ML20214M087

Forwards Input for Resident Inspector Rept for Insp of Facility on 870330-0401.Insp Conducted to Determine Technical Reasons for Four Broken inner-external Closure Springs on Msivs.List of Sequence of Events Also Encl

ML20214M087
Person / Time
Site:	Fermi
Issue date:	05/27/1987
From:	Merschoff E Office of Nuclear Reactor Regulation
To:	Greenman E NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION III)
References
REF-PT21-87 NUDOCS 8706010164
Download: ML20214M087 (9)

Text

--

y ss ua

'g UNITED STATES 8 o NUCLEAR REGULATORY COMMISSION g p WASHINGTON, D. C. 20655

\ *****

/ May 27, 1987 Docket No. 50-341 MEMORANDUFi FCR: Edward G. Greenman, Deputy Director Divisicn of Reactor Projects Region III FR0h: Ellis W. Merschoff, Acting Chief Vendor Inspection Branch Division of Reactor Inspection and Safeguards Office of huclear Reactor Regulation

SUBJECT:

FERMI 2 BROKEN INTERNAL - EXTERNAL CLOSURE SPRINGS ON ATWOOD & MORRILL MAIN STEAM ISOLATION VALVES Enclosed is our input for the Fermi 2 resident inspector's report for the inspection conducted by J. C. Harper on March 30 - April 1,1987 of Fermi 2.

The inspection was conducted in order to determine the technical reasons that four inner-external closure springs on the MSIV's were broken.

During the inspection it was determined that the inner springs (Heat #8067703) failed due to inadequate heat treatment. At the time of manufacture of the subject springs, the spring manufacturer, Duer, had apparently lost control over their heat treating operation. Contrary to the Femi 2 LER 86-011, both inner and outer springs are identical materials therefore they are potentially affected provided that they were heat treated during the same time frame as Heat #8067703.

The Fermi 2 Part 21 evaluation only addressed the inner-external springs and did not take into account that the outer springs (same material as inner springs) may have suffered temper embrittlement and quench cracking. There is a concern for the material integrity of the outer springs. The General Electric reconnended proof testing of the inner springs (or outer springs) at 105% of load in order to sort out any defective springs was proven to be inadequate by f racture mechanics analysis and reoccurring inner spring failures.

ff Ellis W. Merschoff

,/

ting Chief l

l Vendor Inspection Branch l Division of Reactor Inspection and Safeguards Office of Nuclear Reactor Regulation

Enclosure:

Inspection Findings cc: J. J. Harrison, RIII W. G. Rogers, SRI Fermi 2 8706010164 870527 , r y PDR ADOCK 05000341 -'

,1 0 PDR ,

,}

7

,, Inspection Finding The inboard and outbcard main steam isolation valves (PSIV's) failed their local leakage rate test (LLP.T) during a required IC month surveillance test en April 28, 1986. According to LER C6-011, the leakage rate of each set of valves tested (4 sets of valves were tested) was greater than 42.30 standard cubic feet per hour (SCFH). The acceptance criteria established in Technical Specification 3.6.1.2.c for leakage rate is less than er equal to 100 SCFh for all four sets of valves combined when tested at a pressure of 25.0 pounds per scuareinchgauge(psig). The total leakage rate fcr all four sets of valves tested combined to exceed the acceptance criteria of 100 SCFH by at least 69.5 SCFH. All eight hSIV's were disassembled and inspected to determine ar.d correct the cause of failure. According to Detroit Edison Ceviation/ Event Report Number NF86-0170, "MSIV-LLRT Failure -Steam Lines A, E, C, D," the failure of the valves to pass the LLRT was apparently due to excessive wear to the poppets and pilots seats and guides. Wear caused the poppets and pilots to position themselves improperly in their seats. As part of the investigation Fermi 2 discovered on 5/11/86 that four inner external MSIV springs (17/32" Dia x 4-5/8" 00) on the outbuard valves were broken. Of the springs that broke, three had one break approximately 2/3 the distance frcm the bottom of the springs and the remaining spring broke in several places. According to Atwood & Morrill

~

Company, Inc. (A&M), the outer springs were also supp1ied from DueTSpring oil the same purchase order. A&M claims that the outer springs were from a

~ _

different heat number.

Three broken springs were found on the upper units and the remaining broken spring was found on the lower unit. As stated previously, four inr.er external MSIV springs of two outboard valves were discovered broken (17/32" dia. x 4-5/6 0.D.). Valve B2100F028A contained one broken spring and valve B2100F02EB contained three broken springs. These springs were AISI 5160 material, purchased with valves from A&M, Part #33134-824-3219. The springs were purchased by A&M from Duer Spring (A&h Furchase Order #AM5243 Heat No. 8067703, Oven #20224-R5, test report dated January 6, IS72). Twc of the broken springs and three non-failed springs were submitteo to the Detroit Edison Engineering Research Division (DER) for retallurgical examination. The outer springs (6-1/4 inch 0.0.) were not examined. From their investication DER concluded that the root cause of the spring failure (Heat #8067703) was three fold in nature, temper embrittlerrent, quench cracking, ar.d surface imperfections.

Acccrding to DER, "The metallurgical results revealed that failure was attribut-able to pre-existirs seams and quench cracks (induced during heat treatment) which propagated as brittle fracture due to temper embrittlement of the spring material. The three non-failed springs were examined and quench cracks were found in twc cf the three springs; all three springs exhibited temper embrittle-ment. Based on these findings, it was recommended that all inr.er springs be replaced.

From independent review of the DER metallurgical report, the NRC inspector concurs with the DER conclusions. The material, AISI-5160 used in the failed springs, is highly susceptible to 500 F embrittlement (or temper embrittlement) because the alloy contains substantial amounts of Cr, Mn and P. Temper embrittle-r..ent is caused by tempering a susceptible alloy in the 400-700'F range, thereby I 1

I l

i*

r '

allowing Mn and P to segregate at the grain boundaries and subsequently render

the material brittle. As reported by DER, Scanr.irg Electron Microscopy (SEM) l of the fracture surfaces revealed predominate intergranular type fracture and numercus grair. bcundary cracks (this was reported to have been observed on the failed spring sample away from the quench cracks). The fracture surface SEM micrograph reported by DER is good evidence that the spring material was temper embrittled during heat treatment (HT). The CER investigation stated that the

non-failed inner springs examined wert also temper embrittled. As a result of destructive irrpact testing it was determined that the nonfailed springs had a similar grain boundary fracture as the failed springs.

During forming, the bar stock was heated to 1700-1750'F and the springs coiled.

l The springs were then quenched in oil to 4CC'F. Subsequently, the springs were tempered in a furnace at 650'F for 1.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />. According to A&M during the time

! frame when the subject springs were manufactured, Duer's heat treating processes were being carried cut manually. Manual heat treating is subject to many nere

, errors in prccess control than autcrrated, continuous heat treating processes.

It is conceivable that a loss of control during manual heat treating caused problems with the failed hSIV springs (quench cracking and temper embrittlement).

In this situation, overheating during austenizin i

quenching rate and medium (water instead ofare oil)g and/or likely using causes an improper of quench cracking. however, it is more probable that time delays between the austenizin5/

quenching process and the tempering process caused the quench cracking, since a manual HT process transfers materials from one HT station to the next in a

ncncontinuous manner. It is known that time delays before tempering and tempering a material that is below recommended tempering temperatures promotes quench cracking. At present, Duer employs an automated, continuous HT process.

) Detroit Edison contracted independent consultant, Failure Analysis Associates i (FAA) te determine whether fatigue or stress corrosion cracking were probable causes of failure for the springs. According to FAA, "In its position as acvisor to Fermi 2 Nuclear Proouction, FAA did not perform an independent metallurgical evaluation of the failed springs. However, FAA dio review some of the metallurgical evidence prepared by Detroit Edison ERD." However, FAA did perform a fracture mechanics analysis and based on these results cercurred with General Electric's (GE's) positico that a 105% cverload test on the springs

woulc te a suitable means of separatin5 defective springs. For EWRs in operation,

, the GE Service Informaticn Letter, dated July 18, 1986, titled " Inspection ot Atwood & Verrill MSIV External Springs" recomn. ended replacemer.t cf ary cracked J,

or broken springs following a 5% overicad test performed with springs installed or as a bench test. The springs were to be stroked ten times to 105% of the r.ornial stroke at a cc. pression rate less than one inch per second. According to GE, "A spring which passes this (ti:e 105 percent cf load) test is expected to provide satisiactory service at normal loads." The FAA justification of GE's i reconsendation vas based on their fracture mechanics analysis.

l l

l

___-___-___________-_____-_____--___________-___-_-_-_____-__A

I.

Powever, it should be rcted that the FAA fracture mechanics analysis is in contradictico to the cor.clusions reached in the EER fracture trechanics stucy.

DER concluoed from their fracture mechanics analysis that the 105% overicad test woulc not be an adequate means cf sorting cut defective springs. The FAA fracture trecFanics analysis did not take into account that the inner springs had been temper erbrittled, therefore their analysis was carried cut using literature fracture toughness parameters of properly heat treated AISI 4340 spring material instead of a comparable material that was temper embrittled.

On the other hand DER's fracture trechanics analysis utilized fracture tought.ess values for the embrittled springs (30-40 ksi-ini) in order to calculate a critical crack size of .03 inches (using the A&F reported service stress of 87.6 ksi - valve in open position). The actual critical crack sizes n.casured cr the tested material by DER varied from .044 .077 inches. DER estimated that in a 5% overload conditicn a critical crack size ci .027 inches would resuit, as opposed tc a critical crack size of .030 inches in the nonoverloadeo condition (assuming embrittled traterial). Lased on these calculations it was correctly concluded that a proof load test would not be an ef fective neans of sorting out defective springs.

On 6/2/66 a Part 21 evaluation was corrpleted by Fermi E. The evaluaticr. was limited to the inner-external springs and did not adoress temper-embrittlement as teing a controlling tactor in the root cause of the failure. According te the evaluation "... All,nonfailed springs in all eight of the MISV's were proof tested according to a plan outlinec by Cetroit Edison consultants, FAA and ccncurred with by GE and Eetroit Edison Engineering." As pointed out previously, Detroit Ediscn Engineering did not ccncur with proof testing of the springs.

On 6/13/06, A&h confirmed that the failure cf the MSIV springs were caused by quench cracks developing during the tranufacturing process. A.t that time

/SP was of the cpinion that ene or trore failed springs could cause slower closing speeds in the nern.al forward flow directions, and increase the difficulty tc pass the local leak rate test in % R units and also affect valve closure in the reverse flow cirection in PWF units sir.ce the springs 61one provide the coly external closing assistance. A&V. recommended that the external cicsing springs en all MSIV's be magnetic particle (MT) inspected as soon as possible. l Ferri 2 was provideo with a procedure fcr performing NT from Duer. I f tor, ar crerations star.dpoint it is Detroit Edison's position that the springs are a seconcary system designed tc assist the primary (air activatico) systems in closing the valves. GE estimated that if three out of eight outer springs or all eight inner springs failed, the valve would take about one more seccnd to close, also all eight BSIV inner springs ccntribute 10% of the total closing force. EF-i-FUP states that in a con.plete sudden stean line break outside primary containtnent the fuel barrier is prctected acainst loss of cooling if MSIV cicsure takes 10.5 seconds or less. EF-2-Technical Specifications require closure times ci 3 to 5 seconds.

j 0 1

Conclusion As a result of the inspection, the hRC inspector concludes that at the time of manufacture of the subject MSIV springs, Heat #8067703, Duer had lost control over their heat treating operation. Both the inner and ou_ter sprinSs are identical materials of construction, therefore they are both potentially i

affected by improper heat treating if they_ were' heat treated during the sa_me

~

perio6 et time as Heat #8067703. DER evaluated both failed and nonfailed fnner

~~

a springs' and cetermined the springs to be temper erbrittled as well as containing quench cracks. According to the DER metallurgical report and the Fermi 2 part 21 evaluation, the outer springs were not evaluated. In fact, the_, Fermi 2 LER 66-011 informed the hRC that the cuter sprines were of a different material of construction than the inner springs, therefore dismissing any possible potential heat treating problems with the cuter springs. There is a concern for thi~katerial integrity of"the outer springs. The GE reconmended proof test for sorting out defective springs was proven to be inadequate by the DER fracture mechanics aralysis and since there were reoccurring spring failures following the proof test acceptance.

Fermi 2 shculd clearly describe the af fect that the inner and outer springs have on valve closure times. To date the hRC inspector has received written statements frcm both GE and A&h stating their positichs, the MSIV's can acequately fulfill their safety function in their present ccndition. DER has submitted hand written notes to the NRC in attempts to justify the use of the outer springs. However, there has not been a formal metallurgical analysis completed on the suspect outer springs, nor a detailed calculation on the effects that a failure of the inner ano outer external springs have on desired valse closure times of 3 to 5 seconds as stated in the Fermi 2 technical 1

specifications. As cf 4/13/87, Ferni E conAitted to gather data relating to 1

the cuter-springs' heat treatment and heat treatrcent time frame as well as to rhysically inspect the springs atter every five cycles until such time that a destructive test can be perforned (cn the outer springs).

I I

f

s

. ATTACHMENT 1 CETROIT EDISON 00CUMENTED SEQUENCE OF EVEhTS Event Date Recorded

1. LLRT failure of inboaro and outboard MSIV's - 4/28/86 Deviation / Event Report issued NP-860170

2. Corrective Action Review Board Approval - Deviation 4/30/86 Event Report LLRT Failure (NP-860170) for Wear of Poppet's and Pilot's Seats and Guides

3. Broken inner external springs found on MSIV 5/11/86 outboard valves (Heat 8067703)

4. Corrective Action Review Board Approval - Deviation / 5/14/86 Event Report Broken Inner Springs on MSIV Outboard Valves (hP-86-182) 1

5. Broken springs evaluation completed Deviation / Event 5/16/86 Report (NP-06-182)

6. GE letter stating their position en the MSIV broken 5/21/86 spring safety relevance

7. LER No.86-011 to the hRC describing NP-86-0170 and 5/28/86 NP-86-182

8. Part 21 evaluation completed by an assigned Detroit 6/2/86 Edison Fermi 2 systems engineer

9. Part 21 evaluation conmittee approval of the Fart 21 6/13/86 evaluation report ccmpleted on 6/2/86

10. Detroit Edison Engineering Research Report (Netallurgical) on the broken springs

11. Detrcit Edison Operations Plant Experience Report 7/15/86 surrr6rizing the history of the safety significance of the broken MSIV springs

12. GE Information Letter summarizing the safety 7/18/86 significance of the broken MSIV springs

13. Root cause evaluation of the LLRT failure oescribed 7/25/86 ir. Deviation / Event Report NP86-0170 completed

14. Corrective action con.pleted for Deviation / Event 7/31/86 Report NP 80-0170

15. Corrective action approved for Deviation / Event 8/8/86 Report NP 86-0170

s Event Date Recorded

16. Failure Analysis Associates (Metallurgical and 9/86 Fracture Analysis Report issued)

17. lwo more broken inner-external springs found 28B/28C 3/24/87 from Heat #8067703

9..

4

. J. J. Harrison May 27, 1987 DISTRIBUTION:

RIDS - Code -IE:09 VIB Reading DRIS Reading JPartlow BGrimes EMerschoff JStone JHarper d Slek 98 h 2.

3 ,n

~ ~ ~

0FC :V10:DR15 i3C/ VIB:DRIT~~iKE'C7V'lh:DRIS: ~ T'~~~~  :

. . . . . : . . . . . . . . g .". : ..:.

..e  :

NAME :JHarper :JStone  : ' schoff  :  :  :  :

.....:...... ... ...:........ .../i............:..... . ....:........... :............:...... .

DATE :$j 1/87  :$/91/87 [ .:_ffa_ -/87  : ._.__ _ _

f' k J. J. Harrison -2 May 27, 1987 DISTRIBUTION:

RIDS Code IE:09 VIB Reading DRIS Reading JPartlow BGrimes EPerschoff JStone JHarper d Steba. . 9 8 Fec-i 2.

~ ~ " ~ ~ " ~

OFC :VIB:Dkl5 :lC/VIB:DRIT~:WQVill:bRIS:  :  :  :

.....:........ g .,:....... ..:.

j(f.......:................................. . ..:........

NAME :JHarper :JStone *

• schoff :  :  :  :

.....:........... ...:.......... 7 /f............:............:..................................

DATE:$]y,\/8,7,,_,,,.:$/d/87 8 / : f/2 -f 87  :  :  :  :