A Condition-based Preventive Maintenance Policy for Markov Deteriorating Systems

Abstract: A condition based preventive maintenance (CBPM) policy is proposed for a continuously operating device whose condition deteriorates with time in service. The model incorporates both deterioration and random failures. Deterioration is modeled as discrete states process. The system undergoes random inspections to know the condition, mean time between inspections are exponentially distributed. If the observed condition at an inspection exceeds the threshold deterioration level, Preventive maintenance (PM) is performed. Else no action takes place, continue to run the system. Each PM makes the system τstages ( 1   ) younger. The proposed models consider an accumulated deterioration based increasing intensity for the random failures. A continuously increasing failure rate (for example Weibull) is converted into a stepwise increasing failure rate using stair-step approximation. An exact recursive algorithm computes the steady-state probabilities of the system. An operating unit time (hour, day, week, etc.) based cost function is defined using different cost rates for the different types of outages. Based on maximum availability or minimum costs, optimal solution of the model is derived.