Making an optimal inspection-free maintenance decision for a coherent system subject to hidden malfunctions

A system may experience hidden malfunctions as its components degrade over time. To ensure system performance during operation, it is essential to conduct timely maintenance based on the health of the components. However, when monitoring the components’ states is technically or economically infeasible, the operator must be more proactive in making an optimal maintenance policy. Unfortunately, state-of-the-art maintenance policies (e.g., age-based and block-based) often lead to suboptimal decisions under such circumstances. This study investigates proactive inspection-free maintenance policy optimization for a coherent system subject to multi-level hidden malfunctions. We prove that the globally optimal maintenance policy always takes a periodic form at the component level when the duration of a maintenance action is negligible. When the maintenance duration is non-negligible, we develop two algorithms that employ explainable procedures for merging maintenance actions toward an enhanced maintenance policy. We prove that under most circumstances, the proposed maintenance policy is periodic at the system level and aperiodic at the component level. Moreover, sufficient conditions extending the proposed methodologies to scenarios with state-specific maintenance costs for components, stochastic dependency among components, and dynamic loading levels are explored. A case study on pipelines illustrates that both the proposed maintenance policies and solution algorithms result in superior performance over the benchmark alternatives.