clustering of condition-based maintenance activities with imperfect maintenance and predication signals

Combining maintenance activities for multi-component systems is called opportunistic maintenance or maintenance clustering, which is known to be cost-effective, especially for multi-component systems with economic dependency. every operating system is subject to gradual degradation which ultimately leads to system failure. each level of degradation can be represented by a state. the state of a system can be estimated through condition monitoring, albeit with uncertainty. in this paper, we present a clustering model that factors in uncertainty in alerting and lifetime distribution and considers the possibility of using the imperfect maintenance approach. this model is developed for a system with three levels of warning (signal, alert, alarm), which combines inspections and condition monitoring to avoid unnecessary inspections and thereby achieve better cost-efficiency. our analysis and results provide a general view of maintenance clustering to minimize costs and maximize system availability. different policies for clustering maintenance activities are proposed and compared. numerical investigations performed with matlab software show that maintenance clustering can result in as much as 80% cost saving compared to no clustering policy. moreover, alert clustering policy costs less than the other policies, and this cost difference becomes larger as the initial probability that the system units is weak increases.