Quasi-Periodic Replacement Policy for a Two-Unit Production System

Abstract

Maintenance optimization of periodic replacement for mono-unit production system is a widely discussed topic. In practice, the replacement for a parallel production system with multi-unit served by one repairman is universal. The replacement action is often implemented in a maintenance window at random, although the replacement period appears as a quasi-periodic event. To optimize these maintenance policies, this study analyzes a bivariate quasi-periodic replacement policy (T, W) for a two-unit parallel production system with a maintenance window and one repairman. One unit of the system is replaced when either a failure occurs or a replacement plan comes in the maintenance window, whichever comes first. If one of the two units is in the replacement, the other may continue operating or waiting until the replacement is completed. An optimal replacement window [T*, T* + W*] can be obtained by jointly considering the system’s long-running cost rate and the availability using the genetic algorithm. This study also introduces three examples of the production system with different types of units to illustrate the proposed policy.