ML20086S828

Ro:On 910822,indefinite Primary Containment Isolation Sys TS Limiting Condition for Operation Occurred After Unsuccessful Attempts to Quantify Leakage on-line.Caused by Mfg Flaws in Lower Wedge of Stanchion.Upper & Lower Wedges Replaced

ML20086S828
Person / Time
Site:	Brunswick
Issue date:	12/27/1991
From:	Spencer J CAROLINA POWER & LIGHT CO.
To:	NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM)
References
NUDOCS 9201060019
Download: ML20086S828 (5)

Text

p 1 CP&L Carolina Power & Ught company Brunswick Nuclear Project P. O Box 10429 Southport, N.C. 28461 0429 DEC 271B1 FILE: B09-13510C U.S. Nuclear Regulatory Commission

-ATTN: Document Control Desk Washington, D. C. 20555 BRUNSWICK STEAM ELECTRIC PLANT UNIT 1 DOCKET NO. 50-324 LICENSE NO. DPR-61 BRUNSWICK INFORMATION REPORT Gentlemen:

- The following event, while not reportable, was considered of possible interest to the Nuclear Regulatory Conunission.

l.

.Very truly you: ,

) .,

. '. Spencer neral Manager nswick Nuclear Proj ect l

cc; 'Mr. S. D bbneter -

Mr. N. B. Le BSEP NRC Resident Office i

l 300073 I

9201060019 911227 PDR ADOCK 05000324 'j#

v.

l.

l 5 PDR

i At 0440 on August 22, 1991, Unit I was at 100% reactor power. The Reactor Water Cleanup (RWCU) system was being isolated to perform maintenance surveillance testing. When the first RWCU system isolation valve 1-G31-F001 (RWCU Inboard Isolation Valve) was closed, 'the control operator noticed that the system pressu:e remained constant until the redundant isolation valve was also closed.

After attempts to quantify the leakage on-line were unsuccessful, the line was isolated and placed under clearance. This placed Unit 1 in an indefinite Primary Containment Isolation System Technical Specification Limiting Condition of Operation (LCO).

The 1-G31-F001 is an Anchor Darling 6" double disc gate valve which seals via an upper and lower wedge mechanism. When the valve torques shut, the wedges spread the two discs against the in-body seats (see Attachment 1 for a view of the disc pack assembly). When the valve was disassembled, it was found that the lower wedge landing stanchions were of unequal length (see Attachment 2) . This resulted in the lower wedge hitting the body bridges (contact points) prematurely on one _

side, causing the wedge to cock. The resulting binding action between the lower and upper wedges, as well as the discs, resulted in poor disc contact with the in body seats. Contributing to the bind:ng was the clearance between the upper and lower wedge assemblies. This clearance was less than the 1/8" recommended as n ninal by the vender. The vender agrees that the upper / lower wedge fit up was the probable cause of the valve leakage at low preasures, but as the wedge sets are individually hand ground there is not a generic concern. The wedge assemblies may have been able to accommodate the binding caused by the unequal stanchion lengths if this clearance had been greater. The vendor feels that combined with the low differential pressure (DP) across the valve when it was closed, the upper / lower wedge fit-up was sufficient to prevent satisfactory sealing. The valve had been satisfactorily production tested by the vendor at the design differential pressure but the low pressure seat tightness, as would be seen during Local Leak Rate Testing (LLRT), was not a valve procurement specification requirement.

These findings may explain the inability of the 1-G31-F001 to seal under low DPs (i.e., LLRT or when the valve is closed on-line against little or no DP). The vendor has indicated that a higher DP may be necessary to force the disc against the in-body seat to provide satisfactory sealing. This may account for the inconsistent results of the pre-operational LLRT's performed just prior to disassembly on September 4, 1991 (the first and third tests did not pressurize, whereas the second test did, although the leakage rate was unsatisfactory) . -

Markings cn the wedges, dhes , and body bridge confirm the impinrement which occurred between the wedges and discs. The bridge which mates with the longer wedge stanchion had slight indention marks (.003" deep) which indicated it had been hitting harder thr.n the opposite, shorter stanchion. The bride.e which corresponds with the shorter stanchion had markings which indicated only partial Z contact. The markings showing the contact area between the upper and lower wedges indicate these components were binding during closing of the valve. Typically, the contact points of the wedges would be in the same plane (approximately the mid point of the wedge guides) . Also, markines on the back of the downstream disc and lower wedge indicate that the disc had Ueen binding against the back of the lower wedge which resulted in poor disc-to-seat contact. Additionally, the stem had slight scuff marks on one side which were caused by the stem contacting the bonnet stuffing box bore during stroking as a result of the binding (i.e., the stem to upper wedge cennection is rigid as shown in Attachment 1).

A blue check of the dernstream in-body seat and disc seat iace showed 100%

contact. This substantiated the theory that the leakage was due to the disc not uniformly contacting the in-body seat and not due to the seats themselves.

Though replacement vedges were not a cause of this event it was noted that the wedges must be individually hand ground to make a matched net. Procuring separate components instead of a r atched assembly could cause future problems, or at least

i result in delays, as CP5L does not have adequate expertise to grind these wedges and must send them to the vender for grinding. Replacement parts _for the upper i and lower valve wedges will inLthe future be procured as assemblies (one upper .:

and one lower wedge), as:the individual wedges require grinding to make matched .;

sets prior to installation. Existing stocks of wedges will be fit-up into matched assemblies and restocked as assemblies.

The root cause of this event is considered to have been a combination of manufacturing flaws. The manufacturing of the lower wedge with one leg / landing stanchion longer than the other, combined with the less than tvminal clearance

, between the upper and lower wedge assemblies, resulted in the poor disc to in-body seat contact.

Corrective- actions for the.1-G3)-F001 included replacing the upper and lower vedges as well.as the stem. The downstream disc was lapped as a conservative measure to ensure a satisfactory sealing surface. The valve was reassembled and a satisfactory LLRT performed before returning the valve to service.

The safety significance of this event was minimal. When the 1 G31-F001 failed to-seat, the Technical Specification LCO Action statement was followed by isolating the affected line with the 1031-F004 and placing it under clearance. The failure of the 1 G31-F001 valve to adequately seal due to the manufacturing problems with the wedges is considered- to be an isolated case that would not affect the operability _of the redundant isolation valve (1+C31-F004) . The 1-G31 F001 valve did close and while it leaked by the seats it did not serious 1v compromise the Primary Containment boundary for this line as the redundant iso'lation valve did seal.

This report only discusses.the August 22, 1991, event which was caused by a manufacturing problem, Other operational problems with Anchor Darling double disc gate valves are still being evaluated by CP&L with Anchor Darling. Related valve i problems were reported in LERs' 2 91-011, 1-91-016 and 1 91 018.

4 F

5 4

. . , _ - - - - - - - - --_n-,-- - ---._,.-_

0 1

ATTAC10 TENT 1

_ 16_ _ stm c >

_0JSC RETAlWEA

.h i p

.U? pin wt006 k

-=Ql'iC ASSEM8LY 0 . LOWER WEDGE

• 8 "

3 1 I

-BODY

/

< sis' e t

0

=

. BRIDGE j

i r---- '-'

llr ii1 '-r . _,

, ~

^ .

ATTACHMENT 2

( h -

\ /

LDWGR WGbGE iFRDMT VIEW)