ML20070F128
| ML20070F128 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 07/08/1994 |
| From: | Broughton T GENERAL PUBLIC UTILITIES CORP. |
| To: | NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM) |
| References | |
| C311-94-2089, NUDOCS 9407180187 | |
| Download: ML20070F128 (4) | |
Text
.
s GPU Nuclear Corporation u
Route 441 South 0
Uh hf P.O. Box 480 Middletown, Pennsylvania 17057-0480 (717)944 7621 Writer's Direct Dial Number:
717-948-8005 July 8, 1994 C311-94-2089 U. S. Nuclear Regulatory Commission Attention: Document Control Desk Washington, D.C.
20555 Gentlemen:
Subject:
Three Mile Island Nuclear Station Unit 1, (TMI-1)
Docket No. 50-289 Operating License No. DPR-50 GPU Nuclear Review of the NRC's Preliminary Accident Precursor (ASP) Program Analysis of the TMI-l Event Reported in LER No. 289/93-002 This letter is in response to the NRC's May 10, 1994 letter which requested that GPU Nuclear review and comment on a preliminary Accident Precursor (ASP) Program Analysis of the January 29, 1993 event at TMI-l which resulted in simultaneous bypass of cooling water to both Decay Heat Removal (DHR) coolers for approximately 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />. The event was reported as TMI-l Licensee Event Report (LER) No. 289/93-002. GPU Nuclear has reviewed the preliminary ASP analysis and found that additional information beyond that provided in the LER, Individual Plant Evaluation (IPE), and Updated Final Safety Analysis Report (FSAR) is needed to establish modeling assumptions (Section 0.1.3 of the ASP analysis) which can then be used to calculate a core damage probability estimate.
The LER evaluated the worst case scenario, that of a Large Break Loss Of Coolant Accident (LBLOCA). Because the analysis of a Small Break LOCA (SBLOCA) is bounded by the LBLOCA, the SBLOCA is not described in the LER.
The LER conclusion that the LBLOCA is bounding is based on the fact that the earliest challenge to Engineered Safeguards (ES) equipment (as the result of depleting the Borated Water Storage Tank (BWST) and requiring transfer of the suction of the Low Pressure Injection (LPI) Pumps from the BWST to the Reactor Building (RB) sump) comes as the result of this LBLOCA scenario.
This scenario allows the least amount of time for recognition by the operating crew that a DHR Service Cooler is bypassed.
Although the conclusions of the LER remain the same regardless of the break size. certain SBLOCA break sizes would require operation of High Pressure 9407180187 940708 t
PDR ADDCK 05000289
/ [f ?/l
.LUu J ~
._ _PDfC GPU Nuclear Corporation is a subsidiary of General Public Utilities Corporation g
e
)
C311-94-2089 Page 2 of 4 1
Injection (HPI) at fluid temperatures higher than the TMI-l FSAR fluid design temperature of 200*F for continuous operation. Additional information concerning a SBLOCA is needed to facilitate discussion of the modeling assumptions of the ASP analysis.
Therefore, GPU Nuclear offers the following comments that directly affect the modeling assumptions, as discussed in Section 0.1.3 of the ASP analysis. Our comments, which have been discussed with Mr. Patrick O'Reilly of the NRC in a telephone call on 1
June 24, 1994, are as follows:
1.
Initiating Event Freauency r
The SBLOCA initiating event frequency used in the preliminary ASP i
analysis (See Table A-13 in Enclosure 5.) appears to have been the i
BWR frequancy (2.0e-2/yr) instead of PWR frequency (1.5e-2/yr).
t The generic SBLOCA initiating event frequency is based on data for i
the early 1980s and is not specific to THI-1. The TMI IPE uses a i
SBLOCA mean frequency of 6.7e-7/hr (5.87e-3/yr) which was evaluated by Bayesian update of the generic SBLOCA mean frequency with TMI-l specific data of zero SBLOCA events in 10.25 years of operation.
l Assuming a 3 hour3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> unavailability of HPR and 0.43 non-recovery
[
probability, as stated in the ASP analysis, use of the TMI-l SBLOCA initiating event frequency rather than the generic SBLOCA frequency would result in a core damage frequency of:
6.7e-7/hr x 3hr x 0.43 - 8.6e-7.
2.
Unavailability of HPR Initiation of RB sump recirculation is not expected to occur until 3-10 hours after the start of a SBLOCA. As indicated in the LER, action was taken by TMI-1 operators to correct the condition 2.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> after the coolers were bypassed.
Since recovery from the cooler (ES) bypass would have taken less than 30 minutes, the action would have been completed by the time actual recirculation was initiated.
Therefore, the SBLOCA probability should be applied to 2.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> unavailability of HPR instead of 3.0 hours0 days <br />0 hours <br />0 weeks <br />0 months <br />.
Also, operators are trained to assess the status of ES systems in the event of an accident.
If there had been a LOCA during the time that the DH Service Coolers were isolated, it is likely that the control room (if not the individual who isolated the coolers) would have recognized immediately that a surveillance had been started which I
could have affected the Decay Heat Closed Cooling Water System (DCCW).
Even if control room communications were assumed to be poor at that time, it is likely that the Shift Supervisor would have r
recognized that an Operations Surveillance had been started which could have affected DCCW cooling and immediately taken action to restore the system on receipt of the high temperature alarm.
i l
i
I C311-94-2089 Page 3 of 4 3.
High Pressure Recirculation Non-recovery Probability The preliminary ASP analysis assumed a 0.43 non-recovery probability for High Pressure Recirculation (HPR). Assuming an allowable response time of 2 hrs based on continuous operation of HPR and an actual recovery time of 20 min to correct the valve alignment, the non-recovery probability would be reduced to 0.01.
Although GPU Nuclear is unable to provide what we would consider to be the appropriate non-recovery probability within the time constraints allowed by your schedule, we believe that 0.43 is too high.
The following is provided for discussion purposes.
a.
HPI Based on engineering judgement and information from the HPI Pump seal manufacturer, the HPI seals can accommodate operation at elevated process water temperature (250*F) for a period of two hours.
The TMI plant design utilizes three HPI pumps that would be running following ES actuation with offsite power available.
The third HPI pump would not have been affected by the simultaneous bypass of cooling water to both DHR Service Coolers.
In this case, given a SBLOCA, that pump would have continued to operate provided a suction source were available to it from the BWST or from the LPI pump's discharge.
b.
LPI LPI pump operation at elevated process water temperatures is not a concern. The LPI pumps are designed for operation at 300"F.
Evaluation of LPI pump operation at elevated cooling water temperatures indicates that the pump and bearings would have continued to operate for at least two hours. Motor operation under these conditions would need to be confirmed.
GPU Nuclear has declined to pursue further evaluation of the LPI pump motor because of: the vendor's estimate of five to seven weeks to complete the work, the cost, and the other factors discussed above which result in a core damage probability below 1.0e-6.
However based on engineering judgement, we feel that it is likely that LPI pump motor operation would continue for some period of-time under LOCA conditions.
As the result of accounting for each of the above mentioned factors,
. including 2 hour2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> availability of LPI which is not confirmed, GPU Nuclear has calculated'that the core damage probability for this event would be 1.67e-8,-
which is well below the ASP threshold of 1.0e-6.
Without confirmation of LPI pump motor operation under LOCA conditions, our calculations resulting from the other factors discussed above still result in a core damage probability below 1.0e-6.
Considering only the change discussed in item no. I above, use of the THI-1 SBLOCA initiating event frequency rather than the generic SBLOCA frequency would result in a core damage frequency of I
}
4 C311-94-2089 Page 4 of 4 8.6e-7 which is still below le-6.
Therefore, the TMI-l event reported as 1
LER No. 289/93-002 is not an accident sequence precursor event as defined in I of your letter and should not be included in the 1993 ASP Report.
If you have any questions regarding any of the assumptions used in our analysis, please call Bob Knight at TMI (717) 948-8554.
Sincerely, b%%q bd^'
T. G. Broughton Vice President and Director, TMI MRK cc: Administrator, Region I i
l TMI-l Senior Resident Inspector TMI-l Senior Project Manager
)