Abstract
This paper deals with a practical approach for the analysis and modelling of preventive maintenance (PM) data for repairable systems which have an increasing failure frequency and/or increasing severity. The concept and testing for the trend of severity of corrective work (CO) and PM are discussed. A framework for statistical analysis of interarrival times and downtimes due to CO/PM is proposed. A generalized non-stationary model for scheduling PM to maximize availability is suggested. The effect of severity on scheduling PM activities is shown through sensitivity analysis. Copyright (C) 2002 John Wiley Sons, Ltd.