ML17059C625
| ML17059C625 | |
| Person / Time | |
|---|---|
| Site: | Nine Mile Point  |
| Issue date: | 04/12/1999 |
| From: | Kachnik L, Kirchner B, Knoll A NIAGARA MOHAWK POWER CORP. |
| To: | |
| Shared Package | |
| ML17059C624 | List: |
| References | |
| ES99-75, SAS-99-20, SAS-99-20-R, SAS-99-20-R00, NUDOCS 9904220288 | |
| Download: ML17059C625 (18) | |
Text
ANALYSIS GROUP TECHNICALREPORT SAFETY AND AVAILABILITYASSESSMENT
'MP1 Tornado Missile Risk-Diesel Generator Rooms
. [SAS-99-20]
Revision 0 ES99-75 April7, 1999 PREPARED:
Leo Kachhik PREPARED:
Alex Knoll REVIEWED:
Jim oody ooJ REVIEWED:
Bob Ki chner iZ<) ( fired~ yy(Z/fl APPROVED:
Ted Kulczycky ~
(
9904220288 9904i2 PDR ADQCK 05000220 P
PDR Niagara 5)/X'>
Mohawk'f
TABLEOF CONTENTS 1.0 Background and Objectives 2.0 Approach 3.0 Results and Conclusions 4.0 Sensitivity Cases 5.0 Analysis Method 1 6.0 Analysis Method 2 7.0 References Attachment 1 RISKMANPRA Calculation
t NMP1 EDG Room Tornado Missile Risk 1.0 Background and Objectives The risk from high winds and tornadoes was assessed in the NMP1 IPEEE'. The risk of missiles was judged as bounded by the analysis ofwind; no quantitative analysis of missiles was provided. NRC has asked specifically that NMPC determine the contribution to core damage frequency (CDF) from tornado missiles hitting one or both EDG rooms.
The objectives ofthis analysis are as follows:
~
Provide a simplified analysis ofCDF due to tornado missiles to show that this risk is less than 1E-6/year as judged in the IPEEE.
~
Use the simplified analysis to support sensitivity analysis and decision-making with regard to further protecting the diesel generators from turbine missiles.
2.0 Approach Two different analyses are presented here. These support and compliment one another.
The first analysis, presented in section 5 is a simplified analysis based on engineering judgment and data from Reference 4 and the IPEEE. This method was used to support sensitivity studies. The second analysis is based on a data from a detailed TORMIS analysis ofUnit 2. This was performed as a "sanity check" for the first analysis.
This analysis excludes the impact of tornadoes on areas of the plant other than the EDG room doors. This analysis also excludes the impact ofeffects other than tornado missile strikes.
3.0 Results and Conclusions The base case analysis in Section 5 indicates that CDF is less than 1E-6/year due to tornado missiles. The total core damage is 3.5E-7/year.
In Section 6, another approach was taken to evaluate tornado missile risk utili'zing an NMP2 analysis. In this analysis, the results are similar.
Based on these analyses and the sensitivity cases in section 4, the risk oftornado missiles is judged acceptably small (<1E-6/year CDF).
4.0 Sensitivity Cases Sensitivity cases were run with the model in Section 5 as summarized below:
1.
Failure probability of top event Door was increased to 0.1 (versus 0.01), 0.5 (versus 0.1), and 0.9 (versus 0.5). The resulting CDF due to Missiles is 8.5E-7. This case shows that the conclusions are not sensitive to the probability of a missile hitting a door (e.g. CDF is on the order of 1E-06 or less).
1 of7
1
~ ~
NMP1 EDG Room Tornado O'Iissile Risk Failure probability ofEDGI and 2 were increased to 0.5 (versus 0.1), 0.9 (versus 0.3),
and 1.0 (versus 0.9). The resulting CDF due to Missiles is 8.8E-7. This case shows that the conclusions are not very sensitive to diesel fragility.
3.
Same as Case 1 above plus EDG2 was set to 1.0 when EDG1 fails. The resulting CDF due to Missiles is 2.0E-6. This case shows that the conclusions are sensitive to increasing the probability ofdoor hits combined with increasing the common cause failure probability ofthe diesels.
During a recent plant walk-down, trucks and barrels were observed in the area adjacent to the diesel rollup doors. Since these objects are parked here in preparation for a refueling outage, the applicable fraction oftime during the year must be considered as well as the impact on top events "Door" and "EDG" in order to evaluate the CDP associated with this condition. In our judgment, the sensitivity cases above envelope these temporary conditions.
5.0 Analysis Method 1
A simplified quantitative analysis oftornado missile risk associated with hitting the diesel generator rollup doors is provided below. The event tree below is used to represent key aspects and uncertainties ofthe event and show the frequency (freq) ofeach scenario (4).
Then, based on the impact ofeach scenario (Impact), a conditional core damage probability (CCDP) is calculated from the NMP1 PRA'. This is documented below and in Attachment
Tornado S eed Door EDG1 EDG2
¹ fre Im act CCDP CDF 0.001 0.30 0.017 0.003 0.01 O. IO 0.50 0.10 0.30 0.90 O. lo 0, lo 0.30 0.30 0.90 0.90 I
2.97E-04 2
2.43E-06 3
2,70E-07 2.70E-07 5
3.00E-08 6
1.53E-OS 7
8.33E-07 8
3.57E-07 9
3.57E-07 io 1.53E-07 I I I.SOE-06 12 I.SOE-08 13 1.35E-07 14 1.35E-07 i 5 1.22E-06 losp losp losp. cdg" losp, edg I losp, edg I&2 losp losp losp, edg2 losp. edg I losp, edg 1 &2 losp losp losp, cdg2 losp. cdg I los, edlt I&2 Total CDF 4.00E-04 4.00E-04 7.88E-03 6.90E-03
~ 40E-OI 4.00E-04 4.00E-04 7.88E-03 6.90E-03 2.40E-OI 4.00E-04 4.00E-04 7.88E-03 6.90E-03 2.40E-OI 1.19E-07 9.72E-10 2.13E-09 1.86E-09 7.20E-09 6.12E-09 3.33E-IO 2.81E-09 2.46E-09 3.67E-08 6.00E-IO 6.00E-12 1.06E-09 9.32E-10 2.92E-07 4.74E-07 Total CDF Missiles 3.47E-07 Tornado = tornado frequency Speed = severity of tornado by speed 2of7
NMP1 EDG Room Tornado Missile Risk Door = probability tltat tornado & n>issile(s) does not impact EDG door area EDG1 = probability tliat missile(s) does not disable EDGl02 EDG2 = probability that missile(s) docs not disable EDGl03 The above event tree and analysis represents our judgment and present state of knowledge regarding CDF risk. Each event tree top event is described below with regard to scenarios, top event probability, and how they impact the plant and the PRA calculation of CCDP.
Tornado tornado fre uenc The frequency of a tornado in a 1' 1'rea in which NMP1 is located is 1E-3/year according to Rutch
. The above model assumes irrecoverable loss of normal AC power (LOSP impact) for this event, which is conservative because every tornado occurrence
'oes not cause loss ofoffsite power. Also, not all LOSP events are unrecoverable.
S eed severit oftor'nado b s eed There are three branches in this simplified analysis to represent tornado severity as summarized below:
Top branch is the conditional probability that tornado speed is between 80 and 150 MPH. This represents relatively smaller tornadoes.
~
Middle branch is the conditional probability that tornado severity is such that speed is between 150 and 190 MPH. This represents a more severe tornado.
~
Lower branch is the conditional probability that tornado severity is such that speed is greater than 190 MPH. This event is a very severe tornado.
The conditional probability for each branch is based on the following data from Rutch:
Wind S eed MPH 80 150 190 Probabilit ofExceedance 0.32 2.0E-2 3.0E-3 This simplified set ofbranches are set up to obtain a more reasonable probability ofthe tornado effecting the target (EDG rollup doors) versus the size or severity ofthe tornado.
The probability oftop events "Door" and "EDG1 & EDG2" below are dependent on tornado severity.
Door robabilit that tornado and missile s does not im act EDG door area The failure branch represents the conditional probability that the tornado and its missiles impact the EDG rollup door area. This is a localized area between Unit 1 and Unit 2 structures. The more severe and larger the tornado, the more likely it is to effect the EDG rollup door area on site. Each of the three branches in the event tree are summarizes below:
3 of7
a
NMP1 EDG Room Tornado Missile Risk
~
The top "Speed" Branch: The likelihood of a smaller tornado or its missiles moving between the two Units (a localized area protected by both Unit 1 and Unit 2 structures) is unlikely. The likelihood ofreaching this localized area is small. The probability value used in the event tree is based on judgment.
~
The middle "Speed" Branch: The likelihood ofa more severe tornado or its missiles effecting the area between the two Units is judged more likely. The probability value used in the event tree is an order of magnitude higher and based on judgment.
~
The lower "Speed" Branch: The likelihood of a severe tornado or its missiles effecting the area between the two Units is judged even more likely since such a severe event could engulf the site. The probability value used in the event tree is based on judgment.
This simplified analysis is not based on a detailed analysis oftornadoes and missiles; it is based on judgment to determine an order of magnitude importance ofthe scenarios.
Overall, it is judged reasonable or conservative.
EDG1 2 EDG2 robabilit that missile does not disable EDG The failure branch represents the probability that missile(s) penetrate and fail the EDG or damage the door preventing it from opening (door jams and diesel loses ventilation).
Each EDG rollup door is treated as independent random events in the analysis, given the preceding conditions in the event tree.
The rollup doors are closed during a tornado per procedure
. The rollup doors are constructed ofheavy steel. In the EDG room, there is a steel barrier just inside the rollup door that protects the lower 4 feet ofthe rollup door. This structure provides protection that appears capable ofpreventing heavy objects from getting into the room. The more severe and larger the tornado, the more likely it is to generate significant missile(s) that could penetrate or damage the rollup door. Damage to the roll-up door could prevent adequate ventilation to the room. The room could exceed the safe operating temperature limitfor electrical equipment. This is difficultto analyze because ifthe door is damaged enough to jam it, preventing opening, it could also be buckled sufficiently to allow adequate ventilation. The likelihood ofgetting through the door and damaging a diesel is judged unlikely, but again dependent on tornado severity. The EDG fragilityused in the three branches ofthe event tree is based on judgment. For the lower "Speed" branch (most severe tornado), failure is essentially guaranteed (0.9). The other branches are reduced in probability by a factor ofthree, with 0.1 as the lowest value for the least severe event.
This simplified analysis is not based on a detailed analysis oftornadoes, missiles, and the capacity ofthe rollup door; it is based on judgment to determine an order of magnitude importance ofthe scenarios. Overall, it is judged reasonable or conservative.
Failure at this top event is treated as an irrecoverable failure ofboth EDGs in the PRA calculations ofCCDP.
4of7
A
6.0 'nalysis Method 2 The approach used here is to evaluate diesel room roll-up doors by analogy to an analysis performed for the NMP2 Reactor Building. This is also similar to the analysis performed for the IPEEE, which was never formally documented. The results are summarized in the table below.
2 3
7 8
9 10 11 4
5 6
Tornado Tornado Tornado Occurrence Intensity Speed Rate (MPH)
(/s m/ r)
Path Area (s
m)
Tornado Prob of Cond Prob Prob Strike Missile Missile
(/ r)
Strike Strike 4.00E-03 3.92E-07 6.00E-03 9.49E47 1.20E-02 1.32E46 7.00E-03 3.15E-07 3 40E-02 3.89E-07 2.46E-O I 3.24E-07 3.69E-06 FI 73-103 1.36E-04 0.72 9.79E-05 F2 103-135 8.50E-05 1.86 1.58E-04 F3 135-168 2.95E-05 3.74 I. IOE-04 F4 168-209 6.23E-06 7.22 4.50E-05 F5 209-277 8.48E-07 13.48 1.14E-05 F6 277-360 6.24E-08 21.12 1.32E-06 Total 4.24E-04 Cond Prob EDG Im act O.l 0.2 0.3 0.5 0.8 I
9.79E49 4.74E48 9.93E48 3.94E-08 7.77E-08 8.11E-08 3.55E47 3.92E-08 0.25 1.90E-07 0.25 3.97E-07 0.25 1.57E-07 0.25
(/ r)
This method contains some conservatism and some non-conservatism as follows:
~
The target area used for the Reactor Building is 13,423 ft. This is conservative relative to the roll-up doors because these doors have a smaller area.
~
The number of missiles in the Reactor Building analysis was judged to be conservative compared to the number available to strike the diesel roll-up doors.
~
The Reactor building target area is in the upper portion ofthe structure (60 feet above the ground). This is non-conservative compared to the roll-up doors. A larger percentage of missiles would be expected to reach the doors.
~
The. influence ofintervening structures is ignored. This is conservative for the roll-up doors because the roll-up doors are shielded to a large degree by other structures.
~
The analysis assumes that the same missile(s) impacts both roll-up doors simultaneously and fails both diesels. This is conservative.
Columns 1 through 7 are from.Reference 6. The probabilit'y of a tornado strike is conservatively derived in this reference. The NMP2 USAR reports a probability of 3.75E-05/year of a tornado striking NMP2 (this should also apply to NMP1).
The table presents conditional probabilities and resulting frequencies for various tornado intensities. Conditional probabilities of diesel failure are relatively high and are based on uncertainty associated with diesel room cooling ifthe roll-up doors are damaged (jammed).
The total CDF for this method is 3.55E-07/yr.
5 of 7
4
NMP1 EDG Room Tornado Missile ask 0 'eferences 1.
Francisco et al., Nine MilePoint Nuclear Station Unit 1 Individual Plant Examination for External Events (SAS-TR-96-001), August, 1996 2.
NRC Letter dated May 18 RAI relative to NMP1 IPEEE (File NMP1L 1318).
3.
Kirchner et al., Nine Mile Point Nuclear Station Unit 1 Individual Plant Examination, Rev. 0, Niagara Mohawk Power Corporation, July 1993 4.
Rutch, Christine, et.al., "Tornado Risk Analysis" July 1992, Pages 134-139 in Plant/Operations Progress (Vol. 11, No 3) 5.
NI-SOP-10, Revision 5, High Winds, July 1998
Tornado Missile Simulation and Design Methodology and Computer Manual Vol.
1 Simulation Methodology, Design Applications and TORMIS Computer Code, Electric Power Research Institute (EPRI), Report NP-2005, August, 1981 6of7
NMPl EDG Room Tornado Missile kisk Attachment 1 RISKMANPRA Calculations Case I - LOSP 1.0/year Initiator CCDP for sequences I, 2, 6, 7, 11, and 12 are based on LOSP initiator set at 1.0/year and the following model changes; CDF quantified with a IE-12 cutoff:
Top event OGR in SUP I 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 2 LOSP 1.0/year Initiator and EDG102 failed CCDP for sequences 4, 9, and 14 are based on LOSP initiator set at 1.0/year and the following model changes; CDF quantified with a IE-12 cutoff:
Top event OGR in SUP I event tree set to failure (OGRF) to ensure no recovery Top event A2 in SUP I event tree set to failure (A2F)
Top event EDG rules in SBO event tree set to ensure only I EDG is recoverable Added EDGF ifD2=F Deleted EDGI. EDG2, EDG4, and EDG8 (these allow lof2 EDG recovery)
Revise EDGA, EDGB, EDGC, and EDGD as follows EDGA CD I EDGB CD2 EDGC CD4 EDGD CD8 Top event OSP in SBO event tree set to failure (OSPF) to ensure no recovery Result; CCDP = 6.0E-03 Case 3 LOSP 1.0/year Initiator and EDG103 failed CCDP for sequences 3, 8, and 13 are based on LOSP initiator set at 1.0/year and the following model changes; CDF quantified with a IE-12 cutoff:
Top event OGR in SUP I event tree set to failure (OGRF) to ensure no recovery Top event A3 in SUP I event tree set to failure (A3F)
Top event EDG rules in SBO event tree set to ensure only I EDG is recoverable Added EDGF ifD I=F Deleted EDGI, EDG2, EDG4, and EDG8 (these allow lof2 EDG recovery)
Revise EDGA, EDGB, EDGC, and EDGD as follows EDGA CD I EDGB CD2 EDGC CD4 EDGD CDS Top event OSP in SBO event tree set to failure (OSPF) to ensure no recovery Result; CCDP = 7.9E-03 Case 4 LOSP 1.0/ ear Initiator and EDGs Failed CCDP for sequences 5, 10, and 15 are based on LOSP initiator set at 1.0/year and the following model changes; CDF quantified with a IE-12 cutoff.
Top event OGR in SUP I event tree set to failure (OGRF) to ensure no recovery Top event A2 in SUP I event tree set to failure (A2F)
Top event A3 in SVPI 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 = 0.24 7 of7
1 1