Dynamic Repositioning of Condition-Based and Opportunistic Maintenance Resources

The physical condition of many assets degrades stochastically over time leading to increased expenditures or, in the absence of properly timed interventions, failures. Recent advances in sensor technology have increased the use of condition monitoring and condition-based maintenance (CBM) strategies, which prescribe interventions (e.g., inspection, repair, replacement) dynamically as a function of observed asset health parameters (e.g., vibration, sound, heat). In many CBM applications, the assets being maintained are geographically dispersed and the maintenance resources travel to, from and in between the assets’ locations. These applications yield complex decision spaces with inherent tradeoffs between the timing of maintenance interventions and the positioning/routing of the maintenance resources. We propose a stochastic dynamic programming approach to generate and analyze CBM policies performed by one or more mobile maintenance resources on a set of geographically distributed assets. Our research objective is to dynamically determine the optimal positioning of the maintenance resource(s) and the optimal timing of the interventions that they perform. These decisions are made as a function of the conditions of the assets and the current location of the resource(s) to minimize total expected costs, e.g., downtime, travel, and maintenance expenses. Our analysis will be both analytical and numerical in nature; involve both optimization and simulation; compare the performance of multiple heuristic policies using various metrics; and lay the foundation for a larger scale follow-on proposal to NSF