Optimizing fixed-interval maintenance periods for mobile asset sub-systems operating in remote locations

Abstract

Fixed-interval planned maintenance is the predominant strategy for looking after mobile asset fleets such as cars, trains, and trucks. To understand the impact of maintenance interval selection on the fleet's performance it is necessary to mathematically link these planned maintenance tasks to the asset's subsequent reliability performance and the ability to complete planned journeys without incident. Failures, particularly those happening in remote locations, are costly to manage, cause significant operational disruption, and increase safety exposure for maintainers. A typical mobile asset has some number of sub-systems, each with its own maintenance interval and reliability. Our challenge is to set appropriate maintenance intervals for each asset sub-system to minimize the probability of an unplanned failure, without limiting unnecessarily the availability of the asset. To do this we generate sets of maintenance intervals using a genetic algorithm and test them using a discrete event simulation (DES) model of the operations and maintenance functions. We are motivated by two industry examples of fleets in remote locations: long-distance freight trucks, and heavy-haulage rail locomotives. In the truck case, the model found optimal intervals similar to those used by the operator. The locomotive case is more complex, but the model suggests improvements are possible in interval selection, maintenance practices, and data collection. Each model is conceptualized for its specific context; this process identifies assumptions that need to be considered when linking maintenance and operations models.