ML18198A387
| ML18198A387 | |
| Person / Time | |
|---|---|
| Issue date: | 07/13/2018 |
| From: | Office of Nuclear Reactor Regulation |
| To: | |
| Wu A | |
| Shared Package | |
| ML18198A393 | List: |
| References | |
| FAQ 18-0018 | |
| Download: ML18198A387 (7) | |
Text
FAQ Number 18-0018 FAQ Revision 0 (Draft B)
FAQ Title Electrical Non-Suppression Probability (NSP)
Plant: Various Date:
Contact:
Beth Meade Phone: (508) 532-7139 Mark Schairer (508) 532-7317 Rob Cavedo (301) 938-0397 Email: bah@epm-inc.com mvs@epm-inc.com robert.cavedo@exeloncorp.com Distribution: (NEI Internal Use)
FPRA TF BWROG PWROG Purpose of FAQ:
This FAQ provides an update to the non-suppression probability (NSP) for electrical equipment based on Bayesian approach separating the data before Jan 1st, 2000 and after.
Relevant NRC document(s):
NUREG/CR-6850 NUREG/CR-6850 Supplement 1 (FAQ 08-0050)
NUREG 2169 Details:
NRC document needing interpretation (include document number and title, section, paragraph, and line numbers as applicable):
See list of relevant NRC documents Circumstances requiring interpretation or new guidance:
The non-suppression probability for electrical fires provided in NUREG 2169 Table 5-1 are considered overly conservative due to the equal treatment of fire data over a prolonged period. As a result, the risk associated with electrical fires may be artificially high.
Detail contentious points if licensee and NRC have not reached consensus on the facts and circumstances:
FAQ Number 18-0018 FAQ Revision 0 (Draft B)
FAQ Title Electrical Non-Suppression Probability (NSP)
Although recent guidance has sought to provide a better estimate of non-suppression probability, in the case of fire events over a prolonged period, the probability of non-suppression is believed to be overly conservative.
Potentially relevant existing FAQ numbers:
FAQ 08-0050, Manual Non-Suppression Probability Response Section:
Proposed resolution of FAQ and the basis for the proposal:
The electrical fire NSP is adjusted using a Gamma-Poisson Bayesian model using the data before Jan 1st, 2000 as a prior distribution. There have been numerous improvements to plant process controls that affect both the likelihood, control, and severity of fire scenarios. This includes smoking controls, foreign material exclusion, cutting and welding improvements, combustion controls, brigade training, etc. It is for these reason among others, that the conclusion in NUREG 2169 is that data from 2000 and later is the most applicable in fire analysis. The data complied from the NUREG 2169 is as follows:
Before 2000 After 2000 Combined Alpha 112 65 177 Beta 1185 628 1813 Mean 0.095 0.104 0.0976 5th% 0.0803 0.0833 0.0859 50th% 0.0942 0.1030 0.0974 95th% 0.1097 0.1255 0.1100 It seems reasonable to assume that the data before 2000 would have some influence on the data 2000 and beyond but certainly not more influence. It is useful to view the data prior to 2000 as to the degree it could affect data after 2000. For a lognormal distribution a range factor is a useful metric. Although not directly applicable to a gamma distribution, it is a good tool for measuring the degree of influence of the prior distribution on the posterior distribution. For a range factor of 1.5 between the 95% tile and median, the resultant NSP would be:
FAQ Number 18-0018 FAQ Revision 0 (Draft B)
FAQ Title Electrical Non-Suppression Probability (NSP)
Gamma/Poisson Weak Prior Bayesian Model Before 2000 After 2000 Combined Alpha 15 65 Alpha 80 Beta 159 628 Beta 787 Mean 0.095 0.104 Mean 0.102 5th% 0.058 0.083 5th% 0.084 25th% 0.077 0.095 25th% 0.094 50th% 0.092 0.103 50th% 0.101 75th% 0.110 0.112 75th% 0.109 95th% 0.138 0.125 95th% 0.121 RF 1.5 1.2 RF 1.2 This equates to an NSP of 0.102. With a range factor (RF) of 2.0, the NSP only changes to 0.103:
Gamma/Poisson Weak Prior Bayesian Model Before 2000 After 2000 Combined Alpha 5 65 Alpha 70 Beta 53 628 Beta 681 Mean 0.095 0.104 Mean 0.103 5th% 0.037 0.083 5th% 0.083 25th% 0.064 0.095 25th% 0.094 50th% 0.088 0.103 50th% 0.102 75th% 0.119 0.112 75th% 0.111 95th% 0.173 0.125 95th% 0.124 RF 2.0 1.2 RF 1.2 The final distribution is fairly insensitive to the RF of the data prior to 2000. The recommend electrical fire NSP parameters are based on a RF of 1.5. The recommend distribution parameters are:
Electrical Fire NSP Alpha 70 Beta 681 Mean 0.103 5th% 0.083 25th% 0.094 50th% 0.102 75th% 0.111 95th% 0.124
FAQ Number 18-0018 FAQ Revision 0 (Draft B)
FAQ Title Electrical Non-Suppression Probability (NSP)
If appropriate, provide proposed rewording of guidance for inclusion in the next Revision:
The following are proposed revisions to NUREG 2169:
FAQ Number 18-0018 FAQ Revision 0 (Draft B)
FAQ Title Electrical Non-Suppression Probability (NSP)
Table 5-1 Probability distribution for rate of fires suppressed per unit time, (Originally, Table P-2 from NUREG/CR-6850)
Number Total Rate of Fire Suppressed ()
of Events Duration Suppression Curve 5th 50th 95th in Curve (minutes) Mean Percent Percent Percent Electrical fires 70 681 0.103 0.083 0.102 0.124
Page 6 of 7 Page 7 of 7