Joint decision making on preventive maintenance and reconfiguration in complex manufacturing systems.

Abstract

During the operation of a flexible or reconfigurable manufacturing system, maintenance and reconfiguration interact in a dynamic and complex way. The decision making on these two issues are always treated separately in both literature and practice. However, there is a benefit to combining them, because reconfiguration can be an alternative to maintenance actions and maintenance information can be used to make more accurate reconfiguration decisions. Hence, we developed integrated reconfiguration and maintenance strategies, which model maintenance in dynamic environments and explore reconfiguration considering component reliability and degradation process. Formal frameworks have been developed to obtain cost-minimizing control-limit type integrated polices. First, operation transfer are incorporated into age-based preventive repairs for a parallel-serial system. The expected total cost is estimated and minimized using a simulation-based optimization procedure. Second, production dispatching and maintenance scheduling are jointly determined for a multi-product, multi-stage complex manufacturing systems, e.g., semiconductor fabs. A two-phase hierarchical decision-making is developed, in which long-term job dispatching and maintenance actions are planned jointly on station level, while short-term detailed maintenance schedules are determined on system level. Last, dynamic load-allocation and condition-based maintenance are integrated for a multi-unit load-sharing deteriorating system. A Markov decision model on a joint state space is built to minimize the operating cost rate over an infinite horizon. Numerical experiments demonstrate that the proposed integrated policies result in lower total operating cost than those preventive maintenance policies without reconfiguration. Further, integrated policies require less frequent preventive repairs and yield larger utilization of the component residual life.