A tool for evaluating repairable systems based on Generalized Renewal Processes

Abstract

Abstract Generalized Renewal Processes -GRP- have been in the kernel of modelling repairable systems. Via a virtual age function, they extend classical reliability engineering formalisms, such as renewal and Poisson processes. Via point estimation, GRP practitioners have evaluated intervention crews as well as forecasted the occurrence of undesirable events underlying the system. Assuming an interval estimation perspective, this paper introduces a hypothesis testing framework for GRP. Thus, it makes possible to measure the uncertainty when inferring the stage of the system (e.g. whether stable or deteriorating) and the quality of the interventions. The approach focus on the Weibull-based GRP, notably the main GRP found in the literature. The usefulness of the method is illustrated via real world cases involving offshore, windshield, and transformer facilities. Simulated cases are also studied.