Characterization of optimal age-replacement policies

The intention of the effort presented is to characterize the solution to a general class of preventive maintenance optimization problems. Successive lengths of operating periods and the successive repair times for a piece of equipment are each assumed to be i.i.d. Weibull random variables. The equipment is subjected to an age replacement preventive maintenance policy and the successive preventive maintenance times are assumed to be i.i.d. Weibull random variables. An optimal age replacement policy is a policy, T_age*, that maximizes A_avg(T_use), the average availability of the equipment over its useful life (T_use<∞). The relationship between T_age* and the parameters of the previously mentioned Weibull distributions is approximated by a first-order polynomial function. The model is formulated through regression analysis of data obtained through simulation-optimization techniques. Model validation is presented. In general, the model predicts an optimal T_age close to the true optimal. The model, however, tends to prescribe less frequent preventive maintenance than the true optimal policy and is not useful in some in situations where the true optimal T_age is very small