Maintenance optimization of a production system with buffer capacity

We consider a deteriorating installation, which supplies input to a subsequent production system. The installation is provided with a buffer to cope with unexpected failures of the installation, which may cause interruptions of the production process. By preventive maintenance on the installation a further improvement of the production process can be obtained. In this paper we consider a preventive maintenance policy, which is based not only on the information about the age of the installation but also on the content of the subsequent buffer. For this integrated maintenance–production problem, we analyse a class of control limit policies which are nearly optimal and easy to implement. The analysis is based on the embedding technique from Markov decision theory. In addition, we provide a characterization of the overall optimal policy and report on numerical comparisons of the best control limit policy with the overall optimal policy and other reference policies.