On Optimality Criteria for Age Replacement

Abstract

Age replacement is a well-known topic in the literature of operational research and reliability. Traditionally, the probability distribution of a unit's failure time is assumed to be known, and the cost criterion is derived via the renewal reward theorem, which implicitly assumes that the same preventive replacement strategy will be used over a very long period of time. As an alternative, a one-cycle criterion can be used, aiming at minimization of costs per unit of time only over the period that one unit is in place. These two criteria are discussed and possible alternatives are also considered. Recently, the authors have presented a non-parametric predictive approach to age replacement, which is based on rather minimal assumptions for the failure time distributions, and provides full flexibility to the information from the process. The main conclusions from this research, where both the renewal criterion and the one-cycle criterion were also considered, are summarized. Further aspects related to age replacement are discussed, highlighting several interesting topics for further research.