ML20137F327

Requests Addl Info on AP600 Draft Rept Concerning thermal- Hydraulic Uncertainty in Probablistic Risk Assessment Success Criteria Analyses

ML20137F327
Person / Time
Site:	05200003
Issue date:	03/28/1997
From:	Huffman W NRC (Affiliation Not Assigned)
To:	Liparulo N WESTINGHOUSE ELECTRIC COMPANY, DIV OF CBS CORP.
References
NUDOCS 9704010066
Download: ML20137F327 (7)

Text

. - . ._. ._

. . March 28, 1997 Mr. Nich31as J. Liparulo, Manager Nuclear Safety and R:gulatcry Analysis Nuclear and Advanced Technology Division Westinghouse Electric Corporation P.O. Box 355 Pittsburgh, PA 15230

SUBJECT:

REQUESTS FOR ADDITIONAL INFORMATION (RAls) ON AP600 DRAFT REPORT CONCERNING THERMAL-HYDRAULIC UNCERTAINTY IN PROBABLISTIC RISK ASSESSMENT (PRA) SUCCESS CRITERIA ANALYSES

Dear Mr. Liparulo:

4 In support of the evaluation of the AP600 thermal-hydraulic (T/H) uncertainty associated with PRA success criteria and the passive system reliability concerns, Westinghouse letter NSD-NRC-97-4928 dated January 2, 1997, submitted a draft report on T/H uncertainty issue resolution and requested staff feedback. Based on review by the Nuclear Regulatory Commission staff and its contractor of the Westinghouse draft report, the staff has concluded it needs additional information to complete an assessment of the approach Westinghouse is using to bound the AP600 T/H uncertainties. Enclosed with this letter are the staff's RAls on the draft T/H uncertainty repart.

The staff notes that the draft report still does not provide a plan approach or schedule for addressing T/H uncertainties associated with long term cooling. This was scheduled to be resolved by Westinghouse by January 17, 1997. This area has potential to impact the staff's design certification review schedule.

If you have any questions regarding this matter, you can contact me at (301) 415-1141.

l Sincerely, original signed by:

William C. Huffman, Project Manager Standardization Project Directorate Division of Reactor Program Management Office of Nuclear Reactor Regulation Docket No.52-003

Enclosure:

As stated cc w/ enclosure:

See next page DISTRIBUTION:

Docket File PDST R/F TTMartin i PUBLIC MLSlosson TRQuay TKenyon DJackson BHuffman JMoore, 0-15 B18 JSebrosky WDean, 0-17 G21 p' h OJ I ACRS (11) JLyons, 0-8 E23 ACubbage. 0-8 E23 Alevin, 0-8 E23 NSaltos, 0-10 E4 GHolahan, 0-8 E2 JFlack, 0-10 E4 MRubin, 0-10 E4 DOCUMENT NAME: A:TH-UNCRT.RAI T3,eceive e copy of tNo docunient, Indicate in the bes: 'C' = Copy without attachrnent/encMe, *E' = Copy with attactwnent/ enclosure 'N' = No copy 0FFICE PM:PDST:DRPM SRXB:pSSN! SPSB:[6SAV)) D:PDST:DRPM l NAME WCHuffman: MFW Alevih WN JF1acktWF TRQuay 1)N DATE 3 C2'j97 3hi/11/ \ 's h7/971 3 / 3 /97 9704010066 970328 '^"'^ll RECORDjCOPY PDR ADOCK 0520 3

. . i l

l Mr. Nicholas J. Liparulo Docket No.52-003 Westinghouse Electric Corporation AP600 cc: Mr. B. A. McIntyre Mr. Ronald Simard, Director Advanced Plant Safety & Licensing Advanced Reactor Programs Westinghouse Electric Corporation Nuclear Energy Institute Energy Systems Business Unit 1776 Eye Street, N.W.

P.O. Box 355 Suite 300 Pittsburgh, PA 15230 Washington, DC 20006-3706 Ms. Cindy L. Haag Ms. Lynn Connor Advanced Plant Safety & Licensing Doc-Search Associates Westinghouse Electric Corporation Post Office Box 34 Energy Systems Business Unit Cabin John, MD 20818 Box 355 Pittsburgh, PA 15230 Mr. James E. Quinn, Projects Manager LMR and SBWR Programs Mr. M. D. Beaumont GE Nuclear Energy Nuclear and Advanced Technology Division 175 Curtner Avenue, M/C 165 l Westinghouse Electric Corporation San Jose, CA 95125 One Montrose Metro 11921 Rockville Pike Mr. Robert H. Buchholz Suite 350 GE Nuclear Energy Rockville, MD 20852 175 Curtner Avenue, MC-781 San Jose, CA 95125 Mr. Sterling Franks U.S. Department of Energy Barton Z. Cowan, Esq.

NE-50 Eckert Seamans Cherin & Mellott 19901 Germantown Road 600 Grant Street 42nd Floor Germantown, MD 20874 Pittsburgh, PA 15219 Mr. S. M. Modro Mr. Ed Rodwell, Manager Nuclear Systems Analysis Technologies PWR Design Certification Lockheed Idaho Technologies Company Electric Power Research Institute Post Office Box 1625 3412 Hillview Avenue Idaho Falls, ID 83415 Palo Alto, CA s*su3 Mr. Frank A. Ross Mr. Charles Thompson, Nuclear Engineer U.S. Department of Energy, NE-42 AP600 Certification Office of LWR Safety and Technology NE-50 19901 Germantown Road 19901 Germantown Road Germantown, MD 20874 Germantown, MD 20874

!~

! AP600 REQUEST FOR ADDITIONAL INFORMATION 1

! 492.33 The draft report is referred to as the " resolution of T/H uncertainty l issues for AP600 passive system reliability " The staff believes this j mischaracterizes the purpose of the report. It is the staff's understan-3 ding that the purpose of the report is to evaluate and bound (primarily i through a margins assessment) the T/H uncertainties associated with the focused and baseline PRA success paths. Westinghouse should revise the  !

report accordingly.

i

, 492.34 Westinghouse indicates that the results and insights from the implementa- l t tion of the approach to the focused PRA will be used to show how T/H l uncertainties impact the results of the baseline PRA. Westinghouse '

states that this can be done because, due to the presence of the active

systems, the frequency of a typical " success" accident sequence (except i for large LOCA sequences which are identical in both the focused and the baseline PRAs) is at least an order of magnitude smaller in the baseline PRA than the corresponding sequence in the focused PRA. According to

. Westinghouse, this compensates for the lower values of CDF and LRF in the baseline PRA. Therefore, if it is shown that T/H uncertainties do not

impact certain sequences in the focused PRA, they do not impact the

corresponding baseline PRA sequences either. The only exception to this, 1 according to Westinghouse, are large LOCA sequences which do not include l active systems. Westinghouse concludes that T/H uncertainties associated j with LLOCA sequences can have a significant impact on the baseline PRA results and this will be assessed.

, The staff believes that this may require more effort than what Westing-i house proposes to do. Westinghouse should ensure that this is done systematically to address issues such as the following:

• in addition to large LOCA sequences, safety injection line break

sequences are identical in both the focused and the baseline PRAs.

there are " success" sequences in other event trees, such as medium LOCA and CMT line break, which do not include failure of active

systems.

l

• Westinghouse assumes an one-to-one correspondence between the focused PRA and the baseline PRA event trees'(this may not be true

, because of assumptions and simplifications made in the focused

PRA, e.g., about AC power availability). This assumption could l 1 eave out important sequences of the baseline PRA.

• Initiating event frequencies and failure probabilities (of both l active and passive features) are used to screen out sequences as not risk significant. Westinghouse's approach should provide for j the assessment of- the impact of any future changes in the values i of these quantities (e.g., due to closing of PRA open items) on d

the screened out sequences.

Enclosure J

i

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

I 4

j i j 492.35 The "0K" sequences are described as sufficiently similar to DBAs that

they can be considered as having no core uncovery. Is this explicitly i

demonstrated by analysis, or is it based only on the assumption that qualitative similarity to a DBA sequence is sufficient to conclude that

there is no core uncovery?

! 492.36 Have all the sequences in the ten OK categories been analyzed as no core uncovery with MAAP4 or does Westinghouse plan to only analyze a represen-j tative sequence from each category? If only a representative sequence is to be analyzed, how does Westinghouse intend to demonstrate the chosen l sequence is representative?

492.37 Westinghouse should clarify in the report how MAAP4 is used to support the results of the T/H uncertainty evaluation. In addition, a descrip-tion of the MAAP4 benchmarking process and the related conclusions from j this effort should be included.

) 492.38 The staff has the following questions regarding the general statements Westinghouse makes about the method of categorization in Section 5.0 of the report:

a. On p. 21, at the bottom, item I states that "...some success paths... fit the definition of multiple categories." If the development of categories was done to allow selection of a "repre-sentative" sequence (or sequences) for analysis, the above state-ment would appear to. imply that the selection of categories, for some sequences, was arbitrary,

b. Item 2 (which continues onto p. 22) raises the question of "most probable" vs. "most limiting". While it may be true that the sequence with 3/4 ADS-4 may be more probable, the sequence with 2/4 may be more limiting in terms of T/H consequences. Westing-house should demonstrate the assertion that the fraction that is 2/4 ADS-4 is " negligible."

c. Item 3 is unclear about the way in which ADS-1 operation is accounted. How is the operation of stage 1 " estimated based on information on stages 2 and 3?"

492.39 The lack of containment isolation appears to be a significant contributor to the uncertainty of success of many T/H sequences. What assumptions are made with regard to containment isolation? What equipment is assumed to be functional? Is containment pressure explicitly evaluated? What failures must occur for failure of containment isolation? Are these failures accounted for in the PRA event probabilities?

492.40 With reference, in part, to Table 5-1, how is PCCS operation accounted for? (See also discussion on p. 50, Categories UC-6, 7). Containment isolation is now considered as part of some sequences. Thus, containment

l 3 pressure may have an impact on time to recirculation, and consequently on decay heat. Also, why are IRWST lines not included in Table 5-17 Availability of two (vs. one) line may also affect the time to transition to recirculation.

492.41 In Section 5.3, Westinghouse states, "The most risk-significant (long-term cooling) scenarios are anticipated to be the ones with up to 1 single failure." This needs to be explained in more detail. Certainly l the scenarios with the highest frequencies have "up to 1 single failure," '

but it is not completely clear how these also constitute the most risk-significant sequences, if they are already subsumed by DBA analyses

(which do not show core uncovery or core damage).

492.42 The statement at the bottom of p. 64 that " risk significance for the T/H uncertainty process is defined as increasing the Focused PRA CDF or LRF  !

by at least 1 percent if the success category were counted as core damage" means that the definition of significance depends on how the "UC" categories are chosen. A more objective criterion that is based on sequence contribution to CDF (rather than an arbitrary grouping of

, sequences) would be preferable. Also, the staff raised the issue of why  ;

the 1 percent criterion was chosen several months ago, and the question really has never bee:n answered by Westinghouse. Again, this is a dependent criterion (on the categorization of sequences and the number of sequences) and could change if the categories were changed. This l criterion needs to be justified. '

492.43 In Section 8.3, Westinghouse addresses the selection of representative cases to evaluate low-margin, risk-significant scenarios. The sequences are chosen by examining the categories; this assumes that the sequences in each category exhibit similar behavior, an assumption that needs to be demonstrated. It appears that Westinghouse proposes to select the sequence having the highest frequency in each category as the representa-tive sequence of that category. Such a sequence may be the least limiting in terms of peak clad temperature (PCT) since it has more systems available. The staff believes that, in a " margins" approach, the representative sequence of a category should be the one which is the most limiting in terms of PCT but with frequency above an acceptable cutoff (based on risk significance considerations). Westinghouse should provide additional information on how the representative sequence is chosen from each category.

492.44 Westinghouse needs to demonstrate the assertion that more than 3 system failures decreases a sequence frequency to the point where it need not be further expanded.

492.45 What is the total frequency of all the success sequences after the total frequency of the "0K" sequences is subtracted? This and the total frequency of the sequences considered " risk significant" can provide a measure of the completeness of the approach.

f.

l -4_

l 492.46 In Table 8-2, for example, there are many sequences " embedded" in the event tree that is summarized in each line. For instance, for LLOCA08, the ADS 2-3 block has at least 4 different branches, representing different numbers of ADS-2 and -3 valves. Does the frequency represent j .. the sum of the multiple sequences? i 492.47 The report does not contain sufficient information to check sequence frequencies. Please include a table with the probabilities of each of

the events shown in the event trees. Also, please explain the reason j that in some sequences the probability of having a component failed is higher than the probability of having it operating (e.g., ADS stage 2&3 l valves in sequences #7 and #8 of event tree in p. 15).

t 492.48 Referring to Table 4.2:

l

a. Although break size and location were not considered in expanding the event trees, these factors may have an impact on T/H uncer-i tainty assessment. Westinghouse should provide additional discus-sion on how T/H uncertainty associated with these factors will be considered within each' sequence categorization.

l l b. The third line notes that the small size of the ADS-1 lines means j that they have an insignificant impact on the accident. Are these 4

lines considered as failed in the analyses? Do the sequence fre-quencies include failure of the lines?

l

! 492.49 Figure 7-1 needs better labeling and/or explanation to make clear what is j being shown.

i 492.50 Westinghouse is using a risk-based approach to show how T/H uncertainties  !

l impact both the focused and baseline PRA results. This approach is using l 4 initiating event frequencies and failure probabilities to screen out '

! sequences as not risk significant. This implies that the conclusions of i

the T/H uncertainties study may be sensitive to uncertainties associated with initiating event frequencies and failure probabilities used in the PRA. Westinghouse should address these concerns in Chapters 11 and 12 of the report (not completed yet) where results, insights and conclusions will be summarized.

492.51 The discussion at the top of p. 51, for Category UC7, is confusing. The lack of " refinement" of the event tree is given as a reason for the category frequency being high, but in the next sentence the frequency is claimed to be "non-risk-significant." It is not clear how a "high" frequency can be "non-risk-significant." Please clarify the use of "high" in the context of the discussion.

492.52 Sequences #34, #35 and #36 of event tree IEV-SGTRW in p.18 (SGTR with PRHR available) are shown as OK (category OK9). These sequences involve failure of both CMTs which requires manual ADS actuation. On what basis b-

were these sequences categorized as OK7 Are these DBA or "similar" sequences? Were these sequences analyzed with T/H uncertainties? Please explain with appropriate documentation 492.53 The definition of " low margin" as given on p. 21 appears to be based (implicitly) on as-yet uncompleted analyses. Westinghouse has yet to provide only T/H analyses to the staff to allow us to judge if the assumptions made in assessing the outcomes of these sequences are, in fact, bounding. In the staff's view, this is a critical issue.

l