The Variability Associated With Heuristic Adaptive Replacement Policies

In the literature on replacement/maintenance policies for systems of stochastically failing machines, it is generally assumed that the failure time parameters are known and the objective is to find a replacement policy that minimizes expected cost per unit of time. In practice, however, the variability of costs can be an important consideration. In addition, most managers would want to adapt to new information obtained while operating the machines. In this paper, a class of policies is analyzed where the manager is allowed to adapt the replacement policy according to the statistical information obtained while operating the machines. It will be demonstrated that it is feasible to explicitly calculate both the expected cost per unit time and the variance associated with this class of replacement policies. This research has been partially supported by grant #0003658-472 from the State of Texas Advanced Technology Program, National Science Foundation Grant #DMC-8910378 and a Babcock Research Grant