A discretizing approach for time-dependent reliability modeling

Abstract

Reliability modeling of time-dependent systems has been the subject of many research development efforts. Even within the context of simple series or parallel systems, the determination of the mean time to failure at the system level can be complex and can exceed the available resources for analytical development and computation. The approach presented is suitable for application in the engineering design of systems for which the failure times of the individual components can be reasonably assumed to be normally distributed. Further research extensions include a more comprehensive analysis of the approach. In particular, the authors conduct a robustness analysis of the approach when departures from the normality assumption are encountered and apply the approach to complex networks.