Component rearrangement and system replacement for a system with stochastic degradation processes

Abstract

Consider a system of multiple functionally-interchangeable components. These components degrade following non-homogeneous stochastic processes that are related to different workloads, usage rates, or operating environment. The unbalanced degradation levels of components affect the overall performance of the system. This paper studies a preventive maintenance policy based on component rearrangement, and the goal is to prolong the useful time of a system before its replacement or overhaul. In this model, the optimal component rearrangement is planned to execute at optimal time. The stochastic degradation models and reliability functions that incorporate the component rearrangement are built. Then a mixed binary nonlinear programming model is established to optimize the component rearrangement plan, rearrangement time, and system replacement time. The models are derived for each of the Wiener process, gamma process, and inverse Gaussian process, which are three most widely used stochastic degradation processes. The specified models are derived for k-out-of-n:F, series–parallel, and parallel–series systems. Furthermore, a hybrid approach is proposed that integrates permutation-based heuristic global search for solving a combinatorial subproblem and nonlinear optimization method for solving constrained continuous subproblems. The analytical results and numerical experiments demonstrate the efficiency of the models and solution approach and provide the insights of the component-rearrangement based maintenance policy.