ML20214A035

Forwards Remaining Response to 861008 & 23 Requests for Addl Info Re Emergency Planning Sensitivity Study.Several Previous Responses to Requests for Addl Info,W/Minor Editorial Changes & Corrections,Resubmitted

ML20214A035
Person / Time
Site:	Seabrook
Issue date:	11/17/1986
From:	Devincentis J PUBLIC SERVICE CO. OF NEW HAMPSHIRE
To:	Long S Office of Nuclear Reactor Regulation
References
SBN-1234, NUDOCS 8611190171
Download: ML20214A035 (59)

Text

r

~

l SEABROOK STATION Engineering Office

.) c-LJ

_3 Pub 5c Service of New Hampshire November 17, 1986 SBN-1234 New Hampshire Yankee Division T.F.

B7.1.2 United States Nuclear Regulatory Commission Washington, DC 20555 Attention:

Mr. Steven M. Long, Project Manager PWR Project Directorate No. 5 Division of PWR Licensing - A

References:

(a) Facility Operating License NPF-56, Construction Permit CPPR-136, Docket Nos. 50-443 and 50-444 (b) USNRC Letter, dated October 8, 1986, " Request for Additional Information for Seabrook Station, Units 1 and 2, Emergency Planning Sensitivity Study", S. M. Long to R. J. Harrison (c) USNRC Letter, dated October 2 3, 1986, "Reques t for Additional Information for Seabrook Station, Units I and 2, Emergency Planning Sensitivity Study", S. M. Long to R. J. Harrison (d) PSNH Letter (SBN-1225), dated October 31, 1986,

" Response to Request for Additional Information (RAIs)", J. DeVincentis to S. M. Long (e) PSNil Letter (SBN-1227), dated November 7, 1986,

" Response to Request for Additional Information (RAIs)", J. DeVincentis to S. M. Long

Subject:

Response to Request for Additional Information (RAIs)

Dear Sir:

Enclosed herewith are the remaining responses to the Requests for Additional Information forwarded in References (b) and (c). Previous responses were oubmitted in References (d) and (e).

In addition, we are resubmitting several previous responses to RAIs with minor editorial changes and corrections. Attachment A identifies the remaining responses that are included in this transmittal and the responses that are being resubmitted. The responses are provided in Attachment B.

8611190171 861117 r

PDR ADOCK 05000443 9

P PDR d

ho Il Seabrook Station Construction Field Office. P.O. Box 700. Seabrook, NH O3874

United States Nuclear Regulatory Commission SBN-Attention:

Mr. Steven M. Long Page 2 Very truly yours, j' m Att n John DeVincentis Director of Engineering Attachment cc:

Atomic Safety and Licensing Board Service List Director, Office of Inspection and Enforcement United States Nuclear Regulatory Commission Washington, DC 20555 r

I i

i t

l 1

Diana Curran, Esquire Pater J. Mathews, Mayor Hsrmon & Weiss City Htll 2001 S. Streat, N.W.

Newburypo rt, MA 01950 Suite 430 Washington, D.C.

20009 Judith H. Mizner Silvergate, Gertner, Baker, Sherwin E. Turk, Esquire Fine, Good & Mizner Office of the Executive Legal Director 88 Broad St.

i U. S. Nuclear Regulatory Commission Boston, MA 02110 Tenth Floor Washington, DC 20555 Calvin A. Canney City Manager Robert A. Backus, Esquire City Hall 116 Lowell Street 126 Daniel Street P. O. Box 516 Portsmouth, NH 03801 Manchester, NH 03105 Stephen E. Merrill, Esquire Philip Ahrens, Esquire Attorney General Assistant Attorney General George Dana Bisbee, Esquire Department of the Attorney General Assistant Attorney General Statehouse Station #6 25 Capitol Street Augusta, ME 04333 Concord, NH 03301-6397 Mrs. Sandra Gavutis Mr. J. P. Nadeau Chairman, Board of Selectmen Selectmen's Office RFD 1 - Box 1154 10 Central Road Kensington, NH 03827 Rye, NH 03870 Carol S. Sneider, Esquire Mr. Angie Machiros Assistant Attorney General Chairman of the Board of Selectmen Department of the Attorney General Town of Newbury

l One Ashburton Place, 19th Floor Newbury, MA 01950 Boston, MA 02108 Mr. William S. Lord Senator Gordon J. Humphrey Board of Selectmen U. S. Senate Town Hall - Friend Street i

Washington, DC 20510 Amesbury, MA 01913 (ATTN:

Tom Burack)

Richard A. Hampe, Esquire Senator Gordon J. Humphrey Hampe and McNicholas 1 Pillsbury Street 35 Pleasant Street Concord, NH 03301 l

Concord, NH 03301

( ATTN:

Herb Boynton) i Thomas F. Powers, III H. Joseph Flynn, Esquire Tosn Manager Office of General Counsel Town of Exeter Federal Emergency Management Agency 10 Front Street 500 C Street, SW Exeter, NH 03833 Washington, DC 20472 Brentwood Board of Selectmen Paul McEachern, Esquire 1

RFD Dalton Road Bhtthew T. Brock, Esquire Brentwood, NH 03833 Shaines & McEachern i

25 Maplewood Avenue Gary W. Holmes, Esquire P. O.

Box 360 4

Holmes & Ells Portsmouth, NH 03801 47 Winnacunnet Road Hampton, NH 03842 Mr. Ed Thomas Robert Carigg FEMA Region I Town Of fice 442 John W. McCormack PO & Courthouse Atlantic Avenue Boston, MA 02109 North Hampton, NH 03862

ATTACHMENT A Responses to these RAIs were forwarded by SBN-1225, Reference (d):

1 12 24 44 61 2

13 25 45 62 3

14 26 46 63 4

15 28 49 64 5

16 33 50 67 1

6 17 34 51 68 l

7 19 35 53 69 8

20 40 55 70 9

21 41 57 71 10 22 42 59 72 11 23 43 60 73 Response to these RAIs were forwarded by SBN-1227, Reference (e):

18 48 75 27 52 30 54 32 56 36 58 37 65 39 66 47 74 Responses are included in this transraittal for the following RAIs:

29 31 38 Responses to the following RAIs are being resubmitted with minor editorial changes and corrections:

21 59 61 62 l

l l

L

l t

ATTACRMENT B RESPONSES TO REQUEST 9 FOR INFORMA!!ON l

RAI 21

What is the impact on risk from accidents during shutdown and refueling when the containment function may not be available?

RESPONSE 21 All work performed to date to identify and to assess the risk of potential occidents at Seabrook station has concerned itself primarily with scenarios that could initiate at or near full power operation.

In the original full scope PSA (Reference 4), the coverage of accident sequences in terms of initiating events, the possibilities for system success and failure states, and the treatment of dependent events met or exceeded those of other published PSAs. This coverage was certainly greater than was possible during the seventies when the Reactor Safety Study was performed. A judgcent normally made in a PSA, and made in the SSPSA, is that the level of risk associated with accidents that could initiate during full power operation, however small, 2l-l

5 is substantially greater than that associated with accidents that occur during shutdown. There are many reasons to support this judgment including tne fact that at full power there is a greater level of RCS stored energy, after-heat level and inventory of radionuclides than the case with plant shutdown. There is also generally more time available to recover from adverse situations during shutdown.

Several years after the SSPSA was completed a research project was performed for the Electric Power Research Institute in which the risk of accidents at the Zion nuclear plant during plant snutdown and RHil system operation was assessed (Reference 5). The only risk parameter quantified in this study was core melt frequency. The results in comparison with the results of the Zion plant PSA (Reference 6) for power operation events that had been completed previously by the same PSA team are as follows.

Core Damage Frequency Description Mean Median Cold Shutdown (Reference 5) 1.8 x 10-5 2.6 x 10-6 Power Operations (Reference 6) 6.7 x 10-5 5.0 x 10-5 Hence, the core melt frequency from cold shutdown events at Zion is less likely but more uncertain than that from power operations. The Zion cold snutdown study did not address consecuences of nese events; it only addressed tne frequency of core damage events.

2I-Z

Tne Zion cold shutdown study examined plant shutdown and startup procedures in detail to identify a wide spectrum of potential accident sequences that could originate and develop during plant shutdown.

It also made use of an in-depth review of in-plant records and information that covered 10 refueling outages, 24 maintenance outages, and some 27,888 hours0.0103 days <br />0.247 hours <br />0.00147 weeks <br />3.37884e-4 months <br /> of RHR system operations. Several person-years of effort went into the Zion investigation.

It is of interest in this note to address the risk from plant shutdown events at Seabrook Station, which like the Zion plant, is a four-loop Westinghouse PWR with a large dry containment.

In the brief time available, it is not possible to complete the kind of in-depth examination that was described in Reference 5.

On the other hand, for the purpose of addressing the implications on emergency planning, it is not sufficient to measure risk simply in terms of core damage frequency.

Witn this perspective in mind, the objectives of this response are to:

o Provide an order of magnitude estimate of the frequency of core damage events that could initiate at Seabrook Station during plant shutdown.

e Estimate the frequency of the aoove events tnat result in containment bypass, containment high leakage, or containment intact end states.

e Account for important specific and unique features of tne Seacrook plant hardware anc procecures.

1447P102986 I~

i e

Provide a suitable allowance for uncertainties associated with a preliminary level of analysis through the appropriate use of conservative assumptions.

Provide for a reasonable level of accountability of operating e

experience with events that have occurred in similar plants during plant shutdown.

APPROACH The approach taken to address shutdown loss of cooling events at Seabrook Station was first to review the Zion study (Reference 5), to compare the design and operational features of Zion and Seabrook, and to identify key differences important to the determination of shutdown cooling risk.

Based on this review and the key differences that were identified, a determination was made of the extent to which all or part of the NSAC-84 results for Zion could be applied to Seabrook.

In cases where Seabrook specific features indicate a reduced level of risk, appropriate corrections were made to the Zion results. Finally, a quantification was made of sequences that could occur at Seabrook Station at a higher frequency than that assessed for Zion.

In summary, the risk of snutdown cooling events at Seabrook Station was evaluateo as folicws:

Seabrook Risk = Zion Risk per NSAC-84

- Portion of Zion Ris< Not A;;11caole to Seaccook

- Portion of SeaDrock Risk Not Applicable to Zicn 1447P102986

-n

,.s in other words, there are some design and operational f eatuees common to Zion and Seabrook and some unique to each plant. The enhanced features of Seabreak were accounted f or by reducing the risk contribution of selected Wriinant -

sequences in the Zion results. This resulted in a reduction of the cora damage f requency evaluated in NSAC-84. Then, the enhanced features e,f. Zion were accounted for by adding to those results a separate Seabrook specific analysis of accident sequences that were not important in the Zion results because of its unique enhanced features.

A The above process resulted in a balanced and unbiased albeit cor..)ervative assessment for Seabrook Station that was especially designed to makJ maximum

~

and appropriate use of the Zion results for core damage frequency..Then, all the resultant core damage sequences were evaluated to determine the trequeny of three types of core damage release states: core damage with intact containment, small bypass, and large bypass. Finally bounding estiraates were made of the contributions of shutdown loss of cooling events to the 200-rem dose versus distance curves in References 1 and 2.

COMPARSION OF ZION AND SEABROOK DESIGN FEATURES The ability to respond to this question was quickly f acilitated by the f act that key plant and systems analysts of the Zion PSA team played major rriles on the Seabrook PSA team. The design f eatures of the respectivo plants vare compared f rom two perspectives.

First, the major dif ferences between the !.wo plants were noted based on our general understanding of the plants, systene, components,

~

and PSA results. Second, the 34 dominant accident sequences for Zion in the shutdown loss of cooling, Z I - 6I

s,

~

p s

.x

/.

's I

cy:nt listed in NSAC-84, Table e-1, w;rs revicwed for applicability to r.

i f /

u Seabrook Station.

I While many similarities between the two plants were

^

-r.oted, the following key differences were identified that could be ol, e

f,

1

' 31gnificant'in the determination of shutdown cooling risk.

m

.r 1.

RHR System Configuration.

c~.,

Both plants employ a similar two train RHR I

"3 '

system with each train consisting of a' 1007, motor-driven pump, heat a'

exchanger, and a minimum flow recirculation line.

Each train in both

~,x, : plants supplies cooled RHR flow to the four RCS cold legs and takes succion from hot leg piping. 8cth plants have comparable-instrwentation and alar;n systems for the RHR system.A major

' Yi?fer:ence is that Seabrook has a separate hot l eg s'uction pipe for 1

each'RHR pump train; whereas Zion has a single suction pipe that feeds both R,4R pump trains. - This is a very important difference and an advantage for Seabrook because single failures (valve closures) in the suction path were four:d in NSAC-84 to be the ranking contributors to shutdown loss of cooling frequency.

Such events have occurred frequently at Zion and over half of the shutdown core damage frequency at Zion was found to be initiated by such an event.

At p

Seabrook, a single valve closure in the RHR suction valves would need to be accoinpanied by the unavailability of the second pump train to 3

cause a total loss of RHR flow.

Hence, the frequency of such events i

would be sipificantly lower at Seabrook Station.

2.

Support System /Frontline System Interfaces.

There are significant differences between the two4 plants in the electric power, service water, and component cooling water systems and in the interfaces 5

1447P102986 4

i

(

._--,w----,

---

,.-.-n,

i

'e s,

'S t

C n

bstwaan these systems and the f rontline systems, such as the RHR system.

For one thing, there are ways to utilize equipment on Unitr2 for Unit 1 and vice versa at Zion that are not possible at Seabrook.

These differences stem from the fact that modern design criteria, to which Seabrook was designed and Zion was not, call for a strict physical separation betw'een redundant trains of safety-related systems and greatly reduce the opportunities for lining up cross-train pump and heat exchanger combination.. In othere words, there are more success paths in the older plants such as Zion.

Ironically, tha introduction of these more restrictive design criteria in Seabrook produces a relative advantage for Zion in this regard. Theref ore, we would expset to see a higher contribution from sequences involvong cross-train combinations of electric power, s'ervice water, component cooling water, and RHR systems at Seabrook, relative to Zion.

1

(

l t

3. Other Plant Differences.

The remaining plant differences that were identified could be significant in the determination of the risk of power operatloa events, but are not found to be significant with respect to shutdown cooling risk.

These differences include those in the containment heat removal systems (different c'o nf igura t io ns of containment s, pray and f an cooler systems), use of solid state versus relay technology in the safeguards actuation system at Seabrook and the ability to utilize Unit 2 e'quipment for Unit 1 and vice versa at Zion. There is a high degres of similarity between Zion and Seabrook in the procedures that govern shutdown operation. Of the differences in this area, there are distinct advantages to Seabrook (e.g. some of the local manual valve operations at Zion are perf ormed remote manually from the control room at Seabrook.

y Ill-7 m

1

i UTILIZATION OF NSAC-84 RESULTS FOR SEABROOK Following the design and procedures review and comparison, the dominant sequences from Table 6-1 in NSAC-84 were reviewed for applicability to Seabrook Station. The following conclusions were reached.

Because of the similarity between tne plants and the procedures, the e

dominant sequences from Table 6-1 are generally applicable to Seabrook.

The NSAC-84 sequences would be expected to occur at the same e

frequency at Seabrook, except for those sequences involving inadvertent closure of RHR suction path MOVs and those involving combinations of support system faults and RHR train failures.

The sequences involving suction patn MOV closures uould occur at a lower frequency at Seabrook because Seabrook has a separate suction path for each pump. The frequency of valve closures was calculated as part of Top Event RM in NSAC-84 This top event asks whether RHR cooling is maintained during maintenance and refueling outages.

Ine cause taole for tnis event is snown in Tacle 1 (acapted frcn Tacle 5-5 of NSAC-da). Also

't 1447P102986 2)-g

snown in the taole is a correction factor tnat snows tne effect of two drop lines at Seabrook in lowering the frequency of " hardware failures" and " human errors." The derivation of the correction factors is explained below.

For spurious valve closure to cause a loss of RHR cooling at Seabrook Station, it is necessary to postulate either a common cause event involving one valve in each suction patn, or a coincidence of a single valve closure and maintenance being performed on the other RHR train (these could also be maintenance in a support system of the other RHR train, but tnese sequences are separately accountea for below). The correction factor for this cause of RM is given by i

SMOV + (1-SMOV) (.5) QRHRM =.072 where SMOV = MOV Common Cause Parameter =.043 from SSPSA Section 6 QRHRM = Maintenance Unavailability of a Single RHR Pump Train During Shutdown

= 6.1 x 10-2 Based on Zion Data in NSAC-84 The factor of.5 is the chance that the maintenance is being done in one of two specific trains.

The correction factor for errors in invertar switching is given by

.5 Q

=.031 RhRM 1447P102986 St)-9 w--

y

. - - - y ym,

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

.e The result of the above corrections to this top event is a reduction in the failure frequency to a factor of.145.

This factor was applied to applicable sequences in Table 6-1 and the following results were ootained:

Core Melt Frequency Results Mean Median NSAC-84 Results for Cold 1.8 x 10-5 2.6 x 10-6 Shutdown Results Corrected for two RHR 7.6 x 10-6 1.1 x 10-6.

Suction Paths

• Estimated as source factor reduction as calculated for mean results.

Hence, because of the dominance of the valve closure events in the NSAC-84 results, the effect of having two suction paths is a reduction of core damage frequency of the NSAC-84 sequences at Seabrook by a factor of about 2.

ANALYSIS OF SEABROOK SUPPORT STATE SEQUENCES Because of differences in the support system interfaces with tne RhR system and-cecause tnese particular differences are unfavoracle for Seaorook, separate event tree analyses were performed to cover these events for Seabrook. The following initially events were selecteo for this analysis.

2L l ~ hO 1447 P I(170M

Dc31gnator Initiating Event LOSP Loss of Offsite Power L1RH Loss of One RhR Train L1PC Loss of One PCC Train L2PC Loss of Both PCC Trains LISW Loss of One Service Water Train L25W Loss of Both Service Water Trains As shown in Figure 1, these initiating events were first analyzed in support system event trees whose sequences result in one of five different plant states. The plant states denote the numoer of RHR trains and safety grade AC power trains rendered unavailaole by the combination of the initiating event and support system failures. These states together with the sequences borrowed from NSAC-84 as corrected for Seabrook were then fed into a frontline system event tree, which considers additional events needed to resolve the end states of the event sequences in terms of release categories. This main line event tree is based in the event sequence diagram in Figure 2.

In tnis analysis, tne NSAC-84 sequences were assigned to support state R2E0 (loss of botn trains of RHR with botn trains of AC power available).

1447P1029a6 Z I -il

=

The event tree quantifications for L1RH, L1PC, LISW, and LOSP are snown in Figures 3, 4, 5, and 6, respectively. The quantifications were based on the SSPSA and RMEPS results for the support systems and initiators, except for maintenance unavailability. Train B of all systems was assumed to be unavailable for maintenance with a conservative value of unavailability of 0.1.

This more than accounts for the higher chance of maintenance during plant shutdown. The initiating events L2PC and L25W are assigned directly to support state R2E0 because of a very small chance of electric power failure with no loss of offsite power. The results of the analysis up to the point of support state are presented in Table 2, which is organized into tnree types of events:

Type 1 is events with one RHR train unavailable (RIE0, R1E1); Type 2, with two RHR trains unavailable (R2EO, R2El, and R2E2), and Type 3 is tne set of NSAC-84 sequences. When the sequences are combined according to supoort state, the following results are obtained.

Support State Mean Frequency (events per reactor-year)

R1E0 1.7 x 10-1 R1El 3.3 x 10-5 R2E0 2.0 x 10-3 R2E1 2.0 x 10 7 R2E2 3.8 x 10-7 RXEY = Sequence with X RHR trains and Y electric power train unavailaDie.

Ine even: sequence 213; ram in Figure 2 cefines tre aossiale progress 1:n for Type 1, 2, and 3 support state sequences. For Type 1 and

2. ) -t 2.

1447P102985

I 2 sequences, consideration is given to operator recovery to prevent core melt.

On the other hand, such consideration is not made for Type 3 because such actions are already considered in NSAC-84.

Next, the RO event questions whether the RCS pressure boundary is open initially; i.e.,

vessel head or steam generator manway cover is removed.

For RO closed sequences, the ESD tracks the possible developement of interf acing LOCA conditions either through check valve closures or RHR system repressurization via Top Events CV, RV, and MC.

For pressure boundary open sequences or all other nonbypass sequences, consideration is given to whether large and small penetrations are initially open and, when open, whether or not operator actions to secure these penetrations are successf ul.

Since the containment sprays are not tracked in the ESD for simplicity, successf ully isolated sequences could result in either a containment intact (S5) or delayed overpressurization (S3). Those with large or small bypass sequences are assigned to S6 and S2, respectively.

I l

i I

i i

21-13

EVENT TREE QUANTIFICATION With reference to Figure 1, the event trees were quantified in two stages.

Fi rs t the support system event trees were quantified for each initiating event resulting in the quantification of the uncon-ditional frequencies of 5 different RHR support states.

RIEO, RIEL, R2EO, RZEl and R2E2 (where RXEY is the state in whcih X RHR trains and and Y trains of safety grade AC electric power are rendered unavailable.

Then, the main line event tree was quantified 6 times, one for each RHR state and a separate quantification for the sequences borrowed from the NS AC-84 results. The derivation of the event tree split fractions for each event tree quantification is described below.

Support System Event Trees (Figure 3, 4, 5 and 6)

The support system event trees were quantified for the following initiating events.

LIRH - loss of 1 RHR train LIPC - loss of 1 PCC train L2PC - loss of both PCC trains LISW - loss of I service water train L2WS - loss of both service water trains LOSP - loss of of f site power LIRH The f requency of loss of 1 RHR train during shutdown was estimated using the f ollowing model.

hLlRH=2 RHR RHR hLIRH f requency of the initiating event where

=

(events per reactor year)

)(RHR f ailure rate of 1 RHR train

=

(dominated by the RHR pump) the number of hours per year in shutdown t RHR

=

Note the omission of MOV closure events in the above is by design, these events are included in the " type 3 events" borrowed f rom NSAC-84 and corrected for Seabrook having 2 RHR suction paths from the RCS.

The time on RHR, t RHR, is estimated using Zion experience, which is viewed as a conservative assumption for Seabrook. The reason for this view is that the zion experience is worse than average for PWRs and does reflect G7 \\ ~ l 4-

the generally higher availability factors of the Yankee system of plants (Maine Yankee, Vermont Yankee, Yankee Nuclear Power Station).

In the 16 reactor-years of experience of zion i and 2 there were 12 first refueling outages of average duration 1,992 hrs and 3.05 maintenance outages per unit-year of average duration 488 hrs.

12 ref uelings x 1992 hrs + 3.05 maintence events / reactor-year g

16 reactor-years x 488 hrs / outage = 2982 hrs / year Hence, in the first 16 reactor-years at zion the plant was shut down about 347. of the time. We conservatively assume the same value for the plant lifetime at Seabrook.

The support system event tree quantification for LIRH is shown in Figure 3.

In this event tree, it is assumed that the plant is initially being cooled with RHR train A and train B is in the standby.

Becuase of the strict train-wise dependencies at Seabrrok Station, critical operation of RHR train A precludes unavailability of service water and PCC trains A, since both are needed to operate RHR train A.

In normal power operation, the unavailability of single service water and PCC trains is very low-no greater than 10-3 to 10 -2 per train. However, during plant shutdown, the unavailability due to maintenance is generally higher. For example, at Zion the plant specific data shows single train maintenance unavailabilities of RHR, SW and PCC in the range of.03 to.06.

It is conservatively assumed in the analysis that all safety grade non-operating subsystem have shutdown maintenance unavailabilities of 10-1 This is greater than any shutdown maintence unavailabilities observed in the data.

LIPC The loss of one PCC train initiating event is analyzed in Figure 4.

The initiating event frequency is estimated with the following model.

2ke tRHR s +} pc LLPC

=

a where

= f ailure rate of 1 PCC pump = 3.4 x 10-5 per hour from pc PLG-0300 Section 6.

= (.34) (8760) based on LIRH analysis tRHR hs

= failure rate to start of standby PCC pump

= 2.4 x 10-3 f rom PLG-0300 Section 6.

= mean time to repair the initially running pump

= 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> per PLG-0300 Section 6.

6.3 x 10-4/ reactor-year kLIPC Hence

=

o Z l -15

=

The LIPC event tras is quantified using information for tha RMEPS and SSPSA for service water train A and the same train B maintenance assumptions as with the LIRH event.

LISW The configuration and failure rate of the SW system are comparable to the PCCW system, i.e, each train has an operating and a standby pump. The event tree for LISW is quantified in Figure 5. the initiating event frequency is the same as that for LlPC.

Note that this analysis includes the tunnel system only.

The SW cooling tower system is considered in the subsequent recovery analysis.

LOSP The loss of offsite power event tree is quantified in Figure 6.

The initiating event f requency is estimated using the following model:

E LOSP LOSP RHR where

]kLOSP = f requency of LOSP =.135 events / year (SSPSA Section 6)

= time on RHR =.35 (see LIRH above) tRRR EPR

= f requency of non-recovery of LOSP before core damage =.01 (assumed)

The above assessment for EPR can be compared with EPR-1 in the SSPSA, a value of.03 for f ull power operation. The.01 value is viewed as conservative in comparsion with EPR-1 since the time constants for core recovery are much longer during plant shutdown.

The event tree split f raction f or LOSP in Figure 6 are based on the results of the SSPSA and RMEPS for train A and 107. maintenance unavailability for train B used for all shutdown looss of cooling events.

I i

L2PC and L2SW I

The initiating event f requency for loss of both trains of PCC is estimated for the following model.

hL2SW=hL2SW h2PC=

h L2PCC RHR

'W

'W where h = frequency of same event during operation from SSPSA C

hours of power operation assumed in SSPSA (8760) po =

2L I - 16

Top Event RH For the LIRH and other RIA0 support states sequences, it is assumed that train A is initially used to provide RRR and train B is in standby.

Hence, it is the train A subsystem that is involved in the initiating event.

For these conditions, the failure to provide continued RHR cooling is estimated from the following model.

hRH=

a + h ps +

p where fm=RHRpumgtrainmaintenanceunavailabilityduringshutdown

= 6 x 10-from Zion data in NSAC-84

),,ps=standbypumpfailuretostart

= 3.3(-3) from SSPSA Section 6 1pr=runningpumpfailuretorun

= 3.4 x 10 5/ hour from SSPSA Section 6

[p=meantimetorepairtheinitiallyfailedpump=21hoursfrom SSPSA Section 6 hRH=6(-2)+3.3(-3)+3.4(-5)(21)=6.4(-2)

For RIAL, the same model is used except the standby pump must run longer to cover the repair time of a diesel generator - assumed to be one week.

Hence, for R2EO, R2El, and R2E2, Q RH = 1 j

jRH = 6(-2) + 3.3(-3) + 3.4(-5)(168) = 6.9(-2)

Top Event OM RIE0 represents the most ideal conditions and minimum stress levels of those considered for OM for these conditions, OM is estimated from:

OM = DEI + SC + BF where del = operators f ail to recognize that RCS heat removal should be restored after running RCS pump stops.

SC = operators fail to align and restart a core cooling system l

l BF = operators fail to provide long term makeup to chargining i

system l

l Using appropriate values f rom Table 5-6 of NSAC-84, the following quantification is made:

OM = 1.0(-5) + 5.0(-4) + 1.0(-5) = 5.2(-4) i As shown in Table 3, higher values are used for the remaining support i

states to reflect dif fereat and progressively greater stress and comparison levels going through the sequence RIEO, R2E0, RIEL, R2El, R2E2. For type 3 events, OM = 1 to avoid double counting recovery already considered in NSAC-84.

I i

11-17

Top Event RO The fraction of time the pressure boundary is open is keyed to the time 7

assumed for RHR shutdown cooling.

For consistency, since Zion data was used to quantif y the latter, it must be used to quantif y the former.

From Table 3-4 of NSAC-84, the total time the RCS is opened during maintenance outages is 5,014 hours1.62037e-4 days <br />0.00389 hours <br />2.314815e-5 weeks <br />5.327e-6 months <br />. From table 3-1, the RCS open time during refueling outages is above 3,000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br />.

Hence, overall the 31,687 hours0.00795 days <br />0.191 hours <br />0.00114 weeks <br />2.614035e-4 months <br /> of the Zion outage experience, the RCS was opened (8014/31,687) =.25 of the time.

Top Event CV The frequency of CV is quantified per our response to question 47, hCV=5.5x104 Top Event RV This event is quantified in RMEPS; it is estimated using:

jRV=21RV=4.8(-5) wherelRV is the f ailure rate of each RHR relief value = 2.4(-5) from SSPSA Section 6 Top Event MC For both trains of AC power available, the top event is estimated using the following model.

MC = 2(k MOV2+B D10V)+4(kCV2+BhCV)

MOV

}MOV = f ailure rate (f ail to close on demand)

= for MOVs - 4.3(-3) per SSPSA hCV=failurerate(failtocloseondemand) f or check valves = 5.5(-4) per question 75 response.

b MOV;) CV = Beta f actors for each type of valve =.1 hence q

-'(5.5 x 10 4)2 jMC=2((4.3x103)2+ 4.3 x 10 3)(,t)t

+4

+ (5.5 x 10 4)(.1) = 1.1(-3)

~

For one AC power bus available (RIEL and R2EI) there is only one MOV on each suction path potentially available. For these states hMC=21MOV+4(1CV2 $ CV1 CV)

= 8.8(-3)

For two AC power bases unavailable OMC = 1 Z t -18

Top Evant LI Thi,s avant quistions whsthar any largs panstrations are opsn initially. Largs is dafinsd as a total equivalent single opsning of greater than 3" in diamter.

Examples of such penetrations are equipment hatch, parsonnel hatch, and containment purge penetrations. These penetrations may be opened during shutdown unless fuel is being moved.

The chance that large penetrations are opened is highly dependent on the reason for the shutdown.

If the reason is refueling, steam generator manintenance or othare maintenance on reactor coolant system components, it is likely that large penetrations such as the equipment hatch will open.

If on the other hand, the outage occurs due to need to repair or maintain equipment outside the containment, (e.g. turbine generator related maintenance) there would not be a compelling reason to open up large penetrations in the containment.

To reflect the above considerations, LI is assessed as a function of the status of event RO.

If RO is true (reactor coolant system is opened), it is assumed that LI is true (large penetrations are open) 90% of the time. If RO is not true (reactor coolant system is closed), it is assumed that LI is truc 10% of the time.

Note that at Zion, of the 8,014 hours1.62037e-4 days <br />0.00389 hours <br />2.314815e-5 weeks <br />5.327e-6 months <br /> during shutdown that the RCS was opened, the fuel was being shuffled for 1600 hours0.0185 days <br />0.444 hours <br />0.00265 weeks <br />6.088e-4 months <br /> (roughly 160 hrs per refueling outage).

Hence 80 percent of the time that the RCS was opened, it would have been permitted by tech specs to have the equipment hatch open.

Top Event OL Given a large penetration is opened initially, the event questions whether the operators successfully close the penetrations before a potential release situation could develop.

The probability of successful recovery is assessed as dependent on the RRR support state, i.e. the combination of the initiating event and the response of the plant support systems. At dif ferent suppo rt states there would be different levels of stress and confusion to inhibit operator recovery actions.

To provide an indication of the amount of time available to close the equipment hatch or other large penetrations, the time to core damage taken as the time to uncover the core was estimated f or the f ollowing cases:

Cases Time to core uncovery (hr) 1.

Reactor vessel head open with water level at hot leg nozzle midplane A.

Loss of cooling at 2

days after shutdown 0.8 B.

Loss of cooling at 30 days af ter shutdown 2.6 2.

RCS filled at pressure < 425 psig with A.

Loss of cooling at 1 day after shutdown 5.4 B.

Lees of cooling at 10 days after shutdown 14 C.

Loss of cooling at 30 days af ter shutdown 22 3.

Water at refueling level with A.

Loss of cooling at 5 days after shutdown 72 B.

Loss of cooling at 30 days af ter shutdown 162 t

i 2L i - l9

It is not known how quickly tha equipmsnt httch can ba secured.

Our currant information is it would take several hours to attach, and up to 8-12 hours to secure all the bolts and establish a tight seal. Just how quickly this process can be accelerated is uncertain. To address this uncertainty, a base case and a bounding case are performed.

In the base case, it is assumed that the mean time to close the hatch is 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />.

In the bounding analysis, it is assumed that the hatch remains open with a probability of 1.

From the Zion data in NSAC-84, there were 10 refueling outages and 24 forced maintenance outages resulting in an average outage duration of about 39 days.

Of the entire 31,687 hours0.00795 days <br />0.191 hours <br />0.00114 weeks <br />2.614035e-4 months <br /> of outages, roughly 1% of the time the RCS was drained, 5% of the. time refueling was taking place, 5% of the time the plant was not on RHR, and in most of the remaining 89% of the time, the reactor system was filled on RHR.

Based on the above recovery time, and assuming a 4-hour hatch recovery time, it is seen that with the RCS drained to the hot leg nozzle midplanes, the chances of successful hatch recovery are not very high.

While under all other conditions, the chances are high. Therefore, the base case and low stress levels, a value of.01 is used for f ailure to isolate large penetrations.

For degraded RHR states, this value is increased en correspond with higher stress levels, as indicated in Table 3.

In the bounding case, a failure frequency of 1 is assumed for all states.

Top Events SI and OS It is conservatively assumed that small penetrations are open 90% of the time and the chances of recovery are assessed at levels comparable to those for OL, even though all small penetrations can be isolated quickly.

Results The results of this preliminary analysis of shutdown loss of cooling events are shown in Table 4 for the base case assumptions on event OL.

To bound the consequences of these events, accident sequences were assigned to the existing PSA release categories, even though the release f ractions for shutdown events would be expected to be considerably lower than those calculated for power operation events.

Based on what is believed to be a very conservative set of assumptions in this base case, the impact of shutdown events is assessed to result in no greater than a 14% increase in core melt frequency, and an 18%

l increase in category S6 frequency.

For the bounding case of no credit for event OL, the frequency of S6 would increase to about 5 x 10-6/ year.

The impact of these bounding estimates of shutdown events on the dose vs distance curves for 50 rem and 200 rem whole body gamma doses are shown in Figures 13 and 14 for the base case OL and bounding case OL assumptions, respectively. Our best current statement of risk levels is represented by Figure 13.

As seen from this figure, the addition of shutdown events impacts the right tails of these curves, but the combined results at 1 mile are still less than the i

NUREG-0396 values at 10 miles. Even with no credit for equipment hatch recovery as assumed in Figure 14, the combined shutdown and power operation results fall i

below the NUREG-0396 10 mile levels at less than 2 miles.

Hence, even a very conservative analysis of these events does not impact the conclusions of the sensitivity study.

It is expected that a more detailed investigation of these events would result in much lower levels of risk than either set of results presented here.

21-2o

-._~_

ggygvasum_

Pickard, Lawa and Garrick, Inc., Westinghouse Electric Corporation, 1.

and Fauske and Associates, Inc., "Seabrook Station Risk Management and Emergency Planning Study," prepared for Pub 1tc Service Company of New Hampshire, New Hampshire Yankee Division, PLG-0432, December 1985.

2.

Pickard, Lowe and Garrick, Inc., "Seabrook Station Emergency Planning Sensitivity Study," prepared for Public Service Company of New Hampshire, New Hampshire Yankee Division, PLG-0465, April 1986.

3.

Letter from S. M. Long, U.S. Nuclear Regulatory Commission Staff, to R. J. Harrison, Public Service Company of New Hampshire, " Request for Additional Information...," Docket Nos. 50-443 and 50-444.

4 Pickard, Lowe and Garrick, Inc., "Seabrook Station Probabilistic Safety Assessment," prepared for Public Service Company of New Hampshire and Yankee Atomic Electric Company, PLG-0300, December 1983.

5.

Bley, D. C., and J. W. Stetkar, " Zion Nuclear Plant Residual Heat Removal PRA," EPRI/NSAC Report NSAC-84, July 1985.

6.

Pickard, Lowe and Garrick, Inc., Westinghouse Electric Corporation, and Fauske & Associates, Inc., " Zion Probabilistic Safety Study,"

prepared for Commonwealth Edison Company, September 1981.

p -26 1

Table 1.

CORRECTION OF NSAC-84 RESULTS FOR RHR LOSS TO ACCOUNT FOR 2 SUCTION PATHS Seabrook l

Revised Failure Cause Mean Value Duminant Contributor Correction Failure l

Factor Frequency liardware Failures 6.08-2 Spurious Closure of Rl!8701 or Ril8702

+.072 4.38-3 lini n t enance 7.37-3 Running RilR Pump Falls with Standby Pump x1 7.37-3 Out for Maintenance iluman Errors 6.00-2 Errors During TSS 15.6.36 or Inverter x

.031 1.86-3 Switching (Ril8701 or RH8702 close)

Suppo rt System 2.94-6 Component Cooling Water lleat Exchange r x1 2.94-6 Fail ures Failures Dependent Cunponent 6.03-3 RilR Pumps Fall During Operation x 1 6.03-3 Fail ures l

l l

l Total 1.34-1 1.96-2 l

l N

Tw N

TABLE 2.

CLASSIFICATION OF SUPPORT MODEL ACCIDENT SEQ MAIN LINE MODEL QUANTIFICATION CASES Impact vector Power Classif catfom A

8 A

8 Type 1 Events - One RHR Trafn Mede Unavailable LIRH 1.7-1 X

R1E0 L1PC4R 5.0-5 X

R1EO L OSP*G8M*01R 2.0-5 X

X R1El LOSP* gal *01R 1.3-5 X

X RIE1 LISW*SR 5.1-6 X

RIED LOSP*P8M9 R 3.4-6 X

RIE0 L OSP*W8M*SR 3.8-7 X

X R1El LOSP4A2*PR 8.9-8 X

R1E0 LOSP*WA3*SR 5.6-8 X

X RIE1 L1PC*WA3*SR 1.8-9 X

R1E0 Type 2 Events - Two RHR Trains Made Unavailable L1RH*P8M9R 1.8-3 X

X R2E0 LIRH*W8M*SR 2.0-4 X

X R2E0 L1PC*P8M9R 5.5-6 X

X R2E0 LISW*W8M*SR 6.3-7 X

X R2E0 LISW*P8M*SR 5.7-7 X

X R2E0 L OSP* cal *G8M*02R 3.5-7 X

X X

X R2E2 LOSP*CA1 98M*01R4 R 1.4-7 X

X X

R2E1 L OSP4A2*G8M*01R 6.0-8 X

X X

R2E1 LOSP* gal

• WPM *01R*SR 1.5-8 X

X X

X R2E2 L OSP9A298M*PR 9.8-9 X

X R2E0 LOSP*WA3*W8M*SR 6.9-9 X

X X

X R2E2 LOSP*WA398M*SR 6.2-9 X

X X

R2E1 LOSP*WA3*C8M*01R*SW 3.9-9 X

X X

X R2E2 LOSP *PA2*W8M*SR 1.1-9 X

X X

R2E1 LIPC*WA3*W8M*SR 2.2-10 --.

X X

R2E0 L1PC*WA398M*SR 2.0-10 X

X R220 Type 3 Events - From 7.6-6 X

X NSAC-84 R2E0 NOTE:

Exponential notation is f.e.,1.7-1

• 1.7 x 10-I. indicated in abbreviated form; 1448P102886 2t-23

TABLE 3.

SUMMARY

OF SPLIT FRACTIONS FOR MAIN LINE EVENT TREE Event Tree Ouantification Cases Main Line Event Tree Type 1 Type 2 Type 3 Top Event (NSAC-84)

R1E0 R1El R2E0 R2E1 R2E2 EUT 6.4-2 6.9-2 1

1 1

1 UTI 5.2-4 1.0-2 2.0-3 1.0-1 1

1 RIf

.75

.75

.75

.75

.75

.75 C7 5.5-4 5.5-4 5.5-4 5.5-4 5.5-4 5.5-4 TOT 4.8-5 4.8-5 4.8-5 4.8-5 4.8-5 4.8-5 FRI 1.1-3 8.8-3 1.1-3 8.8-3 1

1.1-3 ETlR0

.90

.90

.90

.90

.90

.90 ElliRT

.10

.10

.10

.10

.10

.10 IUT 1.0-2 3.0-2 1.0-2 3.0-2

.10 1.0-2 IT'

.90

.90

.90

.90

.90

.90 Uli 1.0-2

.10 1.0-2

.10 1

1.0-2 X = Event Success I = Event Failure NOTE: Exponential notation is indicated in abbreviated form; i.e., 6.4-2 = 6.4 x 10-2, laanpin? Ann 7, g - 2 4

l TABLE 4.

KEY RESULTS OF SHUTDOWN SEQUENCES FOR SEABROOK STATION Release Category Event Tree S5 or S3 S2 S6 R1E0 5.6-6 5.0-8 1.9-8 R1El 1.9-8 1.7-9 2.0-10 R2E0 4.0-6 3.5-8 1.4-8 R2E1 1.8-8 1.8-9 1.9-10 R2E2 3.2-8 3.3-7 1.2-8 Type 3 7.5-6 6.8-8 2.6-8 Total for Shutdown 1.7-5 4.9-7 7.1-8 Events Total for Power 1.1-4 2.0-5 3.2-7 Operation Events Percent Increase 13.3 2.4 18.2 l

with Shutdown l

Events NOTE: Exponential notation is indicated in abbreviated form; i.e., 5.6 5.6 x 10-6, I

l l

l l

1448P102886 gg.gf

IIHfl PLANT STATES SUPPORT SYSTEMS (STAIUSOF RHR AND MAIN LINE EVENTSEOUENCE INITIATING EVENTS EVENT 1REES AC FUWFR SYSTEMS)

EVENT TREE END/ RELEASE STATFS

• LOSS OF 1 RHR TRAIN L1RH RIE0 y

NOCORE I

LOSS OF 1 PCC

_p TRAIN L1PC R1El DAMAGE WITH 35 LOSS OF 2 PCC OFFSITE FOWER TflAINS L2PC AVAILABLE (FIGURES 3,4. AND 5)

EVENTTREE CORE LOSS OF 1 SW I

OUANTiflED FOR DAMAGE WITH TRAIN LISW R2CO y

m dL EACH RHR PLANT SMALL BYPASS STATF S2 (FIGURES 7 THilOUGH 12)

LOSS OF 2 SW 4

TRAINS L2SW DAMAGE WITH R2E1 LARGE BYPASS LOSS OF SUPPORT TREE FOR S6 OFFSITE POWEfl

> LOSS OF OFFSITE LOSP fOWEfl (FIGUflE 6)

APPLICABLE CORE

+

R2E2 DAMAGE EVENTS FilOM NSAC-84 N

I flGURE 1.

ACCIDENT SEQUENCE EVENT TREE MODEL FOR SEABROOK STATION SHUTDOWN COOLING EVENTS

[

O g

.il!

n Ne$

Est E'

1. 55 li!!=

!"E

_s v

ig

'il 483 is

!E N, !"

1 I

-ifl=

3

_ii-L ili!

ll!!

ss i

al II 1

4 i

n ni! W i!!! -

v

(

Gl/

i,

!!!l l iii

, e

,li>llil!i e

i e

r

,i 6

i-i

.i q,.i i,i,p.

1; 3lill,l1!;!

lii isii; G

!!i ill d

nl

=

l_!

.gli e

m

!!Il!!i

-e

g

i in

s e-ai

((i.

D-U

t l

INITIATING SENCE SENCE PRIMARY PRIMARY EVENT WATER WATER m

m NUMW 9

L1RH TRAIN A TRAIN 8 A

B

'"I mg YEAR)

UNAVAILA8LE 20/ YEAR a 1

m

.90 m1 a

.90 1

.185 1

ma ma ro r,

1

.10 2

1.8 (-2) 2 0

3 0

1 m

s,<

4 0

2

.10 m1 5

2.0 ( 21 2

_, g, 0

6 0

2

_g, 0

Q 7

o 9

8 0

2 g_ _ _ g_

9 0

2 GF = GUARANTEED FAILURES FIGURE 3.

SUPPORT SYSTEM EVENT TREE QUA 7IFICATION FOR LOSS OF ONE RHR TRAIN (L1RH)

2. t - 2 6

I f

f t

1 INITIATING SERVCE SERVCE I

PRIMARY EVENT WATER WATER PRIMARY t

FREQUENCY NU M OF COWONENT COWONENT L1PC TRAIN A TRAIN 8 k[

(EVENTS PER RHR TRAINS COOUNGA COCUNGB WA WB REACTOR PA PB MADE YEAR)

UNAVAILABLE 6 3 f.4) n - 1 m 90 3<

60 a.90 1

0 we Te xg 0

.10 2

0 1

1

m. 90 3

5.0 f.4) v 1

i

.10 4

5.5 f.5) 2

.10 0

_ _, y, 5

0 1

1

_y_

6 6.2 (.5) 2 3 5 f.4) 90

,y, 90 7

1.8 I 7) 1

_10 8

2.0 ( 8) 2 10

_ _ y_ _ _ y_ _

9 2.2 I 8) 2 l

GF = GUARANTEED FAILURES FIGURE 4.

SUPPORT SYSTEM EVENT TREE QUA'.TIFICATION FOR LOSS OF ONE PCC TRAIN (L1PC;'

2. t - 2 9

- =_

e l

INITIATING SEFMCE SERVICE PRIMARY PRIMARY y

NU M OF EVENT WATER WATER m

m g

(EVENTS PER RHR TRAINS L1SW TRAIN A TRAIN 8 COOUNGA COOUNG8 C

NM WE WA We pg pg 5

YEAR)

UNAVAILA8LE 6.3 ( 4) n0

^

^

m 1

0 o

^#

rs ss 2

0 1

m 3

0 1

4 0

2 5

0 1

^

.y_

6 0

2 j

-g_

1

^ 90

.90 7

5.1 ( 4) 1

,,, g,,

1

.10 8

5.7 f 5) 2

.10

_. g 9

6.3 ( 5) 2 GF = GUARANTEED FAILURES l

l FIGURE 5.

SUPPORT SYSTEM EVENT TREE CUANTIFICATION FOR LOSS OF ONE SERVICE WATER TRAIN (LISW)

N 21-30

...~

i ELECTRC ELECTRC SEPMCE SO NCE PRIMARY PRMARY INIT1ATING SECtJeCE t

PONet PONWt WATER WATER CX:WCNENT CCMONENT FWEQJENCy NUMBEROF TRAIN A TRAIN 8 TRAIN A TRAIN 8 CCOUNG N

(EVENTS PER RHR THAths N

GA TRAIN A TRAIN 8 i

GS WA WR REACT 3q MADE pA pS YEAR)

UNAVAILA8LE 4 6 f.4) m 93 a 90 m.98 n, 90 m-1 m 90 t

we we s,

s s,

3.1 r.4) o s,

.10 2

3 4 f.5) 1 2 9 f 3) a 90 3

8 9 f.7) t PA2 N'

10 4

9 8 f.9) 2 10

^~1

_ _ _ op, 5

3 8 f,5) t 2 9 f.3)

PA2

,g, 6

11 f.7) 2

' 8 '- 21 90 Q

_.g_

90 7

5 6 f.6%

t to 8

6 2 f.7) 2 to 9

. gg,,

7,_

6 9 f.7) 2 to Q gg

,, g,,

1

,,,,, e,,

to a1 f.5) 1 2 9 f.31 11 PA2

_ _, 77, 12 f.7) 2 1 8 (.21

_, cp, _,, gg, _,, cg, 12 7 7 f.7) 2

'7 'a,,' c-

- - cF -O '

--es- < '

2 5 ' 51 to 14 2 7 f.61 2

'O

- _ cp.

15 3 0 r.61 2

'O

- cp -

cp., - cp t6 3 4 r.4) 2 LOSP = LCSS CF CFFSITE PCWER WITH NO REcov2RY S EFCRE CCAE GF = GUARANTEED FA;LLRE FIGURE 6.

l SUPPORT SYSTEM EVENT TREE QUANTIFICATION FOR LOSS OF 0FFSITE POWER (LOSP) l l

Zs-31

ACTse' Test open gesaLL Opema90s a

'infTlaftnD

1. 685 OP9eaTOs aCs mue owe own uses s's me TO pgweve vaget aC?tosus te ease eveafff COOL 8Me aCTtCas to petmueG CwtCit K eLie 7 SUCTHDW

1. 9ssGTW.. ?tems gI"fw" N

suspent esaservasmes peevest 30 Yo*0 a a f v+Lwee v4 Lyse esevo sweMLLY Close uses wetiLLv v

CL pga,ose sasaLL gy;gg yraft aseLT opte CLOSS OpfaLis Clos 8 stCunes eg nete.tsosse BeCumes gre.g,0see TTpt $

, 8 88 Ces NO CV av tsC L8 OL ss og g

va L.

e e

t m

^ 2'

---

C

---

C

30 m,,.'

4 DS 9 7 f.79 t e s.=

n 1,..

m,. s m,,. t.

=

se m..

=

t.

in e.M e

n on e.e i

i I.

e,e te is

..t e-M g.'

g.1

__g=

_ _ _ g.t.

=

m.s it se

.sim rv I.0 f.2 tt et 43 f.es m,, =1 m,,.14 v3 38 0.? t41

.te JG m9 to SS S.8 f.M

,e t 0 f.B 18 ft 914 4D 4

10 f.B to 38 3 7 f4 m,.6 m,,.Se

{.te et es tJ g. set 4 A 148

=

m.s

=

u e tB t 012 te et tJ t.es m, e.t m, e.10 20 as 174.92 te ft 38 1A f.stl De

,m 9

e t O f.ft 22 12 13112 10 f.21 23 se 2.7 f.1M 1.9 f.3%

24 as 30f tM S S f.at 24 SB 3 9 f.88

=

AttyLTS E SS = FJ (4)

ZE3=88tM I 80

• 34180 FIGURE 12. EVENT TREE CUANTIFICATION FOR TYPE 3 EVENTS - EVENTS TAKEN FROM NSAC-84 THAT ARE APPLICABLE TO SEABROOK STATION 7\\-32

.o,.-vum

,.. u,.

.o.,., -,o,i.

c s eu..

.ve.r.io..

u,= sno,=.

c.

.o,

r.o,.

wou==

..m r

.e.t.o v

ev o

.e.t.o,,.,,.,o,,,,75

<- > ou. tv

.caeso evio

,o ou u,

.,4

..co.ca.o.e eu m,..u

.<.m.u.o,

...n. u.o tv u,.

eu=.

,,u

'=

w tr,a,,,,

v.,.

o,.o.a u.

..co, m

u o.6 e

r o

ev tv c

u os

==

u.

o.

's i.a se 3

3 o.

u.

m,-.

.4 6.m m

8 3m g,

c*

es v.*

i a

n -,

e n

,s e.n i. e.e 3,.

m

=.

30.

=

=

u e.

g,,,

n u s.

.. e m

esi.

i.

m sa e.n

=

m -.

g -,

_gw

_ _,_ g..

a

=

s,

• 6 f.39 il P1 SAf 14 5

9 8.f=4__

n, -9 8

e. g.B 1

t ta. _

t. f.S 9

4333 t,

t a f.s vg 4 A f.48 a,, = 0 m,, A.

=.

-. {.9 m, -.

i.

=

w t.1.B t.

.9 1 A f.t 3 2.

W 3.ft3 m, at a,.t.

.9 3

w M

,=9 11 M

t. $.9 9) t. l.25 21 s.

e a f. 3.

e. f.n r3 I s.

r. e.,n l

h r

1.tl.31

?4

1. fot.I

= = -.

2.

f.f '

.. f.at a.but,1 ZM e. A 1 I Za. ef7f I Las *.4f l

FIGURE 7.

EVENT TREE QUANT!FICATICM FCr R1E0. EVENTS THAT FA!L CNE RHR TRA!1 AND NO TRAINt 0F tAFETY GRACE AC POWER 2.1-33

i awe OP9wtf08 i

pe.86 I g ae.LA 848

. met 27,,ug 4 485 f**e E***

Op.o r.t sasatt Orge*?

P's

.C.t.o.es'ess.os gygg #

COOums actiong fc must w as'32 m@Jf, puC?i@m faxficas8 aCT vg pCasever?*0auf d CT'Om.om 8

gu, sags,yy pee.ve. arf gewa.ss..a.av asmeerv ames vdtyse

?

i g

ps

• =stu.t.s.v

.t.h.o.se. ansa.g.t 1 5,eog

.v.4.gv e.

ine fu.t t ?

esovo

r. tan..t.s.

,1.

t eton acu

.. t.

i econ neu ee Afg se,v,eaweg K989 2 88 Oes no CV WV tac ta 04 IS 00 3

ggang i

I g-l es s.m 3 -.

o=

32 '*

~'s '*z m g

g '8 C

88 m, 39 4

SS 4A s.eg

,s

'8 8

99 S.9 f.o.1 m

e, 3.ie e

e se AS s, SS e

SS 44840

- \\

19 0

39

.A6 e s e r-a is i.

m

_ _g a

_ g...

ie s, se i,

se

.s n

, e, m,,..

u s,

N

=

,=

s, n

34 s.n i.

=

e....

.see g-.

g es

_,_,_g...

m

,4, m, m i

'e

's i

I l

e,,

e, m

m s,

=

, s.

3, m

.4 e.

?

v

.ie 22 v e t.ee 3 0 f.25 73

=

23ffm I

i

(

8 8 f.39 M

=

,3 f.te l

3 E

'd 0*'I assutre 3

E SS*IJ(4 E S2

• 9.7 f 4 E 80 = 2A 4.f40 FIGuat 8.

EVENT TREE QUANTIFICATICN FOR RIE1. EVENTS THAT FAIL CNE RHR TRAIN AND CNE TRAIN CF SAFETY GRACE AC PCWER 21-34

r

=u==

<u

,e,..,.,.

w oous.=et-

=

r=

==

u.es

.o.s :,

n e,= w o.

.os,e= +vo,.a., c=

revei suer.o..

e o.,e. ;

,es.o g

.,o.e i e,e.o,, 'y's.pi ece,

n emtio ev o v.

,.,v,,.19.,. e=

ui

^

.+.m m

,etom ue st Ario.e secus sia e.i.o,

eau..

vasvu vuva

-ov

.neuur

..nwu..e.

eo.en.e, v

e ev ig ev ev

=c u

o.

si os a

's0l*

• cLoss oee;ite cLose secuneo

. we.

ne we.st c.a.e.

pr,: re-o

,9 se

=

=

e p,e

=

u<

t mv s

a esse

_ _ _ __ _ _ g, ' e

.. _ _ g's us m an i

er e -o a

w ess. ;

'T w

e n

t.

e en n<a

.,.t e m

m e

a

a. m e

=

ne.

. ei.

ie

=

tv e n

.=

m -.

g.,

__g

_ _ _ g.ie _

m ii

=

u e.

%W 9 e t.3 11 et SA s.de 93 en 33e4 Me n*

m, a.se 3r 84 en f.7 f.?g n,a 9

se 9 e f-19 9e et 1.7 f4 1 e t=tt 1

N I

t.e t4 I

e.

.te 17 N

9J f=4 99 4884 C' n

37 3r es

.9 to en 12099 me e.1-2 to re

'tJ699 to le se O A s.13 te e =9 m

mv me en m.0 ft M

938899 l

we

! e f-2 21 ft 12 f t?t l

f t e f.ft 23 es 9 A 0 83B 3e se e a s. tsp f f f.23 1

3e es it ? f.e e e M8

(

I l

avuta I

Z M*este Z 81

• 331.el Z se e t Al-el FIGURE 9.

EVENT TREE QUANTIFICATION FOR R2E0 - EVENTS THAT FA!L TWO RHR TRAIN $ AND NO TRAINS OF SAFETY GRADE AC POWER Z l -36

w e,.g., u,,,.,x,.

tanee 88 8.D Gu sv.e =, y eusq,rn.

sesm.4tr

.C.t e.e,s.t, e.ecu.t.t.

.e,,.,.,,,,,,

ie n.e===

• e vi gggg gg ere's i s.v e

reu

..w usu ee n.v.o

. e

.v.A.t.w.asn ctose a

esaserEs e coeu.

witwe.e o.,

y

.we.

e u

os u

eg u,,s sp, eta.

r

.n ev

==

o

.we o.

p,e

=

e c

s a

u s.

g

• _ _ _.c

• m, m A

n n,,.

ee e

n u na

=

g g...

m

=

m,,.

m n

u e

es ue.

w 1

I c-'

c-'

- - -C "

is a

,, e

=

e is a

u i.,i

,s a

u sa n,, u n,,.w

.n u

a e

=

m e

es es ne.

=

u

=

.'.L'dt c '

c*

-C

m,,.

=

.. n m.

m,, m m,,...

io si n

m,, e a

n n

_n

=

m 34 m

f 4 f.t#F t

f N ge t s f.tet 3 s f.at

n u utre In - 3.7(4 ISS = SJ 6 75 Iso-uem EVENT TREE QUANTIFICATION FOR R2E2 - EVENTS THAT FAIL TWO TRAIN FIGURE 11.

TWO TRAINS OF SAFETY ORADE AC POWER 2I-36

U s

.a, =,

===se

.u.

u

.mm.,

seu a.

c a.uet eu.sm==m.L? Y

.C.*088L. 68 sw.M s.t.L.F

.am-=

n O.90 0.h...a.r k

:,m

• ne ce w.

=== ce

' N UOUEF

,n,t.,. as.

et g

gy q cne. v.

8h*T'.ee on.G8. re w

e.v vikvta vJLW

.sff s.SG ws.7 gu,,s.g.r.7

.e.sr nacu mario=

neu.

,g etas.

eton oc u

os ev

.ni o.

t

=

p,.

=

=

ts un a

n t

=

a es s."'

gm

_.c,

• _ _ _ _c '

t

=

mm

=

u e.t.

.t.

n 19,

=

te m

u t.

wr wr s.

...e

,, m

=

te e

n t.

e

=

u e t.

.. n ie a

t.e g-t g-1

_ _g a

_ _ _ g. t.

.=

te n te it SS t.1 f4 se to 11 St (Jte 18 m J7 e.50 93 90 9 3 s.te es n og 1.

As tJf4 I.9 10 39 9 3 f.t.

3 0 t*19 10 N

26899

.8f4 m -1 m 88 te 97 90 ca t.t.8 3r

=r a

30 n SS 10 00 SJ 8.tse mr SJ 0.te 90 19 H

.to n 37 m.te 30 gg ta t.sg we we

_90 m 30 ft SS B.9 f t.

se to 11 St 03 f.tm

$ 0 f.19 13 33 22 f.te e

_eD e

___.gg gun.

te 33 3 3 f.119

$ $ f-.l

.tsytr3 ISS

• t a t49 IS3

• t a t-Gl Igg e g3 3.gm FIGURE 10. EVENT TREE...Q.UA.N,TIFI.C.A..T. ION, F,O.R R2E1.ac.N,.TS THAT FAIL TWO RHR TRAINS EVE a

ce, r onuce E l -3 7

acy,maton ope sestTt *.ttese mens CPtst :Ti3m - me owe owe owe LAsce opsetfos anstLL svenfft C00tsas8 ACT10'uB f0 poemune cuccit t esiSp 8vCT W Posseve> nesse acticasero esseevn:.viosse 0sse m aa

• N t.t.f

.C. Lose.tA,m,.e.s m..e fU s.k.f,C,L,,o,e,g e, sat-s, 3Sm e

esassavAesses pa.gve,ss? gou.s. sea.nv V asy t.e

.v.4 6Ytf

.=um c

com eto.

n cto W

ym ses.

no CV Lv asc ba 06 as os Oe 3

8888 i

m.

,r.

8 88

'884 g*

g

A, M 9

38

,=9 4

OS 9.7 f.M

'A84 S

M e994 e 99 0

OS 9 f f.it m, a l 80 ir i

88 7

95 9984

,r 1A6 3 e

et 344 9

N 1.1 f4 928 8

... w g..

g..

__g.

_ _ _ g...

m,,..

. e s.m is n

. e4 se u nd m,,.,

m,,.i.

,e a

u e.n n,,.,

se

. e.s n

... e4 ned vee n a

u.

.. :-ei,..

m,, m

_ _ _ c...

.e a

u u.m n,,.,

ee l

o i.n n

u...

m a

n e...

m,,.,

f so si a

ue n m,,

.m I o 125 72

$1 1J f.838 9 D f-21 23 38 2.7 f.'ft to SB 3A f.939 t.11 38 76 m

3.9 f-se s e i et metytts Z SGefA(4 Z St = 4 A l-N Z & = 2 A (-49 EVENT TREE QUANTIFICATION FCR TYPE 3 EVENTS. EVENTS TAKEN FROM N i

FIGURE 12.

THAT ARE APPLICABLE TO SEABROOK STATION Z i -36

t i

i i i i i

i isig i

.... i i_

NuREG osse m

---

SEN8mVRYSTUDY g

FOR EVENTS AT POWER

~

~

~

SHUTDOWN EVENTS

.-...m.mo l

ei 01 :

T 2

5x bh b~~%

x wf

\\

4

\\

\\

50 REM gg

\\

\\

so

\\

\\

oW

\\

g

$$ 0.01 7 g

7

\\

\\

bg l<

1 I

b$

l 5,so Re%

QM/

!g i \\*[

go,lNg 200 REM i,cm '

I

I 0.001 1

10 100 1,000 DISTANCE (MILES)

F,ca.2 / 3 COMP RISON OF 200 REM AND 50 REM 00SE VERSUS DISTANCE CURVES WITH CONTRIBUTIONS FROM SHUTDOWN EVENTS 2I-39 1

......i NUREG 0396 s

sENamvrrystuoy g

FOR EVENT. ATPOWER

~

sauroOWu tvEurs e H.

WITH NO CREDIT FOR EQUIPMENT 4

HATCH RECOVERv o'

~

=

g<

3 l5 F-s

~

mg

\\

\\

\\\\

50 REM 88

%..\\s0ggy

/

~

8@

\\ \\

en oo' i

-\\ \\

bg

\\

\\\\

j<

I

\\\\

\\

a i

u s

gg I

u s

88

/ ~u s

l REM 200

=

"g ]

\\"g

~

l,/

200 REM 0.001 I

10 100 1,000 OISTANCE(MILES)

&})ej t'd COMPARISON OF 200 REM AND 50 REM DOSE VER CONSERVATIVE ASSUMFTION OF NO CREDIT FOR OPERATOR RECOVERY OF O EQUIPMENT HATCH 21-40

RAI 29

The S2W release category isotopic distribution listed in the Table 4-3 in PLG-0465 shows release fractions of cesium and tellurium that exceed the release fraction for noble gases and greatly exceed the release fractions for elemental and organic iodine. Please justify the isotopic distribution of the S2W release category consistent with WASH-1400 source term methodology.

RESPONSE 29 Table 4-3 in PLG-0465 "SEABROOK STATION EMERGENCY PLANNING SENSITIVITY STUDY" contains a misprint for the release f raction of the noble gases (XE) in release category S2W. The XE release fraction for the third puff release (S2W-3) should read 0.23 instead of 0.023.

The total release of XE should also be corrected to read 0.33 instead of 0.123.

The remianing release f ractions for release category S2W have been verified against the original CORRAL code calculation performed in the SSPSA.

The release f ractions for release category S2W have also been compared to the release fractions calculated for release category S6W which has some similar features. S6W represents accident sequence with a dry containment (no injection, no spray) where an 8-inch diameter containment on-line purge valve fails to close while the plant is in the purge mode. Thu s,

in this case a 50 square inch leak path exists at the time when the release of activity from the primary system to the containment begins. For this size leak, the release duration is a few hours.

Release category S2W represents accident sequences where a small leak of 1.5 square inches develops early in the accident sequence, resulting in a release duration in excess of 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />.

The release of noble gases for S2W is approximately one third of that for S6W. This is consistent with the leak size for the two cases.

In S6W the entire release occurs within the 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> period, whereas for S2W only about one third of the release occurs in the first 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />.

The CORRAL code treats organic iodine essentially as a gas and the same ratio is indeed observed for the two release categories.

The release of molecular iodine (I-2) in CORRAL is governed by a direct release f rom the fuel to the containment during fuel melting, wall deposi-tion in the containment and release to the environment. The release of particulates is governed by a direct release from the fuel to the con t ainme nt, aerosol settling in the containment and release to the envi ro nmen t.

Table VII J-l in Appendix VII of W4SH-1400 (page VII-203) gives representative deposition rates for the natural deposition processes without sprays in the CORRAL code.

For molecular iodine the deposition rate is 1.38 per hour. Furthermore, corral models the particulate disposition as a two particle size process which results in higher deposition rates at early times and somewhat smaller deposition rates at later times. However, even the larger deposition rates for particulates at early times are significantly smaller than the molecular iodine deposition rate.

In comparison the removal rate constant for leakage at elevated pressure (choked flow) is about 1.0 per hour for S6W and 0.02 per hour for S2W respectively.

For S6W, the leakage flow is choked only for about 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br /> after vessel breach and therefore the average ef fective leakage rate constant for S6W is only about 0.5.

The CORRAL modelling differences between molecular iodine and particulates and the dif ferences in leakage constants between S2W and S6W explains the differences in the release fractions. For the S2W release of molecular iodine, the wall deposition process rapidly depletes the I-2 concentration before any substantive leakage has occured, yielding a release fraction of 0.007.

The leakage rate for S6W on the other hand is comparable to the wall deposition rate, particularly for the first few hours where the I-2 concentration is still large. This results in an 1-2 release f raction of 0.2, significantly larger than for S2W. This value in about 50% of the particulate release fraction because for I-2 the wall deposition removal rate dominates, whereas for particulates the leakage removal rate dominates. The particulate release f raction for S2W appears somewhat higher than might be expected at first.

However, for S2W the leakage flow remains choked for the entire 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> period and af ter a few hours the ef fective particulate settling rate in CORRAL becomes very small.

The net effect is that particulate settling for S2W only reduces the particulates release fraction to about 50% of the maximum possible release fraction indicated by the noble gas release.

The consequences for S2W have been recalculated with the corrected release fraction for the noble gases. The consequences have increased by the following percentages:

Acute Fatality Increase 0.66%

Acute Injury Increase 0.23%

Latent Cancer Fatality Increase 0.16%

This increase will have no discernible impact on either the risk profiles or on the dose versus distance profiles. Therefore neither the results nor the conclusions of the WASil-1400 sensitivity study documented in PLG-0465 are af fected.

RAI 31

Provide a typical calculation to demonstrate that small diameter penetration sleeves do not punch through the containment wall under the worst pressure condition assumed in the analysis.

RESPONSE 31 A typical calculation is attached.

D i

l I

l l

l

m n/c/ SG X./

L PuHc^Ho ur CR 6LC LvCU T Fi;I IL U e E s^

FOR 7 Y/)E& NL D~

/:J '/'H i Ed ^J 3

~

i Cons:cleepuncleo4 ec 6Jousou /

fa,w,a I

,br e ff 2e piatra%.

A.

PE NE T2i? T IdH i-2G L. EvAt u RATE L. cave 75 Ancinre R nis ldELAS 6ardae 1/-

Insicle Cnmt-cone p

e sf e

,/_ S/eee P

~.

pn= ng N&S f/Of49G Sleeve anchorage syste~ -

S*

Plates & g //

'et/s.

W Eslima fe Loct cf Sleeve DE.

12 /4 (Dm.F805C76) l 3

l fress

&cce f7 2to pu-( = (210) f(12.7Sf g,812.ff, l

l Bje Area.

Sch.

/60 h,h.

f'hg Ar* h h

~ 5969 l

= 3. 5'3 2 0 i

l i

txtvany imposs;Ue) flic voy tso,, S/

cauda //cc, ocuves if fl.g pife is of v/fi<nck /A lenStore in stWe

/b e Confain m enf,anc/ af O/ fem.afe in compress to r 00(Sic l&

(h e Con /Qin m enb.

$n (hiS CR S t, (2-)l S. 5 5 2.) { 7 5 ) =S 2 9. 8 k,vo F,,,pt

=

f I.

Y nin pec. 2.T.S.

' rfe,

x'- s ec t.

0<<a Gn2)

1. c;;mp,,

Sirseyt'l duc do

'/C, L fil/d welc/g af c en h <

cleeve :

(2.)[.7)[5/Ic) 1r 12. 7 5 70

/; 2. 2 7 4,po 1

=

~

~

t i

m (wui) 1 It<;m eter tktd9'~f mm.

RZcLe Sitt fhicat l

nominal sn e

.~. At Iow.,'cle L-nd = L227 Il >= >

l Aeol:e<.0 Lead = E29.9 Il;P:

Addilion o( w eM t'o cafar con /ribufe uen m o ce l

$ (Cn}Y.

YO pbld%

hr*Qnc er'rihJ OW l

/ro s/eere 60 con cate.

m l

full (;n< t frafedet w t/c/ a f collOr as/So ex,p ecfecrl l~

i (ohave enovy t capa cit,y tc' carey any i

i (cac/s on its o wn.

l i

i l

I i

=.

r

?'))

//%/24 3

Cou side r

- W a<

/o~ i tu ve in (ort crete ovf-1 msia w 'd e<d m /.

w /.

[

e--

.gg %~ f u +2(c + D)

T

&~ u.

aia.

0+. con e 7

..e H = O 0.

of aselo 3

l r

4

-\\

[q 9

^

l C y _D g fleSisfie.g /buct Jvt /c c'l u /c. n s, n/ /}rdssord O 'u tcn e-f = p (([b' + 2t'C + O ) t

-N' j oir,,,sa ~ /dy Tak p

/GO psi

- 0. tG L>d l et. <

u,6 e or i swr e fo, r &26 C+D-27i~

(Aa(/' Watt Twx" 5, a ssu ~ < Of)

M 14

=

c] = 0, /G (((7 8)*_ (z4 )' ] <

6 92 40,o_ _ _

2F

= 52c/. O

h. 2).

APPLI EQ I

c a,c, d e A eo,- ces & lo-e a)

= =Js=4 reoct de opf befl~cQ.

i

+0 l

A c kf &, d c e c is i o r' e " ' s P rc'> i cla@, Dy?

l a.lbwel aclibit o$ febarS, l

l l

l

h. Shear :n L'n er i/s'tl' l

\\

c. Jh car in VT sec don s.

l l

i

)

No pu u in the wo a vca i

l i

i i'

I

RAI 38

Quantify the leak areas associated with other containment f ailure modes as discussed in Section 5 of Appendix H to the PLG report #PLG-0300. Also, asses the impact on risk by assuming these failure modes to be type A rather than type B failures including the effect of simultaneous occur-rence of various failure modes.

RESPONSE 38 Assessment of Leak Areas of "other Containment Failure Modes" Discussed in Section 5 of Appendix H.1 in the SSPSA A number of additional containment failure modes were considered in the SSPSA in the probabilistic evaluation of the containment capacity.

These include:

1.

Liner tearing due to friction and adhesion of the liner to the concrete 2.

Hicrocracking due to imperfections in the liner 3.

Impe rfections in the welds 4.

Penetrations not considered in detail 5.

Failure at Electrical Penetrations 6.

Leakage at personnel or equipment hatch Although f ailure probabilities for these additional modes were not calculated explicitly, an estimate of these probabilities was given, and they were assigned to a Type B containment failure (leak area of 2 to 60 square inches).

In the following, the issue of leak areas associated with these additional f ailure modes is addressed.

Failure Modes Invloving the Liner Plate or the Liner Weld (Items 1-3)

The first three categories involve failure of the liner plate or the liner welds.

For a given stress level in the liner, there is a critical crack size, larger cracks will continue to tear open, while smaller cracks do not grow. The critical crack length for the liner is expected to be on the order of 1 inch at the stress levels being considered.

Appreciable leak areas will only form for cracks which exceed this critical size. The probabilities given in Section 5 of Appendix H-1 in the SSPSA may be interpreted as the probability that either the length of a crack smaller than the critical size grows suf ficiently to prevent any further pressure buildup, or at least one crack exceeds the critical size. The following discussion focuses on the leak area that would result if a crack exceeded the critical size.

In all cases, the postulated cause of failure is a tear in the liner.

If the size of such a tear exceeds a stress dependent critical value, the tear will propagate.

It may be arrested when it reaches a weld between two liner plates, or it may continue.

The maximum tear length is given by the entire length of the cylindrical shell. The T-sections which anchor the liner to the concrete limit the width of the tear to about 3 inches.

This leads to an upper bound tear area of 5,400 square inchas.

Based on an estimated 50% probability that a tear will be arrested at the weld between two liner plates, the vertical dimension of a liner plate is considered a reasonable median value for the tear length. The corresponding tear width was estimated from the elastic recovery in the liner, including some allowance for plastic deformations at both ends of the tear. The median tear area thus obtained is 100 square inches.

Assigning a 0.1% probability for crack growth to the upperbound tear area or larger yields a logarithmic standard deviation of 1.3 for the tear area.

If the entire tear in the liner occurs over an open crack in the concrete, the leak area will be equal to the tear area.

In most cases, however, the leak area will be much less than the tear area because liner tears and concrete cracks may not occur at the same locations.

The actual expected leak area will depend to some extent on the cause of failure in the liner.

For case 1, failure in the liner is due to friction and adhesion of the liner to the concrete.

In this case failure can be expected to occur at the location of a crack in the concrete. However, the tear in the liner may propagate away f rom the crack in the concrete. As a result the leak area may still be less than the tear area. How much less is difficult to predict.

In addition, there is a somewhat increased probability that a tear might form in the heat affected zone, where the T-sections are welded to the liner. This is also the most likely location for cracks in the concrete. Therefore, assuming that for this case the leak area equals the tear area is considered to lead to a realistic overall prob-ability distribution for leak area.

For cases 2 and 3, f ailure is due to an imperfection in a liner plate or a liner weld.

Such imperfections may occur at random locations, and in most cases, will not coincide with the locations of cracks in the concrete.

On the other hand, there may be a preference for a tear in the liner to form close to a crack in the concrete, because the stresses are expected to be slightly higher at such locations.

For given tear dimensions and cracks in the concrete, the mean leak area can be calculated, assuming that the tear is equally likely to occur at any point in the liner.

Increasing this mean leak area by a factor of 2 to account for the preference for a tear to form close to a crack in the concrete leads to a mean leak area of 15% of the tear area. An upper bound to the leak area is again the tear area. On this basis a median leak area of 6 square inches with a logarithmic standard deviation of 1.5 was estimated

for f ailure of the liner due to an imperfection in the weld or the liner itself.

Penetrations not considered in detail (Item 4)

Completness of penetrations covered is addressed by the answer to RAI 31 and 39.

Failure at Electrical Penetrations (Item 5)

As discussed at a meeting in Brookhaven, the NRC will be reviewing the Sandia test results to address this topic.

Leakage at the Personnel or Equipment Hatch (Item 6)

The equipment hatch consists of a spherical cap inside the containment with a diameter of 28 feet. The hatch is bolted to a flange with two 0-rings embedded in the flange to stop any flow in between the flange and the hatch.

The sleeve is 3.5 inches thick. 32 swing bolts provide the hatch to sleeve seating force, in addition to the tendency for the containment pressure to push the hatch against the sleeve. These swing bolts are expected to prevent opening of a gap in between the flange and the sleeve even if the sleeve deforms in an out of plane mode.

Ovalling deformations of the sleeve on the other hand can be accomodated either by ovalling deformations of the flange and the hatch, or by some slip on the mating surfaces.

In either case, no substantial leaks are expected to form.

Another possible cause of leakage at the equipment hatch is buckling of the spherical cap.

The ratio of the membrane stress in this cap at 210 psi containment pressure to the buckling stress for a perfect spherical cap (generally known as the classical buckling stress) is 6.4.

However experiments show that spherical caps are sensitive to imperfections and tend to buckle at stresses below the classical buckling stress.

Dis-tortions of the flange have the same ef fect as imperfections. Also, bending stresses associated with such distortions as well as other stress concentrations reduce the ef fective stif fness of the material. This is accompanied by a proportional reduction in the buckling stress.

Finally, elevated temperatures also tend to reduce the effective modulus of elasticity of the material. As a result, buckling of the spherical cap due to the combined ef fect of the direct pressure induced stresses and distortions of the flange is considered to be a credible failure mode.

Such a failure would be expected to lead to a la rge leak area in excess of 60 square inches.

Since this f ailure mode would most likely result in a Type C containment failure, it is not necessary to estimate the leak area.

An upperbound would be substantial fraction of the barrel area. The same assesment of leak is assumed for the personnel lock.

In summary, the leak areas for the failure modes which were not individually quantified in the SSPSA area assessed as:

Median Leak Longitudinal Area Standard Failure Modes (Square Inches)

Variation 1.

Liner / Concrete Friction 100 1.3 2.

Liner Imperfections 6

1.5 3.

Liner Weld Imperfections 6

1.5 4.

Other Penetrations See answers to RAI 31,39 5.

Electrical Penetrations To be based on SANDIA test data 6.

Equipment / Personnel Hatch Large, Type C It is concluded that two of the six f ailure modes in this group would involve leak area which would constitute Type C (gross) containment f ailures, namely Items I and 6.

The definition of Type C failures adopted by Brookhaven National Laboratory in NUREG-CR/4540 for Seabrook Station would only assign Item 6 as a Type C containment failure.

In all instances in the SSPSA, the RMEPS and the WASil-1400 sensitivity study where use could have been made of the Type B,

Type C distinction in containment failure leak areas, the releases were actually treated as Type C (gross) failures. This decision was made because the long delay for the late overpressure failures in release categories S2, S3 and S4 rendered the consequences insensitive to release duration, and because the Type C f ailure assumption is conservative. Therefore, the above shift in 1 or 2 of the failure modes from a Type B to a Type C failure has no impact on any of the results and conclusions in the SSPSA, the RMEPS and the WASil-1400 sensitivity study.

In fact, this statement is true even if all Type B f ailure modes were changed to Type C.

1

RAI 59

An item under consideration for advanced nuclear power plants is the ability to monitor pressure on the low pressure side of check valves.

This could provide early warning of check valve leaks and would pro-vide monitoring capability to help assure check valves were operating properly.

The same monitoring capability with respect to RHR suction line valves could identify if individual valves were mispositioned or malfunctioning. Would such a system for Seabrook be of significant benefit in reducing risk in a reduced size emergency planning zone?

RESPONSE 59 The f ailure model and quantification for leaks greater than 130 gpm is summarized below for the four cold leg injection paths (two check valves

.in series) and the two hot leg suction paths (two MOVs in series).

4 COLD LEG INJECTION PATHS CHECK VALVE NO. OF FAILURE CONTINUOUS PATHS MODEL QUANTIFICATION ( ANNUAL TESTING)

MONITORING 4

21(hT/2) 4((4X10-4)2+ Variance]

3.5 K 10-6 o

=

4 2 )3t d 8X(4X10-4 ) X(2.7X10-4 )

0.9 K 10-6 0.9 X 10-6

=

2 HOT LEG SUCTION PATHS NO. OF FAILURE PATHS MODEL QUANTIFICATION (18 MONTH TESTING) 2 2MXT/2) 2h(4X10-4)2+ Variance)XI.5-2.7 X 10-6 2.7 X 10-6 2

2 )Q(d 4X(4X10-4 )X(2.7X10-4 )

0.4 X 10-6 0.4 X 10-6

=

2 11 g 2X(4X10-4)X(1.1X10'4) 0.1 X 10-6 0.1 X 10-6

=

7.6 X 10-6 4.1 X 10-6 The last column above assumes perfect continuous testing (monitoring) of the two series check valves.

As shown the total f requency changes very little (less than factor of 2).

However, quantification of the other failure modes and treatment of MOV discs as check valve discs are believed to be conservative.

" Perfect continuous" leak testing of the RCS series check valves would require significant modification and is not practical.

Frequent, periodic testing can be performed to verify that excessive leakage is not occurring through the inboard RCS isolation valves.

Excessive leakage through the outboard check will possibly be detected by increased pressure in the RHR system and reduced accumulator level depending on the magnitude of the leakage.

Currently the normal RHR suction motor-operated valves cannot be tested as described above because permanent test lines are not connected to the process lines which the valves isolate.

o In addition, as discussed in the response to RAI 48b, the interf acing LOCA event contributes approximately 12% to the total early release f requency.

Therefore, if this was reduced to zero it would be a minor reduction to total release frequency. There is no significant benefit to reducing risk.

I i

i l

l i

l l

l l

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

RAI 61

What is the f requency of f ailures in the pipe tunnel that is mentioned on page 3-23, and which led the authors to conclude they are very low?

= c.

RESPONSE 61 As noted on page 3-24, top event PI in the VI and VS event trees.

represents any failure of piping or the heat exchanger due to the RHR system high pressure challenge.

Any piping failures are assigned to plant damage state 1 FV.

As Table 4-17 shows, damage state IFV always asps to release category SI-the most severe release.

In other words, any RHR pipe failure is assumed to result in the most severe release, esgardless of failure location; therefore results are insensitive to the pipe failure location.

l

RAI 62

Page 3-27 references. situations where the combined sump pump capacity is sufficient to remove leaks and keep the vaults from flooding.

In these cases, the RHR, SI, and CS pumps are assumed not to be impacted by flood-ing. What consideration was given to failure of one (or both) sump pumps?

RESPONSE

As shown in Figure 3-4 of PLG-0432, the sequences referred to in this quastion have f requencies on the order of IE-8 or less; these f requencies, whsn multiplied by the chance of sump pump failure clearly make such saquences unimportant contributors.

.