A model for determining condition-based maintenance policies for deteriorating multi-component systems
conference paper
We discuss a method to determine strategies for preventive maintenance of systems consisting of gradually deteriorating components. A model has been developed to compute not only the range of conditions inducing a repair action, but also inspection moments based on the last known condition value so as to minimize expected costs per time unit. ln our cost optimization, we include costs of inspection, repair, failure and operation, while accounting for savings that can be obtained by combining inspection or repair of two or more components. Deterioration is modelled as a stochastic process; an independence assumption allows us to model the decision problem as a (semi-)Markov decision process. The solution procedure for the system as a whole is a heuristic one as we use only aggregate information about the other components while determining a maintenance strategy for a particular component. This strategy is of the opportunistic control-limit type: upper limits induce mandatory actions, lower limits allow anticipatory action if a combination with a mandatory action on at least one other component is possible. ln this paper we describe the principles of this heuristic. We show how other performance indicators can be calculated besides costs and give a numerical example of a system of components made of concrete.
TNO Identifier
94643
Publisher
KEMA
Source title
SRE Symposium - Reliability: A competitive edge, October 4th - 6th 1993, Conference Centrum Elektrum, Arnhem, The Netherlands. Part 2
Editor(s)
Gestel, P.J. van
Roseboom, W.
Roseboom, W.
Place of publication
Arnhem
Pages
358-371