Development of probabilistic models to estimate fire-induced cable damage at nuclear power plants

Numerous Probabilistic Risk Assessments (PRAs) have shown that fire is a major contributor to Nuclear Power Plant (NPP) risks. However, prediction and estimation of the likelihood of fire-induced damage to electrical cables and circuits and their potential effects on the safety of the NPPs are still a practical challenge, particularly because of the lack of physics-based models with which to perform consistent and objective assessments. This paper contains a discussion of two models - the heat transfer and the IR "K-factor" models - to estimate the likelihood of fire-induced cable damage given a specified fire profile. The results of this research will help to (1) develop a consistent framework to estimate the likelihood of fire-induced cable failure modes, and (2) develop some guidance to evaluate and/or reduce the risks associated with these failure modes in existing and new NPPs. The models are developed (i.e.. their parameters are estimated) based on the test data from various fire damage tests sponsored by the nuclear industry and the U.S. Nuclear Regulatory Commission (NRC). Among the models evaluated, the physics-based heat transfer model is promising because it takes into account the properties and characteristics of the cables and cable materials and the characteristics of the thermal insult. This model can be used to estimate the probability of cable damage (P-CD) under different thermal conditions. (C) 2009 Elsevier B.V. All rights reserved.