LD-96-042, Submits Misleading Statement by BNL on C-E Reactor Coolant Pump Seals.Statement Correction Listed

Submits Misleading Statement by BNL on C-E Reactor Coolant Pump Seals.Statement Correction Listed

ML20129E587
Person / Time
Site:	05200002
Issue date:	09/26/1996
From:	Brinkman C ABB COMBUSTION ENGINEERING NUCLEAR FUEL (FORMERLY, ASEA BROWN BOVERI, INC.
To:	Shaukat S NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM)
References
LD-96-042, LD-96-42, NUDOCS 9610010118
Download: ML20129E587 (2)

Text

.- - - __

A45

~

September 26,1996 LD-96-042 Docket No.52-002 Document Control Desk U.S. Nuclear Regulatory Commission Washington, DC 20555-0001 Attention: S. K. Shaukat

Subject:

Misleading Statement by BNL on CE's Reactor Coolant Pump Seals

References:

1) Brookhaven National Laboratory letter to NRC, dated October 5,1995

2) "A Combustion Engineering Review ofNUREG-1032; Evaluation of Station Blackout Accidents at Nuclear Power Plants," ABB-CE Report NPSD-340, performed for the CE Owners Group, March 1986

Dear Mr. Shaukat:

The Reference I letter was recently placed in the Public Document Room and our review shows that it includes an erroneous statement related to the reliability of our reactor coolant pump seals.

Item six on page two of that letter (attached) identifies "CE's use of nearly zero seal failure probabilities" as an example of a "quite non-conservative" seal model used in industry IPEs.

ABB-CE uses a low seal failure probability because the reactor coolant pump seal on ABB-CE designed NSSSs has unique design features which have resulted in demonstrated high reliability through both normal operation and tests.

A high reliability (and consequent low-failure probability) does not equate to being non-conservative. Rather, the three-stage mechanical seal design, the last one of which is capable of retaining full reactor coolant system operation pressure, leads to a failure probability that is both low and conservative. This is true for all ABB-CE NSSS designs (see Reference 2), but especially so for the System 80+m Standard Plant Design for which there are two independent seal cooling systems (Component Ccoling Water and Seal Injection) with a dedicated backup system for seal injection that may be used during station blackout events.

30Mi0 ppF ABB Combustion Engineering Nuclear Systems I Combustion Engnemng inc 2000 Day Hdi Road Telephone (MO) (A81911 PO Box 500 Fax (E) ?ttS5203 mnosa. ci ocmscsoo 0 960926 2 h00 A

K 0520

4

, September 26,1996 LD-96-042  ;

> Page 2 l j

ABB-CE requests that BNL's statement not be repeated or referenced in future documents. If

you wish to discuss this further, please call me or Stanley Ritterbusch at (860) 285-5206.

Sincerely, l.

COMBUSTION ENGINEERING, INC.

i .

}

C. B. Brinkman Director

. Nuclear Licensing i

ser/lw cc: C. Serpan (NRC/RES) i 1 N. Wilson (NRC/NRR) j C. .Ruger (BNL) 4 9

h 4

I i

.i

I Attachment 4 BROOKHAVEN NATIONAL LABORATORY j E ASSOCIATED UNIVERSITIES,INC. P.O. Box 5000 I '

Upton. New York 11973-5000 l

a .

TEL (516)282 2107 ,

r FAX (516) 282 5569 _

Departrnent of Advanced Technology E Mall Building 130 October 5,1995 S. Khalid Shaukat RES/DET/GSl8 U.S. Nuclear Regulatory Commission 11545 Rockville Pike M/S 10C9 Rockwlie, MD 29852 2738 i

j

Dear Khalid:

4 j This letter provides the latest information on our activities related to the issue of a Reactor Coolant i

Pump (RCP) seal model for use in probabilistic risk assessments (PRAs). Information is included I l here based upon work performed at BNL and by our subcontractor, Robert Bertucio of NUS. This  ;

i material is by way of follow up to the July 25,1995 technical meeting at BNL, in which you

participated. .

! During the July 25,1995 meeting, Bob Bertucio was asked to apply the corrected seal LOCA model developed by BNL to the Sequoyah NUREG 1150 PRA and to estimate the significance of j the prior NUREG 1150 errors both on this plant and on a genenc basis for any PWR. He was also  ;

requested to estimate the level of effort required to correct the errors.

l

! Attached is a letter report from Bob Bertucio of NUS, dated September 1,1995, which roeponds l to these requests. On page 2 of his letter, item 7 discusses the corrected seal LOCA model,

! developed by BNL. This indicates that the seel LOCA paths and probabilities are already

! developed. This is true for the four pump model (Sequoyah), but the three pump seal model (Surry)

! has not been completed to date.

' This cover letter provides our review and suggestions for future actions concoming the RCP seal

! model and related PRA models.

i

1. The NUS letter report confirms that the seal model in Volume 2 of NUREGICR-4550 was not

! used by Bertucio in Volumes 3 (Surryl and 5 (Sequoyah) and that the final calculations for

COF in the NUREG 1150 program erroneously did not include the probability of seal failures l which start at time zero, after a loss of seal cooling (LOSC). A few other errors (unrelated to

, esel leakage probatetieel were also discovered in the 1150 model during the current review

by NUS.~

F g t90ers-960sts -

• me r ."'%

nux eamieemoos ce anocaus uncmv r-4

e

. Letter Ruger to Shaukat 2- October 5,1995 L

2. The correction of the error in RCP seal modeling alone resulted in about a 20% increase in the conditional core damage probabality (CCDP) for station blackout (580) seal LOCA sequences. The CCDP was noted to be very senwelve to the assurrptions and probabilities associated with the nun-recovery of AC power (NRAC).

3. While loss of service water (LOSW) sequences were not important in these two NUREG 1150 plants, they are important in some other plants. For those plants BNL had previously esti-mated that the seal model correction will result in an increase in CCDP for LOSW sequences of 250% (2.5 times). These results will also be very sensitive to NRAC probabilities.

4 The final effect of the seal modeling errors on the lelal CDF will be very plant specific and depends heavdy on the importance of the S80 and LOSW sequences relative to other sequences.

5. There are a number of other issues currently bemg evaluated (e.g., fire protection. GI 106. GI-

57) or requering evaluation in the future which may have important contributions from RCP seal failures. Proper quantification of these issues requires an accepted, reasonably accurate l seal model. ,

l

6. Industry IPE's have used widely different seal models which very from.very conservative I (faPore probability of 1.0 on LOSC) to quite non-conservative (e.g., CE's use of nearly zero M. l seat failure probabilities). Some IPE's may be using the incorrect NUREG 1150 seal model. = I j (

y Based on these conclusions and concems there appears to be a need by both industry and the

NRC for a consistent and reasonably accurate RCP seal failure model under LOSC conditions.

l Therefore, we believe that industry should be notified of the error in the current NUREG 1150 seal

model. Further, a relatively quick way to address the need for an accurate model would be to complete and publish the combined BNUNUS corrected model so that future analysis can be based on a standard, accurate model.

! If you have any questions or wish to discuss any of these issues further, please contact me.

l 3 Sincerely,,

L

Charlie Engineering Assessment & Technical Training CJR

sf Attachment cc: R. Bari t R. Bertucio. NUS J. Cadwell l M. Cunningham, NRC, w/stt.

4 R. Hall

! J. Higgins P. Kohut G. Martiner W. Pratt

C. Serpen, NRC w/stt.

I I

4

--