A parametric mixture-model for common-cause failure data [of nuclear power plants

Abstract

This paper introduces a statistical reliability model for common-cause failure data from the nuclear industry. To achieve target reliability, many components in power plants are placed in parallel systems. The benefits of redundancy can be negated if multiple component failures occur due to a common external event. To model the possibility of multiple failures, a mixture-model based on the binomial failure-rate model is derived using reasonable assumptions of multiple failure events at a nuclear power plant (NPP). In many applications, the original binomial failure-rate model fits failure data poorly, and the model has not typically been applied to probabilistic risk assessments in the nuclear industry. This mixture-model fits better. This paper presents a least-squares solution to the mixture-model parameters and the model fit is investigated. Methods developed here are motivated by, and illustrated with, discrete failure data collected from several US NPP since about 1980.