Clustering condition-based maintenance for manufacturing systems with both perfect and imperfect maintenance actions

Abstract

This paper integrates maintenance and product quality for a manufacturing system with three failure modes. The system is monitored and can be alerted by a prediction signal before any failure mode. Due to the high setup cost of maintenance activities, their clustering can be beneficial. To reduce the frequency of failures and their impacts on the quality of output products, a clustering model with perfect and imperfect maintenance actions is applied. The objective is to investigate the impacts of clustering maintenance actions on non conforming production costs and to compare cost saving in perfect and imperfect maintenances. Our analysis and results yield valuable insights into when and how to cluster maintenance operations so that the cost for maintenance and non conforming units be minimized. Clustering perfect maintenance can be more optimal than the imperfect one for several reasons, e.g., because perfect maintenance actions restore the system completely to the as-good-as new state and the time between two maintenance is larger, and thus the frequency of maintenance decreases.