ML20202G243

Discusses Unsettling Implications of Water Hammer Incident. Uncoupling of Util Licensing Engineering from Operations Seems to Ignore Lessons Learned from TMI

ML20202G243
Person / Time
Site:	San Onofre
Issue date:	03/29/1986
From:	Catton I Advisory Committee on Reactor Safeguards
To:	Ebersole J NRC
References
ACRS-CT-1840, NUDOCS 8607150346
Download: ML20202G243 (1)

Text

_

CT- /g40

• O TO: Jesse Ebersole 29 March, 1986 FROM:

SUBJECT:

Ivan Catton L. . .._, _ ,,;_ ,3 Westinghouse Water Reactors Subcommittee-Meeting Q

I 2 b ary, 1986 -

.b The SONGS-1 water hammer incident ha,a some rather unsettling implications. Aside from the obviously poor maintenance practices, the uncoupling of the SCE licensing engineering from operations l seems to ignose the " lessons learned" from TMI. The operations staff at the plant have, or at one time planned to, put in place devices to sense whether the feed water lines were water or steam filled with appropriate alarms. The devices were actually purchased. The SCE licensing staff knew nothing about them. It seems to me that the poor maintenance record implies a need for another level of protection like the warning system considered '

by the SONGS-1 operating staff. Further, the incident should put water hammer in a different perspective than that now held by the NRC staff. It should also bear on our perceptions of the V sequence core melt initiator.

l l

8607150346 860329 PDR DESIOUATED ORIGINAL

-- 18 1 Certified By

+ . l ci-It'll 7-l TO: El Igne . 29 86 FROM: Ivan Catton C){ @I

(

SUBJECT:

MetalComponentsSubcommitteeMeeting'~g.l'Q,[-

H street I

27 February 1986 gg

• q At the subcommittee meeting I told you I had some comments on the PTS Draft Reg Guide. Although tardy, my comments follow. Actually I'm trying to clear the decks of promised reports before I go off to Holland to give a series of' lectures.

The claim is made that " conservative assumptions" make the results bounding (?) for any PWR. If assumptions are made one l must know what they are and be able to list them. I can find no l such list. The only conservative ass;umption I can find is the GO deg F added to the RTNDT.

The Reg Guide states that a number of calculations must be made f if the calculated frequency of thru-wall cracks is greater than SE-06. I calculated SE-05 for Oconee using numbers from staff sponsored research reports. What next? It seems to me that either the screening criterion needs adjustment or at a minimum some re-calculation is in order.

The rule specifies an across the board RTNDT of 270 deg F for axial and 300 deg F for circumferential welds. My review has led me to believe that either the numbers are too large or there ought to be some plant specific caveats and some attention paid to what happens following a through the wall crack.

I don't understand all the "shoulds" in the first paragraph of chapter 1. The chapter does describe an overall approach not "should describe" .

This carrying the Reg Guide philosophy a bit too far. Chapter 2 describes briefly what is required to describe plant systems. Saying it should do so is silly. It either does or it doesn't.

Under topics covering operator actions, I would think requiring something like the ORNL contribution to the SEP program would be very valuable. It was essentially an operating history of the plant with some conclusions about maintenance practices and operations.

In the chapter discussing T/H calculatios, the reader is referred to Calvert Cliffs and HB Robinson PTS calculations base on TRAC and RELAP5. Can the reader use the code that impacts least on his plant? Recall that TRAC and RELAPS will yield dramatically different vessel conditional failure probabilities.

The section on T/H models (4.2) is too loose. For example:

a)Dne doesn't discuss nodalization schemes, one uses standard techniques to estimate errors.

b)You don't easily estimate multi-dimensional behavior from one-dimensional calculations without full scale experimental data.

O ' +1 M # o7 c j

_v9 i ' \ i m (p tr31GUATZD ORIG ~UAL CedMici 13y -

u .

. - 4 #., .

o c)There is no requirement for accurate prediction of condensation except in the pressurizer. We all know the codes do not handle condensaticon very well.

Will the applicant T/H modeling have to be better than that found in TRAC and RELAPS? Are there no requirements for code verification?

I have no comment on chaprter 5 and chapter 6 looks good.

The chapter on uncertainties (7) is weak. There is no way a Monte Carlo analysis will help determine the uncertainty resulting from poor or physically incorrest modeling. I think one should limit the use of Monte Carlo methods to where they make sense. This is the section where requiremenmts for nodalization studies belongs.

Under corrective actions (chap 8), I see no mention of pressure control, say by a " Reed system sever".

j

..