ML20105B157

Raises Three Issues Re NRC Review of AP600 PRA Covering Frequency of SG Tube Rupture Accidents,Frequency of Reactor Vessel Rupture Accidents & Frequency of Accidents Arising from Earthquakes (Seismic Risk)

ML20105B157
Person / Time
Site:	05200003
Issue date:	09/08/1992
From:	Sholly S MHB TECHNICAL ASSOCIATES
To:	Hasselberg F Office of Nuclear Reactor Regulation
References
NUDOCS 9209180203
Download: ML20105B157 (5)

Text

_ _ _ _ _ _

l Tec,hnical Associates S T - v e.S d oJ G76 s Consultants on Energy & the Ens honment s

8 September 1992 Mr. Frederick W. Hasselberg lead Project Manager, Westmghouse AP600 Standardization Proj,ect Directorate Office of Nuclear Reactor Regulation Mail Stop 11 H3 U.S. Nuclear Regulatory Commission 11555 Rodville Pike 6 Rockville, MD 20814 RE: AP600 Probabilicaic Risk Assessment Review Issues -- Frequency of Et. cam Generator Tube Ruoture and Reactor Vessel P.aglwre initiating Esents. and Seismic Core Dam _agt Freouency Dear Mr. Hasselberg.

Wcstinghcase Electric Corporation recent:y (25 June 1992) submitted an application for Final Design Approval and Design Certification the AP600 PWR standard plant design. Two key parts of the AP600 application are the Standard Safety Analysis Report (SSAR) and the Probabilistic Risk Assessment (PRA).

An important selling point (ooth commercially and in terms of public opinion) and design goal

~

for the AP600 is a frequency of one in a million per year or less for exceedence of a 25 Rem effective d e equivalent at the site boundary within 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> without any emergencv protective etion. The purpose of this letter is to raise three issues in this regard for the NRC's review of the .AP600 Prebabilistic Risk Assessment -- the frequency of steam generator tube rupture (SGTR) accidents, the frec uency or reactor vessel rupture accidents, n id the frequency of accidents arising from earthquakes (seismic risk).

.SGTR Initiating Esent Freauency The AP600 design is whLt I would character as iemi-passive"in nature. Some safety functions still require successful performance of active components in order to assure safety, while other com onents are more passive in nature. It would be reasonable, in my estimation, t 3 expect that A 600 could demonstrate superior performance for most accident scenario types when compared with conventional Westinghouse PWRs, which rely almost entirely on safety systems with active components.

Westinghouse claims that N 6AP600 PRA study supports an analyzed fcequency of a 25-gem site boundary dose of 3 x lif per reactor-year for all events,in comparison to the 1 x 10' per reactor-year goal.1/ In order for Westinghouse to sustain this claim, Westinghouce m able to demonstrate that the frequency of SGTR core melt eccidents is less per than 3 x 10 gst be reactor-year since it is apparent that the site boundary do from an SGTR core melt accident

)

1/ Westinghouse, APGM Probabilistic Rhk Aucssment, Section 17.3, " Release Requency*, Rev. O, 6/26/92, page i7 L 170051 Ao I 1723 Hamilton Avenue-Suite K, San Jose, CA 95125 Pnone (408) 266-2716 Fa (408) 266-7149 9209180203 92090 s PR p,600.]f. 057

. _ - - - - Y

J I

hfr. Frederick W. Hasselberg

8 September 1992 Page2 l

4 1

would easily exceed 25 Rem with or without protectjve action. In fact, Westinghouse estimates an SGTR core melt frequency of 2.6 x 10' per reactor-year. 2_/

)

Existing peer-reviewed PRA studies of conventional Wes.inghouse PWRs suggestJhat a typic.al core melt )robabilite for SGTR accidents is typically approximately 2 x 10 per-reactogyear. 3/ e SGTil initiating event frequency in these PRA studies is approximately 1 x 10' to 3 x 10' per reactor-year. 4/ In order to demonstrate com )liance with both the l AP600 design goal and the Westinghouse AP600 PRA result, Westin cuse will have to be able to demonstrate either an improved system response (a combina ion of hardware and human response) and/or a reduced initiating event frequency for SGTR sequences.

Westinghouse implicitly calculates a factor of 1,000 improvement (i.e., the difference between the AP600 result and the NUREG-1150 results) overa,1 for SGTR accidents, i

Since most of the details of the Westinghouse accident sec.uence analysis and quantification

> for th- AD600 PRA a:e considered to be proprietary (I wiLi be coistnunicating with you soon concerning the lack oivalidity of the Westinghouse proprietary claim), I will confine me remarks at this juncture to the initiatin event frequency, for which details were submitted on

a non-proprietary basis. For the AP , Westinghouse estimates a single-tube SGTR.

initiating event frequency of 5.3 x 10 per reactor-year. 5/ The AP600 value is roughly a factor of two improvement over the NUREG-1150 value, i Westinghouse derived its SGTR initiating event value in Attachment 1 ;o Appendix A of the AP600 PRA. The factor of two reduction is achieved by a three-step process: (a) assessing the existing histo 5 cal experience and eliminating a number of the historical tube ruptures from the data base, (b) calculating the frequency of tube rupture based on an individual tube l

basis, and (c) multiplymg this frequency by the total number of SG ' tubes in the AP600 steam

, generators.

This process, while not irrational, raises four questions. h_g, the SGTR initiating event 4 frequencies for previous Westinghouse plant PRAs were calculated based on the number of actual tube ruptures divided by the number of reactor-years of operation. Thifis a different

calculation than performed for the AP600 PRA, and as a result the AP600 SGTR results

cannot be directly compared with the SGTR results of PRA studies of existing Westinghome plants. Secc.nd, eliminating historical tube ruptures from the calculation is non-conscrvative, and this non-conservatism cannot be sustained since there is not yet any AP600-specifie operating experience with which to support the arguments which Westinghouse uses to reduce

f Westinghouse, AP600 Probabilistic Risk Assessment, Table 8-1, " Initiating Events Contributing to Core Damage (Base Case - At Power)", Rev. O,6/26/92, l page 8-7.

4 2/ {or example, the NUREG-1150 estimates were 1.9 x 10 pei reactor-year for Surry and 2.0 x 10' per reactor-year for Sequoyah. Sec, NUP.EG/CR-4550, Vol. 3, Rev.1, Part 1, Table 4.10-4; and NUREG/CR-4550, Vol. 5, Rev.1, Part 1, Table 5-3. ,

4_/

The NUREG 1150 studies for PWRs used an SGTR initiating event frequencv of 1 x 10-2 per reactor-year. Scc, for e. ample, NUREG/CR .050, Vol. 3, Rev.1, Part 1, page 4.9-4.

ff Westini; house, APm0 Probabjlistic Risk Assessmenl, Section B.2.5.3, " Steam Generator Tube

, Ruptur#, Rev 0,6/26/92, page B-3. It should be noted that the actual calculation, carried oft in Attachtr;cnt I to Appendix B of the AP600 PRA,is 5.2 x 10'3 per reactor-year, not 5.3 x 10' . per

reactor-) car as cited elsewhere in the AP600 PRA. (Compare page B-3 with page B-20 of the AP600 PRA, for example.)

a k

c

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

i Nfr. Frederick W. Hasselberg 8 September 1992 l Page 3

! the number of historical tube rupture events. Although Westinghouse in some cases presents plausible arguments for eliminating historical tube ruptures, actually eliminating them from the data base presupposes that operating experience to date with conventional Westinghouse PWRs has identified all of the prmcipal contributors to tube ruptures. No evidence has been presented by Westinghouse to suggest that this is the case, and I am aware of none which i supports such a hypothesis. Third, the Westinghouse calculation ignores the 10% tube plugging limit for the AP600 design. Thit,latter point is important since each plugged tube i- slightly raises the SGTR initiating event frec.uency since it is calculated on a per-tube basis.

Fourth, Westinghouse assesses a factor of 0.3 to account for plant outages. This is excessive l for existing Westinghouse plants; since this is effectively a capacity factor correction, a value of 0.65 is more realistic.

To correct for these factors, I have recalculated as follows:

f Number of Tube-Years i

10,500,000 tube-years times 0.65 capacity factor times 0.9 tube

! plugg,mg factor; yields 6,142,500 tub,:-years (compared with Westmghoue value of 7,560,000 tube-years).

l Num.ber of Tube Ruotures i Westinghouse counts 3.1 t'Ac y ures; I count 8 tube ruptures in Westinghouse plants (i.e., nom eliminated from histarical

experience).

f &mber of Runtures Per Yube-Year

! Westinghouse calpi.e 4.1 x 10-7 ruptures / tube-year. I

calculate 1.3 x 10~ _ uptures/ tube-year.

i AP600 SGTR Initiating Event Frequency Westinghouse calculates bas d on 12,614 tubes, and estimates an 4 SGTR trequency of 5.2 x 10 per reactor-year. I calculate based 2

on a 10% tube plugging factor, or a total of 11,353 tubes.

i Accoroingly,I calcgilated an SGTR initiating event frequency for AP600 of 1.5 x 10' per reactor-year, or a factor of almost three greater than the Westinghouse value. s/ 2/.

~

fi/ For comparison purposes, the three Surry steam generators have a total of 10,42 tubes.

Assuming a 10% tube plugging allowance, .this smounts to 9,036 tubes. The Surry SGTR frequency, using the Westinghousc method, would be 3.7 x 10' per, reactor-year; using my j variation as set forth above, the Surry SGTR frequency w ild be 1.2 x 10" per reactor-year.

If I haw carried out a separate quantification based on accounting for historical operating experience of atl PWRs and PFlWks (CANDU units), and account:ng for all twelve histgrical l

tube ruptures. Based on this data base, I estimate a tube rupture frequency of 5.4 x 10' per reactor. year (t1 tube ruptures in more than 2,200 reactor-years of experience through February

i. 1991). This is a broader experience base than Westinghouse has used.

4-l 4

d

- + _ - . _ , -- a

-s- 4 ,, ,n, w w --

m- ,-,. --~-, - n,w -~en, , ,,

~

'Mr. Frederick W. Hasselberg 8 September 1992 Page 4 >

I believe that this calculation is more realistic, and I recommend it to the NRC staff for use in its AP600 PRA review. gwould noteghat this would increase the AP600 SGTR core damage frequency from 2.6 x 10'7 to 7.5 x 10' frequency from 33 x 10' to 3.35 xper 10'per reactor-year. reactor-year, and overall It also increases AP600 core the SGTR damage -

contribution from 0.8% to 2.2%

i Reactor Vessel Ruoture Frequency Regarding reactor ve sel rupture, Westinghouse cites the WASH-1400 value of 3 x 10'7 per reactor-year. Westinghouse has apparently forgotten (as it is common to do) that the WASH-1400 value is a mrdian, not a mean. The mean value, as pointed out in its review of the Oconee PRA, isper 1.1reactor-year.

x 10'prookhaven National Even taking Laboratory Westinghouse's factor of ten reduction at face value (it cannot be said that Westinghouse went to very great efforts to justify this factor of ten), this result)in e reactor vessel rupture contribution to AP600 core damage frequency of 1.1 x 10' pey reactor-year. This increases the AP600 core damage frequency from 3.3 x 10'7 to 4.1 x 10' per reactor-year, and increases the contribution of vessel rupture from 9.1% to 26.8% (It should be noted that this sumes, as does Westinghs r e, that vessel rupture automatically leads to core melt. ,This ruay be  ;

conservative for the AP600 c figuration -- the ir,-containment RWST could flood the reactor cavity and cover the core. In view of the contribution of vessel rupture to core damage frequency for AP600, this may be worth investigating.)

These two changes

• increase the AP600 core damage frequency from 3.3 x 10-7 to 4.2 x 10~7 aer reactor gear. The chgnges also increase the 23-Rem site boundary dose frequency slightly from 3 x 10' to 3.5 x 10' per ceactor-year (assuming that SGTR sequences contribute on a 1 for 1 basis and that vessel rupture does not contribute).

Seismic Risk I note that Westinghouse has elected to perform a seismic margin evaluation m lieu of a seismic PRA for AP600. While such an approach is permitted m response to_NRC Generic Letter 88-2P (the IPE Generic Letter) for an o crating plant, it is nol permitted under NRC regulations which require a PRA for standard lant designs for which certification is sought under Part 52;ge 10 CFR 52.47(a)(5). H/ W tinghouse is required to perform a full seismic PRA for AP600. Given the existing AP600 internal events PRA and the AP600 seismic margin analysis, extendi,ng these analyses to a seismic PRA should not be time-consuming, difficult, or very expensive. The NRC staff should communicate this requirement to -

Westinghouse promptly.

Conclusion Now that at least some of the details of the AP600 PRA have become available,it is clear that there are problems with portions of the analysis which suFgest that the results are optimistic. I do expect, however, that unless a serious design error has been made regarding seismic or fire events, the AP600 design . hould demonstrate an improvement on core damage frequency from existing PWR desigm 4

8/ i note that the NRC staff required General Electric to include a seismic PRA for the GESSAR-Il

' FDA.

i

i

. , , Str. Frederick W. Ilasselberg 8 September 1992 I

Page 5 in order to promote additional public interaction on the AP600 application, I believe Aat the NRC staff should undertake an early review of the AP600 PRA with an eve toward publicly i

releasing as much of the documem (and the related fault and event treesj as is possible while still protecting Westinghouse's legitimate proprietary interests. The sooner the phii:is able -

to review and understand the risks posed by operation of plants employing the AP600 i standard design, the sooner the safety issues mvolved in the design can be identified, publicly 4 aired, and resolved. This will permit' timely issuance (if justified) of a Final Design Approval for AP600 and publication of a proposed Design Certification rule. If, however, the NRC staff waits untillater in the process before carrying out its review of the validity of 3roprietaiy i claims, it is predictable that delays will occur. Such delays are entirelv avoida ale at this stage  !

l of the review; the same will not be true at a later date.

I would be pleased to discuss these matters with the NRC staff if there are any questions. I wish the staff the best ofluck on its AP6r0 PRA and SSAR rt views, and expect to communicate further with the staff as the reviews progress.

Sincerely, i

1/m -

d Steven C. Sholly Senior Consultan,t l cc: Mr. Thomas Kenyon, NRR/PDST i-f I

I 4

i i

4

_.-,a - - - - - -,, , me -- --p-- .

.r- - w,-e< a.y,-