Replacement decisions for reliability centered maintenance

Abstract

We study reliability centered proactive maintenance of a complex dynamic system consisting of aging components. System reliability can be estimated from the interactions of the components. Maintenance of the system is realized by replacing components in any period. Our aim is to minimize total replacement cost in a discrete horizon such that system reliability never falls below a predetermined threshold value. We represent the problem with dynamic Bayesian networks (DBNs) and develop an algorithm for determining when and what to replace within the DBN framework. We prove that under the existence of a predetermined threshold, this algorithm assures optimum replacement times. Four approaches are proposed to choose the component to replace and are tested on a complex dynamic problem. © 2020 Elsevier B.V., All rights reserved.