Nuclear Power Plant Fire Ignition Frequency and Non-Suppression Probability Estimation Using the Updated Fire Events Database: United States Fire Event Experience Through 2009(NUREG-2169, EPRI 3002002936)

On this page:

Download complete document

Publication Information

Date Published: January 2015

Electric Power Research Institute (EPRI)
3420 Hillview Avenue
Palo Alto, CA 94304-1338

EPRI Project Manager
A. Lindeman

U.S. Nuclear Regulatory Commission
Office of Nuclear Regulatory Research (RES)
Washington, DC 20555-0001

U.S. NRC-RES Project Manager
N. Melly

Availability Notice


This report documents the development of updated fire ignition frequencies (FIFs) and non-suppression probability (NSP) estimates as potential improvements for nuclear power plant fire (NPP) fire probabilistic risk assessment (FPRA) applications. This research follows prior Electric Power Research Institute (EPRI) research aimed at providing an updated methodology to estimate FIFs and collect more recent fire event data.

Fire ignition frequencies and non-suppression probabilities were previously developed in the NUREG/CR-6850/EPRI 1011989 and revised in Supplement 1 to NUREG/CR-6850/EPRI 1019259. In this report, the FIF estimation benefits from an enhanced methodology and incorporates updated data from EPRI's updated Fire Events Database (FEDB). The report also updates low-power and shutdown (LPSD) FIFs from those published in NUREG/CR-7114. NSP estimates are calculated using the existing methodology and have been updated with new fire event experience.

The fire ignition frequencies and non-suppression probabilities published in NUREG/CR-6850 and Supplement 1 incorporate fire event experience through the year 2000. The research presented in this report incorporates U.S. NPP fire event experience through the year 2009. The fire event data are split and analyzed into three distinct periods—1968–1989, 1990–1999, and 2000–2009. The 1968–1989 data are used to develop a diffuse empirical prior to account for operating experience in that period as a starting point for the Bayesian update. Sparse bins (<2.5 fire events) use a 20-year update period, whereas medium and dense bins use only data from 2000–2009.

Approximately 400 of the events from 1990–2009 are of significance for FPRA applications. Included in the estimation of manual NSPs are 442 fire events from 1981 to 2009. The insights conclude that the data from 2000–2009 are the most complete and accurate for characterizing and estimating FIFs for FPRAs.

Page Last Reviewed/Updated Tuesday, March 09, 2021