Maintenance scheduling of geographically distributed assets with prognostics information

Abstract

Maintenance scheduling for high value assets has been studied for decades and is still a crucial area of research with new technological advancements. The main dilemma of maintenance scheduling is to avoid failures while preventing unnecessary maintenance. The technological advancements in real time monitoring and computational science make tracking asset health and forecasting asset failures possible. The usage and maintenance of assets can be planned more efficiently with the forecasted failure probability and remaining useful life (i.e., prognostic information). The prognostic information is time sensitive. Geographically distributed assets such as off-shore wind farms and railway switches add another complexity to the maintenance scheduling problem with the required time of travel to reach these assets. Thus, the travel time between geographically distributed assets should be incorporated in the maintenance scheduling when one technician (or team) is responsible for the maintenance of multiple assets. This paper presents a methodology to schedule the maintenance of geographically distributed assets using their prognostic information. Genetic Algorithm based solution incorporating the daily work duration of the maintenance team is also presented in the paper.