Modeling and analysis of cluster of failures in redundant systems

Abstract

Redundancy is an important method to improve the reliability of repairable systems. In this paper, we consider clustering of failures in redundant systems due to parallel type of carryover effects, in which the event intensity of a recurrent event process is temporarily increased after event occurrences in other processes. Our main goal is to develop formal test procedures for the assessment of such effects in redundant systems with repairable components connected in parallel and subject to recurrent failures. We develop partial score tests for testing the presence of parallel carryover effects in recurrent event settings. Asymptotic properties of the test statistics are discussed analytically as well as through simulations. A simulated data set including failure times of diesel power generators operating in remote and isolated communities is analyzed to illustrate the methods developed.