ML20043G400

Discusses Reduction in Risk from Addition of Hardened Vents in BWR Mark I Reactors.Estimates Obtained for Reduction in Core Damage Frequency Excluded from Consideration Any Possible Changes in Core Damage Frequency from ATWS

ML20043G400
Person / Time
Issue date:	10/19/1989
From:	Sheron B NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:	Thadani A Office of Nuclear Reactor Regulation
References
NUDOCS 9006200242
Download: ML20043G400 (18)

Text

{{#Wiki_filter:{cy i. U

w in nto

= = - [o UNITED sT ATEs - j -g NUCLEAR REGULATORY COMMISSION g 3 y /.-r W ASHING TON, D. C. 20555 6 1. / 007 1 9 1389 1 MEMORANDUM FOR: Ashok C. Thadani, Director } Division of Systems Technology-Office of Nuclear Reactor Regulation FROM: Brian W. Sheron, Director-Division of Systems Research Office of Nuclear Regulatory Research

SUBJECT:

REDUCTION IN RISK FROM THE ADDITION OF HARDENED. VENTS IN BWR MARK I REACTORS This memorandum and its enclosure present the results and analysis for'the reduction in core damage frequency from the addition of hardened containment vents in BWR Mark I reactors. In addition, for Peach Bottom, the estimated population dose to 50 miles and to 1000 miles are presented for an accident of the type averted by the addition of hardened vents. In' order to estimate the reduction in core damage frequency for each Mark I reactor, the reactors were placed in groups based on similarity with respect to the design of the containment cooling systems. These groups and the respective reduction in severe core damage frequency (cdf) were: 1, Reactors without isolation condensers, two RHR heat exchangers per loop, and two.RHR pumps per loop capable.of feeding the heat exchangers reduction.in cdf = 2.3E-5 per year (generic plant) = 3.6E-6 per year (plant-specific initiating event frequency and AC, DC bus configuration for Peach Bottom) 2. Reactors without isolation cond'ensers, one RHR heat exchanger per loop, and two RHR pumps per loop capable of feeding the heat exchangers reduction in cdf - 4.5E-5 per year 3. Reactors without isolation condensers, one RHR heat exchanger per loop, and only one RHR pump per loop capable of feeding the heat exchangers reduction in cdf - 6.3E-5 per year 4. Reactors with isolation condensers reduction in cdf - 1.4E-5 per year 9006200242 891019 h PDR TOPRP EMVGENE C PNU -[(/

7.,

l' r 3 2 for Group 1, values are given specifically for Peach Bottom and for a generic plant. Peach Bottom has lower initiating event frequencies than are given by generic data. Also, it has a better AC and DC bus configuration than the minimum AC and DC bus configuration assumed for the generic plant. The only plant in Group 3 is Hope Creek. Table 5 of the enclosure provides a list of the reactors in each category. There are two important assumptions underlying the:e analyses: 1. The analysis assumes that there are sufficient sources of alternate injection of water into the core such that emergency core cooling can be continued even after containment venting has begun (even for LOCAs). In many plants the pumps which take L suction from the suppression pool may fail after venting has begun. L 2. The analysis also assumes that core melt will occur on loss of I containment cooling and containment failure. This is not necessarily the case. At Peach Bottom, in certain cases, credit L was given for operation of the control rod drive pumps (located in L the turbine building) even if venting failed and the containment L leaked or ruptured. The core damage frequency reduction for Peach Bottom in the table above does not include this effect; giving credit for averting core damage after containment failure would i reduce the benefit of containment venting to 2.4E-6 per year for Peach Bottom. Such considerations are very plant-specific and l would take extensive resources to investigate. The estimates obtained for the reduction in core damage frequency exclude from consideration any possible changes in the core damage frequency from ATWS or station blackout sequences. We now consider the population dose estimates for Peach Bottom, for the type of sequence averted by the addition of hardened vents. These popuiation dose estimates may be used as a basis for estimating the expected population dose for other Mark I reactors by multiplying by the reduction in core melt frequency from the addition of hardened vents, scaling by reactor power, and taking into account differences in population density for the different reactors, perhaps by the use of the Sandia siting study, NVREG/CR-2723. The estimates for Peach Bottom were obtained from Arthur Payne (principal investigator for the Peach Bottom PRA performed in support of NUREG-ll50) of l. Sandia National Laboratories. No sequences with loss of containment cooling and containment failure followed by core damage survived the truncation limit, and consequently the consequences for such sequences were not calculated. A surrogate ATWS sequence was used where the containment fails by drywell rupture either before core damage or during core damage but before vessel breach. Other sequences for which population dose estimates were obtained were ATWS sequences in which the containment fails by wetwell venting bef.gre_ core damage, and sequences in which the containment fails by drywell head rupture, but these sequences are considered less applicable. The most m

E i t M 3 probable means of containment failure is through the drywell; if failure is through the drywell head (to the refueling floor), core damage may'not occur. Also, the estimates of offsite consequences for failure by drywell head rupture or by drywell rupture are quite similar. Our estimates of the population dose are: Population dose to 50 miles: 4.3E6 person-rem Population dose to 1000 miles: 1.7E7 person-rem This analysis was performed by Arthur Buslik of the Probabilistic Risk Analysis Branch, with support from Dale Rasmuson of PRAB. If you have any questions, please feel free to contact Dr. Busiik. (N, AW Brain W. Sheron, Director Division of Systems Research Office of Nuclear Regulatory Research

Enclosure:

As stated cc: W. Beckner J. Kudrick J. Ridgely s 4 mm m-

Enclosure REDUCTION IN SEVERE CORE DAMAGE FREQUENCY FROM INOLUSION OF HARDENED VENTS IN BWR MARK I REACTORS

1.0 INTRODUCTION

In this document estinates are provided on the reduction in severe core danage frequency from the use of hardened vents in BWR Mark I containments. In order to estimate the reduction in core damage f requency f rom the installation of a hardened vent capability we will determine the core damage frequency f rom sequences which require f ailure of containment cooling for core damage, and assume that these sequences will be completely eliminated by the addition of a hardened vent. That is to say, we will assume that the unreliability of the hardened vent is reasonaby low, say 10%. We will not consider any benefit of venting for ATWS sequences or station blackout sequences. The analysis assumes that if there is failure of the containment cooling systems ' (e.g., the power conversion system and the RHR system) in a BWR Mark I reactor without hardened vents then severe core damage will follow. This is not necessarily the case. For example, the NUREG/CR-4550 Peach Bottom PRA gave credit for avoiding core damage in certain cases even if containment cooling were to fail, and the containment failed by leaking or rupture. At Peach Bottom the control rod drive pumps are in the turbine building, and even if the containment were to fail credit was given (in NUREG/CR-4550) for avoiding severe core damage by injecting water into the reactor vessel by the control rod drive pumps; the heat was of course removed from containment through the leak or rupture. One however has no assurance for a BWR reactor that failure of the containment will not fail all core 'njection equipment (either because the equipment is located in the reactor ouilding or because control centers for operation of the equipment are located in the reactor building) and lead to severe core - damage. Thus, we will assume that failure of containment cooling will lead to severe core damage. If the conventional modes of containment cooling fail, but containment venting is available, severe core damage may be averted. Given successful venting, one must be certain that one has the ability to cool the core. Many of the conventional ways of cooling the core take suction from the suppression pool and may fail if the pool becomes saturated. The assumption made here is that there are adequate alternate injection sources, even for large and medium LOCAc; at Peach Bottom, for example, given successful venting one may have the high pressure service water system (HPSW) or'the condensate system available, for LOCAs, and the HPSW, condensato system, and control rod drive pumps available for transients. By w;y of terminology, we will use the term W sequences for sequences which would not be core damage sequences if containment cooling did not fail. These sequences include LOCAs, as well as transients. It is entirely impractical, within the time frame and resources of the present study, to perform a plant specific analysis for eac,h r_eactor. Rather, the approach taken was to group the BWR Mark I reactors into 4 categories, and take, for each category, an existing PRA as a starting point for the analysis. The next section discusses how the grouping was done,

~ ' ' c 2.0 CATEGORIZATION OF THE BWR MARY,- I REACTORS . For the purposes of this study, the BWR Mark I reactors were grouped based on sindlarity with respect to the design of the containment cooling systems. These groups were: 1. Mark 1 reactors with an RHR consisting of two trains, with two pumps por train, and with two heat exchangers (HX) per train. 2. Mark 1 reactors with an RHR consisting of two trains, with two pumps per train, and with one heat exchanger per train. 3. Mark 1 reactors with an RHR consisting of two trains, with one heat exchanger per train, and with one pump per train capable of being aligned to the RHR heat exchanger for that train. 4. Reactors with isolation condensers It would have been desirable to separate the last group into reactors like Nine Mile Point 1, with two redundant loops for their isolation condensers, and reactors like Millstone 1, with only one loop for the isolation condenser. However, because of the unavailability of a PRA for a reactor like Nine Mile Point (or Oyster Creek), this was not possible. For the first group,=the Peach Bottom PRA NUREG/CR-4550 was used as a-starting point. For the second group, the NUREG/CR-4767 shutdown decay heat removal study for the Cooper BWR was used as the starting point. For the third group,

• which contains the Hope Creek reactor as its single member, modifications of-

~ the Cooper PRA were used, and for the fourth group, the Millstone 1 PRA was used. An attempt-was made to be consistent with the Peach Bottom PRA insofar as the-data which was used. Consequently the group 1 analysis will be considered first. 1, 3.0 ANALYSIS FOk THE GROUP 1 TYPE OF REACTOR (2 PUMPS, 2RXs, PER RHR LOOP) 3.1 Estimate in the Risk Manacement Implications Report As mentioned above, the NUREG/CR-4550 Peach Bottom study, done in support of - NUREG-ll50, was used as the starting point of the analysis. The " Risk Management Implications of NUREG-ll50" report (NUREG/CR-5263) was also consulted. This report estimated the benefit of venting for the Peach Bottom plant. However, there were-certain deficiencies in the analysis. In particular, i 1. Peach Bottom already has a hardened vent system, and credit for this system was given in the NUREG/CR-4550 PRA. None of the sequences in which venting was an applicable recovery action survived the truncation limit of lE-8 per year. Hence there was no completely straightforward way of estimating the benefit of venting at Peach Bottom without redoing the calculations. What was done in the Risk Management Implications report was to take the upper bound on the (point) estimate of each truncated sequence in which venting was a possible recovery action, and take this upper bound as the point estimate. In many cases this upper bound was the truncation limit of IE-8 per year, but in other cases the

(l estimate was lower than lE-8 per year, because, in the truncation analysis, the sequence was tracked until the sequence fell below lE-8, and a record of this value was kept. The use of the upper bound introduced an overestimate in the benefit of venting in the Risk Managenent Implications report. 2. The Risk Management Implications report, in estimating the benefit of venting, took out the recovery 1 action of venting, but added recovery of the power conversion system (PCS)' as a means of containment heat removal. However, in the original NUREG/CR-4550 analysio credit for recovery of the PCS was taken when necessary to bring a sequence f requency below the truncation lindt of lE-8 per year. Hence, there was a double counting of recovery in some sequences, in the Risk Management Implications report. Since NUREG/CR-4550 did not for each sequence specify whether recovery of the PCS.was given credit for in reducing the sequence frequency below the truncation limit, it would be very dif ficult to correct this double counting. This effect introduced an underestimate in the benefit of venting in the Risk Management Implications report. 3. The Risk Managenent Implications report correctly noted that recovery of the Power Conversion System is not a valid recovery action for medium or large LOCAs, because of the heat added to the containment through the break. However, the Risk Managenent Implications report considered recovery of the Power Conversion System as a valid recovery action for transients which lead to a medium or large LOCA by the sticking open of two or three relief valves, and this appears to be incorrect. In addition to the above possible difficulties with the use of the results of the Risk Management implications report, there is another difficulty: the lack of homogeneity of the plants in this category, so that Peach Bottom may ~ not' adequately represent all plants in this category

• In particular, the frequency of losses of the power conversion system, either directly or consequentially by loss of the main feedwater syster, is less at Peach Bottom than in a generic plant, as will be seen below. Moreover, the AC and DC distribution systems at Peach Bottom are perhaps better than in other plants in this category. The approach taken below will be to estimate the reduction in core damage frequency from the hardened vent system at Peach Bottom, and to also estimate the reduction in core damage, frequency from the installation of hardened vents at a generic plant in this category. The generic plant will have generic initiating frequencies and will have a minimum AC and DC system, to be described below.

It is recognized that any particular plant in this category may have a better AC and/or DC system, but it is impossible within the time frame and resources of the present study to take these differences into account. The assumption made in the NUREG/CR-4550 Peach Bottom PRA that loss of a safety-related 4160V AC or 125V DC bus will cause loss of the PCS will be made here, for all plants considered. This assumption may be false for a given plant if there is adequate separation between safety and non-safety related loads. 3.2 Estimates Used in the Present Work In the present work, the dominant cut sets for the W sequences were identified by inspection of the data, by looking at the dominant cut sets for other PRAs, and by inspection of the f ault trees and the dependency diagrane for Peach Bottom in NUREG/CR-4550. The following observations were made: .. - -- l wa

h 1. To lose conventional containment cooling it is necessary to fail all applicable modes of RHR containment cooling (the suppression pool cooling mode, the shutdown cooling mode, and the containment spray mode for transier.tsi the suppression pool cooling mode and containment-spray modes for LOCAs). One can focus therefore on RHR system failures which fail all applicable modes of RRR, for the given initiator. We note further that the success criterion for RHR containment cooling requires only 1 RHR pump. 2. Inspection of the data and the fault trees shows that if electric power is available to both trains of the RHR, then the dominant mode of failure of the RHR is common mode failure of the RHR pumps. Second in importance is common. mode failure of normally closed motor operated valves in the High Pressuro Service Water (HPSW) system. Third in importance is common mode failure of the HPSW pumps. (These are standby pumps at Peach Bottom. ) According to NUREG/CR-4550, Vol. 1, the common mode failure probability of four RHR pumps is 3E-4 (beta factor of .1). The common mode f ailure probability of three RHR pumps is 3.3E-4 per demand. The common mode failure probability of two RHR pumps is 4.5E-4 per demand. 3.. Because there are two heat exchangers per train, and two RHR pumps per train, the maintenance unavailability of a single train is smaller than it would otherwise be. However, there is appreciable maintenance unavailability condng from the fact that maintenance on valves MV26A or MV39A in the RHR system requires blocking flow from both heat exchangers in a loop. It will turn out that this is important for the generic plant in this category, with a minimum AC and DC system, but not for Peach Bottom itself. 4. Peach Bottom 2/3 has four 4160 VAC busse 6, each of which supplies power to both unit 2 and unit 3; Similarly, there are four -125 VDC busses, shared between the two units. There are two loops of RHR, loop A and loop B. Loop A of the RHR contains pumps A and " C; pump A is fed by the A busses of AC and DC, and pump C by the C e busses. Hence, loss of a single bus would not fail both pumps in loop A of the RHR. Loss of bus C would fail the ability to make the proper valve alignments for use of loop A of the RHR in the suppression pool cooling mode or the containment spray mode; it would not however fail the shuLdown cooling mode for loop A, unless there were an independent failure. Loss of bus A would fail the shutdown cooling mode, but not the other two modes'of RHR cooling, provided pump C were available. A symmetric situation exists for loop B of the RHR, which is fed by busses B and D. Thus, in general, at Peach Bottom, loss of a single bus will not fail an RHR loop in all three possible RHR modes, without an additional single failure, However, at another plant in this category with only two safety-related 4160V AC busses, and only two safety-related DC busses, the situation would be different. Loss of bus A (either AC or DC) would fail loop A of the RHR, and loss of bus B would fail loop B. As noted earlier,another assumption of importance is that the loss of the bus causes loss of the power conversion system.

3-t e s At. Peach Bottom, because of the AC and DC power system arrangement, loss of a single AC or DC bus is not much different than any;other means of losing the power conversion system. One RHR pump'will not be available, but the common mode failure of three RHR pumps (3.3E-4 per demand) is almost the same.as for four-RHR pumps (3. 0E per demand), so there is not a big difference here. One does however have to consider any possible dif ferences in the probability of nonrecovery in the tine available for recovery. The time available for recovery of the power conversion system is the time to containment failure from overpressure, in the transient (as opposed to LOCA) case. This time, at Peach Bottom, is about 40 hours. (The nondnal-failure pressure of the containment is 148 psig, and the containment pressure tine trace for a TW sequence, as given in Figure A-4 of NUREG/CR-4550, vol. 4, shows failure at about 38 hours.) For a loss of PCS in general, the NUREG/CR-4550 study used about.01 for the nonrecovery probability. Losses of busses can have considerably long repair times, on the other hand. For example, if the loss of bus is caused by a loss of a conductor (cable and associated joints), then IEEE std 500-1984, on p. 742 gives a nominal out-of-service tine of 445 hours. Even taking into account that out-of-service times include delays before starting repair, and that emergency types of repairs might be made if it were necessary, the repair times can be quite long. Of course, if the loss of a bus is a result of a circuit breaker failure repair times can be much shorter. Nevertheless it appears not unreasonable to take a value of.25 for the probability of nonrecovery of an AC or DC bus in the time available. In the NUREG/CR-4550 study, the probability of nonrecovery of the PCS for a loss of AC or DC bus was taken as the same as for any other loss of PCS (private communication, Alan Kolaczkowski to A. Buslik). The reason is that only one of four AC busses (or one of four DC busses)' was* lost, and, even if the bus could not be restored, the fault could be isolated and enough of the PCS restored to obtain success for the containment heat removal function. e Discussion of Results for Peach Bottom Table 1 gives the results for the Peach Bottom plant, using initiating event f requencies which are plant-specific to the Peach Bottom plant, and using plant-specific inforumtion on the AC and DC bus configuration. One sees that the W sequences are estimated to contribute about 3.6E-6 per year to the core damage frequency. This is therefore the benefit from the use of hardened vents, under the assumptions of this study. This e stinate assumes that f ailure of the conventional containment cooling systens leads to core damage in the absence of hardened vents. However, in the NUREG/CR-4450 PRA -for Peach Bottom, credit was given for avoiding core demage even if the containment leaked or ruptured, since one expects at least one control rod drive pump to be available (even on loss of an AC bus initiator). The control rod drive pumps are in the turbine building; although valves for the control rod drive system are in the reactor building I learned from Arthur Payne of Sandia that these valves fail safe on loss of air, and that a spurious closing signal would have to exist for more than two hours (after containment failure) for core damage to occur. Hence failure of the control rod drive pumps from harsh environments in the reactor building was i =. i ' considered unlikely, in comparison to other failure modes. Thus the actual . benefit of venting at Peach Bottom is less than 3.6E-6 per year, if one gives credit for avoiding core damage after containment failure. However, the f control rod drive pumps cannot supply sufficient water if one has an intermediate.or large LOCA. Hence the sequences involving a trip followed by two or three stuck open relief valves still remain, and, if one were to take credit for avoiding core damage af ter containment f ailure, the reduction in core damage frequency at Peach Bottom from the addition of hardened vents nould be about 2'.4E-6 per year. The detailed results obtained were the following, from Table 1: 1. W sequences initiated by a loss of offsite power contributed 5.5E-7 per year to the severe core damage f requency. The frequency of loss of offsite. power (LOSP) used was 1/yr, and the probability of non-recovery in the time allotted (LOSPNR13HR) was estimated at.01. (The probability of nonrecovery of of fsite power does not change much af ter 13 hours.) The same common cause failures of the RHR as appeared in, say, a loss of PCS transient were important here. Failure of a diesel generator combined with failure of a single train of the RHR was less important. The common cause failures of the RHR were common cause failures of the RHR pumps (RHR-CCF-LF-MDPS), common cause failure of high pressure service water motor-operated valves to open (HSW-CCF-LF-MOVS), and common cause failure of high pressure service water pumps (HSW-CCF-LF-MDPS) to start. 2. W sequences initiated by either a loss of the PCS (T2), or the loss of main feedwater (T3B), or the loss of an AC or DC bus, contributed 6.4E-7 per year to the severe core damage f requency. (T3B is assumed to lead directly to T2; also loss of an AC or DC bus was lumped together with other losses of the PCS, because of tne minimal impact that loss of a single

• bus had on the RHR unavailability.)

s 3. Large LOCAs (A) and intermediate LOCAs (SS1) were not important contributors. However, transients leading to an intermediate LOCA by causing two relief valves to stick open had a dominant contribution of 2.2E-6 per year. The probability of two relief valves sticking open, given a transient,-was denoted by PP2 in Table 1, and has the value.002 per demand in the NUREG/CR-4550 data base. The PCS is assumed lost on an intermediate or large l LOCA. l l fstimates for a Generic Plant in This Catecorv The plant specific initiating event f requencies for Peach Bottom dif fer f rom I the generic initiating event frequencies as given in Table 8.2-4 of NUREG/CR-4550, volume 1 Rev. 1, the internal events methodology report for the analysis of core damage frequency. In particular the plant specific value for the frequencies of losses of main feedwater is.06/ year, while the generic -value for BWRs is.6/yr. The plant specific value for the f requency of loss l of the power conversion system (not indirectly caused by loss of main l feedwater) was.05 per year, while the generic value was 1.7/ year, i Apart from differences in the initiating event frequencies between Peach __ l. L

= - 80ttom and the generic plant in this category, the other difference is in the AC and DC bus configuration. For the generic plant, two 4160V ac busses and two 125v DC busses are assumed, one bus of AC and of DC feeding each loop of .the RHR. Loss of an AC or DC bus fails one loop of the RHR. The important cut sets for loss of a bus initiator are the loss of the bus combined with common cause failure of the motor-operated discharge valves on the high pressure service water side of the RHR heat exchangers in the opposite loop, or combined with common cause f ailure of the RHR pumps in the opposite loop, or combined with the opposite loop being unavailable because of test and maintenance. A probability of nonrecovery of a bus of.25 was assumed. The results are given in Table 2. The sum total of the W sequences is 2.3E-5 per year, which is therefore our estimate of the reduction of core damage frequency from the addition of hardened vents for a generic plant in this category. In contrast, the core damage frequency reduction was 3.6E-6 per year for Peach Bottom. For the generic plant in this category, the W sequences with the loss of PCS initiator (The T2T3B sequences in the table) are more important than with Peach Bottom plant-specific data; the frequency of these sequences increased f rom 4.7E-7 per year to 9.8E-6 per year, a factor of 21. For the generic plant, these T2T3B sequences were the most important. The next most important W sequences were initiated by loss of either an AC bus or a DC bus. There are four different busses (two AC and two DC) which cause loss of one of the two loops of RHR. Quantification can be accomplished by estimating the core damage frequency from the loss of only of these busses and multiplying by four. The multiplication by four can be accomplished by using an initiating event freguncy equal to four times the initiating event frequency for loss of a single bus. The frequency of loss of an AC or DC bus is SE-3/yr, according to the data base in NUREG/CR-4550. (Generic estimates wero used for these initiating event frequencies in the NUREG/CR-4550 Peach Bottom PRA.) A non-recovery factor of.25 was assumsd,- The effective initiating event frequency is the product of the number of busses (4), times the initiating event frequency per bus (SE-3/yr), tines the non-recovery s - ~ factor (.25). Thus the effective initiating event frequency is SE-3/yr. These loss of bus sequences contributed 8.4E-6 per year to the core damage f requency (given no hardened vents,. and given that core damage occurs on containment failure). The most important cut set consists of the initiator, which fails one train of the RHR, combined with test and maintenance on either valve MV26A cr MV39A in the RHR system (RHR-LOOP-UTM); when maintenance is performed on these valves both heat exchangers in the associated loop are made unavailable. 4.0 ANALYSIS FOR THE GROUP 2 (COOPER) TYPE OF PLANT For this group of plants the starting point was the Shutdown Decay Heat Removal Study report NUREG/CR-4767 (A-45 study). However an attempt was made' to be consistent in analysis assumptions with the analysis for Peach Bottom, including the generic data used and the treatment of recovery. The principal changes to the A-45 study weze: 1. The A-45 study used a value of.5 per year for loss of main feedwater frequency (T3D), and did not consider any other way of losing the power conversion system. Thus there was an increase in

i I the initiating event f requency for the loss of PCS in the present study over what was used in the A-45 studies. The initiator T2T3B=T2+T3B has a frequency of 2.3 per year in the present study as opposed to the value of.5 per year in the A-45 h?udy. 2. The A-45 study did not give credit for recovery of loss of offsite power in the W sequences. 3. The probability of nonrecovery of the power conversion system was assumed. to be.16 in 24 hours in the A-45 study, while it was.01 in the NUREG/CR-4550 study. 4. The component data, and the data for common cause failures, was made consistent with the NUREG/CR-4550 generic data. 5. Certain common cause f ailures were included, which were not included in the A-45 study. In particular, cort..on mode failure of the service water outlet valves for heat exchanger A (SWS-652) and for heat exchanger B (SWS-653) was included. The joint failure of these valves was included in the A-45 study, but the failures were treated as if they were independent. 6. Loss of an AC or DC bus coupled with failure of the service water outlet valve for the heat exchanger in the opposite RHR loop appears to be a valid cutset, but was apparently not included in the A-45 study. This cutset+was included in the present study. A cutset which was included in the A-45 study which was not included in the present study consists of loss of an AC or DC bus coupled with = a service water inlet valve for the heat exchanger in the opposite loop failing closed (manual valve, normally open, fails closed.) The reason for not including this latter cutset was its lower probabili.ty. 7. The A-45 study did not consider transients with two or three stuck open relief valves, while the present study did. s The: result obtained was 4.5E-5 per year for the W -sequences, so that this is the benefit f rom venting for the plants in this group (Cooper group). More detailed results are given in Table 3. .The major difference between the generic plant of category 2 (2RHR HXs, 2 RHR pumps), and the Cooper type plant comes f rom the loss of an AC or DC bus initiator. The maintenance unavailability is somewhat higher for the Cooper type configuration, and more importently a single f ailure of the service water system outlet valve for a heat exchanger coupled with the loss of the bus is a valid cutset, while common mode failure of two such valves was required in the Peach Bottom plant, because thote are two heat exchangers per loop in that plant. 4.0 ANALYSIS FOR THE GROUP 3 (HOPE CREEK) TYPE OF REACTOR The pertinent design difference between this group of plants (Hope Creek is the only member) and the Cooper type of plant is that there is only one RHR pump per RHR train which can be aligned to the RHR heat exchanger for that train. Thus, on loss of an AC or DC bus, failure of a single RHR pump fails conventional containment cooling, while common modo failure of two pumps is required for the groups 1 and 2 plants.

[ i The not benefit from a hardened vent is here calculat'ed as 6.3E-5 per year. The detailed results are given in Table 4. 5.0 ANALYSIS FOR THE GROUP 4 (ISOLATION CONDENSER PLANTS) TYPE OF REACTOR Here, all that was done was to use the results for the core danage sequences in Amendment 2 of the Millstone 1 PRA performed by Northeast Utilities. Almost all of the intermediate and late core melts would be eliminated by hardened vents, provided credit was given for an alternate injection source. The only exception was a core melt occurring at an intermediate time (not early, not late) with a small_ break initiator, in which the operator failed to maintain reactor pressure vessel level. This sequence was combined with one in which the containment cooling fails, but it was possible to identify the individual" contributions. Of the 4.6E-6 per year for this particular sequence, 2.lE-6 per year came from failure of containment cooling. In addition, the small break LOCA f requency was only lE-3 per year, whereas the NUREG/CR-4450 data base uses 3E-3 per year. Correcting for this as well, one obtains an estinate cf 1.4E-5 per year for the benefit of venting. Because of time and resource limitations no further attempt was made to modify the Millstone 1 PRA data to be more consistent with the NUREG/CR-4550 data. This estimate of 1.4E-5 per year w'as also applied to the plants, such as Nine Mile Point 1, which have two redundant loops of isolation condensers. No PRA is available for such plants and considerable effort would be required to obtain a better estimate. Despite the two redundant loops for the isolation condensers, the plants may be no better than Millstone 1 and may be worse. For example, on a small LOCA the isolation condensers are not considered adequate for the removal of heat f rom containment. If this is also true for e transients with one stuck cpen relief valve, and if the only system which can give success for the containment heat removal function on such a transient is the containment spray systum, then, using generic data, one could make a rough estimate of the contribution of these sequences to core damage at Nine Mile Point as: Transient Frequency (Loss of MFW/PCS) : 2.3/yr / Stuck open Safety / Relief Valve: .016 (from NUREG/CR-4550, vol.1, Table 8.2-5 ) Failure of containment spray system: 4E-4 Sequence Frequency: 1.5E-5/yr If there were alternate core injection sources this sequence could be eliminated by the use of a hardened vent from containment. Additional sequences such as small break LOCAs and inadvertant opening of relief valves must be added. The extremely rough estimate of the failure of containment spray in the above sequence estimate comes primarily from common mode failure of the four containment spray pumps, 6.0

SUMMARY

OF RESULTS AND ASSIGNMENT OF PLANTS TO GROUPS Table 5 summarizes the results and indicates which plant belongs to each group. s _g_

iih *~ t: s TABLE '1. PLANT SPECIFIC RESULTS F0H PEACH BOTTOM--W SEQUENCES DATA INITIATING EVENTS. FREQ. PER YEAR Basic Events PP2 0.002' LOSP 0.1 HsW-CCF-LF-MOVS 9.60E-05 ] T-AC 0.04 RHR-CCF-LF-MDPS 3.00E-04 { T2T3B 0.11 HSW-CCF-LF-MDPS 2.90E-05 5 0.003 LOSPNR13HR 1.30E SS1 0.0003 PCSNR13RR 1.00E-02. A 0.0001 HSW-CCF-LF-2HOV 1.47E-04 i T3A 2.5 HSW.CCF-LF-2MDP 7.80E-05 l RHR-LOOP-UTM 1.00E-03 } TX 1.50E-01 RHR CCF-LF-2MDP 4.50E-04 l l T AC IS TRIP FROM LOSS OF AC BUS OR DC BUS. ADDED TO T2T3B FOR PEACH DOTTOM PLANT SPECIFIC T2T3B IS EITHER LOSS OF PCS OR OF MFW LEADINO TO LOSS OF PCS 5 IS A SMALL lhCA SS1 IS AN S1 thCA ] A IS A LARCE lhCA j T3A IS FREQ OF REACTOR TRIPS NOT LEADING TO LOSS OF PCS j PP2 IS AN S1 IhCA FROM TWO STUCK OPEN RELIEF VALVES (CONDITIONAL ON A TRANSIENT) i TX=T2T3B+T-AC SEQUENCES FOR PEACH BOTTOM PLANT SPECIFIC i 1.25E-07 +SLOSP'4HSW-CCF-LF-MOVS*$LOSPNR13RR

3. 90E-07 + 4LOSP' $RHR-CCF-LF MDPS * $LOSPNR13HR j

3. 77E-08 + $LOSP' $HSW-CCF-LF-MDPS* $LOSPNR13HR' 1

.5.53E-07 )

1. 4 4E-07 ($TX) * $ HSW-CCF-LF-MOVS'4PCSNR13HR l

I 4.50E-07 ( $TX) * $ RHR-CCF-LF-MDPS' $PCSNR13 HR 4.35E-08 ( $ TX ) * $ HSW-CCF-LF-MDPS' $ PCSNR13 HR { 6.37E-07 l v

9. 60E-09 +4A'8HSW-CCF LF-HOVS

] 3.00E-08 +$A*$RHR-CCF-LF-MDPS

2. 90E-0 9 + $A' $ HSW-CCF-LF-MDPS 4.25E-08 i

2.88E-08 +$SS188HSW-CCF-LF-HOVS 9.00E-08 + $S51*$RHR-CCF-LF-MDPS 8.70E-09 +$ss1*$HSW-CCF-LF-MDPS 1.27E-07

5. 01E-07 ($T3A+ $T2T3B) * $PP2 * $HSW-CCF-LF-MOVS d

h

1. 57E-06 ($73A+ $T2T3B) * $PP 2' $RHR-CCF-LF-MDPS 1.51E-07 ( $ T3 A + $ T2T3Bl * $PP2' $HSW-CCF-LF-MDP S 2.22F-06 3.58E-06 -SUM OF ALL ABOVE SEQUENCES d

i i o a u l l f a ~ TABLE 2.' ' RESULTS FOR A CENERIC PLANT WITH 2 kHR HXs AND 2 RHR PUMPS PER RHR LDOP W SEQUENCES DATA 1.E.*S Basic Events LOSP 0.1 HSW-CCF-Lr-MOVS 9.60E-05 T-AC 0.005 RHR CCF-LF-MDPS 3.00E-04 T2T3B. 2.3 HSW-CCF-LF-MDPS 2.90E-05 l S 0.003 LOSPNR13HR 1.30E-02 881 0.0003 PCSNR13HR 1.00E-02 A 0.0001 HSW-CCF-LF-2MOV I.47E-04 T3A 2.5 HSW-CCF-LF-2MDP 7.80E-05 PP2 0.002 RRR-LOOP-UTM 1.00E-03 l RHR-CCF-LF-2MDP 4.50E-04 ? T-AC IS TRIP FROM LOSS OF AC BUS OR DC BUS T2T3B IS EITHER LOSS OF PCS OR OF MFW LEADING TO LOSS OF PCS S IS A SKALL LOCA j SSI IS AN $1 LOCA. L A IS A LARGE !OCA T3A IS FREQ OF REACTOR TRIPS NOT LEADING TO LOSS OF PCS PP2 IS AN S1 LOCA FROM TWO STUCK CPEN RELIEF VALVES (CONDITIONAL ON A TRMISIENT) SEQUENCES FOR PEACH BOTTOM TYPE P! ANT

1. 25E-07 + $LOSP * $HSW-CCF-LF-MOVS * $LOSPNR13HR

3. 90E- 0 7 + $ LOSP * $ RHR-CCF-LF-MDPS * $ LOSPNR13HR

) - 3. 7 7E-08 + $LOSP * $HSW-CCF-LF-MDPS * $LOSPNR13HR 5.53F-07 2.21E-0 6 + ST2T3B* $HSW-CCF-LF-M3VS* $PCSNR13HR

6. 90E-06 + $T2T3B* $RHR-CCF-LF-MDPS * $ PCSNR13HR l

6. 67E-07 + $T2T38* $HSW-CCF-LF-MDPS* $PCSNR13HR

( 9.77E-06 . 7.3 *F-07. + $T-AC* $HSW-CCF-LF-2MOV l

2. 2 5E-0 6 + $T-AC* $ RRR-CCF-LF-MDPS j

3. 90E-07 + $T-AC* $HSW-CCF-LF-2MDP 5.00E-06 + $T-AC' $RHR-LOOP-UTM 9,60,E-09 +4A*$HfW-CCF-LF-MOVS-3.00E-Ob +.SA*0RHR-CCF-LF-MDPS

2. 90E-0 9 + $A* $HSW-CCF-LF-MDP3 4.25E-08 1

)

2. 8 8E-08 + $SS1 * $HSW-CCF-LF-MOVF i

9. 00E-0 8 + $ $ $1 * $ RHR-CCF-LF-KDPS i

8.70E-09 +$ss1* $HSW-CCF-LF-MDPS l 1.27E-07 9.22E-07 ' ($T3A+ $T2T3B) * $PP2 * $ HSW-CCT-LF-MOVS

2. 8 8E- 0 6 (f T3 A+ S T2T38) * *PP2 * $RHR-CCF-LF-MDPS

2. 7 BE-07 ( $T3A * $ T2T3B) * $PP2 8 4hSW-CCF-LF-MDPS 4.08E-06 2.30E-05 aSUM OF ALL ABOVE SEQUENCES e

1 _. __. i

g. h 3 ?^ 1 h. 'g-5 i g' ns ' TABLE 3. - RISVLTS FOR COOPER TYPE PLANT--W SEQUENCES l(IE, FREQUENCY IS CENERIC, T2T3B FROM :NUREG/CR-4550, VOL.1). p- ' DATA INITIATING EVENTS (FREQ. PER YEAR) Basic Events LOSP 0.1

SWS-CCF-LF MOVs.9.60E-05 T-AC 0.005 RHR-CCF-LF-MDPS 3.00E+04

' T2738 2.3 SWS. CCF-LF-MDP S 2.90E-05 S 0.003 LOSPNR13HR 1.30E-02 $$1 0.0003 PCSNR13HR-1.00E-02 s A 0.0001 SWS-CCF-LF-2MOV 1.47E-04 T3A 4.67 SWS-CCP-LF-2MDP 7.80E-051 - PP2 0.002 SWS-652-VCC-LF 3.00E RHR LOOP 2-UTM-2.00E-03 RHR-CCF-LF-2MDP ~ 4.50E-04 p ~ ~ - T-AC IS TRIP FROM LOSS 0F AC BOS OR DC BUS T2T3B IS EITHER LOSS OF PCS OR OF MFW LEADINO To LOSS OF PCS > $ IS A SHALL LOCAL SS1 IS AN S1 LOCA t A IS A LARGE LOCA T3A IS FREQ OF REACTOR TRIPS NOT LEADING TO LOSS OF PCS PP2 IS ~ AN 81 LOCA.FROM TWO STUCK OPEN RELIEF VALVES - (CONDITIONAL ON A TRANSIENT) SEQUENCES,FOR COOPER TYPE PLANT

1. 91E-07 + $LOSP* $5WS-CCF-LF-2MOV* $ LOSPNR13HR l

^3.90E*07 4$LOSP'4RHR-CCF-LF-MDPS'$LOSPNR13HR 5.81E-07 -l - 3.38E-06 + $72T3B'$$WS-CCF-LF-2MOV* $PCSNR13HR

6. 90E-0 6 + $T273B* $RHR-CCF-LF-MDPS* $PCSNR13HR 1.03E-05 t

! 1.00E-05 + $T-Ac* $RHR-LOOP 2-UTM s 1.50E-05 +$T-AC'$$WS-652-VCC-LF-

2. 25E-0 6 + $f-AC' $ RHR-CCF-LF-2MDP.'

.2.73E-05

1. 4 7E-08 + SA* $sWS-CCF-LF-2MOV 3.00E-08 + SA* $RHR-CCP-LF MDPS 4.47E-08

. j 'l '4.41E-08 +$$$1*$5WS-CCF-LF-2MOV [- 9.00E-08 +$58188RHR-CCF-LF-MOPS 1.34E-07. .' 2.05E-0 6 ($T3A+ $T213B) * $PP2* $5WS-CCF-LF-2MOV - 4.10E-0 6 ' ($T3A+ $T2T3B) * $PP2* $RRR-CCF-LF-MDPS 6.23E-06 4.45E-05 -SUN OF ALL ABOVE SEQUENCES .J 3 L 1 i ~ ]

'l l

l 'l i 'I' + t i

-(C o 4 i f -_ * :

4 A

' TABLE 4.- RESULTS FOR A HOPE CREEK TYPE PLANT -W SEQUENCES (IE FREQUENCY IS OENER3C. T2T3B FROM NUREG/CR-4550, VOL.1) ~ ' DATA: 1 INITIATING EVENTS (FREQ. PER YEAA)'- Basic Events .,LOSP-0.1 SWS-CCF-LF-MOVS 9.60E-05 [ T-AC 0.005-RHR-CCr *?-2MDP 4.50E-04 T273B 2.3 -SWP-CCF-LJ-MDPS 2.90E-05 S.- 0.003 LOSPNR13HR 1.30E-02 S$1 0.0003 PCSNR13HR '1.00E-02 l A 0.0001 SWS-CCF-LF- !MOV-1.47E-04 T3A 4.67-SWS-CCF-LF.2MDP 7.80E-05 SWS-652-VC.-LF -3.00E-03~ 'i RHR-LOOP

• UTM 2.00E-03 RHR-FT*-LF-MDP 3.00E-03 PP2 0.002 T-AC IS TRIP FROM LOSS OF AC DUS OR DC BUS T273B IS EITHER LOSS OF PCS OR OF MFN LEADING TO LOSS VF PCS -

S IS A SMALL LOCA 551 IS AN $1 LOCA A IS A LARGE D3CA - T3A IS FREQ OF REACTOR TRIPS NOT LEADING TO LOSS OF PCS ' . PP2 IS AN 81 LOCA FROM TWO STUCK OPEN RELIEF VALVES (CONDITIONAL ON A TRANSIENT) SEQUENCES FOR HOPE CREEK TYPE PLANT

1. 91E-07 + $LOSP* $SNS-CCF-LF-2MOV* $LOSPNR13HR

5. 8 5E-07 + $LOSP * $RHR-CCF-L?-2MDP* $LOSPNR13RR l

7.76E-07 3.38E-06 +$T2T3B* $SWS-CCF-LF-2MOV* $PCSNR13HR 1.03E-05 +$T273B* $RHR-CCF-LF-2MDP* $PCSNR13HR 1.37E-05 1.002-05 +$T-AC'$RHA-LOOP 2-UTHI 1.50E-05 +$f-AC**SWS-652=VCC-LF 1.50E-05 +ST-AC'$RHR-FTS-LF-MDP 4.00E-05 ' 1. 4 7E-08 + $A* $5WS-CCF-LF-2MOV p L'

4. 50E-08 + $A* $RRR-CCF-LF-2MDP 5.97E-08

4. 41E-08 + $$518 8 8WS-CCF-LF-2MOV 1.35E-07 +$SS1*$RHR-CCF-LF-2MDP 2.05E-06 ($T3A + $T2T3B) * $PP2* $SWS CCF-LF-2MOV N

i

6. 27E-06 (ST3A+ $ T2 T3B) * $PP2 * $RHR-CCF-LF-2MDP 8.32E-06 6.31E-05 -SUM OF ALL ABOVE SEQUENCES t

M4 mm L n

N ~ l ABLE 5.

SUMMARY

REDUCTION IN CORE DAMAGE PREQUEN0Y FROM THE USE OF HARDENED VENTS i

NOTE: DELTA CDF=THE REDUCTION IN CORE DAMAGE FREQUENCY FROM THE USE OF HARDENED VENTS i A. PLANTS WITH RHR AND NO ISOLATION CONDENSERS c l 1. PLANTS WITH 2 HEAT EXCHANGERS PER RHR LOOP AND TWO PUMPS PER RHR LOOP (Poach Bottom type) Browns Ferry 1,2,3 Peach Bottom 2,3 Vermont Yankee Pilgrim DELTA CDP =2.3E-5 PER YEAR, generic Initiating Event frequency =3.6E-6 PER YEAR, for Peach Bottom, using plant specific 1 I.E. frequency and AC,DC bus distribution information-2. PLANTS WITH 1 HEAT EXCHANGER PER RHR LOOP AND TWO PUMPS PER RRR LOOP (Cooper type) Hatch 1,2 Montecello Fitzpatrick Quad Cities 1,2 Cooper Duane Arnold Brunswick 1,2 Fermi 2 ' DELTA CDF=4. 5E-5 PER YEAR, GENERIC I.E. FREQUENCY l 3. PLANTS WITH 1 HEAT EXCHANGER PER RHR LOOP AND ONE PUMP PER RHR LOOP WHICH CAN BE ALIGNED TO THE HEAT EXCHANGER Hope Creek DELTA CDF=6.3E-5 PER YEAR, GENERIC I.E. FREQUENCY 14 a 0

y-wwo;; 7,p..gr. m.r ~-. - = - - --- - - - - - - -, - - -- - - -. - - -., -s3< s ,...j_ j- ? k: ' c s , o- ( ' >;/f- ,1 pH90, w s

p

') o

m..

TABLE 5..(CONTINUED) _

L r

r -ff p t j t. Bi_ PLANTS 'WITH ISOLATION CONDENSERS. -(Group; 4 plants) w.

t n-l'

.1. Plants with two loops'for isolation'conde7sers .{ v -. 4 Nine' Mile Point * ~ ( oyster. Creek 1'-: M 2. Plants with ono. loop for isolation-condenser L Dresden 2"and 3-t; Millstone.1, 3 ={ 1 .e 3 1 g =For-these. plants we will use an estimate based on'the Millstone 1 PRA .j + performed by Northeast Utilities. This-estimate is: DELTA CDF=1.4E-5 PER YEAR'. + =r 0 ..'t ..y .f j t.:, - si. 1 E s S,'? ' = I -.( p 2 +i l j. -{ s s 4 I -( 'I l-1 i )f 'i G r >t l c

• • en6 ms I

g -]l-t n 4 e y e m- ..-nv ,-,-:m ~ n-w s -,-enm --}}