|
|
| Line 18: |
Line 18: |
|
| |
|
| =Text= | | =Text= |
| {{#Wiki_filter:_ - - - - - - - - - | | {{#Wiki_filter:}} |
| EPRl/WCG ANALYSIS OF DECAY HEAT REMOVAL RISK AT POINT BEACH NSAC-Il3 Final Report, October 21,1987 Prepared by Science Applications International Corporation 5150 El Camino Real, Suite C 31 L>s Altos, CA 94022 and Westinghouse Electric Corporation Haymaker and Nonhern P' Monroeville,PA 15146 (s
| |
| '.\
| |
| \
| |
| i .s y
| |
| hcipalInyktkat W.J. P'arkinso AIC E.M. Dougheny, S AIC
| |
| .F. Paddleford, Westinghouse Contributors:
| |
| R.K. Hanneman, Wisconsin Electric Power J.R. Olvera, Wisconsin Electric Power S.E. Guokas, Wiscensin Electric Power W.S. Gough, SAIC R.J. Lutz, Westinghouse N.L. Bums, Westinghouse Prepared for Nuclear Safety Analysis Center Electric Power Research Institute 3412 Hillview Avenue Palo Alto, California 94303 and Westinghouse Owners Group NSAC Project Managers G.L. Vine J.J. Haugh 8712110086 871204 PDR FOIA WEISSG7-714 PDR w
| |
| | |
| NSAC PERSPECTIVE Shutdown Decay Heat Removal has been one of the most analyzed and debated nuclear safety issues of the 1980s. 'Ile Nuclear Regulatory Commission (NRC) elevated this issued to " Unresolved Safety Issue" status in 1980, with the designation USI A-45. Regulatory analysis of decay heat removal (DHR) has involved studies of DHR capabilities, Freening criteria, and plant-specific vulnerabilities and modifications. Six plant specific case studies of DHR risk were conducted by Sandia National Labora-tories and published earlier this year. These case studies were conducted using Probabilistic Risk Assessment (PRA) techniques. One of the first of the six probabilistic studies was performed on the Point Beach Nuclear Power Plant, owned and operated by Wisconsin Electric Power Company.
| |
| The Nuclear Safety Analysis Center (NSAC) initiated a study of decay heat removal in 1982, with emphasis on operating experience analysis and DHR during the " Residual Heat Removal" phase (see IS AC Reports 52,83, 84,88). This phase, a subset of the much broader DHR issue,is concemed with reaching and maintaining cold shutdown. In 1984, an NSAC representative was selected to serve on a peer review group organized by Sandia to review their DHR research. In November 1985, drafts of the first two DHR case studies were provided to the peer review group for review, one of which was the Point Beach case study. The EPRI review of those draft case studies revealed a large number of significant conservatism that biased the studies toward concluding that expensive hardware backfits might be cost-effective.
| |
| After the EPRI comments were sent to Sandia, it became apparent that a reanalysis of the DHR risk at )
| |
| one of the case study plants would be an effective way of quantifying the degree of conservatism in the f Sandia analyses. EPRI embarked on this project to reanalyze DHR risk at Point Beach in June 1986, in cooperation with Wisconsin Electric Power (WEP) and the Westinghouse Owners Group (WOG).
| |
| SAIC and Westinghouse were selected to be the contractors. Sandia published the six case studies in July 1987. Some of the peer review comments had been considered in the f' mal version of the Point Beach study.
| |
| }
| |
| j Both the NRC-sponsored case study and the EPRI/WOG reanalysis evaluated DHR risk for a broad range of initiating events and accident types, including almost all intemal plant sequences except large ]
| |
| j
| |
| ,; ipe ruptures and plant transients with a failure to scram (both of which are accidents outside the scope 1
| |
| iii
| |
| | |
| of decay heat removal), and including all external events (or "special emergencies") that were consider-ed credible. Both studies evaluated a range of potential modifications that had been proposed by Sandia.
| |
| These proposed improvements ranged from small hardware backfits to large expensive add-on system backfits such as the dedicated Shutdown Decay Heat Removal (SDHR) system which would cost almost 5100,000,000 per unit.
| |
| Both studies showed that this SDHR backfit concept is not justified on a cost-benefit basis. In the NRC study, despite conservative analysis assumptions that would tend to bring this ratio closer to unity, the SDHR was shown to provide about 1/2 cent of benefit in terms of averted doses for every one dollar spent installing it. In the EPRl/WOG study, this same ratio was about 0.002 cents benefit for every one dollar spent. In addition, the EPRI/WOG best-estimate analysis revealed that in terms of doses receivec, t ersonnel would receive 150 times more dose from installinc and maintaining the system than would be averted offsite throuch proiected risk reduction. Industry reviews of the SDHR also revealed a number of questions about the operational complexity and system interaction problems that the SDHR would introduce. Arguments about the "non-quantifiable benefits" of the SDHR do not withstand close scrutiny. These reviews show that the SDHR concept is inappropriate for Point Beach.
| |
| The above items, in combination with the equally minute cost-benefit ratios for the SDHR at the five other case study plants, indicate that the SDHR is inappropriate for all plants analyzed in the A 45
| |
| ,arograrn. Barring unique circumstances that might exist at a particular plant, the inappropriateness of the SDHR concept should be considered a generic conclusion of the DHR research to date. Considering the approximate 100 operating nuclear power plants in this country, this conclusion can save the U.S.
| |
| electric utility ratepayers, taxpayers, and stockholders approximately TEN BILLION DOLLARS in avoided costs for capital investments that cannot be justified.
| |
| Although our EPRl/WOG reanalysis provided further confirmation of the inappropriateness of the SDHR concept, this was not the primary purpose of our study, since the NRC case studies form an ade-quate basis for that conclusion. The primary purpose of this study was to provide a best-estimate analysis of DHR risk at a selected A-45 case study plant,in order to help resolve questions about the degree of conservatism in the NRC case studies. A better appreciation of this question of excess conservatism is expected to be an important contribution to the ultimate resolution of A-45. The Nuclear Management and Resources Council (NUMARC), representing the U.S. nuclear industry in its efforts to resolve generic regulatory issues, has endorsed this effort and supported a detailed review of the analysis.
| |
| The results of this reanalysis were: an approximate factor of thirty reduction in core-melt frequency for he sequences included in the scope of the NRC study; an approximate factor of seven reduction in the iv I
| |
| | |
| offsite consequences of these sequences, over and above the cere-melt frequency reduction; and an proximate 50%-400% increase in the estimated cost of the various backfit proposals.
| |
| The EPRI/WOG reanalysis concurred with the NRC case study conclusion that none of the proposed backfits are cost effective. While it is possible that some improvements could be useful at Point Beach, these particular Sandia proposals cannot be justified on the basis of public risk-based value-impact analysis.
| |
| The core-melt frequency estimate in this reanalysis is 1.0 x 10-5 per reactor-year. This estimate. is a factor of ten lower than the core-melt frequency target in the NRC's Safety Goal. This observation should be accompanied by two caveats: the scope of analysis includes most, but not all risk sequences; and the analysis methods do not qualify as a full " level one" PRA.
| |
| Many of the design and operational features that account for this reduced estimate of risk at Point Beac have been instituted by WEP since the TMI 2 accident. The implementation of WOG sponsored Emergency Response Guidelines, fire protection, and the addition of seismically-qualified batteries are particularly significant. Ongoing programs such as these (and in particular, the A-44 Station Blackout Program and A-46 Seismic Verification) will continue to identify opportunities for further
| |
| :enprovements that can be cost-effective. Thus,it appears that the optimum resolution strategy for the
| |
| . 45 program would be to take maximum advantage of the many programs already addressing the specific issues that are subsets of the larger DHR issue. Any remaining potential " outliers" would be identified and corrected by actions pursuant to the Commission's Severe Accident Policy.
| |
| Gary L. Vine DHR Issue Manager I
| |
| 1 l
| |
| i V
| |
| | |
| l .t ACKNOWLEDGEMENTS During the course of this project numerous individuals and organizations made important contributions to this industry-sponsored review and evaluation of decay heat removal risks. Their collective efforts have been appreciated.
| |
| Critical review of the six NRC-sponsmed Decay Heat Removal (DHR) Case Studies was performed by
| |
| ~
| |
| the NUMARC Working Group on DHR: G. Neils (Northern States Power .Co.), Chairman, DHR Working Group; J. Jeffries and R. Oliver (Carolina Power and Light Co.); R. Newton and J. Olvera (Wisconsin Electric Power Co.); D. Helwig and Gl Beck (Philadelphia Electric Co.); A. Ladieu (Yankee Atomic Electric Co.); G. Swindlehurst (Duke Power Co.)'; D. James and M. Schoppman (Florida Pcwer
| |
| - and Light Co.); M. Meisner (Louisiana Power and Light Co.); L. Taylor and T. Enos (Arkansas Power
| |
| ' and Light Co.); X. Polanski (Commonwealth Edison Co.); and D.' Reeves (Nebraska Public Power
| |
| 'istrict).
| |
| The preparation of this reanalysis of the Point Beach Case Study was assisted greatly by the contribu-tions made by the following individuals: G. Neils (Northern States Power Co.), Chairman, NUMARC Working Group on DHR; W. Layman and C. Stepp (EPRI); A. Ladieu (Yankee Atomic Electric Co.), D.
| |
| Dube (Northeast Utilities Service Co.), and G. Swindlehurst (Duke Power Co.) - Westinghouse Owners Group; R. Newton (Wisconsin Electric Power Co.), J.F. Riley and E.M. Grubb (Science Applications International Corp.); and B.G. Cassidy, W.S. Lapay and K.G. Davenport (Westinghouse Electric Corporation).
| |
| l l
| |
| /
| |
| | |
| 'M CONTENTS Section P;Is 1 INTRODUCTION AND
| |
| | |
| ==SUMMARY==
| |
| l-1 Overview 1-1 Background 12 NRC Case Study Results 1-4 Objectives and Scope of this Report 19 EPRI/WOG Analysis Overview and Results 1 11 Pnmary Differences Between the EPRI/WOG and NRC Case Study Analyses 1-14 2 OVERVIEW OF THE APPROACH 21 Methodology 21 Overview of the Significant Accident Types as Determined by the NRC Case Study 25 Accident Types and Potential Plant Modifications frorn the NRC Case Study 28 3 EPRI/WOG INTEGRAL PLANT MODEL 31 Transient Event Sequences 31 Small LOCA Event Sequences 3-17 Containment Safeguards and Support System Success Criteria 3 20 ;
| |
| 4 HUMAN RELIABILITY AND RECOVERY ANALYSIS 41 Introduction 4-1 Comparison of EPRI/WOG Study and NRC Case Study 45 EPRI/WOG Study Analysis of Key NRC Case Study Human Error Probabilities 4-7 EPRI/WOG Study Analysis of Recovery Actions 4 10 5 INITIATING EVENT AND COMPONENT DATA ANALYSIS 51 Intmduction 51 Comparison of EPRI/WOG and NRC Case Study 5-3 Review of Other Significant Events 59 Review of Common-Cause Data 5-10 EXTERNALEVENTS ANALYSIS 61 Intmduction 6-1 vii
| |
| | |
| CONTENTS (Continued)
| |
| Section h Fire Analysis 6-2 Internal Flood Analysis 6-10 Seismic Analysis 6-11 7 ANALYSIS OF THE NRC'S DEDICATED SHUTDOWN DECAY HEAT REMOVAL SYSTEM 7-1 Description of the SDHR 7-1 Fault Analysis of the SDHR for Significant Accident Sequences 73 Sensitivity Analysis of Potential Adverse Public Risk Impacts 7-10 Comparison of SDHR to Existing Point Beach Equipment and Procedures 7-13 8 CORE-MELT SEQUENCE RESULTS 8-1 Summary and Conclusions 81 9 PUBLIC HEALTH CONSEQUENCE ANALYSIS 9-1 Introduction 9-1 Overview of NRC Case Study Approach 9-2 Overview of the EPRI/WOG Approach 9-3 10 VALUE-IMPACT ANALYSIS 10-1 Summary and Conclusions 10-1 Overview of NRC Case Study Results and Approach 10-7 EPRI/WOG Value-Impact Analysis and Example Application 10-8 11 REFERENCES 11-1 APPENDICES A HUMAN RELIABILITY AND RECOVERY ANALYSIS DETAILS A-1 B INTERNAL EVENTS ACCIDENT SEQUENCE ANALYSIS DETAILS B1 C PUBLIC CONSEQUENCE ANALYSIS DETAILS C-1 D VALUE-IMPACT ANALYSIS DETAILS D-1 E DETAILS OF SDHR INEFFECTIVENESS AND UNAVAILABILITY E-1 l QUANTIFICATION I viii i
| |
| _-___________-__-__---___-___--_-O
| |
| | |
| Section 1 INTRODUCTION Aht
| |
| | |
| ==SUMMARY==
| |
| | |
| OVERVIEW This Nuclear Safety Analysis Center (NSAC) report presents an analysis of Decay Heat Removal (DHR) risk at the Point Beach nuclear power plant. It is a trport of work sponsored by the Electric Power Research Institute (EPRI) and the Westinghouse Owners Group (WOG), and conducted by prin-cipal investigators from Science Applications Intemational Corporation (SAIC) and Westinghouse Electric Corporation (Westinghouse). Major assistance has been provided by representatives from Wisconsin Electric Power Company (WEP), the owners and operators of Point Beach. ;
| |
| Point Beach is a two-unit site on it e shore of Lake Michigan, about 90 miles north of Milwaukee. Each of the two units is a two-loop Pressurized Water Reactor (PWR) designed by Westinghouse. Point Beach went into commercial operation in 1970, and each unit has a net electrical output of 497-MWe.
| |
| DHR risk at Point Beach has been analyzed by Sandia National Laboratories (Sandia) and reported in NUREG/CR-4458 (1). That case study analysis was sponsored by the Nuclear Regulatory Commission (NRC) as part of the program plan to resolve " Shutdown Decay Heat Removal Requirements." This issue was designated as Unresolved Safety Issue (USI) A-45 in late 1980, and has undergone intensive study by NRC; the A-45 prime contractor, Sandia; and other A-45 subcontractors.
| |
| The Sandia analysis of Point Beach was one of six plant-specific case studies sponsored by NRC, all conducted by Sandia. The six plants were selected carefully to represent a broad range of reactor and DHR system designs, so as to form the basis for a consistent set of generic or group-generic new licensing requirements for DHR at U.S. reactors.
| |
| As mentioned briefly in the NSAC perspective, these six NRC-sponsored case studies were considered to be unduly conservative in their analysis assumptions and methods. Following the preparation of j extensive industry review comments that identified excessive conservatism in drafts of the first two case studies (Point Beach and Quad Cities), a decision was made to sponsor a reanalysis of one of those case study plants in order to quantify the degree of conservatism. Point Beach was selected for this analysis. Early drafts of the NRC-sponsored case studies were redone by Sandia in parallel with the l
| |
| l l
| |
| l 1-1
| |
| | |
| EPRI/WOG effort. These final versions of the NRC-sponsored case studies accounted for some com-
| |
| 'ents by industry while this reanalysis was in progress.
| |
| Since many organizations have been involved in the analysis and review of these two studies of Point Beach,it was decided for the sake of brevity to identify them by their sponsoring organizations. Thus, the terms "NRC Case Study" and "EPRI/WOG study" are used throughout this report.
| |
| Section 1 of this study discusses the purpose and scope of both the NRC's A-45 program and the NRC Case Study results. The objectives and scope of this EPRI/WOG study are then described. A discussion follows of the results of this study, both in terms of the quantitative results and why they differ from those given in the NRC Case Study.
| |
| BACKGROUND Even after the fission chain reaction in a nuclear reactor has been halted, the continuing radioactive decay of fission products and irradiated core materials produces a significant amount of heat. At a typical 1000-MWe nuclear power plant the value of this decay heat is about 20-MWth at 24 hours after shutdown from full power.
| |
| 'arious systems are provided on both PWRs and Boiling Water Reactors (BWRs) to remove this decay neat under both normal and emergency conditions. Immediately after shutdown, decay heat typically is removed on both PWRs and BWRs by continued steaming via the turbine bypass valves to the main condenser, and by continued feedwater addition. Other backup systems can be used to remove steam and to add feedwater under both normal and emergency conditions, including Atmospheric Steam Dump Valves (ASDVs), and makeup to steam generators via Auxiliary Feedwater (AFW) systems on PWRs; and steam relief to the torus / suppression pool via Safety Relief Valves (SRVs), and reactor makeup via Reactor Core Isolation Cooling (RCIC) systems on BWRs. During the late stages of plant cooldown, the low pressure Residual Heat Removal (RHR) system is connected to the primary system to complete the cooldown process to cold shutdown conditions, and to maintain the desired cold shutdown core temperatures during maintenance or refueling. Rese systems require support systems such as AC and DC power, service water, component cooling, instrument air, and so forth, to function properly.
| |
| NRC'S Shutdown Decav Heat Removal Requirements Program (A-45) ne objective of the NRC's USI program on DHR is "to evaluate the adequacy of cturent licensing requirements for ensuring that nuclea power plants do not pose an unacceptable risk to the public as a result of failure to remove decay heat."(2) This issue was designated USI A-45 on December 24,1980.
| |
| n estimated $7 million has been spent on research to resolve this issue over a seven-year period. A rinal technical summary report and a cornpanion regulatory analysis repon are expected to be available j 1-2
| |
| | |
| for public comment in March 1988, and actions taken by the Commission to resolve the issae are
| |
| ;heduled for completion by October 1989.
| |
| At the time the A-45 program was staned there was much interest at the NRC and arnong some other nuclear safety experts in following the lead of a few European countries in backfitting new add on,inde-pendent, bunkered DHR systems. A September 1981 repon to President's Nuclear Safety Oversight Committee (NSOC) (1), for example, recommended top priority for safety research on DHR. In panicu-lar, the repon to NSOC suggested that new research be conducted along the lines of the German and Swiss approach; i.e., the addition of new, redundant, independent, safety-grade systems capable of removing all post shutdown decay heat while adding sufficient coolant and emergency feedwater at full system pressures. In suppon of that recommendation the repon specifically noted that, "Several countries, including Germany and Switzerland, have required not only highly reliable shutdown heat removal systems but in addition, a dedicated bunkered system which provides backup to a loss of all normal offsite and onsite emergency power, and also protection against fire or sabotage'." (1)
| |
| In the early years of the program the A-45 Task Action Plan (TAP) called for the publication of 14 mile-stone reports to establish the technical decision-making methodology to meet the program's objectives and to established the ground rules for plant-specific analyses. Only two of these milestone reports have een published. The requirement to produce the other 12 repons was deleted in December 1985 by l
| |
| Revision 5 to the TAP. A requirement for seven plant specific DHR case studies replaced the previous requirement for milestone repons.
| |
| DHR Case Studies. In 1984 the A-45 program characterized the DHR systems and functional capabili-ties of almost every operating U.S. reactor plant and organized those individual plants into eight groups ,
| |
| 1 of plants with similar DHR characteristics. The gmuping plan was derived from another A-45 sponsor-ed study by Brookhaven National Laboratory (.4.). 4 From those eight groups, seven representative plants were selected to be analyzed in detail: Point Beach, Quad Cities, Turkey Point, C% , ANO-1, St.
| |
| Lucie and Trojan. The Trojan analysis subsequently was dropped, so that only six case stumes were fin-ally published.
| |
| Each plant-specific analysis or case study addresses nearly the same breadth of issues as a full plant Probabilistic Risk Assessment (PRA). Although the number of systems considered in the analysis is quite comprehensive, the level of detail in plant specific system modeling and plant data do not qualify the case studies as full Level 1,2 or 3 PRAs. Each case study places heavy emphasis on the evaluation of "special emergency" vulnerabilities such as earthquake, fire, intemal and external flooding, tomado I
| |
| )
| |
| tnd sabotage. Each study analyzes all internal sequences considered pan of the DHR issue, and therefore includes all transients, losses of offsite power, small break LOCAs, etc. The only major sequences not included in the scope of A-45 analysis are Anticipated Transient Without Scram (ATWS) 1-3
| |
| - - _ - \
| |
| | |
| l and Large-Break Loss of Coolant Accident (LBLOCA). Each case study includes an evaluation of off. )
| |
| 1
| |
| -ite consequences and a vaiety of alternadve modifications to reduce risk. Among these modifications l 1
| |
| is a Dedicated Shutdown Decay Heat Removal (SDHR) system. l l
| |
| Dedicated Decav Heat Removal Systems. A study of alternate decay heat removal concepts, conducted by Sandia for the NRC Office of Research was published in April,1981 as NUREG/CR-1556. That pre-liminary study of add-on DHR concepts was used by Sandia as a basis for developing a more comprehensive three-volume study, which was published in June,1983 (fi). Although the latter study was sponsored by the NRC's Office of Research in suppon of its severe accident risk reduction pro- i gram, it was considered by the NRC to be "an important contribution to the resolution of one aspect of the NRC's USI A-45 (... evaluation of alternative means for improving DHR system reliability)." (fi)
| |
| {'
| |
| Although the study was somewhat critical of the value of European approaches to alternative DHR sys-tems, it did present some non-European altemative designs which were shown to improve core melt probability by a factor of ten on some U.S. plants. These designs were estimated to cost between 540 and 590 million to install, including power replacement costs. The study included a cost-benefit analy-sis to determine which of six PWR and three BWR add-on designs was best. These top two designs were then used as the basis for the dedicated SDHR in the NRC case studies. The PWR add-on design is a single train of high pressure tractor coolant makeup water and a single train of emergency feed-water, with dedicated power and water sources.
| |
| NRC CASE STUDY RESULTS The overall core melt frequency for Point Beact was detennined by the NRC Case Study to be 3.13E-4.
| |
| The breakdown of this core melt frequency in terms of sources of risk is given in table 1-1. The dominant accident sequences and their contributing core melt frequencies are given in table 1-2.
| |
| Sandia identified 11 potential vulnerabilities frotn the internal analyses, after having examined the i
| |
| above-listed dominant accident sequences and having conducted a recovery analysis. The top seven I internal vulnerabilities, ranked in order of importance, and the corresponding proposed modification to correct that vulnerability are provided in table 1-3. A summary of the proposed modifications for the major vulnerable areas or components identified in the special emergency analysis is provided in table 1-4.
| |
| i
| |
| )
| |
| After developing a proposed modification for most internal and external vulnerabilities, Sandia lumped them into four " alternatives" for evaluation with value impact (V I) analysis (table 1-5). The four l alternatives are explained in table 15.
| |
| .andia then calculated the change in core melt frequencies that would result from the installation of each of these alternatives. V 1 analyses were performed on each alternative, using both a $1,000 per 1-4 1
| |
| l
| |
| | |
| Table 1-1
| |
| . NRC Case Study Results (from Table 9.2 of Reference 1)
| |
| Source of Risk Core Melt Freauency Per Year Intemal 1.4E-4 Total Special Emergency
| |
| * 1.74E-4 Seismic 6.1E-5 Fire 3.2E Internal Flood 7.7E-5 External Flood 1.9E-8 Extreme Wind 4.0E-6 Lightning 5.8E-8 Total Risk 3.13E-4
| |
| % Contribution Internal 44 Special Emergency 56
| |
| * The terms "s;xcial emergency" and " external events" are used interchange 31y in this introduction. Subsequent chapters of the EPRI/WOG study use external events, the term most.
| |
| commonly used in PRAs. The NRC Case Study primarily uses "special emergency" because some of these events are not strictly externalin nature.
| |
| /
| |
| 1-5
| |
| | |
| y i
| |
| Table 1-2 NRC Case Study Internal Accident Sequences (from Table 2.1 and page 2 20 of Reference 1) r Cn Melt Probability Rank Dmninant Accident Sequences ner fear (AfterRecoverv) 1 Small LOCA and failure of ECC recirculation 4.7E 5 (S2MH1'H2')
| |
| 2 . Long-term station blackout (LTSB) 3.6E-5 3 Transient induced LOCA and failure of ECC 2.5E-5 recirculation (T3QHl'H2')
| |
| 4 Small LOCA and failure of ECC injection 8.7E-6 (S2MDID2) 5 Loss of offsite power transient; and failure 6.7E-6 of main feedwater, auxiliary feedwater and F&B (T1MLE) 6 Transient induced LOCA and failure of ECC 4.6E-6 injection (T3QDID2) 7 Loss of power conversion system and failure of 3.5E-6 main feedwater and ECC recirculation (T2MQH1'H2')
| |
| All Others M Total 1.34E 4 o See Table 2-1 for NRC Coe Study acronyms 1-6
| |
| | |
| Table 1-3 NRC Case Study Intemal Modifications (from pages 2-11 through 2 20 of Reference 1)
| |
| Core Melt Probability Intemal Vulnerability Per Year Prooosed Modification
| |
| : 1. Long term station blackout 3.6E-5 Install a turbine-driven generator to caused by depletion of the supply vital AC and DC loads, and an station batteries or the con- additional 270,000-gallon condensate densate storage tank (I.V. #11) storage tank shared between units 0.M. #3 & #11)
| |
| : 2. Failure to switchover from 3.1E 5 Add a more prominent alarm to warn that emergency core injection to switchover to recirculation is necessary recirculation (I.V. #1) and imminent (I.M. #1)
| |
| Common mode failure of safety 2.4E 5 (No proposed modification) system pumps (AFW, CCW, LPI, SW, HPI) 0.V. #6)
| |
| : 4. Failure of the low pressure 2.2E-5 Install a third independent low pressure injection system in the re- train with an addinonal suction line from circulation mode (I.V. #8) the sump and cross-ties to the original trains 0.M. #8)
| |
| : 5. Failure of ECC recirculation 1.5E-5 Install parallel, manual valve-to-valve due to RHR pump cooling fail- XV 30 in the RHR component cooling ute by a valve failure (I.V. #4) water line, and check the valve positions /
| |
| once per shift 0.M. #4)
| |
| : 6. C7mmon mode failure of safety 1.3E 5 Oncluded in Modification #8) syLtem valves 0.V. #7) l I
| |
| : 7. Failure of the AFWS turbine- 1.0E-5 Installindependent diesel-driven AFWS driven pump 0.V. #9) pump in parallel to the turbine-driven pump, but sharing the suction and dis-charge lines 0.M. #9) 1-7
| |
| - _ _ - _ _ _ _ _ _ l
| |
| | |
| f.
| |
| Table 1-4 NRC Case Study Special Emergency Modifications (from Appendices C.D E,F,G and H of Reference 1)
| |
| Core Melt Initiating Vulnerable Probability Ey3,nt Area /Comoonent ner Year Proposed Modification Seismic Refueling water Use spent fuel pool as a Seismic Category storage tank I backup water source for HPl or LPI (S.V. #1) (S.M. #1)
| |
| Electrical cabi- Provide additional anchorage for the nets and battery 4160V safeguards buses,480V trans-racks (S.V. #2, transformers and transformer buses, SI
| |
| #3,#4,#5) pump buses,instmmentation power supply inverters, battery chargers and battery racks (S.M. #2, #3, #4, #5)
| |
| PORV instrument air Install an additional Safety Class 3 system (S.V. #6) nitrogen bottle system for each PORY (S.M. #6)
| |
| Seismic Total = 6.1E-5 Fire AFW pump room 1.3E-5 Install automatic dry-pipe preaction type (F.V. #1) of water suppression system, and water-proof pumps and control circuit boxes in 1
| |
| the room (F.M. #1) 4160V switchgear 2.0E-5 Relocate main battery distribution bus (F.V. #2) DO 1 and its charger and invenerin another fire zone (F.M. #2) i i
| |
| Intemal Service water 7.7E-5 Extend dividing wall between pumps to j Flood pump room (SPRAY the ceiling (SPRAY M.#1) '
| |
| V. #1)
| |
| Extemal Turbine, auxiliary 1.9E-8 None Flood and SW buildings Extreme DG exhaust stacks 4.0E-6 Improved exhaust stack structural Wind (W.V. #6) supports (W.M. #6) l 4
| |
| Lightning DC bus 5.8E 8 None 1-8 L__ _ ___
| |
| | |
| Table 1-5 NRC Case Study Alternatives Installation and O&M Alternative Cost MRC est3 Modifications included 1 5 7,568,000 Intemal Mods #1, #2, #4; Seismic Mods #1-6; Spray Mod; and Fire Mods #1 and #2 2 14,876,000 All of Altemative 1. plus Internal Mods #3 and #11 3 24,806,000 All of Alternative 2, plus Intemal Mods #8 and #9; and Wind Mod #6 4 64,164,000 Dedicated Decay Heat Removal System person-rem cost to-dose ratio and a conservative formula for calculating on-site avened costs (consis-ting of replacement power costs, loss of investment costs, and site cleanup costs). Following NRC guidance, Sandia calculated V-I ratios by considering all costs (installation, operations and maintenance, replacement power during installation, and all onsite avened costs) as a positive or negative " impact,"
| |
| and all doses (onsite and offsite avened doses, installation and operation doses) as a positive or negative "value."
| |
| A summary of the NRC Case Study V I measures for Point Beach is provided in table 1-6. Note that none of the V-I ratios approaches unity,i.e., no alternative is shown to be cost-effective. The V-1 ratio for the dedicated SDHR is 0.007, based on $1,000 per person rem for offsite avened costs. When onsite avened costs and installation and O&M doses are included, the V-I ratio becomes 0.004 OBJECTIVES AND SCOPE OF THIS REPORT The objectives of this EPRI/WOG study were to review the NRC A-45 Case Study of Point Beach from an industry perspective and to assess the impact of the review results on the conclusions in the hTC study. The conclusions obtained from this industry review should provide valuable input to the process of resolving USI A-45.
| |
| Because resolution of a generic issue is involved but a plant-specific study is being used, another objec-tive of this study is to identify wherever feasible those areas of the study which could impact other hPC case studies. Therefore, plant-specific results often will be described in terms of their generic implica- ,
| |
| I tions. j i
| |
| .he scope of this effon includes review of those issues which could increase the NRC's estimate of risk, ,
| |
| I as well as those that could decrease it. However, even this reanalysis still cannot provide the same level 1-9
| |
| | |
| I1lilil ll1 ; I :
| |
| _ s r
| |
| a m n l
| |
| e R 3 3 6 5 l
| |
| P 3 8 7 2 ore r s D p p- 2 1 6 3 )
| |
| t s D 3 9 7 2 )
| |
| 4 1 8 n
| |
| oC o V de )
| |
| +
| |
| ei t 3 )
| |
| ss 6 e an t i0 n
| |
| 8 3
| |
| 3 8
| |
| 6 3
| |
| 0 0 V
| |
| (
| |
| s o
| |
| BO f 1 e V 3 7 8 6 f o
| |
| d i sd t nx B 2- 8 7 6 l a
| |
| l sn ya ee$
| |
| NB( N
| |
| - 1 5
| |
| ht n
| |
| o 4 ae r o
| |
| i t
| |
| a ni t r
| |
| Asff w e 1 .
| |
| t p n
| |
| VO e o
| |
| * s e c n 9 3 3 4 e c I t i
| |
| R 7 6 3 0 r p
| |
| i vr
| |
| - a I 1 0 0 0 (
| |
| ) VR V 0 0 0 0 - )
| |
| e s
| |
| e ); ss -
| |
| u I t n
| |
| l a T Vo c i
| |
| V )
| |
| + e d
| |
| n l
| |
| a 2 l a r s V Ib t
| |
| n r
| |
| a m R o 8 8
| |
| 6 7 9 7 6 0 f o
| |
| V
| |
| ( t r a )4 no C)e) s o r l
| |
| l e r P 7 0 5 3
| |
| * f e Vi t a
| |
| s D pe p-t 1 0 2 7 ov+l l (1
| |
| s( o D I 2 3 7 h t
| |
| * a 3 at r
| |
| e8 r5 u
| |
| C o htr iteVs ( n s4 t e w o n - s i 1
| |
| w o )2 tan N V l
| |
| a4 i ) t 1 e- f s 6 n 2 0 9 7 t -
| |
| MC R f O
| |
| t i0 f
| |
| o 1
| |
| 9 1 5 7 eV s n aVo/' N l
| |
| t t/
| |
| c f e 1 V 1 4 3 6 e/e r s o+(
| |
| T= 'V V/
| |
| eG I
| |
| pI e pE o
| |
| s t nx ee$ B 7- 4 1
| |
| 4 2
| |
| 3 6- ( Tr p -
| |
| (t 1 N NN 6- -t s
| |
| R i s
| |
| NB( N - -
| |
| = = ( 1 V4 V= = =
| |
| 1 l iU y o o= T (= nn l
| |
| e eN u ,
| |
| l a
| |
| n V BP R 'V I V3I BP
| |
| = +RnVR b
| |
| NDNNNVVND l
| |
| a a1 A
| |
| T Vl 1 fI o e V o o 0 9 2
| |
| 8 7 0
| |
| yl 1 t i
| |
| a R
| |
| I 5
| |
| 0 0 1
| |
| 0 0 rba VR V 0 0 0 0
| |
| : a. T i
| |
| r.i :
| |
| e uc Sr u yo d d du(S e n n t
| |
| l r a esi t
| |
| eo R 8 7 9 5 ) a S
| |
| t ovoa t D 7 8 8 9 1 s n
| |
| e s TADR A 0 0 0 0 3 2 23V o C
| |
| a m e x x+3 i t
| |
| a r
| |
| C r-yPs2 me e p
| |
| n R
| |
| eed ) oi t Aox d e o, N t t m 2 sl r i xd s g 1 eb eod fsvo- esree i
| |
| pa0 n 1 f r V 2 4 4 1 1 6
| |
| 0 3 b0 r t i r
| |
| ~
| |
| /
| |
| O A D (p 6 7 8 r o51 eev s e e )
| |
| 7 e 2 prp5 a a n V c i 0 e e= =es e 0 t l g f n
| |
| n 0,m s s a e- 4-1
| |
| - o ob o e )s*o i t
| |
| ,t 4- 4- 4- $ ed d( a nsh a cioete/r o ot
| |
| ~
| |
| t e - r cr E E E E i c r -
| |
| u otbtay m t t t o dCl e
| |
| RiMPb n eol ri i
| |
| A P 6 1
| |
| 2 1
| |
| 5 2
| |
| 0 6
| |
| 2 2
| |
| 9 2
| |
| d teinf t
| |
| n sf s V i )2 g al ew .t l
| |
| a nt c
| |
| a n gddoo+l . in a
| |
| u e s n e l
| |
| a n e e 1 e et se v at r rt Vul ai n pr eh e e( l a o a( t sCv v r- e ei v e
| |
| s = AA =n=V T( m=o o =R D m=2D1 R l
| |
| t t a o =
| |
| ann 1 2 3 4 PI
| |
| * AVVAVT
| |
| - I I
| |
| V 5o lll' u
| |
| | |
| of confidence in risk results as could a level 1,2 or 3 PRA. Within the scope of the EPRl/WOG study I
| |
| nphasis was placed on reanalysis of those areas where risk is mort sensitive to assumptions, input data or modeling techniques. These areas are described in more detail in the next section.
| |
| 1 Furthermore, the EPRI/WOG study has other limitations due to the availability and traceability of infor-mation in the NRC Case Study. In a few areas documentation of the NRC report is either limited or contradictory, and a detailed review was not possible. An example of contradictory documentation can be found in the common cause data. 'Different values are reponed in the text,in the NRC Case Study Appendix B summary table, and in the cut set lists. In this particular case, as in many others, the limited documentation and/or contradictory information was resolved by reconstructing the final results from the accident sequence cut set lists. The EPRI/WOG study began with a baseline case study model which differed only a few percentage points from the results reported in the NRC Case Study.
| |
| As with the limitations inherent in the available data, the NRC Case Study does not include certain acci-dent sequences. LBLOCA, reactor vessel rupture, interfacing systems LOCA, and ATWS events were excluded. Furthermore, pressurized thermal shock and reactor coolant pump seal leakage were not considered because separate analyses of these events are being performed under separate safety issues.
| |
| These limitations in scope were maintained as wellin the EPRI/WOG study.
| |
| EPRI/WOG ANALYSIS OVERVIEW AND RESULTS This subsection provides an overview of the major results and conclusions of this repon. Discussions of the methodology and detailed results can be found in later sections.
| |
| The Point Beach plant model was loaded into SAIC's Risk Management Query System (RMQS) (2).
| |
| This software package was used for the majority of the reanalysis. The initial plant model and data were very close to the Sandia model and data, as confirmed by a verification run before revised data and inodeling were introduced. Both the model and data were then modified to provide a best estimate of I
| |
| Point Beach risk for the same analysis scope defined by the NRC Case Study. More realistic assump-tions and recovery actions were permitted. The same modifications as proposed by Sandia were then evaluated for their effects on core melt frequencies. However, the modifications were evaluated indivi-dually, without grouping them together in " alternatives." V-I analysis of each modification was conduc-ted using the more logical approach of considering averted doses (and, for academic comparison only, onsite averted costs) as values (all positive); and installation costs, O&M costs and installation doses as impacts (all positive).
| |
| The results of this reanalysis show that the probability of core melt at Point Beach is about thirty times lower than the estimate in the NRC Case Study. The overall offsite consequences are shown to be about 200 times lower than the NRC estimate due to a further factor of seven or greater reduction in public 1-11 t
| |
| | |
| risk, due to more realistic source term estimates. The SDHR was shown with high confidence to not be
| |
| 'ost-effective. The most cost-effective modification, more than any of Sandia's proposals, was found to
| |
| .>e a new set of Seismic Category I batteries installed by WEP last year.
| |
| Based on these results, it was concluded that existing DHR is adequate at Point Beach and that no modifications are necessary for providing adequate protection of public health and safety. Although no modifications are evaluated as cost-effective in terms of public protection,it is possible that WEP may consider some modifications to be attractive foi purposes of improved plant performance and availability.
| |
| The values for core melt frequency for each source of risk are give in table 1-7. NRC Case Study results are included for comparison.
| |
| Table 1-7 EPRI/WOG Study Results Core Melt Frequency per Year Reduction Source of Risk h3C ffB1/WOG Factor IEemal 7 4 e
| |
| 1.4E-4 2.6E-6 54 Seismic 6.1E-5 7.4E-6 8 Fire 3.2E-5 6.3E-8 500 IIntemal F1 7.7E-5 <1.0E-8 . .>7700 Extemal Flood 1.9E 8 <1.0E-8* (>>2)
| |
| Wind 4.0E-6 <1.0E 8* (>>400)
| |
| Lightning j,,H.Ed <l.0E 8* (>> 6) i Total 3.1E-4 1.0E-5 31 0 Core melt frequency reduced to <1.0E-8 by explanation of errors in NRC Case Study without RMQS requantification. (See section 8). j The dominant internal accident sequences and their contributing core melt frequencies are given in table ;
| |
| l-8. NRC Case Study results are included for comparison, with key reasons for change listed. The !
| |
| EPRI/WOG analysis of Point Beach resulted in significant reductions in core melt frequency due to ecial emergencies. In the case of the seismic analysis, the niajor changes occmred as a ruult of considering a variety of recovery actions for earthquakes less than three times the Safe Shutdown 1-12
| |
| | |
| Table 18 EPRJAVOG Internal Accident Sequences Core Melt Frequency per Year Secuence Name* ER_C EPRIAVOG Kev Reasons for Change TlMLE 6.7E-6 7.7E-7 New battedes S2MH1'H2' 4.7E-5 5.8E-7 Small LOCA frequency CCW success criteria Operator actions LTSB 3.6E-5 5.4E-7 Refill of CST; diesel generator failure rates TIQDlD2 <1 E-8 1.2E-7 Relief valve LOCA more likely to occur T3QH1'H2' 2.5E 5 N/A Relief valve LOCA cannot occur for this sequence All Others 2.0E-5 5.5E-7 (Vadous)
| |
| See table 1-2 for sequence desedptions and table 2-1 for NRC Case Study acronyms.
| |
| Earthquake (SSE). The primary change occurred with the use of newly installed Seismic Category I batteries, together with the use of many alternative sources of makeup water available onsite should the Refueling Water Storage Tank (RWST) or the Condensate Storage Tanks (CSTs) fM1. Also significant j were reductions in the seismic hazard curves, primarily due to the improved modeling of attenuation. I In the case of fire analysis, the core melt frequency due to fires caused by " transient combustibles" was - j reduced dmmatically due to elimination of the unreasonable assumpthn that ten-gallon spills of scetone {
| |
| would exist continuously in every space in the plant. However, the EPPJAVOG study included cable tray and electrical panel fires caused by means other than transient combustibles, which the h%C Case Stady did not. This broader analysis scope tended to compensate for the reduction in transient combus-tible fire risk. Overall, the fire risk decreased in the EPRIAVOG study as a tesult of a more balanced i and realistic treatment of transient combustibles, initiator frequencies, exailability and recoverability of the turbirie-driven AFW pump, and the installation of new batteries.
| |
| l 1-13
| |
| | |
| In the case of the intemal flooding analysis, the contribution to risk from service water pump room
| |
| 'oods was reduced dramatically. This reduction resulted from an improved method of calculating flood
| |
| .requency and change in the success criteria for cooling of the High-Pressure Injection (HPI) system.
| |
| PRIMARY DIFFERENCES BETWEEN THE EPRI/WOG AND NRC CASE STUDY ANALYSES A large number of factors contributed to the more realistic estimate of risk in the EPRI/WOG study.
| |
| These factors include the treatment of initiating events and data, human reliability and recovery analysis, analysis of SDHR effectiveness, consequence analysis, and V-I analysis.
| |
| I Lnitiatine Events and Data Table 1-9 summarizes the differences in initiating event frequencies and data for equipment failure.
| |
| Table 19 Key Contributors to the Lower Core Melt Frequency at Point Beach Initiating Events and Data (Partial List - See Table 51)
| |
| Reduction in Core-Melt NRC Case Frequency
| |
| ,nitiatinc Event Study Value EPRI/WOG Value (per vear)
| |
| Flood in service water room 2.2E-3 3.7E-5 7.6E-5 Small-break LOCA 2.0E-2 3.0E-3 4.5E 5 All other transients 7.1 2.4 1.9E-5 Loss of offsite power 8.4E-2 6.2E-2 1.1 E-5 pther Significant Events and Comoonent Failure Data Relief valve /PORV LOCA 1.4E 3 1.1E-4 2.9E-5 Diesel generator faults 3.8E-2 2.2E-2 1.0E-5 Common Cause Data Motor-operated valves 4.0E-4 8.0E 5 1.7E-5 Batteries 9.6E-4 4.0E-4 1.7E-5 AFW pumps 2.0E-4 3.0E 5 8.2E-6 i
| |
| l l
| |
| l 1-14
| |
| | |
| Huma3 Reliability and Recovery Analysis he EPRI/WOG reanalysis does not arbitrarily limit the number of recovery actions to one in the first two hours (and two in total) as in the NRC Case Study. In the EPRI/WOG analysis the number of !
| |
| recovery actions permitted was based on the number of options available, the time required to perform each action, and their presence in the Point Beach procedures and training program (Westinghouse Owners Group Emergency Response Guideline (ERGS) (E), or plant-specific procedures)
| |
| The following recovery actions have been added to the NRC Case Study model:
| |
| o RWST refill using water from the spent fuel pool or the chemical and volume control system, or, using the RHR or charging pumps, drawing suction directly from the spent fuel pool.
| |
| o Cross-connecting AC or DC buses for initiating events caused by the loss of an AC or DC bus (included in Section 6 analysis).
| |
| o Manual control of the turbine-driven auxiliary feedwater pump.
| |
| o Provision of a backup supply of feedwater using either service water or CST refill from the diesel-driven fire water system.
| |
| o Use of the charging system for loss of feedwater sequences.
| |
| o Balancing loads on the service water system to respond to low system output when less than three pumps are available.
| |
| o Recovery from common mode failures.
| |
| The findings listed below describe changes in the Human Reliability Analysis (HRA) quantification:
| |
| o Reduce failure to initiate feed and bleed from 9E-3 to IE-3,i.e., a value consistent with the Point Beach operator's lack of hesitation as well as with most other PRAs and NUREG/CR-1278 (2).
| |
| o Reduce failure to initiate sump recirculation from 3E-3 to 1E-4, i.e., a value consistent with NUREG/CR-1278 and most other PRAs.
| |
| o Reduce failure to depressurize froto 1.5E 2 to 3E-3.
| |
| o Change the human recovery time aliability correlation from NUREG/CR-2787
| |
| (.lD) to the model contained in NUREG CR1278, thereby yielding reductions in human recovery failure figuns by up to a factor of about 50.
| |
| i Table 1-10 is a partial list (see table 4-3) of humu reliability and recovery analysis changes and their i l
| |
| impact on NRC Case Study core melt frequencies. '
| |
| j
| |
| )
| |
| EPRI/WOG Analysis of the Dedicated SDHR !
| |
| l
| |
| _PRl/WOG conducted a realistic analysis of the SDHR system. The NRC Case Study assumed, without analysis, that the SDHR and the structure contaming it were capable of withstanding all special 1 1
| |
| 1 1-15 i
| |
| | |
| Table 1-10 Summary of Human Reliability and Recovery Analysis Changes and Their Impact on NRC Case Study Core Melt Frequencies (Panial List--See Table 4-3)
| |
| Reduction in Core-Melt Frequency Deteription of Model Chances per Year Recirculation failure recovery 7.2E-5 CST inventory recovery - station blackout 3.5E-5 Recirculation switchover operator error 2.8E 5 Loss of offsite power riming study 1.7E-5 Alternative to RWST (seismic) 1.5E-5 CST alternative (seismic) - SWS 1.5E-5 emergencies. The NRC Case Study employed different rules in defining SDHR event initiator requency and success criteria than were applied to the existing plant. For example, the NRC Case Study assumed that large acetone spills existed continuously all over the plant as a fire source, except inside the dedicated SDHR structure. The Case Study assumed an excessively high frequency of pipe ruptures in all safety piping at Point Beach except in the SDHR, which was assumed to be immune to pipe break. The SDHR was assumed to be immune to earthquakes, yet it is designed to the same level of seismic design criteria as the existing plant, which is assumed to fail at much lower levels of j j
| |
| earthquake intensity. Based on the EPRI/WOG analysis, this last assumption is potentially the most l
| |
| significant in terms ofits impact on risk.
| |
| These biases and other modeling errors and omissions are corrected in the EPRI/WOG analysis. A sig-nificant problem that was encountered in attempting a more rigorous analysis of the SDHR was a lack of ]
| |
| design information for the hypothetical system, particularly in the area of initiation and control logic. l The system was assumed to initiate automatically and function properly during a ten hour transient with I no operator assistance. This ten-hour period would involve many system control complexities caused by changing plant parameters and interface problems with other safety systems. The IGC case study did not provide an adequate explanation of how the system starts, operates, and controls itself during this ten-hour period. The EPRIAVOG study was forced to make a variety of assumptions about the NRC system in order to be able to analyze it.
| |
| 1-16
| |
| \ _
| |
| | |
| In addition to *.h: above problems, the EPRI/WOG reanalysis considered infant mortality, a broader nge of SDHR components, common cause effects between normal plant safety systems and the i SDHR, and the euntual requirement for recirculation, which the proposed SDHR system cannot perform. Since the system was assumed to be capable of performing for ten hours with no operator assistance and because access to the system is restricted so that it pmtects the plant against sabotage, the EPRI/WOG study assemed that no system recovery was possible for ten hours.
| |
| In contrast to the ebavn. the more realistic analysis of the SDHR also led to some results that were more favorable to the SDHR concept than were credited in the Case Study. For example, the NRC Case Study assumed that both the single train of primary injection and the single train of emergency feed-water must function for the SDHR to succeed. In the EPRI/WOG analysis, the SDHR was considered to succeed for transients not resulting in LOCAs if either of these trains functioned properly, since the ,
| |
| former provides adequate DHR via feed and bleed, and the latter provides adequate DHR via secondary heat removal.
| |
| Two serious weaknesses in the system design that could result in an overall increase in risk should the SDHR be installed are investigated: (1) bypass scenarios through the SDHR'S atmospheric dump valve, and (2) lack of a long-term water supply to ensure cooling after high-acceleration seismic events. These veaknesses are discussed further in section 2. Since we assumed that designers would correct these problems if an SDHR was ever actually built in response to these reports, we did not include these problems in the base case analysis but addressed them in a sensitivity study.
| |
| Consequence Analysis ;
| |
| The EPRI/WOG study of Point Beach uses an approach to consequence analysis similar to the NRC Case Study, but uses a different value for the source term in these calculations. The Industry Degraded j l
| |
| Core Rulemaking (IDCOR) Program (11) source terms were used, whit.h on the average are at least a factor of seven lower than the " central estimate" used in the NRC Case Study.
| |
| These lower source terms are shown to be quite reasonable for this Point Beach analysis, by favorable comparison with two NRC-sponsored studies, BMI 2104 (12) and NUREG-0956 (Draft) (11). For late containment failure scenarios, the source terms derived from the two NRC sponsored studies predict slightly lower doses than those derived from IDCOR. For early containment failure scenarios, the two NRC-sponsored studies would predict slightly higher source terms and doses. Overall, the results using the EPRI/WOG or the NRC source terms yield doses within a factor of two. These lower source term estimates, in combination with much lower core melt frequencies in the EPRI/WOG study, show that he offsite DHR risk at Point Beach is roughly 200 times lower than the NRC Case Study estimate.
| |
| l l
| |
| 1-17
| |
| | |
| Value Imnact Analysis le EPRI/WOG approach to V-I analysis is significantly different than the NRC Case Study approach.
| |
| First, the NRC convention to consider all doses as " values" and all costs as " impacts" has been dropped in favor of a more logical approach. In the EPRI/WOG method, all beneficial factors (avened doses and avened costs) are considered values, and all detrimental factors (installation, operations and maintenance costs and doses) are considered impacts. (Both the NRC Case Study and the EPRl/WOG analysis used the standard method of quantifying the dollar value of avened doses at $1,000 per person-rem.)
| |
| Second, each propoted modification is evaluated individually instead of being lumped together with other modifications. The EPRI/WOG study selected this approach so WEP would have the ability to evaluate each proposal on its own merit. The lumped alternatives created by the NRC Case Study tend to obscure the fact that some modifications within an alternative are much more cost-effective than mothers.
| |
| In 1985 WEP innalled a new set of Seismic Category I batteries, designed specifically to provide backup DC power for diesel starting and vital plant instrumentation. Since the system was installed when this study was conducted, the batteries were included in the plant model and assumed in the base tse for V-I analyses. However, in order to evaluate the V-1 ratio of these new batteries, a revised model was created that reflected the Point Beach plant prior to battery installation. This evaluation indicated that the new batteries had reduced the core melt probability by over 50%. In comparison to other modifications evaluated against the base case, the new batteries had the most favorable V 1 ratio.
| |
| The results of this EPRI/WOG analysis of the SDHR show clearly that the dedicated system is not cost-effective. The V-I ratio for this system, considering avened offsite doses only, is 2.0E-5. This means that the system might provide 0.002 cent in' safety benefit for every one dollar spent installing and maintaining the system. If onsite avened costs are included, the system might provide 0.002 cent in safety benefit for every one cent spent installing and maintaining the system. If one considers offsite avened dose only, personnel will receive 150 times more dose from installing and maintaining the sys-tem than will lx: avened offsite through projected tisk reduction.
| |
| Table 1-11 presents the V-I results of this EPRI/WOG analysis of Point Beach, i
| |
| 1-18
| |
| __ _ _ J
| |
| | |
| l l1 ,
| |
| )l t
| |
| c a
| |
| p es -
| |
| mio hee I t a 335566
| |
| - - - - - 6667-- - 78 - - - t iu nela r
| |
| /eR EEEEEEEEEEEE ot avt u 4. 0. 0 5 73 7 5. l. 9. 8 3 l
| |
| a 1 1 21 44 322971 dbr e e
| |
| V seh ahg
| |
| * bt ih ry h $
| |
| )
| |
| dhe et t o t n rd(
| |
| oee l awc ti a
| |
| r-u i Wte o r s 3331
| |
| + + + + +01 1
| |
| + + + - -202 1
| |
| 1 4
| |
| u lcd i
| |
| d vD at e n t -
| |
| 1 S n eAe EEEEEEEEEEEE cad i
| |
| e sf t 2.0 9 1 4 7 5. 9.4 7. 4 6 sl ul l
| |
| u a e os r
| |
| i f
| |
| 521 31 754364 5 d e
| |
| i a nv u
| |
| o P
| |
| - V f
| |
| O s ui oews t
| |
| s aret
| |
| )
| |
| i si f ca us l
| |
| (
| |
| r is e adnr l
| |
| at 667666775466 e y oo i
| |
| te )o i
| |
| t c Tp oa + + + + + + + + + + + +
| |
| EEEEEEEEEEEE 5 miat he t
| |
| s f a t
| |
| : 7. 9. 5 0. l. 85 9.6. 0.7. 2 3 ii sc s fR m 3 1 923 1 1 1 1 754 1 hi f ei Otc I f
| |
| t dr ga np o e
| |
| o f oe mta t
| |
| i y f e dm ui r
| |
| a ht M
| |
| i l ur b aheto l
| |
| l cnu/e m r ola & *)
| |
| t n vt u a
| |
| Il a
| |
| m u WlaO$ ( 4 5 3 4 l e oh t
| |
| sde + + + + p hl i t
| |
| l tsV S t nnns EE E E e~
| |
| v I
| |
| l Aw 1 uy s
| |
| t c eiao . 69 . . . . . 0 0 i - k sf 1
| |
| - eb Rd e a
| |
| p eo r nDo 01 4 000000201 t c P e E s.
| |
| s r
| |
| t m P i f e i e t ck I t f e Wih r c an n yt e a l
| |
| b pa a ht
| |
| ~
| |
| e T I mR ( r olai$ o) n AbtaB . db
| |
| - y e t el un Wt t (
| |
| ola s 667666775 466
| |
| + 4 + + + + + + + 4 4 +
| |
| ml l e ahio en aO t
| |
| l tTl n aso t
| |
| EEEEEEEEEEEE r
| |
| - st P Vs ef t
| |
| : 7. 9. 5 0. l. 8 5 968 72 n n t t
| |
| sosC e n 31 9231 654 oiunoo Gso P r I 1 1 1 s
| |
| r dh e nt d ey .
| |
| OC p ai Wd
| |
| / e r d ws anc I t n e p nsekb e u Rre Pv it ly) necr 0 i gi d r teq EA o iMe n ae 5668
| |
| - - -77778
| |
| - - - -68 0 s 0 eh e
| |
| dcuala t rf t uy EEEEEEEEEEEE 1 ceqr ur e l. 83 3 01 4 6 6.0.7. 9 $ ne alv e doe e C Fr (p 1 284 21 57374 5 gni oB e m t
| |
| R
| |
| * is at cnne e-
| |
| ) u ii r s i f oeo s
| |
| i6 e r d P bc r
| |
| e e - a of d n o moa i l i t 1 l f
| |
| im da ta 2 21 91 41 8 d at nR g e hs c l l l l l cl Mn oN BiR wml a a aa ai a 21 mrnmmmmns r t o r eeHa n Cio ei si eee e ee ei e NeDiin r r t
| |
| * t t t d
| |
| e samDhe ssS dc Rt a ( SSFFI Inn I InI nSI et n t r ie Nc e v esha rse n n t t at o a f s n
| |
| cB GI o Go e w O .i r
| |
| t a
| |
| Oir t e a eWd et r 28I 092651 374 /l Wcab s Nll c
| |
| /im 1 I 1 e s eRt ap a DohPsnm I f iu Rd T Eii P oN EM
| |
| * 7E
| |
| | |
| I Section 2 OVERVIEW OF THE APPROACH .
| |
| This EPRI/WOG study uses the NRC A-45 Case Study of Point Beach (D as the starting point for its reevaluation of risk at Point Beach and subsequent reassessment of the value-impact of various modifi '
| |
| cations, including the dedicated shutdown heat removal system. For simplicity and ease of comparison, the EPRI/WOG study used the same overall approach as the NRC Case Study. That is, in both cases the likelihood and consequences of accidents were calculated using probabilistic risk assessment (PRA) techniques and insights, modifications were proposed for the vulnerabilities indicated by the risk results, installation costs and other impacts were determined for proposed modifications, and value impact measures for each modification were calculated.
| |
| The objectives and scope of the EPRI/WOG analysis did nQ1 include a completely independent analysis of Point Beach, e.g., a Level 1 PRA, followed by a comparison and discussion of the results of the rudies. Rather, this EPRI/WOG study is a limited reanalysis.
| |
| The majority of this section is devoted to an overview of the NRC results. Those results are described in terms of dominant accident sequences and proposed plant modifications. The intent of this discussion is to allow the reader to understand the starting point for the EPRI/WOG study. Details of this EPRI/WOG effort are given in subsequent sections, and significant differences between the two studies are identified. These relevant details--which may relate to data, assumptions or models -are compared to the appropriate NRC Case Study details in the appropriate section.
| |
| METHODOLOGY At a high level, the analysis methodology used in the NRC Case Study and that used in the EPRI/WOG studies are the same. Figure 2-1 (figure 1-6 from the Case Study) illustrates the overall methodology used to assess the value and impact of the SDHR and other potential plant modifications.
| |
| Both the intemal events analysis and the special emergency analysis use methods developed in the~PRA community to determine the frequency and type of accident that can result from various initiators.
| |
| rhese results can then be used to identify potential plant modifications. Given specific modifications, oth value and impact analyses are performed.
| |
| 2-1
| |
| | |
| - l l
| |
| I 0
| |
| _ 1 C T I A
| |
| - R t
| |
| _ C n sE 4
| |
| - t 6 f 5 s
| |
| _ S 0 i
| |
| / ? I C s .
| |
| _ E C S E y)
| |
| U A Y R l 9
| |
| _ L P L a -
| |
| A M A n1 V I N A
| |
| . A e
| |
| _ l g aa
| |
| _ vp o
| |
| m et R r t p o
| |
| ae S
| |
| eR 3 H T 1 I y C S E S yd A Y U Y au P L L L ct A A M A V N eS I N A D A e ns wa oC
| |
| - d t h u c h a S e
| |
| - r ot B
| |
| S f n E i MV no SI aP IT l LA BN P m AR TE C r ST RF EL N(
| |
| A 1
| |
| - 2 e
| |
| - i r
| |
| u g
| |
| F Y
| |
| L S L C S A I A N E N S R Y E L I
| |
| C ESV E W L T A P Eq a
| |
| _ S N S f m E A I A l
| |
| YM
| |
| | |
| 1 The value analysis uses the PRA models to determine how much the likelihood of various accidents are
| |
| , wiuced and how much the offsite consequences (doses) are reduced for a particular alternative. These s.cident likelihwds can be translated into dollars of "bcracSt". by calculating $1000 per person rem for accident doses (i.e., offsite costs), together with a dollar estimate of the cost to replace the plant in the event of core damage (i.e., onsite averted costs). He impact analysis determines installation costs, operations and maintenance.(O&M) costs, installation and O&M radiation doses, and impacts on plant availability for each alternative. All of these items are " costs" of performing the modification. The value-impact analysis then compares the cost of modifying the plant to the value-benefit of reduced risk.
| |
| Various measures, e.g., a value-impact ratio, are calculated to measure the effectiveness of each modifi-'
| |
| cation, including the SDHR. ..
| |
| Briefly, the following was performed as part of the EPRI/WOG study: 1) the PRA was reanalyzed for j the internal analysis and special emergency analysis; 2) in addition to the alternatives developed in the NRC Case Study, the EPRI/WOG study evaluated a Wisconsin Electric Power (WEP) modification to plant batteries; 3) WEP reviewed the impact analysis and calculated costs for the various modifications;-
| |
| : 4) a value analysis was performed by EPRI/WOG using the reanalyzed PRA models; 5) the SDHR was reanalyzed for potential failure modes; and 6) the EPRl/WOG study reanalyzed the value impact analy-sis, using both new values and impacts, as well as different combinations of terms and formats for pre-sentation of results.
| |
| The principal focus of the EPRI/WOG study was a reanalysis of the PRA. The general approach taken is to:
| |
| : 1) Check assumptions and applicability of the models and data;
| |
| : 2) Compare them to industry PRA and value-impact studies; and
| |
| : 3) Incorporate appropriate changes into the models, and obtain new bottom-line estimates of nsk and value impact.
| |
| Figure 2 2 (figure 2.1 in the NRC Case Study) illustir.es the overall PRA process used in both the Case ,
| |
| Study and the EPRI/WOG study. The left half of the figure depicts fault trees and both transient and L,0CA event trees. The Case Study solved these models to obtain accident-sequence cut sets. (An acci-dent sequence cut set is a set of component or human failures which cause a core-damage accident sequence to occur.)
| |
| .P The EPRI/WOG study reanalyzed the logic model provided in the NRC Case Study, with a new model obtained by modifying the accident sequence cut sets updated in the Case Study. He following two umples illustrate the EPRI/WOG methodology for creating a new plant model. In the case of sump I 2-3
| |
| | |
| _ 5 I
| |
| _ - f D
| |
| o i
| |
| k
| |
| _ I s - - s- n e sio 4
| |
| y I i i R
| |
| g r , et o
| |
| 3 o , era I g , l ul l
| |
| o 2 e , t s u t , bap YO t n ne I a q ueo em C , PMP do 1
| |
| 3, e
| |
| ine I e ,do s
| |
| ced a gC cho e e APC l r e u R l
| |
| i a
| |
| F t
| |
| n e
| |
| m n s n o e e i ri ias nt ut t e oc l i CM nd oo
| |
| - t n
| |
| e m
| |
| ie tt co er tP l l ai Fb
| |
| - a eb n o t o it r y ar an Pt DP t e i nd ev oi ri y s Cc ut t m / c sc i e ltA se sa l t
| |
| s esc eo ri i
| |
| b y Mmn pd a S ee ra l et u eR i t r sq v a ne oye Ot v ee CSS n a mr t e n nf nd U ei iat mc e -
| |
| t n nc c ne ov CE - - )
| |
| iA a
| |
| t t ns ao sn ea r n ue l t r r
| |
| e w
| |
| o M CF - ino C air P
| |
| ( tF ar n ME C t s et K M y K M l t e vnd r m A, O C c O C enc E eno er Men nat t e eiu d e copt r a st ya r cq oce ismse aoep SW CAS BCT O t e r c oi e
| |
| e r
| |
| Mg pv pr ue SS f
| |
| t s e n s r t e
| |
| v e u, ss c l og E c l nn u a d i A S r e ed C c oa O k h u re L d H n s
| |
| ' I t
| |
| ee ee g rr
| |
| ' n e tT e
| |
| e r
| |
| " n K c 0 M
| |
| C i
| |
| s nA tt in T ' . ~ aC rO
| |
| .u ve t rE n TL V e
| |
| v E ]
| |
| t , I8 l ,,',
| |
| n ie I s
| |
| n a g) r nE T iI
| |
| (
| |
| t(
| |
| a it t n i e IE nv iE V wL l
| |
| l
| |
| | |
| recirculation failure after a small LOCA, the EPRI/WOG study modified the sequence cut sets by con-dering the refilling of the RWST as an additional and necessary failure mode. (The addition of this tailure mode to the cut set assured the evaluation of its success or failure). In the case of component cooling water (CCW)-induced failures of the HPI system, those cut sets including CCW were deleted altogether when it was learned that CCW cooling is not required f or HPI.
| |
| The EPRI/WOG study reevaluated the success criteria and accident sequence timing and developed new event trees to illustrate the differences (section 3 of this report). The study performed a human reliability assessment and recovery analysis consistent with The Handbook for Human Reliability Analysis, NUREG/CR 1278 (2), and industry techniques (section 4). The study obtained plant- or vendor specific data for initiating events and some key components, and reevaluated common-cause failure probabilities (section 5). The EPRI/WOG study also reevaluated special emergencies (section 6).
| |
| The right half of figure 2 2 illustrates the approach taken by the NRC Case Study and also applied in the EPRI/WOG study in their overall approach to the sorr:e term and consequence analysis. The EPR1/
| |
| WOG study used a containment system event tree much like that used in the NRC Case Study, except that success criteria were relaxed to be more consistent with the increased containment and containment ystems capacity found in the recent source term work (section 9). On the other hand, containment-
| |
| .ailure modes and release categories were developed for the EPRI/WOG study based on newer source term information, including both Industry Degraded Core Rulemaking (IDCOR) (11) and NRC work (12, ll) (section 9). Funhermore, the The EPRI/WOG study did not use a consequence code to explicitly determine public health doses for various source terms. Sensitivity studies, as well as analysis performed for the Case Studies, were used to infer doses for modified source terms (section 9).
| |
| OVERVIEW OF THE SIGNIFICANT ACCIDENT TYPES AS DETERMINED BY THE NRC CASE STUDY The NRC Case Study identified a variety of significant accident sequence types and accident initiators.
| |
| As mentioned previously, these accident types are used as the starting poim for the EPRI/WOG study.
| |
| Intemal accident initiators were limited to small LOCAs and five types of transient initiators. Four types of extemal events were also analyzed. Of these, seismic events were found to exhibit the potential to cause both transients and small LOCAs, whereas the other three external events resulted solely in transients.
| |
| Functionally, the external event and internal event initiators lead to similar types of accidents. A transient, whether extemally or internally initiated, can be mitigated in either of two ways. First, if econdary heat removal is available (i.e., steam generator cooling using main or auxiliary feedwater and secondary relief valves or the main condenser), the plant can be shut down without any other operable 2-5 l
| |
| t
| |
| | |
| 1 safety function. Altematively, if secondary heat removal fails, feed and-bleed can prevent core melting
| |
| ' implemented in time. Feed-and bleed requires that both of the pressurizer power-operated relief ealves (PORVs) be opened to depressurize the RCS (although one of two PORVs would be sufficient in many situations). Depressurization allows the intermediate pressure HPI system to inject Emergency Core Cooling System (ECCS) water. Feed-and-bleed cooling also requires that a long-term cooling path be established. If secondary heat removal is recovered and the PORVs closed before HPl flow eventually empties the RWST, high pressure recirculation (HPR) from the containment sump can be avoided with continued cooling provided through secondary heat removal.
| |
| The NRC Case Study postulates a non-isolatable LOCA at full power conditions and assumes that nor-mal RHR cooldown is precluded. This type of small LOCA or a transient-induced LOCA can be mitigated in two ways, regardless of the type of initiator. First, ECCS injection and circulation is credited in mitigating the accident. Second,if HPI fails to inject high-pressure water, credit was given to secondary heat removal for depressurizing the reactor so that Low Pressure Injection (LPI) and recircu-lation could mitigate the event.
| |
| in summary, accident sequences take on the following general form:
| |
| o Transient with failure of secondary heat removal and failure of feed-and-bleed.
| |
| o Transient with failure of secondary heat removal and successful feed-and-bleed in injection mode, but failure of ECCS recirculation (HPR mode).
| |
| o Small LOCA with failure of ECCS recirculation.
| |
| o Small LOCA with failure of ECCS injection and failure of either secondary heat removal or LPI.
| |
| o Transient-induced LOCA and failure of ECCS recirculation.
| |
| o Transient-induced LOCA and failure of ECCS injection (and failure of either secondary heat removal or LPI).
| |
| The NRC Case Study evaluated these general accident sequences for specific initiating events. For the internal events analysis, five initiators were significant. For the extemal events analysis, three seismic initiators, two fire initiators, one flood initiator and one wind initiator were significant.
| |
| Significant internal event initiators in the NRC Case Study were:
| |
| o General transient with no impact on mitigating systems.
| |
| o Loss of main feedwater transient.
| |
| o Loss of offsite power transient.
| |
| o Loss of a safety grade AC bus.
| |
| 2-6 l
| |
| L _ _ _ _ _ _ _ _
| |
| | |
| o Loss of a safety-grade DC bus.
| |
| Other than the general transient, each initiator was evaluated because of its unique impact on the capa-bility of the plant to prevent fuel damage. For example, loss of main feedwater will affect secondary heat removal reliability, loss of offsite power will result in loss of the main feedwater system and will affect the reliability of AC power, an important supporting system. The loss of a safety grade AC bus initiator was based on the assumption that the bus failure would prevent operation of about half of the AC power-dependent . safety equipment; in particular, main feedwater would trip (one train is restorable), and a PORV block valve would not open if required. (As will be described subsequently, the EPRI/WOG study found that the bus loss would not trip the plant.) Loss of a safety grade DC bus will also affect about half of the safety equipment as well as the main feedwater, and will fail one of the two PORVs.
| |
| Significant external event initiators in the NRC Case Study were:
| |
| o Seismically-induced plant trip without loss of main feedwater.
| |
| o Seismically induced plant trip and loss of offsite power (and main feedwater).
| |
| o Seismically-induced small LOCA.
| |
| o Fire in the vital switchgear room.
| |
| o Fim in the auxiliary feedwater pump room.
| |
| o Flood in the service water pump room.
| |
| o High wind induced loss of offsite power and failure of the diesel generator exhausts.
| |
| Seismic-induced accidents were important in the NRC Case Study because the Refueling Water Storage Tank (RWST), the condensate storage tank (CST), and the electric power systems were assessed to have poor ability to withstand seismic activity. Accordmg to the NRC analysis,50% of the scenario damage ^
| |
| resulted from the failures of the RWST and the two CSTs. (The other 50% was due to emergency and offsite power failures.)
| |
| i l
| |
| Fires in the switchgear room and auxiliary feedwater room were important incidents covered by the NRC Case Study because nearly all plant safety equipment uses cables which go through these rooms.
| |
| Fiscs were assumed to fail everything except the turbine-driven auxiliary feedwater pump. In the switchgear room fire incident, the turbine-driven pump was assumed to fail due to the lack of DC power for instrumentation.
| |
| 27
| |
| | |
| The NRC's internal flood scenario resulted from a pipe break in the service water pump room which as assumed to fail all six pumps due to water spray. Failure of service water causes failure of most !
| |
| front-line safety systems, again leaving only the turbine-driven auxiliary feedwater pump for plant cooldown. {
| |
| t 1
| |
| The NRC's high wind scenario results from failure of offsite power and failure of the dinel generator i exhaust supports, which was assumed to result in a station blackout. Again, only the turbine-driven pump would be left available for plant cooldown. i 1
| |
| ACCIDENT TYPES AND POTENTIAL PLANT MODIFICATIONS FROM THE NRC CASE STUDY Because the EPRI/WOG study began with the NRC Case Study as a starting point, it is important that the reader 12 introduced to the results of the NRC Case Study. The following two subsections discuss the significant accident types and potential plant modifications resulting from the ins'ights obtained in the Case Study.
| |
| Significant Accident Tvoes The previous section identified various accident types. The following discussion identifies the key initi-aors for the significant accident types. The EPRI/WOG results, which are derived in subsequent sections, are presented here for the purpose of comparison. Those relevant sections also contain the EPRl/WOG study assumptions. (In some instances, conservative rather than best-estimate assumptions were utilized.) The aeronyms in parentheses are described in table 2-1.
| |
| Transients with Failure of Secondary Heat Removal and Failure of Feed-and Bleed Core-Melt Frecuency Per Year NRC Case EPRl/WOG Study Study Loss of offsite power (TIMLE) 6.7 E 6 7.7E-7 I.oss of safety DC bus (T5MLE) 9.1 E-7 1.3E-8 Loss of main feedwater (T2MLE) 6.6 E-7 1.0E-7 Loss of safety AC bus (T4MLE) 6.2 E-7 N/A Seismically-induced plant trip (TMLE) 3.6E-5 4.4E-6 AFW pump room fire 1.3 E 5 2.0E-8 f
| |
| Switchgear room fire 2.0 E-5 4.3E-8 l Service water pump room flood 7.7 E-5 <l.0E-8 !
| |
| t Long-term station blackout 3.6 E-5 5.4E-7 !
| |
| I i
| |
| i 2-8 !
| |
| ___ _ - ____ l
| |
| | |
| i-Table 2-1 :
| |
| NRC Case Study Acronyms for Accident Sequences initiating or System Event Description Initiating Events S2 Small LOCA < 2 Inches in Diameter TI I.oss of Offsite Power Transient T2 less of Power Conversion System (Main Feedwater)
| |
| T3 Transients with Main Feedwater Initially Available T4 Loss of an AC Bus T5 Loss of a DC Bus System Events ,
| |
| M Failure of Main Feedwater Q Failure of SRVs or PORVs to Reclose Given that they were Opened
| |
| , L Failure of the Auxiliary Feedwater System E Fdicte ef Feed and-Bleed Mode l l
| |
| D1 Failure of HPIS !
| |
| D2 Failure of LPIS H1' Failure of HPRS without RHR Heat Exchanger H2' Failure of LPRS without RHR Heat Exchanger H1 Failure of HPRS with RHR Heat Exchanger 1
| |
| 2-9 ;
| |
| | |
| l i
| |
| Transient with Failure of Secondary Heat Memoval and Successful Feed and Bleed in 14ection Wode, but Failure of High Pressure Recirculmion Core Melt Freauency Per Year NRC Case EPRI/WOG Study - Study Loss of feedwater (T2MLHI) 2.0 E.8 1.0E-7 SmallLOCA with Failure of ECCS Recirculation Small LOCA (S2MH1'H2') 4.7 E-5 5.8E-7 SmallLOCA and Failure ofECCS injection
| |
| ~
| |
| Small LOCA (S2MD1D2) 8.7E-6 9.5E 8 -
| |
| SmallLOCA with Failure of ECCS injection and Failure of Either Secondary Heat Removal or Low Pressureinjecdon Small LOCA (S2MXD1) 5.7 E-7 1.0E-8 Seismically-induced LOCA 2.5 E-5 3.0E 6 Transient-induced LOCA and Failure ofECCS Recirculation Plant trip (T3QH1'H2') 2.5 E-5 N/A Loss of main feedwater (T2iAQH1'H2') 3.5 E-6 1.9E-7 l Transient induced LOCA and Failure of ECCS Igection (and Failure of ESher Sv:ondary Heat Removalor Low Pressureidecdon)
| |
| Plant trip (T3QD1D2) 4.6 E-6 N/A less of main feedwater (T2MQD1D2) 6.6 E-7 4.1E-8 Station blackout (TIQDID2) <1.0 E-8 1.2 E-7 ,
| |
| Potential Plan Modifications Plant modifications were proposed and evaluated by the NRC Case Study as a ruult of accident sequence results. The following briefly discusses each modification. The accident nquences the modifications are intended to mitigate are listed in terms of the acronyms in table 2-1 or desevibed if the i
| |
| 2-10
| |
| | |
| acronyms do not apply. The modifications are numbered to reflect vulnerabilities determined in the NRC Case Study. In some cases, vulnerabilities were not addressed by modifications (e.g.,5,6 and 7).
| |
| InternalModification 1 This modification consisted of adding a more prominent alarm warning thai 3witchover fmm injection to recirculation is necessary and imminent. The Case Study concluded that operator failure to switch to recirculation was an important contributor to core-damage risk, resulting in a core-damage frequency of 3.1E-5 from the following sequences:
| |
| S2MH1'H2' T2MQH1'H2' T3QH1'H2' The more prominent waming would be designed to alert the operator to this important action.
| |
| InternalModification 2 This modification involved installation of dedicated diesel generator startup batteries. The Case Study concluded that loss of station baneries was a principal contributor to station blackout accident equences, resulting in a core damage frequency of 4.9E-6 from the following sequence:
| |
| TIMLE The dedicated diesel generator batteries would be designed to allow starting of the diesel generators after a loss of offsite power, gas turbine failure, and failure of the station batteries.
| |
| InternalModification 3 This modification entailed installation of a steam turbine-driven generator operating off the main steam j
| |
| system. The system would be capable of providing power to a motor-driven AFW pump, one charging l
| |
| pump, and vital instrumentation. The Case Study concluded that failure of both diesel generators resulted in a core-damage frequency of 5.5E-7 from the following sequence:
| |
| l l
| |
| TIMLE The generator would be designed to allow imrnediate feedwater addition and control in the event of a station blackout. A long-term source of feedwater would still have to be aligned after the CSTs emptied.
| |
| 2-11
| |
| _____-______-__ l
| |
| | |
| j l
| |
| InternalModification 4 hi'. modification would install a parallel manual valve to component cooling water valve number 30 so that nc single valve failure could disable component cooling to the RHR pumps. The Case Study deter-mined that the failure or unavailability due to maintenance of the manual valve resulted in a core-damage fre.quency of 1.5E 5 from the following sequences:
| |
| S2MH1'H2' T3QH1 H2' T2MQH1'H2' The new valve would provide a parallel flow path, preventing single valve failure from causing the loss of the RHR pumps.
| |
| InternalModification 8 This modification would install a third independent low pressure ECCS train with an additional suction line from the sump and crossties to the two original trains. The Case Study concluded that the failure or unavailability due to maintenance of both existing trains of HPR resulted in a core-damage frequency of 2.2E-5 from the following sequences:
| |
| S2MH1'H2' T3QH1'H2' T2MQH1'H2' The third train would mean that thne failures, rather than two, would be required to fail sump recirculation.
| |
| InternalModification 9 This modification involved installation of an independent diesel-driven auxiliary feedwater pump in parallel with the turbine-driven pump, but with shared suction and discharge lines. The Case Study concluded that the failure or unavailability due to maintenance of the turbine-driven pump resulted in a core-damage frequency of 1.0E-5 from the following sequences:
| |
| TlMLE T4MLE T5MLE 2 12
| |
| _ _ _ _ _ _ _ - _ I
| |
| | |
| l l
| |
| [ l l
| |
| The diesel-driven pump would provide additional redundancy for loss of offsite power or AC or DC bus itiating events. A long-term feedwater source would still be required. Additionally, control and power cables would still be routed through the switchgear room and would still be vulnerable to fire.
| |
| i InternalModification il l
| |
| This modification would be the same as modification 3, however an additional CST would be added to provide a long-term fec/4ter supply for either the turbine-driven feedwater pump or the motor-driven feedwater pump powered by modification 3's generator. This modification would affect the long term station blackout sequence, which had a frequency of 3.6E-5.
| |
| Seismic Modification 1 This modification would provide a high seismic capacity water source attemative to the RWST by adding piping and valving to the spent fuel pool. This modification would affect sequences which had a core damage frequency of 3.1E-5. The sequences affected include both LOCAs and transients for which RWST failure appears alone or in conjunction with failure of the CSTs.
| |
| Seismic Modifications 2 6
| |
| 'his modification would provide additional anchorage for the 4160V and 480V safeguard buses,480V
| |
| . transformers, Safety injection (SI) pump buses, instrumentation power supply inverters and battery l
| |
| chargers. This modification would affect sequences which had a core-damage frequency of 2.9E-5. The sequences affected include both LOCAs and transients which lead to core damage as a result of a seismically-induced station blackout.
| |
| Fire Modification 1 This modification would provide an automatic water suppression system for the AFW pump room. It would affect AFW pump room fire sequences which had a core-damage frequency of 1.3E-5.
| |
| Fire Modification 2 This modification would relocate the main battery distribution bus, charger and inverter to another zone and remove cabling from the 4160V switchgear room. It would affect switchgear room fire sequences which had a core-damnge frequency of 2.0E 5.
| |
| InternalFlood Modification rhis modification would erect a fire wall to function as a water spray barrier in the middle of the service ater pump room. The barrier would prevent a pipe break frem simultaneously failing all 6 service 2-13
| |
| | |
| water pumps. This modification would affect internal flood sequences which had a core-damage equency of 7.7E-5.
| |
| The above modifications were combined and considered as three altematives. Alternative 1 included internal modifications 1,2 and 4, as well as seismic modifications 1 and 2 6, and both the fire modifica.
| |
| tions and the intemal flood modification. Alternative 2 included all attemative 1 modifications plus internal modification 11. Alternative 3 included all of alternative 2 plus internal modifications 8 and 9 and the wind alternative.
| |
| Alternative 4 is the SDHR. This modification is discussed further in section 7.
| |
| A summary of all the modification names and altemative assignments is shown in table 2 2.
| |
| l
| |
| ?.-14 i
| |
| | |
| n i
| |
| o t
| |
| a t
| |
| n e
| |
| m t
| |
| mr e
| |
| u t
| |
| s s
| |
| y I n
| |
| S d l n e va as
| |
| _ l p
| |
| n a ol u
| |
| m e n s P P mse e i o
| |
| s e r C R iD s rn s i r e t t
| |
| e n o D t e d ea r o t
| |
| n n t a e t x
| |
| e g it o p o
| |
| t m a wk a ce na Hf sp uE n
| |
| i t
| |
| p B d n e a r
| |
| o oB t l
| |
| l g i r
| |
| r o F e T e h y e S a v cn Pr r c a re s W i is c t a e t
| |
| s s e r y g i
| |
| t e e t u
| |
| A e e r a a nar A e igr a D r a an dB h e t
| |
| e D n v i o r t l
| |
| v t e a l l e ne F h x i r
| |
| o mG V i t t
| |
| wI E hS p ux S e A m mC i
| |
| l t r o y a h a l r r
| |
| _ n S l
| |
| e R mA t p o yd oa uo er r
| |
| e A s e I a Ti o r e u t
| |
| g t Pu ne esn SWEu qpR a S h ea rne cu I t l
| |
| A t l e
| |
| v D i R R v t
| |
| t t
| |
| e d n t i r d n R l mB d C e r d
| |
| H D o c ca P u e e t
| |
| _ 2- n L e a t d R t t c GS a T a nu e e Ci a a i mles k ea l i r
| |
| wsmt cleWcod 2 s ic e e S ic r s s e e e l i F e e e ie n Wd DR d S a iDNS E A R D S DI i t
| |
| _ l b m e e p a a R T N n
| |
| o it a
| |
| c if id o
| |
| M f
| |
| o #
| |
| y erp s
| |
| r a alteu m carGo 1 1 1 3 3 2 1 1 1 1 4 /3 A 1 m Cy N S
| |
| u Rde uv NtSa i t
| |
| n s
| |
| 6 i e
| |
| - r i
| |
| f e 1 e eim sd 1 2 4 8 9 1 1 2 t t
| |
| a Ca o Na l l a a a a a a l l l l c c i i R B M
| |
| C yon mmmmmmmm1 e e e e e e s s2 i i r e eHD ewn r
| |
| d y i par Rdi Nu ac t
| |
| I t
| |
| n n n n n n S I
| |
| t I
| |
| t I
| |
| t I
| |
| t I
| |
| t e eS F F S NWS i i t
| |
| S Gno Oir e t
| |
| Wcab
| |
| / im f
| |
| 1 2 3 4 5 6 7 8 9 0 1 1
| |
| 1 2 3 4 1 1 1 I
| |
| iu Rd PoN .
| |
| EM .
| |
| | |
| l Section 3 EPRI/WOG INTEGRAL PLANT MODEL l
| |
| The previous section provided an overview of the types of accident sequences considered in the NRC study, together with a measure of their significance. This section discusses these sequences in more l detail and describes the best-estimate plant response, as considered in this EPRI/WOG study. Focus is placed on the sequence of events that the operators must face, their potential actions to recover, the time within which they must recover, and the equipment available to them.
| |
| The EPRI/WOG integral plant model includes both transient and small LOCA event trees, containment safeguards, and support system success criteria. These trees include key recovery actions, for case of display of our models and assumptions. A description of each of the various scenarios is provided, with I' emphasis on the operator actions as specified in the procedures. Also discussed are the importance of i specific initiating events, plant design features, and both intemal and external event sequences and initi-
| |
| . ors. Finally, containment safeguards and support systems are discussed.
| |
| TRANSIENT EVENT SEQUENCES Four types of transient initiators, each of the which has a unique impact on the likelihood of core damage, were considered in this study. These unique impacts are highlighted in the following discussion. The four initiators are:
| |
| l T1- Loss of offsite power i T2- Iess of Power Conversion System (PCS) l T3- Reactor / Turbine trip T5- Imss of a DC bus Except for loss of a single AC bus (T4), these initiators are identical to the NRC Case Study initiators.
| |
| In the case of loss of a single AC bus, the plant would not automatically trip if an essential AC bus was lost. Ioss of redundancy in the Service Water or Component Cooling Water systems would probably )
| |
| initiate an orderly plant shutdown using main feedwater operated off the non-safety buses. If a non- I essential bus were lost, a reactor trip would occur, however, there would be no impact on safety
| |
| / stems. Any important loads on the non-essential bus could be switched to the essential bus.
| |
| I 31 ,
| |
| I
| |
| | |
| Therefore, loss of a non-essential bus can be included in the T transient 3
| |
| category for reactor / turbine ips, i.e., reactor trips without a unique safety impact.
| |
| In the case of reactor / turbine trip (T3), this initiator is insignificant in the EPRI/WOG study due to a change in modeling of PORV actuation and the tesulting potential for transient induced LOCAs. The change in modeling is discussed later in this section. 'Ihis change caused one of the most significant dif-ferences in results between the two studies.
| |
| Event trees were used to identify accident sequences resulting in core damage. The event tree for the four transients considered in this study is illustrated in figure 3-1. Along the top of the event tree are those events which can functionally affect the progression of the accident. That is, success or failure of the event affects the need for other systems or functions required to prevent fuel damage. In the tree, failure of an event is designated by the downward branch under that event. The upward branch indicates successful achievement of the function or event. Each of these events is discussed below in teims of operator actions and the equipment which must function (i.e., the success criteria) as well as the time period over which it must function (i.e., the mission time).
| |
| The various combinations of success and failure events is termed an accident sequence. For clarity and ase of reference, only the failure events are included in these designators. The various types of acci-
| |
| .sent sequence end states fall into three categories: core melt (CM), non-core melt (NCM), and a transfer to the small-break LOCA (SBLOCA) tree. Not every path includes a branch for every event in ' I the event tree. In these cases, the success or failure of an event is either predetermined or irrelevant in evaluating the end states ofinterest.
| |
| \
| |
| \
| |
| Transient initiators result in a demand for a number of safety functions and, therefore, a number of l potential accident sequences. The first of these is the requirement for the reactor protecion system (RPS) to terminate the nuclear chain reaction. This study assumes success of this function, as does the NRC Case Study. Primary system integrity must be maintained, ensuring enough mass to allow for long-term heat transfer to the steam generators. Secondary heat removal (i.e., steam generator cooling using main or acxiliary feedwater and secondary relief valves or the main condenser) is required to prevent boiloff of pnmary system inventory. If steam generator cooling is not available, then feed and-bleed cooling can remove decay heat directly by releasing primary coolant to the containment through the PORVr and replacing it with water from the HPI system. Once the feed and-bleed cooling mode has been established, either secondary heat removal must be recovered or high pressure recirculation (HPR) must take place.
| |
| ailure of primary system integrity can result from relief valve challenges occurring at about the time of the reactor trip, or through subsequent failure of the RCP seals. (The latter event,ifit occurs,is delayed 3-2
| |
| | |
| <lll r
| |
| o 3
| |
| R ii 3
| |
| R _
| |
| et i i l t 2
| |
| c e f f 2 a3 2 B3 R2 nn eg RRRCC G ui 1 RRREE,E E E EE i i 1 1
| |
| * i 1 1 1 1 q s RRRRt RRt t Rt e e 9D 1
| |
| LLLLULLUULUe T T T T T T T T T T T T T T e
| |
| ws
| |
| - . o e
| |
| l C GWaOMsGMaWasG5 .
| |
| E 1 y u r c e $
| |
| r v R i no ,
| |
| coc ei e Rt R i' i' e .
| |
| e Ml
| |
| ; r u
| |
| c i H
| |
| e r
| |
| h ; in T g ; c o HW R t i ei t n
| |
| e y - v rd E
| |
| ee 2 t v e r R n oF e c t e ef a R c w - ~ i' i
| |
| s n
| |
| g o a ne e r i d p C T gi s r v y a o0 d h r8 C P 1 - u t
| |
| s n S h e p
| |
| t G aeO O pY 2 R k G W Y S ci /
| |
| R I -
| |
| O rt P eS - R P
| |
| E d
| |
| n a E 1 d
| |
| d o -
| |
| eo 3 el F B s
| |
| - e e r y - nt u -
| |
| rd i u g ee n i i v e ri R F oF em c t ef a0 R o w3 yr re at i e l w L
| |
| i d xe ue A F e
| |
| nd e enBd p a e O
| |
| y oo t s S Vl nl Mrnc t a a e P ECR f -
| |
| t od es e e n
| |
| i eF T oe r nL ne e it '
| |
| r/e a T W M w wO I1 l ll 1 ll m
| |
| | |
| until some time after the trip.) Neither this EPRI/WOG study nor the NRC Case Study considers conse-
| |
| 'nential seal leakage or seal LOCAs since they are being considered in the A-44 program.
| |
| Failure of secondary heat removal results in primary system pressuri:ation and loss ofinventory through the PORVs. If this state persists until steam generator dryout without recovery of main or auxiliary feedwater and without successful feed and bleed cooling, the pressurizer PORVs and potsibly the safety relief valves (SRVs) will open. If those valves were to stick open, recovery of auxiliary feedwater no longer would be adequate by itself; thus,in this study core damage'is assumed to occur. If makeup was also recovered, successful core cooling may be possible but is not cmlited. If the valves reclose or are l isolated, full charging flow (i.e., feed and bleed with charging pumps and safety valves) or recovery of feedwater will prevent core damage. If feed-and-bleed cooling does succeed, recovery of secondary heat removal followed by successful closure or isolation of the PORVs will prevent core damage. If secondary heat removal is not recovered or the PORVs cr.n be neither closed nor isolated, high pressure recirculation will be attempted. If high pressure recirculation fails, makeup to the RWST or another water source will be required to sustain feed and bleed cooling.
| |
| EVENT Q . PORY OPENS EARLY AND CANNOT BE RECLOSED OR BLOCKED Transient induced LOCAs can result from either of two causes, a PORV sticking open or a loss of eactor coolant pump seal integrity. Failure of RCP seals is not considered in this study.
| |
| Pressure relief for the primary system at Point Beach is provided by two PORVs and two pressurizer i SRVs. 'Ihe set point of the PORVs is 2335 psig and the set point of the SRVs is 2485 psig. If instrument air or DC power is unavailable, the PORVs will not open. If AC power is unavailable, the block valve on each PORV cannot be opened or closed. About one half of the time, both PORV block valves are open and about one half of the time one block valve is closed. About one percent of the time, both block valves are closed.
| |
| Pressure relief occurs if the primary system cannot transfer sufficient heat to the secondary system.
| |
| Failure to transfer sufficient heat could occur shortly after reactor trip during a core power feedwater mismatch. Event Q addresses the potential for loss of RCS integrity at this time. Ioss of RCS integrity occurring near the time the steam generators boil dry (i.e., if secondary heat removal is completely lost and not recovered) is considered in event Q2.
| |
| Only a portion of transient initiators will result in a sufficient core power-feedwater mismatch ta cause a relief valve to actuate. Based on many thermal hydraulic analyses by Westinghouse,it isjudged that of j he five transient initiators considered in this study, only loss of main feedwater (T2) and loss of offsite l power (TI) would result in PORV actuation. (It should be noted that even for those initiators the l 3-4
| |
| _ - _-_ _ _ _ =_-
| |
| | |
| PORVs may not be actuated.) Furthermore, even if both PORVs fail to open or are blocked, pressure is tiikely to increase to near the SRV setpoint. Thus,in no transient initiator assessed in this study is an SRV expected to open during the time period considered in event Q.
| |
| i During this early period of plant response, steam relief from the PORVs will be sufficient to compensate for the temporary mismatch in RCS heat removal. Experience with relief valve failures indicates that they are much more likely to reclose under steam relief conditions rather than liquid relief.
| |
| The EPRI/WOG study evaluates transient induced LOCAs resulting from one of two PORVs stickir.g open under steam relief conditions shortly after a loss of main feedwater or a loss of offsite power. The block valve not only must be open, but also must fail to close. Consequently, the LOCA occurs short'y after reactor trip.
| |
| Modeline Assumptions Neither the EPRI/WOG nor the NRC study explicitly considers that either a PORV or SRV could stick open due to overpressurization from over charging. Because it is likely that operator actic,n will prevent overcharging from occurring and that HPI would be available anyway, this scenario is insignificant to risk. Further, such failures are considered in the basis for the small break LOCA frequency. This ssumption is consistent with the conclusions of the analysis performed for the Oconee PRA L13) in i which such LOCAs were explicitly considered and found to be insignificant toirisk.
| |
| Comoarison of EPRl/WOG and NRC Studies The NRC Case Study evaluates transient-induced LOCAs using the following methodology. In the NRC Ccse Study 7% of the time when a reactor trip occurs, an unspecined relief valve will open, and there is a 2% chance that one of these relief valves will stick open. Once this happens, the valve cannot be isolated. Since the NRC Case Study assumes that the block valves on the PORVs are closed, they assumed that the SRV is the stuck-open valve, that isolation could not occur, and that the relief valve would open at about the time of reactor trip.
| |
| The differences between the EPRI/WOG and NRC studies are substantial. First, two of the fxr risk sig-nificant sequences in the NRC Case Study (T3QH1'H2' and T3QDID2) are not possible, eccording to the EPRI/WOG approach. These two sequences begin with a transient occurring with both main feed-l water and offsite power available. In this condition, relief valve setpoints would not be reached.
| |
| l l
| |
| Second,in the EPRI/WOG study, the likelihood of a PORV failing open considets that the valve can be wlated by closing the block valve (precluded in the NRC Case Study). For those cases where the block valve is available, such as loss of main feedwater or loss of a DC bus initiator, the overaillikelihcod of a ;
| |
| transient induced LOCA is lower in the EPRI/WOG study. For those cases where the block valve is !
| |
| 3-5
| |
| | |
| unavailable, i.e., loss of offsite power initietor with station blackout and consequential loss of an AC bus ir that PORV's block valve, the relative likelihood of a transient-induced LOCA is higher in the EPRI/
| |
| WOG study than the NRC Case Study. This changes the transient-induced LOCA/ station blackout sequence from a non-dominant sequence in the NRC Case Study to the third highest internal event sequence in the EPRI/WOG study.
| |
| EVENT L - AUXILIARY FEEDWATER This event addresses the likelihood of AFWS failure; recovery is addressed in subs:quent events. Point Beach has three AFW pumps available for each unit. A turbine-driven pump is dedicated to each unit and can provide flow to either steam generator in that unit. Each of the two motor-driven pumps can provide flow to a single steam generator in either unit. Flow from any AFW pump to either steam gen-erstor is sufficient for adequate secondary heat removal.
| |
| 1 The most restrictive condition for AFW is a station bla:kout, under which conditiori only the turbine j driven pump can operate. Further, the long-term source for feedwater supply requires refilling the CST using either a diesel or a portable fire pump. Under other conditions, feedwater supply can be provided frorn the hotwell or the service water system. Use of the service water systerr. or the fire pumps requires the use of lake water. If DC power is also lost, the turbine- driven pump must be controlled locally, and ne steam admission salves must also be opened locally. Finally, cooling to the AFW turbine-driven pump beariig is nornully and automatically pmvided by service water. In the event of a station black-out, diesel-driven water normally used for fire fighting or other local actions is available to ensure cooling of the pump. These manual actions are called for in the procedures or training.
| |
| AFW operation is also particularly important for the two fire scenarios censidesed in this study. For both fires, only the turbine-driven pump is available for cooling 25 minutes af:n the fire starts if the fire is in the AFW pump room, and after 60 minutes if the fire is in the switchgear room. Assunung the fire is diagnosed quickly as expected and that the severity of the fire will require the operators to trip the plant, either rnain or auxiliary feedwater will operate for some time before fire damage leaves only the turbine driven pump. Additionally, the DC-operated steam admission valves will have opened, i.e.,
| |
| manual action will not be required.
| |
| 1 Modeline Assumptions {
| |
| It is assumed that there is sufficient secondary steam relief capacity such that the failun of steam relief can be assumed to be of negligible probability. Also,it was judged that a stuck open relief valve will Se an insignificant risk contributor to failure of the AEW turbine-driven pump. This latter assumption is most imponant for station blackout scenarios where the turbine-driven pump is the only available pump. j Since steam for the turbine can be provided from both unm generators, the operators would pmvide 3-6 l
| |
| | |
| sterm frorn the intact generator. Check valves and/o: ? fain Steam Isolr. tion Valves (MSIVs) would event blov,dowa of both steam generators or stc vin , of the turbine. The likelihood that valves on each saam generator fail open is small.
| |
| Comoarison of EPRl/WOG and NRC Studies ne two studies are similar except for two important considerations. First, the NRC study does not con-sider the manual actions necessary for. AFW operation without AC power. Given the likelihcod that the operators will take manual action and the limited impact of the dependency, this difference is not significant to the final results. Second, the EPRI/WOG study would consider the impact of temporary feedwater success for the fire scenarios, except these scenarios trsult in very low frequencies for other reasons.
| |
| EVENT R1 - RECOVERY OF FEEDWATER IN 30 MINUTES The normal response after a reactor trip is for the plant tc stabilize at hot shutdown conditions, with heat removal provided by the steam generators.
| |
| Failwe of event L results in the loss of both main and auxiliary feedwater. (Main feedwater is iost l
| |
| stially during trips which lead to risk-significant accident sequences.) The T3 initiator does not result in main feedwater interruption; it is not analyzed further smce it does not adversely affect any safety systems. The R1 event considers recovery of main feedwater after initial loss, which is expected to occur in most cases. The subsequent timing of events will depend primarily on how and when main feedwater is lost, as well as if auxiliary feedwater functions for a short period of tirne before failing.
| |
| The most restrictive conditions occur when the loss of main feedwater initiates the plant trip and auxili-
| |
| )
| |
| 1 ary feedwater never functions. Under less restrictive conditions, the timing of events will be extended, thereby providing more time for any required operator actions. The remaining discussion for this event i
| |
| }
| |
| assumes the most restrictive conditions. 1 If main f:edwater is lost to the steam generators while insufficient auxiliary feedwater flow is being I provided, Emergency Operating Procedure - Critical Safety Procedure - H1 (EOP-CSP-H1), entitled (
| |
| " Response to Loss of Secondary Heat Sink," directs the operator to take the steps described below. f The operators would be verifymg the status of varioes systems. Rey would likely enter the procedure a few minutes after plant trip. By this time, the RCS conditions would have stabilized. The initial pres-
| |
| ! suritation caused by the mismatch between power and feed flow may have caused the PORV to open. l
| |
| 'f the PORV later becornes stuck open,it i.;likely that operators would block it before serious loss of coolant inventory. Any energy loss out the PORV would delay the time when the PORVs would next 3-7
| |
| | |
| h need to open. While the steam generator tubes could begin to uncover, the RCS would remain stable
| |
| , 'nce steam generator heat removal capability would not yet be significantly affected.
| |
| The fcilowing five. steps from EOP-CSP-H1 describe operator actions to recover the secondary heat sink by initiating auxiliary feedwater from the control room or locally, by opening the main feedwater isola-tion valves and providing feedwaer, or by depressurizing the steam generators and recovering the condensate system.
| |
| : 1. Check if Secondary Heaf Removalis Required Before implemendng actions to restore flow to the steam generators, the operator is directed to check if
| |
| =madary heat removal is required. The operator is also directed to evaluate whether or not conditions would allow placing the RHR system in service. For the transient event tree it is assumed that RCS con-diuons would not allow operation of the RHR system. Also, heat removal via the steam generators would always be preferred in liep of feed ar.d-bleed. In the case of small LOCAs and other initiators for which EOP CSP-H1 applies, secondary heat removal may not be required since heat would be removed through the break.
| |
| : 2. Try to Establish AFW Flow to at Least One SG 1e first attempt to restore feed flow is through the automatic operation of the AFW system. If AFW fails, operators are instructed to check Control Roorn indications for potential causes of AFW system failure to provide feed flow. If the cause of the failure cannot be corrected from the Control Room and/or minimum AFW flow cannot be reestablished, an operator is dispatched to continue AFW system restoration locally while the Control Room operators move on to step 3.
| |
| Successful operation of this step will result in recovery for that portion of the accident sequence cut sets I which include pump and MOV control circuit faults, as well as recoverable common-cause faults. This I step should be completed with high reliability if the auxiliary feedwater equipment faulu are recover-able. Assuming that within 1015 minutes after a reactor trip an operator reaches the failed equipment, another 15-20 minutes is available prior to steam generator dryout. If no pressurizer relief valve sticks open,45-50 minutes is available before core uncovery begiris. l
| |
| : 3. Stop All Reactor Coolant Pumps (RCPs)
| |
| RCP operation results in heat addition to the RCS water. By tripping the RCPs, the effectiveness of the remaining water inventory in the SGs is extended. This also extends the time in which operator action to initiate feed and-bleed inust occur. This extension provides an additional 5 minutes for the operator restore feedwater flow to the SGs. This action will also increase by a few minutes the time between PORV opening (a symptom for actuating feed and bleed) and the time of steam generator dryout (the 3-8
| |
| | |
| /
| |
| time by which feed and-bleed must be initiated). The probability that this step would be conducted roperly is expected to be very high. This step would have occurred automatically for loss of offsite power sequences.
| |
| : 4. Attempt Restoration of Main Feedwater Flow to at Least One SG Main Feedwater (MFW)is the next source of high-pressure water readily available for operator use to reestablish secondary heat removal. Prior to restoring MFW to the SGs, the operator verifies condensate system operation to ensure a source of water to the MFW pumps. Then the status of MFW isolation is checked. If feedwater isolation has occurred, various actions may be required (depending on the logic for Feedwater Isolation (FWI), to reset Safety Injection (SI) and FWI signals, and reopen the FW1 valves. If neither the condensate system can be placed in service nor any FWI valves opened, the l operator is directed to check if secondary heat sink condidon.: require initiation of feed-and-bleed.
| |
| If the condensate system is operational and FW isolation valves are open, MFW is established by the operator. If MFW cannot be established, the operator is directed to step 5 to attempt to establish flow to the steam generators using condensate purnps only. This MFW recovery step will succeed for roughly 90% of the loss of main feedwater events. For loss of offsite power events, roughly 50% of the time off-site power (and auxiliary or main feedwater) would be recovered before steam generator dryout. If the litiator was a loss of a DC bus, the electric MFW pumps could be restarted easily and FWI valves opened. Recovery probability would be limited only by the same time available for recovery. Recovery could be faster and more likely than assumed if the necessary action could be initiated from the control room.
| |
| : 5. Try to Restore Feed F:ow from Condensate System To use the condensate system for direct feed into the SGs, the pressure in at least one SG must be reduced below the shutoff head of the condensate pumps. In order to depressurize at least one SG to less than the shutoff head pressure (550.pcig) of the condensate pumps, the RCS must be depressurized below 1950-psig to allow blocking of the low steamline pressure SI and low pressurizer SI signals. If 1 these signals were allowed to actuate, feedline and steamhne isolation actuation rignals would close the feedwater isolation valves thereby complicating this recovery process.
| |
| Auxiliary pressurizer spray is used to depreseurize the RCS because normal spray is not available (RCPs I
| |
| have been tripped). Auxiliary spray is provided by the charging system. It is the best method for depressurization because it provides maximum cooling to the primary system while allowing no loss of primary water inventory. If auxiliary sprays are net available, pressurizer PORVs are used.
| |
| 3-9
| |
| | |
| c-___--.
| |
| Depressurization of the SG(s) is accomplished through the SG steam dump valves or turbine bypass
| |
| .lves to the condenser, the SG atmospheric dump valves, and the turbine-driven AFW pump. If SG depressurization cannot be accomplished or the condensate system cannot be placed into operation, then the operator is directed to check if secondary heat sink conditions require initiation of feed-and bleed.
| |
| Condensate movery by itself is not credited in the EPRI/WOG study due to the difficulties in determin-ing the percentage of loss of main feedwater transients for which this recovery would be applicable.
| |
| That is, loss of main feedwater initiator data generally does not contain sufficiently detailed infonnation to identify the percentage of transients for which main feedwater is not recoverable but condensate is.
| |
| Comparison of EPRI/WOG and NRC Case Studies The pnneipal difference between the two studies involves the loss of DC bus initiator (TS). Recovery of either DC power or main feedwater dramatically reduces the frequency of this event below the fre-quency assessed in the NRC Case Study. Otherwise, since both studies include about 90% effectiveness for recovery of main feedwater and recovery of signal and control faults in MOVs and pumps in the AFW system, there is not a significant difference between the two studies for the T1 and T2 initiators.
| |
| (There is some difference due to recovery modeling of time-dependent human reliability.)
| |
| EVENT E: FEED AND BLEED COOLING If secondary heat removal is not recovered, wide range steam generator level indication will decrease.
| |
| Heat transfer degradation will occur after about 20 minutes when most of the tube bundle is uncovered i
| |
| and the RCS begins to heatup. The RCS heatup will also be indicated by the increasing pressurizer level and pressure caused by RCS fluid swell. 'Ihe PORVs will open before SG dryout and the end of this phase; steam relief will then occur. If the PORVs opened early in the event, they will open again later in this period.
| |
| RCS Feed and Bleed Initiation Criteria If any of the following conditions occur, the operators are directed to trip all RCPs and immediately per- )
| |
| fonn the feed-and bleed procedure. l A. Wide Range Steam Generator Level - Less than 10% in both SGs.
| |
| B. Pn:ssurizer Pressure - Gnater than 2335 psig due to loss of secondary heat removal.
| |
| I l ae time at which each of these initiation criteria occurs will depend on the plant response and, to a les-l ser degree by instrument uncertainties. If the wide range SG level instrumentation is reading accurately 3-10 l
| |
| w_____--____-_
| |
| | |
| i or low, this indication probably will be reached first, perhaps soon after entering this time interval. If Ne instrumentation is readir.g high, then the pressurizer pressure signal may be reached first. Whether J I
| |
| or not the PORV had opened earlier will also influence the relative times of these two signals.
| |
| The most limiting condition would be a situation where 1) the PORVs opened immediately after trip, and therefore would not open again until late in this period; and 2) wide range stea n generator level read higher than actual (but within its uncenamty range). In this case, feed-and bleed would have to be initiated within a few minutes.
| |
| In the best estimate scenado, feed and bleed initiation criteria would occur on wide range level indica-tion well before feed and-bleed was actually required.
| |
| In the most limiting case feed-and bleed would be initiated as based on the above criteria at about the time of steam generator dryout and would be requimd a few minutes later according to licensing calculations. In the best estimate case, feed and-bleed might be initiated about 10 minutes before steam generator dryout. Actual steam generator dryout (i.e., versus indicated) would occur at about 35 minutes. Therefore, feed and bleed initiation should occur between 25 and 35 minutcs after a reactor trip involving immediate les of secondary heat removal.
| |
| A the best estimate case, the remaining steam generator inventory will help reduce RCS repressurization and allow more safety injection flow into the reactor. The core will be uncovered for a shorter period of time. If only one PORV opens, feed and bleed may still prevent extended core uncovery; however, the EPRI/WOG study conservatively assumes core damage if only one PORV opens. In the most limiting case, delay in initiation of feed and bleed beyond a few minutes may lead to extended com uncovery.
| |
| Feed and bleed cooling is 3 accomplished with the following steps:
| |
| : 1. Actuate Safety injection (SI)
| |
| Actuating the SI signal ensures that high pressure injection (HPI) flow is available to provide RCS makeup for heat removal. It also ensures that the cc,ntamment is isolated and cooled so as to confine reactor coolant releases resulting from extended RCS bleed flow though the PORVs.
| |
| : 2. Verify RCS Feed Path A. Verify that at least one HPI pump is running.
| |
| B. VeCy SI valve alignment.
| |
| 3-11
| |
| : 3. Reset S1 i order to realign safeguards equipment and to provide sustained instrumen: air for the pressurizer PORVs, a deliberate action must be taken to reset the SI signal. For Point Beach, instrument air is needed to maintain pressurizer PORVs in the open position for an extended period of time. The SI signal needs to be reset as part of reestablishing instrument air to containment. l l
| |
| l
| |
| : 4. Reset Containment Isolation ,
| |
| The purpose of this step is to remove the " locked in" signal, which keeps all contamment valves closed. l This allows equipment to be realigned. Operator action is required to remove the "CLOSE" signal. No I valve will reposition upon actuation of these resets, but subsequent control actions will open the valves.
| |
| These valves should remain closed, unless necessary process streams are being established. The j isolation signal needs to be reset as part of reestablishing instrument air to containment.
| |
| : 5. Reestablish Instrument Air to Containment The purpose of this step is to restore sustained compressed air supply to allow control of air-operated equipment inside containment (i.e., charging and letdown valves, pressurizer PORVs, etc.).
| |
| The following actions are taken at Point Beach:
| |
| A. Open both instrument air isolation valves.
| |
| B. Verify that one air compressor is running.
| |
| : 6. Establish RCS Bleed Path A. Verify that both pressurizer PORV block valves are open.
| |
| B. Open both PORVs.
| |
| : 7. Verify Adequate RCS Bleed Path j A. PORVs are open.
| |
| B. Only one PORV is open.
| |
| If two PORVs are not available, the operators are directed to open reactor vessel head and pressurizer l vents to either the Pressure Relief Tank (PRT) or the containment, depressurize at least one SG below the shutoff head of the condensate pumps using atmospheric steam dump, and align any available water 1
| |
| source to depressurized SGs. For the transient event tree no credit was given for this alternate depres-surization procedure.
| |
| 1 3 12
| |
| | |
| l hindeling Assumptions ORV failures dtie specifically to instrument air losses are not considered in this study. Instrument air compressors are powered from the safeguard buses and will be available unless a station blackout occurs. In the event of a complete blackout, HPI pumps will not be available for feed-and-bleed. Also, it is assumed that the valves necessary to establish instrument air to the containment will not fail closed.
| |
| Local operation is possible. Further, failure modes of the valve which would prevent local operation, e.g., disk separation, would not have allowed them to close in the first place on the containment isolation signal accompanying the SI signal.
| |
| Comoarison of EPRI/WOG and NRC Studies There is little difference between the modeling in the two studies. It should be noted however, that the EPRI/WOG study believes that failure of one PORV causing failure of feed and-bleed may rot result in core melting. In almost every case where one PORV has failed, DC operated vent valves will be available and actuated per procedure. These will provide additional depressurization capability. Also, the calculations which indicate two PORVs are necessary use licensing based assumptions, e.g.,1207c decay heat, rninimum steam generator inventory due to instrument uncertainnes, extended core uncovery unacceptable, no charging flow available, and no action until time of steam generator dryout.
| |
| By opening the PORVs earlier at 307c of wide range SG level, a single PORV should be adequate.
| |
| /hile the EPRI/WOG study scope did not include performing plant-specific thermal hydraulics analysis with best estimate codes, this study believes a conservative bias exists for this event's success criteria.
| |
| One important difference between the NRC and EPRI/WOG studies may be that the NRC study allows feed-and bleed to occur as late as 60 minutes after reactor trip. While this is not explicitly stated, the recovery model allows 60-minute recovery for these scenarios. As indicated in the previous discussion, the EPRI/WOG study indicates that the RCS may not depressurize enough to allow sufficient HPI flow shortly after steam generator dryout (approximately 30-40 minutes). If the recovery action also included feedwater recovery, e.g., recovery of offsite power, this would be a moot point. As is discussed below, the EPRI/ WOC study allows only feedwater recovery after steam generator dryout.
| |
| EVENT Q2. PORY PATHS OR SRVS STICK OPEN Shortly after steam generator dryout, feed-and bleed may no longer succeed in preventing core damage.
| |
| PORV capacity may not be sufficient to depressurize the RCS enough to allow sufficient HPI water injection. Feedwater recovery is assumed to be the only available option. It can only succeed by itself if RCS integrity is maintained. RCS integrity could be threatened if a PORV or an SRV opens and icks open.
| |
| 3-13
| |
| | |
| If MFW is lost at the time of reactor trip and AFW fails to operate, the steam generators will boil dry
| |
| 'fter about 30 minutes. Shortly before this time, the PORVs will open. Shortly after this time, the
| |
| ;RVs may open. If the steam generators do boil dry, RCS pressure will quickly rise to the PORV set-point. At first only steam relief will occur. However, as the pnmary system repressurizes, the liquid level in the pressurizer will swell. Eventually, liquid relief through the PORVs will occur, and by this time, relief through the SRVs will occur. This EPRI/WOG study assumes that PORVs or SRVs will open at this time and that all could subsequently fail to reclose. Failure probabilities are based on the more restrictive conditions of liquid relief. If during either time period, a PORV fails to reclose, its block valve can be used to isolate the break (provided it does not fail to operate).
| |
| If RCS integrity is lost, further recovery is not credited. Loss of RCS inventory will alter the timing of the event, possibly accelerating the need ''r recovery to before 60 minutes after reactor trip. If both AFW and HPI are recovered prior to core overheating, recovery will succeed. However, because the EPRI/WOG study did not have explicit information regarding this sequence of events, recovery was not credited.
| |
| Modeline Assumptions As discussed above, all PORVs and SRVs are assumed to be demanded to op:n and pass subcooled or tturated liquid (which increases their failure probability). Consistent with the Oconee PRA, a probabil-sty of 0.1 is used for the likelihood of a loss of RCS integrity at this time.
| |
| Comparison of EPRI/WOG and NRC Study The NRC study does not consider the effect of a stuck open relief valve occurring at the time of steam generator dryout. If this situation was to occur,it would cause auxiliary feedwater recovery to not be effective if ECCS was not also available. If a relief valve sticks open at this point, recovery of feedwater alone cannot prevent core damage. Such a scenario is probably most important for a station blackout, where recovery options would be limited to the turbine driven feedwater pump. The actual differences in quantification are evaluated in section 4.
| |
| EVENT C - CHARGING PROVIDES 180-GPM Three charging pumps operating to provide 180-gpm can provide an alternative water source for feed-and-bleed. Their combined flow rate equals decay heat at about or shortly after the time of the steam generator dryout. Furthennore, since these are positive displacement pumps the flow can be delivered at l the safety ve.lve setpoint making this backup to feed and bleed independent of PORV operation. At the ne of dryout, the operators would initiate feed-and-bleed which would call for maximum charging.
| |
| l Therefore, the failure to initiate this option would be accounted for by the human ciror to implement 3 14
| |
| | |
| feed-and bleed in the previous event. Since success of this event requires three operating charging umps, quantification of event C's failure probability must account for maintenance of any pump, failure of any pump, as well as the unavailability of an AC bus.
| |
| One or two operating charging pumps will extend the time of the event and proside more time for operators to implement normal feed and bleed or to recover feedwater. The allowable additional time for initiating feed-and-bleed with one or two operating charging pumps is not included in the study, and I thus is a conservative analysis assumption.
| |
| EVENT R2 - RECOVERY OF FEEDWATER Recovering secondary heat removal after feed and-bleed has been attempted can be impcrtant whether feed and bleed succeeded or failed. In the event that feed and-bleed fails or is initiated too late, recovery of feedwater will be the only means of pteventing significant core uncovery. Recovery would be required within about 60 minutes after reactor trip. (Successful feedwater recovery at this point requires that PORVs are blocked and that if SRVs are open, they do not stick open.)
| |
| In the event that RCS feed-and bleed cooling had been initiated successfully, the operators will still con-tinue to attempt to regain feedwater flow to the steam generators to be able to terminate feed and-bleed.
| |
| ivent R2 is included in the transient event tree to account for the possibility of late MFW or AFW recovery and subsequent termination of feed-and-bleed cooling prior to the need for high pressure recirculation.
| |
| Comoarison of EPRI/WOG and NRC Case Studies Recovery of feedwater at this time is effectively considered the same in both studies. The NRC Case Study assumes that feedwater can be recovered within 60 minutes to prevent core damage as does the EPRI/WOG study. The NRC Case Study also assumes that recimulation from the containment sump is not required for transients. Therefore, these accidents should not appear in the list of dominant accident I
| |
| sequences. However, one sequence of this type initiated by a loss of main feedwater does appear, but at a frequency of about 2.0E-08. The EPRI/WOG study determined that this event will be less than 1.0E-08 due to differences in recovery modeling. This or other transient sequences requiring recirculation but not resulting in LOCAs were not considered further in this study.
| |
| EVENT H1 - HIGH PRESSURE RECIRCULATION in the event that RCS heat removal cannot be restored before the RWST is depleted, an altemate source f suction for the HPI pumps must be established to maintain core cooling. This mode of cooling I involves using the LPI pumps to take suction from the reactor coolant and RWST inventory that has 3-15
| |
| | |
| collected in the containment sump as a result of RCS leakage through the PORVs. In this mode, the
| |
| 'PI pumps are aligned to tale suction from the discharge of me LPI/RHR heat exchangers. This mode of cooling could then be sustained indefinitely or until feedwater delivery to the SGs could be reestab-lished.
| |
| The EPRI/WOG and NRC Case Study models are the same.
| |
| EVENT R3- RECIRCULATION RECOVERY I
| |
| As the level in the RWST drops during the safety injection mode, the operators will begin to consider implementation of high-pressure recirculation if auxiliary feedwater has not been recovered.
| |
| Emergency procedure EOP 1.3 directs the operators to take preliminary steps in the lineup for sump recirculation when the RWST is 60% full. Many of the problems that could cause failure of recirculation, e.g., common-cause failure of MOVs, would be identified at that time. The procedure also directs the operators to cross-connect the RHR and SI pumps at the 28% level by lining up only one ECCS train at a time. His procedure then directs the implementation of recimulation, one train at a time at 10% RWST level.
| |
| If sump recirculation is unavailable at either of these times, ECA 1.1 advises that the operators should ,
| |
| y to restore the faulted equipment and, failing that, to refill the RWST. The RWST can be refilled at a rate of about 160 gpm using:
| |
| o Water from the spent fuel pit canal using the holdup tank recirculation pump; o Water from the chemical volume control system (CVCS) such as the two 100,000 gallon reactor makeup water tanks; or o Water from the unaffected unit's RWST using the refueling water circulation pump.
| |
| The procedure also directs that HPI and containment spray flows should be reduced (note that contain-ment sprays would not be operating for these sequences) and that maximum charging should be pmvid-ed. Charging does not require the RWST.
| |
| Given that recirculation is not expected to be required untillong after the accident has begun, decay heat removal should be satisfied by either charging or RWST refill. De amount of flow from the charging system if only two pumps are operating, i.e.,120-gpm, would be capable of matching decay heat in a l time period of approximately three hours after trip. Therefore, quantification of this event yields higher probability of success than event C.
| |
| 3-16 j q
| |
| _ _ _ _ _ i
| |
| | |
| Comoarison of EPRI/WOG Study and NRC Case Smdy ne EPR1/WOG study allows recovery from recirculation failure for all situations where charging pumps are available, as well as for those situations where failed equipment can be manually recovered to establish the normal recirculation flow path. The NRC Case Study allows only manual recovery to establish the recirculation flow path, (i.e., refilling the RWST or using charging is not considered).
| |
| SMALLI OCA EVENT SEQUENCES This EPRI/WOG study considers two small LOCA initiators, a LOCA whose size is less than an equiva-lent 3" diameter pipe and a transient-induced LOCA through a PORV or an SRV. The event tree for these two small LOCAs is illusuated in figure 3-2.
| |
| Like the transient event tree, success of the reactor protection system (RPS)is assumed. Since im entory is being lost, the first safety function required is that HPI and/or charging succeed in inventory makeup.
| |
| If this function fails, feedwater can be u:ed to depressurize the RCS to the point where accumulators and LPI can be used to maintain core cooling. If cooling is established early,it must contmue over the long-term. If the LOCA is either isolable or above the core, the reactor can be depressurized and placed in RHR cooling mode. If not, core cooling requires recirculation through the containment sump or covery actions which provide a long-term cooling capability.
| |
| EVENT D1 - HIGH PRESSURE INJECTION / CHARGING Most small LOCAs will initially depressurize the primary system enough to cause a reactor trip and safety injection signal. The remaining small LOCAs (LOCAs of about 0.5" diameter or less) will require a normal cooldown to RHR entry.
| |
| The safety injection signal starts the HPI pumps and aligns the pumps' suction from the RWST. The flow of one HPI pump will be adequate to prevent core uncovery for all small LOCAs.
| |
| Decreasing the pressurizer level willinitiate full flow from the charging system. Charging system flow is 60-gpm per pump. The size of the break will determine the number of charging pumps required to operate to provide sufficient flow. For this study,it is assumed that charging from three purnps,i.e., full flow charging, will prevent core damage for a small LOCA with a diameter of 3" or less. This assump-l tion is consistent with the NRC's Case Study, which assurnes that the 200-gpm shutdown decay heat f removal (SDHR) system will succeed for these small LOCAs.
| |
| l
| |
| , i
| |
| } gnificant small LOCAs in this study include random RCS pipe breaks, stuck-open and unblocked PORVs, and seismically-induced RCS pipe breaks The seismically induced break is of particular 3 17 L____ _ _
| |
| | |
| R r 2E o R HH et EE 1 1 ca MH HH nn ii S332L eg Ht D000M ui S3S1 1 1 1 1 i 2E DDDDDDDDD qs ee EE E E E 22 2 SD SSSSSSSSSS _
| |
| e c
| |
| n es .
| |
| us qe MMM e _.
| |
| CMC C C MMM e el S C MMM C
| |
| N NCNNNCCC T r
| |
| e y t
| |
| l u r n
| |
| c e R e r v v i no E coc A Oie t R C y
| |
| i' O
| |
| e1 L l H e .
| |
| u- 2 l r
| |
| c wu H l a
| |
| r os 1 m
| |
| inLs H co e s eirr Rt oP l L ,I d
| |
| y u
| |
| n t w S o 3 d 0 G dt O l u W oh /
| |
| CS I R
| |
| P E
| |
| 2 D .
| |
| S 2 I -
| |
| P 3 L
| |
| y r e r
| |
| ret re u g
| |
| o1 a . i 1 w M F n1 d i xe eue MA F -
| |
| /
| |
| n eoS
| |
| - .in i ti
| |
| : cg D hg er
| |
| .jo i .nh H7 I C -
| |
| t -
| |
| rn -
| |
| oed t e 2 1 ec 5 1 nuA sedC ernO U
| |
| sTI L Y ;;i l lllIlIl llIl,
| |
| | |
| interest because of the comparatively low capacity of the RWST for seismic events. If the RWST is lamaged, other Seismic Category I equipment is available to provide flow to the RCS. A stuck open PORV may be a significant event since it could occur during a station blackout when neither the PORV could be blocked nor HPI or charging operated to make up leakage.
| |
| Modeling Assumptions Success criteria assumptions for charging system flow are mentioned in the text above. A particularly imponant assumption is the failure mode of the RWST under seismic acceleration. 'Ihis study assumes that at low accelerations the tank would leak rapidly rather than rupture catastrophically. The technical basis and further discussion of this assumption can be found in section 6.
| |
| Comoarison of EPRl/WOG and NRC Case Studies There is little difference between the two studies for this event except for the consideration of the charging pump flow. Because obtaining full charging flow is less likely than having one of two HPJ pumps succeed, this assumption has a small effect on the results.
| |
| EVENT ML - FEEDWATER If HPI and/or charging fail to provide sufficient makeup to prevent core damage, an altemative means of cooling can be provided. If secondary heat removal is available, the operators can cooldown and depressurize the RCS until the accumulators and LPI can provide suf5cient water injec. ion to prevent core damage. Therefore, the question of feedwater availability is considered at this point in the accident sequence. ,
| |
| A small LOCA will cause a safety injection signal which, in tum, will close the feedwater isolation and bypass valves. Use of main feedwater will require the operators to reset the safety injectior and feed-l water isolation signals. AFW will be acruated on low steam generator level.
| |
| Comoarison of EPRIAVOG and NRC Case Studin The studies differ only in assumptions about main feedwater availability. The NRC study does not consider the case of recovery of main feedwater after a safety injection signal. However, the EPRI/WOG study did not quantify the main feedwater recovery. Auxiliary feedwater is very reliable and no statistics were available for main feedwater recovery after a safety injection signal. Both studies require depressurization of the steam generators. Because of other EPRI/WOG changes to the NRC Case Study models and data, quantification of this event would not result in a noticeable change in the reponed EPRI/WOG results.
| |
| 3-19
| |
| | |
| EVENT D2 - LOW PRESSURE INJECTION SYSTEM
| |
| ' the RCS can be successfully depressurized to the low-pressure injection point, inventory makeup for the small LOCA can be provided by LPl. This study requires only one LPI pump to function, whereas the NRC Case Study requires two LPI pumps. A single pump is fully capable of providing the few hundred gpm required. This difference between the two studies yielos a noticeable, but not significant, difference in che final core melt frequency results.
| |
| EVENT D3 - COLD SHUTDOWN This event represents the opportunity to place the plant in cold shutdown despite the presence of a RCS break. In all PWR experience, recirculation has never been required despite the occurrence of a few reactor coolant pump seal LOCAs, instrument weld failures, and stuck-open valves. Like the NRC Case Study, this study did not explicitly credit the ability to achieve cold shutdown. The EPRI/WOG study did, however, consider this event in its determination of a cmall LOCA frequency. 'Ihat is, very small breaks were assumed to result in cold shutdown before the RWST empties. Larger sized small break LOCAs were assumed to require recirculation, but they are represented by a correspondingly smaller frequency (since they have never occurred). Section 5 contains additional discussion of this effect.
| |
| '. VENT H1H2 - RECIRCULATION As in the transient event tree, recirculation from the sump must be established if the plant cannot be placed in cold shutdown before the RWST empties. The NRC and EPRl/WOG models are similar.
| |
| EVENT R3 - RECIRCULATION RECOVERY As in the transient tree, failure of recirculation can be recovered by refilling the RWST or by using the charging pumps. This difference between the EPRI/WOG and NRC Case Studies is the most significant difference in internal event modeling. The dominant NRC Case Study internal event sequence becomes a small contributor in this EPRI/WOG study.
| |
| CONTAINMENT SAFEGUARDS AND SUPPORT SYSTEM SUCCESS CRITERIA In addition to the event trees described above, the modeling of contamment safeguards and support systems can play an important role in the overall risk profile of the Point Beach unit. The following describes the findings of the EPRI/WOG study.
| |
| Containment systems were modeled using success criteria consistent with recent IDCOR findings 01),
| |
| ; well as the consideration that only small LOCAs and transients are modeled in the study. Because !
| |
| l recent IDCOR work has found that large dry containments have capacities greater than twice the design 3-20 1
| |
| 1 ---_.___________j
| |
| | |
| 1 pressure and because the RCS leakage from transients and small LOCAs does not place a significant oad on containment, containment safeguanis were not required to operate beginning at the time of accident initiation.
| |
| l The principal role of the containment safeguards (i.e., sprays and fan coolers) is preventing containment failure in the long term. In this role, the safeguards need only remove decay heat. A single Point Beach j fan cooler will rernove the decay heat being generated about three hours after trip. Two fan coolers are sufficient shonly after trip. A single containment spray train will remove all decay heat from the point of accident initiation. A single containment spray train will also remove radionuclides rapidly enough to provide successful radioactivity removal for all accident scenarios.
| |
| The NRC Case Study placed significantly more restrictive success criteria on containment safeguards.
| |
| Those success criteria were based on FSAR success criteria for a large LOCA design basis accident.
| |
| Despite this significant conservative bias in the NRC Case Study success criteria, most containment safeguanis failures were caused by failures of suppon systems, in particular, total loss of AC power.
| |
| Because these suppon system failures cause failure of all the operating trains of either coolers, sprays, or both, there was not a significant difference between the final results as far as initial availability of containment safeguards were concerned. However, the NRC Case Study did not consider recovery of
| |
| .ontainment safeguards, e.g., after a station blackout-imtiated core melt. Since many hours ne available for recovery before containment failure, the probability of containment safeguard recovery is higher in this EPRI/WOG study than in the NRC Case Study for many sequences.
| |
| Support system success criteria for the EPRl/WOG study were the sarne as the success criteria for the NRC Case Study except for two important systems, Component Cooling Water (CCW) and Service Water (SW). A review by WEP found that CCW is not required for HPI operation in the injection mode since CCW provides seal rather than bearing cooling. This EPRI/WOG study assumed that recirculation of hot sump water would cause seal degradation and eventual pump failure unless cooling from CCW was available. The NRC Case Study assumed that CCW was required for HPI. Removal of this o
| |
| conservatism alone reduced core damage frequency by about 3.6E-5 for related sequences.
| |
| The service water system success criteria were also changed frota the PRC Case Study. First, the requirements for SW cooling of CCW, which in turn cooled the HPI pumps, were removed for the same reasons as discussed above. Second, discussions with WEP indicated that only one or two service water pumps were required to cool equipment such as diesel generators if operators take manual action to provide flow only to needed equipment. The EPRI/WOG analysis required recovery actions if less than 4 tree of six SW pumps operated. The NRC Case Study did not allow this recovery; it required three of
| |
| .x SW pumps at all times. -
| |
| 3-21
| |
| | |
| Section 4 HUMAN RELIABILITY AND RECOVERY ANALYSIS INTRODUCTION The EPRI/WOG human reliability and recovery analysis took into consideration both human actions as identified in the NRC Case Study and recovery actions identified by a thorough review with plant opera-tors and emergency responders. For many of these recovery actions, the discussion in this section is an elaboration of information given in the previous section. The tirne available for operator action in all cases is based on the plant response discussed in : hat section.
| |
| Since TMI, nuclear utilities have expended considerable resources to strengthen their accident manage-ment capabilities. Consequently, a PRA must realistically and credibly reflect these effons through the use of a Human Reliability Analysis (HRA).
| |
| This discussion is a critique rather than a complete HRA A complete HRA cannot be retrofitted onto an existing PRA without more resources and significantly more information than that available in the basic NRC Case Study repon. Nevenheless, a number of issues can be addressed in a critique, and important insights gained. This analysis began by addressing those human errors identified in table 4-1 and the human recovery model given in table 4-2. The following objectives were pursued:
| |
| o Search for omissions of those events that could affect detennination of the actual l potential of the Point Beach operations, personnel to respond to an event. In par-ticular, look for additional recovery accons.
| |
| o Reevaluate the probabilities of events that were included in the NRC Case Study for those cases that could significantly affect the risk results or conclusions.
| |
| o Account for Point Beach implementation of post-TMI modifications, including ,
| |
| the implementation of the Westinghouse emergency response guidelines (ERGS), '
| |
| a safety parameter display system (SPDS), training in emergencies and contingen-cies, and the response team m the Technical Suppon Center.
| |
| o Evaluate the impact of the findings on the Point Beach risk profile.
| |
| In structuring this reanalysis, the basic ground rule was realism. Post TMI modifications were to be accounted for, recovery and response activities were to realistically reflect Point Beach operations irocedures, training, and policies; and the HRA methodology was to be consistent with recent industry and NRC methods.
| |
| 4-1
| |
| | |
| Table 4-1 Human Failure Events EPR1/WOG Review Comments NRC Case Study NRC Failure . Failure Failure Analysis . Updated
| |
| , Fault ID Dagginial Pmbabihty Mod.c - Adeaunte ProbabihtY
| |
| : 1. VOC tJIM MOV notitstored after T or M 8.3E 5 slip yes -
| |
| : 2. XOC UTM manual valve not restored 8.3E 5 slip yes --
| |
| aher T t. M
| |
| : 3. NOC UTM me.4iydraulic valve 8.3E 5 slip yes -
| |
| : 4. NCO-UTM pneumatic / hydraulic valve 8.3E 5 slip yes -
| |
| : 5. PMD-UTM pneumatic / hydraulic valve 8.3E-5 slip yes -
| |
| : 6. XCC-OE operator opens wrong manual SE-3 slip yes -
| |
| valve .
| |
| */. VCC-OE operator opens wrong MOV 3E 3 slip no IE-4
| |
| : 8. HPRF MANACT operator fails to manually 3E-3 rule no IE-4 initinic HPR 9.OPF.MACT-FNB operator fails to rnanually 3E-3 decision no IE 3 initiate feed and bleed
| |
| : 10. OPF-OPEN- operator fails to manually open 3E 3 slip yes FNB PORVs for feed and bleed 11.OPF SI RESET operator fails to reset SIS signal 3E-3 slip yes
| |
| : 12. OFF-MACT. operator fails to manually 3E 3 rule no "
| |
| CSR initiate CSR
| |
| : 13. OPF MACT- operator fails to manually start 3E-3 diagnosis no AFW 7P3 the AFW turbine-driven pump
| |
| : 14. OPF-MACT- operator fails to manually start 3E-3 slip no HPA OPF- HPIpumps for feed and bleed MACT HPB 15.OPF MOF'2N- operator fails to manually open 3E-3 slip no MOVA OPF MOV for feed and bleed MOPEN MOVB
| |
| : 16. SUMP VCC-OE operasor fails to realign properly 1E 3 rule no IE-4 for recirculation from the sump l
| |
| The NRC Case Study assumes this to be an independent failure mode, which it is not. The result is double or even triple counting of the same failure. These " modes" should be coalesced l into one, i.e., eliminated (see page 4-8).
| |
| l ''
| |
| 'Ihese eve: s did not appear in dominant internal event sequences in the EPRI/WOG reanalysis, because they are expected to be still lower in probability, they were not explicitly evaluated in this review.
| |
| 4-2 t
| |
| | |
| .i L
| |
| ! . Table 4-1 (Continued)
| |
| , , . . Human Failure Events EPR1/WOG Review Camments NRC Case .
| |
| ! Study NRC Failure ' Failure Failure Analysis Updased Fanit ID Desenption Probability Meds Adeauate hgbabdity
| |
| : 17. DESGENA- diesel generator not restored 83E-5 slip yes -
| |
| GEN TM after T or M
| |
| .DESGENB..
| |
| GEN TM 18.TM BA'ITA- battenes not restored after T 83E-5 slip yes -
| |
| -DO5TM. or M BATTB-DO6 19.SWSMP 54- service water pumps not 83E-5 slip yes -
| |
| UTM restored after T or M i
| |
| Table 4-2 Non-recovery Probabilities Used in the NRC Case Study Available time in min. un to:
| |
| Recovery of: E 10 20 3Q 40 60 20 120 24Q Loss Of RA-1 0.7 0.7 0.5 0.4 0.3 0.3 0.2 0.1 Offsite Power
| |
| ! Loss Of Main RA-2 1.0 1.0 1.0 0.9 0.6 0.1 0.03 0.01
| |
| . Feedwater Battery RA-8 1.0 1.0 1.0 1.0 1.0 1.0 0.9 0.8 Common Cause Battery Fault RA-9 1.0 1.0 0.9 0.8 0.7 0.7 0.5 0.2 Diesel RA-10 1.0 1.0 1.0 1.0 0.9 0.9 0.9 0.7 Common Cause Diesel Fault RA-11 1.0 1.0 1.0 1.0 0.9 0.9 0.9 0.8 i
| |
| Other Failures RA-6 0.3 0.1 0.05 0.03 0.03 0.01 0.01 0.01 In Control Room q Other Failures RA-7 1.0 0.3 0.1 0.05 0.03 0.03 0.01 0.01 i
| |
| ht of Control Room i
| |
| 4-3 l
| |
| | |
| n 2 * ?
| |
| 4 The information bases for the critique were:
| |
| : 1. The NRC Case Study for Point Beach (1), in particular:
| |
| l
| |
| , - Appendix A - Systems Descriptions and Simplified Fault Trees l
| |
| - Appendix B - Intemal Event Sequence Analysis (and HRA) l - Appendix D - Fire Analysis
| |
| ! - Appendix E-Intemal Flood Analysis i
| |
| i
| |
| : 2. The IREP study of Ark'sasas Nuclear One-1 @.
| |
| : 3. The Seabrook PSA @.
| |
| : 4. The WEPC implementation of the Westinghouse ERGS (2).
| |
| : 5. An interview of Point Beach operations staff on August 13,1986.
| |
| : 6. Answers to questions directed to the project contact at WEPC.
| |
| : 7. The NRC handbook on HRA, NUREG/CR-1278 (1Q).
| |
| The basic criteria used to evaluate any particular postulated human failure event were:
| |
| : 1. Is the event risk significant?
| |
| : 2. Does the event make sense from human reliability and operational perspectives?
| |
| : 3. Is the assessed probability of the event supported by data or the latest HRA methodology?
| |
| I These criteria were used in conjunction with the experience gained in performing Westinghouse plant HRAs (,11,12) and three other I.evel 1 or higher HRAs (16,12 and 20), together with the HRA technol-egy developed by SAIC (21). Two categories of human failure events were identified and analyzed in this critique: (1) risk significant events included in the NRC Case Study but not assessed properly, and (2) events that could be risk significant but were not included in that study.
| |
| There are two basic human failure modes-mistakes and slips. Mistakes are failings in decision making, general diagnosis, or diagnosis using rule-based procedures. Slips constitute the failure to detect the need for action, or the failure to execute a decision. The former mode is predominantly due to a failure in the conscious thinking processes, while the latter mode is predominantly caused by a failure in ettention allocation or some other subconscious process. l
| |
| *Ihe review process utilized here examined the events included in the NRC Case Study analysis, those nitted from that analysis, the failure modes for the events, and the quantification methods used. The
| |
| .eview then proceeded on a sequence cut set basis, top-down from those intemal sequence classes considered most significant to core melt frequency.
| |
| 4-4
| |
| | |
| Wherever a human failure event need:d quantitative evaluation or reevaluation, this review used public-omain methods. These methods include the techniques documented in the final version of NUREG/CR-1278 and modifications, plus a time reliability correlation system developed to account foi events and human factors not included in that NUREG.
| |
| COMPARISON OF EPRI/WOG STUDY AND NRC CASE STUDY The principal differences are summarized below. The impact of these differences on the NRC Case Study core-melt frequency is provided in table 4-3. The following findings describe changes in the HRA quantification:
| |
| o Reduce failure to initiate feed and bleed from 9E 3 to 1E-3,i.e., a value consistent with most other PRAs and NUREG/CR-1278.
| |
| o Reduce failure to initiate sump recirculation from 3E-3 to 1E-4, i.e., a value con-sistent with NUREG/CR-1278 and most other PRAs.
| |
| o Reduce failure to depressurize from 1.5E 2 to 3E 3, i.e., a value consistent with most other PRAs.
| |
| o Change the human recovery time reliability correlation from NUREG/ CR-2787 to the model contained in NUREG/CR-1278, thereby yielding reductions in human recovery failure figures by a factor of about 50.
| |
| The following recovery actions have been added to the NRC Case Study model:
| |
| o RWST refill using water from the spent fuel pool and the chemical volume and control system.
| |
| o Aligning the CCW system to the other unit's system.
| |
| o Cross-connecting AC or DC buses for initiating events caused by the loss of an AC or DC bus (included in Section 6 analysis). I o Manual control of the turbine-driven auxiliary feedwater pump.
| |
| o Provision of a backup supply of feedwater using either service water or CST refill, o Use of the charging system for loss of feedwater sequences.
| |
| o Balancing loads on the service water system to respond to low system output i when less than three pumps are available.
| |
| o Restoration of main feedwater (included in Section 6 analysis).
| |
| 1 o Recovery from common mode failures.
| |
| o Recovery from fire and flood scenarios.
| |
| 4-5 l ..
| |
| | |
| Table 4-3 Summary of Human Reliability and Recovery Analysis Changes and Their Impact on NRC Case Study Core-Melt Frequencies
| |
| * Reduction 1n Core-Melt Frequency Description of Model Changes Per Reactor Year Feed-and-Bleed Operator Error 1.5E-7 Recirculation Switch Operator Error 2.8E 5 Operator Opens Wrong MOV 1.0E-6 CCW Recovery 8.0E-7 Manual Operation of AFW 4.8E-6 Cross-connecting AC or DC Buses 1.4E-6 Recirculation Failure Recovery 7.2E-5 Alternative to RWST(Seismic) 1.5E-5 CST Alternative (Seismic)- SWS 1.5E-5 LST Inventory Recovery - Station Blackout 3.5E-5 Service Water Load Balancing 4.4E-6 Diesel Start with New Batteries 4.8E-6 Loss of Offsite Power Timing Study 1.7E-5
| |
| * Totals are not provided because some recovery actions are adundant.
| |
| 4-6
| |
| | |
| l Further, the EPRI/WOG reanalysis does not limit the number of recovery actions to one in the first two ours and two in total, as in the NRC Case Study, unless, for example, the recovery analysis indicates specifically that sufficient time is available for only one action.
| |
| }
| |
| EPRI/WOG STUDY ANALYSIS OF KEY NRC CASE STUDY HUMAN ERROR PROBABILITIES i
| |
| As noted previously, the NRC Case Study identified three key human failure events for responding to {
| |
| off-normal events. The NRC Case Study quantification did not account for the diagnostic and decision-making elements of either the pacific sequences or the general events. The study also did not credit backup of incorrect actions by multiple crew members and other emergency response team personnel.
| |
| Table 4-4 shows the corresponding values from other recent PRA sources for the three major NRC Case Study human failures: failure to initiate or implement feed and bleed, failure to switch to (high pressure) recirculation, and failure to depressurire the steam generators and add secondary coolant if HPI fails.
| |
| Table 4-4 Human Failure Values from other PRA Sources NRC NUREG/ Case Event CR-1278 ANO-1 Zion (21) Indian Pt (22) Catawba Study Failure in 1.6E-3 SE-4 1.3E-4 SE-3 2E-2 9E 3 (6) feed & bleeda Failure 3E-4 IE-4 1.3E-4 3E-4 4E-4 3E-3 in HPRC Failure in second- -- .. .d .e ..f 1.5E-2 ary depressurization a "Ihe Millstone 3 PRA (21), the IPE screening valve, and the Seabrook PRA use 10-2 and the PUN PRA (2_4_)4 uses 6.4E-4 at 30 minutes.
| |
| b Three failure modes with 3E-3 for each.
| |
| c HPR is high presstue recirculation.
| |
| d Depressurizanon following a steam generator tube rupture was assessed at 6E-3.
| |
| e Depressurization following a steam generator tube rupture was assessed at SE-3.
| |
| f This mode of failure was considered too unlikely to quantify.
| |
| Feed-and-Bleed
| |
| .leed-and-bleed is an altemative to cooling the core when all secondary heat removal is lost. The value of this altemative was recognized as the result of the TMI accident and is now a rule in every PWR's 4
| |
| 47 I
| |
| _ _ _ _ _ _ _ _ _ _ _ . i
| |
| | |
| emergency procedures. The feed-and-bleed option is a last recourse that would be used only when econdary heat removal cannot be recovered. It is not a desired option because an extensive plant outage and cleanup of radioactive water in the containment building would be required. One could question the likelihood that the SRO or shift supervisor would or should make the decision to initiate feed-and-bleed. For this reason, this action may be considered uncertain, no matter how clearly indicated in procedures or trammg.
| |
| Problems related to decision making are not treated in NUREG/CR-1278 (1), the source book for this HRA, as well as all of the PRAs mentioned in table 4-4, except for Catawba. The SAIC methodology (20) is a system of time reliability correlations (TRCs) that considers:
| |
| : 1. The time available from a leading cue to act to the point that the action will not be effective;
| |
| : 2. Whether the action is explicitly a rule in the new symptom-based emergency pro-cedures;
| |
| : 3. Whether the decision to act is hindered by goal conflict, burden or other sources of hesitancy;
| |
| : 4. Whether the plant / utility management clearly removes the hesitancy associated with conflict through assurances to operators; and
| |
| : 5. A myriad of other possible influences, reflected in a Success Likelihood Index (SLI) that ranges from a best of 1.0 to a worst possible of 0.0.
| |
| The Catawba analysis used the first four steps. Feed-and-bleed is clearly a rule in all Westinghouse-based procedures. Conflict in feed-and-bleed is clearly real in all PWRs. The Catawba operators exhib-ited hesitancy to initiate feed and-bleed in simulator exercises and interviews. Thus, the failure probability was assessed 0.02 (the available tine was assumed from analysis to be 20 minutes).
| |
| Two Point Beach operators were queried regarding a change to feed-and-bleed and they showed no hesitancy. 'Ibey claimed that management clearly supported the use of feed and-bleed as specified.
| |
| FRG H.1 of Point Beach's emergency procedures cues feed-and bleed initiation on both neam generator levels indicating low (55" wide range). This act is to supersede but not stop any activity to recover some means of providing FW flow. Since operators have good access to the AFW equipment, a recov-l crable fault in AFW would likely be determined within 10 minutes and reported back to the control room. Therefore, the recoverability of AFW will be known quickly and thus less likely to cause hesitan-cy to implernent feed-and-bleed.
| |
| The thermal hydraulic analysis for low-pressure plants (1) indicates that about 15 minutes is available bom the first " cue" for feed and bleed (low level in both steam generators) to the time of core uncovery.
| |
| l If the first cue for diagnosing the need for feed and-bleed is considered to be indication of total loss of 4-8
| |
| | |
| feedwater, at least 30 minutes would be available before feed and bleed would no longer prevent fuel damage. Table 4-5 indicates the resulting feed-and-bleed failure probabilities with these two different available times and a SLI of 0.5 (assuming nominal human factors) versus a SLI of 0.7 (assuming fair-to-good human factors). The effects of assuming hesitancy and no hesitancy are also indicated. The results range from 0.002 to 0.05 for 15-min available time and from 6E-5 to 0.01 for 30-min available time. The Point Beach plant specific information suggests using a SLI of 0.7, and to assume non-hesitancy. This yields a range of failure pmbabilities from 0.002 for 15-min available time,i.e., cue on low steam generator level, to 6E-5 for 30-min available time, i.e., cue on total loss of feedwater I indication.
| |
| I Considering the above range of values (2E-3 to 6E-5), that given in table 4-4 (2E-2 to 1.3E-4), and that given in table 4-5 (SE-2 to 6E-5), the EPRI/WOG study selected 1.0E 3 for its feed and-bleed failure probability. The reduction in core melt probability obtained by using the value for feed-and-bleed error is only 1.5E-7. This small change indicates the relatively weak dependence of core melt probability on this parameter.
| |
| Table 4-5 Spectrum of Feed-and-Bleed Failure Probabilities SLI = 0.5 (nominal human factors) 15-min 30-mm no hesitancy 0.006 0.0003 hesitancy 0.05 0.01 SL1 = 0.7 (fair-to-cood human factors) 15-min 30 min no hesitancy 0.002 0.00006 hesitancy 0.02 0.006 A subsidiary problem is tnat the NRC Case Study has three events that are treated independently but are obviously mutually dependent. These are OPF MACT-FNB, OPF SI RESET, and OPF-OPEN FNB, which are presented in table 4-1. These events are all part of the integrated, successful executior of feed and bleed. They are evaluated at the same probability and thus, together, triple the contribution to one ent. This practice is a remnant of early applications of NUREG/CR-1278 and is not current practice.
| |
| Because of the presence of sufficient personnel, both to identify the need for and to perform feed and-4-9
| |
| ]
| |
| - _ _ _ _ _ _ d
| |
| | |
| V ne t
| |
| bleed, the plant's acceptance of the feed and bleed requirement and its consequences, and operator dning, the only credible failure mode is due to the decisional component.
| |
| Recirculation Failure Recirculation switchover is also a rule in any PWR's emergency procedures and is a simple action based on an obvious indication (the low-level alarm for the RWST). From a human perspective, the action l
| |
| would occur many hours following a successful injection cooling mode. Small LOCAs and transients do not require much makeup and, therefore, slowly deplete the RWST. At this time, there would be many, people watching and verifying its implementation. Therefore, there is no basis for assessing a high human failure probability. Table 4-4 bears this out, yielding probabilities 10-30 times lower than the NRC Case Study estimate. A value for SUMP-VCC-OE of IE-4 was used in this EPRI/WOG study.
| |
| Steam Generator Deoressurization There is a tendency in the operation of PWRs to try to stay at normal operating pressure unless a leak is
| |
| - uncontrollable or some fault manifests itself in high pressure injection. In this case, an available AFW cooling loop and steam generator atmospheric steam dump valves can be used to rapidly cool down and ;
| |
| depressurize the steam generators. This cools down the primary side. If the primary system is cooled.
| |
| 'own sufficiently, more core cooling equipment is available (i.e., LPI and RI-IR). The bounding value ;
| |
| ,f 1.5E-2 used in the NRC Case Study dominates that assessment of failure to depressurize. The value 3E 3 used in this study is more reasonable and consistent with actions specified in Point Beach procedures and training. The effect of this change is negligible.
| |
| EPRI/WOG STUDY ANALYSIS OF RECOVERY ACTIONS The fellowing is a summary discussion of the recovery analysis details which are provided in Appendix A. The recoveries discussed are those added to the model. The appendix also describes the time reliability correlation used in this study for the recoveries already identified in the NRC Case Study.
| |
| The correlation is the same as that used in the development of the human error probabilities quoted
| |
| - below for the new recoveries.
| |
| Mitientine a Imss of All Service Water or Comoonent Cooline Water The number of interconnections of a typical two unit Westinghouse PWR often allows for altemative sources of water or rerouting of flows to satisfy the cooling missions of the two systems. Point Beach Abnormal Operating Procedure (AOP) AOP 9A applies to service water (SW) system malfunctions and
| |
| ' OP 9B applies to component cooling water (CCW) system malfunctions. These procedures direct the
| |
| -perator to secure equipment not required and to reroute flows from the other unit.
| |
| 4-10
| |
| | |
| .l Specific service water losses are discussed in another event. Component cooling water losses were important to the NRC Case Study since CCW was assumed to be required for HPI oumps and therefore -
| |
| .or feed-and-bleed cooling. As this restriction should not apply to Point Beach, it reduces the importance of this event for the EPRl/WOG study.
| |
| Operating the Turbine Driven AFW Pump WithoutDC The loss of DC power, coincident with a loss of all station AC, means that the operators can monitor the steam generator water level only with the instrumentation powered by new batteries, installed after the original Point Beach NRC Case Study. It also means that the steam supply motor-operated valve for the turbine-driven AFW pump must be controlled manually at the valve's loca: ion with communication with the control room provided by security system radios, if necessary. In this situation, the operators can open the control valve and open the outlet valves locally. The operators stated that they could control the turbine-driven AFW pump successfully guided by the instrumentation which is powered by the new batteries.
| |
| Emergency Procedure ECA 0.0 directs the control room operators to dispatch an operator to monitor or control the turbine-driven AFW pump immediately upon loss of all AC or DC. Operators estimated that it would take 10-15 minutes to manually start the pump, including local operation of the DC steam
| |
| 'dmission valves. The time available for recovery would then be about 15 20 minutes before steam denerator dryout and PORV or SRV opening, and about 45-50 minutes before it is assumed that core damage could not be averted by feedwater only. Based on the time reliability correlation presented in table 4-6 (taken from NUREG/CR 1278), the non recovery probability would be 0.03 for recovery j before PORV or SRV opening and 0.0003 before core damage could no longer be averted by feedwater !
| |
| only. These two probabilities resporcM the entries in the table for 15 minutes and 45 minutes.
| |
| Based on these numbers, the conditional probability of recovery between PORV and SRV actuation to avert core damage would be 0.01 if auxiliary feedwater recovery was successful for all scenarios in that time period. However, the EPRI/WOG analysis assumes that this conditional probability is limited to 0.1 (the probability that a small LOCA requiring primary system makeup would occur in that time period). The 0.1 probability is the EPRI/WOG estimate of the likelihood of a stuck-open SRV or PORV after they have passed liquid. This probability is based on the Oconee PRA review of EPRI relief valve test data. Consequently, the overall non-recovery probability for this es ent is 0.03 times 0.1 (or 0.003).
| |
| The remaining recovery values reported in this section were calculated in a similar manner.
| |
| Cross-connecting DC or AC Se desif,n of both the AC and the DC trains of electrical powerincludes ways to cross-connect the AC :
| |
| of one t ain to the AC or DC of the other train, through a cross-connect charger. This recovery can be 4 11
| |
| | |
| Table 4-6 Time Reliability Correlation for Human Errors During Recovery (from NUREG/CR-1278)
| |
| Human Available Non Recovery Time (Minutes)* Probability 5 0.3 10 0.1 20 0.01 30 0.001 60 0.0001
| |
| * Logarithmic interpolation used for times not listed in the table.
| |
| executed by a manual closing of one or more circuit breakers. However, these breakers open automati-cally on loss of one bus. This recovery would be possible for the loss of one AC bus but not the whole C grid, or for the loss of one DC bus but not both. The time needed to effect such a recovery is about 15 20 minutes. A recovery factor of 0.05 would be consistent with the NRC Case Study model for actions outside the control room. A value of 0.003 would be consistent with the previously-mentioned correlation in NUREG/CR-1278. However, this review uses 0.005, a value which accounts for a 0.002 unreliability of the breakers opening and a 0.003 value for failure to recover.
| |
| Recirculation Failures The RWST can be refilled if a sump recirculation path it not available. Additionally, charging pumps taking suction from the two 100,000-gallon reactor makeup water tanks can also be used to recover from recirculation failures even if HPI or LPI pumps were damaged due to cavitation. Emergency procedure ECA 1.1 directs the operators to attempt both of these recoveries. For its recirculation recovery calcula-tion, the EPRI/WOG study assumes that containment sprays are operating and that recirculation failure occurs early. A non recovery probability of 0.05 is used to account primarily for equipment failures as well as human failure to diagnose the event and take the correct action.
| |
| Increasing the rime available by terminating containment sprays would not reduce this assessment j significantly since the above value assumes a significant contribution from equipment unavailability. A vare detailed systems analysis might indicate lower equipment failures and allow reducing this value; l
| |
| .owever, the required analysis was beyond the scope of this study.
| |
| 4-12
| |
| | |
| -r Recovery of Feedwater on Imss of CST Inventorv Ass of CST inventory can result from two important causes. First, either tank can fail due to a seismic event, causing both to leak. Second, inventory will eventually be depleted in the event of a station
| |
| -blackoat. A means of feedwater recovery is available for each of these sequences. Both are directed in cmergency opemting procedures, many of which state:
| |
| "If CST level decreases to less than 4.$ feet (the tech spec limit), the alternate witer sources will be necessary."
| |
| Operator training addresses means of obtaining altemate sources including service water and by refill of the CST using the diesel-driven fire water pumps. Appendix A discusses these two modes of makeup and their role in mitigating seismic events and long tenn station blackout events.
| |
| Balancine Loads in the Event of Less than Three Service Water Pumps As was discussed previously ir. Section 3. if the service water success criteria requiring three operating ,
| |
| pumps is not satisfied, operator actions can be taken to balance loads to the critical components, usually the diesel generators. AOP-9A provides direction in this regard. Appendix A discusses this cperation in general, as well as specifically addressing recovery for recovery of CCW and recovery for operation of te diesel generators when normal station power has been lost.
| |
| Use of New Seismic Catenorv I Station Batteries These betteries provide the operator with instrumentation in the event of loss of normal station DC.
| |
| 'These station batteries provide an altemate source of DC power for startmg the diesel generators.
| |
| Further, they provide extended DC power in the event of a long term station blackout.
| |
| Recovery of Common-Cause Failures Experience reported in section 5 indicates that most common-cause failures would be recoverable within 30 60 minutes. Accordingly, common-cause failures are reduced from the original data to account for this possibility.
| |
| Loss of Offsite Power Timian Study This discussion in appendix A and the modeling used in the EPRI/WOG study considers the time depen-dency of station power loss and recovery. Florida Power Corporation's Crystal River Unit 3 PRA (lE)
| |
| +udy has shown that explicit consideration of time dependency in the recovery of station power will
| |
| ,,ield a reduction of about 40% in station blackout core-melt risk.
| |
| 4-13
| |
| | |
| 'W.
| |
| Section 5 INITIATING EVENT AND COMPONENT DATA ANALYSIS ,
| |
| l L INTRODUCTION
| |
| , Generic or plant-specific data is used in the integral plant model of Point Beach in four areas: compo-nent failure and unavailability values, initiating event frequencies, some event tree top events, and common-cause (multiple, dependent) failure rates.
| |
| [ It is highly preferable that plant-specific data be used in a study whose objective is to identify plant modificati3ns; plant specific data most significantly changes how the results are used to define the modifications. Point Beach has 16 yeans of operating experience, anr1 this plant specific data and root I cause information would be valuable input to determining how reliability improvement might substitute for hardware changes. Unfortunately, however, the narrow scope of the EPRI/WOG study limited the '
| |
| plant specific data that could be obtained to initiating events only. The much larger task of collecting omponent data was not possible. Therefore, as mentioned in section 2, this review focused on factors that would change the results most significantly, and thus change the determination of the need for a -
| |
| plant modification.
| |
| Initiating event data were obtained by reviewing studies which summarized Westinghouse plant data, e.g., NUREG/CR-3862 (21), an up-to-dcte source of plant specific initiator frequencies. NSAC-103 (26) was used as a source for offsite power loss and recovery. Significant differences were found between generic data and plant-specific initiating event data for two categories, loss of main feedwater
| |
| . and all other transients. Loss of offsite power data for Point Beach were similar to generic data. Loss of DC bus initiating event frequency was significantly lower than the NRC Case Study's value; however, this reduction was due to consideration of reco tery rather than use of plant-specific data.
| |
| 'Ihe small LOCA frequency used in the EPRI/WOG study is consistent with other PRAs and with Westinghouse plant experience. The internal flood frequency was developed using a specific pipe break / leak model. Finally, fire frequencies were developed based on recently collected industry experi-ence and evaluation of the probability that transient combustibles would be present.
| |
| 5-1
| |
| | |
| The EPRI/WOG study used the same component data for independent events as was used in the NRC Case Study, except in the case of diesel genentor faults. In this instance the data set used was NS AC-108 (22), since it was more recent and carefully constructed.
| |
| Certain event tree top event probabilities differed, however, including Event Q in the Transient Tree (a transient induced LOCA), and Event X in the Small LOCA Tree (depressurization of the steam gener-ator). The EPRI/WOG study also examined Westinghouse plant experience applicable to stuck-open relief valve LOCAs. A systematic modeling approach yielded results more consistent with the Westinghouse experience. Basic assumptions about block valve status were based on WEP information about Point Beach operating experience. Common-cause failure data were estimated using methods consistent with current industry practices. The NRC Case Study values were used as a starting point, compared to other PRAs, and further updates were based on pimt design and experience as well as opportunity for recovery.
| |
| It is often difficult to pmvide an adequate basis for dependent event probabilities. Common-cause data frorn industry experience should be compared to plant design features to help define plant-specific coinmon-cause failure probabilities. The approach can be found in EPR] NP-3967 (21), but the resotuces available to this study did not permit it to be taken. Alternatively, then, insights from other applications were used to obtain estimates for plant specific values. As was the case with component data, an industry and plant specific data review for the root enuses of common-cause failures applicable to Point Beach system designs would not only yield more de fensible prot ability estimates, but identifi-cation of the root causes would lead to more practical and credible fixes for potential common-cause failures. Modifications which add another train of equipraent may not provide any additional safety margin if the underlying causes of common cause failure am not addressed.
| |
| The EPRI/WOG study also used the insights obtained f7orn the review of common-cause failure oper-ating experience. That experience, contained in AEOD Report C504 (22), indicates that common cause losses of safety function are often recovered within one hour. This review thus included recovery in its Essessment.
| |
| Overview of the NRC Case Study Anproach
| |
| 'Ihe NRC Case Study used the following sources of data:
| |
| Initiating events and most component data NUREG/CR-2728 (10)
| |
| Batteries - both independent and dependent NUREG-0666 (31)
| |
| Diesel generators dependent failures NUREG 1032 (12) 4 5-2 l
| |
| | |
| \
| |
| i Pump common-cause failures ASEP Q1)
| |
| MOV common-cause failures LaSalle PRA (M)
| |
| The only area in which modifications were made to the data used by the NRC Case Study was in the development of common-cause failure probabilities. For example, after using an Accident Sequence Evaluation Program (ASEP) beta factor for two motor-driven auxiliary feedwater pu:nps yielding a value of 5.2E-4, the NRC Case Study.then determined that 2E-4 was reasonable when accounting for the turbine-driven feedwater pump. Herefore, while ASEP and other efforts were references for some of the beta factors for common-cause failure probabilities, the NRC Case Study applied other judgment factors as to how they should be utilized.
| |
| COMPARISON OF EPRI/WOG AND NRC CASE STUDY Table 5-1 summarizes the impact on core melt frequency of the data differences between the two studies. These differences are discussed below.
| |
| Small LOCA Frequency The NRC Case Study of Point Beach included only one LOCA, the small (<2") break LOCA. The fre-luency used for this small LOCA is from the ANO-1 IREP (JD) and is dominated by RCP seal ruptures.
| |
| This seal LOCA frequency is based on an NRC memo to D. Eisenhut from T. Murley. This memo includes an LER search for seal leakage events, most of which show leak rates on the order of 1-10 gpm. Some events in the LER data did yield estimated leak rates greater than or equal to 50 gpm (Oconee in 1974 with 90 gpm. Arkansas Nuclear One Unit 1 in 1978 with 350 gpm, Robinson in 1975 with 400 500 gpm, Indian Point 2 in 1977 with 75 gpm, and Brunswick in 1975 with 50 gpm). Aware
| |
| < of this data, another NRC analysis (the ASEP study) decided that these seal LOCA initiators could be ignored for a large containment Westinghouse PWR. For small leakage rates, containment spray would not actuate and makeup could be maintained from the RWST for at least 20 hours since tank inventory would be depleted at only a few hundred gpm. Thus, recirculation would never be needed and recirculation failures, as in the NRC Case Study sequences, would not be valid. (In over 400 reactor years of PWR plant experience, recirculation has never been required.) ne Oconee PRA frequency for small LOCAs is 3E 3 and is also the result of an examination of actual industry experience. his fre-quency is consistent with other PRAs as well. It includes an isolatable small LOCA which occurred at Zion, and is therefore conservative in its application to this study. (As a comparison, WASH 1400 used 1E-3).
| |
| 53 ,
| |
| j
| |
| | |
| a Table 5-1 Initiating Event and Component Data Changes and o . Their Impact on NRC Case Study Core-Melt Frequency Model NRC Case < EPRI/WOG >
| |
| Reduction in Parameter Studv Value Value Core-Melt Frequency -
| |
| Initiatine Events Small LOCA 2.00E-2 3.00E 3 4.5E Loss of Offsite Power 8.40E-2 6.20E 2 1.1E-5 All other Transients 7.1 2.4 1.9E-5 Loss of AC Bus 6.00E-3 N/A 5.1E-7 Loss of DC Bus 3.00E-3 1.50E-5 8.1E-7 Flood in SW Room 2.20E-3 3.73E-5 7.6E-5 I Fire in AFW Room 2.38E-3 6.5E 4 9.5E-6 Fire in Switchgear 4.96E-3 3.5E-3 5.9E Total for Initiators is a Reduction of 55% or: 1.7E-4
| |
| %er Significant Events X - Depressurization 1.5E-2 3.8E-3 Negligible Q - Relief Valve / 1.4E-3 Cannot occur for 2.9E-5 PORV LOCA most transients Block Valve Status Always Closed Sometimes Closed 1.4E 6 Diesel Generator Faults 3.8E-2 2.2E-2 1.0E 5 Total for Other Significant Events is a Reduction of 13% or: 4.0E Common-Cause Data )
| |
| Batteries 9.6E-4 4.0E-4 1.7E 5 l AFW Pumps 2.0E-4 .3.0E-5 8.2E 6 l 1
| |
| SWS Pumps 2.0E 5 4.0E-6
| |
| ' 1.0E-6 CCW Pumps 8.0E-5 3.0E 5 2.lE-6 LPI, HPI, and CSI Pumps 1.0E-4 3.0E-5 2.9E-6 Sump MOVs 4.0E-4 8.0E-5 1.OE-5 Allother MOVs 4.0E-4 8.0E-5 7.0E-6 DieselGenerators 1.5E-3 5.0E-4 4.3E 7
| |
| 'otal for Common-Cause Data is a Reduction of 14% or: 4.3E-5 5-4
| |
| | |
| ?
| |
| Loss of Offsite Power (TI)
| |
| 'oint Beach has experienced only one complete loss of offsite power in roughly 16 years of plant oper-ating experience. The consideration of this experience results in a slight reduction over the generic experience used in the NRC Case Study (i.e., from 0.084 to 0.062 reactor years).
| |
| Loss of Main Feedwater (T2)
| |
| The loss of main feedwater frequency used in the NRC Case Study was originally 1.0/ year. After the following comments were submitted by industry, both the NRC and the EPRI/WOG studies use 0.1/ year for an unrecovered loss of main feedwater. He value of 1.0 was quoted from NUREG/CR 2728, the IREP Procedures Guide, table 5.3-1. His table lists the initiating event frequencies used in the Arkansas Nuclear One Unit I (ANO-1) IREP analysis. The frequency in the ANO-1 report was determined from EPRI repon NP 801, and includes the following categories for PWRs which would cause the total interruption of main feedwater:
| |
| ,Categorv Dmeription 16 Total Loss of Main Feedwater 17 Full or Partial Closure of MSIV (1 loop) 18 Closure of All MSIV <
| |
| 20 Increase in Feedwater Flow (all loops) 21 Feed Flow Instability - Operator Error 22 Feed Flow Instability - Mechanical Causes 24 Loss of Condensate Pumps (allloops) 25 Loss of Condenser Vacuum 29 Sudden Opening of Stearn Relief Valves 30 Loss of Circulating Water In the ANO analysis, it was assumed that approximately one-half of the closure of one MSIV events would result in a total loss of main feedwater. An event that was not considered in the ANO analysis is an inadvertent safety injection signal. In most Westinghouse plants, a safety injection signal causes feedwater isolation. However, main feedwater can be provided by clearing the signal and opening the valves.
| |
| l l \
| |
| 55
| |
| | |
| )
| |
| I l
| |
| I EPRI report NP-801 was used as the data base for the NRC Case Study initiating event frequency, )
| |
| 'though this report is old and several others have been issued since that time. In particular, EPRI report
| |
| )
| |
| NP-2230 was published in January of 1982 and NUREG/CR-3862 was published in May,1985. )
| |
| When NUREG/CR 3862 is used as an updated data base, the initiating event frequencies can be deter-mined with greater confidence resulting from the larger experience base, i.e.,418.19 years of operating cxperience for all PWRs, 274.63 years of experience for Westinghouse plants, and 24.28 years of experience for Point Beach Units 1 and 2. If a loss of main feedwater event is defined by the above categories and an inadvertent safety injection signalis included, then the frequencies cabulated are:
| |
| For all US PWRs Frequency = 384 events /418.19 years = 0.92 per reactor year j For Westinghou:e Frequency = 399 events /274.63 years = 1.45 per reactor year For Point Beach Frequency = 22 events /24.28 years = 0.91 per reactor year Recovery of Main Feedwater Categories 17,20,21,22, and 29 contain 307 Westinghouse plant events for which it is anticipated that main feedwater recovery is probable following the transient. If the ratio of recoverable to total number of events is considered, then the fraction of recoverable events is:
| |
| 307/399 = 0.77.
| |
| For a plant with electric main feedwater pumps, categories 18 and 25 would also be events in which main feedwater is expected to be recoverable. In this case, the fraction recoverable is:
| |
| 353/399 = 0.88.
| |
| Thus, for a plant with electric main feedwater putnps like Point Beach we would expect the frequency of non recoverable interruptions of main feedwater to be in the range of 0.11 per plant year (399-353/399).
| |
| Funhermore, for transients initiated by loss of a single nmin of either AC or DC power, the NRC Case Study has not included partial recovery of main feedwater as a viable means of heat removal. In these cases, since feedwater - condensate flow and control of breakers and feedwater control are trainwise separated, flow from the intact train can be restored from the control room.
| |
| All Other Transien.n f
| |
| The initiating event frequency for the category defined as All Other Transients in the hTC Case Study is quoted from NUREG/CR 2728 table 5.3-1 as 7.1 occurrences per reactor year. 'Ihis frequency is
| |
| }
| |
| 5-6 a _ _ _ _ _ _ _ __
| |
| | |
| i i
| |
| 4 4 based on EPRI report NP-801'(21). Newer data bases have been established since this report was pub-
| |
| 'ished, namely EPRI reports NP 2230 (31) and NUREG/CR 3862. The events that make up this q category are, according to NUREG/CR-2728:
| |
| Categorv- Descriorion 1
| |
| Loss of RCS Flow (1 loop) 2 Uncontrolled Rod Withdrawal 3 CRDM Problems and/or Rod Drop 6 Iow Pressurizer Pressure
| |
| , 10 Containment Pressure Problems 14 Total Loss of RCS Flow 15 Loss or Reduction in Feedwater Flow (1 loop)
| |
| ,17 Full or Partial Closure of 1 MSIV (1 loop) 33 ,
| |
| Turbine Trip, Throttle Valve Closure, EHC Problems s
| |
| 34 Generator Trip of Generator-Caused Faults 37 Loss of Power to Necessary Plant Systems 38 Spurious Trips - Cause Unknown 39 Automatic Trip - No Transient Condition 23 Loss of Condensate Pumps (1 loop)
| |
| If these categories are used to determine the initiating event frequencies using NUREG/CR-3862, the following frequencies, are determined:
| |
| For all US PWRs Frequency = 2413 events /418.19 years = 5.77 For Westinghouse Frequency = 1573 events /274.63 years = 5.73 For Point Beach Frequency = 59 events /24.28 years = 2.43 '
| |
| Thus, the frequency used in the EPRI/WOG study is 2.4 for all other transient events per year, versus the 7.1 events per year used in the NRC Case Study.
| |
| loss of a Sinnie DC Bus (TS)
| |
| The EPRI/WOG study and the NRC Case Study both use the same frequency for the loss of one vital DC bus. However, as described below, the EPRI/WOG study includes recovery of the DC bus in its initiating event frequency.
| |
| 5-7
| |
| | |
| Recovery of Vital AC or DC Buses lanual cross-connecting of AC or DC buses is accornplished easily. As discussed in Appendix A, a recovery factor of 0.005 is used. Main feedwater recovery would also add increased margins for the T5 initiator. Using only the 0.005 factor, an unrecovered loss of a vital AC or DC bus becomes 3E-5/ year and 1.5E-5/ year, respectively. However, these events are not significant risk contributors, since AFW would have to fail as well. Also, the loss of a vital AC bus, the T4 initiator, was determined by WEP to not cause plant trip. For this reason, the initiator was removed from the EPRl/WOG study and this recovery action for the AC bus was not important.
| |
| Extemal Event Initiat2EE The extemal events considered significant enough to model were: (1) a break in a fire main in the ser-vice water pump room that spray damages all six SW pumps and the two fire pumps; (2) a fire in the auxiliary feedwater pump room that disables all turbine-driven and motor-driven pumps and the service water cabling that passes through the room; and (3) a fire in the 4160V switchgear room that disables all vital AC and DC power.
| |
| The service water pump house flood scenario requires a break to occur in one specific T-joint in the fire mam The NRC Case Study uses a generic roorn flood frequency for auxiliary building moderate oods. The EPRl/WOG study uses a frequency derived for the specific failure which causes the flood.
| |
| A correlation by Thomas Q2)is used that extends the notion in WASH-1400 that pipe break frequency is proportional to the length of pipe. This correlation is:
| |
| P, = (P /PL) * (Qp + A
| |
| * S
| |
| * Q,)
| |
| * BF
| |
| * P where P, = probability of break over the pipe length P/Pt = percent of breaks out ofleaks Qp = ratio of the product of the pipe diameter with length of pipe to the square of the pipe thickness A,S = factors related to weld quality Q. = same as Qp but for the welds BF = dynamic loading factor P = global pipe failure rate per Q Thomas suggests PdPL to be 0.06, A to be 50, S to be 1, and BF to be 2 for quality piping. The global falltue rate is 1E 8/yr/Q. The net result for 10"-diameter,1/2"-thick, and 3 ft. length of pipe is 5.93E-6/yr. Ifleats were sufficient to damage the pumps,i.e., the P/Pt factor were removed, the failure rate would be 9.88E-5/yr. Further, since no such leak has occurred in 16 years at Point Beach, a
| |
| 'ayesian update would reduce this frequency by 37%, yielding a failure rate of 3.73E-5/yr for a break, I
| |
| .ae value used in this EPRI/WOG study.
| |
| 58
| |
| | |
| Both the NRC Case Study and this review use fur frequencies deduced from industry wide data com-
| |
| : led by type of room and type of initiating fire, e.g., an auxiliary building and cable tray fire. This
| |
| >RI/WOG review has chosen an approach for developing fire frequencies based on the Limerick PRA (31), op: of the NRC Case Study's referenced " fire PRAs." Additional basis for fut frequencies can be found it section 6.
| |
| REVIEW OF OTHER SIGNIFICANT EVENTS Three events that are not initiators but were not modeled in detail by the NRC Case Study are: X, the failure of depressurization following the loss of high-pressure injection; Q, the failure of the pressurizer relief valves to close given that they are open; and block valve status.
| |
| For event X, the only difference between EPRIsWOG and the NRC Case Study is the human reliability analysis portion. The EPRI/WOG HRA recommended 3.0E-3 as the value to use in the derivation of the probability of X. As a result, EPRI/WOG calculated a value for X of 3.8E-3 for all events other than loss of offsite power, versus 1.5E-2 used in the NRC Case Study.
| |
| Event Q is a spurious PORV opening which is not isolated, or 3 failure of the pressurizer safety relief valves (SRVs) to close given that they are demanded to open. Evt.nt Q's conditional probability was stermined as (7E-2) * (2E-2) = 1.4E-3. One portion of that NRC Case Study value,7E-2, is derived {
| |
| .com experience stated to show that PORVs may inadvertently open though not required. The data base for this experience is neither provided nor referenced as part of the NRC Case Study documentation, and thus could not be checked.
| |
| However, a survey of Westinghouse plants was conducted in 1981 (12) to determine the operational data for pressurizer PORVs and safety valves in order to respond to NUREG-0737 (4Q), Item II.K.3.2 (WCAP 9804, "Probabilistic Analysis and Operational Data in Response to NUREG-0737 Item II.K.3.2 !
| |
| for Westinghouse NSSS Plants," D.C. Wood and C.L. Gottshall, February 1981). No failures of PORVs i 1
| |
| were reported at domestic Westinghouse plants. This included 163 operational openings of PORVs.
| |
| l The report also stated that no operational openings, and therefore no failures of safety valves, have occurred at Westinghouse plants.
| |
| As discussed in section 4, the EPRI/WOG study assumes that both PORVs will open for loss of offsite power, loss of feedwater, loss of an AC bus and loss of a DC bus. Westinghouse thermal hydraulic cal-i culations show that reactor and turbine trips will not result in PORVs opening. One percent of the time l a PORV will stick open (SE 3 for each PORV). If the PORV sticks open, the block valve must be alosed to terminate the I.OCA. The block valve failure probability (frorn the NRC Case Study data I .se)is 8E-3. Operator failure to close the valve within 30 minutes is IE-3. (See section 4.) The l
| |
| l l 5-9
| |
| | |
| I transient induced LOCA conditional probability then would be 9E-5, i.e. considerably less than the -
| |
| TRC Case Study value of 1.4E-3.
| |
| However, power will not always be available to the block valve. A station blackout could occur, in-which case the block valve cannot be closed. The transient induced LOCA probability then is 1E-2 given a station blackout. The core-damage frequency for this case is much lower since the probability of a station blackout is low. The effect of these changes to transient-induced LOCA modeling is signif-icant. First, the four dominant internal event sequences in the NRC Case Study are either impossible or l insignificant. Sequences T3QH1 'H2' end T3QD1D2, with frequencies of 2.5E-5 and 4.6E-6, cannot occur. Sequences T2MQH1'H2' and T2MQD1D2, with frequencies of 3.5E-6 and 6.6E-7, will be reduc-ed by a factor of 16, i.e., their sum will be less than 3E-7. For station blackout events, considering these additionr1 sequences increases the EPR]/WOG station blackout core-damage frequency by about 25%.
| |
| With regard to the block valves, plant operators indicated that only roughly 1% of the time were bath block valves closed. Roughly 10-50% uf the time, one valve was closed and the remaining time no block valves were closed.
| |
| Cut sets indicating failure of only one block valve were reduced by a factor of 4 (50% chance one valve was closed and a one in two chance it was that valve). Cut sets with failure of both block valves weic reduced by a factor of two (50% chance that a block valve was closed).
| |
| REVIEW OF COMMON-CAUSE DATA The basic e<ent probabilities for equipment were quantified in the NRC Case Study using the IREP Pro-cedures Guide data from EG&G (M). This data bank is a reasonable choice for generic purposes, and Point Beach data fo. independent failures do not indicate any reason to challenge the generic data.
| |
| However, the data to be applied to common-cause (dependent) events are derived from component data and so-called beta factors. The process used in the NRC Case Study can be challenged by reviewing the basis for generating the beta factors.
| |
| 'Ihe common cause events and their updated values are listed in Table 5-2.
| |
| 'Ihe NRC Case Study methodology omits two very significant considerations in developing beta factors.
| |
| First, the study faih to use a plant specific design review to determine which failure modes from generic data are applicable. Second, the study does not adequately address recovery analysis.
| |
| 3efore considering these omissions, the NRC Case Study can be compared to other studies. Significult differences and similarities are noted below. Although the NRC Case Study method for applying beta factors could be challenged, it was decided for this study simply to choose values consistent with other
| |
| , 5-10
| |
| _ _ _ _ _ _ _ _ _ _ _ _ _ _ a
| |
| | |
| T' Table 5-2 1 Common-Cause Failure Rates NRC Case Study EPR1/WOG ,
| |
| Event Description Probability Probability '
| |
| i BATT-CM Dependent failure of all batteries 9.6E-4 4.0E-4 AFWP-CM Dep. failure of both MDPs and TDP 2.0E-4 3.0E 5 SWSP-CM Dep. failure both running SWPs and 2.0E-5 4.0E-6 one standby CCWP-CM Dep. failure of running and standby 8.0E-5 3.0 5 5
| |
| .of CCW pumps LPI, HPI, CS-CM Dep. failure of other pumps 1.0E-4 3.0E-5 MOV-CM Dep. failure of MOVs 4.0E-4 8.0E-5 DG CM Dep. failure of diesel generators 1.5E-3 5.0E-4 a studies. In this manner, the EPRI/WOG measure of the significance of common-cause failures would be consistent with the other studies' measures of the significance of common-cause failures.
| |
| The common cause failure probability used for the diesel generators in the NRC Case Study was 1.5E-3.
| |
| This was based on the local faults of the diesel generators multiplied by a common-cause factor. The local faults were made up of failure to start and failure to run probabilities. His information was based on NUREG/CR 1032. The common-cause failure rate parameter estimates in Table B.3 of that NUREG show that the probability of failure of two diesel generators due to common cause is 5.0E-4 using the Multiple Greek I.etter (MGL) method, and 7.1E-4 using the Binomial Failure Rate (BFR) method from NUREG/CR-2989. The failure rate for diesel generators due to common cause used in the Millstone-3 PRA (based on the BFR method) was 2.59E-4. Rus, the common-cause probability used in the NRC Case Study is two to three times larger than the other values. This EPRI/WOG study uses 5.0E-4 as a reasonable mean value of the altematives. De NRC Case Study value is high also if the failure to run probability is considered excessively high. (See the discussion on common-cause diesel run failures in EPRI report NP-3967 (21), Section 4.)
| |
| The common-cause failure rate used in the NRC Case Study for motor-operated valves is 4E-4 using the a factor method. The failure rate for common-cause faults used in the Millstone PRA study (21)is EE-5. This is based on the BFR method. His is roughly a factor of 5 lower than the value used in the 5-11
| |
| | |
| t l
| |
| NRC Case Study. The Millstone 3 value will be used in this study. (The Millstone-3 value can also be l itained by applying the same factors that are used in the pump analysis described subsequently.)
| |
| i The common-cause failure rate used for batteries is based on NUREG-0666; the value for a two battery
| |
| . system is calculated to be 9.6E-4. Backup by additional batteries is considered sepantely, since they perform different functions.
| |
| J A comparison was made between the common-cause failure rates used for the various pumps in the NRC Case Study and the Millstone-3 PRA study. Table 5-3 shows the cornparisor..
| |
| Table 5-3 Common. Mode Failure Rates (NRC Case Study and Millstone-3)
| |
| Typs NRC Case Study Millstene 3 AFWS 2E-4 1.49E-4 SWS 2E 5 2.8EE 5 LP IE-4 1.4E-4 HP IE-4 1.43E-4 CCW 8E-5 NOT ANALYZED CS lE-4 NOT ANALYZED As can be seen from the comparison, the values are in relatively good agreement.
| |
| Cmis.on-Cause Desien Review
| |
| , EPRI report NP-3967 encourages common-cause failure analysis to employ a design review when deter-mining beta factors. 'Ihis design review process has yielded results of a factor of two lower in EPRI's application than beta factors calculated using all industry experience,i.e., as was done in the NRC Case Study.
| |
| As a supponing example, this study reviewed a prior study of auxiliary feedwater reliability performed 4
| |
| by Northern States Power for their Prairie Island plant (_42). A review of industry and plant data was 5-12
| |
| | |
| perfortned in that study to assess common-cause failure probabilities. The review identified the follow-ig probabilities for pump and valve common-cause failures:
| |
| Pumps 2.3E-5 Valves 4.3E 5 These values are approximately a factor of 10 lower than the NRC Case Study estimates.
| |
| l In considering pump failures, a factor of 3 reduction is used as a compromise between the factors of 2 and 10 reported in the tv.o design reviews. It is reasonable to assume that a design review would reduce common-cause probabilities, since no common-cause events have occurred at Point Beach.
| |
| Common-Cause Failure Recoverv To account for recovery from common-cause failures, AEOD repon C504, entitled " Case Study: Report on Loss of Safety System Function Events" (29), was reviewed to determine the percentage of events which were recovered in a~ timely manner. The review indicated that roughly 60% of both human and hardware failures were recovered within one hour. This factor was applied in this EPRl/WOG study to generate the probabilities for dependent failures reported in table 5 2.
| |
| , The NRC Case Study method for determuung component data supports the assessment of 60% recover-able. The majority of pump and valve failures in the NRC Case Study data base were control signal failures. Since these control signal common-cause faults would be recovemble within one hour, the 60% reduction was judged to be reasonable and consistent with the NRC Case Study data base.
| |
| L l
| |
| \
| |
| l 5-13 ,
| |
| l
| |
| | |
| Section 6 EXTERNALEVENTS ANALYSIS INTRODUCTION The Point Beach NRC Case Study performed analyses which identified three externally initiated acci-dent types with frequencies greater than IE-5: seismic, fire, and intemal flood. The EPRI/WOG reanal-ysis of these events could not be cornprehensive in the sense of identifying other scenarios. For example, while the EPR]/WOG study concluded that the NRC Case Study fire scenarios were not risk significant when reanalyzed, it did not attempt to determine if other scenarios were significant. Further-more, while the review could conclude that the NRC seismic analysis did not account for other recovery actions, it .was impossible to assure that these recovery actions would be possible without re-exercising the codes used in the Case Study. We believe that these recovery actions are appropriate and have esti-mated new core melt fmquencies accordingly. On the other hand, new seismic fragilities were not determined and integrated into the model for the EPRI/WOG reanalysis.
| |
| This section provides an overview of the NRC Case Study approach and a discussion of the EPRl/WOG reanalysis. The principal differences between the two studies are given below:
| |
| o Transient combustible fire frequency is lower in the EPRI/WOG study by a factor of close to 10.-
| |
| o The EPRI/WOG study includes cable tray and electrical parel fires, while the NRC Case Study does not.
| |
| o Conditional probabilities reflecting the fire's exact required location were not included in the NRC Case Study, but were treated in the EPRI/WOG study.
| |
| o The cleanliness of the plant and the specific combustible control procedures are not reflected in the NRC Case Study analysis.
| |
| o Manual operation of the turbine-driven pump is considered possible in the EPRI/
| |
| WOG study for the switchgear room fire scenario, and more realistically esti-mated for the AFW pump room scenario.
| |
| o The internal flood risk given in the NRC Case Study scenario is essentially elimi-nated.
| |
| o Seismic risk sec.nario recoveries are accounted forin the EPRI/WOG analysis o The new batteries considend in the EPRI/WOG study significantly change seis-mic and switchgear room fire scenarios.
| |
| a
| |
| | |
| o In the NRC Case Study tank failuies are conservatively treated in terms of failure mode, and possibly in terms of failure probability.
| |
| o Small LOCA frequency for seismic events may be conservative; at a minimum, modeling should consider two size ranges.
| |
| o The EPRI/WOG seismic hazard curve reduces seismic risk from low acceleration earthquakes by a factor of two, and seismic risk from large acceleration canh-quakes by a factor of five.
| |
| The significance of each individual finding is presented in table 6-1. De final sequence frequencies are presented in section 8.- The results of these findings reduce both seismic and fire core-melt frequencies to well below the NRC Case Study estimate and internal flood risk is essentially eliminated.
| |
| FIRE ANALYSIS Because this EPRI/WOG study reanalyzed the fire sequences presented in the NRC Case Study,it is useful to describe that analysis first. The following provides an overview of the NRC Case Study analysis and a description of the EPRI/WOG analysis, as well as a comparison of the two effons.
| |
| Overview of NRC Case Study Fire Analysis
| |
| ''he approach that was used in the NRC Case Study is reponed to be a "... scaled down version of the imerick (31), Millstone 3 (23), and Seabrook (15)) PRAs". As with the EPRI/WOG reanalysis report-ed here, a comprehensive fire analysis was beyond the scope of the NRC Case Study. Basically, the NRC Case Study performs the following steps to detennine fire induced core-melt frequencies:
| |
| : 1. Survey of critical areas to detennine scenarios where fire damages all safety sys-tems or all but a single component, e.g., turbine driven auxiliary feedwater pump. 1
| |
| : 2. Assessment of key fire locations within the screened critical areas.
| |
| : 3. Calculation using COMPBURN (43) to determine threat fmm two fire types and associated times for manual suppression and human action.
| |
| : 4. Quantification of accidents using fire frequency data, manual suppression data, and automatic suppression system failure experience reponed in other fue PRAs, together with component and humt.n ermrs.
| |
| Exercising these steps lead to the creation of two significant fire scenarios, namely a fire in the switch-gear room and a fue in the auxiliary feedwater pump room. Rese two scenarios suggested fire core-melt sequences on the order of 1.0E-5 for each case.
| |
| t I
| |
| 6-2
| |
| | |
| Table 6-1 Extemal Event Changes and Their Individual Impact on NRC Case Study Core-Melt Frequency Event Probability Reduction
| |
| * in NRC Case EPRI/WOG Core-Melt Frequency Model Parameter Study Value Study Value (oer reactor year)
| |
| Fire in AFW Pump Room 3.38E-3 6.5E-4 9.5E-6 Fire in Switchgear Room 4.96E-3 3.5E-3 5.9E-6 Conditional Probability 1.0 0.3 9.1E-6 AFW Room Fire Location AFW Pump Room Fire: 0.1 0.03 9.1E-6 Ma .aal AFW Pump Actuation All Fires: 0.2 0.06 2.7E-5 !
| |
| Halon System Reliability {
| |
| AFW Pump Room Fire: Not Modeled Modeled 1.0E 5 Extra Halon System I
| |
| Switchgear Room Fire: 1.0 0.03 1.9E 5 Manual AFW Pump Actuation j SW Pump Room Flood Frequency 2.20E-3 3.73E 5 7.6E-5 CST Altemative (SWS)- (Not Credited) Credited <3 SSE) 1.5E-5 Seismic New (Seismic Grade) (Not Credited) (Credited <3 SSE) 3.1E 5 Baneries RWST Altemative (CVCS)- (Not Credited) (Credited <3 SSE) 1.5E 5 Seismic
| |
| * Reductions assume all other prameters remain the same.
| |
| 6-3 1 :
| |
| : l. .. . . . . . . . . . . . . . _ . . . . . . . . . _ . . . . . . . . . . . . . . . . . . . .. . . ..
| |
| | |
| l I
| |
| EPRI/WOG Reanalysis and Comparison to NRC Case Study
| |
| .e EPRI/WOG fire analysis shares some of the same limitations of the NRC Case Study analysis in that neither is a comprehensive fire analysis. The EPRI/WOG effort is also limited in the sense that specific fire growth calculations (e.g., using COMPBURN) were not performed. This limitation is important because the EPRI/WOG study found that the NRC Case Study did not analyze the risk-dominant fire-initiated accident sequences. Because the EPPJ/WOG study did not perform fire growth calculations using COMPBURN,its ability to produce alternative risk estimates for the scenarios not considered in the NRC Case Study is limited. Because of its limitations, the EPRI/WOG results are most appmpriate as an indication of the degree of conservatism in the NRC Case Study, rather than as an independent assessment of fire risk.
| |
| The NRC Case Study scenarios focused on transient combustible fires. The EPRI/WOG analysis dem-l onstrates (as described below) that transient combustible fires generally are not dominant fire scenarios.
| |
| l Other PRAs, including the " fire PRAs" referenced by the NRC Case Study (li,23,31), conclude that fins initiating in cable trays or electrical panels are those which lead to dominant risk fire scenarios.
| |
| l The EPRI/WOG analysis provides its interpretation of the same fire PRAs referenced in the NRC Case Study and their implication to fire-initiated accident risk at Point Beach. The EPRI/WOG interpretation dicates that the NRC Case Study includes important conservatism which are not contained in the referenced " fire PRAs". Also, differences in human reliability analysis and,in particular, success criter-ia and plant models are incorporated into the EPRI/WOG reanalysis, which concludes that fire risk at Point Beach is much lower than that indicated in the NRC Case Study.
| |
| Figure 6-1 illustrates the elements of the EPRI/WOG fire analysis after the critical fire areas are select-ed. The following elements illustrated by the event tree figure are discussed below:
| |
| o The type and the frequency of the fire in the room.
| |
| o if applicable, the probability that the fire is in the necessary adverse location. j 1
| |
| o The probability that automatic fire suppression fails.
| |
| o The probability that manual fire suppression fails.
| |
| o The probability that the operators fail to manually initiate the auxiliary feedwater (AFW) turbine driven pump.
| |
| o The probability that the AFW turbine-driven pump is unavailable. ;
| |
| o The probability that the charging system is unavailable.
| |
| ne EPRI/WOG study considers three types of fire initiators for the auxiliary feedwater pump room and the switchgear room. These initiators are: l l
| |
| 6-4 i l
| |
| | |
| I a
| |
| r o
| |
| et _
| |
| ca nn og LD _.
| |
| ui NMN AA A A gs ee CCCCC SD FF P FFF e
| |
| c i .
| |
| r ee e e
| |
| us qo MUMM r el 4c;MWM l
| |
| i N ;;
| |
| : C T 9C t n
| |
| e v
| |
| E E
| |
| I L i s
| |
| S. s WPLI y W
| |
| F I NDL A l a
| |
| AT PF i' n A
| |
| RDP O e r
| |
| OTAW O i
| |
| F T F ASYA y RLR 4 d EI A 9 u .
| |
| P OFS# AT E t S
| |
| G B
| |
| O LE N A
| |
| 3 U
| |
| P N WI
| |
| /
| |
| t f
| |
| A L IPO R MSS P E
| |
| C .
| |
| I - 1 T5 -
| |
| AE A 6 MU e OPN r T FOI u -
| |
| UL g A SS i F
| |
| N t N Op t C TW AI INI M
| |
| I F
| |
| C ON LI t
| |
| p X F Eu F
| |
| f I N FI g*
| |
| h
| |
| - l!! !
| |
| | |
| o Fire starting in a cable tray, o Fire starting in an electrical panel, and o Fire staning in a source of transient combustibles.
| |
| The frequencies for the first two initiators were obtained based on the methodology and data presented in the Limerick fire analysis (18.), which repons that nuclear plant experience indicates that cable tray fires have a frequency of about 5.3E-3 per year per auxiliary building. The likelihood of a cable tray fire is calculated for a panicular room by multiplying by the ratio of the amount of cabling in that room and the total amount of cabling in the auxiliary building. Using the ratios provided in the NRC Case Study yields the following initiating event frequencies:
| |
| Cable tray fire in the AFW pump room 2.6E-4 Cable tray fire in the switchgear room 5.5E-4 The frequencies for fires starting in electrical panels are obtained in the Limerick study by taking the total number of such fires in a plant and assuming 40 panels per plant. The resulting electrical panel fire frequency is 2.2E-4 per panel. There are two panels in the auxiliary feedwater pump room. One is an instrumentation panel with low voltage in the turbine-driven pump cubicle. The other is a panel with pical voltage; however, this panel is located in the end cubicle in the room. Both areas where the panels are located were not specifically identified as adverse in the NRC Case Study. The EPRl/WOG study conservatively assumes that either panel in the AFW pump room can initiate the fire scenario in the NRC Case Study. A more detailed analysis may conclude that fires in these two panels are either much less likely to occur (due to lower voltage) or much less likely to fail redundant trains of safety equipment (due to the location). There are 13 panels in the switchgear room. (These numbers of p9els are similar for the corresponding fire areas in the Limerick plant.) The initiating event frequencies ased in the EPRI/WOG study are-i l
| |
| Electrical panel fire in the AFW pump room 4.4E-4 '
| |
| Electrical panel fire in the switchgear room 2.9E-3 Fire scenarios initiated by a transient combustible fire have not been dominant contributors for many PRAs, including the Limerick (33) and Millstone (21) studies referenced in the Case Study. Those l PRAs specifically noted the lack of experience in the nuclear industry with transient combustible fires of the magnitude analyzed by the Case Study,i.e., a ten gallon acetone fire. Correspondingly, the Limerick and Millstone studies assessed a very low initiating event frequency for this type of fire. l l
| |
| l 6-6 t _ _ - - - - _ -
| |
| | |
| l l
| |
| The Limerick fire analysis reviewed plant fire data and found that no large trar.sient combustible fires had occurred (they would certainly have been recorded due to theirintensity). Conservatively, the anal-
| |
| )
| |
| ysis assumed one such occurrence to determine a frequency for the plant. (This is a standard analysis technique when no failures have occurred.) ne Limerick study then determined a transient combustible fire initiator fr quency by apportioning the probability equally among the various fire areas in the plar,t.
| |
| Using the same methodology for Point Beach, one can obtain the given result, as follows. First, there have been 666 compartment years of auxiliary building experience, and roughly the same for the switch-gear room. Assuming that one fire has occurred yields a frequency of about 1.5E-3 per year. The Point Beach plant has nine fire areas in the auxiliary building and two switchgear rooms. The frequencies for a large transient combustible fire are:
| |
| o Auxiliary building AFW pump room 1.7E-4 per area-year o Switchgear room 7.5E-4 per area-year i
| |
| These ficquencies are much smaller than the ininatirg event frequencies used in the NRC Case Study, which are:
| |
| o Auxiliary building AFW pump room 2.38E-3 per room-year o Switchgear room 4.96E-3 per room-year It is clearly difficult to assess a transient combustible frequency from industry experience. Point Beach experience and practices lead one to conclude that the transient combustible fire frequency would be lower than the industry average. The plant is kept noticeably clean, specific procedures are followed, and walkdowns are performed to ensure that such fires do not occur. Each fire area at Point Beach is checked three times each day for transient combustibles or other potential fire problems. Further, transi-ent combustibles must be signed out and in during use. After sign-in a check is made by the contractor or maintenance person, as well as a supervisor, to ensure that no combustibles remain in the area.
| |
| 1 Given that a transient combustible fire occurs in a room, the EPRI/WOG study estimates the conditional l 1
| |
| probability that the fire would be in the right location. He NRC's AFW pump room fire scenario calcu- l lation with COMPBURN indicates that cables will be ignited by the transient combustible after about )
| |
| four minutes; however, the NRC Case Study indicated that the acetone wou'l bum approximately three I
| |
| minutes at peak intensity, Based on this information, the EPRI/WOG study judged that the probability of ignition is only marginal. Furthermore,if the fire is not di"ectly below the cables,it would be unlike-j ly to cause sufficient damage to destroy all the cables in the room. Based on a walkdown of the room, l : EPRI/WOG studyjudged that a fire starting in a th'.rd to a fifth of the area of the rtom only would be sufficient to ignite the cables. Similar weighting factors were used in both the referenced Limerick and 6-7
| |
| | |
| Seabrook PRAs, i.e.,0.25 and 0.1, respectively for dominant sequences. Hence, this review uses a 0.3
| |
| :onditional probability that a transient combustible fire in the AFW pump room will be sufficient to damage all the required cables. I In the switchgear room fire analysis, the NRC Case Study did not indicate either the assumed location for the COMPBURN calculations, the timing ofignition versus duration of the fire, or a descripion of the sensitivity of the scenario location. Hence, it is difficult to establish a conditional probability for the switchgear room. 'Iherefore, the EPRI/WOG study notes but does not quantify this conservative bias.
| |
| No such factors were applied for the other initiators in either room.
| |
| Tne EPRI/WOG study used the basis provided in the NRC Case Study for automatic suppression. For manual suppression, however, the EPRI/WOG study notes that two of the three fire analyses (i.e.,
| |
| Millstone-3 and Limerick) find higher reliability for manual suppression thin does the NRC Case Study.
| |
| In the Millstone 3 analysis a logic model (figure 6-2) was u,ed to quantify the various means by which manual suppression could fail. Applying this logic model to Point Beach yields the following results:
| |
| Failure of manual suppression for AFW pump room 0.15 Failure of manual suppression for switchgear room 0.024 The Limerick study used the same historical data referenced in the NRC Case Study but derived different values as a function of time. The Limerick study results were as follows:
| |
| Failure of manual suppression in 25 minutes (AFW) 0.15 Failure of manual suppression in 60 minutes (switchgear) 0.04 This EPRI/WOG study uses the slightly more conservative Limerick values.
| |
| Manual suppression is credited both in this study and in the NRC Case Study but questions are raised in the NRC Case Study about whether water spray would fail additional components. In the switchgear room, non vital AC buses could be used if vital buses were damaged. In the AFW pump room the turbine driven AFW pump is far enough fmm the motor-driven pump so that spray should not be a problem.
| |
| Other differences between the two studies involve the modeling of human reliability. The EPRI/WOG study analyzes manual operation of the turbine-driven AFW pump consistent with the HRA discussed in section 4. A factor of 0.03 is used for operator failure (see discussion in section 4 and Appendix A).
| |
| This value accounts for the operators being required to open a steam admission valve in another area.
| |
| 6-8
| |
| | |
| H T
| |
| W VN O
| |
| e,
| |
| , ET I WA FT S S E
| |
| SE ES RO PH P
| |
| E 2 S U O S f 0 H 0
| |
| HN TO 1
| |
| 0 r .WHITA 0 E
| |
| o T 6
| |
| ES U
| |
| 0
| |
| )
| |
| E 2 L E S
| |
| r L 1 A F O o B .
| |
| H A 0 1 l T E I OR SN I
| |
| TOP t(S AS F
| |
| M
| |
| %rr 1 0
| |
| I UO OA T T K
| |
| I T
| |
| S f
| |
| P -
| |
| P E t U O t E S R A 4 C 1 F 2 H 0 T W
| |
| V r
| |
| E R
| |
| o R U
| |
| 5 E
| |
| A s n l T E 1 e i n et unw F
| |
| C R EF I
| |
| . F T O o mnI E S DS ititimn co e e
| |
| OM ' t n53 ka TP dge t i P nt e E
| |
| R U U S 1 dimra-n aa va s LO AP 4
| |
| E eme n l. ot i FA N l
| |
| l o oi A
| |
| O.
| |
| 1 it mri l S
| |
| MN UO ia o pa t
| |
| o b l N T I i r n euo mb
| |
| , C i h prp EE B T e
| |
| r t 2
| |
| U E LD ihWr f
| |
| c o M
| |
| 2 A naFr r 0 F eeAe er en f T
| |
| C O w oh a t t E e s n umt T
| |
| E bes t i l T D r te n u nI I
| |
| O T
| |
| r i ui o .
| |
| nmtitam m5so o l
| |
| L F
| |
| 1 1
| |
| A 0 -
| |
| 0l er s r F E 1 aoa e
| |
| E M
| |
| J s e o nhc g 6 2
| |
| OR AO miheh udt it t
| |
| c ST sdfi C s aow3 E a -
| |
| NT yan se e sA LFK uhle L l A adrtb t
| |
| nona F Safi t
| |
| ~
| |
| e t
| |
| o N
| |
| | |
| in the switchgear room analysis, manual operation of the turbine-driven auxiliary feedwater pump is
| |
| .so credited in the EPRI/WOG study. At the time of the original NRC Case Study, the new batteries, which power certain instrutnentation in the event of loss of normal station batteries, had not been installed. This instrumentation permits manual operation of the turbine-driven AFW pump during a loss of switchgear room.
| |
| Finally, the NRC Case Study and EPRIAVOG study differ in their assessment of Halon system effectiveness. First, the AFW pump room and switchgear room share the same redundant (two train)
| |
| Halon systems. The NRC Case Study analysis of the AFW pump room fire credited only one Halon system train. Second, Halon system unreliability used in the NRC Case Study does not reflect recently reported experience. Data collected by the Department of Energy (DOE) found no Halon system failures in 17 fires (44). EPRI/WOG used an estimate of 1/17, or about 0.06, for Halon system unreliability versus 0.2 in the NRC Case Study.
| |
| The conservatism in the NRC Case Study not analyzed in the EPRI/WOG fire analysis included consideration of the effects of early reactor trip. Each of the fire scenarios would occur for a reasonable period of time before complete system failure occurs. If the plant is tripped upon identification of a severe fire, or initial ignition of cables cause plant trip, plant shutdown would have begun prior to full re damage. The steam-driven auxiliary feedwater pump may have started, thus negating the difficulty of manually opening the DC steam admission valve. The motor-driven AFW pumps would also run for some period of time. More importantly, however, steam generator dryout will be significantly delayed due to the heat removed prior to main feedwater failure. This delay will be sufficient to allow two of three charging pumps (versus 3 of 3) to cool the core. Since the CVCS does not require service water or component cooling water, its reliability will be due only to CVCS equipment unavailability, or 4E-2.
| |
| (The EPRI/WOG study assumes that if the operators fail to initiate AFW, they will fail to initiating charging; hence, CVCS would only be considered in the event of AFW equipment failure.) This addi-tional safety margin was not included in the EPRI/WOG study because it could not conclude whether CVCS would be damaged by the fire. In the case of th'e switchgear room, CVCS cabling was in the floor and its status could not be verified for the conditions given.
| |
| Sections 8 and 10 present the core melt and value-impact analysis results for fire scenarios and their associated modifications, incorporating the above comments in a new quantification.
| |
| INTERNAL FLOOD ANALYSIS As documented in section 5, the initiating event frequency for the internal flood scenarios differ signif- ,
| |
| antly between the EPRIAVOG and NRC studies. Reducing the initiator frequency to the EPRl/WOG {
| |
| value eliminates this scenario as a concern. Because CCW is not required for HPI, the scenario modeled 6-10
| |
| | |
| l
| |
| 'yould not lead to immediate core melting anyway since feed and bleed would be a viable option until l circulation. Auxiliary feedwater recovery or charging pumps will prevent core melting at that time.
| |
| SE1SMIC ANALYSIS Because the EPRI/WOG study reanalyzed the seismic sequences presented in the NRC Case Study, it is useful to describe the NRC analysis before describing the EPRl/WOG analysis. De following provides an ovrview of the NRC Case Study Analysis and then a description of the differences between the EPRI/WOG analysis and the NRC Case Study.
| |
| Overview of the NRC Case Study Seismic Analysis The NRC Case Study seismic analysis uscs techniques for seismic risk assessment which were devel-oped during the Seismic Safety Margins Research Program (SSMRP)(41). A site specific hazard curve was developed. Component fragilities were either ccimponent specific or generic. Component specific fragilities were developed if the site visit indicated some potential vulnerability together with perceived importance to the plant logic model. These important components included battery racks, AC and DC buses, and tanks, he resulting model was quantified and results were presented in terms of core melt risk as a function of multiples of the SSE. Cut sets were also presented (but without numbers) for the iur dominant accident types.
| |
| The dominam accident sequences included two transient and two small LOCA sequences. The two tran-sient sequences both included failure of all feedwater plus failure to feed and-bleed. The scenarios were initiated by either of two transients (i.e., with main feedwater initially available or with main feedwher initially unavailable). All seismic events are assumed to result in a transient of some kind unless a LOCA occurs. The two small LOCA sequences both include failure of HPSI. The more likely sequence included successful auxiliary feedwater and plant depressurization, but failure of LPI to suc-ceed in cooling the core. The other sequence included auxiliary feedwater failure, so the LPI option is not available.
| |
| Three of the four contributors are dominated by two scenarios:
| |
| o Failure of both CSTs and the RWST o Station blackout caused by loss of offsite power and loss of the battery racks i
| |
| I The other contributor, the small LOCA without HPSI but with AFW success, is caused by failure of the l RWST. !
| |
| /
| |
| 6-11
| |
| | |
| J l
| |
| l I
| |
| For those sequences that included station blackout, containment safeguards were lost. Sequences with l
| |
| tank failures resulted in containment spray failure but fan cooler success. These containment safeguards j states were integrated into the consequence analysis. l l
| |
| )
| |
| FfRI/WOG Reanalysis I
| |
| (
| |
| There are six principal differences between the NRC Case Study and the seismic risk reanalysis per-formed by EPRI/WOG:
| |
| : 1. The consideration of new Seismic Category I batter.es in the EPRI/WOG study significantly changes the seismic sequences.
| |
| : 2. Tank failures are conservatively treated in terms of failure mode and possibly in terms of failure probability in the NRC Case Study.
| |
| : 3. The seismic hazard curve is conservative by a factor of two for low acceleration earthquakes, and by a factor of five for high acceleration earthquakes.
| |
| : 4. The NRC's Case Study small break LOCA frequency may be conservative; as a minimum its modeling should be considered in two size ranges, as in the EPRl/
| |
| WOG stuc'y.
| |
| : 5. Failure of one CST rather than two as in the NRC Ca:;e Study will cause loss of inventory because the tanks are not separated by check valves.
| |
| : 6. The NRC Case Study seismic analysis does not consider AFW recoveries which can be implemented from the control room.
| |
| Each of these issues is diffictCt to incorporate in terms of bottom line estimates. The seismic analysis uses a convolution of probability distributions which is somewhat difficult to reproduce without using the model to recalculate. Also, some of the comments above identify the need to consider the seismic j fragility of other components. For this reason, changes in seismic risk are estimated more qualitatively )
| |
| than in other areas of the study. The following describes each of the above comtnents and its implica-tion to plant modeling differences in the EPRl/WOG study.
| |
| New Batteries The new batteries provide a seismic grade capability for starting the diesel generators. Roughly 40% of the seismic core-melt frequency in the NRC Case Study was from station blackout events induced b" battery or DC bus failures. 'Ihese battery and bus failures would not prevent diesel starting with the new !
| |
| batteries in place, although difficulties exist because service water would have to be manually initiated l shonly after the diesels were started As described in Appendix A, a human error of 0.3 was applied by EPRI/WOG to these scenarios to account for operator failure to establish service water to the diesels before they failed from overheating.
| |
| 6-12
| |
| | |
| RWST Failure Mode and Canacitv he RWST is a significant weak point in Point Beach's seismic response according to the NRC Case Study. The RWST was found by the NRC Case Study to have low seismic capacity and high uncer-tainty because ofits high aspect ratio, both of which contribute to it being involved in about 60c7c of the seismic core-melt frequency. As in most seismic analyses, methods depend significantly on estimates of design strength. A limited review of actual tank earthquake experience was tabulated in a letter to VEP j from EQE Incorporated (4fi), the pnmary consultant to the Seismic Qualification Utility Group (SQUG). I l
| |
| This limited review and a more in-depth review performed by EQE for the Electric Pow:r Research i I
| |
| Institute (EPRI), have found that a catastrophic failure of anchored tanks (such as an RWST), with rapid i loss of contents, is improbable. That is, some period of time (e.g.,30 minutes) would be available for operators to establish other flow paths, since the tanks' function would not be lost immediately.
| |
| Seismically qualified water level indication in the RWST would support diagnosis of the need for recovery actions.
| |
| The operators are well aware of the seismic grade alternatives to the RWST. The Boric Acid Storage Tanks (B ASTs), two reactor makeup water storage tanks with 100,000 gallons of water each, and other portions of the CVCS provide some additionalinventory for immediate use. In the intermediate term, the spent fuel pool can be aligned for use to provide flow rates in a range sufficient for decay heat
| |
| 'emoval or small small break LOCA leakage rates after a period of about 30 minutes. While difficult to establish in the confusion of a seismic even: sufficient to fail major equipment such as the RWST, these alternatives are nonetheless available and known to the operator. A factor of 0.1 was used in the quantification of this recovery as described in Appendix A. This factor considers both human and equipment failure but does not consider dependent seismic failure. It should be noted, however, that most equipment (i.e., pipes, valves, and altemative water supplies) are substantially stronger than the RWST at Point Beach, as demonstrated by the generic component frequencies reported in the SSMRP, the NRC Case Study report, and the ongoing SQUG analysis. The factor is applied only at accelerations less than 3 times the SSE.
| |
| Seismic Hazard Curve The seismic hazani curve developed for the NRC Case Study uses techniques developed during the SSMRP program that have been subsequently modified. As part of its nuclear safety program, EPRI has performed extensive reviews of similar seismic hazard curves. EPRI has developed altemate means for preparing site specific seismic hazard curves. Discussion with EPRI personnel (4.1) indicates that an industry review would tend to yield lower hazard curves and hwer uncertainties primarily due to differences in the modeling of attenuation. Mean values for hazard curves developed using EPRI's methodology would be likely to be lower by about a factor of two or more at low accelerations, and five 6-13
| |
| | |
| or more at high acceleration canhquakes. However, as was mentioned previously, it is difficult to eval-iate the significance of this difference without a full uncertainty analysis. This is particularly v dven the large uncertainty assessed for the RWST capacity. For these reasons, the EPRI/WOG study reduces seismic risk by two for less than three times the SSE and by five for greater than three times the SSE.
| |
| Small Break LOCA Frecuency and Modeling Small break LOCA frequencies for the NRC Case Study corne directly from the SSMRP literature.
| |
| They include two size ranges, a small break LOCA and a small-small break LOCA. The small-small break LOCA, from 0.5 to 1.5 inches, requires significantly less injection for makeup. Also, for this size break auxiliary feedwater can play a significant role in delaying core uncovery while means of recovery are pursued. Over 90% of the NRC Case Study small break LOCA frequency below three times the SSE is this size LOCA (<1.5 inches). This size was judged small enough for recovery by the altemate source of water for the RWST. In two-thirds of these scenarios, feedwater and the capability to 1 depressurize will be available to enhance heat removal and provide additional time and margin for recovery. This factor could affect about 1.0E-5 of the seismic hazard.
| |
| CST Failure and its Imoact on AFW A single CST failure (rather than two, as indicated in the NRC Case Study) will like?y fail the AFW system since there are no check valves to isolate a single tank frorn the other tank. As with the hazard curve, it is difficult to assess the exact impact of this finding on the model since both tanks failed in a manner where dependencies were undoubtedly involved. This modeling difference is not expected to change the results significantly because experience indicates the dependencies will likely dominate.
| |
| When the tank failure mode is considered (i.e., slow evacuation as with the RWST), failure of one CST will still allow a long time for recovery of AFW suction from the service water system. The service water system provides a seismic grade backup to CST failure which can be actuated from the control room (and which would be isolated by check valves from the leakmg tanks). As described in Appendix A, an alternative source of water to the CST from service water during a seismic event was given a 0.1 recovery factor. This factor is only applied to scenarios less than 3 times SSE. Since electric power problems occur with an increasing likelihood for increasing acceleration, electric power failures would hamper this recovery.
| |
| In summmy, this review has identified conservatism in the NRC Case Study seismic analysis. As with l other ponions of the plant model, assumptions and failure to consider all recovery actions significantly affects the results of the NRC Case Study. His review estimates the impact of those recovery actions on the seismic risk. The actual quantification of seismic risk changes is provided in the discussion in cetion 8.
| |
| l 1
| |
| 6-14 o___________ !
| |
| | |
| Section 7 ANALYSIS OF THE NRC'S DEDICATED SHUTDOWN DECAY HEAT REMOVAL SYSTEM The NRC Case Study describes and analyzes a conceptual arrangement for a dedicated shutdown decay heat removal system (SDHR) for Poin: Beach, the add-on emergency feedwater train and makeup train described in NUREG/CR-2883 (6). The EPRI/WOG review evaluates the effectiveness of this system for the dominant accident sequences described in this reanalysis. This section briefly describes the SDHR proposed for Point Beach, performs a fault analysis of the system fx significant sequences, performs a sensitivity analysis of potential adverse public risk impacts and then cornpares the SDHR to the existing hardware and procedures at Point ' Beach which perform similar functions but were not credited in the NRC Case Study analysis.
| |
| DESCRIPTION OF THE SDHR The following is the SDHR description provided on page 4-1 of the NRC Case Study report:
| |
| "The SDHR system is a backup system designed to provide emergency core cooling in the unlikely event of a failure of the existing safety systems during a SBLOCA or tran-sient. Core cooling is accomplished by injection of emergency feedwater into the steam generators and release of steam via dedicated atmospheric dump valves. Natural circula-tion in the reactor coolant loops ensures flow and heat removal in the core. The reactor coolant system pressure is mamtained above saturation pressure by the use of one group of pressurizer heaters in conjunction with the alternate makeup system. Offsite power will be used when available, otherwise the add on system is corr.pletely independent of the plant system. The system consists of a high pressure emergency feedwater type :
| |
| pump, a high pressure makeup pump, a storage tank for borated water, a storage tank for feedwater and the required piping, valving, instruments, and controls for initiation, moni-
| |
| )
| |
| toring and operation of the system. This system is automatically actuated given failure of j the primary systems. The structmes and connections to the existing system are designed i to Seismic Category I specifications. A complete description of this system can be found i in reference 11 (NUREG/CR-2883) and in the impact analysis (appendix J of the Case Study repon)." 1 A simplified diagram frorn NUREG/CR-2883 is included as figure 7-1. For simplicity, this repon will l
| |
| refer to the entire system as the SDHR system; and refer to its two trains of add-on components as l l
| |
| " Emergency Feedwater" (EFW-SDHR), and " Makeup" (makeup-SDHR)."
| |
| ( \
| |
| l 7-1
| |
| | |
| nS G2 3/
| |
| B
| |
| _ 35 89 2 -
| |
| - B R
| |
| Ce W
| |
| / r Gu
| |
| _ Eg oS Ri G1 Q
| |
| UF N
| |
| - e6 c1 Wf W f n-0 1
| |
| W5 ,
| |
| F i M
| |
| ~
| |
| e2 r
| |
| ee
| |
| - f g ea RP 4 O 5 O ht R
| |
| s p,
| |
| f '
| |
| 9 1
| |
| L 2 NL t
| |
| 5 O I 9 L 8 4 1 2 3
| |
| 8 7 2 1
| |
| 0 iI g 8 I ,g i g g g I l ,8 I a I 4 I 1 S I ' I I I 3 l ' I A 2 6
| |
| N 1 m 2
| |
| r 1
| |
| 6 5 2 1
| |
| 4 0 5 1 2 4 O 3 O p 1 L L
| |
| m u
| |
| ppm m U p p r g g ee k t0 p0 yMW2 aa0 Wm0 F u2 EP1 s
| |
| e 3 O c L
| |
| 2 O a 1 L f r
| |
| e t
| |
| n I
| |
| l a 2
| |
| $ 1 ' O n
| |
| g ~, l 1 - s T
| |
| a g
| |
| &r dm S0 de w0 At T0 s R 0, S0 ey T0 C 0, rS L2 1 T0 e A1 hg L
| |
| L0 A2 Wn i tt s
| |
| &O
| |
| _ \ ns g-
| |
| . ii
| |
| _ om
| |
| _ PE
| |
| _ e
| |
| - /
| |
| JL l
| |
| I
| |
| | |
| FAULT ANALYSIS OF THE SDHR FOR SIGNIFICANT ACCIDENT SEQUENCES ine NRC Case Study evaluated the SDHR system for faults under two types of conditions, namely with and without offsite power. The system was not evaluated for any extemal events, but was assumed to be invulnerable to all such special emergencies. The SDHR is evaluated in this EPRI/WOG study for a 5 ariety of accidents, including cases where the SDHR is either more and less reliable than the NRC Case Study. The EPRI/WOG study considers:
| |
| o Infant mortality.
| |
| o Failure to consider all required components.
| |
| o Common-cause effects between normal plant safety systems and the SDHR.
| |
| o Events which could be mitigated by either train of the SDHR system,i.e., by the makeup train or by the emergency feedwater (EFW) train, o Actuation of SDHR.
| |
| o Seismic response of SDHR.
| |
| The following discussion addresses these points and provides a summary of the effect of the SDHR on core-melt sequence frequencies. Of particular importance is the adverse impact of the SDHR on risk for tam generator tube ruptures and steam line breaks inside containment, as well as seismic events at high acceleration resulting in loss of containment heat removal. These adverse impacts, if not corrected before SDHR installation, could cause an increase in public risk.
| |
| Infant Mortality With no design information nor operations experience available, the EPRI/WOG study judged that it was not realistic to credit SDHR availability based strictly on the component reliabilities presented in the NRC Case Study.
| |
| The SDHR should be expected to experience a long period of infant monality because of the lack of testing and operational experience available for Point Beach when compared to existing safety system designs. While this SDHR design uses standard components, the design philosophy - to reach safe shut-down of the plant independent of operator action for ten hours - significantly differs from existing safety systems. Therefore, this EPRI/WOG study anticipates most of the additional unreliability to result from problems with inidation and control of the system. The importance ofinfant mortality is underscored by two points. First, assuming that the only opportunities for full functional tests would be during a 9 fueling outage, the SDHR will be tested at best twenty times during the remaining years of cmrently
| |
| . censed plant life. The time value of hypothetical averted costs in the value-impact analysis would tend to increase the importance of the SDHR reliability early in its life.
| |
| 73
| |
| | |
| l l
| |
| l l
| |
| The EPRI/WOG study increases for SDHR infant mortality by a factor of two, which is somewhat '
| |
| Tlaller than that evidenced by nuclear power plant trip data (17.6 first year trips verses 5.8 average per EPRI repon NP-2230(M) and section 5 of this report).
| |
| l Consideration of Additional Comoonents i
| |
| The SDHR as described is intended to be self-contained. That is, cooling water systems, room cooling j systems, and possibly others would be required to meet the design criteria described in the NRC Ca',e Study. The EPPJ/WOG study estimated that such systems, by approximately doubling the number of components, would increase the unavailability of the SDHR by a factor of 2.
| |
| Common.Cause Between the SDHR and the Existine Plant Systems NUREG/CR-2883 notes that one of the deficiencies of this SDHR design is in the similarity of the components used in both it and the plant. This similarity is of concem for common-cause failures.
| |
| Since this concern is noted in the design basis for the SDHR used in the NRC Case Study and since data exists for such common faults, the EPRl/WOG analysis has assessed common-cause failure rates for the system for certain key accidents.
| |
| he EPRI/WOG study assumes common-cause dependency for sequences which include either diesel common cause failures or single diesel failures. That is,if two diesels fail due to a common cause, we credit the diesel generator in the SDHR only with the conditional probability obtained by using a common-cause model for three diesels failing in a three-diesel system. Also, if a single diesel fails and a core-damage sequence results from other failures, we assume that the diesel in the SDHR will fail with the probability that two diesels would fail due to common cause in a three-diesel system. These common-cause analysis considerations change the effectiveness of the SDHR enly for certain station blackout sequences. These changes increase the unreliability cf the SDHR noticeably but not dramatically. (The EPRIAVOG sttidy used the same diesel generator failure rates for independent and common cause as were used in the remainder of this study, which is about two times lower than the failure rate assumed in the NRC Case Study.)
| |
| Accidents That Can be Midcated Bv Either Train of SDHR The NRC Case Study assumes that failure of either train of the SDHR will result in functional failure.
| |
| According to the EPRI/WOG ant. lysis integral plant model, the SDHR will succeed if either the t
| |
| makeup-SDHR or EFW-SDHR functions for accidents with RCS integrity maintained. If RCS integrity fails, either because a small LOCA initiated the accident or because a transient-induced LOCA occurs, l .ly makeup-SDHR will successfully prevent core damage.
| |
| 7-4 1
| |
| i____________._ _ -_
| |
| | |
| The amount ofinjection provided by the makeup-SDHR train will be sufficient to keep the core covered 1d remove decay heat for either transients or SBLOCAs. Two hundred gallons per minute will more dian compensate for decay heat as is indicated in section 3 of this report. The amount of EFW-SDHR will also be adequate for providing sufficient secondary heat removal for transients.
| |
| However, the EFW-SDHR train will not be adequate on its own for SBLOCAs. If only feedwater were available, secondasy cooling would delay core uncovery and possibly allow operator intervention to depressurize the primary system and provide low pressure injection. However, SBLOCA risk was dominated more by makeup failures than failure to depressurize and use low pressure injection.
| |
| Therefore, feedwater alone will have only marginal impact on LOCA-related risk at Point Beach.
| |
| To summarize, the following is the EPRI/WOG model application of the SDHR:
| |
| o SB LOCAs - the unavailability of the SDHR is equal to the sum of the unavaila-bilities of actuation, electric power, and makeup-SDHR.
| |
| o Transients - unavailability of the SDHR is equal to the sum of the unavailabilities of actuation and electric power, plus the product of the unavailabilities of EFW-SDHR and makeup-SDHR.
| |
| 5cruation of SDHR The engineering details of the SDHR initiation and control design are not given in the NRC Case Study.
| |
| However according to NUREG/CR-2883, the SDHR will be actuated as follows: :
| |
| o The EFW-SDHR train will be actuated on low steam generator level.
| |
| o The makeup-SDHR train will be actuated on EFW-SDHR train stan as well as on low RCS pressure.
| |
| o The atmospheric steam dump valve (ASDV) for controlling steam generator pressure will be actuated on high RCS temperature.
| |
| By themselves, the existence of these three actuation signals are not sufficient to describe the actuation and control philosophy for the SDHR. In panicular, insufficient information is provided to comprehensively evaluate the operation of the SDHR for tne dominant intemal event sequences reported in the NRC Case Study, namely SBLOCAs followed by recuculation failure and long term station blackout events.
| |
| Problems were encountered by the EPRI/WOG review in evalurting the systein to the level of detail necessary to accurately assess the benefit of the SDHR. The problems included:
| |
| o At what steam generator level will the EFW-SDHR actuate, i.e., at the same level as auxiliary feedwater or m a level below it?
| |
| 7-5
| |
| | |
| o- Will the makeup-SDHR pump be isolated when RCS pressure falls below the ' .
| |
| makeup train actuation serpoint during a normal shutdown or during depressurization following a successful cooLdown following a SBLOCA?
| |
| o Can the operator terminate the system after it is actuated, either spurious!'y or as a l result of a demand? -
| |
| ,, o Can the operator initiate the system from the control room, locally, or both?
| |
| o Will high temperature actuation of the SDHR ASDV occur even if core damage has occurred?-
| |
| o Will the SDHR 'ASDV reduce pressure below that necessary for AFW turbine-driven pump operation?
| |
| o At what RCS pressure will the SDHR actuate and will that pressure be reached for the SBLOCAs considered in this study?
| |
| ' o How does the system control or terminate flow to prevent overfilling the pressuri-zer or steam generator? l The following discussion describes some of the risk significant issues raised by these questions. While l
| |
| design of an SDHR is not within the scope of this study, assumptions were made by the EPRl/WOG study to complete the assessment of the SDHR. Consistent with the philosophy of both the NRC Case Study and the EPRI/WOG study, conservative assumptions were made when information was not available.' In some cases these assumptions reduced the effectiveness of the SDHR over a situation where an independent design analysis might have identified different and more favorable'information.
| |
| EFW SDHR should actuate on all transients.that require a demand. However, if EFW SDHR is actuated at the same level as APN, EFW-SDHR may require manual termination to avoid overcooling.
| |
| If we assume thai :;ubsequent operation of the EFW-SDHR would require manual actuation, then all i run related failures of AFW would require manual initiation of the EFW-SDHR system. For station blackout events, the dominant sequence of importance for AFW is that the normal AFW water supply will eventually be depleted. He SDHR would then require manual initiation. Since human error is an important cause of failure to provide long-term AFW supply, the addition of an SDHR requiring manual initiation would have e limited effect for reducing station blackout risk considered in these studies.
| |
| However,if EFW-SDHR is actuated at a lower steam generator level than AFW, these problems may not occur. An attetsment would be required to determine if EFW-SDHR actuated before the PORVs opened and to determine the corresponding impact on SDHR reliability if one of the PORVs stuck open.
| |
| This study conservatively assumes initiation at the same level as AFW.
| |
| If makeup-SDHR is scruated on low RCS pressure,it will have actuated for all SBLOCAs before ECCS circulation is mquired. As in the case with accumulators, the system would most likely have been manually isolated to prevent actuation when not required during a normal cooldown. D erefore, 7-6
| |
| | |
| l 4
| |
| i
| |
| , ;{
| |
| ( . makeup SDHR would have to be actuated manually for all HPI recirculation failure sequences. Because 1
| |
| deup SDHR could be actuated from the control room without local manual actuation, the SDHR ;
| |
| I ould be more reliable than RWST makeup by using the CVCS. However, CVCS could also be actu-ated from the control room under these conditions without local manual action. Further, it is likely that purely because the CVCS is a normally operating and reliable system, operators would use the CVCS before the SDHR. Consequently, the SDHR systetn provides meager additional safety margin for I SBLOCAs or transient induced LOCAs that result in recirculation failure of ECCS. These sequences I were the dominant intemal event sequences in the NRC Case Study.
| |
| l 1
| |
| The EPRI/WOG study assumes that the ASDV, if stuck open, will prevent the AFW turbine-driven pump from operating. However, the EPRI/WOG study does assume that the high RCS temperature set-point for the ASDV would be high enough to prevent actuation of the ASDV for sequences where AFW l
| |
| initiates automatically. If AFW requires manual actuation, e.g., the two fire scenarios, the EPRl/WOG study does assume that the ASDV will open. It could stick open with a probability of SE-3 (the same as EPRI/WOG assurned for the PORV).
| |
| In the event of a fire, if the EFW-SDHR fails, the EPRl/WOG study assumes the AFW turbine driven pump will also fail if the ASDV sticks open. That is, steam generator level will continue to decline,
| |
| 'igh RCS temperature will result, and the ASDV will open. If it sticks open before the turbine-driven amp can be operated to prevent high RCS temperature, EPRI/WOG assumes steam pressure decays I rapidly enough to fail the motive force for the turbine. Since the AFW turbine-driven pump is about 8 times more likely to fail than the ASDV is to stick open, this consideration does not have a significant impact on SDHR success for fire scenarios. (The EPRI/WOG study does assume that the SDHR is designed and installed to be undfected by the two fire scenarios. Further, it assumes that an independent means of cooling will be established before the SDHR condensate and makeup tanks fail.) i i
| |
| I The EPRI/WOG study assumes that if a transient-induced LOCA occurs and feedwater is available (either from existing systems or from the SDHR), the RCS pressure will be reduced enough to actuate the makeup-SDHR train before core uncovery begins. The combined effect of mass loss out the break and secondary heat removal would be expected to r: duce RCS pressure to below the normal ECCS initiation setpoint. If feedwater were not available from existing systems, the makeup-SDHR train would be actuated on EFW-SDHR actuation on low steam generator level.
| |
| In summary, the SDHR actuation model is assumed to function as follows:
| |
| o For SBLOCAs or transient-induced LOCAs resulting in a failure of ECCS recir- l culation, the SDHR is given no credit unless charging is unavailable. Charging is expected to be available for all cases because only two pumps would be required 7-7
| |
| | |
| a
| |
| - at this time in the accident sequence sad electric power would be available since -
| |
| . HP1 had been operating (i.e., the SDHR is given no credit).
| |
| o For SBLOCAs resulting in ECCS injection failure, the SDHR should actuate on low RCS pressure.
| |
| I- ,
| |
| o For transient induced LOCAs, SDHR is credited for all accidents since we assumed either the low RCS pressure setpoint will be reached or EFW-SDHR ]
| |
| ll (and therefore makeup-SDHR) would be initiated on low steam generator level. j Seinmic Resnonse of SDHR Finally, EPRI/WOG raises another concern not considered in its base case analysis. De NRC Case
| |
| . Study assumes that the SDHR will withstand all seismic events with 100% effectiveness; yet the study requires only Seismic Category I design for the system. (See Appendix J of the NRC Case Study which Leontains the' impact, i.e., cost analysis.) Point Beach is currently. designed to Seismic Category I.
| |
| Therefore, it is obvious that potential weak points in the SDHR could exist if they are assessed to exist in the plant as currently designed, nese weak points could reduce the effectiveness of the SDHR for all or a portion of the seismic sequences.
| |
| Summary Fault Analysis and Results for Secuence Ouantification
| |
| ' A summary of the ineffectiveness of the SDHR is. contained in table 7-1. The results of the EPRI/WOG analysis indicate that the SDHR is twice as ineffective as assumed in the NRC Case Study. That is, the EPRI/WOG assessment of the average ineffectiveness of the 'SDHR is 0.18, while the average ineffectiveness determined by the NRC Case Study is 0.084. The table also indicates a wide range of ineffectiveness for the system depending on the specific scenarios for which the SDHR is applied. The following provides an overview of how this ineffectiveness was calculated. Specific details are provided in appendix E.
| |
| In the case of long-term station blackout, the high SDHR ineffectiveness is due to a reduction in the effectiveness of the CST refill recovery. The CST xefill recovery is assessed to be less likely for a plant design which includes an SDHR. The principal reason for this change in the EPR1/WOG assessment of the CST refill recovery is the assumption that the operators would be tesipv.w3y divened by failure of I the SDHR. 'Ihis diversion would cause a loss of valuable time, thereby reducing the likelihood of suc-l cessful diagnosis of the need for CST refill. 'Ihe actual availability of the SDHR for these sequences varies from 0.096 to 0.22 depending on the susceptibility of the SDHR to common-cause failure.
| |
| 7-8
| |
| | |
| < , -1
| |
| (. -
| |
| j
| |
| . - ., s
| |
| - Table 7-1 'i
| |
| )
| |
| Summary of SDHR Impact on Core Melt Frequency {
| |
| and Comparison of EPRI/WOG and NRC Case Study - !
| |
| Assessment of SDHR Unavailability (Base Case EPRI/WOG Assumptions for SDHR) '
| |
| i 1
| |
| EPRI/WOG EPRl/WOG Core Melt Core Melt Fre- NRC Case ~
| |
| Frequency for quency with Average inef. . Study Assumed Existing Point SDHR lastalled factiveness! Ineffe ;veness2 Core-Melt
| |
| , Sennence Tyne Esash Plant /vr (Base Caue) of SDHR ofSI./HR Long-Term Station Blackout 5.4E-7 4.(E 7 0.74 0.1 Short-Term Stadon Blackout 8.9E 7 1.9E 7 0.21 0.1 3.3E 7 0.49 0.052 52MH1'H2' and S2MD1D2 6.8E-7 OtherInternal Events 4.5E 7 1.7E-9 3.7E 3 0.052 Fire ' 6.3E 8 2.3E-10 3.7E-3 0.052 Seismic
| |
| : l. < 3 x SSE l
| |
| 'SBLOCA 1.7E 6 - 2.6E-7 0.15 0.1 transient 3.5E-6 3.4E 7 0.096 0.1
| |
| > 3 x SSE SBLOCA 1.4E-6 2.0E 7 .15 0.1 transient 8,0E-7 8.1E-8 f!M QJ Totals / Composite 1.0E-5 1.8E-6 0.18 0.084 I
| |
| : 1. Ineffectiveness is defined as the fractional reduction in core-melt frequency with and without the SDHR installed (column 2 divided by column 1) 2 NRC Case Study ineffectiveness is assumed to be equal to SDHR unavailability.
| |
| 1 l
| |
| 79 L--____-________ j
| |
| | |
| The ineffectiveness for the SBLOCA sequences was also increased because of sequence-dependent
| |
| : tors. The EPRl/WOG study assumes that the SDHR must be manually initiated for recirculation failures. If human action was the principal cause of recirculation failure, the SDHR is assumed to not be initiated. The EPRI/WOG study applied the same restriction to recirculation recovery using RWST refill. Consequently, the SDHR is not effective for the dominant cause of recirculation failure sequences.
| |
| Other ineffectiveness factors can be explained as follows. For short-term station blackout sequences, common-cause failure between plant diesel (s) and the SDHR diesel increases SDHR ineffectiveness.
| |
| For fire sequences and transients with offsite available, either makeup or emergency feedwater can successfully prevent core melting and the SDHR diesel generator is not required. This effect allows the SDHR to work with a high degree of reliability. For seismic sequences, SDHR ineffectiveness is assumed to be equal to its unavailability. For SBLOCA sequences, both the SDHR diesel and the makeup train must succeed. For transient sequences, the SDHR diesel and either the makeup train or the emergency feedwater train, must succeed.
| |
| SENSITIVITY ANALYSIS OF POTENTIAL ADVERSE PUBLIC RISKIMPACTS
| |
| ""he EPRl/WOG study identified two issues with potential adverse impact on public risk from
| |
| .stallation of the SDHR as designed. The following describes the two issues and then the sensitivity study perfonned to consider the range of effect on SDHR value.
| |
| ASDV Actuation After Core Damace Automatic actuation of an Atmospheric Steam Dump Valve (ASDV) on high RCS temperature could have significant adverse safety impacts for some scenarios that proceed to core damage. We must assume that the ASDV would open if core damage was in progress. because the RCS temperature would certainly be high. If this is the case, an accident initiated by a steam-line break inside a containment would result in a containment isolation failure. (The leak path would be from the containment, through the main steam-pipe break,into the main steam system, and out the ASDV to the environment.) If a steam generator tube rupture-initiated core melt accident occurred, a leak directly from the pnmary systera to the environment (i.e., a containment bypass) could result.
| |
| The Oconee PRA (14) calculated steam generator tube rupture (SGTR)-initiated core melt frequency as 2.7E-6, while the Indian Point PRA (22) calculated it as 1.6E-6. Using these numbers as a guide, an
| |
| ~
| |
| EPRI/WOG sensitivity analysis assumed that incorporating the SDHR into the Point Beach design
| |
| >uld result in an early containment failure-type accident release with a fmluency of 4.4E-7. (The aDHR is assumed by EPRI/WOG for convenience to reduce SGTR risk by 80% in the development of these numbers.) This effect is not included in the base case, but is included in a sensitivity study. It was 7-10
| |
| | |
| .4 not included in the base case because we assumed that the problem, now identified, would be corrected y the SDHR designer prior to any installation.
| |
| SDHR Ooeration After Dedicated CST and RWST Empty Durine Hinh Acceleration Seismic Events Because the SDHR does not have a recirculation capability, it can only delay core damage until other systems are recovered. If recovery does not occur, and if containment beat removal is a failure, the -
| |
| SDHR may delay core melting until containment conditions are conducive to large releases. The impact of this concern is evaluated in the following paragmphs. (Note, the following adverse impact is not
| |
| . included in the base case EPRI/WOG evaluation. 'Ihat base case evaluation assumes that potential adverse impacts will be removed from the SDHR design before installation.)
| |
| The basic assumption regarding the system's mitigation capability is that enough equipment can be recovered to cool the core after the SDHR runs out of water. The SDHR would no longer function after about 10 hours. This EPRI/WOG study assumes that the normal plant safety equipment will be recov--
| |
| cred in most scenarios, e.g., offsite power will be recovered or other sources of water will be found for recirculation, etc. The EPRI/WOG analysis identified many instances where recovery would occur even without the SDHR. In cases where the SDHR succeeds, even more time would be available for these recovery actions.
| |
| However, the assumptions about recoverability of equipment within 10 hours become more questionable if the accident occurs as a result of a high acceleration earthquake. In this case the EPRI/WOG study I
| |
| considers in a sensitivity study that the system would only delay core damage for 10 hours, rath:r than l
| |
| prevent it. As a consequence, containment pressure would continue increasing if containment heat removal had also failed. At the time of core melting, containment pressure would be higher than if core melting began early. As a result, the margin will be reduced between containment pressure at the time of vessel failure and the estimated containment failure pressure.
| |
| If the reactor vessel eventually fails after core melting, an increase in containment. pressure w!!! result from the blowdown of the vessel and the contact between core debris and water inside the containment.
| |
| The resulting rapid steam pressurization, when considered in light of a reduced margin to containment ' ,
| |
| failure, would be a larger threat to containment ir.tegrity. Most important] , that threat would occur l relatively early compared to when fission products are released from the core. Such a concern regarding an adverse impact of a plant modification has been considered before. (See the Oconee PRA G4).) a i
| |
| If containment heat removal is available, this threat of early containment failure would not exist. Since 6e EFW-SDHR will remove heat, it was also assumed to be sufficient to prevent this type of threat. )
| |
| Therefore, for those scenarios with successful makeup-SDHR but failed EFW-SDHR, the EPRI/WOG l
| |
| 7-11
| |
| | |
| k study then assumes a higher potential for early containment failure. In general, source term studies have ot evaluated these scenarios since they were not risk significant in any PRAs. For this reason, the
| |
| ]
| |
| dPRI/WOG sensitivity study used an early containment failure probability of 0.1 from the Oconee PRA, together with the EPRI/WOG source term for early containment failure. (See section 9 and Appendix C.)
| |
| The range of values of adverse risk will be affected by the following factors:
| |
| o The frequency of SGTR initiated core-melt accidents.
| |
| o The ineffectiveness of the SDHR in response to an SGTR.
| |
| o The release magnitude for SGTR-initiated core melt with the ASDV for the SDHR open.
| |
| o The survivability of the SDHR, including both SDHR-EFW and SDHR-makeup events greater than 3 times the SSE. >
| |
| o The recoverability of the plant or the SDHR after the ten hour mission of the SDHR is completed.
| |
| o The degree of heat removal provided by the EFW SDHR if functioning.
| |
| o The containment capacity.
| |
| o The containment loading shortly after fission product release.
| |
| This sensitivit; study assumes that the SGTR initiated core-melt frequency is 4.4E-7 based on a core-melt frequency of 2.2E-6 without the SDHR and an SDHR ineffectiveness of 0.2. The study also assumes that the release for an SGTR-initiated core-melt accident with an open ASDV is the same as the early containment failure source term for the EPRI/WOG model.
| |
| This sensitivity study assumes that the SDHR will survive an earthquake greater than three tirnes the SSE, but that neither the SDHR nor the other plant systems will be recovered within ten hours. The study assumes that the EFW-SDHR will not remove an appreciable amount of heat for the core melt accident. At three times the SSE, greater than 60% of the core-melt frequency results from SBLOCA sequences, for which EFW-SDHR would not be very effective at removing heat because the RCS pres-sure would drop below steam generator pressure. Finally, the study assumes that the conditional probability of a containment failure which would cause the previously-mentioned IDCOR carly release l
| |
| is 0.1.
| |
| The results of this sensitivity study indicate that public health risk from the expected value of offsite isc wenld increase by about 4-person-rern per year over the remaining 23 years of licensed plant life.
| |
| sihe EPRI/WOG study calculated an expected value of dose for the remaining plant years of about 7-12
| |
| | |
| 7-person-rem per year.) Effective dose is defined as the dose from potential accidents weighted by its
| |
| 'robability of occurrence within the remaining plant life. Two-thirds of the increase is due to the SGTR-initiated risk and one-third due to the seismic risk. The SDHR decreases the expected dose by fhout 5 person rem per year for its positive effects, but increases dose by about 9 person rem for its adverse impacts. (The NRC Case Study calculated an expected value of dose for the remaining plant years of about 39-person-rem per year.)
| |
| l COMPARISON OF SDHR TO EXISTING POINT BEACH EQUIPMENT AND PROCEDURES The EPRI/WOG study identified a variety of existing plant-design features and operational practices which can be considered similar in effect to the SDHR. These plant design features include: >
| |
| o Seismic Category I CVCS, including pumps and water sources for both long-term and short-term requirements o AC power-independent long-term makeup supply for condensate from the diesel-driven fire pumps o Seismic Category I makeup supply for condensate from the service water system The basic concept for the SDHR is to provide makeup and EFW backup in the event of a significant !
| |
| sccident. These features piovide a similar capability that currently exists at the plant and is known to
| |
| .he operators and the emergency responders. Table 7-2 compares and contrasts these two systems,i.e.,
| |
| the SDHR and Point Beach existing equipment. Note that the existing equipment mentioned above was not credited in the NRC Case Study.
| |
| 7-13 i
| |
| | |
| k Table 7-2 COMPARISON OF SDHR AND EXISTING POINT BEACH EQUIPMENT NOT CONSIDERED IN THE NRC CASE STUDY Existing Point Beach Design Features Not Credited in the NRC Case Study SDHR Makeup Capab!!ity o 180 gpm fmm three charging pumps o 200 gpm charging pump o Seismic Category I o Seismic Category I o > 200,000 gallons in existing tanks o 120,000 gallon new RWST o makeup from spent fuel pool also available o automatically actuated on low RCS pressure o manual actuation always required o manual actuation required ifisolated after low RCS pressure signal, e.g., for recircula-tion failure after SBLOCA Emergency Feedwater Capability no " dedicated" pumps but diesel- o additional electric powered pump providing driven fire pumps provide capability 1200 gpm o manual recovery required (covered in training) o automatically initiated on low steam o unlimited supply of water from lake either from desel-driven fire pumps o Seismic Category I to CST or from service water direct to AFW (Seismic Category I) o new CST with 200,000 gallons l
| |
| t ;
| |
| i 7-14
| |
| | |
| l l
| |
| Section 8 CORE-MELT SEQUENCE RESULTS
| |
| | |
| ==SUMMARY==
| |
| AND CONCLUSIONS This review's reevaluation of the NRC Case Study sequences indicates a significant reduction in core-melt frequency (i.e.,30 times less). The two principal reasons for this reduction are:
| |
| o As a result of corrections made for previous use of conservative methods, assumptions, and data when performing probabilistic risk assessments.
| |
| o The safe operating practices of Wisconsin Electric Power, including incorporation of symptom-based procedures and training for recovery.
| |
| The conservative assumptions and data used in the NRC Case Study general approach provide the most important reason for the significant reduction in core-melt frequency. For example, both the SBLOCA
| |
| 'requency and transient-induced LOCA frequency in the NRC Case Study are significantly higher than
| |
| .or other PRAs, whether industry or NRC. The human error data utilized in this study is also conservative. In point of fact, the techniques used are inconsistent, i.e., more stringent, than those recommended in NRC's Handbook for Human Reliability Analysis (NUREG/CR-1278) (2). i Another important factor in obtaining this reduction (especially in light of the NRC Case Study conser-vatisms regarding recovery and data) resulted from the safe operating practices of Wisconsin Electric Power (WEP). The Point Beach plant was originally selected by NRC for study because a qualitative design screen indicated some potential weaknesses. Because WEP's awareness of its problems led to various oagoing corrective programs in traming and procedures, this study has been able to take credit for many recovery actions. These safe operating practices are also demonstrated by Point Beach's tran-sient frequencies relative to other plants.
| |
| The scope of the NRC Case Study does not address some aspects of plant risk which frequently yield dominant accident sequences. Hence, the low results found in this EPRVWOG review are not necessar-ily inconsistent with other studies. Since this EPRVWOG study did not have the resources to redo the analysis, such sequences have not been specifically identified and quantified; however, potential non- ,
| |
| conservatism have been mentioned. Examples ofimportant sequences from past PRAs not included in )
| |
| ie NRC Case Study are as follows: {
| |
| }
| |
| o Fires in electrical equipment (rather than transient combustible fires). j
| |
| (
| |
| 1 8-1 l
| |
| - - - - - - - - - .------___.----_j
| |
| | |
| i i
| |
| )
| |
| o RCP seal LOCAs resulting from station blackout (excluded from study scope). I l
| |
| o Transients initiated by loss of service water or component cooling water (not j evaluated in the NRC Case Study methodology). .
| |
| l There is reason to believe that these scenarios may not be dominant at Point Beach. New O-ring seal materials are being evaluated for installation in a Westinghouse Owners Group program which has also developed a more realistic seal LOCA PRA model. Component cooling water is not required for HPSI or charging pumps. However, since this study was basically restricted to evaluating the NRC model rather thaa reevaluating the plant, we are limited in our ability to categorically rule out other scenarios.
| |
| Gt should be noted that although some aspects of accident risk have been increased,in each case other mitigating factors have combined to reduce overall accident sequence frequency.)
| |
| Comparison of NRC Case Study Results and EPR1/WOG Results Tables 81 through 8 6 provide an overview of the differences between the NRC Case Study results and the results of this EPRI/WOG study. Table 81 provides a summary of total internal event frequency and individual external event frequency, and indicates that the most significant reduction in frequency occurred for internal floods, fires and other internal events. The fire analysis is conservative and based '
| |
| on the results of other PRAs. The internal events are thus reduced dramatically,largely because of data hanges for initiating event frequencies, but also because of credit for equipment and procedures not considered in the NRC Case Study.
| |
| The seismic scenarios were also reduced, the principal changes being the incorporation of a few key recoveries at lower ground accelerations, and adjustments to the hazard curve making it consistent with findings of EPRI seismic margin studies. The reduction in the mean seismic hazard curve led to sub-stantial reduction in overall seismic risk. In addition, WEP has also installed new seismic htteries that ,
| |
| provide a redundant source of DC power for starting the diesel generators and providing instrumentation. Rese changes led directly to a significant reduction in the frequency of loss of all AC power sequences.
| |
| This study also considered potential means of recovery from a seismically-induced RWST failure, an area which has not been considered in other PRAs. %e bases for incorporating the recovery in this study are the reevaluation of a conservative assumption about the tank's failure mode and a seismic alternative injection capability that appears reasonable because of the low accelerations involved.
| |
| (When compared to RWSTs at other plants, the Point Beach RWST is more likely to fail at low accelerations. This makes more plausible the less conservative failure mode assumption and the ailability of the CVCS, which is of Seismic Category I design.)
| |
| 8-2
| |
| | |
| i Table 8-1 Total Core Melt Frequency (Per Year) Summary Core-Melt Freauenev Per Year NRC Accident Tvoe Case Study EPRI/WOG Study Intemal Events 1.4E-4 2.6E 6 Seismic 6.1E-5 7.4E-6 Fire 3.2E-5 6.3E-8 Intemal Flood 7.7E-5 <1.0E-8 Wind 4.0E-6 <1.0E 8 Extemal Flood 1.9E-S <1.0E-8 Lightning 5.8E-8 <1.0E-8 Totals 3.1E-4 1.0E-5 Table 8-2 i
| |
| Internal Events Summary Core Melt Freauency Per Year NRC J Tvoc of Accident Scenario Case Study EPRI/WOG Study 1
| |
| Total Long-Term 3.6E 5 5.4E-7 I Station Blackoutl Total Shon-Ter n 6.7E 6 8.9E-7 ,
| |
| Station Blackoud l Other Internal Ever,ts3 9.2E 5 1.1E-6 1
| |
| Total Inta.rnal Events 1.4E-4 2.6E-6 Frequency
| |
| : 1. See Table 8-4
| |
| : 2. See Table 8-3, sequences TIMLE and T1QD1D2
| |
| : 3. See Table 8-3, sequences other dian TIMLE and T1QD1D2 8-3 l
| |
| _ _ _ _ _ _ . _ _ _ _ - _ - _ _ _ /
| |
| | |
| Table 8 3 Comparison of Internal Event Sequence Results (Not Including long-Term Station Blackout)
| |
| Cg-Melt Frecuency Per Year Key Reasons Sequence Name NRC Case Studv EPRI/WOG Study for Change S2MH1'H2' 4.7E-5 5.8E-7 - Small LOCA Frequency CCW Success Criteria Operator Actions TIMLE 6.7E-6 7.7E-7 New Batteries T3QH1'H2' 2.5E-5 N/A Relief Valve LOCA cannot occur T2MQH1'H2' 3.5E-6 1.9E-7 Relief Valve LOCA Prob.
| |
| Operator Actions S2MD1D2 8.7E-6 9.5E-8 Small LOCA Frequency CCW Succ:st Criteria
| |
| '3QDID2 4.6E-6 N/A Relief Valve LOCA cannot Occur T2MLE 6.6E-7 1.0E 7 Main Feedwater Recovery Alternative to 1 PORV T2MQDID2 6.6E-7 4.1E 8 Relief Valve LOCA Prob.
| |
| CCW Success Criteria S2MXD1 5.7E 7 1.0E-8 Small LOCA Frequency CCW Success Criteria T5MLE 9.1E 7 1.3E-8 DC Bus Cross Connect Main Feedwater Recovery T4MLE - 6.2E-7 N/A loss of AC Bus does not d p Pl ut T2MLH1 2.0E-8 ' .0E-7 t Operator Actions Main Feedwater Recovery J t
| |
| i T1QD1D2 <1.0E-8 3.2E-7 Relief Valve LOCA more likely to occur <
| |
| /
| |
| 8-4 l
| |
| - _ _ _ _ _ _ _ _ - _ _ - _ _ _ _. l
| |
| | |
| f Table 8-4 Long Term Station Blackout Quantification l
| |
| l Initiator Total Frequency Probability EPRl/WOG (less of Equipment Faults of Blackout Not of Failure to Core Melt Offsite Causing Causing Early Provide Long- Frequency Power Station Blackout! Core Damage Tam Cooline Per Year T1- DG-CM (Diesel Common 5.0E-6 0.032 1.5E-7 Mode)
| |
| T1 (DG-II)2 (Diesel Local 1.8E-6 0.032 5.4E 8 Fault)
| |
| T1 DG-LF x DG UTM x 22 1.0E-6 0.032 3.0E 8 T1 DG-LF x SWP-LF x 1.3E-6 0.14 1.3E 7 SWSLDBAL5 MIN 3 x 6 (4 of 6 Service Water -
| |
| Pumps Unavailable)
| |
| Tl DG-LF x SWP-UTM x 1.5E-6 0.14 1.5E 7 SWSLDBAL5 MIN x 4 (4 of 6 Service Water Pumps Unavailable)
| |
| . T1 DG-UFM x SWP-LF x 3.5E 7 0.14 3.5E-8 SWSLDB ALSMIN x 6 (4 of 6 Service Water Pumps Unavailable)
| |
| Total EPRI/WOG Core Melt Frequency (per year) 5.4E-7 for 'long Term Station Blackout'
| |
| : 1. Numbers indicate the number of combinations of similar equipment faults, e.g., DG A local fault and DGB maintenance plus DGB local fault and DG A mamtenance l
| |
| 1
| |
| : 2. See Section 4 and Appendix A for description. Recovery requires refilling CST using diesel fire pumps !
| |
| cr m~u water. ;
| |
| : 3. See Secnon 4 and Appendix A for description of this event. Rmvery requires operators to balance loads in service water system to ensure proper cooling of the diesel generators.
| |
| : 4. Assume that 4.dssy in human actions makes it inappropnate to apply two recovery factors.
| |
| '!herefore, must remove 0.3(SWSLDB AL5 MIN) from cut sets.
| |
| l I
| |
| l 85
| |
| | |
| i I
| |
| Table 8-5 .1 Seismic Analysis Summary i
| |
| Core-Melt Frequency ner year oer Seismic Acceleration Level Dominant Seismic l Sequences Description 1-2 x SSE 2-3 x SSE Other Imal S2/LD1/XD2 NRC Case Study 6.3E-6 6.7E-6 3.0E-6 1.6E-5 EPRI/WOG 6.5E-7 6.5E-7 6.0E-7 1.92-6 E2LD1 NRC Case Study 3.0E-7 4.1E 6 3.3E-6 7.7E-6 EPRI/WOG 3.0E-8 4.1E-7 6.6E-7 1.1E-6 T2UP/QE NRC Case Study 1.2E-5 8.4E-6 4.3E-6 2.5E 5 .
| |
| EPRI/WOG 1.2E-6 8.0E-7 8.1E-7 2.8E-6 T3MUP/QE NRC Case Study 6.3E-6 2.4E-6 2.9E-7 9.0E 6 EPRI/WOG 6.5E-7 2.4E-7 3.0E 8 9.2E-7 Other seismic contributors NRC Case Study 3.0E-7 1.0E-6 5.0E-7 3.0E-6 EPRI/WOG 1.5E-7 5.0E-7 1.0E-7 6.5E-7 Total Seismic Core-Melt Frequency NRC Case Study 2.5E-5 2.3E-5 1.2E 5 6.lE-5 EPRI/WOG 2.7E 6 2.6E-6 2.2E 6 7.4E-6 As discussed in the ext, due to the uncertainty regarding the availability of ec uipment for recovery, recovery only applied to cut sets below 3 x SSE. Besides recovery analysis, tie only other quantified difference between EPRI/WOG and the NRC Case Study is a reduction in EPRI/WOG hazard curve by a factor of two for 1-3 x SSE and by five for greater than 3 x SSE (i.e., "Other" column).
| |
| i 8-6 w________
| |
| | |
| Table 8-6 Fire Analysis Summary EPR1/WOG Core-Melt Secuence Totals Fire Fre- Manual Operator or Core-Melt initiator quency Key Auto Sup- Suppres. AFW TDP Sequence Iyg per Year i ncarion pression siQD faull Total (ner year)
| |
| AFW CF 2.6E-4 14.A. 3.6E-3 0.15 0.07 1.0E 8 AFW-PF 2.2E-4 N.A. 3.6E 3 0.15 0.07 8.5E-9 AFW TC 1.7E4 03 3.6E-3 0.15 0.07 2.0E-9 SW CF 5.5E-4 N.A. 3.6E-3 0.04 0.07 5.6E 9 SW.PF 2.2E-3 N.A. 3.6E 3 0.04 0.07 3.0E 8 SW-TC 7.5E-4 N.C. 3.6E 3 0.04 0.07 7.7E 9 Comparison of EPRI/WOG and NRC Fire-Induced Core-Melt Frecuency Results by Room Core Melt Frequency Per Year NRC Case Studv EPRl/WOG Study AFW Pump Room 13E-5 2.0E-8 Switchgear Room 2SE-3 43E-8 Totals 33E-5 63E-8 Ety AFW = Auxiliary Feedwater Pump Room CF = Cable Fire SW = Switchgear Room PF = Panel Fire N.A. = Not Applicable TC = Transumt Combusutte N.C. = Not Calculated Due to lsk of Information in NRC Case Study
| |
| '!DP = Turbine Driven Pump 8-7
| |
| _ _ _ _ _ _ _ _ _ . - _ i
| |
| | |
| i i
| |
| , Internal Events i lable 8-2 summarizes internal event risk in terms of the following three components:
| |
| o Long-term station blackout, i.e., auxiliary feedwater succeeds until CST emp-13es, o - Short term station blackout,i.e., no power and AFW failure.
| |
| o Other internal events.
| |
| Table 8-3 presents internal event risks in more detail, utilizing the terminology presented in the NRC 3 a
| |
| Case Study and mentioned in section 2 of this study. De most important reasons for the differences in j
| |
| ' the results of the NRC Case Study and of this review are presented here:
| |
| o HPI does not require CCW.
| |
| t o SBLOCA and transient-induced LOCA frequencies.
| |
| 1 o low Point Beach-specific transient frequencies.
| |
| o AC and DC bus cross connects.
| |
| o Miscellaneous operator recovery actions or human errors in performing normal actions, e.g., initiate feed-and-bleed or switch to recirculation.
| |
| Long term station blackout is the dominant contributor to internal event risk, as assessed by both EPRl/
| |
| WOG and the NRC Case Sady. However, it is the ability to recover that provides the principal reason for the difference found between the NRC Case Study and EPRIAVOG. The condensate storage tank (CST) can be refilled using either the diesel-driven or portable diesel fire water pump. This operation is part of operator training, and dry runs have been performed. Table 8-4 outlines the long term station blackout quantification.
| |
| Short term station blackout is the second most important accident type. However, installation of the new Seismic Category I batteries has significantly reduced the core melt frequency of these events.
| |
| ' Appendix B provides more detail on internal event risks, including a table of the NRC Case Study domi-nant accident sequence cut sets and the conservatism used in each. Dere is more than one conserva-tism in each of these cut sets. This presentation indicates qualitatively that core-melt risk is lower than that presented in the NRC Case Study, by significant margins.
| |
| Seismic Analysis
| |
| .able 8-5 provides an overview of the seismic analysis quantification for both the NRC C se Study and EPRi/WOG. Of the four dominant sequences listed, the last three are split evenly between station 88 i .. . . .. . .. . .. . .
| |
| | |
| )
| |
| i l
| |
| l blackout accidents and simultaneous failures of the RWST and both CSTs. The first sequence, S2/L 1/X D2,is dominated by the RWST as a single event.
| |
| The frequencies for seismically induced accidents were reduced by a factor of two at levels below three times the SSE, and by a factor of five for higher accelerations. The seismic recovery analysis quantifi-cation for this review deals only with accelerations below three times the SSE. It was judged that above this acceleration level, recovery actions employing the new station batteries or the CVCS and Service ]
| |
| I Water Systems could also be susceptible to failure.
| |
| l 4
| |
| The four dominant sequences were each reviewed for possible recovery actions. The first sequence could only be recovered by providing an alternative to injection from the RWST. This substitute procedure is discussed in appendix A. A 0.1 recovery factor was applied for successful recove- ;n about 60 minutes. Since the RWST is assumed to drain over a period of 30 minutes (based on a ,ta review referenced in section 6), about 30 minutes or more would be available to recover the altemative source. Auxiliary feedwater is successful, so further delay in core uncovery would result from the heat l l
| |
| I removal it provides. Recovery was only applied to Small-Small Break LOCAs (SSBLOCAs), which were conservatively estimated based on tables in the NRC Case Study to be 90% of all SBLOCAs for one to two times the SSE, and 80% of all SBLOCAs for two to three times the SSE. These fractions vere based on both the SBLOCA and SSBLOCA frequency data from SSMRP (41), the source for hTC
| |
| _ase Study initiator frequencies.
| |
| The last three sequences involve two different recoveries, depending on whether a station blackout or tank failures have occurred. If a station blackout has occurred, the operator must use the new batteries to start the diesel generators. Since the service water pumps have tripped and normal DC is not avail-able, the operators would have to manually close the SW pump breakers and ensure that diesel generator cooling is p ovided within about 5 minutes. Because of the difficulty of this process,it was assumed
| |
| ~
| |
| that they would fail 30% of the time. Appendix B contains adGt onal details.
| |
| If tank failures have occurred, the operators must recover either the RWST or a source for AFW. The senice water pumps can provide the feedwater source and the RWST recovery would be as discussed !
| |
| before. About 60 minutes is available before core damage becomes unavoidable. While two potential recoveries were available only one was credited, since it seemed that the operators would have time only to effectively implement one action. A non-recovery factor of 0.1 was applied for this recovery sequence.
| |
| l 8-9 L
| |
| | |
| Eire Analysis section 6 outlines many of the problems this review found with the fire analysis. The corresponding quantification is provided in table 8-6. As can be seen, the principal quantifiable differences include:
| |
| o Fire initiator type and frequency.
| |
| o Conditional pmbability that the fire is in an adverse location.
| |
| o Manual suppression probability.
| |
| o Operator non-recovery pr$ ability for manual AFW turbine pump operation.
| |
| o Availability of AFW for switchgear room fires and CVCS for both fires.
| |
| There is a slight difference in AFW turbine pump availability which cannot be explained, but it appears that the NRC Case Study did not include turbine maintenance in its estimates.
| |
| The new batteries play a significant role in the case of a fire in the switchgear room, allowing the opera-tors continuing access to the instrumentation needed to guide their operations. Use of the instruments-tion is important, for without it the operators might begin to underfill or overfill the steam generators. If the steam generator is allowed to overfill, water in the steam lines will cause an overspeed trip of the ,
| |
| trbine-driven AFW pump. The NRC Case Study did not include the new batteries. The assumption !
| |
| I was that this scenario would occur, thus playing a significant role in the frequency of switchgear room j fires. l l
| |
| l Internal Flood Analysis l
| |
| As was discussed in section 6, internal floods were significantly overestimued, due to improper !
| |
| I calculation of flood frequency. Historical data for service water pump room floods was applied without consideration for the specific scenario. For external event scoping studies of this kind, internal flood ,
| |
| risk has fitquently been overestimated. The Oconee PRA is a good example, where the scoping and detailed studies yielded different resul3. !
| |
| 1 I
| |
| The EPRIAVOG study used a pipe failure frequency model and conservatively assumed that leaks i rather than bn:aks were sufficient to cause the identified spray failure. The EPRI/WOG study also did .
| |
| not use the arbitrary 0.1 factor applied for floods causing failure of all pumps, nor did it credit recovery l (as described in appendix A). Finally, because CCW failure does not fail HPI, and because the plant j would be shut down before recirculatic,n was required, the EPRI/WOG analysis climinated this scenario.
| |
| 8 10
| |
| | |
| Wind. Linhtnina. and External Flood Analysis i ind, lightning, and external floods make up a small portion of the risk calculated by the NRC Case Study, i.e., about one percent. For this reason, these scenarios were not reevaluated as part of the EPR'dWOG study. The EPRI/WOG study believes the risk of these events to be below 1.0E 8 because of conservatism in the analysis identified in Reference 48, and because WEP has strengthened the diesel genemtor exhaust supports, i.e., the wind vulnerability identified in the NRC Case Study.
| |
| I 1
| |
| 8-11
| |
| | |
| Section 9 PUBLIC HEALTH CONSEQUENCE ANALYSIS INTRODUCTION Consequence analysis is used in both the EPRI/WOG study and the NRC Case Study to determine doses for the accidents identified in the integral plant model. Details on the EPRl/WOG approach can be found in Appendix C. De likelihood and mode of containment failure is determined for each accident I
| |
| type. For each accident type and mode of containment failure, the amount of radioactivity released is determined. Many of the combinations of accident types and containment failure modes result in similar releases. Similar releases are grouped into release categories. The likelihood of accident types and contamment failure modes can be combined to calculate release category frequencies.
| |
| For each release category, calculations are made to determine the dispersion of the radioactivity and the means by which the public can incur dose. Total population dose in terms of person-rem were calcula-
| |
| .ed or estimated for the Point Beach site. These calculations are used to measure public health conse-quences for use in the value-impact analysis. The value impact analysis determines how much offsite dose (in person-rem)is averted if plant safety is improved through modification. Such improvements are a value of the modification. The value of the SDHR is evaluated in this way.
| |
| Three alternative sourt e . . ms were created for this review. The first is derived fmm the results of NRC contractor work contained in BMI 2104 (12) and Draft NUREG-0956 (H). De second is derived from the results of the IDCOR p.ogram And the third attempts to reflect a representative source term for a dry cavity PWR, using the reru!ts of NRC contractors and engineering judgments as to the significance of the dry cavity design. A dry cavity means that water does not cool core debris when it is expelled into the containment. The expected value of accident dose for these three somte terms and the NRC Case Study source terms are as follows: ,
| |
| Expected Dose ]
| |
| (Person-rem / !
| |
| Source Term Reactor Year) {
| |
| j EPRI/WOG 1 (Interpretation of NRC Calculations) 0.08 EPRI/WOG 2 (IDCOR Source Terms) 0.16 EPRI/WOG 3 (Dry Cavity) 0.7 NRC Case Study 1.2 9-1 l
| |
| | |
| All expected doses given in the foregoing table were obtained using the EPR1/WOG accident equencies. Use of the NRC Case Study fnquencies results in an increase of about a factor of 30 for the NRC Case Study source terms, i.e., yielding a total of about 40 person-rem per reactor year. This factor of 30 is just the difference in core-melt frequency reported in section 8.
| |
| The EPRI/WOG source term compares favorably to the source term developed from the two NRC-spon-scred studies. For late contamment failure scenarios, the source terms derived from the NRC-sponsored studies predict slightly lower doses than those derived from IDCOR. For early contamment failure scenarios, the NRC-sponsored studies would predict slightly higher source terms and doses. Overall, the results using the EPRI/WOG or the NRC source terms yield doses within a factor of two.
| |
| The EPRI/WOG study used the IDCOR-derived source term for a base case. EPRI/WOG predicts consequences about a factor of seven lower than the NRC Case Study, and by a factor of two hundred lower when core-melt frequency is included.
| |
| OVERVIEW OF NRC CASE STUDY APPROACH The NRC Case Study consequences analysis employs four steps. First, a containment systems event tree is drawn using the two functions of containment overpressure protection and post accident radioac-vity removal. Second, the sequences developed in the event tree are combined with the timing of the core melt, i.e., early or late, to generate accident types. These accident types are assigned probabilities for five containment failure modes, and the cornbination of type and containment failure mode are assigned to WASH-1400 (_42) 4 release categories. *ltird, source tenns are developed based on WASH.
| |
| 1400 release categories and Reactor Safety Study Methodology Applications Program (RSSMAP) (10) source term calculations. These upper bound source terms are then reduced by a factor of three to be more consistent with recent source term evaluations. Some of the resulting source terms, e.g.,
| |
| ruthenium, revert to their WASH-1400 levels when the higher RSSMAP-based source terms were divided by a factor of three. Finally, the person-rem for each release category is cal:ulated for a 50-mile limit using the CRAC2 code (11.).
| |
| The consequence analysis is integrated into the value impact analysis in the following manner. The core melt sequence cut sets are combined with the containment systems sequences to generate a probability of each accident sequence type. (These pmbabilities are described for both intemal and extemal events in their corresponding appendices.) The accident sequence type probabilities are then multiplied by their appropriate containment failure mode probabilities (step 2) and assigned to a release category.
| |
| Summmg all of the accident sequence type and containment failure mode pmbabilities for a particular lease category will yield a release category probability. The expected value of person rem for that release category is then multiplied to yield an expected value'of consequences on a per reactor year 9-2
| |
| | |
| basis. This latter value is then multiplied by the effective (i.e., discounted) number of remaining perating years and by $1000 per person rem, to obtain an expected dollar value to be used in a value impact model.
| |
| OVERVIEW OF THE EPRI/WOG APPROACH ,
| |
| The first source term model (EPRI/WOG 1) is based primarily on the published BMI 2104 and the draft NUREG 0956 for PWRs. The model includes early and late containment failure releases for various initiating conditions together with their conditional probabilities, based on the discussion provided in i
| |
| these NRC references. The second is based on the published IDCOR Technical Summary Report (11)
| |
| . and the supporting Task 23 reports (52), and again, source terms and conditional probabilities are determined.
| |
| The third source term analysis addresses the impact of an important Point Beach plant design issue not analyzed by the NRC Case Study or reflected in EPRI/WOG 1 or 2, the two generic source terms. Spe-cifically, the reactor cavity design influences how much core debris exits the cavity after vessel failure and the coolability of that debris. The Point Beach cavity is expected to be dry unless water can be injected into the vessel after it fails. Further, little core debris is expected to be dispersed to flooded areas of the containment, even during a high pressure blowdown of the vessel. A dry cavity will allow
| |
| .ignificant core concrete attack, which will in turn generate additional hydrogen and release additional fission products. However, because it is believed that this attack will most strongly impact the contain-ment loading, i.e., through hydrogen burning, only containment failure probabilities are calculated. In general, a dry cavity will increase risk since it increases the likelihood of containment failure due to hydrogen burmng.
| |
| Source terms are translated into public health consequences using the CRAC2 code in the NRC Case Study. This EPRI/WOG study generates doses for its source terms by using tle relationships between source term magnitude and public health consequences presented in NUREG/CR 2239 (53), Siting Study for Nuclear Power Plants.
| |
| 9-3
| |
| | |
| i l
| |
| (
| |
| i Section 10 l 1
| |
| VALUE-IMPACT ANALYSIS l l
| |
| | |
| ==SUMMARY==
| |
| AND CONCLUSIONS The V-I analyses for the EPRI/WOG study and for the NRC Case Study differ considerably. Not only I are the inputs different, but the two studies combine the terms differently when calculating V-1 measures and evaluate plant modifications with different levels of detail. Also the EPRI/WOG study considers one additional plant modification. Despite the differences in methodology and input data (to be 1 described in the following discussion), both the EPRI/WOG and NRC V-I analysis results are similar in )
| |
| that neither shows the dedicated SDHR to be cost-beneficial.
| |
| This section presents the results of the EPRI/WOG V-I analysis, incorporating the core melt frequencies from section 8 and the consequences from section 9. It compares the results and approach to those of the NRC Case Study.
| |
| Table 10-1 contains the V-I ratios calculated for this study, using averted offsite costs only. Within the EPRI/WOG ranking, the new seismically-qualified batteries installed by WEP have the highest V-1 ratio, albeit still much smaller than unity. Seismic modifications 2-6 proposed by the NRC Case Study are a close second. All other modifications have a V-I ratio lower by nearly two orders of magnitude.
| |
| Table 10-2 presents the ratios for offsite and onsite costs. Table 10-3 provides a shon description of each of the modifications. (More detailed descriptions can be found in section 2 of this repon and appendix J of the NRC Case Study.) The new batteries continue to have the highest V-I ratio. Seismic modifications 2-6 drop slightly in ranking to third place when the offsite and onsite costs are combined.
| |
| The relative rankmgs of the other modifications change significantly. For example, NRC Case Study internal modification 1 is the second-highest ranked modification, as compared to ninth when only offsite averted costs are considered. The change in rankings results if the modification changes the core-melt frequency more than the consequences (i.e., if the modification changes a sequence with containment safeguards available).
| |
| Many factors combine to yield these greatly reduced V-I ratios for offsite avened costs only: core melt equency was lower by a factor of about 30, WEP's cost estimates were roughly 50-400% higher (table 10-4), and offsite public dose was lower by about a factor of seven. When onsite avened costs are 10-1
| |
| | |
| a es p -
| |
| mio 33 h iee r u
| |
| - 5666667
| |
| - - - 8- - - -
| |
| t it a EEEEEEEEEEEE ne la
| |
| /eR t u ot av y l 4 0.05 7. 3 7 5 l. 9. 8 3 dbr r a 1 1 21 4 4 3 22 971 e e a V s h aheig m bt
| |
| * h m
| |
| u ht )
| |
| $ d.dhe e et t S t i a
| |
| e rd(
| |
| oee s u l awc Wter os u u i
| |
| _ l a
| |
| 3331 1
| |
| + + + + + 01 + + + - -
| |
| 1 1 2 + 0 -2 i s cd edn vD l
| |
| _ t
| |
| _ V n eAe EEEEEEEEEEEE 2 091 4 s at i r
| |
| a cad sfi t 7. 5 9. 4 7.4 6 e sl ul u eosf r
| |
| 5 21 3 1 7543 64 5 y ia nv o P f 5 oew O 3 i t s 1 a et rl
| |
| )
| |
| f of i s es i ca u
| |
| $( e d nr f oo e )o l
| |
| at t c oa 667 6667 75
| |
| + + + + + + + +
| |
| EEEEEEEEEEEE
| |
| +466+ + + i l
| |
| mi e t h
| |
| n scat s t i t i
| |
| fs a t
| |
| Tp m 7. 9. 5 0. 1 1 181 15 9. 6 0. 7 2 a ii fR 3 1 923 75 4 l p hf ie Otc I g f t id r ote ga np n o mta i y in ur b e duIm r l
| |
| cnu/e a
| |
| m ht r
| |
| ola&*
| |
| M iamvt lah ou et e e ohit i l s a m
| |
| u Wla^ t t sd . .
| |
| ( h. 4 5
| |
| + +
| |
| 3
| |
| +
| |
| 4
| |
| +
| |
| r d h l'
| |
| l t V ~ S nnn s EE E E e '
| |
| I l A w 1 uy t c eI ao 69 . . . . . 0 . 0 t n . k sb sf 0
| |
| e a p e r o o nD 01 4 0000002 01 u P s Rd e o c E s. e i r
| |
| 1 P i l
| |
| e t ck I m t s
| |
| id yt r Wih ea c b an pa a ht
| |
| 'T mR ( r n A btaBe db I
| |
| y ol ai$ o) .
| |
| le en t
| |
| Wtola el un t (
| |
| s 667 6667 7 54 66
| |
| + + + + + + + + + + + +
| |
| ml e ah io t l
| |
| aO nTl t
| |
| as t
| |
| o EEEEEEEEEEEE r n
| |
| t s
| |
| nt P
| |
| Vs t ef t s oI nC s 7. 99235 01 8 5 968654 72 oiunoo e 3 1 1 1 1 1 Gos P r s r
| |
| e dh nt d ey .
| |
| OC p a i ws c Wd
| |
| /e r d e ns ekbe an l t r n p Re i t l y) 0 irgid eq Pv necr na 5668 0 s t EA o iMe uye - - 77778
| |
| - - - - - - 0 ehar f t
| |
| EEEEEEEEEEEE 1 dcu c
| |
| ureqr l. 8 3 3 01 4 6607 9 $ ne alalv doe p e r 1 28 421 5 73 74 5 g oBem n i ReCF(
| |
| t i
| |
| s fcn at ne er
| |
| )
| |
| s u ii oeo e s i r dP bc r 6 i
| |
| a of i
| |
| f im e t t
| |
| e -
| |
| a 2 1
| |
| 21 91 41 8 l
| |
| l o
| |
| d at nRg moM hs da c l l l l l cl oN BiR 21 mmrn a a a a ai a mmmm o r eeHa Mn wm Heee e e e ei e t
| |
| Cio ei NeOiin s
| |
| r r t t t n n n et n t t s
| |
| de amDh s s In I I I SI iee sS d t
| |
| Rt a ( SSFFI r e
| |
| Nc v r esise n n t t aia t o a f c BGI io Gno w O.t e
| |
| r Oir e t a eWdr et Wcab 28 1 092651 374 s e N/l ac l l
| |
| / i m
| |
| 1 1 1 s eRt p Do T hPsnm f
| |
| I i u Rd P oN Eii EM
| |
| * 8
| |
| | |
| - 222223
| |
| - - - - 4- 4 - -444 es -
| |
| mie - - - ht iu ee I t
| |
| - a EEEEEEEEEEEE 6 r
| |
| : 2. 5 3 8 04 3 3 eR 7 l. l. n e a l
| |
| l u 6532 1 21 98 531 ot av t
| |
| a .
| |
| dbr V e e se h btah ig
| |
| ) h
| |
| $ 5 3435 53 4 4332 dhet t e l (
| |
| a + * + + + + + + + + + +
| |
| t e ou EEEEEEEEEEEE 5 5 6 6.0 9.5 8 2038 l awc t ia u
| |
| Tla 2861 1 1 21 1 1 1 4 cd e id n V -
| |
| l at i cad ul y sl u r d ia o a t e )
| |
| nv oew m r 5343 ee$ + + +5 +53+ 44232 i
| |
| s m vi t ( + + + + + ++ t arlet S
| |
| u Ant s s s EEEEEEEEEEEE -
| |
| i ca us e l aOoC 4 02 6 0. 8 4 72 535 f is e te )o u t 28 61 1 1 2 1 1 91 4 dnr oo i i l o fsta fR V.
| |
| a T mie tat h
| |
| Otc ht ii sc s ht a rd o e *)
| |
| hfie t
| |
| idre op 423 1 3 3 1 221 1 1 Bm Wtr$ e( + + + + + + + + + + + + f o ot g/
| |
| I t n ve EEEEEEEEEEEE e mt a n e eAs o 3 5 9. 9 3 7. 2 7 3 0. l. 9 ur b i u sf 1 23 4 3 78 54 642 l et dl u Va eoD r
| |
| ah vt u o l
| |
| c P e oh ny hl it I b T l A w 2- s d t
| |
| ) -
| |
| . k 0
| |
| l e $(
| |
| 65646767 7 666 P s suk l at + + + +++++++++ E s. i r e n 1
| |
| EEEEEEEEEEEE e i
| |
| e n RaR t c Tp oa 7. 6. 9 0 7 5 8 95243021 Wihyt ea r c t
| |
| a -m 3 1 1 75 9l l 1 bt aB e T c a (s t I ~
| |
| ps . edb t mo C
| |
| * ml l en i
| |
| y ht d'n ) e ah io r t t ed r a$( - s P ue r a o ne ; n nt un l t 43 5 4 o Wos i
| |
| s dnthoo ar e m + + + +
| |
| Vv m t t i o EE E E r e d .
| |
| GA u naD .60 9 0 . p a i ey Oe S el s aM l .
| |
| 001 204 00001 0 r d wsa c.r t
| |
| et e ekbe Wi
| |
| / n s
| |
| t c
| |
| a Pn&O r s 0
| |
| p ns gir de qu I p I i s e RO )
| |
| 0 0 eh t
| |
| dcuar Pd I m h tl $ 1 f
| |
| En a rl (
| |
| oa s 6 $ ne alal t
| |
| e Wtsn os + 56467677666 g n oBvem t
| |
| E + + + + + + + + + + +
| |
| EEEEEEEEEEE i
| |
| t -
| |
| t I nl C 7698 75 8 95 021 i s at ne ea u fcn er st eoo n 3
| |
| * 1 1 6591 1 1 243 s i ii r dP bc oeo a
| |
| PrTi a
| |
| )
| |
| t s l of dn i e o mo hs a l
| |
| 6 d at nRg e i
| |
| - r i e a - 1 o r f t n
| |
| im da t i a 241 291 8 t e eH a Blai claic e samDh l 1 l l d
| |
| oN a a a a e sesS dc Mn wmmmmRmmm2 s s He e e eee m1 t r
| |
| e i rsen Cio Rt a Nte
| |
| ( I SI n ei e n t SSIeI I eiDt n n t
| |
| ni t
| |
| ni r
| |
| FI F t r v esh a n t at t
| |
| Nc o a f s c
| |
| e BGI o i n r w O.t a Go a eWdr et Oir t e 2 1 8 3 71 25 6049 s e Nl/llca a Wab
| |
| /iiuc mf 1 1 1 s eRt p o hPsnm l
| |
| Rd D'Eii I PoN EM *
| |
| * 5w
| |
| | |
| n
| |
| _ i o
| |
| t t
| |
| a n
| |
| e m
| |
| t mr e s u
| |
| t s
| |
| y I n
| |
| S d l n p l e v a a m n o s Pa n l
| |
| u me o s
| |
| e P e s i t
| |
| s n
| |
| i e
| |
| r r
| |
| e n C R ie D D ts r
| |
| nse e n t t
| |
| t a e g
| |
| o d e ao r t p xo t
| |
| m o a wk a i
| |
| c t n p E it B d n r H f n
| |
| g ip r e a - o e a t y
| |
| s u l a e S a l
| |
| is s
| |
| r c t o Fe e T e v cnh PmB r c e r s W i
| |
| t s a y g i
| |
| e e A e r e
| |
| t a t u
| |
| n e r ar n A e r g D r a a d t
| |
| e D v a l v t e l a
| |
| i l
| |
| r t o e ne F h i a n B hx e r
| |
| mG i t i it o V p ux S t
| |
| w1 o yr E r hS e A m mC a - r l h a l o y d see R mA S l p
| |
| m o r e u g oa ou er t
| |
| A u n sa Ti t
| |
| l t te i S uq R iD R RP evr end t
| |
| t l e t h e r t A WE p a S a ne u c t
| |
| v n R a umB e ed C e r t i 3- d e d n I l n L e a I c GS t
| |
| 0 1
| |
| a s T ac d t
| |
| S i u ne er R
| |
| D Ci l
| |
| o cmtir Wcc P t a a oic t
| |
| i me l e
| |
| ie ws c e e s d s se k Wd e ed ap l a
| |
| R DR S DNSe elEe AF R l
| |
| m i e e e i t
| |
| b a
| |
| N a DS iDIn T
| |
| n o
| |
| it a
| |
| i c
| |
| f i
| |
| d o
| |
| M f -
| |
| o -
| |
| y er p r
| |
| a s eut al o m
| |
| m car Cy G 1 1 1 '3 3 2 1 1 1 1 4 /3 A i N
| |
| S u Rde uv NtSa i t
| |
| n s
| |
| 6 i e
| |
| ie - r f 1 e
| |
| sdm ei ao a l 1 2 4 R 9 l l l l 1
| |
| l 1 2 c c t
| |
| t a
| |
| C N n n mmmmmm1 a a a a a a i i s s e 2 e R HB C Myon d y r r wn a
| |
| _ e e e e e e i i r R di S eS F F DSe NWS r r Nua t t
| |
| n n n n n n e t t t t I
| |
| t i i i
| |
| p
| |
| , I I I I I t
| |
| Sc
| |
| _ Gon Oir t e Wcab
| |
| /im f 1 2 3 4 5 6 7 8 9 0 1 1
| |
| 1 2 3 4 1 1 1 l
| |
| iu Rd P oN EM f
| |
| l
| |
| | |
| y e/
| |
| s 2 e
| |
| - om 0 0 3 2 s r o e Dre 0 0 0 0 0 0 0 1 0 0 0 0 0 0 Dp s* M n-
| |
| &sr o M%
| |
| , te *s Oep &5 t
| |
| _ aes Oa imo D
| |
| ( d d nt e
| |
| _ e t
| |
| s s an Ed o) nu yan Dme oo d nr it c s
| |
| _ ut s o-n 0 0 2 0 0 0 0 5 0 0 6 0 0 08 1 1 lai t
| |
| s .
| |
| d So i a o t 4 l a s seC l l
| |
| s r
| |
| t s a r ae n e aM s (p I
| |
| C t
| |
| y
| |
| - & n e I h 3
| |
| _ C R O t 2
| |
| ~d ,
| |
| Nrfo 2 0 0 7 7 7 2 3 0 0 7 0 0 0 n e.
| |
| a i
| |
| ) 5 6 6 3 5 3 3 t ,
| |
| M($ 8, 6, 6, 0, 8, 3, 0, s s ora
| |
| &ts 1 6 6 7 2 01 3 1 7 9 7 Ce
| |
| . OoC 3 My
| |
| &ev
| |
| - i s
| |
| s y Oic t l
| |
| a eef n . h f t
| |
| e A de y ht 5 t
| |
| c ) Ct uta 0 0 00 0 00 0 0 0 0 00 0 00 00 00 /50 00 A 0 0 iw3 a
| |
| p $
| |
| ( RSm i 5, 0, 0, 0, 0, 9 0 7 5 4 3 0 6 0, 0, 0, 0, 0, 0, 7
| |
| N 7, 0, 5 8 e y 1
| |
| 4- I m dt ns Net s s 4 57 2 4 0, 6 4 6 7 4 8 A 1 1 1 7 t
| |
| ab
| |
| _ e ao C aE 8, 7, 5, 5 3, 9 f9 1 md
| |
| : d. l u nC og 2 5 6 3 1 5
| |
| iet si l
| |
| ,. a Ep it n b V lair t slt i
| |
| a r e ou T ee l
| |
| f o a t
| |
| t a Cm s n m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 a ni g 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Pe Ea r t i I t a n s 0, 0, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, D E E 5 0 7 0 0 0 0 0 0 5 0 0 0 0 Ws e
| |
| *t P 3 0 6 2 0 0 3 0 8 7 0 9 3 9 es c E 1 8, 8, 0, 5, 6, 8, 0, 9, 0, 6, 7 h o a ~l 3 8 4 5 3 0 3 t d
| |
| p W 1 1 1 1 9 gd n na I
| |
| m i
| |
| # m s er p s eu mts al t uo sc CA m A y Cy G 3 3 2 4 /3 bM 1 I 1 1 1 1 1 1 Rde N d&
| |
| NtSa uv i t
| |
| e nO
| |
| : i. a n
| |
| s bmt oe r i
| |
| f e 6 i e e n -
| |
| ei r sdm a o a 1 2 4 8 9 1
| |
| 1 1 2 t
| |
| t e bo ns a r l l l l l l c c a c p e CMN a a a a a a mmmmmmmm1 e e e e e e s s 2 e R eHBwn a i i C yon d y ts r R di i i r ce ap S F F DSe NWS r r
| |
| _ Nua c t t
| |
| I t
| |
| n n n n n n S I
| |
| t I
| |
| t I
| |
| t I
| |
| t I
| |
| t e e i i i
| |
| p p
| |
| 0 S
| |
| i m0 0 G1$
| |
| _ Gon Oir t e Ot Wa
| |
| /d Wab
| |
| / i I
| |
| f c
| |
| m 1
| |
| 2 3 4 5 6 7 8 9 0 1 1
| |
| 1 2 3 4 1 1 1 l e Ru r Plae
| |
| - Rdi u Evy E PoN EM
| |
| _ 5"
| |
| | |
| included, the differences betweer. the NRC Case Study and EPRI/WOG study V I ratios are more diffi-uit to discern due to differences in method of calculation. The example problem later in this section
| |
| , vill provide further insight. For the principal modification considered by A-45, namely the SDHR, the I V I ratios were 2.0E-5 for avened offsite cost only, and 2.0E 3 for both offsite and onsite costs. Such low V-I ratios make uncenainty irrelevant; hence, it is not included S this assessment.
| |
| No modification was found to be cost-effective considering only avened doses at $1,000 per person-rem. That is, further modification of the plant solely for public health and safety is not cost beneficial. !
| |
| When onsite avened costs are considered, V-I ratios increase dramatically, but not sufficiently to make )
| |
| any modification cost beneficial. These two types of V-I caletlations clearly indicate that the most sig- f nificant benefit is to be attained by reducing utility economic risk. Onsite averted costs are about a ;I factor cf 20 to 30 higher than the present wonh of avened dose, as based on the results of the EPRl/
| |
| WOG study, and would be about a factor of seven higher, based on NRC Case Study consequences )
| |
| (based on tables 10-1 and 10-2, and the NRC Case Study results shown in tables D-1 through D 8 in appendix D).
| |
| Three other conclusions were drawn by examining these tables. First, a different V-I measure could be l
| |
| obtained by comparing installation and O&M doses to total averted offsite dose (table 10-1). Installa- '
| |
| don and O&M doses are much higher than the expected avened offsite dose for every modification that
| |
| .is a " measurable" O&M dose. For the SDHR, EPRI/WOG calculates the ratio of these two doses to H nearly 150. That is, the SDHR is expected to " cost" 150 times as much dose to install than it is projec-ted to save in offsite exposure by preventing accidents. If onsite avened doses are included in the savings, the SDHR would cost nearly 40 times more. 'Ihe NRC Case Study V-I methodology does not include installation and O&M doses whea considering only public health risk (i.e., avened offsite dose).
| |
| Since the installation and O&M doses would result if the modification is made, whether or not it actu-ally avens any accident, they are real contributors to the totalimpact of the modification. These doses should be included whether or not avened onsite costs are included as well.
| |
| Second, the difference in the level of detail when considering plant modifications can be imponant. The NRC Case Study approach groups large numbers of individual modifications into an " alternative." This process significantly distons the perceived value to impact. Using the NRC Case Study numbers, indi- '
| |
| vidual modifications within alternative gmup 1 differ in their V-I ratio (offsite only) by a factor of roughly 60. Funher, the wind modification was placed in Alternative 3 despite having the fourth highest V-I ratio. Clearly, this aspect of the NRC Case Study appmach obscures the decision making process. l l One final difference in V-I analyses is the methodology for selecting plant modifications. Wisconsin
| |
| :ctric Power added a new battery system after the NRC Case Study models were " frozen." The EPRll WOG study considers the battery system as a utility-initiated plant modification.
| |
| l l 10-6 1
| |
| | |
| l l
| |
| By providing both diesel stan capability and instrumentation power, the new battery system affects a
| |
| 'ide variety of sequences associated with three ongoing r,afety issues:
| |
| o Switchgear room fires (appendix R) o Seismic station blackout sequences (A-46) o Battery common mode during loss of offsite power (A-44)
| |
| The EPRI/WOG analysis calculated a reduction of 1.1E-5 in core-melt frequency. The majority of the reduction resulted from the batteries improving AC power reliability during seismic events. The V-1 ratios for the batteries are:
| |
| o Offsite only - 1.4E-3 o Offsite and Onsite - 6.7E-2 While these ratios are less than 1.0, they are the highest for any of the modifications considered in this study. The principal reason for this ranking is that the WEP-sponsomd change affects three of the more important core-melt sequences analyzed by the EPRI/WOG study, but at less than one-twentieth of the cost of the SDHR. The NRC Case Study developed modifications to address only a single vulnerability.
| |
| he following parts of this section provide an overview of the hTC Case Study results and approach and an example calculation using the EPRI/WOG approach. During the example calculation, comparisons to the NRC Case Study are made and sensitivity studies reponed.
| |
| OVERVIEW OF NRC CASE STUDY RESULTS AND APPROACH The NRC Case Study approach calculated V-I ratios (using its input data and methodology) ranging from 18-0.4% for a V-I ratio including avened offsite doses, avested onsite costs and doses, and installation and O&M doses, and from 0.5-0.7% for avened offsite doses only, for alternative 1 and the SDHR system, respectively. Modifications to the plant were identified after reviewing the core. melt risk results (summarized in section 2 of this report). The analysts first identified vulnerabilities and then potential modulations to limit the significance of those vulnerabilities. These modifications were then grouped into three cumulative sets of modifications called " alternatives," and a fourth alternative, the SDHRS.
| |
| Each of the four alternatives identified was evaluated using a methodology patterned after NUREG/CR-3568 (51), NRC office Letter 16 (14. ), and NUREG/CR-3971 (51). In the NRC Case Study methodol-ogy, the values are the person rem averted by the accident, including the avened offsite dose and the verted onsite dose to cleanup after core damage. Any radiation exposure from installing or operating 10-7
| |
| | |
| the modifications was considered a " negative" value. The value of each alternative is equal to the net
| |
| 'fect of all modifications grouped in that alternative.
| |
| The impacts of the alternative are the engineering, installation and O&M costs, as well as any replace-ment power costs necessitated by the installation. Negative impacts include averted onsite costs such as replacement power, loss of investment, and cleanup costs associated with accident prevention. Whether or not onsite averted costs (and onsite averted doses) should be included in regulatory decision making is a topic of ongoing debate.
| |
| The installation and engineering as well as O&M cost impacts for the four alternatives were calculated in the NRC Case Study to be (in 1985 dolla s):
| |
| Altemative S in Millions 1 7.5 2 14.9 3 24.8 4 64.1 The avened onsite costs included replacement power, loss of remaining value of the utility's investment, and cost of plant and site cleanup. Replacement power costs were calculated to be $293 million based a an annual cost of $38 million for a replacement period of 10 years and a 5% discount rate. The loss of capital investment from core damage was estimated to be S429 million in 1985, and to decrease by 1/23 for each of the remaming 23 years of plant life. The cost of plant and site cleanup was estimated to be $1.2 billion. A tottl cost of $1.92 billion per accident is calculated. Using a 5% discount rate and a 23-year plant life, the expected present value of onsite averted costs from an accident is roughly one billion dollars.
| |
| The NRC Case Study and the EPRI/WOG study both use a 5% discount rate. However, NUREG/CR-3568 recommends that a 10% discount rate be used as a base case and 5% in a sensitivity study. EPRI/
| |
| WOG chose to use the same value as the NRC Case Study because: (1) 10% may be too high for constant dollar calculations, i.e., without inflation; (2) the two studies' results would be easier to compare; and (3) since the EPRI/WOG study did not perform uncertainty analysis, the more conserva-tive value was chosen. The impact of using 10% would be to further reduce the calculated V I measures for both studies.
| |
| EPRI/WOG VALUE-IMPACT ANALYSIS AND EXAMPLE APPLICATION The EPRI/WOG study V-I analysis differs from the NRC Case Study V-I analysis in three ways. First, ere are differences in the input data, e.g., the core melt frequency and source terms differ. They differ 10-8
| |
| | |
| from the NRC Case Stug for the plant, both as it currently exists and as it would after modifications td been implemented.
| |
| Second, the EPRI/WOG study combines the intermediate terms to calculate different V-I measures using a method that is different than used in the NRC Case Study. V-I ratios, net benefit and dollars per thousand person-rem are still calculated; however, the differences in the equations may yield different values. For er. ample, consider the calculation of V-I ratios. The NRC method considers all doses (both
| |
| " good" averted doses and " bad" installation and O&M doses) as " values" and all costs as " impacts."
| |
| Thus, the NRC approach yields a " dose-dollar" ratio; however, this measure is redundant to the dollar per person-rem measure except for differences in discounting. This review's approach presents all dollar costs and doses associated with the installation and operation of the modification as impacts. For example, dose during installation is multiplied by $1000 per person-rem and included as a cost. In the NRC Case Study approach, all doses are kept on the value side of the equation. As a result, installation dose is incorporated as a negative value.
| |
| Third, the EPRI/WOG study evaluates modifications at a more detailed L vel than does the NRC Case Study. For example, the EPRI/WOG study calculates V I measuru individually for each plant modification proposed in the NRC Case Study. The NRC Case Study, on the other hand, grouped these
| |
| ' modifications into " alternatives" and then calculated V-I measures for these altematives.
| |
| The points indicated above are illustrated best by the following explanation of the overall approach to V-I used in the EPRI/WOG study, and a step-by step comparison with the NRC Case Study. To make the illustration more tangible, the SDHR is used an an example.
| |
| .Calqnlation ofImpacts anO'21 pas The previous paragraphs referred to " intermediate terms." The intermedia (e terms, listed below with their NRC Case Study variable names, apply to each modification:
| |
| o Design, fabrication, and installation cost 01) o Occupational dose during installation (V3A) o O&M costs (12) l o Occupational dose during O&M (V4A) o Averted offsite public dote (V2A) l o Averted cost of post-accident site cleanup 053) o Averted occupational dose from post-accident site cleanup (VI A) l l
| |
| 10-9
| |
| | |
| l l o Averted cost of post-accident replacement power (immediately after the accident)
| |
| (15 1) o Avened cost of post-accident replacement power (long-term power generation to replace lost investment in damaged plant) (152)
| |
| A discussion follows on each of these terms, including the equations and input data used in the~
| |
| calculation of each. The variable name used in subsequent equations, tables and units is also presented in each case. The value and basis for the example is presented together with a comparison of EPRI/WOG and NRC Case Study approsches and results.
| |
| 1 l
| |
| {
| |
| II - Desien. Fabrications. and Installation Cost ($1. The NRC Case Study used the Handbook for Cost !
| |
| 1 Estimating as a guideline for estimating modification costs. An architect-engineer making a major con- 1 tribution to the handbook made the specific estimates used in the NRC Case Study. The EPRl/WOG analysis used the results of the NRC Case Study as a starting point for estimating these costs. An exper-ienced cost estimator from WEP reviewed the estimates in the NRC Case Study (appendix J) and adjus-l ted those estimates where appropriate. The basis for adjustments in cost is presented in Appendix D. q Comparison of EPRIlWOG and NRC Case Study. Individual modifications were evaluated by EPR1/
| |
| 'VOG to cost between 50-400% more than estimated in the NRC Case Study. The basis for each increase is presented for each modification in appendix D. Tangible evidence exists in suppon of each of these increases. ' Die NRC Case Study estimated that a dedicated diesel generator battery system would cost $750,000. The actual cost of a new battery system installed at Point Beach and designed for backup capability to start the diesel generators, as well as for providing backup power for half of the plant's critical safety instrumentation, was $3,690.000. The actual cost for installation of a dedicated battery modification, based on the one similar modification available, supports the higher estimate of
| |
| $1,800,000 provided by WEP and used in the EPRI/WOG study.
| |
| The general reasons for cost differences vary depending on the modification. For expensive modifica-tions, most of the differences in cost inchided: (i) failure to consider some design requirements, e.g.,
| |
| seismic, for spec!fic aspects of the modification; (2) failure to account for existing structures and/or buried piping or cabling at the site; and (3) failure to consider costs of iteration between initial design and final installation, especially when construction of supports or structures end excavation were invol- j ved. For the inexpensive modifications, one important difference involved the fixed cost of paperwork j of $10,000 for any modification. l l
| |
| Example application. The SDHR was estimated by WEP to cost $90 million. The construction of the !
| |
| DHR was estimated to take four refueling outages if no planj downtime was desired. (These values i l
| |
| i 10-10 !
| |
| | |
| i
| |
| '1 l
| |
| compare with $59 million and two outages given in the NRC Case Study.) The value for 11 in the EPRl/ {
| |
| OG study is then 59E+7.
| |
| I Sensitivity studies. He effect of spreading the cost over four outages possibly totaling up to four and j one-half years was not accounted for in the EPRI/WOG study. Discounting at 5% over this period would decrease the present value of the cost of the modification (II), but only by about 77c if one assumes equal expenditures and the first outage at time zero. (This effect would also decrease the overall benefit of the modification since no safety benefit could be credited until the system was [
| |
| installed.)
| |
| The effect of retainir g the NRC Case Study two-outage installation, but adding an additional 120 days of replacement power, can also be estimated. Spreading costs over only 18 months would reduce their present value by about 4%. The increased cost of replacement power would increase the modification by about $17.9 million, assuming again that costs were incurred in two installments 18 months apart.
| |
| The present value of replacement power is obtained by calculating a daily cost of $160,000. The daily )
| |
| cost is based on an annual cost estimate of $38 million from the NRC Case Study and NUREG/CR-4012 (16), a 5% discount rate and a 65c5 capacity factor. The resulting present value of the cost of the SDHR (11) would increase to $102 million, or by about 137c.
| |
| Note that the NRC Case Study refers to installation replacement power costs as variable 13, and O&M replacement power costs as variable I4. The NRC Case Study determined that these values were zero in every case. The EPRI/WOG study assumed, as did the NRC Case Study, that work would be performed
| |
| .during normal outages. The previous paragraph indicated the sensitivity of this assumption for the example application. EPRI/WOG simplified its methodology by not using a separate variable for replacernent power costs.
| |
| V3A - Occupational Dose Durine Installation (oerson rtin). He EPRI/WOG study attempted to use the same values for occupational dose during installation as the NRC Case Study. The NRC Cast Study estimates were developed in their appendix J, using the guidelines provided in the Handbook for Cost Estimating. Appendix J was used to infer the occupational dose for each individual modification, based on the discussion of the alternatives. No specific mealculation of doses was made by the EPRl/WOG study.
| |
| Only one situation was identified where the cost review by WEP identified a difference in assessment of I radiation conditions for the working environment, i.e., Seismic 1, the altemative to the RWST. The dif-rence in dese was not estimated. In general, the installation dose was found not to be significant to the verall V-I calculation. The installation dose evaluated at $1K per person-rem is much less than the l installation cost of any proposed Point Beach modification.
| |
| l 10-11 L _ _ _ _ _ _ _ _ _ _ _ _
| |
| | |
| i l
| |
| Example application. The NRC Case Study estimated that the SDHR would require 486 person-rem of )
| |
| cupational dose to install, resulting from in-containment work to connect the system to the existing l teedwater and primary systems. j 12 - Operation and Maintenance Costs (5/vear of olant life). The EPRI/WOG study used the same val- l ues for O&M costs as did the NRC Case Study. The NRC Case Study estimates were developed in their appendix J, using the guidelines provided in the Handbook for Cost Estimating. No specific attempt to {
| |
| reassess these values was made. The largest contribution to overall cost frorn this term was 3E Example application. The NRC Case Study estimated that the SDHR required about $380,000 per year for O&M costs.
| |
| V4A- Occupational Dose Durine Operations and Maintenance (oerson-rem /vear of clant life). The EPRl/WOG study attempted to use the same values for occupational dose during O&M as used in the NRC Case Study. The NRC Case Study developed estimates in their appendix J, using the guidelines provided in the Handbook for Cost Estimating. Appendix J was used to infer the occupational dose for )
| |
| each individual modification, based on the discussion of the tc.tals for each alternative. No specific recalculation of doses was made by the EPRI/WOG study. This term contributed even less to the verall V-I analysis results than the installation doses did. j Example application. No estimate was provided by the NRC Case Study for O&M doses for the SDHR.
| |
| 1 V2A - Averted Offsite Public Dose (oerson-rem). Averted offsite public dose was calculated using the l EPRI/WOG model of Point Beach. A base case dose was calculated using the results of the existing plant model. A revised dose was calculated by esthnating the effect of the modification on plant risk. l The averted dose, then, was the difference between the two. j 1
| |
| l The EPRI/WOG model base case results for accident sequence frequencies were reported in section 8.
| |
| Consequences for each accident sequence wen: developed from the information reportedin appendix C.
| |
| By combining the base case accident sequence frequencies with the corresponding consequences, a base case dose for each year of plant operation was calculated. 'Ihe total base case dose was obtained by multiplying this base case dose by the number of years of remaming plant operation.
| |
| f The effect of each modification on averted dose was obtained by adjusting the frequency of each l accident sequence impacted by the modification and multiplying the new frequencies by their corres-l anding consequences. The new accident sequence frequencies for the EPRI/WOG assessment were l
| |
| 10-12
| |
| | |
| obtained in different ways for the SDHR, the other NRC Case Study modifications, and for the new
| |
| 'ation batteries.
| |
| For the SDHR, the new accident sequence frequencies were calculated by assessing the effectiveness of the system in detail. The bases for these calculations and their results are reported in section 7 and appendix E.
| |
| The new accident sequence frequencies for all other modifications contained in the NRC Case Study were determined by adjusting the accident sequence frequencies in the same manner as reported in the NRC Case Study. For example, the probability of operator failure to switch to recirculation was reduced by a factor of ten to assess the impact of installing new instrumentation for RWST level. That is, the EPRI/WOG study did not perform a separate human reliability analysis or other analysis on each modifkation.
| |
| The new accident sequence frequencies for the batteries installed by WEP were determined in a slightly different manner. The base case results of the EPR1/WOG study determined the results ofinstalling the system. Hence, the base case model was adjusted by removing the benefits of the new batteries, thereby increasing risk. The averted dose was obtained by subtracting the base case dose from the dose that ,
| |
| 'vould be obtained assuming the batteries were not present.
| |
| Example application. The SDHR results in a total averted dose of 3.3 person-rem over the remaining 23 years of plant life.
| |
| 153 - Averted Cost of Post-Accident Cleanup (S). The EPRI/WOG study has used the same value for post-accident cleanup costs as the NRC Case Study. That value,1.2 billion dollars for a core melt acci-dent, is specified in NRC Office Letter 16. The cost has been assumed to occur in the year of the l accident. Spreading the cost over a few years has only a minor impet on the results. This term repre-sents most of the dollar value of averted onsite costs. ;
| |
| i Exampic application. EPRl/WOG calculated the averted cost of post accident cleanup to be 1.3E+5 dollars for the SDHR. The value is obtained by multiplying the NRC Case Study value of $1.2 billion !
| |
| l l by 13.5 effective years and the change in core melt frequency from SDHR installation, i.e., 8.3E 6. ;
| |
| This assumes instantaneous cleanup. Spreading the cost over five years and discounting at 5% per year !
| |
| yields a present value of $1.lE+5. l i
| |
| VI A - Averted Occupational Dose From Post-Accident Cleanup (person-rem). Again the EPRI/WOG j
| |
| ; dy has used the NRC Case Study value for averted occupational dose from post accident cleanup. ,
| |
| ihe value includes three terms, dose to operating perso'nnel on site at the stan of the accident, dose !
| |
| 1 10-13 i l
| |
| 1 t ___________________________________a _
| |
| | |
| received during cleanup, and the dose to personnel op: rating the undamaged unit after the accident. The stal dose was calculated to be 51,500 from values of 0 for dose to operating personnel (assumes an arly and effective evacuation),40,000 person-rem for cleanup dose (14), and 11,500 person-rem for dose to personnel operating the second unit for the remainder of its life.
| |
| l Example application. The averted occupational dose was calculated to be 9.8 person-rem. The value is obtained by multiplying the NRC Case Study value of 51,500 person rem by 23 years and the change in l
| |
| core-melt frequency, i.e., 8.3E-6.
| |
| 151 - Averted Cost of Post-Accident Replacement Power (Immediate) ($). The EPRI/ WOO study has used the NRC Case Study value forimmediate post-accident replacement power. The NRC Case Study calculated using NUREG/CR-4012 that the annual cost of replacement power for Point Beach would be
| |
| $38 million. It was assumed that a new plant could be constructed in 10 years and replacement power was required in the interim. The average value of replacement power was discounted over the ten years to obtain a value of $293 million for an accident.
| |
| The EPRI/WOG study notes that replacement power costs would be limited to the remaining plant life if an accident were to occur in the last 10 years. That is, the normal replacement plant would be used for
| |
| " replacement power" at the end of normal plant life. This " extra" replacement power in the NRC Case udy is small because it occurs at the end of life, and is discounted significantly. Most of the onsite averted costs come from post-accident cleanup; therefore, this difference is not significant to the final conclusions of either study.
| |
| Example application. The value for 151 for the SDHR was calculated to be 3.3E+4 dollars. This cor-responds to the NRC Case Study value of $2.93E+8 times the 13.5 effective years times 8.3E-6 (change in core-melt frequency).
| |
| 152 - Averted Cost of Post-Accident Replacement Power (Lone Term) ($). The EPRI/WOG and NRC Case Studies use the same method for calculating the remaining plant investment value. While calculat-ing both replacement power costs and lost plant investment over the same ten year period after an accident is double counting, the value of 152 contributes less than 10% to the total onsite averted costs, i.e.,152,151 and 153. The difference, then,is not significant to the conclusions of either study.
| |
| Example application. The value for 152 for the SDHR was calculated to be 1.7E+4 dollars. This value is obtained assuming an NRC Cr.se Study value of $0.25 billion for the replacement plant value.
| |
| 1]culation of Value-Impact Measures The EPRI/WOG study calculated four V-I measures for each modification. The measures were:
| |
| 10-14
| |
| | |
| o V-I ratio (VIRN and VIRO) l o Net benefit (NBVO and NBVh) o Dollars per person-rem (DPRO and DPRN) o Net dose (DIAO and DIAN)
| |
| The study also considered an incremental V-I ratio as a measure for evaluating the SDHR.
| |
| The equations for each of these measures are presented below in terms of the impacts and values men-tiotied previously.
| |
| VIRO and VIRN - Value-Impact Ratios. The equation for the EPRI/WOG V-I ratio depends on whether onsite averted costs are included. The equation when onsite averted costs are act included is:
| |
| VIRO = V2B / 01 + 12 + V3B + V4B) where V2B is the present worth of V2A, V3B is the present worth of V3A, and V4B is the present worth of V4A.
| |
| .he equation when onsite averted costs are included is:
| |
| VIRN = (V2B + 151 + 152 +153 + V1B) / 01 + 12 + V3B + V4B) where VlB is the present worth of VI A.
| |
| Comparison of EPR1/WOG and NRC Case Study. EPRI/WOG classifies a value for a plant modifica-tion as a reduction in health risk to the public or economic risk to the utility. An impact is a cost in dollars or occupational dose required to attain the risk reduction. The above equations reflect that approach.
| |
| The NRC Case Study uses the following equations:
| |
| VIRO = V2B / 01 + 12), and VIRN = (V2B + V1B - V3B - V4B) / 01 + 12 - 151 - 152 - 153) .
| |
| The NRC method considers all doses (both " good" averted doses and "bau' installation and O&M doses) as " values" and all costs as " impacts." Thus, the NRC approach yields a " dose-dollar" ratio;
| |
| > wever, this measure is redundant to the dollar per person-rem measure except for differences in dis-
| |
| .ounting. This review's approach presents all dollar costs and doses associated with the installation and operation of the modification as impacts. For example, dose during installation is mul:iplied by $1000 10-15
| |
| | |
| per person-rem and included as a cost. In the NRC Case Study approach, all doses are kept on the value ide of the equation. As a result, installation dose is incorporated as a negative value. The h1 C dpproach omits installation and maintenance doses when only offsite averted doses are calculated.
| |
| The EPRI/WOG approach allows the reader to more easily compare modification impacts to ave-ted cost and dose values. This representation may have fewer inconsistencies. Por example, consider table 10-4, the summary of EPRI/WOG and NRC Case Study results. The NRC Case Study V-I ratio for the SDHR with avened onsite costs (VIRN) and the V I ratio without avened onsite costs (VIRO) differ only slightly. In fact, VIRN is less than VIRO. The reader could easily conclude that the onsite avened costs were insignificant. However, the onsite avened costs are fifteen times the offsite avened costs (tables D-5 through D-8 in appendix D). In fact, they are the single most significant benefit.
| |
| Example application. The EPRI/WOG study calculates the V-I ratios for the SDHR to be 2.0E-5 for offsite costs only and 2.0E 3 when onsite avened costs are also included.
| |
| NBVO and NBVN - Net Benefit (S). The equations for the EPRI/WOG net benefit measures depend on whether onsite avened costs are included. The equation for net benefit when onsite avened costs are n03 included is:
| |
| NBVO = V2B-Il-12-V3B-V4B .
| |
| The equation for net benefit when onsite avened costs are included is:
| |
| NBVN = V2B + 151 + 152 + 152 + VlB 12 - V3B - V4B.
| |
| Example application. The EPRI/WOG study calculates the net benefit to be -9.5E+7 for both NBVO and NBVN.
| |
| Comparison of EPR1/WOG and NRC Case Study. Because net benefit is obtained by subtracting impacts from values, the differences between positive and negative terms found in the V-I ratio calcula-tion cancel. The only significant difference in the two studies was mentioned previously, namely the NRC Case Study's lack of consideration ofinstallation and maintenance doses when only offsite avened costs are included.
| |
| DPRO and DPRN - Dollars Per Person-Rem ($ per oerson-rem). Again, the equations for the dollars per person rem measure depend or, whether onsite avened costs are included. As with the previous measure, the principal difference between the NRC and EPRI/WOG studies is in accounting for 4 10-16
| |
| | |
| installation and maintenance doses when only offsite costs are considered. The equation for dollars per erson rem when onsite costs are n01 included is:
| |
| DPRO = 01 + 12) / (V2A - V3A - V4A) .
| |
| The equation for dotlars per person-rem when onsite costs are included is:
| |
| DPRN = 01 + 12 - 151 - 152 - 153) / (V2A + VI A - V3A - V4A) .
| |
| While utilizing this methodology raises some of the same concems for the NRC Case Study V-I ratio, EPRI/WOG chose this measure for the.following reasons: First, this musure provides a net dollar versus net dose measure, when the NRC Case Study equation for offsite averted dose is modified to include installation and O&M doses. Second, this measure avoids converting doses to dollars and discounting doses.
| |
| Example application. The EPRI/WOG study calculates the dollar per person rem for the SDHR to be 1 2.9E+7 for DPRO and 2.0E+5 for DPRN.
| |
| DIAO and DIAN - Net Dose (oerson rem). The EPRI/WOG study added a measure which it defined as the net. dose. Net dose is the dose avened by a modification minus the dose incurred from the modification. This measure was added because the analysis indicated that more dose would be incurred by some modifications than would be averted by them. The equation for the net dose when onsite doses are HD.1 included is:
| |
| DIAO = V2A - V3A -V4A .
| |
| The equation for the net dose when onsite doses are included is:
| |
| DIAN = V2A-+ VI A - V3A - V4A .
| |
| Example application. The EPRIf#OG study calculates the net dose for the SDHR to be -483 person-i rem for DIAO and -473 person-rem for DIAN.
| |
| l Incremental Value imanet Ma2=res. Another aspect of V-I analysis that is important to the presentation and interpretation of results is consideration of incremental V-I measures. When a variety of plant modifications are being considered and when some of those modifications have overlapping effects, the installation of one modification may affect the benefit of another modification. An incremental measure can be vsed to measure the effect of subsequent modifications given the previous stallation and safety impact tiothers.
| |
| 10-17
| |
| | |
| I Using the EPRI/WOG results and methods, an incremental ratio was calculated assuming that the sismic 2-6 modification had been implemented. If the SDHR is implemented after seismic 2-6
| |
| . modifications,its V-1 ratio for averted offsite costs drops from 2.0E 5 to 2.5E-6,i.e., nearly an order of magnitude. The V-I ratio for averted offsite and onsite costs drops only slightly from 2.0E-3 to 1.4E-3. I
| |
| ; These results are explained by the nature of the impact of seismic 2-6 on risk. Because the modification improves the reliability of ernergency power after an earthquake,it reduces the frequency of accidents l
| |
| where containment safeguards are also lost. These accidents result in higher public doses. Therefore, i
| |
| the ave.rted offsite dose V-I ratio is affected most significantly. The V-I ratio for averted onsite and l
| |
| offsite costs varies directly with core melt frequency since onsite averted costs dominate this measure. I This example illustrates the significance of incrementally evaluating modifications and emphasizes the underlying weakness of the NRC approach of grouping alternatives. The finer the detail of the V-1 analysis, the more plant modifications can be focused to maximize benefit and the more wasted expenditures can be minimized.
| |
| l Example Application Summary Table 10-5 presents the results for the values, impacts and V-I measures discussed and defined previous-
| |
| * The total r.lso illustrates the root cause of the differences between the two studier. First, the values aiffer between the two studies much more significantly than the impacts. Second, the methodology a difference results in similar V-I ratios for averted onsite and offsite cost despite the difference in inputs.
| |
| This similarity indicates the weakness of the NRC V-I ratio method. Third, the difference in sign in the dollar per person rem estimates illustrates the sensitivity of the dose versus dollar comparison technique. The most significant conclusion, however, is that neither study indicates favorable V-1 measures for the SDHR.
| |
| Selection of Plant Modifications The EPRI/WOG study evaluated all the plant modifications proposed in the NRC Case Study, plus the new batteries which were installed after the NRC Case Study models were " frozen." The EPRI/WOG study also used the NRC Case Study estimates for the reliability and effect from each of the NRC Case Study modifications except for the SDHR (see section 1 10-18
| |
| | |
| l Table 10-5 SDHR Example Application - Comparison of ]
| |
| EPRI/WOG and NRC Case Study Values, 2 Impacts, and Value-Impact Measures (Non Zero Entries Only)
| |
| EPRI/WOG NRC Case Studv Values V2A* 3.3E+0 8.3E+2 4
| |
| V2B 1.9E+3 4.9E+5 VIA 9.8E+0 3.5E+2 V1B 5.8E+3 2.0E+5 15 1.8E+5 7.4E+6 Impacts Il 9.0E+7 5.9E+7 12 5.1E+6 5.1E+6 V3A 4.9E+2 4.9E+2 V3B 4.95 r5 4.9E+5 l
| |
| Value-Impact Measures Value-Impact Ratio VIRO 2.0E-5 7.0E 3 VIRN 2.0E-3 4.0E 3 Ne Benefit NBVO -9.5E+7 -6.4E+7 (Odlars)
| |
| NBVN -9.5E+7 -5.6E+7 Dollars Per DPRO 2.9E+7 7.7E+5 Person-Rem .
| |
| DPRN -2.0E+5 8.2E+5 Net Dose DIAO -4.8E+2 3.4E+2 ,
| |
| DIAN -4.7E+2 6.9E+2
| |
| * Variables defined in text except fer 15, which is the sum of 151,152, and 153.
| |
| 10-19
| |
| | |
| Section 11 REFERENCES
| |
| ' 1. W. R. Crammond, et al. Shutdown Decav Heat Removal Analysis of a Westinghouse 2-Loop Pressurized Water Reactor. Case Studv." Marc.o 1987. NUREG/CR-4458.
| |
| : 2. ' A. Marchese. " Shutdown Decay Heat Removal Requirements." Task Action Plan frAP)- F.B_C
| |
| ]Jnresolved Safety Issues A-45, rev. 3. USNRC, March 1984.
| |
| : 3. N. Rasmussen. Report of the Reactor Safety Research Review Group to the President's Nuclear Safety Oversight Committee. NSOC, September,1981. ,
| |
| : 4. Grouoing of Light Water Reactors for Evaluation of Decay Heat Removal Capabilitv.
| |
| Brookhaven, MA: Brookhaven National Laboratory, June 1984. NUREG/CR-3713. .,
| |
| l 5, D. L. Berry and P. L. Bennett. Study of Alternative Decay Heat Removal Concepts for Light Water Reactors - Current Systems and Proposed Ootions, SAND 80-0929. USNRC, April 1981.
| |
| NUREG/CR-1556.
| |
| i 6. Study of the Value and Impact of Alternate Decay Heat Removal Concepts for Linht Water i Reactors, SAND 82-1796. Sandia National Laboratories, June 1983. NUREG/CR-2883.
| |
| : 7. J. Riley and B. Putney. "The Risk Management Query System." In Proceedings of the Annual Reliability and Maintainability Symposium. Las Vegas, NV: January 1986.
| |
| : 8. Westinghouse Owner's Groun Emergency Resnonse Guidelines - Low-Pressure Version, rev.1,
| |
| ~ ' ~
| |
| September 1,1983. 4
| |
| : 9. A. D. Swain and R E. Guttmann. Handbook of Human Reliability Analysis with Emohasis on Nglpar Power Plant Aeolications, final report. USNRC, August 1983. NUREG/CR-1278.
| |
| 1
| |
| : 10. G. J. Kolb, et al. Interim Reliability Evaluation Program: Analysis of the Arkansas Nuclear One
| |
| - Unit 1 Nuclear Power Plant. USNRC, June 1982. NUREG/CR 2787.
| |
| : 11. Nuclear Power Plant Resoonse to Severe Accidents. technical summary report. Technology for Energy Corporation, November 1984.
| |
| : 12. J. A. Gieseke, et al. Radionuclides Release Under Soetific LWR Accident Conditians, vols. V and VI. Battelle Columbus Laboratories, July 1984. BMb2104.
| |
| : 13. M. Silberberg, et al. Reme=*== ment of the Technical Bases for Estimatine Source Terms, draft report for comrtznt. USNRC, July 1985. NUREG 0956.
| |
| : 14. Nuclear Safety Analysis Center and Duke Power Company. Oconee PRA: A Probabilistic R_igh Assessment of Oconee Unit 3. Palo Alto, CA: Electric Power Researen Institute, June 1984.
| |
| NSAC-60.
| |
| . s. Seabrook Station Probabilistic Safety Assessment. Pickard. Lowe and Garrick,Inc., December 1983.
| |
| 11-1 c _ _ _ - __-____- _-_
| |
| : 16. Risk Assessment of McGuja, Duke Power Company,1984.
| |
| : 7. Risk Assessment of Catawba, Duke Power Company,1986.
| |
| : 18. PRA of Crystal River-3, Florida Power Corporation.1986.
| |
| : 19. United States Department of Energy. Technology for Energy Corporation. Clinch River Breeder Reactor Plant erobabilistic Risk Assessmani. Washington, D.C.: Govemment Printing Office, September 1984.
| |
| : 20. E. M. Dougheny, J. R. Fragola and E. P. Collins. Human Reliability Anr.lvsis. Science Applications Intemational Corporation, April 1986. SAIC/NY-1-86-OR.
| |
| : 21. Zion Pmbabilistic Safety Studv. Commonwealth Edison Company,1981.
| |
| : 22. Indian Point Probabili; tic Safety Study. Power Authority of the State of New York and Consolidated Edison Company of New York, Inc.,1982,
| |
| : 23. Millstone Unit 3 Probabilistic Safety Study. Northeast Utilities Service Company, August 1984.
| |
| : 24. PUN PWR Nuclear Standard Power Plant Probabilistic Safety Study. 1984. WCAP 10590.
| |
| : 25. Development of Transient Initiatine Event Frequencies for Use in Probabilistic Risk Assessment.
| |
| Idaho National Engineering Laboratory,1985. NUREG/CR-3862.
| |
| : 26. H. Wyckoff. Loss of Offsite Power at U. S. Nuclear Power Plants - All Years throuch 1985.
| |
| May 1986. NSAC-103.
| |
| : 7. H. Wyckoff. The Rstiability of Emercency Diesel Generators at U. S. Pt:; lear Power Plants.
| |
| September 1986, hSAC-108.
| |
| : 28. K. N. Fleming, et al. Classification and Analysis of Reactor Ooeratine Exoerience Includinc Dependent Events. Palo Alto, CA: Electric Power Research Institute June 1985. NP-3967.
| |
| : 29. Office for Analysis and Evaluation of Operational Data. QLie Study Rcoort on Loss of Saferv System Function Events. Washington, D.C.: Govemment Printing 0%ce, December 1985.
| |
| AEOD/C504.
| |
| : 30. Interim Reliability Evaluation Procram Procedures Guide. Sandia National Laboratories, SAND 82-1100. January 1983. NUREG/CR-2728.
| |
| : 31. P. W. Baranowsky, et al. A Probabilistic Safety Analysis of DC Power Suppiv Requirements for Nuclear Power Plants. April 1981. NUREG-0666.
| |
| : 32. Evaluation of Station Blackout Accidents at Nuclear Power Plants. Technical Findines Related to Unresolved Safety Issue A-44. In draft, May 1985. NUREG-1032.
| |
| : 33. Summary of Accident Secuence Evaluation Procram (ASEP) Work for Senior Consultant Groun Review. Prelimmary draft, April 13,1984. (Reference taken from (1).)
| |
| l
| |
| : 34. LaSalle Probabilistic Risk Assessment. Sandia National Laboratories. To be published. (Refer-ence taken from (1).)
| |
| ' 5. F. L. Leverenz. ATWS: A Reaooraisal. Pan II: Frmuency of Anticipated Transients. Palo
| |
| ~
| |
| Alto, CA: Electric Power Research Institute, July 1978. NP-801.
| |
| l i
| |
| 11-2
| |
| : 36. A. S. McClymont and B. W. Poehlman. ATWS: A Reanora! sal. Pan III: Frecuency of Antici-pated Transienu. Palo Alto, CA: Electric Power Research Institute, January 1982. hT-2230.
| |
| : 37. H. M. Thomas, " Pipe and Vessel Failure Probability," Reliability Encineerinc, Vol. 2.1981, pp.
| |
| 83-124.
| |
| : 38. Severe Accident Risk Assessment Limerick-Generatine Station, main report, chapt. 4.
| |
| Philadelphia Electric Company, April 1983. Report #4161.
| |
| : 39. D. C. Wood and C. L. Gottshall. Probabilistic Analysis and Operational Data in Resoonse to
| |
| ~ ~
| |
| NUREG-0737, item II.K.3.2. for Westinghouse NSSS plants. February 1981.
| |
| : 40. Clarification of TMI Action Plan Requirements. November 1980. NUREG-0737.
| |
| : 41. Reliability of Emercency AC Power Systems at Nuclear Power Plants. Oak Ridge National Laboratory, July 1983. NUREG/CR-2989.
| |
| : 42. Prairie Island Units 1 & 2 Auxiliary Feedwater System Reliability Study, NSPNAD-8606P, 4 Revision 0, April 1986.
| |
| : 43. N. O. Siu. COMPBRN - A Computer Code for Modeline Compartment Fires. Los Angeles.
| |
| CA: University of California, May 1983. NUREG/CR-3239, UCLA-ENG-8251.
| |
| : 44. Walter Maybee, Department of Energy, Personal cominunication.
| |
| : 45. M. P. Bohn, et al. Aeolication of the SSMRP Methodolecv to the Seismic Risk at the Zion
| |
| ~
| |
| Nuclear Power Plant. Livermore, CA: Lawrence Livermore National laboratory,1983.
| |
| NUREG/CR-3428.
| |
| i
| |
| : 46. P. Yanev, EQE Incorporated. Letter to R. K. Hanneman, Wisconsin Electric Power Company, January 27,1987.
| |
| : 47. C. Stepp, Electric Power Research Institute. Personal communication.
| |
| : 48. G. Neils Northern States Power Company, Letter to D. Ericson of Sandia National Laboratories, June 22,1987.
| |
| : 49. Enttor Safety Study - An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants, NASA 1400. USNRC, October 1975. NUREG-75/014,
| |
| : 50. Reactor Safety Study Methodolon Applications Procram: Calvert Cliffs #2 PWR Power Plent, vol. 3, SAND 80-1897. Sandia National Laboratories, May 1982. NUREG/CR-1659.
| |
| I
| |
| : 51. CRAC2 Model Description, SAND 82-0342. Sandia National Laboratories, March 1984.
| |
| NUREG/CR-2552.
| |
| : 52. IDCOR Task 23.1 - Interrated Containment Analysis. Zion Nuclear Plant. Technology for Energy Corporation December 1984.
| |
| : 53. Technical Guidance for Sitine Criteria Development, SAND 81-1549. Sandia National Laboratories, December 1982. NUREG/CR-2239.
| |
| : 54. H. R. Denton. " Regulatory Analysis Guidelines". In memorandtan: NRR office letter no.16, rev. 2, October 30.1984. -
| |
| 11-3 L_- __- --- - - -
| |
| : 55. J. R. Ball, et al. A Hand %ok for Cost Estimating: A Method for Develonine Estimates of Cost for Generic Actions for Nuclear Power P]arits. October 1984. NUREG/CR-3971.
| |
| : 56. . Replacement Energy Costs for Nuclear Electricity Generatine Units in the U.S. Argonne Nationallaboratory, October 1984. NUREG/CR-4012.
| |
| L l
| |
| 11-4
| |
| | |
| j l
| |
| Appendix A HUMAN RELIABILITY AND RECOVERY ANALYSIS DETAILS i
| |
| This appendix expands upon tae discussion in Section 4 of the EPRI/WOG appsoach and provides additional detail regarchng the pmcedures, implementation time, and other factors which affect operator response. Tables 4-1 and 4-2 are repeated below as Tables A-1 and A-2. Table A-1 is aferred to in the discussion of human failures which are not of the recovery type, e.g., a failure to restore after test or !
| |
| maintenance. Table A-2 is referred to in the discussion of human recovery.
| |
| Tl e review and reevaluation of the NRC Case Study Human Reliability Analysis HRA presented here and in Section 4 is based on a half-day interview with three members of Point Beach's operating staff and numerous follow-up phone calls.
| |
| Key operator events such as feed and bleed and implementation of recirculation are discussed only in lection 4. Other events such as failure to restore equipment after test or maintenance is discussed only-in this appendix. Recovery events are discussed in both Section 4 and in this appendix.
| |
| OVERVIEW AND DISCUSSION OF NRC CASE STUDY HRA The NRC Case Study (1) considered three types of human failure events:
| |
| : 1. Failure to restore equipment following maintenance as a contributing factor in equipment unavailability (denoted by TM or UTM in the fault ID).
| |
| : 2. So called errors in manipulating equipment or actuating systems in re'sponse to an j accident (usually denoted by OE or OPF).
| |
| : 3. Failures in recovery activities (denoted by RA). l 1
| |
| i 1
| |
| The above has become a fairly standard categorization scheme. The NRC Case Study HRA quantified the events in each category of human failure in a generic and often highly conservative manner. The ]I self-explanatory Table A-1 lists the events of types 1 and 2 identified for in:ernal event sequences by the NRC Case Study. Table A-2 lists the events of type 3 identified for internal event sequences. I r !
| |
| he " Failure Mode" column in Table A-1 classifies each failure as to its dominant mode according to ;
| |
| .ne EPRI/WOG classification scheme. The EPRI/WOG classification scheme uses four failure modes, j namely slips and three types of mistakes. nese events are defined as follows:
| |
| i" A-1 r
| |
| | |
| u slip - a failure to execute an intended action.
| |
| Dils - a failure to act in time or correctly according to the new ERG rules.
| |
| 1 diagnosis - a failure to identify the plant state or its response requirements.
| |
| decision - a failure to decide to initiate a diagnosed action.
| |
| l Table A i HUMAN FAILURE EVENTS i
| |
| EPRI/WOG Review Comments NRC Case Study NRC Failure Failure Failure . Analysis Updated Fault ID Description Probability Mads Adeauate Probability
| |
| : 1. VOC UTM MOV not restored after T or M 8.3E-5 slip yes --
| |
| L 2. XOC UTM manual valve not restored 8.3E.5 slip yes --
| |
| dwTwM
| |
| : 3. NOC UTM pneumaticAydraulic valve 8.3E 5 slip yes -
| |
| : 4. NCO-UTM pneumaticAydraulic valve 8.3E 5 slip yes -
| |
| l- i
| |
| : '. PMD-1JTM pneumaticAydranL: vaive 83E-5 slip yes -
| |
| l l
| |
| l
| |
| : 6. XCC-OE operator opens wrong manual SE-3 slip yes -
| |
| valve l
| |
| : 7. VCC OE operator opens wrong MOV 3E 3 slip no IE-4 j
| |
| : 8. HPRF MANACT operator fails to manually 3E 3 rule no IE-4 initian HPR
| |
| : 9. OPF-MACT-FNB operator fails to manually 9E 3 decision no IE-3 initiate feed and bleed
| |
| : 10. OPF-OPEN- operator fails to manually open 3E 3 slip yes FNB PORVs for feed and bleed 11.OPF 51 RESET operator fails to reset SIS signal 3E 3 slip yes
| |
| : 12. OPF.MACT. operatw fails to manaally 3E 3 rule no CSR initiate CSR
| |
| : 13. OPF-MACT. operator fails to manually start 3E-3 degnosis no AFWTP3 the AFW turbine-driven pump The NRC Case Study assumes this to be an independent failure mode, which it is not. The result is double or even triple counting of the same failure. These " modes" should be coalescM into one,i.e., eliminated.
| |
| These events did not appear in dominant intemal event sequences, where expected to be still lower in probability, and hence were not explicitly evaloated in this review.
| |
| j A-2
| |
| | |
| Table A-1 (Continued)
| |
| HUMAN FAILURE EVENTS EPRl/WOG Review Comments NRC Case Study NRC Failure Failure Failure Analysis Updated Fanh1D Desenpuon Probabihty Medt Adequate ProbabihtY 14.OPF MACT. opeator fails to manually stan 3E 3 slip no HPA OPF- HPIpumps for feed and bleed MACT HPB 15.OPF MOPEN- operator fails to manually open 3E 3 slip no MOVA OPF MOV for feed and bleed MOPEN MOVB -
| |
| : 16. SUMP VCC4E operator fails to realign properly 1E-3 rule no IE 4 for recirculation nom the sump
| |
| : 17. DESGENA- diesel generator not restored 8.3E-5 slip yes --
| |
| GEN TM after T or M DESGENB-GEN-TM 18.TM BATTA- batteries nc4 restored after T or 8.3E 5 slip yes --
| |
| DOS TM- M BATTB-DO6
| |
| : d. XOC OE 7 0.0 7 7 --
| |
| : 20. SWSMP-S4- service water pumps not 8.3E5 slip yes -
| |
| UTM restored after T or M Table A-2 NON RECOVERY PROBABILITIES FROM THE NRC CASE STUDY AYailable time in min. un to-Recovery of: ID.; 10 20 $ A0 60 E 12Q 24_Q Ioss Of RA-1 0.7 0.7 0.5 0.4 0.3 0.3 0.2 0.1 )
| |
| Offsite Power '
| |
| Imsa Of Main RA 2 1.0 1.0 1.0 0.9 0.6 0.1 0.03 0.01 Feedwater Battery RA 8 1.0 1.0 1.0 1.0 1.0 1.0 0.9 0.8 Common Cause Bauer/Fault RA-9 1.0 1.0 0.9 0.8 0.7 0.7 0.5 0.2 Diesel RA-10 1.0 1.0 1.0 1.0 0.9 0.9 0.9 0.7 Common Cause l
| |
| l A-3
| |
| | |
| l
| |
| - Table A-2 (Continued)
| |
| NON-RECOVERY PROBABILITIES FROM THE NRC CASE STUDY Available time in min.' un to:
| |
| Recoverv of: Q.1 JD 20 30 AQ 60 20 12Q 24_Q Diesel Fault RA-11 1.0 1.0 1.0 1.0 0.9 0.9 0.9 0.8 Other Failures RA-6 0.3 0.1 0.05 0.03 0.03 0.01 0.01 0.01 In Control Room Other Failures RA-7 1.0 0.3 0.1 0.05 0.03 0.03 0.01 0.01 Out of Control Room An example of a slip is the VOC-UTM failure type, where the correct valve and its correct status is in a known procedure, and the failure amounts to an uncormeted error in carrying out the procedure. An ex-ample of a rule failure is the failure to initiate ECCS recirculation, which is a specific symptom action rule in the LOCA ERG (and other places). An example of a diagnosis failure is the failure to initiate HPIin those LOCAs so small that SIS is not initiated. An example of a decision failure is the failure to initiate feed and bleed due to a wait to restore secondary cooling.
| |
| All NRC Case Study type 1 human failure events were classified by EPRI/WOG as slip mode, while type 2 events were classified as either slip or any of the 3 mistake failure modes, and type 3 as any of the 3 mistake failure modes. The following discusses test and maintenance errors, event response errors, and recovery errors. The remaining errors in Table A 1 are discussed in Section 4. The Case Study approach is discussed followed by presentation of an EPRI/WOG result.
| |
| QUANTIFYING TEST AND MAINTENANCE HUMAN FAILURE EVENTS (NRC CASE STUDY TYPE 1)
| |
| Human feilure events that occur in normal operations prior to the accident initiator, e.g., VOC-UTM, were quantified as a contribution to the unavailability of a piece of equipment, i.e., added to the expec-ted downtime for the component. Events 1-5,17-18, and 20 in Table A-1 are such events. A single model was used to quantify this contribution. Within this event the following assumptions were made:
| |
| component restoration wcs assumed to occur once per year; the component was checked monthly; the check was assEmed to be perfect, i.e., the component was assumed to be successfully restored if found in an undesired state; and a check always finds the fault.
| |
| A basic human failure probability was taken from NUREG/CR-1278, table 20-7, item 1. This failure robability is 0.001 and applies to an error made following a shon list procedure with (2) checkoff pro-vision. According to the NUREG, the probability ranges from 0.003 to 0.0003.
| |
| A-4
| |
| | |
| The unavailability contribution is cornputed as the expected downtime for a year (in months) divided by 2 months / year. For the data above, this unavailability works out to be:
| |
| (1 maintenance act/vr) x (1E-3 failure /act) x (1 mo downtime / failure) 12 mo/yr or 8.3E 5. The value for basic events (those ending with "-UTM") was then obtained by adding the de-rived unavailability value to the expected downtime due to test and to maintenance for reasons other than restoration failure. Component that included this failure mode were valves, emergency diesel gen-erators, and the Service Water (SW> purnps. No consideration was given to the potential for common-cause failures of multiple train or multiple system equipment due to failures in test and maintenance procedures and practices.
| |
| The NRC Case Study notes that many PRAs have assumed that it is not necessary to model restoration failures explicitly. In reality, the restoration failure contribution is already incorporated in the data used to generate component failure rates and unavailabilities. Its inclusion may therefore constitute double counting. Nevertheless, the magnitude of this contribtnion is judged to be insignificant.
| |
| On the other hand, the lack of events intended to handle multiple equipment failures (i.e., human initi-ted common causes) by test and maintenance may constitute a significant problem. To assess the actual degree of significance, however, would require a review of all test and maintenance procedures, which is beyond the scope of this critique. Furthermore, recovery will usually make these common.
| |
| cause failures insignificant to the final results, even though they may be significant before recovery is considered.
| |
| QUANTIFYING EVENT RESPONSE HUMAN FAILURE EVENTS (NRC CASE STUDY TYPE 2)
| |
| Human failures in responding to an event are basically of two types:
| |
| : 1. Those in which the activity was not performed successfully within the available timeframe because of hesitancy, uncertainty, too little available time.
| |
| : 2. Those in which the wrong equipment is manipulated, either inadvertently or intentionally (as the irsult of a nusdiagnosis).
| |
| The latter type of failure is often called a commission error, and HRA is notably lacking in a quantification technique. Oconee Q) and Seabrook (4) used a confusion matrix concept in an effort to quantify the commissions that are based on misdiagnosis. But the matter becomes almost moot when l
| |
| i considering that the new Westinghouse ERGS should render these types of failures very rare.
| |
| Furthermore, the inadvertent commissions are usually correctable either by the perpetrator, another l perator, or the response team. The NRC Case Study did identify one (apparent) commission failure, XOC-OE, item 19 in T able A-1, the closing of a normally open valve, but assigned 0 as its probability.
| |
| A-5
| |
| | |
| l The other type of OE event was quantified by using NUREG/CR 1278, Tables 20-7 and 20-13. The 1C Case Study should have used Table 20-1 for this event, since this table was desigr.ed to account ror sotne types of diagnosis. The NRC Case Study ignored the potential decision problems associated with going to feed and bleed following a loss of all feedwater, OPF-MACT-FNB. This event is quantified with a probability of 0.003, using Table 20-7, item 1, of the NUREG, as are most of the remainder of the OE events. The event represented by improper alignment of the LPR pumps' suction l
| |
| to the sump at the low RWST alarm is quar.tified with a probability of 0.001, which is the lower bound for the 0.003 value. OE events involving manual valves are quantified with a probability of 0.005, using Table 20-7.
| |
| Even without considering the magnitude of the quantities derived for these events, such a generic assignment is not credible. The OE events are highly dependent on the characteristics of the sequences in which they are postulated to occur. Fer example, operator initiation of feed and bleed inay differ by up to two orders of magnitude between specific plants. As a result an interview of the Point Beach operations personnel and a review of the emergency procedures was conducted to elicit the qualitative information necessary to quantify these OE events.
| |
| QUANTIFYING RECOVERY EVENTS (NRC CASE STUDY TYPE 3) he NRC Case Study also identified and quantified human failure events in performing recovery, def'med as an activity performed in the restoration of failed equipment or the use of altemative equipment to achieve similar function. The NRC Case Study generally did not credit the restoration of failed or unavailable equipment, with the exception of AC and DC buses. The use of non-safety or other alternative equipment to substitute for failed equipment was considered on a case-by-case basis. Table A-2 lists the kind of recovery events that were assigned non Ircovery probabilities. The electrical equipment recovery model is from NUREG/CR-3226 (5.) and me non electrical is from NUREG/CR-2787 (6).
| |
| Time available to recovery is the only determining parameter in these models.
| |
| This general approach for recovery HRAs is standard. As in the case for the OE events, however, the NRC Case Study numbers are derived without much sequence context, available time being the only exception. In this EPRI/WOG study, all recoveries were reassessed based on plant- and sequence-specific information.
| |
| l Discussion of EPRI/WOG Recoverv Analysis l s noted previously, the NRC Case Study recovery analysis method is fairly standard now and not subject to methodological criticism. However, a key to the recovery analysis is the assumption made on A-6 L_ ___
| |
| | |
| 3 the success criteria and definition of failure of systems. Many of the recovery events described below
| |
| .re discussed in Sections 3 r.nd 4 in terms of success criteria.
| |
| Mitigating a Loss of All Service Water or Comoonent Cooline Water The NRC Case Study includes a number of sequence cut sets resulting in the loss of all CCW and the loss of all SW. The number of interconnections of a typical two unit Westinghouse PWR often allows for alternative sources of water or rerouting of flows in order to satisfy the cooling missions of the two systems. A review of the, ' abnormal and emergency procedures at Point Beach revealed possible alterna-tive sources and pointed out the necessity of protecting the equipment affected by the lost cooling. The interviews with the Point Beach operators revealed that the loss of CCW could be recovered using CCW from the other pump or the other unit.
| |
| Abnormal Operating Procedure (AOP) AOP-9A applies to SW system malfunctions. AOP-9A addresses the three most likely causes of a loss of all SW: loss of electrical power (to the pumps);
| |
| screen plugging most likely due to icing; and a rupture in a main header. Since a loss of electrical power to all SW pumps is most likely to result from a loss of station power, the " Loss of Station Power" emergency operating procedure takes precedence in this case. Icing of the SW inlets has its own procedure, AOP 13A, which specifies deicing activities. For a rupture in the main header, AOP-9A i
| |
| : alls for the following actions:
| |
| o Isolate the system to find the rupture, o Secure pumps and other equipment associated with or cooled by the affected header, o Transfer selected equipment to altemative cooling, e.g., the emergency diesel generators.
| |
| o Use the header still available to cool at least a train of vital equipment.
| |
| AOP-9B applies to a loss of component cooling. 'Ihis procedure notes that the standby pump shauld j actuate automatically and that, if plant trip occurs, Emergency Operating Procedure EOP- O should be consulted. Recovery activities called out in that procedure include: I o Secure noimal letdown and reduce charging flows, o Maintain reactor coolant pump seal injection.
| |
| l o Secure all auxiliary equipment that uses CCW for cooling.
| |
| l o Shift the source of CCW to the refueling' water storage tenk,if appropriate.
| |
| o Use the unaffected unit's CCW by opening the crossconnects. i o If the fault is a rupture, isolate it whenever possible.
| |
| A7
| |
| | |
| Even with common-cause faults of the unit 1 CCW pumps, the second unit can be used for recovery and
| |
| , clearly called out in the appropriate procedure. Since core cooling pumps would not be needed until
| |
| , core heatup, or 30-60 minutes down the sequence timeline, a non recovery probability in this case of 0.05 is consistent with the NRC Case Study model for actions outside the control room. The EPRI/
| |
| WOG study, however, uses 0.005. This is consistent with the time-reliability corrlation presented in Table A-3 (taken from NUREG/CR 1278).
| |
| Table A 3 TIME-RELIABILITY CORRELATION FOR HUMAN ERRORS DURING RECOVERY (From NUREG/CR-1278)
| |
| Human Available Non-Recovery Time (Minutes? Probability 5 0.3 10 0.1 20 0.01 l
| |
| 30 0.001 l
| |
| 60 0.0001
| |
| * Logarithmic interpolation used for times not listed in the table.
| |
| Cperating the Turbine-Driven AFW Pumo Without DC The loss of DC power, coincident with a loss of station AC, means that the operators can monitor the steam generator water level only with the instrumentation powered by the new batteries installed after the original Point Beach NRC Case Study. It also means that the steam supply motor-operated valve for the turbine-driven AFW pump must be cor: trolled manually at the valve's location via communication with the control room provided b'y security system radios, if necessary. In this situation, the operators an open the steam supply valves and open the outlet valves locally. The operators stated that they A-8
| |
| | |
| could control the turbine-driven AFW pump successfully with control guided by the instnunentation
| |
| 'hich is powered by the new batteries.
| |
| Emergency Procedure ECA 0.0 directs the control room operators to dispatch an operator to monitor or control the turbine-driven AFW pump immediately upon less of all AC or DC. Operators estimated that it would take operators 10-15 minutes to manually start the pump, including local operation of the DC j steam admissiortva?ves, he time available for recovery would then be about 15-20 minutes before steam generator dryout and PORV or SRV opening, and about 45-50 minutes before it is ascumed that core damage could not be averted by feedwater only. Based on NUREG/CR-1278, non-recovery probability would be 0.03 for recovery before PORV or SRV opening and 0.0003 before core damage could no longer be averted by feedwater only.
| |
| Based on these numbers the conditional pmbability of recovery between PORV and SRV actuation and averting core damage is 0.01,if auxiliary feedwater was successful for all scenarios in that time period.
| |
| However, the EPRI/WOG analysis assumes that this conditional probability is limited to 0.1, the proba-bility athat a small LOCA requiring pnmary system makeup would occur in that time period. The 0.1 probability is the EPRI/WOG estimate of the likelihood of a stuck-open SRV or PORV atter they have passed liquid. This probability is based on the Oconee PRA review of EPRI relief valve test data. Con-equently, the overall non recovery probability for this event is 0.03 times 0.1, or 0.003. The remaining secovery values reported in this section were calculated in a similar manner, if appropriate.
| |
| Crossconnecting DC or AC The design of both the AC and the DC trab.s of electrical power include ways to crossconriect the AC of one train to the AC or DC of the other train through a crossconnect charger. His recovery can be exe-cuted by a manual closing of one or more circuit breakers. However, these breakers open automatically on loss of one bus. This recovery would be possible for the loss of one AC bus but not the whole ac grid, or for the loss of one DC bus but not both. The time needed to affect such a recovery is about 15-20 minutes. A recovery factor of 0.05 is consistent with the NRC Case Study model for out of control room actions. A value of 0.003 is consistent with NUREG/CR-1278. However, this tryiew uses 0.005 to account for both the reliability of the breakers opening and the recovery.
| |
| Recirculation Failures :
| |
| The RWST can be refilled if a sump recirculation path is not available. Additionally, charging pumps
| |
| : t. tak.? suction from the two 100,000-gallon reactor makeup water tanks can also be used to recover from I
| |
| recirculation failures even if HPI or LPI pumps were damaged due to cavitation. Emergency Procedure CA 1.1 directs the operators to attempt both of these recoveries. For its recirculation recovery calcula-l tion, the EPPJ/WOG study assumes that containment sprays are operating and that recirculation failure A-9
| |
| | |
| occurs early A non recovery probability of 0.05 is used to account primarily for equipment failures, as vell as human failure to diagnose the event and take the correct action.
| |
| Increasing the time available by terminating containment sprays would not reduce this assessment signi-ficantly since the above value assumes a significant contribution from equipment unavailability. A more detailed systems analysis might indicate lower equipment failures and allow reduction of this value; however, the required analysis was beyond the scope of this study.
| |
| The sump recirculation path consists of two symp outlet lines to the suction lines for the LPI pumps.
| |
| There are two sump valves in each sump oatlet line, and the failure of any pair u of moderate significance. 'Ihe operators pointed out that the MCCs of the valves are outside of containment and the valve bodies are in a channel just outside of containment. The valves closest to the sump,850A and B, are motor-hydraulic valves, each with a hand-hydraulic system to manually open the valves.
| |
| For recovery actions when sump recirculation is faulted, emergency procedure ECA 1.1 advises operators to try to restore the f aulted equipment and, failing that, to refill the RWST. The RWST can be refilled at about a total rate of 160-170 gpm using o Water from the spent fuel pit canal using the holdup tank recirculation pump.
| |
| o Water from the Chemical Volume Control System (CVCS).
| |
| o Water from the unaffected unit's RWST using the refueling water circulation pump.
| |
| This same procedure directs the operators to throttle both SI flow (i.e., reduce to one operating pump) and containment spray to the minimum level required for operation. Exercising these actions would buy the time; needed to refill the RWST and to attempt to restore recirculation. The manual valve realign-ments required to refill the RWST would typically require 5 minutes, but achieving full flow may require as much as 15 minutes according to the operators interviewed. Operators are familiar with these valves because they are normally used in routine operations.
| |
| The procedure also directs the operators to test sump availability at the 60% RWST level (remaining) and when implementing recimulation at the 28% level, to start only one LPI pump first. This strategy assures that the operators would become aware of the unavailability of the sump before cavitating and damaging both LPI pumps. The 28% level would typically mean that for a 2" break and a few hundred gpm makeup rate, some four hours of water are left in the RWST,i.e., until the 10% level is resched,
| |
| 'roviding much more than enough time to initiate RWST refill. If containment sprays were operating,
| |
| ,arJy abom 20 minutes would be left.
| |
| i l
| |
| A-10
| |
| _ - _ _ _ _ _ _ )
| |
| | |
| 1 I
| |
| 1 In addition to the 20 minutes'until 10% level is reached, time would still remain before the core i l
| |
| .mcovered. The NRC Case Study and the EPRI/WOG review both assume an additional 30-40 minutes, l if the loss of ECCS occurred immediately. Since the fuel would be cooled and decay heat would be lo'wer,it is expected that more than 30-40 minutes would be available to recover the RWST and prevent core recovery. :
| |
| i Assuming (conservatively) that 30 minutes were required to fully align and verify that RWST refill occurred, the operators would have at least 20-30 minutes for diagnosis and decision making. An operator failure probably of 0.005 from NUREG/CR-1278 would seem reasonable. However, since a number of other components are required, a value of 0.05 was selected to account for the potential for additional equipment faults.
| |
| ' Recovery of Feedwater on Loss of CST Inventory Loss of CST inventory can result from two important causes. First, either tark can fait due to a seismic event, causing both to leak. Second, inventory will eventually be depleted in the event of a station blackout. A means of feedwater recovery is available for each of these sequences. Both are directed in emergency operating procedures, many of which state: "If CST level decreases to less than 4.5 feet (the technical specification limit), the altemate water sources will be necessary." Operator training addresses he means of obtaining alternate sources, including service water and refilling of the CST using the diesel-driven fire water pumps.
| |
| 4 If a seismic failure of a CST was to occur, service water could be used as a backup source. Valves can be opened from the control root. to quickly provide a backup feedwater source. If offsite power and station batteries were lost sir altaneously, time would be required to provide AC power for the service water pumps. The followi' g steps are required:
| |
| o Manual', close the SW pump breakers (DC).
| |
| o Star' ; diesel generator using the new (backup) batteries. 1 o F. .n a SW pump to provide flow to the diesel generator.
| |
| o Start the remaining generator and balance SW loads, o Initiate SW backup.
| |
| o Start motor-driven AFW pumps for both units.
| |
| L This vocess must occur within 60-90 minutes,i.e., before complete evacuation of the CST tanks and the j ''s,equent loss of AFW has resulted in either a LOCA or core uncovery. This study judged that the CSTs would empty in 30 minutes. The PORV and/or SRV actuation would o: cur at least 30 minutes i later, followed by core uncovery in about another 30 minutes. An operator failure corresponding to 20-A 11
| |
| | |
| 4 L*
| |
| 30 minutes would be reasonable for preventing a relief valve LOCA, i.e.,30-40 minutes is assumed to be required for this action. The corresponding non-recovery probability would be 0.003.
| |
| c It should also be noted that in the event of CST seismic failure, RWST seismic failure is likely due to its lower capacity. Additional actions would be required to align an altemative path. -ne possibility exists that the operators would recover the RWST first, or in conjunction with AFW. 'Ihis concern over whether the operators would attempt to recover the RWST first merits an increase in the non-recovery probability to 0.1, the value derived earlier for the RWST recovery action. In effect, this quantification assumes the operators would be diverted by the RWST recovery for about 20-30 minutes.
| |
| l In the second case, a lorcerm station blackout from other causes, operator recovery would be more
| |
| ; straightforward. Emergency Procedure ECA 0.0 calls for refilling the CST. Traming identifies the diesel-driven fire pump or the portable diesel fire pump as a possible means of refilling the CST. At least two hours would be available to prepare for this action; however, the operators would hope to recover AC power first and avoid filling the steam generator with lake water. It was conservatively assumed that recovery does not begin until after loss of CST inventory. Assuming 50 minutes to dryout at this late time and 30 minutes spent aligning the pumps and hoses, the non recovery pmbability for 20 minutes, i.e.,0.01, would seem appropriate. It was assumed that the unavailability of the pumps and hoses would dominate the failure probability. A probability of 0.03 was deemed reasonable. Allowing for a failure probability of 0.02 essentially considers the two redundant fire pump sources equivalent to another train of hardware.
| |
| Balancine Loads in the Event of Less than Three Service Water Pumps This action could be performed quickly from the control room according to procedure. Two time frames are considered. First,in the event ofloss of offsite power and station emergency power prob-lems, service water to the diesel generators would have to be provided within a few minutes to prevent diesel damage. For this case, the NUREG/CR-1278 value of 0.3 for recovery within 5 minutes is used.
| |
| The second time frame corresponds te loss of CCW and its impact on recirculation. In this case, balan-cing loads would not have to occur until recirculation, i.e., after a significant period of time. A value of 0.0001 was deemed reasonable.
| |
| Use of New Station Batteries WEP has installed seismic grade batteries which provide redundant power to vital inst.umentation and for starting the diesel generators. These batteries affect the plant model t'y allowing:
| |
| A-12
| |
| | |
| o The turbine-driven auxiliary feedwater pump to be controlled during the loss of normal DC power and vital AC backup, i.e., by not having to operate the pumps blind (long- and short-term loss of offsite power, fire, and seismic scenarios).
| |
| o The diesel generators to be started when both normal DC power and offsite power have been lost (loss of offsite power and seismic scenarios).
| |
| Use cf New S tation Batteries WEP has indicated that with load shedding the life of normal station batteries would be eight hours, s rather than the two hours indicated in the NRC Case Study. This extra battery life will extend the time during which the runmng AFW turbine-driven pump can be supported by DC powered instrumentation.
| |
| Extended battery life would allow extended station blackout sequences for which the diesel-driven fire ;
| |
| pump or portable diesel fire pump were used to refill the CST to be recovered with instruments powered )
| |
| by the main station batteries still available. However, extended battery life for the main station batteries will make little difference to the EPRI/WOG plant model because the new batteries could power neces-sary instrumentation after the main station batteries are depleted. The new batteries automatically 5
| |
| provide instrumentation for the operator to control steam generator level and AFW flow in the event of loss of the mair, station batteries.
| |
| Recovery of Common Cause Failures -
| |
| Experience indicates the most common cause failures would be recoverable within 30 to 60 minutes.
| |
| (See Section 5.) Accordingly, probabilities of common cause failures are decreased in the original data for this review.
| |
| Lack of Containment Sprav After a Small Break LOCA orTransient When Containment Fan Coolers
| |
| _Att Available l
| |
| Tuming off containment sprays will conserve RWST inventory, delay the need for recirculation, and I thereby both reduce the likelihood of recirculation failure and increase the likelihood of recirculation recovery, if necessary. WEP procedures require the operators to tum off containment sprays if adequate heat removal is provided by the fan coolers, once containment pressure is below 25 psig, the contain-ment spray initiation setpoint.
| |
| The small break LOCA sizes considered in this study will not lead to initiation of containment sprays. I The transient accident scenarios will not lead to containment spray initiation unless containment fan coolers fail.
| |
| This action is not credited specifically in either the EPRI/WOG or NRC plant models. The long time criod available to initiate recirculation should reduce operator error rates and increase the success A-13
| |
| | |
| F probability for RWST refill. The time for initiation of recirculation would be many hours, rather than
| |
| 'he two to four hours assumed in the NRC Case Study.
| |
| Loss of Offsite Power Timing Studv, An additional significant internal events issue which is not considered in the NRC Case Study is timing
| |
| ~ analysis for long-term station blackout scenarios. Florida Power Corporation's Crystal River Unit 3 PRA study G.) has shown that explicit consideration of time dependency in the recovery of station -
| |
| power will reduce station blackout risk by about 40%.
| |
| De NRC Case Study analyzes "run failures" of diesel generators for an eight hour mission time. How-ever, acovery of offsite power is considered only for one hour. In a real situation, if only one diesel generator were operating for a certain number of hours, additioval time for recovery of offsite power would be available. If a diesel were to run for four hours, recovery of offsite power would increase in -
| |
| . likelihood by a factor of five. After a certain effective time, diesel run failures would not be important since recovery of offsite post would be extremely likely. In the Crystal River study, this effective diesel run time was about two hours. Since this Crystal River timing analysis would be roughly appli-cable to other PWRs, Point Beach station blackout core-melt frequency was reduced by about 40%,
| |
| when considering time dependencies.
| |
| 5
| |
| /
| |
| A-14
| |
| | |
| 'GFERENCES FOR APPENDIX A ,
| |
| : 1. " Shutdown Decay Heat Removal Analysis of a Westinghouse 2 Loop Presurrized Water Reactor, Case Study," NUREG/CR-458, Carmmond, W.R., et. al., March,1987.
| |
| : 2. Handbook of Human Reliability Analysis with Emphasis on Nuclear Power Plant Applications, Final Report, NUREG/CR-1278, USNRC, August,1983, Swain, A.D., and Guttmann, H.E. 4
| |
| : 3. Nuclear Safety Analysis Center and Duke Power Company, Oconee PRA: A Probabilistic Risk Assessment of Oconee Unit 3, NSAC-60, Electric Power Research Institute, June,1984.
| |
| : 4. Seebrook Station Probabilistic Safety Assessment, Pickard,I.oure and Garnick,Inc.,
| |
| December,1983.
| |
| 5.
| |
| : 6. Interim Reliability Evaluation Program: Analysis of the Arkansas Nuclear One - Unit 1 Nuclear Power Plant, NUREG/CR 2787, K016, G.L. et. al., USNRC, June,1982.
| |
| : 7. Florida Fower Corporation, PRA of Crystal River-3,1986.
| |
| h t
| |
| A-15
| |
| _ _ _ _ _ _ _ _ _ _ _ - _ _ _ _ _ _ _ _ _ _ a
| |
| | |
| a Appendix B INTERNAL EVENTS ACCIDENT SEQUENCE ANALYSIS DETAILS i
| |
| Table B 1 provides a summary of the EPRI/WOG changes to the NRC Case Study (1) core melt model that affected the results obtained for the EPRI/WOG study. Dese changes correspond to those identified in the previous appendices and whose significance was summarized in the first table of each section.
| |
| Table B-2 is a list of the top 30 dominant accident sequence cut sets, nese cut sets comprise roughly 75fc of the intemal event risk presented in the NRC Case Study (not including long term station blackout). In the right hand column are comments corresponding to entries in Table B-1. The comments indicate that there are a significant number of changes applicable to each sequence. Both qualitatively and
| |
| , quantitatively, this study has shown that the NRC Case Study sequences are conservative.
| |
| Tables B-3 and B-4 present the corresponding top 30 and top 15 sequences resulting from this review's ,
| |
| reevaluation. The results are presented with and without the new batteries.
| |
| l l
| |
| i I
| |
| t i I
| |
| l B-1
| |
| | |
| \
| |
| l l
| |
| l l
| |
| l EFERENCES FOR APPENDIX B l
| |
| : 1. " Shutdown Decay Heat Removal Analysis of a Westinghouse 2-Loop Presurrized Water i Reactor, Case Study," NUREG/CR-458, Carmmond, W.R., et. al., March,1987.
| |
| l B2
| |
| .___________ _ a
| |
| | |
| Table B 1
| |
| | |
| ==SUMMARY==
| |
| OF PLANT CHANGES AND KEY TO COMMENTS IN TABLE B 2
| |
| : 4. Success Criteria and Accident Seouence Timing
| |
| : a. Open DC Vent Valves if 1 PORV Fails
| |
| : b. Small LOCA Requires only 1 LPT Pump
| |
| : c. Less Restrictive Containment Success Criteria
| |
| : d. HPSI and LPIdo not Require CCW
| |
| : e. SWS can be Imad Balanced if < 3 Pomps Work
| |
| : f. Refill RWST and/or Provide Altemate Flow Path
| |
| : g. Full Charging Flow if 1 PORV Fails
| |
| : 5. Human Reliability and Recovery Analysis
| |
| : a. Improve HRA Modeling of Diagnostics
| |
| : b. Recover CCW
| |
| : c. Menual AFW Turbine Pump Operation
| |
| : d. Cross-Connect AC or DC Buses
| |
| : c. Recirculation Recoverv
| |
| : f. Alternative to Failed RWST (Seismic)
| |
| : g. PORV Recovery w/DC Vent Valves, Charging
| |
| : h. Alternative to Failed CST (Seismic) i Refill CST on Long-Term Blackout l i.
| |
| : j. Balance Service Water Leads to Diesel
| |
| : k. Start Diesels with New Batteries
| |
| : 1. Recover MOV Control Circuit Faults
| |
| : m. Recover Pump Control Circuit Faults
| |
| : n. Loss of Offsite PowerTiming Study
| |
| : o. Improve General Recovery Model
| |
| : 6. Initiatine Ever.t and Comyanent Data
| |
| : a. Small LOCA Fmquency
| |
| : b. Loss of Offsite Power Fmquency
| |
| : c. Loss of Feedwater (with Recovery) ,
| |
| : d. All OtherTransients
| |
| : c. Loss of AC Bus (with Recovery)
| |
| : f. less of DC Bus (with Recovery)
| |
| : g. Diesel Generator Reliability B-3
| |
| | |
| i Table B 1 (Continued)
| |
| | |
| ==SUMMARY==
| |
| OF PLANT CHANGES AND KEY TO COMMENTS IN TABLE B-2
| |
| : 6. Initiating Event and Composent Data (Continued)
| |
| : g. Flood in Service Water Pump Room
| |
| : h. Fire in AFW Pump Room
| |
| : i. Fire in Switchgear Room
| |
| : j. Depressurization (X)
| |
| : k. SRV or PORV LOCA (Q)
| |
| : 1. Block Valve Status
| |
| : m. Battery Common Mode
| |
| : n. AFW Pump Common Mode
| |
| : o. SWS Pump Common Mode I
| |
| : p. CCW Pump Common Mode
| |
| : q. LPI and HPI Pump Common Mode
| |
| : r. Sump MOV Common Mode
| |
| : s. All Other MOV Common Modes
| |
| : t. Diesel Generator Common Mode i
| |
| j l
| |
| l B-4
| |
| | |
| l TABLE B-2 TOP 30 NRC CASE STUDY INTERNAL SEQUENCES WITHOUT LONG-TERM STATION BLACKOUT PLANT PER-DAMAGE SEQUENCE FREQ. CENT CLASS NAME MEASURE (%) ACCIDENT SEQUENCE EVENTS COMMENTS
| |
| ..............................._...................m....... .. ................ .--
| |
| AB 2.00E-05^ 14.99 S2 M SMPVCCOE Sa, 5e,6a NR B1 19BB 9.94E-06 7.45 73 Q R Sa,5e,6d,6k SMPVCCOE NR B19B 2B B.00E-06 6.00 S2 H NR 5e,6a,6:
| |
| HRMVSUMPCM B1 IB ?.60E-06 5.70 S2 M NR 4d,Se,6a CWXV3Du,'97M B1 17F 4.85E-06 3.64 71 M o 5k,5c,6m BATCM GT.
| |
| RA1F RAB)
| |
| F17AC RA20H/RA1F AFWMAN 3C 4.00E-06 3.00 S2 M NR 4d,5e,5b,6a CWMPP@ CM C3CCW 1B 4.00E-06 3.00 52 M NR 5e,6q,6a LIMPPC CM B1 19BB 3.98E-06 2.98 T3 Q R 5e,6d,6k,6r NR HRMVSUMPCM B19B 19BB 3.78E-06 2.83 T3 O R 4d,5b,5e,6d,6k NR CWXV30XOCUTM B19B 1B 2.00E-06 1.50 S2 M NR 4d,5b,5e,6a CWXV30XOCLF B1 19BB 1.99E-06 1.49 T3 Q R 5e,6d,6k,6g NR LIMPPCCM B19B 21BC 1.96E-06 1.47 T3 O R 4d,5b,5e,6d,6k
| |
| -X NR 6p CWMPPCCM C21CCW 37A 1.79E-06 1.34 T4 M -Q 4a,5d,5g,5m,6f NR AWTP3PTDLF AWMPlPE LF A37 20F2PORV 27A 1.57E.06 1.18 T2 M -Q 4a,5g,6c,6n NR AWSPCM BKVLVALF 20F2PORV i
| |
| A27 6 27A 1. 5'i E- 0 6 1.18 72 M -Q 4a,5g,6c,6n NR AWSPCM f BKVLVBLF 20F2PORV l A27 3C 1.51E-06 1.13 S2 M NR 4d,5b,5e,5m,6a CWMPAP@ LF CWMPBP@ LF l C3CCW ,
| |
| B-5
| |
| | |
| TABLE B-2 (Continued)
| |
| TOP 30 NRC CASE STUDY INTERNAL SEQUENCES WITHOUT LONG-TERM STATION BLACKOUT PLANT PER-DAMAGE SEQUENCE FREQ. CENT CLASS NAME MEASURE (%) ACCIDENT SEQUENCE EVENTS COMMENTS 1B 1.51E-06 1.13 S2 M NR 5e,5m,6a LIMP 1PMDLF LIMP 2PMDLF B1 19AB 1.40E-06 1.05 T2 M Q 4d,5a,6c,6k R SMPVCCOE NR 1B 1.39E-06 1.04 S2 M NR 5e,51,5m,6a LIMP 1PMDLF HRMV38VCCLF B1 1B 1.39E-06 1.04 S2 M NR 5e,5m,51,6a j' LIMP 1PMDLF HRMV5VCCLF B1 3B 1.39E-06 1.04 S2 H NR 5e,51,5m,6a LIMP 2PMDLF HRMV4VCCLF B1 1B 1.39E-06 1.04 S2 M NR 5e,51,6a LIMP 2PMDLF HRMV36VCCLF B1
| |
| .B 1.28E-06 .96 S2 M NR 5e,51,6a HRMV36VCCLF HRMV5VCCLF B1 1B 1.28E-06 .96 S2 M NR 5e,51,6a HRMV4VCCLF HRMV5VCCLF I B1 I aB 1.28E-06 .96 S2 M NR 5e,51,6a i HRMv36VCCLF HRMV38vCCLF j B1 AB 1.28E-06 .96 S2 M NR HRMV4VCCLF HRMV38vCCLF 5e,51,6a B1 37A 1.18E-06 .88 T4 M -Q 4a,5d,5g,6e,6n.
| |
| AWSPCM NR A37 20F2PORV 47F 1.03E-05 .77 T5 M -Q $d,5j,6f NR AWTP3PTDLF SNMPDPMDUTM F47SNS 27F 1.03E-06 .77 T2 M -Q 4d,5e,6c,6o NR SWPCM AWTP3PTDLF F27SWS 47F 1.03E-06 77 75 M -Q 5d,5j,6f NR AWTP3PTDLF SNMPEPMDUTM F47SWS I
| |
| TOTAL: 9.64E-05 72.27% of CM Total Frequency 1.33E-04 l
| |
| I B-6
| |
| | |
| TABLE B-3 TOP 30 SEQUENCES FOR INTERNAL EVENTS WITHOUT LONG-TERM STATION BLACKOUT EPRI/WOG REPRESENTATION WITHOUT NEW BATTERIES PLANT PER-DAMAGE SEQUENCE FREQ. CENT CLASS NAME MEASURE (%) ACCIDENT SEQUENCE EVENTS p 17F 3.57E-06 65.70 T1 M -Q j BATCH GTF RA1F RABF F17AC RA20H/RA1F AFWMAN DGNEWBAT 17F 3.01E-07 5.55 T1 M -Q GTF RAIF AWTP3PTDLT DSGENAGENLF DSGENBGENTM F17AC RA20H/RAlr 1B 3.00E-07 5.53 S2 M SMPVCCOE B1
| |
| , 17A 1.BBE-07 3.46 T1 M -Q GTF RA1F AWTP3PTDLF DSGENAGENLF DSGENBGENLF F17AC RA20H/RA1F 17A 1.75E-07 3.23 T1 M -Q GTF RA1F DGCM AWTP3PTDLF RA10F F17AC RA20H/RA1F 26A 9.9BE-08 1.84 T2 M -Q A26 HRXCCOE AWSPCM IB 6.00E-08 1.11 S2 M RECIRCREC HRMVSUMPCM B1 1B 5.70E-08 1.05 S2 M RECIRCREC CWXV30XOCUTM B1 17F 5.65E-08 1.04 T1 M -Q GTF RA1F LFBATTADOS LFBATTBDOo RA20H/RA1F AFWMAN F17AC DGNEWBAT 17F 3.40E-08 .63 71 M -Q i GTF RAIF DGCM RA20H/RA1F I
| |
| AWTP3PTDUTM F17AC 17F 3.27E-08 .60 T1 M -Q GTF RA1F DSGENAGENLF DSGENBGENLF RA20H/RA1F AWTP3PTDUTM F17AC B-7
| |
| _ _ _ _ - - _ - _ __ . - . 1
| |
| | |
| TABLE B (Continued)
| |
| TOP 30 SEQUENCES FOR INTERNAL EVENTS WITHOUT LONG-TERM STATION BLACKOUT EPRI/WOG REPRESENTATION WITHOUT NEW BATTERIES PLANT PER-DAMAGE SEQUENCE FREQ. CENT CLASS NAME MEASURE (O ACCIDENT SEQUENCE EVENTS 87F 3.01E-08 .56 T1 M -Q GTF RA1F AWTP3PTDLF DSGENBGENLF DSGENAGENTM F17AC RA20H/RA1F 3C 3.00E-08 .55 S2 M RECIRCREC CWMPPMDCM C3CCW 1B 3.00E-08 .55 S2 M RECIRCREC LIMPPMDCM B1 19BB 2.69E-08 .50 T3 0 R SMPVCCOL B19B 17F 1.95E-08 .36 T1 M -Q RA1F AWTP3PTDLF SWPCM F175WS RA20H/RAIF JF 1.80E-08 .33 S2 M SSLOGCMTM RA6D F3 SIS RA20H/RA6D 17F 1.59E-08 .29 T1 M -Q GTF RA1F LFBATTBDO6 TMBATTADOS AFWMAN F17AC RA20H/RA1F DGNEWBAT 17F 1.59E-08 .29 T1 M -Q GTF RA1F LFBATTADOS TMBATTBDO6 AFWMAN F17AC RA20H/RA1F DGNEWBAT 8B 1.50E-08 .28 S2 M RECIRCREC CWXV30XOCLF B1 17F 1.48E-08 .27 T1 M -Q GTF RA1F AWTP3PTDLF DSGENAGENLF F17ACSMS SNMPEPMDUTM RA20H/RA1F SMSLDBAL5 MIN 17F 1.48E-08 .27 71 M -Q l GTF RA1F l
| |
| AWTP3PTDLF DSGENAGENLF .
| |
| F17ACSWS SWMPDPMDUTM RA20H/RA1F SWSLDBAL5 MIN l
| |
| l B-8 L
| |
| | |
| L l TABLE B-3 (Continued) 1' TOP 30 SEQUENCES FOR INTERNAL EVENTS WITHOUT LONG-TERM STATION BLACKOUT EPRI/WOG REPRESENTATION WITHOUT NEW BATTERIES PLANT PER-DAMAGE SEQUENCE FREQ. CENT CLASS. NAME MEASURE (%) ACCIDENT SEQUENCE EVENTS 17F 1.29E-08 .24 T1 M -Q GTF RA1F l AWTP3PTDLF DSGENBGENLF l RA20H/RA1F SNMPBPMDLF F17ACSWS SWSLDBAL5 MIN 17F 1.29E-08 .24 71 M -Q GTF RA1F AWTP3PTDLF DSGENAGENLF RA20H/RA1F SWMPDPMDLF F17ACSWS SWSLDEALSMIN 17F 1.29E-08 .24 71 M -Q GTF RA1F AWTP3PTDLF DSGENBGENLP RA20H/RA1T SNMPAPMDLT
| |
| : l. F17ACSWS SWSLDBAL5 MIN 7F 1.29E-08 .24 T1 M -Q GTF RAIF AWTP3PTDLF DSGENAGENLF RA20H/RAlr SNMPEPMDLF F17ACSWS SWSLDBALSMIN 1B 1.14E-08 .21 S2 M RECIRCREC LIMP 1PC LF LIMP 2PC LF B1 PUMPCC 3C 1.14E-08 .21 S2 M PUMPCC CWMPAPMDLF CMMPBPMDLF C3CCW RECIRCREC IB 1.04E-08 .19 S2 M RECIRCREC LIMP 2PMDLT HRMV4VCCLF B1 PUMPCC MOVCC IB 1.04E-08 .19 S2 M RECIRCREC LIMP 2PMDLF HRMV36VCCLF B1 PUMPCC MOVCC TOTAL: 5.20E-06 95.75% of CM Total Frequency 5.43E-06 B-9
| |
| | |
| TABLE B-4 TOP 15 SEQUENCES FOR INTERNAL EVENTS WITHOUT LONG-TERM STATION BLACKOUT EPRI/WOG REPRESENTATION WITH NEW BATTERIES PLANT PER-DAMAGE SEQUENCE FREQ. CENT CLASS NAME MEASURE (%) ACCIDENT SEQUENCE EVENTS 87F 3.01E-07 16.01 T1 M -Q GTF RAlr AWTP3PTDLF DSGENAGENLF DSGENBGENTM F17AC RA20H/RA1T 1B 3.00E-07 15.94 S2 M SMPVCCOE B1 17A 1.88E-07 9.97 T1 M -Q GTF RAlr AWTP3PTDLT DSGENAGENLF DSGENBGENLF F17AC RA20H/RA1F 17A 1.75E-07 9.32 T1 M -Q GTF RA1F DGCM AWTP3PTDLF RA10F F17AC RA20H/RA1F 17F 1.07E-07 5.68 71 M -Q BATCM GTF RA1F RA8F F17AC RA20H/RA1F AFWMAN DGNEWBAT 36A 9.98E-08 5.30 72 M -Q A26 HRXCCOE AWSPCM 1B 6.00E-08 3.19 S2 M RECIRCREC HRMVSUMPCM B1 1B 5.70E-08 3.03 S2 M RECIRCR.EC CWXV30XOCUTM B1 17F 3.40E-08 1.81 71 M -Q GTF RAlr DGCM RA20H/RA1F AWTP3PTDUTH F17AC 17F 3.27E-08 1.74 T1 M -Q GTF RA1F DSGENAGENLF DSGENBGENLF RA20H/RA1F AWTP3PTDUTM F17AC 17F 3.01E-08 1.60 T1 M -Q GTF RAlr AWTP3PTDLF DSGENBGENLF DSGENAGENTH F17AC RA20H/RA1F B 10
| |
| | |
| 1 I
| |
| TABLE B-4 (Continued)
| |
| TOP 15 SEQUENCES FOR INTERNAL EVENTS
| |
| .WITHOUT LONG TERM STATION BLACKOUT EPRI/WOG REPRESENTATION WITH NEW BATTERIES Pl. ANT PER-DAMAGE SEQUENCE FREQ. CENT CLASS NAME MEASURE (%) ACCIDENT SEQUENCE EVENTS 3C 3.00E-08 1.59 S2 M RECIRCREC CWMPPMDCM C3CCW IB 3.00E-08 1.59 S2 M RECIRCREC LIMPPMDCM B1' 19BB 2.69E-08 1.43 T3 Q R SMPVCCOE B19B 17F 1.95E-08 1.04 T1 M -Q RA1T AWTP3PTDLT SWPCM Fl?SWS RA20H/RA1F TOTAL: 1.49E-06 79.24% of CM Total Frequency 1.88E-06 3
| |
| l 1
| |
| /
| |
| B-11
| |
| | |
| Appendix C PUBLIC CONSEQUENCE ANALYSIS DETAILS INTRODUCI'lON Three EPRI/WOG source terms have been developed. Following this discussion of their development, l
| |
| i the EPRI/WOG method for estimating doses is delineated.
| |
| FIRST ALTERNATIVE: NRC SOURCE TERM MODEL This appendix uses BMI 2104 (1), Volumes IV and V, and a draft version of NUREG-0956 (2) as a basis for developing an attemative source tenn model closest to current NRC work. The reports are reviewed to determine unique accident sequence types, then their source terms and potential contain-ment failure modes are obtained, and finally, probabilities are erdmated for each. Because the NRC Case Study (1)is limited to small break LOCAs and transient-initiated accidents, a source term is devel-
| |
| 'oed only for these accident types.
| |
| The following discussion is broken into four subsections. The first identifies the factors important to defining accident sequence types for an NRC source term; it also discusses the limitations associated with applying the NRC Case Study sequence models thereto. The second discusses accident sequence types in more detail, based on containment conditions. Containment failure mode probabilities are pre-sented as well. The third discusses how pnmary system conditions affect releases. And the fourth presents a table of NRC source terms.
| |
| Eactors Important To Definine Accident Sequence Tvoes Individual core melt sequences and containment safeguard states are assigned to accident sequence types. By definition, these accident types result in similar containment failure probabilities and similar source terms. Conditions both in the pnmary system and in the containment can affect containment fail-ure probabilities and source terms. The NRC Case Study uses an approach based on WASH 1400 (4) to define accident sequence types (table K.2 in appendix K, seen here as table C-1). WASH 1400 had determined that primary system conditions have no direct effect on source terms. Employing new NRC source terms requires that defming accident seguence types consider primary system conditions.
| |
| C1
| |
| | |
| Table C-1 Identification of Containment Failure Mode Probabilities and Release of Category Assignments for Point Beach Accidents Containment Failure Release Mode Probability Cateaory-Accident Tvoe 1
| |
| : 1. Core melts with contain- a= .0001 7
| |
| ment sprays and heat c = 0.25 removal always available. y+6, = 0.014 3 j
| |
| &g = 0.18 5 )
| |
| l 8 = 0.002 5 Early core melts with a= .0001 1 2.
| |
| 6 containment heat-removal e = 0.25 always available. Sprays Y+6, = 0.014 3 succeed initially, fail &g = 0.18 4 8 = 0.002 5 in recirculation
| |
| : 3. Core melts with containment a= .0001 1 heat removal available, c = 0.25 6 no sprays at-time of y+6, = 0.014 2 core melt 6g = 0.18 3 8 - 0.002 4 Early core melts with no a= .0001 1 4.
| |
| containment heat removal, c = 0.25 7 sprays always available y+6, = 0.014 3
| |
| &g = 0.18 4 8 = 0.002 5
| |
| : 5. Late core melts with no a= .0001 1 containment heat removal, c = 0.25 6 sprays succeed early, y+6, = 0.014 3 fail late 6g = 0.18 4 8 - 0.002 5 l
| |
| : 6. Core melts with no a= .0001 1 containment heat removal, c = 0.25 7 1 I
| |
| sprays succeed prior to y+6, = 0.014 2 melt, fail at seit 6g = 0.18 3 8 - 0.002 4 a= .0001 1
| |
| : 7. Core melts with no 6 1 containment heat removal c = 0.25
| |
| )
| |
| or sprays v46, = 0.014 2 4g = 0.18 3 8 - 0.002 4 Containment Failure Modes a In-vessel steam explosion y Rydrogen burn overpressure 6, Ex-vessel steam spike 8 Containment leatage c Base mat neit through
| |
| &g steam and noncondensible gas overpressure C-2
| |
| | |
| l Primary System Conditions. Recent source term analysis, and NRC source term analysis in particular, Mdicate a relationship between reactor coolant system leakage rate and pressure and the amount of fis- i sion products retained in the primary system at the time of vessel failure. Small-break LOCAs and iransients differ in both of these respects. While engineering judgment and indications from some anal- !
| |
| yses point to differences between transient-induced LOCAs and small break LOCAs, i.e., fission product removal in the quench tank pathway, these accident types have been conservatively grouped f bectuse of a lack of detailed information to support these differences' quantitative impact on source terms.
| |
| The timing of core melting, e.g., ECC injection versus recirculation failure, does not appear to make sig-nificant differences in NRC analyses. However, this conclusion may be a result of the small number of sequences analyzed.
| |
| Containment Conditions. Containment conditions will affect the probability of early and late contain-ment failure and the source tenn. The principalimpact on the source term results from the operation of j containment sprays. Operation of sprays for even a short period of time will significantly reduce fission products in the containment.
| |
| Both sprays and fan coolers will affect containment heat removal and therefore containment failure probability. If containment heat removal succeeds, steam overpressurization is unlikely to occur. How-ever, heat removal also removes steam from the containment which may create favorable conditions for hydrogen burning. Hydrogen burning is judged by NRC analysis to be a credible mode of containment failure. Containment heat removal appears to have little affect on NRC assessments of other contain-ment failure modes, i.e., direct heating.
| |
| The NRC Case Study notes a difference between early and late post-accident radioactivity removal, based on whether containment spray operates in both injection and recirculation modes, orjust the injec-tion mode. If containment spray operates while the core is still being cooled, it has no significant effect on source terms. This tmung effect is also accounted for in this study.
| |
| Limitations in Application of NRC Case Study Sequences to NRC Models. The NRC Case Study does not account for a spray tuning effect on containment heat removal. If the core melts late, i.e., due to an ECCS recirculation failure, timing for heat removal can be a significant factor in determining contain-ment failure probability. If sprays were to have succeeded in the injection mode, containment pressure would be low at the time core melting began.
| |
| , however, sprays and fan coolers had never operated, then containment pressure would be high due to the continued steaming from the core while ECC was operating. The difference in pressure between the C-3
| |
| | |
| two cases could lead to a significant difference in containment failure probability at the time of vessel
| |
| 'ilure. For example, in NSAC's Oconee PRA (1) (page 10-204), the difference in early containment
| |
| .ailure probability for these two types of sequences was three orders of magnitude. Table C-1 indicates that in the NRC Case Study, all containment heat removal failures are grouped together (accident type 7),i.e., that no difference is assessed between early and late core melts for this containment condition.
| |
| Section 2.3.1 in appendix B of the NRC Case Study indicates the same.
| |
| It should be noted that this non conservatism is not expected to be significant to the conclusions of this review because of the low probability of this accident type. The probability of four independent fan cooler failures is very low, i.e., much less than the three order of magnitude potential risk increase.
| |
| Dependent fan cooler failures,i.e., from support systems, would be expected to fail the ECC system as well. In general, multi-system failure sequences require dependent support system failures to have sig-nificant probability.
| |
| The significance of this finding follows from a lack of carefulintegration or consequences and systems analysis encouraged by the use of WASH-1400 source term analysis. It is possible that a modification could be proposed which could actually increase risk by increasing the likelihood of this type of sequence, while decreasing overall core damage frequency. Using tnore appropriate source terms and
| |
| 'erforming a more careful integration with the sequence analysis needs to be done if the NRC Care
| |
| ,tudy is to support value impact analysis of plant modifications.
| |
| In section 4.1 reference was made to conservatism in the containment systems success criteria and to its impact on integrating new source terms into the existing analysis. These lirm. "ons are discussed below.
| |
| Definition of Accident Tvoes Based on Containment Conditions The containment event tree in the NRC Case Study delineates six sequences. The six sequences can be described as two without either heat or fission product removal at the time of core melting, two with heat removal but without fission product retnoval at the time of core melting, and two with heat removal and fission product removal at all times.
| |
| In this subsection, containment failure modes and the probabilities are identified for each contamment safeguard state. As will be discussed, the modes and probabilities can be defined based only on contain-ment conditions because of the insensitivity of the NRC results to primary system conditions. l l
| |
| "ontainment safeguard state 7_, C2,i.e., a failure of both sprays and fans, results in immediate failure of l
| |
| . cat and fission product removal, according to the NRC Case Study success criteria. Actually, the state includes conditions where one, two or three fan coolers are operating. If any fan coolers are operating, J i
| |
| l C-4
| |
| | |
| l l
| |
| l the state /Z C2, i.e., successful fans and failed contamment sprays, would yield more appropriate source ms. Z C2 could result in the following containment failure modes and the central estimate and condiConal probabilities (according to NUREG-0956, table B.5):
| |
| Conditional Embabilitics Central Containment Failure Mosig Estimate none (NCF) 0.38 !
| |
| I basemat melt-through (MT) 0.09 j l
| |
| late overpressurization (LOP) 0.32 late hydrogen burn (LHB) 0.20 carly steam spike and/or vessel depressurization (EOP) N.A.
| |
| early steam spike and/or direct heating (DH) 0.005 in-vessel steam explosion (SE) N.A.
| |
| large isolation failure or preexisting leak (CI) 0.002 The pessimistic estimate probabilities of NUREG-0956 are not considered to be appropriate for best estimate value impact decisions.
| |
| Failure of all heat and fission product removal at the time of recirculation,i.e., spray failure coincident with ECC failure and previous fan cooler failure, is represented by Z /C2 F'. This sequence end state should be very similar to Z C2 in that containment heat removal fails when core cooling fails. A conservative approach was used in the EPRI/WOG interpretation of an 'NRC' source term to group these two sequence types, and assign the same containment failure modes and pmbabilities. Nete also that this sequence is unhkely.
| |
| Successful heat removal provided by operating fans with immediate failure of all fission product heat removal caused by failed sprays is represented by state /Z C2. This end state describes a situation where only fan coolers are operating (all 4). The contamment failure modes are similar to the previous cases with the addition of early steam spike and hydrogen burn (EHB). This additional failure mode exists
| |
| 'secording to NUREG-0956) because the fan coolers prevent steam inerting of the containment. In the
| |
| .evious scenarios, steam inerting prevents hydrogen from burning near the time of vessel failure. The containment failure mode probabilities (according to NUREG-0956, table B.3) are:
| |
| C-5
| |
| | |
| Conditional Probabilities Central Containment Failure Mode Estimate none (NCF) 0.49 basemat melt-through (MT) 0.28 late overpressurization (LOP) 0.18 late hydrogen burn (LHB) 0.05 carly steam spike and/or vessel depressurization (EOP) N.A.
| |
| early hydrogen bum (EHB) N.A.
| |
| early steam spike and/or ,
| |
| direct heating (DH) 0.01 !
| |
| in vessel steam explosion (SE) N.A.
| |
| large isolation failure or preexisting leak (CI) 0.002
| |
| 'ontainment safeguard state /Z /C2 F' includes spray success up until loss of core cooling and continued fan cooler operation. At that time, containment sprays fail for the same reason that ECC failed. The containment responds much the same as if an early core melt occurred and sprays were immediately unavailable but fan coolers were available. The containment failure mode probabilities r=e the same as
| |
| /Z C2.
| |
| Containment safeguard state /Z /C2 /F' guarantees that both sprays and fan coolers rie available throughout the accident and, therefore, that heat and fission product removalis provided. Because the i NRC analysis does not distinguish between sprays and fans with regard to heat removai, the contain-ment failure mode probabilities can be assumed to be the same as the previous case. The principal difference between these cases and the previous cases will be the effect of fission product removal on the source term.
| |
| Containment safeguard state Z /C2 /F' also guarantees that fan coolers have failed but that sprays are available throughout the accident. However, there is some uncertainty as to the availability of heat j removal, i.e., the state includes both sprays operating with and without RHR coolers. Without heat removal, the containment would eventually fail. However, with fission product removal available
| |
| ,urce terms would be extremely small. Also the probability of heat removal recovery before this C-6
| |
| | |
| l a
| |
| delayed containment failure would be very high. It was judged that assigning this end state to the same l
| |
| onditions as /Z /C2 /F' was a reasonable assumption. l I
| |
| The above mentioned probabilities need to be considered in light of one important assumption, namely regarding the effect of primary system conditions on containment failure probabilities. As mentioned previously, reactor coolant system leakage rate can affect the amount of hydrogen produced during core ineltdown. The amount of hydrogen produced can in turn affect the probability of containment failure due to hydrogen burning. Also, the RCS pressure at the time of vessel failure can affect the amount of ;
| |
| debris dispersion and, therefore, the probability of containment failure due to direct heating. )
| |
| )
| |
| The problem obtaining probabilities for containment states for different primary system conditions is that NUREG-0956 does not provide probabilities for all the combinations of states, e.g., transient cases include only loss of both heat and fission product removal. Tables B.3, B.4, and B.5 in NUREG 0956 provide probabilities for three different RCS leakage rates. The tables indicate very little sensitivity in hydrogen burmng and direct heating probabilities. Therefore,it was judged reasonable to use the same probabilities for each containment failure mode regardless of primary system conditions.
| |
| p_dirlitipp of Accident Tvoes Based on Primary System Conditions Nmary system conditions are important however for predicting source terms using an NRC-based approach. The RCS leakage rate plays a significant role in determining primary system fission product retention according to the BMI-2104 reports. The larger the leakage rate, the greater the amount of fis-sion products released to the containment.
| |
| The NRC Case Study analyzes accidents with three types of RCS leakage conditions, namely small-break LOCAs, transient induced LOCAs through the pressurizer, and transients with cycling relief valves. The following discussion explains how this EPRI/WOG interpretation of NRC source terms were estimated using the NRC information available in the NRC references such as NUREG 0956.
| |
| First,it was assumed that the most appropriate source terms to use were those found in NUREG-0956 since it contains the most recent data. However, NUREG-0956 does not provide sufficient details to allow differentiating between the three sequence types contained in the NRC Case Study. Further, it would not be accurate to use only the NUREG-0956 results because some effects, namely increased source terms for small-break LOCAs, could not be accounted for and risk would be underestimated.
| |
| Therefore, BMI 2104 was used for guidance to adjust the NUREG-0956 source terms to account for pri-mary system conditions.
| |
| ae following describes how source terms were derived for accidents characterized as follows: carly or f
| |
| late core melt, no active fission product removal systems, i.e., no sprays; and early and catastrophic !
| |
| C-7 l
| |
| | |
| containment failure, i.e., EOP, DH, EHB and SE. 'Ge final source terms are then presented in a table ithout further detailed discussion; however, a similar pattern of analysis was used.
| |
| NUREG-0956 includes a PWR source term for an early containment overpressure failure without sprays for a transient with a cycling relief valve. The accident is calculated for the Surry plant and is called TMLB' - delta. The source term is contained in table 4.13 on page 4-40 in NUREG 0956.
| |
| Throughout most of this report only iodine (I) and cesium (Cs) will be discussed. They serve as reason-able measures of all isotope release magnitudes for the accident types considered in this study. In other words, if cesium releases are lower by a factor of 5, it is very likely that ruthenium releases will be lower by very close to 5.
| |
| Table 4.13 of NUREG-0956 indicates the I release fraction to be 0.07 and the Cs release fraction to be 0.058 for a transient with cycling relief valve accident sequence resulting in core melt without contain-ment safeguards. This release will then be the basis for determining the releases for transient induced LOCA and small break LOCA core melts.
| |
| BMI-2104 Volume V calculates source terms for large and small break LOCAs and for transients. The
| |
| -esults indicate that transients with cycling relief valves exhibit the best fission product retention in the amary system due to low leakage rates during melting and the tortuous release path through the pres-surizer and associated piping to the containment. About 85% of the I and Cs are retained in the primary system. (See table 7.16 of BMI-2104.)
| |
| BMI-2104 also calculates primary system retention for two small break LOCAs, a hot leg break and a j cold leg break. About 50% of the I and Cs are retained for the hot leg break and 75% for the cold leg break. Looking at it in another way, about two to three times as much material is released from the primary system for a small break LOCA than for a transient with a cycling relief valve.
| |
| Taking a closer look at table 7.16 for the transient indicates that nearly 16% of the released fission prod-ucts are retained in the pressurizer and associated piping to contairanent. This 16% number represents about half of the leakage passing through the structure. Also, the leakage occurs after core slumping when flow rates are quite high through the pressurizer relief valves, i.e., conditiens similar to a transient-induced LOCA. Since a small hot leg break would release three times as much as a cycling relief valve transient, a LOCA from the pressurizer might be expected to release about half of the amount released l by a hot leg break. Hence, it appears reasonable to assign small or transient-induced LOCAs a release i
| |
| of about twice as rub as a transient with cycling relief valves. The reasonableness of this argument umes that all LOCAs will not be hot leg breaks,in which case LOCA releases should be increased by I a factor of 3. l C-P L
| |
| | |
| 1 1
| |
| To calculate releases to the environment based on releases to the containment, this analysis assumes that environmental releases will be proportional to containment releases. Strictly speaking, this is not true xcause the more mass that is present in containment, the more effective natural removal mechanisms should be. Therefore, the assumed method of interpolating releases from the existing NRC references may yield high estimates.
| |
| To summarize the effect of primary system conditions on releases, transients with cycling relief valves without active fission product retnoval, i.e., sprays, resulting in early core melts will be assumed to yield releases consistent with the TMLB' accident in table 4.13 in NUREG-0956. Transient induced and SBLOCAs will be assumed to yield releases a factor of two lower.
| |
| NRC Source Term Results The previous two sections identified qualitative and sometimes quantitative relationships between acci-dent sequence types and source terms. This section presents in tabular form the source terms for each accident type and each containment failure mode. In essence, table C-2 represents the results of step 3 in the NRC Case Study approach described in Section 4.1. Table C-2 provides the corresponding infor-mation to tables K.2 and K.3 in appendix K of the NRC Case Study report.
| |
| 'ECOND ALTERNATIVE: IDCOR SOURCE TERM MODEL This sec' ion provides similar types of information for risk estimates as those presented Lr. the previous section; however, IDCOR is used as a basis for development of the accident sequence types, source terms, and containment failure probabilities. The IDCOR references include the Technical Summary Report (TSR) (5) and the Task 23.1 Report on Zion (1). The IDCOR analysis yields a much simpler source term model. This simplification results from the conclusion that the containment will fail only because of an isolation failure or because of a long term loss of containment heat removal. Therefore, the multiplicity of failure modes in the NRC analysis is reduced to those two, either of which can only occur if all containment safeguards fail and remain unavailable for a long period of time.
| |
| IDCOR has determined that for a large dry PWR like Zion and for all accident sequence types consid-cred in the NRC Case Study, the following holds true: (1) a steam explosion failing the containment could not occur (TSR, section 6.3.5); (2) a hydrogen bum would not occur with sufficient overpressure to fail the contamment either early or late (TSR, sections 6.3.2,6.3.6,6.4.4 and 6.4.5); (3) direct heating resulting from debris dispersion at vessel failure could not occur in sufficient magnitude to fail the con-tainment (TSR, section 6.4.3 and Issue Resolution Report); (4) early overpressure failure could not result because the containment capacity is much gn:ater than any possible steam spike from vessel fail-e and debris dispersion (task 23.1 Zion); and (5) melt through of the basemat will be prevented by the
| |
| .ventual cooling of the core debris (TSR, section 6.4.2).
| |
| C-9
| |
| | |
| _ 4- 7 4 4- 7 -
| |
| 4 a 4 0 0 - 4 -
| |
| 0 0 -
| |
| - 2 -
| |
| _ L. 3 - 1 2 - 3 - 4 -
| |
| o M -
| |
| 3- 9 -
| |
| 3- 3- 9- 2-
| |
| - u 6 - 3 3 - 6 - 6 1 e
| |
| p R 2
| |
| - 3 5 - 2 - 6 1 -
| |
| y T
| |
| - e a 2- 5- 2- 2 5- 1 D B - -
| |
| o r 0 - 8 8 0 -
| |
| 6 2 -
| |
| t o
| |
| S S 2
| |
| - 1 5 - 2 - 3 1 I
| |
| _ Y 4
| |
| B b S- 1 4 1 1 4 -
| |
| 2- 1 2-
| |
| - s - - - -
| |
| 0 n e 1 8 3 1 3- 1 6 4 2 4 1
| |
| o 2- i t
| |
| T 1 1 7 1 7 1 3 1 2 1 I c a
| |
| M r b 5-B F 5- 6- 2- 6- 5- 2 6-e R -
| |
| 1 1
| |
| - 5 -
| |
| d s s 2 5 1 3 7 2 0 0 6 4 n a a
| |
| l e C 1 3 5 1 4 1 7 1 2 9 6 e 5 R 9
| |
| 0 - r 1
| |
| 5- 2- 1 5
| |
| 5 -
| |
| 2-G B- 4 6 8 2 6- 4 2 6 4 5-E I 1 3 7 2 5 1 7 5 4 1 R
| |
| 2 U C
| |
| N K r
| |
| 2 -
| |
| e m e 5 5 5 5 3 5 5 5 5 4 l
| |
| b o
| |
| r X 8, 8 8 8 7 8 8 8 8 1 a f T d e
| |
| r r
| |
| f e
| |
| I n e s d s o e m
| |
| r M it e li T mdi nb e
| |
| c luaa i
| |
| eb r
| |
| u at o Far o t mP S nil et sa 5 20 8 20 8 C men 0 . . 32 9 2 3 5 0 . . 32 9 2 3 o
| |
| R N
| |
| nl ai i
| |
| t 0.0==0 0. ==
| |
| 0 0 0 =
| |
| 0.0==0 0. ==
| |
| 0 0 0 =
| |
| iard t t nnn =BP= PB =
| |
| = F =BP= PB =
| |
| = F oeo Hl OE l OH T 1 C HHOE OI I T 1 C CCC DEES LL M C N DEES LL M C N l
| |
| - l n a d a
| |
| r g mtn n t
| |
| n smisA e a 'F e
| |
| o m ci dn 'F l
| |
| n v f m ya2 cec maC2 f rO C T / t eorL7/
| |
| t t
| |
| n l
| |
| t h vZ l l
| |
| etia r dr d
| |
| e mwvo mAeo i
| |
| c mt f
| |
| e 2 eCc2 u c oel niC rOdC o
| |
| CLiZn A Cisre Z 2.
| |
| 1 g
| |
| | |
| 4- 7- 4 4- 7- 4
| |
| _ a 7 0 0 4 - 0 0 L 1 - 1 2 - 3 - 2 4 o
| |
| M -
| |
| 3- 9- 3- 3- 9 -
| |
| 2-u 3 3 3 6 - 6 1 e
| |
| p R 1 - 3 5 - 2 - 6 1 y
| |
| T a e 2- 5- 2- 2- 5- 1 p B -
| |
| t o r 0
| |
| 8 8 0 -
| |
| 6 2 o S 1 - 1 5 - 2 - 3 1 s
| |
| I y
| |
| 4 B b S- 2- 4- 1 1 4- 2- 1 2 0 s 3-n e 5 8 1 3 1 6 4 2 4 1
| |
| 2- o T 5 1 7 1 7 1 3 1 2 1 it I c M a r b 5-B F_ R 2- 5- 6- 2- 6- 1 5 5 2 6 d e s
| |
| 8 5 3 7 6 0
| |
| 0 6 4
| |
| n s 1 a a C 5 3 5 1 4 1 7 1 2 9 e
| |
| 6 5
| |
| le 9 R_
| |
| 0 - r 5- 2- 1 5- 5 2-
| |
| ) - -
| |
| d e
| |
| G B- 2- 6 8 2 6 -
| |
| 4 2 6 4 5 u E I 7 3 7 2 5 1 7 5 4 1 n R i
| |
| t n
| |
| U o N K r
| |
| 2 C
| |
| ( m e 5 5 5 5 3 7 7 7 7 4 o
| |
| r X 8 8 8 8 7
| |
| 2- f 1 1 1 1 1 C d e
| |
| le r r
| |
| b e a f T I n e s d s o
| |
| m r M ie it e edl T r nb i
| |
| e l uaa c
| |
| at eb i
| |
| r u o Far o t mP S nilt C e s a n 8 5 9 85 9 mEo . . 1 0 2 1 0 2 1 8 4 1 8 4 R nl i 0.0= 0=0 . 2 0 0.0= 0.=0. 2 0 N iai t
| |
| == 0 = == 0 =
| |
| ard t t nnn =BP= PB T
| |
| =
| |
| = F C
| |
| =BP=
| |
| i l PBI T
| |
| =
| |
| = F C
| |
| oeo Hl OE l OI I M
| |
| I N
| |
| f DEES lOE OI l CCC DEES LL C LL M C N l
| |
| - l n a d a '
| |
| mtn n' t
| |
| r g F seaF n
| |
| c o mid2 l c nC misA2 n C v mya/ o aC/
| |
| r T f ceZ f rOZ t
| |
| h v/ r eorL/
| |
| t t t l l
| |
| ~
| |
| n etia l
| |
| r d
| |
| e mwv o f2 mAe do c2 i
| |
| c c
| |
| et roel nieC eCuC r
| |
| oOd A Cisr/e Z CLi/nZ 3 4.
| |
| Q~
| |
| l l
| |
| | |
| 4- 4 -
| |
| 5 0 -
| |
| a - - -
| |
| l 4 - - - 9 - - -
| |
| o M -
| |
| 3-u 4- - -
| |
| ep R_ 5 - - -
| |
| - 1 - -
| |
| y T a e
| |
| p B 2- 2 -
| |
| o r
| |
| - 5 - - 0 - - -
| |
| t - - - - - -
| |
| o s
| |
| S 1 - - - 3 - - -
| |
| I y
| |
| B b 6 8 8 6 8- 8 4 s S - - - - -
| |
| 0 n e 3- 8 6 6 2- 6 2 2 T
| |
| 1 2- o 5 1 3 3 1 3 7 7 it I c a
| |
| M F r b H R 7- 8 8 7 -
| |
| 8 -
| |
| 8 -
| |
| d e s
| |
| 5- 5 3 3
| |
| 4- 0 6 6 n s a
| |
| a C 5 3 2 2 1 7 4 4 e
| |
| 6 le 5
| |
| 9 R 0 7- 8 8 7- 8 8
| |
| ) - r - - - -
| |
| d G B 3- 6 5 5 2- 2 0 0 e E I 5 3 2 2 1 7 5 5 u R n
| |
| it n
| |
| U r o N K 2-C m e 5 5 5 3 7 7 7 4
| |
| (
| |
| o r X 8 8 8 7 1 1 1 1 2 f C d e
| |
| l e r r
| |
| b e a f T I n e d s s o e m
| |
| r M it i
| |
| e edl T r uaanb i
| |
| e il at eb c
| |
| r o u Far o t mP S nil et sa 8 5 9 8 5 9 C men o 1 . . 1 0 8 2 4 1 . . 1 0 8 2 4 R
| |
| N nl iai i
| |
| t 0.0==0 0. ==
| |
| 2 0 0 =
| |
| : 0. =0 =0 0. ==
| |
| 2 0 0 =
| |
| ard t t nnn
| |
| =
| |
| BP= PBI T
| |
| =
| |
| = F =BP= PB1 T
| |
| =
| |
| = F C
| |
| oeo l l i l OE OI M
| |
| I C
| |
| C N
| |
| HHOE DEES O1 M
| |
| 1 C N CCC DEES LL LL l
| |
| - l
| |
| ' a mtn dn/ 'F n F a
| |
| r g /
| |
| 2 r, e a2 t
| |
| n C e mlidC c n// misA/ n p
| |
| y oyaf r maCZ tCrr /
| |
| T f cer f t
| |
| rLo t l t
| |
| hvo l l
| |
| eo n et a' m Ade F'/
| |
| i d
| |
| e mwvF f /
| |
| icc et e 2 eCc2 r
| |
| r oelniC/ oOduC A CieZ sr C L i 'In/
| |
| 5 6 g"
| |
| | |
| L To summarize containment conditions, if containment heat removal succeedr,, the containment will not 1'mless failed already due to containment isolation failure. If containment heat removal fails, the wntainment will fail in the long term due to overpressure unless already failed due to containment isola-tion failure. Finally,if the containment has already failed due to a containment isolation failure, it will be necessary to distinguish between cases where containment sprays are available or not. This conclu-sion implies that the same three containment safeguard states defined for the NRC source tenn will be applicable for the IDCOR source term.
| |
| l i I With regard to primary system conditiens, IDCOR's pnmary system retention modeling indicates that !
| |
| l nearly all of the fissz;n products released during core melting initially deposit in the primary system.
| |
| Therefore, the difference between P. small-break LOCA or transient is insignificant to final source terms.
| |
| )
| |
| Defining accident types for IDCOR source terms need only consider containment conditions. !
| |
| I The containment isolation failure mode probability from IDCOR for Zion is 0.005 (TSR, table 10-3, page 10-21). Other containment failure mode probabilities are either 1 or 0 (TSR and task 23.1 Zion),
| |
| depending on the status of containment systems. If containment heat removal has failed (i.e., NRC Case Study states Z C2 and Z /C2 F'), Long-term Overpressure (LOP) probability is 1.0. If containment heat removal has succeeded (i.e., all other containment states), LOP is 0.0.
| |
| .se source terms for accident types can be taken directly fmm table 10 3 of the TSR except for one case. No results are presented for containment isolation (called impaired containment in the TSR) when sprays are available. This case is estimated to result in releases about a factor of 150 less than the con-tainmer,t isolation case presented. The factor of 150 is derived from table C 2, i.e., from the NRC s.ource tenn analysis. This factor was obtained by calculating the average reduction in person-rem due to presence of sprays for containment isolation failure. These source term! are shown in table C-3.
| |
| TRANSLATING NRC AND IDCOR RELEASE FRACTIONS INTO PERSON REM The two principal references besides the NRC Case Study report used to make this translation are NUREG/ CR-2239 (3.); and EPRI 217-2-6, Sensitivity Assessments in Reactor Safety Analysis (2).
| |
| These references provide sensitivity studies which allow changes in source terms to be reflected in terms of person rem.
| |
| Tables 2.3.2-1 and 2.3.2-2 from NUREG/CR-2239 (shown here as tables C-4 and C-5) provide esti-mates of reductions in latent cancer fatalities for reductions in overall source term magnitude and for reductions in the magnitude of specific isotopes. By noting that latent cancer fatalities are roughly pro-vtional to person rem, one can use these tables to predict the person rem for any release. (The tables
| |
| .: provided here for convenience.)
| |
| C-13
| |
| | |
| 5- 5 a ! 5 -
| |
| 5- !
| |
| 3-L < 6 2 < 9 o
| |
| M -
| |
| 5- 5-u ! 5- 5
| |
| - ! 6 5 -
| |
| e_
| |
| p R < 6 2 <
| |
| y T a e
| |
| p B 5- 5- 9 o r
| |
| ! 4 -
| |
| 5- ! 7 0
| |
| t s
| |
| o S < 6 2 <
| |
| I y
| |
| 3 B b S 5 2 s -
| |
| 9 k n e 5- 4 6 ! 6 s
| |
| a o T 2 3 9
| |
| < 0 it T c a
| |
| d r n
| |
| a fes Rb -
| |
| 3 2- 4-5 2 s R a s - !
| |
| e S e C 2 1 3 < 7 s T le a R R le e
| |
| r O 5 y C B r
| |
| 3 2 4 -
| |
| 5 l D -
| |
| 2 1
| |
| 3
| |
| - I
| |
| < 7 r
| |
| a I I c
| |
| e .
| |
| 3 ht e g
| |
| - r r C m K 5 a l
| |
| e o r e 0 0 0 !
| |
| 0 l
| |
| r b f X 1 1 1 < 1 f
| |
| o a d T e r it y
| |
| r f
| |
| e l i
| |
| n b i
| |
| a e b s
| |
| d o r
| |
| m o
| |
| * p M y r
| |
| e it l
| |
| T etel i a
| |
| n e r u ab o c a i r
| |
| u imb l
| |
| ai o it d
| |
| o Fs r t
| |
| n S t EP o R nl l c eaa 0 0 e O mt r n 5 5 t n nio e a C iet 1
| |
| 0 0
| |
| 0 0
| |
| 1 l
| |
| m D aCd i = = b a it I t P = = F l s
| |
| ndn on o O I I C
| |
| * ia e CaC L C C N
| |
| * v l a ar t t o
| |
| n n ec F 2 s 2 / e 7 CIC 2 2 R t a 0 C r/C W mt u I
| |
| 'F Z d C P i e o t
| |
| p d2 dn na d d dZoZ 0 s
| |
| e ba y nC n rn n/ r aoa a or 'F o a/ 0 c T a 'F t 4 ,
| |
| e Z t t t i t y
| |
| - 2 'F t l l l l 1 s
| |
| n r e2 e 'F 'F e' -
| |
| l d
| |
| e mo mC mZ 2 /
| |
| mFC2 /2 l
| |
| i ima l
| |
| r i
| |
| c e2 e/
| |
| rZ eC r/C cC r ZC S A
| |
| i s e n
| |
| c rC o o r oZ / // s e e A C 7, Co C/Z o/rZ CZo/ W P G 1 2 3 4. *
| |
| * gE
| |
| | |
| EPRI 217-2-6 also provides valuable information for estimating consequences without the direct use of
| |
| 'he Calculations of Reactor Accidents - Version 2 (CRAC2) code (1Q). Table 3 in that report (shown
| |
| .iere as table C-6) provides the percentage of person-rem attributed to nine key isotopes. This informa-tion is panicularly applicable to situations where changes in source term magnitude are not uniform, and where they occur in the isotopes that are not listed in table 2.3.2-1 of ref. (E). In panicular, table 3 is valuable in rectifying the differences in the source terms fmm WASH 1400 and current estimates, speci-fically with regard to ruthenium, strontium, barium and lanthanum.
| |
| The usefulness and applicability of this approach can best be judged by comparing the sensitivity studies perfonned in the NRC Case Study with those contained in the two references. Table K.4 in the NRC Case Study (summarized hem as table C-7) provides person rem estimates for the seven WASH-1400 release categories for six different conditions, namely within 50 miles and total together, with the release category source terms divided by 1,3 and 10. Both of the reference studies use PWR 2 as the base case for sensitivity analysis. Comparing the latent cancer numbers in table 2.3.2-2 for 10% SST1, i.e, one tenth of PWR 2, to the total dose for table C-5's upper bound, i.e. full PWR 2, ar.d its lower bound, one tenth of PWR 2, one can see the ratios are very close.
| |
| Under more careful consideration, this confirmation of the acceptability of the numbers does not entirely hold true. First, the NRC Case Study uses as its value measure the population dose within 50 miles, not atal population dose. Second, population dose within 50 miles does not change at the same ratio for the two cases. Hence, there may be some problems in application for cases within 50 miles.
| |
| The NRC Case Study has noted that doses within 50 miles are less sensitive to source term changes than total doses. This particular relationship is caused by the effect of decontamination and interdiction actions on doses. Decontamination and interdiction actions can be characterized as 1) occurring close to the plant, i.e., within 50 miles, and 2) having a significant effect on total dose only when large releases occur.
| |
| A WASH-1400 source tenn will probably result in significant amounts of decontamination within 50 miles of the plant. If such a source term is reduced by a factor of three,it willin tum lead to less decon-tamination. (The economics of decontamination and interdiction exhibit threshold effects and are l therefore non-linear.) Less decontamination and interdiction will mean that a higher percentage of the released fission pmducts will remain to expose the population. 'Iherefore, as source terms are reduced linearly, person rem or latent cancers will be reduced more slowly. This relationship is indicated clearly l by table C-5. I l
| |
| I owever, as the magnitude of the source tenn is reduced and interdiction and decontamination play l smaller roles in the dose reduction process, the relationship between source term reduction and doses l
| |
| C-15 )
| |
| l !
| |
| | |
| Table C-4 Sensitivity of Mean Consequences to Reductions in SST) Release Fractions ofIodine, Cesium, and Telluriuma ,b (Table 2.3.21 from NUREG/CR-2239) i l
| |
| 1 Latent Acute Dose C Accident Early Early Cancer Area of Land selease Fatalities Injuries ratalities Bone Marrow Thyroid Interdiction SST)
| |
| (Standa rd ) 100 b 200 100 100 100 100 504 1 75 75 98 85 60 100 lot I 60 55 95 70 30 100 Cl 1 50 55 95 65 20 100 506 Cs 95 95 90 95 100 55 104 Cs 90 95 75 90 100 15 04 Cs 85 90 60 90 100 1 504 Te 75 65 95 85 90 100 lot Te 50 45 90 70 80 100 04 Te 45 40 90 65 80 100 504 1,Cs 70 70 90 80 60 55 lot I,Cs 45 55 70 60 30 15 On I,Cs 40 50 55 55 20 1 50% 2,Cs,Te 40 45 85 60 50 55
| |
| : a. Assumptions: 1120 MWe reactor, Indian Point site, New York City meteorology, Summary Evacuation.
| |
| : b. All consequences normalized to 100 for source term SST1.
| |
| : c. Relat:ve doses are approximately independent of distance. l
| |
| )
| |
| C-16 l
| |
| | |
| da ee t r cA 5 0 5 1 i 0 5 1 dd 0 rn 1 y ea g tL o n l o
| |
| I r
| |
| o e
| |
| d t i e o 0 0 5 1 m r 0 5 1 c y 0 y sh 1 t i .
| |
| eT C s s e o k s Dw r a Ib T a, o o . g Ss er Y . e t r Se s ua 1 c e na cM 3 6 1 8 w T n l iG A 0 5 1 1 e S a b s
| |
| n le9
| |
| )
| |
| e 0 N S t o n 1 s n iob3 o2 o ,
| |
| m i t
| |
| c e r d t uN2 -
| |
| B -
| |
| t e p dt R i t f e e pC S o c Re/ c r e x oxGEE e t c t e t
| |
| cs n r n 5- s sR et ne i u e s C c nU ai o o d e n
| |
| e eN Ct 4 2 9 5 P s n p ic um qem i 0 7 3 1 e p
| |
| o b
| |
| e l o tl 0 n r t a sEr na 1 a o e o T nl f
| |
| et i f d s ol I t a d n i C fA-2 aF n 0 i L I 0 l no3 a 1 y l e s2 n , l a Mole s r o e f it b e o t t r ocaa yi 5 4 2 1 t a o yrT l r 0 3 c d m f i F(
| |
| t ru 0 a e i iv es aj 1 e z x d t
| |
| En r i o e ia s e I l r c nl ee e a p u f
| |
| m p d SR 9 M r a e s . o r e 3 0n n e i 2 0 2o r s yt b . . 1i s a n li 0 0 1 0 0 1t e o rl 0 3 a c s i aa 1 u n e t Et : c e s c a sa u o a F nv q d r oE e f i s e t y n v e
| |
| ) pr o i s t d d d d d ma c t a ne r i i i i um a e es la T T T T sm su l l e
| |
| l e
| |
| da Td S S S S l ie Sn S S S S AS A R R cl Sa ce t % % % t AR S 0 0 5
| |
| .I . . .
| |
| ( 5 1 a b c d Ob l
| |
| | |
| Table C-6 Results of Parametric Calculations for PWR Release Category 2 (Table 3 from EPRI 217-2 6, Sensitivity Assessments in Reactor Safety Analysis)
| |
| Sv. e, Oe.cri,iion ave,..e e.. < pro.setisi, Pes. *
| |
| .0i'1 Probability Averste)
| |
| (Pef Ac cidect )* 1 Lost W a p - B en Acute Lost Wa n. p ge. AChte 0 ts106) Tets11 ties ($a109) (mle ) Fateltties'$n10 )
| |
| 1 Reference 3.1 62 1.8 31 2300 5.6 1.9 R 2 No Ral' 3.4 64 R R 2 Deposasaan
| |
| %elocity n .2 59 64 .58 63 1200 3.0 4 Deposttnen belocity a 6 98 61 1.8 9.3 1100 8.1 S No Plume Rase 2.9 140 1.6 26 2800 5.6 6 No Plume Base S.6 from Heat 32 65 1.8 33 2000 7 Plume Rase Fror 66 Racapactisnty 2.0 106 1.7 27 2700 8 No Ivacuation 3.3 140 8 36 9100 8 9 Evacuation Rate 4200 8 a .6 3.2 90 8 38 10 Esacuation Rate 1200 a2 3.1 37 R SI R Il pessh al 53 R 30 1600 8 1sacuers * .0 31 12 Ressw el Esscuees * .2 3.1 71 8 31 3100 R 13 Isacuation to 33 70 P 36 4S00 R 10 Males 14 twarustaan to R 30 unles 26 8 9 15 R 0 0 0 0 0 0 15 Er Cal /
| |
| 16 Br only .45 1.0 1.7 9.4 49 63 Ru Only .14 .64 0 1.6 30 0 17
| |
| .22 2.3 .09 2,6 77 .17 18 Te On1) 1 Only .33 11 .41 3.3 330 1.5 19 0 .34 0 n 20 1e Only .01 0 21 Cs Only .65 1.5 1.2 7.2 61 3.8 Ce only .12 0 .01 1.4 0 .32 22 Pu Only 26 0 .01 2.9 0 .32 23 24 All Other 0 7.3 140 0 Elseents .66 S.6 25 Population 1 (1%) 11.9 360 3.0 25.1 B &1 26 Population 1-3 (551 12.0 250 3.1 R R R 27 Populatson 1.4 (10%) 9.4 230 2.8 R R R l
| |
| 1 1
| |
| 1 e
| |
| I i
| |
| i I
| |
| C-18
| |
| | |
| l t
| |
| Table C-6 (Continued) 4 Results of Parametric Calculations for PWR Release Category 2 (Table 3 from EPRI 217-2-6, Sensitivity Assessments in Reactor Safety Analysis)
| |
| .w, me., rie n en . e,s. c,.6 er,.,n. iii., ee.s .
| |
| t*e r aceseca't* ^^: revdatstti, a erer. i UnN as ute t .CEr% a. . i .
| |
| jtu i tal0 6) Densittle, ilsin81 'aleo) es+ . l i t i es. S = 10' )
| |
| i sa Popuistson 3-s (60%) 6.3 310 3.2 R R R 39 8mseldsng Factor * .1 70 61 R 30 2000 9 30 Seleiding 7setoe
| |
| * 1.0 36 #10 R SS 4000 8 31 Damage Thre4-holds a .6 R R 34 R R 10 32 Damage Teres.
| |
| Ronas a 2 R R .97 9 R 31 33 Release Reasht
| |
| + 3Sy 3.3 34 l'3
| |
| , . 37 140 4.8 34 Release Neagnt
| |
| * 200m 3.4 24 96 10 140 40 SS db Day temasse tapress F R 1.f R R 40 36 390-Day kanamue Espenses R R 2.0 R R ?.!
| |
| 37 ho Lapenses R R 13 m R 27 38 Ishalation Onli 2.9 de R 29 1400 R 39 Snane Onlp .09 25 P . 94 16 9 40 Depositsen On!) .33 3.8 A 34 140 R R keterence
| |
| 'These numbers ref er to t he Ref erence Acc acent only , PWR Release Category e2.
| |
| l t
| |
| C 19
| |
| | |
| Table C-7 Source Terms for Release Categories from Table K.4 in Point Beach NRC Case Study Pooulation Dose Release Categorv Unoer Bound baseline Imwer Bound 1 7.9E+5 6.6E+5 4.4E+5 2 8.0E+5 7.5E+5 5.6E+5 3 9.6E+5 6.2E+5 3.7E+5 4 5.4E+5 2.7E+5 1.3E+5 5 2.2E+5 1.0E+5 4.8E+4 6 5.3E+4 2.2E+4 9.8E+3 7 2.6E+3 1.7E+3 1.4E+3 l
| |
| C-20 l
| |
| | |
| l l
| |
| released fission products will remain to expose the population. Therefore, as source terms are reduced
| |
| 'early, person rem or latent cancers will be reduced more slowly. This relationship is indicated clearly
| |
| -y table C-5. .
| |
| l i
| |
| However, as the magnitude of the source term is reduced and interdiction and decontamination play i 1
| |
| smaller roles in the dose reduction process, the relationship between source term reduction and doses I will become more linear. The closeness to linearity is exhibited by the last three rows in table C 5.
| |
| Funber, if one compares the ratio of the baseline source term dose within 50 miles to the lower bound source terra dose within 50 miles in table C-7 to the ratio of latent cancers for a 10% SSTI and a hypothetical (i.e., interpolated) 33% SSTl release, one will see that the ratios are quite similar.
| |
| Therefore, as releases are reduced below the level of one third of a PWR 2 release, the source term ratio-ing method proposed here will become accurate for doses within 50 miles. Because the NRC and IDCOR source temis are all below this threshold, the simplified consequence method should prove quite accurate, especially when compared to other uncertainties in the study.
| |
| The following tables (tables C 8 and C-9) present the containment failure probabilities and person rem estimates for both the NRC central estimate and IDCOR source terms for each of the previously defined cident types. The containment failure probabilities for NRC source terms are presented for two ntral estimate cases. Cases are presented with and without direct heating. The first case,i.e., central estimate without direct heating, is used as the base case for Point Beach's NRC source term. Point Beach's reactor cavity appears to lack pathways for dispersion of sufficient debris to make direct heating an issue.
| |
| The IDCOR source term table is much simpler due to the smaller number of containment failure modes determined to be possible. Both the IDCOR and NRC results can be compared to the NRC Case Study results in Tables 2.2,2.3, and 2.4 of that study.
| |
| THIRD ALTERNATIVE: SOURCE TERM MODEL FOR DRY CAVITY PWRs The following reviews and provides alternative estimates for containment failure likelihood, release category assignment, and source terms. As opposed to the previous source terms, this alternative includes dry cavity cases.
| |
| First, an alternative mapping ef accident sequences to release categories is presented. The hPC Case Study is inconsistent with the findings of the Containment leads Working Group (NRC-directed and l ,
| |
| l 1
| |
| l C-21
| |
| | |
| Table C-8 NRC Containment Failure Probabilities and Person-Rem by Accident Type (Summarized from Text)
| |
| Central Central Estimate Estimage w/o Direct w/ Direct Acrident Tvoe Heatine Heating Person-Rem Accident Type 1 Early Containment Failure 0.00 0.005 SE+5 Acident Type 1 Late Containment Failure 0.52 0.52 3E + 3 Accident Type 1 Basemat Melt-Through 0.09 0.09 3E + 3 Accident Type 1 Containment isolation Failure 0.002 0.002 SE+5
| |
| .ccident Type 1 Design Leakage, i.e., No Failure 0.38 0.38 SE+1 Accident Type 2 Early Containment Failure 0.00 0.005 8E+5 Accident Type 2 Late Containment Failure 0.52 0.52 6E + 3 Accident Type 2 Basemat Melt-Through 0.09 0.09 6E + 3 Acx:ident Type 2 Containment Isolation Failure 0.002 0.002 8E+5 Accident Type 2 Design i nhge, i.e., No Failure 0.38 0.38 1E+2 Accident Type 3 F.arly Containment ilure 0.00 0.01 SE+5 C-22
| |
| | |
| Table C-8 (continued) l NRC Containment Failure Probabilities and Person-Rem by Accident Type (Summarized from Text)
| |
| Central Central Estimate Estimage w/o Direct w/ Direct Accident Tvoc Heatine Heating Person-Rem Accident Type 3 Late Contamment Failure 0.23 0.23 3E + 3 Accident Type 3 Basemat Melt-Through 0.28 0.28 3E + 3 Accident Type 3 Containment Isolation Failure 0.002 0.002 SE+5 Accident Type 3
| |
| 'esign Leakage,
| |
| .e., No Failure 0.49 0.49 5E+1 Accident Type 4 Early Containment Failure 0.00 0.01 8E+5 Accident Type 4 Late Contamment Failure 0.23 0.23 6E + 3 Accident Type 4 Basemat Melt Through 0.28 0.28 6E + 3 Accident Type 4 Containment Isolation Failure 0.002 0.002 8E+5 Accident Type 4 Design leakage, i.e., No Failure 0.49 0.49 1E+2 Accident Type 5 Early Containment Failure 0.00 0.01 SE+4
| |
| :cident Type 5
| |
| , ate Contamment Failure 0.23 0.23 3E + 3 C-23
| |
| | |
| Table C-8 (continued)
| |
| NRC Contamment Failure Probabilities and Person Rem by Accident Type (Summarized from Text)
| |
| Central Central Estimate Estimage w/o Direct w/ Direct Accident Tvoe jinting Heatinc Person-Rem Accident Type 5 Basemat Melt-Through 0.28 0.28 3E + 3 Accident Type 5 Containment Isolation Failure 0.002 0.002 3E + 3 Accident Type 5 Design Leakage, i.e., No Failure 0.49 0.49 SE+1 Accident Type 6 Early Containment c ailure 0.00 0.01 1E+5 accident Type 6 Late Containment Failure 0.23 0.23 6E + 3 Accident Type 6 Basemat Melt Through 0.2.8 0.28 6E + 3 Accident Type 6 Containment Isolation Failure 0.002 0.002 6E + 3 Accident Type 6 Design Leakage, i.e., No Failure 0.49 0.49 1E+2 C-24
| |
| | |
| Table C-9 IDCOR Containment Failure Probabilities and Person-Rem by Accident Type (Summarized from Text)
| |
| Containment Failure Accident Tvoe Mode and Probability Person-Rem Accident Type 1 Late Overpressure = 0.995 SE+4
| |
| ' ecident Type 2 Containment Isolation = 0.005 SE+5 Accident Type 3 Containment Isolation = 0.005 4E + 3 Accident Type 4 No Containment Failure = 0.995 SE - 1 i
| |
| i
| |
| ' l l
| |
| C-25
| |
| | |
| l l
| |
| i l
| |
| These two assumptions will ensure that even an adiabatic quench of the core debris ejected from the reactor vessel at the time of vessel failure will not result m containtnent failure. j 1
| |
| : 3. For high RCS pressure at the time of reactor vessel failure,50% of the core debris goes to the containment where water is available for quench and short-term cool-ing, and 50% remains in the reactor cavity area.
| |
| : 4. For low RCS pressure at the time of reactor vessel failure,100% of the core debris remains m the reactor cavity area.
| |
| 1
| |
| : 5. The reactor cavity area is greater than 15 square meters so that a coolable debris bed can be established if water is present; the 1200 kw/m2 dryout heat flux can be established at 1% decay power. l
| |
| : 6. The containment configuration is such that direct air heating of the containment is unlikely or physically impossible.
| |
| This leaves only the question of water availability to the reactor cavity as an uncertainty to be dealt with.
| |
| In some plants, water from spilled coolant or RWST injection would flow to the cavity after topping a relatively shallow curb. In others, the structures preclude water flow to the cavity. Normally for all high-pressure cases, at least the accumulator volume would discharge to the cavity. Three distinct cases can be considered which bound all possible configurations:
| |
| o Cavity is always dry o Cavity is always wet o Cavity is wet when RWST is injected; otherwise dry.
| |
| Man.h intermediate cases can exist; these configurations are used as bounding in order to simplify further considerations.
| |
| Table 7.2 in the NRC Case Study report can be constructed as follows, with the generalized conditions as listed below:
| |
| : 1. For all cases, the in-vessel steam explosion mode (ot) is maintained at 10-2 (jow pressure) and 10-4 (high pressure).
| |
| : 2. The ex-vessel steam explosion (Se)is set at 10-2 for all cases based on assump-tions a, b and f, and the work of the NRC's Contamment Loads Working Group.
| |
| : 3. For all cases, the probability of impaired containment isolation function ( ) is taken as 10-3 It should by noted faat these estimates for steam explosion represent the pessimistic estimate in draft
| |
| 'UREG-0956. At these values, they are significant contributors to the overall source term.
| |
| C-26
| |
| | |
| Interviews with WEP personnel indicate that the Point Beach cavity would remain dry without direct zater injection. Direct water injection can be provided by containment sprays or by injection of water into the vessel after vessel failure. It appears that the refueling canal would direct containment spray away from the cavity. Because the area for direct spray to the cavity is small, spray is assumed to not impact cavity conditions.
| |
| Direct injection to the vessel can occur from a number of sources. The most likely sources are the accu-mulators. For most of the Point Beach scenarios, the pnmary system will remain intact during core melting at a pressure which will prevent the accumulators dumping until vessel failure. After vessel failure, the accurnulator water inventory will delay cavity dryout for more than an hour. For a station blackout scenario, recovery of power would allow HPSI to inject into the cavity. The time between ini-tial core uncovery/ damage and cavity dryout implies a factor of between 3 and 5 for recovery prior to cavity dryout.
| |
| For that portion of core melts which occurred despite the availability of 1 or 2 charging pumps, cavity dryout might also be prevented. However, for the 1 charging pump case, decay heat might outpace flow for some time leading to temporary dryout.
| |
| The following analysis prescnts containment failure modes and release category assignments for both
| |
| , vet and dry cavity conditions. The case of temporary dryout is not considered, but is bounded by the dry cavity case. Source terms for the release category assignment are discussed subsequently.
| |
| Case A - Always Drv. In all cases, the containment leak failure mode (I - for intact containment) is nearly zero since the basemat always is predicted to be penetrated for these cases (conservative assump-tion on basement penetration in the absence of water). The basemat melt mode is 1.0 - (n+ +y, Se+Sl+I) and the release category is always 7 since basemat failure is always late, and at these times the release is independent of containment post-accident activity removal status due to the efficiency of natural proces-ses, given long time periods.
| |
| From this point, the y, Se and Si probabilities are determined from the sequence descriptions. For long-term overpressure, only Z or /Z are considered. For /Z, the probability of Si is very small; for Z, con-tainment overpressure failure is likely to occur before basemat penetration for HP (high pressure)
| |
| I sequences and about a 50-50 split for LP sequences, based on Millstone-3 (H) and Seabrook (H) analy-ses. These Si can be assigned a release category 6, irrespective of fission product removal systems due to the late failt_re time and action of natural removal processes. The y, Se mode is more difficult to
| |
| :stablish without exact analyses, but the important considerations provide some bounds:
| |
| : 1. For /Z sequences, the magnitude of the H2 bum / steam spike must be greater than for Z sequences in order for containment failure to occur, since containment l
| |
| 1 l C-27
| |
| [ .
| |
| | |
| pressure at vessel failure will be 1 to 3 bars higher for Z (vs./Z). However, the higher base pressure for Z means more steam is in the containment, and thus steam inerting will occur more quickly. For high pressure (RCS) vessel failure states, the containment would quickly steam inert due to quenching outside the cavity and accumulator water discharge onto the debris. For low pressure cases, there is no quenching (because core debris would remain in the cavity which would be dry) and no subsequent accumulator water discharge; inening is much less likely.
| |
| : 2. In vessel hydrogen production is independent of event sequence (small differ-ences amongst sequences, compared to the total quantity generated in-vessel).
| |
| Ex vessel hydrogen production is only significant from core-concrete interaction.
| |
| This leads to the following conclusions:
| |
| : a. For /Z (which denotes successful containment heat removal by either fan coolers or spray in the NRC study nomenclature) with LP vessel failure, an early contamment failure is very improbable, but a late failure due to H2 burn may be likely due to H2 generation from concrete attack, if it is permitted to accumulate beyond its flammability limit. However, it would not appear diff-crent in release category from a St . Thus, a probability of <0.1 might be reasonable with a release category of 6 with sprays opera-tional (///C2) and possibly as high as category 4 when no syays are operational (/ZC2) at time of core melt.
| |
| : b. For /Z with HP vessel failure, an early failure may be slightly more probable due to the quenching spike combined with an H2 burn.
| |
| The late H2 burn failure is the same as the /Z LP case.
| |
| : c. For Z (which denotes containment heat removal not successful) i with HP vessel failure, the containment quickly becomes steam inert and no y or y/Se failure is likely,
| |
| : d. For Z with LP vessel failure, steam inerting does not occur imme- j diately. A short term y/St failure is unlikely due to the low Se i component. A long term y failure is questionable due to the potential for steam inerting prior to flammabdity (steam and H2 from core-concrete interaction). Thus, the use of a 0.5 for y is ,
| |
| used. j Thus, table 7.2 of the NRC Case Study can be written as shown in table C-10. 1 Case B - Always Wet. This is the simplest case because basemat penetration does not occur, thus mini-mizing large H2 generation and basemat failure (c). The absence of large H2 generation from concrete decomposition implies that large hydmgen burns are unhkely (y failure mode). In these cases, the l
| |
| amount of water in the containment may be important in that with only RCS and accumulator values, some concrete penetration will occur late in time after the water volume is vaporized. However, for cases where the RWST is also injected, concrete penetration would always be after containment failure l
| |
| ' v overpressurization.
| |
| l C-28 l f
| |
| L- -
| |
| | |
| f.
| |
| Table C-10 i Cor:tainment Failure Mode, Probability and Release Category Assignments for PWRs with Potentially Dry Cavities (Case A: Dry Cavity) i Containment Systems RCS Sequence Pressure g Q 16e he y $1 g 1
| |
| /Z gp IE-2 1E-3 IE-3 1E-2 1E-1* IE-4 9E 1 1E 2 1- 5 2 2 4/6 6 7 8 HP 1E-4 1E-3 1E-2 IE-2 1E-1 IE-4 9E-1 1E-2 1 5 2 2 4/6 6 7 8 Z/C2 Lp 1E-2 1E 3 IE-3 1E-2 5E 1 2.5E-1 2.5E-1 1E-4 1 5 2 2 5 6 7 8 HP IE-4 1E 3 IE-3 1E-2 IE 2 8E-1 2E-1 1E 4 1 5 2 2 5 6 7 8 ZC2 LP 1E-2 1E-3 .IE-2 1E-2 SE-1 2.5E-1 2.5E-1 1E-4 1 5 2 2 4 6 7 8 HP 1E-4 1E-3 IE-2 1E-2 1E-2 8E-1 2E-1 1E 4-1 5 2 2 4 6 7 8 For y modes the release category is 6 with spray and heat removal,5 with spray but no containment heat removal, and category 4 with no spray at time of core melt, i
| |
| C-29 I i
| |
| _________________u
| |
| | |
| For cases with long-terrn containment cooling available (/Z), containment failure is predicted to be rare nless the hydrogen production exceeds 75 to 100% Zr-H2O reaction.
| |
| For cases without long-term cooling (Z), the ultimate containment pressurization transient is nearly identical for all cases, except maybe for timing of failure. Sufficient steam is produced due to debris cooling in order to render the containment steam inerted at times past the immediate vessel failure transients; (yis very nearly zero). Only the combined steam spike and hydrogen burn are likely to cause any appreciable contamment failure (y, Se). If we use the same probabilities for this case as the Case A; Z; HP; C2 (or C2), we obtain the version of table 7.2 shown in table C-11.
| |
| Case C - Wet if RWST Iniection. Otherwise Drv. This source term is a combination of the previous two cases but requires distinction of RWST injection by either sprays (C2 or /C2) and ECC. In the latter case, the early rnelt (EM) is taken to denote no ECC injection of RWST, whereas the late melt (LM) is taken to denote RWST injection by RWST. Table C-12 presents these results.
| |
| Review of NRC activities regarding containment failure probabilities for severe accident sequences reveals a general agreement with these values. In the Seabrook evaluation (NUREG/CR-4540. 2/86),
| |
| for the case of containment heat removal available (/Z) the dominant failure mode was I with 0.99.
| |
| Basemat melt-through (c) was given a 0.40 probability for sequences without containment heat removal
| |
| .d no RWST injection (Z/C2 and ZC2 for HP and LP late melt). The probability of late overpressure failure was given as 0.60 for these cases.
| |
| In the Millstone 3 evaluation for cam with containment heat removal (/Z), the no-failure probability (I) is given as 0.95 with a 0.05 probability of basemat failure (c). For events without containment heat removal (Z), the dominant failure mode was given as late overpressure (61 ) with a probability of 0.90 for the TMLB sequence; a 0.1 probability of basemat failure (c) was used for the other failure mode.
| |
| I Fission product releases for a variety of NRC and industry studies are presented in table C-13.
| |
| From this information, a logical conclusion for multipliers to WASH 1400 release fractions to account l for new research since 1975 would be:
| |
| Early Failures:
| |
| Upper Bound = 0.5 x WASH-1400 (based on Seabrook early failure)
| |
| Baseline = 0.1 x WASH-1400 (based on BMI 2104 carly failure)
| |
| Lower Bound = 0.01 x WASH-1400 (based on IDCOR)
| |
| C-30 L_---- --- _
| |
| | |
| i I
| |
| Table C-11 Containment Failure Mode, Probability and Release Category Assignments for PWRs with Potentially Dry Cavities (Case B: Wet Cavity)
| |
| Containment Systems RCS Sequence Pressure g S yle he y bJ E ]
| |
| /Z IE 2 1E-3 1E-3 1E-2 1E-4 1E-4 1E-4 IE-0 LP 1 5 2 2 4/6' 6 7 8 1E-4 1E-3 IE-2 1E 2 1E-4 1E 4 1E-4 1E-0 HP 2 6 7 8 1 5 2 4/6 Z/C2 IE-2 1E-3 1E-3 1E-2 1E-4 1E-0 1E-4 1E-4 y'p 1 5 2 2 5 6 7 8 IE-4 IE 3 1E 3 1E-2 1E-4 1E-0 1E-4 1E 4 HP 2 2 5 6 7 8 1 5 ZC2 IE-2 1E-3 1E-2 1E 2 1E-4 1E 0 1E 4 1E-4 Lp 1 5 2 2 4 6 7 8 1E 4 1E-3 1E-2 1E 2 1E 1E-0 1E-4 1E-4 HP 2 2 4 6 7 8 1 5 Release category 6 with containment spray, category 4 without spray.
| |
| C-31
| |
| | |
| l Table C-12 Containment Failure Mode, Probability and Release Category Assignments for PWRs with Potentially Dry Cavities (Case C: Wet Cavity After RWST 1njection)
| |
| Containment Systems RCS Core Sequence. Pressure Mill n E yle Se y 51 E I
| |
| /Z/C2 Lp Ag 1E-? 1E-3 IE 3 1E-2 1E-4 IE-4 1E-4 1E-0 1 5 2 2 6 6 7 8 1E-4 IE-3 1E-2 1E-2 1E-4 1E-4 JE-4 IE-0 HP All 1 5 2 2 6 6 7 8
| |
| /ZC2 1E-2 1E-3 1E-3 1E-2 1E-1 1E-4 9E-1 1E-4 LP EM 2 2 4 6 7 8 1 5 gp gy .IE-2 1E-3 IE-3 1E-2 1E-4 1E 4 IE-4 1E-0 1 5 2 2 4 6 7 8 1E-4 1E-3 IE-2 1E-2 1E-1 1E-4 9E-1 1E 4 HP Ehi 1 5 2 2 4 6 7 8 1E-4 1E-3 IE-2 1E-2 1E-4 1E-4 1E-4 1E-0 HP Lh1 1 5 2 2 4 6 7 8 Z/C2 IE-2 1E-3 1E-3 1E-2 1E-4 1E-0 1E-4 IE-4 LP All 1 5 2 2 5 6 7 8 IE-4 1E 3 IE-3 1E-2 1E-4 1E-0 1E-4 IE-4 HP All 1 5 2 2 5 6' 7 8 ZC2 IE-2 1E 3 IE-2 1E-2 SE-1 2.5E-1 2.5E-1 1E 4.
| |
| LP EM 2 4 6 7 8 1 5 2 1E-2 1E-3 1E 2 1E-2 1E 4 1E-0 1E-4 IE-4 LP LM 1 5 2 2 4 6 7 8 IE-4 IE-3 1E-2 IE-2 1E-2 SE-1 SE-1 1E-4 HP EM 1 5 2 2 4 6 7 8 IE-4 1E-3 IE-2 1E-2 1E-4 1E 1E-4 IE-4
| |
| & LM 1 5 2 2 4 6 7 8 C-32
| |
| | |
| Table C-13 A Comparison of Fission Product Release Fractions from Various Studies 1
| |
| lodine C s- Rb '' TeISb' ~ Ba-Sr Ru La NRC-Point Beach Early Fail - PWR-1 0.7 0.4 0.4 0.05 0.4 3(-3)
| |
| Late Fail - PWR-2 0.7 0.5 0.3 0.06 0.02 4(-3)
| |
| No Fail - PWR-9 1(-7) 6(-7) 1(-9) 1(-11) 0 0 Basemat - PWR-6 8(-4) 8(-4) 1(-3) 9(-5) 7(-5) 1(-5)
| |
| NRC - Millstone 3 (9/85) i Early(2) 0.5 0.6 0.2 0.07 0.02 6(-3)
| |
| Late (7) 9(-3) 3(-1) 3(-1) 3(-2) 2(-2) 4(-3)
| |
| No fail (12) 6(-6) 1(-6) 9(-7) 2(-7) 8(-8) 1(-8)
| |
| Basemat (10) 8(-4) 8(-4) 1(-3) 9(-5) 7(-5) 1(-5)
| |
| NRC - Seabrook (2/86)
| |
| Early 57V 0.31 0.31 0.32 0.034 0.02; 4(-3)
| |
| Late 53 53V 2.4(-2) 2.4(-2) 3(-2) 2.6(-3) 2.3(-3) 3.9(-4)
| |
| No fail S5 3.5(-8) 3.5(-8) 6.1(-9) 4.0(-9) 1.2(-9)1.2(-10)
| |
| Basemat S6 3.6(-3) 3.6(-3) 6.7(-4) 4.4(-4) 1.3(-4)1.3(-5)
| |
| BMI-2104 - Surry (Vol Vill, Tb 7.21)
| |
| Early TMLB'-6 7(-2) 6(-2) 1(-1) 5.8(-2) 5.3(-3)1.0(-3)
| |
| Late None Basemat TMLB c 2.B(-3) 6(-4) 8.5(-2)1.7(-2) 2.4(-5)4.4(-4) 10COR - ZION Late 1.7(-3) 1.7(-3) 2(-5) 1(-5) 1(-5) 1(-5)
| |
| J C-33
| |
| | |
| i Late Failures:
| |
| Upper Bound = 0.1 x WASH 1400 (based on Seabrook late failure)
| |
| Baseline = 0.01 x WASH 1400 (based on extension of BMI-2104 early failure)
| |
| Lower Bound = 0.003 x WASH-1400 (based on IDCOR)
| |
| No Failure:
| |
| All cases = WASH 1400 release category 9 Population doses can be estimated by linear extrapolation (see earlier discussion). The doses correspon- ,
| |
| l ding to these source terms are contained in table C-7.
| |
| l
| |
| )
| |
| C-34
| |
| | |
| REFERENCES l
| |
| : 1. J. A. Gieseke, et al. Radionuclides Release Under Soecific LWR Accident Conditions, vols. IV and V. Columbus, OH: Battelle Columbus Laboratories, July,1984. BMI-2104.
| |
| : 2. M. Silberberg, et al. Reassessment of the Technical Bases for Estimating Source Terms, draft report for comment. USNRC, July 1985. NUREG-0956.
| |
| : 3. W. R. Crammond, et al. Shutdown Decay Heat Removal Analysis of a Westinghouse 2-Looo Pressurized Water Reactor Case Study. March,197. NUREG/CR-4458.
| |
| : 4. Reactor Safety Study - An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants, WASH-1400. USNRC, March 1975. NUREG-75/014.
| |
| : 5. Nuclear Safety Analysis Center and Duke Power Comoany. Oconee PRA: A Probabilistic Risk Assessment of Oconee Unit 3. Palo Alto, CA: Electric Power Research Institute, June 1984. I I
| |
| NSAC-60.
| |
| : 6. Nuclear Power Plant Response to Severe Accidents. Technical Summary Reoort. Technology for Energy Corporation, November 1984.
| |
| : 7. IPCOR Task 23.1 -Interrated Containment Analysis. Zion Nuclear Plant. Technology for Energy Corporation, December 1984.
| |
| : 8. Technical Guidance for Sitine Criteria Development. Sandia National Laboratories, SANT81-1549, December 1982. NUREG/CR-2239.
| |
| >. J. E. Kelley, et al. Sensitivity Assessments in Reactor Safety Analysis. Palo Alto, CA: Electric Power Research Institute, February 1976. EPRI 217 2-6.
| |
| : 10. CRAC2 Model Description. Sandia National Laboratories, S AND82-0342, March 1984.
| |
| NUREG/CR-2552.
| |
| : 11. NUREG/CR-4143.
| |
| : 12. NUREG/CR-4540.
| |
| I C-35
| |
| | |
| Appendix D VALUE IMPACT ANALYSIS DETAILS This appendix presents a series of tables and backup information to the value impact analysis in section 10.' Originally it was envisioned that this portion of the report would include a variety of sensitivity.
| |
| studies (e.g., discount rates). However, since this review's results generated such small V-1 ratios, the value of sensitivity studies seemed insignificant. Also, sensitivity studies indicating the impact of modeling changes were presented in each of the previous sections. These assumptions, data, and methods are much more important to the conclusions of the study than changes in the V-I model input -
| |
| such as the discount rate.
| |
| The first part of this appendix presents a reevaluation by Wisconsin Electric Power of the cost estimates made by the NRC. These revised estimates provide new cost data for this review (see table D-1). They
| |
| . include "two-unit effects," yet two units were not evaluated in the EPRI/WOG core melt analysis. Com-letely incorporating costs of these modifications to two units would also require considering other topics in the core damage model such as:
| |
| o The effect of core damage on the other unit and the resulting loss of power genera- j tion. I o Whether the sequences sould affect both units or not such that the risk should be doubled.
| |
| o The probability of core damage of one unit given core damage on the other unit, 1 e.g., due to complications and potential site habitability problems.
| |
| This level of detail would be appropriate for a complete V-1 analysis, but the uncertainty in assumptions and the additional requirements on plant modeling precluded accurate consideration of two-unit effects.
| |
| The second part of this appendix presents value impact analysis results using the NRC Case Study (1) representation. Tables D-2 through D 5 recreate the NRC Case Study results for the evaluation of i individual modifications. Tables D 6 through D-9 provide our results in the format of the NRC Case Study representation. Table D 10 is a key to the tables.
| |
| [
| |
| Table D 1 compares estimated costs from the NRC Case Study to those by an experienced WEP estima-ir. These WEP estimates are for both units and take credit for those situations where a common ,
| |
| i modification can be shared. By comparison, the NRC Case Study estimates appear to be somewhat low. ;
| |
| D-1
| |
| | |
| l Table D-1 l Comparison of NRC Case Study and Wisconsin Electric Power Cost Estimates ALTERNATIVE COST ESTIMATES (Thousands)
| |
| Sest Est imate by i Estimates usin9 SNL Procedure Foe Site (2 Units) Experienced WEP Estimator Generic Local Estimated ALT.1 29 x 2 = 58 22 x 2 = 44 60 x 2 = 120 MOD 501 RWsT LLA 423 x 1 = 423 333 x 1 = 333 800 x 1 = 800 BOD 302 D-G Batteries 16 x 2 = 32 72 x 2 = 24 30 x 2 = 60 MOD 803 Redundant RHR Viv.
| |
| 2.155 x 2 = 4,310 1.582 x 2 = 3.164 2,500 x 2 = 5,000 000 806 SFP to RHR MOD 107 Anch, Elec. Equip. 182 x 1 = 182 146 x 1 = 146 500 x 1 = 500 133 x 2 = 266 110 x 2 = 220 300 x 2 = 600 MOD 508 PORV's 94 x 1 = 94 79 x 1 = 79 350 x 1 = 350 CD 109 SWP Wall 436 x 1 = 436 1,000 x 1 = 1,000 MOD 810 AFWP Fire Prot'. 579 x 1 = 579 596 x 1 596 1,200 x 1 = 1,200 MOD 211 Cnar9er & DC 'B' 808 x 1 = 808 6,752 5,042 9,630 Total direct Cost Alt. 1 21,260 Total Cost Alt. 1 14,906 11.324 ALT. 2 1,835 x 2 = 3,670 3,000 x 2 = 6.000 M D 815 TD Generator 2.178 x 2 = 4.356 MOD 818 CST 1,477 x 1 = 1,477 1.162 x 1 = 1,162 2,500 x 1 = 2,500 12.585 9,874 19,030 Tot al Direct Cost Alt. 2 42,764 Total Cost Alt. 2 28,281 22.609 ALT. 3 MOD B16 Spare RHR Pump 1,162 x 2 = 2.322 909 x 2 = 1,818 1,700 x 2 = 3,400 3.242 x 2 = 6,484 2,606 x 2 = 5.212 8,000 x 2 = 16.000 MOD 817 Diesel AFWP 21,391 16,904 38,430 Total Direct Cost Alt. 3 86,590 48,198 38,809 Total Cost Alt. 3 l
| |
| j ALT. 4 l
| |
| 131.584 118.094 100'000 Total Cost Alt. 4 l
| |
| D-2
| |
| | |
| c__---_________
| |
| m m m e w w w c er c w w en en O O O O O O O O O O O O O he
| |
| * + + e e + e + e e e +
| |
| * W $ 2 a e K c,
| |
| WW r~
| |
| WWWWWWWWWWW e r- r~ .o e- e N m e w Om e*
| |
| ** k u O e i 0 t Q g &
| |
| t O O e w. N. O. m. O. O. e e r. O. e e I
| |
| M # ee imi * *t m - - N m cW w @
| |
| e > s M
| |
| w we en en w w in en w in r* Ww@W 8 D w g
| |
| O
| |
| + + + e e O O O O O o Oe O o Oe Oo e O O * *O G
| |
| Id # WC em > Id W W W W 14 m e GD N ei==
| |
| Ed W W I4 W W E2
| |
| ** ew e-o e en GD O N 6 6 $ 0 M CD r* N e w O. e= e N W s'* *=
| |
| p Z $ *=' Z e . . m. e e . . e. o. en e e M
| |
| M 2 m ,= e-o sw en W em N en en N e= W P*
| |
| I 2 O e e a e e e i e i w i
| |
| * a N N N N m N m N m m e E I >Z O E O O Oe Oe Oe O E
| |
| eC O
| |
| ** 2 e a s O e e O eO sO O O O #
| |
| fd W Ed e e JC e4 y g W (d (d Ed e- W We- (aJ (d W .id ew O m E E 2 W g ea en e m e > O w
| |
| -= W. en O.
| |
| t*
| |
| D
| |
| > E > 84 e
| |
| W w e w e f*e Ne ene e o CP e O.
| |
| e g 3 g we es e er gP M M ** w en ew N m N .4 E 2 2 I 8 I t i g CC E b, = > b
| |
| @ la
| |
| > E o c O K
| |
| O z y
| |
| *** O ** E W O T 6 g O su 3 6 v
| |
| w e w w w m e m m e en w w w m O O O O O O O O O O O O O V M s w & * * * * * * * * * * + + e W C e e w Ed Ede-LdwIder fd td Id Ed e- W W in Ed er W mW ar-4g. 80 4 > O w e N ew a 4 W D CD N w M & mO > e
| |
| * e W es m e @ eq @ Oe Er6 e en
| |
| * 2 e * * * * * * - e T M he w 0 == ed N
| |
| * N == ar r- ea - N ** N @
| |
| O M & O O e 8 E $ en e e w e e e e e- w w w w 4 oe O* O 8"g*
| |
| E
| |
| (
| |
| a U + O
| |
| * O
| |
| * O + O e O + O + O + + O O4 Oe g
| |
| p 9 W W m r- es e cw en m W W W W W Wr= Wen W e W eW m W m M s 4 & N e a w e N & e& e 8 e Ee= z e e . e e e O. e. N.
| |
| **"" > E e m m M w @
| |
| * N D D N e
| |
| @ @N
| |
| > O e i t are T
| |
| C a==e m M M er w w en e en m en en w OO t'3 O O O O O O O O O Oo e
| |
| + e o e o e e o e e e Cn C m o E
| |
| WWWWWWW w O w r oc == .s W ** W er W en W oc W en W er 6 w ee o, e C.
| |
| e,-
| |
| g e=
| |
| 9 e-o E a Q
| |
| O e e
| |
| * =. . w w *
| |
| .= N w ce en cp .= em N ew m an e e a. . . e e
| |
| D e e-a he &.
| |
| b
| |
| > C 0 e w e O L 4 w w en er e c w e e- ww@@ t--
| |
| p w C O O Oe O O O O O O O O O ** C O e e o e o e e e e p l tw P** a 0 W e o +
| |
| W W W cd W cd e-W OW e<derWar su W % >
| |
| N 3 m =
| |
| > c= w .a - N O e o E C %
| |
| 4 w am e- ew m cv N **
| |
| g e *= 4 M CC rwe CP w. e. er.
| |
| P-
| |
| : m. . er. e. m. . e > > p O U p u C = 3 e en en - ** e- @ ewt e Ie mt a e I r= f=
| |
| O'. m 6 4 0 1 O t e p B a
| |
| @ 4 2 C 1 8 9
| |
| - - N N N m m m m m m e w 0 0 0
| |
| }e V k o Oe eO Oe Oe eO O Oe Os O Oa Oe O O e me e a e
| |
| > m c - O e i
| |
| > Z Q ha a E WWWW Ed W le] Ed Ed W W Ed e- w e w O en 10 b= CL e a= w N c,n h ,er.. e,. O,r os
| |
| > ,, se
| |
| . e.
| |
| B E > g ,n O. e e O. .
| |
| N, e .
| |
| >"a e4 (P c h I en e N e v e r= c w N ==
| |
| T
| |
| @ M 3 3 h E P- O r s u e u o
| |
| *=
| |
| O 0
| |
| ==
| |
| : h. 4 t
| |
| m
| |
| ;lne fd O T C ay T> g g
| |
| H 3 4 *=
| |
| a w w gr e rw m m er m en m m m m h eo O 4 3 "O M W
| |
| a na C L 6
| |
| * O* O* O+ O* O* O* O* O+ O* O* O* O* ~ a Wm U
| |
| g y le. & > E W Ed Ed W Ed 64 W W W (d Ed feJ r= Cd m a M em - O O m e
| |
| * e w ev en en ed M
| |
| O m 4 6 &
| |
| : b. w 0 e m hd G fw w ene e
| |
| P.
| |
| e en O e eN Oe e e a f*
| |
| e e-e
| |
| . e
| |
| -m w
| |
| e --
| |
| C
| |
| > e 6e ee e
| |
| 2 b c= w -
| |
| * N r=
| |
| g O A o O 4 > ea == *= w e w Y O w w e e o e . e - . . . . o o e e M W M O+ + O *O eO +O *O+ O+ O O O Oe +O +Oe m 5
| |
| =. g
| |
| ==
| |
| 6.
| |
| Q e 4
| |
| EG s=e 4 9 O
| |
| = W id f4 W W W W W W Id td 14 W o an e e O en O .* N e er w Q 3
| |
| g O
| |
| O v 4 e em e e en en e- w e ew cw >g g .
| |
| cr* an w
| |
| o E a=
| |
| ~ H P-
| |
| .
| |
| * e e e e e c m.
| |
| e . m. A y ,c T s &
| |
| g e N f* ed ** r= w N de w m ,= P- f=
| |
| M T C u
| |
| > @ -4 U g mm O eJ @
| |
| m m
| |
| Q d -a -
| |
| * N N N N ew ee .s N .a N O 9 3 4 M T 4 D C O O O Oe Oi eOe Oe eO e O O O 4 s D 4 6 % W e 4 0 h- M b ca w # #
| |
| W W (d Ed Id W e e F e- Id WanWeed enW D W M 80 h
| |
| : k. M e w 9 6 e R e w 4T O C==
| |
| s @ ei a e a dP O O O w -e t
| |
| 0 > 0 e O O Oe ** *
| |
| * dP w EP * *
| |
| * W *M* m e me m* w* *er mm g a p g o gD C 4 M M N & cee ee m A M M e
| |
| e C eb 3 0 D 3 U @
| |
| & T a
| |
| * ed
| |
| * O M O ee O N ed O -4 O N O.
| |
| h w M s @ E em a x g w 6 O
| |
| + O+ O+ O+ O+ O+ O* O* O* O* Oe O+ O+ o a a a. 6 e n. T ws @ m e d W W W Ed r-en t- O en Wm WW e- W en W e W .a W en W N wW e C
| |
| > N C w > 0 C. N 8% EP Ne m e N O -e T e O 4 O ~ > en en im e o e a m me e a e=8 e e e e e a C c a e g N N N M ** e N W N e4 w M r* C m he b & .M g ai em 3" C s he W W an @ W en W in e en e en w a j A e
| |
| == e-e O an w 0 g &
| |
| $ C 4 Ot O OI OI O O OI Oe O e e 4 i OI Oe O i O e 4 a p w
| |
| @ E 6 @ W W W W W W W W W W W W W
| |
| & D > M O EP O en eo ev en en w O en ,.
| |
| C & F O* ** en 0% e e a *e
| |
| * e* e e=e em W gr es ce O hs @ h
| |
| * * *
| |
| * a + e
| |
| * he W as Jll 0 U U les a E.
| |
| Q e e f* N ar N r- /*=e N m em f*n m n
| |
| $ e m-w mN e w O ** r= c% e in o na O e a e* == e= x 4 E D-3
| |
| | |
| en to W w an in W en P= WW WW b OO O O O O U e e O. O. O. O. O. O. + + + + O.
| |
| 4 M-E WWWWWWWWWWWWW H
| |
| W L
| |
| & ~
| |
| m em - W N - m e e e e m m P= N w e N P* P* e K me e e en. a. e e e e e e O. in. N*
| |
| e-o em eo w W e N en eN .e W P=
| |
| 1 i 1 e en in W w in to in in W en in cw I O O O O
| |
| .3 b M O. + O. O. O. * *
| |
| * O. O. O. O. O.
| |
| 4 E W b - en 14 ed W W h3 W W Ed Id h3 80 h3 tal b W e.J M M M O e. e m e e m w m en m O W t
| |
| h3
| |
| .3 O > c Q b ( 4 O w O O W w O e & O e e e N N ee W m w en w e e e e e wmPeoNc-W a e P*
| |
| e P=
| |
| e m
| |
| i b 4 i E eJ in in W w w in in en W in en in w I ta) be Q O O O O O O O O
| |
| > 0 . O. O. + + O.
| |
| * O. * * + +
| |
| * 4 3 De sa3 W W h3 h3 W En3 h3 En D m W r= m e en m W 6A e3 no to h3e hJ e W e O u C e in w w N W e en N e W .= m a en
| |
| (=e C g g a e. M = . e e e a e e e a e a e
| |
| @ es & M M- -4 == me w M M M N er en ** an en b g3 m is em O =
| |
| T U b M U U D y Q he a
| |
| : 0. M C M w w w e m m e w w c w w w M U O em O OO O O ea e e,. UM he t O. O. O. + O. + + O. O. + O. O. + he a em y 4 la O u h3 ta3 kJ Gd h3 M h g b M e N ta3WWen w En3e=W eW wh3e(d .h3 en e N N w U T U W a O e t- a .mJ - en
| |
| : e. e e e a w in P4 e w m in O e g ==s O m > c m e og e e e e e e e e . e e e m .D J U op.
| |
| ]C g0 -* == ** en
| |
| * w a m in W W W P* La3 m E W NN G
| |
| - - - _Ce m e
| |
| *Q b M e w w c w w w w w W so e c w em g e o 4 2 6 O O O C U C c Q U O. O. O. * *
| |
| * O. O. O. O. O. O. O. o E 80 g b e == W I4 W W ta) W h ta) 14 W ta) tw EC U U -P 2, ed a @ w e P- O O in N m e' O m W em e= e= et E e-.
| |
| 4 C 3 m e- = in We Q *= e We D m = z to pe-g & L 0 0 ~ e e e e= m m e a 6 m N en e e e e N.
| |
| * 3 o E & ce O m m m e= e= .= w me ea &
| |
| D h 4 4 e
| |
| *C & a & T erm ed s w w an er en en W no WWWW g
| |
| ** .3 b
| |
| U O. O. O. O. O. O. O. O.
| |
| P=
| |
| O.
| |
| O
| |
| + O. O. O.
| |
| Te =C
| |
| & 4 4 e e w
| |
| "O 4 4 e. m W W Ed 14 W W tu h: W ta3 h: ta3 kJ O T 3 L g b 4 m b O e e e O.c O - N e w w O 4 6 3
| |
| *"' O I w P= w r= to ao f= w e w e m C h in m b e= e e e e e e *
| |
| * e e e N. M. he 4 a p w N P= wie ** P= W N e W M em Pm i rm 4T O
| |
| @ m == .C g -
| |
| m O mM f @
| |
| NO "" U g C 17 4
| |
| & 3 O D i M a ha 4 U C Q Q ha C
| |
| 4 3
| |
| O
| |
| ==
| |
| O O O O O O O O O O O O O 0 O O O O O N
| |
| == 4 b
| |
| I g n
| |
| 3 0 > - O.
| |
| * O. O. *
| |
| * O. O. + O. O. O a U N Q. he *= w tu h: b; LC Ed I &n) ta; h3 W EC (a) 54 & C T P'- W a @ ==-
| |
| 3 e == 0 o O O O O O O O O O O O > N C
| |
| ,Q Q C eJ M he
| |
| * w O O O O O O O.
| |
| O O O O O O
| |
| * T I e g & C e e e e e e e e e e .J C O
| |
| >-- C-m e
| |
| o L ~
| |
| C O O e Oe O O O O O O O O O O g g C e a m
| |
| w 6 = w e -
| |
| .E 4 w
| |
| O g #
| |
| w 3
| |
| - :)
| |
| A e
| |
| - = C O O O O O O O O O O O O O - C g 4 h M a e o - 4 se e he oc a 6 s -
| |
| * m O. O. O. O. O. O. O. O. O. O. O. O. O.
| |
| 3 O E m as - 84 W to W W I4 taJ 84 ti; I4 te h: W e=. ==
| |
| ** et M C O O O O O O O O O O O O == N p C ==
| |
| b == OO O M 4 e e O. O. O. O. O. O. O. e se g,)
| |
| U 0e O Oe O Oe O O O O O O O O O O O m 4
| |
| V L e
| |
| ~ l o 4 C Z L) m e m C a sJ O C 19 O E -s ==. O O w o O O w in O W w O in W CC b ue eg U 3~ O g .4 he e., O. O. O. O. O. O. O. O. + O. O. O. O.
| |
| 3 he $ w M (a) b) W (d W ty (al Es3 b m T U w N O O O O O e O O .$s2
| |
| = O OWO(a}w (d (n3 M W w Q b M
| |
| e C C e g O in O O O e.
| |
| e e e o e . O.
| |
| ** en e e O. O. e O O N O O O e m O in N O en see h O L I > 0 p M 4J =e C
| |
| * 9 we W e 9ll he w w in w an in W .c em W W W te 3 >
| |
| m a.d* ** C @
| |
| g , O. O. O. + +
| |
| O O O. O. O. + +
| |
| OO O. O. O.
| |
| a e. w M b O g C e - h3 h3 Id Ed ta) W W W W W kJ W h3 M
| |
| ** w C == a e o in e e O W in - O W w w W m M m O & m w ed P- e c= in no P= 0 m e in m P= + en O C == C 0 M e e e e
| |
| * e
| |
| * e * * *
| |
| * a. w 6e M y hJ U N w ou M P= in N e in m es te in m M +
| |
| + N in C & > 4e W en te W to 9 an w in en en te N M e
| |
| ==* ** O he Q O O O O O O O O O O O O M M k C e t # # 6 e e i 4 e i 4 e s +
| |
| $ E & b
| |
| & > E WWWWWWWWWWWWW m O e C in e O N en e e o an
| |
| +
| |
| .4 f
| |
| C g w 9 Q. O ** en e e s e e e e4 s'e
| |
| *=e N w we in w h e a e e e e e e a e e a w w p g O he e e P= N w N P* *=e N M eW m M V U L B S 8 e em eJ e m pee g m N g w O M Po e )g gm g=e en ed
| |
| -4 0 m ese em ew ** 4e == 3 4 E D-4
| |
| | |
| r l
| |
| m E w
| |
| bb I w 0 t gL w w w m w w w w in , er e m C 6 % o* *o* o* *oo o o o o. o o o o b 4 4 as o C * * * * * *
| |
| * e e3 ha o N Iu W W W W to: W ta3 W si3 s.3 tiJ W "I' 6 b 4 c - N w M w w o -& w N w m 3 6 O > - > *a .* e w w e. p 3 .e a G. 3 4 M e . e . . o. sc. . e . e. in. M J N N w N - N w w m N = N w 4 4 Da m O g 6= G O e
| |
| e == =a ** rw tw N rw tw == ca N N O
| |
| 'pw % U o oo o oo o eoe oo o oo m a e e e i e e e e i e e z w@@W 2 E
| |
| WWWWWW e o o o w WWWWWWW r~ e er ao e sn o m =n 6 e e a e
| |
| > 0 9 0 o o == w w w aC O go O 4 e e EP. . cP. e. m. m. . m. m. . er.
| |
| * C en ** ** rw *= en e ee EP ca en ca e C Z6.
| |
| O e= 0
| |
| *** T *= ***N o ed == ed N == * .* ==
| |
| a 6 E o* *o
| |
| * o* o ooo o o o o o o E u 6 aC 4 T be W ha N W in3 la taJ W W Ia3 la3 W W Ia3 W Es1 U $ e 6 - N e e w w - e c e e e -- w 4, ep. > 0 (L > w w in. gP.
| |
| w o. go.
| |
| : m. an. e. o. ev. p g y C o- . . . .
| |
| m m na w -e m mW W w w on .= m g T 3 l
| |
| : 4 en. e -
| |
| J O C==
| |
| u.
| |
| 6 E r** O T g w gr er eg e m er m in m m m m I g gg 3 & he o* *o* o* *o
| |
| * oo o o o o o o o tr a e
| |
| : a. e- W -WW e O 4 o14 o I4 m ti: eo W e-fu sc W W ga: ra: W
| |
| & m L w rs* e o. e he E@ % C m
| |
| ".O b > 0 N w in 4 9 m=
| |
| y a sC t o > e r. e. o. ec. rN. O. r=. . e. e. ==. s 4 m o ea == ** w ec w e= e sa r N f* w T
| |
| * b tem O == == N C l
| |
| T EL 0 Q M > e e
| |
| }
| |
| C Q
| |
| * m w m 9 .
| |
| m t - 4 C' Cn m- eu - 3 s C rw .O M = -
| |
| M e 4 == u
| |
| .C.
| |
| , Is) t= 4' N f% N m N #N fw rw == N rN rw fN N t= h. C d **
| |
| * O o o o o o o o oo o o o e s 3 M T e e e a e e i e e e i a e a 4 en C C" W 4 % O f4 ts: W W to: W ta: En: En: tal En: W W m O M e b 6 s C w N c w
| |
| * o e r- in e w rw o O O e q g O be. @ t & E w W O -EL -
| |
| 8 0 4 m m an c CP. o. o. . es. . r. . m. r=. m. so. fNe $ T g > 0 se O 4 > en EP m w rN w tw a m .a m tw 0 m T CD sC O CC C e g & C N D b c 4
| |
| > c e U 4 3 P"D" m 9 E - 6 J - - - o - o .= o es .a 0 - o e E N m. . -
| |
| g gy T - o* *o* o* *o
| |
| * o* +o +o+ e o o o o o cc N - > s % ~ r.
| |
| n f- > 6 E * * * - > U m g w= w & W t: W W W W W W sc w
| |
| * e 5 t o p a. t 6 aC W te W .te in W o = r* m 6 in cP -. c rs er %% == s I >
| |
| rw E Wm a c'. . e e tv. q. N=
| |
| 6 m e es in an ch - m - EP iP
| |
| > 0 > * *
| |
| * e
| |
| * o. rs - = = - F M 8C O -4 r% N fN == .* e rv w rN == w es r* > > > > c & C C
| |
| == m - -
| |
| C e e e e e e C g, C U m --*
| |
| C 2 E O 6- 't b .C es U O G e E
| |
| 'O w E N E -
| |
| o-c m a e L g 6. E m m w m m e m er e w m ar m aC > sc > -= 2 0 > 6 &. .
| |
| p M
| |
| O T 3 &
| |
| e, 6
| |
| w e
| |
| *********oeoo o o o o oo o o o W W In: W sd W En: W Ec id W W W 3
| |
| >I 4
| |
| T
| |
| & c g C. C > U m
| |
| a he CL go r- rw r- - r= - rN w eo == cv rw o .O u 3 == == 9 w C C.
| |
| b ee N Ce - &
| |
| a (J C, 4 > o > r%. M. fN. O. m. e. w. rv. CP. fN. e. N. . T C O E g w w in rg e - en .= .= rw P= ts es t - m 1 4 -
| |
| M e 4 & o e & E **
| |
| @ W 4 m o U he *** O - O T - & w T
| |
| .D3 EL 0 Q M e Q. 4 U b & C m O b e U aC
| |
| > == h 4 T O O m i
| |
| & fa.)
| |
| 6
| |
| - es we oo .* o N ,a ** c M Q. T
| |
| = s &
| |
| g gd ee T o
| |
| .s e e m
| |
| * *o *o+ o+ o o *o
| |
| * o* o* *o *o oo
| |
| > M @ E 4 k k 3 w E a 4 et *
| |
| * o e C E k4 W (d W W Ed ta3 (d W W fa) id o a O ==
| |
| * O en Ed e W N u @ w **
| |
| O en > P- so cP w w eD w w e P. e o U e> I E D N' mo et e
| |
| oe o e a. w* r~. an. ew. *.=m r~ m e. o ,* N m 5 3 M
| |
| aC Q - * * * *
| |
| * e. mrw M ** M @ C C 4h b O, gh ca ev e m eN e rw m m es m
| |
| * mm O
| |
| @ 4 4 3 b N - M M WC
| |
| > ,C 4 4 s i
| |
| M 3 C & >=
| |
| ** -a e C e U 6 w w in w w in W in e en in in W M M W s E E O O g6 EL oo mm
| |
| *T M C e g C. be 6
| |
| & & W D '**
| |
| o o
| |
| a 6 k E D o ooo s a e o o o eooo 6 e e e oooe a e M M T e 11 g 6 @
| |
| w $ in in D.
| |
| >=
| |
| E 6
| |
| la) W W W id fd fa) la) fa) Id 10 fa3 ta) o o a a ** O ** m m e he g h A, m o en e o tw in e er o en w o o be h aC Eh. en e4 **
| |
| e in in ey &>
| |
| 4: O n.
| |
| y y go, c. o. .o=. deen. e. am.we. e. e. e. c.e **N. . . - .a -
| |
| rw se e r= rw w tw r. == N m ** M m e en at at ,
| |
| * e e e a ..
| |
| I & m M > b e -= CD CD s a s m e w m tw so e o ed r= EP W in == es tw rv 0 e= g O. =9 ** .a * * * * > > > > z aC C E l
| |
| D-5 l 1
| |
| __- _ - __ _ _ __ _ _ _ _ _ _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _. _ _ _ _ _ _ _ __ __ _ _a
| |
| | |
| V h
| |
| i 8 E i 6 4 e e u O s O 4 w w e m w m m ** en e ar w m N
| |
| O O O O O O O O O O O O O in e e e * *
| |
| * e o e e e e e D.
| |
| &C D TC s O CD
| |
| > W tal tal ta? W ta: W W In2 W h3 W W T N w & et w w to e N N en w m 3 6 4 a E e .=
| |
| a y be .O.* e e e. d. .= e m
| |
| e e. @. cae e eo ee n e e w
| |
| q/g
| |
| &@ m. t M ** N W f% ,=e w m ==e es N .a N w U D an 0 > # ==* @
| |
| a* b 3 4 m he o e e a h 9
| |
| > um U sh *C
| |
| .=
| |
| * N r: .= ** N c. - e 5 t &
| |
| =
| |
| O O O 63 O O O O O O O O O m ea U E e o e * * * * + * * * *
| |
| * en) m g E w 4 .4 W h: W tset W eIa2 Li2 h2 En1 h2 h3 W W E 6@ he > N in e w on N e e == eo C &
| |
| g m 6 z w w .e
| |
| * w == m em i > 0 hw * . . si .a e. e. O. e. en. e. t .- O. a e e M I C O m m .* g a ne e* N N ce w *e en **
| |
| e C I C D C O =*
| |
| ,,O. e uO e-t a s M a 4 im E
| |
| == E ==.
| |
| h
| |
| @ e sr G 3 0 g 6 w
| |
| * b G
| |
| 1 O ** & 8 b4 @
| |
| * I 3 *= m Q. ID
| |
| .D M .C *p M m O O C. O w w O en O O O O es e 0 g O N w O O O Ci O O O O O O O O O 'O C as ev=
| |
| g g as M Q O Cm O
| |
| > * * * * * * * * * * * *
| |
| * 6 = @ *D i fa! fa3 la; W In} W W In3 in) h3 Is3 is) fa3 m a 'Q l @C e O + 0 O O O O r= N O w O O O O O T D u=
| |
| 4 mM 6 ** O O O e g e 3D e O. O. O. O. in. O. e e. O. O. O.
| |
| 4 & D e h= en e % *d m N O O O O ** ** O ar O O O O O.
| |
| s 4 e &
| |
| p t A O e > C C 50 s
| |
| l 9
| |
| & T..
| |
| "D e e s m n 4 s; e C e .D t O e C C O O O O O ca a O N O O O O 3 O D
| |
| .T,. I la w O O O O O O O O O O O eO O e
| |
| > g .ee CT ==Cc > e e e * * *
| |
| * e o e e N O O &
| |
| w w 0
| |
| I m 4 a m g g O +
| |
| taJ ta3 W W (42 h2 ta: W In3 W W tal la3 e= 4 4 as C h. O Q O O O O N m O w O O O O O 4>
| |
| T e O O O w O co o O O t C T E e mO 4 e O+ O= * *
| |
| * O. * *
| |
| * O. * > N C I- C == =d m WOO O* O *= ** O w O O O O == T e 9 t ** as 4 > 4J C O 6 g g M
| |
| & e e *>
| |
| C .C. 6.
| |
| ''" E t =s 3 W@4 .G, 3 4 m M Om h
| |
| - O g 4 y en= == 0 t C 4 p 6.
| |
| so O > Q L C h % O O O C C N N O O O O O O M & 4 6
| |
| * O O O O D O O O O O O O O
| |
| .=
| |
| N gg) y m a m w o e o e C. * * *
| |
| * e * *
| |
| * g g C
| |
| & 6e i O CC C W W h; W W W W ta: W W h: la: e.
| |
| he O C *a e in O O C O e N O O O O O O m O P'"
| |
| 80 == A 3
| |
| * O e > -
| |
| a O+
| |
| O C O e &
| |
| * - O=
| |
| O
| |
| = O* O. O.
| |
| O a
| |
| 6
| |
| * W ** O O O O O O gp > i I +>
| |
| 9 4
| |
| ** O O C- O O I he O 2
| |
| ".O U i ===
| |
| no > s >
| |
| p .e. I h. 0 *= m O O C: O O == 0 O O O O O p B 4 m e Oe Oe C C O O O O O O O O O I L/) 6e g. 4 * *
| |
| * s * * *
| |
| * e e C I t Om 4 WWW =EWWWWWWWWW p C w ge 6 3
| |
| *a @M- CO j
| |
| : t. > O 0
| |
| O O c O O O O O O O O O C. O. O. O. N. Oe O. O.
| |
| O O O. O.
| |
| E I N O O O O ca m O O O O O O e e 8
| |
| y b I e + E D 4 &
| |
| I i $ w 43 f b+ g m a g r. s o ,. o c. m O O O O w w O .n O O O O t C ** C3 % O Oe O O O O O O eO oO Oe *O
| |
| * O 6 41 q Q e e e e o e e Oil M f C e M W f C @ eJ.
| |
| w C O m D D h3 W in3 ta: In2 ta: la3 la: W W En: ta3 W gg e.
| |
| 6 c=
| |
| CC m
| |
| O O O O O O O O w O O O O 4 == O C
| |
| * O U .D.3 3 a g6 3 em 4J w
| |
| < 1 g
| |
| > e O. O. O. O. an. O O e e e e O. O. O. O.
| |
| > N O O O O ** ~ O w O O O O Q i ==.
| |
| g fa3 i e I
| |
| > I & **
| |
| S e.* I m ** O O O O O en *an O N O O O O eh e
| |
| Da C 4 C E O O O O O O O O O O O O O m 4 4 ea a O W 6 * * + + * * * * * * * * * *= (D g O e == m a 4 h) Ea3 ta3 In3 W En3 fa3 h3 W W In3 ta3 ds3 e W C & O q m O O O O O O O O w o oO O M >
| |
| R em 4 O 4 > O O O O O h e e e e e en. O. O. e. Oe O eO Oe e oe g b N O O O O *
| |
| * O w O O O O O >
| |
| O ID M i m e >
| |
| 1 i
| |
| .s > m g
| |
| C
| |
| ** ** O he O sc > a i W C 4 O = t 10 E p W W en W @ en W en w en en en W D m N pm O O O O O O O O O O O O O e e e e e en N N >
| |
| 4 t.
| |
| h he
| |
| # 4 W h3 Is3 In3 W ta3 W In) b3 h3 In3 W is3 e o e i e 4 e
| |
| > e 4 0 h @ E m @ O en e O N en ED w O en 31 +
| |
| U U lae & 4 .O e* ** ** w w O.
| |
| =e
| |
| . en. e. e. . e. e e.e . . N. .
| |
| m >
| |
| g
| |
| ,D.
| |
| e I
| |
| e ee N w N r= e=* N m es m m > == > >
| |
| 4 8 9 5 3 3 3 4 he >
| |
| t b ** 80 10 to 4 *s aJ e m M w mN e w O *as e e W en m w > >
| |
| e e* 4 0 e e= *= = **
| |
| g d C E > > z 3 D-6 j
| |
| | |
| en en en r= r= r= em r= en er en en O
| |
| E O. 4 O. O. O. O. O. O. O. O. O. O.
| |
| % i e E E c.
| |
| W W e=
| |
| W W W W W W W W W W cv c= N - a w e w w ao e M Q. W O @ e M N e e. O. e e a O. e. en. e. e. o.
| |
| ee sw N ** 4e e=e ew me m m e w 0 1 e i em 4
| |
| =*
| |
| w te c= w w w P=
| |
| * an w w @
| |
| O es.4 ene E O. O. O. O. O. O. O. O. O. + O. O.
| |
| g O & > la3 fa3 la2 P= le3 la) W tal la! W W &a) In:
| |
| 4 C E w in w e O O w ce es e e C 9 &W w. e an e Eb.
| |
| Q W E EQ 3
| |
| e o. e en. e en. e=. en e .
| |
| a me
| |
| .,. 4. e en ** *= w en w g , g m. c.a e m. e.s . , e,4 e i e e e . ~
| |
| g to O
| |
| t m m m en w en en en m ew w m 3
| |
| .,. E O O O O O O O O O O O O O w d b O =* 2 e e e e e e e i e e o a E 6 >
| |
| ke a a h3 in! W ta) W Ede Ww W Is2 sd W W % E se= id 5 e ,*
| |
| w N w te es e= w w go ew tc CD %
| |
| y e
| |
| =
| |
| > EC > w w O. O. ev. en. e. e.
| |
| w sc.-e. w. > > ==
| |
| S M
| |
| . . . z a a m te en m e w N N ew N en fN
| |
| & E e e e a a a a O
| |
| , E z z O E E IC E E E w > b 3 aC s.
| |
| a E > 2 O O T @
| |
| ".U.
| |
| . Ed 3 6 w
| |
| > 9= 9 w w en se e.e == N eu ev m m e
| |
| ~ u 6 &. Q. O O 0 en- M c y e% +
| |
| * O. O. O. O. c.= O. O. O. O.
| |
| *Q ' W P O -
| |
| 10 la3 ta3 fa3 W tal W Ia3 la3 W Ea2 Ee3 g E e aC eO se e e m w - O w ev en w -
| |
| Q @ mO E
| |
| >=0 as and o ** N. ed. c=. O. e. N. m. e=. en. e. N e O.
| |
| E A O O ce ew ** w w ew a w en N em m s=e g Q l 6 8 1
| |
| & O
| |
| *** W to e e- @ w w e= e= en w w w 3 m a
| |
| < u O. O. O. O. O. O. O. O. O. O. O. O.
| |
| Y C 9 = WWW en w e-WWWWWWWWW O M ws a$ 2 w w N e O O w ew em c e g w 2 en e e. . o.- O. e. en. o.- en. en. in. .
| |
| g in m w e ce m e em .= ee W cw
| |
| >=
| |
| N a E 4 2
| |
| in in
| |
| * P= e en so ao e e en O==
| |
| g 4 m.
| |
| 6 o O. O. O. O. O. O. O. O. O. O. O. .
| |
| U **
| |
| * e E 6
| |
| E A
| |
| WWWWWW e w ew o in WWWWWW r= 0 == en en e O c e == b a O w c.e . am.a. N. m.
| |
| w m c= e w w Q, > 0 0 # e e e e . .
| |
| 0 Q- * 'a * * * '" "" "' ' ' " ' ^ ' ' '
| |
| E 7 C=
| |
| I w w r= w w w e= e= c w w w e gj w s O.,
| |
| O -e O. O. O. O. O. O. O. O. O. O. O.
| |
| 8 O e=d W W W Ia2 Ea2 Id Ea: W W W Ea! W O P"" m & o ao O - r= ec o O w o en c e a c > w o- w c- .co e.
| |
| g y a. g e. .n. o. O. so. in. . =.
| |
| m 6
| |
| 3 4 o 2 C 2 t
| |
| m ** e sa m *=
| |
| 4 1 i e I I a e=
| |
| t e=
| |
| I e
| |
| we enw e e e w e 6*
| |
| p i c%
| |
| dC "O
| |
| e b 9 % >
| |
| c "O t ec%
| |
| p p E m m e en e te w s e r= e= e es ev p D as o C O O O O O O O O O O O a e e e e e e e a e e a M = w g .E tu Ed W se: W W W W ta; W W W s e a e en w - w ab a r= w w m m o
| |
| g
| |
| > E > w O e e O.
| |
| c w e . es. e. c.
| |
| .er.
| |
| . e. e. m. oMo g
| |
| g
| |
| == e= es ,= w w m ew N de c= *= e.= E C 9 M > Z C O ''
| |
| M '
| |
| 4 %
| |
| U B O E s.
| |
| 6 O l z W O T O le 3 @ e=
| |
| w s e m m m *= ** O e. == *= ew O N M M he O O c
| |
| or=
| |
| W c o e
| |
| & > E O. O. O. O. O. C. C. O. e O. O. e la. CC W W W W W Ia2 W W W ta7 En: 80 g Q m aC @ @ N N P= ** c w ed O m e, an m c=
| |
| D E n.
| |
| m he > ev e e. O.
| |
| w e=. en. m. . .
| |
| W w w en.
| |
| es. O e e a . .
| |
| O A O O Q. en ed *= m ca em an w m w w in M
| |
| e''"
| |
| w w - w w w- - c w w w M M O 3 4 me gu O. O. O. O. O. O. e O. O. O. O. O.
| |
| Sa IC m 60 la3 80 ta) W ta! s.) te3 ed 10 ed W U M fp $* e O ** e e o O w O en en e M O E to e E w p oe a e in. e. O. e. in. e. to. e=. to. m.
| |
| M m ** e sw m en e.e ow we w en w
| |
| @ a e i i
| |
| dC e w .4 .s ev N ev e4 .4 e4 N en cv j I T 4 O O O O O O O O O O O O 4
| |
| 6 % Q E e a e e a a e e e e e e ee E W dal le2 Id W W W tal W W W ta) we o u. 6 G 6 o O m e e m e en P= en m e e m Cit' e e <
| |
| ch. >
| |
| e u 0 IeO e 0 en e =. w. c=. ew w w ** O. e.
| |
| . . . . . e. e. i Lg,J b 4 Q S Q # e e et ce m se ev e4 N en ce l b ee .=. O O O ew en av c. ew w w m in i O w 6 s O O O O O O O O O -
| |
| -e a Ee O. O. O. e e s e s e e e e b a he $ be 4 b) W W W $d 1a3 (d teJ ld la) Id the p e N O en e ew .$d. e- O m m en e b and > 0 > w w o<O .e e. m. N. *=.e . M. m. e. e. ed. en. .
| |
| e I
| |
| to m m en ev .= e e en a e=
| |
| M ==
| |
| C
| |
| .M. O
| |
| >=
| |
| : h. in w w a so e- e= e= e= e w e
| |
| @ C e O O O O O O O O O O O O
| |
| {E v
| |
| tD @be E t i a e e i e e e i e e lsJ W tes 80 ta) (d Ed 10 ta) 80 ta) h3 C &a= F L w e e D .w e O w N m e e 4 6 he O em en w O w a
| |
| : f. O he y a . N. m. O. O. er. * . e a a uQ te. sh ed N so e ev es en P m P= w en e N e
| |
| * O eN w e a M M w
| |
| ** O e .* .4 **
| |
| dC E
| |
| %.T
| |
| | |
| en w w wh f* w w w w
| |
| * w w en O O O O O O O O O O O O O O e e o e * * *
| |
| * e o e e *
| |
| * f*
| |
| y en W W te3 W h3 le) W h3 h3 In3 W N O e a w O an e e e-da' e en e O se3 fa3 4 E en en en e= me ene en e oe e. w w e e H A e e o e e e . e e e WZ e= e-a w or ,= ce c .= m .= e m N e-R. w g m m m m w e ew N rN en en N N t OO O O O O O O O O O O O C J H M e
| |
| * O.
| |
| * e e e e e e * * *
| |
| * O 4 E W H
| |
| >= eJ WWWWWWWWWWWWWW m e w == e O w w N w w en tw #N ep 'M O E C m en m W W == en n e m N N
| |
| 'N In3 #= 4 et e.ne e e e e e . D. e e e e e e a C 3 m N e-o en e= ,= eo w w e == N fN fN u g; 8" >= 4 8 at: a m m m N w m w e N N en in en ce
| |
| *e" En3 he C. O O O O O O O O O O O O O O ag > 0 l3 * * * * * * + + + * * * *
| |
| * WWWWWWWWWWWWWW 4 3 C e
| |
| @ e s m e m w w es w ch e O w w tw N rs M & 4 m en r* ed an N P* en W M e m f* w we C -e m o e= a. e a e e e e e e e e e e ps. 6 M Q Q in =e ,= e e ee w m w e= cv ,= e=e fa3 m D G 3 O k
| |
| 'O &
| |
| es.e en - is.* e ce ce ev en m m m m e. e. w w e.e .=
| |
| > ta O +3 O O O O O O O O O O O O O O U to et M
| |
| * e + * * * * * * * * * * *
| |
| **'' 4 9"' m W aJ eJ In3 8d fa3 hJ (*3 h3 la) h3 ta3 fa3 ta3 W In3 la3
| |
| 'O b M m > C e fu e m O w w w w > e w w w E o e m e.n O.* en w C = w tw
| |
| : e. .=. 0 e Oe f-C w ,oJ ee y ee e e e . in. e e e e e
| |
| '"* r= rw en ce ** eo e an w e es N fw tw In3 h3 oc > >
| |
| ~ ~
| |
| e u t- M G
| |
| O m ce ew m m w w .= ew w w - e==
| |
| * 8C E 4 ha m O O O O O O O O O O O O O O 3 0 Q W
| |
| s S w
| |
| =C. C 3 m *=
| |
| e * * *
| |
| * e e * * *
| |
| * id la3 fa3 Ea3 W Ed le3 (a3 14 W Id ta3 W W e * *
| |
| =g 2 E
| |
| & O C Coe U en e=e .
| |
| m w e m e w w O N O eD Ch w w e fN f* M CP pe ce fN *e e e O e e e e e e e. e O. e=. e e e e oc e= w #N ce rw ce e == m e m m M
| |
| e w w w n > w w w W W w w c O O O O O O O O O O O ep H e e O. O. * *
| |
| * O. e e e
| |
| * e e U J O = W W W W W W W W W W W W W OWC 4 8C >= 0 O en op w O en O e m - C C H ^ w e M M e # w e C e W w Q e a e oE e e e e
| |
| ~ ~ * ' ~ ~ * * * * * ***
| |
| e e o e e e e e e e E y b ~
| |
| ** o i m Q t% G E e 3 a- '& 6 O O O O O O O O O O O O O r2 O " o a u o O O O O O O O O O O o C z e - - e e O. O. e e e e *
| |
| * e e e
| |
| * g > s b > 3 W b
| |
| =
| |
| W 10 t W W W EJ W W hJ W te W E4 0 C D C C C O O O O O O O O F O eO h C
| |
| t
| |
| +3 C M
| |
| @~ O e Oe Oe Oe O. O. O. O. Oe O. O. e Oe Oe it) "O M g O O O O O O O O O O O O O O W O s s $
| |
| M b u M w 4
| |
| ** 4 O O O O O O O O O O O O O O g to fe, sw C O O O O O O O O O O O O E 4 g o
| |
| * O. O. e e e * * * * * * *
| |
| * O O E a - m WWWWWWWWWWWWWW C M m +e M O O O O O O O O O O O O O O y *- C
| |
| .e.e O
| |
| e O O Oe O O Oe O O. O. O O. O. O oc ee o e e e e e e g u O O O O O O O O O O O O O O CE 4 g M i C
| |
| m 6 m C W so
| |
| ,C. o C e
| |
| :c o e- e O O c w en w w O O w O O O g >=
| |
| e.d
| |
| -a 4- U 4 E 3
| |
| b O O O O O O O O O O O O O O D 3 he s - W te sia is sa: s.3 t3 W h3 sc En3 se te3 fe3
| |
| =e O O O m T U Pu O O O O or O O Eb O O to en p
| |
| g es m
| |
| C C g g en O O a e * *w w O en e O O re Oe O O e e e e e e e e e O fN O O M eo in fN @ O O an O O O U U C b c E De o tr
| |
| >e +# == C
| |
| *e as e 8'". W .=e e T be en w w w f* P= w w w w P= w w en mie eos g .m#e 4 C $
| |
| e g O O O O O O O O O O O O O O
| |
| * * * * * * * * * * * * *
| |
| * er De D 4 CW In3 W 8e3 ta3 de3 fe3 in3 te3 b3 W id h3 W w e=e a C *e a in o en tw O an m o e ep O O O m M e P e M m ee e e a ew w e o ee Oe @e O e e e
| |
| * e C .O C o w e e e e e e e m w w a sn3 y ce .4 w m e-o ,= en we m we em me m e=s +
| |
| * f%
| |
| W C a >e r- a e e e- P- w w e e
| |
| * in e e N w o 46
| |
| ** == 0 he O O O O O O O O O O O O O O m M 4 C 4 I 4 6 6 I s 6 1 8 9 I 6 4
| |
| * WWWWWWWWWWWWWW t
| |
| O
| |
| * i
| |
| $ $ ew e- m e w O e = w O e w O O w w c- n
| |
| =
| |
| g E. w O O e .n O m N D O O ,e e.e g k e . b C e e e e e e e e e o e e e e M M p eC O 6e M == P" en P* en er Pe fw e e *"* e= e=*
| |
| W D b f 9 I M
| |
| i ,e) e e O 4 E e
| |
| Pe f9 m e in 01 P= m e O e==
| |
| s'a f% M w
| |
| ** *=* *=
| |
| e-o ta eo 2 en an eo Do8
| |
| | |
| 4 G
| |
| e E O 6' rw a ** == ew rw m m * == m = a ..
| |
| m O ha o O O O O O O O O O O O O O w e .* * * * * * * * * * * * *
| |
| * C s 0 T, Q. W W W W W W W W W h3 W W W W O e s N - e O = 0
| |
| * O w m w w O O C,
| |
| s e, a O m tw w c. w w p w cw in. . e. . e. m. cv. e. e. O. . .
| |
| l m, w s, O O e. ev e w w in w m m rw e r- .=
| |
| 4 s.
| |
| 6 O > *a >
| |
| U 0, 3 aC e ec= .3
| |
| >=
| |
| - p - ~ N N - - - ~ ~--m -
| |
| T r= 6 O O O O O O O O O O O O Oe O e o e e m e e e e e e e e t T 4 W W h3 W W tal e-W h3 W We. W wW .= W W w w 6 % O e e* e m we e w ei a E w e=e ** w w w p a g g e g O. . . e. . . em. N. r=.
| |
| . . in. m. m.
| |
| 6 m e m O e* m rv ee N -= en ei == ev e e w w g > 0 e O 4 4 O W D
| |
| ** es N N ** ** O O ew .-e s= es N ew es-
| |
| > O O O O O O O O O O O O O O e e i e e e + + e e *
| |
| * e e
| |
| **" T = W h3 W fa3 W W h3 W W W We= In3 e' em w w w e m in e W e-W T 6 5 4 EP O w a 6 tw w w C se g n e.=w . . m. e. r . M. in. . e. o. m. *=. m. m.
| |
| s e t > w ed a w e f* in ww .* ** ev rw N
| |
| > O -4
| |
| .C aC O M
| |
| as-.
| |
| I L T #
| |
| w E C o T ** O ew tw -a -= O m - .= m m O O
| |
| .C 3 6 e.
| |
| 6 O O O O O O O.
| |
| OO e e O.
| |
| O O O O
| |
| * *
| |
| * e g w e e e o e *
| |
| * w h. e Q. W W W W tal W W W We- W WWW C e % m N - e t= m o m w en ' h3= rw e e 6 > O rw w P= w en e in w e w O de *N* O* *O
| |
| * m aC 47 O > m r= w in w e* w *ea* = m == en em em -
| |
| p 6 mO sia y so w O ** >
| |
| W D Q M CL 5 - m .
| |
| m ''v & F
| |
| ""0" m w C
| |
| +
| |
| N ia
| |
| @ W w m e e ew N m .= m rN ** en m m N ** w 7 E @ 6 C aa m O O O O Oe O O e O aO O Oe O O eO e e e u D P"" M T g e e o e e e e t' C & N O W Ea3 !a3 W W W W W WP- WesWF h3 W W m O C 0 w tw ew 0 0 9 O m >= O ==
| |
| u 6 4 t O
| |
| a en = m m w O ==
| |
| * en O
| |
| : e. e. m. w O. rw. .rw Q e* m e g a m m O en. .n. O. c. rw e a e . 4 L.
| |
| CD h > c e o aC - m m N N N N e w - e e- mm 4, m T e su T T 4 O E O e U b C
| |
| @ g "o
| |
| # Im I, e C I
| |
| e eN v
| |
| .c p
| |
| w N w e N N m O N N O O O O Oe Oe Oe O eO eO Oi Oe *O *Oe e
| |
| .O O N N m
| |
| g cw - > - e
| |
| @ i E i i i e- > - 6.
| |
| e M u 6 W W W W W W W W W W W h3 W ta3 % +
| |
| @ w & w ( m - m e o e o c w e w O - - %% & 4 tw C >
| |
| Q $ e i e,
| |
| > e. m. *=. . .
| |
| w w
| |
| : m. en. m. e. *=. tv. ch. tw. rw . cw c. .= e 4, U 4> O o
| |
| -a in == == e e e' W m ew em e ea - > > > > C C F
| |
| e , t 6 Of C 2: O 2 m m -
| |
| E rw E rw . h. 9 A O ** O c* @ 4 E C v.
| |
| b E ew ee es ce rw ew m m ** s-a m m O O 4 > 4 > m 4 &
| |
| * O > t= m er= 0 t @ O 3 Oe Oo O eO *O *O* O* O* O* O* O O T a 3 s w e 'e *
| |
| * m e W W t.3 W h3 W W W W h3 h3 la3 h3 W 7 4 O D a. g ee w e e en in w m ew e en w en en C D e s n ee %
| |
| CL
| |
| - w rw rw &
| |
| o - - o e C, g 3 O > e. . e. O. . c. . e. w. se. e. m. se. am. 6 C w L m 4 b O N P= w w en M m *=e ** rw an r* w w C e. U **
| |
| g 6 m o e g &
| |
| D w o ** E w T P""
| |
| 6 n., 0 Q w e. .&= L w C g m U be b e M c*
| |
| W h3 ta T w ***
| |
| * D.3 ** e* N rw ee ee O O N ew O s-* N N N 6 0 0 4
| |
| . em I
| |
| e.
| |
| O O O O O O O O O O iOa Oe eO O a o 0 .c M E
| |
| < e e O q w 6. -
| |
| > > z
| |
| : a. e.
| |
| e i i a e i e e h3 e-W W W W W h3 W h3 h3 W'W W te3 O E 4 ape O 6 e 6 **
| |
| O == >
| |
| M h3 e e s > EP m e in O ne em N ei O w e e in' N w O e tw ** ea a er=
| |
| e.*
| |
| > c aC O Q.
| |
| =*
| |
| N N m e- e e
| |
| * * * * * * * * * * * * *
| |
| * in. mfw e m 3 D W Mm t C C 4.
| |
| g H w es a w G w in m N en e ** ** ** N e4 M N D e=0 O M @@@v b MN@@ 4 4 y a w w 4 3 cO. m a W > E E O O
| |
| .C* es O w Q. Q. O O s e 5 9 C 4 M M e e F M w w e e w @ e e C E @ O O O O O O O O O O O e O eO eO e e N N T e Tg C
| |
| >= e e e e e e e 1 4 in e.m 9 E A,
| |
| WW W W W W In3 e- m e, w C h3 W h3 W W W in!
| |
| e- e w O e w O O O O u O O w a.
| |
| 6 en. m e
| |
| N g e 6 e O 6.
| |
| i.e e w oO e in w w r- O m tw O O O in in e e e rw m. e Ch. * * * * * * * * * * * * *
| |
| * s'* ed > >
| |
| O U h. Q.
| |
| m ee P- an r= in er N rw w e es ** ** in in at 4 e.
| |
| Lh) en B B B 9 gg, e.
| |
| 4 % e O w > C @ W G =* P* e tP O ** rw *) *r ** - N rw o h a a * ** rw m e in e L -= e o e ** * * * * ** *" > > > > z dC C E e
| |
| D-9
| |
| | |
| I' M
| |
| C E O @6
| |
| * m 6 al I O I sc O A tw a m w N N m w == .= in er O - O= O y w % C Oe O +O eO *O *O *O* O* O* O.
| |
| * e e a omO +
| |
| ''" C & O g0 W W W W W to: W W W w tal W W h1 W
| |
| @ E a. O > eu ed in - w o w e=.
| |
| m e w O O 4.e== m W h ** E en.
| |
| fw f*
| |
| : e. O. . m. N. . e. O. r=. N. . ar.
| |
| g @ h. & M e we .e ei O > tw e e* w in w m ew N W er
| |
| .D
| |
| - 0 3 d e e s e e e
| |
| '** a T - .= ** o . . .a oO ev .= N .* N N i
| |
| @ W @ E Oe sO oe Oe O O eO *O* Oe O Oe O* O* e O a 3 E w @ e "O h* @ b. 4 W h2 hJ W h3 h3 W W W Ia2 la2 tal hJ taJ
| |
| @ m # > e o N N ew w w e m an un f*e e
| |
| > 0 Q. E f=
| |
| : m. en. en. e. O. .a. e. m.
| |
| ".>" ( O w N. er. e. O. . .
| |
| *** w ** * * *
| |
| * e r* an P w == w 6 N fw l l l 9 i
| |
| i C
| |
| *==e
| |
| .C .e E g &.,
| |
| n . G.
| |
| h O C l
| |
| *P" R .* m a e o O m or o o ow o o en O o O 3 e O % ar O
| |
| + O* O* O* O* O* O*O *O *O +o +O +O *O
| |
| * M w Q O > h3 W W W W W la: W h2 Y
| |
| C e
| |
| @ C
| |
| * C5 O
| |
| * O O o fw O O o r'= o Id W#.7, w W O Oh2 O ta:
| |
| O m, g
| |
| = h. -
| |
| e+ gc O. O. O. o. O. O. O. en. O. . e. O. O. O.
| |
| ,g m O O fw ** O O O a O O e O O O p he w
| |
| & O e >
| |
| .e ,
| |
| 4 O O Oo Osa O O OO ** oO O O N o O oo p O I C H T C ar o O
| |
| : o. o. O. o. o, V .e C - @ > + o. * * * * * * +
| |
| a0 .== e a m W ta: las h3 W h3 W ta) h: h3 la2 W ta2 W a e O + C C o m o o O rw wOO O w o OO O O n a C e. o h
| |
| O. O.
| |
| O
| |
| * O.
| |
| O
| |
| *o .
| |
| * O* e. *
| |
| * O*
| |
| mC d
| |
| ** C *. -
| |
| e m
| |
| O O N .* O o* o. O ** O O ar O oO 9 wW 3
| |
| m
| |
| @ *> % 3*
| |
| I a O > b O o o N O O O <w o oo o O O Q h C h-o ti E t
| |
| * o
| |
| * O
| |
| * O
| |
| * O
| |
| * o
| |
| * O
| |
| * O O
| |
| * O
| |
| * O
| |
| * o
| |
| * O
| |
| * O a
| |
| *Q u M h.r *
| |
| * g g a
| |
| @ b 8 w
| |
| E W W In: W In:
| |
| o O in fw o Id O
| |
| WO h3 la hJ hl W la3 W e o o O O o O 7 .6 D C' == 0 C
| |
| .O M b 3 == > c. O. a.
| |
| f*
| |
| O. O. O. e. O. O. O. O. O. O.
| |
| gg
| |
| .O .
| |
| g g g O o - - O O o w O O O O O O M w u gg e V o E. s O O m - O o O O O O O O O o e e O O O O eO *O*o* O* O* O* o* O* O* o*
| |
| O M e.' C &
| |
| h *
| |
| * s g I Om 4 ta: W W In3 W W Ia3 W W fa3 Id W fa3 fa)
| |
| C - C w o O O O o O O O OtwO o O O o E M MD > O. O. O. O. O. O. O. . O. O. O. O. o. O.
| |
| @ O O N M o O O *=8 O O O o O O C
| |
| e E O b Q w l W h.
| |
| m e g a O .e o Q.
| |
| w O o m er o O o e O O an O O o C ** Q % o* o* O* O* O* O* O* O*O*O *o *O *O *O
| |
| * D $ E e O 3 to C e s hea C O h3 W h3 W W W h3 fa3 h3 D1 ta2 b3 fa3 h3 g In) O e m 0 o CD m
| |
| O o O O o o O o O O w o O OQ
| |
| *. a= 0 *=
| |
| en. O. O. e. Q. O.
| |
| a c. 3 .C. =, O. O. O. O. O. O.
| |
| 80 .D3 .a e+ > O. .
| |
| > d e O O N ed O O O .=* O O w O O O G *
| |
| > U 1
| |
| * O O O .* O O O ** O O fw O O O e.
| |
| "P* == m W H C e C E Oe O+ O+ o+ O+ O+OeO *O *O *O *O *O
| |
| * O + me 3 d e. y O g g ar g d h2 tal in3 In3 h3 fa3 h3 la3 W W ta3 ta2 ta3 W
| |
| @ 0 b) m =a C a O e m 6.
| |
| m O O O o O O O o o O w o O O si >
| |
| W z += a oA > 0 0 O. O. O. o. O. an. O. O. e. O. O. O. O w e g O O N ** O O O ed O O w O O O O >
| |
| 0 C
| |
| == 8 m e+ >
| |
| m C a >* M > t O
| |
| - - = u hs O .= 4 2
| |
| ** 9 C e r= e e f* r- *D e e a w in e e o m N g E @ f*
| |
| Oe Oe Os Oe Oa Oe Oeooo ce N N >
| |
| g, e
| |
| e eO eO eO eO e a we W e 6. h.
| |
| E h3 W 63 N
| |
| in) in) ta3 Is3 W h3 h3 W h2 fa) ta3 f* m ew O e W tr O e W O O M
| |
| * r O ct. e h.
| |
| 4- U u h, A w* O* O*
| |
| * e *in* w w r- O m ew O O O it
| |
| * ' *
| |
| * m > ""' '''
| |
| O m .* r= an t= en e tw' rw' w' e' e. .= ,* > = > >
| |
| @ 6 a a e D"i e L w > me m 6 == e eO .== N M m w > >
| |
| I a w . ,e rv m e in w P-
| |
| ** * * *a e*r > > E E a e O *9 d C E D
| |
| M D-10
| |
| | |
| ~
| |
| n io t
| |
| t a
| |
| n e
| |
| m mru e t t s syI n
| |
| S dn l
| |
| p l e a a m n v s o u a l e n s
| |
| e P P ms e o
| |
| C R ie i
| |
| r s
| |
| _ t s i r e D s ne n e t a e n o D t t
| |
| r e n o
| |
| t t
| |
| g it d e o p a r o t x
| |
| m n
| |
| i B
| |
| a d n wk a c n Hf p E g
| |
| t p e a r e a o y e S s u l i r e T e h oB t
| |
| l i
| |
| s r c o F v cn r a i r
| |
| a y ge a t
| |
| A s s e m i
| |
| t P r ec e t e s t
| |
| u W D e n ev r a a r A e g rD r
| |
| t a a d o emt e a l v t e a l i
| |
| l n eF i
| |
| n B hx ie i
| |
| t a
| |
| r o mG V p u x r
| |
| i S
| |
| t l
| |
| t h n Cd owI yr rE h h a l e A mp my S m S A se IR mA l
| |
| e I u
| |
| t a Ti i t e uh g a u o o t er t
| |
| e ie n s S u l
| |
| A l
| |
| v D R R t P ver end WEq a S a ne ucr 'p t t
| |
| t e r t n R a umB de C e i
| |
| 0 d e d n I l 1
| |
| n a
| |
| L e a T a t
| |
| dn R I
| |
| D Ci- occP te a a t
| |
| c GS t
| |
| u e e w smtcr Wo cic smes i i D s e S ic l l e
| |
| le m Wd e e dp ai e r
| |
| s e F l d i ie k
| |
| a R D R S DN S E A R D eS eD e i
| |
| b a e e l t
| |
| n a N I T n i
| |
| o t
| |
| a i
| |
| c f
| |
| i d
| |
| o M
| |
| f o #
| |
| y e r- p a
| |
| r s eu t
| |
| m CAal m A m Cy G 1 1 i 3 3 2 1 1 1 1 4 /3 i N
| |
| S u Rde uv NtSa i t
| |
| n s
| |
| - 6 i e
| |
| i eif e 1 - r e
| |
| sdm ao a 1
| |
| l 2 4 8 9 l l l 1 1
| |
| l 1
| |
| c c 2 t t
| |
| a a a a a a a i i CMN C yon mmmmmmmm1 e e e e e e s s e e2 R i i HB wn a d y r
| |
| R di Nua c t I
| |
| t n n n n n n S I
| |
| t I
| |
| t I
| |
| t I
| |
| t I
| |
| t e eS F F D Se NWSp r
| |
| i i r i t
| |
| S Gon Oir et Wcab
| |
| /im f 1 2 3 4 5 6 7 8 9 0 1 I
| |
| I 2
| |
| 1 31 4 1 I
| |
| iu Rd PoN EM gi
| |
| | |
| On the basis of average cost per unit, we have the following companson using the NRC Case Study esti-ue from table 6.3 of the impact analysis and half the estimates in the last column of table D-1. The
| |
| .ollowing summarizes these cost estimates for installation and engineering costs only:
| |
| NRC Case EPRIAVOG Study Estimate Altemative Esnmate (oer unit) 1 $7.4 million $10.6 million 2 14.4 million 21.4 million 3 22.5 million 43.3 million 4 59.0 million 90.0 million The WEP estimates are seen to be higher by 50% to 100%.
| |
| Table D-11 describes for each modification the factors included in the basis for the different cost estimates.
| |
| REFERENCES W. R. Crammond, et al. " Shutdown Decay Heat Removal Analysis of a Westinghouse 2 Loop Pressudzed Water Reactor, Case Study," NUREG/CR-4458, March,1987.
| |
| l l
| |
| l
| |
| /
| |
| D-12 .
| |
| l
| |
| - _ - - - - -------_____-___a
| |
| | |
| Table D-11 Basis for EPRI/WOG Cost Estimstes Internal 1 - RWST Level Alarrn (MOD 501)
| |
| WEP estimates cost as $60,000 versus $22,000 in the NRC Case Study, ne principal differences include a fixed startup cost mimmum of $10,000 for paperwork and processing. Further, this modifica-tion appears to be inconsistent with the human engineering constraints resulting from the control room design review, which aempts to minimize the number of alarms and lights in the control room.
| |
| Intemal 2 - Dedicat; d Diesel Generator Battery (MOD 302)
| |
| WEP estimates costs as $800,000 versus $333,000 in the NRC Case Study. The principal basis for dif-ference derives from experience gained by WEP with the recent installation of new batteries. This design is redundant to the new batteries. Further, the design includes a safety hazard of venting batter-ies, and therefore hydrogen, to the air compressor room.
| |
| Internal 4 - Redundant RHR Valve (MOD 803_)
| |
| WEP estimates costs as $30,000 versus $12,000 in the NRC Case Study. The principal basis is from the recent experience ofinstalling containment isolation valves at $30,000 per valve, which includes fixed cost of,$10,000 per modification, tests, technical specification submittals, and other paperwork and processmg costs.
| |
| .nternal 8 - Soare RHR Pump (MOD 816)
| |
| WEP estimates costs as $1,700,000 versus $909,000 in the NRC Case Study. The principal difference results from the prov.sion of a different power arrangement, i.e., an additional 480V supply and associ-at;d electrical equipment, to ensure proper consideration of the swing bus power problems referenced in the NRC Inspection and Enforcement Information Notice (IEN) 86-79.
| |
| l Internal 9 - Diesel-Driven Auxiliary Feedwater Pumo (MOD 817)
| |
| WEP estimates costs as $8,000,000 versus $2,606,000 in the NRC Case Study. The principal difference is that piping could not go through the non-seismic turbine building. The WEP estimate assumes con-struction of a new seismic building to house the pump.
| |
| Intemal 11 - New Condencate Storare Tank (MOD 818)
| |
| WEP estimates costs as $2,500,000 versus 01362,000 in the NRC Case Study. The principal difference includes the consideration that in:tallation c:atnot be in the area suggested because maintenance shop and offices curantly occupy part of that space, and because of undergmund piping and cabling in the area. Additional cost: include extended piping runs and rework problems for the underground cabling areas. De initial design will not succecc; constmetion will uncover underground cables and pipes, work will necessarily cease, and the design would have to be redone. These rework costs can be sub-stantial.
| |
| D-13
| |
| | |
| Seismic 1 - Seismic RWST Altemative Connection to Soent Fuel Pool (MOD 806)
| |
| EP estimates costs as $2,500,000 versus $1,582,000 in the NRC Case Study. The principal differen-
| |
| ..s include a requirement for two contractors, versus one, and associated interface costs, the necessity of meeting the cable separation criteria in appendix R,ine eased pipe routing for seismic design, additional penetrations and more cable routing. Also, work will occur in high-radiation areas and some activities, such as pipe cutting, will create airborne pollutants necessitating work in full suits and respirators.
| |
| (Note that the RHR pumps can already take suction from the spent fuel pool via a two-inch pipe con-nection.)
| |
| Seismic 2 6 - Electrical Equiement Anchorace NOD 107)
| |
| WEP estimates costs as $500,000 versus $146,000 in the NRC Case Study. The principal basis results from information gained from recent work on new seismic-grade batteries and previous upgrades of elec rical equipment, including station batteries and two MCCs, performed three years ago.
| |
| Fire 1 - Auxiliarv Feedwater Pump Room Fire Protection fMOD 810)
| |
| WEP estimates costs as $1,000,000 versus $436,000 in the NRC Case Study. The principal basis results form information gained from the recent installation of a Halon system in the referenced room. The 51 million Halon system cost included about 75% for detection. The NRC Case Study estimate does not {
| |
| include adequate cost for a detection system. Using the existing detection system would compromise overall system reliability, and is not accounted for u either the NRC Case Study or EPRI/WOG model.
| |
| Further, fire wrapping of critical cables has been performed in this area since the NRC Case Study analysis, and cable wrapping was identified as an altemative to this modification.
| |
| 're 2 - Relocate Bat'erv Charcer B and DC Distribution Panel (MOD 311)
| |
| WEP estimates costs as $1,200,000 versus $596,000 in NRC Case Study. Principal basis is the experi-ence ofinstallation of conduit and wiring for the new batteries. Further, the installation would require shutdown of both units which is not accounted for in either EPRI/WOG or the NRC Case Study.
| |
| SDHR - Dedicated Shutdown Decav Heat Removal System WEP estimates costs as $90,000,000 versus $59,000,000 in the NRC Case Study. Principal basis is the l i
| |
| additional piping and construction required because the referenced tunnel cannot be routed where '
| |
| indicated in appendix J. The costs associated with the iteradve nature of underground construction are included in the EPRI/WOG study as well, i.e., the rework costs discussed for internal 11. EPRI/WOG estimates that the work will require four refueling outages rather than two.
| |
| Wind - Diesel Generator Exhaust suonorts (MOD 119)
| |
| I WEP estimates costs as $30,000 versus $15,750 in the NRC Case Study. Principal basis is the fixed cost of $10,000 per modification for tests, technical specifications submittals, and other costs. Note also that the NRC Case Study appears to ignore existing supports installed for IEB 79-14. .
| |
| Sprav -Intake Structure Shield Wall Extension (MOD 109)
| |
| WEP estimates cost as $790,000 versus $178,000. Principal basis is the cost to disassemble the existing wall and erect a new wall. This work is not included in the NRC Case Study which neglected to onsider that the fire protection spray header over the pumps is seismically supported. Iteration is quired generally for seismic cor.struction, thereby increasing,the cost. WEP estimates do not reflect alat there is a high risk of damaging the pumps during construction and that a two-unit shutdown may be required to maintain pump motor separation.
| |
| D-14
| |
| | |
| Appendix E i DETAILS OF SDHR INEFFEC'ITVENESS AND UNAVAILABILITY QUANTIFICATION SDHR ineffectiveness is defined as the fractional reduction in core melt frequency with and without the SDHR installed. SDHR ineffectiveness may differ from its unavaili. ility if, for example, the SDHR is assessed not to succeed for certain event conditions, or if the presence of the SDHR changes the base case recovery analysis. Section 7 describes the assumptions and considerations which are important to the EPRI/WOG quantification of SDHR unavailability. Table E-1 (also table 7-1 in section 7) presents the results of a calculation of the ineffectiveness of the SDHR.
| |
| OVERVIEW OF THE FAULT ANALYSIS OF THE SDHR The NRC Case Study (1) evaluated the SDHR system for faults under two types of conditions, namely with and without offsite power. The system was not evaluated for any external events, but was assumed to be invulnerable to all such special emergencies. The SDHR is evaluated in this EPRl/WOG study for
| |
| ! . variety of accidents, including cases where the SDHR is both 1: ,
| |
| end less reliable than the NRC Case Study. The EPRI/WOG study considers:
| |
| I o Infant morality.
| |
| o Failure to consider all required components.
| |
| o Common-cause effects between normal plant safety systems and the SDHR.
| |
| o Events which could be mitigated by either train of the SDHR systems, i.e., by the makeup train or by the emergency feedwater (EFW) train.
| |
| l o Actuation of SDHR.
| |
| o Seismic response of SDHR.
| |
| 1 Each of these points was discussed in section 7. The objective of the following discussion is to provide a traceable presentation of how the issues discussed in section 7 are translated into an assessment of ,
| |
| SDHR unavailability and ineffectiveness. The first three considerations in the above list are addressed initially so that the unavailability of the three individual parts of the SDl31 can be calculated. These three parts are:
| |
| E-1
| |
| | |
| Table E 1 Summary of SDHR Impact on Core Melt Frequency and Comparison of EPRI/WOG and NRC Case Study Assessment of SDHR Unavailability (Base Case EPRI/WOG Assumptions for SDHR)
| |
| EPR1/WOG EPRl/WOG Core Melt Core-Melt Fre- NRC Case I Frequency for quency with Average inef. Study Assumed Core-Melt Existing Point SDHRInstalled festiveness! Ineffectiveness 2 Seauence Tvoc Beach Plant /vr mase Case) of SDHR of SDHR Long-Terrr Station Blackout 5.4E-7 4.0E 7 0.74 0.1 Shon Term Station Blackout 8.9E-7 1,9E 7 0.21 0.1 6.8E 7 3.3E 7 0.49 0.052 S2MH1'H2' and S2MD1D2 4.5E-7 1.7E-9 3.7E-3 0.052 O'her Internal Events Fire 6.3E-8 2.3E 10 3.7E-3 0.052 Seismic
| |
| < 3 x SSE SBLOCA 1.7E-6 2.6E-7 0.15 0.1 transient 3.5E-6 3.4E 7 0.096 0.1
| |
| > 3 x SSE SBLOCA 1.4E 6 2.0E-7 .15 0.1 transient 8.02-7 8.1E-8 A26 Ql TotalsComposite 1.0E 5 1.8E-6 0.18 0.084
| |
| : 1. Ineffectiveness is defined as the fractional reduction in core-melt frequency with and without the SDHR installed (column 2 divided by column 1).
| |
| : 2. NRC Case Study ineffectiveness is assumed to be equal to SDHR unavailability.
| |
| E-2
| |
| | |
| EFW-SDHR: The emergency feedwater portion of SDHR.
| |
| l Makeup-SDHR: The high pressure makeup ponion of SDHR.
| |
| l f
| |
| SDHR Suppon: The AC and DC power suppon ' system' for the SDHR.
| |
| Given the unavailability of the three pans of the SDHR, the unavailability of the SDHR as a whole for each accident sequence type can be determmed by considering the other issues in the previous list, namely:
| |
| : 1) Whether or not either EFW-SDHR or makeup-SDHR can prevent core damage,
| |
| : 2) How the SDHR is actuated and, in panicular, the need for manual actuation and its impact on SDHR ineffectiveness, and
| |
| : 3) The seismic capacity of the SDHR. These issues, together with whether offsite power loss requires SDHR suppon, determine the overall ineffectiveness of SDHR.
| |
| The next two subsections discuss the issues affecting primarily SDHR unavailability, i.e., the unavail-ability of its individual parts, and then the issues affecting primarily SDHR ineffectiveness, i.e., the applications of those unavailabilities to specific accident sequences.
| |
| Unavailability of the Parts of the SDHR The unavailability of the three pans of the SDHR is determined in the following manner: First, the NRC Case Study component failure probabilities and the EPRI/WOG diesel generator failure probabili-ties are used to calculate a " component based" unavailability for each pan of SDHR. Then, the EPRI/WOG study considers the effect of infant mortality, other components and common-cause failures on the unavailability of the SDHR pans.
| |
| Component Based Unavailabilities for Parts of SDHR The first step in the calculation of SDHR unavailability is a calculation of the component based unavail-ability for the three pans of the system, namely emergency feedwater (EFW SDHR), high pressure makeup (makeup-SDHR), and the supponing diesel generator and battery system. Section 4 of the hTC Case Study reponed component-based unasallabilities. These component unavailabilities could not be used by EPRI/WOG to recreate the SDHR unavailabilities also reponed in that section. For example, EPRI/WOG calculated a total failure probability for EFW-SDHR and makeup-SDHR of 4.lE-2, while the NRC Case Study calculated 5.2E-2. However, the EPRI/WOG study used their component-based values to calculate the following train unavailabilities rtue to component failures and test and maintenance downtime. (EPRI/WOG values, rather than the NRC Case Study values were used to quantify diesel generator faults.)
| |
| E-3
| |
| | |
| Total Portion of SDHR Unavailability Comoonent Faults T&M Downtime EFW-SDHR 2.0E-2 1.7E-2 3.0E 3 Makcup-SDHR 2.1 E-2 1.6E-2 4.8E-3 SDHR-Suppon 3.3E-2 2.6E-2 7.2E 3 Consideration of Inf ant Mortality. Other Components. and Common-Cause Failure The above component based unavailabilities were used in the EPRVWOG as a startmg point for calcu-lating SDHR unavailability for each of the three parts. The calculation of total unavailability also de-pended on three other factors; namely, infant mortality, other components, and common cause failures.
| |
| As was mentioned in section 7, the impact ofinfant monality on unavailability of each part of SDHR was approximated by increasing the component based unavailability by a factor of two. The factor of two represents a " conservative" application of the factor of 3 found in plant trip infant monality data.
| |
| Test and maintenance unavailability was assumed to be unaffected; however, corrective maintenance would be expected to be higher.
| |
| The impact of other component failures, e.g., room cooling, was performed by doubling both the amponent-based and T&M based unavailabilities. The reanalysis of these two impacts increased the unavailabilities of each pan of the SDHR to the following:
| |
| EFW-SDHR 5.7E-2 Makeup-SDHR 3.7E-2 SDHR-Suppon 9.2E-2 Common-cause failure of SDHR components was evaluated for two conditions, namely common-cause failure within the component of the SDHR and common cause between components in the SDHR and components in the remainder of the plant. Common-cause failure among components in the SDHR was only significant as a contributor to the simultaneous failure of both the EFW-SDHR and the makeup-SDHR. Common-cause failure between the SDHR and other plant components w s judged to be sig-nificant only for diesel generators.
| |
| For common-cause failure between motor operated valves within the SDHR, a value for all MOVs was I obtained by using the EPRI/WOG value of 8E-5 for two MOVs and 20% of that value for three MOVs.
| |
| The factor of 20% is based on Atwood's results (2). Two combinations of two MOV failures and two
| |
| -1 E-4
| |
| | |
| combinations of three MOV failures causing failure of both EFW-SDHR and makeup-SDHR were iden-tified. The resulting unavailability was calculated to be 1.9E-4, i.e.,2 times 8E 5, plus 3 times 8E-5, times 0.2. That value was doubled to 3.8E-4 to account for infant monality.
| |
| These common cause failure factors were applied as follows. If either the EFW-SDHR or the makeup-SDHR could prevent core damage, then the common-cause failure of MOVs were added. These MOV common cause faults accounted for about 10% of the unavailability when the two trains are considered in tandem.
| |
| Common-cause failure between the diesel generators in the plant and the diesel generator in the SDHR was the only type of common cause between plant components and SDHR components considered.
| |
| This limitation was taken because the only significant common-cause failure in EPRI/WOG's dominant accident sequences for Point Beach without the SDHR was for failure of both diesel generators. Since single diesel generator faults were also significant to both short and long term station blackout sequences, common cause failure between one diesel in the plant and the SDHR diesel was also considered.
| |
| Common-cause failure beta factors for a third diesel, generator given failure of two diesel generators was estimated based on a study by Atwood (1). That study indicated that a beta factor of 0.125 should be used for common-cause failure of three diesel generators given common-cause failure of two. The beta factor used for common-cause failure of the SDHR diesel generator given failure of a single plant diesel generator was the same as that used in both the EPRI/WOG study, namely 0.023.
| |
| F The EPRI/WOG accident sequence cut sets were reviewed to determine which accident scenarios were susceptible to diesel generator common-cause faults. For those cut sets with both diesel generators failing due to common cause, the SDHR unavailability was obtained by adding 0.125, the beta factor for three diesels failing given that two in the plant had failed. For those cut sets with single diesel generator faults, the SDHR unavailability was obtained by adding 0.023, the beta factor for a second diesel failing given one diesel in the plant has failed. For those cut sets with two independent diesel generator failures, the SDHR unavailability was obtained by adding two times the beta factor for a second diesel failure since common-cause failure could occur between either diesel, or 0.046.
| |
| EVALUATION OF THE SDHR FOR KEY ACCIDENT TYPES The dominant accident sequences in the EPRI/WOG study are listed in table E-1. The following describes how the ineffectiveness of the SDHR was calculated using the previously described numbers.
| |
| E-5
| |
| | |
| Lone-Term Station Blackout ong-term station blackout evaluation requires consideration of two aspects of SDHR quantification.
| |
| Jne aspect is the calculation of SDHR unavailability using the numbers calculated above. The second aspect is the consideration of the impact of the SDHR on the recovery actions considered in the base case EPRI/WOG accident evaluation.
| |
| The unavailabilities described above were applied in the following manner. First, because offsite power is lost, the SDHR-Support must function for either the EFW-SDHR or the makeup-SDHR to function.
| |
| The SDHR unavailability is then obtained by adding the SDHR-Support unavailability to the product of the EFW and makeup SDHR unavailabilities. (Either EFW-SDHR or makeup-SDHR can prevent core melting for transients with RCS integrity intact.)
| |
| Second, the diesel generator common cause factors are added to the appropriate cut sets in the manner described in the previous subsection.
| |
| The impact of the SDMR on the base case recovery model was calculated as follows. First, as described in section 7, the EPRl/WOG assumptions about EFW-SDHR actuation require that the SDHR be actuatt.1 manually for sequences involving delayed safety system failures. It was assumed further that sanual actuation of the SDHR, and attempts to restart the SDHR should it fail, would divert the sperators from considering CST refill. (C2T refill is the base case recovery action in the EPRI/WOG model for long term station blackout sequences.) The diversion of the operators for sequences where the SDHR failed was assumed to '' cost" ten minutes. The loss of ten minutes reduced the time availath to diagnose the need for CST refill from 20 minutes in the base case to 10 minutes. The human error F portion of the unayallability for this recovery increased from 0.01 to 0.1. ' Die total unavailability of the CST refill recovery action increased from 0.03 to 0.12. (Equipment faults accounted for 0.02 of the unavailability.) This change in the CST refill recovery actually decreases the effectiveness of the SDHR for long term station blackout scenarios by a factor of four.
| |
| Short-Term Station Blackout Short term station blackout quantification was essentially the same as for the long-term etation blackout case, except consideration of adverse impacts on the recovery actions was not required. That is, the SDHR-Support must function, either EFW-SDHR or makeup-SDHR can prevent core melting for transients with RCS integrity, and common cause failures between trains of SDHR and diesel generators is considered. There is a difference however for the short-term station blackout scenarios resulting in a stuck open PORV. In this case, makeup-SDHR must function and EFW-SDHR is not considered.
| |
| E-6 1
| |
| | |
| l 1
| |
| Small Break LOCA Sequq1gr3 The principal small break LOCA sequences are recirculation failures which the EPRI/WOG study dssumes Will require manualinitiation of the SDHR. The requirement for manualinitiation significantly reduces the effectiveness of the SDHR because of dependence between human failures in small break LOCA recirculation failures and SDHR manual initiation. The dominant accident sequence cut sets for the small break LOCA sequences include a failure to implement recirculation or recirculation recovery.
| |
| In the case of failure to implement recirculation, the EPRI/WOG study assumes that SDHR will not be actuated manually. (The EPRI/WOG study also assumed that recirculation recovery would not be performed if the operators failed to implement recirculation, i.e., the human errors were dependent.)
| |
| The basis for this assumption is the strong coupling of human actions for faihire to diagnose the need for recirculation and the need to initiate the SDHR.
| |
| Most of the remaining small break LOCA failures included failure to recover recirculation by refilling the RWST or correcting the faults causing the recirculation failure. Recirculation recovery failure is dominated by equipment faults, i.e.,0.045 of 0.05 failure probability (see appendix A). The human error is determined to be 0.005. However, again assuming the SDHR manual actuation and the failtre to diagnose the means for recirculation Ncovery are strongly coupled, the EPRI/WOG study assumed that the SDHR could not improve the reliability of recirculation recovery by more than a factor of ten,
| |
| .e., the residual human error probability. Therefore, despite the makeup-SDHR train unavailability of 0.058, the SDHR ineffectiveness was taken to be 0.1. For all remaining faults, the SDHR would be actuated automatically and the makeup-SDHR unavailability would apply.
| |
| r Transient Secuences The SDHR unavailability was calculated based on the following considerations. Offsite power is avail-able in all these sequences (station blackout transients are considered separately above) so the SDHR-Support is not required. Either EFW-SDHR or makeup-SDHR could succeed, therefore the SDHR unavailability is obtained by addmg the product of the train unavailabilities to the MOV common cause failure probability.
| |
| Fire Seouences The SDHR unavailability for fire sequences is calculated in the same manner as for the transients sequences.
| |
| Seismic Secuences he SDHR unavailability for seismic sequences is calculated based on the following considerations.
| |
| First, all seismic core melt sequences also have lost offsite power. Therefore, the SDHR Support is l
| |
| E-7
| |
| | |
| i required to operate. Second,if the seismic sequence is a transient with RCS integrity maintained, either EFW-SDHR or makeup SDHR can prevent core melting and the remaining SDHR unavailability is ibtained by using the product of the train unavailabilities. If the seismic sequence results from a small break LOCA, the SDHR unavailability is obtained by using the makeup-SDHR train unavailability (along with the SDHR-Suppon unavailability),
| |
| t 1
| |
| l l
| |
| l l
| |
| t l
| |
| E-8 l
| |
| | |
| i l
| |
| I REFERENCES Crammond, W.R., et al, " Shutdown Decay Heat Removal Analysis of a Westinghouse 2-Loop i Pressurized Water Reactor, Case Study," NUREG/CR-4458, March 1987.
| |
| : 2. J.A. Steverson and C.L. Atwood," Common Cause Failure Rate for Valves," NUREG/CR 2770, USNRC, February 1983.
| |
| : 3. J.A. Steverson and C.L. Atwood, " Common-Cause and Individual Failure and Fault Rates for Licensee Event Reports of Diesel Generators at U.S. Commercial Nuclear Power Plants, NUREG/CR-2099, February 1981.
| |
| i l
| |
| l l
| |
| l 1
| |
| l i /
| |
| l l
| |
| 1 E9 I
| |
| _ _ _ _ _ _ _ _ _ _ _ _ _}}
| |