History of Fire Events in the U.S. Commercial Nuclear Industry

Abstract

Over the past decade, interest in performance-based fire protection has increased within the nuclear industry. In support of this growing interest, in 1997 the Electric Power Research Institute (EPRI) developed a long-range plan to develop/improve data and tools needed to support Risk-Informed/Performance-Based fire protection. This plan calls for continued improvement in collection and use of information obtained from fire events at nuclear plants. The data collection process has the objectives of improving the insights gained from such data and reducing the uncertainty in fire risk and fire modeling methods in order to make them a more reliable basis for performance based fire protection programs. In keeping with these objectives, EPRI continues to collect, review and analyze fire events in support of the nuclear industry. EPRI collects these records in cooperation with the Nuclear Electric Insurance Limited (NEIL), by compiling public fire event reports and by direct solicitation of U.S. nuclear facilities. EPRI fire data collection project is based on the principle that the understanding of history is one of the cornerstones of improving fire protection technology and practice. Therefore, the goal has been to develop and maintain a comprehensive database of fire events with flexibility to support various aspects of fire protection engineering. With more than 1850 fire records over a period of three decades and 2400 reactor years, this is the most comprehensive database of nuclear power industry fire events in existence today. In general, the frequency of fires in the U.S. commercial nuclear industry remains constant. In few cases, e.g., transient fires and fires in BWR offgas/recombiner systems, where either increasing or decreasing trends are observed, these trends tend to slow after 1980. The key issues in improving quality of the data remain to be consistency of the recording and reporting of fire events and difficulties in collection of records. EPRI has made significant progress towards improving the quality of the fire events data through use of multiple collection methods as well as its review and verification. To date EPRI has used this data to develop a generic fire ignition frequency model for U.S. nuclear power industry (Ref. 1, 4 and 5) as well as to support other models in support of EPRI Fire Risk Methods such as a cable fire manual suppression model. EPRI will continue its effort to collect and analyze operating data to support risk informed/performance based fire safety engineering, including collection and analysis of impairment data for fire protection systems and features. This paper provides details on the collection and application of fire events to risk informed/performance based fire protection. The paper also provides valuable insights into improving both collection and use of fire events data.