A real-time solution for preventive and repair maintenance

A deterministic approach to a single-machine maintenance problem is presented which is solved forward in real-time. The importance of real-time solutions is particularly evident in new manufacturing environments, where fast decisions reflecting the most current information are vital to obtain high utilization of expensive equipment and to avoid bottlenecks in the production process. A sequential decision-making process is introduced. First, the model presented is solved at the initial item when the machine is purchased. Then, each time a breakdown occurs, the model is solved again using the updated breakdown information. The initial solution contains the optimal preventive maintenance policy and the optimal breakdown-repair policy is obtained indicating whether a repair should be made or whether the broken machine should be junked. If a repair is made, then the optimal preventive maintenance policy and resale time of the machine are derived. It is shown that if a breakdown occurs following the interval in which preventive maintenance is optimally applied, then a repair is not advocated and the machine is junked. In addition, a method for deriving the future expected times of machine failures is presented. Therefore, in addition to the optimal decisions corresponding to the realized behaviour to the machine, optimal policies are derived which consider its future expected performance.