ML20195G723
| ML20195G723 | |
| Person / Time | |
|---|---|
| Issue date: | 11/02/1987 |
| From: | Murley T Office of Nuclear Reactor Regulation |
| To: | Jordan E NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| Shared Package | |
| ML20154D839 | List: |
| References | |
| REF-GTECI-093, REF-GTECI-NI, TASK-093, TASK-93, TASK-OR IEB-85-001, IEB-85-1, NUDOCS 8711100236 | |
| Download: ML20195G723 (3) | |
Text
{{#Wiki_filter:, / NOV 9 1997 PEWORANnUV FOR: Edward L. Jordan, Director Office for Analysis and Evaluation of Operational Data FROM: Thomas E. Murley, Director Office of Nuclear Reactor Regulation SlN ECT: ISSUANCE OF GENERIC LETTER TO RESOLVE GENERIC ISSUE 43, "STEAM BINDING OF AFW PUSD5" Generic Issue 93 on steam binding of AFW pumps due to backleakage from the main feedwater system has been resolved and no additional requirements bevond those already soecified in IE Bulletin 85-01 need he imposed on the licensees or construction remit holders of pressurized water reactor plants. We intend to issue the attached generic letter, which indefinite 1v extends the monitoring requirements specified in the bulletin to keep the risk of steam hinding acceptably low. The bulletin was reviewed by the Comittee to Review Generic Renuirerents (CGGD) in Septenber 1985. We do not believe further forral review by the CRGR is required on this issue. We request that you affim this position within 10 days. Or IM31 siined by hmes H. Sn;enk /;FThomasE.Murley, Director Office of Nuclear Reactor Regulation
Enclosure:
Generic Letter 87-XX CONTACTS: Alfred H. Spano, RES 492-7476 C. Vernon Hodge, NRD 49?-8105 DISTRIRUTION TEkurlev. M R JHSnierek, NRR RWStarostecki, NRR CERossi, NRR CWRerlinger: NRR CVHodge, NRR EJ99tcher, NDD WDLanning, NQD Central Files NS DOEA R/F OGCB R/F VHodge R/F e t ASpano, 4ES j y ~ Ay D/NRP p TEMurlev O 10/ /87 r i OGCR:DOEA':ND R C/0GCB:00EA:NRR M., AD / CYHodge CHRerlinger ERossi RWStarostecki erek i 10/;3 /87 10/p/87 10//3 /87 10/d/87 /87
- .,f gas
- 7 %'
,&e t b S w ha-<. [it'r' t. s XA "" M K!?'/'l L
UNITED sT ATEs >s ta ggIo NUCLE AR REGULATORY COMMISSION f[1 # 'g nsmNGTON,0 C 20555 s, e %.~.~...f TO ALL LICENSEFS, AFPLICANTS FOR OPERATING LICENSES, AND HOLD CONSTRUCTION PEPMITS FOR PRESSUR17ED WATER PEACTORS Gentlemen: RESOLUTION OF GENERIC SAFETY ISSUE 93, "STEAM BINDING OF AUX 1LIAPY FEE 0 WATER PUFPS" (GENERIC LETTER 87-XX)
SUBJECT:
The purpose of this generic letter is to infom ycu of the staff's resolution iliary of the subject s&fety issue which concerns the potential disabling of aux feedwater (AFW) Durps by steam binding caused by backleakage of main The (MFW) past the isniation check valves interfacing the AFW and MFW syste significance of the issue stems from the potential vulnerability of mo systens to connon modt failure of the redundant AFW pumps due to l This vulnerability is inherent to the pipirg configurations used, which al ow for redundant trains cf AFW to be cross-connected via com suction and discharge sides of the pumps. l To reduce the probability of AFW pump failure as a consequence of ste if backleakage does occur IE Bulletin 85-01 (dated October ?9,1985) re 1 those licensees and CP holdees, wtio had not already done so in response to d s previous NRC comunicat!cns and INPO recocrendction backleakage and for restoring the pumps to operable status shift, compared with the previous once per month check were to occue. in p ep uravailability due to steam binding. ( was made of the backleakage experience in opera f The systentic monitoring (about April 1985 for a majority of the plants). h the results showed a double-humped distribution in regar (a da mean of less than one per reactor year) while the reinaining plants showeNone far higher event rate by a factor of roughly 100. I events during the review period appeared to have resulted in steam bi an AFV pump, indicating that the various monitoring methods emnloyed c highly effective in providing for the prevention of steam binding if CONTACTS: Alfred H. Spano, RFS (301) 492-7476 C. Vernon Hodge, NRR (301) 492-8196
-?- Multiple Addressees [ backleakage occurs. For the plants with a high bacileakage event rate, the l installation of continuous monitoring systems with control room alarms was irstrumental in providing ior early warning to the operator and timely correc-tive action. The results of the regulatory analysis indicated that, within tne framework of the existing Bulletin 85-01 requirements, the contribution of AFW pump steam binding to core melt frequency and public risk was sufficiently les to warrant the finding that no new requirements beyond those specified in IE Bulletin Therefore, the staff has concludeo that the monitoring 85-01 need be imposed. requirements of the bulletin must be continued to keep the risk low. Although the staff has concluded that the currently assessed "isk associated with this issue is reasonably low, the risk is nevertheless considered important in light of the generally unsatisfactory check valve reliability experience obtained to date in operating plants. Plant operators nust continue to be alert to the pos:ible development of nalfunctionirg check valves, especially as the plant ages, and must be prepared to increase the monitor *ng frequency as needed to ensure that stean binding failure of the AFW pumps does not occur. No written response or specific action is required by this letter; therefore, no clearance from the Office of Panegerent and Budget is required. If you have any questions on this matter, please contact your project manager. Sincerely. Frank J. Miraglia, Associate Director for Proiects Office of Nuclear Reactor Regulation l w
Original: Sheron cys: Beckjord UNITED STATES [ NUCLEAR REGULATORY COMMISSION ginners ~ g s ) nAsmoTon, o, c. mes Kniel Mazeus AUG 14 g M pano MTHORANDUM FOR: Eric S. Beckjord, Director Office of Nuclear Regulatory Research FROM: Thomas E. Murley, Director Office of Nuclear Reactor Regulation
SUBJECT:
RESOLUTION OF GENERIC SAFETY ISSUE 93, "STEAli BINDING OF AUXILIARY FEE 0 WATER PUMPS" This necerandun is in response to your request, included ia nemoranda of May 1, 1987 and June 2E, 1987, that hRC Inspection Procedures be modified to include monitoring of Auxiliary Feedwater Systen (fFWS) backleakage and that Licensing Project Panagers ensure that the annuel licensee upcetes of FSARs include decur.entation cf ar.y added AFWS backleakage monitorinc systes. IE Bulletin 85-01, i:; sued Octcber 29, 1985, requirJJ that licensees develop and implement procecures covering the sub,iects nf perirdic menitoring of AFWS temperatures, recognizir.g steam binding, and recouring f rom a steam binding event. Temocrary Instruction 2515/69 issued December 20, 1985 (expired April 30, 1957) defined the inspection requirements to verify that licensee personnel understood and were using the new procedures. The generic letter ) discussed in your May 1, 1987 menorandum, requires licensees to centinue to implement the precedures required by IE Bulletin 85-01 on a routine basis. g .I Current NRC Inspection Procedures (IPs) require periodic review of licensee acherence to appreved procecures and surveillance tests. Inspection Procedure 71707, "Operational Safety Yerification," which is part of the minimum er basie j hRC Light-Uster Reacter Inspection Program for the Operations Phase, provides ( cuidar.ce that the auxiliary feedwater system should be includec as part of the periodic operability verification of engineered safety features (ESF) trains j l (71707-03c). Additionally, the precedural requirements for inspection of ESF systems require weekly inspection by the resident inspector of any general ccr.dition that enight prevent fulfillment of their functional requirements. On this basis, we believe there is no need to add a new spscific inspection requir nent for inspection of AFWS pumps for steam binding. However, we will l l revise IP 71707-03c to include the matter of monitoring the AFNS pumps for steem binding as an example of a recurring operational event that should be periodically checked by the NRC inspectors. Any changes affecting the FSAR are required to be submitted by the licensee to the NRC by 10 CFR 50.71. AFVS backleakage monitoring systems which meet the recuirements of 10 CFR 50.71 would be covered by this requirement. We do MP f09*W N. d __....m.
I Eric 5. Beckjord AUG 141987 not consider it appropriate to initiate a special requirement for review cf such licensee submittals to determine if the licensee's program for AFWS backleakage monitoring systems is includeo. It is our belief that the current licensing and inspection requirements, particularly after the revision to IP 71707 guidance, will be sufficient to provide adequate assurance that licensees continue to address steam binding of Auxiliary Feedwater Pumps. Yd" Thomas E. Murley',' Director Office of Nuclear Re6ctor Regulation i i i f
l y[p u%( l UNITED STATES o, NUCLEAR REGULATORY COMMISSION g wassiaotoN. o. c. rosss f j JUN 2 61997 r MEMORANDUM FOR; Thomas E. Murley, Director Office of Nuclear Reactor Regulation FROM: Eric 5. Beckjord, Director Office of Nuclear Regulatory Research
SUBJECT:
RE50'.UTION OF GENERIC SAFETY ISSUE 93, "STEAM BINDING OF AUXILIARY FEE 0 WATER PUMPS" Your memorandum of June 10, 1987, concurring in the proposed resolution of Generic Issue 93, suggested that the draft generic letter be revised to delete the paragraph permitting a relaxation in the check valve'backleakage monitoring frequency. We concur in the letter as revised. We also wish to repeat our original request that inspection modules be modified to include monitoring of AFWS backleakage and that project managers ensure that the annual FSAR updates include documentation of any added AFWS backleakage monitoring systems. It was also indicated that the modified generic letter be forwarded as part of the CRGR submittal to close out the issue. As we have discussed with your staff, the generic letter is informational in character and the resolution of i the issue involves no new requirements; CRGR consideration is not necessary, i Accordingly, this issue is resolved and only the issuance of the letter by HRR to licensees and applicants is needed to implement the resolution of the issue. EricS.BeckjordDirector l Office of Nuclear Regulatory Research ~ 1 cc: J. Sniezek F. Miraglia R. Starostecki i J Blaha J. Funches F. Hebdon l ir ,z,,n ,a- ,4 l Qi
g68#fC, + c, UNITED STATES j' s y,(/( j NUCLE AR REGULATORY COMMISSION c W e usmNotoN. o. c. rosss 1, dI June 10, 1987 MEM0FANDUM FOR: Eric S. Beckjord, Director Office of Nuclear Regulatory Research FROM: Thomas E. Murley, Director Office of Nuclear Reactor Regulation
SUBJECT:
RESOLUTION OF GENERIC SAFETY ISSUE 93 "STEAM BINDING OF AUXILIARY FEE 0 WATER PUMP!" Your memorandum of May 1,1987 proposed resolution to Generic Issue 93 "Steam Binding of Auxiliary Feedwater Pumps". We concur with the proposed resolution.. We have also revised the generic letter on this subject to delete the paragraph which pemitted a relaxation in the check valve backleakage monitoring frequency. Given the persister.t problems with check valve leakage, particularly as the valves degrade with age, we see no reason to deviate from tht current check valve backleakage monitoring frequency that has been implemented by licensees in response to IE Eulletin 85-01. The present frecuency.(once per shift) does not impose a significant burden on the licensees, and acceptably ersures that i potential auxiliary feedwater pump steam binding will be detected and corrected in a timely manner. We propose this nodified generic letter be forwarded as part of the CRGR submittal to close out generic safety issue 93. /lM%d Aub 'L mas E. Murley, Dir cto Of ice of Nuclear Reactor Regulation
Enclosure:
1 As stated cc: F. Miraglia J. Blaha i J. Funches F. Hebdon 1
Contact:
W. LeFave X27798 4 i 'ff ,,m. j y \\
"? ENCLOSURE TO ALL LICENSEES, APPLICANTS FOR OPERATING t.1 CENSES AND HOLDERS OF CONSTRUCTION PERMITS FOR PRESSURIZED WATER REACTORS Gentlemen:
SUBJECT:
RESOLUTION OF GENERIC SAFETY ISSUE 93, "STEAM BINDING OF AUXILIARY FEEDWATER PUMPS" This letter is to infore you of the staff's resolu' ion of the subject safety t pumps by stean binding caused by backleakage of main feedwater (MF issue which concerns the potential disabling of auxiliary feedwater past the isolation check valves interfacing the AFW and MFW systems. The significance of the issue stems from the potential vulnerability of most AFW systems to cow en mode steam binding failure of the redundant AFW pumps, this vulner-l ability being inherent in the piping configurations used, which allow for crest connections between trains via comon headers on the suction and discharge sides of the pumps. To reduce the probability (of pu p failure by) steer binding i' beckleaksge does occur, IE Bulletin BE-01 dated October 1985 required those licersees and CP holders, who had not alreaty done so in response to previous NRC and INPO recorrendations, to implement procedures both for monitoring the AFW piping teeperatures as an indication of possible backleakage anc for restoring the punps to operable status if steam binding were to occur or aopeared likely to do so. The Bulletin's recomended ronitoring frequency of once per shif t, 1 compared with the previous once per month check when the pumps were tested, provided for a factor of about 90 (3 shifts / day x30 days / month) reduction ir purr average unavailability due to steam binding. l 4 As a basis for the staff's regulatory analysis of this issue, a limited review l j was rade of the backleakage experience in operating plants since the start of I l systematic monitoring (about April 1985 for a majority of the plants). The results showed a double-hurped distribution in recard to backleakage, with the l )l dominant majority of plants showing low incidence of backleakage occurrences ] (a mean of less than one per reactor year) while the remaining plants showed l a far higher event rate by a factor of roughly 100. None of the backleakage events during the review period appeared to have resulted in steam binding of an AFW pump, indicatine that the various monitoring methods employed can be highly effective in providing for the prevention of steam binding if backleakage ~ occurs. For the plants with a high backleakage event rate, the installation of continuous monitoring systers with control room alarrs was instrurental in j providing for early warning to the operator and tir$ely corrective action. l 1 y .---n--
1, 2 The results of the regulatory analysis indicated that, within the framework of the existing Bulletin 85-01 requirements, the contribution of AFW pump steam birding to core melt frequency and public risk was sufficiently low to warrant the finding that no new requirements beyond those specified in IE Bulletin 85-01 need be imposed. Although the staff has concluded that the currently assessed risk associated with this issue is reasonably low, it is nevertheless considered irportant in light of the generally unsatisfactery check valve reliability experience obtained to date in operating plants, that plant operators continue to be alert to the possible development of malfur.ctioning check valves, especially as the plant ages, ard be prepared to increase the monitoring frequency as needed to ensure that steam binding failure of the AFV pumps does not occur. No written response or specific action is required by this letter; therefore. no clearance from the Office of Maragement and Budget is required. If you 1 have ary questions on this matter, please contact your project manager. = ) I i I l f l .I l i i I i i
O. 01 kM 87 MEMORANDUM FOR: Thomas E. Murley, Director Office of Nuclear Reactor Regulation FROM: Eric S. Beckjord, Director Office of Nuclear Regulatory Research
SUBJECT:
RESOLUTION OF GENERIC ISSUE 93, "STEAM BINDING OF AFW PUMPS" Generic Issue 93 is concerned with the potential disabling of the auxiliary feedwater (AFV) pumps by steam binding caused by backleakage of main feedwater (MFW) past tne isolation check valves interfacing the AFW and MFW systems. The key safety significance of the issue lies in the potential vulnerability of most AFW systems to common mode steam binding failure of the redundant pumps. This vulnerability is inherent in the piping configurations used, which allow for cross connections between trains via common headers on the suction and discharge sides of the pumps. To reduce the probability of pump failure by steam binding if backleakage does occur, IE Bulletin 85-01 (October 198b) required those licensees and CP holders, who had not already done so in response tc previous NRC and INPO recommendations. to implement procedures both for monitoting the AFn piping temperatures as an indication of possible backleakace and for restoring the pumps to operable status if steam binding were to occur or appeared likely to do so. The Bulletin's recommended monitoring frequency of once per shift, compared with the previous one of effectively once per month (when the pumps were tested), provided for a f actor of about 90 reduction in pump unavsila-bility due to steam binding. Generic Issue 93 has been resolved and no further study is required. The basis f or this conclusion is described in the attached Regulatory Analysis. To pro-vide an empirical basis for this analysis, a review was made of the backleakage experience obtained in cperating plants since the start of systematic monitoring (about April 1985 for a majority of the plants). The Task Manager contacte:i cach of the operating PWRs via the resident inspectors to determine the status of this concern at each plant. The results showed two distinct groups in regard to backleakage, with the dominant majerity of plants (about 90%) 55owing)a low incidence ot backleakage occurrences (less than one per reactor year while the other plants showed a far higher event rate by a factor of roughly 100. It is important to note that none of the backleakage events during the review period resulted in steam binding of an I.FW pump, because the various monitoring metherts employed appear to be ef fective ir preventing steam binding with a high degree of assurance. For the plants with a high backleakage event rate, utility installation of a continuous monitoring systeru with control room alarm was instrumental in providing for early warning to the operator and timely corrective action. As shown in the enclosure, conservative estimates of the contribution of steam binding to core melt frequency indicates this to be smail: a frequency of about 1x10 7/RY for the 90% group of plants and about 1x10 5/RY for the second ] group. The corresponding public risks, integrated over the remaining plant O P' f l -4f/MWMM j
1 .g. 'I lifetimes, are estimated to total 36 person-rems and 44 person-rems for the I respective groups. These results do not support any new requirements or further study of this issue, Because the risk is low, the plants that have not experienced backleakage over j ] at least several r.1onths of operation could be allowed some relaxation in l l monitoring frequency (f,or example, to once per week). This would be beneficial in permitting a shift of effort to other areas of plant maintenance. l Conversely, the second group of plants that have backleakage problems should l continue efforts toward resolving their specific leaky check valve problems. Generic Issue 93 is resolved and no further stJdy is required. However, to 1 keep the risk low, it is in.portant for the plant operators and our inspectors l to continue to be alert to the possible develop'nent of leaky check valves, L j especially as plants age, and be ready to increase the monitoring frequency as needed to assure prevention of pump steam binding. In this connection, it is j suggested that inspection modules be modified to include monitoring of possible j i backleakage degradation that may develop. The coordination of such findings j { with the relevant sectors of HRR, or of information gained by the resident l inspectors on the progress made by the second group of plants in resolving l 1 'their specific backleakage problems, would provide the basis for continuing surveiliance by NRC on this issue. Additionally, it would be appropriate for t the PWR project managers to ensure that the annual FSAR updates include docu- ] mentation of any plant instrumentation changes made in regard to the monitoring J of AFW system piping temperatures so that inspectors will have a basis for i inspecting this instrumentatio1. A draft memorandum covering these points is l enclosed for your consideration. i r i I i i Eric 5. Beckjord, Director I ,i Office of Nuclear Regulatory Research l l
Enclosure:
1 Regulatory Analysis of GI-93 I cc: V. Stello, EDO i j J. Zerbe, CRGR i l R. Fraley, ACRS F. Miraglia, NRR ) R. Starostecki, NRR j E. Jordan, AEOD j J. Allan, Reg. ! I j J. N. Grace, Reg. II i J. G. Keppler, Reg. III ) R. D. Martin, Reg. IV l 1 J. B. Martin, Reg. V I \\ l 1 l l
/p *8%% UNITED STATES [' NUCLE AR REGULATORY COMMISSION ?, m wmorow, c. c. rosis MEMORANDUM FOR: Frank J. Miraglia, Associate Director G r Projects, NRR Richard W. Starostecki, Associate Director for Inspction and Technical Assessment, NRR FROM: Thomas E. Murley, Director 4 Office of Nuclear Reactor Regulation
SUBJECT:
RES0'.UTION OF GENERIC ISSUE 93, "STEAM BINDING i 0F AFW PUNPS" Generic Issue 93 on steam binding of AFW pumps due to backleakage from the t main feedwater system has been resolved and no new requirements other than those already provided by the issuance of IE Bulletin 85-01 need be imposed on the licensees or CP holders of pressurized water reactor plants. The regulatory analysis of this issue is included in the RES remoranoum from E. Beckjerd to me, copies of which have been sent to you and to the Regional Administrators. Within the framework of the existing Bulletin 85-01 require-ments, the assessed risk associated with this issue is found to be low. I However, because of the known experience regarding the general un eliability of check valves in nuclear power plants it is important that the project managers be alert, via the resident inspectors, to possible AFW check valve leakage problems that may develop in their plants, in order to ensure that the plant I response to the proolem is such as to assure the prevention of AFW pump steam binding. l i The attached generic inforsation letter shoA d be sent to all licensees and CP holders within your scope of responsibility to advise them of the resolution of Generic Issue 93. Furthermore, all PWR project managers are to ensure that the annual FSAR updates include documentation of any instrumentation changes l made in rwgard to the monitoring of AFW systems piping temperatures so that inspectors will have a basis for inspecting this instrumentation. I Thomas E. Murley, Director Office of Nuclear Reactor Regulation d Enclesure: Draft Letter h All Licensees 4 and CP Holders.or PWRs I i
~ 1 . 'p* *
- s jo, NUCLE AR REGUL ATORY COMMISSION m
UNITED ST ATEt ,s, l ,e I % Al*4 t NG T oN. c. C. Posts . J Enclosure s E., o / ose* 0 RAFT TO ALL LICENSEES, APPLICANTS FOR OPERATING LICENSES, AND HOLDERS OF CONSTRUCTION PERMITS FOR PRESSURIZED WATER REACTORS Gentlemen:
SUBJECT:
RESOLUTION OF GENERIC SAFETY ISSUE 93, 5 TEAM BINDING OF AUXILI ARY FEE 0 WATER PUMP 5" 4 This letter is to inform you of the staff's resolution of the subject safety issue which concerns the potential disabling of auxiliary feedwater (AFW, 4 pumps by steam binding caused by backleakage of main feedwa',er (MFW) past the isolatien check valves interfacing the AFW and MFW systems. The significance of the issue stems from the potential vulnerability of most AFW systems to l comon mode steam binding f ailure of the red'indant AFV pumps, this vulner-ability being inherent in the piping configurations used, which allow for cross connections between trains via common headers on the suction and I 1 discharge sides of the pumps. To reduce the probability of pump failure by steam binding if backleakage does occur, IE Bulletin 85-01 (dated October 1985) required those licensees and CP hoicers, who had not already dene so in response to previous NRC and INPO recomendations, to implement procedures both for monitoring the AFW i piping temperatures as an indication of possible backleakage and for restoring j the pumps to operable status if steam binding were to occur or appeared likely { The Bulletin's recoscended monitoring frequency of once per shif t, l to de so. compared with the previous once per month che:k when the pumps were tested, j l provided for a f actor of about 90 reduction in pump unavailability due to l steam binding. 1 As a basis for the staff's regulatory analysis of this issue, a limited review was made of the backleakage experience in operating plants since the start of systematic eenitoring (about April 1985 for a majority of the plenh ). The results showed a double-humped distribution in regard to backleakage, with the i dominant majority of plants showing a low incidence of backleakage occurrences f
4 1 - ta mean of less than one per reactor year) while the remaining plants showed a fcr higher event rate by a f actor of roughly 100. None of the backleak3ge events during the review period appeared to have resulted in steam binding of an AFW pump, indicating that the various monitoring methods employed can be highly effective in providing for the prevention of steam binding if backleakage occurs. For the plants with a high backleakage event rate, the installation of continuous monitoring systems with control room alarms was i instrumental in providing for early warning to the operator and timely 1 corrective action. r l The results of the regulatory analysis indicated that, within the frsme-work of the existing Bulletin 85 01 requirements, the contribution of AFW i pump steam binding to core melt frequency and public risk was sufficiently low to warrant the finding thtt no new requirements beyond those specified in IE Bulletin 85-01 need be imposed. Further, the staff concludes that for plants that have not experienced backleakage over at least several (e.g., four) months of operation, some reduction in the frequency of monitoring (e.g., from once per shift to once per day or to once per week) alght be beneficial in persitting a shift of effort to other areas of plant saintenance. Conversely, for those plants that have had frequent backleakage it is considered important that plant ef f orts to resolve specific AFh check valve leakage problems receive the j degree of emphasis appropriate to a safety system of high risk importance. j Furthersore, the FSARs should be updated to include documentation of any l ]. instrumentation changes made in regard to the monitoring of AFW systems i i piping temperatures, l j Although the staff has concluded that the currently assessed risk associated with this issue is ' reasonably low and that some relaxation in monitoring f I frequency might be appropriate under certain conditions, it is nevertheless j considered, in light of the generally unsatisf er. tory check valve reliability I experience obtained to date in operating plants, that plant operators continue to be alert to the possible development of malfunctioning check vahes. l [ especially as the plant ages, and be prepared to increase the monitoring t frequency et needed to ensure that steam binding failure of the AFV pumps does i j t not occur. 2 \\,
T O No written response or specific action is required by this letter, therefore, no clearance from the office of Management and Budget is required. If you have any questions on this matter please contact your project manager, j i I i t l I l t f f l l 1 ? l t i t 6, l 1
) ENCLOSURE ) 4 d 1 RE'GULATORY ANALY515 OF GENERIC SARITY ISSUE 93
- STEAM BINDING OF AUX!LIARY FEEDWATER PUMP 5" i
l l I j L i i i t 1 i i t i j i I i ] ) ) i t I I i A. M. Spano j 50$!/0620 l{ February 1987 1 l i l 1 i 1 t I i 3 i I l 37P P ' t i 9 l
- WI l
i d i
i t o l CONTENT 5 4 I 1. STATEMENT OF PROBLEM 4 1.1 Description of Issue 12 Historical Background 2. 08L1ECTIVE t i 3. CURRENT SAFETY ASSESSMENT OF AFW PUMP STEAM $1NDING 1 3.1 Updated Review of Plant Experience on Backlenkage f i 3.2 Risk Significance i 4 ALTERNATIVE RESOLUTIONS 4.1 Proposals ) 4.2 Consequences j 4.2.1 Alternat.ive 1 No Action l ) 4.2.2 Alternative 2 - Continuous Monitoring Systen Sackfit i 5. CON:LUSIONS l l 6. REFEREN:E5 Appendix 1 Appendix 2 Apptadia 3 l l l i j l a i 1 i I l I
Reculatory Analysis for Generic issue 9), "Steam Bindino of Auxilisry Feedwater Purps" I 1. STATEMENT OF PROBLEM 1.1 Description of Issue l In a pressurized water reactor the Auxiliary Feedwater (AFW) system supplies i feed ster to the steam generators whenever the main feedwater (MFW) flow is ) interrupted. In the event of an abnormal condition resulting in the loss of MFW, the AFW system serves as a vital backup safety system for ensuring the renoval of decay heat. Under normal operating conditions, the idle AFW system is kept isolated from the high pressure (s1000 psig) steam system by a nutter of check valves and, in some systems, closed, remotely operated valves. Generic Issue 93 is concerned with the potential disabling of the AFd pumps by steam. binding as a result of the backleakage of hot water or steam past the iso-lation check valves interfacing the AFW and MFW systees. In the low pressure environment of the AFW system, the leaking subcooled water flashes into steam, and a backflo. rixture of steam and het water ray deveicp that forces itself upstream past other lesking check valves to one or more of ths RFV pumps. There, the continued buildup of the steam void content can lead to pump cavitation and consequent f ailure when the pumps are started up. The key significance of the issue arises from the potential vulnerability cf sost AFW sys'.ees to comen mode steam binding f ailure of the redundant pannps of the system. The potential for such f ailure is inherent in the typical piping configurat. ions used, which allow for cross connections between trains via comon dischargo headers, suction headers, and recirculation lines, with usually only a single check va've to prevent backleakege to the ascend or thirc pump (Figure 1). l l i 9 Ne emesmee se e
2 l Given the occurrence of backleakage, the probability of one or more of the pumps becoming steam bound will depend upon the effectiveness of the upstream ) check valves in stopping the backleakage. in this regard, check valve l reliability involves correct design application of the type and size of check j va14e selected for installation in each AFW discharge line in order to ensure j compatibility of valve perforeance with the local hydraulic conflitions. Many l of the check valves employed in existing systems rely solely on pressure l I difforential (AP) for sealing. For the check valves interf acing the AN and t MN systems, the AP across the seat is large and a good seal can normally be expected, if the disc butts up squarely against the seat. In this con-l nection, the need for periodic valve maintenance makes it important that the l check valve design allow for correct disc to seat aligneent in a relatively simple straightforward manner, and in a way that can be checked before the l valve is reassembled. For the upstream check valves, where minimal AP conditions obtain, unless the check valve incorporates a mechanical (spring / gravity) load to assist in sealing, the capability of the valve to stop the backflow of steam and het water from a leaky inte.*,f acing valve becomes questionable, as evidenced by the nuterous instanc'en of steam binding observed in AFW systems with multiple check valves in series. In this connection, an ) upstrea gate or globe valve operated normally closed can be expected to ( l provide significantly better protection against backleakage than a non-rechanically loaded check valve. Approximately 4 % of the operating PuRs l j (mostly CE and B&V p) ants) have a normally closed remotely operated valve in l the discharge lines. (Westinghouse plants typically operate with the remotely-j operated control valves nerva11y open, which rey reflect the vendor philoscphy ( j of sirple cor teci system design and reliance on the operator for subsequent l I throttling of AN flow, in contrast to the approach taken in other plants where sophisticated control systems are employed for th4 programed opening i of the closed valve in contro11ing AN f1w.) J f L l While t.hese and cther f actors governing the likelihood of backleakange and i potertial steat. binding are comples and plant specific,, the sisple fact that l backleakage has occurred in a system is readily evidenced by the effected AN i i i 1
4. 1 3 pipes becoming hot. Thus, appropriate monitoring of the AFW piping temperature can alert the operator to the occurrence of backleakage and to the need for sitigative action to offset possible vapor binding of the pumps. 1.2 Historical Backaround The number of AFW pump steam binding events reported during the period 1981 1964 led te the issuance of a number of information notices and reports by NRC snd the industry: In January 1984, IE Inforsatica Notice 84 06 and INP0 Significant o Event Report 5 84 were issued, describing the circumstances relating to approxirately 18 steam binding events that had occurred at H. B. j Robinson, Far ley 1 and -2, Crystal River-3, and D.C. Cook-2 since 1981. At Robinson and Farley, procedures were initiated for the periodic ronitoring of pump casing temperatures and for the venting and I cooling of pumps when required. i In April 1984, INP0 issued a Significant Operating E%nt Report (50ER e ) i, 64 3) analyzing the safety ir.plications of the reported events at i farley, H. B. Robinson, and Surry 2, and making recoerendations on the j need to: (1) periodically ponitor and record the AFV piping temperatures I at least once per shif t, or preferably on a continuous, instrumented l basis because of the possibility of backleakage occurring rapidly (as had l been observed at Farley); (2) review the capability of AFW system check valves to seat with low pressure differentials; and (3) include guidance on operating procedures and training for identifying and restoring the ATV system to full operability, including special actions to be taken l when one or rore of the check valves are known to leak repeatedly. ) l In July 1984, AE00's case study of the safety implications of backleakage o to the AFW system analyzed 22 LER-reported instances of steam binding cr i backleakage that occurreu during 1951 1983 at six plants (five Westing-4
4 house and one B&W designed plants) out of a total of approximately forty-seven PWRs operating during the period surveyed.II) While thise statistics would suggest that the backleakage problem, with its potential for puep steam binding, is associated with a relatively small fraction of the PWRs, the AE00 report did note that the number of events reported eay not be a reliable indicator of the actual number of events occurring, in that such occurrences might not have been deemed reportable by the plant j technical specifications. Thus, under the post-January 1964 LER reporting requirements (10 CFR 100 Part 50.73(a)(2)(vi)), a steem bound pump is not l considered reportable if the redundant pump (s) are operable and available to pe-form the required function. } In the AE00 review of the backleakage experience no pattern or single i rajor cause of check valve leakage could be identified, with the causes l differing between recurrent eventh at a given plant as well as between plants, and with check valve leakage recurring even af ter a valve hsd s 1 been repaired or replaced. The report pointed out that the operating j i experience supported the conclusion that AFW systems operating with the i remotely operated valves run norrally open (as in cost Westinghouse j olant.s) ray be vore susceptible to steam binding than are the other plants. Note was also made that some plants had already adopted f procecures for the routine surveillance of the AFV piping torperature; the report went on to recommend that such monitoring be made a general requirement. 1 i i o $ team binding of AFW pumps was assessed to be a generic issue of high i priority, and authorization for work on the issue was provided by [ rer.orandue f rote H. R. Denton to R. Bernero in October 1964. l o In Apri) 1985, t,o detereine 1.he need for short term corrective action related to the problee of steam binding, IE requested the regitnal j offices (Teeporary Instruction 2515/67-03) to conduct, a survey of l licensee responses to previcus NRC and industry reccovaehdations regarding ) i I
(1) the monitoring of pipe temperatures at least once per shift and (ii) the availability of procedures for detect.ing and correcting a steam binding condition. Of the 58 units surveyed at the time, approximately half had both precetures and related training in place, while the others 4 lacked either certain procesares cr training or both. 3 o On the basis of this survey, IE Bulletin 85 01 (issued October 29,1985) requested 28 licensees and all CP holders to develop and implement pro-cedures for monitoring AFV piping temperature on a recommended once ) per shift basis, for recognizing steam binding, and for restoring the i ANS to operable status should steam binding occur. The Bulletin also re4uired that procedural controls were to remain in effect pending the l aception of an appropriate hardware fix substantially reducing the likeli-hood of steam binding, or until superseded by n,ction implemented as a result of resolution of Generic Issue 93. The utility responses to Bulletin 85 01 indicated that various methods are being used to monitor piping temperatures. In most cases, the method involves sirple touching of the pump casing or pipe, such that if it j is "hot" to the touch, the operator or shift supervisor is notified, und recovery procedures are inititted (e.g., venting of pump casing, operating the pu p and flushing out the affected discharge lines). At a nu-ber of p16nts, quantitative temperature readthgs are obtained using l contact pyroreters, temperature sentitive color tape, or other persanently attachec temperature instruments with local readout. The sonitoring l j frequency is generally once per shift, although some plants, depending en [ j their previous backleakage experience, may conduct surveillances every 4 i hours. Still other plants have installed a continuous, instrumented i monitoring system, with a control room alarm to alert the operat,or as to when the pipe temperature has risen above a given setpoint. t l l 1 1 4 j i 1
-6* While use of the hand is as reliable as any of the methods in regard to detecting e sensibly hot pipe, the continuous monitoring system is citarly the most effective one in regard to minimizing operator warning time and the time to start cooling procedures, and, thereby, minimizing the con-ditional probability of pump steam binding, given the occurrence of back-leakage. The usefulness of the system is further enhanced by locating I the temperature reading locations appreciably downstream of the pump dis-charge points, in the 'ticinity of the interfacing check valves, 4 i The advantage of the continuous monitoring system over the manual j approach is also evident for the case where an interfacing check valve i leaks repeatedly and severely enough to heat up the discharge path back to the pump in a time period short compared with en 8* hour surveillance period. Under such nditions, a pump could become steam bound long befere the next shif t check, with the probability for common mode failure of the redundant purps increasing the longer it takes to commence cooling procedures. As discussed further below, for the relatively sas11 number l of plants where there have been multiple recurrences of backleakage, the installation by the utility of a continuous monitoring syatem has been instrumental in the plant nperator being able to prevent pump stone r binding pending repsir of the leaky check valves at the next saintenance
- outage, i
a An additional development affecting the issue of AFW pump steam binding o l l was the establishment, in the wake of the November 1985 water hammer event at San Onof ra 1,(2) of an industry sponsored effort to deal with l the general problem of safety related check valve failures in reactor { systers. The Owners Groups Task Force (OGTF) concerned with this issue pet with the NRC in November 1996 to present its propotep program of work. 3 This OGTF prograe essentially involves the issuance of the following two { j docueents for use by the individual utilities in imp 1teenting their own j i check valve reliability improvement programs: I i
(a) An INPO report (50ER 86 3), issued October I!:, 1966, to provide general guidance to the utilities on: (i) the 6,inds of check valve misapplication problems that can occur (e.g., selection of a valve 1 size that does not match system flow conditions, valve installation l with wrong orientation, wrong valve type); (ii) the detectici of valve degradnion or failure (e.g., preventive maintenance wit.r. disassembly and visual inspection, periodic testing, use of acoustic / vibration analysis techniques, radiographic methods); and (iii) personnel training (e.g., in utilizing various maintenance / testing methods). As a followup to 50ER 86 3, the industry program I calls for the utilities to implement appropriate chsnges in their check valve raintenance and testing programs, based i.m the 50ER ) guidance. It is also proposed that each utility identi1y the high risk check valves in its plant (e.g., AN system check valves) and I l include them in its progrta. i l l l (b) An "Applications Guide" document to be prepared by the OGTF, so l provide detailed inferiraticn on the appropriate selection of l various types of check velves, their physical location in the syst.em, the ef fect of flow conditione in regard to the sizing of check valves, etc. The Applications Guide is intended to serve es a basis for check valve design reviews by the individual utilities. These would then be followed by utility implementation of appropriate design modifica' ions, as indicated by the results of their design reviews. The Guide is scheduled for issuance by l June 1957. The OSTF has indicat.ed to tne staff that the end date for cocplete ieplems.t.ation of the inowstry progra'n is expsJted to be 1991. m
r l 3 l' t 2. OBJECTIVE Section 10.4.9 of the SRP requires that an acceptable AFW tystem have an unreltahi11ty of less than 10 4 per 6emand, as estimated using ?,he methods and data presenttd in NUREG 0611 and NUKEG 0635. Those methods did not r include specific consideration of steam binding of the AFW paps as one of the coneon sodt failurv contributors to system unreliabi18ty. The regulatory objective sought in connection with resolution of Generic issue 93 is that step binding should be a non-significant contributor i,o the Werall AFW systea unavailability, which say be interpreted here to mean a contribution of, sa;<, less than 10% percent of the overall system unavcilability. Resolution of this issue is viewed from the following perc;ective: Gi<en the fact that interfacing check valves separating the AFW and P.FW systets can be expected to f ail on occasion, with the resulting b.scklenkage of steam end het water posing a challenge to the isolatability of the st&ncty AFV system, I we consider cost-effective actions that can asse,qe i ut the operit y can be apprcpriately alerted in tire to prevent stest bi'ndt a of the puns, or, if stes. binding has already eccurred, to restore the system to full operability on an appropriately timely basss. In this regard, the isuu n',e of 3F Bulletin 85 01 requesting the developeent and ierleeentation of procedures for *,he periodic sonitoring of the AFW piping toeperature and for system restorttion constituted an important step fomard in dealing with the potential for stone vinding. In comparison with tha i earlier survei) lance period of effectively once a month when the pueps vere tested, the Bulletin requiree'ents for a once a shif t monitoring period provided for a reduction in tbe steam binding re16ted average unavailadi)ity ci the i pumps by a f actor of about, $s (3 shif ts/ day x 30 days /sont,h), On the basis of the operational experience ch backloakage obtained 4,ince systerstic vanitoring of the AFW pipes was started in 1984 1995 by oost of i the licensees, we analysed the adequacy of the Bulletin requirements in setting the above stated objective and whether additional requirements are needed. -~ ~' ~ ~ ;_ ~ _
3. CURRENT SAFETY ASSESSMENT OF PUMP STEAM BIN')ING 3.1 Updated Review of Plant Experience on Backleakage A search of the LER and Nuclear Pint Reliability Data System (NPRDS) files for information on steam binding-related failures of AFW pumps indicated a total of just two events (occurring in early 1984) that were not reported in the AE00 study. However, the absence of such reported events is not considered meaningful in the light of the post January 1,1984 LER rule not requiring the rep ting of individual component f ailures, or in the light of the voluntary basis for utility repotting of component failures to the NPRDS. To obtain a current picture of plant experience on AFV backleakage and steam binding as a basis for an assessment of the risk posed by steam binding under conditions where the AFW piping temperature is monitored on a systematic basis, pertinent information on recent backleakage occurrencea du *ng the perioJ since monitoring started was obtained via an informa'. survey of the NRC resident inspectors at plants, data derived from staff visits to various plants, as well as by several telephone conversations with plant engineering personnel, as arranged by the PMs. The information obtained is presented in Appendix 1. These data show that the backleakage experience varies widely among the plants surveyed, with the dominant majority of the operating PWRs indicating a low, backleakage event frequency of from zero to a few events per year, and a much smaller group of about a half dozen plants indicating a significantly higher annual backleakage event frequency. It is noted, in part'cular, that although sorre backleakage has been experieaced in about 20% of the operating plants during the survey period, no steam binding of AFW pumps appears to have resulted, indicating that the monitoring procedures were ei.'ective in catching these backleakages early enough to prevent any subsequent steam binding of the AFW pumps.
. Io. 3.2 Risk Significance In assessing the safety implications of the backleakage experience observed in plants with monitoring procedures, the survey results suggest an appropriate division of the operating PWRs into two categories: (A) plants experiencing a low frequency of backleakage events (defined as one involving the detection of a hot pipe or pump, followed by operator action to cool the pump), and (B) plants exhibiting a relatively high frequency of such events. Approximately 56 PWRs, or about 89% of the operating plants surveyed are found to fit into the low frequency group, i.e., less than about I event / year. In i comparison, the remaining group of seven PWRs appears to have had a back-i leakage event frequency that is greater by a factor of 10-100. In the following value-ittpact analysis we will consider these two groups separately. Category A Plants, A ceasure of the public risk posed by steam binding of the AFW pumps can be obtained on a core melt frequency risk level by examining the dominant accident sequences affected by unavailability of the AFW system. An estimate is obtained of the increase in core melt frequency (ACHF) due to the increase in A7W system unavailability resulting just from steam binding of the pumps. For each accident initiation of interest, we have:
- AQ O
ACHF4=F4 AFWS pg 1 trhe re, Accident initiating event frequency for each initiator i of primary F = j interest, i.e., direct loss of all main feedwatei ( WW), loss of the offsite power (LOSP), and loss of all AC (Station Blagkout). The last is a much lower probability event than the other two (by a 2 factor of at least 10 ), and will accordingly be neglected in this analysis; ) 1
B . AQAFWS = Increase in AFW system unavailability due solely to steam binding, assuming non-recovery of the steam bound AFV pumps within a 30-min. time period to prevent core uncovery, as discussed below; Qpg Unavailability of feed and bleed (F&B) as an alternative method = for decay heat removal, given that steam generator cooling is not available. With the loss of main feedwatar the steam generators start to boil dry, with the rate of fall-off of the steam generator water level depending on the initiating event. In the LOSP-induced LMFW case, the reactor and reactor ecolant pumps (RCPs) are automatically tripped, stopping the further production of heat from fission and RCP operation that has to be removed via the steam generators. The resulting rate of depletion of the steam generator secondary water is consequently significantly lower than it would be for the direct LMFW event, where tne reactor may not be tripped until pressurizer pressure reaches a preset high point, or steam generator level drops to a preset low point, generating a safety signal that scrams the reactor and actuates the AFW system. Operation of the RCPs following a LMFW event may continue until manually tripped by the operator later in the transient. The steam generator dryout time also depends on the steam generator inventory, which for the B&W once-through design is a factor of 3 to 5 times smaller than that for U-tube steam generators for equivalent plant sizes. If the AFW system fails to respond adequately because the pumps are steam bound, the steam generator secondaries will dry out, at which point the transfer of decay heat from The primary system ceases and the system temperature rises to the saturation temperature, corresponding to the pressure setpoint where the PORVs open, releasing steam to the pressurizer relief tank. Unless F&B is initiated in time, or unless MFW or AFW is recovered, the i continued loss of primary system coolant through the PORVs will lead to core l uncovery. Reference 6 shows the estimated times to core uncovery for station i l i J
blackout sequences, where the AC-independent AFW pumps are not available (e.g., because of stean binding). These uncovery times vary between about 100 minutes for Westinghouse and CE plants down to about 45 minutes for B&W plants. To preclude core uncovery, it is necessary tisat an auxiliary feedwater pump be started about the time of steam generator dryout for low v/ad safety injection pumps, and somewhat later for plants with high head safety injection pumps. This results in a limiting time of approximately 30 min. for both CE and Westinghouse plants with low head pumps and for &&W plants with high head safety injection pumps. Thus, it will be assumed that, if F&B ia not available, core uncovery can still be avoided if at least one of the steam bound AFW pumps can be recovered within 30 minutes after loss of all f*edwater. Evaluation of each of the factors in the 6CHF equation above 's obtained as follows: (c) Loss of HFW Event Frequency: Most of these events constitute either short term or partial losses of feedwater (ex., loss of one MFW pump, with another pump available for continued feedwater operation). A staff estimate ( } based on a search of the LER files indicated a frequency of between 1 and 3 LMFV events per reactor year (RY), with a small fraction of these being non-recoverable in time to prevent steam generator dryout: specifically, a frecuency in the range 0.1-0.4/RY 'or such non-recoverable LMFW occurrences. A different analysis of the experience on total losses of MFW in 36 PWRs over a period of 213 reactor years indicated a mean annual frequency of 0.15 events /RY.(4) A different type of ertimate for this quantity was derived from a detailed fault tree analysis of the Oconee MFW system.(5) The results yielded a frequency estimate of 0.64 non-recoverable LMFW events per RY, which is about a factor of three to four higher than that derived from plant data. For purposes of this analysis, we shall use the Oconee value as a conservative generic estimate for PWRs in general.
s) LOSP Event frequency: On loss of offsite power, the motor-driven c MFW pumps are lost and there is a prompt trip of the reactor and reactor l coolant puops. In the context of the 30-min. time constraint assumed above for recovery of the steam bound AFW pumps, the frequency of a 30-min. LOSP event is about 0.045/RY.(6) (c) AQgpg3: For the Category A pisnts, an upper bound estimate of the unavailability of the AFW system for 30 min, or more caused by steam binding of the pumps is provided in Appendix 2. The result is AQAFWS < 4x10 7/d. (d) Unavailability of Feed and Bleed Cooling: An analysis of the credit to be given to the use of feed and bleed techniques as an alternite means of decay heat removal has been provided as part of the regulatory analysis of Generic Issue 124 on AFV system reliability, ) and the results obtained have been utilized in the present analysis of the related issue of AFW steam binding. In this analysis, an examination is made of the following factors affecting the failure probability of feed and bleed techniques for various PWRs: (1) Hardware f ailure, including for different vendor-designed systems questions of the failure probability of the relevant high pressure injection (HPI) systems, the adequency of the HPI pump dis-charge pressure for lifting the pressurizer safety valves, and failure probability of the PORV components that may be available; (2) Thereal-hydraulic failure, which relates to the time window available for feed and bleed before steam generator dryout occurs or primary system saturation is reached; (3) Decir,ional human error probability, which arises in connection with the decisional conflict between operator reluctance to us1 feed and bleed methods and the need to initiate feed and bleed during the tire window available to the operator;
. l (4) Procedural human error, which can arise in the implementation of feed and bleed procedures following the decision to use such methods. The calculatinnal results obtained for the net feed and bleed failure probability for two Westinghouse, two CE, and three B&W plants varied between 0.42 and 0.53, with an average value of 0.47. In the present analysis of the risk impact of AFW steam binding, a generic value of 0.5 is used. (f) Core Melt Frequency: Combining the above evaluated factors making up the LOMF and LOSP dominant accident sequences, one obtains for the Category A plants an upper bound for the AFW pump steam binding contribution to CHF of: ACHF = (0.64/RY + 0.045/RY) * (4 x 10 7/d) * (0.5) = 1.4 x 10 7/RY, which is negligible in the context of a safety goal criterion of 1x10 */RY. Category B Plants,, This group of plants has experienced multiple instances of backleakage e, into the AFW system; it is comprised of the two Farley units, the two McGuire units, the two Catawba units, and Diablo Canyon-2 (see Appendix 1). As a result of the recurrences of valve leakages, continuous monitoring systems with control room alarns were installed in these plants. Since installation i of this equipment, there apparently have been no occurrences of actual steam ) binding of the AFW pumps, the instrumentation acting to provide sufficiently early indication to the optrator of the onset of backleakage and need for recovery action, such that steam binding can be prevented with a high degrek of assurance. (a) Farley Units 1 and 2: The problem of backleakage in these two reactors goes back to 1983. As a result of the numerous instances e,f back-leakage tnat occurred in 1983, the utility initiated procedures for monitoring the AFV piping temperature by a rover every four hours, in addition to the ] l continuous monitoring provided by the installation of temperature detectors located both in the vicinity of the interfacing check valves and nest the
pump discharges. The monitoring procedures have been effective in preventing steam binding of the pumps in spite of recurrent instances of backleakage. Valve leakage repairs in one or more o' *.ne eight interfacing check valves in each unit have been performed at a rate of about four repairs a year, in a continuing effort to resolve the underlying problem of check valve leakage. Progress in this direction appears to have been made recently as a result of vendor proposed modification in the valve maintenance procedures used at F riey, in which the valve relapping method was changed to provide a circular line stating area rather than an angular band, and the hinge pin bushing tolerances were tightened to ieduce disc play. In the two-month period since the revised maintenance procedures were put into effect in Farley-1 late in 1936, no backleakage has been observed in that reactor. (b) McGuire Units 1 and 2: Backleakage problems have been experienced in these reactors for several years, with numerous instances of steam binding of the AFW pumps occurring especially in 1984. The incorrect installation of the turbine driven pump discharge check valves at 90' to the correct direction, which led to an increase in the probability of pump steam binding given the occurrence of backleakage, was corrected in 1985. The installation in 1985 of a continuous monitoring system with the temperature sensors located near the interfacing check valves allowed for early corrective action by the operator in the event of backleakage. The incidence of valve leakage at McGuire appears to be associated with the monthly testing of pumps, whereby an opened interfacing check valve may not seal af ter the pump is secured. If this occurs tne ensuing backleakage is detected by the monitoring system and f the operator attempts a reclosure of the valve. The operators are sensitive to the possibility that a valve may not close properly and accordingly are i prepared to irplement corrective action. No pump steam binding events have apparently occurred since early 1985, although there have been continuing instances of valve leakage. ,c._,
. (c) Catawba Units 1 and 2: These are sister plants to the McGuire reactors, with similar AFW systems and check valves, and with temperature sensors for continuous monitoring installed near the interfacing check valves. In both Catawba units there has been a pattern of almost continuous leakage through one or more of the interfacing check valves, and this has necessitated extended periods of operation of the AFW pumps to flush out the backleaked hot water and cool the discharge lines. At Catawba, there is the growing realization that one cf the major causes of the valve leakage appears to be related to the bonnet-hinged design of some of the swing check valves used. These require very precise alignment of the bonnet to the body, both vertically and in angular orientation, in order to obtain proper seating. Unfortunately, the design of the valva precludes making a check that the disc is seating correctly until the valve has been reassembled and installed, and the system placed in operation. At Catawba a program is underway to replace the installed check valves with valves of a different type. I (d) Diablo Canyon-2: Here some backleakage occurred over a period of several months as a result of a small crack in the seal weld surrounding the disc of one of the interfacing check valves. The leakage was stopped following a welding repair of the valve performed during the plant outage in late 1985. In this as in the other cases of plants with recurrently leaky check valves, the availability of the continuously monitoring system with control room alarm allowed for timely mitigation of the effects of backleakage and prevention of pump steam binding. For the Category B plants, an upper bound estimate of the steam binding contribution to core melt frequency can be obtained on the basis of the analytical model described above, modified with regard to two factors: the hot pipe event frequency, Agp, and the probability that a pump becomes steam bound, given a hot pipe event, p(SB/HP) (see Appendix 2).
As shown in Appendix 2, AQAFWS=4x10.s/d, and accordingly, ACMF=(0.69/RY)* (4x10.s/d)*(0.5)=1.4x10 8/RY, which is a factor of 10 greater than the ACHF estimate for the Category A plants, but which is also small compared with the safety goal criterion of 1x10 4/RY. 4. ALTERNATIVE RESOLUTIONS 4.1 Proposals Two alternative resolutions were considered by the staff in reaching its proposed resolution of Generic Issue 93:
- Alternative 1 - No Action:
In this case, where the objective stated above in Section 2 has been achieved within the framework of the existing Bulletin 85-01 requirements, it is proposed that the Bulletin requirements remain in place, with allowance for appropriate modification as discussed below in Section 5, but that no further requirements be defined and that the issue be closed out.
- Alternative 2 - Backfit Requirement:
This would seek to reduce the steam binding contribution to AFW system uravailability by requiring all plants to install e continuou: monitoring system with control room alarm. At the present time eleven operating PWRs and two cps (South Texas I and 2) have such monitoring systems in place, so that the proposed requirpment would impact about 52 licensees and 16 applicants. 4.2 Consequences 4.2.1 Alternative 1 - No Action As shown in Appendix 3, estimates of the public risk posed by steam binding of the AFW pumps may be made based on an approach similar to that employed in the analysis for Generic Issue 122.1.(8) These results, together with those i i s See +
1 - for tne estimated impact on AFWS unavailability and core melt frequency are presented in Table 1. Tabic 1 Current Assessment of Risk Impact From AFW Pump Steam Bindina Category A Plants Category B Plants Risk Level _ (55 PWRs) (7 PWRs) AFWS Unavailability <4x10 7/D 4x10 8/D Core Melt Frequency 1.4x10 7/RY 1.4x10.s/RY Puolic Risk
- 36 person-rem 44 person-rem t
"Integrated over remaining lifetime of all PWR plants. These results indicate that the the risk to the public arising from steam binding of the AFW pumps is negligible for both the Category A and B operating plants. 4.2.2 Alternative 2 - Continuous Monitoring System Backfit (a) Risk Reduction Benefit I An estimate of the reduction in AFV system unavailability provided by installation of continuous monitoring systems in 52 of the operating PWRs can be obtained as follows. As described in Appendix 2, the probability, j p(SB/HP), for operator f ailure to prevent pump steam binding given detection l of a hot pipe, can be reasonably set at 0.1 for the typical Category A plant case where the pipe is monitored locally near the pump once each shift, and at 3x10 8 for the Category B case, where there is a continuous monitoring system in place, with control room alare and detection point near the interfacing check I J
valve. This factor nf 30 reduction in p(SB/HP) is also reflected in a factor of 30 reduction in AQAFWS, ACMF, and in the public risk. Thus, the residual risk af ter backfit would be about I person-rem, with the risk reduction benefit amounting to about 35 person-rems. l (b) Costs Two major costs elements are considered in this analysis: the cost for the monitoring system instrumentation and installation, and the related mainte-nance costs integrated over the remaining plant lifetime. It is assumed that the backfit is implemented by the plant personnel and not by an AE, that there are four trains of instrumentation, that the signals from the temperature sensors can be tied into an existing plant computer system and existing trouble alarm in the control room, and that the installation does not affect plant operation time. (i) Cost of Monitoring System Engineering support $15,000 (overhead, design; vendor spect drawings; etc.) Equipment 530,000 (temperature sensors, including one full set of replacements; electronic signal transmission) Installation 540,000 1 (field materials, labor, testiag) { Total = $85,000 per plant (ii) Maintenance 1 J
It is assumed that the additional effort required for maintenance of the monitoring system amounts to 20 hours per year at a cost rate of 20 $/ hour, i.e., an annual personnel cost of 400 $/ year. At the current time, the average remaining lifetime of the licensed plants is estimated to be 30 years, so that assuming a discount rate of 4%, we have for the lifetime personnel cost, Cpg = (400 $/ year) (1.04)30-1 = $6.9x108 per reactor. (0.04)(1.04)30 Thus, the total system plus maintenance cost per reactor adds up to about $92,000 per reactor. For a total of 52 operating reactors to comply with the backfit requirement, the total cost to the industry for the licensed plants alone would amount to about (52)(92,000), or about 5 million dollars. (c) Cost / Benefit Ratio S. The overall cost benefit ratio is seen to be about $5H/35 person-rems or about 140,000 dollars per person-rem averted. This value does not justify the instal-1ation of continuous monitoring systec:s at the Category A plants. (We note too that consideration of averted onsite costs would not significantly affect this conclusion. Thus, assigning an onsite cost of 4 billion dollars per core melt, and using ACHF=1.4x10 7/RY, a mean remaining lifetime of 30 years, and a total Category A population of 56 reactors, we obtain an estimated industry averted cost of: (1.4x10 7 CH/RY)x(30 Y)x(56 R)x(4x10' $/CH)=$0.9H, which is only a small part of the direct cost of $5H for the proposed fix.) 5. CONCLUSIONS The foregoing results on the estimated risk due to steam binding of the AFW pumps were derived using a conservative, upper bound analytical approach based on the backleakage experience obtained in operating PWRs ov2r the past one to two years, during which time systematic monitoring of the AFV piping I ,,_<,-e c w
w--
temperature was performed. The results show clearly that for both categories of plants the steam binding contribution to AFW system unavailability and the related risk to the public are currently at a negligible level. In this regard, we note, too, that for an appreciable number of plants, the AFW systems characteristics (e.g., the long runs of uninsulated piping, the use of separate auction lines from the water source to the pumps, the use of mechanically-loaded upstraam check valves, ce the possibility for cross connecting the AFW systems between units as a safety backup) may be such as to provide for an intrinsically low vulnerability of the AFW system to potential failure from steam binding. Operation wi,h the remotely-operated valves run normally closed say be expected to provide some additional assurance against the possibility of pump steam binding if an interfacing check valve did leak. For plants that operate the valves normally open, a shift in the valve mode of cperaticn would necessitate plant-specific evaluations of the potential negative effect on the AFV function of valves failing to open on demand, as well as the impact on AFW control system operations. In any event, in the light of the low steam binding risk level obtained in the approximately thirty-seven plants that do operate with the valves norn. ally open, there is clearly little risk reduction incentive for those plants to consider changing tneir valve mode of operation simply on the Dasis of potential steam binding. The staff believes that the cost-benefit results associated with the backfit alternative considered above clearly support a recommendation for selecting the alternative rer.olution of no action, as defined aoove, and to close out j this issue with no additional expenditure of NRC resources, Finally, in the light of the staff's findings on the current low level risk posed by Generic Issue 93, for plants that have not exper'.enced backleakage over a period of at least several months of operation, utility review of the problem may indicate that an appropriate reduction in the frequency'.of monitoring, (ex., from once per shif t to once per week) would be useful from an overall safety point of view in allowing increased emphasis on other areas
,. of maintenance and surveillance. In this case, to keep the risk low the plant operators should continue to be alert to the possible development of a leaky check valve or valves, especially as the plant ages, and be ready to increase the monitoring frequency as appropriate to assure that the AFW pumps do not become steam bound. The probability of unexpected check valve led, aga problems can be expected to be reduced when the proposed industry program on improvement in check valve reliability is implemented. l L 1 i i j n 1 h I
HFW ?&m& SB $!n$$r'Ysy e .1 IS MG c N i n eMaro*w ust
- Q Sg.A v
t-{4 ',) (h SG D %.r M dm 'Itr'" ~ 4*-g* ww -*4:-t+N
- f Guer m prw
^%%,a v seg neg qe S I % $G.A
- M r
M e n nov 5 'i 's s n CII W I gg y WA uSG E p e-n n s O h- $H4 X i + - - SS-A m.,,, t ',: 0 _. y' g N 4 'b'"o,, M SG B + lt MLo n,4,.-,.. 3 EM 'y. u V][ ,M $ msul Mver rr +5 (B&W) mm..,:Hy,, c--N " Nl J %.- a--- n ro msen r Tigure 1. Sirplified Arv system Configurations l Line thickening to t.he right of the interfacing check velves denotes high pressure.
6. REFERENCES (1) "Steam Bind.ir.g of Auxiliary Feedwater Pumps," AEOD/C404, W. D. Lanning, July 1984. (2) "Loss of Power and Water Hammer Event at San Onofre, Unit 1, on November 21, 1985," NUREG-1190, Jaunary 1986. (3) Memorandum from A. Thadani to 0. Parr, "Auxiliary Feedwater System - CRGR Package," Nover.ber 9,1984. (4) "ATVS: A Reappraisal. Part 3: Frequency of Anticipated Transients," Electric Power Research Institute EPRI NP 2230, January 1982. (5) "A Probabilistic Risk Assessment of Oconee Unit 3." Electric Power Research Institute, NSAC-60, June 19B4. (6) "Evaluation of Station Blackout Accidents at Nuclear Power Plants," NUREG-1032, May 1985. (7) Backfit Analysis - Auxiliary Feedwater System Reliability, Generic Issue 124, Namorandum from S. Diab to W. Minners, Septer.ber 23, 1986. (8) Generic Issue 122.1, "Potential Inability to Remove Decay Heat", NUREG-0933, June 1986. (9) "Handbook of Human Reliability Analysis with Emphasis on Nuclear Pover Plant Applications - Final Report," A. D. Swain, H. E. Guttman, NUREG/CR 1278, August 1983. 1 l l 1
APPENDIX 1 Survey of AFV System Backleakage Experience ID) Monitoring Number of IC) Connerical Monitoring ") Frequency Hot Pipe T I Plant Vender Operation Method (Per Shift Occurrences (Years) ANO-1 B&W 07/74 t 1 0 1.7 AND-2 CE 12/78 LR0 1 0 1.7 Beaver Valley-1 W 06/76 t 1 0 1.7 Byron-1 W 09/85 t 1 0 1 Callaway W 12/84 t 2 0 1.7 Calvert Cliffs-1 CE 01/75 LR0 1 0 1.7 Calvert Cliffs-2 CE 12/76 LR0 1. 0 1.7 Catawta-1 W 01/85 CR0 Continuout Multiple 1.7 Catawba-2 W 08/86 CR0 Continuous Multiple 0.6 Crystal River-3 B&W 01/77 t 1 0 1 Davis Besse B&W 08/77 t 1 0 1.7 D.C. Cook-1 W 02/75 t, LRO, C 1 Few (15) 1.7 D.C. Cook-2 W 03/78 t LRO, C 1 Few (1 ) 1.7 5 Diablo Canyon-1 W 11/84 CR0 Continuous 0 1.7 Diablo Canyon-2 W 10/85 CR0 Continuous Multiple 1.7 Farley-1 W 08/77 CR0 Continuous Multiple 1.7 Farley-2 W 05/81 CR0 Continuous Multiple 1.7 Ft. Calhoun CE 08/73 LR0 1 0 1.7 t Ginna W 12/69 t 1 0 1 Haddam Neck V 08/67 CR0 Continuous Few (15) 1.7 Indian Point 2 W 06/73 t 1 0 1.7 1 l Indian Point 3 W 04/76 t 1 0 1 Kewaunee W 04/74 t 2 ,0 1.7 i l I c , ~.,,., - - - _ ,,,, _.. - ~.
Survey of AFW System Backleakaae Experience (Cont.) ID) Monitoring Number of IC) Commerical Monitoring (*) Frequency Not Pipe T Plant Vender Operation Method (Per Shift Occurrences (Years) Maine Yankee CE 11/72 t 1 0 1.7 McGuire-1 W 09/81 CRO,C Continuous Multiple 1.7 McGuire-2 W 05/83 CRO,C Continuous Multiple 1.7 M111 stone-2 CE 11/75 t 1 0 1 Millstone-3 W 02/86 CRO,t Continuous Few (15) 0.7 N. Anna-1 W 04/78 t 1 0 1.7 N. Anna-2 W 08/80 t 1 0 1.7 Oconee-1 B&W 05/73 t 1 0 1.7 j Oconee-2 B&W 12/73 t 1 0 1.7 Oconee-3 B&W 09/74 t 1 0 1.7 Palisades CE 12/71 t i,, 1 0 1.7 Palo Verde-1 CE 06/85 t 1 0 1 0 0.3 Palo Verde-2 CE 09/86 Point Beach-1 W 11/70 t 1 0 1 Point Beach 2 W 08/72 t 1 0 1 Prairie Island 1 W 12/73 t 2 0 1 Prairie Island-2 W 12/74 t 2 0 1 a Rancho 5 co B&W 10/74 t 1 0 1.7 Robinson-2 W 09/70 t 1 0 1.7 l Salem 1 W 12/76 LR0 1, 0 1.7 Salem 2 W 06/81 LR0 1 0 1.7 San Onofre-1 W 07/67 t 2 0 1.7 j San Onofre-2 CE 09/82 t 1 0 1.7 San Onofre-3 CE 09/83 t 1 0 1.7 Sequoysh 1 W 07/80 t 1
- 0 1.7 a
Survey of AFW System Backlea_kane Experience (Cont.) I Monitoring Nnber of(b) Comnorical Monitoring *) Frequency Hot Pipe T _ Plant Vendor Operation _ Method (Per Shift Occurrences (Years 3 IC) Sequoyah-2 W 12/81 t 1 0 1.7 St. L9eie-1 CE 05/76 t 1 0 1.7 St. Lucie-2 CE C6/83 t 1 0 1.7 Sunner W 11/82 LR0 1 0 1.7 Surry-1 W 07/72 LR0 1 0
- 1. 7 Surry 2 W
03/73 LR0 1 0 1.7 THI-1 B&W 06/74 LR0 1 0 1 Trojan W 12/75 LR0 1 0 1.7 Turkey Point-3 V 11/72 t 1 0 1.7 Turkey Point-4 W 06/73 t 1 0
- 1. 7 Waterford-3 CE 03/85 LR0 1
0 1.7 Wolf Creek-1 W 06/85 CR0 Continuous 0 1.7 Yankee Rowe W 11/60 t 1 0 1.7 Zion 1 W 06/73 t 1 Few (1 ) 1.7 Zion-2 W 12/73 t 1 5 Few (15) 1.7 Explanatory Notes (a) t = hand touch; LR0 = local temperature readout; CR0 = control room readout; c = check performance af ter each use of an AFW pump. (b) A "hot" AFW pump / pipe occurrence is defined as one resulting in the oparator to vent, flush, or otherwise cool down one or more of the affected AFW pump discharge paths. discussion for each plant case); "Few" refers to the reported "few""M occasions especially associated with monthly pump tests, when a check valve may,not seal correctly, as revealed by a post pump test ( ature check c pipe occurren(ce)s.for these cases; "few" is interpreted to mean five hot few occasions when backleakage was obtained in goin ation under conditions where the steam generator pressure levet is low a low Ap condition across the check valve may allow some leakage of s (c) the start of systematic monitoring called for unde represents the period going back from the end of 1986 to at least the time when the regional survey was conducted in April 1985. 4 1
~- APPENDIX 2 Analysis of Steam Bindina Contribution to AFW System Unavailability An upper bound estimate of AQggg3, the contribution to AFW system unavail-ability caused by steam binding of the pumps, may be derived from the backleakage experience obtained in the Category A plants. For this, we have ASAFW3
- 9p CM pNR, P
where, assuming a constant failure rate model. t is the AFV pump unavailability due to steam binding
- q,= 0.5ASB averaged out over the surveillance period t(= 8 hour shift), A is the 3g pues steam binding failure rate, with A tul; 3g
- P is the probability for common mode steam binding failure of the CM redundant AFW pumps, given that the first pump is steam bound; and
- P represents the probability of non-recovery of at least one steam gg bound pu?p prior to dryout of the steam generators.
An estimate of the pump steam binding failure rate, A may be obtained in 3g terrs of the measureable rate of occurrence of a hot pump, p, using ASB
- E(II/N')AHP where p($9/HP) is the probability that a pump becomes steam bound, given that a hot pump (or hot pipe downstream of the pump) is detected, and A is the number J
gp I of hot pipe detection events per year (Hgp) divided by the number of hours in a year (8760), i.e., Agp = Ngp/8760 per hour. i I 1
~ 29 - If a pipe is found to be "hot," pror.edures are in place whereby the operator can initiate timely action to cool the pump and related discharge paths. The threshold temparature for initiating such action varies from plant to plant, with the threshold temperature typically about 20 to 30F above ambient, i.e., a sufficient rise in temperature as to clearly signal the occurrence of back11akage. Under these conditions, the detection of a hot pipe would not in l itself signify the existence of a pump steam binding condition, especially if j the monitoring neint is appetciably down stream of the pump region. Accordingly, the probability of the operator failing to restore the pum.p and discharge lines to an ambient temperature condition before the pump becomes vapor bound may be expected to be reasonably low. This expectation is i reflected in the fact that the plant survey results show that while backleak-age occurrences have been numerous, there have been no reported instances of pump steam binding during the period since systematic monitoring started. It l is judged that this failure probability may be conservat.vely set at less than i one chance in ten, so that l l p(SB/HP) = 0.1. l Fcr the 56 Category,'. plants enhibiting a low backleakage frequency, the survey results (Appendix 1) indicate a total of 30 hot pipe detections over a total period of 73 reactor years (RY)( which yields hp = 30/(73/RY)(8760 hours per year) = $=10 5/hr, so that A3g = (0.1)(5x10 5/hr) = 5x10 8/hr, and q =(0,5) j p (5x10 5/hr) (8 hr)=2x10.s per demand. j With regard to p H, an estimate for the comon mode failure probability may C be based on the approach used in the prioritization of the steam binding issue, l 1 l This used the f act that of the 13 puep steam binding events reported in 1983, three involved the comon mode f ailure of a second pump, out none of a third pues. It was arbitrarily assumed that the comon mode steam binding failure of a third pump would bc 0.1, given that the first two pumps were already steam bound. In this analysis we will assume that the probability of f ailure of the third pump is unity. Hence, P,= (3/13)(1) = 0.2. g 1
t [ h l. With regard to PNP, two different estimates were obtained for the probability of not recovering at ieast one of the steam bound pumps within the 30-minute ties period specified in Section 3.2: one based on estimates provided by reactor operators from four different plants., and one based on an approach along the lines of Swain's method (') for human reliability analysii (HRA). The estimated times obtained from four plants were 10 min.,15 min.,15 min., ) and 20 min., from which one obtains a mean time of 15 min., with a standard deviation of 3.5 min. Assuming thase times fit a Gaussian distribution centered about a mean of 15 min., and essuming conservatively that 3 to 5 min. are expended before the recovery procedure is started, the probability that recovery would not be completert within the remaining time to 30 minutes is .s seen to be P 12x10 NR For the HRA based estimate, the event description assumes that the AFV pumps have been started up on a safety actuation signal subsequent to a loss of MFW transient, and that at time zero the reactor operator trips all running AFW pumps in response to the control panc1 miscations of low suction pressures and tarperatures, low discharge pressures, and rapid fluctuations in pump motor current, all of which cre, for the experienced operator, symptoratic of vapor binding of the pu ps. l The total loss of all feccNater to the steam generators constitutes an emergency situation, for which the operator would direct two or three people (e.g., a process engineer and one or two auxiliary operators) to proceed 1 i directly to the nearby AFW pump rooms, ascertain that the pump casings and piping are hot, and, if so, proceed with the recovery procedures. It is estimated that no more than about 2 minutes are typically needed for the ] auxiliary operators (AQs) and process engineer to reach the pump rooms, including the time to unlock various secured doors by using their badge keys and entering the appropriate identification numbers into the control console. With two or three people f 6reing the recovery tease, the probability of a key badge not bring available or the wrong identification number being used say be expected to be negligibly small (i.e., <10-6). It is noted that consnunication )
with the control room operator can be readily accomplished by means of a nearby extension telephone. The established recovery procedures are simple and few in number, typically involving such tasks as: (i) venting the pumps end piping at various locations and flushing out the pumps by opening the pump discharge drain lines which empty into floor sumps; (ii) closing the vents and drain lines af ter ascertaining that only water is exiting from the vents and drains; (iii) inferring the operator that the venting is complete and that the pumps i are refilled; (iv) the operator starting the pumps and checking to see that the discharge pressures and pump motor currents ire at the correct level and holding steady; and (v) the operator slowly opening the dischstge MOVs to cool down the AFW lines 6nd provide feedwater to the steam generators. It should also be noted that the process engineer and A0s perfoming these tasks are highly trained and skilled, and have previously perfomed these tasks not infrecuently in other instances of pump cavitation by air or vapor binding. I Utilizing Swain's tables :,r a two branch event tree with the options of: (a) the A0 uses procedures but not the check-off feature and fails to open or close the vents / drains, or (b) the A0 uses the procedures and check-off features but still fails to open or close the vents / drains, one obtcins corresponding human error probabilities of 6x10 and 2x10 for the two options, or an overall estimated probability of about 3x10'3 for not carrying out the recovery procedures correctly. l Goth kinds of estimates are seen to be of the order of 10~3 and comparable in magr.itude; we judge that error factors of the order of 10 to be app 1'icable for each estimate. In any event, for purposes of this analysis of AFW system j unavailability due to steam binding, we shall conservatively assuee an upper
~ bound probability for non-recovery of r. pump within 30 minutes equal to pg= 0.1. Hence, for the Category A plants we obtain AFW5 (2x10 5)(0.2)(0.1)=4x10 7/d, AQ c which indicates that the steam binding contribution to overall AFW system unavailability for '.hese plants is negligible compared with the SRP unreli-ability criterion of $1x10 8 It is to be noted that this upper beund estimate of AQ py3 applies to both g two pump and three pump AFV systems, in that it incorporates the assumption 1 bat the t.omon mode failure probability for the third purp is unity. For the Category B plant, we will assume that a hot pipe event occurs as often as once per shift, so that Agp = 0.13/hr, i.e., a..,f requency about 1000 times greater than that for the Category A plant. With regard to p(SB/HP), if account is taken of (1) the availability of the continuous monitoring system for automatically alerting the operator when backleakage occurs, (ii) the placement of the temperature sensors near the point of valve leakage, which allows for earlier initiation of corrective action than if the detection point were nearir the pump, and (iii) the anticipatory attitude of the operator for taking needed action for an event that occurs fregrently, a reasonably conservative assumption for probability of operator f ailure to take action in time to prevent pump steam binding is judged to be in the range of 10.a go go.e, or taking a geometric mean, we have p(SB/HP) equal to 3x10.s. Accordingly, the failure rate for pump steam binding is A3g = (3x10.s) (0.13/hr) = 4x10 */hr. l Under conditions where there is e continuous monitoring system in place, the time period, t, used in calculating the mean pump steam binding untrysilability, J SB, can be interpreted to be the tire between the sounding of the q,= 1/2 A t control room alare and the completion of the restorative action by the operator. Estimates of this time obtained from plant operations personnel indicate a l l
i 1, pean of 15 minutes for such corrective action. For present purposes, we will asseme this time to be equal to I hr., whence q, = 2x10 */ demand and QAFWS * (2x10 4/d) (0.2)(0,1) = 4x10 8/d. In summary we have: ' arameter Catecory A Plants Catecory B Plants P Hot pipe frequency, A 6x10 5/hr 0.13/hr NP p($B/HP) 0.1 3x10.s Steam binaing frequency, A 5x10**/hr 4x10 4/hr SB 2x10 8/ demand 2x10 4/hr Pump unavailability, q, Common mode failure probability, P, 0.2 0.2 g Pump non-recovery, P 0.1 0.1 NR AFWS 4x10 7/ demand 4x10 */oemand I AFWS unavailability, AQ 1 I
a s APPENDIX 3 Estimate of Offsite Health ConSeQuertes The risk estimate developed below follows along the some lines as 5,revious staff estimates for related safely issues involving failure of the AFV system (e.g., Reference 6). The LOMF and LOSP accident sequences associated with AFW system failure both 1 involve a core melt with no large breaks initially in the reactor coolant pressure boundary, and until the core melts through the lower head, the reactor is likely to remain at high prassure with a steady discharge of stease and gases emanating froe the PORV. These are the conditions likely to produce significar. levels of hydrogen generation and combustion.. The Zion and Indian Point PRA studies used a 3% probability for contains>ent failure due to hydrogen burn (the "gassa" failure soot of WASH-1400). We shall also use a value of 3%. With regard to probatill:ty of containment failure to isolate (the "beta" failure mode), the Oconee PRA(5) figure of 0.53% will be used here. If the.ontainment does not f ail by hydrogen burn or non-tsolation, it will be assumed to fail by base sat me)t-through ("epsilon" failure). The PWR release categorie. for the different failure modes t*e as defined in WASH 1400. The whole body dose is esiculated using the CRAC code, based on an assuesd unifore population density of 340 pbrsons per square mile (the mean for U.S. sites), a 50 mtie radius, and a central, midwest plains meteorology. The results are: i Failure Release CRAC Dose Percent Consequences Mode , Category (Pe",on Rem) Probability (Person Rem) gansta PWR-2 4.tx10 3% 1.4x10 beta PWR-5 1.0x10 0.5% 0.05x10 3 epsilon PWR 7 2.3x10 96.51 0.02x10 5 Weighted Mean consegur.nces per core melt = 1.5x10 p,7,,n,7,,, i i
35 - i ~ For the 63 PWRs currently operational, the estimated mean remaining lifetime is 30 years. (For a total of 80 licensens and CP holders, the estimated mean j remaining lifetime is about 32 years). For the integrated exposure we obtsin: Remaining Consequences Integrated ACNF Reactor Years Per Core Melt Exposure Category A 1.4x10 ?/RY 1.7x108/RY 1.5x105 pe rson-rees 36 person-rems (56 plants) Category 8 1.4410.s/RY 2.1x108/RY 1.5x105 person-rees 44 person-rem (7 plants) i n ..}}