Cost-Effective Updated Sequential Predictive Maintenance Policy for Continuously Monitored Degrading Systems

The importance of maintenance optimization has been recognized over the past decades and is highly emphasized by today's competitive economy. In this paper, an updated sequential predictive maintenance (USPM) policy is proposed to decide a real-time preventive maintenance (PM) schedule for a continuously monitored degrading system that will minimize maintenance cost rate (MCR) in the long term, by considering the effect of imperfect PM. The USPM model is continuously updated based on the change in the system state to decide an optimal PM schedule. Mathematical analysis of the proposed USPM model demonstrates the existence and uniqueness of an optimal PM schedule under practical conditions. The results validate that: 1) the proposed USPM model yields PM schedules that are consistent with the change in the system states and 2) the USPM model is able to quickly react to drastic degradation of the system and provide an optimal PM schedule in real time. The proposed maintenance policy can provide significant benefits for real-time maintenance decision making. Note to Practitioners-This paper is motivated by the gap that scheduling of commonly applied imperfect preventive maintenance (PM) (e. g., adding lubrication, partial replacement, etc.) scarcely considers a system's operating condition which is highly correlated with machine health and failures. The updated sequential predictive maintenance (USPM) policy developed in this paper outlines a framework for real-time PM scheduling in a cost-effective way. To implement the proposed method, it is necessary to: 1) monitor a performance variable (e. g., pressure, temperature, etc.) that well indicates the system state; 2) estimate the system lifetime distribution; 3) quantify the PM work orders; and 4) measure the maintenance cost. Although the proposed maintenance policy is based on the objective of minimizing maintenance cost rate (MCR), it can be easily revised according to other practical optimization objectives, i.e., maximizing system availability.