ML20206N279

Estimated Safety Significance of Generic Safety Issue 61

ML20206N279
Person / Time
Site:	Pilgrim
Issue date:	06/30/1986
From:	Economos C, Lehner J, Perkins K BROOKHAVEN NATIONAL LABORATORY
To:	Office of Nuclear Reactor Regulation
References
CON-#487-5035, CON-FIN-A-3793 2.206, BNL-NUREG-51986, NUREG-CR-4594, NUDOCS 8607010423
Download: ML20206N279 (119)

Text

-

NUR5G/bR-4594 BNL-NUREG-51986 Estimated Safety Significance of Generic Safety Issue 61 Prepared by J. R. Lehner, K. R. Perkins, C. Economos Brookhcven National Laboratory Pr psted for U.S. Nuclear Regulatory Commission i

8g70 g 860630 CR-4594 R PDR

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

NOTICE Availability of Reference Materials Cited in NRC Publications Most documents cited in N RC publications will be available from one of the following sources:

1, The NRC Public Document Room,1717 H Street, N.W.

Washington, DC 20555

2. The Sucerintendent of Documents, U.S. Government Printing Of fice, Post Office Box 37082 Washington, DC 20013-7082

3. The National Technical Information Service, Springfield, VA 22161 Although the listing that follows represer.ts the majority of documents cited in NRC publications, it is not intended to be exhaustive.

Referenced documents available for inspection and copying for a fee from the NRC Public Docu-ment Room include NRC correspondence arid internal N RC memoranda; NRC Office of Inspection and Enforcement bulletins, circulars, information notices, inspection and investigation notices; Licensee Event Reports; vendor reports and correspondence; Commission papers; and applicant and licensee documents and correspondence.

The following documents in the NUREG series are available for purchase from the GPO Sales Program: formal NRC staff and contractor reports, NRC sponsored conference proceedings, and NRC booklets and brochures. Also available are Regulatory Guides, NRC regulations in the Code of Federal Regulations, and Nuclear ilegulatory Comminion issuances.

, Documents available from the National Technical Information Service include NUREG series

! reports and technical reports prepared by other federal agencies and reports prepared by the Atomic l Energy Commission, forerunner agency to the Nuclear Regulatory Commission.

l Documents available from public and special technical libraries include all open literature items,

! such as books, journal and periodical articles, and transactions. Federal Register notices, federal and i state legislation, and congressional reports can usually be obtained from these libraries.

l l Documents such as theses, dissertations, foreign reports and translations, and non-NRC conference l proceedings are available for purchase from the organization sponsoring the publication cited.

Single copies of NRC draft reports are available free, to the extent of supply, upon written request to the Division of Technical Information and Document Control, U.S. Nuclear Regulatory Com-mission, Washington, DC 20555.

Copies of industry codes and standards used in a substantive manner in the NRC regulatory process are maintained at the NRC Library, 7920 Norfolk Avenue, Bethesda, Maryland, and are available l there for reference use by the public. Codes and standards are usually copyrighted and may be l purchased from the originating organization or, if they are American National Standards, from the l American National Standards Institute,1430 Broadway, New York, NY 10018.

I l

M NUREG/CR-4594 BNL-NUREG-51986 Estimated Safety Significance of Generic Safety Issue 61 Manuscript Completed: June 1986 1 Date Published: June 1986 Prepared by J. R. Lehner, K. R. Perkins, C. Economos Brookhaven National Laboratory Upton, NY 11973 Prep: red for Division of Safety Review and Oversight Offics of Nuclear Reactor Regulation U.S. Nuclear Regulatory Commission Wcshington, D.C. 20555 NRC FIN A3793

ABSTRACT The potential threat posed by transient initiated accident sequences in-volving,BWR systems capable of releasing steam into the wetwell airspace and thereby pressurizing the containment has been examined. This study estimates the likelihood of a rupture in one of a number of high pressure steam lines which pass through the wetwell airspace before entering the suppression pool.

If the broken steam line is connected to an active steam source, such as a stuck open relief valve, which supplies steam at a high enough rate and over a long enough interval, the suppression pool bypass may result in containment overpressure failure. The three BWR plant systens identified as having compo-nents whose failure could lead to a steam discharge into the wetwell airspace and pressurization of the containment are: the Main Steam Relief Valves and associated discharge lines, the High Pressure Coolant Injection turbine ex-haust and the steam condensing relief lines in the Residual Heat Removal Sys-tem. This study outlines the postulated accident sequences, estimates their frequency, and calculates containnent response to the proposed steam dis-charges using the computer code CONTEMPT 4. Based on the predicted containment response, the investigation estimates the probability of core melt, the possi-ble fission product release and the associated consequences. Mitigating actions designed to prevent containment failure as well as actions designed to prevent core melt given containment failure are also discussed.

The study concludes that, while some of the sequences could potentially result in core melt and cause significant releases of fission products, the frequency of these sequences are judged to be sufficiently small to remove them as significant contributors to public risk.

i NUREG/CR-4594 111 June 1986

f CONTENTS Page ABSTRACT............................................................... iii LIST OF TABLES......................................................... vii LIST OF FIGURES........................................................ viii ACKNOWLEDGMENTS........................................................ xi LIST OF ACR0NYMS....................................................... xiii 1.0 EXECUTIVE

SUMMARY

................................................. 1

2.0 BACKGROUND

AND PRIOR WORK

SUMMARY

................................. 3

3.0 DESCRIPTION

OF SAFETY ISSUE....................................... 8 3.1 Accident Sequences Selected for Analysis..................... 9 3.2 Estimated Frequency of Occu rrence. . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 3.2.1 Sequence No. 1 - MSRV Di scharge Li ne Break . . . . . . . . . . . . 11 3.2.2 Sequence No. 2 - HPCI Turbine Exhaust Line Break...... 14 3.2.3 Sequence No. 3 - RHR Steam Condensing Mode Relief f Line Break............................................ 14 3.3 Containment Survival......................................... 15 3.3.1 Mass and Energy Source Terms.......................... 15 3.3.2 Limi tati on of CONTEMPT Cal cul ati ons. . . . . . . . . . . . . . . . . . . 15 3.3.3 Results of CONTEMPT 4 Calculations..................... 19 3.3.4 Containment Failure Modes............................. 20 3.4 Possible Mitigating Actions.................................. 21 3.4.1 Mitigating Actions to Prevent Containment Failure..... 21 3.4.2 Mitigating Actions to Frevent Core Damage............. 22 3.5 Sequence Frequency Results................................... 23 4.0 ESTIMATED SAFETf SIGNIFICANCE..................................... 51 4.1 Estimated Core Camage Probability............................ 51 4.2 Estimated Fission Product Release and Consequences........... 51

5.0 CONCLUSION

S....................................................... 54

6.0 REFERENCES

........................................................ 55 APPENDIX A. BWR Owners Group Comments on BNL Informal Report BNL-NUREG-31940........................................... A-1 APPENDIX B. Estimated MSRV Failure to Close Frequency and MSRV Discharge Line Rupture Frequency.......................... B-1 NUREG/CR-4594 v June 1986

l i

CONTENTS (Cont'd)

Page APPENDIX C. Estimated Consequences for Steam Bypass Accidents Which Result in Rapid Pressurization of Containment....... C-1 APPENDIX D. CONTEMPT Modeling for Steam Discharge Line Breaks in the Wetwel1 Airspace....................................... D-1 l

i i

i a

i i

i 1

NUREG/CR-4594 vi June 1986

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

LIST OF TABLES Table Title Page 3.3-1 Cases for Mass and Energy Data............................. 16 3.3-2 Mass-Energy Release Data................................... 17 3.5-1 Sequence No. 1 (MSRV Line Break)........................... 24 3.5-2 Sequence No . 2 (HPCI Li ne Break) . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.5-3 Sequence No. 3 (RHR Reli ef Li ne Break) . . . . . . . . . . . . . . . . . . . . . 26 4.2-1 Estimated Environmental Release Fractions for a Core Melt Accident Resulting from a Steam Discharge Line Break.. 53 A-1 BWROG Comment on BNL-NUREG-31940 and Response.............. A-5 C-1 Estimated Environmental Release Fractions for a Core Melt Accident Resulting from a Steam Discharge Line Break.. C-6 C-2 Summary of Input Assumptions for the CRAC-2 Consequence Calculations for a Core Melt Accident Resulting from Steam Di scharge i nto the Wetwell Ai rspace. . . . . . . . . . . . . . . . . . C-7 C-3 Total Estimated Risk for the Three Sequences Which May Result 'in Steam Discharge in the Wetwell Airspace.......... C-8 NUREG/CR-4594 vii June 1986

h LIST OF FIGURES Figure Title Page 2-1 Containment Response to S0RV + SRVDL Rupture from BNL-NUREG-31940........................................... 6 i 2-2 Simplified Transient Event Tree Used in BNL-NUREG-31940... 7 3-1 Mark I Pressure Suppression Containment................... 27 3-2 Mark II Pressure Suppression Containment.................. 28

3-3 HPCI System Schematic..................................... 29 3-4 -RCIC System Schematic..................................... 30 3-5 Schematic of RHR Heat Exchanger in the Steam Condensing t Mode...................................................... 31 3-6 Simplified Event Tree for MSRV Discharge Line Break....... 32 3-7 Simplified Event Tree for HPCI Turbine Exhaust Line Break..................................................... 33 3-8 Simplified Event Tree for Line Break During Steam Con-densing Mode.............................................. 34 i

3-9 Mai n Steam Rel i e f Val ve Fl ow Rate . . . . . . . . . . . . . . . . . . . . . . . . . 35 i

3-10 Typical CONTEMPT 4 Pressure Response Calculation for Peach Bottom using BWROG Transient........................ 36 >

3-11 Typical CONTEMPT 4 Pressure Response Calculation for Limerick Usi ng BWROG Transi ent . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 3-12a CONTEMPT 4 Pressure Response for Peach Bottom Using BNL-NUREG-31940 Transient ~................................ 38 3-12b (Same as Figure 3-12a but different scale)................. 39 3-13a CONTEMPT 4 Pressure Response for Limerick Using BNL-NUREG-31940 Transient............................................ 40 3-13b (Same as Figure 3-13a but different scale)................. 41 3-14 CONTEMPT 4 Pressure Response Without Pool Surface Conden-sation. Same MSRV DLB Steam Source as Used in Reference -

1. Peach Bottom........................................... 42 3-15 CONTEMPT 4 Pressure Response Without Pool Surface Conden-sation. Same MSRV DLB Steam Source as Used in Reference

1. Limerick............................................... 43 l

NUREG/CR-4594 viii June 1986

LIST OF FIGURES (Continued)

Figure Title Page 3-16 CONTEMPT 4 Pressure Response Without Pool Surface Conden-sation. Envelope of BWROG Steam Sources for MSRV DLB.

Peach Bottom............................................... 44 3-17 CONTEMPT 4 Pressure Response With Pool Surface Condensa-tion. Envelope of BWROG Steam Sources for MSRV DLB.

Peach Bottom............................................... 45 3-18 CONTEMPT 4 Pressure Response Without Pool Surface Conden-sation. Envelope of BWROG Steam Sources for MSRV DLB.

L i me r i c k . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 3-19 CONTEMPT 4 Pressure Response With Pool Surface Condensa-tion. Envelope of BWROG Steam Sources for MSRV DLB.

Limerick................................................... 47

! 3-20 CONTEMPT 4 Pressure Response Without Pool Surface Conden-sation for HPCI Turbine Exhaust Line Failure. Li me ri ck . . . . 48 3-21 CONTEMPT 4 Pressure Response Without Pool Surface Conden-sation. Steam Condensing Mode Transient. Li me ri ck . . . . . . . . 49

, 3-22 CONTEMPT 4 Pressure Response With Pool Surface Condensa-l tion. Steam Condensing Mode Transient. Limerick.......... 50 B-la Normal Stress and Strength Distributions Showing Overlap... B-10 B-lb Di f f e rence Di st ri but i on . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-10 B-2 Pipe Failure Rates Estimated by the Stress Strength Over-lap Method................................................. B-11 D-1 CONTEMPT-LT/028 Input for Limeri ck Contai nment . . . . . . . . . . . . . D-5 0-2 CONTEMPT 4/M004 Input for Limerick Containment . . . . . . . . . . . . . D-6 D-3 CONTEMPT-LT/028 Heat Structure Initial Temperatures........ D-7 i

D-4 CONTEMPT 4/M004 Heat Structure Initial Temperatures......... D-8 1

l l

l i

NUREG/CR-4594 ix June 1986

l ACKNOWLEDGMENTS I

l This work was performed for the Reactor Safety Issues Branch of the Divi-( sion of Safety Review and Oversight, NRR/NRC. The considerable input and dir-l cction provided by A.W. Serkiz, the NRC manager for this program, have been of

, great benefit. The authors are also grateful for the helpful comments re-ceived from M.W. Hodges, G.R. Mazetis, and W.C. Milstead of the NRC, as well as T.B. Powers of PNL, and N. Hanan of ANL.

l Several discussions with other DNE staff members at BNL have also bene-l fitted this work. In particular the authors wish to thank W.T. Pratt, i C.C. Lin, R.A. Bari and W.J. Luckas for their comments regarding various sec-l tions of this report.

l l The authors would also like to express their appreciation to Ms.

l S. Flippen for her excellent typing and accurate assembly of this document.

l We are also grateful to Ms. E. Mitchell for her typing of sections of this report.

1 NUREG/CR-4594 xi June 1986 i

[

f LIST OF ACRONYMS ADS Automatic Depressurization System

! SCL Batelle Columbus Laboratory l BNL Brookhaven National Laboratory BWR Boiling Water Reactor

, BWROG BWR Owners Group CR0 Control Rod Drive l CSS ~ Containment Spray Systems

! DLB Discharge Line Break l ECCS Emergency Core Cooling Systems l EPG Emergency Procedure Guidelines FSAR Final Safety Analysis Report GE General Electric

! HPCI High Pressure Coolant Inection l LPCI Low Pressure Coolant Injection l LPCS Low Pressure Core Spray j MSIV Main Steam Isolation Valve l MSRV Main Steam Relief Valves l NRC Nuclear Regulatory Commission l PMD Probabilistic Mechanical Design PRA Probabilistic Risk Assessment PUAR. Plant Unique Analysis Report i PWR Pressurized Water Reactor RCIC Reactor Core Isolation Cooling

! RHR Residual Heat Removal RPV Reactor Pressure Vessel RY(Rx-yr) Reactor Year SDL Steam Discharge Line SLC Standby Liquid Control SNL Sandia National Laboratory SORV Stuck Open Relief Valve SRV Safety Relief Valve SRVDL Safety Relief Valve Discharge Line l

l i

xiii June 1986 l

NUREG/CR-4594

1.0 EXECUTIVE

SUMMARY

Brookhaven National Laboratory (BNL) has conducted a study which quanti-fies the potential threat to containment integrity of steam bypass posed by BWR systems which route high pressure steam through the wetwell airspace in response to anticipated transients. This study was co,npleted for the U.S.

Nuclear Regulatory Commission (NRC) in support of Generic Safety Issue 61,

" Steam Discharge Line Break in Wetwell Airspace of Mark I and Mark II Contain-ments". This issue was identified as a potential generic safety issue at the April 29, 1982 meeting of the ACRS Subcommittee on Hydrodynamics and was for-mally proposed as a generic safety issue by NRC's Generic Issues Branch on August 20, 1982.

The present study postulates a potential steam discharge line break in the wetwell airspace coupled with an active associated steam source such as a stuck open relief valve (SORV). Also included is the assumption that the steam source (e.g. S0RV) remains active for a significant period of time, thereby allowing steam to bypass the pool and could result in containment overpressure failure in a short time (15-20 minutes). Failure to condense the steam, resulting in loss of containment integrity, could lead to loss of reactor coolant injection capability and possible core damage. Thus, the sequence of events postulated has the potential for core melt and release of fission products into a failed containment.

Three BWR plant systems were identified which have components whose fail-ure could lead to a steam discharge into the wetwell airspace and pressuriza-tion of the containment: the Main Steam Relief Valves (MSRVs) and associated discharge lines, the High Pressure Coolant Injection (HPCI) turbine exhaust, and the RHR steam condensing relief lines. This investigation outlines the postulated accident scenarios involving these systems, estimates their fre-quency of occurrence and calculates estimated consequences.

Brookhaven National Laboratory evaluated the potential for the MSRV acci-dent sequence and associated consequences in 1982 (BNL-NUREG-31940) and con-cluded that the potential for occurrence was relatively high and that contain-ment failure consequences could be potentially severe. This initial BNL report was subjected to NRC staff review and also review by the BWR Owners Group (BWR0G), who felt many of the assumptions used in the analyses were overly conservative. The decision was made to reevaluate this generic issue based on comments received, as well as more recent severe accident studies and containment overpressure survivability studies. This report contains the re-suits of this reevaluation.

The current investigation also looks at the potential threat posed by ac-cident sequences involving the other two BWR systems, HPCI and RHR Steam Con-densing, mentioned above. The significant findings can be summarized as fol-lows:

NUREG/CR-4594 1 June 1986

Estimated Core Melt Estimated Risk Accident Sequence Frequency (1/Rx-Yr) (Person-rem /RY)

MSRV Line Break 2.5E-7 5.0 HPCI Line Break <1E-8 0.16 Steam Condensing Mode Line Break <1E-8 0.16 MSRV Line Break refers to a stuck open MSRV whose associated discharge line has ruptured in the wetwell airspace. HPCI Line Break designates a se-quence where the HPCI turbine exhaust line has broken in the wetwell airspace and HPCI operation causes steam to escape into the airspace. Steam Condensing Mode Line Break refers to a sequence where a steam condensing valve failure during decay heat removal via steam condensation in the RHR heat exchanger causes a relief valve to divert steam to the suppression pool and this dis-charge pipe has failed somewhere in the airspace. These three postulated pool bypass sequences are judged to be those most likely to result in rapid con-tainment overpressurization, i.e. less than 30 minutes.

The conclusion which can be drawn from the estimated core melt frequen-cies and potential releases is that the safety significance of this safety is-sue is lower than previously estimated. The principal reasons for this reduc-tion in safety significance are:

1) The conditional probability of core melt, given containment failure, was reestimated from the previous value of 1.0 to a value of less than 0.1.

2) Estimates of the probability of MSRV Discharge Line failure are now calculated as 4.2E-5 per transient for the upper bound based on oper-ating experience and 1.E-10 per transient using PMD methods.

The sections which follow discuss prior work and comments received, the safety issue accident scenarios, estimated probabilities of occurrence, con-tainment survivability, possible mitigating actions and estimated conse-quences. Appendix A summarizes comments received and action taken relative to those comments. Appendices B, C and D contain development of analytical mod-els to predict pipe break and releases, and illustrative calculations.

NUREG/CR-4594 2 June 1986

2.0 BACKGROUND

AND PRIOR WORK

SUMMARY

I As noted above, this safety concern was designated Generic Safety Issue 61 in 1982 and evaluations were undertaken by BNL; the results were reported by Economos et al. in BNL-NUREG-31940,1 October 1982. That investigation used a limited Probabilistic Risk Assessment (PRA) methodology to develop an acci-dent sequence, evaluate the frequency and assess the consequences of a postu-lated MSRV line break in the wetwell airspace of Mark I and Mark II contain-ments. No actual consequence calculations were performed. The severity of the accident seguences were characterized in terms of a " release category" using WASH-1400 terminology. Only the most dominant sequences that could be associated with MSRV pipe failure were considered. A Mark I plant, Peach Bottom, and a Mark II, Limerick, for which PRAs were readily available were chosen for evaluation so that the accident sequences developed involving a MSRV pipe break could be compared to other sequences previously evaluated in terms of overall risk.

Containment overpressurization (due to the postulated steam bypass) was calculated using the containment assessment code CONTEMPT-LT/028.3 Estimates of mass flow and energy input as a function of time, provided by the NRC, were utilized for the hypothesized stuck open MSRV and the pressure transients calculated in Reference 1 are shown in Figure 2-1. Containment design pres-sures were calculated to be exceeded in 4 minutes. Containment rupture pres-sures (150-175 psia) were calculated to be reached in about 10 minutes Ide-pending on the steam bypass fraction). The conclusion was reached that con-tainment failure was likely to occur due to the short time available for taking mitigating action.

The event tree developed by Economosi is shown in Figur3 2-2. The se-quences TPD, TPDW and TPDZ were judged to be the most likely sequences to lead to containment failure. These sequences deal with the following scenarios.

The initiating event for all three sequences was assumed to be a turbine trip (with bypass) (the event T), generally considered to be the transient with the highest frequency of occurrence. It was assumed that the reactor scrammed successfully and that the MSRVs actuated normally. Economos postu-i lated that as the reactor pressure was reduced, one of the MSRVs would fail to reseat (the event P). It was further assumed that the discharge line associ-ated with this valve had ruptured and allowed steam to bypass to the wetwell airspace (the event D). For the TPD sequence it was then assumed that the Containment Spray System (CSS) was actuated before containment rupture by overpressure and that coolant injection and heat removal functions operated successfully to bring the reactor to a cold shutdown. This scenario or se-quence did not result in a degraded core condition. The frequency of the TPD sequence was calculated as 1.0E-6.

The TPDW Sequence proceeded exactly as the TPD sequence described above but the additional failure to remove the residual heat from the containment was postulated (the event W). This ultimately led to pressures which were sufficiently high to rupture the containment. Rupture was considered to lead, with probability near unity, to a core meltdown because of failure to maintain coolant inventory. The frequency of this sequence was calculated as 1.5E-9.

While this was a relatively low frequency, Economos t indicated that this NUREG/CR-4594 3 June 1986

accident would lead to core melt with large fission product release, some of which would bypass any scrubbing action of the suppression pool.

In the TPDZ Sequence, it was postulated that the CSS was not actuated before the containment ruptures (the event Z*). The containment failure was assumed to cause loss of injection which would lead to degraded core condi-tions. This sequence was found to have the relatively high frequency of 1.0E-6. Because containment rupture occurred only ten minutes after reactor scram for this sequence, consequences were expected to be more severe than any of the sequences involving the W event, since, in those cases, containment failure did not occur until 20-25 hours after scram. The severity was further compounded by fission product bypass of the suppression pool. This sequence was expected to contribute significantly to risk.

Reference 1 concluded that the accident sequences studied, involving failure of a MSRV to reseat combined with the rupture of the associated SRVDL, occurred at frequencies comparable to those of many sequences considered to be significant contributors to risk.4 Since containment failure was assumed to lead to core melt, it was concluded that the consequences of the postulated sequences were potentially severe. The study also pointed out three major areas of uncertainty associated with quantification of the examined accident sequences: a) failure rates assigned to MSRVs and SDLs, b) operator reliabil-ity, and c) the effect of exceeding design pressure and temperature on safety systems. While the first of these could be refined by additional work, it was felt the other uncertainties involved too many plant unique features to be treated successfully on a generic basis.

In a response to an NRC Request for Information relating to Generic Issue 61, the BWROG5 commented on the analysis and findings in BNL-NUREG-31940.1 Basically, the thrust of the comments were that the BNL study was too conser-vative for several reasons. First, the mass and energy releases used in the study were too high. Second, the overall event frequency per year for these sequences would become so small as to remove them as significant contributors to risk if less conservative probabilities were chosen for event elements making up the sequences. The most significant differences between the BWR0G estimates of event elements and those of NUREG/CR-31940 are shown in the fol-lowing table. For completeness the revised estimates calculated in this re-port are also listed.

• Note that the symbol Z, chosen to represent failure of the CSS, is not a con-ventional notation.

NUREG/CR-4594 4 June 1986

BNL-NUREG-31940 1 BWROG5 Current Report Estimate Estimate Estimate Frequency of.a Stuck Open MSRV Per Demand 6.7E-3 5.6E-4 2.5E-2/L (L - No. of valves actuated)

Frequency of a Broken SDL Per Demand 7.4E-5 1.E-10 4.2E-5/L (upper bound) (upper bound) 7.5E-7 1.0E-10/L (PMD method) (PMDmethod)

Conditional Probability of Core tielt Given Containment i Failure 1.0 .01 <0.1 The BWR0G also provided steam release rates for six SORV scenarios which they felt were more realistic yet still conservative. These steam release rates were later used by BNL for containment pressure response calculations involving stuck open MSRVs (see Section 3.3). Even under worst case assump-tions, these BWR0G release rates extended the time to containment failure by 30 to 50% beyond that of the BNL-NUREG-31940 calculations.

Appendix A provides a comprehensive review of BWROG comments received, action taken and where such information can be found in this report. As an example, the BWROG mass-energy release data was used to recompute containment pressure rise.

NUREG/CR-4594 5 June 1986

I _

I I K

K R R A A M M _

0 4

g _

E R

U _

S S .

/ E R

1 _

0 5

P g S D

e c

E N N n R

U G

I O

C e

r _

S S E e S

E R

P\

E R

\

5

/ / E D

5 ly g

S 0

0 0

1 f

R i

e n

7 U d T e P

U ' s .

R 0 U 1 # y e N

O I

T

/

/ t r

t R

L u

p u

C n D A e V R h v R

F E S

S g S n +

A i P

Y B

/ t i

i t

a n

V R

O S

o I

t m e o s r n F o i 0 p 0 e s 1

i m e

- T R t

n e

m n

i a

t n

o C

1 2

e r

u g

i F

- - - - - 0 1

0 0 0 0 0 0 0 6 2 8 4 2 1 1 2Go.]5 Ec. 5i cl GR 2EEQb%

E A

R n

7 Make Limit Maintain Maintain Remove M Event or Initiating Reactor RCPB RCPB Limit Containment Coolant Decay

$ Function Event Subcritical Pressure Integrity Pressure Inventory Heat Sequence' Frequency Type of FW Designator (a)

Event or . Anticipated CRD/SLT .

SRV SRV VSS CSS HPCI PCS System Transient (0 pen) (Close) LPCI RHR Symbol T C M P D I Q/U/V W Point of departure of this study from earlier work i

I

" e t

I P TPD T*P*D(1-Z) 1

/ W 8

I

' )

T TPDW T*P*D*(1-Z)*W 1-Z '

8 I D

i Z TPDZ T*P*D*l E'

$ (a)The symbols T, P, etc. are used to designate both the event and its frequency.

G Figure 2-2. Simplified Treisient Event Tree Used in Reference 1.

3.0 DESCRIPTION

OF SAFETY ISSUE The most significant system in terms of potential mass and energy re-leases is the reactor overpressure protection system consisting of the main steam relief valves (MSRVs) and associated discharge lines (SRVDLs). The MSRVs, which are mounted on the main steam line in the drywell, are opened during certain operational conditions to allow steam to escape from the reac-tar vessel and divert it into the suppression pool. The SRVDLs run through the wetwell airspace before entering the pool . If one of these SRVs fails to reclose after being activated and the associated discharge line has ruptured somewhere along its run in the wetwell airspace, most of the escaping high pressure steam will not be condensed and the containment pressure will in-crease. If this situation remains uncontrolled, the potential exists for eventual containment overpressure failure. Such a containment failure may severely impair the Emergency Core Cooling Systems (ECCS) and possibly lead (under extreme circumstances) to core melt and the potential for significant radioactive releases. Figures 3-1 and 3-2 illustrate where potential SRVDL breaks leading to bypass of the suppression pool could occur in Mark I and II type containments.

Another BWR system (see Figure 3-3) with components routing stean through the wetwell airspace is the High Pressure Coolant Injection System (HPCI).

The HPCI turbine, driven by high pressure steam tapped off the main steam lines, has its exhaust line routed to the suppression pool. A break of this line in the wetwell airspace could lead to containment pressurization during HPCI operation in a similar manner to that for the MSRVs mentioned above.

The Reactor Core Isolation Cooling (RCIC) System Turbine also exhausts steam into the suppression pool when in operation (Figure 3-4). However, the RCIC turbine exhsust can be dismissed as a threat to containment integrity be-cause of the relatively low flow rates. The maximun capacity of this line in a typical BWR (Limerick) is less than 33,000 lbs/hr. This is less than five percent of an MSRV flow and will at most lead to a gradual pressure rise over several hours with ample time for mitigative measures (containment spray, venting, or RCIC isolation). Therefore, development of steam bypass accident sequences involving the RCIC exhaust lines is not necessary.

Finally, some BWRs have the capability of operating their Residual Heat Removal System (RHR) in the Reactor Steam Condensing Mode (see Fig. 3-5).

During this mode, should a steam condensing pressure control valve failure oc-cur, a relief valve diverts steam to the suppression pool via piping in the wetwell airspace. If a failure of this pipe occurred in the wetwell airspace while the plant was in the Steam Condensing Mode containment pressure could rise with the potential of containment failure and possible core melt.

A comment regarding the vacuum breakers on any one of these systems should be added since a failed open vacuum breaker also provides a potential path for steam bypass, and has a greater probability of occurrence than a pipe ruptu re. Vacuum breakers on the MSRV Discharge Lines are located in the dry-well of the containments and so cannot pressurize the wetwell airspace. Vac-uum breakers on the HPCI exhaust line are typically one-third or less the di-ameter of the exhaust line. Then flow through any stuck open vacuum breaker would at most be one-ninth that of the exhaust line - not nearly enough to significantly impair containment integrity in a short time. Similarly, vacuum NUREG/CR-4594 8 June 1986

breakers on the RHR Steam Condensing Mode pressure relief line have relatively small flow rates. Maximum flow through this vacuum breaker for Limerick is given as 7,200 lbs/hr. Therefore accident sequences involving failed open vacuum breakers were not considered further in this study.

Thus, the safety significance of this issue becomes dependent on:

1) The probability of a stuck-open safety relief valve (SRV).

2) The frequency with which the MSRVs, the HPCI and the RHR (in Steam Condensing Mode) reactor systems are activated.

3) The probability that a break will occur in the wetwell airspace sec-tion of the associated discharge line.

4) The containment servival during containment pressurization due to by-pass of the suppression pool .

5) The availability of mitigation systems such as containment spray or wetwell venting which could prevent containment overpressure failure.

6) The survival of ECCS and other coolant injection systems (e.g., ser-vice water) after containment failure.

i These considerations are discussed in detail in the sections which follow.

~

3.1 Accident Sequences Selected for Analysis Since this safety issue is driven by a discharge line break in the wet-well airspace, those reactor systems whose actuation by anticipated transients could result in significant steam discharge to the suppression pool via pipes passing through the wetwell airspace become prime candidates for considera-tion. A review of such systems was carried out utilizing the Boiling Water Reactors System Manual used by th several Mark I Plant UniqueAnalysesReports(PUAR's),9NRCTrainingCenter,6 10 Mark I and Mark II Final Safety Analyses Reports (FSAR's) 11,12 as well as pertinent General Electric 13,27 and utility documents.5.14.15 Anticipated transients involving the MSRVs are the most obvious sources of concern since these valves (in both Mark I and II containments) have maxi-mum steam flow design capacities of 900,C00 lb/hr and the steam discharge lines (SDLs) have extensive pipe lengths with numerous welds within the wet-well airspace.

The accident previously analyzedl dealt with an SRVDL break. This report also analyzes the effect of line breaks in the HPCI, and RHR (steam condensing mode) systems. The reason for extending the evaluations to these systems is that their steam discharge rates, while less than that for an SRVDL break, are still significant and may lead to potential overpressurization of the contain-ment.

The accident sequences involving the MSRV Discharge Lines were discussed extensively in Reference 1, and still remain the sequences of primary inter-est. This postulated scenario is as follows:

NUREG/CR-4594 9 June 1986

a) A turbine trip with steam bypass (the event T) occurs.

b) The reactor is scrammed and MSRV actuation follows in a normal fash-ion to limit reactor pressure.

c) When reactor pressure is reduced below SRV setpoints, one SRV fails to reseat (event P),

d) The corresponding SOL ruptures or has ruptured previously (the event D) in the wetwell airspace and steam bypass commences to pressurize the containment.

A simplified event tree for a MSRV discharge line break is shown in Fig-ure 3-6 and is the basis for the subsequent analysis regarding MSRV discharge line breaks.

The assumed events for the HPCI system sequence consists of these steps:

a) The HPCI turbine is activated as anticipated by an appropriate initi-ating event.

b) The turbine exhaust line has ruptured in the wetwell airspace allow-ing steam to escape.

c) The automatic turbine shut-off triggered by high exhaust pressure fails and steam continues to bypass the pool and pressurize the con-tainment.

A simplified event tree for this sequence is shown in Figure 3-7.

The third postulated scenario, involving the RHR system in the Steam Con-densing Mode can be described with the following events:

a) An appropriate initiator causes the operator to configure the RHR system in the Steam Condensing Mode.

b) The pressure control valve on the line routing steam to the heat ex-changer fails, causing the safety relief valve to open. (Alterna-tively, the safety relief valve opens spuriously, i.e. fails open).

c) The discharge line routing the excess steam to the suppression pool

, has ruptured in the wetwell airspace allowing steam to escape and i pressurize the containment.

Figure 3-8 shows a simplified event tree for this sequence of events.

3.2 Estimated Frequency of Occurrence As discussed in Section 3.1, three sequences were selected to assess the significance of this safety issue. For convenience, these sequences will be referred to as Sequence No.1, 2 and 3 in what follows, where:

l i

i i

NUREG/CR-4594 10 June 1986

1) Sequence No. I designates a turbine trip event followed by MSRV failure to close, an SRVDL break and ultimately con .

tainment failure.

2) Sequence No. 2 designates an HPCI turbine exhaust line break which also has the potential to overpressurize the contain-ment.

3) Sequence No.-3 designates a relief line break when the RHR system is being used in the steam condensing mode which can also lead to containment overpressurization.

Sections 3.2.1, 3.2.2 and 3.2.3 provide estimates of the frequency of events involved in Sequence 1, 2 and 3, respectively. The events are the ones shown in the event trees of Figures 3-6 through 3-8.

The three sequences and the estimated frequencies of occurrence are sum-marized in Tables 3.5-1, 3.5-2 and 3.5-3.

3.2.1 Sequence No.1 - MSRV Discharge Line Break This is the principal sequence of interest s.ince turbine trips and MSRV actuations are relatively routine occurrences and the mass and energy releases

-involved are large. As discussed in Section 2.0 and shown in Fig. 2-2, from the point of discharge line rupture and containment pressurization, Economos et al .1 developed three separate sequences depending on operator actions: The TPD sequence previously judged to cause containment design pressure exceedance but not rupture estimated in Reference 1 at 1.0E-6. The TPDW sequence which led to containment failure and core melt but whose probability as estimated by Economos t was relatively low (1.54E-9), and the TPDZ sequence also leading to containment failure and core melt but with an estimated frequency of 1.0E-6 ir.

Reference 1.

The emphasis in the study of postulated nuclear reactor accidents has in-creasingly been placed on sequences leading to core melt with a probability greater than 1.0E-7. From this perspective it becomes apparent that the TPDZ sequence, where mitigating operator action is not taken in time to prevent rupture pressure from being reached, is of primary concern. Since the TPD se-quence did not lead to rupture and the TPDW sequence was conservatively esti-mated at a frequency of 1.5E-9, neither of them appear to be a significant contributor to risk. Therefore, the TPDZ sequence will be the only one con-

-si6ered in this study. Figure 3-6 shows a simplified event tree development of the TPDZ sequence. This figure differs from Fig. 2-2 in that an extra-column has been added for the probability of core melt given containment rup-ture.

A-brief sumary of the event frequencies for Sequence 1 now follows.

.3.2.1.1 Estimated Frequency of Turbine Trip An estimated turbine trip frequency of 4 (based on Reference 4) was uti-lized in BNL-NUREG-31940. Based on the values in Table 4.1 of NUREG/CR-4050,30"A Review of the Shoreham Nuclear Power Station PRA," we feel this is still a good best estimate value but that 8 trips /Rx-yr should be used as an

NUREG/CR-4594 11 June 1986

upper bound. (This frequency will also cover other events such as MSIV closure which could lead to MSRV actuation.) .

3.2.1.2 Estimated Failure of MSRV to Reseat Appendix B discusses the development of estimated MSRV failures to reseat and is based on an updated EG&G report.16.17 The later EG&G estimate 17 is 3.1E-3 per demand, with a 1.5 multiplier to obtain 95% confidence limits. The -

previous estimated failure frequency 16 was 4.5E-3. We feel there are poten-tial non-conservatisms in the data analysis of Refs.16 and 17. For example, the authors of Refs.16 and 17 assumed that for BWRs all sc ams reported and analyzed resulted in pressure transients that lifted all of the plants' safety-relief valves. A review of related LERs however indicated only a frac-tion of the plants' valves lifted.

The number of valves opened by a specific transient in a particular plant depends on the number of MSRVs in the plant, their individual set points and the severity of the overpressure transient. To avoid the non-conservative assumption in References 16 and 17 regarding number of valves lifting, the data in Reference 17 was used in the following way. Reference 17 reported 20 MSRV failures to reseat for 814 ' demands' on BWR plants during which tne MSRVs were actuated. In order to obtain a generic estimate of failure to reseat we assumed an ' average' number of valves lifted, L, per plant demand. Then the failure to reseat per (valve) demand is given by 20/814L and the failure per transient where L valves lift is 20/814 or 2.5E-2.

Although the BWROGs cites NED0-24951 13 as evidence of improved SRV de-sign, and therefore recommends reduction of the estimated failure rate, NED0-24951 discusses only in qualitative terms the possible reductions in future valve failures due to redesign and retrofitting. No new quantitative data is provided. Therefore, the present report utilizes a failure rate of:

PSRV = 2.5E-2/L per demand or, an estimated SRV failure rate of 2.5E-2 per transient.

3.2.1.3 Estimated SRV Line Break Probability Two methods were utilized in calculating the probability of an MSRV dis-charge line break. An estimate was developed using Probabilistic Mechanical Design (PMD) methods, and an upper bound based on operating experience was de-rived by assuming that, since no lines have ruptured to date, the next SRV line actuation would result in line rupture. Appendix B provides more details on these calculations.

The assumed line break could have occurred during a previous actuation or during the accident sequence actuation either before or after the failure of the valve to reseat. Since pipe break during a previous actuation is far more likely, no attempt has been made here to restrict the pipe break to the same actuation during which the valve fails to reseat.

NUREG/CR-4594 12 June 1986

To obtain an upper bound of MSRV line break probability based on operat-ing experience we proceeded in the following manner: As discussed in Appendix B, data from Reference 17 can be used to show that there have been approxi-mately 2400 events to date in BWRs which have caused MSRVs to lift with no known line breaks. If we assume the next actuation causes a MSRV line to rup-ture, the break probability is 1/2400L or 4.2x10-4/L per demand where L is the number of valves lifted. The probability per transient is L times the event frequency or 4.2x10-4 Since the broken pipe and stuck open valve must coin-cide in order for a steam bypass accident to occur, the break frequency for the accident sequence should be divided by P, the plant MSRV population.

Assuming a typical plant population of 10 valves the upper bound break proba-bility becomes 4.2x10-5/L per demand or 4.2x10-5 per transient.

As Appendix B shows, using the PMD method results in an estimated MSRV line break probability of 1.0E-12 to 1.0E-10 per transient.

3.2.1.4 Failure of Containment Spray Activation by Operator Prior analyses (Ref.1) postulated that due to the rapid rise in contain-ment pressure to rupture levels (i.e.10 minutes), the probability of failure to activate containment spray was estimated as 0.5. While operator training and the existence of specific procedures such as the Emergency Procedure Guidelines (see Section 3.4) are designed to assure successful spray opera-tion, the short time to containment overpressurization imposes very stressful conditions on the operator. Therefore successful spray activation was judged as likely to occur as not in Reference 1. Tne revised containment pressure rise calculations (see Section 3.3) even under the conservative assumption of no pool surface condensation show an increase in time of 30 to 50% before rup-ture pressures are reached. Thus, with conditions somewhat less stressful, a conservative upper limit estimate of failure to activate CSS can be reduced to 0.3. In reality, some pool surface condensation would take place, thereby providing considerably longer time for the operator to turn on containment sprays. Thus, the best estimate value for operator failure used in this re-port is 0.1, which was also the value estimated by the BWROG5 based on their operator Emergency Procedures Guidelines (EPGs).

3.2.1.5 Frequency of Core Damage Given Containment Failure The probability of core melt given containment failure was assumed to be 1.0 in the previous analyses.1 More recent studies related to containment failure and subsequent core cooling are discussed in Sections 3.3 and 3.4.

These reduce our estimate of core melt probability to less than 0.1 as a best estimate, with 0.1 as an upper bound. The BWROG5 advocates a probability of

.01 for core melt once the containment has failed.

3.2.1.6 Overall Frequency of the MSRV Steam Bypass Sequence The elements discussed in Sections 3.2.1.1 through 3.2.1.5 can be com-bined to yield an updated estimate of the likelihood that the TPDZ sequence of Reference 1 and this report will lead to core meltdown. If only the most con-servative estimates of the event element failures are chosen, the sequence frequency is calculated as:

8.0 x 2.5E-2 x 4.2E-5 x 0.3 x 0.1 = 2.5E-7/Rx-yr.

NUREG/CR-4594 13 June 1986

f.

3- Current "best". estimates give a value of 4.0 x 2.5E-2 x.1.E-10 x 0.1 x .<0.1 = <1.0E-13/Rx-yr.

d 3.2.2 Sequence No. 2 - HPCI Turbine Exhaust Line Break

The HPCI system is activated whenever Level 2 is reached. The frequency for such events can be approximated from NUREG/CR-4050 30 for the Shoreham Mark j 11 plant. We will conservatively use 2 per reactor year as the generic fre-quency of this sequence initiator. As discussed in Section 3.1, and shown in

z Figure 3-7, rupture of'the turbine exhaust is needed for steam bypass to occur. .An estimate of.the failure probability of this line can be obtained from the data presented in Appendix C of Reference 28. -There the failure

, probability per reactor year for nuclear piping 6 to 10 inches in diameter is cited as 4x10-5 while piping 10 to 16 inches is assigned a failure probability of 3x10-5 .The failure postulated for our case consists of a total rupture, not just a leak,'of a pipe line only pressurized a small portion of the time and only a' part of which runs through the wetwell airspace. Based on these considerations and the failure rate data cited in Reference 28, a failure on demand consisting of a rupture on this line in the critial location of the wetwell airspace during HPCI operation can be conservatively estimated as p

having a probability of 4.0E-5. To continue containment pressurization the automatic turbine shutdown due to high exhaust pressure must fail. While no failure probabilities of this specific event were found the following reason-1 ing was used to assign 1.0E-2 per dema'nd as a conservative value to this fail-ure. Data in Reference 4 shows that, typically, instrumentation failure, such

+

as a containment pressure sensor f ailing to operate, is of the order of 1.0E-3 or 1.0E-4 per deman.d. Allowing an error margin of an order of magnitude leads to a conservative estimate of 1.0E-2 per demand. Additional supporting data

- for this estimate is provided in Reference _29 which cites 1.0E-2 per year as the probability of inadvertent start up of HPCI. The instrumentation and

. mechanics of HPCI startup is deemed to be at least as complex as-that of shut-i down due to high exhaust pressure, so a demand value of 1.0E-2 for failure to

shut down appears reasonable but conservative. Next, since aslthe result pre-sented in Section 3.3.1 for this sequence shows, pressures rise relatively i 3 ._ slowly, the operator has ample time to take mitigating actions such as turning

.on the containment spray. Therefore the probability.of operator failure to

, activate spray or take other heat removal action can be estimated at 2.0E-2 as

a "best" estimate and at.0.1 as-a conservative upper bound. Finally, if core melt'given containment failure is again estimated at <0.1 (See 3.2.1.5) then

. the overall 'best' estimate frequency for the HPCI tu'Fbine exhaust bypass fre-l quency can be calculated as:

I:

' 2.0 x 4.0E-5 x 1.0E-2 x 2.0E-2 x <0.1 = <1.6E-9/Rx-yr ,

with- 8.0E-9/Rx-yr as an upper bound.

3.2.3 Sequence No. 3 - RHR Steam Condensing Mode Relief Line Break For those Mark I and Mark 11 plants which can utilize the Steam Condens-i ing Mode (Peach Bottom does not), an estimate of the frequency of using this mode is approximately 1.0 per reactor year with 2.0 per reactor year as an upper bound. Once the steam condensing mode is used, either a pressure con-trol or relief valve failure mue now be postulated to proceed with this NUREG/CR-4594 14 June 1986

~

J

. w . er . _n_ . . - _ . . . . . . - , ._.

.---_--__e -- .m.__ _ _ _ _ . . . . u.- . .y.._-~4--

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

sequence. Based on data in Reference 17 an estimate of 1.0E-2 per demand is a conservative value for such a failure. Valve failure must be followed by an SDL rupture, which can be conservatively estimated at 4.0E-5 per demand based on the same data and considerations as discussed for the HPCI turbine exhaust line previously. Failure to get the containment spray on in time is conser-vatively estimated at 0.1 based on the pressure rise cal;ulations in Section 3.3. With the probability of core melt, given containment failure, again at

<0.1 the overall sequence frequency becomes:

1.0 x 0.01 x 4.0E-5 x 0.1 x <0.1 = <4.0E-9/Rx-yr ,

with 8.0E-9/Rx-yr as an upper bound.

3.3 Containment Survival Containment survival is dependent on two factors: a) the ultimate pres-sure level the structure can sustain before rupturing and b) operator inter-vention (e.g. turning on containment sprays to condense steam from a pipe break). This latter aspect, operator mitigating actions, is discussed in Sec-tion 3.4. The sections which follow deal with recalculation of containment pressure rise based on mass and energy source terms received from the BWR0G.s 3.3.1 Mass and Energy Source Terms s

As mentioned in Section 2 of this report, one of the BWR0G comments re-garding BNL-NUREG-31940 was that the mass and energy releases used by BNL in the prior studyl were too high. Subsequently, at the request of the NRC, the BWR Owners' Group provided mass and energy releases appropriate for transients involving the actuation of one or more MSRV's. Figure 3-9 shows the six tran-sients (plus the original estimate used in BNL-NUREG-31940) which the Owners provided as being representative of "a number of cases for various BWR Product Lines for Mark I and Mark 11 containment types."5 Table 3.3-1 shows the case description that the BWROC provided; Table 3.3-2 provides mass and enthalpy values versus time for the transients of Figure 3-9. As can be seen from Fig-ure 3-9 all six transients are substantially less than the one used by Economos l with some at only half the mass flow of the original. For the present report, calculations involving MSRV discharges were done with an im-proved version of the containment performance code, CONTEMPT 4,18 for all six transients provided.

For sequences involving HPCI turbine exhaust and Steam Condensing Mode relief valves, constant flow rates corresponding to the maximum capacities of the involved SDL's were chosen as a bounding calculation.

3.3.2 Limitations of CONTEMPT Calculations When containment response calculations with CONTEMPT 418 using the BWR0G supplied transients were made for Peach Bottom and Limerick, pressure results as shown in Figures 3-10 and 3-11 were obtained. When compared with the Reference 1 results, using the original transient and CONTEMPT-LT/28,3 shown in Figure 2-1, not only were pressures much lower but the pressure curves  ;

eventually " turned over," unlike the steadily increasing pressures of Figure 2-1. To ascertain if these dif ferences were solely due to the reduced mass and energy releases of the new transients, the original transient of Reference NUREG/CR-4594 15 June 1986

Table 3.3-1. Cases for Mass and Energy Data CURVE CASE DESCRIPTION A BWR 4, MK I One SORV with reactor at power B BWR 4, MK I Similar to Case A - slightly different scram time C BWR 4. MK II HPCI shut off when vessel water level reaches 45 ft.

NOTE: SRV flow rates vs. time are altered if ADS is manually actuated. The flow rates are substantially reduced upon actuation of ADS.

D BWR 3, MK I One SRV fails open and cannot be closed E MK II One SORV with isolation at power NEDO NED0-24708A (pg 4-642)

BWR 4 MK I Loss of feedwater flow & SORV, HPCI/RCIC on - at 97 Sec one 50RV 1

BNL BNL-NUREG-31940 SORV following a turbine trip with about 30% bypass to the main condenser 4

NUREG/CR-4594 16 June 1986

Table 3.3-2. Mass-Energy Release Data 1

1 I

NEDO A B 27 I C D 24708A E BNL MASS ENTHALP7 w h 'w h' "w h' w w w TIME (w) (h)

(SEC) LBM/SEC BTU /LBM 0 225 252 10 196 205 207 200 232 20 189 199 201 200 234 30 187 198 200 197 40 184 1194 199 1193 199 1193 188 50 181 201 197 188 234 60 176 201 195 186 70 172 1196 201 1193 191 1195 184 1196 187 214 80 167 200 189 181 90 163 199 186 176 100 158 198 183 174 173 184 252 200 111 1204 168 1199 155 1200 149 1201 130 121 221 300 77 123 1202 129 1203 127 1202 98 82 193 400 58 1203 87 110 1204 113 1204 75 68 172 500 58 79 113 103 1205 75 157 600 58 82 115 94 81 147 700 58 84 1205 116 1204 86 137 800 58 86 116 81 128 900 57 1203 87 116 74 1205 90 120 NUREG/CR-4594 17 June 1986

1 was calculated using CONTEMPT 4. Figures 3-12a,b show the result for Peach Bottom, Figures 3-13a,b for Limerick.* It is apparent that more than the re-duced transients were responsible for the differences in calculation. Compar-isons were made between CONTEMPT-LT/28 used in BNL-NUREG-31940 and CONTEMPT 4 used for the current study. Several significant differences were found and these are discussed in greater detail in Appendix D. However, by far the most significant difference was in the amount of steam Londensing on the suppres-sion pool surface as calculated by the two codes. As the transient pro-gressed, and the wetwell atmosphere became almost entirely made up of steam, CONTEMPT 4 predicted a large amount of this steam would be condensed on the pool surface while CONTEMPT-LT/28 predicted very little condensation. The significant condensation with CONTEMPT 4 led to lower containment pressure, ex-plaining the differences in pressure prediction by the two codes. Further in-vestigation showed that the heat and mass transfer coefficients for pool sur-face condensation built into the version of CONTEMPT-LT/28 used for the pre-vious study were extremely conservative to the point of almost eliminating pool condensation. Corresponding heat transfer coefficients in CONTEMPT 4 were several orders of magnitude larger, and appear to be more realistic under nor-mal pool conditions. It is easily shown that the difference in pool surface condensation is indeed responsible for the difference in pressure prediction by the two codes by exercising the user option in CONTEMPT 4 which turns off the suppression pool surface condensation model. As Figures 3-14 and 3-15 show, the CONTEMPT 4 pressure predictions are slightly more conservative with no pool surface condensation than the old CONTEMPT-LT/28 predictions (since some limited condensation still took place in these old LT/28 calculations).

However, this does not mean that one should jump to the conclusion that the large pool surface condensation predicted by CONTEMPT 4 for the steam bypass sequences is phenomenologically correct. The problem is in the " single node" nature of the CONTEMPT codes which model all the water in the suppression pool as being at the average pool temperature without accounting for any of the thermal stratification which would in reality take place. For the steam by-pass sequences we are considering here, the pool remains quiescent and so any condensate will form a hot layer at the pool surface. This hot layer will in reality significantly reduce subsequent condensation from what would be pre-dicted if the condensate were evenly distributed throughout a pool with one average temperature as modeled by the CONTEMPT codes.

One of the significant findings of this study is that the CONTEMPT codes in their present form provide only a very limited tool for investigating con-tainment response to the accident sequences with steam release in the wetwell ai rspace. Predictions for the same event can vary drastically between a rapid pressure rise leading to relatively quick containment rupture if no pool sur-face condensation is assumed and a slower pressure rise terminating below the failure pressure if the well-mixed single node pool model is used with best-estimate surface condensation correlations.

• Figures 3-12a,b (Figs. 3-13a,b) show the same data. Figure 3-12a uses a log scale time axis for ease of comparison with Fig. 2-1 which came from Ref.1.

Figure 3-12b uses a linear scale for comparison with other plots in the cur-rent report.

NUREG/CR-4594 18 June 1986

3.3.3 Results of CONTEMPT 4 Calculations To bracket the response to the six BWR0G supplied transients, an envelope of these transients was constructed and used as input in CONTEMPT 4 calcula-tions for both Limerick and Peach Bottom. For each plant, one calculation was done with pool surface condensation turned off and another with surface con-densation in place. Figures 3-16 and 3-17 show the results for Peach Bottom, while Figures 3-18 and 3-19 show results for Limerick. As Figures 3-16 and 3-18 show, without surface condensation, containment failure pressures are reached relatively rapidly. Using the conservative calculations, the new transients add approximately five minutes to the time before containment fail-ure is reached for both plants. In the calculations with pool condensation (Figures 3-17 and 3-19), the containment failure pressures are never reached.

To calculate containment response to a postulated HPCI turbine exhaust pipe failure and steam bypass, a constant source term with maximum exhaust flow rates of 235,000 lbs/hr was used for the Limerick plant.12 (Peach Bottom HPCI flow rates are lower and containment failure pressure is higher.) The pressure result of a very conservative calculation with no pool surface con-densation is shown in Figure 3-20. As the figure shows, even under the con-servative assumption of constant maximum flow and no pool surface condensa-tion, design pressure is not exceeded until 900 seconds (15 min.) and even after 2,000 seconds containment rupture pressure has not been exceeded. Pool surface condensation would lower pressures still further. Therefore, unlike MSRV discharge line failures, the HPCI turbine exhaust does not provide a large enough steam source to seriously threaten containment integrity in a short time span.

The Limerick containment response to the Steam Condensing Mode transient is shown in Figures 3-21 and 3-22. Results of Figure 3-21 are with no pool surface condensation while 3-22 shows pressures with pool surface condensa-tion. The source for both cases was conservatively chosen as a constant 400,000 lbs/hr, the maximum flow rate of the SDL involved. With no condensa-tion, rupture pressures are reached in about 1,200 seconds (20 minutes) while rupture pressure is barely reached in 2,000 seconds with pool surface conden-sation allowed.

One other comment regarding pool surface condensation should be made: In their remarks on BNL-NUREG-39140,1 the BWROGs stated that data from the 1962 Bodega Bay tests showed complete steam condensation occurring for the case of steam being blown through downcomers which ended two feet above the pool sur-face. The Owners Group argue that this means condensation on the pool surface should be very effective for the SDL breaks in the wetwell airspace we are in-vestigating here.

This argument however, is not convincing if very similar data frcm the Humboldt Bay Tests h is examined closely: In those tests, downcomers, cut off two feet above water level, directed steam jets directly at the pool surface, i.e. the steam jet direction was perpendicular to the pool surface, only two feet below. In the scenarios we have postulated here, the odds of the escap-ing steam being directed right at the pool surface are obviously extremely small. Moreover, the Humbold Bay data showed pressure rises even with a steam jet directed perpendicularly toward the pool, if the steam escaped from a point higher than two feet above the pool surface. Certainly in our accident NUREG/CR-4594 19 June 1986

sequences the discharge lines involved are for the most part much more than two feet above pool level. Therefore, the unconditional conclusion that, for the discharge line breaks we are analyzing here, a substantial fraction of steam is condensed on the pool surface is not warranted.

3.3.4 Containment Failure Modes Unlike postulated accidents where containment rupture is preceded by core melt, in the scenario considered here, containment failure does not necessari-ly mean large fission product releases. While the escaping steam will have some radioactivity, this represents a negligible risk to the public. Only if containment failure were so catastrophic as to disable all ECCS systems, and thereby make subsequent core damage a high probability event, would large fis-sion product releases become likely for the accident sequences considered here. Most experimental and analytic evidence to date indicates that failure of containments similar to the types used in Mark I and Mark II BWR's will not disable all ECCS injection. Even in the early Reactor Safety Study,2 descrip-tion of containment failure for BWR's envisioned a rupture which was suffi-ciently large to rapidly depressurize the containment but was not expected to lead to castastophic structural failure of the containment. Subsequent stud-les have reached similar conclusions: A recent study of the likelihood and magnitude of containment leakage 19 for various reactor containment types con-cluded that the potential for significant leakage before reaching containment capability pressures was greater for BWR's than for PWR's, with such leakage being due primarily to f ailure of nonmetallic seals for containment penetra-tions.

A 1982 study by Marciniak et al.20 which looked at containment structural responses to overpressurization specifically for Limerick and Browns Ferry, concluded that large meridional gaps in the structures are the most likely failure modes. A study by Murray 21 combining experimental evidence and analy-sisforanexistingprestressedconcretecontainmentshowsfailurebyleakage rather than catastrophic bursting for such a containment. Rizkalla et al .2 report on an experiment where a thin walled prestressed concrete secondary containment structure (1:14 scale model of a CANDU containment) was pressur-ized. Concrete cracking and reinforcement yielding was observed with increas-ing pressure. While the first cracks were observed at 40 psi, failure did not occur until above 145 psi and then by opening a gap not by catastrophic bijrst-ing.

Most evidence to date suggests that failure in a real BWR containment would not be so catastrophic as to disrupt all coolant injection capability.

Since a BWR has many systems available for core cooling and even a very modest amount of flow (- 200 gpm) will prevent core damage, it is clear that only very severe containment failure will make core damage and subsequent releases likely. BWR systems available for core cooling are listed in Section 3.4.2.

Currently SNL is performing an uncertainty analysis in support of NUREG-1150,31 called the Limited Latin Hypercube23 (LLH) study. The LLH study is using an expert panel to estimate the probability of:

1) catastrophic failure of a PWR subatmospheric containment given a rapid pressurization transient, and NUREG/CR-4594 20 June 1986

2) ECC injection failure given a catastrophic containment failure.

The panel has concluded 23 that the probability of Item 1 is 0.6 to 1.0, and that the probability of Item.2 is 0.01 to 0.1, with most of the experts supporting the higher level. Thus although gross failure of a PWR subatmo-spheric containment was judged relatively high, total loss of ECC injection was judged to be relatively low.

Utilization of NUREG-115031 results (PWR design experiments) to estimate Mark I and Mark II containment overpressurization failures necessitates intro-ducing considerable engineering judgement. First of all, the reinforced con-crete structures utilized in Mark Is and Ils are not the same as the thin (scaled down) steel structures employed in the SNL experiments. In addition, BWRs have two levels of containment. Thus for this study, the combined proba-bility of containment failure (due to rapid overpressurization) and total loss of ECC systems is judged to have a value of less than 0.1 (see Appendix C).

3.4 Possible Mitigating Actions Actions which plant operators can undertake to mitigate the severity of the accident sequences considered here can be roughly separated into two broad categories: First, actions to prevent containment failure and second, actions to prevent core damage. The most probable mitigating actions of both types which could be undertaken are specified in the Emergency Procedure Guide-lines 24 (EPG's) which BWR operators are trained to use.

3.4.1 Mitigating Actions to Prevent Containment Failure One of the generic symptomatic emergency procedure guidelines is the Pri-mary Containment Control Guideline whose purpose is to maintain primary con-tainment integrity and protect equipment in the primary containment. Among the items which cause entry into this guideline are a drywell temperature, containment temperature, and/or primary containment pressure above its high operating limit.

Operator actions regarding monitoring and control of drywell temperature (DW/T) call for (among other actions) initiation of drywell sprays (DW/T-3).

Operator actions regarding monitoring and control of containment tempera-ture (CN/T) call for initiation of suppression pool sprays (CN/T-2).

Operator actions to monitor and control primary containment pressure (PC/P) include the initiation of either suppression pool sprays (PC/P-2), or drywell sprays (PC/P-3), or both (PC/P-6).

If the suppression chamber pressure exceeds the Primary Containment Pres-sure Limit, the operator is called on to vent the primary containment to re-duce and maintain pressure below the Primary Containment Pressure Limit (PC/P-7).

High drywell and/or containment temperature, as well as high primary con-l tainment pressure can also cause the operator to initiate Emergency RPV De-pressurization (Contingency #2 of the EPG 24) (CN/T-3, PC/P-4) or RPV Flooding (Contingency #6 of the EPG24) (DW/T-2, CN/T-4, PC-P/5).

NUREG/CR-4594 21 June 1986

As was demonstrated by the calculation; in BNL/NUREG-31940,1 if contain-ment sprays are activated successfully in time, pressure and temperature levels immediately drop sharply for the accident scenarios under analysis here. Controlled venting is another means to reduce pressure loads. However, venting, because it may require the overriding of some protection systems can be a relatively lengthy and untried procedure for some plants and would have a low probability of success for the pressurizations we are considering.

Emergency RPV depressurization would rapidly reduce the source of the high pressure steam escaping from the postulated ruptured discharge lines and so provide another means of mitigating the steam bypass accidents under dis-cussion. According to Contingency #2 of the EPG,24 such rapid depressuriza-tion can be accomplished by:

. Initiating the isolation condenser.

. Opening all ADS valves (or other SRV's totalling the number of SRV's dedicated to ADS).

. If less than 3 SRV's are open, other systems can be used to aid in de-pressurization. These include the main condenser, RHR steam condens-ing, the HPCI and RCIC steam lines, any other steam driven equipment, the main steam drain lines, the head vent and the isolation condenser side vent.

As the above discussion shows, a BWR reactor and containment provide the operator with a considerable number of options to prevent containment failure if initiated in time.

3.4.2 Mitigating Actions to Prevent Core Damage Since for these low frequency sequences containment f ailure without core damage does not constitute a significant public risk, the operator can miti-gate these bypass accidents, by keeping the core cooled with any of the avail-able injection systems still operating after containment failure.

The RPV Control Guideline of the EPG's 24 requires the operator to execute the following steps concurrently irrespective of entry condition:

. Monitor and Control RPV water level (RC/L)

. Monitor and Control RPV pressure (RC/P)

. Monitor and Control Reactor Power (RC/Q)

For the RC/L actions the EPG's list the following systems as practicable aids in restoring and maintaining RPV water level:

. Condensate /feedwater systems

. Control Rod Drive (CRD) system

. RCIC system

. HPCI system

. Low Pressure Core Spray (LPCS) system

. Low Pressure Coolant Injection (LPCI) system l l

NUREG/CR-4594 22 June 1986

If RPV water level cannot be maintained with the above, the EPG's call for entry to Contingency #1, Level Restoration. Action Cl-2 calls for lineup for injection and start of pumps in 2 or more of the following injection sub-systems:

. Condensate

. HPCS

. LPCI-A,B,C

. LPCS-A,B If less than 2 of these subsystems can be lined up, the operator should line up as many as possible of the following alternate subsystems:

. RHR service water cross tie

. Fire system

. Interconnections with other units

. ECCS keep-full systems

. Standby Liquid Control (SLC) (test tank)

SLC (lower tank)

Other pertinent EPG contingencies are Contingency #4, Core Cooling With-out Level Restoration, Contingency #2, Emergency RPV Depressurization, dis-cussed in subsection 3.4.1, and Contingency #6, RPV Flooding. The latter lists the same subsystems and alternate subsystems discussed for Contingency

1. 1 above, as useful for RPV flooding.

The conclusion to be drawn from all of this is that a BWR contains many subsystems which the operator can use to keep the core covered and the RPV pressure below damage limits. While some of these systems have a degree of interdependency, the likelihood of all of them being completely disabled by a containment rupture would appear to be small.

3.5 Sequence Frequency Results The discussions and analysis described in Sections 3.1 through 3.4 are quantified in the following tables, which list best estimates as well as upper bounds of the event frequencies which make up each of the three se-quences under discussion. The best estimate numbers are the ones explained in Section 3.2. Some upper bound numbers are quoted in 3.2 also but for the most part these numbers are varied from the best estimate based on the discussions in Section 3.3 and 3.4. The event frequencies are combined to give an esti-mate of the core damagl> frequency for each sequence.

I l

NUREG/CR-4594 23 June 1986 l

Table 3.5-1. Sequence No.1 (MSRV Line Break)

Estimated Frequency Sequence of Upper Best*

Events Bound Estimate Turbine Trips per Reactor Year 8 4 MSRV Failure to Close per Turbine Trip 2.5E-2 2.5E-2 SRVDL Break per MSRV Failure to Close 4.2E-5 -1.0E-10 (1.0E-9 to 1.0E-12**)

Containment Failure per SRVDL Break (CSS Not Actuated) 0.3 0.1 Core Damage per Containment Failure 0.1 <0.1***

Core Damage per Reactor Year 2.5E-7 <1.0E-13

• Upper bound event frequencies were retained where lack of in-formation made best estimates problematic.

• • See Appendix B for a discussion.

• • • See Appendix C for a discussion.

1 NUREG/CR-4594 24 June 1986 1

Table 3.5-2. Sequence No. 2 (HPCI Line Break)

Estimated Frequency Sequence of Upper Best*

Events Bound Estimate Demand _for HPCI per Reactor Year 2 2 HPCI Exhaust Line Break per Demand for HPCI 4.0E-5 4.0E-5**

Turbine Shutoff Failure per HPCI Exhaust Line Break .01 .01 Contaiment Failure per Turbine Shutoff Failure (CSS Not Actuated) 0.1 .02 Core Damage per Containment Failure 0.1 <0.1 Core Damage per Reactor Year 8.0E-9 <1.6E-9

• Upper bound event frequencies were retained where lack of information made best estimates problematic.

• • This is an upper bound value. If stress and strength information were available for this pipe, a PMD cal-culation (as done in Appendix B for MSRV SDL's) would very likely lead to a value several orders of magni-tude lower.

I NUREG/CR-4594 25 June 1986

~

-Table 3.5-3. Sequence No. 3 (RHR Relief Line Break)  ;

-Estimated Frequency '

Sequence of Upper Best*

Estimate  :

Events- Bound Demand for. Steam. Condensing Mode (SCM) per Reactor Year 2 1 Failure of Relief or Pressure Control Valve per Demand for SCM .01 .01 Relief Line Break per Relief or PC Valve Failure 4.0E-5 4.0E-5** -

' Containment Failure l per Relief Line Break >

~

, (CSS Not Actuated) 0.1 0.1 l- l Core Damage per Containment Failure- 0.1 <0.1 Core Damage per Reactor Year 8.0E-9 <4.0E-9 ,

• Upper bound event frequencies were retained where lack of information i made best estimates problematic.

i

! **This is an upper bound value. If stress and strength information were available for this pipe, a PMD calculation (as done in Appendix B for

• MSRV SDL's)-would very likely lead to a value several orders of magnitude ,

lower. ,

1 5

l l

l l

l l

NUREG/CR-4594 26 June 1986

(L REACTOR PRESSURE VESSEL

/ \

I DRYWELL

,p::::: 4 g

!! SV lM -

i VACUUM BREAKER " TORUS qTORUS *t9, gg DOWNCON * '

\

,f.f '

b( MAIN VENT 3 SITE OF '

,,h , ilh ,., POSTULATED BREAK _'

QUENCHER STIF F ENERS EARTHCUAKE TIE Figure 3-1. Mark I Fressure Suppression Containment.

NUREG/CR-4594 27 June 1986

u 1 UE 1

I 1

. J;# m Ir C1 / \ b* -

Q -

fp

>.c

).O- . .4 d$ ' .. .

.p e-i 1

o ,o .

p

~

E..[  ; e ~'..

h/f. l',

3

+

REACTOR k,

M MAW STE AM UNE I. J, is

'I '#

fz e SAFETY REUfF

o.

q'/

C

'(

m cca-wc.gu c m.ah.- ,.

• C VALVE

• D.

.

• b o- .

h' .

6 -

m* '

SAFETY RELIEF / 33 +W N=

VA1.VE DISCHARGE (@

LINE

\ ,-

gb t

,, w w,u ]

e' e g.,

VACUUM o p:. .s ,

_ p ._

,.( $ ({L.f*.WD ?2_khl{M[?

fN ' *j p

r Ps D

p 4

M 7 om.coutR y.c D SITE OF POSTULATED % "o W 3 S 0" BREAK hs s A.. M; g ,~ -

. *c yos<

g

. - ~ ~ - - n g - - :::::_ y n ~ ~ - n SAFETY REUEF 'd' - v - "0

., - M ^ '

VALVE ** g T; ' " "

.g.

DISCHARGE h, . , ,,, # ,y 15

• < s. ,3 J.

7 LJ otvics ,p >

• 1 T

touu4cHenl%g ._ _ _

g __

jyyPPO o t w a w x

/.y y F.x. , ,

{1 g:.. .

h. ey4 I

.Y Y* .

T Y h. .

Figure 3-2. Mark II Pressure Suppression Containment.

NUREG/CR-4594 28 June 1986

( _.

O

_ g. ...

I i

/N .

W:::'

1.-.

_ SITE dF PdSTULATED " -

l 9

I s' ,

\VbH b ..

1.

I o

O O E 1

i ,.

xg sj

j. ,

l Q fy i . . ...

~

'O JE_

q> SP +44 psp M gt i

-n . . . &

e & $,.

__% . - , . , , , _ _ _ e,,

m Y o

}

l Figure 3-3. HPCI System Schematic.

! (From Limerick FSAR.)

NUP.EG/CR-4594 29 June 1986

d.?. (

i 3= .,

e  :

g l

E I

D m SITE OF POSTULATED BREAK

"~d 2 . $_~  ? , ts fI

'I, (

x@ g a x  :

2

+--EU1.

D

e.  :>p D" ._c 2 cSo

,.. m_

-i .:a e:.:

% P E 1

?  % W-N~Y I .u u Gq.

U. _.. . <e p-) N.. *

. . . . . T I l l 09-Figure 3-4. RCIC System Schematic.

(From Limerick FSAR.)

i NUREG/CR-4594 30 June 1986 l

I r

SITE OF POSTULATED BREAK SRV SUPPRESSI POOL

%w U

T TUBE SHEET

1. j WATER LEVEL

_ .W a

LEVEL I D< _

FROM PCV REACTOR '% P pq y TO RCIC SUCTION JL V SERVICE SERVICE WATER WATER IN OUT Figure 3-5. Schematic of RHR Heat Exchanger in the Steam Condensing Mode.

NUREG/CR-4594 31 June 1986

' 1l r )

eo d 3 ct e 1 W 71 na n Z ensgn . D D -

ugaiif P P EED s e T T P qi es s R 5 21

0. T S e A (

D y

n t i i a r .

t eg k nre 1 a iot e aCn 0 r M I B

e e vyt oaa SR n mce CH PR W i L

eeH RD y

l l W

eg n r r ito V a ant II / h t an WCC U c nl e FPP / s iov HL Q i aon D MCI l t V n R e S ) S m S Z 31 M n C i e 00 r ar ( o t u f ns )

os 0 e Ce 51 e r - - r tP S EE T i S 0 20 m V 41 t i

L ( n e

y v n t ) E i i e d aBr Vs 2 tP g Ro P - e nC e Sl E i i Rt C 5 f a n ( i M I 2 l p

e r ) i m

t u n S iBs Ve mP s Rp M ice S0 .

LR r (

P 6

l l

3 a e rc T r oi L ett S u g

k ci / C aar D i F

Mec R Rb C u

S l d

g et n t n i ae pi )

tt an is T 00 ie cn 84 t v ia

(

iE t r n nT I A rn r ooi f om l o e o tt tt b nc ens m en pey y vu yvS S EF TE EAO@$$ U E' 8 G&

i

E g Initiating HPCI Exhaust Pipe Automatic Containment Core Sequence R Event Startup Integrity Turbine. Shutoff Spray Integrity Frequency 2

s. '

~ 1.0 4.0E-5 m 2.0

" .01 0.1

(.02) 0.1 8.0E-9 (1.6E-9)

I E

8

! G s Figure 3-7. Simplified Event Tree for HPCI Turbine Exhaust Line Break.

iI)lj il y ) )

ec 99 99 cn - - - -

ne EE EE eu uq 00 00 qe 84 er 84 (

(

SF e

d y o M _

t i g er rg 1 1 n i

oe O 0 s Ct n n e I d n

o C

t n m e

my a na e 1 1 t ir S ap tS 0 0 g n n o i C r u

D e k n a i

Ly e r

t B ei gr 5 e rg 5 ae - - n i

ht cn E. E. L si 4 4 r i

D o f

e f e e r i

T l

t ee n Rv e l

ya 1 v 0 E tV f

e d a i e

S f i

l p

l o '

i m

r S t

n o .

Ce 8 v 1 -

el 0 3 ra uV e s r s u e g r i P F g

n )

i tt 00 an 21 ie tv (

iE n

I 2kRn a2 g c5- *= _

uom

' T a <

o 8 8 -

m I

I l - 8

~

8

= .

I 3

1 a

8

_ 8 c l/ 8 G%

s U e bl 2

/l /

gM %x 5

~

/ g

/ / 0

/ / -

g i

/

/ a A

- /  ?

/ -

8 a l./ / "

C

/ ~

S b/

1 y / o

, 6 o 8 6 $ $

" 2 -

33s/ng3 I

5 O 5 iI' M NUREG/CR-4594 35 June 19g5 l

w INRK I EDMETRy 30TTOM SRy LINE BRERK - CASE 8 9 ,

y

= 3

~g -

N 8

We W

N '

Ya s

W 8

v2 -

h4 a

h -

.R 8

5 m

e i , . , . , i i i I E e too 400 a00 800 1000 1200 1400 1e00 1s00 at

[ TIME ISEC) cn Figure 3-10. Typical CONTEMPT 4 Pressure Response Calculation for ?each Bottom Using BWROG Transient.

l

! NFRK II DEONETRY - LIMERICK SRV LINE BRERK - CASE B .

i .

g; - . ,

e i i i i i i s

9 $$

~

L a _bg -

N

~

z -

EB E

0 $ -

It!

2 8

m -

EI d

s _

.R 8

5 E

c e t I f I ' ' ' '

d 200 400 800 600 1000 1200 1400 1C00 1200 E

_ TIME ISEC1 E

Figure 3-11. Typical CONTEMPT 4 Pressure Response Calculation for Limerick Using BWR0G Transient.

~ . _ _ .- ,

g.

l g

- MARM i 040 METRY - PEACH 80TTOM SRV LINE BREAM . . . . . . .. . . . . . . . . -

g . . . . ..... . . . . . . . . . . . ......

N s

m -

~

fa -

N

=

g =

W .

z -

E8

- E _

vs a

4 Y -

A -

1 3*I A

at

.g ! .

4 - .

. r.

Ni

= A A a & 3 3 AAA a a R A A A A A..

A A A A A A AAA A A B A A A AA&

1 A A A A A AAA 3 8 0 8 2 10 13

• 10 10 10 t

f lMEI SEC l f

Figure 3-12a. CONTEMPT 4 Pressure Response for Peach Bottom Using BNL-NUREG-319401 Transient.

l l

1 38 June 1986 NUREG/CR-4594

- - .. . - . . . . . - - -- - x .-

e MARK I GEOMETRY - PEACH 80TTOM SRV LINE BREAK I E E 5 I I I I E

k_

2 co .

s g

5 N

8 I

i 7 .

4~

1

.E3 a

/

s^

/

/ .

?3 l

< /

s l' 1

>. t

b. A i t

' i i R R i t i 4

vn 200 isoo isoo isoo imn

( rn am 800 imo f lNE I SEC I Figure 3-126. (Same as Figure 3-12a but Different Scale) 39 June 1986 NUREG/CR-4594

f 4

i MARK !! GEOMETRY - LIMERICK SRv LINE BRE AK . . . . . . .

,3 . . . . ..... . . . . . . . .

9 I

e de -

N g

zg . -

E-3 /

/

v2 " -

/

i~ <

/

5 r -

ta!

c. ,

W '

/

5 /

~," / -

5 i' R a a a a aaman a a a a aaaan a a a a a aman a a a a aaaan a a a a aaan

' :1'I lY 10 3

80 2

10 I

10'

!!Mt l 5EC 1 Figure 3-13a. CONTEMPT 4 Pressure Response for Limerick Using RNL-NUREG-31940 1 Transient. ,

i

. NUREG/CR-4594 40 June 1986

.__ _ _ _ . . . _ _ _m. -

i d

MARM 11 GEOMETRY - LIMERICK SRV LINE BREAM y 3 y I g g g y y 9

e q - -

a N

- a i

$3-W is

< /

/

c i. /

, .O. i /

[.j  ?'

I E.

I 1 i t i I g R g 3

< rn m m soo toca iroo isoa isoa isoo .am TIME ISECl Figure 3-13b. (Same as Figure 3-13a but Different Scale)

NUREG/CR-4594 41 June 1986

_~

8 MARK

- I GEOMETRY

. PEACH 80iTON SRV LINE BREAM N s a 5 /

Se

~~

. /

/ .

/

"o /

2 -

/

7 .

_a Failure Pressure #

zw E~. _

E E

ER .

v, - -

5

• 8 c

Re '

e 0 ld0 2d0 3d0 4d0 00 600 7b0 8b0 9do TOIO Figure 3-14. CONTEMPT 4 Pressure Response Without Pool Surface Condensation. Same MSRV DLB Steam Source as Used in Reference 1. Peach Bottom.

NUREG/CR-4594 42 June 1986  ;

l

i I

1 PuutM II GEMTRY - LIERICM SRV LIE BREAK

_g . . . . . . . . .

9 R

e de - -

n g

~

/

.- /

zg -

/ -

E- Failure Pressure /

I -

e. r e ~ ~

{S d

E

.9 a A R A 3 3 3 g 3 a a too acc m ano sco eco a ano suo icoo TIME ISEC1 Figure 3-15. CONTENPT4 Pressure Response Without Pool Surface Condensation. Same MSRV DLB Steam Source as Used in Reference 1. Limerick.

NUREG/CR-4594 43 June 1986

g NARK I GEOMETRY

- PFACHBOTTON'SRV LINE 8 REAM M i . , , ,

q T /

58m

- /

/ .

~ /

S

~

/

7 .

/

??

/ -

a E Failure Pressure /

$~8 N

$8 -

?

E

' * **

• eco i000 ib , ), ,

N IlME ISECl Figure 3-16. CONTEMPT 4 Pressure Response Without Pool Surface Condensation. Envelope of BWROG Steam Sources for MSRV DLB. Peach Bottom.

NUREG/CR-4594 44 June 1986

o NARK I GEOMETRYe - PEACH 80TTOM SRV L INE BRE AK

_$ ' i i a i i .

E So -

-g .

m k '

-o ~

Failure Pressure h$ -

E

<1

,~

? -

U

.R

?}8" -

Eo .

.e .

E I A I I I e 3 g ,

O O O lM lM l 4% gg gg flNE ISECl i

Figure 3-17 CONTEMPT 4 Pressure Response With Pool Surface Condensation. Envelope of BWR0G Steam Sources for MSRV DLB. Peach Bottom.

NUREG/CR-4594 45 June 1986 1

IWWIM II E(NETRY - LifERICK SRV LIE IWtEfM

_g . . . . . . . . .

E dg -

/

~ /

p -

E /

/

5

/

/ .

Y /

/

g - Failure Pressure / .

ga d

a

~

is Q

a a a a a e a a a af 21 5 W M M IM 1253 le IM IM 2GXI TIME ISEC)

Figure 3-18. CONTEMPT 4 Pressure Response Without Pool Surface Condensation. Envelope of BWROG Steam Sources for MSRV OLB. Limerick.

NUREG/CR-4594 46 June 1986

PWWIN 11 GE(NETRY - LilERICM SRV Liff OREf58 -

_g . . . . . . . . .

9 S . .

-I

~

g s!

E Failure Pressure i

Y.E

' ~

m ~ ~

Ta 8 /

5 ~

/

~

$9 Q

a a a a a a a a a ce 255 6 M M 1012 175) le im IM NERI flME 15EC l Figure 3-19. CONTEMPT 4 Pressure Response With Pool Surface Condensation. Envelope of BWR0G Steam Sources for MSRV DLB. Limerick .

NUREG/CR-4594 47 June 1986

1 i

MARM 11 MOMETRY - LIMERICK HPCI LINE BREAN -

, _g . , , , . > > > .

St e

dg - -

~

g s . .

WI Failure Pressure i

a e

B -

i a i A 3 3 g 3 3 3 g ca 2G3 400 000 000 1000 t200 1400 t400 1000 2000 flME ISECl l

4 1

' Figure 3-20. CONTEMPT 4 Pressure Response Without Pool Surface Condensation for HPCI Turbine Exhaust Line Fail-ure. Limerick.

l l

NUREG/CR-4594 48 June 1986

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

_g PHIM !! OEDETRY

- LIERICK RP8t LIE IIRERN -

9 C

/

Eg -

/-

/

u /

R

" / ~

,/

~

/

' /

$3

- / "

Failure Pressure IYE '

{s ~

d E

[s '

I a . . .

a zoo a son eso iaco uno = ison ion nao Tit 1E t SEC I Figure 3-21. CONTEMPT 4 Pressure Response Without Pool Surface Condensation. Steam Condensing Mode Transient.

Limerick.

NUREG/CR-4594 49 June 1986

MARM II rEOMETRY - LIERICK RMR LIE GRERN - POOL COM) ON

_g . . . . . . . . .

9 t

S*

~

u Es .

y .

Failure Pressur '-

og -

a A

e ~

~

{S d

E5 ls "

E a a a a a a a a a a ce reo ano eco suo sono arco ieon ison 4

isao reno i flME ISEC1 i

Figure 3-22. CONTEMPT 4 Pressure Response With Pool Surface Condensation. Steam Condensing Mode Transient.

Limerick.

NUREG/CR-4594 50 June 1986

1 4.0 ESTIMATED SAFETY SIGNIFICANCE

. 4.1 Estimated Core Damage Probability Section 3.5 summarized the predicted outcome of the accident sequences analyzed in terms of the probability of core damage (or melt). The estimated core melt frequencies are distributed as follows:

t Core Melt Probability (1/Rx-yr)

Upper Bound Best Estimate Sequence No.1 2.5E-7 <1.0E-13 Sequence No. 2 8.0E-9 <1.6E-9 l Sequence No. 3 8.0E-9 <4.0E-9 Thus from the viewpoint of predicted core melt frequency, Safety Issue 61 can be judged to be of low safety significance.

4.2 Estimated Fission Product Release and Consequences Containment overpressurization due to SRVDL breaks, plus HPCI and RHR (steam-condensing mode) line breaks was discussed in Section 3.3. For those sequences in which the containment spray systems were assumed to be not actu-ated, the CONTEMPT calculations indicated that the containment rupture pres-sure (approximately twice the design pressure) could be reached in 15-30 minutes. Although containment failure of itself would not result in core i melt, such failure could disrupt core coolant injection capability and RHR operability. Thus containment failure resulting in total loss of primary sys-tem makeup water is the key consideration in determing whether or not core damage will occur.

i The probability of core melt given containment failure of less than 0.1 estimated in Section 3.3.4 indicates that 90% of all potential SRV discharge line breaks would not lead to core melt since the operator could use several different ECC systems to maintain a covered core. For the fraction of remain-ing SRV line breaks, core melt cannot be ruled out. In that case, core melt

would progress rapidly and neither "in-vessel" nor "ex-vessel" releases would be scrubbed (See Appendix C).

Release fractions were estimated utilizing Peach Bottom (a Mark I con-tainment) calculations 2s and the Source Term Code Package (STCP).26 Since a 2 pool bypass path has been provided (i.e., the assumption that SRVDL break has overpressurized and failed containment) no credit is given for pool scrubbing.

The resulting estimated release fractions are given in Table 4.2-1 and repre-sent a combination of the V-sequence fraction (interfacing system LOCA) for "in-vessel" release and the TC1 sequence (anticipated transient without scram and containment failure prior to core melt).

The release fractions shown in Table 4.2-1 correspond roughly to a BWR-2 i release (using WASH-14002 nomenclature). Consequence estimates (using the CRAC-2 code) for a " typical" eastern reactor site (100 persons per square mile) results in a release of approximately 2x10 7person-rem.

NUREG/CR-4594 51 June 1986

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

Although there is some uncertainty in the " source term" (Table 4.2-1),

under the assummed conditions of pool bypass into a failed containment, the uncertainty in risk is dominated by the estimated uncertainty in core melt frequency. In other words, further reduction in the source terms to reflect "better estimates" would not alter the conclusions of this study. Thus, no attempt has been made to bound the release estimated given core melt.

7 Combining this consequence estimate (2x10 person-rem) with the estimated

core melt frequencies summarized in Section 4.1, results in the following es-4 timates of release

Risk (Person-rem /Rx-yr)

Upper Bound Best Estimate ,

t Sequence No. 1 5.0 <2.0E-6 l- Sequence No. 2 0.16 <.03 i

Sequence No. 3 0.16 <.08 a Thus, based on the estimated fission product releases and core melt prob-abilities for the sequences analyzed, Issue 61 can be judged to have a very small potential impact on the public health and safety.

i i

! i

! I i

i l

i i

NUREG/CR-4594 52 June 1986 l

t Table 4.2-1. Estimated Environmental Release

' Fractions for a Core Melt Acci -

dent Resulting from a Steam Discharge Line Break 1

, Species Environmental Release Fraction I .28 Cs .24 Te .24 Sr .49 Ru 1x10-6 La .012 Ce .022 Ba .39 1

1 i

k i

l

'i

) NUREG/CR-4594 53 June 1986

5.0 CONCLUSION

S The present study estimates the potential threat posed by transient ini-tiated accident sequences involving BWR systems capable of releasing steam into the wetwell airspace and thereby pressurizing the containment. Based on our analysis none of the sequences identified, involving three separate BWR systems: MSRV, HPCI and RHR Steam Condensing, present a significant contribu-tion to risk.

While some of the sequences could potentially result in core melt and are predicted to have the potential for significant releases of fission products, the f requency of these sequences is estimated to be sufficiently small to remove them as significant risk contributors.

The conclusions regarding the frequency of MSRV discharge line break se-quences differ from those of the previcus work by Economos et al.1 for two main reasons:

1) The probability of core melt, assuming containment failure, was re-estimated fro:n the previously assumed value of 1.0 to a value of 0.1.

2) Estimates of probability of MSRV Discharge Line failure are now cal-culated as 4.2E-5 per transient for the upper bound based on operat-ing experience and 1.E-10 per transient using PMD methods.

These changes are based on information obtained since the publication of Reference 1. Other sequence event frequencies were also changed based on additional data. Reduced mass and energy releases provided by the B'JROG and used for containment pressure calculations also decreased the overall estimat-ed risk in the current study. However, the updated probabilities for core melt given containment f ailure and for pipe rupture, represent the principal causes for the MSRV frequency reduction f rom the value in Reference 1.

In addition to the postulated MSRV discharge line breaks, other systems whose failure could cause steam bypass of the pool have been considered in the present analysis. The only contributors to the possibility of rapid steam discharge to the wetwell airspace were found to be the HPCI turbine exhaust and the steam relief valves from the RHR heat exchangers during decay heat removal in the steam condensing mode. The frequency of aese events have been calculated on a conservative basis as described in Sect- 3.2. Even with these conservative frequency estimates the estimated ris given in Section 4.2 is still inconsequential.

54 June 1986 NUREG/CR-4594

6.0 REFERENCES

1. Economos, C. et al., " Postulated SRV Line Break in the Wetwell Airspace of Mark I and Mark II Containments - A Risk Assessment," BNL-NUREG-31940, Ocotober 1982,

2. NUREG 75/104 (WASH-1400), " Reactor Safety Study. An Assessment of Acci-dent Risks in U.S. Commercial Nuclear Power Plants," October 1975.*

3. Hargroves, D.W. and Metcalfe, L.J., " CONTEMPT-LT/028, A Computer Program for Predicting Containment Pressure-Temperature Response to a Loss-of Coolant Accident," NUREG/CR-0255, March 1979.

4. Philadelphia Electric Co., " Limerick Generating Station Probabilistic Risk Assessment," Rev. 3, April 1982.

5. Attachment to BWR0G Letter No. BWR0G-8511 dated March 13, 1985 from J.M.

Fulton, BWR Owners Group, to R. Wayne Houston, USNRC, entitled "BWR Own-ers Group Response to NRC Request for Information Relating to Generic Issue #61, October 22, 1984."

6. USNRC Technical Training Center, " Systems Manual Boiling Water Reactors BWR/4 Design," Volume 1 and 2.

7. Bechtel Power Corp., San Francisco, " Peach Bottom Atomic Power Station Units 2 and 3, Mark I long Term Program Plant Unique Analysis," April 1982.

8. NUTECH Engineers Inc., San Jose, "Monticello Nuclear Generating Plant Unique Analysis Report," Revision 1, November 1982.

9. Teledyne Engineering Services, Waltham, " Plant Unique Analysis Report of the Torus Suppression Chamber for Vermont Yankee Nuclear Power Station,"

Rev. 2, TR-5319-1, November 30, 1983.

10. NUTECH Engineers. Inc., San Jose, " Hope Creek Generating Station Plant Unique Analysis Report," January 1984.

11. Philadelphia Electric Co., " Peach Bottom Atomic Power Station, Units 2 and 3. Final Safety Analysis Report," April 31, 1970.

12. Philadelphia Electric Co., " Limerick Generating Station, Units 1 and 2.

Final Safety Analysis Report," Rev. 3, March 1982.

I 's . General Electric Co., "BWR Owners Group NUREG-0737 Implementation:

Analysis and Positions Submitted to the USNRC," NE00-24951, June 1981.

14. Attachment 1 to Mississippi Power & Light Letter No. AECM-82/353 dated August 19, 1982 from L.F. Dale, MP&L, to H.R. Denton, USNRC.

• NUREG and NUREG/CR documents are published by the U.S. Nuclear Regulatory Commission.

NUREG/CR-4594 55 June 1986

- _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . l

15. Attachment to Illinois Power Company Letter No. V-0714 dated May 25, 1984 f rom 0.1. Her born, IPC, to A. Schwencer, USNRC.

16. Hubble, W.H. and Miller, C.F., " Data Summaries of Licensee Event Reports of Valves at U.S. Commercial Nuclear Power Plants," NUREG/CR-1363, June 1980.

17. Miller, C.F. et al ., " Data Summaries fo Licensee Events Reports of Valves at U.S. Commercial Nuclear Power Plants, NUREG/CR-1363, Revision 1, Octo-ber 1982.

18. Lin, C.C. et al., " CONTEMPT 4/M004 A Multicompartment Containment System inalysis Program," NUREG/CR-3726, March 1984.

19. " Containment Performance Working Group Report," Draft Report for Comment, NUREG-1037, May 1985.

20. Marciniak, T.J. et al., " Resolution of Containment Structural Response Issues Under Degraded Core /xcidents," NUREG/CP-0033, Proc. of the Work-shop on Containment Integrity, Vol .1, pp. 47-62, October 1982.

21. Murray, D.W., " Experiments and Analyses to Predict Behavior of Pre-stressed Concrete Containment Structures," NUREG/CP-0033, Proc. of the Workshop on Containment Integrity, Vol . 2, pp.181-196, October 1982.

22. Rizkalla, S., "A Test of a Model of a Thin-Walled Prestressed Concrete Secondary Containment Structure," Trans. of the 5th International Confer-ence on Structural Mechanics In Reactor Technology, J4/2, August 1979.

23. Sandia National Laboratory, " Limited Latin Hypercube Uncertainty Analy-sis," Appendix A1, February 1986.

24 Attachment to BWROG Letter No. BWROG-8262 dated December 22, 1982 from T.J. Dente, BWROG, to D. A. Eisenhut, USNRC, entitled: BWR Emergency Pro-cedure Guidelines, Revision 3 (Prepublication Draft dated December 8, 1982).

25. Denning, R.S., BCL, " Report on Radionuclide Release Calculations for Selected Severe Accident Scenarios," January 31, 1986.

26. Silberberg, M. et al., " Reassessment of the Technical Bases for Estimat-ing Source Terms," USNRC Accident Source Term Program Of fice, NUREG-0956, Draft for Comment, July 1985.

27. General Electric Co., " Additional Information Required for NRC Staff Gen-eric Report on Boiling Water Reactors," NED0-24708A, Rev. 1, December 1980.

28. NUREG-0869 Rev.1, "USI A-43 Regulatory Analysis," October 1985.

29. Electric Power Research Institute, EPRI NP-2230, "ATWS: A Reappraisal,"

Part 3: Frequency of Anticipated Transients. January 1982.

NUREG/CR-4594 56 June 1986

30. - 11 berg, D. et al., "A Rev-lew of the Shoreham Nuclear Power Station Probabilistic Risk Assessment," NUREG/CR-4050, November 1985.

31. " Evaluation of Severe Accident Risks and the Potential for Risk Reduction," NUREG-1150 (To be published).

NUREG/CR-4594 57 June 1986

Appendix A BWR Owners Group Comments on BNL Informal Report BNL-NUREG-31940 r

l l

l l

I NUREG/CR-4594 A-1 June 1986

NUREG/CR-4594 A-2 June 1986

Appendix A

'BWROG Comments on NUREG-31940 and Action Taken Two sets of BWROG comments on NUREG-31940 were transmitted to BNL by the USNRC: The first was an informal letter with two enclosures from David R.

Helwig to Farouk Eltawila dated October 22, 1984. The second was Letter No.

BWROG-8511 plus attachments dated March 13, 1985 from J. M. Fulton, BWROG Chairman to R. Wayne Houston, Assistant Director for Reactor Safety Division of Systems Integration, USNRC. This second transmittal, entitled, "BWR Owners Group Response to NRC Request for Information Relating to Generic Issue #61",

in addition to repeating and formalizing the comments of the first transmit-tal, also provided six mass and energy releases during transients involving the actuation of one or more safety relief valves, as requested by the NRC.

The following items made up this transmittal:

1) Letter from J. M. Fulton to R. Wayne Houston containing two comments on NUREG/CR-31940.

2) Two tables and a figure describing six mass and energy releases.

3) A section entitled, "BWROG Comments on BNL-NUREG-31940" divided into comments on event frequency and event consequences.

4) Three additional comment sections entitled:

A. Overview Comments: Impact of the Postulated Sequence.

B. Misinterpretations in the Report.

C. Conservative Bias.

5) A section labeled: " Enclosure II - Summary of GE Analysis on Transi-ent Sequences with S0RV".

In Table A.1 below, each BWROG comment in the BWROG-8511 transmittal is identified as to its content and where in the transmittal it appeared. For each comment the action taken and the relevant sections of the current report addressing the item are stated.

BWROG comments on a preliminary draft of the current report were received in a letter from J.M. Fulton, BWR0G to A.W. Serkiz, USNRC, dated May 27, 1986. While these comments were reviewed prior to publication of this final draft of our report, they arrived too late to be included in this Appendix.

NUREG/CR-4594 A-3 June 1986

b NUREG/CR-4594 A-4 June 1986

E Table A.1 gj R

S

1. Response BWROG Comment on BNL-NUREG-31940 gj Comments in letter from Fulton to Houston:

The BNL report identified the frequency of the T?. : current report reevaluates all the event fre- i l

transient event (turbine trip with bypass) followed quencies in the MSRV sequence based on the latest by SRV actuation and fail tre to close and discharge information. The rationale detailed in subse-line break as 10-6 per reactor year. Based on the quent BWROG comments was taken into considera-rationale in the attached discussion, a much lower tion. A lower accident frequency than that of frequency can be justified, primarily due to the BNL-NUREG-31940 was calculated.

low pipe break probability.

The BNL report indicates that containment failure The new study used mass and energy flows provided would occur in ten minutes followed oy a core by the BWROG (Section 3.3.1). The rate of steam

?" condensation in the wetwell airspace is highly o' melt. This is unrealistic for the reasons provided in the attachment. Use of more realistic values uncertain under pool bypass conditions. The for flow and steam condensation in the pool would present study treats the condensation rate as a result in cony 31nment failure at greater than sensitivity parameter as discussed in Sections thirty minutes, if at all, and no core damage. 3.3.2 and 3.3.3.

Comments in Section entitled "BWROG Comments on BNL-NUREG-31940 On the basis of EPRI NP-2230, the frequency per This statement is in contradiction with a later year of the initiating transient should be reduced BWROG comment which suggests a frequency of 7 from 4 to 0.3. transients per year. As discussed in Section 3.2.1.1, a frequency of 4/yr is still used for the best estimate.

E' 5 Based on better SRV Design (NED0-24951), the As discussed in Section 3.2.1.2, NE00-24951 does

frequency / demand of a stuck open MSRV should be not present any quantitative data and so does not g; reduced from 6.74E-3 to 5.6E-4. support a frequency reduction. However, based on data in Revision 1 of NUREG/CR-1363, the frequen-cy of a stuck open MSRV is changed to 2.5E-2 per transient.

2 Table A.1 (Continued) 55

'A

BWROG Comment on BNL-NUREG-31940 Response 7 Based on updated stress and strength data, the

$; The frequency per demand of a broken MSRV discharge y? line should be reduced from 7.4E-5 to 1.0E-10. best estimate value for a discharge line break frequency by the PMD method is calculated to be 1.0E-10. The method used in BNL-NUREG-31940 to calculate the 7.4E-5 value is recognized to give a conservative upper bound. Section 3.2.1.3 and Appendix B give details on discharge line break frequency calculations.

Operator error probability should be reduced from Because of the reduced mass and energy releases 0.5 to 0.1. of the transients used in the current study, the operator will have more time to assess the situa-tion and take proper action. Therefore, the best estimate of operator error has been reduced to 3 0.1 (See Section 3.2.1.4).

b Based on the containment failure modes discussed Probability of core melt given containment failure should be reduced from 1.0 to approximately 0.01. in Section 3.3.4 and the mitigating actions the operator can take to keep the core cooled as dis-cussed in Section 3.4.2, the conditional proba-bility of core melt given containment failure has been estimated at less than 0.1 as indicated in 3.2.1.5.

The overall event frequency per year of TPDZ The best estimate value of the TPDZ sequence sequence becomes negligibly small, based on a reassessment of the latest data is calculated to be 1.0E-13/Rx-yr.

Comments on Event Consequences E'

8 The BNL report indicates that containment failure The energy input into the containment is now that

r. would occur in ten minutes followed by a core provided by the BWROG. Operator actions guided 8 melt. This is unreasonable because the energy by the EPG's and designed to prevent both con-

• input to the containment is much less than used by tainment failure and core damage are discussed in BNL; operator action guided by emergency procedures Sections 3.4.1 and 3.4.2. The Bodega Bay data could terminate the event before containment does not prove that higher steam condensation failure and core melt is an unlikely consequence. rates prevail. See Section 3.3.3 for details on The energy input to the containment is at most a this latter point.

gg Table A.1 (Centinued)

=

E? .inHU)3 Comment on BNL-NUREG-31940 Response y

f factor.of two below the BNL assumption as shown in

. $ Figure 1. In addition, data from Bodega Bay tests ,

.2 in 1962 show that complete steam condensation oc-t L curs for the case of downcomers two feet above the pool' surface. For a discharge line, the exiting steam flow is ~at a much higher velocity than the downcomer case, so the condensation on the pool  ;

should be effective even from a greater distance.  :

These factors combine to significantly extend the time to reach containment failure pressure.  !

In the BNL report, high containment pressure and As the frequency estimate in Section 3.2.1.5 of i ultimate failure is assumed to lead directly to- this report shows,. core melt given containnent core melt. There is no mechanistic-link between failure has been estimated to be small based on i

containment failure by overpressure and core melt. the operator actions available to prevent core 2 Probabilistic risk assessments for Mark I and II damage as discussed in 3.4.2.  :

la BWRs suggest that the most likely containment fail-ure location will be above the suppression pool, thus ensuring a water supply for injection into the vessel if other water sources are unavailable. l Comments in the Section Entitled:  ;

A. Overview Comments: Impact of the t Postulated Sequence The assumed RPV pressure during the postulated The BWROG supplied reduced mass and energy re- .t transient does not appear to be consistent with GE leases have been used in this study (Section

• analysis (see attached Enclosure II). As seen by 3.3.1). The added time for operator action has

,_ the comparable GE analysis, the RPV pressure de- been considerd in the revised estimate for opera- ,

Ef creases much more rapidly than assumed in the BNL

. tor error probability in Section 3.2.1.4.

5 work. This more rapid RPV pressure decay curve re- ,

- sults in a slower rate of containment pressuriza-  !

o$

tion than is calculated in the BNL report. This in  !

turn leads to additional time. for operator action to successfully mitigate the accident. <

?

t

-e,-,, - - . - .-r- ,,-r ,_-m ,% , --, , - - - m- c-,.- - ---- - -,,- --., y

, Table A.1 (Continued)

E y BWR0G Comment on BNL-NUREG-31940 Response A

? The " bypass fraction" is not defined by BNL. A This comment is apparently based on a

$; transient initiator with scram causes 3 to 9 SRVs misinterpretation and is not reproduced here in jl to open, all relieving steam pressure to the sup- its entirety. Bypass fraction in BNL-NUREG-31940 pression pool. Therefore, most of the initial merely referred to the amount of steam escaping steam flow does not bypass the pool (even with an from the broken line compared to the amount of S0RV) but rather is condensed in the pool. This flow in the line. (A double ended guillotine initial injection of steam flow can represent a break means 100% bypass.)

substantial amount of the assumed steam bypass flow which the BNL report has apparently modeled to be deposited directly in the containment wetwell air-space (i .e., for the 100% bypass case).....

Comments on the Section Entitled:

B. Misinterpretations in the Report 3 (page numbers' refer to the pages in so BNL-NUREG-31940)

P. 9 The use of four turbine trips / year as a scaling This contradicts a previous BWROG comment which factor or figure of merit is improper. Seven (7) suggested 0.3 transients /yr. As discussed in transients per year is a better estimate to use in 3.2.1.1, a frequency of 4/yr is the best estimate the evaluations. used in this report.

P . 10 Limerick unavailability of RHR is quoted as This comment is not relevant to the current 3.5E-8. This is not true. This point estimate un- study.

availability corresponds to the total containment Er heat removal function over approximately 30 hours3.472222e-4 days <br />0.00833 hours <br />4.960317e-5 weeks <br />1.1415e-5 months <br /> g for selected accident sequences. Therefore, it in-

~ cludes the main condenser as a viable heat removal gg sink, plus repair of the RHR system. These values os are inappropriately quoted by BNL.

Table A.1 (Continued)

! BWROG Comment on BNL-NUREG-31940 Response E

P. 10

$ Similarly, WASH-1400 is incorrectly quoted to have This comment is not relevant to the current

$ portrayed RHR as having a reliability of 10-6/ study.

demand. The WASH-1400 value is more appropriate for all containment heat removal mechanisms, not only RHR.

P. 12 The TPD sequence is not a core melt and is not a As noted in Section 3.2.1, the TPD sequence has significant challenge to containment. Comparing been dropped from consideration.

this sequence with a large LOCA is totally inappro-priate. Such a comparison is incorrect and de-tracts from the principal thrust of the report to identify risk significant containment bypass scen-p arios.

e P . 14 The determination or assessment by BNL that a se- This comment is in agreement with the discussion quence that is " conservatively" quantified at 1.5 x of Section 3.2.1 of the current report. The se-10-9/yr. is a "significant contributor to risk", is quence referred to in the comment has been not consisent with either WASH-1400 or Limerick dropped from consideration.

where the risk of latent fatalities is dominated by sequences in the range of 10-s/yr and the risk of early fatalities is dominated by sequences in the range of 10-7 per year.

P. 22 E' The list of " dominant accident sequence frequen- This list was not intended to be complete.

5 cies" for WASH-1400 and Limerick is not complete.

~

y P. A-3 The containment spray operation is contrary to the The containment spray system was simulated in a BWR0G EPGs. simplified manner to determine mitigative capa-bility and does not reflect the detailed actua-tion procedures in the EPGs.

g Table A.1 (Continued) b BWROG Comment on BNL-NUREG-31940 Response Comments on the Section Entitled: ,

C. Conservative Bias (page numbers refer to the pages in BNL-NUREG-39140)

P. 5 Number of SRV demands over 40 years is approximate- These demand numbers are used only to allow for ly 2500 - 4000 based on a realistic estimate rather fatigue in the SRV discharge line strength than 7000 - 14000 which is similar to a design val- parameter. The accident sequence frequency is ue. not significantly influenced by which set of numbers is used.

h P. 5 SRVDL pipe failure rate of 7.4E-5/d is described as As Section 3.2.1.3 makes clear a best estimate an upper bound but is used in the BNL analysis. value of 1.0E-10 per transient is used in the

, current analysis, while an estimate of 4.2E-5 per transient is used as an upper bound.

P. 8 & 6 Containment high pressure is assumed to lead di- As Section 3.2.1.5 shows core melt frequency rectly to core meltdown. This was purely an as- given containment failure has been estimated as sumption made for conservatism in the WASH-1400 an- small based on available operator action as alysis. More recent analyses (Grand Gulf, discussed in 3.4.2.

Limerick, GESSAR, etc.) have pointed out there is no mechanistic link between containment failure by p overpressure and core melt, o

[ P. 7 & 8 TPD is a sequence with no core damage, no contain- As stated in Section 3.2.1, the TPD sequence has ment pressure above ultimate but BNL still assumes been dropped from consideration, releases of radioactivity.

1 g Table A.1 (Ccntinued)

BWROG Comment on BNL-NUREG-31940 Response n

[ P. 10

$ Multiplier of 1.5 is used for " conservatism" in the The estimate of SRV failure to close frequency assessment of the failure probability of an SRV to used in the current study avoids the need for the close after. activation. 1.5 multiplier.

P. 9 The data base for SRV failure to reseat is outdated As discussed in 3.2.1.2, NED0-24951 does not and does not reflect the improvements which have give any quantitative data. Revision 1 of been achieved in the last few years. This topic NUREG/CR-1363 has been used as updated source for has been addressed by the BWROG in response to TMI SRV failure to reseat.

Item II.K.3.16 (see NED0-24951).

p P.11

~

The judgement that manual actuation of the CSS is Based on the extended time available for operator "as likely to occur as not" is overly conservative action with the new transients, likelihood of CSS given the existence of specific procedures and op- actuation has been increased to a best estimate erator training. of 0.9 and to 0.7 for the conservative bound as Section 3.2.1.4 shows.

Discussion in Enclosure II, " Summary of GE Analysis As discussed in Section 3.3.1 of the current on Transient Sequences with SORV" regarding mass report, the transient mass and energy releases and energy flows into the containment. provided by the BWROG are used in the present analysis.

E 8

t; 8

I NUREG/CR-4594 A-12 June 1986

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

i- i

(

l

Appendix B ,

1

', g  :

1 i Estimated MSRV Failure to Close Frequency and  ;

[- MSRV Discharge Line Rupture Frequency  :

i .

4 1

i ,

f i~

,t l

t-I s

, i

1 l 6 1- t 1

l

?

i i

i 1

k i

.f-.

,i 1

4 1

1

! ~.

i l  !

t i i L

I B-1 June 1986 f lNUREG/CR-4594 k

i

,--nr,~ ---,n. . . - - , - - - - , - , - ,,---v---- ,. . - - - - , , .,-c, . , . -- ~ - - - - - , - - - - - - - - - - - - - - - - , - - - - >

l l

NUREG/CR-4594 B-2 June 1986

B.1 Frequency of MSRV Failure to Close When this investigation was initiated, it was recognized that one of the important factors was an accurate quantification of Main Steam Relief Valve failure rates. In BNL-NUREG-31940, an MSRV failure rate per demand was esti-mated based on NUREG/CR-1363,1 " Data Summaries of Licensee Events Reports of Valves at U.S. Commercial Nuclear Power Plants, January 1,1976 to December 31, 1978". For the current study, valve failures and valve actuation fre cies, reported over the years in " Nuclear Plant Operating Experiences,"2 quen-and summarized in NUREG-1363 Revision 1,7 " Data Summaries of Licensee Events Reports of Valves at U.S. Commercial Nuclear Power Plants, January 1,1976 to December 31, 1980" were used. The demand failure rate of BWR primary relief valves failure to reseat is given as 3.1E-3 in NUREG-1363, Rev. 1, based on 20 failures in 6524 demands. This is slightly lower (30%) than the failure fre-quency reported in the original NUREG-1363 and used (with a 1.5 multiplier) in BNL-NUREG-31940. A 1.5 multiplier is given as the increase necessary to ob-tain the upper 95% confidence limit for the failure rate in both the original NUREG and Revision 1.

To obtain the number of demands needed to calculate the failure rate, the authors of NUREG-1363 Rev. 1 make the assumption that all scrams, both auto-matic and manual, caused pressure transients that lifted the entire population of a BWR plant's safety-relief valves. A check of specific LERs shows, how-ever, that for many scrams only a portion of the valve population actually lifts. Therefore, the stated assumption introduces a non-conservatism into the failure rate calculation. For the present study this nonconservatism has been offset in the following manner.

The data presented in NUREG-1363 Revision 1 showed 20 individual MSRV failures to reseat during 814 events which caused safety-relief valve actuation in BWR plants. By assuming an average number of valves, L, lift during each of these 814 ' plant' demands we can obtain the failure rate to re-seat per (valve) demand as 20/814L. The failure per transient during which L valves lift is then 20/814 or 2.5E-2. Note that the demand failure rate per valve of 3.1E-3 given in NUREG-1363 Rev.1 implies approximately 8 valves lift during each transient which seems high when compared to individual LERs which give number of valves actuated. By using the failure rate per transiept, 2.5E-2, we avoid having to assume a specific number of valves lifting per ac-tuation.

B.2 Frequency of MSRV Discharge Line Failure Most available data on piping failures in general and nuclear plant pip-ing in particular deal with process pipin 8 often exposed to a corrosive environment.g which is under By contrast, constant MSRV load piping discharge and is really standby piping which is under no significant stress except for the short bursts of temperature and pressure loads which occur during MSRV actua-tion. Therefore, process piping failure data is not very appropriate for MSRV discharge lines and other means of estimating failure frequency must be found.

In BNL-NUREG-31940, two different methods for estimating failure rates were used: An evaluation of operating experience similar to that used in WASH-140010 to estimate LOCA frequencies gave a conservative upper bound estimate

-(See BNL-NUREG-31940 Appendix C). Another estimate was obtained based on Probabilistic Mechanical Design Methods (PMD) using available stress and NUREG/CR-4594 B-3 June 1986

t t

material strength data. (See BNL-NUREG-31940 Appendix B). The current re-port, as indicated in Section 3.2.1.3 uses estimates obtained from updating both of these methods: The evaluation of operating experience has been up-dated to reflect the additional accumlated experience since BNL-NUREG-31940

, was written. The PMD method has been revised to include new stress data and,

more importantly, new and more accurate material data.

, 8.2.1 Estimate of MSRV Pipe Failure Rates from Operating Experience i

The approach used here to obtain a bounding estimate of SRVDL failure j rate is similar to that used in WASH 1400 H to estimate the probability of a

" catastrophic" failure of the Reactor Coolant Primary Boundary leading to a 4

design basis LOCA. What was done there was to reason that since such an event

! had not occurred during the total number of plant years in which BWR's were operating, to postulate one such failure during this period would represent a bounding of the frequency of this potential failure. In our case, we have developed this failure rate on a "per transient" basis. Specifically, we have used the results cited in Reference 7 which indicate that there were 814 tran-

! sients actuating MSRVs during the 106 reactor years covered by that study or approximately 8 transients involving SRVDLs during a reactor year. We have also estimated from the information supplied in Reference 11 that, through 1981, the total reactor years of BWR operation (all Mark I) is 210. Since then an additional 4 years times 22 operating Mark I reactors or 88 reactor years have elapsed. (Several Mark 11 reactor years of operation have also el apsed.) These figures imply that a total of 2400 transients challenging SRVDLs have occurred during BWR experience without a single reported. failure.

We thus conclude that an estimate of a bounding f ailure rate can be obtained by postulating one such failure during this period yielding 1/2400L failures per demand where L is again the average number of valves (and therefore lines) challenged per transient. The SRVDL failure rate per transient is then simply 1/2400 or 4.2E-4. This is the value used in Section 3.2.1.3 of this report for the conservative upper bound MSRV discharge line failure per transient estimate.

i

The probability that the pipe break and the stuck open valve are on the i same discharge line is 1/P where P is the number of MSRVs or lines in the l pl ant . The 22 operating Mark I plants range from one plant with only 4 valves to three plants with 13 safety and relief valves. Of the five operating Mark II plants three units have 18 valves and two units 16 valves. The average number of valves per plant for all Mark I and Mark II units is 10. The median 7

is 11. To obtain a generic failure rate we have estimated P as 10 and lumped this together with the pipe break probability to obtain 4.2E-5 as the upper 3

bound failure estimate per transient.

B.2.2 -Estimate of MSRV Pipe Failure Rates Obtained from Probabilistic Mechan-l ical Design B.2.2.1 Description of Method The concept of a safety factor in design implies that there is a certain i separation of the strength of a particular component from the stresses applied to that component. What must be remembered is that the strength and stress in

question are mean values (or conservative estimates of mean values) of a com-I plete strength and stress distribution. For any reasonable distribution,

NUREG/CR-4594 B-4 June 1986

including the normal, an overlap of stress and strength, however small~, is un-avoidable (Figure B-la). Rather than use a safety factor, the adequacy of a component for which strength and stress distributions are known can be deter-mined from the probability that strength exceeds stress (reliability), or the probability that strength is less than stress (failure). By definition, reli-ability equals one minus the failure probability. Given the variable nature of the parameters encountered in the physical sciences, the probabilistic approach to design is a natural choice.

The amount of overlap of the strength and stress distribution is a mea-sure of the probability of failure. If the distributions are known, the reli-ability or, alternately, probability of failure, can be computed. As de-scribed in Reference 12, the salient points in the calculation are the follow-ing:

Let s, the stress, and S, the strength, be normally distributed random variables with density functions f(s) and f(S). Let 6 = S-s. Since f(s) and f(S) are normally distributed, so is f(6). Therefore, 1 1 (6-0 f(6) = a /2w **P YI o I 6

(6) -

where 3 = 5-s and o6 *

"S + "s A bar indicates the mean and a the standard deviation (or standard deviation estimator) of the variables indicated.

Figure B-lb shows a sketch of f(6). The density of 6 to the right of zero is the reliability, while the failure density is represented by the por-

' tion less than zero. In other words, the reliability R is the probability that 6 > 0 and so R =f"f(6)d6 and the failure probability Pf is given by 1-R to or Pthe make = f Eran(6)d6.

f To evaluate either of these integrals, one has only sformationgwhich relates 6 to the standardized normal variable Z and look up the integral values which are tabulated in standard tables of normal functions. The transformation is given by Z = 6-3 Since the Z coor-dinate of interest is the one where 6 = 0, Z = =

g (o +o

\g s )I or Z=

f 2+o 2y 1/2 og gj Therefore, if the distributions of S and s are known, Z can be computed and the reliability or failure probability found from tables of the standard normal function. (See Reference 12 for a more detailed discussion.)

NUREG/CR-4594 B-5 June 1986

B.2.2.2 Obtaining Parameter Values If stress and strength are assumed to be normally distributed, then the mean and standard deviations for each can be used to compute the probability of failure.

The stress data for this study was obtained from measurements made in the SRV lines of the Caorso nuclear plant in Italy during SRV discharge tests.

Details of measurement location, strain gage arrangement and data interpreta-tion can be found in Reference 13. Although many measurements were made, Ref-erence 13 cites only 7 data points for which all stresses, i.e., pressure, thermal, etc., are given. Two additional data points can be found in Refer-ence 14. These nine values lead to a mean stress of 23,700 psi, with a stan-dard deviation of 2560 psi. Additional stress data from Monticello 1s and Kuoshengis in-plant SRV tests was also reviewed. However the details of the stress measurements reported were not as explicit as those in References 13 and 14, i.e., it was not clear whether all stresses were included in the val-ues cited. Therefore this data was not used in our final computations.

Implicit in the use of the Caorso data to determine failure rates for the present purpose is the assumption that the strain gages in Caorso were located on the most highly stressed part of the SRV line.

To obtain the strength parameters, the SRV pipe material properties and their variability must be established. In BWR wetwells, MSRV discharge lines are usually 10-inch, Schedule 80 piping, with the material specified as A106, Grade B, carbon steel. The yield strength for this material is listed as 35 ksi, while the ultimate strength is given as 60 ksi. These values were taken to be conservative estimates of the means of the assumed normal yield and ul-timate strength distributions in the BNL-NUREG-31940 study. For the present investigation accurate estimates of the actual material properties were used instead: The United States Steel Corporation, a leading U.S. manufacturer of 10-inch A106 Grade B Schedule 80 pipe, supplied us with yield and tensile strength data for their entire 1985 production of this material. This produc-tion consisted of 16 heats with each heat supplying about 160 tons of product.

The mean yield st;ength of the 16 heats was 45.0 ksi with a standard deviation of 4.15 ksi. Mean ultimate strength was 73.4 ksi with a standard deviation of 1.09 ksi.

A choice must be made regarding the value assigned to the strength at failure, 5, i.e., the plastic collapse load. Choosing the yield strength would be overly conservative, since the carbon steel pipe can undergo substan-tial plastic deformation before rupture failure. Construction quality can also be considered high since these pipes are designed and f abricated to spe-cific code requirements. One may be led to conclude then that the ultimate strength would be an appropriate choice for 5. This may be too optimistic an assumption however, since the potential failure at smaller loads due to weld-ment imperfections or other stress concentrations must be recognized. There-fore, a reasonable value for failure strength 5 is judged to be a stress level halfway between the yield and ultimate strength of A106 Grade B. This choice of 5 agrees with the plastic collapse load chosen by General Electric for a generic evaluation of Mark I SRV discharge line integrity.18 The philosophy behind choosing this average of yield and ultimate stress is the same as that expressed for BWR containment failure criteria in Reference 10.

NUREG/CR-4594 B-6 June 1986

l Therefore, if 5 = (ultimate stress + yield stress)/2, the corresponding standard deviation.as is then computed from:

"S * " yield * " ultimate For the yield and tensile strength of A106 Grade B pipe cited above 5 = 45,000 + 73,400 or S = 59,200 psi o s=(41502 + 10902 or os = 4,290 psi .

B.2.2.3 Sample Calculation

. A particular probability value is computed using the parameters indicated earlier as (i.o )s = (23700, 2560) for the MSRV line stress and (5,oS) = (59,200, 4290) for the strength.

Then Z= 5-s/yo32 + "s or Z= 59200-23700/k42902 + 25602 = 7.1.

From a standard table forl' f(Z)dZ the probability of failure is found to be Pf = 7.0E-13 or -1.0E-12 pdr demand.

B.2.2.4 Fatigue and Corrosion Considerations As stated in NUREG-0661,17 SRV piping must be evaluated in accordance with ASME Class 2 Rules. The ASME Class 2 Fatigue requirements provide for a stress range reduction factor which is a function of the number of alternating stress cycles as shown in Table NC-3611.2(e)-1 of the 1977 ASME Boiler and Pressure Vessel Code,Section III, Division 1, Subsection NC. Below 7,000 cycles no reduction is called for. Best estimates for the number of MSRV actuations over 40 plant years range from 2,500 to 6,000.19 Therefore 7,000 is a very conservative design value.

Assuming that the entire applied stress can alternate, one can get an es-timate of the effect of fatigue on the failure probabilities by applying the stress range reduction factor directly to the mean value of the strength, 5.

For example, using values from the table cited above and the "as built" value for 5 of 59.2 kips, between 7,000 and 14,000 cycles S = 0.9x59.2 kips or 53.3 kips. Taking the strength standard deviation og as a fixed percentage of S, i.e., 4,290/59,200 = 3,860/53,300 = 7.0%, one can now calculate a failure probability taking into account fatigue by using the modified strength values and the previously used applied stresses.

NUREG/CR-4594 B-7 June 1986

I Corrosion can be accounted for by assuming a certain reduction in pipe thickness during the life of the plant. Using a general rule of conventional design practice, one can assume a 1/6 reduction in pipe thickness due to cor-rosion over 40 plant years. This change in wall thickness will modify the applied stress values obtained from References 13 and 14. For a thin walled cylinder, like the SRV pipe, thermal stresses are relatively unaffected by thickness, while pressure and bending stresses are inversely proportional to the wall thickness. Modifying the stress components from References 13 and 14 appropriately then leads to a mean stress value s = 25,100 psi and a os =

2,690 psi. Now a demand failure rate accounting for corrosion can be found by using the modified stress parameters and the original strength parameters.

Obviously, one can evaluate failure rates using both modified stress and modified strength parameters thus accounting for both corrosion as well as fatigue at the same time. The values for several cases are given below.

Failure Rate Pf "As Built" 1.0E-12 Fatigue (more than 7,000 cycles) 7.0E-11 Corrosion 1.0E-11 Fatigue and Corrosion 1.0E-9 Based on the above values of Pf and the consideration that the SRV Dis-charge Lines will most probably experience less than 7,000 cycles but still suffer some fatigue deterioration over plant life, a best estimate of 1.0E-10 was determined for a discharge line failure probability.

B.2.2.5 Comments on Assumptions and Accuracy Any discussion of this PMD stress-strength overlap method would be incom-plete without mentioning its sensitivity to the assumptions regarding distri-bution shapes and parameter values.

The high sensitivity of the failure probability to variations in the standard deviation is shown in Figure B-2. Three different strength standard deviations taken as 4, 7 and 10% of the mean failure strength give vastly dif-ferent failure probabilities as shown in the figure. That such small varia-tions in the standard deviation a lead to such large differences in the fail-ure probabilities is not so surprising since only the extreme tails of the distributions overlap to determine the probabilities and these tails are sen-sitive to o. While Figure B-2 shows the effect of varying the standard devia-tion of the strength, a similar effect would be achieved by varying the stress standard deviation.

So far, only a normal distribution has been considered. The sensitivity of the probabilities to distribution shape is obviously also great since it is the extreme tails of the distributions which determine the probabilities.

Even if a distribution appears close to normal for moderate distances away from the mean, the shape of the extreme tails may differ much more, thereby changing the probability from one predicted using a normal distribution as-sumption.

NUREG/CR-4594 B-8 June 1986

An interesting and thorough discussion of the failure probability's sen-sitivity to assumptions of distribution shape and parameters can be found in Reference 20.

It should also be noted that with probabilities as low as 1.E-10 for pipe rupture as estimated here, other effects, such as earthquake induced indirect pipe failure may actually dominate failure estimates. The probability for such a failure occurring in the primary loop piping of a Mark I containment (Brunswick) was estimated by one source 21 as ap per reactor year. (90th percentile confidence limit 5x10 per ,reactor proximately year). 2x10 8 The uncertainties discussed above provide part of the impetus for also considering a very conservative pipe break estimate based on the considera-tions in B.2.1 for a bounding calculation as shown in 3.2.1.6.

l NUREG/CR-4594 B-9 June 1986

f(S) h Y

T i Strength Stress 1 f(s) m [ggg) m s ,S (a) Normal Stress and Stre igth Distributions Showing Overlap f( 6) $

Failure Density

-6 0

(b) Difference Distribution Figure B-1 NUREG/CR-4594 B-10 June 1986

l l ~ ~

10-4 l

- a/T = 10% -

10 Demand _ _

Failure Rate

-8 _ -

10 a/T = 7%

-10 _ _

10 f

• /5 = 4%

, I i i 10-12 30 40 50 60 70 5 (ksi)

Figure B-2. Pipe Failure Rates Estimated by the Stress Strength Overlap Method.

NUREG/CR-4594 B-11 June 1986

I APPENDIX B REFERENCES

1. Hubble, W.H. and Miller, C.F., " Data Summaries of Licensee Event Reports of Valves at U.S. Commercial Nuclear Power Plants," NUREG/CR-1363, June 1980.

" Nuclear Power Plant Operating Experience: 1976," NUREG-0366, December 2.

1977.

Beebe, M.R., " Nuclear Power Plant Operating Experience: 1977,"

3.

NUREG-0483, February 1979.

Beebe, M.R., " Nuclear Power Plant Operating Experience: 1978," NUREG 4.

0618, December 1979.

" Nuclear Power Plant Operating Experience: 1979," NUREG/CR-1496.

5.

6. " Nuclear Power Plant Operating Experience: 1980," NUREG/CR-2378.

7. Miller, C.F. et al., " Data Summaries for Licensee Events Reports of Valves at U.S. Commercial Nuclear Power Plants, NUREG/CR-1363, Revision 1, October 1982.

8. Bush, S.H., " Reliability of Piping in Light Water Reactors," Internation-al Atomic Energy Agency, IAEA-SM-218/11, October 1977.

9. " Report of the USNRC Piping Review Committee, Investigation and Evalua-tion of Stress Corrosion Cracking in Piping of Boiling Water Reactor Plants," NUREG-1061, Vol. 1, August 1984

10. " Reactor Safety Study. An Assessment of Accident Risks in U.S. Commer-cial Nuclear Power Plants," NUREG-75/104 (WASH-1400), October 1975.

11. NUS Corporation, " Commercial Nuclear Power Plants," Edition No.12, January 1980.

12. Haugen, E.B., Probabilistic Approaches to Design, John Wiley & Sons, New York, 1968.

13. General Electric Proprietary Report, "Caorso Safety Relief Valve Dis-charge Tests, Phase I Test Report," NEDE-25100-P, May 1979.

14. General Electric Proprietary Report, "Caorso Safety Relief Valve Dis-charge Tests, Phase 11 Test Report," NEDE-24757-P, May 1980.

15. General Electric Proprietary Report, " Mark I Containment Program Final Report, Monticello T-Quencher Test," NEDE-21864-P, July 1978.

16. NUTECH International, " Final Test Report, Safety Relief Valve Discharge Test, Kuosheng Nuclear Power Station, Unit No.1," ZTP-06-310, August 1982.

17. " Safety Evaluation Report, Mark I long Term Program, Resolution of Gener-ic Technical Activity A-7," NUREG-0661, July 1980.

B-12 June 1986 NUREG/CR-4594

18. Attachment to letter from R.H. Buchholz of General Electric to D.G.

Eisenhut of USNRC dated May 2,1980 entitled " General Evaluation of Mark I Safety / Relief Valve Discharge Line Integrity."

19. Attachment to BWR0G Letter No. BWR0G-8511 dated March 13, 1985 from J.M.

Fulton, BWR Owners Group, to R. Wayne Houston, USNRC, entitled "BWR Own-ers Group Response to NRC Request for Information Relating to Generic Issue #61, October 22, 1984."

20. Bari, R.A. et al ., " Reliability of CRBR Primary Piping: Critique of Stress-Strength Overlap Method for Cold-Leg Inlet Downcomer," BNL-NUREG-21642, April 1976.

21. Holman, A. et al ., " Pipe Ruptures in BWR Plants," f rom NUREG/CP-0072, Vol . 3, February 1986.

NUREG/CR-4594 B-13 June 1986

NUREG/CR-4594 B-14 June 1986

Appendix C Estimated Consequences for Steam Bypass Accidents Which Result in Rapid Pressurization of Containment l

NUREG/CR-4594 C-1 June 1986

NUREG/CR-4594 C-2 June 1986

Estimated Consequences for Steam Bypass Accidents Which Result in Rapid Pressurization of Containment As discussed in Section 3.2 of this report there are three steam bypass sequences which may lead to containment pressurization and failure. Contain-ment failure, of itself, does nut cause a core melt but it may disrupt core coolant injection capability or residual heat removal. For these sequences, the pool will be subcooled and will not boil during containment pressuriza-tion. Thus the ECCS pump suction head would not be seriously degraded and the ECC pumps would continue to perform even af ter depressurization.

Since a steam discharge line break in the wetwell airspace and associated containment pressure rise does not prevent the HPCI, RCIC and CRD pumps from operating, the main contributor to core melt, appears to be a large sudden rupture of containment resulting in physical disruption of all coolant injec-tion systems. SNL is performing an uncertainty analysis in support of NUREG-11509 and has used an expert panel to estimate the probability of:

1) Gross containment failure (a rupture area corresponding to 7 square feet or more) given a rapid pressurization transient for a PWR sub-atmospheric containnent.

2) ECC injection failure given a gross rupture of containment.

The experts conclude l the probability of a gross rupture of containment under these conditions to be 0.6 to 1.0. ine probability of ECC injection failure ,

given catastrophic containment failure is estimated to be 0.01 to 0.1 with most of the experts tending to the higher value. Since BWR containments are physically different from PWR containments, the above conclusions do not apply directly, Recent structural analyses 2indicate that the Mark I and Mark II containments' ultimate pressure is about 130 to 190 psia. For the Browns Ferry containment the structural analysis indicates that the failure location will be in the drywell above the drywell floor at the junction of the sphere and cylinder. For Peach Bottom the analysis 2 indicates that the wetwell wall is equally likely to f ail. Even if a gross rupture failure (as opposed to a self-limiting small fracture or leak) occurs in the dome it is unlikely to disrupt all injection lines. However, a failure of the wetwell wall may dis-rupt injection supply lines as well as cause a loss of the primary makeup source (the wetwell pool).

For the Mark 11 containment the conclusions are similar. For Limerick 2 overpressure f ailure (if it occurs) is anticipated in the drywell wall.

Cracks in the concrete structure may develop to relieve pressure buildup and gross rupture failure may not occur. Ilowever for other Mark 11 containments, failure is expected3 to be equally likely at the base of the wetwell pool.

Although cracks will also develop in this case, pressure relief by water leak-ing cut is expected to be insufficient to stop the pressure rise and a severe failure may result. If the failure occurs at the base of the wetwell even without a Severe f ailure, the pool will be displaced and may flood ECC pumps.

We conclude that for scme BWR plants the possibility of ECC f ailure as a direct result of overpressure failure of the containment may be eliminated if a complete structural analysis can demonstrate that the containment f ailure location is remote from essential equipment. However, on a generic basis I

NUREG/CR-4594 C-3 June 1986

failure in the wetwell, disruption of the injection lines, displacement of the pool and/or flooding of the ECC pumps are possibilities that cannot be elimi-nated without detailed plant specific information.

The risk rebaselining effort at SNL has obtained a concensus judgement" on the possibility of a large containment failure given a rapid pressurization of a Mark I containment. Under the high temperatures characteristic of a core melt accident the probability of seal failure and small leaks is expected to be high. But under the relatively low temperature conditions imposed by a steam bypass accident the probability of a large failure in either the wetwell or drywell is estimated to be greater than 80%. The BWR risk rebaselining effort has not addressed the issue of ECC survivability given containment failure. However, a BWR has more injection systems than a PWR and the struc-tural analyses (cited above) indicate that the containment failure is likely to be in the drywell where ECC suction lines and pumps would be relatively unaffected. Thus, there is substantial reason to believe that the survivabil-ity of the BWR ECCS in the event of containment failure is at least as good as a PWR. Butthereisverylittlequantitativedatatosupgorttheextremely low failure probability suggested by the BWR Owners Group (i.e.,0.01).

Based on the engineering consideratior.s outlined above, we believe that for these sequences the ECCS failure probability is no higher than 0.1 and we have adopted this fraction as a conservative estimate of the conditional probabil-ity of ECCS failure. Thus, 90% of the steam discharge line breaks are not expected to result in core melt (not including credit for containment spray or wetwell venting).

For the fraction of the remaining steam discharge line breaks, core melt cannot be ruled out. Although the operator can turn to the Emergency Proce-dure Guidelines (EPGs) and activate core sprays, controlled venting, etc. to minimize pressure buildup, the short calculated time available suggests that (for calculational purposes) a high operator failure probability be assigned (i.e.,10%).

For those cases in which the operator fails to stop the pressure rise in containment and the containment failure causes loss of injection, the decay heat would boil off the remaining primary system coolant inventory. Eventual-ly the core would uncover and core meltdown would occur into a failed contain-ment. The consequences of such a meltdown have been estimated using the new source term methods described in the NRC's source term reassessment program.6 One of the major findings of the NRC study was "that the containment behavior is the most important single factor in determining the source term for many of the accident sequences. If the containment fails early in the sequence and gas flows are high, airborne fission products will be carried from the con-tainment."

Under the present assumed scenario during the core meltdown phase the volatile fission products would be released within the reactor vessel but would escape out the broken steam discharge line without scrubbing. The re-lease fraction for the fission products has been estimated based on calcula-tions for an interfacing LOCA and an ATWS with early containment failure per-formed by BCL7 in support of NUREG-1150. For an interfacing LOCA the fission products released within the vessel also bypass the pool and are released within the reactor building without being scrubbed by the pool .

NUREG/CR-4594 C-4 June 1986

For the present scenario with a postulated break in the steam discharge line the ex-vessel fission product release would also bypass the pool since the containment has already failed. For this portion of the accident the pro-gression is similar to an anticipated transient without scram (ATWS) with an early containment failure. After the core has melted and failed the reactor vessel the new source term methods 6 predict that the core debris will reheat and a substantial fraction of the less volatile fission products will be re-leased during a sustained period of core concrete interaction. Since the pool is assumed to be bypassed during both phases of the present scenario, the re-lease is estimated to be the combination of unscrubbed volatile and non-vola-tile fission products. This estimated release fraction is given in Table C.1 and forms the basis for the following consequence calculations.

The CRAC-2 code 8 was used to calculate offsite consequences for the

" source term" given in Table C.1. The input conditions are summarized in Table C.2. The total health effects for this typical eastern site are found to be 2x107 person-rem over 30 years. Although there is some uncertainty in the source term given in Table C.1 and the health effects may change depending on site specific weather patterns and population, the uncertainty in risk is dominated by the uncertainty in core melt frequency. Thus, no attempt has been made to provide an uncertainty range on the consequence estimates. It would also be inappropriate to obtain worst case consequence calculations and combine them with worst case frequency estimates.

The total risk is estimated using the range of core melt frequencies given7 in Section 3.2 of this report along with the calculated consequence (2x10 person-rem) and is given in Table C.3. Even the upper bound estimates for these three sequences are at most only 5 person-rem / reactor year and appear to have a very small impact on the overall risk of a nuclear power plant.

NUREG/CR-4594 C-5 June 1986

Table C.1. Estimated Environmental Release Fractions for a Core Melt Accident Resulting from a Steani Discharge Line Break' Environmental Species Release Fraction I . ?_8 Cs .24 Te .24 Sr 49 Ru 1x10 6 La .012 Ce 022 Ba .39 NUREG/CR-4594 C-6 June 1986

Table C.2. Summary of Input Assumptions for the CRAC-2 Consequence Calculations for a Core Melt Accident Resulting from Steam Discharge into the Wetwell Airspace Population distribution Uniform

(

Population density 100/sq. mi. ,

Weather conditions Typical eastern wind rose Accumulator period 30 years l

Evacuation delay 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> Warning time 2.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> L

k NUREG/CR-4594 C-7 June 1986

Table C.3. Total Estimated Risk for the Three Sequences Which May Result in Steam Discharge in the Wetwell Airspace

~

Risk (Person-rem /Rx-yr)

Upper Bound  !!est Estimate MSRV Line Break 5.0 <2.0E-6 Turbine Exhaust Line Break 0.16 <.03 Steam Condensing Relief Line Break 0.16 <.08 C-8 June 1986 NUREG/CR-4594

APPENDIX C REFERENCES

1. Sandia National Laboratory, " Limited Latin Hypercube Uncertainty Analy-sis," Appendix A1, Draft Report, February 1986.

2. Griemann, L., Fanous, F. and Bluhm, D., " Containment Analysis Techniques -

A State-of-the-Art Summary," Ames Laboratory, NUREG/CR-3653, March 1984.

3. Science Applications, Inc., " Appendix M: Ultimate Pressure Capacity of Shoreham Primary Containment," SAI-372-83-PA-01, June 1983.

4. Benjamin, A. et al., " Proceedings of the Sandia/NRC Phenominological Issues Meeting," Rockville, MD, February 13 and 14,1986.

5. Attachment to BWROG Letter No. BWR0G-8511 dated March 13, 1985 from J.M.

Fulton,- BWR Owners Group, to R. Wayne Houston, USNRC, entitled "BWR Owners Group Response to NRC Request for Information Relating to Generic Issue

1. 61, October 22, 1984.

6. Silberberg, M. et al., " Reassessment of the Technical Bases for Estimating Source Terms," USNRC Accident Source Term Program Office, NUREG-0956, Draft for Commert. July 1985.

7. Denning, R.S. et al ., BCL, "Radionuclide Release Calculations for Selected Severe Accident k.enarios: Vol .1, BWR, Mark I Design," Draf t Report Sub-mitted to USNRC, November 18, 1985.

8. Ritchie, L.T. et al ., "CRAC2 Model Description," Sandia National Labora-tory, NUREG/CR-2552, March 1984.

9. " Evaluation of Severe Accident Risks and the Potential for Risk Reduc-tion," NUREG-1150 (To be published).

NUREG/CR-4594 C-9 June 1986

i l

NUREG/CR-4594 C-10 June 1986

Appendix D CONTEMPT Modeling for Steam Discharge Line Breaks in the Wetwell Airspace NUREG/CR-4594 D-1 June 1986

i NUREG/CR-4594 D-2 June 1986

CONTEMPT 4 Modeling for Steam Discharge Line Break in the Wetwell Airspace As mentioned in Section 3.3.2 of this report, when comparisons of calcu-lations were made between CONTEMPT-LT/0281 as used in BNL-NUREG-31940 and 2

CONTEMPT 4/M004 used for the current study, several significant differences were found.

One immediate difference occurs in the way heat structure temperatures are initialized in the two codes. Figure D-1 shows a portion of the input for a modeling of the Limerick containment using CONTEMPT-LT/028.1 Figure D-2 shows a similar section of the input for the same containment using CONTEMPT 4/

M004.2 By comparing the two figures, one can see that wetwell, drywell and outside air conditions are specified identically. Heat structure material properties are also the same in both. Nodalization is somewhat different in the two inputs but points are spaced closely enough so as not to affect the results in either. Figure D-3 shows the initial temperature gradient through the six heat structures as calculated by CONTEMPT-LT/028.1 Figure D-4 indi-cates the initial temperature distribution in heat structures 1 through 6 as calculated by CONTEMPT 4/M004.2 As these last two figures show, several heat structures show slightly different temperatures between the two codes but structures 4 and 6 show very noticeably different gradients.

Closer inspection of the codes revealed the reason for the difference.

While the user inputs bulk temperature conditions for both codes, the codes calculate heat structure surface and interior temperatures based on the heat transfer conditions. If the Uchida ,2 heat transfer condition is specified l

(as it was in both codes for the cases considered), CONTEMPT-LT/028 1 uses only 2

Uchida to calculate initial temperatures whereas CONTEMPT 4/M004 will check compartment conditions and will use Uchida or natural convection heat transfer as conditions indicate. For all subsequent time steps compartment conditions are checked by both codes to determine whether Uchida or natural convection should be used, but for initialization CONTEMPT-LT/028 1does not make the check while CONTEMPT 4/M0042 does. This leads to a difference in initial teta-perature distributions which may allow the same structure to absorb more energy in a model using one code than in a model using the other. For the cases considered in this report, subsequent checks showed that these initiali-zation differences did not cause major differences in predicted results.

Far more impact on the predictions of the two codes regarding containment pressures and temperatures was caused by the different pool surface condensa-tion rates found in the two codes. As discussed in Section 3.3.2 of this report, pool surface condensation rates are much higher in CONTEMPT 4/M004 2 than in CONTEMPT-LT/028.1 The following table shows just how different the rates of vapor condensation on the suppression pool surface are. Results are for Limerick using the mass and energy releases used in BNL-NUREG-31940.

I NUREG/CR-4594 D-3 June 1986

Condensation Rate (lbs/sec)

Time (sec) on Pool Surf ace at Time Indicated 2

CONTEMPT-LT/028 1 CONTEMPT 4/ MOD 4 100 0.47 6.07 300 2.02 29.0 600 4.34 57.3 800 5.66 69.9 2

The CONTEMPT 4/M004 rates appear to be more accurate for the code calcu-lated pool and atmosphere conditions. The pool is 95.0 F throughout, includ-ing its surface, at 100 seconds and only rises to 99.8 F at 800 seconds since the hot condensate is mixed uniformly into the existing cold reservoir. Mean-while the wetwell atmosphere conditions change from 362 F and 43.7 psia at 100 seconds to 461'F and 143 psia at 800 seconds (CONTEMPT 4 values). As pointed out in Section 3.3.2 this single node pool model with its surface temperature equal to average pool temperature is not accurate for these conditions since a hot layer of condensate is expected to form on the pool surface and thereby decrease the pool to atmosphere AT and reduce subsequent condensation. The dramatic difference such a layer can make, is qualitatively illustrated by the CONTEMPT 4/ 2MOD 4 results in the drywell where a pool of hot condensate forms.

At 800 seconds this condensate liquid has a temperature of 326 F while the at-mosphere is at 472*F and 142 psia, similar to the wetwell atmosphere condi-tions. The surface condensation on this hot pool is only 0.35 lbs/sec however compared to 69.9 lbs/sec on the surface of the cold (99.8 F) suppression pool.

In reality, the condensate on the suppression pool would be cooled from below and so would have a lower temperature than a pool of condensate only, in the drywell . However the surface layer temperature of the suppression pool would be well above the average pool temperature. The dramatic effect pool surface condensation has on the containment pressure and temperature is indicated by some of the figures refarred to in Section 3.3.2 and 3.3.3 of this report.

While other minor differences between CONTEMPT-LT0281 and CONTEMPT 4/M004 2 exist, they were not found to have significant influence on the calculations done for this report.

l NUREG/CR-4594 D-4 June 1986

BRIT 1 UNITS USED FOR IPFUT Am OUTPUT WHERE UNITS ARE UNSPECIFIED.

E m PROBLEM ElO TIME = 1.000000E+00 fftS NO. HEAT STRUCTLRES = 6 PRESSURE SUPPRSSION OPT. = 1 g OUTSIDE AIR TEMPERATURE = 8.000000E +0.1 F PRESSURE = 1.470000E401 PSIA HJ4 tDITY = 5.000000E-01 y CONSTANT TEMP. FOR EAT SLABS = 0. F STEP. WATER TO [RY WELL = 0. LBM , WITH STEP ENERGY = 0. BTU g PRIMARY SYSTEM END-OF-BLOWOOWN WATER CONTENT = 0 LEM,WITH ENERGY = 0 BTU y CALCULATE EVAPORATION OR CONDENSATION RATE BETWEEN ATMOSPHERE AND POOL DURING BLOWDOWN.

PROBLEM EXECUTION STOPS WHEN TOTAL PRESSURE EXCEEDS 154.700 PSI A.

SPRAY-ECX; HEAT EXCHANGER, NO. I TYPE = 0 EAT TRANSFER AREA = 0. OVERALL H.T. COEFF. = 0. COOLANT INLET TEMP. = 0 INLET MASS FLOW = 0 PRESSURE FOR SPRAY ON AND OFF = 0. , O.

SPRAY-ECC EAT EXCHANbER, NO. 2 TYPE = 0 HEAT TRANSFER AREA = 0 OVERALL H.T. COEFF. = 0 COOLANT INLET TEMP. = 0 INLET MASS FLOW = 0.

PRESSURE FOR SPRAY ON AND OFF = 0 ,O.

?

ui C04P. = 2 VOL. = 2.800000D+05 LIQ.VOL = 1.186000E405 VAPOR YOL. = 1.614000D+05 HtMIDITY = 1.0000 TOTAL PRESSURE = 1 .5450000401 VAPOR TEMPERATURE = 9.5000000 +01 LIQ. TEMP. = 9.500000D+01 SURF. AREA = 5.700000D+03 HEAT TRANS. MULT. = 1.000000D+00 MASS TRANS. MULT. = 1.0000000400 C04P. = 3 VOL. = 2.4840000405 LIQ.VOL = 0 VAPOR VOL. = 2.484000D405 HlMIDITY = .2000 TOTAL PRESSURE = 1 .5450000401 VAPOR TEMPERATURE = 1.500000D+02 LIQ. TEMP. = 1.5000000+02 SURF. AREA = 5.2660000+03 HEAT TRANS. MULT. = 1.000000D+00 MASS TRANS. MULT. = 1.000000D+00 DRYWELL TEMPERATURE FLASH MODEL SELECTED.

ARBITRARY AIR ADDITION TABLE, TIME, AIR ADDED, TEMP.

O. O. O. 3.600000E409 0 O.

OUTSIDE AIR CONDITIONS TABLE, TIME, TEMP., HEAT TRANSFER COEF. (24144. CYCLE)

O. 8.000000E+01 1.000000E+00 2.400000E401 8.000000E+01 1.000000E+00 c

5 THEf44AL CONDUCTIVITY AIO YOLLMETRIC EAT CAPACITY TABLE

• C04 POSIT lON NO., THE144AL CONDUCTIVITY, HEAT CAPAC1TY G 1 2.900000E+01 5.300000E401

$ 2 1.052300E+00 2.341500E401 Figuro D-1. CONTEMPT-LT/028 Input for Limerick Containment.

I 1

l

....... 3* 3*

s:  : ._ I I g,

c, 3

'  : 8 8 8 b:  : IE-og- .

! :  : * * = ,

g::  : ge o l

W 8 8 8

: I -( 3. 8. 3.

: tf - . _

: :s *

: . I 3 8 3 #

j j i} E. 8. 8. 5

.  : =- - - -

E

:. -r c s -

. 1 e r r m i i e-I 5!58'd #

o

: 5- , , u i

.a

I tctct- u a

i - w~E -s- sa i g g

.  :. a- '

a3 i i aavsaa  ; e

! &>v ss f f a 3

.  :  : g , , , M saa e i !

[; e

-r f -

~8

~8

-8 h

2 h

O S ses . . a i !:.-r

't i *

! s ers

! s -< g.

- s c gi :  : 5 eses  !:vi

-=  : e

.g

.  : t .s  ! s a  !-:: o av o

-:: =

r

s ,

• y-

[5:: *

=* :

o I.s ass r

s. :  : e c_

re .  : ;f- -

II z

_- - - - w gi:

g5 j 8 8  ? sgI--- @

x :  : 4 st,:- I I a :ly f o o

f:

s:  :

- 9 g l .

8; t:  : - - or g* i

, i t' -

A E 8

Id m

[tg - i o

y~ i i ih - ~ - . :-

• i as

t. - -

.i. t.

i; p

u l

l NUREG/CR-4594 D-6 June 1986

' THE FOLLOWING DEFINITIONS WILL BE USED AT TIMES FOR COMPARTMENT IDENTIFICATION 0 = OUTSIDE ATMOSPHERE 1 = PRIMARY SYSTEM -

'2 = WET WELL

'3 = ORY WELL 4 = ANNULAR COMPARTMENT HEAT STRUCTURE 1 BETWEEN COMPARTMENTS 2 AND 0 UPPER SURFACE OF WETWELL (2 TO 0) iTO LEFT COMP. END STEP BTU /HR- . STEP BTU TOTAL NET BTU TO RIGHT COMP. ENO STEP BTU /HR STEP BTU TOTAL NET BTU 2 VAPOR -2.0129460+04- O. O. . O VAPOR 2.0129460+04 0. O.

LEFT FILM COEF. H = 6.232125E+00 BTU /HR.FT2.F FIRST MESH K = 2.900000E+01 BTU /HR.FT. BULK TEMP = 9.500000E+01 F RIGHT FILM COEF. , H = 1.000000E+00 BTU /HR.FT2.F LAST MESH K = 1.052300E+00 BTU /HR.FT.F BULK TEMP = 8.0000000+01 F

. MESH POINT TEMPERATURES (F), LEFT TO RIGHT 9.4622940+01 9.4622100+01 9.4621260+01 9.4597940+01 9.4574620+01 9.4507710+01 9.4440850+01 9.4374040+01 9.4307270+01:

9.4240540+01 9.4173870+01 9.4107230+01 9.4040640+01 9.3974100+01 9.3907600+01 9.3708380+01 9.3509550+01 9.3311130+01

'9.3113110+01 9.2915490+01 9.2718270+01 9.2521440+01 9.2325000+01 9.2128960+01 9.1933310+01 9.1499910+01 9.1068420+01 9.0638820+01 9.0211090+01 8.97d5220+01 8.9361180+01 8.8938960+01 8.8518560+01 8.8099940+01 8.7683100+01 8.7106810+01

- G.6533880+01 8.5964270+01 8.5397940+01 8.4834860+01 8.4275000+01 8.3718300+01 8.3164740+01 8.2614290+01 8.2066900+01 HEAT STRUCTURE 2 BETWEEN COMPARTMENTS 2 AND 2 CONCRETE STRUCTURE IN WETWELL (2 TO 2)

TO LEFT COMP. END STEP BTU /HR STEP BTU TOTAL NET BTU TO RIGHT COMP. END STEP BTU /HR STEP BTU TOTAL NET BTU 2 VAPOR -4.0241620-08 0. O. 2 VAPOR 0. O. O.

LEFT FILM COEF. . H = 6.232125E+008TU/HR.FT2.F- FIRST MESH K = 1.052300E+00 BTU /HR.FT. BULK TEMP = 9.500000E+01 F RIGHT FILM COEF. . H = 6.232125E+00 BTU /HR.FT2.F LAST MESH K = 1.052300E+00 BTU /HR.FT.F BULK TEMP = 9.5000000+01 F MESH POINT TEMPERATURES (F), LEFT TO RIGHT 9.5000000+01 9.5000000+01 9.5000000+01 9.5000000+01 9.5000000+01 9.5000000+01 9.5000000+01 9.5000000+01 9.5000000+01 9.5000000+01 9.5000000+01 HEAT STRUCTURE 3 BETWEEN COMPARTMENTS 3 AND'2 00WNCOMERS (3 TO 2) cTO LEFT COMP. END STEP BTU /HR STEP BTU TOTAL NET BTU TO RIGHT COMP. END STEP BTU /HR STEP BTU TOTAL NET BTU 3 VAPOR -2.7127460+06 0. O. 2 VAPOR 2.7127460+06 0. O.

LEFT FILM COEF. , H = 5.671260E+00 BTU /HR.FT2.F FIRST MESH K = 2.900000E+01 BTU /HR.FT.- BULK TEMP = 1.500000E+02 F RIGHT FILM COEF. . H = 6.232125E+00 BTU /HR.FT2.F LAST MESH K = 2.900000E+01 BTU /HR.FT.F BULK TEMP = 9.5000000+01 F MESH POINT TEMPERATURES (F), LEFT TO RIGHT 1.2086220+02 1.2084440+02 1.2082670+02 1.2080900+02 1.2079140+02 1.2077390+02 1.2075640+02 1.2073890+02 1.2072150+02 1.2070420+02 1.2068690+02

' HEAT STRUCTURE 4 BETWEEN COMPARTMENTS 3 AND 2 O!APHRAGM SLAB (3 TO 2) 70 LEFT COMP. ENO STEP BTU /HR STEP BTU TOTAL NET BTU TO RIGHT COMP. END STEP BTU /HR STEP BTU TOTAL NET BTU 3 VAPOR -8.3695330+04 0. O. 2 VAPOR - 8.3695330+04 0. O.

'LEFT FILM COEF. , H = 5.671260E+00 BTU /HR.FT2.F FIRST MESH K = 2.900000E+01 BTU /HR.FT. BULK TEMP = 1.500000E+02 F RIGHT FILM COEF. . H = 6.232125E+00 BTU /HR.FT2.F LAST MESH K = 1.052300E+00 BTU /HR.FT.F BULK TEMP = 9.5000000+01 F MESH POINT TEMPERATURES (F). LEFT TO RIGHT 1.4735280+02 1.4734740+02 1.4734210+02 1.4719350+02 1.4704490+02 1.4661690+02 1.4618890+02 1.4576090+02 1.4533290+02 1.4490490+02 1.4447690+02 1.4404890+02 1.4362090+02 1.4319290+02 1.4276490+02 1.4148090+02 1.4019690+02 1.3891290+02 1.3762890+02 1.3634490+02 1.3506090+02 1.3377690+02 1.3249290+02 1.3120890+02 1.2992500+02 1.2667330+02 1.2342170+02 1.201701D+02 1.1691850+02~ 1.1366690+02 1.1041530+02 1.0716370+02 1.0391210+02 1.0066050+02 9.7408930+Ci HEAT STRUCTURE 5 BETWEEN C0d ARTMENTS 3 AND 3 CONCRETE STRUCTURE IN DU WELL (3 TO 3)

TO LEFT COMP. END STEP BTU /HR ' STEP BTU TOTAL NET BTU TO RIGHT COMP. ENO STEP BTU /HR STEP BTU TOTAL NET BTU 3 VAPOR 3.1989380-08 0. O. 3 VAPOR 4.5680200-08 0. O.

LEFT FILM COEF. . H = 5.671260E+00 BTU /HR.FT2.F FIRST MESH K = 1.052300E+00 BTU /HR.FT. BULK TEMP = 1.500000E+02 F RIGHT FILM COEF.', H = 5.671260E+00 BTU /HR.FT2.F LAST MESH K = 1.052300E+00 BTU /HR.FT.F BULK TEMP = 1.5000000+02 F MESH POINT TEMPERATURES (F). LEFT TO RIGHT 1.5000000+02 1.5000000+02 1.5000000+02 1.5000000+02 1.5000000+02 1.5000000+02 1.5000000+02 1.5000000+02 1.5000000+02 1.5000000+02 1.5000000+02 HEAT STRUCTURE 6 BETWEEN COMPARTMENTS 3 AND 0 DRYWELL WALLS (3 TO 0)

STEP BTU TOTAL NET BTU TO RIGHT COMP. END STEP BTU /HR STEP BTU TOTAL NET BTU VO LEFT COMP. END STEP BTU /HR 3 VAPOR -1.7480960+05 0. O. O VAPOR 1.7480960+05 0. O.

LIFT FILM COEF. H = 5.671260E+00 BTU /HR.FT2.F FIRST MESH K = 2.900000E+01 BTU /HR.FT. BULK TEMP = 1.500000E+02 F RIGHT FILM COEF. , H = 1.000000E+00 BTU /HR.FT2.F LAST MESH K = 1.052300E+00 BTU /HR.FT.F BULK TEMP = 8.0010000+01 F MES'8 POINT TEMPERATURES (F), LEFT TO RIGHT

'i.4824610+02 1.4824250+02. 1.4823890+02 1.4814050+02 1.4804200+02 1.4775840+02 1.4747480+02 1.4719130+02 1.4690770+02 1.4662410+02 1.4634050+02 1.4605690+02 1.4577340+02 1.4548980+02 1.4520620+02 1.4435540+02 1.4350470+02 1.4265400+02 1.4180320+02 1.4095250+02 1.4010170+02 1.3925100+02 1.3840020+02 1.3754950+02 1.3669870+02 1.3480820+02 1.3291770+02 1.3102710+02 1.2913660+02 1.2724600+02 1.2535550+02 1.2346490+02 1.2157440+02 1.196a380+02 1.1779330+02 1.1500870+02 1.1222410+02 1.0943940+02 1.0665480+02 1.0387020+02 1.0108560+02 9.8300960+01 9.5516340+01 9.2731720+01 8.9941100+01 Figure D-3. CONTEMPT-LT/028 Heat Structure Initial Temperatures.

NUREG/CR-4594 D-7 June 1986

9 8

!*s 3[! ggy fife III tij 4:s ssi ESE8  !!! 6?i

!! i i I I I i'Ej!Sijji 5 Ei 8Ii !Illi E!Ei J ES!* 2888 558 2 4:s ssss o I f %u

?

- I I E

'jjj8ljlg 2- EliBI EIIII ESI8 g 28!! SEE g s:s ssss E!E8 r

~

S .i f I I. S -

'[j!Sijjj5 EHR3 EII!! Effi 3 aus ssss E5E8  !!!8 635 g i i  ?" I I $

'I!jSjjjg 5 E EEEE !ilil E IR) h a:s ssis E518 8888 !?5 t=

I E

i i  ?! i

{E 'jjj 4:n ssisSjjjj5 Ellii E"!!

Elis5 I E???  !!!8 6E? x Z

gs , , ,, , ,

o

-E I s1 1 1

3s *!

1 gig ir i!!5 3i555 i 35EE85 3ffli g I;fasifirssI ge r fe?!!!

e e 18!!8 r 5555= r o-

s. s s s s s 5 5

ri i i_-__ f i,.., i i- e SF35!Isilli3,>.335,- s IIII 83I3 5

}; rass vssss r5 rig!! rgg82 rEES u

. t n - a a t 1l:

g, -

e r

a

!)! I  !!  ! I e

i. 7 .

? V

. .V V y

V h,

[

NUREG/CR-4594 June 1986 D-8

APPENDIX D REFERENCES

1. Hargroves, D.W. and Metcalfe, L.J., " CONTEMPT-LT/028, A Computer Program for Predicting Containment Pressure-Temperature Response to a Loss-of-Coolant Accident," NUREG/CR-0255, March 1979,

2. Lin, C.C. et al., " CONTEMPT 4/ MOD 4 A Multicompartment Containment System Analysis Program," NUREG/CR-3726, March 1984.

T l

b NUREG/CR-4594 D-9 June 1986

NRCPORM3JS U S. huCLE A3 E EGULATctr.Y COMMIS$liN Rtromi NUMBEFs #4 ss ,a.a e, rioc, .ar ver ee. .r enys 3E/' BIBLIOGRAPHIC DATA SHEET G 5 986 SEE INSTf:uCTIONS ON T E REVER$t 2 YITLE AND 5uefiTLE 3 LE AVE SLANet

" Estimated Safety Significance of Generic Safety Issun 61" /

[ 4 C ATE REPORT COMPLETED

, EAR f.ONT. l

. AuY RO;,s, Jup 1986 J.R. Lehner, K.R. ins and C. Economos I + oa'5 at'oa' '55um MONT. vtAR l

7 PE~FORasNG ORGAN 12AYiON NAME AND MA ING ADDRESS (farfue le Com/

/ une ROJECTIT ASK/ WORK UNIT NUMBER 1986 Department of Nuclear En gy r' ' " ' " ' " ' ' ' " " * * * "

Brookhaven National Labor ory William Floyd Parkway Upton, New York 11973 FIN A-3793 to Sv0N50 RING ORGANil ATION NAME AND MAILING AO E55 (sactue le Conf lia TYPE OF REPORT Division of Safety Review and ersight -

Final Office of Nuclear Reactor Regul ion , , , ,c, ,,, ,, _

U.S. Nuclear Regulatory Commissi Washington, DC 20555 n sumEMEu AR, NOTES

" '""$["p'o~Ie*Et'i al threat posed by tra ient i iiated accident sequences involving BWR systems capable of releasing steam into e wet 11 airspace and thereby pressurizing the containment has been examined. This study ti es the likelihood of a rupture in one of a number of high pressure steam lines which s through the wetwell airspace before enter-ing the suppression pool. If the broken stea ine is connected to an active steam source, such as a stuck open relief valve, which sup 'es steam at a high enough rate and over a long enough interval, the suppression pool p may result in containment overpressure failure. The three BWR plant systems ident ie s having components whose failure could lead to a steam discharge into the wetwell 'irspac and pressurization of the containment are: the Main Steam Relief Valves and ass iated d charge lines, the High Pressure Cool-ant Injection turbine exhaust and the ste condensi relief lines in the Residual Heat Removal System. This study outlines th postulated cident sequences, estimates their frequency, and calculates containment re onse to the p osed steam discharges using the computer code CONTEMPT 4. Based on the redicted contai ent response, the investigation estimates the probability of core melt, e possible fissi product release and the asso-ciated consequences. Mitigating action . designed to prevent ntainment failure as well as actions designed to prevent core melt g en containment failu are also discussed.

The study concludes that, while s e of the sequences coul potentially result in core melt and cause significant releases of fission products, the f uency of these sequences are judged to be sufficiently small remove them as significa contributors to public ri14 @ .

DOCwMENT ANALY5tS - e et E YWORDS DE SCRiPTOR5 15 AV A6LAssLIT y STATEMENT GSI 61, SRV break, wetwell airspace, R, containment pressurization core damage, CONTEMPT 4, transient, H I, steam condensing mode Unlimited i

16 SECURITv CLASSIFICATION 9 <r...,

m iOENTinERs:0,iNENotoTERus j Unclassified j o r:. . .m

$ Unclassified 17 NUMSER OF PAGES IB PRICE

UNITE] STATES -sescauouarwetass nave NUCLEA] RE"_'ULATORY COMMISSION- postagg,7s raio WASHINGTON, D.C. 20566 w

,,Asyg, OFFICIAL BUSINESS -

PENALTY FOR PRIVATE USE, $300 120555078877 1 1ANIAI11S19Al US NRC ADM-DIV 0F TIDC POLICY & PUB M GT BR 3DR NUREG

, W-501 W ASHI ri3 TON DC 20555 N.