Text

Review of IPE Level 2 Draft Repts for Cpses
	ML20116C962
Person / Time
Site:	Comanche Peak
Issue date:	07/16/1992
From:	Torri A; TEXAS UTILITIES ELECTRIC CO. (TU ELECTRIC)
To:	;
Shared Package
ML20116C954	List: ML20116C952; ML20116C962;
References
	NUDOCS 9608010232
	Download: ML20116C962 (105)
	v • d • e

1 REVIEW OF THE IPE LEVEL 2 DRAFT REPORTS FOR THE COMANCHE PEAK STEAM ELECTRIC STATION Reviewed by Dr. Alfred Torri RISK AND SAFETY ENGINEERING ENCINITAS, CALIFORNIA 1

Reviewed for TEXAS UTILITIES ELECTRIC COMPANY DALLAS, TEXAS July 16,1992 9608010232 960726 PDR ADOCK 05000445 P

pop

TABLE OF CONTENTS l

t 1.

G EN ER AL CO M M ENTS..................................

1 II.

S PECIFIC CO M M ENTS..................................

2 i

DOCUMENT A.

PLANT DAMAGE STATES 2

DOCUMENT B.

CONTAINMENT FAILURE CHARACTERIZATION 4

DOCUMENT C:

CET STRUCTURE, FAULT TREE STRUCTURES, AND CET DEPENDENCIES..................

5 DOCUMENT D.

MAAP BASELINE CALCULATIONS 11 i

DOCUMENT E.

CPSES CONTAINMENT PERFORMANCE DURING l

SEVERE ACCIDENTS...................... 12 1

i l

l

)

I,

GENERAL COMMENT

S G1 A number of the comments and observations apply to several sections of the report. For example, comments on basic event probabilities may apply for more than one PDS group, which may be describes in separate sections. Dese comments not repeated, unless there is a different flavor.

Resnonse': No response.

G2 I found it difficult and time consuming to trace the specific conditions under which a basic event probability is actually applied in the decision trees. This is due in part to the repetitiveness of elements of the trees. I recognin that this is a feature of the basic methodology, not the CPSES implementation of it.

l Response: Dat's right. Some experience in applying the method is necessa.y in order to understand

)

the purpose of each BE in the context of the overall methodology, i.e. in order to go from the detail to the big picture.

G3 It would be desirable to apply the stress-strength overlap method to calculate BE probabilities more often and to use the engineeringjudgement method with Table 2-1 only when no other alternative exists. This would require a more formal quantification of the uncertainties in the MAAP calculations or in the ancillary calculations.

Response: That approach was applied as often as it seemed to be warranted. Due to the large size of the CPSES containment, large number of heat sinks and absence of a curb, many early failure mechanisms such as alpha events and rocket mode failures for which the approach is most suited are difficult to quantify. In order to retain these mechanisms for puposes of showing they were indeed considered in the CPSES IPE, their probabilities had to be assigned by engineering judgement.

However, in all cases the judgement is based on NUREG-1150 assessmente for either Zion of Surry, depending on the issue. In a few cases the NUREG-Il50 judgements are adjusted for CPSES using our own judgement and basing it on specific differences between our plant and that being referenced at the time. In all cases the rationale for the adjustments is presented.

G4 The documentation of the accident sequence bascline and the dominant sequence analyses is i

very good and comprehensive.

Resnonse: No response.

G5 I recommend that you develop the frequency and the dominant contributors for the intact containment sequences. You should do this using the CAFTA and GTPROB methodology, and not by taking the difference to the sum of the containment failure sequences. You should check that for each PDS the sum of the frequencies of all the cutsets/ sequences going to all

' Responses are by Dr. Hugo da Silva, TUElectric, author of CPSES LEVEL ll portion of the IPE. Responses as of DECEMBER 1992.

TUL2-CoM.RTU 1

RSE 11/27/1992

I the CET end states is the same as the PDS frequency, i. e. make sure that no frequency is

" generated" and no frequency is " lost" in the CET quantification. I have seen applications of the methodology where this has happened, but I do not know what caused it to happen.

1 Response: This was done once for verification purposes and we found that with CAFTA quantification that problem happens indeed. However, when we use GTPROB the end state i

probabilities all add up nicely to 1.0. That is why we use GTPROB, otherwise we would used CAFTA for the quantification.

]

i G6 You should review NUREG/CR-5282 before considering changes that would reduce the HPME early containment failure probabilities.

Response: I have reviewed NUREG/CR-5282. However, that documer,t has no addional information on probability of the various pressure rises which we are currently using from NUREG/CR-4551.

1 i

4 G7 Numerically, the. greatest impact on the result could be due to comments C51 and C53.

I l

Response: Rose comments have been addressed and implemented. De impact has not been as large

)

as on might expect because recovery is only considered after succesful depressurization, j

1 1

i 1

l I

1 l

4 W

2 TUL2 CoM.RTU 2

RSE 11/27/1992 i

i

II.

SPECIFIC COMMENTS DOCUMENT A.

PLANT DAMAGE STATES (RXE-LA-CPI /0-002)

Al Page 2, bottom: A pressure range at vessel breach (VB) from 200 psia to 2000 psia is very wide for the purpose of sorting out the phenomena at vessel breach. I would recommend either four pressure ranges, <200,200 to 600,600 to 2000 and >2000 psia, or three pressure ranges <600,600 to 2000 and > 2000 psia. De principal distinction between

< 200 psia and 200 to 600 psia is that for the higher pressur3 range the debris can be swept out of the cavity, but without a curb most of the swept of.dris may return to the cavity anyway, and the cavity floor seems to be large enough for debris cooling without dispersal so there is not a large difference in containment response at CPSES.

Response

The intermediate pressure range is wide 200 to 2000 psia only in name. In practice, this range is collecting the larger seal locas which constitute most of the frequency here and some PORV failures to reclose. The seal LOCAs are 250 gpm/pmp leak rates and lead to RCS pressures at VF around 700 psi. The PORVs lead to pressures at VF around 500 psi. Another way of looking at it is to say that if we were to break up the 200 to 2000 range into a 200 to 500 and a 500 to 2000 range it would be found that most of the frequency would be in the 500 to 2000 range because of the large seal locas. Your alternative suggestion to lump the 200 to 500 range with the less than 200 range is reasonable but since the frequencies in the 200 to 500 are not that high it is simpler to leave things as they are. His is theoretically conservative as higher pressure (200 - 500 psia range) sequences are binned with the 500 - 2000 psia range, which is more severe. In practice this is a moot point because the 200 - 500 psia range is comparatively unpopulated.

A2 Table 2-2: The PDS for the SGTR do not distinguish the availability of secondary side heat removal or feedwater makeup. If the tube rupture is below the secondary side water level in the SG, substantial fission product scrubbing would occur. Is this neglected?

Response

Most of the SGTRs leading to core melt at CPSES result from failure to depressurize the RCS where ECCS injection is successfull until the RWST is depleted. Shortly after that the faulted SG inventory above the ruptured tube elevation drains back into the RCS because RCS and faulted SG pressure are essentially the same and the elevation differential forces the drainback. In this situation there is no scrubbing since core damage occurs after the drainback.

A3 Table 2-2: De PDS for containment isolation failure do not distinguish leak size. If the leak size is less than about a 3 inch diameter hole, a much slower release would occur. Are all isolation failures considered large?

Response

Yes. All isolation failures are considered large because the the total frequency of i

isolation failures is small (9.9e-09) and thus it does not warrant making two classes. Note that SGTR and V-sequences are binned separately from isolation failures.

A4 If the containment isolation failure is either the cavity sump or the containment sump discharge line, then the pressure in the containment can push the water on the containment floor or cavity floor out of the containment and convert a wet hbris sequence to a dry debris sequence. Whether this is a significant consideration depends on the probability of this type of isolation failure and how the radwaste system would handle it.

TUL2 CoM.RTU 3

RSE 11/27/1992

1 1

Resoonse:

There are two sets of lines leaving the containment sumps / cavity: (a)

RECIRCULATION LINES AND (b) DRAIN LINES (a) RECIRCULATION LINES:

During normal operation and injection, the containment isolation sump discharge valves are normally closed. Furthermore, they are interlocked with their respective RWST pump suction valves, meaning they cannot be opened without the RWST discharge valves to the pump suction closing, which cannot occur in injection mode or during normal operation, as it would be an error of comission. Therefore this kind failure is not considered an isolation failure. However the possibility of valve leakeage or failure is accounted for as an interfacing systems LOCA in that group of accident sequences.

l During recirculation, those valves are open but check valves and the RWST discharge valves, which must be closed due to the interlock, prevent this kind of isolation failure in this mode.

(B) DRAIN LINES:

l These are normally open air operated valves which can fail to isolate. Such a failure is considered in the containment isolation failure calculation (SYSTEM NOTEBOOK: CONTAINMENT ISOLATION RXE-SY-CP1/1-024C) and correponds to approximately 25% of the total containment isolation failures. 'Iherefore 25% of the containment isolation failures could potentially be of this type.

However, since the total containment isolation failure probability is so low (9.9e-09, 25% of which is 2.5 e 09) a separate PDS or release category is not warranted, as explained in A3 above. The reasons this probability is so low are: (1) two fail closed valves in series have to fail open (2) alarms for failure of isolation have to fail (3) flow is totalized on the control board and failure would also be detectable that way (4) valves and pumps are controlled within a control band.

l A5 Table 2-4: 23 PDS is a lot to analyze separately in Level 2. You could try to reduce the l

number by conservative condensation.

Resoonse:

It is a large number. However, all 23 PDS are analysed. DOCUMENT D "MAAP BASELINE CALCULATIONS" presents those analyses.

l A6 Page 19 to 23, Table 3-1, ete: There is a difference between what the Level I usually calls a small LOCA and what is classified as intermediate pressure in the PDS. For PDS l

consideration, any LOCA size greater than 2 inches in diameter will yield a low pressure l

(<200 psia) at vessel breach. Figure 2-2 of the MAAP Baseline Report shows the pressure at vessel breach versus leak size. During the review meeting we realigned the Level 1 sequence binning to the PDS needs. This now needs to be consistently implemented l

throughout the several reports.

Resoonse:

YES.

A7 Table 3-1 column heading: What is CT (CS?)

Resoonse:

CT is the correct abbreviation for containment spray.

A8 Table 3-1(C): Containment bypass sequences with a small leak area, such as valve leakage inducing an RHR pump seal ISLOCA, are probably much more likely than the large multiple i

TUL2-CoM.RTU 4

RSE 11/27/1992

valve disc rupture ISLOCA leading to a catastrophic piping failure. Have these smaller ISLOCAs been considered?

Response

No they had not been at the time of this review but now they have been. Review of the ISLOCA calculation including your comments has resulted in an increase in the previous frequency of the ISLOCA core melts. This is because while the larger pipe breaks are less likely, recovery actions which exist were not credited. Thus, seal leaks with recovery are expected to have a similar probabilty as pipe breaks without recovery. The smallere seal leaks represent around 80% of the frequency. However, due to that low overall frequency fof the V-sequence (1.2 e-7) all cases are assumed to have the more severe source term and scrubbing is not credited.

A9 Table 3-2(A): All these sequences are low pressure at VB not intermediate.

Response

YES. 'Diat has all been taken care of in the A6 response.

A10 Page 26, last para: If all isolation failures are binned to the same PDS, how are you distinguishing source terms from small and large isolation failures? If it is done with the last CET top event, have you made sure that the correct branching ratio, based on the relative frequency of isolation failures for small versus large penetrations is used?

Response

No distinction is made, as explained in A3, all are considered to be large because the total frequency is so small (9.9e-09).

All Table 2-2: Long-term containment heat removal can be accomplished by either the fan coolers or by removing heat through the RHR heat exchangers. I read this table to mean that only the PDSs with the containment safeguards bin letter "F" have long-term containment heat removal, l. e. spray injection and recirculation means also that the RHR heat exchangers are j

cooling the water on the containment floor. This is the only way that late containment failure i

can be prevented.

Response

YES. Safeguards bins H and E will ultimately result in containment failure, althour,,h the time frame can be significantly long. Fan coolers are addressed as a sensitivity issue and are effective in preventing containment failure (4 out of 4 fan coolers are needed). However, the cases which would benefit most from fan coolers, namely PDS 1H and 3H (with about 45% of the containment failures), don't have them available per Level I.

TUL2-CoM.RTU 5

RSE 11/27/1992 L

1 i

j 1

)

DOCUMENT B.

CONTAINMENT FAILURE CHARACTERIZATION (RXE-LA-cpl /0-004) 1 B1 Page 11, Section 2.3.2,1st and 2nd para: Can the large 48 inch diameter purge lines really I

be open during normal operation? I thought they are only used during shutdown.

Resnonse:

It is open once a week for three hours as a part of the normal ventilation program.

His note is added to text.

B2 Page 13,2nd para,1st line: " flued" head, not " fluid" head.

j Resnonse:

YES B3 Page 29: Based on the containment ultimate strength analysis for the Seabrook plant, a 1%

strain at failure is a rather conservative failure criterion for a reinforced containment structure.

Response

This is good to know.

B4 Page 29: ne equation assumes that the liner and the rebar are at the same temperature. At failure the liner is probably around 350 to 400'F, and the rebar is around 80*F. His means that the liner is at a lower strain than the rebar, and assuming the rebar to carry the full 1 %

strain load is somewhat non-conservative. However, this is far outweighed by the conservatism identified in comment B3, i

Response

No response.

i B5 Page 29: he design pressure P, cancels out of the equation. Why show it?

Resnonse:

In order to illustrate the rationabfor the formula more clearly, i.e. to show that it is an extrapolation formula.

B6 Page 31: The strength values for f, and f,, are probably lower bound values not mean values, so the result would be closer to a lower bound pressure capacity.

Resnonse:

That is correct and it is another reason why the analysis is conservative.

B7 Section 4: The method referenced for ductile liner failure is not generally accepted by structural engineers knowledgeable in this field. Some of the concerns I am aware of are: 1) the liner is always at a higher temperature and therefore at a lower uniform strain level than the concrete rebar structure,2) when the liner tears, the load that was carried by the liner is transferred to the rebar in the section behind the tear. Liner tearing can only yield a stable leak failure mode if the rebar can absorb this added load without rebar failure.

Response

Structural engineers knowledgeable in this field developed the method for EPRI.

EPRI has sponsored many studies to derive this approach and EPRI stands behind their findings. As a practical matter, in terms of partioning containment failures into rupture or leakeage the assumption we make is that if pressurization is quasi-static such as by steam or non-condensibles there is a 99.5%

probability that it will be by liner tear, which is a leakeage failure mode. If pressurization is rapid such n in HPME and/or burn the split is assumed to be 50%. If the pressurization is in the TUL2 COM.RTU 6

RsE 11/27/1992

l structure's dynamic response time frame such as alpha, the rupture mode is assumed to be certain.

Response

Dere are two separate arguments which should not be confused with each other. One is that leakeage occurs before rupture. He other is that the onset of the occurence of the prevailing failure mode taken at a lower pressure is a conservative position. The argument of page 63 is only that taking a failure at a lower pressure limit is conservative. De relative position of the failure r

modes, with leakeage occuring first is established elsewhere. He fact that leak occurs before rupture, is a result of the EPRI studies as well, and has nothing to do with selection of property values. In any case, applying best estimate property values to obtain the rupture strength would yield a value which would be higher than the best estimate property liner strength, so the leak mode occurs j

before the gross rupture mode if we use consistent sets of property values in both analyses. It does i

not make sense to use conservative values in the liner rupture analysis and then use best estimate values in the rupture analysis as suggested above. Finally, comparing best estimate property liner tear strength to conservative property liner tear strength alone does not prove that liner tear occurs before gross rupture. However, when both rupture and leakeage best estimate strengths also show the same relative position the argument tl at liner tear occurs before rupture can be made with even more confidence.

B9 Figure 21: The probability distribution for the containment failure pressure should express the uncertainties in the data and in the analysis method used. Hat means that the calculated value is more like a 10 %-ile and not a mean, and the standard deviation should be significantly larger. The stress data alone represents a 10% variation and the value used is 13% lower than the mean. In addition, the very approximate analysis method should yield j

another 10 to 20 % contribution to the coefficient of variation. Therefore, the COV should be more like 0.2.

Response

The basis for the 10% value mentioned above is purely judgemental. Furthermore, if conservative property values are used, which are lower than the mean by say 13%, that does not make the uncertainty at least 13%. On the contrary, it rather reduces the uncertainty of failure at pressures below the value calculated with the conservatism. Hence, the basis for the 0.2 COV is also purely judgemental and appears to be counter intuitive. In any case, a rigorous statistical treatment for developing a fragility curve is out of the scope of the IPE. The strength estimate is a best estimate and hence assumed to be a mean value. The 7% coeficient of variation is based on structural engineering experts judgements, namely:

Fardis, M. N. A. Nacar, and M. A. Delichatsios, Reinforced Concrete Containment Safety Under Hydronen Exnlosion Loading, Massachusetts Institute of Technology, Department of Civil Engineering, NUREG/CR-2898, September 1982.

B10 he leak areas associated with the leakage failure modes should be discussed.

Response

The leak areas associated with leak failue modes are such that containment pressurization would be arrested and very slow depressurization would follow. He size of such areas could vary depending on whether the pressurization is by steam or non condensibles because the TUL2-CoM.RTU 7

RSE 11/27/1992

pressurization rates are substantially different. M AAP calculation to support DOCUMENT E indicate that these leak areas about 2 to 3 inches in ec,utvalent diameter.

1 TUL2-CoM.RTU 8

RSE 11/27/1992

DOCUMENT C: CET STRUCTURE, FAULT TREE STRUCTURES, AND CET DEPENDENCIES (RXE-LA-cpl /TBD)

C1 I find it confusing to have the generic EPRI CET and logic trees up front in this report, to find out that the tree is modified and adapted to CPSES only much later. I recommend to move the EPRI CET and logic models to an appendix, if it is needed at all, or simply reference the EPRI report. I would show the CPSES CET and logic trees where the EPRI CET and logic models are now. All the discussion should be pertinent to the CPSES model that is actually quantified.

Resoonse:

Agreed.

C2 Page 7, Item 2: Only (bypassed) or (bypassed or not isolated)?

Response

BYPASSED OR UNISOLATED.

C3 I understand why under the Generic Letter rules operator actions to recover injection before VB are not credited if they are not formalized by procedures and operator training. However some " recoveries" are automatic. For example in sequences where HPI is failed and a hot leg creep rupture occurs, the LPI system can start injecting automatically and recover the debris in vessel. Has this type of non operator recovery also been neglected?

Response

YES, for three reasons:

(1) They are not statistically significant to the CPSES Level II analysis because the affected fraction of high pressure events PDS would reduce by little the late containment failures.

(2) The early failures are unaffected by this recovery, since they can only occur after the risk of early failure has already been reduced by depressurization.

(3) The extent of core damage that will have occurred at the time of a hot leg failure for example is such that recovery in vessel is not considered likely, i.e melt progression is judged to be too far along.

However, the automatic injection of LFi after vessel failure is considered when dealing with debris coolability (BE SNOLPI), but it was also found to be statistically insignificant. See question C28.

C4 Page 14, 8th line and Figure 8-4, VF: It is possible and maybe even likely that with debris in the vessel bottom, both ex-vessel cooling and in-vessel covering of the debris with water is necessary to prevent vessel failure, unless the water level on the outside covers the side of the vessel to a sufficiently high elevation in order to absorb the radiative heat flux from the debris surface on the inside of the vessel.

Resoonse:

At TMI vessel failure did not occur, although there was no external cooling of the vessel. If water is present over the debris inside the vessel, coolability of the debris in vessel is determined by issues such as crust formation, ammount and composition of debris, etc... to a greater extent than whether there is also water on the outside of the vessel. By no means is it required that there be water inside and outside to obtain in vessel debris coolability.

C5 Page 16: CFE should include consideration of hydrogen burns before vessel breach.

Resoonse:

It does. BE PRWCP-PULT & PRDCP-PULT represent the convolution of pressure TUL2-CoM.RTu 9

RSE 11/27/1992

- ~ -

- - - ~. -

rise due to HPME with the containment fragility curve. The pressure rise cumulative probability distribution for the HPME is from NUREG/CR-4551 for Zion and includes effects ranging from blowdown only, to burn only to DCH to DCH plus burn. In addition to that approach the issue is considered outside the tree because, the maximum ammount of in-core generated Hydrogen, which i

becomes available just after vessel breach is such that burns (see discussions for basic event PRPRHB2 under CFL for each PDS) where only in-vessel and no CCI Hydrogen is available cannot cause containment failure. His implies that burns prior to vessel failure cannot cause vessel failure and therefore we omit the question at this time. See also the discussion in the Level Il submital in the sensitivity studies section and in CIS below.

I C6 Figures 8-5 (CFEI) and 8-6 (CFE2): How does this logic tree determine whether DCH, ex-vessel steam explosions or hydrogen burns occur? It seems that the basic events PRCP-l PULTL, PRWCP-PULT and PRDCP-PULT should have been developed further to address j

the possibility of these events, because they each may or may not occur, and some events may occur in combination with others. I realize from our discussions that the quantification assumes that DCH occurs in HPME sequences, and that may be conservative.

j

Response

EX-VESSEL STEAM EXPLOSIONS are considered in basic events (BEs) PRALPHL j

and PRALPHAH. The probability of occurrence and the probability of containment failure given the occurence of the steam explosions are combined into that same BE. In other words, ALPHA EVENT 4

OCCURS means a steam explosion occurred and failed the containment. In view of the uncertainty j

associated with this event and its low probability it was not felt necessary to break up the BE into an i

occurrence part and an effectiveness (i.e. causing failure) part.

FOR HYDROGEN BURNS ALONE, as discussed in C5 and C15, it is found that for CPSES, burns right after vessel failure, i.e where only Hydrogen from Zircalloy Oxidation is available cannot fail the containment. Nevertheless, blowdown forces, Hydrogen burns alone and in combination with DCH are considered in PRWCP-PULT and PRDCP-PULT, where the HPME pressure rise fractiles f

of TABLE 2-2 include all HPME effects. Herefore, this event is included in the tree for CFE1 AND CFE2. TABLE 2-2 reflects NUREG-4551 considerations for HPME ranging from blowdown j

only to burn only, to DCH only, to DCH plus burn and includes the probability of occurence.

1 C-7 Page 18, middle: He pressure rise at HPME/DCH can also be augmented by hypergolic recombination of hydrogen (a forced and complete burn not limited by flammability limits) because of the very high temperatures caused by the direct heating of the atmosphere.

{

Response

Yes it can and that effect is considered in the probabilities for the various pressure rises taken from NUREG/CR-4551. Furthermore, in DOCUMENT E, MAAP calculations are made forcing this phenomenon to occur simultaneously with DCH itself and only then is the CPSES j

containment actually challenged. A line is added to the text here for completenes. Furthermore, as j

mentioned in C6 above, the DCH pressure rises of TABLE 2-2 include the effect of a simultaneours burn.

1

)

C-8 Page 20, top of 2nd para: Hypergolic recombination is not subject to flammability limits.

i

Response

ADDED the following: In cases where and if the temperatures are high enough for l

hypergolic recombination, then the usual inerting flamability limits do not apply.

C-9 Page 22, end of 1st para: Recent experiments at SANDIA of HPME into wet Zion like

]

cavities have shown highly energetic behavior (NUREG/CR-3916).

TUL2-CoM.RTU 10 RSE 11/27/1992 l

.. ~,

_m e

I l

l

Response

Our logic trees consider the possibility that steam explosions can fall the containment.

Response

He discussion here is generic. Section 3 has elements of the discussion you are talking about. In particular, TABLE 3-3 has an estimate of how much debris in terms of mass can be expected and the associated debris bed thickness. If the debris were expelled from the cavity it would likely be well dispersed because of the high gas flow rates required to entrain the debris out of the cavity. For that reason the CET has been changed to show that if the containmentfails early then the debris must be coolable. See also discussion in C49.

l

.C-11 Figure 8-1 and logic tree CFM2: He CET considers early leak containment failures. It does not consider the possibility that a late hydrogen burn could turn the leak failure into a gross failure.

Response

CFM2 is a containment failure mode associated with early containment failure. Bus, it will be mostly due to DCH and simultaneous burn. If the initial cause for the failure, which is usually DCH and a burn, did not cause a rupture, it is not likely that a future burn without additional l

hydrogen would, by itself, do it.

1 l-Another way oflooking at it is that the BE PR-RUPWCFE includes the probability of a future burn converting a leak into a rupture, where this incremental probability is negligible, since the containment's response to a challenge more severe has already been a leak mode failure once.

C12 Figure 8-1: Recovery of injection systems (REC) is not asked for high pressure sequences.

Since only SBO recovery of offsite power is addressed in this top event, its success would seem equally likely for sequences with or without hot leg creep rupture.

Response

Yes but the ECCS cannot get into the RCS at 2250 psia due to an alignment set which sends CCP water back to the RWST if system pressure is near 2250 psia.

C13 Figure 8-1: CFE1 is applied to both recovered and unrecovered sequences with vessel breach. For recovered sequences the steam spike can be much larger, because the entire core is quenched, whereas without recovery only the debris discharged at vessel breach is quenched. Derefore, CFEl(Rec) should be larger than CFEl(No Rec).

i

Response

De pressure spike associated with quenching the entire core is on the order of 15 psi for CPSES. Herefore, it is not worth it to make two events out of CFE1 by reducing the spike in the case of a partial discharge of molten material, since the containment failure probability associated with the upper bound discharge is already insignificant.

l C14 Top event DC is not asked for sequences with low pressure, rec failed and CFE failed, and it is assumed that the debris is not covered. His path includes both injection failures and TUL2-CoM.RTU 11 RsE 11/27/1992

l recirculation failures. For recirculation failures the debris is covered and cooled long term.

He subsequent top event FPR(4) assumes that there is CCI. His is ok for injection failures since the debris will uncover fairly soon, but it seems conservative for recirculation failures.

Response

Along the low pressure, no recovery, early failure path the question is not asked and the debris is assumed to be coolable. The reason is that the only possible cause for the early containment failure along that CET branch at CPSES would be an alpha event. If that happens (and that probability is already very low, i.e 8e43) it is further virtually certain that the debris would be so finely distributed in the containment that it would not get hot enough to cause core concrete attack even if it were not covered by water. As a practical matter, in restrospect it is felt that the debris j

coolability issue need not be addressed at CPSES in cases where the containment fails early, since the only containment failure modes are HPME and alpha. Therefore, they have been eliminated from the l

CET and all early containment failures are assumed to be coolable for the reasons given above. Non-5 coolability is be addressed outside the tree as a sensitivity issue.

C15 CFL1, CFL3 and CFL5 logic tree and Section 2.6.2: There are two hydrogen burn issues, a hydrogen burn occurring early and a hydrogen burn occurring late. An early hydrogen burn i

should consider any burn up to about 3 to 5 hours5.787037e-5 days 0.00139 hours 8.267196e-6 weeks 1.9025e-6 months after vessel breach, because if this burn l

fails the containment, it causes an early release. His should inc!ude any hydrogen and CO i

generated as the debris is quenched in cases where the debris is cooled. A late hydrogen burn would occur many hours after vessel breach, mostly driven by the accumulation of additional 1.

H2 and CO from the CCI, which extends over many hours if the debris is not cooled. For the same reasons, a hydrogen burn failing the containment at the start of CCI would be an j

early containment failure. A hydrogen burn after extensive CCI could be worse, because it has more H2 and CO. De tree structure does apply to late hydrogen burn after CCI 1

generation of H2 and CO. Events IHB2-FAIL,3HB2-FAIL and 5HB2-FAIL should be i

redefined as " LATE H2 BURN" instead of "H2 BURN AFTER VESSEL BREACH" and event PRPR2 should be redefined "...TO LATE H2 BURN... Instead of "...H2 BURN AT l

START OF CCI...". He three events xHB2-FAIL, PRPRHB2 and PRPR as defined now should be transferred and added to the CFE logic tree. His comment applies to all fault trees for late containment failure.

i 5

2

Response

I agree that there are two times for the burns: (a) early and (b) late.

)

1 (a) early burn:

The case of the early burn has been discussed in C5 for burns prior to vessel failure and it was pointed out there that these burns cannot fail containment at CPSES. The case of early burn at vessel j

failure, as pointed out in C5 and C6 and as demonstrated in the discussion of Section 2.6.2, j

Hydrogen burn alone, without DCH, when only the Zircalloy-generated Hydrogen is present, cannot fail the containment. De discussion of Section 2.6.2 already considers 500 Kg of H2 produced in the oxidation a reaches a maximum final pressure of 77 psi in the containment. Another calculation i

discussed in the Level II submittal for an AICC of 1000 Kg of H2 (100% clad oxidation) leads to a i

final pressure of 100 psia, where the containment probability of surviving a load of 105 psia is 0.999.

l Furttermore, when you were here you calculated the maximum pressure in the containment from the j

complete combustion of all hydrogen resulting from 100% clad oxidation at around 80 psi. Thus, a 4

burn involving only Zircalloy-produced Hydrogen can't fail the containment. This burn question is j

asked under the CFL event, i.e. it is the xHB2-FAIL (x= 1,2,3,4,5) top event because it could i

involve additional Hydrogen generation in the cavity before CCI. This Hydrogen would come from j

further oxidation in the cavity and would occur a few hours after vessel failure. However, because it j

has already been verified that all Zircalloy-Hydrogen cannot fail the containment the question is moot.

i TUL2-CoM.RTU 12 RSE 11/27/1992 1

d Finally, if the Hydrogen burn at vessel failure did not fail containment with all the violent pressure spikes that occur then, it is not likely to do so shortly after, as things settle down in the containment.

Therefore, for these reasons we do not place the HB2-FAIL event under the CFE top event but rather under the CFL top event. By the way, the 3 to 5 hour5.787037e-5 days 0.00139 hours 8.267196e-6 weeks 1.9025e-6 months after vessel failure for separating early versus late is excessive at CPSES. His time frame is to allow for evacuation after a General Emergency is declared. We feel 2 hours2.314815e-5 days 5.555556e-4 hours 3.306878e-6 weeks 7.61e-7 months is sufficient since by CPSES procedures a general emergengy is declared very early in severe accident progressions. In summary, the only way Hydrogen burns can cause i

early containment failure at CPSES is if they occur simultaneously with DCH during HPME. Rose l

case are considered in the CFE tree under BEs PRWCP-PULT and PRDCP-PULT.

(b) late burn:

l The late burn is impossible for CFL1, CFL3 and CFL5 because (a) these events are on the path of a coolable debris, i.e. steam overpressurization which inerts the containment or (b) in this path the debris is cooled, thus additional hydrogen is not produced and it has already been tested in PR-HB2 l

whether the Zirconium oxidation Hydrogen could fail the containment. In the case of the non-l coolable debris or CCI path, the late burn issue is addressed in HB3-FAIL also under CFL2, CFL4 i

or CFL6 as it should be.

C16 CFL2 fault tree, event 2BMMTl: Basemat meltthrough should be an independent top event L

in the CET, because the source term is very different from an atmospheric leak containment failure source term. Equatin;, basemat failure source term with atmospheric leak source term is conservative, except for liquid pathway component.

Response

I agree with your assessment of conservatism. The conservatism does not affect CPSES much because basemat meltthrough is not a likely event at CPSES, which due to its large cavity area (70m2) is more likely to fail by non condensible gas generation in dry sequences.

C17 Fault tree for CFL2, CFL4 and CFL6: De portion of the tree under event 2HB2-FAIL, 4HB2-FAIL and 6HB2-FAIL should be deleted here and moved to the fault tree for early containment failure (CFE). A hydrogen burn failing the containment within a few hours of vessel breach would be an early containment failure compared to a late overpressure failure, l

which typically occurs 15 to 30 hours3.472222e-4 days 0.00833 hours 4.960317e-5 weeks 1.1415e-5 months after vessel breach. Tree structure already addresses late hydrogen burns under CCI related effects. See also comment C15.

l

Response

CIS response addresses this point as well.

C18 FPRI fault tree: The labels for containment spray should be reversed.

Resnonse:

YES. DONE.

i C19 All FPR fault trees: For late containment failure the "No fission product scrubbing" also requires that long term aerosol settling in the containment after vessel breach fails. This distinct on, which differs from the early containment failure, is not made.

i l

Response

Aerosol settling is not considered as a question because it always occurs in the late failure sequences. "No fission product scrubbing" does not imply that settling did not occur.

I C20 PRHLSLOKI for PDS 1 and 2: His value of 0.95 covers a wide pressure range from 200 l

psia (or 400 psia) to 2000 psia. For the low pressure end the probability is probably even TUL2-CoM.RTU 13 ase 11/27/iss2

\\

i closer to 1.0. At 2000 psia and stainless steel hot legs the probability should be in the range of 0.3 to 0.5, according to the Seabrook analysis. If the hot legs are carbon steel, the probability should be slightly lower. For the stuck open PORV cases and for the large seal

)

LOCA (about 400 psia with repressurization),0.95 is ok.

Response

As discussed in Al, while the actual possible range is wide, most small breaks are either large seal LOCAs (250 GPM/PMP or about 1 inch) resulting in pressures at vessel failure around 700 psi or a stuck open PORV (1.6 inch) with vessel failure pressures around 500 psi, so 0.95 is ok.

C21 Operator opening PORV on high core outlet temperature (> 1200*F) should be identified as an accident management issue for avoiding DCH potential.

Response

It already is included in the procedures for inadequate core cooling, the FRH series. I have also marked it as an issue for accident management.

If recovery after operator error to switch to recirculation is considered in the future, the l'

C22 probability would have to be evaluated conditional upon whether there already has been an e

operator error, i.e. that the operators have misinterpreted the sequence. With a previous operator error, the probability of recovery success would be considerably smaller.

i

Response

Yes, that is why we are not doing that, i

C23 Late recovery of power in an SBO is only applied to the fraction of the SBO that has a seal i

LOCA < 0.6 inches. This could be very conservative, i.e. most SBO sequences may not I

actually have recovery of power credited.

I l

Response

Based uppon the seal LOCA analysis it is expected that 70% of the induced seal i'

LOCAs will be less than 0.6 inches for the fast station blackouts, which are about 29% of the station l

blackouts and 71% will be larger seal LOCAs for the protracted SOBS. Thus, it is true that we don't benefit as much from recovery as we are entitled to. However, station blackouts have already had j.

their frequencies significantly reduced by Level I reocvery actions. As a result, very little recovery i

potential is left to the back-end because the recovery curve is nearly flat in the time interval of interest.

t' i

C24 SNOSPRAY1 for PDS lE: Spray injection will stop before VB. Therefore this should be reassigned to 1.0.

Response

The question is whether spray injection will Ean before vessel breach, in order to assess water conditions in the cavity. See CAV-DRY event in the DCI logic tree. However, for SNOSPRAY2 IE and SE you are correct and the values have been changed to 1.0.

C25 This is a nice analysis of the split fraction for containment failure. You should check to see whether the NUREG-4551 distributions for the pressure rise are based on the assumption that DCH always occurs, or whether some credit is given for the possibility that the pressure rise is due to blowdown only. If the underlying distribution assumes that DCH always occurs, then the CPSES result may be conservative. A way to account for this would be to give some weight to the MAAP calculated pressure rise with and without a forced hydrogen burn, for example:

TUL2 CoM.RTU 14 RSE 11/27/1992

9 1

NUREG-1150 weight 0.50 MAAP without hydrogen burn weight 0.25 MAAP with forced hydrogen burn weight 0.25 l

l l

Response

De NUREG distributions considers all the mechanisms from blowdown only, to burn only, to DCH only, to DCH and burn and includes the probability of occurrence as well.

C26 nis may be a misinterpretation of the expert opinion data. My interpretation of the statement that a rocket event is less likely than an alpha event includes the probability of a circumferential failure, causing a missile that fails the containment. This would be consistent with the interpretation of the alpha event probability. Therefore, PRROCKET would be 0.001.

Response

De interpretation that you see in the text is not mine, but rather it is that of the experts which developed the EPRI methodology and includes the EPRI review. They stand by their interpretation.

C27 I agree with the magnitude of the debris cooling probabilities, but I think they should be backed up with some estimates of debris depth in the cavity. My rough estimates are:

Qsg Debris Deoth Debris of 25 % of core in cavity 3 inches Debris of 50 % of core in cavity 6 inches Debris of 75 % of core in cavity 9 inches Debris of 100 % of core in cavity 12 inches With debris from 50% of the core discharged at VB and half of that dispersed out of the cavity in the EVSE, that leave a debris bed of 3 to 4 inches, for which a 95% probability of coolability by overlaying water is reasonable. However, the debris dispersed out of the cavity is only coolable, if the RWST has been injected or if the debris spreads into a thin layer of less than 1 or 2 inches, so it self-cools by radiation and convection heat losses to the atmosphere and conduction to the floor without CCI.

Response

Estimates of debris bed depth can be obtained from TABLE 3-3. They would be similar to what you have outlined above. Specifically,100% would be 18 inches, because that table considers the core barrel and upper internals as well. In our analysis, debris coolability requires water and a coolable geometry. This question only addresses the geometry or configuration issue.

C28 This seems an unnecessarily conservative assumption for sequences where the vessel is not depressurized and the high pressure injection has failed. De cut sets for these PDS (IH and 3H) allow a direct determination whether the LPI system would start to deliver water to the cavity automatically after VB.

l l

Response

The cutsets for 1H and 3H have been examined to determine this SNOLPI for those l

PDS. That has been determined as (1.0 - 2.le-5) for 1H and (1.0 - 6.4e-4) for 3H. These values are l

TUL2-COM.RTU 15 RSE 11/27/1992

probabilistically insignificant, i.e. results are effectively the same using 1.0 as has been done.

C29 His discussion and approach for a slow pressurization arrest recovery is good, but actually taking credit for this recovery in the numerical analysis is likely to require a commitment by TUE for an accident management activity to analyze the dominant steam pressurization sequences, identify recovery possibilities and develop TMC guidelines to implement these recovery actions. It may be strategically better not to take credit for these recoveries and to commit to an accident management investigation if the late steam overpressure failures turn up to be uncomfortably high.

Response

Perhaps. However, the approach is conservative and extremely useful. The usefullness of the approach comes from quantitatively crediting the fact that wet sequences are preferable to dry sequences even when dry sequences only lead to extremely protracted containment failure times. The advantage is that a wet sequence is guaranteed to respond to cooling water introduction where dry sequence may not. This approach brings out the benefit of the wet sequence in a way which is easy to see. A committment of the sort you mention is inherent in our participation in an accident management program, but even without that, given common sense and the tim involved it is hard to question the approach.

C30 In sequences with RWST injection, the cavity does not dry out before containment hilure.

i l

Response

That is true but a non-coolable debris bed could still form even with and overlying pool. This possibility of formation of an insulating crust is what is being considered here.

i C31 Page 90, top: Items 1) and 2) are early containment failure and should be asked in the CFE logic trees.

Ramonse:

No. This has already been discussed in connection with items C5, C6, CIS, C17.

4 t

i i

C32 Table 24, Early, Spray always on: This is what happened at TMI-2. The probability of a burn should not be considered so unlikely.

Response

The probability of occurrence of a burn has been increased. However, the probability i

that it would fail containment is still zero as discussed in C5 and C6, so the point is effectively moot.

However, TABLE 24 has been modified to read "likely" instead of " highly unlikely".

C33:

Table 24: The logic for hydrogen burns is hard to follow. What is labelled early and late all contributes to early containment failure. This table seems to argue that different burns occurring at different times during the early time frame can reduce the likelihood of a big burn failing the containment early. This may be true, but where is the probability that these

" mitigating mild burns" do not occur considered? It seems hard to me to try to do that in a logical and complete way without developing the logic as an expanded logic tree.

i Resnonse:

The table is just to list the various possible H2 burns and to help situate them in the CET. He actual assessment of a value to a given BE probability is made in the text with a J

i discussion for each BE and for each PDS. Also the BE discussion refers to the table to point out which kind of burn is being quantified. He table explains how each burn is treated in the logic trees j

(LTs) but it is neessary to go from the BEs to LTs to the table in order to understand the connection.

j Using the table by itself without following the text is confusing, but that is not its intended use. It is TUL2-CoM.RTU 16 RSE 11/27/1992 i

1

1 i

also not intended to make the point mentioned in C33 above nor does it use that assumption at all.

The table is merely a road map.

i i

C34 Page 90, SACSPREC-lE = 1: These are SL sequences with injection failure, and containment spray is available in the injection mode only. It takes about 2 hours2.314815e-5 days 5.555556e-4 hours 3.306878e-6 weeks 7.61e-7 months to vessel l!

breach and the time frame considers hydrogen burns until 3 to 5 hours5.787037e-5 days 0.00139 hours 8.267196e-6 weeks 1.9025e-6 months after vessel breach. If at any time in these 5 to 7 hours8.101852e-5 days 0.00194 hours 1.157407e-5 weeks 2.6635e-6 months containment spray comes on it depletes the RWST in about I'

O.5 hour5.787037e-5 days 0.00139 hours 8.267196e-6 weeks 1.9025e-6 months . Derefore. for more than 90% of the time frame considered here there is actually no containment spray. How has this been accounted for?

f

Response

This BE is used for evaluating the probability of occurence of a burn. If sprays actuated even if only for a short period the atmosphere is considered to be de-inerted and a burn is possible. It doesn't matter that sprays are not operating throughout the entire time frame.

l Nevertheless, it should be noted that SNOSPRAYl-lE has been corrected (C24). Also SACSPRECIE=0 AND SACSPREC5=0 these are complements of SNOSPRAYl.

C35 Page 91, Sprays available: Where is the probability of 0.05 actually used in the logic tree?

j

Response

Nowhere. See C36.

l C36 Page 91, Sprays unavailable: As I read the logic tree, this condition is not considered under HB2-FAIL. He hydrogen concentrations should be significantly higher than the 2%, because at this time almost all the RCS water is in liquid form in the cavity and on the containment j

floor and not as steam in the atmosphere.

Response

That is correct. Both the condition discussed in C35 (sprays available) and this one (sprays unavailable) are discussed for completeness but the worst case is the case where sprays are recovered and that is worst case and its probability (0.995) is what is used.

l As to how much steam is in the atmosphere a hand calculation in DOCUMENT A page 11 determines that ammount as 1.24E5 Kg instead of the 1.95ES Kg used here. That moves the hydrogen mole fraction up to 2.5% from 2%: not significant.

C37 Page 106, Table: Where does Figure 2-3 come from? It must take credit for some form of recovery to arrest containment pressurization. According to the generic letter 88-20, the NRC will assume / require that a formal procedure and operator training are in place for all operator 3

i actions (including recovery actions) for which the model quantification takes credit. The only passive mechanism for this would be heat conduction through the containment wall, and that would be insufficient for many weeks.

j 1

1 i

Response

The derivation, rationale for and purpose of FIGURE 2-3 is discussed in pages 78 through 83. The figure itself is referenced in page 82. All we are taking credit for is the assumption made in NUREG-ll50 for Surry that if the containment had not failed in 72 hours8.333333e-4 days 0.02 hours 1.190476e-4 weeks 2.7396e-5 months in a wet sequence there was a probability of 95% that it would not fail. That curve is merely a way of interpolating that probability for other times, say 70 hours8.101852e-4 days 0.0194 hours 1.157407e-4 weeks 2.6635e-5 months or 48 hours5.555556e-4 days 0.0133 hours 7.936508e-5 weeks 1.8264e-5 months . The approach is important and I don't want to scrap it for the reasons discussed in C29.

C38 Page 106, bottom: I could not find the analysis in Reference 3, but I question the statement.

j After all the RCS and accumulator water boils off, the decay heat from the debris is radiated

^

to the containment atmosphere an the pressure continues to go up, but more slowly, while the TUL2-CoM.RTU 17 RSE 11/27/1992

I I

l temperature increases more rapidly. As the liner temperature and the inner tendon l

temperature increases, the containment failure pressure decreases. Eventually it has to reach

{

the containment failure pressure.

Response

I stand by the statement. Before the mechanisms you mention take place, CCI will occur. The probability of a steam overpressure failure involving only RCS plus accumulator inventory, i.e. If the RWST is not injected into the containment, is nil. In these sequences the failure will come after cavity dry out, will come after CCI, and will be driven by non-condensibles generated in the CCI on top of the initial high steam pressure. Dat will happen with a probability of i

one.

C39 Page 107 bottom,108 top: Considering the redefinition of the small LOCA size for Level 2 binning, the RCS residence time is not so short. Furthermore, no fission product retention is

]

desirable if the alternative is likely revolatilization later. Derefore, the probability of J.

QRCS-RET = 0.95 may be optimistic.

i -

Response

. In terms of diameter we have gone from 4 inches to 2 inches. That is not a l

sufficiently large difference to suggest that fission product retention is significantly more likely.

[

Derefore, we keep the value.

C-40 Page 110, top: Considering comments B7 and B8, I would consider the value of j

PR-RUPWCFL = 0.995 optimistic. Furthermore, late containment failures due to a late j

hydrogen burn should be treated as a likely rupture failure as in the early containment j-failures.

}

Response

Considering the response to comments B7 and B8 we stand by the 0.995 value l

because it results from the findings of the EPRI containment structural studies referenced in DOCUMENT B. As you seem to imply in your question, those EPRI studies also document that j

there is a static and a dynamic behaviour of the structure depending on the nature of the challenge.

j De dynamic response would most likely be a rupture type failure and the quasi-static mode would i

most likely be liner tear or leakage, ne dynamic or rupture pressure mode threshold is related to the j

speed of sound in the structure. A value of 500 psi /sec is quoted as the threshold for dynamic behaviour. Hydrogen burns involve pressure spikes on the order of 50 psi over tens of seconds. This l

is two orders of magnitude below the requirement for the structure to respond dynamically.

nerefore, a burn is also a quasi-static load as far as the containment is concerned. Only alpha 4

failures are considered dynamic. The rocket failure is considered a rupture because it is a missile-l type failure, not because of a dynamic or quasi-static overpressure response, in early failure cases a

{

50% rupture /leakeage split fraction is used because of the uncertainty associated with HPME rate of j

pressure rise. However, that is felt to be conservative because those rates are close to those of burns which are not dynamic challenges.

C41 Page 112, CFE: The probabilities for early containment failure may increase somewhat when hydrogen burns up to 3 - 5 hours5.787037e-5 days 0.00139 hours 8.267196e-6 weeks 1.9025e-6 months after vessel breach are moved to the early containment failure logic trees.

Response

Not so for the reasons discussed in C5 and CIS responses.

TUL2-CoM.RTU 18 RsE 11/27/1992

}

i l

C42 Page 115-116, CFL = 0 for IF and 2F: With sprays operating, there should be a non-zero 4

contribution from late hydrogen burns.

l l'

Response

In a generic sense that would be possible and it is considered in HB2-FAIL.

)

i However, in the case of CPSES this is impossible for the reasons discussed in C5 and CIS, namely that burning of Zircalloy oxidation-generated Hydrogen alone, even 100%, cannot cause containment failure.

i C43 Page 110 - 127: Good discussion of results.

l

Response

No response.

C44 Page 147 and Fig. 8-2: If an SRV sticks open, the logic tree transfers to the right side, still questioning hot leg rupture, but presumably with lower probability. A stuck open SRV and a failed RCP seal would be likely to depressurize the RCS out of the high pressure range into j

the medium pressure range.

i Response.

Yes that.is correct.

I i

C45 Page 148, PRHLSLOK2 = 0.5: Why is this value different from PRHLSLOKI = 0.957 According to the text, both seem to apply to medium pressure conditions.

Response

PRHLSLOK1 =0.95 was defined for medium pressure PDS 1&2, where the pressure j

goes down fairly rapidly. PRHLSLOK2=0.5 is defined for high pressure PDS 3&4 a!beit under i

medium pressure conditions, i.e. following an SRV stuck open or an SGTR or RCP seal failure, j

which may occur after many hours at high pressure. While both do apply to medium pressure 1

conditions, PkHLSLOK2 applies to the higher end of the medium conditions because these l

mechanisms are not all that effective for depressurization. In addition there may be a large time at j

high pressure prior to SRV/SGTR/ SEAL failure.

l C46 Page 157, top para: The Table 2-2 results for debris mass hillow and hole size hillow should be probabilistically weighted, since they depend on assumptions in MAAP.

l'~

Furthermore, it seems that results for the low case already exceed by far what could be expected from

Response

That refinement is not warranted by the current understanding of these phenomena.

these pressure rises for CPSES. Calculations using MAAP even when forcing all parameters in the direction of maximizing DCH and including a hypergolic burn do not reach such high levels.

i C47 Table 3-1 needs some discussion. Why do all six cases have exactly the same pressure and l

cavity water depth?

l

Response

In the 3 cases PDS 3E,F,H the sequence is the same. 'Ihe only difference is spray operation which is injection only, injection and recirculation and failed, respectively. However, sprays only inject after vessel failure so the level at vessel failure is the same in these 3 cases. In the j

other 3 cases 4E,F,H the sequence is similar to the PDS 3 cases except the TDAFW operates.

However, injection does not take place until after vessel failure in both cases 3 and 4 so water levels i

in the cavity at the time of vessel failure are all the same. The si:uation for pressure in the i

containment at the time of vessel failure is similar. The pressures listed are immediately prior to j

vessel failure. If we were interested in pressures shortly after vessel failure the pressure for cases 3H l

and 4H would be 36 psia instead of 21 psia.

l 4

TUL2-CoM.RTu 19 RSE 11/27/1992

C48 Table 3-2, Figure 3-1: This approach of stress-strength interference calculation of BE probabilities could be applied to many other BEs, such as containment failure due to hydrogen burns, debris coolability, containment failure mode, etc. It makes the range of conditions accounted for much more visible compared to the point estimate approach, or the engineering judgement approach based on Table 2-1.

Response

De approach is only warranted when there is a finite probability of the challenge exceeding the containment failure pressure. In the case of burns, had we applied this approach we -

would not have seen a failure probability, the judgement was used to get some late burns to cause containment failure. In the case of debris coolability and containment failure mode the approach doesn't apply because it doesn't involve comparison of a challenge distribution to a limit distribution.

C49 De density of a molten debris mixture is typically about 7 kg/1. Why is the zirconium contribution negligible compared to the same mass from the support plate? The core barrel and upper internals are not normally considered in the mass of ex-vessel core debris. You can use this table to show that the debris depth is not likely to exceed the NRC " guaranteed" coolability limit of 25 cm.

Response

Good point, except that MAAP analyses show we get CCI anyway, so we had to go with that. However, we could use your point in the sensitvity analyses. In order to examine that possibility consider adding all the masses in TABLE 3-3 except core barrel and upper interr.als. This 2

yields 145721 Kg or 21m' using a density of 7000Kg/m'. For a cavity area of 70m the debris bed is 30 cm, just 5 cm thicker than the NRC " guaranteed" coolability limit of 25 cm.

C50 Page 172, BE PRHEATUP: I would think that this probability is at least 0.5. What keeps the fission products from revolatizing?

Response

The probability includes also the flushing of revolatized products into the containment from the RCS. His would be difficult because of the heat in the cavity. Since these were previously high pressure sequences there are no other holes in the RCS for the revolatized fission products to exit.

C51 Page 207, BE SACPOWER: The probability values correspond to the probability of recovering power within the time window. The numerical value for SACPOWER should be the complement,1. e. the probability that power is not recovered. Also injection would have to be started at least 20 to 30 minutes before vessel breach to quench the debris in-vessel and prevent vessel meltthrough.

Response

Nice work. It is reversed and will be fixed. The 20 to 30 minute business just goes into the uncertainty. In any case, most of the recovery probability has been used up in the Level I analysis and the available recovery margin is negligible.

C52 Page 52, PRCDB-LPSE: In the CPSES design without a curb, the debris that is dispersed out through the slanted instrument tube opening can return to the cavity by gravity or by dispersed water flushing back to the cavity.

Response: (Page 216 not 52) Perhaps some of it will, but not a significant amniount. Note that as Tut 2-cou.nTu 20 RSE 11/27/1992

i l

debris blows out into the lower compartment it has to go past a rather tortuous path including small room and a door. Dat will cause it to scatter once it passes the door and will not return easily. In any event this probability is just saying that the debris will be unlikely not to be coolable if a steam explosion occurs. As we have seen from C49, all we need is for at least 17% of the debris not to return since we can goarantee the debris will occupy the entire cavity area after an explosion.

C53 Page 246, Table 4-4: De values for REC 1 do not seem correct. They should be the complement of the value listed. See comment C51. Also this table would be easier to follow if the top events were listed in the same order as they appear in the CET.

Response

The response to CSI covers this issue as well, since this is merely a calculated result j

based on that input.

l C54 Page 205, SLP-SIS 1, SHP-SISI: After power recovery, the recovered systems have to start j

and run for at least several days and probably weeks. The probability of failure after

]

recovery should be accounted for because it may not be trivial.

j

Response

That would apply to all success states, e.g if we have RHR cooling it could later be lost. Unless there is a reason to consider this only for SBO and not for all success states it does not j

seem reasonable. In any case, recovery is already negligible for r,tation blackouts and has not been credited for any other PDS.

f C55 Page 267, M : I cannot reproduce this number. With an initial pressure of 14.65 psia, a free containment volume of 2.985E6 ft' (FSAR Table 6.2.1-1) and an initial containment j

temperature of 100*F, I obtain an air mass of 3315 moles, or 17% more.

Response

Differences in relative humidity can account for that.

C56 Page 300, middle: Numerically, this implicit taking credit for long term flooding of the containment by an unspecified recovery action is not important, but according to the IPE ground rules it is also not permitted.

Response

It is unimportant. Nevertheless, BE probabilities have been adjusted to make this probability 5 e43. This makes it insignificant. Herefore, that discussion has been deleted.

C57 The discussion of Section 8.1 should be moved up to the beginning. See comment C1.

Resoonse:

Agreed.

TUL2-CoM.RTU 21 RSE 11/27/1992

DOCUMENT D.

MAAP BASELINE CALCULATIONS (RXE-LA-cpl /0-003)

D1 This is a very comprehensive accident progression baseline report.

Resoonse:

Thank you.

D2 Page 148 top: Many people in the PRA business consider an accident sequence a no core l

melt success only if at 24 hours2.777778e-4 days 0.00667 hours 3.968254e-5 weeks 9.132e-6 months a stable and permanent core cooling configuration is established. If the core is predicted to melt a few hours after 24 hours2.777778e-4 days 0.00667 hours 3.968254e-5 weeks 9.132e-6 months , they would consider it a core melt sequence. He 24 four success criterion is only used to establish the mission time for the system run time unavailability.

R-onse:

That is true and the SGTR success criteria have been changed because of this point.

He SGTR section here will be rewritten.

l l

I l

1 i

TUL2-CoM.RTU 22 RSE 11/27/1992 1

l DOCUMENT E.

CPSES CONTAINMENT PERFORMANCE DURING SEVERE ACCIDENTS i

i i

El Figure 2-3: The label on the figure is likely to be misinterpreted. As I understood the analysis the 90.6% of early containment failure probability is due to "Either DCH (HPME) or a hydrogen burn", but not "DCH (HPME) with a simultaneous hydrogen burn.

1

Response

Dat is a fine point. The HPME challenge pressure distribution from NUREG/CR-i 4551 includes alli effects from blowdown only to DCH only to burn only to various combinations of l

all these phenomena. It is not possible to say that one single phenomenon which caused the failure.

In the CPSES MAAP analyses and in the most recent SURTESY tests a burn simultaneous with DCH j

is required to cause the failure.

j E2 Page 12: If comment C51 and C53 is correct, the frequency of SBO core melt sequences and the frequency of early containment failure could go up more than a factor of 10. Based on i

comments C7 and C46, the conditional probability of HPME leading to early containment failure could also increase.

Resoonse:

Comments C51 and C53 have been implemented but the effect has been negligible.

E3 Page 15,16: De input changes for 'ITRX (1.) and NVP (6.) may be outside the normal MAAP recommended values, but based on the results from the TMI-2 examination they appear quite reasonable.

1

Response

No response.

E4 Page 16, TAUTO: If DCH is postula'N1 to occur, the a complete hypergolic recombination of all hydrogen is expected to occur, so this is not an extreme assumption for DCH (See NUREG/CR 48% and NUREG/CR-5282). The discussion about deinerting should be revised, it is not relevant if DCH is pcstulated.

Resoonse:

Agreed and done.

E5 Page 19,27 top event CFE: NUREG/CR-5282 documents a large number of DCH sensitivity calculations for ZION with containment pressures in the range of 107 psia to 246 psia. You should review this report before making changes to reduce the CFE probabilities.

Resoonse:

It is based on NUREG/CR-4551 calculations that the probabilities are calculated.

However, there is other evidence, from the SURTESY tests that these pressures might be significantly lower. For the time being we are not revising downward. We continue to use the NUREG/CR-4551 ZION estimates for pressure increases.

E6 Page 26, HOP-DP: If you take credit for an operator action to depressurize the RCS, the NRC will expect that you have a formal procedure in place and that the operators are trained in the use of the procedure.

Resoonse:

Such procedure exists and is in place (FRC 0.1). An evaluation of these probabilities has also been performed by Level I analysts for each of their relevant functional sequences. nose probabilities are weight-averaged over all the functional sequences in the PDS to obtain the PDS TUL2-COM.RTU 23 RSE 11/27/1992

i s

l values used. In addition a sensitivity study has been carried out as discussed in the sensitivity studies section of the IPE Level II submittal. he difference in early containment failure probabilities with and witout the depressurization by operators is not significant because HPME and ALPHA failure modes are in competition at CPSES so that depressurization lowers HPME probabilities but increases ALPHA probabilities with minimal net gain.

E7 He MAAP plots should be in the same order as they are listed in Table 3-1.

Response

Document is not ready yet.

l E8 A table listing the definition of the MAAP parameters plotted should be included, j

Response

Either that or figure legends will be implemented.

l E9 Section 5.2: In most IPEs I am familiar with, the release category from intact containment l

leakage is listed.

l

Response

We could list it but it is not a risk contributor, as LEVEL IIIs have shown.

l l

1 l

l i

l l

I t

TUL2-CoM.RTU 24 RSE 11/27/1992

+4 ATTACHMENT 2 TO TXX-96427 GENERIC DATABASE USED IN THE IPE 1

4 i

i Database used in the IPE Page 1 Of 10 Table 3.2.2-1: GeDeric Data For CPSES Unit 1 COmpODent Failure Rates COMPONENTI FAILURE MODE FAILURE UNFI' DATA SOURCE CODE RATE ROTATING EQUIPMENT CHR1ERS (CU)

PAIL DURING OPERA 110N 9.445 4 H

PLA400 VOL2 FAILTO START S.07E4 D

PuMS00 VOL2 COMPammame Ag yAg. DERMO OPERATION 9AIE45 E

PuMS00 VOL2 (PA)

FAILTO START 3.295 4 D

F1AH500 VOL2 DIESEL OENERA10868 PAIL TO START 2.145 4 D

PMM500 VOL.2 (DO)

FAIL DURING FRST HOUR 1.705 4 H

PuM500 VOL2 FAIL AFTER PRST HOUR 2J154 H

PuM500 VOL2 FAN, SMALL FAIL DURING OPtatATION 7.885 4 H

PLO 0500 VOL2 (VENTILATION)(PN)

FAILTO START 4.84544 D

PuM500 VOL2 i

M-D PUMPS.

FAIL DURINO OPERATION 3.365 4 H

PLO 0500 VOL2 OPERATING (PO)

FAILTO START 2.355 4 D

PuM500 VOL2 M-D PUMPS, FAIL DURING OPERATION 3.425 05 H

PMM500 VOL2 STANDSY (FM)

FAILTO START 3.295 4 D

PuM500 VOL2 PUMPS TURBINE.

FAIL DURING OPERATION 1.03E4 H

PuM500 VOL2 DRIVEN (Fr)

FAE.TO START 3J1EC D

PuM500 VOL.2 AR COOLER (AC)

FAIL DURING OPERATION 1.00545 H

IDCOR IPEM 2.4-1 A FAILTO START 2.935 4 D

IDCOR IPEM 2.4-I A VALVES AND DAMPERS DAMPERS, MANUAL TRANSFER OPEN/CLDSE 4.205 08 H

PMM500 VOL2 (DX)

DAMPERS, MOTOR-FAIL ON DEMAND 4.30E4 D

PWM500 VOL2 OPERATED (DM)

TRANSFER OPEN/C1.OSE 9.27E 08 H

PMM500 VOL2

DAMPERS, FAE.ON DRdAND I.52E4 D

PuM500 VOL2 PNEUMATIC (DA)

TRANSFEk OPEN/CLDSE 2.67E 07 H

PUM500 VOL2 VALVES, AR.

FAR.ON DEMAND 1.52E4 D

PuM500 VOL2 OPERATED (VA)

TRANSFER OPEN/CLOSE 2.67E47 H

PuM500 VOL2 VALVES, CHIiCK FAIL ON DEMAND 2.69E 04 D

PuM500 VOL.2 (OTHER THAN ETOP VALVES)

OROSS REVERSE L2AKAGE 5.36547 H

PuM500 VOL2 TRANSPER C1DSID/ PLUG 1.04E 08 H

PuM500 VOL.2 i

1

}

I 3-147 1

--n-w,-

., ~, - - - -

Database used in the IPE Page 2 of 10 Table 3.2.2-1: Generic Data FOr CPSES Unit 1 CompODent Failure Rates COh00NEpfr/

FAILURE MODE FAILURE UNTI*

DATA SOURCE CODE RATE VALVES, CHECK FAIL ON DEMAND 9.13544 D

PuM500 VOL2 (WTOP VALVES)

(VT)

CROes REVERSE LEAKAGE 5.365 47 H

PUNK 500 VOL2 1RANSPER CLOSED / PLUG 1.04548 H

PUNK 500 VOL.2 VALVES, ELECTRO-FAIL ON DEMAND 1J2543 D

PuMk500 VOL2 HYDRAUUC (EXCEPT TUREINE STOP/COff!ROL TRANSPER OPEN/ CLOSED 2.67547 H

PuMIS00 VCI. *,

VALVES)(VM)

VALVES, MANUAL TRANSPER OPEN/ CLOSED 4.20548 H

PUNkl00 VOL2 (VX)

FAILTO REPOSFFION 1.00E 04 D

IDCOR IPEM 2.4-1 A VALVES, MOTOR-FAIL ON DEMAND 4.30543 D

PLO4500 VOL2 OPERATED (VM)

TRANSPER OPEN/CIASE 9.27548 H

PuMk500 VOL2 VALVES, RELIEF (2 FAILTO CLOSE 8.885 43 D

PLO 0500 VOL2 STAGE TARGET ROCK)(VK)

FAILTO OPEN 9.05543 D

PUNKS 00 VOL2 VALVES, RELIEF FAIL TO OPEN 2.42E45 D

PuMkl00 VOL2 (OTHE THAN FORY OR APETY VEM 1RANSPER OPEN 6.06E46 H

PUNK 500 VOL2 VALVES, (PORV) (VP)

FAILTO CLOSE 2.50541 D

PUNK 500 VOL2 FAE.TO OPEN 4.27545 D

PUNkSCO VOL2 TRANSPER CLOSED /OPEN 2.67547 H

EXPERT OPINION SadILAR10 AOV VALVES,8AFETY FAIL TO OPEN 3.28544 D

PuMIS00 VOL2 FAILTO RESEff AFTER 2.87543 D

Punk 500 VOL2 STEAM PM.IEF FAIL TO RESEAT AFTElt 1.00E41 D

PUNKS 00 VOL2 WW2R REUEF FAILTO RECLOSE 1.00E42 D

IDCOR IPEM 2.4-1 A VALVES, SOLRNOID FAL. ON DEMAND 2.4353 D

PLO4500 VOL2

""~

(VS)

,%Ar: war OPEN/CIASED 1.275 06 H

PLO4500 VOL2 VALVES, TURBINE FAIL ON DEMAND 1.25544 D

PuM500 VOL2 STOP/ CONTROL (VU)

TRANSPEIt CLOSED 2.88E45 H

PLO4500 VOL2 TRANSFER OPEN 1.24E45 H

PIA 4500 VOL2 3 148

Database used in the IPE Page 3 of 10 l

l Table 3.2.2-1: Generic Data FOr CPSES Unit 1 COmpOnset Failure Rates COMPONENT /

FAILURE MODE FAR.URE UNT!*

DATA SOURCE CODE RATE PREBBERE VESSELS ACCUMULATORS, RUFFURE/ LEAK 2.46546 H

PLO4500 VOL.2 SCRAM HEAT EXCHANGERS RUFFURE/ LEAK 1.95546 H

PLO4500 VOL.2 I

FWEB (CREATER RUPTURE / PLUG (PER SECDON) 8.60E 10 M

PLO4500 VOL.2 THAN 3-INCH DRAMETEE)(PP) i PIPES (1388 THAN 3-RUPTURE / PLUG (PER SECrlON) 8.605 09 H

PLO4500 VOL.2 j

DeCH DiAMEnBo (Po I

TANKS, STORADE i RUPTURE /12AK 2.66548 H

PLD4500 VOL.2 l

(HQ IXPANSION HMNT RUFIVRE 8.605 09 H

CRYSTAL RIVER 3, (BI)

FRA FIDW ELEMENT WS PLUGRUFIURE 3.00548 H

CRYSTAL RIVER 3 FRA FLEXMLE HOSE (FH)

PLUG / RUPTURE 3.00E47 H

CRYSTAL RIVER 3 FRA l

FLOW ORIFICE (OR)

PLUO/RUFTURE 3.005 08 H

IDCOR IPEM 2.4-l A NOZZLES, FTRADERS, SUMPS, AND FILTERS i

FILTERS, AR PLUG 5.835 06 H

PLD4500 VOL.2

FILTERS, FLUG 3.54EM5 H

PL44500 VOL.2 COMPRESSED AR FILTERS, OIL PLUG 1.76E45 H

PLD4500 VOL.2 REMOVAL i

FILTERS, PLUG 1.07E.06 H

FLO4500 VOL.2 l

VENnLATION (PV) l NOZZLES.

FLUG 7.06E.0s N

PLO4500 vot.2 l

CONrADMENT sUILDING SPRAY (ONE TRAIN)(NZ) i STRADIERS. SERVICE PLUG 6.21E46 H

PLO4500 VOL.2 WATER (FL)

SUMPS.

PLUG 1.00545 H

PLO4500 VOL.2 CONrADMENT(SU) 1RAVELDioSCREEN FAILTO OPERATE 2.71545 H

PLD4500 VOL.2 (TE) 4 2

i

)

3-149

Database used in the IPE Page 4 Of10 t

Table 3.2.2-1: Generic Data FOr CPSES URit 1 Component Failure Rates l

COMPONBfr/

FAILURE MODE FAILURE UNTr*

DATA SOURCE CODE RATE E12CTRICAL EQUIPMENT BATTERIES,125V DC FAR.ON DR4AND 4.84544 D

FIA4500 VOL2 (NT)

FAR. DUES 40 OPELATION 7.53547 H

PIA 4500 VOL2 BATTERY CHARGERS FAIL DURING OPERATION 1.86545 H

PIA 4500 VOL.2 (EC) i BWTABLES (BI)

PAIL ON DhiAND 3.89547 D

FIA4500 VOL2 SPURIOUS OPERATION 2.21546 H

PLA4500 VOL2 j

BUSES (BU)

FAIL DURING OPERATION 4.98547 H

Pl40500 VOL2 CABLES, COffTROL FAIL OPEN OR SHORT 4.64546 H

PIA 4500 VOL2 (CA)

CEtcurr=== Arm yAILTO C1DSE 1.61E.03 D

FIA4500 VOL2

(> =400V AC)(BA)

PAILTO OPEN 6.49544 D

PLS 0500 VOL2 TRANSPER OPEN 8.28547 H

PLA4500 VOL2 CRCUTr ammame FAILTO CLOSE 2.27544 D

PLO4500 VOL2

(<480V AC)(BB)

FAILTO OPEN 8.39E44 D

PLO4500 VOL2 1RANSFER OPEN 2.68547 H

PLD 0500 VOL2 l

FUSES (PU)

FAIL OPEN 9.20E47 H

PLO4500 VOL2 INVERTERS (IV)

FAIL DURING OPRA110N l.83545 N

FIA4500 VOL2 MOTOIL FAIL DURING OPERA 110N 3.59545 H

PIA 4500 VOL2 i

OENERATORS(MG)

POWER SUPPLJES FAIL DURING OPERATION l.33E44 H

PIA 0500 VOL2 i

(

(+ 120V DC ESFAS)

(PD)

POWBt SUPPLES PAIL DURD4G OPERATION 5.33E45 H

PIA 4500 VOL2

(+5V OR +25V DC ESFAS)(PS) l' REACTOR 1 RIP FAIL ON DEMAND 1.77E43 D

PLO4500 VOL2 mREAKERS (SB)

SHUNT TICP COILS FAIL ON DEMAND 1.40544 D

PLO-0500 VOL2 i

on UNDERVOLTAGE FAR. ON DEMAND 2.75E45 D

PLG4500 VOL2 COR.8 (Un RELAYS (RM)

FAIL ON DEMAND 2.41E44 D

PLO4500 VOL2 FAIL DURD40 OPERATION 4.205 07 H

PLO 0500 VOL2 i

l 3-150

,1

Database used in the IPE Page 5 of10 Table 3.2.2-1: Generic Data For CPSES Unit 1 Component Failure Rates COMPONENT /

FAILURE MODE FAILURE UNT!'

DATA SOURCE CODE RATE

SWFTCHES, FAIL ON DEMAND 2.406 @

D FIA400 VOL2 PUSMBUTTON TRANSPORMERS FAIL DURING OPERATDN 1.56E4 H

PtG@00VOL2

(> =4.16KY)(FR) 11tANSPORMERS FAIL DURING OPERATION 6.873W7 H

PLA400 VOL2

(<4.16KV, > =480V)

GM)

TRANSPORMERS, PAIL DURING OPERA 110N 1.55546 H

PLD&00VOL2 INSTRUMENT

(<480V, > =120V)

(TD AUTOMA11C FAILTO11tANSPER 2.305 06 H

SHEAROM MAIULIS 1RANSFliR UNTT(AT)

RELAY CONTACT FAILTO OPENCLOSE 3.00544 D

IDCOR IPEW 2.41 A (CN)

SPURIOUSLY OPEN/CLOSE 2.40E@

H OCONEE FRA TABLE B-1 TIME DELAY RELAY FAILTO TRANSFER 3.005-04 D

IDCOR IPEM 2.41 A (ItT)

TRANSFIIR PREMATURELY 6.00E46 H

IDCOR IPEM 2.4-I A RELAY COE.3 (RY)

FAILTO 3.00E46 D

IDCOR IPEM 2.4-1 A DEIDGRGIZE/ENEROIZE SPURIOUSLY DEENERGIZE 3.00E-06 H

IDCOR IPEM 2.4-I A TEltMINAL BOARD SHORT/OPEN CIRCUTT 3.00547 H

IDCOR IPEM 2.4-1 A (TB)

ELEC11 TONIC EQUIPMENT SIGNAL MODIFIsts FAIL DUIUNG OPERATION 2.94E46 H

P14-0500 VOL2 (MS)

TRIP LOGIC FAIL ON DEMAND S.52E4 D

P144500 VOL2 MODULES (SS)

FAIL DUIUNO OPERATION 2.70E46 H

PLo@00 VOL2 INTTRUMENTATION

SWFTCHES, FAIL ON DEMAND 2.69E44 D

Pt4 0500 VOL2 PRESSURE (SP)

OPERATE SPUIuoUSLY 3.40E 07 H

OCONEE FRA TABLE S1 TEMPEllATUltE NO OUTPUT 3.41E-06 H

P14 0500 VOL2 MONITOR 140PS TRANSMTITBts.

FAIL DURING OPEllAT10N 6.25E46 H

F14-0500 VOL2 FLOW (TF) 3-151

Database used in the IPE Page 6 of 10 Table 3.2.2-1: Generic Data FOr CPSES Unit 1 Component Failure Rates j

l COMPONENT /

FAILURE MODE FAILURE UNtr*

DATA SOURCE CODE RATE TRANSMr!TERS, FAIL DURING OPERATION 1J7545 H

PIA 4500 VOL2 12 VEL (TL)

TRANSMITTERS, FAIL DURING OPERATION 7.60546 H

FIA4500 VOL2 PRESSURE (TP)

IAdrr SwirCH (30 FAILTO OPERATE 1.00544 D

IDOOR M 2.4-I A OPERATE SPURIOUSLY 4.70546 H

OOONER PRA TABLE B1 12 VEL swr!G FAILTO OPERATE 2A0544 D

OCONEE FRA TABLE l

(SL)

B1 i

l OPERA 11 SPURIOUSLY IJ7546 H

SHEARON HARRIS MANUAL SWFTCH FAILTO OPERATE 3A10545 D

IDCOR IPEM 2.61 A (SM)

OPERATE SPURIOUSLY 1.30546 H

OCONEE FRA TAB 12 B-1 TORQUE SWrrCH FAILTO OPERATE 1.00544 D

IDCOR IPEM 2.41A (SQ)

OPERATE SPURIOUSLY 3.40647 H

OCONEE FRA TABLE B-1 TEMPERATURE FAILTO OPERATE I.00E44 D

IDCOR IPEM 2.4-1 A l

SWrrCH (ST)

OPERATE SPUIGOUSLY 3A0547 H

OCONEE FRA TAB 12 B1 TEMPERATURE FAIL BOGH/ LOW RESPOND 3.00546 H

ECOR IPEM 2.4-1 A '

TRANSMrrrER (TT) l SCRAM RODS l

SINGLE SCRAM ROD FAIL ON DEMAND 3.30545 D

PLG4500 VOL.2 (PWR) (SC) l l

i i

l

[

3-152 i.

. _. - - - =

Database used in the IPE

- Page 7 of 10 j

Table 3.2.2-2: Generic Data For Cp8e8 Unit 1 Maintanance Frequencie8 and Duration 8 l

i NO.

COMPONBfr TBCIOGCAL MADfrENANCE MAIN 7ENANCE SPECIPICATIONS PREQUENCY DURATION GeOURS)

SVEN7MIOUR) 1 CHILLERS NONE 1J019544 466.7 s 24 HRS 1.3819544 6J 1

i de OR 72 IRS IJ819504 13.1 les OR 3M HR8 13819544 37.2 2

COMPRB880RS NONE 2.9311544 MJ i

s 24 HR$

2.93115 44 6J 48 OR 72 HRS 2.9311544 13.1 168 OR 3M HRS 2.93115 44 37.2 3

LARGE PANS NOME 1.4727544 38J 5 24 HRS 1.4727544 6J 48 OR 72 las 1.47275 44 13.1 les OR 336 HRS 1.47275 44 37.2 4

SMALL FAN 8 NONE 2.08875 44 38.5 s 24 HRS 2.0887544 6J de OR 72 HRS 2.08975 44 13.1 l

168 OR 336 HRS 2.08975 44 37.2 5

DESEL OENERATORS NONE 1.02705 4 M.5 s 24 HRS 1.02105 4 6J 48 OR 72 HRS 1.02705 4 13.1 168 OR 336 HRS 1.02105 4 37.2 6

HEAT EXCHANGERS NONE 4.1453545 583.1 s 24 HRS 4.1453545 GJ l

48 OR 72 HRS 4.1453545 13.1 168 OR 3M HRS 4.1453545 37.2 l

7 OPERATING SERVICE MONE 3.34505 44 266J l

WATER PUMPS 5 24 HRS 3.3459544 7J 72 HRS 3.34595 44 11.1 168 HRS 3.3459544 1s.7 8

UTIER OPERATING NONE IJ790544 266.3 PUMPS s 24 HRS 1J790E44 7J 72 HRS IJ790E04 11.1 168 HRS 1.57905 44 28.7 9

STANDBY MOTUR-NOME 1.16705 44 266J DRIVEN PUMPS 5 24 HR8 1.1670E44 7.5 72 HRS l.16705 44 11.1 168 HR$

1.1670E44 28.7 10 STANDBY TURBINE-NONE 4.1928544 264.3 I

DRIVEN PUMPS s 24 HRS 4.1928544 73 72 HRS 4.19285 44 11.1 168 HR$

4.1928544 28.7 11 PostlTVE NONE 6.3703E44 266J DISPLACEMEPfr s 24 HRS 6.37035 44 7.5 PUMPS 72 HRS 6.37035 04 11.1 i

l 168 HRS 6.37035 0e 28.7 t

l f

3-153 i

Database used in the IPE Page 8 of 10 Table 3.2.2-2: Generic Data For CPSES Unit 1 Maintenance Frequencie8 and Durations (continued) d NO.

COMPONENT TEClOGCAL MADrTENANCE MADrTENANCE i

SPECIFICA110NS FREQUENCY DURATION (HOURS)

)

1 CFfff M NOME I.3819E44 460.7 5 24 HRS 3.3819E44 6.3 48 OR 72 HRS 1.38195 04 13.1 i

168 OR 3M HRS 1.3819E44 37.2 12 VALVES NONE 2.7382545 132.3 s 24 HRS 2.7382545 4.1

{

72 OR 168 HRS 2.7382E45 18.9 13 BATTERIES. BATTERY NONE 2.4948545 38J CHAROERS.AND s 24 HRS 2.49485 45 6.3 1

INVERTERS 48 OR 72 HRS 2.49485 45 13.1 j

168 OR 3M HRS 2.4948E45 37.2 14 BUSES NONE 2.6546546 38.5 s 24 HR$

2.6586E46 6.3 48 OR 72 HR$

2.6586546 13.1 1

168 OR 336 HRS 2.6546546 37.2 15 11LANSFORMERS NONE 4.40375 06 38.5 s 24 HRS 4.40375 06 6.3 l

48 OR 72 HRS 4.4037E 06 13.1 168 OR 3M HRS 4.4037E.06 37.2 16 STILADERS NONE 9.2738E45 38.5 i

s 24 HRS 9.2738E 05 6.3 l

48 OR 72 HRS

, 9.27385 45 13.1 168 OR 3M HRS 9.27385 05 37.2 17 OAS TURBDIES NONE 1.9213E44 38.5 s 24 HRS 1.9213E44 6.3 l

48 OR 72 HRS 1.9213E44 13.1

)

168 CR 3M HRS 1.9213E44 37.2 i

i t

4 4

i i

l 3-154 k

l

)

l 1

Database used in the IPE Page 9 of 10 i

t Generic Data Used for the Initiating Event Categories:

1.

Excessive LOCA (> > 6').

Freq. = 2.66E-07 yr8 (Volume 6, Ref.1).

2.

Large Break LOCA (> 6').

Freq. = 2.03E 04 yr' (Volume 6, Ref.1).

3.

Medium Break LOCA (> 4" and < = 6').

Freq. = 4.65E-04 yr8 (Volume 6, Ref.1).

4.

Small Break LOCA (> 2" and < = 4").

Freq. = 5.83E 03 yr (Volume 6, Ref.1).

5.

Vary Small Break LOCA (< = 2").

Freq. = 1.26E-02 yr8 (Volume 6, Ref.1).

6.

Loss of Condenser Vacuum.

Freq. = 1.18E-01 yr8 (Volume 6, Ref.1).

9 7.

Steam Generator Tube Rupture..

Freq. = 1.18E-01 yr' (Volume 6, Ref.1).

8.

General Plant Transients. Including:

Reactor Trip.

l Turbine Trip.

Excessive Feedwater Flow.

Closure of One MSIV.

Inadvertent Closure of All MSIVs.

Core Power Excursion.

Loss of Pnmary Flow.

Freq. = 2.90E-00 yr8 (Volume 6, Ref.1).

4 3-155 t

}

Database used in the IPE Page 10 of10 1

9.

Inadvertent Safety Injection Signal.

d Freq. = 2.99E 02 yr (Volume 6, Ref.1).

i l

10.

Main Steam Line Break. Including:

Steam Line Break Outside Containinant.

Steam Line Break Inside Containrnent.

I Inadvertent Opening of Main Steam Relief Valves.

Freq. = 1.07E-02 yr5 (Volume 6 Ref.1).

11.

14ss of Main Feedwater. Including:

Total Loss of Main Feedwater.

l l

Partial 14s of Main Feodwater.

t 4

l Freq. = 1.29E40 yr8 (Volume 6, R# 11 l

12.

Loss of a DC bus.

l Freq. = 3.35E-02 yr8 (Volume 6, Ref.1).

l 13.

Iass of Off-Site Power.

Freq. = 3.50E-02 yr8 (NSAC/166, Ref.11).

14.

Iass of Non-Vital AC Bus.

Freq. = 8.23E 02 yr8 (Volume 6, Ref.1).

l 15.

less of Safeguards Bus.

Freq. = 8.36E-02 yr8 (Volume 6, Ref.1).

l 16.

Loss ofInstrument Air.

l Freq. = 2.02E-03 yr8 (Volume 6, Ref.1).

l r

3-156 i

I

n.

ATTACHMENT 1 TO TXX-96427 CPSES UNIT 1 INDEPENDENT REVIEW 0F IPE STUDY BY OUTSIDE CONSULTANTS

CPSES Unit 1 Independent Review of IPE Study by Outside Consultants No.

Comment

Response

Reviewer 1.

This is a carefully done and documented Thank you.

Gaertner IPE. TU people understand it. It has

}

potential to satisfy all NRC IPE l

requirements without significant change.

i 2.

There is an adequate basis to dismiss any Good.

Gaertner concern on A-45 issue. A sensitivity run without high seal failure rate will i

help see how good your secondary heat removal is. (See item 7 below).

i i

3.

Be careful when running importances to Muliple recovery actions, with the Gaertner present to NRC. When multiple exception of recovery of faulted recoveries are used in a cutset, then equipment, are not allowed.

l importances are distorted.

i 4.

To summarize sensitivity runs of value, We do not currently plan on running any Gaertner consider sensitivity runs. In the future, however, i

a.

Effect of ATWs we do plan on doing some sort of j

b.

Effect of LOSP recover uncertainty quantification. We will use c.

Effect of seal failure importance factors to evaluate the d.

Effect of operator recovery current runs.

e.

Effect of faulted system recovery f.

Effect of more (or less) conservative HVAC dependencies 5.

A discussion of uncertainty is needed, See above.

Gaertner although I believe a big effort to quantify uncertainty is unnecessary.

6.

Emphasize that the CM frequency is We will.

Gaertner distributed somewhat uniformly among initiators and sequence types.

Therefore, there is no unique sequence that is a vulnerability at CPSES.

7.

NRC staff uses 3 categories for A-45 Good.

Gaertner l

DHR vulnerability at CM frequency <

$E-5 due to DHR, CPSES is in the best i

ca y ory.

2 j

Page 1 of 47

}

\\

i l

CPSES Unit 1 Independent Review of IPE Study by Outside Consultants s

4 No.

Comment

Response

Reviewer 1.

Cutsets are grouped by event tree We believe that the sequences are Gaertner sequence at this time (e.g. #1SCM3, functional sequences, and will therefore

ACM2). They should be regrouped be reported as such.

along the functional lines of the NUMARC Severe Accident Closure report for NRC presentation.

2.

ATWS core damage frequency and We believe that the modeling of ATWS Gaertner i

number of cutsets is high compared to at CPSES is realistic. Other PRAs have I

i other PWRs.

not been so diligent.

l 3.

Induced LOCAs (RCP seal failure)

Gaertner dominates CM risk frequency, especially from LOSP. Since the seal design will 2

be changed, consider running a sensitivity study with low seal failure rate to seal change in sequence types that are important.

I 4.

LOSP recoveries are low compared to We believe that the methodology used Gaertner other PRAs such as Surry and Sequoyah generates reasonable values for LOOP NUREG 1150. His is due to the recoveries.

convolution (which is good) and the use of Weibull distribution (which is in question) and use of old data (recent NSAC data is newer). Consider a sensitivity study to see impact of more i

conservative assumptions.

5.

Failure probabilities of pump trains from We believe that the failure probability of Gaertner loss of HVAC are treated as independent a pump on loss of HVAC is failures. ney are not. Consider independent. The dependencies in the treating as totally dependent HVAC system itself are the only (conservative) or requantify with a dependencies.

realistic dependence.

6.

Use of faulted equipment recoveries, NSAC-161 gives us a strong basis for Gaertner especially EDGs, results in significantly using faulted equipment recoveries.

reduced CM frequency. Be prepared to l

defend these numbers or adjust upward.

l l

Page 2 of 47

l l

CPSES Unit 1 Independent Review of IPE Study l

by Outside Consultants l

l l

No.

Comment

Response

Reviewer 7.

LOCAs (A, M and S) sequences from We will be.

Gaertner CCF or valves and pumps are low in frequency. Be prepared to justify these l

low valves of CCF.

l l

8.

Many cutsets have multiple recoveries The recoveries that are actually dynamic Gaertner applied. Resulting frequencies are very actions have been imbedded into the low. Also, its not clear that the HRA model. Any addition of a dynamic assessment was done with the knowledge recovery will be a single event that will l

that multiple (series or parallel) actions account for the dependencies of multiple are underway. Consider re-evaluating actions.

key combinations that affect overall CDF.

9.

In the course of discussion, it was found Justification has been added to the Gaertner that the UPS HVAC is a critical system.

accident sequence calc. under the loss of l

Justify excluding it from the initiator list.

HVAC description.

10.

Cutset #21 ir tves recovery of LOSP The recovery factor applied does not Gaertner after CCF of 4 pumps. LOSP recovery consider the fact that the D/Gs have run is 0.12 which seems relatively high for 4 hours4.62963e-5 days 0.00111 hours 6.613757e-6 weeks 1.522e-6 months . May need to add in the compared to earlier cutsets with fewer

future, failures (e.g. #15) please check.

11.

Cutset #22 has $SFSMALL1, ne ne $SFLARGE1 cutset was recovered.

Gaenner i

mirror image cutset with $SFLARGE1 is missing. Perhaps the recoveries are applied differently in an incorrect way.

Please check. This is also true of cutset

24.

12.

Cutset #39 is unrecovered " slow transfer All recovery possibilities have not been Gaertner l

I of train B". I suspect it is recoverable.

considered. Only the dominant cutsets.

13.

After functional sequence definitions are The sequeces that we quantify are Gaertner l

determined for the IPE report, dominant functional sequences. The functions are cutsets should be regrouped into these less granular than other IPEs.

categories and discussed by type, he small event tree method hides many of i

the actual failures (e.g., loss of IEDI l

bus fails components in AFW, SI, CCW,..) which makes cutsets difficult to read and understand herefore, insights are hidden.

i I

l I

Page 3 of 47 v

- - -e yer-

9 i

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 14.

Cutsets are completely traceable through Thank you.

Gaertner documentation because of high quality quantification notebooks, accident sequence notebooks, system and data notebooks, HRA notebooks. Also good is complete basic event description on cutset lists.

15.

With CAFTA knowledge, it is The quantification calculation will be Gaertner reasonable to expect that the results are

" fattened up" to include sufficient reproducible. Not present in the information such that someone documentation are the flag files, CAFTA knowledgeable in CAFTA can reproduce macros, and log of actual quantification the results.

runs (with file names) so that it can actually be done. 'Ihe IPE team should verify that this information is available, accurate, and in good order to assure reproducability.

16.

Insights are not developed yet. To get Ok.

Gaertner them, several steps are still required:

1)

Sensitivity studies are suggested in comments 3.4.17.

2)

Grouping by functional sequences and looking at those above 104/yr (or 104 BP) and 5% contribution.

3)

Digging out the actual failures from the dominant cutsets.

17.

ATWS contributions to CM freq. are Yes. We will consider.

Gaertner high and probably more conservative than other sequences. Some IPEs consider these as screening numbers and handled separately. Consider a sensitivity study with ATWS removed to see the change to CM contributors.

Page 4 of 47

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer -

18.

Design and Procedural enhancements are Will present as part of the submittal.

Gaertner not presented. We know from presentations that Unit 2 cross-connects and joint TSs have already been identified and transmitted to the plant.

We also know that RCP seals are schcduled for upgrade. Other potential enhancements are evident from the cutsets (e.g. cutset #47 shows that dependence of RHR receive on operator restoring 1A after S1 is an amendable to a fix).

19.

Cutset listings are traceable and easy to Again, these are our functional Gaertner read. Dey do not identify dominant sequences, sequences by NRC criteria yet, but this can be done. Here must be a roadmap from the current sequence designators to the eventual functional sequence definitions. Many comments above address this issue. Also see comments in accident sequence section.

20.

Not all assumptions are documented.

Added more assumptions.

Gaertner For example, why is depressurization not modeled in VS and R event trees.

Generally, however, assumptions are well documented.

21.

Use of generic frequencies for T1 and Noted.

Gaertner T6 is clearly conservative. As the models now stand it will have most effect only on ATWs, but reducing ATWS contribution is worthwhile.

Page 5 of 47

CPSES Unit 1 Independent Review of IPE Study i

by Outside Consultants No.

Comment

Response

Reviewer 1.

End state characterizations can support ne method of characterizing endstates Gaertner the backend analysis, but the will be added to the sequence containment systems (isolation and quantification notebook.

containment spray) have not been linked in the material available to us for review. (Later presented to us by l

Hugo.) Continued as comment 17.

2.

SGTR event tree has no " fails to Added it.

Gaertner l

depressurize" top event. Therefore, H

SGTR contribution to CM freq. is ~ 1E-l 9/yr. Be prepared to justify this missing l

sequence since it is typically present in j

l PRAs and because it is a bypass l

sequence, (i.e., important at 104/yr).

l 3.

Combining 611 EDG faults into one Split the modules out into their BEs in Gaertner model per EDG has its drawbacks. The the tree.

l recovery factor doesn't relate to the "cutset", and the timing (start or run failures) of important sequences is lost.

Consider separately start and run as two modules.

4.

Availability of CST suction for 24 hours2.777778e-4 days 0.00667 hours 3.968254e-5 weeks 9.132e-6 months This is a base assumption that will be Gaertner is not explicitly modeled even though placed in the AFW system model.

equipment / operators must function.

Justify this omission or add to model.

5.

Several ATWS cutsets (e.g. #8, #20)

De impact of %T6 is significantly Gaertner initiate with T6 (freq = 1.29). It greater that %T1, in that MFW is lost.

appears they can also initiate with T1

%T1 would require an additional failure (freq = 2.9) but are missing. Please to lose MFW.

explain or change.

6.

He event tree for very small LOCA Added the recirculation requirement.

Gaertner l

does not require recirculation in 24 L

hours, but it does not model cooldown l

either. A strongjustification for assured cooldown is needed.

l I

r Page 6 of 47

1 i

CPSES Unit 1 Independent Review of IPE Study by Outside Consultants 4

j No.

Comment

Response

Reviewer i

7.

Cutsets are very long (e.g. cutset #37 The depth of the cutset increases Gaertner has 9 terms). This is burdensome to understanding, and increases i

understand and probably limits the cutoff quantification time. The multiple frequency. Some of these events would recoveries will be combined.

i be better if combined (e.g., multiple i

HRA recoveries, HRA with faulted system recoveries, conditional i

probability of failure given HVAC loss).

l I believe this approach will also improve l

accuracy by considering dependencies.

I 8.

Use of small event trees places burden Acknowledged.

Gaertner on utility to explain cutsets to NRC and i

plant staff. Cutset #47 has many insights not evident without explanation l

(e.g. instrument air trip on SI signal, need to throttle "WRV" HVAC valve.

9.

Cutsets are generally not solved deep Will quantify all sequences to IE-09.

Gaertner enough to see all important cutset types.

4#

(maybe) and to get the accurate i

combined frequency of CM.

l 10.

I was able to find very few errors (ie, Thank you.

Gaertner wrong modeling, wrong data, omitted l

events, dumb cutsets) relative to other IPEs. His indicates that the models are mature and well understood by utility j

IPE team.

11.

This IPE assumes success after 24 hours2.777778e-4 days 0.00667 hours 3.968254e-5 weeks 9.132e-6 months The assumption is that after 24 hours2.777778e-4 days 0.00667 hours 3.968254e-5 weeks 9.132e-6 months Gaertner even if a safe stable state is not reached sufficient manpower and equipment (vs LOCA not needing recirc.). This is resources will have been made available non-standard IPE assumption that must such that it is highly unlikely that the be justified.

accident will progress to core damamage, given that it has not occurred in the first 24 hours2.777778e-4 days 0.00667 hours 3.968254e-5 weeks 9.132e-6 months .

Page 7 of 47 I

4 CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 12.

All ET transfer points are handled by the The method is not the problem as much Gaertner induced LOCA ' tree, and they appear as tl'e fact that the probability for correct. The quantification approach is inducing a seal LOCA is very high.

lir.Jting the ability to solve the model to lov/ transaction levels. This does not appear to be a problem, however because many cutsets are coming up at the current transaction.

13.

There are some differences between the Depressurization is now required for Gaertner CPSES models and some other PRAs.

SGTR.

The very small LOCA, no depressurization for SGTR, and the detailed ATWs quantification are different.

14.

Not all success criteria b.assi are All TH bases should be available.

Gaertner documented in the material available for review. However, there were none observed that appeared unreasonable or non-conservative.

15.

ISLOCA is at about SE-8 and is not Ok.

Gaertner reportable at this level which is j

reasonable.

Page 8 of 47

CPSES Unit 1 Independent Review of IPE Study by Outside Consultants No.

Comment

Response

Reviewer L

16.

Recoveries are properly included in the Mulitple recoveries, with the exception Gaertner accident sequence models with some of recovery of faulted equipment, is not exceptions:

allowed.

1)

Per comment 4 in quantification of results, LOSP recoveries are i

probably optimistic and difficult l

to interpret.

l 2)

Per comment 6 in quantification of results faulted equipment recoveries are used.

3)

Per comment 8 in quantification of results, multiple recoveries are used.

l l

l Dese comments should be considered.

17.

Although the safeguards states are crude Ok.

Gaertner i

l (no recoveries except power, no coolers, l

etc.) my experience is that the resolution is adegaate for what is needed for IPE i

Level 2.

18.

Systems, functions, and operator actions Ok.

Gaertner seem to be well modeled except as noted

[

in above comments (e.g. comment 13) 1 l

19.

Inclusion of the SRO on the IPE team Ok.

Gaertner and review of results by operators seems to have eliminated any potential inconsistencies with procedures. All issues were clearly explained.

l 20.

De use of NSI to eliminate nonsense he models have been changed to Gaertner I

cutsets is confusing. De reviewer must eliminate the need for the NS recoveries.

l chase these down through the l

quantification notebook to be sure these l

"modeling errors" are treated correctly.

1 It would be better to eliminate them from the model.

i Page 9 of 47 i

CPSES Unit 1

)

Independent Review ofIPE Study i

by Outside Consultants 4

)

No.

Comment

Response

Reviewer 21.

Regarding earlier comment 5, it appears Yes, see response to #5.

Gaertner that the T1 counterpart tc T6 in cutset 20 would not result in coremelt because feedwater would not trip. 'Ihat cutset is

11 and many others from the

ATCM7. CUT file.

22.

Consider providing a more detailed Will consider.

Gaertner support system matrix to satisfy NRC request for IPE submittal. Current one is not instructive. For example, it notes

~

that SI has a chilled water dependency, but it does not specify it. Without system notebooks, it is impossible to identify the dependence.

23.

Not including fan coolers during Unfortunately, this is the way the Gaertner sequences is unfortunate, since they procedures and the plant design are.

prevent spray actuations. This is more realistic and less conservative.

Page 10 of 47

CPSES Unit 1 Independent Review of IPE Study.

by Outside Consultants No.

Comment

Response

[

Reviewer 1.

AFW notebook; assumption #3 is not Yes, but a flag in the ATWS.FLS file Wakefield correct for ATW/ events whose MD and prevents the TDAFWP from failing on TD pumps are all required. (Section overfill on an ATWS.

V.1).

2.

4.2 of AFW notebook; the impact of Will consider in future model Wakefield each Human action listed should be improvements.

mentioned (e.g. so what if operators fail to control AFW flow to 4565 Fail TDP7).

3.

AFW system - latent HE contributers Latent HE values are all screening Wakefield seem to be too high. Hey dominate the values. They may be reanalyzed in the failure of AF1000; also very important future.

for AF6000.

)

4.

AFW system - for AF6000, event His event is also disabled on ATWS in Wakefield

&SGLXAFWXYY is probably not a the ATWS.FLS flag file.

failure for ATWS if SG would not overfill until much later when all 3 pumps are no longer required.

i 5.

Human action AFSGTR would also This event is only the isolation of the Wakefield cause secondary not to be isolated: so steam supply to the TDAFWP. He how is this modeled in the sequence other isolation requirements are included quantification?

in &SGTR01 and &SGTR02.

6.

AFW, sectiorr6.0- a basis for not There is a basis. It is the fact that they Wakefield modeling CCF across all 3 pumps (not are significantly different types of l

drivers) should be given.

pumps.

7.

AFW, section 6.0- I don't understand He factor is calculated based on the Wakefield the " factor" column. Why (1,2) for formula prescribed in the data cale. for some and (.0667) for others?

CCF.

i 8.

IAss of chilled water as in IE is modeled He AC bus failure as an IE would have Wakefield as a single basis event in the sequence impacted the SSW system preferentially.

quantification; this does not allow the See the support system initiator cale.

l full impact of say an AC bus failure which caused the loss of chilled water, being modeled.

l Page 11 of 47

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 8.

For chilled water, operator action to his is the screening value. De current Wakefield manually start the standby train is too value has been significantly reduced.

high 1.e., = 0.2. Perhaps its in the HRA section; but the allowable time or success criteria for each action should be noted.

9.

Chiller system modeling assumptions We do recognize this.

Wakefield

13. Doubling maintenance outage on train A may lead to an asymmetric importance ranking for train A/B. This maybe ok if the trains are symmetric

]

when considering their impacts on the last of the plant, so long as tb analyst recognizes it.

10.

Electric power system (EPS) need to ne HVAC to the battery charges is Wakefield show contributors of HVAC for battery explicitly modeled.

chargers.

]

11.

EPS - also need to document an ne dismissal as an IE is included in the Wakefield agreement as to why loss of all HVAC accident sequence discussion of %X' to UPS room as an initiating event is so Due to the plant configuration, the low in frequency that it need not be system is not very recoverable. He modeled, or include a recovery action reliability after a LOOP is explicitly because loss of all UPS should be a core modeled.

melt due to failure of all operator action.

Also consider the HVAC reliability under LOSP condition.

12.

EPS - De F.O. transfer pump looks like ne number of demands has been Wakefield i

it models only 1 pump start, yet more changed. (also for SW and CW screen starts may be required within 24 hours2.777778e-4 days 0.00667 hours 3.968254e-5 weeks 9.132e-6 months .

wash systems) 13.

EPS - the discussion of results for Hey accurately depict the dominant Wakefield electric power is currently incomplete, it unavailability of EP. Will add more only looks at 6.9KV AC.

discussion to future revision.

Page 12 of 47

I-i I

CPSES Unit 1 Independent Review ofIPE Study j

by Outside Consultants No.

Comment

Response

Reviewer 14.

Ref. comment on RHR miniflow in CCW as a support for events in which Wakefield -

section 2.1 - De miniflow line goes the system was required to maintain l

back to the suction of the pump, with no miniflow for long periods, should have CCW aligned to the HX, the pump could been included. The model will be overheat before swapover to changed (or this will be calculated recirculation. At Diablo Canyon, we away), However, room cooling is estimated / calculated the time to overheat required for success, and it would not at 4 hours4.62963e-5 days 0.00111 hours 6.613757e-6 weeks 1.522e-6 months . I disagree with the available if CCW were unavailable.

conclusion as stated. Reaction to first Herefore, this does not need to be turnoff the pump should be included.

rerun.

Caution in FRC4.1 is too late to be of any use.

15.

For RHR LP injection; shouldn't the One of 4 with one auto faulted is 1 of 3.

Wakefield success criteria RBD be 1 of 3 cold leg injection paths; i.e. with the other path being the source of the break. Yes, analysis agrees in section 4, but doesn't mention in 2.1 and 3.1.

16.

RHR system - is miniflow assumed It is, conservatively.

Wakefield required for large LOCAs too? It should not be.

17.

RHR System - The manual action for Yes. Also the manual action of swap Wakefield sump swap over is said to be excluded.

after autoswap failure was imbedded.

Yet for SLOCA, it appears in cutsets "and" with late swapover, per discussions; the SLOCA actions will be combined into one action.

18.

CCF of miscalibration oflevel The defense of the development of this Wakefield j

transmitter; this value of 2.56E-6 seems probability is in the HRA notebook.

i very low; especially as a screening valve.

19.

I can see why you don't have to model The discharge head of the RHR pumps Wakefield closure of the RHR RWST suction valve will overcome the static head of the for HPR; but what of out the CCP/SI RWST to ensure a suction supply. Also, l

RWST pump suction valves 7 there are check valves that isolate flow l

back to the RWST.

Page 13 of 47

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants l

No.

Comment

Response

Reviewer 20.

Credit for providing makeup to the Added discussion to the accident Wakefield RWST to obviate the need for sequences.

recirculation is not modeled. This point l

needs to be expanded in the sequence i

model; especially for SG tube ruptures.

)

i 21.

RHR system - FCV-611 is listed in These valves (610 and 611) are MOVs Wakefield segment 2 of RBD but is not listed in that are initially open and are not

{

section 2.1. table of supports required, required to close for success. Derefore, How does FCV-611 fail on loss of they do not require their 480V source i

control power; probably a 118v for success.

instrument bus? Does it fail open or closed?

j f

22.

RHR system - page 26 - Is it really true Yes. The second will run out due to too Wakefield that failure of 1 RHR pump and to many flowpaths.

separate the train results in failure of the other pump?

23.

RHR system - I can't see why ne model has been changed to prevent Wakefield misaligning 1-8840 open in RHG1 would this. The latent His are screening.

fait low pressure injection? He misalignment HI's are very high.

24.

For RHR system - the inadvertently he latent HIs are screening.

Wakefield disabled value for valve 1-8811 A in RHSEGA10 is very high; greater than FTO in demand number. This is excessive.

25.

RHR system - the top logic for the RHR Re top events ! cave the model Wakefield system must combine the top events separately, ani are combined in the considered there to complete all the logic accident sequences.

for say HPR. It would be nice to have a sheet showing how all the tops come together to complete this function; in the system notebook.

)

I i

l Page 14 of 47 1

l CPSES Unit 1 Independent Review ofIPE Study by i

l Outside Consultants No.

Comment

Response

Reviewer 26.

De CCF terms for RHR system event They do not include double event Wakefield RHCCFCLR seem to include double failures. Dere may be some double event failures inve'.vmg CCF events and counting in that some valves are counted other train failures. However, the in RHR CCF, and in high head injection frequency of these CCF-events are CCF.

j.

probably already accounted for by using I

the total failure rates in the non-CCF part of event tree; it looks like some of l

the double cutset CCF events used were l

double counted.

l 27.

Containment isolation - If the It is assumed that this will not occur.

Wakefield containment sump pump discharge line See new contalment isolation fails to isolate, are there calculations to assumption.

t l

show that the potential loss of sump water during a LOCA would not fail l

recirculation i.e. could failure to isolate i

the line cause failure of recirculation?

28.

If the containment pressure relief line is It is conservatively assumed not to Wakefield initially open and failed to close at the impact the containment pressure start of the accident; how would this excursion, and lead to containment impact the sequence progression?

isolation failure.

l 29.

Preexisting leaks are not modeled in the This has been acknowledged by Hugo, Wakefield containment isolation model. This was and not dominant at CPSES.

found to demonstrate in the Seabrook EP7 study.

30. of containment isolation We do not credit equipment failures for Wakefield analysis. The exclusion of the RCP seal successes. Attachment 3 of the analysis return line is unusual. A picture would debates it away, based on system be helpful to show the flow path and configuration pressure limits. Is it fair to say that the relief valves on this line would limit the containment overpressure value if this line were not isolated in a station l

blackout? During the initial stages of an l

RCP seal failure, could the relief valves i

lift and possibly not reseat due the l

relatively large flow rates?

Page 15 of 47

\\

CPSES Unit 1 Independent Review of IPE Study by Outside Consultants No.

Comment

Response

Reviewer 31.

Service water - Has the possibility that No, because the failure of the Unit 2 Wakefield the reason U1 fails is because a pump pump was not considered, because it is stops and its check valve faih to seseat swamped by the HI Also, the check leading to diversion of flow from the valve failure would have to be other pump? Under this condition, compounded by the MOV failure to would cross-ticing to U2 do any good?

actually divert flow, l

i Page 16 of 47

^

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 1

1.

Dependence between dynamic action More discussion will be added.

Wakefield i

1 HEPs is not considered in attachment 3 screening tree. Dese is very little discussion of this issue.

2.

The time available to complete each Will add.

Wakefield dynamic action should be listed in the screening tables; and be consistent with the success criteria and assumed plant impact if the action succeeds or fails.

3. - &CHSTART assigned The procedures were changed after it Wakefield value was found to be of great importance, which was after the interview with the is not consistent with operators operators.

Judgements. Why?

De opening of the block valve is an For RC8000 A/B; I don't think automatic, trained action upon a you can credit operators opening transient.

PORV block valve in time to mitigate RCS pressure peak during an ATWS. Believe that step is later in the FRS procedure.

4.

An example of dialogue with NRC in the Ok, Wakefield issue ofloss of SW with consideration of cross-tie with unit 2 and alternate cooling to changing pumps could be obtained from PG&E or PLG, for Diablo Canyon.

5.

Action SSXTIE is discussed, but there is Will add reference to the discussion.

Wakefield no reference to a procedure.' All such actions should have procedures, else NRC will balk.

6.

De &TURBTRIP action should be It is, effectively.

Wakefield dependent on the status of the action to backup manual reactor trip.

Page 17 of 47

l-..

i t

l l

4 l

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants i

l No.

Comment

Response

Reviewer l

1 j

7.

Actions following loss of UPS HVAC They are.

Wakefield without recovery of HVAC, should be given very litt!< credit; i.e. without instrument:, tion, all bets are off.

(

l l

1 I

l I

I 1

l i

Page 18 of 47 i

i CPSES Unit 1 Independent Review ofIPE Study l

by l

Outside Consultants i

s l

No.

Comment

Response

Reviewer l

1.

Not all initiating events alone have to ne support system initiating events Wakefield j

cause a plant trip. Loss of an were calculated including the failure of a l

emergency AC bus, combined with other bus. Although they are not transferred, failures, would cause a plant trip. e.g.

the effect of loss of service water, is just loss of an AC bus leading to CCW loss as impactive.

J is an IE that is currently not included in i

the model because the impact of the bus l

failure is not transferred from the IF to the accident sequence fault trees.

l 2.

Other HVAC systems failures may be of He loss of UPS hvac is debated away in Wakefield interest than event YY e.g. loss of the accident sequences in the Loss of l

HVAC to the UPS systems should be HVAC section considered.

3.

Excess LOCA should be defined as The write up has been changed.

Wakefield requiring greater makeup than the large LOCA design basis event; i.e. a LOCA in the Reactor Vessel itself of size sufficient to preclude case cooling.

j 4.

The basis for saying pressurizer PORVs hisis Wakefield would not be challenged in the event of a turbine runback as from change or RCP trip should be given, when discussing IE grouping.

5.

Shouldn't RT & 'IT be separated in T1 ney do not have to be. Correction Wakefield i

i.e. for ANS events or is credit for factors are added to the ANS cutsets.

neither RT or TTs assumed for general transients? Simple Reactor trips should i

not be counted toward ATWS sequences.

6.

Section 7.0 reference for review of Hey were reviewed by the original Wakefield 1769LERS should be identified.

analyst.

s 7.'

Initiating event fault tree is a unique and H anks.

Wakefield useful tool for identifying IE. Good Job.

i Page 19 of 47

i CPSES Unit 1 Independent Review of IPE Study by Outside Consultants No.

Comment

Response

Reviewer i

8.

Suggest that IE fault tree be Will add to a future revision.

Wakefield supplemented by FMEA of support systems / trains specific for Comanche Peak. ie. list each support train or identify what direct impact would occur N

at plant. This would help explain why its ok to group all DC buses in 1 event.

Question naturally revises as to whether the impacts are symmetrical. Right now, the completeness of the support system IEs is suspect.

9.

For steamline breaks; it would be nice to Will add to future revision.

Wakefield i

see the different impacts of where the break is i.e. into which building or room. He current discussion does not say how the environmental impacts are modeled by 1 category.

10.

I wonder if the medium LOCAs don't Rey require 1 high head pump AND Wakefield also require 2 high head pumps or 1 one RHR pump.

RHR pump during the injection phase.

11.

The list of potential initiators generated Yes. They were documented for future Wakefield in the system analysis is not documented uses.

in this section.

12.

Based on the sequence modeling, I don't It has greater impact than the nominal Wakefield see why loss of condenser vacuum is trip in that it disables the steam dumps modeled as a separate IE group.

and the Main Feedwater Pumps.

13.

The discussion on requirements for Hey have been beefed up.

Wakefield different site LOCAs should be beefed up. I don't see now why very small and small LOCAs are in different initiator groups.

Page 20 of 47

1 I

CPSES Unit 1 1

Independent Review ofIPE Study by Outside Consultants i

No.

Comment

Response

Reviewer i

14.

When considering recovery of SW as an This does not apply as the fan coolers Wakefield initiating event; it may not be are tripped on a LOCA.

appropriate to recover by cross-tieing if these is a substantial heat load on the CCW system (i.e. after much review at Diablo Canyon, we had to concede that CCW may be lost in a short period of time if a LOCA to containment resulted in heating of CCW via the containment fan coolers). This may not apply to you.

15.

Interfacing LOCA. Table of Typo.

Wakefield j

penetrations say blowdown from the SG-Blowdown?

I 16.

Interfacing LOCA (RHR Lines) The The relief in containment goes to the Wakefield ECA procedures for LOCA outside PRT, so they would not see high containment may not be implemented containment radiation. The Aux, because, like at Seabrook, the EOPs first Building rad monitors would be going aimed operator to EOP-1.0 because of off to show outside containment l

the initial relief to containment, and problems, and the recovery is not there is no transfer from these back to credited at this point because calculations ECA-1.2A. Statement in section 9 may of the MOV torque for full RCS be inaccurate.

pressure, have not been done.

l 17.

ISLOCA-RHR - The capability to close We agree that if the operator action is Wakefield MOVs on cold-leg injection against high credited, the capability must be DP must be established.

demonstrated. At this point we are not l

crediting this action.

I i

i f

Page 21 of 47

i i

CPSES Unit 1 Independent Review ofIPE Study i

by l

Outside Consultants A

No.

Comment -

Response

Reviewer i

18.

ISLOCA-RHR - His analysis looks Yes, and all equations have been Wakefield very thorough and uses latest reviewed by our reviewer.

information. However,I'm familiar l

with the C-93 term, but I had trouble following its deviation in the reference, can you? I don't have time to check all aspects of this package but I'll close with a few questions:

Has the relief valve capabilities been adjusted to account for the high water temperatures / steam He relief capacities are design values, that it will be passing?

Other components than pipes may break (i.e. RHR HX was found to be limiting for Davis Besse The heat exchangers would fail into a closed, monitored system. He pump seal failures are the only concern in the RCP seal leak area?

RHR and SI pumps.

j RCP seal leaks w/ Containment Isolation failures are addressed elsewhere.

If the RHR system remains pressurized, would both RHR They would not be deadheaded in that pumps dead head on a cold leg the impeller does not introduce a injection leak and therefore not significant DP.

be available for LOCA mitigation 7 i

Page 22 of 47

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants 1

6 No.

Comment

Response

Reviewer 1.

Cutset 11 - Why does the PORV cycle This has been changed.

Wakefield 100 times? Is bleed and feed guaranteed failed?

2.

I believe the DG recovery values from Hey have been increased.

Wakefield NSAC-161 (G.15), are optimistic.

3.

Some uncertainty calculations are We plan on doing this in the future.

Wakefield needed; at least sensitivity results are required.

4.

T/H calculation notebook still sup 2 A new calculation supercedes this.

Wakefield PORVs are needed for B/F.

5.

T/H calculation 8 states that containment His is not assumed in the accident Wakefield pressure would not reach the spray sequences.

setpoint for a 2 4" diameter break. Has this been verified? With no fan coolers?

I believe it will.

Page 23 of 47

4

(

i CPSES Unit 1 l

Independent Review of IPE Study ~

by l-Outside Consultants l

No.

Comment

Response

Reviewer l

1.

For tube ruptures; an argument should Added the requirement for Wakefield be made to note that fail"re to depress depressurization.

the secondary, without continuing to feed the ruptured SG, is very low l

frequency.

l 2.

For tube ruptures; need an argument to Added.

Wakefield justify feed and bleed as a success path; since makeup to the RWST will be required eventually. His could be an alternative requirement for recirculation under F/B conditions.

3.

Large LOCA why isn't there a large Bey are swamped (and minimal to) cold Wakefield LOCA sequence with failure of hot leg leg recire, failures.

recire. failure?

4.

Medium LOCA - with failure of Only if you require the ECCS flows to Wakefield RHCCFCLR should lead to core damage remain high. After RWST injection, the at (4.65-4/yr) (3.64-4) = 1.69-7/yr.

amount of water required to prevent core Dere should be no recovery for RWST damage significantly decreases. He makeup because these is probably too recovery action of extending the little time w/ spray actuated.

injection phase (which includes shutting down containment spray) was included.

5.

PRA comparison (attachment 3), I It is not a top event on the RT tree.

Wakefield believe Seabrook did have consideration of control rod binding, though not as a separate top event in the event trees. It should be included in top event RT system model.

6.

I would think that the PDSs should His information is post processed onto Wakefield distinguish melts with and w/o the cutsets using CUTCLASS.

containment heat removal.

7.

Section 5. 0 - there needs to be a Dere is, the event tree has them listed.

Wakefield relationship between the PDS's defined in Section 4 and the end states for each sequence.

Page 24 of 47

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 8.

Dependency Table - I'm surprised that These are all correct. Pumps fail on loss Wakefield SI and CS depend on CH (chilled of room cooling.

water). Are these dependencies actually modeled? What about the dependence of The dependency matrix has been CVCS on CCW7 Operation of FW changed.

j should depend on TM (MSIVS) for flow path back to condenser. Suggest putting ESFAS is cooled by Control Room Air in " "'s for the diagnosis boxes.

Conditioning, which is cooled by CCW.

Do CCW pumps (CC) really fail on loss of room cooling via CH7 Is ESFAS only cooled by CCW, not chilled water?

9.

Success criteria (App. 2/4)

They will fail anyway.

Wakefield Wouldn't RCPs vibrate and fail seals of CCW is lost to RCP motor bearings and the pumps not tripped by operator. Is this action modeled?

l SF5000 - What is basis for only needing accumulators?

This gate is used in concert with an RHR gate.

AF4000 - For ATWS, isn't AFW flow to all SG's required for the 100% flow case?

For some ATWS cases, only half AFW is required.

" Noun Name" in table, should be Gate description.

Ann 3 - too event success criteria INCXX02: Conditions when pressurizer PORV's are challenged should be noted.

They are described in the induced LOCA section.

10.

For %T1 - I would not require hot leg Hot leg recire is not required for Wakefield recire. for B/F. IF it occurs is it melt?

success anymore.

With no recovery? Both ST2RHR hot i

leg recirculation required?

f Page 25 of 47

1 CPSES Unit 1 Independent Review of IPE Study by Outside Consultants No.

Comment

Response

Reviewer 11.

Why is containment spray asked only for It used to be required to distinguish PDS Wakefield smaller size LOCA's? Spray actuation bins. It no longer is. Spray actuation is might be anticipated for all core melts?

assumed to occur on all LOCAs, unless the system or its supports are unavaialable.

12.

I didn't think secondary heat removal is It is no longer asked.

Wakefield required for >2" size LOCA; ie for "small break LOCA". Why is it asked?

13.

Shouldn't the secondary heat removal Secondary heat removal is no longer Wakefield question for VS and S initiators be required for S.

different? Especially since for VS they must cooldown to avoid the need for recirculation.

j 14.

The assumption that recirculation is not Changed the tree to require recirculation.

Wakefield needed for LOCA's up to 2" in diameter is not standard, nor is it consistent with NUREG-1150 work. Make sure you have backup documentation. I believe its possible, but cooldown and depressurization should be required, and possibly also closed loop RHR.

15.

For very small LOCA initiators, how do Changed the tree. The PDS bin reflects Wakefield you know that 2 of the sequences only that SBO, or a combination of equipment via station blackout? (PDS3SBO, failures similar to an SBO.

PDS4SBO).

16.

Suggest that for %R, feed & bleed Feed and bleed requires recirc., or Wakefield success scenarios require makeup to the makeup, as modeled.

RWST. These should be added to the event tree.

17.

For %R, SGXXO13 should require It was changed to require this.

Wakefield secondary depressurization or RWST makeup for long term.

Page 26 of 47

CPSES Unit 1 I

Independent Review of IPE Study by Outside Consultants No.

Comment

Response

Reviewer 18.

Tree [NONLOCAINIT:]

The seals fail even if the pump is not Wakefield tripped.

$NLXXO4 should include tripping the RCPs in the event The events are listed, and include any that CCW to RCP notes bearings event leading to a RCS pressure

fail, excursion.

I can't tell which transients

$NIXXO2 requires PORVs to lift, or $NLXX03 requires SRV's to lift.

19.

I believe the sequences in The order of the tree has been changed Wakefield NONLOCANIT tree should include to check largest to smallest.

branches under PORV/SRVs even if a seal LOCA due to loss of cooling occurs. Medium LOCA could occur and recovery timing and feasibility may be different. Seal LOCA is not this problem if you've lost all secondary heat renoval.

20.

Tube rupture with induced LOCAs are Any LOCA required injection. This Wakefield j

not modelled; but could develop from would not change anything.

j loss of CCW or failure to reclose PZR 1

PORV after RCS depressurization.

21.

Operator action for SGTR This tree has been changed.

Wakefield depressurization should be different 4

value if all HPI to lost. Did HRA j

analysis assume HPI was failed?

Because that's the only place its used in the event tree.

22.

Were is ARTXX01 A1 computed for It is not computed. It is a gate with all Wakefield ATWS7 Not in IE section; initiators ORed.

23.

Does this plant have AMSAC installed?

Yes, it is credited under $TURBTRIP.

Wakefield Where is it modeled? For ATWS, doesn't RTXXO6 success criteria need to Yes, and it does.

consides the PRZ SRV's?

Page 27 of 47

CPSES Unit 1 Independent Review of IPE Study t

by Outside Consultants a

i No.

Comment

Response

Reviewer a

24.

For ATWS, not all of the questions are It depends on what question you are Wakefield asked for AFW and containment spray, answering. We are statusing the l

one cannot just do a " combine" to Containment spray and containment determine the status of these function.

isolation systems on core damage. This j

'Ihe resulting core melt cutsets must be is the correct method for this.

"AND" with the AFW and spray logic to 1

determine the proper PDS assignment.

l 25.

ATWS trip; you should ask MFW The MFW success criteria defined would Wakefield j

everywhere, even if turbine fails to trip not be able to limit the pressure l

because MFW would limit the RCS excursion if the turbine did not trip, j

pressure increase.

The success criteria would have to be revised.

f 26.

Simple RT and TT events should be Correction factors were added to the Wakefield modeled separate from the RT initiator, cutsets to correct for this.

"~~

at least for ATWS, where credit for both of the events can be credited for AFWS j

mitigation.

27.

Event tree headings - titles should either Need to clean these up.

Wakefield event success or failure consistently. I i

believe RT2000 is stating event success j

where as all others are event failures.

l 28.

Why don't the top event logic include Because FTAP would generate tons of Wakefield j

the success events? e.g. sequence success events. DELTERMing out the

ACMl?

successes makes the sequences quantify j

faster.

29.

The independent failures of containment The independent failures are very low Wakefield spray and isolation of containment must compared to the support system failures, i

be considered for each core melt which are already in the cutset.

f sequence.

f 30.

Level B page 4 - for F/B after Changed success criteria for SBLOCA.

Wakefield SBLOCA, why do the PORVs have to Don't require F&B.

i be opened? Its greater than > 2" j

diameter already.

j 1

i; l

1 Page 28 of 47 i

2 J

l CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants i

No.

Comment

Response

Reviewer 31.

I don't think a LOMFW or a LOOP (or Yes, these are conservative.

Wakefield loss of condenser vacuum) would always give a PORV i.e. without additional failures. I think the assumptions on page 8 oflevel B top logic are conservative. These are certainly conservative for SRV challenge on page I of top logic level B.

32.

I don't see why hot leg recire.

It is a seperate system function. It has Wakefield switchover for intermediate pumps been removed from $RCXX01 for other (SI3000), is separated from all others reasons.

$RCXX01 in page 18 if top logic event level B.

~

33.

I didn't look at RCS system, so I'll just It is very difficult to recreate the TH Wakefield add the question to make sure that the analysis done in the W study. Deviation large size of your PORVs is considered from success criteria requires recreation relative to the PZR SRV sizes to of the base <leck. This recreation would determine how many PORV, and SRVs generate a different initial reactivity 4

are needed to mitigate each ATWS insertion, which would probably sequence pressure rise. Your PORVs generate the same success criteria.

are probably larger than those analyzed j

in WCAP study.

1 34.

Pages 30 of top logic level B - why are It has been changed. Only RHR pumps Wakefield high pressure pumps needed? This is are now required.

very conservative licensing criteria? I believe only RHRs needed for large LOCAs, realistically.

35.

Does RT3000 event on page 33 of top It does now. It is in the MS gate.

Wakefield logic level B include failure of trip valves. Where is failure of trip valves included.

36.

Shouldn't FL-SGHH on page 35 include It should, but it is used to fail AFW Wakefield SG overfill events from AFW?

after failure of FWI. SG overfill events due to AFW control failure fail it directly.

Page 29 of 47

CPSES Unit 1 Independent Review of IPE Study by Outside Consultants No.

Comment

Response

Reviewer 1.

Multiple Human actions in some of the There are no multiple human actions in Zamani sequences. This might cause HRA cutsets (that I have reviewed) any dependencies on other actions. Even longer.

sometimes using early or late recovery for actions? Might not be conservative.

2.

In quantifying some of the sequences the Yes. But we cannot resolve this now.

Zamani conditional probabilities for tops are larger than 0.2 (see seq. #37 for j

example). Since CAFTA doesn't consider the complement (only assumes i

rare events) aren't you penalizing l

yourself.

[

3.

Are you planning to perform an Some sort of uncertainty will be done, Zamani uncertainty analysis? It seems to me but not prior to submittal.

l some uncertainty analysis may be i

expected based on 1335 report.

l e

i i

i I

f l

l Page 30 of 47 i

- -.=.

i CPSES Unit 1 Independent Review ofIPE Study l

by 4

Outside Consultants No.

Comment Re6ponse Reviewer 1.

In cale sheets RXE-SY-CP1/1-103 on One PORV is required for success. The Zamani page 8, Feed & Bleed section #), the write up has been changed, second paragraph assumes success path as one PORV being available, However in paragraphs 4 & 5 it seems that 2 PORVs must be available. Check for consistency.

2.

Report RXE-SY-cpl /1-103 5.11, Loss The room cooling effects are Zamani of CCW (X6)It is assumed that the probabilistic. Everything else is equipment cooled by CCW are not going assumed to fait directly.

to fail and only increases the failure probabilities. Has a heat load been done to see this is a good assumption for the mission time.

3.

What is the contribution of CDF from The current truncation limit is very Zamani unaccounted cut sets? Since you used a reasonable. The CDF cited is from all truncation value it might be a good idea sequences above IE-09.

to do some sensitivity runs with different truncation limit to just see how sensitive they are to this value.

4.

I was expecting to see some sequences he successes are DELTERMed away.

Zamani from #1SCM2. (since it is a success That left no sequences.

branch of #ISCM3).

5.

He dependency matrix is not detailed Will consider in a future revision.

Zamani enough for the reviewer to go through without asking questions. A more detailed matrix is suggested.

Page 31 of 47

{.

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 6.

In VSLOCA event tree, a plant damage PDS4SBO is used for SBOs that lead to Zamani state of PDS4SBO is assigned to this core damage, OR a combination of branch point.

equipment faults that results in similar equipment unavailability.

However, some of the scenarios may get there, without SBO, and through combinations of support system failures and independent failures. Even though the likelihood of these scenarios are very small however, it might be misleading, especially if some sensitive decisions may be concluded based on this find of assumptions. The same comment is applicable to PDS35BO and others if used in other events.

j 7.

To make it easier for the reviewer We are considering ESDs for the Zamani especially for the IPE summary report submittal.

l that you plan to submit to NRC, the documentation of the sequences and the way you present them to NRC is very important and sensitive. I am not too sure how you want to put so much info.

int 200 pages. May be one way to do this efficiently is using ESR (Event Sequence Diagram). That might help.

8.

In ATWS event tree loss of main feed Only if the success criteria for MFW is Zamani water should be questions when turbine 100%. It was not developed that way, fails, since if MFW success would help in preventing or at least delay core damage.

Page 32 of 47

i-l CPSES Unit 1 i

Independent Review of IPE Study by Outside Consultants i

No.

Comment

Response

Reviewer 1.

What about loss of ventilation to some Loss of HVAC is not so impactive Zamani critical rooms.

except for Safety Chilled Water.

2.

What is the criteria for binning the They are grouped by similar plant Zamani Initiating Events?

responses.

3.

Make sure the PLG generic data reflects It does not. We will update in the Zamani

~

the last few years, or at least update future.

your IE with more recent data if possible. My understanding in PLG generic data is several years old, since then the trend of plant # of trips has been going down due to better utility j

practice.

l 4.

He generic data should be screened for his only works if you add events that Zamani j

each initiator to remove the events that are specific to TU, but not elsewhere.

j do not apply to TU. This might reduce Realistically, it all evens out.

j some of the IE events. Lis screening may be performed for most dominant events. However, the values used by TH can be considered conservative, and screening only can help to remove some of the conservatism.

Page 33 of 47 j

i

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 1.

In CCW system the piping rupture is not Pipe ruptures are not considered in our Zamani modeled the pipe rupture, frequency IPE, except in flooding. For CCW, this (failure rate) can be equivalent to would have to be compounded with the component (one from each train) failing failure of the trains to separate and simultaneously, isolate.

j 2.

CCW path 1A to non-safety. Misaligned Yes. All His have been assigned values Zamani close.

use HRA techniques.

Did you used a human reliability t

analysis for this action? The value of 0.005 seems low.

i 3.

In CCW I did not see any relief valve or All vents and drains are very small Zamani drains to be modeled. Failure of these compared to piping size. Additionally, components if not recovered may cause the mission time failure of a manual j

system failure or at least a train failure.

valve is very small compared to anything i

else.

4.

In CCW the relay which actuates and The relay is considered to be part of the Zamani starts PMPA is not modeled here. Is the breaker, which is modeled as part of the relay modeled somewhere else or not at pump. The ESFAS relay is modeled 1

all?

separately.

5.

Has contamination of IA been modeled?

No. It is considered to be part of the Zamani initiator Loss of IA.

6.

A brief description of the systems can This will be provided in the submittal.

Zamani help the reviewer to understand systems better.

7.

Loss of ventilation to switch gear and It was modeled. Then the SBO room Zamani UPS room should be modeled in heatup calculations showed that the electrical systems or a dependency.

temperature would not exceed the EQ limit after 24 hours2.777778e-4 days 0.00667 hours 3.968254e-5 weeks 9.132e-6 months .

- - = =

x -

- = ~ - - - - - = = = -

=_m-..===--

= - - - - = = = -

!amminammessumamu No.

Comment

Response

Reviewer 1.

The dependencies in Human actions They have been.

Zamani should be reviewed in detail to make sure you are not taking credit where you should not.

Page 34 of 47

.-... ~ -. - - -. -..

i 1

1 i

CPSES Unit 1 l

Independent Review ofIPE Study by

+

Outside Consultants i

I f

4

)

No.

Comment

Response

Reviewer i

2.

It would have been more accurate to get he PSFs were discussed implicitly.

Zamani

'l

}

the operators opinions at the PSF level l

and then use their results to come up with the final value. Rather than ask them to guess the final values form the j

sequences you discuss with them.

i 3.

The human action value for the basic De discussion effectively arrives at this Zamani

{

events ESCCFMISCAL in 2.5*104 value.

l Dis value seems to be very low.

4.

What is the basis for probability value Dey have been compared to other Zamani l

selected in screening value Decision studies via NUCLARR and shown to be j

j Tree, and whether these values have conservative.

j been benchmarked?

5.

For action &CHSTART the operators ne procedures were modified after the Zamani gave the rang;. from 1/5 to 1/100 (1/5-operator interviews.

j 1/40 and 1/10 -l /100) i l

De selected value, however is 1/100 =

l 1*104 his value seeing to be not t

conservative, f

6.

In human action analysis report Changed.

Zamani j

references 8 & 14 are the same? (this is a minor comment, just to be consistent).

i 7.

Recovery of faulted equipment should be ne fact that the recovery factors are not Zamani used very wisely, make sure you only 0 accounts for the recoverability factor.

l recover the recoverable portion of any i

failure and do not use the given values in l

EPRI publications blindfully. NRC may have a good reason to question, given j

that you did not study each faulted event carefully, i

4 1

4 i

)

Page 35 of 47 1

i

. - - - - _ =. -

\\

}

i CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants i

No.

Comment

Response

Reviewer i

l 1.

The flooding impacts on top events or Will consider.

Zamani

)

trains are very difficult to follow without an impact matrix. I suggest to develop a matrix for ease of review a better presentation.

?

2.

The sequences and cutsets were not True.

Zamani available during the review period.

3.

Recovery actions and HRA were not True.

Zamani available during the review period.

4.

Flooding analysis report page 15, states No.

Zamani that human action in-control room were changed to 1. If flood started on propagated to the control room. What about the components that can be controlled from the control room, but the flooding already has failed them?

Did you consider any recovery on those?

i l

)

Page 36 of 47 J

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 1.

Will the op actions to arrest the flood Both will be considered.

Hubbard be necessary or will the non-flood recovery numbers already built into the i

model be enough to make the core damage frequency manageable? Without these flood recoveries the impact even from closed system flood may be too conservative. No such actions are apparent either in this report or in the human actions report. If such actions are added they must be carefully coordinated with the non-flood recoveries to account for any dependencies between actions.

2.

You may need to make the frequencies We have changed this.

Hubbard depend on break size as you include the operator actions to mitigate. Most breaks that have occurred have been

" leaks" not ruptures. The size of the break will considerably impact both the time available for op, response and the impact of the event on plant equipment.

3.

What is your conclusion, can the normal Only if the room and the path to the Hubbard dynamic and recovery actions be room is not affected.

performed during the flood? I suggest that you also eventually include in this report your conclusions on this issue.

Page 37 of 47

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 1

4.

It was difficult to determine from the Will consider.

Hubbard information available in the draft report what the impact of each flood was, i.e.

what initiating event resulted and what PRA systems were affected. Something like a dependency matrix would help here. Something which shows what fails as the flood reaches successively higher levels in the aux. bldg., would also help.

Presumably the flood recoveries will stop the flood at various levels with various frequencies. Depending on what the times are involved with each level the op. action failure rate will vary. I know you have very carefully cataloged all the equipment in each fire zone, therefore you could easily do this.

5.

More details of the drain system need to Will consider, but the write up is based Hubbard be provided here in order to review the on the system's design capacity of 50 accuracy of their incorporation into this gpm.

analysis:

drain sizes valve types and locations pump sizes and starting logic, etc 6

The temporary high levels caused by

'Ihis addressed by faulting equipment 1 Hubbard water flowing through rooms needs may foot and lower.

fail equipment not affected by where the water eventually collects. I could not identify and review your assumptions about this effect.

/

Page 38 of 47 i

t CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 1.

Human actions, especially recovery We will be ready.

Hubbard actions, are more important to the outcome of this IPE than to any of the others that I am familiar with in detail.

[My scope of detailed knowledge does not include any other Putney stjie linked fault tree.] However, I sense that the NRC is sensitive to the impact of HRA on the IPE results. (See NUREG-1335, page 2-8, item 6). Whether this level of operator actions is common to other PRAs or not it will surely attract the NRCs attention and it is my recommendation that TU be ready to deal with this issue. Dealing with it certainly means listing the 50 sequences requested in 1335.

2.

Because of the importance of operator We plan on increasing the work in the Hubbard actions to the calculated core damage HRA field in the future.

frequency, I believe that the operator interviews need to be more structured and bater h-ted than what is shown in the current draA of this report.

Otherwise it will be difficult to justify the wide spread use of these numbers.

j Page 39 of 47 J

O I

i l~

l CPSES Unit 1 l

Independent Review of IPE Study by Outside Consultants i

e f

No.

Comment

Response

Reviewer l

3.

More detailed operator action All sections of the HRA cale are being Hubbard descriptions should be made which beefed up, describe the following:

I time available procedure steps used f

applicable scenarios j

i l

instrumentation needed i

support requirements for the instruments their importance l

relative to the plant hardware in controlling CDF at CPSES and their higher relative subjectively i

both suggest that op, actions should be documented in mote detail than they are now and analyzed in comparable detail.

l

[See attachment for a suggested format].

4.

For my taste, operator action values are The values used are meant to reflect to Hubbard l

used to generically. Take B&F as an most conservative application of the HI.

example:

]

after TOLFW

{

after Inadvertent "S" signal after failing to control afw flow i

After having said this, I think you can probably justify this usage, but I think it

{

will be necessary to document this i

justification more than is in the report i

draft which I have reviewed.

i e

Page 40 of 47 l

i

4

)

4 CPSES Unit 1 J

Independent Review ofIPE Study j

by Outside Consultants i

l No.

Comment

Response

Reviewer 5.

Multiple operator actions which occur We believe that they are accounting for Hubbard j

following the same plant trip are in the dependencies as modeled.

l general dependent on each other, ie. the i

operator may be less likely to perform i

one manipulation correctly if he has, previously in the same scenario, failed to j

perform so other action another.

4 Dealing with such dependencies may require multiple versions of the same l

action which are conditioned on the j

success or failure of the previous action, j

All operator actions used should be j

categorized according to where they are used and what actions precede them.

i 1

Page 41 of 47

t i

i CPSES Unit 1 i

Independent Review of IPE Study by l

Outside Consultants 6

i No.

Comment

Response

Reviewer i

j 1.

He thermal fragility factors mentioned Only if you don't believe in statistics.

Hubbard i

in top cutset should probably apply to he values we use are calculated based two systems at once because if the room on physical processes in the motor.

temperature is known then the only Hey are statistically independent.

I unknown is failure temperature at which j

the pumps quit working. His failure j

temperature would most likely be pretty i

similar to both.

I I

2.

Discussion of success criteria for B&F It has been revised.

Hubbard on sheet 8 should be enhanced up. It is I,

confusing.

1 3.

Is piggy back recirculation required for No. So do we.'

Hubbard j

large LOCA7 Other plants which have i

this arrangement recirculate with only j

the low pressure pumps.

4.

How was the overcooling potential of Overcooling itself is a threat to fuel clad Hubbard

}

scenarios like turbine trip failure dealt damage, but not to core melt. With clad with? Only SLB is mentioned?

damage, a small fraction of fission Overcooling is potentionally important products, those in the clad gap, are for three reasons:

released. It takes fuel melt to get l

significant releases.

j 1)

Low temperature overpressurization i

l 2)

"S" signal generation l

3)

Requirement for throttling high j

pressure injection pumps before i

they force water out of the l

primary safety valves.

5.

Are the environmental impacts of the Indirectly. De flooding analysis does Hubbard downstream steamline breaks treated in consider the steamlines to be a source of the internal flooding analysis?

the flood.

l 6.

What about stuck open MSIV and/or afw Stuck open MSIVs are only a problem Hubbard pump run out? Where is the impact of with SGTR. AFW pump runout is either of one these failures dealt with7 prevented by flow orfices.

1 i

Page 42 of 47 4

l

l i

CPSES Unit 1 Independent Review of IPE Study by Outside Consultants i

No.

Comment

Response

Reviewer 7.

It should be clearly stated whether you We do not credit reflux boiling directly.

Hubbard I

are crediting reflux boiling heat transfer We consider failure to be when the core or not. The peak cladding temperatures is uncovered, not when PCT exceeds 3

may be considerably higher, but if you 2200 F.

don't worry about maintaining j

pressurizer level and the RCS subcooled, then reflux boiling is the alternative.

l 8.

It appears that AFW control requires The action does require the operators to Hubbard 4

controlling the feed to all four SGs.

control all, but the logic only fails if the Wouldn't that assumption produce operator fails SG #1 or 4.

conservative results? Only two SGs feed j

the steam driven pump therefore only i

these two could send water to the pump i

turbines.

9.

There seems to be a case of circular Too much steam relief is not considered Hubbard reasoning here relative to the ADVs.

a failure, as stated above.

i ne system analysis states that ADVs will not be questioned to open because there is so much redundant relief capacity. However, that is no guarantee that they don't get open, and need reclosing.

Page 43 of 47 m

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 10.

Why does very small LOCA not need It does. It needs both.

Hubbard safety injection if secondary system heat removal is available? A 2" break would probably empty the RWST within 24 hrs especially considering that the -

sprays are actuated on high containment pressure. Secondary system heat removal, if it only removes decay, and not latent heat, will not preclude the necessity of injection. Other things would be needed for cooldown to cold shutdown including:

TBVs and ADVs operator action to cooldown some way of depressurizing like aux. sprays or PORVs 11.

Somethiag needs to be done to gracefully Recirculation is now required.

Hubbard I

end very small LOCAs. The second highest frequency scenario is a small LOCA with recirculation failure yet very small LOCAs don't require recirculation.

If s.s. heat removal works. Since the IE frequency. is a factor of 10 greater, the question about what happened to s

recirculation. begs to be asked. Sprays coming on will make very small LOCA need recirculation whether or not B&F is needed. Hugo says spray actuation setpoint is reached in 3.6 hours6.944444e-5 days 0.00167 hours 9.920635e-6 weeks 2.283e-6 months for a 2" diameter break.

Page 44 of 47 a

4 l

CPSES Unit 1 Independent Review of IPE Study by i

Outside Consultants l

No.

Comment

Response

Reviewer 12.

Inadvertent ESFAS ("S") or overcooling It is considered as a source of an Hubbard leads to the following things that might induced LOCA.

be important:

i overfill and shutoff HPSI loss of instrument air Less of instrument air in turn will:

fail PORVs after accumulator. (1 l

hr) make ADVs unavailable (5 hr) l disable aux. spray??

It is not clear how all of these l

complicated scenarios from overcooling l

have been dealt with in each potential case of evercooling.

13.

On SGTR because ofloss of instrument The modeling takes care of this.

Hubbard j

j air on "S" you don't have TBVs. Be

)

careful not to credit them in the modeling oflong term heat removal following SGTR.

14.

Can't do b&F after an hour without PORVs are nitrogen powered.

Hubbard restoring air to the PORVs unless it is j

done with the safeties. Is restoring air 1

included in the operator action for B&F7 l

15.

It looks to me as if AF1000 doesn't have This requirement is modeled from the Hubbard any ADVs or TBVs in it so it can't be SGTR side, not the AFW side.

l used as a cooldown top or as a top for SGTR which maps in the supports l

required for cooldown using the secondary system. For instance, if you i

want to track the supports required for TBVs.

I i

l l

Page 45 of 47 l

CPSES Unit 1 Independent Review ofIPE Study by Outside Consultants No.

Comment

Response

Reviewer 16.

With such a relatively abbreviated set of ESDs are being drawn.

Hubbard scenarios, something more needs to be done to record what happened to those that are "normally" considered but don't appear explicitly in the model. Your small event tree model seems to need ESDs or some such thing that records each scenario and what happened to it.

(To my knowledge all LET-SFT PRAs have qualitative event sequence diagrams besides quantification structures, usually event trees.) It would help the reader a lot if they also contained the recovery actions added on a cutset by cutset basis.

All the containment safety features should appear in these qualitative pictures of the sequences.

17.

"Dus recirculation. would have only a ne phrase has been revised.

Hubbard marginal cooling effect on the RCS or containment." What does this statement mean? It needs clarification.

18.

It seems to me that a more extensive Will consider revising. Coupling Hubbard dependence matrix than the one in fractions don't mean anything in SET-Appendix 1 will be required to satisfy LFr models.

the NUREG 1335 request for one. In your case it could also contain the

" coupling fraction" instead of just using "x's.

19.

Within 24 hrs the operator will have to ne tree has been changed to require Hubbard do something else to avert core damage break flow termination.

than is shown on the SGTR event tree.

Will this all be handled with manually inserted recovery action?

20.

" Consideration is also given to the ne initiating event analysis grouped Hubbard analysis performed to develop the LOCAs by success criteria, not by size.

I initiating event frequency, Herefore..."

This was subsequently modified to be a i

What does this statement refer to?

combination of both.

l l

Page 46 of 47 J

i r

\\

CPSES Unit 1 l

Independent Review ofIPE Study I

by Outside Consultants i

No.

Comment

Response

Reviewer l

Even with a medium LOCA the Agree. This is pretty conservative.

Hubbard l

accumulators and the low pressure l

pumps will probably not be required, l

Think of a boiling pool - you only need enough water to replace the water being boiled off by decay heat. Only one or two high pressure pumps would makeup l

decay heat. Level 2 analysts would probably have a hard time getting core damage out of a scenario which is sent to melt only because there weren't two j

accuraulators available>

22.

I am unable to conclude definitively They do. We classify the resulting Hubbard whether the plant damage end states are cutsets and include containment spray adequate to this job. They appear to not and containment isolation status.

transmit from level I model the same i

amount of information as is transmitted l

from the Level 1 of other PRAs with l

which I am familiar. I believe that your I

plant damage state assignments should l

include isolation valve closure information and information about stuck open steam generator safety valve after a steam generator tube rupture.

23.

Instead of heroic actions to recover We have revised the sequences to Hubbard recirculation, it would be wiser to try include recovery of MFW.

reestablishing MFW and preclude the necessity for so much B&F. Since the operator will be trying to do this anyway it is confusing not to at least account for his success or failure. 'Ihe outcome l

might adversely affect the scenario and should be accounted for in projecting the likelihood of subsequent op, actions being performed successfully.

1 l

4 Page 47 of 47

T

\\

i 1

i i

l l

l i

ATTACHMENT 3 TO TXX 96427 DRAFT REVISED PLANT SPECIFIC DATABASE FOR LIVING IPE/ UPDATE 1

l i

l

\\

I t

i l

1 l

l l

l 1

REVISED DATABASE FOR LIVING IPE/ UPDATE

\\

I t

I l

l

l l

Revised Database for Living IPE/ Update Page1of18 R&R-PN-008 3

Rev.0 DRAFT 6/28/96 l

I TABLE 4.1.1. CPSES UNIT 1, COMPONENT FAILURE DATA BASE J

I COMPONENT /

FAILURE MODE FAILURE UNIT

DATA SOURCE i

i CODE RATE 1

ROTATING EQUIPMENT CHILLERS (CU)

FAIL DURING OPERATION 4.20E-05 H

PID4500 VOL.2 FAIL TO START 1.69E-03 D

PLO 0500 VOL.2 COMPRESSORS AIR FAIL DURING OPERATION 8.22E-05 H

PID 0500 VOL.2 (PA) l FAIL TO START 2.00E-03 D

PLO-0500 VOL.2 DIESEL FAIL TO START 7.10E 03 D

PLG-0500 VOL.2 GENERATORS (DO)

FAIL DURING FIRSYI:OUR 9.03E-03 H

PID 0500 VOL.2 FAIL AFTERFIRST HOUR 1.16E-03 H

PID 0500 VOL.2 FAN, SMALL TAILDURING OPERATION 1.58E 06 H

PLG 0500 VOL.2 I

(VENTILATION)(FN)

FAIL TO START 4.47E 04 0

PIE-0500 VOL.2 M-D PUMPS, FAIL DURING OPERATION 7.15E 06 H

PLO 0500 VOL.2 j

OPERATING (PO)

FAIL TO START 839E.04 D

PID 0500 VOL.2 M-D PUMPS, FAIL DURINO OPERATION 238E-05 H

PLO 0500 VOL.2 STANDBY (PM)

FAIL TO START 1.25E-03 D

PLO-0500 VOL.2 l

PUMPS, TURBINE-FAIL DURING OPERATION 3.10E 05 H

PID-0500 VOL.2 DRIVEN (PT)

FAIL TO START 2.07E-03 D

PLO 0500 VOL.2 AIR COOLER (AC)

FAIL DURINO OPERATION 9.84E-06 H

PLO-0500 VOL.2 FAIL TO START 4.84E-04 D

PIE-0500 VOL.2 I

VALVES AND DAMPERS DAMPERS, MANUAL TRANSFER OPEN/CLOSE 4.20E.08 H

PLG-0500 VOL2 (DX)

DAMPERS, MOTOR.

FAIL ON DEMAND 430E.03 D

PLD4500 VOL 2 l

OPERATED (DM) l TRANSFER OPEN/CLOSE 9.27E 08 H

PLO-0500 VOL.2 DAMPERS.

FAIL ON DEMAND 1.52E 03 D

PLO-0500 VOL.2 PNEUMATIC (DA)

TRANSFER OPEN/CIDSE 2.67E 07 H

PLO4500 VOL.2 VALVES, AIR-FAIL ON DEMAND 1.52E-03 D

PLO 0500 VOL.2 OPERATED (VA)

TRANSFER OPEN/CLOSE 2 67E-07 H

PIM-0500 VOL2 1

i Page 26 l

Revised Database for Living IPE/ Update Page 2 of18 R&R-PN-008 Rev.O DRAFT 6/28/%

TABLE 4.1.1. CPSES UNIT 1 COMPONENT FAILURE DATA BASE COMPONENT /

FAILURE MODE FAILURE UNIT

DATA SOURCE CODE RATE VALVES, CHECK FAIL ON DEMAND 1.99E 04 D

PIh4500 VOL.2 (OTHER THAN STOP VALVES)

GROSS REVERSE LEAKAGE 5.36E 07 H

PIS 0500 VOL2

)

TRANSFER CLOSED' PLUG 1.04E 08 H

PID 0500 VOL.2 I

VALVES, CHECK FAIL ON DEMAND 913E-04 D

PID 0500 VOL.2 (STOP VALVES)

(yy)

CROSS REVERSE LEAKAGE 5.36E 07 H

PLO 0500 VOL.2 l

6 TRANSFER CLOSED / PLUG 1.04E 08 H

PID-0500 VOL.2 VALVES, ELECTRO-FAIL ON DEMAND 2.28E 03 D

P14 0500 VOL.2 HYDRAULIC (EXCEPT TURBINE STOPCONTROL TRANSFER OPEN/CIDSED 2.67E-07 H

PLO 0500 VOL.2 VALVES)(VH)

VALVES, MANUAL TRANSFER OPENELOSED 4.20E 08 H

PLO 0500 VOL.2 (VX)

FAIL TO REPOSITION 1.00E 04 D

IDCOR IPEM 2.41 A VALVES, MOTOR-FAIL ON DEMAND 1.72E 03 D

PLO-0500 VOL.2 OPERATED (VM)

TRANSFER OPENCLOSE 9.27E 08 H

PLO-0500 VOL.2 VALVES, RELIEF (2 FAIL TO CIDSE 8.88E-03 D

PLO 0500 VOL 2 STAGE TARGET ROCK)(VK)

FAIL TO OPEN 9.03E-03 D

PID-0500 VOL,2 VALVES, RELIEF FAIL TO OPEN 2.42E-05 D

PLO 0500 VOL.2 (OTHER THAN V OR SAFETY TRANSFER OPEN 6.06E-06 H

PLO 0500 VOL.2 VALVES,(PORV)

FAIL TO CLOSE 2.18E 02 D

PLO-0500 VOL.2 (VP)

FAIL TO OPEN 4.16E 03 D

P14-0500 VOL.2 TRANSFER CLOSEDOPEN 2.67E-07 H

PLO4500 VOL.2 VALVES, SAFETY FAIL TO OPEN 3.28E-04 D

PLO-0500 VOL.2 (VF)

FAIL TO RESEAT AFTER 2.87E 03 D

PLO 0500 VOL.2 STEAM RELIEF FAIL TO RESEAT AFTER 1.00E-01 D

PLO 0500 VOL.2 WATER RELIEF FAIL TO RECIDSE 2.87E 03 D

PIA 0500 VOL.2 VALVES, SOLENOID FAIL ON DEMAND 2 43E-3 D

PLO-0500 VOL 2 (VS)

Page 27

Revised Database for Living IPE/ Update Page 3 of 18 R&R-PN-008 Rev. 0 DRAFT 6/28/%

TABLE 4.1.1. CPSES UNTT 1. COMPONENT FAILURE DATA BASE COMPONENT /

FAILURE MODE FAILURE UNIT

DATA SOURCE CODE RATE TRANSFER OPEN/ CLOSED 1.27E 06 H

PLO-0500 VOL.2 VALVES, TURBINE FAIL ON DEMAND 1.25E44 D

PLO-0500 VOL.2 STOP/ CONTROL (VU)

TRANSFER CLOSED 2.88E45 H

PID-0500 VOL.2 TRANSFER OPEN 1.24E45 H

PIB4500 VOL.2 PRESSURE VF3SELS ACCUMULATORS, RUPTURE / LEAK 1.15E 06 H

PLO-0500 VOL.2 SCRAM HEAT RUPTURE / LEAK 1.73E46 H

PID 0500 VOL.2 EXCHANGERS(HX) j PIPES (GREATER RUPTURE / PLUG (PER 8.60E 10 H

P14-0500 VOL.2 THAN 3-INCH SECTION)

DIAMETER)(PP)

PIPES (LESS THAN 3-RUPTURE / PLUG (PER 8.60E 09 H

PLO-0500 VOL.2 INCH DIAMETER)

SECTION)

(PI)

TANKS, STORAGE RUPTURE / LEAK 2.64FA8 H

PLO 0500 VOL2 (TK)

EXPANSION JOINT RUPTURE 8.60E 09 H

PLG-0500 VOL.2 i

(ED 1

i FLOW ELEMENT PLUO/ RUPTURE 8.60E-09 H

P14-0500 VOL.2 (FE)

FLEXIBLE HOSE PLUO/ RUPTURE 8.60E-09 H

PID-0500 VOL.2 (FH)

FIDW ORIFICE (OR)

PLUO/ RUPTURE 8.60E 09 H

PLO 0500 VOL2 d

NOZ71ES, STRAINERS, SUMPS, AND HLTERS FILTERS, AIR PLUG 5.83E 06 H

PLO-0500 VOL.2

HLTERS, PLUG 9.01E-06 H

PID 0500 VOL.2 COMPRFSSED AIR FILTERS, OIL PLUG 1.76E45 H

PLO 0500 VOL.2 REMOVAL

FILTERS, PLUG 1.07E-06 H

PID 0500 VOL.2 VENTILATION 6V)

Page 28

Revised Database for Living IPF/ Update Page 4 Of18 R&R-PN-008 J

Rev.O i

DRAFT 6/28/96 TABLE 4.1.1. CPSES UNIT 1, COMPONENT FAILURE DATA BASE COMPONENT /

FAILURE MODE FAILURE UNIT

DATA SOURCE CODE RATE NO7M M.

PLUG 7.06E48 H

P14-0500 VOL.2 CONTAINMENT BUILDING SPRAY (ONE TRADO(NZ)

STRAINERS, PLUG 621E46 H

PLD 0500 VOL.2 SERVICE WATER (FL)

SUMPS, PLUG 1.00E 05 H

PLO 0500 VOL.2 CONTAINMENT (SU)

TRAVELING FAIL TO OPERATE 4.47EM H

PIA-0500 VOL.2 SCREEN (TS)

ELECTRICAL EQUIPMENT BATTERIES,125V FAIL ON DEMAND 4.84E-04 D

PLO 0500 VOL.2 DC (BT)

FAIL DURD 0 OPERATION 6.65E 07 H

PLG-0500 VOL.2 BATTERY FAILDURING OPERATION 3.85E 06 H

PLG4500 VOL.2 CHARGERS (BC)

BISTABLES(BI)

FAIL ON DEMAND 3.89E-07 D

PLO-0500 VOL.2 SPURIOUS OPERATION 2.21E 06 H

PLD-0500 VOL.2 BUSES (BU)

FAIL DURING OPERATION 4.98E 07 H

PLD 0500 VOL.2 CABLES, CONTROL FAIL OPEN OR SHORT 4.64E 06 H

PLO 0500 VOL.2 (CA)

CIRCUIT BREAKERS FAIL TO CLOSE 1.61E 03 D

PLD-0500 VOL.2

(>=480V AC)(BA)

FAIL TO OPEN 6.49E-04 D

PLJ 0500 VOL.2 TRANSFER OPEN 8.28E 07 H

PLD 0500 VOL.2 CIRCUIT BREAKERS FAIL TO CLOSE 2.27E 04 D

PLO 0500 VOL.2

(<480V AC)(BB)

FAIL TO OPEN 839F 04 D

PLO-0500 VOL.2 TRANSFER OPEN 2.68E-07 H

PID 0500 VOL.2 FUSES (FU)

FAIL OPEN 920E 07 H

PLO 0500 VOL.2 INVERTERS (IV)

FAIL DURING OPERATION 2.24E 05 H

PLO 0500 VOL.2 i

i MOTOR FAIL DURING OPERATION 3.59E 05 H

PLO-0500 VOL.2 i

OENERATORS(MO) j Page 29 i

r I..

a

Revised Database for Living IPE/ Update Page 5 of 18 l

R&R-PN-008 Rev.O DRAFT 6/28/96 i

l TABLE 4.1.1. CPSES UNIT 1, COMPONENT FAILURE DATA BASE COMPONENT /

FAILURE MODE FAILURE UNIT

DATA SOURCE CODE RATE POWER SUPPLIES FAIL DURING OPERATION 133E 04 H

Pw4500 VOL.2

(+120V DC ESFAS) 4 i

(PD) 4 POWER SUPPLIES FAIL DURING OPERATION 533E 05 H

Pw 0500 VOL2

(+5V OR +25V DC Jj ESFAS)(PS) l REACTOR TRIP FAIL ON DEMAND 1.77E 03 D

Pw-0500 VOL.2 BREAKERS (SB)

SHUNT TRIP COILS FAIL ON DEMAND 1.40E 04 D

PLO4500 VOL2 (SY)

UNDERVOLTAGE FAIL ON DEMAND 2.75E 03 D

Pw.0500 VOL2 l

COILS (UY)

RELAYS (RM)

FAIL ON DEMAND 2.41E-04 D

PW-0500 VOL2 FAIL DURING OPERATION 4.20E 07 H

PLO-0500 VOL2 l

SWITCHES, FAIL ON DEMAND 2.40E 05 D

PLO 0500 VOL 2 PUSHBUTTON TRANSFORMERS FAIL DURING OPERATION 1.565-06 H

Pw 0500 VOL2

(>-4.16KV)(TR)

TRANSFORMERS FAIL DURING OPERATION 6.87E 07 H

PLO4500 VOL.2

(<4.16KV, >-480V)

(TM)

]

TRANSFORMERS, FAIL DURING OPERATION 1.55E-06 H

Pw4500 VOLO INSTRUMENT

(<480V, >-120V)(TI) l AUTOMATIC FAIL TO TRANSFER 2.94E 06 H

PLO-0500 VOL2 TRANSFER UNIT (AT)

RELAY CONTACT FAIL TO OPEN/CLOSE D

IDCOR IPEM 2.41 A (CN)

SPURIOUSLY OPEN/CLOSE H

OCONEE PRA TABLE B-1 TIME DELAY FAIL TO TRANSFER 2 41E-04 D

PLO4500 VOL2 RELAY (RT)

TRANSFER PREMATURELY 4.20E-07 H

PLO-0500 VOL2 RELAY COILS (RY)

FAIL TO D

IDCOR IPEM 2.4-1 A

_DEENERGIZE/ ENERGIZE SPURIOUSLY DEENERGIZE H

IDCOR IPEM 2 4-1 A Page 30

l Revised Database for Living IPE/ Update l

Page 6 OfI8 R&R-PN-008 l

Rev.O DRAFT 6/28/96 TABLE 4.1.1. CPSES INT 1, COMPONENT FAILURE DATA BASE l

COMPONENT /

FAILURE MODE FAILURE UNIT

DATA SOURCE CODE RATE TERMINAL BOARD SHORT/OPEN CIRCUIT 4.64E-06 H

PLO4500 VOL.2 (TB)

ELECTRONIC EQUIPMENT SIGNAL MODIFIERS FAIL DURING OPERATION 2.94E46 H

PID 0500 VOL 2 (MS)

TRIP LOO 1C FAIL ON DEMAND 8.52E 05 D

P!D 0500 VOL.2

(

FAIL DURING OPERATION 2.70E4 H

PLO4500 VOL.2 i

INSTRUMENTATION

SWITCHES, FAIL ON DEMAND 2.69E44 D

PID.0500 VOL.2 PRESSURE (SP)

OPERATE SPURIOUSLY 2.21E4 H

PLO4500 VOL.2 TEMPERATURE NO OUTPUT 3.4tE4 H

PLO-0500 VOL.2 MONITOR LOOPS TRANSMITTERS, FAILDURING OPERATION 6.25E 06 H

PLO-0500 VOL.2 RDW(TF) i TRANSMITTERS, FAILDURING OPERATION 1.57E 05 H

PLO-0500 VOL.2 LEVEL CFL) 2 TRANSMITTERS, FAIL DURING OPERATION 7.60E4 H

PLO4500 VOL.2 PRESSURE (TP) 1 LIMIT SWITCH (SI)

FAIL TO OPERATE 2.69E 04 D

PLO-0500 VOL.2 OPERATE SPURIOUSLY 2.21E4 H

PLO-0500 VOL.2 LEVEL SWITCH (SL)

FAIL TO OPERATE 2.69E 04 D

PLO-0500 VOL.2 OPERATE SPURIOUSLY 2.21E4 H

PLO 0500 VOL.2 MANUAL SWITCH FAIL TO OPERATE 2.40E 05 D

PID-0500 VOL.2 (SM)

OPERATE SPUR 10USLY 2.21E-06 H

PLO-0500 VOL.2 TORQUE SWITCH FAIL TO OPERATE 2.40E-05 D

PI4-0500 VOL.2 (SQ)

OPERATE SPURIOUSLY 221E 06 H

PID-0500 VOL.2 TEMPERATURE FAIL TO OPERATE 2.40E 05 D

PLO-0500 VOL.2 SWITCH (ST)

OPERATE SPURIOUSLY 2.21E-06 H

PLO-0500 VOL2 TEMPERATURE FAIL HIGH/IDW RESPOND 3.41E 06 H

PLO-0500 VOL.2 TRANSMITTER (TT)

SCRAM RODS f

Page 31 1

I

c l

Revised Database for Living IPE/ Update Page 7 Of I8 R&R-PN-008 Rev.O DRAFT 6/28/96 TABLE 4.1.1. CPSES UNrr 1, COMPONENT FAILURE DATA BASE l

COMPONENT /

FAILUREMODE FAILURE UNIT

DATA SOURCE CODE RATE SINGLE SCRAM FAIL ONDEMAND 320E 05 D

PLO-0500 VOL2 ROD F#R)(SC)

I I

I l

l l

l i

l l

i Page 32 i

i

l l

Revised Database for Living IPFJUpdate Page 8 of 18 l

l R&R-PN-008 i

Rev.0 l

DRAFT 6/28/96 4.2 Mamnenance Data Base The mamnenance frequency and duranon data are given in Table 4.2.1. This taNe includes the following canegories :

Component Type.

Techmcal Specifications : allowable time out of service.

Mmmennance Frequency (evens per hour).

Mamnenance Mean Duranon (hours).

]

i l

t I

li U

l I

i Page 33

Revised Database for Living IPE/ Update Page 9 of 18 i

R&R-PN-008 i

Rev.O DRAFT 6/28/96 1

i l

TABIE 4 2.1. OENERIC MADTTENANCE DATA BASE NO.

COMPONENT TECHNICAL MAINTENANCE MAINTENANCE SPECIFICA110NS FREQUENCY DURATION (HOURS) j (EVEN"!WHOUR) i I

CN111ERS NONE 1.41E44 187.0 48 OR 72 HRS 1.41E44 49.0 ll 1

4 2

COMPRESSORS NONE 3.85E44 38.5 3

LARGEFANS NONE 3.20E 06 38.5 4

SMAILFANS NONE 3.25E 06 38.5 5

DIESELOENERATORS 48 OR 72 HRS 4.24E44 35.1 i

s l

6 HEAT EXCHANOERS NONE 4.92E 06 S$3.1 s24 HRS 4.92E 06 6.3 48 OR 72 HRS 4.92E46 13.1 168 OR 336 HRS 4.92E 06 37.2 7

OPERATING SERVICE 72 HRS 5.24E 05 35.1 WA17.R PUMPS j

8 OTHER OPERATINO NONE IJ2E44 266.3 PUMPS 72 HRS 1.32E44 35.1 N

Y I

9 STANDBY MOTOR.

72 HRS 5.26E45 27.6 DRIVEN PUMP 5 2

10 STANDBY TURDINE-72 HRS 1.75E44 37.7

)

DRIVEN PUMPS 11 POSITIVE NONE 1.51E44 317.2

+

DISPLACEMENT PUMPS i

j 12 VALVES NONE 6.89E 06 61.4 l

s 24 HRS 6.89E 06 4.1 72 OR 168 HRS 6.89E46 18.9 13 BATTERIES, BATTERY NONE 1.ME45 40.6 CHAROERS, AND s 24 HR8 1.ME45 6.3 i

DWERTERS.

48 OR 72 HRS 1.ME 05 13.1 168 OR 336 HRS 1.ME 05 40.6 I'

- 14 BUSE2 NONE 2.6586E 06 38.5 s 24 HRE 2.6586E 06 6.3

]

48 OR 72 HRS 2.6586E 06 13.1 168 OR 336 HRS 2.6586E 06 37.2 15 TRANSPORMERS NONE 7.57E46 40 d

i Page 34 1

i i

1

- =.. -

1

I Revised Database for Living IPE/ Update Page 10 of 18 R&R-PN-008 Rev.0 1

DRAFT 6/28/96 TABII 4.2.1. OENERIC MAINTENANCE DATA BASE NO.

COMPONENr TECHNICAL MAINTENANCE MAINTENANCE SPECIFICATIONS FREQUENCY DURA 110N (HOLTS)

(EVENTS / HOUR) 16 STRAINERS NONE 9.2738E45 38.5 s 24 HRS 9.2738E 05 6.3 48 OR 72 HRS 9.2738E 05 13.1 un no su une oesser m ss,

n i

i 1

Page 35

Revised Database for Living IPE/ Update Page 11 of 18 R&R-PN-008 Rev. 0 j

DRAFT 6/24/96 4.4 Imaatmg Events Data Base The PLG Initiating Events Data Base is given in this secnon. The initiating events data for the IPE project is smnmarized in Table 4.4.1. less of service water and loss of campnnent cooling water, and loss of the i

HVAC System events are not included in the initiating events data base. The frequencies of these initiators will be determmed by the fault tree analysis in a separate task.

j i

The following categories are included in the initiating evenu data base.

1 l

Gate Name: The name used in Jr event tree analysis, inmanny Event Category : listed events which vald interrupt normal plant operanons, and which when coupled with stW= failures could lead to a degraded core canesian Frequency : estimated mean value of the innaang event Gwy (Events / Reactor year).

PLG Event #(s). The # of the ImnannF events listed on the PLG data base.

o Page 66

Revised Database for Living IPE/ Update Page 12 of18 R&R-PN-008 Rev. O DRAFT 6/24/96 PLG INITIATING EVENTS DATA BASE NO.

INITIATING EVENT CATEGORY DESCRIPTION FREQUENCY CATEGORY (EVENTS / REACTOR YEAR) i LOSS OF COOLANTINVENTORY 1

EXCESSIVE LOCA 2.66E-07 l

2 LARGE LOCA GREATER THAN 6 INCHES 2.03E-04 j

3 MEDIUM LOCA 4 TO 6 INCHES 4.65E-04 4

SMALL LOCA 2 TO 4 INCHES 5.83E-03 (NONISOLABLE) 5 SMALL LOCA ISOLABLE 2.30E-02 (ISOLABLE) 6 VERY SMALL LOCA LEAKAGE FROM CONTROL 1.26E-02 RODS LEAKAGE IN PRIMARY SYSTEM PRESSURIZER LEAKAGE 7

STEAM GENERATOR STEAM GENERATOR 2.84E 02 1

TUBE RUPTURE LEAKAGE

('

l. :

Page 67

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

1 1

Revised Database for Living IPE/ Update Page 13 of18 i

R&R-PN-008 1

Rev. O DRAFT 6/24/96 1

u PLG INITIATING EVENTS DATA BASE l

NO, INITIATING EVENT CATEGORY DESCRIPTION FREQUENCY CATEGORY (EVENTS / REACTOR YEAR) 8 REACTOR TRIP CRDM PROBLEMS AND/OR 1.35E +00 i

ROD DROP i

HIGH OR LOW PRESSURIZER PRESSURE l

PRESSURE, TEMPERATURE, ICWERIMBALANCE MISCELLANEOUS LEAKAGE i

IN SECONDARY SYSTEM j

PRESSURIZER SPRAY j

FAILURE i

SPURIOUS TRIP - CAUSE l

UNKNOWN ALTTOMATIC TRIP - CAUSE l

UNKNOWN t

{

MANUAL TRIP DUE TO FALSE SIGNALS l

' FIRE WITHIN PLANT l

i i

GENERAL TRANSIENTS 9

TURBINE TRIP TURBINE TRIP, THROTTLE 1.07E +00 VALVE CLOSURE, AND EHC PROBLEMS 10 TOTAL LOSS OF MAIN TOTAL LOSS OF FEEDWATER 1.62E-01 FEEDWATER FLOW (AlL LOOPS) i Page 68

1 ^

1 Revised Database for Living IPE/ Update Page 14 of 18 R&R-PN-008 Rev. 0 4

DRAFT 6/24/96 PLG INITIATING EVENTS DATA BASE NO.

INITIATING EVENT CATEGORY DESCRIPTION FREQUENCY CATEGORY (EVENTS / REACTOR YEAR)

I1 PARTIAL LOSS OF LOSS OR REDUCTION IN I.13E +00 MAIN FEDWATER FEEDWATER FLOW (ONE LOOP)

FEEDWATER FLOW j

INSTABILITY - OPERATOR l

ERROR. ONLY THOSE EVENTS INVOLVING INSUFFICIENT FEEDWATER FLOW WERE USED IN QUANTIFYING THIS

(

INITIATING EVENT CATEGORY.

FEEDWATER FLOW INSTABILITY -

MISCELLANEOUS MECHANICAL CAUSES.

ONLY THOSE EVENTS INVOLVING INSUFFICIENT l

FEEDWATER FLOW WERE l

USED IN QUANTIFYING THIS l

INTTIATING EVENT j

, CATEGORY.

LOSS OF CONDENSATE PUMPS (ONE LOOP). THIS l

TRANSIENT REDUCES FEEDWATER FLOW.

LOSS OF CONDENSATE PUMPS (ALL LOOPS). THIS TRANSIENT CAUSE A LOSS l

OF FEEDWATER FLOW.

Page 69 i

Revised Database for Living IPE/ Update Page 15 0f 18 a

i R&R-PN-008 j

Rev. 0 DRAFT 6/24/96 1

PLG INITIATING EVENTS DATA BASE NO.

INITIATING EVENT CATEGORY DESCRIPTION FREQUENCY j

CATEGORY (EVENTS / REACTOR YEAR) 12 EXCESSIVE INCREASE IN FEEDWATER I.68E 01 l

FEEDWATER FLOW FLOW (ONE LOOP).

INCREASE IN FEEDWATER FLOW (ALL LOOPS) 2 FEEDWATER FLOW l

INSTABILITY - OPERA 1T)R i

ERROR. ONLY THESE EVENTS INVOLVING 1

EXCESSIVE FEEDWATER i

FLOW WERE USED IN l

QUANTIFYING THIS INITIATING EVENT a

CATEGORY FEEDWATER FLOW INSTABILITY -

MISCELLANEOUS l

MECHANICAL CAUSES.

j ONLY THOSE EVENTS INVOLVING EXCESSIVE FEEDWATER FLOW WERE j

USED IN QUANTIFYING THIS INITIATING EVENT CATEGORY 13 LOSS OF CONDENSEf t LOSS OF CONDENSER 1.18E-01 VACUUM VACUUM

[

CONDENSER LEAKAGE i

j LOSS OF CIRCULATING WATER 14 CLOSURE OF ONE FULL OR PARTIAL CLOSURE 8.66E-02 MSIV OF MSIV(ONE LOOP)

i 15 INADVERTENT CLOSURE OF ALL MSIVs I.93E-02 i

CLOSURE OF ALL MSIVs a

4 i

1 Page 70 l

4 3

a

Revised Database for Living IPF1 Update Page 16 of18 3

R&R-PN-008 i

l Rev. O DRAFT 6/24/96 PLG INITIATING EVENTS DATA BASE NO.

INITIATING EVENT CATEGORY DESCRIPTION FREQUENCY CATEGORY (EVENTS / REACTOR YEAR) 16 CORE POWER UNCONTROLLED ROD 2.68E 02 4

EXCURSION WITHDRAWAL

)

cvCS MALFUNCTION i

l STARTUP OF INACITVE COOLANr PUMP j

17 LOSS OF PRIMARY LOSS OF RCS FLOW (ONE 1.76E-01 FLOW LOOP) 4 TOTAL LOSS OF RCS FLOW 2

18 STEAM LINE BREAK 4.65E-04 INSIDE CONTAINMENT 19 STEAM LINE BREAK 6.04E-03 OUTSIDE CONTAINMENT 5

20 INADVERTENT SUDDEN OPENING OF STEAM 4.19E 43 OPENING OF MAIN RELIEF VALVES STEAM RELIEF VALVES 21 INADVERTENT INADVERTENT SAFETY 2.99E-02 SAFETYINJECTION INJECTION SIGNAL SIGNAL COMMOM CAUSE INITIATING EVENTS SUPPORT SYSTEM FAULTS 22 LOSS OF OFFSITE 1.40E-Ol*

POWER 23 LOSS OF ESSENTIAL 3.35E-02 DC BUS 24 LOSS OF INSTRUMENT 2.02E-03 AIR 1

Page 71 l

Revised Database for Living IPE/ Update Page 17 of I8 R&R-PN-008 1

Rev. 0 l

DRAFT 6/24/96 PLG INITIATING EVENTS DATA BASE NO.

INITIATING EVENT CATEGORY DESCRIPTION FREQUENCY CATEGORY (EVENTS / REACTOR j

YEAR) 25 LOSS OF VITAL AC LOSS OF POWER TO 8.36E-02 l

POWER NECESSARY PLANT SYSTEMS. ONLY THOSE EVENTS INVOLVING LOSS OF VITAL AC POWER WERE USED IN QUANTIFYING THIS

'l INITIATING EVENT CATEGORY i

26 LOSS OF ICWER TO LOSS OF POWER TO 8.23E-02 i

NECESSARY PLAN NECESSARY PLANT SYSTEMS SYSTEMS (NOTE: THOSE EVENTS INVOLVING LOSS OF VITAL AC POWER WERE TREATED SEPARATELY) i

UNIT IS EVENTS PER SITE YEAR i

i i

i 1

(

i Page 72 4

J

i j

Revised Database for Living IPE/ Update Page 18 of18 j

i R&R-PN-008

]

Rev. O i

DRAFT 6/24/96 TABLE 4.4.1 CPSES UNIT 1, GENERIC INITIATING EVENTS DATA BASE l

Gau Inmanag Event Caegory Inmamir Event

)

Name Frequency (yr')

%A LARGE BREAK LOCA INIT. EVENT FREQ 2.02E 04 j

%CV LOSS OF CONDENSER VACUUM INIT. EVENT FREQ 8.63E-02

{

%M MEDIUM BREAK LOCA INIT. EVENT FREQUENCY 4.63E-04

-l i

l

%R STEAM GENERATOR TUBE RUPTURE INIT. EVENT FREQ l.61E-02 j

%S SMALL BREAK LOCA INIT. EVENT FREQ 7.45E-03

%T1 REACTOR TRIP INITIATING EVENT FREQUENCY 3.00E-00

%T3 INADVERTENT SIAS INITIATING EVENT FREQUENCY 1.79E-02 2

%T4 MAIN STEAM LINE BREAK INITIATING EVENT 9.54E-03 i

FREQUENCY j

%T6 LOSS OF MFW - NO MFW AVAIL. INIT. EVENT 7.67E-01 i

FREQUENCY

]

[

%VS VERY SMALL LOCA INITIATING EVENT FREQUENCY 9.88E-03 f

%XI LOSS OF A DC BUS INIT. EVENT FREQ 2.50E-02

__%X2 LOSS OF THE HVAC SYSTEM INIT. EVENT FREQ Sysean SpectSc i

%X3 IDSS OF OFFSITE POWER INIT. EVENT FREQ 5.83E-02 1

%X4 LOSS OF A NON-VITAL AC BUS INIT. EVENT FREQ l.00E-01

%X5 LOSS OF A SAFEGUARDS BUS INTT. EVENT FREQ.

4.74E-02 e

%X6 LOSS OF COMPONENT COOLING WATER SYSTEM INIT.

Sysum SpeciSc

}

EVENT. FREQ

%X7 LOSS OF STATION SERVICE WATER INIT. EVENT FREQ.

Syssem Speci6c l

%X8 LOSS OF INSTRUMENT AIR INIT. EVENT FREQ l.90E-03 l

l

Units are evens per site-year i

The #5 minatmg event (small LOCA, isolable) is not included in the CPSES initiating events.

l t

1 1

Page 73 i

4 i

4

.. _..

ML20116C962

Text

GENERAL COMMENT

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response

Response