There is ' at least some release of fission products to the containment during the core - degradation process, either through the break or through the cycling PORVs. Some of this material will escape through the isolation failure. SURSOR calculates a release which is approximately half way 1 (logarithmically) between the release from an isolation failure that is followed by VB and the release fm an accident with no VB and no contain-ment failure. Although most of u frequency in the No VB reduced bin represents noVB & noCF, the isolation failure and no VB combination fits better.in the No VB reduced bin that it does in one.of the VB and Early CF reduced-bins. The relative frequency of th's combination is small since the probability of an isolation failure at Surry is 2E-4. Combination 5: SGTR and CF-at-VB SURSOR was designed to treat SGTRs in addition to other failures of the [ containment, so this combination of events poses no special problem for the ' source term calculation. As the SGTR largely determines the early release, and o the early release is more important than the late release, this combinations is placed in the Bypass reduced bin. An Alpha mode failure is also a containment failure at vessel breach, so a SGTR followed by an Alpha event is also placed in the Bypass reduced bin. l -Combination 6: SGTR and CF-Late I This1 combination of events is analogous.to the previous one. Here the SGTR completely. dominates the releases; there is no question that this s combination should be placed in the Bypass reduced bin. Thus, in assigning combinations of events in -the ApET to reduced bins, bypass failures (V and SGTR) take precedence no matter what else happens or doesn't happen. Alpha. mode failures take precedence over other failure modes at VB, and over isolation failures. No VB is above Early CF in the priority list, so isolations failures are placed in the No VB reduced bin. 2.66
CF at VB with no VB is not possible, so that combination will not arise. The seven reduced bins may now be defined as follows: Bypass Includes . Event V and SGTRs no matter what happens to the containment after the start of the accident; it also includes SCTRs that do not result in VB. Alpha Includes all accidents that have an Alpha mode failure of the vessel and the containment except those that follow Event V or an SCTR; it includes Alpha mode failures that follow isolation failures because the Alpha mode containment failure is of rupture size. No VB Includes all the accident progressions that avoid vessel failures except those which bypass the containment. Most of the bins placed in this reduced bin have no containment failure as well as no VB, bit it also includes bins in which the containment is not isolated at the start of the accident and the core is brought to a safe stable state before the vessel fails. ECF-illPr Implies Early CF with the kcS above 200 psia when the vessel fails. Early CF here means at or before VB, so it includes isolation failures and seismic containment failures at the start of the accident as well as containment failure at VB. It does not include bins in which containment failure at VB follows Event V or an SGTR, or Alpha mode failures. ECF LoPr Implies Early CF with the RCS below 200 psia when the vessel fails. It does not include bins in which containment failure at VB follows Event-V or an SCTR,-or Alpha mode failures. Late-CF Includes accidents - in which the containment was not failed or bypassed before the onset of CCI and in which the vessel failed. The failure mechanisms are hydrogen combustion during CCI,'BMT in several daya, or eventual overpressure due to the- failure to provide containment heat removal in the days following the accident. VB-NoCF Includes all' the accidents not in one of the . previous reduced bins. The vessel's lower head is penetrated by the core, but the-containment does not fail and is not bypassed. l 2.67
2.5 Results of the Accident Progression Analysis This section presents the results of evaluating the APET. As evaluating the APET produces a number of accident progression bins (APBs), the discussion is primarily in terms of APBs. Some intermediate results are also presented. Sensitivity analyses are discussed as well. Section 2.5.1 presents the results for the internal initiators. Section 2.5.2 discusses the sensitivity analyses run for the internal initiators. The accident progression analysis results for the fire initiators are presented in Section 2.5.3 and sensitivity analyses for fires are presented in Section 2.5.4. Seismic results are given in four sections. The basic results based on the LLNL hazard curve are presented in Section 2.5.5 and sensitivity analyses us!ng the LLNL hazard curve are presented in Section 2.5.6. The seismic res"' s based on the EPRI hazard curve are presented in Section 2.5.7 and sensitivity analyses utilizing the EPRI hazard curve are presented in Section 2.5.8. The tabler in this section contain only a very small portion of the output obtained by evaluating the APETs. Complete listings giving average bin conditional probabilities for each PDS group, and lijtings giving the bin probabilities for each PDS group for each observation are available on computer media by request. 2.5.1 Results for Internal Jnitiators 2.5.1.1 Results for PDS Groun 1: Slow SBO. This PDS grcup consists of accidents in which all ac power is lost in the plant, but the steam turbine. driven AFWS operates for several hours. The operation of this system keeps the core covered and cooled as long as there is no water loss from the RCS, DC power is available until the batteries deplete. When the + batteries deplete, control of the steam turbine-driven AFWS is lost and !t fails. .This PDS group contains six PDSs: two have the RCS intact at UTAF, two have failure of the RCP seals before UTAF, and two have stuck-open PORVs before UTAF. In four of the six PDSs, the operators depressurized the secondary system before UTAF, and in two PDSs they did not The PDSs in this group are listed in Table 2.2-2. Table 2.5-1 lists the five most probatle APBs for the PDS group, the five most probable APBs that have VB, and the five most probable APBs that have VB and early containment failure (CF). Most probable means most probable when the whole sample of 200 observations is considered; that is, the five most probable bins are the top five when ranked by mean probability condi-tional on the occurrence of the PDS group. In Table 2.5-1, the " Order" column gives the order of the bin when ranked by conditional probability. The " Prob." column lists mean APB probabilities conditional on the occur-rence of the PDS group. That is, this table shows the results averaged over the 200 observations that form the sample. If Bin A occurred with c probability of 0.005 for each observation, its probability would be 0.005 in Table 2.5-1, 2.68
Table 2.5-1 Results of the Accident Progression Analysis for Surry Internal Initiators--PDS Croup 1: Slow 5B0 ro. CF RCS' VB Amt Zr CF A4g Bin Prob
- Occur. Time Sorays CCI Pres. Mode CCI Q3 HERE Size Five hest Probable Bins **
1 HDCDFCDBDFB 0.171 121 No-CF Always No-CCI IoPr No-VB No-CCI Hi No No-CF 2 HDCDFCDADFB 0.145 113 No-CF Always No-CCI IoPr ~ No-VB No-CCI Io No No-CF 3 HDCDFCDADFA O.046_- 41 No-CF Always No-CCI IoPr No-VB No-CCI In No No-CF 4 HDCCFCDBDFA 0.040 38 No-CF Always- No-CCI ImPr No-VB No-CCI Hi No No-CF 5 HrADBCABDFA 0.038 33- No-CF L+VL PraDry loPr Four Large Hi No No-CF m Five Most Probable Bins that have VB" "os
- 5 HFADBCABDFA 0.038 33 No-CF L+VL PrmDry loPr Pour Large Hi No No-CF 10 HFADBCAADFB 0.033 104 No-CF L+VL PrmDry IoPr Four Large la No No-CF 14 HDCDBCDADFB 0.017 113 No-CF Always No-CCI loPr Four No-CCI Im No No-CF 15 HDCDBCDBDFB 0.017 121 .No-CF Always No-CCI loPr Pour No-CCI Hi No No-CF 16 HCADBCABDFB- 0.016 120 No-CF VL PrmDry IoPr Pour Large Hi No No-CF Five Most Probable Bins that have VB and Early CF" 64 DHADDCBADBB 0.0012 50 CFatVB Never PrmDty LoPr Alpha Medium 14 No Rupture 73 DHADDCBBDBB 0.0010 64 CFatVB Never PrmDry LoPr Alpha Medium Hi No Rupture 95 DFACACABACB O.0007 1 CFatVB L+VL PrmDry ImPr HPME Large Hi Hi Isak 145 DFAAACAAABA 0.0004 4 CFatVB L+VL PreDry SSPr HPME Iarge Io Hi Rupture 172 DFADBCAADCB 0.0003 1 CFatVB L+VL PrmDry LoPr Four Large la No leak Mean probability conditional on the occurrence of the PDS.
A listing of all bins, and a listing by observation are available on computer media.
i I If Bin B occurred with a probability of 1.00 for one observation and did ; not occur in the other 199 observations, its probability would also be 0.005. The column headed *No. Occur." gives the number of observations out l of the 200 in the sample in which this APB occurred with a nonzero condi- ] tional probability. ' 1 The remaining nine columns in Table 2.5 1 explain 9 of the 11 characteris- I tics in the APB indicator. The sixth characteristic. SGTR, has been : omitted since none of the 100 most probable bins for this PDS group had T.I SGTR. The last characteristic, RCS ilole, has also been omitted since it is , of less interest than the others. The abbreviations for each APB charac. teristic are explained in Section 2.4 above. The first part of Table 2.51 shows the first five bins when they are ranked in order by probability. Evaluation of the APET produced 1049 bins for the Slow $80 PDS Scoup. To capture 95% of the probability, 138 bins t are required. The four most probable bins have no VB and no CF. The five most probable bins capture 44% of the probability. The five most probable bins with VB all result in no CF, and all have the RCS at low pressure (less than 200 paia) at VB. Most of the APBs that resuit in VB and CF have ! BMT has the mode of CF. Note that each of the five most probable APBs and the five most probable APBs with VB occurred in at least one sixth'of the observations in the sample. This is not the case with the five most probable APBs with both VB and early CF. Two of these bins occurred in only one sample, and another occurred in only four. This indicates that few observations in the sample may be large contributors to CF at VB, and thus to risk. Two of the five most probable APBs with both VB and early CF are Alpha mode failures of the containment, two are llPME and DCil failures, and the other is a low pressure Pour. Early CF means CF before or at VB. CF before VB is improbable at Surry, almost all early CFs are CF at VB. The strenSth of the containment led to the conclusion that CF before VB due to hydrogen combustion was negligible. The fact that the containment'is maintained hiow atmospheric pressure during operation means that initial containment failures are unlikely for internal initiators. In this PDS group, the probability of recovering offsite electrical power in time to arrest the core degradation process and avoid VB is about 0.62. More information concerning the methods used for determining the probabi- ' lity of the arrest of core damage for each case may be found in Appendix A.3.3. Of the fraction of this PDS group which resulted in VB, most had the RCS at low pressure at VB. The fractions of this PDS group which are.in the four pressure ranges at UTAF and just before VB (if it occurs) are: At UTAF Just Before VB , SSPr (2500 psia) 0.54 0.06 liiPr (600 to 2000 psia) 0.13 0.10 ImPr (200 to 600 psia) 0.33 0,19 LoPr (< 200 psia) 0.00 0.65 2.70
The relative f requencies of the "T", "S",3 and "S" PDSs, in conjunction with whether the secondary system has been depressurized while the ATWS is I operating, result in about half the PDS group being at the PORV setpoint pressure when the core uncovers (Question 15). Just before VB , the situation is quite different (Question 23). Five mechanisms for depres-surizing the RCS are considered in the APET. Three of these are quite effective: RCP seal failures, PORVs sticking open, and tem'>erature induced hot leg (or surge line) failures. The result is that the probability of the accident continuing with the RCS pressure boundary intact from UTAF to VB is about 0.10. The determination of RCS pressure at VB is discussed further in Section 2.5.2.1. The mean probability of CF at VB for this PDS group is only 0.008. Note that the 0.992 probability of no CF includes the 62% of the group that had core damage arrest and no VB. The mean probabilities of late and very late CFs due to hydrogen burns are 0.007 and 0.002, respectively. The mean probability of basemat melt through (BMT) is about 0.07, 2,5.1.2 Results for PDS Croup 2 LOCAs. This PDS group consists of accidents initiated by a break in the RCS pressure boundary. Three of the PDSs have A size breaks, and three have S -breaks which are treated at A 3 breaks in this analysis. There is one S and one Sa PDS in this group. These PDSs result in core damage because one or more of the ECCS required to respond does not operate. Four of the eight PDSs in this group have the LPIS operating but not injecting at UTAF. The PDSs in this group are : listed in Table 2.2 2. Table 2.5 2 lists the ten most probable APBs for this PDS group and the five most probable APBs that have VB and early containment failure (CF). The five most probable APBs that have VB are included in the ten most probable bins. Most probable means most probable based on the entire sample. Evaluation of the APET produced 216 bins for the LOCA PDS group. To capture 95% of the probability, 31 bins are required. The ten most probabic bins capture 84% of the probability. Eight of the ten most 1 probable bins have no CF, and four of them have no VB as well. The mode of CF for the two bins with CF is BMT. Most of the AFBs that result in VB and CF have BMT have the mode of CF. Of the five most probable APBs with both VB and early CF, four are Alpha mode failures; the fifth is due to HPME and DCH. Only one of these APBs occurred in very few observations. In the LOCA PDS group, the probability of arresting the core degradation process and avoiding VB is about 0.35. For two of the PDSs, the LPIS is operating at UTAF and the break (A or S ) is large enough by itself to 3 depressurize the RCS to the poinc where the LPIS may inject. These are core damage situations because the success criteria require the accumula-tors (A break) or HPIS (S3 break) to function in addition to LPIS, and these systems fof'ed. For two other PDSs, the LPIS is operating at UTAF, but the initiating break (S2 or S )3 is not large enough to depressurize the RCS so the LPIS can inject. If a temperature induced failure occurs after UTAF that is large enough to depressurize the RCS , then LPIS operation is likely to prevent VB by halting core degradation. 2.71
Table 2.5-2 Results of the Accident Progression Analysis for Curry Internal Initiators--FDS Group 2: IDCAs No. CF RCS VB Amt Zr CF Order Bin Prob.* Occur. .line_ Sorays CCI Pres. Mode CCI QB HPME Size Ten Most Probable Bins ** 1 HDCDFCDADFB 0.168 113 No-CF .Always No-CCI LoPr No-VB No-CCI la No No-CF 2 HDCDBCDADFB 0.166 113 No-CF Always Ns-CCI loPr Four No-CCI lo No No-CF 3 HDCDFCDBDFB 0.136 121 No-CF Always .No-CCI lePr No-VB No-CCI HL No No-CF 4 HDCDBCDBDFB 0.131 121 No-CF Always No-CCI loPr Pour No-CCI Hi No No-CF 5 HDDDBCAADFB 0.081 113 No-CF Always PraDeep loPr Pour Large lo No No-CF 6 HDDDBCABDFB 0.064 121 No-CF Always PrmDeep LoPr Pour Large Hi No No-CF y 7 HDCDFCDADFA 0.028 41~ No-CF Always No-CCI lePr No-VB No-CCI In No No-CF L 8 CDDDBCAADDB 0.027 113' Final Always PrmDeep loPr Four Large In No BMT [ 9 CDDDBCABDDB 0.021 121 Final Always PraDeep IoPr Pour Large Hi No BMT 10 HDCDFCDBDFA 0.020 33 No-CF Always No-CCI loPr No-VB No-CCI Hi No No-CF Five Most Probable Bins that have VB and Early CF** 28 DAFDDCBBDBB- 0.0024 70- CFatVB Early IElyDry loPr Alpha Medium Hi No Rupture 37 DAFDCCBADBB 0.0018 65 CFatVB Early LD1yDry loPr Alpha Medium Lo No Rupture 69 DADDDCBBDBB 0.0004 52 CFatVB Early. PrmDeep loPr Alpha Medium Hi No Rupture 75 DDCBACDBACA' O.0003 1 CFatVB Always No-CCI HIPr HPME No-CCI Hi Hi Isak 77 DADDDCBADBB. 0.0004 46 CFatVB Early PrmDeep LoPr Alpha Medium Lo No Rupture Mean probability conditional on the occurrence of the PDS. A listing of all bins, and a listing by observation are available on computer media. l l l
, . , . - . . . . - _ , .. .~ , , . , , . . . , .
1 The fractions of tt.$ 1;0CA PDS group which are in the four pressure ranges , at UTAF and just before VB (if it occurs) are: l At UTAF Just Before VB SSPr (2500 psia) 0.00 0.00 , HiPr (600 2000 paia) 0.00 0.06 t ImPr (200 600 psia) 0.19 0.06 , LoPr (< 200 psia) 0.81 0.88 l The five most frequent PDSs in this group are " A" or "S " PDSs , so it is 3 not surprising that low pressure in the RCS is likely both at UTAF (Question 15) and at VB (Question 23). , The mean probability of CF at VB for this PDS groun is only 0.006. Note j that the 0.994 probability of no CF includes the 35% of the group that had core damage arrest and no VB. There are no late or very late CFs due to hydrogen burns for this PDS group. Electrical power is available all l
~
along, so ignition is expected shortly after a flammable concentration is reached. Burns at this concentration do not threaten the Surry contain-ment. The mean probability of BMT is about 0.06. 2.$.1.3 Results for PDS Groun 3 Fast SBO. This PDS group consists _of r accidents in which all ac power is lost in the plant and the steam turbine-driven AFWS fails at, or shortly after, the start of the accident. The , Fast SB0 PDS. group consists of only one PDS, TRRR RSR. Table 2.5 3 lists the five most probable APBs for the Fast SB0 PDS group, the ' five most probable APBs that have VB, and the five most probable APBs that have VB and early containment failure (CF).
- The first part of Table 2.5 3 shows the first five bins when they are
-ranked in order by probability. Evaluation of the APET produced 870 bins ,
for the Fast SB0 PDS group, of which 103 are required to capture 95% of the probability. The five most probable bins capture 45% of the probability. The five most probable bins capture 68% of the probability. They all have
- no - CF, and three of them have no VB as well. Three of the five most l probable bins that have VB have no CF; the other two have BMT. Most of the l
APBs that result in VB and CF have BMT as the mode of CF. Three the five i most probable APBs with both VB and early CF are due to an Alpha mode event ' and the other two are due to' HPME and DCH at VB. Two of these APBs occurred in only one observation in the sample, i In this PDS group, the probability of-recovering offsite electrical power in time' to arrest the -core degradation process and avoid VB is about 0.51. I More information concerning the methods used for determining the probabi- l lity of the arrest of core damage for each case may be found in Appendix I l~ A.3.3. 1 2.73
Table 2.5-3 Results of the Accident Progression Analysis for Surry Internal Initiators--FDS Croup 3: Fast 550 No. CF RCS VB Amt ,Zr CF Order Bin' Prob.* -Occur _ Time _ Sprays _.pCI Pres. Mode CCI 23 HPME 111t Five Most Probable Bins ** 1 HDCDFCDBDFB 0.185 121 No-CF Always No-CCI LoPr No-VB No-CC1 HL No No-CF-2 HFADBCABDFB 0.095 '120 No-CF L+VL PrmDry LoPr Four Large Hi No No-CF 3 HDCDFCDADFB 0.083 113 No-CF Always No-CCI loPr No-VB No-CCI lo No No-CF 4 HFADBCAADFB 0.047 104- No-CF L+VL. PraDry loPr Pour large In No No-CF 5 HDCCFCDBDFA 0.042 38 No-CF Always No-CCI ImPr No-VB No-CCI Hi No No-CF Five Most Probable Eins That Have VB** 2' HFADBCABDFBl 0.095 120 No-CF L+VL PreDry LoPr
- Pour Large Hi No No-CF 4- HFADBCAADFB 0.047 104 No-CF L+VL PrmDry loPr Four Large In No No-CF 7 CFADBCABDDB 0.032 120 L+VL Final PrmDry LoPr Four Large Hi No BMT 13 HFADBCAADFA 0.018 40 No-CF L+VL 14 PraDry- LoPr Four Large Lo No No-CF GFADBCAADDB- 0.016 104 Final L+VL PrmDry IoPr Pour Large le No BMT Five Most Probable Bins that have VB and Early CF**
45 DHADDCBBDBB 0.0027 64 CFatVB Never- PrmDry LoPr Alpha Medium Hi No Rupture 71 .DFACACABACB 0.0014 1 CFatVB L+VL PrmDry ImPr HPME Large Hi Hi Imak 111 DHADDCBADBB 50 _.0.0006 CFatVB. Never PrmDry IePr Alpha Medium In No Rupture 116 DFCBACDBBCA O.0005 1 CFatVB L+VL No-CCI HiPr HPME No-CCI Hi Md Leak 168 DAFDDCBBDBB 0.0003 70 CFatVB 'Early LDlyDry IePr Alpha Medium Hi No Rupture Mean probability conditional on the occurrence of the PDS.
** A listing of all bins, and a listing by observation are available on computer media. - _ . . . -. _ __ _ - . - . _ . . _ . _ -_ _ - _ . _ ., .., . , _ . _.__ ___ _ __ _ .__ _ _ _.___ _ _ _=
I i i Of the fraction of this PDS group which resulted in VB, most had the RCS at ! low pressure at VB. The fractions of this PDS group which are in the four - pressure ranges at UTAF and just before VB (if it occurs) are: At UTAF Just Before VB ! SSPr (2500 paia) 1.00 0.03 l HiPr (600 2000 psia) 0.00 0.12 ImPr (200 600 psia) 0.00 0.20 , LoPr (< 200 psia) 0.00 0.66 As the only PDS in this group has the RCS intact at UTAF, the RCS is at the PORV setpoint pressure at that time (Question 15). Just before VB (Ques-tion 23), the probability of being at SSPr is only about 0.03. As : discussed with regard to PDS Croup 1 three of the five depressurization ! mechanisms considered in the APET are _ quite effective: RCP seal failures, , PORVs sticking open, and temperature induced hot leg (or surge line). ! failures. The result is that the probability of the accident continuing with the RCS pressure boundary intact from UTAF to VB is fairly small. The ) determination .of RCS pressure at VB is discussed further in Sections ; 2.5.2.1 and 2.5.2.2. The mean probability of CF at VB for this PDS group is only 0.007. Note I that tho 0.993 probability of no CF includes the 51% of the group that had core damage arrest and no VB. .The mean probabilities of late and very late ' CFs due to hydrogen burns. are 0.017 and 0.0005, respectively. The mean probability of BMT is about 0.09, 2.5.1.4- Results for PDS Grono 4: Event V. This PDS group consists of accidents in which the check valves between the RCS and the LPIS fail, and then the LPIS piping, subjected to pressures much higher than those for which it was designed, also fails. This produces a path from the RCS to l the auxiliary building, bypassing the containment, and is known as Event V. ; Experts considering the break location in the' LPIS concluded that the > l- probability was 0.85 that it.would be low enough in the auxiliary building " that the water from the~ RCS and the RWST, escaping through the break, would-form a pool' covering the break by the time when core de8radation commenced. Table 2.5 4 lists the eight most probable APBs for the V PDS Croup. Evaluation of the APET produced 16 bins for this PDS group, of which eight
- are required to capture 95% of the probability. The four most probable bins capture 81% of.the probability and all of them have the break location under water, There- is no possibility of avoiding VB or CCI in this PDS group.
Due to the size of the containment bypass, containment failure is not of much interest. a 2.75 I
Table 2.5-4 Results of the Accident Progression Analysis for Surry Internal Initiators--FDS Group 4: Event V-No. CF RCS VB Amt Zr CF Order Bin Prob.* Occur. Time Sorays CCI Pres. Mode CCI 9x HPHE Size Eight Most Probable Bins ** 1 BHADBCAADEA 0.268 110 V-Vet Never PrmDry IoPr Four Large la No Bypass 2 BHADBCABDEA .- 0.212 88' .V-Wet Never PrmDry IoPr Four Large Hi No Bypass 3 ~ BHADBCAADDA 0.182 110 V-Vet . Never PrmDry ImPr Four large Lo No BMT 4 BHADBCABDDA 0.144 .88 V-Vet- Never PrmDry IoPr Pour Large Hi No BMT 5 AHADBCAADEA 0.049 110 V-Dry Never PrmDry LoPr Four Large Lo No Bypass 6 AHADBCABDEA 0.036 88- V Dry Never PrmDry IoPr Four Large Hi Nc Bypass'
,o c 7 AHADBCAADDA 0.033 110 V-Dry Never PrmDry loPr Pour Large Im No BMT ,o -8 AHADBCABDDA 0.024 88- V-Dry Never PrmDry lePr Pour Large Hi No BMT Mean probability conditional on the occurrence of the PDS.
A listing of all bins, and a listing by 6bservation are available on computer media. 1
.--ss 7w- T '9e.jy 6 7p * * -,i, g*-_g.w'-e' -
*.me- pey g ""C*iW F.e 9 -9 W6 -*d'2% M7 V esWg-gqrV =, M F "M-- FG
- ps-i wwwe q- -W eu -_ _ - _ ^- _____-_s__ __m---m---n-- --2-* -- _ - .-.d
2.5.1.5 Results for PDS Croun 5' Transients. This PDS group consists of accidents in which the RCS is intact but there is no way to remove heat from the core. The AFWS fails at the start of the accident; bleed and feed is ineffective because the llPIS fails or the PORVs cannot be opened. LPIS is available but the operators cannot open the PORVs or have failed to do so. The Transient PDS group consists of two PDSs, TBYY YNY and TLYY YNY. Table 2.5 5 lists the ten most probable APBs for the PDS group and the five most probable APBs that have VB and early CF. The five most probable APBs that have VB are included in the ten most probable bins. Most probable means most probable based on the entire sample. Evaluation of the APET produced 252 bins for the Transient PDS group, of which 21 are required to capture 95% of the probability. The ten most probable bins capture 89% of the probability. They all have no CF, and the three most probable APBs have no VB as well. Six of the ten most probabl.e bins have VB; three have Pour as the failure mode and three b ve llPi4E as the failure mode. Most of the APBs that result in VB and CF j have BMT has the mode of CF. Of the five most probable APBs with both VB and early CF, two are due to an Alpha mode event, two are due to llPME and DC11 at VB, and the fif th is due to gross failure of the lower head at the i PORV setpoint pressure. Three of the five APBs occurred in only one I observation out of 200. In this PDS group, the probability of a temperature induced failure of the RCS pressure boundary is quite high, almost 0.90. Since the LPIS is always operating for this PDS group, and itPIS is operating as well with a probability of 0.67, the probability of arresting the core degradation process and avoiding VB is high, about 0.77. More detail on the arrest of core damage may be found in Appendix A.3, The fractions of this PDS group that are in the four pressure ranges at UTAF and just before VB (if it occurs) are: At UTAF Just Before VB SSPr (2500 psia) 1.00 0.12 liiPr (600 2000 psia) 0.00 0.002 ImPr (200 600 paia) 0.00 0.09 LoPr (< 200 psia) 0.00 0.79 As both PDSs in this group have the RCS intact at UTAF, the RCS is at the j
~PORV setpoint pressure at that time (Question 15). Just before VB (Question 23), the probability of being at SSPr is only about 0.12. This i probability is higher than PDS Group 3 (Fast SBO) because RCP seal cooling 'is available, thus re Acring the failure of the pumps seals ineffective as a means of depressurization. The PORVs still function in their safety mode, so they may stick open even when hardware failures prevent their being opened from the control room. The two effective depressurization i mechanisms for this PDS group are the PORVs sticking open and the temperature induced hot leg (or surge line) failures. Deliberate opening '
of the PORVs by the operators is inef fective because they cannot open the . PORVs or have already failed to do so. Temperature-induced SGTRs are very ! unlikely according to the expert panel. The determination of RCS pressure at VB is discussed further in Sections 2.5.2.1 and 2.5.2.2. 2.77
g , i e a I ' Table 2.5-5 Results of the Accident Progression Analysis for Surry Internal Initiators--FDS Croup 5: Transients CF RCS VB Amt Zr CF No. Prob,* Occur. Time Sorays CCI Pres. Mode CCI Qs HPME Size Qrder Bin Ten Most Probable Bins ** No-CF Always No-CCI LoPr No-VB No-CCI Hi No No-CF 1 HDCDFCDBDFB 0.501 121 No-CF Always No-CC1 LoPr No-VB No-CCI Lo No No-CF 2 HDCDFCDADFB 0.207 113 No-CF Always No-CCI ImPr No-VB No-CCI Hi No No-CF 3 HDCCFCDBDFB 0.034 22 121 No-CF Always No-CCI LoPr Pour No-CCI Hi No No-CF 4 HDCDBCDBDFB 0.032 No-CF Always No-CCI SSPr HPME No-CCI Lo Lo No-CF 5 HDCAACDACFA 0.026 24 No-CF Always No-CCI SSPr HPME No-CCI Lo Md No-CF 6 HDCAACDABFA 0.025 30 No-CF Always No-CCI SSPr HPME No-CCI Lo Hi No-CF 7 HDCAACDAAFA 0.021 25 No-CF Always No-CCI ImPr No-VB No-CCI Lo No No-CF 8 HDCCFCDADFB 0.019 23 No-CF Always PrmDeep LoPr Pour Large Hi No No-CF om 9 HDDDBCABDFB P.015 121 No-CF Always No-CCI LoPr Pour No-CCI Lo No No-CF 10 HDCDBCDADFB 0.013 113 Five Most Probatte Bins that have VB and Early CF** CFatVB Early LD1yDry LoPr Alpha Medium Hi No Rupture 45 DAFDDCBBDBB 0.0006 70 CFatVB Always No-CCI SSPr HPME No-CCI Lo Hi Rupture 48 DDCAACDAABA 0.0005 1 CFatVB Always No-CCI SSPr BtmHd No-CCI Lo Md Leak 64 DDCACCDABCA 0.0002 1 CFatVB Always PrmDeep SSPr HPME Large Lo Hi Rupture 72 DDDAACAAABA 0.0001 1 CFatVB PraDeep IoPr Alpha Medium Hi No Rupture 75 DADDDCBBDBB 0.0001 52 Early Mean probability conditional on the occurrence of the PDS.
** A listing of all bins, and a listing by observation &re available on computer media.
_t ____ -
1 I o The mean probability of CF at VB for this PDS group is only 0.002. Note that the 0.998 probability of no CF includes the 0.77 of the group that had ! core damage arrest and no VB. There are no late or very late CFs due to hydrogen burns for this PDS group. Electrical power is available all 1 along, so ignition is expected shortly after a flammable concentration is reached. Burns at this concentration do not threaten the Surry { containment. The mean probability of BMT is about 0.013. 2.5.1.6 Besults for Pos croun 6: ATVS. This PDS group consists of l accidents in which neither control rod insertion nor boron injection bring l the reaction under control shortly after the start of the accident. The core continues to generate large amounts of heat and steam until the water level drops far enough below TAF that the loss of the neutron moderating effect of the liquid water is lost for a substantial portion of the core. - The ATWS PDS group consists of three PDSs. one with the RCS intact at UTAF, I one with an S 3 break, and one with an SGTR. In all three situations, the PORVs will be open at UTAF due to the rate of steam generation in the core. ; The LPIS is operating but not injecting in the RCS intact and SCTR PDSs. t Table 2.5 6 lists the ten most probable APBs for the PDS group and the five ' most probable APBs that have VB and early CF or bypass. The five most t probable APBs that have VB are included in the ten most probable bins. Most probable means most probable based on the entire sample. Evaluation of the APET produced 348 bins for' the ATWS PDS group, of which 32 are required to capture 95% of the probability. Table 2.5 6 differs from the preceding tables in that the sprays characteristic has been omitted and the SGTR characteristic included. All the APBs for this PDS group have sprays all.the time. The ten most probable bins capture 78% of the probability. Seven of these APBs have no failure or bypass of the containment, two have SGTRs, and one ' has BMT in the final period. Four of the ten most probable bins have no VB. The two bins that have SGTRs have no VB, due to the operation of the ' LPIS throughout the accident. There may be a significant release in this accident ~ since the core degradation may. not be arrested until- it is quite well advanced and a substantial portion of the fission products are ! released from the core. The last part of. Table 2.5-6 shows the five most probable APBs with VB and early CF or bypass. These APBs all have SGTR and no CF; two of them occurred in only a few observations. Based on the mean core damage frequencies, 0.077 of this PDS group has an SCTR initiator, and so have containment bypass at the start of the accident, The most probable bin with CF at VB is 38th in order with a probability of 0.0016; the CF is l due to an Alpha failure of both vessel and containment. In this PDS group, the mean probability of a-resting cora degrad: tion process and avoiding vessel breach is about 0.41. This comes from the operation of the LPIS following a temperature induced break in the RCS. The water from the RWST injected by the LPIS contains enough boron to shut down the reaction should the core be in a configuration where continued L reaction is possible. l 2.7Y
Table 2.5-6 Results of the Accident Progression Analysis for Surry Internal Initiators--PDS Group 6- ATUS No. CF .RCS VB Amt Zr CF DIder ' Bin Prob.* - Occur. Time Sorays CCI Pres s Mode CCI Qg HPME Size Ten Most Probable Bins ** 1 HDCDFCDBDFB 0.223 121- No-CF No-CCI LoPr No-VB No No-CCI Hi No No-CF 2 HDCDBCDBDFB 0.136 121 No-CF No-CCI IoPr Pour No No-CCI Hi No No-CF 3 HDCDFCDADFB 0.128- 113 No-CF No-CCI LaPr .No-VB No No-CCI la No No-CF 4 HDCDBCDADFB 0.084 113 No-CF No-CCI LoPr Four No No-CCI In No No-CF 5 HDDDBCABDFB 0.067 121 No-CF PrmDeep lePr Four No Large Hi No No-CF (1 HDDDBCAADFB .0.041 113 No-CF PraDeep LoPr Pour No Large In No No-CF-y 7 HDCDFADBDEB 0.035 120 No-CF No-CCI lePr No-VB SGTR No-CCI HL No Bypass a2 8 HDCCACDBCFB 0.024 7 No-CF No-CCI ImPr HPME No No-CCI Hi la No-CF o 9 GDDDBCABDDB 0.022 121 LoPr Pour Final PraDeep No Large Hi No BMT 10 HDCDFADADE3 0.020 101 No-CF - No-CCI LoPr No-VB SGTR No-CCI Im No Bypass Five Most Probable Bins that have VB and Early CF or Bypass ** r
't HDCDFADBDEB' O.035 120 No-CF No-CCI LoPr No-VB SGTR No-CCI Hi No Bypass 10 HDCDFADADEB 0.020 101 No-CF No-CCI LoPr No-VB SGTR No-CCI Lo No Bypass 32 HDCCBADEDEB O.0026 6 No-CF No-CCI ImPr Four SGTR No-CCI Hi No Bypass 34 HDCCAADBCEB 0.0024 7 No-CF No-CCI ImPr HPME SGTR No-CCI Hi Lo Bypass 35 HDCDBADBDEB 'O.0022 118 No-CF No-CCI LoPr Pour SGTR No-CCI Hi No Bypass Mean probability conditional on the occurrence of the PDS.
A listing of all bins, and a listing by observation are available on computer media. 4 v-, , - .-Nrn a- -- ,, e ,- e v --s-- --
<~- - ~s - - - - - - - - - - - - - - - - - - - - - ~ ~ - - - ' - - - - - - -
l l 1 i The fractions of this PDS group that are in the four pressure ranges at UTAF and just before VB (if it occurs) are: At UTAP Just Before VB i SSPr (2500 psia) 1.00 0.005 it!Pr (600 2000 psia) 0.00 0.002 ImPr (200 600 psia) 0.00 0.19 LoPr (< 200 psia) 0.00 0.80 : The RCS is at the PORV setpoint pressure at UTAF (Question 15) because the ! reaction has not been shut down and the steaming rate is high. Just before ! VB (Question 23), t.he probability of being at SSPr is only about 0.005. This probability is lower than in PDS Groups 1, 3, and 5 because _ the operators are allowed to deliberately open the PORVs in this PDS. In the human ' reliability analysis , it was judged that the operators would be too busy trying to bring the reaction under control before UTAF to consider opening-the PORVs, and the PORVs would be kept open by the escaping steam ; in any event. Thus the effective depressurization mechanisms for this PDS group are: the PORVs sticking open, temperature induced hot leg (or surge line) failures, and deliberate opening of the PORVs by the operators. Pump seal cooling is available in the one PDS where it would be effective (the "T" PDS where the RCS is intact), so failure of the pumps seals is ineffective as a means of depressurization for the ATWS PDS group. ; Temperature induced SGTRs are very unlikely according to the expert panel. The mean probability of CF at VB for this PDS group is only 0.003. Note that the 0.997 probability of no CF includes 0.41 that had core damage arrest and no VB. There are no late or very late CFs due to hydrogen burns
- for this PDS group. Electrical power is available all along, . so ignition is expected shortly af ter a flammable concentration is reached. Burns at this concentration do not threaten the. Surry containment. The mean
. probability of BMT is about 0.05. . 2.5.1.7 Results for PDS Groun 7* SGTRs. This PDS group consists of .! t accidents in which the initiating event is the rupture of a steam generator-tube. The reaction is shut down successfully. The SGTR PDS group includes two PDSs in which the RCS is depressurized using. the two unaffected SGs _[ according to procedures, and the SRVs on the main steam lit.es from the - affected SG do ' not . stick open. These accidenta, denoted "C" SGTRs, are indicated by "SCTR" in Table 2.5 7. The most frequent and the least frequent PDSs in~ the SGTR PDS group are accidents in which the RCS is not . depressurized according to procedures, and the SRVs on the main steam-lines from the af fected SG stick. open. These accidents, denoted "It" SGTRs, are , indicated by "SRV0" in Table 2.5 7. Like Table 2.5 6, Table 2,5 7 omits the sprays characteristic to show the SGTR characteristic. All the APBs for' this PDS group have sprays all the time. Most of the fif teen most probable observations occurred in one sixth of the observations, or fever. 2.81
Table 2.5-7 Results of the Accident Progression Analysis for Surry Internal Initiators--PDS Croup 7: SGTRs i No. CF RCS VB Amt Zr CF Order Bin Prob.* Occur. Time Sorays CCI Pres. Mode CCI HPME Size QK Fifteen Most. Probable Bins ** l 1 HHADBBAADEA 0.083 41- No-CF PraDry loPr Four SRVO Large Io No Bypass
-2 CHADEBAADDA 0.057 _ 41' Final PrmDry IoPr Pour SRVO Iarge la No BMT
! 3 0.052 25 No-CF PraDry HHADBBABDEA. loPr Four SRVO Large Hi No Bypass 4 HHACABABBEA 0.042 15 No-CF PrmDry ImPr HPME SRVO Iarge Hi Med Bypass 5 CHADBBABDDA' O.036 25 Final PrmDry IoPr Pour SRVO Large Hi No BMT 6 HHBBBBAADEA -0.026 16 No-CF PraShi
" HiPr Four SRVO Large lo No Bypass l- . 7 HHACABAAAEA 0.022 8 No-CF PrmDry ImPr E 8 HPME SRVO Large In Hi Bypass HHEBABAAAEA 0.022 9 No-CF PrmDry HiPr HPME SRVO Large Hi HL Bypass 9 HDCDFADADEB- 0.021 101 No-CF PraDry LoPr No-Vb SGTK No-CCl lo No Bypass 10 HDCDFADADEA 0.019- 41 No-CF PraDry LoPr No-VB SGTR No-CCI la No Bypass l 11 CHBBBBAADDA- 0.017 16 PraShi Final HiPr Pour SRVO Large Io No BMT 12 HHACABAACEA- 0.017 8 No-CF PrmDry ImPr HPME 13 SRVO Large Im Io Bypass HHEBABAABEA O.017 8 No-CF SDlyDry HiPr HPME SRVO Large 14 In Md Bypass HHEBABAACEA 0.017 9 No-CF SDlyDry HIPr HPME SRVO Large Lo 15 Io Bypass HHACABAABEA- 0.016 6 No-CF PraDry ImPr HPME SRVO Large Io Md Bypass ,
Mean probability conditional on the occurrence of the PDS.
** A listing of all bins, and a listing by observation are available on computer media.
i l
Evaluation of the APET produced 717 bins for the SGTR PDS group, of which . 178 are required to capture 95% of the probability. Since all the APBs for I this PDS group have bypass of the containment, Table 2.5 7 lists the 15 ' most probable APBs. They capture 46% of the probability. Only two of the 15 most probable bins have the SRVs reclosing; the other 13 bins result from the "H" SGTR accidents in which the secondary _ SRVs are stuck open. PDS llINY NXY has a higher frequency than the other three PDSs in this grou> combined. l In this PDS group, the probability of avoiding vessel breach is about 0.06. No ECCS are operable in the "II" PDSs. The LPIS is operating in the two "C" 1 PDSs, but there is an eficctive depressurization mechanism for only one of ) them. This mechanism is the deliberate opening of the PORVs. RCP seal cooling is available, so there are no seal failures. The RCS is not at the l PORV setpoint pressure, so there is no possibility of the PORVs sticking ) open, T 1 hot leg failures, or T-I SGTRs. The fractions of this PDS group which are in the four pressure ranges at i UTAF and-just before VB (if it occurs) are: At UTAF Just Before VB SSPr (2500 psia) 0.00 0.00 tiiPr (600 2000 psia) 1.00 0.30 i ImPr (200 600 psia) 0.00 0.32 LoPr (< 200 psia) 0.00 0.39 As all four PDS in this group have an S3 size SGTR at UTAF, the RCS pressure is in the liigh range at UTAF (Question 15). The four PDSs in this group are llINY NXY, - HINY-YXY, GLYY-YNY, and GLYY YXY. In llINY NXY and r .YXY the operators failed to follow procedures and open the PORVs b .: re UfAF, so no credit is given for their opening the PORVs af ter UTAF. In both HINY YXY and GLYY.YNY the PORVs are open at UTAF as the operators are or were attempting to cool the core by bleed and feed. In HINY YXY, the open PORVs serve only- to reduce the RCS pressure before VB as no ECCS are available. In GLYY YNY, the resulting pressure reduction in the RCS may allow the operating LPIS to inject water and arrest core damage beiore VB. PDSs llINY YXY and GLYY YNY are the least frequent of the four.PDSs in the SGTR group. _ As discussed in Section 2.5.2.1, it was estimeted that with an ~ 3S size break in the system, the low, intermediate, and high pressure ranges were equally likely at VB. The . probabilities of these three pressure ranges given above = vary somewhat from 0.33 due to the open PORVa_ just discussed. For the "H" SGTRs, CF at VB is not particularly significant for risk as the bulk of the fission products escapes through the containment bypass. The MCDF of IIINY NXY is about 75% of the MCDP of the SGTR PDS group. For the group as a whole, the mean probability of CF at VB is' about 0.03. There care no late or very late CFs due to_ hydrogen burns. The mean probability of BMT is about 0.27. 2.83 [ _
I k 2.5.1.8 Core Damage Arrect and Avoidance of VB. It is possible to arrest the core damage procecs and avoid VB if ECCS injection is restored before the core degradation process has gone too far. Recovery of injection is due to one of two events. In the LOSP accidents, recovery of i injection follows the restoration of offsite power. In other types of accidents, the ECCS is operating at UTAF but no injection is taking place because the RCS pressure is too high. Any break in the RCS pressure boundary that allows the RCS pressure to decrease to the point where the ECCS inj ec t is likely to arrest the core degradation process. The break may be an initiating break or a temperature induced break or other failure ' that occurs after UTAF. PDS ALYY YYY has the LPIS operating at UTAF. This is a core damage situation because the success criteria require the accumulators to operate in addition to the LPIS, and the accumulators fail. PDS SgLYY YYN also has the LPIS operating at UTAF; it is a core damage situation because the success criteria require the HPIS to operate in addition to the LPIS, and the llPIS fails. For both of these PDSs, the initiating break depressurizes the RCS sufficiently for the LPIS to inject. In PDS TLYY YNY, the LPIS is also operating but the RCS is_ intact at UTAF. In this situation, injection will commence only if one of the five depressurization means considered in this analysis operates, and if the RCS is depressurized to a low enough level. In PDS TBYY YNY. both LPIS and HPIS are operating, so it is not necessary that the RCS be depressurized to a low level. The five means of depressurizing the RCS after UTAF are: .
- 1. PORVs or SRVs stick open;
- 2. T-I RCP seal failure;
- 3. Deliberate opening of the PORVs by the operators; 4 T I SGTR; and
- 5. T 1 hot leg or surge line failure.
Figure 2.51 shows the probability of halting the degradation of the core before the lower head of the vessel fails, thereby achieving a safe stable ; state with the vessel intact. For the IASP collapsed PDS group, the distribution in Figure 2.5-1 reflects the distributtor._for offsite ac power recovery in the APET "early" period, To avoid a gap in the times for which . power recovery is considered, tia utart of the APr.r. "early" period is the end of the period for which recovery of offsite r.swer was considered in the - accident frequency analysis. This time is e.ominally the onset of core damage, but' for some PDSs this time precedN the current estimates of the onset of core damage (UTAF) by a signifkant amount. The end of the APET "early" period is the expected time of VB. The ' estimated core damage states that different times in this period were used to determine the probability of core damage arrest for each PDS involved, as explained in ., Appendix A.1.1 (see the discussions of Questions 21 and 24) and in Appendix A.3.3. For the ATWS. . Transients , and LOCAs , the distributions for core damage arrest show the combined effects of RCS depressurization that allows ECCS ! injection in those PDSs which have ECCS operating at UTAF. The probabiD ty l of core damage arrest is very high for Transients since one PDS in tt.* l group has LPIS operating and the other has both LPIS and HPIS operating. l l 2.84
SURRY l 00 W = mean 96th. m o median m---+ I th = percentile 0.76-
.6th. .
96t h. b# O
. 96th ,
96th. b l
% m. a.
hk D5
- \ -
t.t t. 38 3 l IO '"-4 m.".:t 1 I at 026- _i j .
,, l ,, ]
stk. l 0 00 , PDS Group 1.0SP ATWS Transients 1.0CA Dypass All Core Damage iYeq 2 8E- 05 1.4E-00 1 BE-06 010-06 340-00 4lE-05 l Figure 2.5 1. Probability of Core Damage Arrest Internal, t i As the probability of one or more of the depressurization mechanisms operating is high, so the probability of core damage arrest is high. 2.5.1.9 Early Containment Failure. For those accidents in which the containment is not bypassed, the offsite risk depends on the probability that the containment will fail before or at VB. There are four possibi-lities:
- 1. Pre existing containment leak; l 2. Isolation failure; l 3. CF before VB due to hydrogen combustion; and
- 4. CF at VB due to the events at VB.
As the Surry ' containment is maintained about 5 psia below the ambient atmospheric pressure during operation, an unsealed hatch or an open vent line would be quickly discovered since the vacuum pumps could not keep the , containment at the desired pressure. Thus the p robabili t' of a pre- , existing leak at Surry is negligibic. There are only two uormally open l 2.85 1'
lines into the Surry containment, one to the vacuum pump and one to the sump pump. The failure of these two lines to isolate on demand is quite low. Because the Surry containment was found to be qu i t.e strong by the structural experts who considered the issue, CF due to hydrogen burns before VB was not considered at Surry. It was estimated to be very unlikely that enough hydrogen would be generated in the vessel before VB to cause a pressure rise, when dispersed to the containment and ignited, that would threaten the Surry containment. This failure mode was included in the APET used for the analysis for Draf t NUREG 1150, and no CFs before VB were found. The main risk for non bypass accidents at Surry, then, comes from CF r.t VB. Almost all early CFs are CF at VB. As used in F16ure 2.5-2, early CF means CF before, at VB, or immediately following VB. Figure 2.5 2 shows the probability distribution for early CF at Surry. The probability is conditional on core damage. All the no VB probability, including a very small fraction that has isolation failures, 3 counted as no early CF (see Section 2.4.3). The conditional probability of early CF 1.IV . SURRY 1.E-1. esth. l voth. . esth. esth. I e l s w _. D 51.E-2. 'Sth* g E: W -* I u ,, , u4 q
.n y esth. ,8 as .
u.. . I gE u . q
- n. 51.E.3. -
1 1
*E 3 ~1 ec ,o 3 J .
_ m. -. -
- n. _I i l
@ gIE .
sth. .. _ g m. g oth. ** _. 6t h. I sin. 3 7 1.0-5. sin . I W - mean rn - rnedlen th a percentile
- 1. E- 0.
- - - - - - - - - - - - - - - - - - l pt e r n al I n it i at o rs - - - - - - - - - - - - - - - - -
PDS Group SD0 AT U S Transients LOCAs liypa ss All hre Core Darnage fYeq 080-0$ 14r-06 100-00 n10 00 3 4 0 - O 't 4 10-05 110-05 Figure 2.5 2. Probability of Early Containment Failure -Internal. 2.86
i t is particularly low for the Transient PDS group because the probability of core damage arrest is quite high. There is no histogram for the Bypass collapsed PDS group. When the containment function is bypassed by Event V or an SGTR, early CF ceases to be very innportant in determining the release of fission products and the offsite risk. Thus, the conditional probabili-ty of early CF was deliberately not plotted for the Bypass group. For , accidents other than Bypass, the mean conditional probability of early CF is on the order of 0.01. This reflects the strength of the Surry contain-ment reintive the loads expected at VB and the probability that the vessel does not fail. 2.5.1.10 Summarv. Figure 2.5 3 shows the mean distribution among the surmnary APSs for the summary PDS groups. Only mean values are shown, so Figure 2.5 3 gives no indication of the range of values encountered. The distribution for core damage arrest is shown in Figure 2.51, and the dis-tribution for early (at or before VB) failure of the containment is shown in Figure 2.5 2. Nonetheless, Figure 2.5 3 gives a good idea of the rela-tive likelihood of the possible results of the accident progression analysis. Except for the Bypass initiators, either no failure of the vessel (safe stable state) or no containment failure are by far the most
SUMMARY
SUMM ARY PDS GROUP ACCIDENT * * " C" ' D* "' ' " " "i " * "* * )
'""~"""""'"~"""*'"*"""""""~"""
PROGRESSION tOSP ATFS Tr a nsiente (DCAs Dypass All Fire DIN GROUP ( 2 at-Os) ( i 4t-Oc) ( i er-Oo) ( e it-Oc) ( 3 4r-06) ( 4 it-05) (i it-os) VD, alpha, 0003 0 003 0.005 0 003 0 005 early CF Vil > 200 psi. 0 005 0 Oct 0 001 0 004 0 013 carly CF VD, < 200 ps1, early CT
~
f VII, INT or late et 0 079 0 040 0 013 0055 0 059 0 292 l [ .,_ Dypass 0003 0 078 0 007 1 000 0 122 _ _ - ._ ~._ VH, No CF 0 310 0 621 0 217 OSM 0 348 0 690 No VD 0 599 0 3$0 0 762 0 352 0 460
. _ . . _ _ . _ _ _ L_ . _ . _ _
Eev IWT
- Dasemat Mclt-Tlircush SURRY
$vnEN.,,,"n"n,'UN u ..o ,Sf" Figure 2.5 3. Mean Probability of APBs for PDSs--Internal and Fire.
2.87
likely outcomes. If VB is followed by CF, then a late failure is more ! likely than failure at or before VB. The late failure may be due to hydrogen ignition some hours after VB, but is more likely to be due to BMT. i Early CF is fairly unlikely, as was indicated by Figure 2.5 2. This is largely due to the robust nature of the Surry containment. F!gure 2.5 3 shows only the mean frequencies for the summary PDS groups and mean conditional probabilities for the summary APBs , where the mean is taken over all 200 observations in the sample. The core damage frequency of each PDS group is different for each observation. Figure 2.5-3A displays the range of mean core damage frequencies for the 200 observations for the seven PDS groups. The frequency range from the 5th percentile to the 95th percentile is about two or three orders of magnitude for all of the PDS groups APBs except Event V. The large range for Event V reflects the large uncertainty in the initiating event frequency for the interfacing system LOCA. 1.E- 3 . SURRY e6t h. sot h., pg_4
- - 66th. 06th ,
l .E- 5. % l u., _.l, 96th. 06th. 96th. 6th.. ]
% ]
- L_, M-* I g 6th ,_ W -. . g_, llc; h' '
6th .
% ! l g I _
l~ 6i% __l
~
c:1'E-7' 6t h - 6th h -
] m.
{1.E-8. _ l0-9. W = mean rn = median th a percentde 1.E-10 '" PDS Group Slow SD0 LOCAs hst SD0 Event V Tra ns. ATWS SGTR Total Figure 2.5.3A. Range of Mean Core Damage Frequencies for 200 Observations for Seven PDS Groups. 2.88
The mean conditional probability of each summary APB may be computed for each PDS group for each observation. When combined with the PDS group frequency, a frequency for each summary APB for each observation is obtained. The distribution of these values is displayed in Figure 2.5 3B, The 95th percentiles of the distributions for VB coincident with early containment failure (the first three distributions) all fall below 1.0E-6/ year. The means are much greater than the medians for these distribu-tions, indicating that the means are largely determined by a small number of observations with high frequencies of vessel breach followed by early CF. The Bypass summary APB includes both Event V and the SGTRs. The lon6 low frequency ' tail' of the distribution for Event V in Figure 2.5 3A is lost when the interfacing system LOCA and SGTR frequencies are summed for presentation in Figure 2.5 3B. The releases from accidents that result in VB and early CF are roughly com-parable to releases from the most severe bypass accidents, and the releases 1.E- 3 , SUHl(Y l l.E-4, i estk. esth. i 1.E- f2, D6t h. 95th. " t. w-
";* L. l .u . _J T 1.E-6_ l l
[ u
. sui. .
.%. su,.J
- a. _I
&l.E-7. u_.
h C:
]
l $1.E-8.
"+
l I 1 e6th.l l I I 1.E- 9. m..
] j
~.
1.0-10 6th. W = mean m = median th = percentile 3.E.3t .a i. ..! l APH Group VB. alpha Vil Vil Vit DMT ltypass VD, No CF Core early CF > 200 psi <. 200 pst or late CL Damage e erly CF carly i F No VH Figure 2.5.3B. Distribution of Frequencies for APB Groups. 2.89
from both of these types of accidents are much larger than non bypass accidents in which the containment does not fail at all or fails some hours af ter vessel breach. Therefore, since Figure 2.5 3B shows that bypass accidents have a much higher frequency distribution than accidents with vessel breach and early CF, it may be inferred that the risk to the offsite population from interrally initiated accidents at Surry is likely to be j dominated by bypass sceidents. l l 2.$.2 Sensitivity /nnlyses for Internal Initiators. This section reports the results of three sensitivity analyses performed for the internally initiated accidents at Surry. The first explores the effects of changing the split between the pressure ranges for the RCS pressure at VB for S3 breaks. The second sensitivity analysis concerned the elimination of temperature induced SCTRs and hot leg failures. The third sensitivity analysio consisted of running a second sample of 200 observations. 2.5.2.1 Pressure at VB for S2 and S 3Breaks. The RCS pressure at VB is important because it largely determines the magnitude of the effects that accompany failure of the lower head. The pressure in the vessel just , before breach may be considerably higher than the pressure during most of the core degradation process. A considerable amount of water is predicted to remain in the lower head until core slump. The steam generated at core slump repressurizes the RCS, at least temporarily. The pressure from core slump decreases at a rate determined by the size of the hole (s) in the RCS pressure boundary. For "T" PDSs, the PORVs reclose and keep the RCS at 2500 psia, For the "A" PDSs, the hole is so large and the depressurization so rapid that the RCS pressure is unquestionably low at VB. For situations with S3 and S breaks, 3 the depressurization is slow enough that VB may occur while the pressure is quite high. Indeed, for small and very small breaks, the RCS pressure at breach depends directly upon the time between slump and breach. The time. between slump and breach varies considerably depending on the lower head failure mechanism postulated. Code predictions for the time between slump or collapse and breach range from a few minutes to over an hour. Not only is chore uncertainty as to the time between slump and breach, but different mechanistic codes predict different peak pressures due to the slump and different pressure decay rates for the same size h ue. Finally, each hole size represents a range of sizes, and the pressure Jecay rate varies considerably for holes within that range. As a result of these I unknown and uncertain parameters, for S3 and Sa breaks, the RCS pressure at VB could not be assigned to -a single pressure range. Indeed, eve n the division of the probability among the ranges was difficult. More informa-tion on this topic may be foand in Volume 2, Part 6, of this report. The division of the branch probability among the pressure ranges was de-termined by Sandia National 1.aboratories and Battelle Columbus Division. For situations with S2 breaks, it was determined that the low pressure branch probability was 0.80 and the intermediate branch pressure was 0.20. For situations with S 3 breaks,-the division of the probability used in the analyses was one third for each of the low, intermediate, and high pressure 2.90 t
branches. A sensitivity analysis was done for the Sa breaks (Case 3 of Question 23) using probabilities of 0.47 for low pressure, 0.42 for intermediate pressure, and 0.11 for high pressure. The high pressure range includes all pressures from 600 psia to 2000 psia, but most of the code results examined tended to have RCS pressure at VB that fell in the 1000 to 1400 psia range. The intermediate range is from 200 to 600 psia, and the low pressure range is below 200 psia. In summary, the branching probabilities for the base and sensitivity cases are: Pressure Rance Sensitivity Sagg liiPr 0.11 0.333 ImPr 0.42 0.334 LoPr 0.47 0.333 Only the most frequent of the internally initiated PDS groups was examined in detail. This group was Group 1, Slow SBO. This PDS group consists of six PDSs, two with no break in the RCS at UTAF, two with an Sa break at UTAF, and two with an Sg break at UTAF - The changes from the base case to the sensitivity case were pronounced only for the high pressure branch in Question 23. The average branching ratios - for a sample of 200 observations for the four pressure ranges were: RCS Pressure at Vessel Breach (Q23) - Pressure Ranne Sensitivity BAgg SSPr 0.055 0,055 liiPr 0.034 0.10 ImPr 0.21 0.19 1.oPr - 0.70 0.65 The realized splits for Case 3 of Question 23 were, for the three non zero branches: liiPr 0.10 0.32 ImPr 0.39 0.33 LoPr 0.51' O 35 i t These realized splits differ from the specified probabilities. given above. l but the variation is not unreasonable given ' that zero one sampling was employed. ~ (Zero one sampling is discussed in Volume 1 of this report and , under Questions 15 and 23 in Appendix A.1.1 of this volume.) The fraction H
. of PDS Group 1 which has an initiating S3 break (all RCP seal failures before UTAF) is 30.24; at Question 23,. 32.0% of this group of accidents went to Case 3. The difference mostly consists of "T" PDSs (RCS intact at UTAF) for which an RCP seal failure occurred af ter UTAF, but there were some T-I SGTRs as well. While the difference in the Case 3 splits la significant, as expected, the difference in the overall Question 23 branch i probabilities is much less. Only two significant figures are r,1ven in these and all other tables, so the figures may not sum exactly to 1.00 itun to roundoff.
2.91
+
It is interesting to see if the change in branch probabilities for the l pressure at VB for S3 breake affected the type of VB. The realized branch probabilities were: 1 Type of Vessel Breach (Q35) Pressure Rance Sensitivity . Base I l PrEj 0.068 0.076 Pour 0.30 0.29 Btmild 0.012 0.016 NoVB or a 0.62 0.62 The differences are not significant. Whether the change in the branch probabilities for S3 breaks had any effect on containment failure at VB is most important. The realized branch probabilities were: Containment Failure at Vessel Breach (Q42) Pressure Range Sensitivity Base CtRp 0.0004 0.0004 Rupture 0.0039 0.0039 Leak 0.0029 0.0040 NoCF 0.99 0.99 The differences are not significant. The number of expected CFs resulting from VB at LoPr is essentially zero. While the number of expected failures , for VB at }! ipr is much larger, it is still small. And the fraction at liiPr at VB only changed from 3.34 to 104. The final question of interest for- this sensitivity analysis is the last question in the APET: Final Containment Condition (Q71) Pressure Range Sensitivity Base CtRp or Rupt 0.0068 0.0067 Leak 0.010 0.011 BMT 0.067 0.066 Bypass 0.0033 0.0033 NoCF 0.91 0.91 The differences are not r.ignificant. Given the results of the previous l- questions, this is to be expected. 2.5.2.2 No T.I SCTRs or Hot 1#r Breaks. A sensitivity analysis was performed to determine the importance and the effects of the temperature-induced (T.I) hot leg (and surge line) breaks and the T-1 SGTRs. These failures occur af ter the core melt and when the hydrogen and superheated 2.92
l' i I i steam leaving the core have heated the hot leg, surge line, and steam generator inlet plenum to temperatures on the order of 1000 K. Aggregate I cumulative failure probabilities for t eseh phenomena were provided by the in Vessel Expert Panel. Their conclusions were that these failures would l occur only if the RCS was at the PORV setpoint pressure (about 2500 psia). The hot leg failures were judged to be relatively likely (mean failure probability about 0.70) while the SGTRs were estimated to be quite unlikely ; (mean failure probability about 0.015). In the sensitivity analysis, these two TI failures were eliminated completely. Note that the distributions used for the other three depressurization mechanisms were not altered in this sensitivity analysis. The deliberate opening of the PORVs is not a partici'larly effective means of depressurizing the RCS, but the sticking 9 n of the PORVs and the failure of the RCP seals are effective. Of the seven internally initiated PDS groups at Surry, three (thCAs, Event V, and SCTRs) are completely unaffected by the elimination of the T I hot
- leg failures and T.I SCTRs because the conditions for these events (RCS at PORV setpoint pressure) are not met. The other four PDS groups were eval-uated in this sensitivity analysis, and the results for PDS Group 1. Slow SBO, will be discussed in some detail.
In the Surry APET, whether T 1 SGTRs occur is Question 19, and whether T I hot leg failures occur is Question 20. Thus, the base case (T-I failures I as specified by the expert panel) and the sensitivity case (no T I ! failures) are identical up through Question 18. I For slov blackouts, the mean RCS condition at the uncovering of the top of active fuel (UTAF) is: No Break 0.541 . S3 Break 0.303 S Break 0.156. l This is the condition of the RCS at the start of the accident progr.n:sion analysis as determined by averaging the 200 observations in the sample. Question 15 determined the - RCS pressure at UTAF. As the RCS pressure depends upon the state of the AWS as well as the condition of the RCS, the mean division among the pressure levels for the Slow SB0 PDS group at Question 15 does not exactly match the division among RCS states: SSPr 0.541 liiPr 0.126 0.333 ImPr LoPr 0.000, where SSPr - 2500 psia (PORV setpoint), li1Pr - roughly 1000 to 1400 psia, but perhaps as high as 2000 psia, ImPr - 200 to 600 psia, and 'a LoPr - less than 200 psia. 2.93
The high pressure range includes all pressures from 600 psia to over 2000 psia, but the detailed mechanistic codes suggest that, during most of the core degradation process, the RCS pressure will be in the 1000 to 1400 psia range. Question 16 is wherner the PORVs stick open. The probability that the PORVs will stick open is 0.50 if they are cycling, that is, if there is no break in the RCS and it is at the PORV setpoint pressure (SSPr). Thus, half of the no 'oreak states become effective S 2 states at this point. Question 17 is whether th:. RCP =cals fail. The mean failure fraction is 0.325, but most of these failuras occur for states in which there is already an S3 or Sa break, and so hwe no effect. As there is no electric power, the operators are prevented '! rom opening the PORVs in Question 18. Question 1% concerns the T-1 SGTR. No SCTRs were computed in the sensiti-vity case vs. 0.0033 in the base case. Question 20 concerns the T.I hot leg (or sutge line) failure. No failures were computed in the sensitivity case vs. 0.197 in the base case. Thus, at Question 23, which determines the pressure in the RCS just before VB, the mean division among the pressure levels is noticeably different for the two cases: RCS Pressure at Vessel Breach (Q23) Sensitivity Base Pressure Rance (No T 1 Breaks) (T 1 Breaks) SSPr 0.25 0.035 HiPr 0.10 0.10 ImPr 0.19 0.19 LoPr 0.46 0.65 These tables give results to only two =ignificant figures, so roundoff may cause the column sums to differ slighcly from exactly 1.00 Since the PORVs stick open half the time for the T " PDSs, and the RCP seals fail about 60% to 70% of the time when there is no pump seal cooling, there are two effective means of depressurizing the RCS in the sensitivity case. About 60% of this PDS group has no pump seal cooling. The stuck open PORVs question alone has converted half the No Break PDSs in the Slow SB0 group to effective Sa breaks. The base case has T.I hot leg breaks as well, and the difference is obvious. As expected, the T 1 hot leg failures and SGTRs affect only the SSPr and LoPr pressure ranges since hot leg failures occur only when the RCS pressure is at the PORV setpoint value. 2.94
The fractions of the Slow SB0 group that went to each case in Question 23 may also be of interest: Sensitivity Base (No T 1 Breaks) (T-I Breaks) Case 1: A Breaks 0.000 0.197 Case 2: S Breaks 0.428 0.428 Case 3: S3 Breaks 0.327 0.320 Case 4: No Breaks 0.245 0.055 , The effect of eliminating the T I SGTRs is negligible, even in Question 23, but the effect of eliminating the T I hot le6 failures is to transfer about 20% of the Slow SB0 group from LoPr to SSPr. The reason the fraction is not greater is that only 54% of the group is in the "No Break" category to begin with, and the stuck open PORVs eliminate half of this category before the hot leg failure question is asked. The type of vessel failure is determined in Question 36 of the Surry APE 1. The realized branching is: > Type of Vessel Breach (Q36) Sensitivity Base
,,,,,1ype o f VB (No T I Breaks) (T I Breaks)
PrEj 0.14 0.076 Pour 0.22 0.29 BtmHd 0.025 0.016 NoVB or a 0.62 0.62 The differences are not larger because the probability is about 0.62 that offsite electric power and coolant injection is recovered before a large portion of the core is molten, and vessel failure is thus averted. It may be noted that the fraction for pressurized ejection is nearly doubled. l Alpha mode failures account for only about 0.2% or 0.3% of the vessel ! failures. L If eliminating the T-I SCTRs and hot leg failures is to increase risk
- significantly, it must do so by increasing the fraction of containment failures at VB. This is determined in Question 43. The mean branch probabilities for the Slow SB0 group are
Containment Failure at Vessei Breach (Q43) i Sensitivity Base CP Mode (No T-I Breaks) (T-1 Breaks) Cat. Rupture 0.0005 0.0004 Rupture 0.0070 0.0039 Leak 0.0049 0.0040 NoCF 0.99 0.99 l 2.95 l
There are slightly more containment failures when the T 1 RCS breaks are set to zero. The number of ruptures is neacly doubled, but the fraction is small in either case. 'The increase in the number of leaks is not as great. Although 19% of the accident was transferred from LoPr at VB to SSPr at VB, the number of containment failures would not be expected to increase pro- l portionally. In the first place, 62% of the accident goes to NoVB since l power is recovered before the core melt has gone very far. And, of the i portion of the accident shifted to SSPr, only a fraction will result in CF l st VB since the load distributions provided by the Containment Loads Expert . Panel are generally lower than the containment failure pressure distribu- l tion provided by the structural experts. This will be diccussed in more I detail below. The state of the containment at the end of the APET is summarized in the final question: Final Containment Condition (Q71) Sensitivity Base Condition (No T.I Breaks) (T 1 Ereaksi Cat. Rupt. or Rupture 0.0095 0.0067 Leak O.012 0.011 BMT 0.058 0.066 Bypass 0.00 0.0030 NoCF 0.92 0.91 The differences are not significant. Given the results of Question 43, this is to be expected. Tables 2.5-8 through 2.5 11 summarize the results of the sensitivity anal-ysis for the four internally initiated PDS groups for which the elitination of the T I breaks have any effect. The Slow SB0 group has alter.ly been discussed. The tables show the mean branch probabilities. The Fat SB0 group results are similar to those.for the Slow SB0 group. The difference in CF at VB, the most important question for offsite risk, is discernible but not significant. For the Transient group, Table 2.5 10, the maj or , difference is in the probability of core damage arrest and no vessel fail-ure. The hot leg failure plays a very important role in depressurizing the RCS so that LPIS injection results. Furthermore, RCP seal cooling is oper-ating in this PDS group, so the RCP seal failure mechanism is ',,ot effec-tive. In spite of the large difference in the probability of the arrest of core degradation, the elimination of the T 1-breaks'does not make CF at VB likely. While the relative increase in the probability of CF at VB is large, the Frobability of CF at VB with no T I breaks is still only about 0.016. For the ATWS PDS group, Table 2.5 11, the differences between the base and the sensitivity cases are not significant. Both the base and sensitivity cases were carried through to risk. There , were no significant differences between the two cases. The following tables and discussion show that although the probability of SSPr at VB increased markedly when the T I hot leg failures and SCTRs were eliminated, the increase in the probability of CF at VB was fairly small. 2.96
The reason for this is that for VB at SSPr, CF results only a small frac-tion of the time. This fact follows from the load and failure pressure distributions determine ( by the expert panels. The relationship of these distributions has been summarized in Table 2.5 12. Table 2.5 8 Comparison of APET Results With and Without T 1 Hot Leg Breaks and SGTRs PDS Group 1 Slow SB0 Fraction With RCS Pressure in Four Ranges: At VB At UTAF At VB Base Case No T-I Breaks SSPr 0.541 0.055 0.245 HiPr 0.126 0.101 0.103 ImPr 0.333 0.190 0.191 LoPr 0.000 0.654 0.461 Base Case Sensitivity Case Fraction With No Vb 0.618 0.618 Fraction With Alpha Mode Failure 0,0028 0.0022 Fraction With CF at VB Total 0.0083 0.0124-Ca t' . Rupture 0.0004 -0.0005-Rupture. 0.0039 0.0070-Leak- 0.0040 0.0049 Fraction With VB, but No CF at VB 0.374 0.370 Fraction With CF- Late Burn 0.0072 0.0068 Fraction With CF -Very Late Burn 0.0024 0.0022 Fraction With the Following Final Contai:.,ent Condition Rupture- 0.0067 0.0095 Leak O.0114 0.0120 BMT 0.0660 0.0577 Bypass 0.0030 0.0000 NoCF (& NoVB) 0.9130 0.9209 2.97
Table 2.5 9 Compatison of API'T Results With and Without
'l-7 Hot Leg Breaks and SGTRs PDS Croup 3 -Fast SB0 Fraction With RCS Pressure in Four Rangest At VB At UTAF At VB P.r.se Case No T I Breaks SSPr 1.000 0.029 0.150 HiPr 0.000 0.116 0.116 ImPr 0.000 0.198 0.197 LoPr 0.000 0.657 0.537 Base Case Sensitivity Case Fraction With No VB 0.508 0.508 Fraction With Alpha Mode Failure 0.0038 0.0036 Fraction With CF at VB Total 0.00/4 0.0130 Cat. Rupture 0.0000 0,0001 Rupture 0.0049 0.0089 Leak 0.0025 0.0040 Fraction With VB, but No CF at VB 0.484 0.479 CF -Late Burn 0.0166 0.0149 CF Very Late Burn 0.0005 0.0006 Fraction With the Following Final Containment Condition Rupture 0.0070 0.0107 Leak 0.0179 0.0180 4 'BMT 0.0870 0.0765 Bypass 0.0021 0.0000 NoCF (& NoVB) . 0.8862 0.8948 1
2.98
i Table 2.5 10 Comparison of APET Results with and without T-I Hot Leg Breaks and SCTRs l PDS Croup 5 Transients Fraction With RCS Pressure in Four Ranges: At VB ; At UTAF_ At VB Base Case No T-I Breaks SSPr 1.000 0.115 0.500 HIPr 0.000 0.002 0.000 ; ImPr 0.000 0.093 0.091 l LoPr 0.000 0.790 0.409 j Hase Case Sensitivity Case i Fraction With No Vb 0.766 0.422 Fraction With Alpha Mode Failure 0.0009 0.0009 ! Fraction With CF at VB Total 0.0022 0.0165 Cat. Rupture 0.0000 0.0000. 1 Rupture 0.0017 0.0108 ! Leak 0.0005 0.0057 Fraction With VB, but No CF at VB 0.232 0.562 Fraction With CF- late Burn 0.00 0.00 Fraction With CF -Very Late Burn 0.00 0.00 Fraction With the Following 1 Final Containment Condition Rupture 0.0016 0.0108 ! Leak 0.0008 0.0590 ' BMT 0.0130 0.0202 Bypass 0.0066 0.0000 NoCF (6 NoVB) 0.9781 0.9631 2.99 W-
n Table 2.5 11 Comparison of APET Results with and without T-I Hot Leg Breaks and-SCTRs PDS Croup 6 -- ATWS Fraction With RCS Pressure in Four Ranges: At VB i At UTAF At-VB Base Case No T-I Breal;g ] SSPr 1.000 0.005 0.025 liiPr 0.000 0.002 0.008
. ImPr 0.000 0.191 0.198 LoPr 0.000 0.802 0.769 i
Base Case Sensitivity Case r-Fraction With No VB 0.406 0.387 , i Fraction With' Alpha Mode Failure 0.0029 0.0039 Fraction With CF at VB Total 0.0029 0.0039 Cat. Rupture 0.0000 0.0000
-Rupture 0.0029 0.0033 Leak O.0001 0.0006-Fraction With VB, but-No CF at VB 0.591 0.609 Fraction With CF -Late Burn 0.00 0.00 Fraction With CF -Very Late Burn 0.00 0.00 Fraction With the Following [
Final Containment Condition Rupture. 0.0029 0.0033 4 Leak, 0.0003 - 0.0008 BMT 0.0472 0.0470 Bypass 0.0761 0.0754 NoCF (& NoVB) 0.8736 0.8736 2.100 f
l, l l- , i I E Table 2.5 12 Probability of CF and-Probability of Caso I for CF at VB for Surry PDS Croup 1 - Slow SB0 l l 1' Core Prob. Prob. Hole RCS Rx Fraction Prob. Case Case Case .1121 Pressure Cavity Ej ection CF Sens. 13.g3 1 Any Lo Any Any .0016 .220 .288 2 Large Im Wet Large .123 .0016 .0016 3 Large Im Wet Medium .071 .0016 .0016 4 Large Im Wet Small .012 .0009 .0009 5 Large Hi Dry Large .287 .0053 .0015 6 Large Hi Dry Medium .173 .0054 .0015 7 Large Hi' Dry Small .022 .0094 .0045 8 Large Hi _ Wet Large .181 .0057 .0049 9 Large Hi Wet Medium .109 .0043 .0029 10 - Large Hi Wet Small .014 .0058 .0052
-I l 11 Small .In ' Wet Large .081 .0034 .0034 L 12 Small Im . Wet Medium .035 .0037 .0036 13 Small Im Wet Small .007 .0037 .0037 1
14 Small Hi Dry Large .152 .0212 .0093 Small Hi
~
15 Dry Medium .073 .0269 .0082 16 Small Hi Dry Small .009 .0201 .0069 l
~17 Small Hi Wet Large .095 .0109- .0084 +
18 Small Hi Wet Medium .043 .0192 .0151 19 Small Hi Wet Small .008 .0109 .0085 L The pressure rise at vessel breach is determined in Questions 39 and 40 in 11 the - Surry APET. These two questions contain 19 non trivial cases for u . pressure rise at VB; they have been numbered sequentially in Table 2.'5-12.- The pressure rise depends upon: 1
- The size of the hole in th- vessel, i
- The RCS pressure t.c VB, L
- The presence or absence of water in the cavity, and
- The fraction of the core ejected at VB.
lL D Columns : 2 through 5 define these quantities for each case. The . cxperts considered the ' HiPr and SSPr ranges to be the same for load purposes, so l the "Hi" designation in the "RCS Pr," column in Table 2.5-12 means roughly 1500 to 2500 psia, although the range actually extends as low as 600 psia, 1; The "Im" range 1 is 200 to 600 psia, and the "Lo" range is less than 200 psie. There turned out to be no significant statistical differences l l. 2.101
between the experts' aggregate pressure rise distributions for the Hi RCS pressure and wet cavity case and the Im RCS pressure and dry cavity case. So Cases 8, 9, 10, 17, 18, and 19 apply to intermediate pressure, dry cavity cases as well as'the cases indicated. The " Prob. CF" column in Table 2.5-12 gives the mean probability of CF if only that case is considered, i.e., if all the frequency went to that case. For example, if the APET were rigged so that Case 11 in Table 2.5 12 was always selected, then CF would be expected about 8% of the time. The values in this column were determined by comparing 10,000 random values from the containment failure distribution with 10,000 random values from each pressure rise distribution. The' comparison used to determine the mean CF probabilities was done in the following manner. The Structural Expert Panel provided an aggregate dis-tribution for the probability of containment failure as a function of pressure, in psig. This was converted to psia by adding 14.7 psi to every point in the distribution. The Containment Loads Expert Panel provided - aggregate distributions for pressure rise at VB for the 19 cases listed. These distributions were converted to absolute pressure by adding to them the baseline pressure in the containment before VB. The "no sprays" base-line pressures were used,-which makes this comparison applicable to the SB0 PDS groups (Groups 1 and 3). With no sprays operating, the baseline pres-sure is 26 psia- for all breaks but the large breaks . (A size) and 37 psia for the large breaks. That is, 37 psia was added to the low pressure pressure rise distribution, and 26 psia to the other 18 pressure rise distributions. The absolute load and failure pressure curves were compared by selecting 10,000 pairs of random numbers between zero and one. One value from each pair determined a point on- the failure curve, and the other determined a point on a load curve. If the load pressure exceeded the failure pressure, it was counted as a CF. Note that for station blackouts, it is possible to have water in'the cavity at VB if the power is restored.before VB, but VB is not averted. In this situation, the sprays would have filled the cavity and reduced the baseline containment pressure to about ambient pressure. This comparison method overstates the CF probability for the " Wet" cases in Table 2.5-12. Examina-tion of.the " Prob. CF" column in Table 2.5 12 shows that the mean probabi. i lity of CF at VB ranges from about 0.29 for the case with the highest pressure rise, Case 5, to less than 0.01 for Cases 13, 16, and 19. It remains. to determine the probability of each of these 19 cases. The " Prob. Case" columns in Table 2.5 12 give the fraction of PDS Group 1 (Slow SBO) that went to each case for the base run and for the sensitivity run in which the T-I hot leg breaks and SGTRs were eliminated. Note that the values in the " Prob. Case" columns only add up to about 0.38 since about 62% of the slow blackout PDS group had core damage arrest and no VB due to the recovery of offsite electric power. As small holes in the vessel are much more likely than large holes, only the small hole cases were selected with a probability over 0.01. As would be expected, Table 2.5 12 shows that eliminating the two T-I breaks decreased the fraction in the low pressure case and increased the fractions in the high pressure cases. The probability of the Slow SB0 group going to the case with the 2.102
i I highest pressure rise, Case 5, is very low, less than 0.01 for both the base and sensitivity cases. The most probable case for the base case, Case 18, is about 10 times more probable than Case 5. The CF probability for Case 18 is 0.043. The most probable case for the sensitivity case, Case 15, is about five times more probable than Case 5. The CF probability for , Case 15 is 0.073, Prom Table 2.5 12 it may be seen why the increase in the fraction with SSPr i i st VB'did not lead to a corresponding increase in the fraction with CF at VB. The load distributions and the failure distribution given by the i experts are such that the more probable pressure rise cases result in CF ; probabilities on the order of 0.01 to 0.10 at Surry. If the entire frac- ' tion that went to VB in the Slow SB0 PDS group went to Case 15, the CF probability would be 0.073 for the 38% that went to VB, resulting in an overall CF probability of 0.027. When the T 1 hot leg and SGTR breaks were eliminated in the sensitivity analysis, for the slow SB0 group, the
- probability of CF at VB increased from 0.008 to 0.012. So the effective probability of CF at VB, conditional on VB, is considerably less than that for Case 15. ;
In summary, the ef fect of eliminating the T-1 hot leg breaks and SCTRs is to increase' the mean probability of PORV setpoint pressure at VB signifi-cantly. For the SB0 PDS groups, this has little effect on the mean pro- ! bability of CF at VB. For the Transient PDS group, eliminating the T I i breaks has a large effect on the mean probability of core damage arrest, j but only a very small effect on the mean probability of CF at VB. When all PDS groups are considered, the effects of eliminating the T-I breaks on , offsite risk are not discernible. The reason there is so little effect can l be found in the containment failure pressure and containment-load distribu-tions determined by the expert panels. That is, Surry has a strong con-tainment relative to the loads expected, and increasing the RCS pressure at VB does not increase the containment failure probability markedly. 2.5,2,3 Second Samole. To test the robustness of @ sampling pro-cess, and to determine which means of displaying the results are uhject to { variation from sample to sample, a second sample of 200 observations was run all the way through to risk for the internal initiators at Surry. The : analyses were identical except for the seed value used for the Ills program that generated the sample values. [ Figure 2.5-4 shows the mean probabilities of the summary APBs for each summary PDS group for the second sample. It may be compared with Figure 2.5 3 which -is the equivalent plot for the first sample. Although the results are not valid to three significant figures, three significant figures are shown so that the differences between the two samples may be discerned. The conditional probabilities for the collapsed APBs differ less than 0.01 from the first sample to the second. These differences are f not significant. Indeed, the largest differen'ces in this figure are in the mean core damage frequencies, which -- are determined in the accident fre-quency analysis, not here in the accident progression analysis. The results for PDS Croup 1, Slow SBO, have been examined in detail. For the two samples, the mean branch probabilities for the question that determines the RCS pressure just before vessel failure are: 2.103
SUMMARY
SUMMARY
PDS GROUP (Mean Core Darnage Fawency) ACCIDENT i
~~~~~~~~~~~~~~~~~~'""*"'"*~~~~~~~~~~~~~~~~~~~
PRO M SS N LOSP ATWS Transients LOCAs Bypass All l BIN GROUP ( 2.9E-05)( l.5E-06) ( 1.8E-06)( 5.9E-06)( 3.4E-06)( 4.2E-05) VD, alpha, 0.003 0.004 0.001 0.004 0.003 early CF VB > 200 psi, 0.006 0.005 0.002 0.002 0 004 carly CF VD, < 200 psi, early CF VD DMT or late CL 0.076 0.046 0.013 0.055 0.058 Dypass 0.003 0.077 0.006 1.000 0.121 VD, No CF 0.312 0.51 r> 0.212 0.58 0.351 , No VB 0.000 0.354 0.767 0.357 0,462 Key: BMT = Basernet Melt-Through SURRY CF = Containment Thilure CL = Containtnent Lenk SAMPLE 2 VB = Vessel Dreat h Figure 2.5-4. Mean Probability of APBs for PDSs--Internal Sample 2. Question 23. Vessel Pressure Just Before Breach? Samnle 1 Samole 2-Setpoint Pressure 0.0549 0.0523 High Pressure 0.1015 0.1166 a- Intermediate Pressure 0.1901 0.1866 Low Pressure 0.6536 0.6445 The differences between the samples are not significant. More significant
-figures are given than. the accuracy of the analysis warrants so that the differences between the two samples are evident.
The mean probability of arresting the degradation of the core and avoiding ' VB is determined in Question 24; the probability is 0.6177 for Sample 1 and 0.6185 for Sample 2, 2.104
G The- mean branch probabilities for containment failure and the mode of failure at VB are: Question 43. Containment Failure at Vessel T '3ach? Samole 1 Samole 2 Catastrophic Rupture 0,00043 0.00008 Rupture 0.00390 0.00629 Leak 0.00328 0.00228 No d iure 0,9917 0,9914 The differences are not significant. T The output bins for PDS Group 1, Slow SBO, were also examined, Some of the significant bins are listed in Table 2,5-13, -Differences are evident only for'eean probabilities below 0,01. Dit tributions for various results were also examined and showed no signi-ficant differences between the two samples. The conclusion from running tw o . sar le s is that the accident progression analysis appears to be very st able d reproducible, t-i ;- g. 2,105 -~
.m m .
Table 2.~5-13' Comparison of. Selected. Bins .for .PDS Group 1, Slow SB0 for Two Samples for Surry Samole 1 Sample 2 Relative' Relative Bin Frecuency. Bin Frecuency Comments HDC-DFC-DBD-FB 0.1713 HDC-DFC-DBD-FB 0.1709 Highest Freq. Bin, No VB, NoCF EDC-DFC-DAD-FB 0.1446' HDC-DFC-DAD-FB 0.1448 2nd Highest Freq., No VB. NoCF HDC-DFC-DAD-FA 0.0455 HDC-BFC-DAD-FA 0.0427 3rd Highest Freq., No VB, NoCF N CFA-DBC-ABD-DB 0.0125 CFA-DBC-ABD-DB 0.0125 P' , hest Freq. Bin with CF, BMT
-o 0.0015 Highest Frequency e EFA-DBC-AAD-CB 0.0029 EFA-DBC-ABD-CB Bin with CF due to Late Burn DHA-DDC-BAD-BB 0.0012 DHA-DDC-BBD-BB 0.0015 . Highest Frequency Bin with CF at VB due to Alpha DFA-CAC-ABA-CB 0.0007 DFA-CAC-AAA-BB 0.0008 Highest Frequency Bin with CF at VB due to DCH s
l
\
2.5.3 Results for Fire Initiators i This PDS group consists of accidents initiated by fires. Fires were found i to be important in four locations at Surry: the emergency switchgear room, i the auxiliary building, the control room, and the cable vault and tunnel. Table 2.2 4 lists the mean CDFs for these four locations. The fires lead to core damage accidents by destroying the electric cables or switchgear necessary to power or control the ECCS. The coolant loss from the RCS is due to RCP seal failures in three of the four PDSs, and to stuck open PORVs in the fourth PDS. The destruction caused by the fire is considered not to be reparable in the timeframe of interest; that is, there is no chance of recovering the ECCS, so there is no possibility of arresting the core degradation process and vessel failure is inevitable for the fire PDSs. Table 2.5 14 lists;the 10 most probable APBs for the PDS group and the five most. probable APBs that have VB and early containment failure (CF). Since ; there i no possibility of core damage arrest for accidents initiated by .
' fires, all the APBs have VB. So, the 10 most probable APBs are listed !
instead of the five most probable and the five most probable that have VB. i Evaluation of the APET produced 754 bins for the Fire group. To capture 95% of the probability, 230 bins are required. Three of the 10 most ' probable bins have BMT; the other seven have no CF. The ten most probable , bins capture 42% of the probability. Of the five most probable APBs with , both VB and ' early CF, two have CF . due to gross bottom head failure . at l intermediate pressure, two have CF due to an Alpha mode failure of both ' vessel and containment, and one has CF due to HPME and DCH with the RCS at i
'high pressure. (Early CF means CF before or at VB.)
l The fractions of the five PDS groups that are in the four pressure ranges at UTAF and just before VB (if it occurs) are:
]
At UTAF Just Before VB l l SSPr (2500 psia) 0.0 0.0 HiPr (600 to 2000 psia) 0.14 0.15 ImPr (200 to 600 psia) 0.86 0.24 l LoPr (< 200 psia) 0.00 0.61 i
'The relative frequencies of the "Sa" and "S"2 PDSs, in' conjunction with whether the secondary system has been depressurized while the AFWS is operating, result in over three fourths of the Fire group being in the high pressure range when the core uncovers (Question 15). Just before VB, the situation - is quite different (Question 23). Five mechanisms for depressurizing the RCS are considered in the APET. Only the deliberate '
opening of the PORVs by the operators is effective for the Fire PDSs. The
. determination of RCS pressure at VB was discussed in some detail in Section 2.5.2.1. See also the discussion of Question 23 in Appendix A.
2.107 1 t u
L l Table 2.5-14 Results of the Accident Progression Analysis for Surry Fire Initiators. PDS Croup Fire No. CF RCS VB Amt Zr CF Order ' Bin Prob.* Occur. Time Sorays .- CCI' fr.m Mode CCI Qg HPME Size Ten Most Probable ~ Bins" 1 HHADBCAADFB.. 0.079 114 - No-CF Never PrmDry' lePr Pour Large lo No No-CF 2 HHADBCABDFB 0.069 92 No-CF Never PrmDry . LoPr Pour Large Hi No No-CF 3 CHADBCAADDB 0.054 .114- Final' Never PrmDry loPr Pour Large Lo No BMT 4 HDCDBCDADFB 0.047 89 No-CF Always .No-CCI LoPr Pour No-CCI Io No No-CF 5 GHADBCABDDB 0.047 29 Final. Never PrmDry loPr Pour Large 'Hi No BMT u 6 HDCDBCDBDFB 0.033 '85 No-CF. Always No-CCI loPr Pour No-CCI Hi No No-CF L 7 HHADBCAADFA .0.027' 38 No-CF Never PraDry loPr Four Large Lo No No-CF S 8 HHADBCABDFA 0.025 28, -No-CF Never PrmDry LoPr Pour Large HL No No-CF 9 HHDDBCAADFB 0.023 82 No-CF. Always PraDeep IoPr Pour Large la No No-CF 10 GHADBCAADDA 0.019 38 -Final Never PrmDry LoPr Four Large 14 No BMT Five Most Probable Bins that have VB and Early CF" 62 DHACCCAABCA 0.0032 'l CFatVB 'Never PrmDry ImPr BtmHd Large Im Md Irak 102 DHADDCBADBB 0.0018. '54 CFatVB Never PrmDry LoPr Alpha Medium No
.Lo Rupture 119 DHACCCABACA 'O.0014 1 CFatVB Never PrmDry ImPr BtmHd Large Hi Hi Ioak 151 DHADDCBBDBB 0.0010 47 CFatVB 'Never PrmDry lePr Alpha Medium Hi No Rupture 159 DDCBACDBACA 0.0009 1 CFatVB Always No-CCI HiPr HPME No-CCI Hi Hi Leak ,
Mean probabilitf conditional on the occurrence of the PDS. ' A listing of al.' bins, and a -listing by observation are available on computer media.
, I.
l, -
.. _ - - , . - - , _ . ,.-n. _ _ , _ .. - . . . _ __ _ _ _ _
The mean probability of CF at VB for this PDS group is 0.018. There are no late or very late CFs due to hydrogen burns for the Fire PDSs. Although power or control to the ECCS has been irreparably lost, normal station power is not failed. It was estimated that there would be enough sources of sparks in the containment in these situations to ignite the hydrogen present whenever a flammable concentration occurred. Deflagrations at the flammable limits are judged to be extremely unlikely to fail the Surry containment, so combustion events in these conditions are not considered. The mean probability of BMT is 0.26. As three of the four PDSs in this group have the sprays irreparably failed, it is possible that an overpressure failure of the containment may occur after several days due to the lack of'CHR. As the Surry basemat is constructed of siliceous concrete, the cause of this CF is not primarily due to the generation of noncondensible gases by CCI, but results from the increase in the temperatures of the containment wall and atmosphere to the point where steam condensation on the walls ceases. Steam then behaves as a noncon-densible gas. -The mean probability of eventual overpressure CF is 0.03. 2.5.4 Sensitivity Analyses for Fire Initiators n No sensitivity analyses were performed for fire initiators at Sur'ry.
, '2.5.5 Results for Eeismic Initiators- LLNL Hazard Distribution e
The seismic risk analysis was performed using two different seismic bazard distributions. This section reports the results using the hazard di:tri-bution developed by LLNL. Results based on the seismic hazard distribution devel'oped by EPRI are presented in Section 2.5.7. The differences between these two distributions are discussed in NUREG/CR-4550, Vol. 3, Part 3. The accidents initiate [ by earthquakes were analyzed in two groups. Those due to seisms with a maximum ground acceleration in excess of 0.6 g were denoted the high acceleration events. .It was judged that the destruction Mn the general vicinity of the plant for . those earthquakes would be so great ' that ovacuation would be ineffective. For the low acceleration events, less .than 0.6 g, it was . estimated that evacuation would be possible, although perhaps more slowly than in emergencies without earthquakes. 2.5.5.1 Results for PDS Group E0 l' LOSP (No SBO)- High Accelera t ion . 'LLNL Hazard Distribution. This PDS group consists of accidents initiated by LOSP, but in which SBO-does not result. The LOSP is due to the earthquake, but the DGs start and supply station power. There
. are five PDSs in this group as listed in Table 2.2-6. Four of them are * 'similar rto the' internally initiated transients; AWS fails at the start, and bleed and feed fails due to HPIS failures or PORV failures. Two of the four "T" (RCS intact) PDSs have LPIS operating, one has- both LPIS and HPIS operating, and the fourth has no ECCS operable. The "S" 3 PDS has an RCP seal failure and failure of all ECCS. Thun, in three of the five PDSs in the group, c' ore damage arrest is possible if a T-I break after UTAF lowers the RCS pressure sufficiently. This PDS group consists of the fraction of these accidents due to seisms with a maximum ground acceleration exceeding 0.6 g.
2.109
Table 2.5 15 lists the 10 most probable APBs for this PDS group and the five most probable APBs that have VB and early CF. As the the five most probable APBs that have VB are contained in the ten most probable bins, these are listed instead of repeating many of the APBs. Evaluation of the APET produced 729 bins for the LOSP, high acceleration PDS group. To capture 95% of the probability, 70 bins are required. The 10n most probable bins capture 73% of the probability. Four of the 10 most probable bins have no VB; the other six have low pressure pour failures of the lower j head. Two of the 10 most probable bins have BMT; the other eight have no CF. Of the five most probable APBs with both VB and early CF, two have CF due to an Alpha mode failure of both vessel and containment, and three have i CF due to llPME and DCH with the RCS at high pressure. Some of the PDSs in this group have ac electric power partially available in the plant. This is denoted by a "P" for the fourth PDS characteristic. Thic indication is used when seinmic failures have rendered part or all of the ECCS, sprays, or AWS inoperable but there is no SBO. The operability of these systems in these situations is indicated by the second, third, and seventh characteristics. In the LOSP, high acceleration PDS group there is the possibility of arresting the core degradation process and avoiding VB since three of the PDSs have some ECCS operating at UTAF. If the RCS can be depressurized, inj ection by the ECCS may arrest core damage for these PDSs. The mean probabilities for this PDS group for the four pressure ranges at UTAF and just before VB (if it occurs) are: At UTAF Just Before VB SSPr (2500 psia) 0,934 0.012 HiPr (600 to 2000 psia) 0.0 0.002 ImPr (200 to 600 psia) 0.066 0.190 LoPr (< 200 psia) 0.0 0.796 Four of the five PDSs in this group are "T" PDSs, which have the RCS intact at the PORV setpoint pressure at UTAF (Question 15). These PDSs account for the high probability for SSPr at UTAF. At VB (Question 23), the probability is only about 0.01 that the RCS will be at that pressure. The Surry APET considers five mechanisms for depressurizing the RCS. Deliberate opening of the PORVs by the operators, the PORVs sticking open, and T-I hot leg breaks are effective for the IAS P , high acceleration PDS group. RCP seal failures are not important for this group since the four PDSs that do not have RCP seal failures at UTAF have seal cooling operating. In the seismic accident frequency analysis, it was judged that the operators would be too busy coping with the effects of the earthquake to open the PORVs before UTAF. Thus, they may do so afterwards since power is available unless the PORVs have failed closed (as they have in one PDS). Note that failures that prevent the PORVs from being opened from the control room do not prevent the PORVs from opening in their relief mode, and thus sticking open. The experts who conside: ed T-I failures determined that hot leg failures were likely if the PCS was at the PORV setpoint l pressure (about 2500 psi), while SGTRs were very unlikely. As the j 2.110
Table 2.5-15 Results of the Accident Progression Analysis for Surry
' Seismic Initiators. PDS Group.EQ 1, LOSP (No SBO): High Acceleration, LLNL Hazard Distribution No. CF RCS VB ' Ant Zr CF Order Bin Prob.* Occur. Time' _Soravs CCI Pres, Mode CCI .jls HPME . Size Ten Most Probable Bins **
1 HHCDFCDBDFB 0.192 123 No-CF Never No-CCI LoPr No-VB No-CCI Hi No No-CF 2 HHCDFCDADFB 0.117 111 No-CF Never -No-CCI LoPr No-VB No-CCI Lo No No-CF 3 HHADBCABDFB 0.108 123 No-CF Never PraDry LoPr Pour Large Hi No No-CF 4 CHADBCABDDB 0.073 123 Final Never PraDry LoPr Pour Large Hi No BMT 5 HHADBCAADFB 0.072 112 No-CF Never- PrmDry LoPr Pour Large Lo No No-CF y 6 HDCDFCDBDFB 0.059 123 No-CF Always No-CCI LoPr No-VB No-CCI Hi No No-CF g 7 GHADBCAADDB 0.049 112- Final Never PrmDry LoPr Pour Large Lo No BMT r* 8 HDCDFCDADFB 0.030 111 No-CF Always No-CCI LoPr No-VB No-CCI Lo No No-CF 9 HHACBCABDFB 0.017 8 No-CF. Never PrmDry ImPr Pour Large Hi No No-CF 10 HDCDBCDBDFB 0.016 123 No-CF Always No-CCI LoPr Pour No-CCI Hi No No-CF Five Most Probable Bins that have VB and Early CF** 30 DHACACABACB 0.0037 1 CFatVB Never PrmDry ImPr HPME Large Hi Hi Leak 34 DHADDCBBDBB 0.0029 76 CFatVB Never PrmDry LoPr Alpha Medium Hi No Rupture 88 DHADDCBADBB 0.0005 71 CFatVB Never PrmDry LoPr Alpha Medium to No Rupture 112 DHFCACABACB 0.0004 1 CFatVB Never LDldDry ImPr HPME Large Hi Hi Leak 122 DHEAACAABCA 0.0003 2 CFatVB Never SDldDry SSPr HPME Large Lo Md Leak Mean probability conditional on the occurrence of the PDS. A listing of all bins, and a listing by observation are available on computer media.
deliberate depress,urization and PORVs stick open questions are asked before the T-I hot leg break question, the probability is high that the RCS will be depressurized before the hot leg failures can occur. Two of the five PDSs in this group have the LPIS operating at UTAF, and one l had both LPIS and HPIS operating. For these PDSs, it is possible to arrest core damage and avoid vessel failure. Due to the depressurization of the RCS before VB, the mean probability of arresting the core degradation process and avoiding VB is about 0.40. The mean probability of having the LPIS operating at UTAF is about 0.55, but not all of the time will the open PORVs depressurize the RCS far enough and f ast . enough to allow LPIS injection in time to prevent VB. The probability that both LPIS and HPIS will be operating is small (about 0.016). The mean probability of.CF at VB for this PDS group is 0.009. There are no late or very late CFs due _to hydrogen burns for the LOSP PDSs si;.cc power was available throughout the accident. The experts considering hydrogen ignition concluded that with electrical power available continuously, combustion would occur shortly after a flammable concentration of hydrogen was attained in the containment. Only deflagrations are possible at the flammable limits, and these are judged to be extremely unlikely to fail the Surry containment. Thus, late and very late CFs due to combustion when electrical power has been continuously available are not considered. l [ In the long term (several days), containment failures due to BMT or high pressures and temperatures in the containment are possible. The mean probability of BMT is 0.19 for the ' LOSP, high acceleration PDS group. BMT is only partially dependent upon the presence of containment heat removal (CHR). The long-term failure of the containment due to high pressures and I temperatures is completely dependent upon CHR. That is, if CHR is ; operating, this mode of CF, usually denoted eventual overpressure, is not possible. As the three most probable PDSs of the five PDSs in-this group have the sprays irreparably failed, eventual overpressure failure of the containment after several days is possible for this PDS group. In most plants the basemat is composed of limestone concrete, so the attack l of the - basemat by the core debris will generate a large amount of noncondensible gases such as CO 2. .The Surry basemat is constructed of siliceous concretu, so the eventual ovepressure CF at Surry is not primarily -due to the generation of noncondensibic gases by CCI. Instead, i it results from the increase in the temperatures of the containment wall and atmosphere over several days. When the wall temperatures reach the-point where steam condensation on the walls ceases, steam then behaves as a noncondensible gas and the pressure rises as steam is boiled off from the sump. The heat source is the decay products contained in the sump water. At Surry, this failure mechanism is estimated to take 5 or 6 days. While there is no question that this mechanism can eventually fail the contain-ment, whether it will occur is primarily a function of whether CHR can be restored _ in this time period. The heat load af ter several days is not high, so almost any ad hoc means of heat removal from the containment, or a means of water addition, will suffice to prevent eventual overpressure failure. Therefore it was estimated to be very likely, even following an 2.112
earthquake, that CHR would be restored before the containment failed. The mean probability of eventual overpressure CF for the LOSP; high accelera-tion PDS group is thus fairly small, about 0.02. 2.5.5.2 Results for PDS Crouc EO 2. SBO' Hirh Acceleration. LLNL Hazard Distribution. This PDS group consists of accidents initiated by 14SP in which SB0 follows. The LDSP is due to the earthquake, and the DGs fail to start due to seismic and random hardware failures. Due to the seismic failures in the electrical distribution system that may be expected, it was judged that offsite power would not be recovered within the timeframe of this analysis. 1hus there is no chance of arresting core damage or avoiding VB in this PDS group. This PDS group consists of the fraction of these accidents due to seisms with a maximum ground accelera-tion exceeding 0.6 g. There are eight PDSs in this group as listed in Table 2.2 6. Three of them are "T" (RCS intact) PDSs. two are "S" 3 (very small brsak) PDSs, one is an "S"2 (small break) PDS, and two are "A" (large break) PDSs. The two " A" PDSs haie failure of the SG or RCP pump supports coincident with the LOSP. These support failures are judged to place sufficient stress on the main steam line (MSL) penetrations - that the containment fails. The failure mechanism envisaged is cracking of the welded steel pressure boundary where the penetration stiffener plate joins the thinner plate. It was estimated that the probability was 0.90 that the failure would be of leak size and 0.10 that the failure would be of rupture size. Only the two "A" PDSs in this group have these " time zero" CFs. In six of the eight PDS in this group, the AFWS operates until the batteries deplete. However, one of these six PDSs is an "A" PDS, and with a large break the operation of the AFWS is irrelevant. Table 2.5 16 lists the ten most probable APBs for the PDS ' group and the five most probable APBs that have VB and early CF. In the "CF Time" column in Table 2.5-16, Early means at the start of the accident, and is used to distinguish "C" bins from "D" bins which have CF at VB. As the offsite consequences and risk are much the same whether the containment f ails - at time zero or at VB, when th. APBs are grouped, or in general discussion, Early means at or before VB, and so would include the CFs at the start of the accident. (In analyses of other plants, the probability of CF before VB due to hydrogen combustion may not be negligible as it is at Surry. In these plants, there tany be a significant chance of CF after UTAF and before VB for internal initiators.) Since there is no possibility of core damage arrest for the seismic SB0 PDS group, all the APBs have VB. So, the 10 most probable APBs are listed instead of the five most probable and the five most probable that hav. VB. Evaluation of -the APET produced 490 bins for this group, of which,109 are required to capture 95% of the probability. Two of the 10 most probable bins have CF at the start due to SC or RCP support failures, three have BMT, and the other five have no CF. The 10 most probable bins capture 61% of the probability. Of the five most probable APBs with both VB and early CF, four have CF at time zero; the fifth has CF at VB due to an Alpha mode event. 2.113
Table 2.5-16 Results of the Accident Progression Analysis for Surry Seismic. Initiators. PDS Group'_EQ 2, SBO: High Acceleration No. CF RCS VB Amt Zr- CF Order ~ Bin Prob.* Occur. Time Sorays CCI Pres. Mode CCI QK llPHE Size Ten Most Probable Bins" 1 HHADBCABDFB 0.147 123 No-CF Never PrmDry loPr Pour Large Hi No No-CF 2 CHADBCABDDB 0.100 123- Final Never PraDry LoPr Pour Large Hi No BMT 3 HHADBCAADFB 0.094 112 No-CF Never PrmDry IoPr Four Large Io No 'No-CF 4 CHADBCAADCB 0.073 112 Early Never PraDry loPr Pour Large In No Imak' 5 GHADBCAADDB 'O.064 112 Final Never PrmDry. IoPr Pour Large Io No BMT 6 CHADBCABDCB
'0.037- 120 Early Never. PrmDry IoPr Four Large Hi No Irak-HHADBCAADFA 0.031 38 No-CF Never PraDry IoPr Pour Large Lo No No-CF p 7 Large Hi No No-CF g 8 HHADBCABDFA 0.025 33 No-CF Never PrmDry IoPr Pour e 9 GHADBCAADDA. 0.021 38 Final Never PrmDry IoPr Pour Large la No BMT 10 HHACACABBFA 0.019 16 No-CF Never PrmDry ImPr HPME Large Hi Md No-CF Five Most Probable Bins that have VB and Early CF**
4 CHADBCAADCB 0.073 112 Early_. Never PrmDry IoPr Pour Large Is No leak 6 CHADBCABDCB 0.037 120 Early Never PrmDry IxPr Pour Large Hi No Imak 17 CHADBCAADBB 0.0081 '112 Early Never PrmDry I/ ipr Four Large lo No Ruptui 27 DHADDCBBDBB 0.0053 76 CFatVB Never PraDry loPr Alpha- Medium Hi No Rupture 39 CHADBCABDBB- 0.0041 88 Early Never PrmDry 2oPr Pour Large Hi No Rupture i Mean probability conditional 'on _the occurrence of the PDS.
" A listing of all bins, and a listing by observation are available on computer media.
^
+
In the SBO, hiS h acceleration PDS group, there is no possibility of arresting the core degradation process and avoiding VB since offsite power is not recoverable in the time period considered in this analysis. The depressurization of the RCS during core melt serves only to reduce the loads placed upon the containment at VB. The mean probabilities for this i PDS group for the four pressure ranges at UTAF and just before VB are: At UTAF Just Before VB
.SSPr (2500 psia) 0.67 0.02 riiPr (600 to 2000 psia) 0.0 0.11 ImPr (200 to 600 pria) 0.21 0.19 LoPr (< 200 psia) 0.12 0.68 Two of the three most probable PDSs in this group are "T" PDSs, so the RCS is intact at the PORV setpoint pressure at UTAF (Question 15) for these PDSs. In the two "S "3 PDSs, the AFWS operates until the batteries deplete and the secondary system is depressurized as well, so the RCS will not be in the high pressure range at UTAF for these PDSs. At VB (Question 23),
the probability is only about 0.02 that the RCS will be at the system l- setpoint pressure. The Surry APET considers five mechanisms for l depressurizing the RCS. RCP seal failures, the PORVs sticking open, and T-l I hot leg breaks are effective for the SBO, high acceleration PDS group. RCP seal failures are important for this group since SB0 causes a loss of I seal cooling. Deliberate opening of the PORVs by the operators is l prohibited by the procedures since there is no electric power. The mean probability of CF at VB for this PDS group is 0.016. This is in addition to the 0.12 probability of CF at the start of the accident. If a rupture occurs at time zero,. CF at VB - is not considered since no further failure is possible. If a leak occurs at time zero, CF at VB is considered since the - fast pressure rise at VB may convert the leak into a rupture or catastrophic rupture. As CF at VB itself is very unlikely, the conversion of a leak. from the start into a rupture at VB is also highly unlikely. There are no late CFs due to hydrogen burns for the LOSP PDSs since power is never recovered during this period. It is possible to have hydrogen burns in the very late period when the containment slowly cools off due to heat transfer to the containment, and enough steam condenses to de-inert ! l the containment atmosphere. The mean probability of very late CF due to a ! burn is slightly less than 0.04 l The mean' probability of BMT is 0.29 as sprays are not operable in the final period -for ' this PDS group. This stems from the inability to recover offsite power when the LOSP is - caused by an earthquake. The lack of sprays, and thus CHR, implies that an overpressure failure of the containment is possible several days af ter the start of the accident. As discussed in the previous subsection, this is unlikely to occur. The mean probability of eventual overpressure CF is less than 0.04 for the SBO--high acceleration PDS group. 2.115
2.5.5.3 Results for PDS Croup EO3. 1DCAs Hich Acceleration. LLNL Hazard Distribution. This PDS group consists of accidents initiated by seismic pipe breaks. The failures in the ECCS required to respond to these breaks are partially seismic and partially random. There is no SBO, but some of the seismic failures are failures in the electrical distribution system, specifically in the parts that supply power to the ECCS, sprays, or AFWS. In these situations, the availability of electrical power, fourth PDS characteristic, is denoted by a "P". This PDS group consists of the fraction of these accidents due to seisms with a maximum ground acceleration exceeding 0.6 g. There are 11 PDSs in this group as listed in Tabic 2.2-6. Three of them are "S" (small break) PDSs, three are "S" 2 (intermediate break) PDSs, and the remaining five are "A" (large break) PDSs. The "A" PDSs are initiated by pipe breaks due to failures of the SG or RCP pump supports. These support failures are judged to place sufficient stress on the main steam line (MSL) penetrations that the containment fails by the formation of cracks around the penetration stiffener plates. The prouability is 0.90 that the failure is a leak and 0.10 that the failure is a rupture. Only the "A" PDSs in this group have these " time zero" CFs. Two of the three "S"2 PDSs.and two of the three "S "3 PDSs have the LPIS operating at UTAF. In the "S3 " PDS s , the RCS will depressurize enough due to the break alone that inj ec tion from the LPIS will commence. VB is unlikely in this situation. Sufficient depressurization from the S 2 breaks is not as certain. LPIS injection and the avoidance of VB is not as likely as for the S breaks. i Table 2.5-17 hsts the 15 most probable APBs for the PDS broup. The overlap betwe .n the five most probable bins, the five most probable bins that have VB, and the five most probable APBs that have VB and early CF is so great that listing the 15 most probable bins provides more information. In the "CF Time" column in Table 2.5-17, Early means at the start of the accident, and is used to distinguish "C" bins from "D" bins which have CF at VB. When the APBs are grouped, or in general discussion, Early means at or before VB, and includes the CF at the start of the accident and CF at VB.
' Evaluation of the APET produced 284 bins for this group, of which, 46 are required to capture 95% of the probability. The 15 most probable bins capture 75% of the probability. Six of the 15 most probable bins have CF at the start due to SG or RCP support failures; four have no VB and no CF, three have VB but no CF, and two have VB and BMT. The two most probable bins;have no VB and No CF. They result from the PDSs with LPIS operating at UTAF.
In. the LOCA, high acceleration PDS group there is the possibility of arresting the core degradation process and avoiding VB since four of the 11 PDSs in the group have the LPIS operating at UTAF. If the RCS can be depressurized sufficiently in a timely manner, injection by the LPIS may , arrest core damage for these PDSs. The mean probabilities for this PDS l group for the four pressure ranges at UTAF and just before VB (if it l occurs) are: l 2.116
Table 2.5-17 Results of the Accident Progression Analysis for Surry Seismic Initiators. ~PDS Group EQ 3, IACAs: High Acceleration, LLNL Hazard Distribution
. No . CF RCS VB Amt .Zr CF Order Bin Tr9b* Occur. Time Soravs. CCI Fres. Mode _ CCI QE HPME Size Fifteen Most Probable Bins" 1 HHCDFCDBDFB~ 0.110 123 No-CF Never No-CCI LcPr No-VB No-CCI Hi No No-CF 2 HHCDFCDADFB~ 0.107 111' No-CF Never No-CCI IoPr. No-VB No-CCI ~Lo No No-CF 3 CHADBCAADCB 0.080 112 Early Never PrmDry Iofr Four Large Lo No Irak 4 CHADBCABDCB 0.066 120. Early Never PrmDry IoPr. Pour Large HL No leak 5 HHADBCAADFB 0.058 112 No-CF Never PrmDry LoPr Four Large Lo No No-CF u
h 6 HHADBCABDFB- 0.050 123 No-CF Never PrmDry IoPr Pour large Hi No No-CF , u 7 CDCDBCDADCB 0.045 112 Early Always No-CCI IoPr Pour No-CCI Io No Leak 8 GHADBCAADDB 0.039 112 Final Never PrmDry IoPr- Pour Large Io No BMT 9 GHADBCABDDB 0.034 123 Final Never PrmDry LoPr Pour Large Hi No BMT 10 HDCDFCDBDFB 0.032 123 No-CF Always No-CCI LoPr No-VB No-CCI Hi No No-CF 11 CDCDBCDBDCB 0.030 95 Early Always No-CCI loPr Pour No-CCI Hi No Leak - 12 CDDDBCAADCB 0.030 _112 Early Always. PrmDeep IoPr Pour Large Lo No Leak 13 HDCDFCDADFB 0.026 111 No-CF Always No-CCI IoPr No-VB No-CCI le No No-CF 14 HDCDBCDADFB 0.020 112 No-CF Always No-CCI LoPr Pour No-CCI Io No No-CF 15 CDDDBCABDCP' O.019 93 Early Always PrmDeep LoPr Pour Large Hi No Leak A listing of all bins, and a listing by observation are available on computer media. Mean probability conditional on the occurrence of the PDS. l l l l l
At UTAF Just Before VB SSPr (2500 psia) 0.0 0.0 HiPr (600 to 2000 psia) 0.0 0.0 ImPr (200 to 600 psia) 0.42 0.01 LoPr (< 200 psia) 0.58 0.99 As all of the PDSs in tt.ts group have an A. S, or S2 break, the RCS i pressure at UTAF (Question 15) is bound to be low or intermediate. The S 2 breaks comprise abou*. 0.42 of this PDS group. As there is electric power available, and op' ning the PORVs is directed by the procedures when the core exit thermocouples show temperatures of 1200'F, it was estimated that the probability was 0.90 that the operators would open the PORVs in time to ensure the RCS was at low pressure by VB. The mean probability of having the-LPIS operating at UTAF is about 0.34 for the seismic thCA PDS group. Four of the 11 PDSs in this group have the LPIS operating. For these PDS, it is likely that the RCS will depressurize sufficiently and quickly enough to allow LPIS injection in time to prevent VB. The mean probability of arresting the core degradation process and
. avoiding VB is just over 0.30, The mean probability of CF at VB for this PDS group is 0.005. This is in addition to the 0.42 probability of CF at the start of the accident. This value is so large because five of the 11 PDSs in this group are "A" PDSs in which SC or RCP support failures at the time of earthquake fail the containment. If a rupture occurs at time zero, CF at VB is not considered l
since no further failure is possible. If a leak occurs at time zero, CF at l VB is considered since the fast pressure rise at VB may convert the leak into a rupture or catastrophic rupture. As CF at VB itself is very unlikely, the conversion of a leak from the start into a rupture at VB is also highly unlikely. There are no late or very late CFs due to hydrogen burns for the 14CA PDSs since power is continuously available. The hydrogen will burn when a flammable concentration is attained, which poses a negligible threat to the Surry containment. The mean probability of BMT is slightly less than 0.10.
.The sprays operate throughout the accident in five of the 11 PDSs in this group, and the sprays are irreparably failed in six - of the 11 PDSs. The l PDSs with failed sprays are more likely than those with operable sprays, so the mean probability for having no CHR is over 0.70. When CHR is not recovered in the long term, an overpressure failure of the containment is possible. As discussed in section'2.5.5.1, this is unlikely to occur. For the IDCA, high acceleration PDS group, the mean probability of eventual overpressure CF is 0.01.
2.5.5.4 Results ~ for PDS Croun E0 1. LOSP (No SBO)* Irv Acceleration. LIRL Hazard Distribution. This PDS group consists of accidents initiated by LOSP, but in which SB0 does not result. The 14SP is due to the earth-l quake, but the DCs start and supply station power. The five PDSs in this group are the same as those discussed in Section 2.5.5.1. This PDS group consists of the fraction of those accidents due to seisms with a maximum ground acceleration less than 0.6 g. 2.118
Table 2. 5-18 ' lis ts the 10 most probable APBs for this PDS group and the five most probable APBs that have VB and early CF. As the five most probable APBs that have VB are contained in the 10 most probable bins, these are. listed instead of repeating many of the APBs. Evaluation of the APET produced 803 bins for the LDSP, low acceleration PDS group. To cap-ture 95% of the probability, 93 bins are required. The 10 most probable bina capture 71% . of the probability, Four of the 10 most probable bins have no VB; the other six have low pressure pour failures of the lower , head. Two of the 10 most probable bins have BMT; the other eight have no CF. Of-the five most probable APBs with both VB and early CF, one has CF due to an Alpha mode failure of both vessel and containment, and four have CF due to HPME and DCl! with the RCS at high pressure. In the LOSP, low acceleration PDS group there is the possibility of arrest-ing- the core degradation process and avoiding VB since three of the PDSs
;have some ECCS operating at UTAF. If the RCS can be depressurized, inj ec-tion by_ the ECCS may arrest core damage for these PDSs. The mean probabi-lities for this PDS group for the four pressure ranges at UTAF and just before VB (if it occurs) are:
At UTAF Just Before VB SSPr (2500 psia) 0,80 0.025 HiPr (600 to 2000 psia) 0,0 0.007 ImPr (200 to 600 psia) 0,20 0.179 LoPr (< 200 pala) 0,0 0,789 Four of the five PDSs in-this group are "T" PDSs, which have the RCS-intact at the PORV -setpoint pressure at UTAF (Question 15), These PDSs account l .for the high probability for SSPr at UTAF, At VB (Question 23), the proba- , bility is - only about 0.025 that the- RCS will- be at that pressure. The effectiveness of the five mechanisms for depressurizing the RCS is the same-l~ for this PDS group as for the LOSP, high acceleration group (see Section
- 2.5.5.1),
Two of the five PDSs in this group have the LPIS operating at UTAF, and one had both LPIS and HPIS operating. For these PDSs, it is possible to arrest core damage and avoid vessel failure. Due to the depressurization of the RCS1before-VB, the mean probability of arresting the core degradation pro- , cess and avoiding VB is about - 0,38. The mean probability of having the LPIS operating at. UTAF is about 0.51, but not all of the time will the open PORVs depressurize the RCS far enough and fast enough to allow LPIS injec-tion.in time to prevent VB. The probability that both LPIS and.HPIS will be, operating is about 0.15. If the HPIS is operating, any break at all ! will depressurize the RCS enough ' to allow injection. However, there is no
-assurance that this injection will occur soon enough or that enough water will-be injected to prevent VB.
2.119
Table 2.5-18 Results of 'the Accident Progression Analysis for Surry Seismic Initiators. PDS Group EQ'1, LOSP (No'SBO): Low Acceleration, LLNL Hazard Distribution No. CF .. RCS VB Amt Zr CF Order Bin Prob.* Occur. Time Sorays CCI Pres. Mode CCI 03 HENE Size Ten Most Probable Bins ** , 1 HHCDFCDBDFB 0.148 123 No-CF- 'Never No-CCI' LoPr No-VB No-CCI Hi No No-CF 2 HHADBCABDFB 0.109 123 'No-CF Never PrmDry LoPr Four Large Hi No No-CF 3 HHCDFCDADFB 0.090 111 No-CF Never No-CCI LoPr No-VB No-CCI Lo No No-CF 4 HHADBCAADFB. 0.080 112' No-CF Never _PrmDry LoPr Pour Large Lo No No-CF 5 HDCDFCDBDFB 0.077 123 No-CF' Always No-CCI LoPr No-VB No-CCI Hi No No-CF w L 6 GHADBCABDDB 0.074 123 ' Final Never PrmDry LoPr Pour Large Hi No BMT $ 7 GHADBCAADDB- 0.055 112 Final Never PrmDry LoPr Pour Large Lo No BMT 8 HDCDFCDADFB 0.044 111 No-CF. Always No-CCI LoPr No-VB No-CCI Lo No No-CF 9 HDCDBCDBDFB 0.018 123 No-CF Always No-CCI LoPr Pour No-CCI Hi No No-CF-10 HHACBCABDFB 0.012- 8 No-CF Never PrmDry ImPr Pour Large Hi No No-CF Five Most Probable Bins that have VB and Early CF** 31 DHACACABACB 0.0039' 1 CFatVB Never PrmDry ImPr. HPME Large Hi Hi Leak 43 DHADDCBBDBB 0.0028 74. _ CFatVB Never- PrmDry LoPr Alpha Medium Hi No Rupture 92 DHEAACAABCA 0.0007 2 CFatVB Never. SDldDry SSPr HPME Large Lo Md Leak 114 DHEAACAAABA- 0.0005 4 CFatVB Never SDldDry SSPr HPME Iarge Io Hi- Rupture 167 .DHFCACABACB 0.0003 1 - CFatVB Never LDidDry ImPr HPME Large- Hi HL Leak Mean probability conditional on the occurrence of the PDS. A listing .of'all bins, and a listing by observation are available on computer media. ,
l The mean probability of CF at VB for this PDS group is 0.011. There are no late or very late CFs due to hydrogen burns for the LOSP PDSs since power is available throughout the accident. The mean probability of BMT is 0.19. The sprays are irreparably failed for the most probable three PDSs in this group. When CHR is not recovered in the long term, an overpressure failure of the containment is possible. As discussed in Section 2.5.5.1, this is unlikely to occur. The mean probability of eventual overpressure CF for the 14SP, low acceleration PDS group is 0.02. 2.5.5.5 Results for PDS Group EO 2. SB0: Low Acceleration. LIEL Hazard Distribution. This PDS group consists of accidents initiated by LOSP in which SB0 follows. The LOSP is due to the earthquake, and the DCs fail to start due to seismic and random hardware failures. There is no recovery of offsite power, and thus no chance of arresting core damage. The PDSs in this group are the same as those described in Section 2.5.5.2. This PDS group consists of the fraction of these accidents due to seisms with a maximum ground acceleration less than 0.6 g. Table 2.5-19 lists the 10 most probable APBs for the PDS group and the five most probable AFBs that have VB and early CF. In the "CF Time" column, Early means at the start of the accident, and is used to distinguish _"C" bins from "D" bins which have CF at VB. Since there is no possibility of core damaga arrest for the seismic SB0 PDS group, all the APBs have VB, 1 So, the 10 most probable APBs are listed instead of the five most probable and the five most probable that have VB. The 10 most probable bins capture 54% of the probability. Evaluation of the APET produced 492 bins for this group, of which, 112 are required to capture 95% of the probability. One of the 10 most probable bins has CF at the start due to SG or RCP support failures, four have BMT, and the'other five have no CF. Of the five most probable APBs with both VB and early GF. four have CF at time zero, the fifth has CF at VB due to an Alpha mode event. In the SBO, low acceleration PDS group the.e is no possibility of arresting the core degradation process and avoiding VB since offsite power is not recoverable in the time period consideted in this analysis. The depres-surization of the RCS during core melt serves only to reduce the loads placed upon the containment at VB. The mean probabilities for this PDS group for the four pressure ranges at UTAF and just before VB are: At UTAF Just Before VB SSPr-(2500 psla) 0.56 0.01 HiPr (600 to 2000 psia) 0.0 0.18 ImPr.(200 to 600 psia) 0.38 0.22 LoPr (< 200 psia) 0.06 0.59 Two of the three most probable PDSs in this group are "T" PDSs, so the RCS is intact at at the PORV setpoint pressure at UTAF (Question 15) for these PDSs. In the two "S 3" PDSs , the AFWS operates until the batteries deplete and the secondary system is depressurized as well, so there 'is no possibi-lity that the RCS will be in the high pressure range at UTAF for these PDSs. At VB (Question 23), the probability is only about 0.01 that the RCS will be at the system setpoint pressure. The Surry APET considers five 2.121
Table 2.5-19 Results o'f the Accident Progression Analysis for Surry Seismic-Initiators. PDS Group EQ 2, SB0: Low Acceleration, LLNL Hazard Distribution RCS VB Amt Zr CF No. CF .- Size Time Soravs CCI Pres. Mode CCI 93 HPME Order. Bin Prob.* Occur. Ten Most Probable Bins ** Never PrmDry LoPr- Pour Large Hi No No-CF 1 HHADBCABDFB 'O.112 123 No-CF Never PrmDry loPr Pour Large le No No-CF 2 HHADBCAADFB 0.080 112 No-CF LoPr Pour Large Hi No BMT 3 GHADBCABDDB 0.076 123 Final Never PrmDry Never PrmDry -LoPr Pour Large lo No BMT 4 GHADBCAADDB: 0.054 112 Final No-CF No-CF Never PrmDry LoPr Pour Large Lo No 5 HHADBCAADFA 0.050 38 w Hi No No-CF 0.039 33 No-CF Never PrmDry LoPr Pour Large L 6 HHADBCABDFA LoPr Pour Large Lo No Leak E! 7 CHADBCAADCB 0.038 112 Early Never PrmDry
- PrmDry Pour Large Lo No BMT 8 GHADBCAADDA 0.034 38 Final Never IoPr Never PrmDry ImPr HPME Large Hi Md No-CF 9 HHACACABBFA 0.032 16 No-CF Never PrmDry LoPr Pour Large Hi No BMT 10- CHADBCABDDA 0.026 33 Final Five Most Probable Bins that have VB and Early CF**
Never PrmDry LcPr Four Large Lo No Leak 7 CHADBCAADCB 0.038 112 Early No Leak Never PrmDry LoPr Pour Large Hi 11 CHADBCABDCB - 0.017 120 Early No Rupture Never PrmDry LoPr Four Large Lo 44 CHADBCAADBB 0.0043 112 Early No Rupture 46 DHADDCBBDBB 0.0042 74 CFatVB Never PrmDry LoPr Alpha Medium Hi Never PrmDry HiPr HPME Lnte Hi Md Leak 55 DHABACABBCA 0.0035 1 CFatVB' Mean probability conditional on the occurrence of the PDS.
** A listing jif all bins, and a listing-by observation are available on computer media.
_ _ . . . . . = . . mechanisms for depressurizing the RCS. RCP seal failures, the PORVs sticking open, and TI hot leg breaks are effective for the SBO, low acceleration PDS group. RCP seal failures are important for this group since SB0 causes a loss of seal cooling. Deliberate opening of the PORVs by the operators is prohibited by the procedures since there is no electric power. The mean probability of CF at VB for this PDS group is 0.019. This is in addition to the 0.06 probability of CF at the start of the accident. If a rupture occurs at time zero, CF at VB is not considered since no further failure is possible. If a leak occurs at time zero, CF at VB is considered since the fast pressure rise at VB may convert the leak into a rupture or catastrophic rupture. As CF at VB itself is very unlikely, the conversion of a leak from the start into a rupture at VB is highly unlikely. There are no late CFs due to hydrogen burns for the LOSP PDSs since power is never recovered during this period. It is possible to have hydrogen burns in the very late period when the containment slowly cools off due to heat transfer to the containment, and enough steam condenses to de inert the containment atmosphere. The mean probability of very late CF due to a burn is slightly less than 0,04. The mean probability of BMT is 0.30 for this PDS group. The lack of sprays, and thus CHR, for this PDS group implies that an overpressure failure of the containment may occur af ter several days. As discussed in Section 2.5.5.1, the Surry basem a is constructed of silicoous concrete and this is valikely to occur. The mean probability of eventual overpressure CF is let a than 0.04 for the SBO, low acceleration PDS group. 2.5.5.6 Results for PDS Group EO 3. LOCAs: Low Acceleration. LLNL Hazard histribution. This PDS group consists of accidents initiated by seismic pipe breaks. The failures in the ECCS required to respond to these breaks .tre partially seismic ano partially random. There is no SBO, but some of the seismic . fallures are failures in che electrical distribution system, specifically in the parts that supply. power to the ECCS, sprays, or AWS, In these situations, the availability of electrical power, fourth PDS characteristic, is denoted by a "P". This PDS group consists of the fraction of these accidents due to seisms wita a maximum ground acceleration less than 0.6 g. The 11 PDSs in this group are listed in Table 2.2 6 and discussed in Section 2.5.5.3 Table 2.5 20 lists the 15 most probable APBs for the PDS group. The overlap between the five most probable bins, the five most probable bins that have VB, and the five most probable APBs that have VB and early CF is so great that listing the 15 most probable bins provides more information. Evaluation of the APET produced 285 bins for this group, of which, 45 are required to capture 95% of the probability. The 15 most probable bins capture 78% of the probability. Five of the 15 most probab?3 bins have CF at the start due to SC or RCP support failures; four have no VB and no CF,
'.our have VB but no CF, and two have VB and BMT. The two most probable bins have no VB and No CF. They result from the PDSs with LPIS operating at UTAF.
l 2.123
.w-
~ Table 2.5-20
.Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Group EQ 3, IDCAs: law Acceleration, LIRL Hazard. Distribution No. CF RCS VB Amt Zr CF order Bin Prob.* Occur. Time Sorays CCI Pres. Mode CCI QK HPME Size Fifteen Most Probable Bins" 1 HHCDFCDBDFB- 0.123 123 No-CF- Never No-CCI IoPr No-VB No-CCI Hi No No-CF 2 HHCDFCDADFB 0.116 111 No-CF 'Never No-CCI. LoPr No-VB No-CCI In No No-CF 3 CHADBCAADCB 0.080 112 Early Never PrmDry LoPr Pour Large lo No Ioak 4 HHADBCAADFB 0.076 .112 No-CF. Never. PraDry IoPr Pour Large lo No No-CF 5 HHADBCABDFB 0.069 123 No-CF Never PrmDry IoPr Four Large Hi No No-CF ;
m L 6 CHADBCABDCB 0.063 120 Early. Never PrmDry LoPr Pour Large Hi No Leak 7 7 - GHADBCAADDB 0.052 112 Final Never PrmDry LoPr Pour Large In No BMT 8 GHADBCABDDB 0.047 123 Final- Never PrmDry LoPr Pour Large Hi No BMT 9 CDCDBCDADCB 0.034 111 Early Always No-CCI loPr Pour No-CCI Lo No Ioak 10 HDCDFCDBDFB 0.025- 123. No-CF Always No-CCI loPr No-VB No-CCI Hi No No-CF 11 CDDDBCAADCB 0.022 ~ 111 ~ Early Always PrmDeep IoPr Pour Large Lo No Leak 12 CDCDBCDBDCB 0.021 93 Early Always No-CCI IoPr Four No-CCI Hi No Leak 13 HDCDFCDADFB 0.020 111 No-CF Always No-CCI IoPr No-VB No-CCI Io No No-CF 14 HDCDBCDADFB 0.017 112 No-GF Always No-CCI IoPr Pour No-CCI Im No No-CF 15 HDCDBCDBDFB ~0.015 -123 No-CF Always No-CCI loPr Pour No-CCI Hi No No-CF Mean probability conditional on tl'e occurrence of.the PDS. A listing of all bins, and a liste 4 by observation are available on computer media.
, . . _ - _ _ _- . , - - . _ ~ _. . _ _ - _ _ _ _ - _ _ = - _ _ . - _ . .-
In the thCA, low acceleration PDS group there is the possibility of arrest- l ing the core degredation pocess and avoiding VB since four if the 11 PDSs ' in the group have .u LPIS operating at UTAF. If the RSS can be uepressur-ired suf ficiently it. timely manner, injection by t. . LPIS may arrest coro damage for these PDSs. The mean probabilities f a this PDS group for the four pressure ranges at UTAF and just before VB (if it occurs) are: ; At UTAF Just Before VB SSPr (2500 psia) 0.0 0.0 HiPr (600 to 2000 psta) 0.0 0.0 ImPr (200 to 600 psia) 0 $1 0.01 LoPr (< 200 psia) 0.49 0.99 As all of the PDSs in this group have an A, St or $2 break, the RCS pres-sure at UTAF (Question 15) is bound to be low or intermediate. The Sa breaks comprise about 0.51 of this PDS group. As there is electric power available, and opening the PORVs is directed by the procedures when the core exit thermocouples show temperatures of 1200'F, it was estimated that , the probabili<y was 0.90 that the operators would open the PORVs in time to ensure the RCS was at low. pressure by VB.
- The mean probability of having the LPIS operating at UTAF is about 0.34 for the seismic thCA PDS group. Four of the 11 PDSs in this group have the LPIS operating. For these PDS, it is likely that the RCS will depressurize sufficiently and quickly enough to allow LPIS injection in time to prevent VB. The mean probability of arresting the core degradation process and f
avoiding VB is about 0.31. The mean probability r f CF at VB for this PDS group is 0,006. This is in addition to the 0.35 probability of CF at the start of the accident. This value is s.o large because five of the 11 PDSs in this group are "A" PDSs in which SG or RCP support failures at the time of earthquake fail the containment. There are no late or very late CFs due to hydrogen burns for the LDCA PDSs since power. is continuously available. The hydrogen will burn when a flammable concentration is attained, which poses a negligible threat to the Surry containment. The mean probability of BMT 'is 0.12. The sprays operate throughout the accident in five of the 11 PDSs in this group, and the sprays are irrepa-rably failed in six of the 11 PDSs. The PDSs with failed sprays are more likely.than those with operable sprays, so the mean probability for having no CllR is over 0.78. When CHR is not recovered in the long term, an over-pressure failure of the containment may occur after several days. As dis-F cussed in Section 2.5.5.1, this is unlikely to occur. For the LOCA, low acceleration PDS: group, the mean probability of eventual overpressure CF is 0.01. 2.125 1
2.5.6 Sensitivity Annivses for S e l s nil e Initiatorst LLNL Hazard Distribution To determine the effect of the initial failures of the containment due to failures of the 50 and RCP supports, the integrated risk analysis using the LLNL hazard distribution was repeated without these Induced seismic failures of the containment at the start of the accident. Except for the initial CFs, the removal of these CFs at the start of the accident has very little effect on the accident progression analysis. There are slightly more failures computed at VB since the probability of the containment being intact at VB increases, and the probabilities of BMT and late above ground failures increase for the same reason. Table 2.5 21 compares the results for CF at VB and for final CF with and without the initial failures of the containment for seismic PDS groups EQ 2 (SBO) and EQ 3 (1DCA). Initial seismic CFs occur only in the "A" PDSs, and as PDS group EQ 1 (14SP, No SBO) contains no "A" PDSs, eliminating the initial seismic CFs has no effects on the APET results for this PDS group. The probability of initial CF does not go to zero in the sensitivity case because the isolation failures remain with a very low probability. 2.$ 7 Results for Seismic Initiators: EPRI Hazard Distribution The seismic risk analysis was performed using two different seismic hazard distributions. This section reports the results using the hazard distri-bution developed by EPRI. Results based on the seismic hazard distr (bution developed by LLNL were presented in Section 2.5.5. The differences between these two distributions are discussed in NUREG/CR 4550, Vol. 3, Part 3. The accidents initiated by earthquakes were analyzed in two groups. Those due to seisms with a maximum ground acceleration in excess of 0.6 g were denoted the high acceleration events. It was judged that the destruction in the general vicinity of the plant for those earthquakes would be so great that evacuation would be ineffective. For the low acceleration ' events, less than 0.6 g, it was estimated that evacuation would be possi-ble, although perhaps more slowly than in emergencies without earthquakes. 2.5.7.1 Results for PDS Grouo E0 1. LOSP (No SBO): High Acceleration. EPR7 Hazard Distribution. This PDS group consists of accidents initiated by LOSP, but in which SB0 does not result. The LOSP is due to the earth-quake, but the DGs start and supply station power. The five PDSs are dis-cussed in Section 2.5.5.1. This PDS group consists of the fraction of these accidents due to seisms with a maximum ground acceleration exceeding 0.6 g. l t l l l 2.126
i Table 2.5 21 ' l Comparison of the Accident Progression Analysis Results for Surry Seismic Initiators. With and Without Containment Failures at the Start of the Accident; LLNL llazard Distribution l High Acceleration Event Probability of Event ) Base Case . CF at t-O lina. Case . No CF at t-0 1 EQ 2 SB0 EQ 3 LOCA EQ 2 SB0 EQ 314CA Initial CF 0.122 0.420 0.0002 0.0002 CF at VB 0.0156 0.0054 0.0158 0.0069 Late CF 0.0 0.0 0.0 0.0
-Very Late CF 0.038 0.0 0.042 0.0 BMT 0.291 0.096 0.337 0.199 Eventual OP 0.036 0.010 0.041 0.020 No CF 0.496 0.472 0.564 0.774 i
Low Acceleration Event Probability of Event Base Case . CF at t-0 Sena. Case . No CF at t - 0 EQ 2 5B0 EQ 3 LOCA EQ 2 .SB0 EQ 3 LOCA 3 Initial CF 0.062 0.355 0.0002 0.0002 CF at VB 0.0189 0.0055 0.0190 0.0074 : Late.CF. . 0.0 0.0 0.0 0.0 < Very Late CF 0.038 0.0 0.039 0.0 .t l BMT 0.304 0.120 0.328 0.213 s Eventual OP 0.038 0.013 0.041 0.022 L No CF 0.538 0.509 0.572 0.757 , 7 s i 2.127
l Table 2.5 22 lists the 10 most probable APBs for this PDS group and the five most probable APBs that have VB and early CF. Evaluation of the APET 1 produced 708 bins for the LOSP, high acceleration PDS group. To capture 95% of the probability, 63 bins are required. The 10 most probable bins capture 74% of the probability. Four of the 10 most probable bins have no VB; five have low pressure pour failures of the lower head; one has llPME. Two of the 10 most probabic bins have BMT; the other eight have no CF. Of the five most probable APBs with both VB And early CF, three have CF due to an Alpha mode failure of both vessel and entainment, and two have CF due to llPME and DCil with the RCS at high pressure. In the LOSP, high acceleration PDS group, there is the possibility of arresting the core degradation process and avoiding VB since three of the PDSs have some ECCS operating at UTAF. If the RCS can be depressurized, injection by the ECCS may arrest core damage for these PDSs. The mean probabilities for this PDS greap for the four pressure ranges at UTAF and just before VB (if it occurs) are: ; At UTAF Just Before VB SSPr (2500 psia) 0.932 0.011 il1Pr (600 to 2000 psia) 0.0 0.003 ImPr (200 to 600 psia) 0.068 0.191 LoPr (< 200 psia) 0.0 0.796 Four of the five PDSs in this group are "T" PDSs which have the RCS intact at the PORV setpoint pressure at UTAF (Question 15). These PDSs account for the high probability for SSPr at UTAF. At VB (Question 23), the probability is only about 0.025 that the RCS will be at that pressure. The effectiveness of the five mechanisms for depressurizing the RCS is the same for the EPRI hazard curve as for the 1.LNL hazard curve for this PDS group (see Section 2.5.5.1).
.Two of the five PDS in this group have the LPIS operating at UTAF, and one had both LPIS and llPIS operating. For these PDSs, it is possible to arrest cote damage and avoid vessel failure. Due to the depressurization of the >
RCS before VB, the mean probability of arresting the core degradation process and avoiding VB is about 0.36. ihe mean probability of havi9F the LPIS operating at UTAF is about 0.51, but not all of the tinie will the open PORVs depressurize the RCS far enough and fast enough to allow LPIS injection in time to prevent VB. The probability that both LPIS and ilPIS will be operating is negligible for the portion of this PDS group that has the peak ground acceleration above 0,6 g. The mean probability of CF at VB for this PDS group it 0.011. There are no late or very late CFs due to hydrogen burns for the 1DSP PDSs since' power " was available throughout the accident. The experta considering hydrogen ignition concluded that with electrical power r.vailable continuously. L combustion would occur shortly after a flammable concentration of hydrogen j was attained in the containment. Only deflagrations are possible at the flammable limits, and these are judged to be extremely unlikely to fail the Surry containment. Thus, late and very late CFs due to combustion when electrical power has been continuously available are not considered. 2.128
Table 2.5-22 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Croup EQ 1. IDSP (No SBO): High Acceleration, EPRI Hazard Distribution No. CF RCS VB Amt Zr CF hskr Bin Prob
- Occur. Time Sorays CCI Pres. Mode CCI h HPME . Size Ten Most Probable Bins **
1 HHCDFCDBDFB 0.166 123 No-CF Never No-CCI IoPr No-VB No-CCI Hi No No-CF 2 HHADBCABDFB 0.139 123 No-CF Never PrmDry IoPr Four Large Hi No No-CF 3 HHCDFCDADFB 0.130 112 No-CF Never No-CCI LoPr No-VB No-CCI In No No-CF 4 CHADBCABDDB 0.095 123 Final Never PrmDry IoPr Pour Large Hi No BMT 5 HHADBCAADFB 0.072 112 No-CF Never PrmDry loPr Four Large Io No No-CF m
~
@ 6 CHADBCAADDB C.049 112 Final Never PrmDry loPr Pour Large la No BMT 7 HDCDFCDBDFB 0.036 123 No-CF Always No-CCI loPr No-VB No-CCI Hi No No-CF 8 HDCDFCDADFB 0.026 112 No-CF Always No-CCI LoPr No-VB No-CCI la No No-CF 9 HHACBCABDFB 0.017 8 No-CF Never PrmDry ImPr Pour Large Hi No No-CF 10 HHACACABCFB 0.015 7 No-CF Never PrmDry ImPr HPME Large Hi Im No-CF Five Most Probable Bins that have VB and Early CF**
24 DHADDCBBDBB 0.0043 72 CFatVB Never PrmDry IoPr Alpha Medium Hi No Rupture 25 DHACACABACB 0.0036 1 CFatVB Never PrmDry ImPr HPME Large Hi Hi Leak 88 DHADDCBADBB 0.0006 72 CFatVB Never PrmDry LoPr Alpha Medium Lo No Rupture 97 DHFCACABACB 0.0004 1 CFatVB Never LDidDry ImPr HPME Large Hi Hi Ioak 102 DAFDDCBBDBB 0.0004 50 CFatVB Never LDidDry LoPr Alpha Medium Hi *o
. Rupture Mean probability conditional on the occurrence of the PDS.
m listing of all bins, and a listing by observation are available on computer media.
i The mean probability of BMT is 0.21. The three most probable PDSs of the five PDSs in this group have the sprays irreparably failed. For the PDSs, the lack of CHR may cause an overpressure failure of the containment after several days. As discussed in Section 2.5.5.1, the Surry basemat is con-structed of siliceous concrete and this is unlikely to occur. The mean probability of eventual overpressure CF for the 14SP, high acceleration PDS group is 0.025, 2.5.7.2 Results for PDS Croup E0 2. SB0: High Acceleration. EPRI
- Hazard Distribution. This PDS group consists of accidents initiated by LOSP in which SB0 follows. The 1DSP is due to the earthquake, and the DCs fail to start due to seismic and randois hardware failures. Due to the seismic failures in the electrical distribution system that may be expect- :
ed, it was judged that offsite power would not be recovered within the timeframe of this analysis. Thus there is no chance of arresting core < damage or avoiding VB in this PDS group. The eight PDSs in this group are discussed in Section 2.5.5.2. This PDS group consists of the fraction of these accidents due to seisms with a maximum ground acceleration exceeding 0.6 g. . Table 2.5 23 lists the 10 most probable APBs for the PDS group and the five most. probable APBs that have VB and early CF. In the "CF Time" column in Table 2.5 23, Early means at the start of the accident, and is used to distinguish "C" bins i. om "D" bins which have CF at VB. As the offsite consequences and risk are much the same whether the containment fails at time zero or at VB, when the APBs are grouped, or in general discussion, r Early means at or before VB, and so would include the CFs at the start of the accident. Since there is no possibility of core damage arrest for the seismic SB0 PDS group, all the APBs have - VB, so the 10 most probable APBs are listed. Evaluation of the APET produced 483 bins for this group, of which, 117 are required to capture 95% of the probability. Two of the 10 most probable bins have CF at the start due to SC or RCP support failures, three have BMT, and the other five have no CF. The 10 most probable bins capture 586 of the probability. Of the five most probable APBs with both VB and early CF, three have CF at time zero; one has CF at VB due to an Alpha mode event; and one has CF at VB due to HPME. In the SBO, high acceleration PDS group there is no possibility of arrest. Ing the core degradation process and avoiding VB since offsite power is not recoverable in the time period considered in this analysis. The depressu. rization of the RCS during core melt serves only to reduce the loads placed upon the containment at VB. The mean probabilities for this PDS group for the four pressure ranges at UTAF and just before VB are: (: At lTTAF- Just Before VB SSPr (2500 psia) 0.60 0.02 HiPr (600 to 2000 psia) 0.0 0.13 ImPr (200 to 600 psia) 0.31 0.21 LoPr (< 200 psia) 0.09 0.64 2.130
Table 2.5-23 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Group EQ 2, SB0: High Acceleration, EPRI Hazard Distribution No. CF RCS VB Amt Zr CF Prob.* Occur. _ Time. Soravs CCI Pres. Mode - CCI Qs HFME Size Order Bin Ten Most Probable Bins ** 0.125 123 No-CF Never PrmDry IoPr Four Large Hi No No-5 1 HHADBCABDFB 0.096 112 No-CF Never FraDry IoPr Four Large Lo No No- G 2 HHADBCAADFB 3 GHADBCABDDB 0.085 123 Final Never FraDry IoPr Pour Large HI No BMT 4 GHADBCAADDB 0.065 112 Final Never PrmDry IoPr Pour Large In No BMT No-CF Never PrmDry IoPr Four Large In No No-CF 5 HHADBCAADFA 0.045 38 6 CHADBCAADCB 0.044 112 Early Never FraDry IoPr Pour Large In No Isak co ~ 0.036 121 Early Never PrmDry IoPr Four Large Hi No Leak ~. 7 CHADBCABDCB $ 8 GHADBCAADDA 0.031 38 Final Never PrmDry IoPr Pour Large lo No BMT 0.029 33 No-CF Never PrmDry loPr Four Large Hi No No-CF 9 HHADBCABDFA 0.021 16 No-CF Never PrmDry ImPr HPME Large Hi Md No-CF 10 HHACACABBFA Five Most Probable Bins that have VB and Early CF** 0.044 112 Early Never PrmDry IoPr Four Large lo No Leak 6 CHADBCAADCB 7 CHADBCABDCB 0.036 121 Early Never PrmDry IoPr Four Large Hi No Ieak 36 DHADDCBBDBB 0.0050 72 CFatVB Never PrmDry IoPr Alpha Medium Hi No Rupture 37 CHADBCAADBB 0.0049 112 Early Never PrmDry LoPr Pour Large In No Rupture 49 DHACACABACB 0.0040- 1 CFatVB Never PrmDry ImPr HPME Large Hi Hi Irak Mean probability conditional on the occurrence of the PDS.
** A listing of all bins, and a listing by observation are available on computer media.
Two of the three most probable PDSs in this group are "T" PDSs, so the RCS is intact at the PORV setpoint pressure at UTAF (Question 15) for those PDSs. In the two "S "3 PDSs, the APWS operates until the batteries deplete and the secondary system is depressurized as well, so the RCS will not be in the high pressure range at UTAF for those PDSs. At VB (Question 23), the probability is only about 0.02 that the RCS will be at the system setpoint pressure. The Surry APET considers five mechanisms for depressu. rizing the RCS as discussed in Section 2.$.5.2. Note that the fractions in each pressure range are not identical with those in Section 2.5.5.2. The relative frequencies of the "T", "S", 3 dS", 3 and *A" PDSs depend on the hazard distribution. The mean probability of CF at VB for this PDS group is 0.019. This is in addition to the 0.09 probability of CF at the start of the accident. There are no late CFs due to hydrogen burns for the LOSP PDSs since power is never recovered during this period. It is possible to have hydrogen burns in the very late period when the containment slowly cools off due to heat transfer to the containment, and enough steam condenses to de inert the containment atmosphere. The mean probability of very late CF due to a burn is slightly less than 0.04. The mean probability of BMT is 0.30. The lack of sprays, and th9s CHR, implies that an overpressure failure of the containment may occur after several days. As discussed in Section 2.5.5.1, this is unlikely to occur. The mean probability of eventual overpressure CF is less than 0.04 for the SBO, high acceleration PDS group. 2.5.7.3 Results for PDS Croun EO 3. LOCAst llich Acceleration. EPRI linzard Dist ibution. This PDS group consists of accidents initiated by seismic pipe breaks. The failures in the ECCS required to respond to these breaks are partially seismic and partially random. There is no SBO, but some of the seismic failures are failures in the electrical distribution system, specifically in the parts that supply power to the ECCS, sprays, or AFWS. In these situations, the availability of electrical power, fourth PDS characteristic, is denoted by a "P" . The 11 PDSs in this group are discussed in Section 2.5.5.3. This PDS group consists of the fraction of these accidents due to seisms with a maximum ground acceleration exceeding 0.6 g. Table 2.5 24 lists the 15 most probable APBs for the PDS group. Evaluation of the APET produced.266 bins for this group, of which, 46 are required to capture 95% of the probability. The 15 most probable bins capture 76% of the probability. Six of the 15 most probable bins have CF at the start due to SG or RCP support failures; four have no VB and no CF, three have VB but no CF, and two have VB and BMT. The two most probable bins have no VB and no CF. They result from the PDSs with LPIS operating at UTAF. In the lhCA, high acceleration PDS group there is the possibility of ar. resting the core degradation process and avoiding VB since four of the 11 PDSs in the group have the hPI'.i operating at UTAF. 2.132
Table 2.5-24 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Croup EQ 3, IDCAs: High Acceleration, E?RI Hazard Distribution RCS VB Amt Zr CF No. CF Time - Sorays CCI Pres. Mode CCI Qg HPME Size Order Bin Prob.* Occur. Fifteen Most Probable Bins" No-CF Never No-CCI loPr No-VB No-CCI in No No-CF 1 HHCDFCDADFB 0.119 112 No-CF No-CF Never No-CCI LoPr No-VB No-CCI Hi No 2 HHCDFCDBDFB 0.108 123 PrmDry LoPr Pour Large Io No leak 3 CHADBCAADCB 0.077 112 Early Never PrmDry LcPr Four Imrge Hi No leak 4 CHADBCABDCB 0.071 121 Early Never No-CF No-CF PrmDry IoPr Four Large In No 5 HHADBCAADFB 0.064 112 Never w Hi No Wo-CF 0.060 123 No-CF Never PrmDry IoPr Pour large L 6 HHADBCABDFB large In No BMT U 7 GHADBCAADDB 0.044 112 Final Never PrmDry IoPr Pour PrmDry IePr Four Iarge Hi No BMT 8 GHADBCABDDB 0.041 123 Final Never Isak Always No-CCI IoPr Pour No-CCI Io No 9 CDCDBCDADCB 0.036 111 Early No-CF No-CF Always No-CCI IoPr No-VB No-CCI to No 10 HDCDFCDADFB 0.034 112 Always No-CCI IePr Four No-CCI Hi No Leak 11 CDCDBCDBDCB 0.028 91 Early Isak Always PrmDeep IoPr Four Large In No 12 CDDDBCAADCB 0.023 111 Early No-CF Always No-CCI IoPr No-VB No-CCI Hi No 13 HDCDFCDBDFB 0.023 123 No-CF Always No-CCI IePr Four No-CCI In No No-CF 14 HDCDBCDADFB 0.018 112 No-CF Always PrmDeep LoPr Pour Large Hi No Isak 15 CDDDBCABDCS 0.018 90 Early Mean probability conditional on the occurrence of the PDS.
" A listing of all bins and a listing by observation are available on computer media.
aA. ll
If the RCS can be depressurized sufficiently in a timely manner, injection by the LPIS may arrest core damage for these PDSs. The mean probabilities for this PDS group for the four pressure ranges at UTAF and just before VB (if it occurs) are: At UTAF Just Before VB SSPr (2500 psia) 0.0 0.0 liiPr (600 to 2000 psia) 0.0 0.0 ImPr (200 to 600 psia) 0.44 0.01 ; LoPr (< 200 psia) 0.56 0.99 , As all of the PDSs in this group have an A, S, 3 or Sg break, the RCS pressure at UTAF (Qunstion 15) is bound to be lov or intermediate. The S breaks comprise 0.44 of this PDS group. As there is electric power available, and opening the PORVs is directed by the procedures when the core exit thermocouples show temperatures of 1200*F, it was estimated that the probability was 0.90 that the operators would open the PORVs in time to ensure the RCS was at low pressure by VB. i The mean probability of having the LPIS operating at UTAF is abr i for $ the seismic IDCA PDS group. Four of the 11 PDSs in this '
. ave the :'
LPIS operating. For these PDSs, it is likely that the RC .1 depres-surize sufficiently and quickly enough to allow LPIS inject in-time to > prevent VB. The mean probability of arrestin 6 the core deg e uion process and avoiding VB is about 0.31. The mean probability of CF at VB for this PDS group is 0.007. This is in addition to the 0.39 probability of CF at the start of the accident. This value is so large because five of the 11 PDSs in this group are "A" PDSs in which SG or RCP support failures at the time of earthquake fail the containment. There are no late or very late CFs due to hydrogen burns for the IDCA PDSs since power is continuously available. The mean probability of BMT is 0.11 The sprays operate throughout the accident in five of the 11 PDSs in this group, and the sprays' are irreparably failed in six of the 11 PDSs.- The PDSs with failed sprays are more likely than those with operable sprays, so the mean probability for having no.CllR is about 0.75. When CllR ' is not recovered in the long term -an overpressure failure of the contain-ment may occur af ter several days. As discussed in Section 2.5.5.1, non- : recovery of CllR in a period on the order of five days was judged to be of ; low probability, even following an earthquake.. Thus, for the LOCA, high i acceleration PDS group the mean probability of eventual overpressure CF is 0.01, t 2.5.7.4 Results for PDS-Grouo E0 1. LOSP (No SBO): Low Acceleration. ' l EPRI ltarard Distribution. This PDS group consists of accidents initiated t- by LOSP, but in which SB0 does not result. The LOSP is due to the earthquake, but the DGs start and supply station power. The five PDSs in this group are the same as those discussed in Section 2.5.5.1. This PDS .' group consists of the fraction of these accidents due to seisms with a maximum ground acceleration less than 0.6 g. 2.134
Table 2.5 25 lists ths 10 most probable APBs for this PDS group and the five snost probable APBs that have VB and early CF. Evaluation of the APET produced 789 bins for the 14SP, low acceleration PDS group. To capture 95% of the probability, 92 bins are required. The 10 most probable birm cap-ture 71% of the probability. Four of the 10 most probable bins have no VB; five have low pressure pour failures of the lower head; and one has HPME. Two of the 10 most probable bins have BMT; the other eight have no CF. Of the five most probable APBs with both VB and early CF, three have CF due to an Alpha mode failure of both vessel and containment, and two have CF due to llPME and DCH with the RC9 at high pressure. In the LOSP, low acceleration PDS group, there is the possibility of arresting the core degradation process and avoiding VB since three of the PDS have some ECCS operating at UTAF. If the RCS can be depressurized, injection by the ECCS may arrest core damage for these PDSs. The mean probabilities for this PDS group for the four pressure ra*1ges at UTAF and just before VB (if it occurs) are: At UTAF Just Before VB SSPr (2500 psia) 0.77 0.029 Il1Pr (600 to 2000 psia) 0.0 0.008 ImPr (200 to 600 psia) 0.23 0.179 LoPr (< 200 psia) 0.0 0.784 Four of the five PDSs in this group are "T" PDSs, which have the RCS intact at the PORV setpoint pressure at UTAF (Question 15). These PDSs account for the high probability for SSPr at UTAF. At VB (Question 23), the proba-bility is only about 0.025 that the RCS will be at that pressure. The effectiveness of the five mechanisms for depressurizing the RCS is the same for this PDS group as for the LOSP high acceleration group (see section 2.5.5.1). Two of the five PDSs in this group have the LPIS operating at UTAF, and one had both LPIS and HPIS operating. For these PDSs. it is possible to arrest core damage and avoid vessel failure. Due to the depressurization of the RCS before VB, the mean probability of arresting the core degradation pro-cess and avoiding VB is about 0.34. The incan probability of having the LPIS operating at UTAF is about 0.31, but not all of the time will the open PORVs depressurize the RCS far enough and fast enough to allow LPIS injec-tion in time to prevent VB. The probability that both LPIS and HPIS will be operating is about 0.15. If the HPIS is operating, any break at all vill depressurize the RCS enough to allow injection. However, there is no assurance that this injection will occur soon enough, or that enough water will be injected, to prevent VB. The mean probability of CF at VB for this PDS group is 0.012. There are no late or very late CFs due to hydrogen burns for the LOSP PDSs since power is available throughout the accident. The sprays are irreparably failed for the most probable three PDSs in this group. The mean probability of BMT is 0.21. The mean probability of eventual overpressure CF for the LOSP, low acceleration PDS group is 0.02. 1 2.135
T&ble 2.5-25 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Group EQ 1, IDSP (No SBO): Iow Acceleration, EPRI Hazard Distribution No. CF RCS VB Amt Zr CF Order Bin Prob.* Occur. Time _Ecrays CCI Pres. Mode CCI Qg HZHE Size Ten Most Probable Bins ** 1 HHCDFCDBDFB 0.129 123' No-CF Never No-CCI loPr No-VB No-CCI Hi No No-CF 2 HHADBCABDFB 0.129 123 No-CF Never PrmDry loPr Pm.tr Large Hi No No-CF 3 CHADBCABDDB 0.088 123 Final Never PrmDry IoPr Pour Large Hi No BMT i 4 HHCDFCDADFB 0.086 112 No-CF Ne.ver No-CCI loPr No-VB No-CCI Lo No No-CF 5 HHADBCAADFB 0.082 112 No-CF Never PrmDry IoPr Four Large Im No No-CF
," 6 HDCDFCDBDFB 0.059 123 No-CF Always No-CCI loPr No-VB No-CCI Hi No No-CF
[ 7 GHADBCAADDB 0.056 112 Final Never PraDry IoPr Pour Large In No BMT
- 8 0.047 HDCDFCDADFB 112 No-CF Always No-CCI loPr No-VB No-CCI Lo No No-CF 9 HDCDBCDBDFB 0.018 123 No-CF Always No-CCI LoPr Pour No-CCI Hi No No-CF 10 .HHACACABCFB 0.013 7 No-CF Never PrmDry ImPr HPME Large Hi la No-CF Five Most Probable Eins that have VB and Early CF** '
29 DHACACABACB 0.0041 1 CFatVB Never PrmDry ImPr HPME Large Hi Hi Leak 31 DHADDCBBDBB 0.0036 75 CFatVB Never PrmDry Alpha IoPr Medium Hi No Rupture ! 69 DHADDCBADBB 0.0012 72 CFatVB Never PrmDry Alpha loPr Medium Io No Rupture 106 DHEAACAAABA 0.0006 4 CFatVB Never SD1dDry SSPr HPME Iarge lo Hi Rupture 126 DAFDDCBBDBB 0.0004 51 CFatVB Early LDldDry IoPr Alpha Medium Hi No Rupture Mean probability conditional on the occurrence of the PDS. ! A listing of all bins, and a listing by observation are available on computer media.
2.5.7.5 Results for PDS Groun EO 2. SB0: Low Acceleration. EPRI IInz ard Distribution. This PDS group consists of accidents initiated by thSP in which SB0 follows. The LOSP is due to the earthquake, and the DGs fall to start due to seismic and random hardware failures. There is no recovery of offsite power, and thus no chance of arresting core damage. The PDSs in this group are the same as those described in Section 2.5.5.2. This PDS group consists of the fraction of these accidents due to seisms with a maximum ground acceleration less than 0.6 g. Table 2.5 26 lists the 10 most probable APBs for the PDS Broup and the five most probable APBs that have VB and early CF. In the "CF Time" column. Early means at the start of the accident, and is used to distinguish "C" bins from "D" bins which have CF at VB. Since there is no possibility of core damage arrest for the seismic SB0 PDS group, all the APBs have VB. The 10 most probable bins capture 526 of the probability. Evaluation of the APET produced 478 bins for this group, of which, 119 are required to capture 95% of the probability. One of the 10 most probable bins has CF at the start due to SG or RCP support failures, four have BMT, and the other five have no CF. Of the five most probable APBs with both VB and early CF, two have CF at time zero; one has CF at VB due to an Alpha mode vent, one has CF at VB due to llPME at intermediate pressure, and one has CF at VB due to gross bottom head failure at intermediate pressure. The llPME failure of the vessel and the complete failure of the bottom head both result in DCil. In the llPME failure mode, the hole results from the melt through of one or a few penetrations. In the SBO, low acceleration PDS group there is no possibility of arresting the core degradation process and avoiding VB since offsite power is not recoverable in the time period considered in this analysis. The depres-surization of the P.CS during core melt serves only to reduce the loads placed upon the containment at VB. The mean probabilities for this PDS group for the four pressure ranges at UTAF and just before VB are'. At UTAF Just Before VB SSPr (2500 psia) 0.53 0.02 liiPr (600 to 2000 psia) 0.0 0.18 ImPr (200 to 600 psia) 0.43 0.24 LoPr (< 200 psia) 0.04 0.57 Two of the three most probable PDSs in this group are "T" PLSs, so the RCS is intact at at the PORV setpoint pressure at UTAF (Question 15) for these PDSs. In the two "Sa" PDSs. the AFWS operates until the batteries deplete and the secondary system is depressurized as well, so the RCS will not b: in the high pressure range at UTAF for these PDSs. At VB (Qut.stion 23), the probability is only about 0.02 that the RCS will be at tte system setpoint pressure. The depressurization mechanisms are the same as those discussed in Section 2.5.5.2. Note that the fractions in each pressure range are not identical with those in Section 2.5.5.5. The relative frequencies of the "T", "Sa". "S", 2 and "A" PDSs depend on the hazard distribution. 2.137
Table 2.5-26 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Group EQ 2, SB0: Iow Acceleration. EPRI Hazard Distribution No. CF RCS VB Amt Zr CF Order Bin Prob,* Occur. Time Sorays CCI Pres. Mode CCI Og HPHE Size Ten Most Probable Bins ** 1 HHADBCABDFB 0.101 123 No-CF Never PrmDry LoPr Four Large Hi No No-CF 2 HHADBCAADFB 0.080 112 No-CF Never PrmDry loPr Pour Large In No No-CF 3 GHADBCABDL3 0.068 123 Final Never PraDry loPr Pour Large Hi No BMT 4 HHADBCAADFA O.060 38 No-CF Never PrmDry Lo LoPr Four large No No-CF ; 5 GHADBCAADDB 0.054 112 Final Never PrmDry IoPr Pour large la No BMT w 6 GHADBCAADDA 0.041 38 Final Never PraDry LoPr Pour Large In No BMT g 7 HHADBCABDFA O.039 33 No-CF Never PrmDry LoPr Pour Large Hi No No-CF co 8 HHACACABBFA 0.031 16 No-CF Never PrmDry ImPr HFME Large Hi Md No-CF 9 GHADBCABDDA 0.026 33 Final Never PrmDry IoPr Pour Large Hi No BMT 10 CHADBCAADCB 0.022 112 Early Never PrmDry Pour IoPr Large In No Irak Five Most Probable Bins that have VB and Early CF** 10 CHADBCAADCB 0.022' 112 Early Never LoPr Four PrmDry Large Io No leak 11 CHADBCABDCB 0.018 121 Never LoPr Early PrmDry Four Large Hi No Leak 51 DHADDCBBDBB O.0039 75 CFatVB Never PrmDry LoPr Alpha Medium Hi No Rupture 54 ~ DHACACABACB 0.0038 7 CFatVB Never PrmDry ImPr HPME Iarge Hi Hi leak 57 DHACCCAABCA 0.0034 1 CFatVB Never PrmDrv ImPr BtmHd Large Io Md Imak ; Mean probability conditional on the occurrence of the PDS. A listin; ^ all bins, and a listing by observation are available on computer media.
The mean probability of CF at VB for this PDS group is 0.020. This is in i addition to the 0.045 probability of CF at the start of the accident. There are no late CFs due to hydrogen burns for the LOSP PDSs since power is never recovered during this period. It is possible to have hydrogen burns in the very late period when the containment slowly cools off due to heat transfer to the containment, and enough steam condenses to de inert the containment atmosphere. The mean probability of very late CF due to a burn is slightly less than 0.04 The mean probability of BMT is 0.31. The mean probability of eventual overpressure CF is less than 0.04 for the SBO, low acceleration PDS group. , 2.5.7.6 Results for PDS Croun EO 3. LOCAs' Low Acceleration. EPRI Hazard Distribution. This PDS group consists of accidents initiated by , seismic pipe breaks. The failures in the ECCS required to respond to these breaks are partially seismic and partially random. There is no SBO, but some of the seismic failures are failures in the electrical distribution , system, specifically in the parts that supply power to the ECCS, sprays, or ' AFWS. In these situations, the availability of electrical power, fourth PDS characteristic, is denoted by a "P". This PDS group consists of the fraction of these accidents due to seisms with a maximum ground accelera-tion less than 0.6 g. The 11 PDSs in this group are discussed in Section 2.5.5.3 Table 2.5 27 lists the 15 most probable APBs for the PDS group. Evaluation of the APET produced 266 bins for this group, of which, 43 are required to capture 95% of the probability. The 15 most probable bina capture 80% of the probability. Six of the 15 most probabic bins have CF at the start due to SC or RCP support failures; four have no VB and no CF, three have VB but no CF, and two have VB and BMT. The two most probable bins have no VB and no CF. They result from the PDSs with LPIS operating at UTAF. In the lhCA, low acceleration PDS group there is the possibility of arrest-int the core degradation process and avoiding VB since four of the 11 PDSs in he group have the LPIS operating at UTAF. If the RCS can be depres. l surized sufficiently in a timely manner, injection by the LPIS may arrest ' i core damage for these PDSs. The mean probabilities for this PDS group for l the four pressure ranges at UTAF and just before VB (if it occurs) are: At UTAF Just Before VB SSPr (2500 psia) 0.0 0.0 HiPr (600 to 2000 psia) 0.0 0.0 ImPr (200 to 600 psia) 0.54 0.01 ' LoPr (< 200 psia) 0.46 0.99 l As 'all of the PDSs in this group have an A, S t or S 2 break, the RCS pres-sure at UTAF . (Question 15) is bound to be low or intermediate. The Sa breaks comprise about 0.54 of this PDS group. As there is electric power available, and opening the PORVs is directed by the procedures when the core exit thermocouples show temperatures of 1200*F, it was estimated that the probability was 0.90 that the operators would open the PORVs in time to ensure the RCS was at low pressure by VB. 1 L 2.139 1
Table 2_5-27 Results of the Accident Progression Analysis for Surry Seismic Initiators. PDS Group EQ 3, IDCAs: Iow Acceleration, EPRI Hazard Distribution No. CF RCS VB Amt Zr CF Order Bin Prob.* Occur. Time Sorays CCI Pres. Mode CCI Ds HPME . Size Fifteen Most Probable Bins ** 1 HHCDFCDADFB 0.129 112 No-CF Never No-CCI LoPr No-VB No-CCI In No No-CF 2 HHCDFCDBDFB 0.118 123 No-CF Never No-CCI IoPr No-VB No-CCI Hi No No-CF 3 HHADBCAADFB 0.084 112 No-CF Never PrmDry LoPr Pour Large la No No-CF 4 HHADBCABDFB 0.080 123 No-CF 'Never PrmDry LoPr Pour Large Hi No No-CF 5 CHADBCAADCB 0.073 112 Early Never PrmDry loPr Four Large In No leak n 6 CHADBCABDCB 0.071 121 Early Never PrmDry IoPr Pour Large Hi No Ioak g 7 GHADBCAADDB O.057 112 Final Never PrmDry loPr Pour Large lo No BMT o 8 GHADBCABDDB 0.055 123 Final Never PrmDry LoPr Pour Large Hi No BMT 9 HDCDFCDADFB 0.026 112 No-CF Always No-CCI loPr No-VB No-CCI In No No-CF 10 CDCDBCDADCB 0.026 111 Early Always No-CCI LoPr Pour No-CCI Lo No Ioak 11 CDCDBCDBDCB 0.020 92 Early Always No-CCI IoPr Pour No-CCI Hi No leak 12 CDDDBCAADCB 0.017 111 Early Always PrmDeep lePr Pour Large In No Ioak 13 HDCDFCDBDFB 0.017 123 No-CF Always No-CCI No-VB No-CCI loPr Hi No No-CF 14 HDCDBCDADFB 0.015 112 No-CF Always No-CCI No-CCI loPr Pour In No No-CF 15 CDDDBCABDCB 0.013 87 Early Always Pour PrmDeep IoPr Large Hi No Leak Mean probability conditional on the occurrence of the PDS. A listing of all bins, and a listing by observation are available on computer media.
l l 1 The mean probability of having the LPIS operating at UTAF is about 0.34 for I the seismic 1hCA PDS group. Four of the 11 PDSs in this group have the LPIS operating. For these PDS, it is likely that the RCS will depresst4' sufficiently and quickly enough to allow LPIS injection in time to prer VB. The mean probability of arresting the core degradation process and , avoidin6 VB is about 0.31. } t The mean probability of CF at VB for this PDS group is 0.008. This is in addition to the 0.33 probability of CF at the start of the accident. This value is so large because five of the 11 PDSs in this group are "A" PDSs in which SC or RCP support failures at the time of earthquake fail the containment. ! There are no late or very late CFs due to hydrogen burns for the LOCA PDSs since power is continuously available. The hydrogen will burn when a fla-mmabic concentration is attained, which poses a negligible threat to the : Surry containment. The mean probability of BMT is 0.13. The sprays ope.
- rate throughout_the accident in five of the 11 PDSs in this group, and the sprays are irreparably failed in six of the 11 PDSs. The PDSs with failed sprays are more likely than those with operable sprays, so the mean proba-bility for having no CllR is 0.81. When CllR is not recovered in the long term, an overpressure failure of the containment is possible after several days. As discussed in Section 2.5.5.1, the Surry basemat is constructed of siliceous concrete and this is unlikely to occur. For the 14CA, low acco- ,
leration PDS group the mean probability of eventual overpressure CF is 0.01, t 2.5.8 Sensitivity Analyses for Seismic Initiatorn! EPRI llazard Distribution i No sensitivity analyses that affected the accident progression analysis were performed for seismic initiators with the EPR1 hazard distribution. L 2.5.9 Comoarison of Results for Seismic Initiators Figure 2.5 5 summarices the results of the accident progression analysis 1 for the seismic initiators. The largest differences between the two hazard , distributions are in the core damage frequencies. The' mean core damage , frequencies that result from using the EPRI hazard distributions are almost an order _ of magnitude lower than the mean core damage frequencies that ; result from using the LLNL hazatd distributions. As the core damage fre- , quency of a PDS group has no effect on the evaluation of the APET, the . accident progression analysis results for the two hazard distributions are very similar. The differences are due to differences in the frequencies of the individual PDSs relative to other PDSs in the group.
' Figure 2.5 5 shows the mean probability of the summary accident progression >
binn for the three PDS groups and both sets of seismic hazard distribu-tions. Based on the mean probability, the majority of seismic core damage accidents result in either no vessel failure or no containment failures. There is no possibility of avoiding vessel failure for the SB0 accidents because the damage to the switchyard is deemed to be not reparable within 2.141 l
the times of interest. The mean probability of early containment failure is on the order of 0.01 if the initial failures due to SG or RCP support failures are excluded. Initial CF occurs only for large breaks, so es-sentia11y all the low pressure early CFs are attributable to 50 or RCP failures. If initial CFs are removed, the low pressure early CFs become negligible due to the strength of the Surry containment relative to the loads expected when the vessel fails at low pressures. The probability of late CF is relatively high for the SB0 accidents because there is no long term recovery of containment heat removal as there is in the LOSP and LOCA accidents. The probability of late CF in the seismic accidents is higher than it is for internal initiators (see Figure 2.5 3) because many of the PDSs (including the most frequent ones) in the LOSP and LOCA groups have the sprays failed. PDS GROUP (Wean con Darnase hviuency) ACCIDENT
" ' "' " " -~~"' * " " " " " ~" " " " " - " " " " " " " - " " " ' " "
PROGRESSION LDSP 3DO LoCAs Total LoSP SD0 LOCAs Total OlN GROUP ( o OE- Os) ( ? et-05) ( 2 3E-05) ( l 91:- 04) ( l on-Os) ( o LE-06) ( 3 ot-Oe) ( 2 et-05) VD, alpha. 0.005 0 000 0 006 0 006 0.006 0 000 0 007 0.006 early CF VD > 200 pel. 0.006 0 Ol2 0 000 0 000 0 014 0 000 early CF VD. < 200 pst. 0072 0 360 0 082 0 OSO O 322 0 050 early CF
._ I _ _ I _
VD. DWT or late CL 0 214 0 377 0 124 0 200 0 232 0 382 0 137 0 203 I itypass 0 001 0 002 0 001 0 001 0 001 0 001 0 391 0 63; O 216 0 435 0 408 0 b40 0 227 0 457 VD. No CF NoYli 0 393 0 30b 0 160 0 346 0 307 0176 {_ . _ . . ._. Key: DMT = llanernet Helt-Through SURRY CF = Containtuent ihture Cl, = Contalbment 1.eak VU = Vessel tireac h Figure 2.5 5. Mean Probabilities of APBs for PDSs -Seismic. 2.142
Figure 2.; 6 displays distribution of the probability of arresting the core melt procesc and avoiding failure of the lower head of the reactor vessel for the seise n initiators. No histogram is shown for the SB0 PDS group as core damage art *st is not possible for this group. Inclusion of the SB0 accidents in the total, however, accounts for the fact that the total dis-tribution shows core damage arrest to be less likely than it is for either the 1DSP or LOCA groups. The differences between the distributions for the hazard distributions are due to differences in the frequencies of the PDSs in which an injection system is operable relative to other PDSs in the group. The probability of core damage arrest for the seismic initiators is lower than it is for internal initiators (see Figure 2.5 1) because the frequencies of the PDSs with one or more injection systems operable are lower relative to the total frequency than they are for internal initiators. Figure 2.5 7 presents distribution of the probability of early failure of the containment for the seismic initiators. Differences between the two sets of hazard distributions are minimal. The relatively high probabil. ities of early CF for the SB0 and bOCAs groups reflect the initial failures of the containment due to SG and RCP support failures. The distributions for early CF for the seismic LOSP initiators are comparable to those for the internal LOSP initiators (see Figure 2.5 2). SURRY l oO W = mean m a median th a percentile 96t h. 96th. 0.75-96th , u 96th.
;": .I Z%4 ?
l I' , O v I 2*
- n. m l
,6 0.50- ,5ig ,
06th. 1 h[ o I $ u ~
]O m--.
3 u _. p OT O I . m_. J m": . o g$. m : : --
]l ] 1
~~
o 00 m m .
... . .- $ ... l . ,,-
..LLNL........ ......-0PRI--------
PDS Croup IDSP LoCAs Total LoSP LoCAs Total Core Damage Req 9 00-05 2 30-05 190-D4 1f.0-05 3 50-06 2 BE-05 i Figure 2.5 6. Probability of Core Damage Arrest- Seismic. 2.143
1.El . SUkRY 1.00 est6 est6 I 96t h.
& Et W -+
3 m.
$';:: f l.E-1. u, u
u v O$ soth. _
- m. 96th. .. I
.8 e _ I m. __
eE m. _ . __ A f 1 p.2, v.. - W -. ! m-+ gj - 6th. ( - cc 6Lk. Y ? k'g , 6tk. l g a ,g 3, 6ie. Y) - 6tk. . .
- m. y no. 6t k.
y 1.0-4.
] hl = rnean in = rnedlan J. og g ,
6tw ,! th = percentile
. . . . . . . . . LL N L- - - - - - - - - ~ ~ - ------~~----tPRl------------
PDS Group LoSP SHO LotAn Total LOSP S 11 0 LorAs Total rote Darnage neq 9OE-OS *t. 0E- 05 2 :tE-Ob 1 DE- 04 10E-05 9SE-06 3 60-00 200-05 Figure 2.5 7. Probability of Early Containment Failure Seismic. 2.6 Insichts from the Accident Procression Analysis For internal initiators, there is a good chance that non bypass accidents will be arrested before vessel failure. The arrest of core damage is due to the recovery of offsite power or the reduction of RCS pressure to the point where e system operating at the onset of core damage can inject successfully. If core damage proceeds to failure of the lower head, the containment is unlikely to also fail at this time. This is partially due to the depressurization of the RCS before vessel failure, and partially due to the strength of the Surry containment relative to the loads expected. If the containment does fail, it is most likely to be BMT many hours after the breach of the vessel. Depressurization of the RCS before the vessel fails is quite effective in p reducing the loads placed upon the containment at vessel breach. The ) effective mechanisms, are temperature induced failure of the hot leg or l surge line, temperature induced failure of the RCP seals, and the sticking open of the PORVs. All of these mechanisms are inadvertent and beyond the control of the operators. The apparent beneficial effects of depressur-izing the RCS when lower head failure is firminent indicate that further investigation of depressurization may be warranted. The dependency of 2.144
containment integrity on failures that occur at unpredictable locations and at unpredictable times is somewhat unsettling. Analysis of the effects of increasing PORV capacity, providing the means to open the PORVs in blackout situations, and changing the procedures to remove the restricting condi-tions on deliberate depressurization might prove rewarding in decreasing the probability of early containment failure at PWRs with large, dry containments. Another factor responsible for the low probability of early failure of the Surry containment is the strength of the containment relative to loads expected. For none of the 19 cases defined by the experts for pressure rise at vessel breach, was the mean of the pressure rise distribution less than 35 psi below the mean of the distribution for the failure pressure of the containment. For the case with the highest pressure rise, the mean probability of containment failure is on the order of 0.30, but this case is improbable. For the more probable cases with the RCS at pressures above 200 psi at low head failure, the mean probability of containment failure is on the order of 0.05. For vessel failure below 200 pai, the failure proba. bility is about 0.002. Although their core damage frequency is relatively low, the bypass acci-dents are important for internal initiators, specifically, Event V. This is due to the low probability of early containment failuro for ' the more frequent accidents, SB0 and LOCAs. The occurrence of Event V is a more likely way to defeat the containment function in a core melt situation than is an SB0 or LOCA followed by CF at VB. Even though Event V creates a bypass of the containment, there is a mean probability of 0.85 that the break location in the interfacing low pressure piping will be covered by a water pool when the releases commence. For fires, there is no possibility of core damage arrest as the initiating fire destroys the ability to provide control of motive power to the coolant injection systems. Early containment failure is unlikely, as for internal initiators, and there are no bypass initiators. Early containmont failure is also unlikely for the seismic initiators, ex-cept for the initial containment failures due to reactor coolant pump or steam generator support failures. There are no bypass initiators for the seismic accidents. Coro damage arrest before VB is not possible in the seismic SB0 accidents because the damage to the switchyard is judged to be to extensive to be repaired in the time frame of interest. Core damage arrest is possible in the LOSP and LOCA accidents, however. On the whole, about one sixth of the seismic core damage accidents are recovered before the vessel melts through. Core damage arrest is not as likely for seismic initiators as it is for internal initiators. This is due to the fact that the frequencies of the PDSs with one or more injection systems operable are lower relative to the total frequency for the seismic initiators than they are for internal initiators. Except for the initial failures of the containment due to SC and RCP sup-port failures, the probability of early CF is about 0.01 for the seismic initiators. Tb initial CF accidents are all large breaks in which the RCS will be at low pressure when the lower head fails. CF at VB is essentially 2.145
1 i j 4 negligible for these accidents. Because of these initial failures of the i containment, , the probability of early CF is much higher for seismic ini. . tiators than it is'for internal initiators. There are no bypass initiators for the seismic accidents, however. If bypass and early CF are considered
- together, then the seismic initiators have a lower probability of causing ,
an accident with an open release path to the environment early in the acci. dent. About one quarter of the seismic accidents have late failures of the. l containment. This is much higher than for internal initiators, and comes i from the inability to recover offsite power for the SB0 PDS group, and from the relatively high frequencies of PDSs in which the sprays are failed in ' the thSP and LOCAs groups, t 4 I h t k [ i 9 f p 5 2.146 f
i 1 1 t I
- 3. RADIOLOGICAL SOURCE TERM ANALYSIS !
The source term is the information passed to the next analysis so that the offsite consequences can be calculated for each group of accident progres-sion bins. The source term for a given bin consists of the release frac-tions for the nine radionuclide groups for the early release and for the late release, and additional information about the timing of the re~ eases, the energy as,sociated with the releases, and the height of the releases. The source terms for Surry are generated by a relatively small parametric computer code: SURSOR. The aim of this model is D9.1 to calculate the behavior of the fission products from their basic chemical and physical
- properties and the flow and temperature conditions in the reactor and the containment. Instead, the purpose is to represent the results of the detailed codes that calculate the fission product behavior by application of the first principles of phvai m, chemistry, and thermodynamics.
A more complete discussion of the source term analysis, and of SURSOR in particular, may be found in NUREG/CR 5360.* The methods on which SURSOR is based are presented in NUREC/CR 4551, Volume 1, and the source term issues considered by the expert panels are described more fully in NUREG/CR 4551, Volume 2. Part 4. , Section 3.1 summarizes the features of the Surry plant that are important to the magnitude of the radionuclide release. Section 3.2 presents a brief overview of the SURSOR code, and Section 3.3 presents the results of the source term analysis. Section 3.4 discusses the partitioning of the thousands of source terms into groups for the consequence analysis. Section 3.5 concludes this section with a summary of the insights gained i from the source term analysis. 3.1 Surry Features Important to the Source Term Annivsis The nuclear reactor of Surry Unit 1 is a three loop pressurized water reactor (PWR) contained in a containment constructed of reinforced concrete with a welded steel liner forming the pressure boundary. Figure 1.1-shows a section through the Surry containment. More detail on the Surry plant in general is contained in Sections 1.2 and 2.1; this material is not repeated here. The design pressure of the'Surry containment is 45 psig, and the structural experts found the- failure pressure to be generally between two and three times the design pressure. The relatively high failure pressure combined with the.large size of the containment (5.1E4 m3 - 1.8E6 f t 3 ) implies that L the containment is relatively resistant to failure by the events at VB or by hydrogen combustion before or af ter VB. This was confirmed by the results of the accident progression analysis.
- H.-N Jow, W. B. Murfin, and J. D. Johnson, "XSOR Codes User's Manual,"
. NUREG/CR 5360, SAND 89-0943. Sandia National Laboratories, (in preparation). l l 3.1 l
Two accident scenarios have been identified at Surry which bypass the containment. These accidents are Event V and steam generator tube ruptures l (SGTRs). In Event V, the check valves which separate the low pressure j injection system (LPIS) from the reactor coolant system (RCS) fail. The , LPIS piping is not designed for full RCS pressure, and it fails outside the l containment. This provides a direct pathway from the vessel to the auxili-ary building. If the failure is at a low elevation, the water from the RCS escaping through the break and the water from the refueling water storage tank (RWST) pumped out the break by the LPIS will cover the break location so that there will be some scrubbing of the fission products released from the vessel. If the break location is not at a low elevation, there may be few effective removal mechanisms between the core and the environment, and releases could be quite high. The magnitude of the source term from an SCTR accident depends on the in- : tegrity of the secondary system and the containment. If the integrity of both is maintained, the releases may be quite small. If the safety relif valves (SRVs) on the secondary system stick open, then a direct path from , the vessel to the environment is created and the releases may be very high. If the; SRVs on the secondary system do not stick open, then the releases depend on the time at which the containment fails (if at all) as in non-bypass accidents. Emergency containment heat. removal at Surry is by spray systems as de-scribed in Section 2.1.3. The sprays are quite effective in removing fission products from the containment atmosphere. The redundancy of the ' spray recirculation system means that there are no accident situations at : Surry in which electric power is available and the sprays are failed due to hardware faults. There are some scenarios in which electric power is avai-lable,but the sprays cannot operate due to luck of water in the sump, but these are all bypass accidents and spray operation has only a minor effect
~
r on the releases at best. Spray operation at and following VB is important i in reducing the release if the containment fails after VB. The pressure inside the Surry containment is maintained about 5 psia below
-ambient atmospheric pressure when the reactor is operating. This makes the existence of pre existing leaks due to an open hatch or airlock negligible. ,
The - likelihood of isolation failure is also low, so, except for certain r seismic events, the probability of a direct failure of the containment at the start of the accident is quite low. l There is no connection between the sump and the reactor cavity at a low l- elevation in the Surry containment as discussed in Section 2.1.5. Thus, L .the cavity is more of ten dry at VB than it would be if the sump and the L cavity were connected. In - general, a cavity full of water leads to less pressure rise at VB, and lower releases should the containment fail at VB, - t than a dry cavity. Further, a dry cavity makes core concrete interaction (CCI) more likely, and the fission products release higher, than if the cavity is full of water. In summary, the Surry containment is relatively robust, which reduces the likelihood of early containment failure. The sprays have a great deal of , redundancy, and their operation in non blackout sequences reduces the 3.2
-yw g-
1 l l l releases from late containment failures (which are few in any event) . In Event V and SGTRs in which the secondary systems SRVs are stuck open, the release path bypasses the containment. 3.2 Description of the SURSOR Code This section. describes the manner in which the source term is computed for each accident progression bin (APB). The source term is more than the fission product release fractions for each radionuclide class; it also contains information about the timing of the release, the height of the release, and the energy associated with the release. The next subsection presents a brief overview of the parametric model used to calculate the source terms. Section 3.2.2 discusses the model in some detail; a complete discussion of SURS0R may be found in Reference 1. Section 3.2.3 presents the variables sampled in the source term portion of this analysis. 3.2.1 overview of the Parametric Model SURSOR is a fast running, parametric computer code used to calculate the source terms for each APB for each observation for Surry. As there are typically a few thousand bins for each observation, and 200 observations in the sample. - the need for a source calculation method that requires a minimum of computer time for one evaluation is obvious. SURSOR is no.t designed to calculate the behavior of the fission products from their basic chemical and physical properties and the flow and temperature conditions in
- the reactor and the containment. The purpose of SURSOR is to provide a framework for integrating the results of the more detailed codes that do i considur these quantities. Since many of the factors SURSOR utilizes to calculate the release fractions were determined by a panel of experts,.the results of the detailed codes enter SURSOR " filtered" through the sxperts.
The 60 radionuelides (also referred to- as isotopes, or fission products) considered in the consequence calculation. are not dealt with individually in the source term calculation.- Some different elements behave similarly enough both chemically and physically in the release path that they can be considered together. The sixty isotopes are placed in nine radionuclide l classes as shown in Table 3.2 1, it is these nine classes which are treated individually in the source term analysis. l L Table 3.2 1 , L Isotopes-in Each Radionuclide Release Class Release Class Isotones Included
- 1. Inert Gases Kr 85, Kr 85M, Kr-87, Kr 88, Xe 133, Xe 135
- 2. Iodine I-131, I-132, I 133, I 134, I 135
- 3. Cesium Rb 86, Cs 134, Cs-136, Cs 137
- 4. Tellurium Sb 127, Sb 129 Te 127, Te 127M, Te 129, Te 129M, Te-131M, Te-132 3.3
Table 3.2 1 (continued) Release Class Isotones included
- 5. Strontium Sr 89, Sr 90, Sr 91, Sr 92
- 6. Ruthenium Co 58, Co 60 Mo 99, Tc 99M, Ru 103, Ru 105, Ru-106, Rh 105
- 7. Lanthanum Y 90, Y 91, Y 92, Y-93, Zr 95, Zr 97, Nb-95, La-140, La 141, 1a 142, Pr-143, Nd 147, Am 241, Cm-242, Cm 244
- 8. Cerium Ce 141, Ce 143, Ce 144, Np 239, Pu 238, P.* 239, Pu-240, Pu 241
- 9. Barium Ba 139, Ba 140 3.2.2 Descrintion of SURSOR Since the largest consequences generally result from accidents in which the containment fails before VB or about the time of VB, the nomenclature and structure of SURSOR reflects failure at VB. There is an early release which occurs before, at, or a few tens of minutes af ter VB, and there is a late release occurs several hours, after VB. In general, the early release is due to fission products that escape from the fuel while the core is ,
still in the RCS, i.e., before vessel breach, and is often referred to as the RCS release. The lato release is largely due to fission products that escape from the fuel during the CCI and is referred to as the CCI release. For situations where the containment fails many hours after vessel breach, the . "early" release equation. is still used, but the release is better - termed the RCS release - After both releases are calculated in SURSOR, they are combined into the late release and the early release is set to zero. The late release includes not only fission products released from the core during CCI, but also material released from the fuel before VB which deposits in the RCS or the containment, and then is revolatilized after VB. For radionuclide class 1, the early or RCS release is calculated from the following equation: . ST(i) - [ FCOR(i)
- FVES(i)
- FCONV(i) / DFE(i) } + DST ( FDCH(i) ) (Eq. 3.1) l And the late or CCI release is calculated from:
STL(i) - [ (1 PCOR(1))
- FPART(i)
- FCCI(i)
- FCONC(i) / DFL(i) )
+ DLATE( FLATE(i) ). (Eq. 3.2)
Both equations are valid for most APBs, but are not complete; there are additional terms, which are either small or apply only to certain types of accidents, that are not shown in this summary for reasons of expediency. 3.4
For example, some of the omitted terms concern releases from Event V and SCTR accidents. There is also an additional term, IATEI, that applies only for the iodine radionuclide class. The complete equations used are ; presented in NUREG/CR 5360. The FORTRAN listing of SURSOR is contained in Appendix B. The meaning of the terms in the equations above is as follows: ST - fraction of the radionuclide in the core at the start of the accident that is released to the environ. tent as part of the RCS release; ; FCOR - fraction of the radionuclide in the core released to the ' vessel before VB; FVES - fraction of the radionuclide released to the vessel that is subsequently released to the containment; FCONV - fraction of the radionuclide in the containment frca the RCS release that is released from the containment in the absence of any mitigating effects; t DFE - decontamination factor for RCS releases (sprays, etc.); DST - fraction of core radionuclide released to the environment due to direct containment heating at vessel breach; FDCl! - fraction of radionuclide in the portion of the core involved in direct containment heating that is released ' to the containment at vessel breach; STL - fraction of the radionuclide in the core at the start of the accident that is released to the environment as part of the CCI release;. FPART - fraction of the core which participates in the CCI; FCCI - fraction of the radionuclide in . the core material- at the : start of CCI which is subsequently released to the ' containment; FCONC - fraction of the radionuclide in the containment from the CCI l release that is released from the containment in the absence of any mitigating effects; DFL - decontamination factor for late releases (sprays, etc.); , D1 ATE - traction of core radionuclide released to the environment due to revolatilization from the RCS late in the accident; and FIATE - fraction of core radionuclide remaining in the RCS which is revolatilized late in the accident. 3.5 .
Only the functional dependence of DLATE on FLATE, and of DST on FDCil, is indicated above, but DLATE and DST also depend on other factors such as FCOR. DST and DLATE are expressed as fractions of the initial core inventory like ST and STL, Cumplete expressions for DST and DLATE and an expanded discussion of them may be found in the ASOR document. Figure 3.21 depicts the parametric equations schematically in terms of a flow diagram. Coming in from the left is all the radioactivity in any radionuclide class. The black arrows represent releases to the environment and the white arrows represent material retained in the RCS or in the containment. This figure is read as follows: the first division of the radioactive material is indicated by FCOR. The top branch, indicated by FCOR, represented the fraction released from the core before VB, and the lower branch, an amount 1-FCOR, represents the amount still in the RCS at VB. The FCOR branch is then split into that wh1ch leaves the RCS before or at VB, TVES, and that which is retained in the RCS past VB, 1 FVES. Of the material retained in the RCS at VB, a fraction FLATE is revolatilized later. Of the revolatilized fraction, a portion is removed by ergineered removal mechanisms such as sprayg factor 1/DFL, and a another portion is removed by natural mechanisms such as A position, factor FCONRL. The part of the revolatilized fraction that is not removed escapes to the environment, DLATE in the equation, as inMcated by the top black arrow in Figure 3.2 1. FCONRL is the containment release fraction for the late revolatilization release, and is set equal to the FCONC value for tellurium. When evaluated as part of the integrated risk analysis, SURSOR is run in the " sampling mode". That is, most of the factors in the release fraction equations are determined by sampling from distributions for that factor, and the value for each factor varies from observation to observation. Most of these distributions were provided by an expert panel. The equations above contain 10 factors. Seven of them were considered by the Source Term Expert panel. The other three were determined either by the expert panel for the previous draft of this report or internally. The eight issues considered by the Source Term Expert Panel are:
- 1. FCOR and PVES
- 2. Ice Condenser DF (not applicable to Surry)
- 3. FLATE
- 4. FCCI
- 5. FCONV and FCONC
- 6. LATEI (not utilized directly for PWRs)
- 7. Reactor Building DF (not applicable to PWRs)
- 8. DCH Releases (DST) ,
Thre6 of these issues are not applicable to Surry. For each issue considered by the expert panel, the result is an aggregate distribution for . the nine radionuclide rele,:se clarses defined in Table 3.2 1. These ( distributions are not necessarily discrete. While the experts provided l separate distributions for all nine classes for FCOR, for other factors, i for example, they stated that classes 5 through 9 should be considered together as an aerosol class. Note that the distributions for the nine 1 3.6
V@dU E N
?
+- i s; ~ A. ' L A' A ( iA i,
\ >a
'g g z.
1 / j i g
*<r J -
% ,It / i ;
3.7 II IEEI.IIWII lllllEIL
radionuclide classes are assumed to be completely correlated. That is, a single Latin Hypercube Sampling (Ills) variable applies to each factor in , the release fraction equation, and it applies to the distributions for all nine radionuclide classes. For example, if the random number for FCOR is 0.777, the 77.7th percentile value is chosen from the iodine distribution, , the cesium distribution, the tellurium distribution, etc. for FCOR. I Many of these factors in the equations above are determined directly by sampling from . distributions provided by a panel of experts, see NUREG/CR-4551, Volume 2, Part 4. Other factors are derived from such values, and still others were determined internally, see NUREG/CR-4551, Volume 2, Part 6 and tho XSOR document.1 A brief discussion of each factor in the equations above follows. FCOR is the fraction of the fission products released from the core to the vessel before vessel failure. The value used in each observation is obtained directly from the experts' aggregate distribution. There are separate distributions for each fission product group for high and low zirconium oxidation in vessel. FVES is the fraction of the fission products released to the vessel tl*at is subsequently released to the containment before or at vessel failure. As for FCOR, the value used in each observation is obtained directly from the experts' aggregate distribution, and there are separate distributions for each fission product group. There are four cases: RCS at system setpoint pressure, RCS at high or intermediate pressure, RCS at low pressure, and Event V. FCONV is the fraction of the fission products in the containment from the RCS release - that is released from the containment in the absence of mitigating factors such as sprays. The expert panel provided distributions for FCONV for four cases, each of which applies to all species except the noble gases, These cases apply to containment failure at or before VB, or within a few hours of.VB. The four cases are: containment leak at or before VB--sprays operating, containment leak at or before VB--sprays not operating, containment rupture at or before VB, and very late containment rupture. Distributions for other levels and times of containment failure are derived in SURSOR from these four distributions. There is a fifth distribution which applies to Event V and which was determined internally. If the containment failure happens a day or more af ter the start vf the accident, none of these distributions is used for FCONV. Because of the long time period for the engineered and natural removal processes to reduce the concentration of the fission products in the containment atmosphere, the fraction of the fission products released before or at VB remaining airborne at the time of containment failure is very small. This fraction was estimated internally to be 1.0E 6, and FCONV is set to that value for final period releases. (The exact value of 1.0E-6 is not important; any very small value would be satisfactory.) This value is used whether the release is due to above-ground failure or basemat melt-through (BMT) . DFE is the decontamination factor for early releases. At Surry the containment sprays are the only mechanism that contribute to DFE for non bypass acci-dents. For Event V, DFE reflects pool scrubbing if the break location is underwater. The Source Term Expert Panel concluded that the distributions 3.8
used for the spray decontamination factors (DF) were less important to determining offsite risk and the uncertainty in risk than whether the sprays were operating and other factors; so, the spray DF distributions (and the pool DF for Event V) were determined internally. There are two spray distributions which apply to the fission products released from the RCS before or at VB: the first applies when the con- tainment fails before or at VB and the RCS is at high pressure at VB; and the second applies when the containment fails after VB or when the con- tainment fails at VB but the RCS is at low pressure. Each distribution applies to all species except the noble gases. Fot failures of the con- tainment in the final period, the value from the distribution is multiplied by 10 to account for the very long period which the sprays have to wash particulate material out of the containment atmos.phere. DST is the fission product releasc (in fraction of the original core inventory) from the fine core debris particles that are rapidly spread throughout the containment in a direct containment heating (DCH) event at VB. The experts provided distributions for the fractions of the fission products that ara relerned. from the portion of the core involved in DCH for VB at high pressure (1000 to 2500 psia) and for VB at intermediate pressure (200 to 1000 psia). There are separate distributions for each fission product group (inert gases, iodine, cesium, etc.). These distributions are utilized only if the containment fails at (or within a few minutes of) vessel failure. For containment failures that occur hours after VB, it was internally estimated that the amount of fission products f rom DCH remaining in the atmosphere many hours after VB would be negligible. FpF T is the fraction of the core which participates in the CCI. The value of this factor is determined in the accident progression event tree (APET). There are three ranges of values. Five por cent of the core is estimated to remain in the vessel indefinitely and is not available to participate-in CCI under any circumstances. SURSOR subtracts the amount of the core par-ticipating in DCH from the amount passed - to it in Characteristic 7 of the accident progression bin. FCCI is the fraction of the fission products present in the core material at the start of CCI that are released to the containment during CCI. The experts provided distributions for four cases that depended upon the frac-tion of the . zirconium oxidized in vessel and the presence or absence of water over the core debris during CCI. There are separate distributions for each fission product group. FCONC is the fraction of the fission products released to the containment from the CCI that is released from the containment. The expert panel provided distributions for FCONC for five cases. There are separate distributions for each fission product group (inert gases, iodine, cesium, etc.). The five cases are the same as for FCONV. None of these cases is used for containment failure in the final period (after 24 h). Since containment failure occurs many hours af ter most of the fission products have been released from CCI, only a very small fraction of these fission products will still be in the containment atmosphere at the time of containment failure. This fraction was estimated internally to be on the order of 1.0E-4 The exact value is determined by utilizing the FCONC 3.9 I m
distribution for Case 3, rupture before the onset of CCI. The ratio of the LHS value from the distribution to the median value of the distribution is multiplied by 1.0E 4 to obtain the value of FCONC used for final period containment failure. This value is used whether the release is due to aboveground failure or BMT. DFL is the decontamination factor for late releases. At Surry, DFL can be due to either the containment sprays or a pool of water over the core debris during CCI. If both mechanisms are operable, only the larger DF is utilized. The pool scruobing DF is obtained from a one of two !nternally determined distributions. One distribution applies to a full cavity and the other to a partially - full cavity (accumulator water only) . The pro-cedure used to obtain the spray DF for the CCI release for final period CF is similar to that used to obtain the value for DFE (discussed above). As for DFE, if the containment fails in the final period, the value from the late CF spray distribution is multiplied by 10 to account for the very long time the sprays have to wash particulate material out of the containment atmosphere. FIATE accounts for the release of radionuclides from the RCS late in the accident. Like DST, it is a fraction of the original core inventory. Fission products deposited in the RCS before VB may revert to a volatile form af ter the vessel fails and make their way to the environment. This term considers only revolatilization from the RCS. ~Revolatilization from the containment is considered to be significant only for iodine, and is discussed below (see 1ATEI). The expert panel provided distributions for the fraction of the radionuclides remaining in the RCS which are revolatilized. The amount remaining in the RCS is a function of FCOR, FVES, and other terms and is calculated in SURSOR. The experts concluded that whether there was-effective natural circulation through the vessel was important in determining the amount of revolatilization. Thus, there aro two cases: one large hole in the RCS, and two large large holes in the RCS. The experts provided separate distributions only for iodine, cesium,.and tellurium. (Revolatilization is not possible for the inert gases as they would not deposit, and the experts concluded that it is negligible for radionuclide classes 5 through 9. ) FLATE is computed in the following manner: the value from the experts' distributions is applied to the fraction of the radionuclide remaining in the RCS to obtain the fraction of the core inventory released to the containment by this mechanism. This is multiplied by the FCONC value for tellurium to determine the fraction that escapes to the environment. The tellurium value for FCONC is considered to be. appropriate fer revolatilized material since it, like tellurium, is slowly released over a long time period. L 1ATEI accounts for iodine in the containment which may assume a volatile form, such as methyl iodide, and be released late in the accident. The primary source of this iodine is the water in the reactor cavity and the
)
containment sumps (which are separate at Surry). This tcrm is added to the late release only for radionuclide class 2, iodine. The experts provided a distribution for the fraction of iodine in the containment which is converted to volatile forms. The method of calculating the amount - of iodine remaining in the containment depends upon FCOR, FVES, FCCI, and other factors and is explained in the XSOR document.
)
3.10
No differentiation is made between the BMT and aboveground leaks in the final period. Even though the release point for BMT is underground, no allowance is made for attenuation or decontamination of the late fission product release. The BMT release is often dominated by the iodine release due to the IATE1 term. The very slow passage of the gases through wet soil with a low driving pressure would undoubtedly result in some reduction in this release. This reduction could be quite large. Although giving no credit for removal in the wet soil is conservative, it is unimportant for the sample as a whole. The total releases from all the BMT failures of the containment are small compared to the releases from accidents and pathways in which the containment fails at or before vessel breach, or where the containment is bypassed. 3.2.3 Variables Samoled for the Source Term Analysis The twelve variables sampled for the source term analysis are listed in Table 3.2 2 below. That is, when SURSOR was evaluated for all the bins generated by the APET evaluation for a given observation, all the sampled , parameters in SURSOR had values chosen specifically for that observation. l i Table 3.2-2 Variables Sampled in the Source Tern Analysis Variable Description FCOR Fraction of each fission product scoup released from the core to the vessel before or at vessel breach. There are two cases: high and low zirconium oxidation. FVES Fraction of each fission product group released from the vessel to the containment before or at vessel breach. There are four-
. cases: RCS at system setpoint pressure, RCS at high or intermediate pressure, RCS at low pressure, and Event V.
VDF Decontamination factor for pool scrubbing for Event V when the break location is underwater at the time of the release. There is=one distribution, which applies to all radionuclide classes except inert gases. < FCONV Fraction of each fission product group in the containment from the RCS release that is released from the containment in- the absence of mitigating factors such as ' sprays. There is one distribution for each case, which applies to all radionuclide classes except inert gases. There are five cases: containment leak at or before .VB with sprays operating, containment leak at or before VB with sprays not operating, containment rupture at or before vessel breach, very late containment rupture, and Event V. Note that FCONV does not account for fission product removal by the sprays. The case . differentiation on spray l operation is to account for differences in containment !- atmosphere temperature and humidity between the two cases, 3.11
Table 3.2-2 (continued) Variable Description FCCI Fraction of each fission product group in the the core material at the start of CCIs that is released to the containment. There are four cases: low zirconium oxidation in the core and no overlyir', water, high zirconium oxidation in the core and no overlying water, low zirconium oxidation in the core with overlying water, and high zirconium oxidation in the core with overlying water. FCONC Fraction of each fission product group in the containmant from the CCI release that is released from the containment in the absence of mitigating factors such as sprays. The five cases are the same as those for FCONV, but there are separate distributions for each radionuclide class. SPRDF Decontamination factor for sprays. Internal elicitation was used to develop a distribution for this variable which was used for all. fission product groups except the noble gases. There is one distribution for each case, which applies to, all radionu-clide classes except inert gases. There are three cases: RCS release at high pressure and CF at VB, RCS releases not covered by the first case, and CCI releases. IATEI Fraction of the iodine deposited in the containment which is revolat.ilized and released to the environment late in the accident. This variable applies only to iodine. FIATE Fraction of the deposited amount of each fission product group in the RCS which revolatilized af ter VB and released to the containment. There are two cases: one large hole in the RCS, and two large holes in the RCS. DST Fraction of each fission product-group in the the core material that becomes acrosol particles in'a direct containment heating event at VB that is released to the containment. There.are two cases: VB at high pressure (1000 to 2500 psia) and VB at intermediate pressure (200 to 1000 psia). FISCFOSC Fraction of each fission product group released from the reactor vessel to the~ steam generator, and from the steam generator-to the environment, in a SGTR accident. There are two separate distributions, FISC and FOSC, each of which has two cases: SCTRs in which the secondary SRVs reclose, and SGTRs in which the secondary SRVs stick open. POOL DF Decontamination factor for a pool of water overlying the core debris during CCI . There are two cases: a completely full (depth about 14 f t) cavity, and a partially full cavity (accumulator water only, depth about 4 f t) . 3.12
These values were selected by the UlS program from distributions that were previously defined. Most of these distributions were determined by the expert panel on source terms. The sampling process works somewhat differently for the source term analy-sis than it does for the accident progression analysis. In the source term analysis, Uls was used only to determine a random number between 0.0 and 1.0 for each variable to be sampled. The actual distributions are con-tained in a data file (listed in Appendix B) that is read by SURSOR before execution. The variables selected by Uls are used to define quantiles in the parameter distributions; the values associated with these quantiles are used as para-meter values in SURSOR. In use, the process works like this. Say UlS selects a value of 0.05 for FCOR for Observation 1. Referring to the data tables in Appendix B.2, it may be seen that, for low zirconium oxidation in vessel, the 0.05 quantile values for FCOR are 0.18 for inert gases, 0.084 for iodine, 0.067 for cesium, etc. There is no correlation between any of the source term variables, but complete correlation within a vari-able. FCOR is not correlated with FVES, FCONV, or any other variable, but the values for the different cases and for the different radionuclide classes are completely correlated. That is, if the 0.05 quantile value is chosen for iodine for low zirconium oxidation, the 0.05 quantile value is also chosen for all the other radionuclide classes and for all values for high zirconium oxidation. As all the source term variables are uniformly distributed from 0.0 to 1.0, and are uncorrelated, there are no columns for this information in Table 3.2-2 as there are in Table 2.3-2. There is a separate distribution for each radionuclide class for each variable in this table unless otherwia.e noted in the variable description. The different cases for each variable are noted in the description. Not all the cases considered by SURSOR are listed in Table 3.2-2; parameter values for other cases are determined internally in SURSOR, often from the values for the cases listed. For example, there is no distribution for FCONV for -late leak. The value of FCONV for late leak is derived from the distribution for another case. (See the listing of Subroutine FCONVC in Appendix B.) The variable identifiers given. in Table 3.5 -2 are used in t.everal ways in the source term analysis. Consider the first variable in Table .3.2-2: FCOR. FCOR in the equation for fission product release is the actual fraction of each fission product group released from the core to the vessel before or at VB for the observation in questi,n. But, FCOR is also used to refer to .the experts' aggregate distributions from which the nine values (one - for each radionuclide class or fission product group) for FCOR are chosen. Further, in the sampling process, FCOR is used to refer to the random number from the Uls which is used to select the values from these distributions. That is, as used'in sampling, FCOR defines a quantile in these distributions. The release fractions associated with this quantile are used in SURSOR as the FCOR values. Thus, in Table 3.2 2, the end use of each variable is given although the actual sampled variable is a random number between 0.0 and 1.0 used to select an actual value. 3.13
Most of the twelve variables in Table 3.2 2 have been described more fully in the preceding section. The distributions for FCOR, FVES, FCONV, FCC1, FCONC, FLATE, and DST were provided by the Source Term Expert Panel. These distributions, the reasoning that led each expert to his conclusions, and the aggregation _ of the individual distributions are fully described in NUREG/CR 4551, Volume 2, Part 4. VDF, SPRDF, LATEI, FISG, FOSG, and DFSPL are discussed briefly below; the distributions for these source tetm factors and more discussion of them may be found in Appendix B. The SGTR accidents with the secondary SRVs stuck open were not known to be significant to risk at Surry when the Source Term Expert Panel met for the last time. Therefore, a special ad hoc panel was convened to consider the factors FISG and FOSG. These factors are discussed briefly below; more detail may be found in NUREG/CR 4551, Volume 2 Part 6. The LATEI factor was considered by the expert panel for the BWRs, but the BWR distributions were not utilized directly for the PWRs as discussed in more detail in Appendix B of this report. VDF is the decontamination factor used fo- Event V when the release location is underwater. These accidents sre referred to as V-Wet accidents. For these tyr~s of accidents, SURSO' sets DFE to the value of VDF. The distribution for VDF was determined by the project staff. The range for VDF is from 1.6 to 5100; the median value is 6.2. VDF represents only scrubbing by passage through the pool of water overlying the break location. Any additional removal in the auxiliary building is accounted for by FCONV. The distribution for VDF is given in Appendix B. SPRDF retors to both the spray decontamination factor for the RCS (vessel) release, DFSPV, and the CCI spray decontamination factor, DFSPC. There is only one value for each of these - DFs; each DF applies to all radionuclide groups except the inert gases. Different spray distributions apply for CF at VB and for late CF. The value selected for DFSPC always is taken from the spray DF distribution for late CF, The value for DFSPV is taken from the early distribution for CF at VB, and from the spray DF distribution for late CF. However, the same random value between 0.0 and 1.0 from the LHS program is used to select both the RCS and CCI spray DF values. That is, the spray DF distributions are completely correlated. The spray DF dis-tributions were determined by the project staff. For the RCS release with CF at VB, there are two distributions for the spray DF. One applies if the RCS was at high pressure before VB. In this case most of the RCS release will escape- from- the - vessel just at VB , and the sprays will be very in-effective. The range of the spray DF distribution is from 1.0 (no effect) to 2.8; the median value is 1.6. For the RCS release with CF at VB with the RCS at low pressure before VB, much of the RCS release will have es-caped from the vessel before VB, and the sprays will be very effective for that portion of the RCS release. The range of this spray DF distribution is from 2.3 to 2800; the median value is 40. The distribution for the CCI spray DF distribution ranges from 6.7 to 3200; the median value is 28. The complete distributions are contained in Appendix B. LATEI refers to the evolution of iodine in volatile form from water in the containment late in the accident. Because of its volatile form (typically organic), this volatile iodine is all released to the environment as it is unaffected by all the removal nechanisms (pool scrubbing, sprays, deposi-3.14 '
tion, etc.). The release fraction determined by 1ATEI applies to all the iodine released from the fuel and retained in the containment in aqueous solution, which is expected to be the bulk of the iodine released from the vessel and remaining in the containment. In Surry, this iodine would be expected to be contained in the water in the sump. The sump water does not play the same role in heat removal that the suppression pool does in the Bk'R, so the results of the expert panel (which apply to Bk'Rs only) were not utilized directly. Instead, the distribution obtained specifically for PWRs in the first draft of this report was used. This is discussed further in Appendix B. The distribution used for LATEI ranges from 0.0 to 0.10; the median value is 0.05. FISC and FOSG are the release fractions used for the RCS release for SGTR accidents. FISO is the fraction released from the core that enters the steam generator; and FOSG is the fraction entering the steam generator which is released from the steam generator to the environment. The actual equation us;d for the early or RCS release is: ST(i) - [ FCOR(i) * ( FISG(i)
- FOSG(i) + [ 1.0 FISG(i) ]
- FVES(i)
- FCONV(i) / DFE(1) ) ] + DST (1). (Eq. 3.3)
As the material passing from the steam generator to the atmosphere bypasses the containment, the factors FCONV and DFE are not applied to this release path. For the SGTRs where the secondary system SRVs reclose, the distribu-tions for FISG and POSG were determined by the project staff. For the SGTRs where the secondary system SRVs stick - open, the distributions for FISG and FOSG were determined by an ad hoc expert panel. The panel provid-ed distributions for the product FISC
- FOSG for iodine, cesium, tellurium, and aerosols. There is no retention in the SGs for the noble gases.
Complete distributions for FISC and FOSG are listed in Appendix B. DFPSL is the decontamination factor for the late pool scrubbing of the CCI release. This DF is applied when the core debris is not coolable and CCI takes place under water. There are two distributions: one applies for a shallow pool (approximately 4 ft deep) that results if only the accumulator water enters the cavity, and the other distribution applies when the sprays are (or were) operating and the cavity is full (14 ft deep). For both the shallow and deep pool distributions, one distribution applies to the io-dine, cesium, barium, ruthenium, lanthanum, and cerium radionuclide classes, and another applies to the tellurium and strontium radionuclide classes. The distributions were determined by the NUREG-1150 project staff and are listed in Appendix B. 3.3 Results of Source Term Analysis This section presents the results of computing the source terms for the APBs produced by evaluating the APET. The APET's evaluation produced a large number of APBs, so, as in Section 2.5, only more likely and more important APBs are discussed here. However, source terms were computed for all the APBs for each of the 200 observations in the sample. The source term is composed of release fractions for the nine radionuclide groups for an early and a late release as well as release timing, release height, and 3,15
release energy. As discussed above, the source terms are computed by a i fast-running parametric computer code, SURSOR. Section- 3.3.1 presents the results for the internal initiators. Section 3.3.2 discusses the sensitivity analyses run for the internal initiators. l The accident progression analysis results for the fire initiators are I presented in Section 3.3.3 and sensitivity analyses for fires are presented in Section 3.3.4. The seismic results are given in four sections. The basic results based on the LLNL hazard curve are presented in Section 3.3.5 and sensitivity analyses utilizing the LLNL hazard curve are presented in Section 3.3.6. The seismic results based on the EPRI hazard curve are presented in Section 3.3.7 and sensitivity analyses utilizing the EPRI hazard curve are presented in Section 3.3.8. The tables in this section present only a very small portion of the output obtained by computing source terms for each APB. More detailed results are contained in Appendix B, and complete listings are available on computer media by request. 3.3.1 Results for Internal Initiators In a manner analogous to Section 2.5.1, the results of the source term analysis for internal initiators are presented for each PDS group. 3.3.1.1 Results for PDS Croun 1: Slow SBO. As discussed in Section 2.5.1.1, this PDS group consists of accidents in which all ac power is lost in the plant, but the steam turbine-driven AFWS operates for several hours. When the batteries deplete, control of the steam turbine-driven AFWS is lost and it fails. This PDS - group contains six PDSs: two have the RCS -intact at UTAF, two have failure of the RCP seals before UTAF, and two have stuck-open PORVs before UTAF. In four of the six PDSs, the operators depressurized the secondary system before UTAF, and in two PDSs they did not.- The PDSs in this group are listed in Table 2.2-2. VB is not inevi-table - for this PDS group as electric power may be recovered before the vessel fails. Small but nonzero releases are calculated by SURSOR in this case as fission products may escape to the containment through the PORVs or a temperature induced break before the arrest of core damage. Even though the containment does not fail in core damage arrest cases, design basis leakage results in a very small release. Table 2.51 lists the five most probable APBs for PDS Group 1, the five most probable APBs that have VB, and the five most probable APBs that have VB and early CF. Table 3.3-1 lists the mean source terms for those same APBs. Although the same bins are shown in both tables, and the structures of both tables are roughly analogous, there are some important differences in the nature of the material presented. First, Table 3.3-1 has two desig-nators for each APB. The first designator is the APB definition initially produced in the analysis of the APET; the second designator is-the rebinned definition which is used as input to SURSOR. Consider the first APB in Table 3.3 1: HDCDFCDBDFB. Following evaluation of the APET, it was re-binned to CDCDFCDBDDB as discussed in Section 2.4.2. The symbols used in these two representations for each APB are given in Sections 2.4.1 and 2.4.2. 3.16
The other difference between the nature of Tables 2.5 1 and 3.3-1 lies in the nature of the information presented. In Table 2.5 1, the bin itself was well defined, i.e., the characteristics of the bin did not vary from observation-to observation. The only item in the table that varied from observation to observation was the probability of the occurrence of the bin itself. Thus, Table 2.51 lists a conditional probability averaged over the 200 observations in the sample. In Table 3.3 1, the bin is still well defined, but, as many of the factors that are utilized in calculating the fission product release vary from observation to-observation, the source term for a specific bin varies with the observation. Thus, the entries in all columns in Table 3.3 1 except the Order and Bin columns represent averages over the 200 observations in the sample. For example, consider the first APB in Table 3.3-1: GDCDFCDBDDB. Of the 200 observations in the sample, 121 had nonzero conditional probabilities for this bin. As source terms are not computed for zero-probability bins, there are 121 source terms associated with APB GDCDFCDBDDB, These 121 source terms were summed and then divided by 121 to produce the mean source , term gisen in the first two lines of Table 3.3-1. The five most probable APBs and the five most probable APBs with VB for PDS Croup 1 did not have containment failure. As a result, the releases associated with these APBs are very small. When there is no containment failure, SURSOR describes releases with a single release segment rather than the two release segments used when there is containment failure. The five most probable APBs with VB and early containment failure have low conditional probabilities (see Table 2.5-1) but larger releases than the APBs vithout containment failure. The mean source terms in Table 3.3-1 can he ased to compare the releases associated with specific APBs. However, as these mean source terms are typically not calculated over the same sample elements, fine distinctions between source terms associated with different APBs may be lost in the averaging process.
. Table 3.3-1 presents mean source terms but does not contain any frequency information. In contrast, Figure 3.3-1 presents information on both source term size and frequency. Figure 3.3-1 summarizes the release fraction CCDFs for - for the iodine, cesium, strontium, and lanthanum radionuclide classes. It indicates the frequency with which different values of the release fraction are exceeded, and displays the uncertainty in that fre-quency. The curves in Figure 3.3-1 are derived in the following manner:
for each observation, evaluation of the APET produced a conditional probability for each APB. When multiplied by the frequency of the PDS group for that observation, a frequency for the APB is obtained. Calcula-tion of the source term for the APB gives a total release fraction for each APB. When all the APBs are considered, a curve of exceedance frequency vs. release fraction can be plotted for each observation. Figure 3.3-1 is a summary presentation of these curves for the 200 observations in the sample. Instead of placing all 200 curves on one figure, only four statistical measures are shown. These measures are generated by analyzing the curves in the vertical direction. For each release fraction on the abscissa, there are 200 values of the exceedance frequency (one for each sample t 3.17 '
element). From these 200 values it is possible to calculate mean, median (50th quantile), 95th quantile and 5th quantile values. When this is done for each value of the release fraction, the curves .n Figure 3.3-1 are obtained. Thus, Figure 3.3-1 provides information . on the relationship between the size of the release fractions associated with PDS Group 1 and the frequency at which these release fractions are exceeded, as well as the variation in that relationship between the observations in the sample. As an illustration of the information in Figure 3.3-1, the mean frequency (yr-1) at which a release fraction of 10-8 is exceeded due to PDS Group 1 is 2 x 10-5, 1 x 10-8, 6 x 10 7 and 4 x 10 7 for the iodine, cesium, strontium and lanthanum release classes, respectively. For a release fraction of 0.1, the corresponding mean exceedance frequencies are 1 x 10-7, 9 x 10-a, 3 x 10-e and < 10-8, respectively. The three quantiles (i.e., the median, 95th and 5th) provide an indication of the spread between observations, which is often large. Typically, the mean curves reach a point where they drop very rapidly and move above the 95th quantile curve. This happens when the mean curve is deminated by a few large observations; this often occurs for large release fractions because only a few of the sample obse rvations have nonzero exceedance frequencies for these large release fractions. Taken as a whole, the results in Figure 3.3-1 indicate that the occurrence of large source terms (e.g., release fractions 20.1) in conjunction with PDS Group 1 is very infrequent (less than 10 e for iodine and cesium and less than 10-a for Sr) . 3.3.1.2 Results for PDS Groun 2: LOCAs. This PDS group consists of accidents initiated by a break in the RCS pressure boundary, as discussed in Section 2.5.1.2. The breaks are of all (A, S, 3 S, 2 and S 3 ) sizes, in t this group. These PDSs result in core damage because one or more of the l' ECCS required to respond does not operate. The PDSs in this group are - l listed in Table 2.2-2. Four of the eight PDSs have the LPIS operating but l not injecting at UTAF, so the arrest of core damage before vessel failure is possible as discussed in Section 2.5.1.2. Even though the containment l does not fail in these core damsge arrest cases, design basis leakage L results in small, but nonzero , rele.as e s . l _ Table 2.5-2 lists the 10 most p obable /PBs for this PDS group and the five l- most probable APBs that have d and early containment failure (CF). The five most probab.J APBs that have VB are included in the 10 most probable bins. Table 3.3 2 lists the mean source terms for these same APBs. The source term consists of the release fractions, the release height and energy, and the times associated with the release. The release fractions give . the early (RCS) and, late (CCI) releases as fractions of the core in-ventory at the start of the accident. However, when there is no CF, or CF only in the final period due to BMT or overpressure, SURSOR sets the early release to zero and places the entire release into the late release por-tion. The three times (all in seconds) in Table 3.3-2 are the time the warning is given to evacuate the surrounding area, the time the release starts, and the duration of the release. The elevation of the release is given in meters, and the energy in watts. 3.18
Eight of the 10 most probable APBs for PDS Group 2 did not have containment failure and the releases associated with these J Bs are extremely small . The other two of the 10 most probable APBs had BMT in the final period. The releases for these APBs are larger than those with no containment failure, but still quita small. The five most probable APBs with VB and early CF have lower conditional probabilities (see Table 2.5-2) but larger releases than the APBs without containment failure. As with the APBs for PDS Group 1 that have VB and CF at VB, some of these APBs give rise to source terms in which the mean release fractions for iodine and cesium exceed 0.10. Figure 3.3-2 summarizes the release fraction CCDFs, and shows that the frequency at which iodine and cesium release fractions of 0.10 are exceeded a.a quite low, even at the 95th percentile. As would be expected, accidents that result in large releases are much more unlikely than those that result in small releases. 3.3.1.3 Results t'or PDS Groun 3: Fast SBO. This PDS group consists of accidents in which all ac power is lost in the plant and the steam turbine-driven AFWS fails at, or shortly after, the start of the accident. As discussed in Section 2.5.1.3, the Fast SB0 PDS group consists of only one PDS, TRRR-RSR. As in the Slow SB0 PDS group, if offsite electrical power is recovered for a Fast SB0 accident before the vessel fails, it may be possible to arrest the core degradation process and avoid vessel breach. Table 2.5-3 lists the five most probable APBs for the Fast SB0 PDS group, the five most protable APBs that have VB, and the five most probable APBs that have VB and early containment failure (CF). Table 3.3-3 lists the mean source terns for these same APBs. The source term consists of the release fractions, the release height and energy, and the times associated with the release. For the Fast S30 PDS group, the five most probable bins all have very low - source terms. All five APBs have no CF, and three of them have no VB as well. Three of the five most probable bins that have VB have no CF; the other two have BMT. The release fractions for the APBs with BMT are low as the failure of the containment occurs only after many days and the release point below ground. Because of the nature of this type of CF, not much effort has been devoted to estimating their value and they may be much lower than shown. As discussed above, for no CF or CF in-the final period, the early release is zero and the late release contains the entire amount estimated to pass to the atmosphere. The five most probable Fast SBO APBs with VB and early CF have lower conditional probabilities (see Table 2.5 3) but larger releases than the APBs without containment failure. Some of these APBs give rise to source terms in which the mean release fractions for iodine and cesium exceed
'O.10, but Figure 3.3-3 shows that the frequency at which iodine and cesium release fractions of 0.10 are exceeded are quite low, even at the 95th percentile.
3.3.1.4 Results for PDS Grouc 4: Event V. As discussed in Section 2.5.1.4, this PDS group consists of accidents in which the check valves between the RCS and the LPIS fail. Failure of the low pressure piplur. 3.19
produces a direct path from the RCS to the auxiliary building, bypassing the containment, and fails the LPIS as well. Experts considering the break location in the LPIS concluded that the probability was 0.85-that it would be low enough in the auxiliary building that the sater from the RCS and the RWST, escaping through the break, would form a pool covering the break by the time when core degradation commenced. This pool can remove and retain a significant portion of the release. When the release is scrubbed by a water pool over the break location, the accident is termed V-Wet, and when there is no pool it is termed V Dry. There is no possibility of avoiding VB or CCI in this PDS group. Due to the size of the containment bypass, I containment failure is not of_much interest. ' Table 2.5-4 lists the eight most probable APBs for the V PDS Croup and Table 3.3-4 lists the mean- source terms for these same APBs. The source term consists of the release fractions, the release height and energy, and the times associated with the release. The four most probable bins are V-Wet and the next four are V Dry. (The probability of the break location being under water is between 0.70 and 1.0.) The V Wet release fractions are considerably lower than the V-Dry release fractions as expected. The release fraction CCDF . summary curves in Figure 3.3-4 shows that the frequency at which iodine and cesium release fractions of 0.10 are exceeded are below 10-e/yr. Because this accident bypasses the containment, if it occurs'the releases are likely to be substantial. 3.3.1.5 Results for PDS Croyn.5: Transients. This PDS group consists of accidents in which the RCS is intact but there is no way to remove heat from the core (see Section 2.5.1.5). The AFWS fails at the start of the accident; bleed and feed is ineffective. In PDS TBYY-YNY, HPIS and LPIS are available but the operators cannot open the PORVs from the control room or have failed to do so before the onset of core damage. In the other PDS in thic group, TLYY-YNY, HPIS is failed but the LPIS is operable. In this l PDS group, the: probability . of a temperature-induced failure of the RCS pressure boundary.is_quite high- almost 0.90. Since the LPIS is operating at- the onset of core damage, the probability of arresting the core degradation process and avoiding VB is high. l Table 2.5-5' lists the 10 most probable APBs for the PDS group and the five most probable APBs that have VB and early CF. The five most probable APBs that have VB are included in the 10 most probable bins. Table 3.3-5 lists the mean source terms for these same 15 APBs. The 10 most probable bins all have no CF, and their release fractions are so low as to be negligible
-in an overall risk context.
The five most probable Transient APBs with VB and early CF have lower conditional- probabilities (see Table 2.5 5) but larger releases than the APBs without containment failure. =Some.of these APBs give rise to source terms in which the mean release fractions for iodine and cesium exceed 0.10, but Figure 3.3-5 shows that the frequency at which iodine- and cesium release - fractions of 0.10 are' exceeded are quite low, even at the 95th percentile. 3.20
3.3.1.6 Results for PDS Grouc 6: ATVS, This PDS group consists of accidents in which automatic control rod insertion fails to bring the nuclear reaction under control. The discussion in Section 2.5.1.6 points out that this PDS group consists of three PDSs, one with the RCS intact at UTAF, one with an S3 break, and one with an SGTR. In all three situations, the PORVs will be open at UTAF due to the - rate of steam generation in the core. The LPIS is operating but not injecting in the RCS-intact and SGTR PDSs. A-temperature induced break in the RCS, however, will allow the LPIS to inject successfully. The water from the RWST inj ected by the LPIS contains enough boron to shut down the reaction should the core be in a configuration where continued reaction is possible. Table 2.5-6 lists the 10 most probable APBs for the PDS group and the five most probable APBs that have VB and early CF or bypass, and Table 3.3-6 lists the source terms calculated for these . same 15 APBs. The six most probable bins have : no failure =of the containment and thus have very low releases. .The seventh and tenth most probable bins have bypass of the containment (SGTR) - and therefore have substantial releases although they have no VB due to the operation of the LPIS throughout the accident. Even
'in the ' absence of VB, SURSOR may calculate significant releases in these SCTR accidents.since the core degradation may not be arrested until it is I j quite well advanced. By this . time , a substantial portion of the fission i products may have been released from the core. The five most probable APBs L with VB and early CF or bypass all have SGTR and no CF. Whether the vessel . f ails - or not does not have a large effect on the computed release fractions. Figure 3.3 6 shows that the frequency at which iodine and i- cesium ; release - f ractions of 3.10 are exceeded are fairly low for this PDS l grcup in spite of the contribution from the SGTR initiators.
l 3.3.1.7 Results for PDS G ro_uo 7: SCTRs. As discussed in Section 2.5.1.7, this PDS group consists of accidents in which the initiating event isz the rupture of a steam - generator tube and the reaction is shut down l successfully. In two of the PDSs in this group, the RCS is depressurized quickly using the two unaffected SGs according to procedures and the SRVs on, the main steam lines from the . affected SG do not stick open. These accidents, denoted "G" SGTRs, are indicated by "SGTR" in the SGTR column of Table 2.5-7. In the other two PDSs in the SGTR PDS group, the RCS is not depressurized in a timely fashion, and the SRVs on the main steam-line from the affected SG stick open. These accidents, denoted "H" SGTRs, are L indicated by "SRV0" in the SCTR column of Table 2.5-7. Since all the APBs for.this PDS group have bypass of the containment, Table 2.5 7 lists-the 15 most probable APBs. Only two of the 15 most probable b % have the SRVs ! reclosing; the -other 13 bins result from the "H" SG% accidents in which the secondary SRVs are stuck open. PDS HINY NXY nas a higher frequency than the other three PDSs in this group combined. Table 3.3-7 lists the mean source terms for the same 15 APBs listed in Table 2.5-7. All - the - most probable APBs have fairly substantial release fractions.- Nots that the start of the release is about 14 h after the start of the accident for the "H" SGTRs. The evacuation warning. time is escimated to be much earlier than this so there is time for the evacuation to be completed. Thus few early fatalities are to be expected'in spite of the mean iodine release fractions over 0.10. Figure 3.3-7 shows that the 3.21
i mean exceedance frequencies for which iodine and cesium release fractions of 0.10 are below 10-s/yr. Because this accident bypasses the containment, if it occurs the releases are likely to be substantial. 1 3,3.1.8 Results for Summary Accident Prorression Bins. The preceding I seven subsections presented the source term results by PDS group. It is l also possible to group the source terms in other ways. These other- i groupings are called summary APBs . These summary APBs generally group accidents by the cause and time of containment failure. The bypass accidents are treated separately. Figure 3.3-8 shows the variation of the exceedance frequency with release fraction for the iodine, cesium, strontium, and lanthanum radionuclide classes for all the APBs which had an Alpha mode failure of the vessel and containment. (In an Alpha mode failure, an extremely energetic molten fuel-coolant interaction in the vessel fails the vessel and simultaneously generates a missile that fails the containment pressure boundary.) None of the APBs included in Figure 3.3-8 were initiated by a bypass event. Figure 3.3 8 shows that the frequency of a sizeable release from an Alpha event is l quite low. The curves for iodine and cesium in this figure indicate that there is a great range in the frequency of the Alpha event ' itself, but that, if the event occurs, the release fraction is likely to exceed 0.01, Figure'3.3 9 shows the variation of the exceedance frequency with release fraction for all the APBs in which the containment failed at VB with the RCS at high (> 200 psia) pressure at the time of VB, CF at VB with the vessel at high pressure indicates that direct containment heating (DCH) made a large contribution to the _ pressure rise in the containment at VB. l- Without a substantial pressure rise due to DCH, CF at VB is quite unlikely as CF due to hydrogen combustion and vessel blowdown alone is improbable. For each of the radionuclide classes, the mean curve lies above the 95th percentile curve. This indicates that the mean is largely determined by a L few observations that have very - large releases. Since the conditional l ;' probability of CF at VB due to DCH was found to be fairly' low in the APET-evaluation, this is not surprising. l Figure 3.3-10 shows the variation of the exceedance frequency with release fraction for all the APBs in which the containment failed at VB with the L RCS at low (< 200 psia) pressure at the time of VB. CF at VB,with the RCS l_ at such low pressures is quite unlikely since the pressure contribution- ?- from DCH is negligible. Figure 3.3-10 shows that the frequencies of CF at VB with the RCS at low pressure'are less than those for CF at VB with the RCS at high pressure. Neithor Figure 3.3-9 nor Figure 3.3-10 include APBs due to bypass events (V or S'JTR) . ! Figure 3.3-11 considers &1l the APBs in which the containment failed some-uurs or days after the vessel failed. Some of thesa railures are due to - hydrogen burns a few hours after VB, or to eventual overpressure due to lack of containment heat removal, but most result from BMT. The figure shows that these types of CF are much more frequent than CF at VB, but that the release fractions are likely to be an order of magnitude or more lower. I 3.22
l Figures 3.3 12 ' and 3.3 13 show the variation of the exceedance frequency with releare fraction for Event V. All the source terms for the V-Dry APBs l were analyzed to' produce Figure 3.312, while all the source terms for the ' V Vet APBs were analyzed to produce Figure 3.3 13. As expected, the V Dry release fractions . are larger than the V Vet release fractions due to the absence of the overlying water pool in the V-Dry accidents . The V Dry releases are, however, about an order of magnitude less likely than the V-Wet releases. Figures 3.3 14 and 3.3-15 consider all the APBs with SGTRs. Almost all < these SGTRs are initiating events; there is a very small portion of these APBs that result from temperature-induced SGTRs following the onset of core damage. (The temperature induced SCTRs are all "G: SGTRs, i.e., the secondary SRVs reclose.) As indicated by the discussion in subsection
- 2. 5.1. 6 - and 3. 3.1. 6, the "H" SGTRs are both more likely and more deleter-ious than the normal "G" SGTRs.
3.3.1.9 Summarv. When all the types of internally-initiated accidents at Surry are considered together, the exceedance frequency plots shown in Figure 3.3-16 are obtained. A plot is not shown for the noble gases since almost all of the noble gases (xenon and krypton) in the core are eventually released to the- environment whether the containment fails or , not. .The mean frequency of exceeding a release fraction of 0.10 for l iodine, cesium, and tellurium is on the order of 10-s/ year. The mean exceedance frequency-for release of 0.10 of the core strontium is somewhat , lower. The second sheet of Figure 3.3-16 shows the release fractions for i ruthenium, lanthanum, cerium, and berium, which are often treated together as aerosol species. The mean frequency of exceeding a release fraction of 0.01_ for these radionuclide - classes is on the order of 10 e/ year. The t' releases for the barium class are slightly higher than those for the other three' aerosol radionuclide classes. 3.3.1.10 gpntributors to Uncertaintv. Figure , 3. 3-16 provides information on the frequency at which release fractions of different sizes will be exceeded, Specifically, mean, median, 95th quantile, and 5th quantile values are given for the frequency at which release fractions will be. exceeded. Thus, Figure 3.3-16 can be viewed as presenting uncertainty
-analysis results for exceedance frequencies. The underlying exceedance frequency curves that gave rise to these results are shown in Section B.3 of Appendix B.
" ~the curves in Figure 3. 3 and in Appendix B.3 show, there is significant uncertainty in the frequency' at which a release fraction will be exceeded. Due to the complexity of the underlying analysis and the concurrent variation of a large number of variables within this analysis, it is difficult to ascertain the cause of this uncertainty on the basis of .!
a simple inspection of the results. However, numerical sensitivity analysis techniques provide a systematic way of investigating the observed variation in exceedance frequencies. This section presents the results of using regression based sensitivity analysis techniques to examine the variability associated with radionuclide releases to the environment. Two dependent variables will be considered 3.23
for individual radionuclide release classes. The first dependent variable is the annual release fraction (units: fraction /yr) for a radionuclide release class. This variable is the fraction of the inventory associated with a release class that is expected to be released each year. For a given sample element, this variable is obtained by multiplying the release fraction associated with each accident progression bin by the bin's fre-quency and then summing these products. This variable can be viewed as the result of reducing each of the curves in Figure B.3-1 to a single number. Further, this variable is analogous to the mean consequence results (units: consequence /yr) presented in Chapter 5. The second dependent variable is the exceedance frequency associated with individual release fractions. As can be seen from the curves in Figure B.3-1, there are 200 exceedance frequencies (i.e., one for each sample element) associated with each release fraction for individual radionuclide release classes. For a given sample element, release fraction and radionuclide release class, the exceedance frequency is obtained by summing the frequencies of all accident progression bins that result in release fractions as large or larger than the one under consideration. The uncertainty analysis techniques used in this study can be viewed as creating a mapping from analysis input to analysis results. The variables sampled in the generation of this mapping are presented in Tables 2.2 9, 2.3 2, and 3.2 2. For convenience, these variables are also summarized in Table 3.3 8 of this section. The variables listed in Table 3.3-8 are the independent variables in the sensitivity studies presented in this section. A . series of correlated variables is represented by one variable. For example, although there arc variables for three cases for core damage arrest, CDARREST, they are all correlated together (with a rank correlation of 1.0) and only -one variable appears in the ser sitivity analysis and in Table 3.3-8. The greatest collapse of this type is for the pressure rise at VB, variable PRISE VB. The 18 variables for :he cases other than low pressure have a rank correlation of 1.0 and appear ss one variable in Table 3.3-8. Note that variable FR-ZROX from Table 2 . .' - 2 does not appear in Table 3.3 8. FR-ZRCX has a rank correlation of 1.0 vith TI HOTLG, so only one of these two variables appears in the sensitivity analysis. As TI-Il0 TLC appears in the APET before FR ZROX, TI-110TLG was chosen to represent both variables. Sensitivity analysis results for annual release fractions are presented in Table 3.3-9. Specifically, this table contains the results of performing a stepwise regression on the annual release fraction for each radionuclide release class. This analysis was performed with the STEP _ program.1 In the analysis, a variable was required to be significant at 0.02 a-level' to enter a regression model and to remain significant at the 0.05 a-level to be retained in a regression model. Further, the behavior of the PRESS crite rion2 and individual R2 values were considered in selecting the actual stopping points for the presented regressions. The PRESS criterion is used to assure that the selected regression model is not overfitting the data on which it is based. The analyses were tried with raw (i.e., untransformed) and rank-transformed data.3 The analyses with rank-transformed data consistently performed as well as or better than the analyses with raw data. Therefore, Table 3.3 9 only presents the results of regression analyses performed with rank-transformed data. ' 3.24
Table 3.3-9 presents the results of a step >ise regression analysis for each radionuclide release class. For each release class, the variables are listed in the order that they entered the regression analysis. Further, the table also gives the standardized regression coefficients for the variables in the final regression model and the R2 values that result with the entry of successive variables into the model. Variable importance is indicated by the order in which variables entered a regression, the change in R2 values with the entry of successive variables, and the size of the standardized regression coefficients in the final regression model. Further, the tendency of a dependent variable to increase and decrease with an independent variable is indicated by a positive regression coefficient, and the tendency of a dependent variable to decrease when an independent variable increases is indicated by a negative regression coefficient. Table 3.3-9 shows that the release rates for the more volatile radionuclide classes to be more affected by the variables that determine the accident frequency (e.g. , V TRAIN, IE SGTR) and for the less volatile radionuclide classes to be more affected by the variables that determine the amounts released (e.g., FCOR, FISGFOSG). For the noble or inert gases, the first seven variables selected in the regression analysis are all related to the accident frequency, and the source term variables have a very small impact. l This reflects the fact that all of the noble gases are released eventually in almost every core damage accident. Thus, it is the frequencies of the accidents - themselves that are important. (DG FSTRT, IE-LOSP, and DG FRUN are important variables in determining the frequency of the SBos.) The other extreme is ruthenium, where 85% of the variability is accounted for i by FCOR and FISGFOSG. This pattern arises because the least volatile
- fission products are-easily removed by both natural and engineered removal mechanisms. Thus the variability in the factors - that determine the amount released are important. For iodine and cesium, the most important variables are FISGFOSC, IE SGTR, V-TRAIN, and FCOR. This reflects the importance of the bypass accidents when the total amount released is the primary measure under consideration.
The fraction of the variability in the annual release rates which the regression models presented in Table 3.3-9 can explain ranges from 0.51 for the noble gases to 0.89 for ruthenium. For most of the radionuclide classes, the regression model accounts for 0.70 to 0.80 of the variability. There is a definite trend for the explained variability to increase as the volatility of the radionuclide class decreases, This implies that a less complex regression model is more suitable for these fission products, which in turn may be due to the importance of removal mechanisms in the RCS. i Nonlinear regression models were also utilized. By including cross-product variables in the regression analysis, the amount of the variability ac-l counted for could be increased moderately for noble gases, iodine, cesium, and tellurium. The important product variables are V-INIT
- FCOR, SG INIT
- FCOR
- FISGFOSG, and DC-FSTRT
- IE LOSP. The product SG-INIT
- FCOR
- FISGFOSG is the first variable chosen for iodine, cesium and tellurium.
For cesium and tellurium this variable alone accounts for 60% of the variability. 3.25
The results given in Table 3.3 9 come from a regression analysis of the annual release rates, which might also be denoted expected release fraction rates. Annual release rates are obtained by reducing each of the curves shown in Appendix B.3 to a single value with the units of fracchn/ year. Forming the annual release rates produces a single value for each obser-rvation and each radionuclide class which incorporates both the ts ize and frequency of the release. Although this single number is useful, in forming it, all information about the relationship between size or the release and the frequency of the release is lost. The release fraction CCDFs shown in Appendix B.3 are more basic results than the annual release rates analyzed to produce Table 3.3-9. Because of the importance of the release fraction CC0Fs, a sensitivity analysis was performed to determine which variables controlled the frequency at which release fractions are exceeded. Partial rank correlation coefficients (PRCCs) were calculated between the sampled variables and the frequency at which release fractions are exceeded.' A partial correlation coefficient (PCC) provides a measure of the strength of the linear relationship between two variables after the linear effects of all other variables in the analysis have been removed.- A PCC close to 1.0 means that both variables increase and decrease together; a PCC close to -1.0 means that one variable increases while the other decreases! a PCC close to 0.0 means that there is little relationship between the behavior of the two variables. Conceptually, a PCC between two variables is obtain-ed by constructing a regression model for the two' variables. The PCC is then defined to be the positive square root of the R2 value multiplied by the sign of the re5ression coefficient. Further, when the two variables l> under consideration are correlated with other variables in the analysis, a i correction is made for this correlation before the regression model is cons truc te d. ' A PRCC instead of a PCC results when a PCC is calculated I from rank transformed data. The PRCC reduces the effects of nonlinearities and outliers relative to the PCC. The PRCC analyses performed for Surry are based on the same data used to
~
l l_ construct the mean and quantile curves for the radionuclide_ classes that l are shown in Figure 3.3 16. The starting point is ' the family of 200 i exceedance frequency curves shown in Appendix B.3. Each curve represents one observation in the sample; so for each release fraction on the abscissa there are 200 exceedance frequencies. For each value on the release fraction axis, the PCCSRC program' calculates the PRCCs between the - exceedance frequency and the individual variables listed in Table 3.3-8. The results of this computation are presented as curves on a plot in which the abscissa is the release fraction and the ordinate is the value of the PRCC. Such curves are presented 1. Figures 3.3 17, 3.3-18, 3.3-19, and 3.3 20 ~ for iodine, cesium, strontium, and lanthanum radionuclide classes. To reduce the number of curves plotted, a variable is included in these plots only if the maximum absolute value of its PRCC exceeds 0.50. Figure 3.3-17 presents the results of the PRCC analysis for iodine. The curves for the important variables have been placed on two plots instead of one to make the results easier to read. For small release fractions, the 3.26
Table 3.3 9 presents the results of a stepwise regression analysis for each radionuclide relcase class. For each release class, the variables are listed in the order that they entered the regression analysis. Further, the tabic also gives the standardized regression coefficients for the variables in the final regression model ard the R2 values that result dih the entry of successive variables into the model. Variable importance is indicated by the order in which variables entered a regression, the change in R2 values with the entry of successive variables, and the size of the standardized regression coefficients in the final regression model. Further, the tendency of a dependent variable to increase and decrease with an independent variable is indicated by a positive regression coefficient, and the tendency of a dependent variable to decrease when an independent variable increases is indicated by a negative regression coefficient. Table 3.3 9 shows that the release rates for the more volatile radionuclide classes to be more affected by the variables that determine the accident frequency (e.g. , V TRAIN, IE SGTR) and for the less volatile radionuclide classen to be more affected by the variables that determine the amounts i released (e.g., FCOR, FISGFOSG), For the noble or inert gases, the first seven variables selected in the regression analysis a.;e all related to the accident frequency, and the source term variables have a very small impact. This reflects the fact that all of the noble gases are released eventually in almost every core damage accident. Thus, it is the frequencies of the accidents themselves that are important. (DG-FSTRT, IE LOSP, and DG FRUN are important variables in determining the frequency of the SBos.) The other extreme is ruthenium, where 85% of the variability is accounted for by FCOR and FISGFOSG. This pattern arises because the least volatile fission products are easily removed by both natural and enginewed removal mechanisms. Thus the variability in the factors that determine the amount released are important. For iodine and cesium, the most important variables are FISGFOSG, IE SGTR, V TRAIN, and FCOR. This reflects the importance of the bypass accidents when the total amount released is the primary measure under consideration. The fraction of the variability in the annual release rates which the regression models presented in Table 3.3-9 can explain ranges from 0.51 for the noble gases to 0.89 for ruthenium, For most of the radionuclide classes, the regression model accounts for 0.70 to 0.80 of the variability. There is a definite trend for the explained variability to increase as the volatility of the radionuclide class decreases. This implies that a less complex regression model is more suitable for these fission products, which in turn may be due to the importance of removal mechanisms in the RCS. Nonlinear regression models were also utilized. By including cross-product variables in the regression analysis, the amount of the variability ac-counted for could be increased moderately for noble gases, iodine, cesium, and tellurium. The important product variables are V INIT
- FCOR, SG-INIT
-* FCOR
- FISGFOSG , and DC-FSTRT
- IE-LOSP. The product SG-INIT
- FCOR
- FISGPOSG is the first variable chosen for iodine, cesium and tellurium.
For cesium and tellurium this variabic alone accounts for 60% of the variability. 3.25
The results given in Table 3. 3-9 come from a regression analysis of the annual release rates, which might also be denoted expected release fraction rates. Annual release rates are obtained by reducing each of the curves shown in Appendix B.3 to a single value with the units of fraction / year. Forming the annual release rates produces a single value for each obser-rvation and each radionuclide class which incorporates both the size and frequency of the release. Although this single number is useful, in forming it, all information about the relationship between size of the release and the frequency of the release is lost. The release fraction CCDFs shown in Appendix B.3 are more basic results than the annual release rates analyzed to produce Table 3.3-9. Because of the importance of the release fraction CCDFs, a sensitivity analysis was performed to determine which variables controlled the frequency at which release fractions are exceeded. Partial rank correlation coefficients (PRCCs) were calculated between the sampled variables and the frequency at which release fractions are exceeded.' A partial correlation coefficient (PCC) provides a measure of the strength of the-linear relationship between two variables af ter the linear effects of all other variables in the analysis have been removed. A PCC close to 1.0 means that both variables increase and decrease together; a PCC close to 1.0 means that one variable increases while the other decreases; a PCC close to 0.0 means that there is little relationship between the behavior of the two variables. Conceptually, a PCC between two variables is obtain-ed by constructing a regression model for the two variables. The PCC is then defined to be the positive square root of the R2 value multiplied by the sign of the regression coefficient. Further, when the two variables under consideration are correlated with other variables in the analysis, a correction is made for this correlation before the regression model is constructed.' A PRCC 'instead of a PCC results when a PCC is calculated from rank-transformed data. The PRCC reduces the effects of nonlinearities and outliers relative to the PCC. The PRCC analyses performed for Surry are based on the same data used to construct the mean and quantile curves for the radionuclido classes that are shown in Figure 3.3-16. The starting point is the family of 200 exceedance frequency curves shown in Appendix B.3. Each curve represents one observation in the sample; so for each release fraction on the abscissa there are 200 exceedance frequencies. For each value on the release fraction axis, the PCCSRC program 4 calculates the PRCCs between the exceedance frequency and the individual variablea listed in Table 3.3 8. The results of this computation are presented as curves on a plot in which the abscissa is the release traction and the ordinate is the value of the PRCC. Such curves are presented in Figures 3.3 17, 3.3-18, 3.3-19, and 3.3 20 for iodine, cesium, strontium, and lanthanum radionuclide classes. To reduce the number of curves plotted, a variable is included in these plots only if the maximum absolute value of its PRCC exceeds 0.50. Figure 3.3 17 presents the results of the PRCC analysis for iodine. The curvea for the important variables have been placed on two plots instead of one to make the results easier to read. For small release fractions, the 3.26
variability in the exceedance frequency is dominated by variables that determine the frequency of SB0 accidents,; but for large release fractions, this variability is dominated by variables that determine the frequency of bypass accidents (V-TRAIN, IE-SCTR) and source term variables (FCOR, FISGFOSG, FCONV) _ The curves for all the PRCCs approach 0.0 as the release fraction approaches 1.0 because the exceedance frequencies approach 0.0 as the release fraction approach 1.0 for all observations in the sample. For some sample elements and for some radionuclide groups, the exceedance fre-quencies reach 0.0 for release fractions substantially below 1.0. The bimodal form taken by the PRCC curve for LATEI in Figure 3.3 17 is very interesting. The high values of the PRCC for IATEI for release fractions between 3 x 10-5 and 3x 10 4 coincide with the highest releases for no containment failure. (For no CF, there is a t. mall release due to design basis leakage even though the containment remains intact.) The largest iodine releases for no containment failure are predominantly due to late revolatilization, since these releases are in volatile forms that are not affected by removal mechanisms. As a result, the uncertainty in the re-lease fractions depends strongly on LATEI in this region. The low values of the PRCC for LATEI between 3 x 10-4 and 3 x 10-3 occur in a range in which very few releases occur. Iodine release fractions tend to be below 3 x 10-4 for no CF, and above 3 x 10-3 for CF. Iodine release fractions for late CF are usually between 3 x 10-3 and 10-1, and iodine release fractions for bypass or early CF are usually between 10-2 and 1. Thus, the high values of the PRCC for 1ATP.I between 3 x 10-3 and 10-1 coincide with the releases due to late containment failure. This agrees with the observation that iodine releases for late CF (late hydrogen burn, eventual over-pressure, or BMT) are often dominated by the late revolatilization release. Figure 3.3 18 presents the results of the PRCC analysis for the cesium radionuclide class. For small release fractions, the variability in the exceedance frequency is dominated by variables that determine the frequency of SB0 (IE IDSP, DG-FSTRT) and by variables that determine the frequency of bypass accidents (V-TRAIN, IE SGTR). SPRDF is also important for the very lowest release fractions, since the sprays are very effective in reducing the size of the releases. The bypass initiator variables continue to be important in the middle range of the release fractions. For the highest release fractions, source term variables are the the most important. Figure 3.3-19 presents the results of the PRCC analysis for.Sr. For small release fractions, the variability in the exceedance frequency _is dominated by variables that dotarmine the frequency of bypass accidents (V-TRAIN, IE-SGTR). FCCI is also important for the very lowest release fractions. In contrast to the situation for iodine and cesium, none of the variables that-determine the frequency of SB0 (IE-LOSP, DG FSTRT) are important for stron-tium. The source term variables FCOR and FCCI are the most important variables at the higher release fractions. Figure 3.3 20 presents the PRCC results for the lanthanum radionuclide class. This figure looks much like that for strontium, but compressed to the left since the fission products in the lanthanum group are less volatile than strontium. For small release fractions, the variability in the exceedance frequency is dominated by bypass initiators; for large 3.27
release fractions FCOR and FCCI are the most important variables. The PRCC curves all drop to 0.0 at a release fraction of about 0.20 since that is , the highest release fraction computed for the lanthanum group in any l observation in the sample. J
' Plots similar to Figures 3.3 17, 3,3-18, 3.3 19, and 3.3-20 can also be generated for standardized rank regression coefficients (SRRCs). Since SRRCs are always less than PRCCs for uncorrelated variables, plots of PRCCs are more spread out and therefore easier to read. As PRCCs and SRRCs have -the same intuitive, although not literal, interpretation, only plots of PRCCs have been - prepared. Additional discussion of the relationship ,
between PRCCs and SRRCs may be found in Reference 4. 3.3.2 Results of Second Samole for Internal Initiators The probabilistic risk assessments in this report use Monte Carlo (i.e., Latin Hypercube)5 techniques in the propagation of uncertainties. It is desirable to have an estimate of the sampling variability that results from propagating uncertainties in this manner. Unfortuntely, a theoretical basis does not exist for obtaining such an estimate for the 'a of Latin , Hypercube Sampling- in conjunction with an arbitrary model. However, it is possible to see the effects of the- sampling process . by performing an analysis with two independently generated samples and then comparing the resulta . obtained with these two samples. This section presents - such a comparison for source term results obtained in the analysis of internal initiators at Surry. The source term results for internal initiators at Surry are summarized in Section 3.3.1,9. Specifically, Figure 3.3-16 gives mean, median, . 5th quantile and 95th quantile curves for exceedance frequencies associated with individual release fractions. In turn, the curves in Figure 3.3 16 are summaries of the more detailed results in Figure B.3-1. The individual curves in Figure B.3 1 result from propagating a Latin Hypercube sample of size 200 through the probabilistic risk assessment for. internal initiators. This sample is based on the variables in Table 3.3 8 (see also Tables 2.2- '9, 2.3 2, and 3.2-2). Each curve in Figure B.3 1 la derived from a single element of the Latin Hypercube Sample, and as a ' result, the curves in -Figure 3. 3 16 - provide a summary of the results obtained with the entire sample. Thus, examination'of the effects of two independent samples on the curves. in Figure 3.3 16 provides a way to investigate the impact of the sampling procedures being used. - To this 'end, n' second sample of size 200 was generated (using the same variables as in the first sample) and then propagated through the analysis. .This resulted in a second set of curves of the form 'shown in Figure B.3-1 ane* .in a corresponding set of summary curves of the form shown in Figure 3.3-16, For ease of comparison, the original summary curves appearbg~in Figure 3.3 16 and the analogous summary curves generated with the second sample are plotted together in Figure 3.3 21. 3.28
l The similarity of the corresponding curves in Figure 3.3 21 shows that very little uncertainty is being introduced into the analysis by the sampling process; Specifically, the median and mean curves are very similar; the 5th and -95th quantiles are also quite similar but show more variability than the median and mean curves. Overall, the median curve is most stable, although the mean curve is also quite stable, with this stability decreasing somewhat for larger release fractions. The 5th and 95th quantile curves are less stable than the median and . mean curves because they are effectively determined by a smaller number of the sample elements. Similarly, the mean curve tends to be less stable for the larger release fractions because the exceedance frequencies for most sample elements are then zero and so the mean value is driven by a small number of nonzero values. A stopwise regression on the annual release fractiot, was also performed for the second sample for - internal initiators. Each radionuclide release class was analyzed.with the STEP programt and results analogous to those in Table 3.3-9 were obtained. The analysis is described in Section 3.3.1.10 above. Table 3.3 10 compares the results of the stepwise regression analyses for the first and second sample for four radionuclide classes: l noble gas, iodine, strontium, 'and lanthanum. As for Table 3.3-9, the analyses were performed.with rank-transformed data.
-Variablo importance is indicated by the order in which variables entered a regression, the change in R2-values with the entry of successive variables, and the size of the standardized regression coefficients .in the final re-grossion ' model. Table 3.3-10 shows that the larger contributors are similar in both samples. There is considerable variation between the samples -for variables whose inclucier. !ncreases R2 by only 0.02 or _0.01. ,
The stepwise regression . analysis accounts for more of the variablility for
- the volatile radionuclide classes for Sample 2 than it does for Sample 1.
For all the radionuclide classes, the first four variables that enter the analysis. for Sample 1 are the same as those for Sample 2, although the order may be- different. as . in the results for noble gas in Table 3.3-10. The. dif ferences in the R2 values due to the , first four variables are pronounced only -- for noble gas - and iodine. Cesium and tellurium are more like strontium is that the R2 values for the first four variables for Sample 1 and Sample 2 are very close. Based'on the release fraction CCDFs, PRCCs'were also calculated for Sample 2 and curves equivalent to those shown in Figures ' 3. b 17 through 3.3-20 were obtained. Figure 3.3-22 compares the PRCC curves for two represent-ative variables for the iodine release fractions. The curves are quite similar. .The reasons for the bimodal form taken by the PRCC curve for LLATEI . ' arc discussed in Section 3.3.1.10. The bimodal form observed- for Sample 1 is closely reproduced by Sample 2. The PRCC results based on analysis of the CCDFs give generally closer agreement between the two sampics than the stepwise ro6ression results-based on the mean annual release. This is due to the fact that much information is lost in forming the annual release values. The annual ~ release value has contributions from the relatively frequent small releasts as well as the very unlikely large releases. Figures 3.3 17 through 3.3 20 3.29
show that the variables which account for the most uncertainty for large releases are usually different than those that are the largest contributors for small releases. Therefore it is not surprising that the agreement between the two samples is closer for the PRCC curves based on the CCDFs than it.is for the stepwise regression coefficients based on the annual release rates. This comparison of results for two independent samples is continued in Section 5.1.2 where final risk results are compared. As in this section, the results for the two sampics are very similar, indicating that little imprecision is being introduced into the analysis by the sampling process itself. 3,30
i coowowowoe oooooooooo NNNNM NNMne
???T"????? ?????????? Y?YY??????
a W M U Mt d W W W W' WWWNWWWNUM WNMMMWMMMW dAdidasada iadidAdadd dddaAddJaa ono-cyoNom ooomononom nemnNNmeau
. ?????.???? ?????????? ??????????
u WWWNWNWMWN WNWWWWWWWM NWWNWWMMMM
&adAsdidad dJ&didaddi dadidadsid ogogonogoo oo-wog???? oso- nmrnNnne-.
T.?.???.?? ?????. wwwwwgwwww
???????*??
e.wwwwwgw-3 w w w w w w w w w g 'r a a * . a a. .a6 ononosoNo-a.a a.a.a. . .9.a cau~ocoso.
- a.
- a . a 9 *. a.
-Nwnw44nes..
oHoNogadoH oHodadoNoM m<M4 NmNnHH
?????.???? ?????????? ??????????
. A w
- a. e. w w" 9 9aaaan w w w. w w w w w w u g w w~ w w c.wa 9 9 a. . a. 99 ggggw*wwwww .aan" "
8 . . . .
- ano-onoNos osonoNoeo. NavnN -N N 9
. oHoHeHoNoo oooaooeooo NNNNHNNMee e u
?????????? ?????????? ??????????
. WWWMWWWWWW WWWWWNWWWW MMMMMWWNEW 8 &adAdaaisi a i a n' a A a A d i AaAaddAdaA 2 ooooooooor oaoaooooom N-NN~Nnn<<
" . ?????????f ?????????? ??????????
" WMNNENWMJM WNWNEMEMWM MWMMMMWWWW addAaddisi diaaddadia AAdidJAdda oSocoooMoS GomomoSoS HNNNM Nd *MM
$ ?Y$f$ $ $ $ $ $ Y ?
m e WWWuMWWnWN MnWWWWWmWW WnWMMWmMMW y daidindAdA diaddAssad adaddiaAda O M ononononov ovo<onomo. M NMNMNN Nmm i H
.$g m
WNWWWMENEW-sasasasasa WWWWWWWWWW sasadasasx WWNNWWNMWM asaaaaaaai
~
aMomoMomoM NHHNHHHMM M
-% o eve?e???e? eMomoM9 e????? ???mom ??????????
l -- A. , U R WNWWWMWWWW WWWWWWWWWW WMMMNNWWMW . dJsiadadas assididdad idaddaddaa a o m0m gh= .O evo<ovowo* ovowo<ovo. N4 N<M < N4 m<
- ::: .?????????? ?????????? ??????????' 3 -
$
- m, A** WWWWWWWWWW WWWWWWWWWW *UNUNWWWNNN j myuy a2 sasasesesa sasasasase naaaaaaaaaa ,
a . m 8 . <<<<<<<<<< <<<<<<<*** he<<<<<<<<<
=j a-1 li a siliEEEE
- isia aa E555555555 jiiiiiiiiii I*
a::s::aia ,aaaaaaaaaa
- C #
9 . oooooooooo oooooooooo eh eh ennsonn B u ih ma
!!!i!!!!!! Bii!!!!!!!! esiliiiiiii sessssssss .s<ssssssss .aaassaassa ii H > >
W W 0 2 M . W o o o o e o o a o o N M M. H H g a
? ? ? ? ? u? ? w? w? w? u?
aw ? r ? ? ? % la w w w w
- a. 9 9 w
a. aw w goo 9a 9- 9 gte '. w 9 w a . w o e o o. o o o o a e e d. t 4 h M M M g a e * * < C4 4 4 4 4 C4 4 < < < d
-g- ,? ? ? ? ? g? ? ? ? V g? ?? ? ?-
e- mmnnn nnnnn n n m. n n H , 3" ja a ' a a *a a a aa ya a ' ' ' 3 a a a .
$2 2" d d $5 2 E m8a t "8 ~S""WW I o2 90 2 o" 2 0 2" N
$ NO $0 $ $isi Oh -
18828888 n la ,18882888888 ,s ollWon !! t. h h @ @ a=hw E N m <n he m o e n e h4 m a n
-a n n N
3 . 3 '.
l l l I T 1 OE-4 C 10E-4 l o , .. ,, ., o i--- ,---- t * ,, , *- s , , t - Percentile .
- 1 _
95th 3 1.0E-5 ',
.o -
$o .10E-5 7 ^ Me~c{ _
o e
, o _ 50th u -.
e u - ------.
. '- - - - - - - . . , ,- - .S.i n - - . _.
g 10E-6 j ., '., , g 10E-6
- a. 3 . .
a ,, y : '
'. s : w : ;
g . .. - w tt LOE-7 7
- 10E-7 '
7
. - t ,
, . Percent.de - .
o - 95th . ' o - c --------- ~ c
~o 10E-8 Weon '. ;, o 10E-B e E
- 50th .
.~
V e e : 5th * *L e . . o ~ - - - - - - - - . .- .- o
*s D t0E-9 - """".'"'>""'~.'.O '
- D t0E-9 . . . " '-' "" .'"' .. "d '
1.0E-610E-510E-410E-310E-21.0E-1 t0E0 w 10E-61.0E-510E-4 LOE-31.0E-21.0E-1 1.0E0 u Release fraction For i Release Froction For Cs M t 10E-4 g _. ..., ,. . . , . . . . . , ...,m m, , .,
$,10E-4 , . , , m ., ., .,
[ : . Percentile : : [ Percentile - go 10E-5
--95th -- _
g o 10E-5 7 --95th - - - E 50th
~
i-u e - 50lh - u e 10E-6 i. 10E-6
,.a.:
=
o p.-... ' r w
, 3 -, :
~, w ,
yt0E-7 , ',-
, y LDE-7 7 j
I e -
'. e * '
' o - o C - C o 10E-8 *
-m - , , o 10E-8 r e -
t ., i e : . e , u , _ e _ . o - D LOE-9 . . " " . . " " . . " " " ' '.- D LOE-9 .'" . .."".."'k. ' 10E-61.0E-5 LOE-410E-31.0E-210E-1 10E0 10E-610E-510E-4 t0E-310E-21.0E-1 LOE0 Release Fraction For Sr Release Fraction for La Figure 3.3-1. Exceedance Frequencies for Release Fractions for Surry Internal Initiators. PDS Group 1: Slow SBO l.
. _ , . . ~ . . , y __ . . . _ ., _ . . _ . . . _ , _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . . _ _ _ _ _ ._ _
i I9I9I9IfIY E9I9IIIff9 EY?Ef9@I?f a exsusmens: smessusmus vesvanesen sidadAiais idiniadasi AndadAddid e I 0808
+ + 0. I 0. I.
I. I 0 I%. I S. gn. twtw.t."?Ia.t.
*5 u susssssnes dasianidad idadisin d seuxwxem e ensusmnens g daaidadidd IEIYIEffI9 Iff9I!IIIE suussusess exsmswssse YEfff??If?
unumenurus aisdadidad saidisaisi didadadinA n SOI.I:I.I0
+ 8. . *?d. 8+ 2. I *. I. . **8.s.'s.VIs.I
?? ?
g sMuneNsNew resNsussJs xWhJunwese
- saiasisdia disidAinsi diAddisidi U
$ u $$$$$$I$I$ $$$$$h$$$$ $$$$$$$I95 e
i a smensessex sidAdi&A&J iaidaddaid everessass menseneses Aaadaadiiu eNnwnNmemN l- A .e ?? " ? ? IecoImoneo????? oofo@eIoIf ??? ??????????
, - ennesusxsn snewswsxsw sessmensve diduidiada adiasidadi aidAdadida coge kY+i'kge o e o gk gYgug~g. NwnwNNanNN i i+k kki k kk O sMsWsusMun sussevsNwM WMwesMWMNN addAsAdian idiuaddasA Adidadadad M e e M e eon N NwNwNNMNNN
* + +i+Y+i+k +Y i+igNon + hi fikYEiikik
$" sWsWsusMsM susMswsNum svWMsMNwWW my aidaddiada iAddisiada saaAssidas p u- EEIhEEEEEE khkk!kIIEE f f f YhEEEEY' a sussenswes sesmesssww memunnesse insisissia saiddA& Add iduadiddia
}l[
.0 e8 ece IttTITIgIt e e
~ m la; + geoeIeoecettit?
+ttt NeNeNeeINeTV
???titt
- i88 lAn~
w MM8M**.45.4M peomoeomoe
**ns.85898 ts.nWW8n9 4n 1 Hj g omomewowoe gNNNNNNwwww ,
OW 3 mu* o eeeveeveee eeeeenenee heeeeeee Ida TiiYi????? TTT??????? YETITTY?geg + T+
-@ M 33" WWWMWMMMNN
.;;;;;;;;;; iais;daaa;WNWMMMWNNW gEWWWWWWsWW i y a= ,aiaeaaaAaa g 1
C g' 3 e t - {s-e IIIIIIIIII IIIIIIIII* esssssssus sessexsxsi IIIIIIIIII
#wsvemensWs 3
- a3 idissisisi siaisidise ,adaAwAAAaa i a 3 o e. - o a w a w- - w 7 IIo g ? I IIw ?? ? uf ? ? ? 2 a
ss s' je s?
+
lS ~
's s s e s s s s N r ,
n i i i i i i i :a a i as A a a A- . , 4 m 4 me e e e e a e e e e cv e e e e M, l-n=annnnn hi ? ? i i ? @ @ ? E i i T i Y e nnnw n nnnnn B s~ .a aa a a w a a a a a sa 4 aaaa :
. lilllllllllllllllllll!Illlllllli !.
a : vf $4 o e N n e n e
- e a ,
- =g g g p g .
11 3.33 1
T,t0E-4 , , ., , , , T,t0E-4 , , , , , x x Fercentile [ [ S 1.0E-5
, p 10E-5 - 95th
- gena- -
g
~
g . u
- scrs-e.
ce 1.0E-6 ------- .. s. 1.0E-6 r . P -- , e a a
' y w : . .
- F 7
r"'"-----------__,,,-
' .
- 10E-7 .
sc' 1.0E-7 r.p -- ; 1 et gg . e . e - - . . o - o 95th , . k E 10E-8 ~ ~ M~ ~ 5 10E-8 7
?* I 5mh ! . 1 7* ,
u -.5.m...--.
. 1, a
" - ." a .-
na 1.0E-9 .. " ' ' ' " - ~ u 1.0E-9 10E-610E-51.0E-410E-310E-21.0E-1 10E0 1.0E-610E-510E-410E-3 LOE-21.OE-1 10E0 w Release Fraction for Cs
'u Release Fraction For i n
B 1.0E-4 g , = , , , , , ,. g t0E-4 -, , , -, -, ,
, h h
i [ : Percentile : [ Percenfire , l S 1.0E-5 r , _ 95th_ _ _ , s toe-5 r 95th 8 Weon - 8 " Wfo6'~ e 50th e f u 10E-6 , e t0E-6 - e 7 l a
- a
- : y M e v -
- 10E-7 ,
,, L.
- 10E-7 :
tt ,. ,_'
~
-- ..__,, : e -
e _ c c '.. o 1.0E-8 O t0E-8 , n . n .
- e '-
e ' e e : o u ..
~
y 1.OC-9 .a , a ..a - . . .a .- y tng-g -.a ...a ...a ...a .- 10E-610E-5 LM-410E-3 t0E-210E-1 10E0 10E-61.0E-51.0E-410E-310E-210E-1 t0E0 Release Fraction For Sr Release Froction For to - Figure 3.3-2. Exceedance Frequencies for Release Fractions for Surry Internal Initiators. FDS Group 2: LOCAs
i
*7.7*
- 2, "7 M. 8+ 0, +,8 : 8 0. ?
- C.YP *?
?* ?*.
- E. U. U./P U. U. **C U. ? ? . U. 1 3
5 E U. k N U. 4.~5. $. U. E. 5. U. T. ...5 0. U. U. 5 U. 5.S.5U. *. E. N N. U. U. ! i
.-o.. . ... ~~-n---.-~ ;
o-o-o-o-oy o-o-non-oo ennne ....n u
?????????. ??i"?????? ?*??t1????
WWWxWxWWWe ExWWWWWeWW WMUNNWUxWW iaididdisi assisJdisi dissiaJais 3
$U8dyd80$3 $%8:$8$0$8 0$yByS8$$2 8.4%%.t%%%4%
omo onouan 8.4%%t4M%8% o.ono onos 48 en-48.448.4%
~n-ann E
I9f9fff9E9 IEEf!YkffY E?E9EfffYY l y WMMWUNWWWx dia d d A d dire WWU$UWENUW disidiaddi NeWWWNWWxe diadaAddid h w n "id.T* 2 8+ ". 8+ 3. *i :. E2.*2.**828*+? T ??+ 2 3. "? ??
" " "? .8 B. U. ,
. WBUNUeWWON WWWWWeNUWN WeMWxeWENW 3
disidisasi disaiadidi AnAidaAsaa E
. EfffffffEY susmenexes ffhEIEUffY exWxensmem EEEEEEEEfY eNxxxemsww -
a s as,a d a a a i saadaAdada diddAAddus
@ kkkkkhkkkh hhhhhh!!hk EEYYEEVYYY m o sW:W:ssuwN nWsusmemes exewVnemen u iAdJaidd&A &JadaAasia addiadidAA e i
- $$$$$?$$$? hhh$$$$$$$ Y?$$$$$$$$
9 excruwevex v m oy wasmsNeeux extweWeneW idiAsiadai iAaAadaAdd iAAAnidiaA uh mMn3 on n non n o n o vso o n n o -n----~~~~ O ?? ? ??? ? iiiiiii? ? ????ii???? A. m O I E. 9 8.on.
... U. 5 8.. E .E. U. E.E5E En-omo-o.
E E 5 5 U. ! 45598.95555
.~~.-.n.-
O ghN . 0 . o . o . p/ T
. o. .o. . N.n.N...N.
M i Tiii ?? hiiii i ?????????? 4 3"e- sesmemusee j e 1 8 m. U$ eNeNwNns N ey4g aa siaisisisi sessensass sisisAsisA gnadanaAAwa o , o
- Ta 5- -: emenemenen meenemenex gwwutuussww y& &* JJJJJJJJJJ JJJJJAJJJA daaaaaaAdd g
.a e .
e o oo ocoa anoonoo n @n men ene. IDa i ?? ???? ??????? ? ??? ???? t
- aa sunnussena mus=smswsm Du? wNsumNumm -
H 5 -& d&iaadadia $adsdissaad ,adidaAidaA l l > .. - e
- 8
- o o o o o o . - . -
a a - - - - 3 i 2 t~ u T ? ? ? ? b U W W W vi ? ? ? ? W U N N W j? u i i i W 8 W W i W 5 1
,d a a a i a a A i A uA e A A A .
m a g
. . . . . . . . . . . . . . . o MIaO "N N N N N "N N N N'N "N N N N N E i
f" *a a a a a *a W a a a ja a a a a a a a .
. ~ u u .
g N A A y a u a y l
$8 I $ I o *
[ E" E" DID E" f 4 <
)- N M . O hN . h n . M - - . .
l a w w - - w* ~
- : a .
3.35 l
i f 1 i S S O O l r- r i r. i ]# P # s. r r r j T s. i T : ' h! r b *i z b
*$ lgh g: $
" . ,I 1 % r 1 %
C i $d 2 - $ u l s c , s u o
,W I 1_so ,s so, O u
i . . -: gc p ge ,- s2 I e ! s3 e 3 3
/
3
- 7E
/ : 7 ga / .
g w@e U.mS , e
- sy 4 q
'l 3
s 4 tuM < l/ hi /
~
,a N
! EE N ..
ed K l e ( l l e e MM
. h l' : b a n.
- t. t. . /2i t. . . t.
$ t i t. 'u i t.
m @ b e0 m
- W @ b W Ch o W b b b b b b b b b b4 O O
,a O O O
,a b
ed O ed O ed m M y O ed ed ed
}I
,a ed ,J ed (JoeA-Jogooo; Jed)'beJJ eeuoposox3 (JDeA-Jo}DDeJ Jed)'beJJ eDVDpeeDX3 ,
*O fr ao S
O S O
$3a i r' >
rr ii r i r rr ' r" i r' ] r-s. ii ~ i r-7 s '." i
'T w
8-l w w . bf' ' O E b O I* r $ on, .- ~ e g$oi n ... ? ~
*5 '
f ,_ a. : ,,. g3 js
,c
- l 1 s u m
r f
/
i: s.d "h l
~
l' 7
$e8 r
[ l "'$ -
"10 r i
,t ' , -3 *" ~j~, 1 -
*g n m l : h6 y
l : hd. , i W
, s$
W k.e I [% > l l N b . 0 ed E p
/ e'
] @s E r e l l - l$
l l : h ,
! y W, ,'
I ,iiLin i t.m i i.1 ii 6.1. iii tm ii <8 u, in i 6. , ,,i j tu. iii & ig
- m @ b ED ci
- O @ 5 to cm b b b b b b b b b' o O O s
b s O
,a s o
.a O
.a 2
s s s s s s (JoeA-JopDeJ Jed)'beJJ e3UOpee0X3 (JoeA-JopoeJ Jed)*beJJ eDU0 pee 0X3 l 3.36
3333331 J 2 2 2 2 2 1. 2 2 0C000? 6C 00000000 _ e EEEEEEEE EEEEEEEE _ B 7 3. !. 6 7 8 3 0 48644161 _ 20302534 14131312 . 434 34 4 4 4 3331 1 333 o 00000000 07t00000
' EEEEEEEE LEEEEEEE .
54005802 9 0 6. t.
; 9663 519158'4 6 374 534 4 3 _
4 34 3 4 4 4 4 434 3 4 3 4 3 _ e 00C00000
- - - - - - - - 00000000 .
L EEEEEEEE EEEEEEEE . 2 3 0 0 2 5. C. 2 28922592 11211826 869584 93 4 44 4 4 4 4 S - 3334 34 34 0000000O 00000000 . u - - - - - - - - - - - - - - - - s R EEEEEEEE EEE1EEEE n 31253028 32703677 o - i 4 371 4 277 21352832 t .
~
- c. 32333333 22222222 r 00000f 00 t 00000000 F r - - - - - - - - - - - - - - - -
S EEEEEEEE EEEEEEEE
. 4101 805 _
- t. 38563759
. 21392635 1514 1312 1 .
32223222 21212222 _ a e 00000000
- - - - - - - - 00000000 .
T EEEEEEEE EEEEEEEE 43254925 45034208 9311911 1 4 1714977 - 23232323 131 313 13 V 00000000
- - - - - - - - 0000C060 s EEEEEEEE EEEEEEEE t C 16964596 56566596 n 7755544 3 37352423 e
v yE 22222222 1212I212 r 00000000
- - - - - - - - 0000C000 r: I EEEEEEEE EEEEEE5E u4 07845299 435071 98 S 'p 7454 54 4 3 33332322 ru oo 121 31213 12131213 4f r 00000000 00000000
- G G EEEEEEEE EEEEEEEE 3 s 3 91919191 91919191 mS 91989198 91989198 3 rD eF e eT . c 3434 3 4 3 4 34 34 343 4
. 00000000 e l
b es
.i)
. t. (s + + + + + + + + 00000000
+ + + + + + + + i EEEEEEEE EEEfEEEE d acr 1 ur 82828282 82828282 e Tro mD' 12121212 12121212 m ut r oa L Si .
. 34 34 34 34 34 34 34 34 t .t r) 00000000 00000000
+ + + + + + + +
ni an .as
- 1. S t(
EEEEEEEE 70707070
+ + + + + + + +
EZEEEEEE 70707070 c w eI a 3131 3131 31313131 n M o . l a l n . s y 65656565 65656565 b r .g) 00000000
+ + + + + + + + 000 00000
+ + + + + + + +
a e .rw et EEEEEEEE EEEEfEEE l i t 1n 97979797 777,77777 a n ' aE 11111112 31313131 v a a, I i s J n o s i 0 0 0 0' 0 0 t ) 0 0 0 0 0 0 0 n a + 0 0 0 i
+ + + + + + + b w
v (s E E E E E E E E l E
- 0 0
0 0 0 0 0. ' 00 0 O. l l s 0 0 0 0 0 a s i n 3 3 3 3 3 ' 3 - 3
'ro ,
3 f n.) i B 0
+
0 0
+
0
+ 0
+ +~ 0 . +0 0
+
s nms ri( e E 3 E 3 E 3 E 3 E E E E m , aT l 3 3 3 3 r e w b 1 1 ' 1 1 ' 1 1 1 1 t a . b e . o AAA4AAAA AAAAAAAA r EDEDDCDC EDEDDCDC u r DDDDDDDD A ABBAABB DDDDDDDD AABBAABB .o n P AAAAAAAA AAAAAAAA i CCCCCCCC CCCCCCCC f B t BBBBBBBB BBBBBE BB o . s DDDDDDDD DDDDDDDD AAAAAAA A AA AAAAAA g o HRRRREME BBBBBBBB BEBRRRRE AAAAAAAA n - M i t s t i l r hg e A d i 1 2 3 4 5 6 7 8 . r O E * ,
".W
; l ; , + ;m! :,3 [l ,t i 0 0 -
E E - 0 O
,
- 1 1
' . ; I.
.1 1 e '
.L 1
- l E-E i
i
... 0 t
nhdt ih 0 1 r o
..' 1 et 0
,- a 2 c5 r9W 5 2 f e , . ' . E- a s
. E- s P " ' ,
n
- 0. C 0. l o -
- 1 r 1 r a.3 o a 3 Fo i tV F c E n .- E- n O o at ~
- 0. io Li t rn ~
1 t a 4 ca Fe a.4 ca v E- r eE E- Fr s 0 '
- 0. Fe a:
- t e a, 1
s e4 a 5 as ' 5 a l
= -
l i e
" El -e '
~
El-e ep t i t h . O e - 0 e R u nhd 1 R rro et t LR ' - c5e0 r9M 5 a6 a.6 oG e _ E- f P~ . E-0 7 . _ 7- 7 ,: 0 sD S
~ c[ r- 7' - I 1
1 eF g 4 5 6 7 8 g i 4 5 6 7 8 - c . E- E- E E- E t- E- E- E- E-0 E-0 g ns 0 O 0 0 O L o t 0 0 1 0 1 t 1 o. 1 er uo 1 L 1 1 1 qt
$,> u28.u 'eavdy egoi*oO t T ,>[ooEu gavd y .ucov eu0 ea ri .
Ft . i . en cI 0 0 E n _ E 0 al , 0 d a - y - , - ,:- - , . . 1
,: : ,: - , : 1 - ,;- __ 1 1
en er
*I ,..'. 1 e -
i: . l . - ce E-i J i f ....' . E xt h6th >.; 0
, 0 1 r5 t 0 1 E n I _
- a.2 c9W 5
- a.2
.y
. E , F ~ ,~. E r O S 4 r r
- 0. i 1 r a.3 Fo
, ,' , L r 3 Fo 3. S u
_ - - 3 E- n
.- . E n e 0 o -
- 0 1i i o r 1i t u a.4 a.4 c a -
-o EF r i g
. . E- R - 0 F D 1 es
._ L es _
e ' a.5 a - a.5 a
-e E- e
_ l - _ i t i i . El . l e n%6h 0 e 0. 1 Fs c560 a.1R e _ t r9W5 6 - a.6 e . , E-P~ E- . . O
. 0 ,E : ._ 7 - : r- - - ,; : ~
, - ,i - r: , ' : [: ~
1 7 _ : L g 4 5 6 8 g 4 5 6 7 8 7- _ 0- E- E- E-0 E-0 t-o E-0 E-0 E-0 E 0 E-0 E-0 0 0 0 1 1 1 1 1 1 1 1 1 t 1 t tg>suo g.u seav&j egoJ*uy T,>.[Rg c gavdy .ucoveeuO
".wa
a E9E9ffEhE9 EnWxWmWnWn Wa0WWWUmWx f9f9fff9f9 99f9E?9U WWmWmWxnux sasidadaia aAaaadadaa aaaadidiaa
" d. 8 0 8 0 8. :. t .*M
? *:I5.*#.T.80
- f. T 0
- 2 0 8. ?" 8 8 8 8. ?' ?
WWWNWNWWWW WWWWWWWMWW WNWWWWWWWx dadadddada saasiadida aiaiadaada
? 8 080.?.*0 8 0. *d 82.*M. ? . 8. d.?" . 8. 2 09..I38..8."?a.
WMWWNMWWWN WNWmWNWmWW WNWWWWWWWN disadadadi saiaiadidi daAcadaAid
%$3 Ed 380yS88088 i # {#jayO)44% @%yM$M8%{d
%44%%44 44444%i4%4 7
444:44.o. oMo-ono o.omo~ocoN MMNoeoN.MM u
$$$$$5$$$Y $$$$$9$$$Y 99$$9$?$99 ;
. WWWeWWWWWe agaidadada aadaidaada WWWnWmWWWu nWWWx0EnWu 5 asiaadaiaa 2
a
$2y2)2$8$2 29y2800000
. 48.48.%.48.4 $$$2$28882.48.4H4%%i$
ohonomodoM oNoNoNoNoh 44 444444% h@N.NN NOWN l i i geooomongo onococomum Nww.eew+NN l C IiiiiiiE+Y iiiiiiiii EYIiiiiiii ! e a WWWNWWWMWW WWWWWNWWWW NxNWWWWWWx aAdssadadi sasiaidida aduaAaadaa i E b$ - 9999999999 9999999999 9999999999 s my ExWWWWWmWm WnWWWWWWWW WWenmWWmWe assasasasa aasaaasasa aaaaaaaaaa N 4 omoMomoM M eMoMoMoM nhD ???????? ? Y??????? ? M N HM MooMNNN
??????????
e a EnWWEnWWWW WWWWWeWWWW weWuMWWxnx L 9go n m essasassss sesssassss sasaassass
.hg e0 gs: F.o.*q.o.g.r e e ,. e v e . e o.g.P.o.o.eever re.e N.No.eN.N.
verveeeeer : qg. - WWWWWWWWWW WWWWWWWWWW EnWWmWWmWn j suego a f sessssssss ssesssssss.$aaaaaaaaaa , m u . .......... .......... s..... .... l
- 7E@@Tiiii@ iTTEEiiiTI iiiii iiiE au -- zeneneneen nuezenzene metWWWWWWWW e et Es
.a~ ;;;;;;;;;; sia;;;;;;; a;aaaaaaaa 4
e U e C N e 5k-oooooooooo cooooooooo Weeen.neneg j 8 W 2WE V????????? ?????????? WWWWWWWWWW WWWWWWWWWW >WWWWNWWWWW mV????it?t+ a T
* && didadadada sidaddadia e aaaadadaaa 5 C > a H e
- 8 2 m
- o o o o o o o o o - s *
- w a Nh k N k k h k k k N k k k $
WWWWW WWWWW
$"W WWWW t~
a ha a a a u & a a s i a a a a & .
+ n w M. . . . . . . . . .
J. . . . . 3 E$m " h$ $ $ $ $ mM M M N N
? $ $ S $ p S S $ $ $
N N N' N N N N M N N E f* .a a a
- a a a '
a *a a a 3 e
, e m , . -
A 0 M M ao ao u 3 M 3
" " # " * *
- O $ I $ N N 3.39
.Ij -! t :s' .I'! -l ;1i r.
a w 0 0 - E E 0 0 m 1 , 1 -
- o. 1 4 _ -
e -- 1 _ le , L E- l -
'., .. - L E i
i _ n h, n t t , a O nh 0 .
, et c5 r9Me -
" 2 c5s et r9u a2 1
r e - - e - . - E o o - P . - - E s P - . 0t f _ 0. 1 C r , 1 r s _ "3 o F a 3 oF n
- o s
_ E- n , E n i 0 o t tn _ 0. io 1
, 1i l c e
" 4 c l
a4 c a i
- -o EF r ,-
EF
-o r r F s 0 -
0 n - 1 es - 1 es e r a _ " 5 a .- a.5- ea sT El-e -
- El a
e: e - 0 e l _ O. 1 R . - 1 R e 5 - _- " 6 _
- - a6 Rp
_ - E 0
. E-0 ro uo r' , ,,_ ,
1
':l E _- "
1 f G r 4 5 6 7 8 9 4 5 6 '
, 8 9
- s S E- E- E- E- E- E- E E E- E- E- c- e D C 0 0 0 0 0 O 0 O O o i P
O. 1 L t 1 1 t 1 L 1 L L t c
. n
?o*w8uo 8*u aeoMT* u eocove oO Cg>:u28euieayT*Ceucoveeoy -
e u s q r e o r t F a i 0 0 t E E ei 0 0 c n 1 ,- _ - , 1 n ' y:~ , _ ,. - - ,: - - - ,, _ _ 7- - , - _ ' _
- aI e _ _ - 1
- le- -
1
- d l
'r i _ _
- E i
t E e a
*. ..... e t
n h_. n th 0 nh 0 n c c5s h_ . t , et 1 r cet 5. oe0 5 t . 2 r9u a.2- x E e T e r 0r_W 5___ -
- e E r t n
1 P _ _ _ E P ,
- 0. S 1
- 0. i .
x .I r
~
_ 1 1 r
. - 3 o
, , . a. 3 o 5 y F ,- . F -
r
- E n ,
. E- n 3 r
- 0 o .
. 0 1i t o 3 u
- " 1i 4 ca t , ~
a4 c S
~ v E- fr
. -a EF r e
r
' i 0
O u g L es 1 es 5 a
-e a.
5 a e i F
- El _ .
E-0 le
' .. 0 e 1 R ,
_ 1R
. " 6 a6
_ - - E- _ . E
. : - ,; : - - ,c:- : 0 g- - r . - r:: - _
,F : - - ' . _
0 c : - - .,i. 1 1 5 6 8 9 4 5 6 7 8 g 4 7- E- E- E- E E- g-E- E- E- E E- E-0 0 0 0 0 0 0 0 0 O 0 o 1 1 1 1 1 1 1 1 1 L 1 t tg>. To*w:uo geu scr*.M7* u. eucoveeo0 : LR8euLeoyW* Leuc 0reeoy w z. i l ilillltlllll;I>lIIlIliIl:Ij,i-,1.
r l
< ?"*.80.*.*0,I0
? ? 82"'f08.:.0.*
+ ?? 2.f*2 82.f*2. f '28 a sunNewswun use==== es? eseem= eses.
dasasssasd saassasaas usadaissai i 0 g"g 8:gdy: 80Sygdy$38 8888888888 , i 4%8.4.%4848.i oNo .-oMoe @444444.M%% ohNooN Mho 4%%%%%%%%% N o t* ... mono
. 8. M. 8. . e 2. 8. M. 8. d. e .d?' e 8. ". *v8. 2. e a.e2.8."e2.I88 v l
- 8. M 4 9 4 8..o
...-o 8 8. 4 M 4o.hoo..
- 8. N 8 8 8.4 4 9 -
4 N498MMM8M8 h a-o-onoh o l i g g: g:389: 8 2808sgassy , i y g:g"g4448.% 8.$848 anoao-once on 4%%.%H%gsH%%% co-on.o %%%%%8.4%%8
.o o-o-o.o l E o-ooo- ooo conoo-oono non no e u
- ?????? ??? ?????????? ??? nonI?? ?Y? <
. seemez=stv samenesNew xeswwwxmms !
i 2e diadassaae daasiadada adaaaassaa comooooom om-oooom-o
. !????????? ?????????? -o-o---M--
??????????
WMWMENUMUN WNNW MUMNW NWNWNMWNMN I sasasasaid saaasdasus asasaaaaai , 8988ag898 98s p 8:8:asema ' o g%44%.%.s%M4 4 o-o. .-.. . 84%g8894%44448
.o-o--o %%4%.%%%4%4
-.- -- -n g
- . S + ? $ $ $
$' @ WNUNOWWWWW WWWWWWOMMM WMMMTEMEUM
.mW u o sJassasasa saaassaaaa aaaaaaaaaa O N O O *******d"
@WU E
?" ? ?"+?
8" UNWWENWWWW WWWWWNNWWW
?" ? ?" ??????vii d"""?????""
???
d***
?? . A. , m WWMWWWWNWW n6 iddadadada saadassaaa asadJaaaaa m i I i i O $8O uuy n=====we== semesse mv esssssures j g-a A.- saassasasa saaaaaaaaa yaaaaaaaaaa ,
mu . .......... .........g s - gs se- ?eerreever neunzemenn ervererer eesennexse. ee:t...P.fe.g.f e .e as: ya &" JJJJJJJJJJ JJaadaJana messs=us=rW nadaaaaada g a . onogogoo o o egoogneo Weoeog gogo. d 5
}a EE $$Esissisi Esisiicili piiilissisi !
a sesssssess esasssssas .asasasasas . i W
- O 2
- o- o o o - o. - - - - - - - a a
k k .N h k I' W E' W 0 5 0~ W U- E W t 8 W W W g a a s a a a a a a a a a a a a a .
*m. a w
. . . . . . . . . C. . . . . &
i T T T T
= g$ 3 Mi i i 'T i *T T ? ? .T l mN MN N N N W N N W "NW l
*a a a: a aN-
* --M M
,a a a a a. A a a a A. 3
- p .
u lillllllllillllllllll!Illlllllli !. @ W
~ N m . . e h e o o h o N q
l 3.41
i l 1-n n o 10E-4 -- -
, o' 'JsE-4 ,
E E r% c n 4;ie -- Percentile t 95th
- 95th : I
-'Mafiii , 310E-5 Weon' ~ _
l Io 1.0E-5 . 7 [5Dlh i g :. . 50th 8
. : * . . _5_t h_ _ _ _
- u _ -~~~-- . . 5_th_____ _
_ u _ L t 10E-6 ; e LOE-6 y '.. , e r g g ~--...... g ; g l
- 10E-7 ,'._ -
.,
- 10E-7 r e '.
g :-. g ; l o 1.0E-8 r '.
;, o 1.0E-8 y ,- t o .... . .,
- m. .
.e . : .. __. .
y g .
"'- w 10E-9 " = m '=' ^ "
1 w 10E-9 A" "" - ' - r - l 1.0E-61.0E-510E-410E-3 LOE-2 t0E-1 1.0E0 1.0E-6 LOE-510E-4 t0E-3 LOE-210E-1 t.,EO r w Release Froctbn For i Release Fraction For Cs 1 , i s- t u Cg 10E-4 , , , _, ,m Cg 10E-4 . , . , , , , j [ :. Percentile : [ : Percentile - _o t0E-5 -
. 95th _. ,. _o 10E-5 ,.
. 95th .. , ,
e. u : 50th i e u 7h I s 10E-6 , u 10E-6 , e 7 o 7 Q. Q. l v - v ' 6 --------- ........, d .._.. ...,,, y 1.0E--7 7
, y10E-7 7 - ..,
, [
e -
. . l v -
u .
- s c -
c - t o 1.0E-S .'. o 10E-8 -. , l 3 3 , 7 -
- u .
u - i O LOE-9 5 10E-9 " . ' - 'O " ~ I 10E-61.0E-510E-410E-310E-21.0E-1 10E0 1.0E-6 LOE-510E-4 LOE-3100-210E-1 10E0 Release Fraction For Sr Release Froction For to i t Figure 3.3-6. Exceedance Frequencies for Release Fractions for i Surry Internal Initiators. PDS Group 6: ATWS
, oc - . r- - _-- . _-- * = .~ ____________________.___a . ~ . . - __
2 8 5 3 3 8 8. 3. 5 8 2. "?U. E. U. s. V" 'e
?"e hU
"?G. S. $. ?" Y . "? f .#8
- xexwmennum verexszews vemmenmemn adaaaadaaa Aaaaaaasas asadaadaad i
- sasacas essssagssy :::sesssay 0
$ k 4 %ik %.%
~~~-n- - $.$- .$.$. ~. .$ $.$ $.5.k .----. k 8 8.4.%.$.$$%%%8 l
.......... me. ...... ..........
. ?????????? ????? ee?e ?e????????
M.M8n Mn M. .-eM M M w.nna M. W 4 8.~o4 4 4o M.n.N-n- M 8. 4 8 MMW.NS.MM.MMW
~nn ;
' ?f9f9f99?f 9??????i?i 9f?ff?V 98 NTN8NN8MN" I
--- 58".888 N#8888MMUN B NW".
----.--... .. ... .~~-.---.-
l N N. N.N.M. N. NMM. N.NONO NMM.. unum. I e n ?????????? *??????e?e ??????*??? xmmemnuman zumxxxvsne nexnensmer 5 .tdadadiaad saidadasas dadadaiaid enne-mne-n nennNu-o-o ennnnu-u-m E . ?????????? ???????e?e ?????????? wwwwemexen unumenness unwennxmum
.adiasadas ididdausas idadiandai ,
--------~~ ----~.-.~. ---~~.-.--
?????????? ???????e?e ?????????? '
3 0 enenWnNNWn xNndeWxsNs NNsexTMWmM G Adaiaauida adaaasasas adaAaadsaa m
-M-N-N-M-N -M-NN.-e-e -N-M-.-.-N B' - - ?????????? ?????????? e?????????
W~ m wummemwx x euxnnnxnes exenwesexx adadaaadad aidadaaaaa addaaaaaa;
* -n-n-nen-n un-n- ~o-o -n-n-e-men 80ou a
???*?????? ??*??????e ??????????
wsensex=vx vezesexuxu wwxxwnxxww "0m n
*$ aAssidadas adaAsaisis a s i d a s d a n's .A.- .0 ....g. . . ...... ... - p~~: ever.eeeve ......eges. verveer.v eeeeeeeeve :
1*8 euu aA MN858N8MMM
-~~-- ---- ---~ 8N8MMN n.-.-848 85MNWNWMMN
-- -- 1 pa w OM 3 mu* . .. . .. . .g g . .. . ...
{ f. a eieeeee?er eeeeeve+e+ iereeeeree 24 ununununun unennessus ununununun sessssssss sessssaaaa ssssssssss o 1 g- 4+ 7 h 3 Y
= it- Effifiiiii III!Iiiiii IIIIIfi!II 3 ;
8 2 "a menussness musssessus esssssssse : !
~ as asasasasas asasasasas asasasasas n
\
I S - - - - - - - - - - - - - -- 3 n e e e e e e e e e e sss e ee~ee sssse 2 1 t~ tussss s a sa m aaaa a a a a a a a a; a g ; y . . . . . . . . . . . . . . . u , om emee a m exx =m.m mm
~
llae I s- je w s s s s s e ; ; x x s s x 4 o .
.e. illililll11 O
... li.ll.il.l'
.u mQ.0 a .
e m u u
- N m . n . u . . O - N y .
q 3.43 1
i , i t :, , , ~ . -! ttl I. - O 0 _ E E C 0 L , ,j.2 - 1 1 1
- l i
e " 1 E E nhchh t
- 0 t 0
' 1 , et c5e0 " 1 r v
- 2 -
9 5 2 o
- e ~M - f
_. E s . F ~ E o N .1;*
- 1.'
- 0. C 1
3 o r 3
....;:..' 10l 3 o r
s n o
\ -
f .' - f i
- - E n .'
- E n t .
- 0. 'e 0 1i o cs
=-
1 4 co t , -
.N
- 4 c t aR rT ~
- E- Fr EF
- ar FG S
e i 0 - 0 _ ' 1 es -
- 1 es s:
5 a - 5 a a7 e ' e l _. El - e
~
El-e l p d-h h t
- 0. Re ,
~ O e LR eu ,
_ i&tM~ L59 0 5 h__ t_
'- 6 1
, ~
~ ' - 6 R o rG r ,
. 5__
,- _. , - - ~
I m" _ E-0 "
~ .
- E
- o fS
- r::. [ - _ : ,! - - r' - , - _ 0, D t 8 ,
4 5 6 7 8 9 4 5 6 7 8 9 sP - e E- E- E- E- E- E- E- E- E- E- E- E- i . 0 0 0 0 0 0 0 0 0 0 0 0 cs 1 1 1 1 1 1 t 1 1 1 1 1 nr
. . eo ng,x uo uo.._ 'e3, eE 0Ei*ud toELo 8*s. eaMoE .ucom.eud ut qa ei rt Fi n
0 0 eI E E c ~ 0 0 nl 1 aa
- pj _ ;- g{- -
1 1
,2 me _
1 : - , _ , . _ 1 d n er
- - .l i . .'*;..*. E- ee
- E ;t
_n .
*t. -
ct -+
- 0 n h_ oh t 0 xn
_ ' t 2
, etc5 e0 h_.
5 t. ' t 2 EI 1,..L-E
,P re9_W 5_. _ .- - - y
_ _ .- E r .r
- 0. i 0 S 7 r 1
3 oF r , 1 r 3 o f
- u E n E- n
- 3. S -
3
- 0. io 0 o
. ' 1 4 c t , .
' 1i 4 c t e r
_ ~
- - ar EF -
." - ar u g
O . EF 0 _ - i L e e '
" 5 a s , -
' 1 es 5 a F
l i _. E- e ,
,.**!...1 E- -
e 0 le t h _ - _ ' - l _ _ nh . 0 e t 1R . 1R etE h__ . _ . c5 0 t , r9 5 6 6 e ~ P ~ 5._.'-
.- - E- . -
- ~
E- _- - - r:t - . rE _
~I : _ 0 < _: .,j_- .
[ :- _ 7- 7 - 0 t ! 4 5 6 7 8 9 4 5 6 7 8 9 E- E- E- E- E- E- E- E E- E- E- E 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 ' 1 t 1 t 1 1 1 1 n'gx uo uo. 6e4oE' 85S*oD .
' Co* lo_8,s. secvaE . . .- cucot.oud
- w. Z -
;!!il1ltiIll\l(il!,-!!.!!!l !;i. .' Iili 1
l ,e A., _p . r l. ? '. , 9 . , . . , e e g 10E-5 , , , . b,10E-5,, , , h b - I t 6 o 10E-6 c . o LDE-6 r o g u - 0 .. . . . . . . . _ _ . . . _ _ . . _ _ _ _ _ _ _ _ _ . , O. .__. _________.......... ... u -
.e. .
st0E-7 ...'.
'. TOE-7 e .
a
- a M :
M : _ 7 7
&* 10E-8 d* 10E-8 f 1
e F e - o Percenfile g : Percentile 95th ;
~
01.0E-9 r -95th Et0E-9 ' g-- ' y r-- Q -- e _ 50th . e o 50th c u .
" - '" % t 51.0E-1' " " ' ' '% '
$ 1.0E-1r "
".0E-2 10E0 10E 10E-4 1.0E-2 1.0E0 1.0E-6 10E-4 1 Release fraction For i Release Frocfion for Cs m
c- e e ut gt0E-5 , . gLOE-5 , , , . . . , . - ,
, - h b -
I 8 - u o 10E-6 h u _ o LOE-6 y , - u - e O O e ..._______... e ~_..... ' ' ' ' ' ' ~ ~ ~ -
~
'"-~'-'. , It0E-7 e -
I.10E-7 _ G Y
.: Y O' -
7 -
' '.,.
- 10E '. _
w2 LOE-8 u 7 i; o
. e
; o ~ Percenfile u - Percentile 95th E 1.0E-9 95th ; ,
E10E-9 - r - ug6- _- . g r- s -- . 7e 50th u 50th
- O u
510E-1d '" K'" ' '
- O t0E w % " '" ' '- ~
1.0E-2 1.0E0 10E-6 10E-4 10E-Z LOCO 10E-6 1.0E-4 Release Froction For Sr Release Froction for to Figure 3.3-8. Exceedance Frequencies for Release Fractions for . Surry Internal Initiators. Alpha Mode
=_
Tg.OE-5 1 .
$, TOE-5 , - , , ,
, : Percentile: [ Percentile . ! [ 0E-6 o 1. g .. 95th _ g ot0E-6 95th
- g --.
[ - g.- 4 E.
~
50th E u 7th - t
- s.t0E-7 y , s. t.0E-7 y ,
1 e e a a r y :___ .. __.._______... __ . _, y w . . - w :__________--...__ .....--- ... .
- LDE-8 *1 '.,
it r
. st .0E-8 -
e
. e '.
o - - o . C -
- C t oLOE-9 r . o t0E-9 * '
v e m e 7 ._ e : : e ; o o . A~
^
01.0E-1r " " ' - - " S toe-Tr - "' ' "" '- t0E-6 10E-4 LOE-2 10E0 LDE-6 10E-4 1.0E-2 10E0 P Release Fraction For i Release Frociion For Cs i J:- l cn
$,t0E-5 .
u , , $,10E-5 -- x _ x i l [ : Percentile: _ Percentile : S LOE-6 y 95th_ _ ,, [10E-6 o r . 95th , _ , ' o .
,_ NN ~ o i WN e.
s 50th - e. s 50th _ I l i t0E-7 P ' s 10E-7 r i y
& _ y
{ : ____...__.' T ___. ._..... __' 7 - 210E-B t._ r t E10E-8 , e - '.~. L e '.. u - y - c -
- 1 c
o 1.0E-9 '. O10E-9 '. v e r 1 o e r e . a ~ y g . u t0E-1r ' ' " ",- w Lor tr " - ' - 10E-6 10E-4 1.0E-2 10E0 LDE-6 1.0E-4 1.0E-2 10E0 i Release Fraction For Sr Release Fraction For to ; i Figure 3.3-9. Exceedance Frequencies for Release Fractions for Surry Internal Initiators. Containment Failure at Vessel Breach with the RCS at High Pressure P l
.-v -
tt0E-5 o s fo.0E-5 t 1 Percentile 1 Percentile 1 : 95th : I 95th 3t0E-6 7
~ ~ lie 6n~ ~, )1.0E-6 y -~ideofr , l
- g 50th
- g l lh 5th -
u . 5_tn_____ u _
- s. 10E-7 - , s. 10E-7 r r
4g : 47 j-----------------------, y t0E-8 , y t0E-8 r o u - c c ' '.'. . l o t0E-9 ,. o t0E-9 r ' y F_.. ]. i ..... ,_ o _ ...........,,_ --.~ ' v .
~~~
~
' - A D t0E-1r " '" - - L-610E-tr - ' ' ' " -
10E0 t0E-6 10E-4 t0E-2 10E0 1.0E-6 LOE-4 s.0E-2 ; Release Fraction For i Release Fraction for Cs w i a v To .0E-5 1 Ct0E-5 o , , ;, , , , ,
. z, ,., , ,,
, .PercentileJ q E _ E L : __95th._ _ - Percentile -
.2LOE-6 ideon [ t0E-6 o - 951h,, _ ,
o hdeon
- g _ 7th '
u
~
w ___s__... - u lib ' u10E-7 r . s.10E-7 r Q. CL Y : y ' 7 - 7 l yt0E-8 7 .. __ __,,, , y10E-8 , o u - c c - Ot0E-9 7 .'.
, ,Ot0E-9 . m
- m. ,
w N tot-te
,*ssa ....a . .a . ~
's N
w tot-se
..- . .a .u e
.. a . . . -
1.0E-6 10E-4 10E-2 10E0 10E-6 10E-4 10E-2 1.0E0 . Release Fruction For Sr Release Froction For to l Figurs 3.3-10. Exceedance Frequencies for Release Fractions for Sutry Internal Initiators. Containment Failure at Vessel Breach with the RCS at Iow Pressure
-, -. - .- .- _ _ _ _ _ _ - _ _ - _ _ _ _ - _ - - - _ _ _ _ _ _ _ _ _ _ - - _ = - _ _
l i i l l
- a bt0E-5
* ....a....,......... , , b, LOE-5 ,
>. Percentile I .
I Q$th
$ toe-6 y I , $ t0E-6 g----- ., "Seon".
8, L......-..........., 8,
', 50th
, u -
. ' .......... - Sih i s.1.0E-7 e ,
,, t10E-7 a
e i , a Y . < M ~ N, T *
* , 7 ,
l E t0E-8 ,
< ,-
- t0E-8 7 .
- , 1 Percentile i Io
t u - 95th i.* ' -
~
! E10E-9 ~~~ideoh~
" 5 .Et0E-9 ";
- i i 50th j ; ] ; '
e o _ .. .s.ih.. . . i.u _ .' O t0E-F -d -" L - - - S t0E-F -- L -" " " ' - t0E-6 t0E-4 t0E-2 1.0E0 t0E-6 t0E-4 LOE-7 t0E0 g Release Fraction For i Release Fraction For Cs c~
$,t0E-5 .
,, ,, $,10E-5 c , . ,
Percentife: [ [10E-6 o o 10E-6 g e r--..-.,
. 95th...,
We~on g r [ .., u . soth . u . i.10E-7 e r '. , 10E-7 ' v" - v a ',
~
- 6 '.
i
- LOE-8 -
. ,
- 10E-5 ,- . _
t t - i e : - o -
~ .
o . - ' ELOE-9 " I E10E-9 " Percentile I ie I ' 3 95th ! _~ o : _ - E-- :
~
. , u O t0E-F - h u' - - '
Y tDE-F " - " " ' - "' - '- '. - 10E-6 LDE-4 1.OE-2 10E0 10E-6 10E-4 10E-2 t0E0 Release Froction For Sr Re'cose
- Froction For La Figure 3.3 11. Exceedance Frequencies for Release Fractions for Sorry Internal Initiators. Late Contairurent Failure
;->i ,l .l 1' !' > el' lii!ll' I h Ii 0 0 E E 0 0 1 1
- <' :e_'
,~ ' - '
,.. i t
nh sh t et 0 c5E 2 r9 5 2
- e ~ I -
r _ E s P ~ *!I;; E o o 0 l f
- 0. C -*
1 r
- " fo 1 r
.*.. ' F o n s _
n o
- ' n -
o i
- o -
t
' 4- co i
'4 i
- t c
- - a
' EF
- m ,
EF r 0 F r 0 - 1 es 1 e e a s
- e , a _
- a s ay l
i _ . le - le er
- t n h_ n h
- e -
e et o t0 R
- R lD c5
, e -
- r9W 5 " 6 a6 RV e - - - , -
- P -
E - E r . 0 os n [- r r : 0 - [. ,:' ^ [ I 1
~
r 1 fr r S 6 7 8 9 o - 5 6 7 8 9 1 t
- - - - st E- E- E- E- E- E- E E E E- E- E ea 0 C 0 0 0 0 0 0 0 0 0 0 ii t t t t 1 1 t 1 1 1 t ct
$,x[o g.u ueO &*d ,oE3eyu 1 .
r ng,x,Io geu'e0YtEu E E ieuD ni en _ uI q el ra F n 0 0 E r E 0 ee . 0 ct -c
, - (: - - ,~: ,: - !, "1 3: ,j - ,, _: 1~ 1 :-
1 nn aI
.1 .i
, *' d
.' U
, l; ,'
ey
.*'.. er cr
' 2 2
xu ES
- E- - . E r
- 0. S Di -
- L r - 1 r .
Fo s
~ o ,- 2
- F _
1 n - n
- i o -
l o 3
- ' 4- ca t -
_ a4 c
-o i
3 e - ET r EF r 0 - O r
- 1 es - L es u g
e _ ' a -
- e - " a i e l e F Eh l
i _ - i t n h_ot nh le - f nh le et R ot t R c5E 05
- c5W r9 e -
0 5 ' 6 r9 e ~ a6
- P -
E- - P ~ E-0 -
,^ : r - - 0 h g- r' : . , : ,[ I r 1
- - ~ F : c 1
r 5 6 7 8 9 t 5 6 8 9 1 7- - - - - - - E E- E E- E- E E E E E E- E 0 0 0 0 0 0 0 0 0 0 0 0 t t t t 1 t t 1 1 1 1 t igw,3g 6 senyt*u
% . r -
OE;eoO S,x8 og,mu 3&*h.uE3*yw t -
' ,e
o o 5 $ d r r- i r' i r-- 4
]d r$I r-- i r' 'a r-- is r-- r i
~
_,,,,..........***=* .c' g . r Vh,. @
,* - k8 r
] ] ~
.e -
y ' a 80 . _[ b o r: r "b 1 "w b 1
- w -
o l -
, u ,
g . g o c: : 7g
/.. : ,
7 *8 g w e l - 8' s / .
, g r!. E. i.
r . ..- y z u.
., , . . :s.
- ,: m > .
,ll b l e= ' l l e W .
,ii m: -
a : s a , L,a a Li L1. Lu . a i u Li L i L. Q
.<, . - . . v .. ..> : t aaasaa
- a
(;o.4_;oioo.; 2.e).e.s;; .ooop..ox3 a s - assaaa
-~s
(;o.4_;oioo.;2.a).e.;3.ovon..ox3
~t 8-
- a r-u W ,
o o u 8 Y ya a
,,,.,,,s s L , - - , c.. ,, r.,,, r., ,- , r- - - , r, , , ,
r.. ,, r. - , j g ...............- i i g@ 3 >
..*- 5 m:s r * '
- r. - -
] u f- 3 m l
[ -
$ I'**',,......... :
h t'5 m' i r e - a <. - s u w r , ao 4 1 - . ] e l :
. : c r e m.
l .9 ,
.9 m
! l 7 5
~
! l 73 8
- ge -: gc a l
r! 1 "% r/ 1 " e* C l 4.' : l} l 4l ' j e I. 1 e l c!! O _l .,f,, c L. i :
+
. . . . L., . . . m ,, m,,,, s , , ,
3 m. . . . . , L., , L, , , , m , , E, n e n . m v n e - e m
- h. h h
b h
- h
- $. h h
h h (Jo. A_Joloo.; J.d) b.;j .ouop..ox3 (Jo. A_;ogoo.; J.d) b.;j .ouop..ox3 . 3.50
g LDE-5 , , ,
-@10E-5 c - , A -
l
.C 3_LDE-6 , . _ _ . . . . _ - _ - - . _ _ _ _ . . . - _ - . . _ _ _ , , , ,,.
o 1.0E-6 g g c_.____....._________...._,,,,.,- u *
.., i._10E-7 s 1.0E-7 r :. . _ _ _ . - - - - _ _ _ _ _ _ _ _ _ . ,
....____.__.._____...., ; 4o- ,
4e- -
;
- 1.0E-8 ,.
- 10E-8 ;
~ Percentile !, ~
Percentile , o 95th -
'(;,
- : c -- --~~-
- o c - - 95th
' ! o 1.0E-9 Weon ;
o t0E-9 Woon 7 50th ; f Mth ! 5 3 . : i.o : ,
- o _. sin----- .
.1
..3m____ -
- 1 ,
a tog-te .-..~a .- ~J. .. Utog w ~..._...a . . . - ..- ... t .a 10E-4 t0E-2 10E0 t0E-2 10E0 10E-6 10E-6 10E-4 Release Fract~en For Cs w Release Fraction for i u, >* ^U,t0E-5 , , ., , ,, , C10E-5 o ,., , , _ _ ,
- Percentile:-
7
~
810E-6 95th ,
*o t0E-6 p.._____________,,,,,,,, --- .-- - ____ - g ---
- 50th -
8 -
~. . .! .
, % 10E-7 7 I.t0E-7 ,.
- v a .
? i__ __ . .
- o-
! [,1.0E-8 - .
m ht0E-8 h ! b. c m.
; u 3 : __9.5t_h _; c .
NO"-- ! ; , 510E-9 1
~ -810E-9 7 J*
P { MM ! ;
- 3
- ; __5th ___* a ." " " " -
E "LDE *
~
-'aL"" " ...a'.. O t0E-fr a . . .
10E-4 10E-2 10E0 10E-2 LDEO t0E-6 10E-6 10E-4 Release Frocfen For la Rhe Fraction For Sr Figure 3.3-14. Exceedance Frequencies for Release Fractions for Surry Internal Initiators. "G" SGTRs (Secondary SRVs Reclosing
n . n g10E-5 - g10E-5 -, , , , p :_. _______________ .....____,,,, - p __...__________..____..__,___, u . u o t.0E-6 . o t0 E-6 -: , r_--------- .-__________, 5,: M, ;______.._, u - u10E-7
, r
. u,10E-7 r a : a :
Y . v w
- w .
'* 1.0E-8 I U
*1 E .0E-8
- i e - Percentile !. - . " Percentile . I o
c 95th o 95th i Woon
. c --------- - -
o10E-9 " o10E-9 " M"m ,,,. i* 50th
- a i* 50th ;i '
.'.. ' ~ , ' . -
o _ _ _5_th_ _ _ _ _ ',,
'_ o _ _ _5_th_ _ _ _ _ -
Q t0E-Y " -" ' L - 01.0E-1n - " ' " - " u, 10E-6 10E-4 1.0E-2 1.0E0 LOE-6 10E-4. 10E-2 LOEC 6 u Release Fraction For i Release Fraction for Cs , e A a '1.0E-5 o . . , .., ., , ., o'- 10E-5 E, :.-----------__..._____, , E, <- _____.... ,,,, Percentile u _ u o 1.0E-6 _'., o 10E-6 Meon 8
~
g
's 7 015-_ ;
L .
. u . Sen
,c 10E-7 r.
, u10E-7 r I
I y a *
. a ;
. . y I v - '. _
v j *
- 10E-8 J 1.0E-8 ,_
~
I . Percentile I i 3 c _ _9_ _5_th_ _.:, 3 : c . i ot0E-9 y i Weon .
, o10E-9 , ;
l* , 50th ' ' i* , i y
. 5th u
I w t0E-y ...a . . .a u . . .a .. .....a .i.. . w toe-tv . .w .. a .
. .a ...a ~
LOE-6 1.0E-4 1.0E-2 10E0 1.0E-6 10E-4 LOE-2 10E0 Release Fraction For Sr Release Traction for la , Figure 3.3-15. Exceedance Frequencies for Release Fractions for ! Surry Internal Initiators. "H" SGTRs (Secondary SRVs Stuck Open) 1
~
F n n
'tE-3 o = , , , ., , , , , o tE-3 , , , , , , , ,
- i
~
E Percentile , e 8-u - - s . 9.5.t.h_ . . stE-4 r.___..._____,,,,. , 31.E-4 y Meon_ . 8 ' g ~>0tb : e - u - --- .__ u stE-5 ._________,'-----____-_-.S.in._._ ttE-5 '. g F-- g Q 7* 1.E-6 3 ' -.., , 2 7 % .
'E LE-6 f
, , , .., N ,
'.o : w;g, , f e
- 95th o "
c - - - - - - - - -
- c .
c 1.E-7 Woon '. otE-7 . v r . v - e : . e
- o
- __ _5_th_ _ __ _
u I, , O LE-e - - - - - U O LE-m -- - - - - 't . . - J ! 1.E-8 1.E-6 LE-4 LE-2 LEO LE-8 1.E-6 LE-4 LE-2 1EO t w Release Frocfion For i Release Froction For Cs u.
" t tE-3, o m , . . = , . = , m, . . , ,.;, .,
ttE-o 3 ., m, , .- m, , , , E _ Percentilej E Percentile s c : __95_t_h_ _ _ : u 8 _ _9.5_t h. . .
- o. f.E-4 e Weon , .o tE-4 _
W-4 8 i 50th - g 7 D_o_n_. th t
- e 5th u - __........__, _ - 5t:
u --... --- ----- t 1.E-5 t e '- --. ..____ , u e t E '------
---. _ 6
- n. N --- .._,
- s. o. w -
i M c _N '. _ M % w l it*1.E-6 7 -- . - - _..,_ 3 w2LE-6 :-__,,. - , i I
' - . . s.
e ' - ... , . u -
, . 1. : u . -
c -
, .- c -
[ 01.E-7 01.E-7 - o r . , y r . . i o u
- ; e
- M
. u X
w1.E-e " "" " " ' ' l w tE _ " ' .' "- ' i 1.E-8 1.E-6 1.E-4 LE-2 LEO LE-8 ".E-6 1 LE-4 LE-2 1.E0 i Release Froefion For Te Release Froction for Sr j i i Figure 3.3-16. Exceedance Frequencies for Release Fractions ) for Surry: All Internal Initiators L l i
go* LE-3,,,, n'- o 1.E-3 -, ., , ;, , , , , Percentile 8 6.
- 8. _ _ 9. 5.t h. _ _
SLE-4 7 , o f.E-4 , . . Mecn _.,.. g : g 50th !
.- .- . . 5_th_____
gtE-5 r , g t.E-5 7---- ---- ,,,,,, ca. o. w w w& 1 e .
*1 it .E-6 r '.
.
- 1.E-6 r '.,
- u. -
. '. e
- v - Percentile . o ". .
c - c . y, 01.E-7 y _. 99h_ __ , o 1.E-7 ;
} Weon ~
j -
. soth . .__*
U u att.n ~.a e ~
.a .. a .. a . . . ..a..' ._ y tt.g ~ a .-s...a .a LE-8 1.E-6 1.E-4 LE-2 tEO LE-8 1.E-6 LE-4 1.E-2 LEO y Release Fraction For Ru Release Troction For La v.
9*o .E-3,,, ,...._,... ,... 1
,,m , ., . . , -,
To .E-3 < ., , , , , , ., 1 Percentile _ E . Parcenfile 8 u : . 95_t_h_ ._ : 8 u . 9. 5_t_h_ _ _ otE-4 - Weon -, otE-4 7 Meon - 8 50th 8 50th c . i __3_tn_ _ _. . . _ u -____, . 5_th_____- u.tE-5 .
. tE-5 r ' - - - - - - - - - - - . ...,- , ,
a o, %. . M -
. Y -
w
=-
*tE-6 P *tE-6 th . 9 it r -
v - '. . - u . c . c . . OLE-7
- OLE-7
- 1
., , i v.
- v. . i o _ . , :
. _ u - . _
[],{.@ ma ..a s. d .. a .. .a . . .U .- (({.M .a ..a ..a . a *, . a . . -l ..a . LE-8 LE-6 1E-4 LE-2 LEO 1.E-8 LE-6 LE-4 LE-2 1.E0 Release Froction For Ce Release Frocfion for Bo Figure 3.3-16. (continued) t i
?
1 , Table 3.3 8 ! Summary of Variables Sampled l Surry: Internal Initiators 1 1 - Variable Analysis Descrintion V TRAIN Acc. Freq. IE - Interfacing System LOCA (Event V) [ IE-LOSP Acc. Freq. IE - Loss of Offsite Power (LOSP) l IE A Acc. Freq. IE large LOCA l IE S1 Acc. Freq. IE Intermediate size LOCA IE S2 Acc. Freq. IE Small LOCA IE S3 Acc. Freq. IE Very small thCA IE T ALL Acc. Freq. IE - Transients that require scram IE-T ilIP Acc. Freq. IE Transients from high power that require scram IE LMFWS Acc. Freq. IE - Transients due to loss of main feedwater IE SCTR Acc. Freq. IE Steam Cenerator Tube Rupture IE DCBUS Acc. Freq. IE Loss of a DC power buss l DG-FRUN Acc. Freq. DG fails to run ! DG FSTRT Acc Freq. DC fails to start UNFV MOD Acc. Freq. Unfavorabic moderator temperature coefficient l AU SCRAM Acc. Freq. RPS fails to scram reactor 1 ! MN SCRAM Acc. Freq. Failure of manual scram AUTO ACT Acc. Freq. Failure of one train of an automatic actuation system CCF-RWST Acc. Freq. CCF miscalibration of RWST level sensors BETA 2MOV Acc. Freq, Beta factor for CCF of two motor operated valves l BETA APW Acc. Freq. Beta factor for CCF of the AFWS motor driven pumps BETA LPI Acc. Freq. Beta factor for CCF of the LPIS pumps APW STM3 Acc. Freq. CCF all AFVS due to steam binding MDP FSTR Acc. Freq. Motor driven pump fails to start (generic) AFWMP PS Acc. Freq. APW motor driven pump fails to start AFWTP FS Acc. Freq. APW steam turbine driven pump fails to start ATP FR6 Acc. Freq. AFV turbine driven pump fails to run for 6 hours ATP FR24 Acc. Freq. AFW turbine-driven pump fails to run for 24 hours PORV BLK Acc. Freq. PORV block valves fail to open LPRS MOV Acc. Freq. LPRS suction MOVs fail to open MOV-FT Acc. Freq. Motor operated valve fails to transfer KNV PC1 Acc. Freq, Plugging of manual valve flow tested every month MOV PC3 Acc Frcq. Plugging of an MOV flow tested every 3 months MOV PC12 Acc. Freq. Plugging of an MOV flow tested every 12 months APW 0CC Acc Freq. CCF - open cross-connect to Unit 2 fails APWS PORV REC Acc. Freq. Pressurizer PORV fails to reclose after opening l SSRVO SB Acc. Freq. Secondary SRV fails to reclose durint,SB0 l SSRVO U2 Acc. Freq. Secondary SRV fails to reclose during SB0 at Unit 2 3.55
i l i Table 3.3 8 (continued) 1 l Variable Analysis Description SOV.FT Acc. Freq. Solenoid operated valve fails to transfer CKV.FT Acc. Freq. Check valve fails to transfer l t ilE FDBLD ec. Freq. Operator error - feed end bleed (pumps and PORVs) llE.PORVS Acc. Freq. Operator error feed and bloed (PORVs only) llE. CST 2 Acc. Freq. Operator error - align AFWS suction to backup CST
.llE. UNIT 2 Acc. Freq. Operator error cross connect AFW from Unit 2 <
llE. SKILL Acc. Freq. Operator error skill based actions RCP.SL.F Acc. Freq. T.I failure of the RCP seals before UTAF V UWATER Acc. Prog. Break location underwater for Event V PORV.0PN Acc. Prog. A prosaurizer PORV or RCS SRV sticks open RCP.SL.P Acc.~ Prog. T.I failure of the RCP seals after UTAF TI.SCTR Acc.-Prog. T.I SCTR TI ll0TI4 Acc. Prog. T.I failure of the hot leg or surge line and fraction of equivalent core Zr oxidized in vessel RCSPR.VB- Acc.-Prog. RCS pressure just before vessel breach ! CDARREST Acc. Prog. Arrest of core damage before VB PR llPME .Acc.' Prog. Fraction of core which participates in llPME ' VB. ALPHA Acc. Prog. P'obability of an Alpha mode CF , TYPE.VB Acc. Prog. Type of VB f VBl{0LSIZ Acc. Prog. Size of the hole in the vessel after ablation PRISE.LO Acc. Prog. Pressure rise at VB + RCS at low pressure or Pour PRISE.VB- Acc. Prog.- Pressure rise at VB . RCS not at low pressure CF. PRES Acc. Prog. Containment failure pressure CF. MODE Acc. Prog. Random number used to select mode of CF llB. SCAL Acc. Prog. Scale factor for pressure rise from a hydrogen burn POWERREC- Acc. Prog. Offsite power recovery FCOR Src. Term Core Release Fraction PVES Src. Term Vessel Release Fraction '! VDF Src. Term Decontamination Factor for V Underwater Releases
~FCONV- Src. Term Containment RCS Release Fraction FCCI Src. Term CCI Palease Fraction FCONC .Src. Term ~ Containment CCI Release Fraction SPRDF Src. Term Spray Decontamination Factor (DFE & DFL)
LATE 1. Src. Term Late Release of Iodine from Water in Volatile Form FLATE Src. Term Late Revolatilization from the RCS DST Src. Term Direct Containment lleating Release Fraction l FISCFOSC _Src. Term Release Fractions for SGTR Accidents
-DTPSL Src. Term = Dacontamination Factor for Late Pool Scrubbing l 3.56
Tabic 3.3 9 i Summary of Rank Regression Analyses for Annual Release Rates (fraction /yr) for Internal Initiatorc Noble Gas Iodine Cesium Sig VAR
- SRCb . R2e VAR SEQ R2 VAR S}1G R2 1 DG-PSTRT 0.35 0.13 FISGFOSG 0.54 0.28 FISGFOSG 0.59 0.36 2 IE-SGTR 0.27 0.21 IE-SGTR 0.34 0.39 FCOR 0.33 0.47 3 IE LOSP 0.27 0.28 V-TRAIN 0.28 0.47 IE-SGTR 0.32 0.58 l
4 V TRAIN 0.27 0.35 FCJR 0.25 0.52 V-TRAIN 0.30 0.67 5 DG-FRUN 0.26 0.42 1ATEI 0.20 0.56 6 LPR-MOV 0.17 0.45 PORV-BLK 0.17 0.58 7 lie-UNIT 2 -0.14 0.47 RCP-SL-F -0.13 0.59 8 FCOR 0.14 0.49 TYPE-VB 0.13 0.61 Tellurium Strontium Ruthenium SIE VAR
- SRCb R2e VAR SEQ R2 VAR SEQ R2 1 FCOR 0.51 0.29 FCOR 0.62 0.41 FCOR 0.84 0.74 2 FISGFOSG 0.50 0.53 FISGFOSG 0.35 0.52 FISGFOSG 0.33 0.85 3 V-TRAIN 0.30 0.62 FCCI 0.30 0.61 IE-SGTR 0.14 0.86 4 IE-SGTR 0.26 0.68 V-TRAIN 0.28 0.69 V-TRAIN 0.09 0.87 5 FCCI 0.12 0.70 IE SGTR 0.14 0.71 FCCI 0.09 0.88 6 PORV-BLK 0.10 0.71 FDCll -0.12 0.72 CF-PRES -0.07 0.89 VB-ALPHA 0.10 0.73 3.57
l Table 3.3 9 (continued) Lanthanum Cerium Barium M VAR
- SRCb . R2e VAR E R2 VAR E R2 1 FCOR 0.62 0.42 FCOR 0.68 0.50 FCOR 0.64 0.44 2 FCCI 0.36 0.54 FCCI 0.32 0.60 FISGFOSG 0.39 0.58 3 V TRAIN 0.28 0.62 FISGFOSG 0.27 0.66 V TRAIN 0.26 0.64 4 FISGPOSG 0.28 0.69 V TRAIN 0.25 0.72 FCCI 0.25 0.71 5 IE SGTR 0.11 0.70 SPRDF 0.10 0.73 IE SGTR 0.16 0.73 6 FDCil 0.10 0.72 FDCil 0.10- 0.74 FDCil 0.11 0.75 7 SPRDF 0.10 0.73 VB ALPilA 0.09 0.76
- Variables listed in the order that they entered the regression analysis, b Standardized regression coefficients (SRCs) in final regression model.
- R2 values with the entry of successive variables into the regression model.
3.58
.m -
Table 3.3-10 Sununary of Rank Regression Analyses for Annual Release Rates (fraction /yr) at Surry for Two Samples for Internal Initiators Iodine Iodine Noble Gas Noble Gas Sample 2 Sample 1 Samnle 2 Sanele 1 _ SRC R2 VAR _SEC R2 3gg_b RC 2 VAR _5RC_ R2 VAR h VARA 0.19 FISGFOSG 0.54 0.28 FISGFOSG 0.60 0.35 1 DG-FSTRT 0.35 0.13 IE-SGTR 0.45 0.34 0.39 IE-SGTR 0.39 0.51 0.21 DG-FSTRT 0.39 0.35 IE-SGTR 2 IE-SGTR 0.27 0.28 0.47 V-TRAIN 0.26 0.57 0.28 V-TRAIN 0.29 0.43 V-TRAIN 3 IE-IDSP 0.27 0.49 FCOR 0.25 0.52 FCOR 0.24 0.63 6 V-TRAIN 0.27 0.35 IE-IDSP 0.25 v DG-FSTRT 0.15 0.65
-0.19 0.53 IATEI 0.20 0.56 g 5 DG-FRUN 0.26 0.42 POWERREC 0.17 0.58 LATEI 0.14 0.67 6 LPR-MOV 0.17 0.45 SSRVO-SB 0.19 0.56 PORV- BIE
-0.13 0.59 IE-LOSP 0.12 0.69 7 HE-UNIT 2 -0.14 0.47 DG-ERUN 0.18 0.59 RCP-SL-F 0.13 0.61 V-UVATER -0.17 0.70 0.4- TYPE-VB 8 FCOR 0.14 AFW-STMB -0.12 0.71 9
RCP-SL-F 0.11 0.73 10
3able 3.3-10 (continued) Strontium Strontium Lanthanum Lanthanum Sample 1 Sample 2 Sample 1 Sample 2 p2c R2 VAR SRC R2 VAR _.2RG R2 Step ._ VARA SRCb VAR SRC 1 FCOR 0.62 0.41 FCOR 0.64 0.40 FOOR 0.62 0.42 FCOR 0.61 0.39 2 FISGFOSG 0.35 0.52 FCCI 0.31 0.50 FCCI 0.37 0.54 FCCI 0.33 0.54 3 FCCI 0.30 0.61 V-TRAIN 0.32 0.60 V-TRAIN 0.29 0.62 V-TRAIN 0.33 0.63 4 V-TRAIN 0.28 0.69 FISGFUSG 0.27 0.67 FISGFOSC 0.28 0.69 FISGFOSG 0.17 0.66 5 IE-SGTR 0.14 0.71 IE-SGTR 0.15 0.70 IE-SGTR 0.11 0.70 VB-ALPHA 0.13 0.68 u, h c' -0.10 0.72 RCP-SL-F 0.11 0.69 6 FDCH -0.12 0.72 VB-ALPHA 0.15 0.72 FDCH 7 VB-ALPHA 0.10 0.73 ATP-FR6 -0.13 0.73 SPRDF 0.11 0.73 FCONC 0.10 0.70 8 0.73 9 0.74 10 0.75 11 0.76 a Variables listed in the order that they entered the regression analysis. b Standardized regression coefficients (SRCs) in final regression model. cR2 values with the entry of successive variables into the regression model.
to ......,,1- ,,,, ,,,,,,,..,,,,,1 ,m, , , m e, , m m, ,,m,,. i l 1 0.8 %._, - N'(,s. 06 - --. ~ , ~
,m, o.4 -
\ -
4-o.2 -
~/ ' [ !
--kY o.o -
j
-o.2 -
i
-a4 LEGEND -
FCOR ; as .
,,,, DG-FRUN,,,,,, .
. -o.s L-
--to
'"""'"""""""""""~d '"'"""'""""
LE-8:1E-7 LE-6 1E-5 1E-4 tE-3 1E-2 tE-1 tEO RELEASE FRACTION:lODitE i E a I I i llTTT' I I I llHT 4 3 1311] ,I 11575 i l 3 ]ITf3 I i l ] III3 l l l g irry i . t iviq o.8 - -
- o.s +
/ -
."l:
f y.
, lg
_ j. o2 - 1 l c ;- ., I
, -8 go s ' %__ v e M.w,% i
.i , .
j k '
-a2 . .
,l \
- 0.4 -
t LEGEND'- s ,, rf -o,s E-EGTR .
..ATEl :
s t s -0.8 E-LOSP ! n7 FCONV .
> N- . o m,mi , m . r. , m m ,m, - , m m,4 , m 'm. , m a , , m . m, , om ,
.to mi LE-8 'tE-7 tE-6 : tE-S 1E-4 1E-3 1E-2 1E-1 LEO-
,4 RELEASE FRACTION LODINE q
l
' Figure'3,3-17. Partial Rank Correlation Coefficients for Exceedance Frequencies for Release Fractions for
,3 Surry Internal Initiators. Iodine +
a.
\. t r p N%T w
to ,,,,, , , u , ,,,,,,,,oi, ,,,,,,",,,om , ",-
~
0.8 -- 0.6 04
,4.\
,b
~_ye - __
on U Alli
. d j, 1
'V -
\"!
-a4 -
LEGEND
-0.6
~
~~
-"U
., _N.J kU1L
-t0
""""""~'""'""""""""""""""
lE-8-LE-7 tE-6 1E-5 1E-4 1E-3 tE-2 1E-l 1E0 ' RELEASE FRACTON: CESUM .
'to ,,u,, ,,,,,,,,,,,,,,,,,,om, ,o, ,o,,, ,,,n ,
na - f_ ,, 0.6 rW -
- 0.4 -
. l.-
l 0.2 -
+}k 0 ,.-.
..;f.i
[ d 0.0
-a2 3
Wl -
- 0.4 -
LEGEND
-a6 V-TRAN -
~ SUN LOS3
-"8
_,f,,C,0NV , _,,, -
' '"""'"""'""""""d """ ' ' " " " ' ""a ""'
-10 LE-8 1E-7 1E-6 1E-5 1E-4 ' 1E-3 1E-2 lE-1 LEO RELLASE FRACTON CESUM
)
Figure 3'.3 18. Partial Rank Correlation Coefficients for Exceedance-Frequencies for. Release Fractions for Surry Internal Initiators. Cesium 1 3.62
to ,,,o, , m m, ,,,m, ,,,,,, ,,,,,,,,,,,,,,,,,m,,, ,,m,,,, 0.8 -
,p., ~. .
o.s l o.4 - [
~
i- 0.2 - a r
-g
,, 0.0 \.a*g i ,
z dv -0.2 i 1 I LEGEND y' -n4 FOOR
+ FSGW, i cE -o.s -
,,,, V-TM N ,,,, .
-02
$~.
"""""="""""""""'"""4 "">"""
-to i
LE-8 1E-7 LE-6 1E-5 1E-4 tE LE-2 1E-I LEO RElIASE FRACilON: STROffTpJM Figure 3.3-19. Partial Rank Correlation Coefficients for
", ~ Exceedance Freque'ncies for Releaso Fractions for ,
. ; Surry = Internal Initiators. Strontium ;
'm . to ...m, ,,n,, i n m.,, in,,,,, , , , , , , , o m, ,,,,,,,,,,u,,,,
0A - [n), i
'l . ; ,
g o.s - - .-
]%
g4 l- . gg . m 'r . .
\.
qt o.2 m'. 4 . ao - (,A g= J!; 3 -0.2 ,y , ; LEGCO - gL --ad. - TORT T T>GF05G,,,,, .
- i. -as . V-TRAN e
-0.s kf .
I , I I t't!! 1 ( l I ? ll t I f l]t11! l l t f if etl f l f I ? t!d f I l 1!!!) ! ll f it'.ll t [tf?u
; tE-8 1E-7 1E-6 1E-5 1E-4 1E-3 LE-2 tE-1 1EO-RELEASE FRACTION: LANTHANNUM ;
. Figure 3.3 20. Partial Rank Correlation Coefficients'for <
.Exceedance Frequencies for Release Fractions for Surry Internal Initiators. Lanthanum 1
3.63 Y l, e ii: , p
.'I' ____ _ _ _ _ _ _ _ _ _ _ _ _ _ . . _ _ _ _ _ _ _ __ _ ,. . . _ , _ . . .
\_ ~ i . l. i.
- l. _
i . l' n e i t I
,h10E S m , ., , , m ., ,. - o
.LOE t -
~ -
y 8 1 l u - u - ! . _o toe-4 g:::==== :- - o
,,T , , . ot0E-4 g' .;
l 0 -
-u t
e
-O..
u s%:::=== - . e~ - utOE-5 _- .u .. ' . _ - g
- r=--.,_,',
. ' t0E-5 w
p Q _N w e, :
- l- c- _ , ..
"' ,_ c-
' *10E-6 ' y1.0E-6
', _ Percentile
; ,. ' *Percen
* * = ..'CIE-* * :::; : - ,
y o c 95th o 95th "' - o10E-7 Meon c - - - - - - - - - - O t0E-7 M60" '\ T* ' 50th
- i* 50th
i o 5th
--------- i, 5th ",.
x
- u ----- - :
y tO{._ ...w .. w e.w . . .w .uw ...w , , . .&.s x .
, a y tg{.A ...w ...a . . is ,..a e i .m n.w U.w . .a' 1.0E-8 LCE-6 10E-4 10E-2 10E0 10E-8 t0E-6 1.0E-4 t0E-2 1.0E0 I
Release Froction For ! ~ Release Froefion for Cs w '. 10E-t . . , m r m .:, , . . , ., c h.0E-3 , -, , , ., ., on ., . 5 ~ Percentile _ Percentile l - 95th 1 o t0M - ---~~- 95th
---~-----
- r- Weon m -o LOC-4 -
Weon - o' E 50th - o 50th
?_:::: 5th 5th
~
ut0E-5 e --- " t :::::...~~~~~~~~~ , u t0E-5 e ,. 2****-::: m __ -----'"_
- c. .**,
M N c.. N _ _ .*,.. . c- - Y _ c-
- 10E-6 *
- toe-6
-3 LE
-g -- , - -----..=_;,,,, .,
.: Lt , - ,_______,
N,, _ e o -
._-, p e _ _ %_ - .
c -
.. . .- u ' < -
s .- *
- vO 10E-7 y **,
- Jt0E-7 *
. .s r s, .
e .. .
* : e ',
- u _
','., u ..
Yt0E-n n' .'"..".."*G'-'" ' 5t0E-n1, uw .ua ,. a
. ~ 't '. u w , .ua e' ~
10E-8 t0E-6 10E-4 10E-2 10E0 t0E-8 10E-6 t0E-4 10E-2 LOCO Release Fraction For Te Release Fraction For Sr Figure 3.3-21. Comparison of Exceedance Frequencies for Two Independent Latin Hypercube Samples Surry Internal Initiators
.g m
l r
^
'oLOE t ., , -, . , , , , . . , -
'LOE O * -, . , , - - , , , , ,
e e Percentile - Percentile >- 1 : 95th - [ 95th s t0E-. , . - yegg- , s t0E-* r ~~Ms66-.s 8 50th -- _ 8 Th 5tn 2 5th
~
E : .--- u 10E-5 J- l"::::=.-..._._, i toe-5 J:- : ~_ !!:n._, -""-~~~, -'-~~"- , l j y-e a - y e a t 7 - tr ~s
&* 10E-6 ,
u210E-6 r t, e-e e.. o a u . c j g - c . o 1.OE-7 ) o 10E-7 , ' ' . . , . , r v , i
.m ,
o e . e . e .
- o .. o
! D LOE-n ~~ ~ ~ ~ ~ ' " - ~~d' o.- Dt0E-n '" . " t c" - . '" '.'t .- l t0E-8 t0E-6 t0E-4 1.0E-2 t0E0 10E-8 t0E-6 10E-4 t0E-2 10E0 L i l Release Frociion For Ru . Release Froction For to u a e
'u e, '- LOE t ttOE . ., . , ..,....,.;,
i i o- ..,..,...,m, , ., , . - . - - - e 1,, , ' >- Percentile >- - Percentile l : 95th . - l 95th a t0E-4 , - M 66- , LOE-d -
- Ms6h- -
8 50th : 8 50th
' ---. _ _ _5_th_ _ _ _ -
_ - ==: - __5_th____ ut0E-s ; - - ' " :: . . . . .__ _ _ ., e t0E-s - *"""::w..__ ,
---s:~.
e ___,- e w
~ .
a . y a , . y ,. o- ~ .. er 's,, wEt0E-6 , '. .. w 2t0E-6 . ~-
. -.n_...-.- -. ....
u - ,. u , . - C * 'a C o t0E-7 . ' _ . -. ,
, o t0E-7 _ ' . .. -
y . y . e +, : * .'. i
. .. _ e ..
u -. .. . u .. Ot0E-a . . . .a ...s \ .. -. .s . .L .. O t0E-a . ..-w . . ..a . . . .a ....s'..a. '
.. a 1. - -
10E-8 10E-6 t0E-4 10E-2 10E0 10E-8 t0E-6 10E-4 t0E-2 - t0E0 Release Traction For Ce Release Fraction For Bo Figure 3.3-21. (continued)
i i DG-FSTRT
. t0 - ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,-
0.8 - - l 0.s -
~~.sy -
j e4 .
\ .
n t 0.2 -
\ ~
s A.] 0.0
-0.2 -
i
'1
-0.4 - -
LEGEND .!
-0.6 .
'(
SAW V 1
-0.8 ~
SAMPLE 2 1. _ l l l l tita f f iIllia
$ l[Illi I f IIIlll l I I !!III O E ! Ill! I I I !!!I !
A, . LE-8 1E LE-6 1E-5 1E-4 1E-3' LE-2 1E--I LEO - RELEASE FRACTON: ODINE LATO
'h to , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,, , , , , , , , ,m,,,, , , , , .
1,~ .. 0.8 - [5 o y L **.- H 0.6 -- - U,c g
. 0.4 -
. I u]ly. '
02 -
/. -
^'
t ;
,- 0.0 g --
1 i O -0.2 . . 4 3 ll 3 -0.4 - J n.
-0.s .. LEGEND ._
. SAMPLE 1 .
-0.8 ~
,' SA M '2 . . . . .
~
I i I t i t ttd t I s ittnd e i t i tinI t t t i ttill i i t i t.fil e i g s tiggl t g g i n t ilj g ', g g i gg, p 1E tE-7 LE-6 1E-5 <E , 1E-3 1E-2 1E-1 LEO RELEASE 'k. "el0N: 10 DINE L; Figure 3.3 22. Comporf4n,of two PRCC Curves for Two Repre at,tativs Va:iables
. % kl
> , s ')
3.3.3 Results for Fire Initiators At Surry, fires were found to be important in the emergency switchgear , roorn , the auxiliary building, the control room, and the cable vault and I tunnel as discussed in Sections 2.2.2.2 and 2.5.3. Table 2.5 14 lists the l 10 most probable APBs for the PDS group and the five most probable APBs l-that have VB and early containment failure (CF). Since there is no possibility of core damage arrest for the fire initiators, all the APBs have VB. Table 3.3 11 lists the mean source terms for the same APBs listed in Table 2.5 14. The structure of tables such as 3.311 is discussed in Section 5.3.1.1. The 10 most probable fire APBs do not have. aboveground containment failure. As a result, the releases associated with these APBs are very small. When there is no containment failure, SURSOR describes releases with a single release segment rather than the two release segments used when there is L containment failure. Thus, the early release is zero for the 10 _ mos t i pobable bins in Table 3.3 11. .The five most probable APBs with VB and early containment failure have' low conditional probabilities (see Table 2.5-14) but larger releases than the APBs without containment failure. The mean source terms in Table' 3.3 11 can be used to compare the releases L associated with specific APBs. Ilowever, as these mean. source terms are l' typically not calculated over the same sample elements, - fine distinctions
- between source terms ' associated with different APBs may be lost in the E
-averaging process.
Table 3.3-11 presents mean source terms but does not contain any frequency l' information. In contrast, Figure 3.3-23, which summarizes the release fraction CCDFs for for the I, Cs, Sr and La radionuclide classes, presents l information on both source term size and frequency. The derivation of the curves in Figure 3.3-23 is discussed in Section 3.3.1.1, 3.3.4 Sensitivity Analyses for Fire Initiators No sensitivity analyses were performed for fire initiators at Surry. 3.3.5 Results for Seismic In Qinters: LLNL llazard Ristribution The seismic risk analysis .was performed using two different seismic hazard distributions. This section i concerns the source term results based on the hazard distribution developed by LLNL, Source term results based on the seismic - hazard distribution' developed by EPRI are presented in Section 3.3.7. The differences between these two distributions are discussed in NUREG/CR 4550, Vol. 3.s
.The accidents initiated by earthquakes were analyzed in two groups. Those E due to seisms with a maximum ground acceleration in excess of 0.6 g were denoted the high acceleration events. It was judged that the destruction ,
in the general vicinity of the plant for those earthquakes would be so great that evacuation would be ineffective. For the low acceleration events, less than 0.6 g, it was estimated that evacuation would be possible, although perhaps more slowly than in emergencies without l earthquakes.
- 1. 3.67 l
l r
The basic results of the source te rn, analysis are the release fraction CCDFs for each observation in the sample, Figures 3.3-24 through 3. 3-26 show four statistical measures of these release fraction CCDFs for the seismic PDS groups based on the L111L hazard distribution. Results are shown for the iodine, cesium, strontium, and lanthanum radionuclide clas-ses: both acceleration ranges are combined in these plots. Figure 3.3 27 contains the release fraction CCDFs for all three PDS groups and both accoleration ranges together for the LillL hazard distribution. , 1 1 3.3.5.1 Results for PDS Groun E0 1. LOSP (No SBO): High Acceleration. LLNL Hazard Distribution. This PDS group consists of accidents initiated by LOSP, but in which SB0 does not result. The LOSP is due to the earth-quake, but the DCs start and run, suppling station power. There are five-PDSs in the seismic LOSP group es listed in Table 2.2 6, and discussed in , Sections - 2. 2.2. 3 and 2. 5. 5.1. This PDS group consists of the fraction of
- these accidents due to seisms with a peak ground acceleration (PCA) exceeding 0.6 g.
Table 2.5-15 lists the .10 most probable AP3s for this PDS group and the five most probable APBs that have VB and early CF. Table 3.3-12 lists the mean source terms for these same APBs. The structure of tables such as Table . 3. 3-12 is discussed it, tion ' 3.' '. The release fractions are low for the 10 most probab. b- e right of them have no CF and the other two have BMT, Four of s.ie five most probable bins with VB and early CF have iodine and cesium release fractions on the order of 0.30. These APBs have mean conditional probabilities (see Table 2.5 15) of 0.004 or less, however.
' Table 3.3-12 presents mean. source terms but does not contain any frequency information. In contrast, Figure 3.3-24 presents information on the ;
frequency with.which different values of the release fraction are exceeded, and displays the uncertainty in that frequency. This figure summarizes the LillL seismic LOSP release fraction CCDFs for the lodine, cesium, strontium,
.and lanthanum radionuclide classes for both acceleration ranges. For . a given release frSction, the exceedance frequency contribution from the high PGA group is roughly an order of magnitude lower than the contribution from the low PGA group., The interpretation and'. generation of figures such as -Figure 3.3 24 is discussed in section 3.3.1.1.
I
!3.3.5.2 Results for PDS Group EO 2. SBO: Hich Acceleration. LLNL
}gzard Distribution. This PDS group consists of accidents initiated by LOSP in which-SB0 follows. The LOSP is due.to the earthquake, and the DGs fa P to start due to seismic and random hardware failures. Due to the-n; c'c failures in the ~ electrical distribution system that may be:
n r ,ed, it was' judged that offsite power would not be recovered within the timeframe of this analysis. Thus there is no chance of arresting core L damage or avoiding VB in this PDS group. The two "A" PDSs in this group
=have-large pipe breaks and initial'CF due to failures of the SG or RCP pump supports that are coincident with the initiating LOSP. The eight PDSs in the'SBO. group are listed in Table 2.2-6 and discussed in Sections 2.2.2.3 and ' 2. 5.- 5. 2. This SB0 PDS group . consists of the- fraction of these accidents due to seisms with a PCA exceeding 0.6 g. .
3.68
i Table ~2.5 16 lists the 10 most probable APBs for the PDS group and the five most probable APas that have VB and early CF. Table 3.3-13 lists the mean source _ terms for these same APBs. Two of the 10 most probable bins have initial failure of the containment due to SG or RCP support failures; for these bins - (the 4th and 6th in order, first characteristic - C) , the releases are much higher than for the others in the 10 most probable APBs. Of the other eight APBs, three have BMT, and the five have no CF. For the five most probable ' APBs with both VB and early CF, the - iodine and cesium release fractions are in _ the range of 0.25 to 0.50. The total mean conditional probability (see Table 2.5-16) of these APBs is ab: m 0.11.
- F16 ure 3.3-25 shows four statistical measures of-the release fraction CCDFs for the. iodine, cesium, strontium, and lanthanum radionuclide classes for the LLNL seismic SBO group for both acceleration ranges. For a given release fraction, the exceedance frequency contribution from the high PGA group is roughly a factor of 5 lower than the contribution from the low PGA group.-
3.3.5.3 Re frui t s for PDS Group EO 3. LOCAs: High Acceleration. . LLNL , Hazard Distribution. This PDS group consists of accidents initiated by seismic pipe bresks. The failures in the ECCS required to respond to these Lbreaks are partially seismic and partially random. There is no SBO, but some of the seiswie- failures are in portions of the the electrical distri-
' bution system that supply power to the ECCS, sprays, or AINS. There are 11 PDSs in this1 group as listed in Table 2.2 6 and discussed in Sections 2.2.2.3 and 2.5.5.3. .The five "A" PDSs in this group have initial CF due to failures of_the SG or RCP, pump supports. This PDS group consists of the Craction of the seismic: LOCA accidents due to earthquakes with a PGA exceeding 0.6 g.
l Table 2.5-17 lists the fifteen most probable.APBs for the PDS group, and Table 3.3' 14 lists the mean source terms for these saine APBs. The six APBs that' have C- as the first characteristic have initial CF. Fot these APBs the _ releases : are much higher than for the others in the 15 most probable .
-APBs. For ~ the 3rd and 4th APBs, which have initial CF and the sprays i l
Inope rable ,- the. release fractions are fairly high. For the 7th, lith, '
-12th,_ and 15 th ' APBs ,- which have initial CF and sprays operating, the release fractions are much lower.- The mean conditional probabilities of the APBs are given in Table 2.5 17 The mean conditional probability of the most likely APB with initial CF is 0.80.
Figure 3.3k26 shows four statistical measures of the release fraction CCDFs fot four radionuclide classes for the LLNL seismic LOCA group for both acceleration ranges. For a given release fraction, the exceedance fre-quency _ contribution from the high PGA group is roughly half the contri- i bution from the low PGA group. 3.3.5.4 Results for PDS Group EO 1. LOSP (No SBO): Low Acceleration. LLNL ~ Hazard Distribution. This PDS group consists of seismic accidents 11nitiated by LOSP, but in which SB0 does not result, and in which the earthquake has a PGA less than 0.6 g. The five PDSs in this . group are discussed in Sections 2.2.2.3 and 2.5.5.1. Similar PDSs with a PCA exceeding _0.6 g are discussed in Section 3.3.5.1. Table 2.5-18 lists ~the
- 10 most probable APBs for this PDS group and the five most probable APBs '
that have VB - and early CF. Table 3.3-15 lists the mean source terms for 3.69 i
these same APBs. The release fractions are low for the 10 most probable 1 bins because eight of them have no CF and the other two have BMT. Four of , the five most probable bins with VB and early CF have iodine and cesium j relssase fractions on t.he order of 0.30. These APBs have mean conditional prebabilities (see Table 2.5-18) of 0.004 or less, however. 1 Figure 3.3 24 sum:narizes the release fraction CCDFs for the iodine, cesium, strontium,'and lanthanum radionuclide classes LLNL seismic IASP group for l
- both acceleration ranges. For a given release fraction, the exceedance frequency contribution from the low PGA group is rough 1; an order of magnitude greater thnn the contribution from the high PGA group.
3.3.5.5 Recults for PDS Groun EO 2. SBO: Low Acceleration. LLNL Hazard Diarribution. This . PDS group consists of seismic SB0 accidents in which the earthquake has a PGA less than 0.6 g. The five PDSs in this group are discussed in Sections 2. 2. 2. 3 and 2. 5. 5. 2. Similar PDSs with a PGA ex-ceeding 0.6 g are discussed in Section 3.3.5.2. Table 2.5-19 lists the 10 most probable APBs for the PDS group and the five most probable APBs that have VB and early CF. Table 3.3 16 lists the mean source terms for these same APBs. One of- the 10 most probable bins has initial failure of the containment; for this bin (the 7th, first characteristic - C), the releases are-much higher than for the other APBs in the 10 most probable APBs. The other nine ' most ~ probable APBs have BMT or no CF, and so have relatively small release fractions. For the five most probable APBs.with both VB and early CF, the. iodine-and cesium' release fractions are in the range of 0.25 to 0.50. The total mean conditional probability (see Table 2.5-16) of these APBs is about 0.06. Figure 3.3 25 shows four statistical ~ measures of the release fraction CCDFs for.the iodine, cesium, strontium, and lanthanum radionuclide classes for. the LLNL seismic SB0 group for both acceleration ranges. For a given release fraction, .the exceedsnce frequency contribution from the low PGA group is roughly a factor of 5 higher than the contribution from ' the high PGA group. 3.3.5.6 Results for PDS Grpuo EO 3. LOCAs: Low Acceleration. LLNL
' Hazard - Distribution. This PDS - group consists of accidents initiated by seismic pipe breaks in which the earthquake has a PGA less than 0.6 g. The five PDSs ini this group are discussed - in Sections 2. 2. 2. 3 and 2.5.5.3.
Similar PDSs with a PGA exceeding 0.6 g ere discussed in Section 3.3.5.3. ' Table 2.5 20 lists the 15 most probable APBs.for the PDS group. Table 3.3-17 lists the mean source terms for these same APBs. The five APBs that have' C as the first characteristic have initial CF. For these APBs the
. releases are much hi B her than for the other APBs in the 15 most probable APBs. For the 3rd and 6th APBs, which have initial CF and the sprays
( inoperable, the release fractions are fairly high. For the 9th, lith, and
~12th APBs, which have initial CF;and sprays - operating, the release L fractions are much lower. The mean conditional probabilities ofLthe APBs 1: are given in Table -- 2.5-20. The mean conditional- probability of the most l likely APB with initial CF is 0.80~
L I Figure 3.3-26 shows, four statistical measures of the release fraction CCDFs L for four radionuclide classes for the . LLNL seismic LOCA group for both L 3.70 l- .. . I
l acceleration ranges. - For a given release fraction, the exceedance fre-quency contribution from the low PGA group is roughly twice the contribution from the high PGA group. 3.3.6 Sensitivity Analyses for Seismic Initiators: LLNL Hazard Distribution To determine the effect of the initial failures of the containment due to failures of the SG and RCP supports, the integrated risk analysis using the LLNL hazard distribution was repeated without these induced seismic fail-ures of the containment at the start of the accident. Eliminating}}