ML17059C623
| ML17059C623 | |
| Person / Time | |
|---|---|
| Site: | Nine Mile Point  |
| Issue date: | 04/12/1999 |
| From: | Kachnik L, Knoll A, Moody J NIAGARA MOHAWK POWER CORP. |
| To: | |
| Shared Package | |
| ML17059C624 | List: |
| References | |
| ES99-74, SAS-99-22, SAS-99-22-R, SAS-99-22-R00, NUDOCS 9904220284 | |
| Download: ML17059C623 (16) | |
Text
AlVALFSIS GROUP TECH1VICALREPORT SAFETY AND AVAILABILITYASSESSMENT NMPI Probable Maximum Precipitation (PMP) External Flood Risk
- Diesel Generator Rooms
[SAS-99-22]
Revision 0 ES99-74 April 8, 1999 PREPARED:
Leo Kach ik PREPARED:
Alex noll
<</i~
REVIEWED:
Jim oody 1</ f APPROVED:
Ted Kulczycky
'I I2 Niagara Qgg~~Mohawk'904220284 9904i2
+~)
PDR ADOCK 05000220~
P PDR~
NMP1 EDG Room External Flood Risk TABLEOF CONTENTS 1.0 Background and Objectives.
2.0 Results and Conclusions.
3.0 Analysis.
4.0 References.
Attachment 1 RISKMANPRA Calculations.
NMP1 EDG Room External Flood Risk 1.0 Background and Objectives The risk from external flooding was assessed in the NMP1 IPEEE(Reference 1). The risk was judged to be near or below the screening criteria of 1E-6/yr although the analysis was mostly qualitative. The NRC asked for additional information (Reference 2) regarding the potential impact on flooding; NMPC response(Reference
- 3) provided additional information. Now NRC has asked (Reference 4) specifically that NMPC re-evaluate core damage frequency (CDF) due to PMP flooding ofthe diesel generator rooms based on their observation that CDF could be approaching 1E-4/yr.
The objectives ofthis analysis are as follows:
~
Provide a simplified analysis ofCDF due to the PMP to more clearly communicate why CDF is not approaching 1E-4/yr.
~
Use the simplified analysis to support sensitivity analysis and decision-making with regard to further protecting the diesel generators during a PMP event.
2.0 Results and Conclusions The base case analysis in Section 3 indicates that CDF is less than 1E-6/year. As described, this is judged to be conservative. Sensitivity analyses were performed to investigate the importance ofassumptions and support a decision on whether to pursue further analysis and/or plant changes.
The followingsummarizes the base case result (see Section 3) and sensitivities to the base case (changes in failure probability shown in base case):
Case Base Case PMP 1E-04 to 1E-05 LOP (0.1 to 0.5)
CUL (0.5 to 0.1)
DOOR (1 to 0.1)
EDG (0.5 to 0.1)
(/ r) 6.0E47 6.0E48 3.0E46 1.2E-07 6.4E-08 1.2E-07 Discussion Extreme flood levels (>1'7" above grade) may be less likely than presented herc, as indicated by values used in a study for another plant ef. 6.
The analysis is very sensitive to the conditional probability of AC power loss.
Site flood levels and impacts are sensitive to culvert blockage assum tions.
Analysis and/or procedures tliat ensure tliat flooding docs not impact EDGs can si nificantlv reduce risk.
More detailed analysis ofcircuit impacts due to PMP could significantly increase the reliabilitvofEDG recoverv.
Based on the results presented here, the CDF associated with PMP floods is less that 1.0E-06/yr.
1 of5
0
~
NMP1 EDG Rooin External Flood Risk 3.0 Analysis I
A simplified analysis ofthe 1E-4/yr PMP event is provided to allow a more quantitative assessment ofcore damage frequency potential as well as assess the sensitivity of assumptions and our state ofknowledge. The event tree below is used to represent key aspects and uncertainties ofthe event and show the frequency (freq) ofeach scenario (0).
Then, based on the impact ofeach scenario (Impact), a conditional core damage probability (CCDP) is calculated from the NMP1 PRA(Reference 5). This is documented below and in Attachment
- o. io 0.50 0.50 1.00 1.00 0.50 0.50 1
4.5E-05 261-3, scram 2
O.OE400
<261-7, scram 3
2.3E-05 261-9. scram 4
S.OE-06 261-3. 1osp 6
O.OE+00
<261-7. iosp 7
2.5E-06 261-9,1osp 8
2.5E-06 261-9, losp'EDGs Total CDF 1,1E-07 1.1E-07 1.1E-07 1.8E-06 4.0E-04 4.0E-04 4.0E-04 2.4E-01 5.0E-12 O.OE+00 2.5E-12
- 4. IE-11 2.0E-09
'.OE+00 1.0E-09 6.0E-07 6.0E-07 PMP = probable n>aximum precipitation coincident with historical maximum lake level LOP = no loss ofoffsite AC power during PMP CUL = culvert blockage is-25% or sufficiently low to prevent EDG impact DOOR = EDG door is closed and/or procedures ensure no EDG impact EDG = EDG is recovered The above event tree and analysis represents our judgment and present state of knowledge regarding CDF risk. The results ofseveral sensitivities are provided in Section 2 above.
Each ofthe above event tree top events is described below with regard to the above scenarios, top event probability, and how they impact the plant and the PRA calculation ofCCDP.
PMP -
robable maximum reci itation coincident with historical maximum lake level This is the PMP initiating event defined as those meteorological conditions necessary to cause external flooding at the emergency diesel'generator doors that exceed El 261-7. El 261-9 was identified as the PMP flood level in the IPEEE, but subsequent modifications and field inspections indicated that electrical connections that impact the ability ofthe diesel to start and run are located on terminal strips located at approximately El 261-7 to 261-8. Thus, for this analysis, PMP is the annual frequency ofexceeding El 261-7.
The PMP frequency used here is 1.0E-04/yr. This frequency was also cited in the original RAI response (Reference 3) and is believed to be conservative.
2of5
~m
NMP1 EDG Room External Flood Risk
'OP no loss ofoffsite AC ower durin PMP The failure branch is the conditional probability that offsite AC power is lost during the PMP. In the PRA, this probability is <1E-3 over 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> given that offsite power was not the initiating event. During the PMP, this probability is higher. A 90% probability of success is provided as the base case.
~
LOP success significantly improves the probability ofsuccessful plant operation or shutdown. The analysis conservatively assumes that the plant is tripped or undergoing a shutdown transient (SCRAM initiating event in PRA). With normal AC power available, failure ofthe EDGs is less important as shown in Sequence 4.
~
LOP failure balance ofplant (main condenser and feedwater) is lost. EDGs and reliefvalves are challenged. The analysis (PRA analysis ofCCDP) conservatively assumes a LOSP initiating event with no recovery. As shown, these sequences are more important.
CUL culvert blocka e is -25% or sufficientl low to revent EDG im act Site flood levels due to the PMP event are sensitive to culvert blockage assumptions.
Therefore, this top event is included to show the sensitivity ofthese assumptions.
Success indicates that blockage foes not cause PMP flood level to reach El 261-7. For example, 25% blockage leads to f561-3 PMP flood level. Failure at this top event indicates
- 74lc, A
blockage is sufficient to ensure that PMP flood level reaches El 261-7. The 261-9 flood level is based on 50% blockage. Storm drains are conservatively assumed plugged for both cases ofculvert plugging. A 50% probability ofsuccess is provided as the base case.
This is judged to be conservative.
DOOR EDG door is closed and/or rocedures ensure no EDG im act Ensuring that the EDG rollup doors are closed or preventing in leakage can mitigate the whole question ofPMP impact on EDGs, Therefore, this top event is included to show the sensitivity ofensuring success (e.g., detailed analysis and/or plant changes) at this top event. Success indicates that flood levels remain below El 261-7. Failure indicates that flood levels exceed El 261-7.
There are leakage paths around the diesel foundation to El 250. It is possible that leakage through this leakage path and around the rollup door to El 250 would prevent a El 261-7 flood level. The PMP event is not a long duration event; thus timing and duration of leakage through the door versus leakage to El 250 is important. Since this analysis has not been performed, a 0% probability ofsuccess (guaranteed failure) is provided as the base case. Also, there is no procedural guidance that requires the doors to be closed during the event.
A sensitivity case is described in section 2 relative to the potential decrease in CDF.
3 of5
t t
NMP1 EDG Room External Flood Risk
'DG-EDG is recovered Given that EDGs fail due to the flood, there is some probability that one can be recove'red in time to prevent core damage. The time available depends on subsequent failures in the PRA. An emergency condenser success path (no dependency on AC power) allows several hours to recover; the PMP event has come and gone. A conservative 50'lo chance ofsuccess is provided as the base case.
Failure at top event EDG is treated as an irrecoverable failure ofboth EDGs in the PRA calculations ofCCDP.
4.0
'eferences 1.
Francisco et al., Nine Mile Point Nuclear Station Unit 1 Individual Plant Examination for External Events (SAS-TR-96-001), August, 1996 2.
NRC Letter dated March 1998 RAI relative to NMP1 IPEEE 3.
NMPC Letter dated May 18, 1998 (NMP1L 1318) - Response to NRC RAI 4.
NRC Letter dated March 1999 RAI relative to NMP1 IPEEE 5.
Kirchner et al., Nine Mile Point Nuclear Station Unit 1 Individual Plant Examination,
. Rev. 0, Niagara Mohawk Power Corporation, July 1993 6.
Haddam Neck Plant, Individual Plant Examination for External Events (HNP IPEEE),
Pages 5-13 to 5-14, December, 1994 by Northeast Utilities Services Company.
4of5
t t
NMP1 EDG Room External Flood Risk Attachment 1 RISKMANPRA Calculations Case 1 SCRAM Initiator CCDP for sequences 1 through 3 is based on SCRAM initiator set at 1.0/year and quantifying CDF with a 1E-12 cutoff Result; CCDP = 1.1E-07 Case 2 SCRAM 1.0/ ear Initiator and EDGs Failed CCDP for sequence 4 is based on SCRAM initiator set at 1.0/year and the following model changes; CDF quantified with a 1E-12 cutoff:
Top event A2 in SUP1 event tree set to failure (A2F) ifnormal AC fails (KA=F)
Top event A3 in SUP1 event tree set to failure (A3F) ifnormal AC fails (KB=F)
Top event EDG in SBO tree set to failure (EDGF) to ensure no recovery Result; CCDP = 1.SE-06 Case 3 LOSP 1.0/ ear Initiator CCDP for sequences 5 through 7 is based on LOSP initiator set at 1.0/year and the following model changes; CDF quantified with a 1E-12 cutoff:
Top event OGR in SUP1 event tree set to failure (OGRF) to ensure no recovery Top event OSP in SBO event tree set to failure (OSPF) to ensure no recovery Result; CCDP = 4.0E-04 Case 4 LOSP 1.0/ ear Initiator and EDGs Failed CCDP for sequence 8 is based on LOSP initiator set at 1.0/year and the following model changes; CDF quantified with a 1E-12 cutoff Top event OGR in SUP1 event tree set to failure (OGRF) to ensure no recovery Top event A2 in SUP1 event tree set to failure (A2F)
Top event A3 in SUP1 event tree set to failure (A3F)
Top event EDG in SBO event tree set to failure (EDGF) to ensure no recovery Top event OSP in SBO event tree set to failure (OSPF) to ensure no recovery Result; CCDP = 0.25 5of5
ENCLOSURE 2