Common cause failure parameter estimation from generic data under masked testing schemes

Abstract

This paper deals with the inconsistency problem when we estimate the basic parameter of common cause failures from generic data in which testing schemes are masked. We propose an approximate confidence interval for the basic parameters under mixed testing scheme including both non-staggered and staggered testing. We also propose a method to obtain the unbiased estimates for the basic parameters from the generic database composed of alpha factor estimates and the total failure probability of a component. The proportion of non-staggered tests is usually unknown, but it can be estimated by stabilizing the estimates for the total probability of any two specific components’ failure for various system sizes. The proposed method will fill the gap between the estimated frequency and the experienced frequency of common cause failures in industry. © 2015 ICIC International.