Abstract: An un-empirical model for probability quantitative analysis of common cause failures is demonstrated that utilizes Nataf transformation approach based on safety margin variables of components in a common cause group. The joint failure probability is figured out by multiplying the products of the individual integration in an independent, normal space. In view of the fact that the data of system low-order common cause failure are known while the system load distribution and components strength distribution are unknown, a practical method for risk prediction of load-dependency systems is put forward. The proposed model and method embody the inherent mechanism of common cause failures and overcome the limitation of traditional model which excessively relies on historical data or subjective experience while neglects the occurrence and action rule of common cause failures. The proposed method provides complicated systems with a new way to risk data analysis and failure probability prediction.

Key words: Common cause failure, Load dependence, Nataf transformation, Risk prediction, System reliability