Importance measures for non-coherent-system analysis

Abstract

Component importance analysis is a key part of the system reliability quantification process. It enables the weakest areas of a system to be identified and indicates modifications, which will improve the system reliability. Although a wide range of importance measures have been developed, the majority of these measures are strictly for coherent system analysis. Noncoherent systems can occur and accurate importance analysis is essential. This paper extends four commonly used measures of importance, using the noncoherent extension of Birnbaum's measure of component reliability importance. Since both component failure and repair can contribute to system failure in a noncoherent system, both of these influences need to be considered. This paper highlights that it is crucial to choose appropriate measures to analyze component importance. First the aims of the analysis must be outlined and then the roles that component failures and repairs can play in system state deterioration can be considered. For example, the failure/repair of components in safety systems can play only a passive role in system failure, since it is usually inactive, hence measures that consider initiator importance are not appropriate to analyze the importance of these components. Measures of importance must be chosen carefully to ensure analysis is meaningful and useful conclusions can be drawn.