A New Concept for Assessing System Reliability Based on Probability Decomposition Method

Abstract

In fault tree analysis, the system failure probability and the component importance measures cannot totally include the contribution of all the component existing states to system reliability. It is for this reason that an ‘equivalent’ failure probability concept is proposed. First, the system existing states are analyzed by probability decomposition method. Then Markov chain method and the expectation theory are used to calculate the expected working number resulting in system failure. And the system equivalent failure probability is finally attained. Analysis shows that: (1) equivalent failure probability not only includes the contribution of critical states of component to system reliability, but also the non-critical states of component are considered; and (2) it may provide a thorough assessment of system reliability and is useful for reliability design.