ML20217G534
| ML20217G534 | |
| Person / Time | |
|---|---|
| Site: | Seabrook  |
| Issue date: | 10/03/1997 |
| From: | Feigenbaum T NORTH ATLANTIC ENERGY SERVICE CORP. (NAESCO) |
| To: | NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM) |
| References | |
| AR#97021637, AR#97022306, NYN-97101, NUDOCS 9710100304 | |
| Download: ML20217G534 (5) | |
Text
.
).
NOF(l)
Nonti Atlantic linngy Sen'im Corporation
- " Atlantic P.O. Box 300 Se hna, Nii 0am I
(603)474 9501 i
The Northeast Utilitice System l
October 3,1997 Docket No. 50 443 NYN.97101 AR# 97022306 AR# 97021637 United States Nuclear Regulatory Commission Attention: Document Control Desk i
Washington, D.C. 20555 Seabrook Station Comments on Preliminary Accident Sequence Precursor Analysis of an Operationni Condition at Seabrook Station in a letter dated August 29,1997 the NRC provided a copy of the preliminary Accident Sequence Precursor (ASP) Analysis of an operational condition which was discosered at Seabrook Station on May 21,1996. North Atlantic has reviewed the analysis and has provided comments in the enclosure.
Should you have any questions concerning this response, please contact Mr. Terry L. liarpster, Director of Licensing Services, at (603) 773 7765.
Very truly yours, NORTil ATLANTIC ENER >Y SERVICE CORP.
ssL2W C. Feig46aum
/
Executive Vice PrMlent and Chief Nuclear Officer cc:
- 11. J. Miller, Region i Adiainistrator g
A. W, De Agazio, Sr. Project Manager pg0 F. P. Ilonnett, NRC Senior Resident inspector
)
9710100304 971003 PDR ADOCK 05000443 t
8
\\
pg
- ; r'_ g
4 COMMEN1S ON PREllMINAl(Y.ACCillEP'T_ SEQUENCE JRECURSOR ANALYSIS OFMDi'ERAllf JAl, CONDITION AT SEA}) ROOK STATV (
'ihe preliminary Accident Sequence Precursor (ASP) Ana.,
documents an opert..ional condition that was discovered at Seabrook Station on hiny 21,1996, and w $ reported in Licensee Event Report 96-003 01. North Atlantic's comments on the ASP analysis are pn 'ded below.
1.
Description of the Analysis Provided by Oak Ridge National 1 nhoratory The following are restatements of the ASP analysis:
EttnLilrnripilun Seabrook was at 100% power on hiay 21,1996, when personnel started the TDEFW pump for its scheduled quarterly surveillance test. The operator tripped the pump locally during the test aner sparks were observed emanating from the outboard mechanical seal area of the pump. The mechanical seal was disassembled and inspected. The sparks were the result of mechanical interference within the seal assembly. The outboard seal gland had 0.007 in, clearance from the top of the shaft sleeve and the throttle bushing inside diameter. The sparks were caused because the shan sleeve rubbed against the inside diameter of the throttle bushing, causing a 0.005 in. gouge in the shaft sleeve and chipping of the l
throttle bushing. Licensee personnel concluded that because of die improper installation of the seal, the TDEFW pump would not have been able to perform its safety function for the required mission time (24h) since the November December 1995 refueling outage.
Aner repairing the TDEFW pump, personnel inspected the mechanical seal of the motor-driven emergency feedwater (hiDEFW) pump and discovered it to have a similar alignment along with the corresponding indications of mechanical rubbing. The h1DEFW pump outboard mechanical seal gland had a 0.0035 in. clearance between the shaR sleeve and the top of the throttle bushing inside diameter.
The h1DEFW pump throttle bushing was not chipped like the throttle bushing was on the TDEFW pump.
Presumably because the h1DEFW pump had not failed a quarterly surveillance test, the system engineer concluded that the htDEFW pump was capable of performing its design function.
The design clearances and tolerances of the TDEFW pump's mechanical seals were insufficient to prevent damage during operation unless the installation technique used non-customary methods (i.e., use of dial indicators and fteler gauges). The design permitted the allowable tolerances to be greater than the available clearance. llence, the design did not preclude the interference between the seal and the shan sleeve. This design deliciency also applies to the h1DEFW pump mechanical seats. Contributing to this event was the failure to adequately incorporate previous knowledge regarding seal installation into maintenance procedures or training. As a result, maintenance personnel were unaware of a prior seal failure (in 1987), or the need to take precision measurements to verify the proper installation of the seal
- assembly, Additional Event.Related Information The emergency feedwater (EFW) system consists of two 100% capacity trains that feed a common discharge header (Ref. 3). One train uses the TDEFW pump and the other train uses the h1DEFW pump.
All four steam generators can be fed by either EFW pump. The TDEFW pump is supplied steam from page 1 of 4
the A and 11 steam generators.11.e MDEFW pump is powered from 4160V emergency bus E6 supponed 1
by the ll' emergency diesel generator (EDG).
Seabrook also maintains a start up feedwater pump with a capacity approximately equivalent to the combined capacity of both EFW pumps (Ref. 3). The start up feedwater pump can be staned from the control room, except during LOOP. Two normally closed motor operated valves (MOVs) must be opened to establish feedwater flow. Following a LOOP, the normal power source to the start up feedwater pump is not supplied power from an emergency bus. Therefore, the normal breaker alignment for the start-up feedwater pump must be altered from 4160V bus 4 to 4160V emergency bus ES (emergency bus ES is powered by the A EDG). The normal and alternate start up feedwater pump breakers are key interlocked, requiring one breaker to be racked out before the interlock key can be removed. The interlock key is required to rack in the alternate source breaker (from bus ES) to the start-l up feedwater pump.
Muddinninuinpuuns Even thougi, previous surveillance tests were successfully completed, the licensee concluded that the TDEFW pun p wo '4 not have been able to perform its safety function Ibr the required mission time (24h) since tl:e Novem,er December 1995 refueling outage (Ref.1,2). llence, the TDEFW pump was considered ir inerable 'ad its failure probability was adjusted to 1.0 (TRUE) for a 3,875 h condition assessment. The a,% h condition assessment is based on the TDEFW pump being required from the end of the outage on December 9,1995, until the discovery of the mechanical seal failure on May 21, 1996. Two days (48 h) were subtracted from the total number of hours that the TDEFW pump was unavailable to account for a reactor trip in January.
The licensee indicated that the MDEFW pump would have performed its safety function for the required mission time. Ilowever, because the outboard mechanical seal on the MDEFW pump had wear similar to that of the TDEFW pump, the potential for a common cause failm,: mereased. The EFW common.
cause factor was developed based on data distributions for mixca pump types contained in INEL-94 0064 Common-Cause Failure Data Collection and Analysis Wstern (Ref 4. Table 919: Alpha Factor Distribution Summary - All Types Fail to Start, CCCG = 2, u2 0.0884). llecause n; is equivalent to the p factor of the multiple Greek letter method used in the Integrated Reliability and Risk Analysis System (IRRAS) models, the common-cause failure probability of the EFW system pumps (AFW FMP CF.
ALL) was adjusted from 3.8 x 10" to 8.84 x 10' based on the common cause failure potential.
The Seabrook Individual Plant Examination (IPE) indicates that the start up feedwater pump is a backup source of feedwater of the EFW system. To credit the use of the start up feedwater pump, a basic event was added to the IRRAS model for the Seabrook plant based on the IPE value for a failure of the stan up feedwater pump to start and run (Ref. 5, Table 7.91) or a failure of the associated valves to open (basic event EFW MDP FC-5FP). 13ecause an operator is requhd to open two normally closed MOVs to establish flow from the start up feedwater system, another basic event was added to account for the failure of an operator to manipulate the required MOVs (EFW XilE XM SFP). Finally, during a LOOP, an operator must realign the supply breaker for the start up feedwater pump to the A EDO. A basic esent was therefore added to represent the failure of an operator to complete this realignment (EFW-XilE-XM 11RKR). This last basic event was based on the assumption that it would take an operator approximately 15 min, following a LOOP, to perform the activity and that approximately 45 min were available before a steam generator dry out would occur, leading to core damage. A lognormal distribution was used to calculate the failure probability for EFW XilE-XM IIRKR.
Page 2 of 4 J
'lhe operator non recovery probability for the EFW system during a LOOP (EFW XI-lE NOREC L) was adjusted from 0.26 to 0.80 because this action is not independent from other operator actions. The operator must utst realign the supply breaker for the start up fecoaater pump to the A EDG (EFW XilE-Xht ilRKR). If the operator fails to realign this breaker, the stan up feedwater pump would not be available in a LOOP scenario (LOOP sequence 15). Further, if the operator does indeed fail to realign this breaker, it is more likely that the operator will fail to recover the EFW system during a LOOP.
Finally, during a Silo, the only source of EFW is the TDEFW pump; therefore, with the TDEFW pump unavailable, there is no opportunity to recover EFW. Ilased on this, the operator non-recovery factor during a S110(EFW XilE NOREC EP)was set to"TRUE"(recoscry not possible).
II.
Comments on the Preliminary AEddenLSnucnge nc. cursor Analysis The following are North Atlantic's specine comments on the Preliminary Accident Sequence Precursor Analysis presented by analysis section:
EuntDescription Paragraph 1, Sentence $ should be replaced by: "The outboard seal gland was making contact with the top of the shan sleeve and the throttle bu hing inside diameter."
Paragrc,h 1, add the following aner Sentence 6: "The inboard seal gland had 0.007 in, clearance between the top of the shaft slees e and the throttle bushing inside diameter."
Paragraph 2, Sentence i should read: " the mechanical seals of the motor driven.. discovered the outboard mechanical seal to have a similar position, along..."
Paragraph 2, last Sentence should be replaced by: "The inspection revealed that the burnishing identined on the outboard mechanical seal of the h1DEFW pump was consistent with normal rubbing experienced during pump startup. The system engineer concluded that the h1DEFW pump was capable of performing it's design function based on the review of the as-found clearance data."
Paragraph 3. Sentence 3 should read: "...between the throttle bushing seal (secondary seal) and the shan sleeve. There was never any contact with tbc primary seal."
Additionallient EclatedJnformattuu No comments.
Modeling Assumpilona Paragraph 2, Sentence 2 refers to the wear on the h1DEFW pump as being "similar" to that of the TDEFW pump. The wear on the h1DEFW pump was due to normal rubbing experienced during pump starts and was not similar to that found on the TDEFW pump. Ilowever, it was concluded that the hiDEFW pump seal was susceptible to the same mechanical rubbing as experienced on the TDEFW pump.
Sequence 39 from the Figure i " Dominant core damage sequence for LER No. 443/96-003" is the risk-dominant sequence in the NRC analysis. This sequence is station blarkout with failure of the turbine ariven EFW pump. In this sequence, no credit is given for recovery of ekctric power, presumably due to 4
Page 3 of 4
the assumption that "45 min were available before steam generator dry out would occur, leading to core damage."
Seabrook has performed analyses of this specine analysis, using the MAAP code, with comparisons to the RiiLAP code results. (Rl!F 1) These codes show that, for this sequence, the time to steam generator l
dryout is about 1.5 brs, with core uncovery at 2.0 hr, and core overheating at 2.2 hrs. As a result, l
substantial time is available for electric power recovery, in addition, the recovery cunes base their highest slope from 0 to 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />, indicating that a few additional minutes of recosery can be signincant for recovery probability. The probability for failure to recover offsite power in 1.5 hrs is approximately 0.25 and the probability for failure to recover at least one of two !!DGs in 1.5 hrs is 0.65. (Riil' 2)'t he overall non recovery probability for electric power is 0.25 x 0.65 = 0.16. This factor should be used in calculating the Station tilackout with !!FW failure sequence.
Rl!FERiiNCli
- 1. " Risk hianagement Actions to Assure Containment liffectiveness at Seabrook Station," PLG 0550, July 1987.
- 2. "STADIC4 Model for Frequency of Nonrecovery nf Electric Power at Seabrook Station at Power and Shutdow n," PLG 0507. Rev. 2 Jan.1990.
Generalfomments North Atlantic agrees with the analysis conclusion that during a LOOP event without the MDlil:W pump and TDEl'W pump available a heavy reliance is placed on operator action to maintain secondary cooling.
North Atlantic licensed and non licensed operators are routinely trained on shifling the startup feedwater pump to the emergency feedwater alignment evolution. Nonh Atlantic is con 0 dent that the operators would have been able to successfully complete this evolution during accident conditions utillring the existing emergency procedure guidanu and the specine training on this evolution. Ilowever, for this event, the engineering review of the MDEFW pump seal as found data concluded that the MDEFW pump would have performed it required safety function, for the required mission time, North Atlantic engineering personnel concluded that the TDEFW pump would not have been able to perform its safety function for the required mission time (24h) because of the improper installation of the seal. This conclusion was based on discussions with the pump manufacturer and engineering judgment.
The exact time that the TDEFW pump became inoperable could not be conclusively determined since the pump successfully completed two prior surveillance runs without any indications of problems related to the mechanical seal degradation. The system engineer evaluated the damage to the seal and conservatively determined that the pump had been inoperable since the mechanical seals were worked on during the November December 1995 refueling outage.
Page 4 of 4