ML20045J176
| ML20045J176 | |
| Person / Time | |
|---|---|
| Site: | 05200001 |
| Issue date: | 06/23/1993 |
| From: | Raftery B GENERAL ELECTRIC CO. |
| To: | Kelly G Office of Nuclear Reactor Regulation |
| Shared Package | |
| ML20045J155 | List: |
| References | |
| NUDOCS 9307230135 | |
| Download: ML20045J176 (1) | |
Text
Juli 25193]O7:fSti; P.11 Pxttit* brand tax transmittal memo 7671 ' e st reees * /
r "GLEA/ KEL.W
'% B RAf7% W V
C*y g g c
- G E A/ 2 -
- Wg)9xg-4/44
-(
+
Not) 504.-2140 ' WlA) 9AS-/4/N j
6 [h t
MEhtORANDUM-(-
~
1 rw To:
Glen Kelly
(
From: Bob Raftery 3
This is in response to your questions regarding the generation of minimal cut sets in 4
performing the ABWR fire risk analysis and justification for why GE did not use importance measures to analyze what were the most important features from the standpoint of fire risk.
i The EPRI FIVE methodolo8y upon which GE based its. fire risk evcluation is' a conservative screening analysis technique. In performing the fire risk evaluation, GE incorporated additional conservatisms and then demonstrated that this very conservative application of the EPRI FIVE methodology would still satisfy the CDF screening criterion of 1.0E-06 per year. Although the FIVE methodology as modified and applied by.GE l
used CAFTA, and cut sets were generated for each of the five fire scenarios evaluated, these were not combined in a composite file of cut sets to produce an overall fire "CDF",
nor were importance measures developed on such a basist This was not done since the assumptions used to obtain the numerical values were exceedingly conservative and were
~
dermed so that they would be bounding values for divisional,' control room, and ' turbine l
building fires. Therefore, summing the fire scenario frequencies does not produce a' fire -
"CDF" which can be meaningfully compared to the ABWR internal event core' damage frequency. Similarly, importance measures developed from such a_ compilation would be highly questionable since~ large and-indeterminate biases could be introduced by the bounding nature of the analysis.
Therefore, capabilities and features important to safety were identified from the fire risk -
analysis as those required to meet the FIVE screening criterion and which were not identified as important from evaluation of the Level 1 internal event PRA. These included 3
the capability to operate RCIC from outside the control room, providing control for four SRVs at the remote shutdown panel, and isolation of the three safety divisions from each -
t other (including necessary support systems) by three hour rated fire barriers.
R. P. Raftery l
9307230135 930630 ADOCK0520g1 PDR
-