ML071020234

Letter from John Lambright, Sandia National Laboratories, to Demetrios Basdekas Re Generic Issue 148, Smoke Control and Manual Fire Fighting Effectiveness.

ML071020234
Person / Time
Site:	Oconee
Issue date:	03/04/1992
From:	Lambright J Sandia
To:	Basdekas D Office of Nuclear Reactor Regulation
References
FOIA/PA-2007-0140
Download: ML071020234 (5)

Text

0 4 1-T)

Sandia National Laboratories tLH March 4, 1992 Mr. Demetrios Basdekas U.S. Nuclear Regulatory Commission 5650 Nicholson Lane Rockville, MD 20852 Re: Generic Issue 148, "Smoke Control and Manual Fire Fighting Effectiveness"

Dear Demetrios:

In my letter of February 5, 1992, I outlined my thoughts concerning the possibility of incorporation of work on Generic Issue 148, "Smoke Control and Manual Fire Fighting Effectiveness" into work in progress on Generic Issue 57, "Effects of Fire Protection System Activation on Safety-Related Equipment." Subsequent to your receipt of that letter, in d:,cussions we held in Rockville, it was determined that while there are some commonalties between these issues, Generic Issue 148 would be better served by treatment in separate analysis.

This letter provides some insights with regard to Generic Issue 148 following my review of the Draft Generic Issue Prioritization Analysis performed by P. M. Daling of Pacific Northwest Laboratory in August 1991. The prioritization analysis looked at a treatment of an Oconee Cable Shaft fire In assessing Generic Issue 148 risk. As we discussed, I found some problems with that analysis. My observations and comments follow:

In the Pacific Northwest Laboratory calculations for the Oconee Cable Shaft area:

4VCM = aux fa

• fs

= (2.3E-2/yr) 9 (0.16) o (0.026) = 9.6E-5/yr  %

Ocm = Frequency of core damage I where:

950207G1,65 92030-

,' S

,.1tK= Fire frequency in the auxiliary building f.= Fraction of fires that occur in equipment room fs Fraction of fires that are large and in the cable shaft area However:

• Manual fire suppression is not taken into account.

" COMPBRN I predicts only a large fire will cause damage, and that the damage will occur at 6.7 minutes from initiation.

• Estimated Fire PRA mean-time-to-fire-brigade response is 14 minutes, based on admittedly limited data (8 fires in the 1970's).

In the Pacific Northwest Laboratory analysis, a value of 11 minutes for fire brigade response was used, having been obtained from the Fire Scoping Study, NUREG-5088. However, in the selection of that 11 minute yvae, the following deficiencies were noted:

• The LaSalle plant layout (source of the data) is entirely different from that at Oconee.

• Fire brigade comparisons between plants were not accomplished, and thus differences are unknown.

• Area specifics such as smoke control systems, area congestion, etc., were not addressed.

• The wrong table in NUREG-5088 was selected, the number 9 minutes should have been used by their methodology.

Because of these deficiencies, I consider the claim that this calculation bounds the manual fire fighting issue to be suspect. Where specific analyses of fire brigade performance have been performed (LaSalle, N-reactor, K-reactor, and to be done at Grand Gulf) a more representative time range for suppression is obtained. Attached is a copy of Figure 4.4-1 "Fire Suppression Time Cumulative Distribution," taken from NUREG-5088. The wide variation in suppression times can be seen. The following ranges for suppression times are available:

• 5 to 60 minutes based on expert (fire protection engineer) analysis of specific plants.

• 2 minutes to 5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> based on LER data.

0' .,,3

.3-

. Using a range for suppression time of 5 to 60 minutes, and the computational approach used by Pacific Northwest Laboratory:

Qcsdtg exp (-g/ts)

Where Qcs,'tg = Conditional probability of fire damage, given a fire under trays tg = Time required for damage rs = Mean suppression time For the case rs = 5 minutes; QMg exp (-6.7/5) = 0.74 For the case Ts = 60 minutes; Qcstg = exp (-6.7/60) = 0.11

. Therefore the probability of suppression varies from 74% to 11%. Substitution of this data into the fire sequence equations used by Pacific Northwest Laboratory:

Fire [1] Best Case 5.OE-7/yr Worst Case 1.7E-6/vr A Core Damage Frequency = 1.2E-6/yr Fire [21 Best Case 2.2E-6/yr Worst Case 7.6E-6/yr A Core Damage Frequency . 5.413-6/yr Fire [31 Best Case 1.5E-6/yr Worst Case 5.1E-6/yr A Core Damage Frequency = 3.61E-6/yr And then Total A Core Damage Frequency = 1.0E-5/yr This number corresponds to the total fire risk predicted by NUREG-I150 fire analysis for Peach Bottom and Surry. However, Total A Core Damage Frequency could

• actually be much larger. Consider acase in which COMPBRN predicted both large and small fires could cause damage then instead of fs for the cable shaft being 0.026,

. it could be as high as 0.26. This would increase the Total A Core Damage Frequency by an order of magnitude to 1.013-4/yr.

In another example, COMPBRN might predict that time to damage cables was 20 minutes instead of 6.7 minutes. Using the same method of calculation as for the first example, this results in a range of fire suppression probabilities for 98% to 28%

and a Total A Core Damage Frequency = 1.IE-5/yr.

It is concluded that if time-to-damage estimates are increased, then Total A Core Damage Frequency from manual fire brigade effectiveness also increases because as shown, an increase of time-to-damage from 6.7 minutes to 20 minutes results in an increase in Total A Core Damage Frequency of 7%.

These calculations represent only one fire area in one plant. There is reason to believe that these manual fire fighting effects could be correlated for all critical areas for any given plant. Therefore Total A Core Damage Frequency could be higher than predicted in this example.

As I recommended in my letter of February 5, 1992, Generic Issue 148 requires additional detailed analysis, as the issues have the potential to lead to core damage frequencies on the order of 104 to 10-5 per reactor year. Sandia National Laboratories is ready to conduct the analysis required for resolution of this issue.

Sincerely yours, Joh Lambright

Attachment:

Figure 4.1-1, NUREG-5088 Copy to:

6419 S. Nowlen 6419 M. P. Bohn

80 4.)

700 60 a-50 L.

a-

~40 30

320 E

3 U 10 0

1 2 5 10 20 50 100 200 500 Time (min)

Figure 4.1-1 Fire Suppression Time Cumulative Distribution