Maintenance Optimization based on Mathematical Modeling

Due to an increasingly competitive marketplace and an ongoing shift from labor intensive to technology intensive industries, the importance of effectively planned and performed maintenance activities is growing. Because consequences of equipment failures are often severe, it is generally preferred to perform maintenance activities preventively. However, performing preventive maintenance too often is also undesirable and costly. A balance between the preventive maintenance frequency and the risk of failures therefore has to be found. Preventive maintenance actions can be based on the time that equipment is in service (time-based maintenance), or on condition information (condition-based maintenance). This thesis contributes to the current research on maintenance planning and optimization. The methodology that we use is that of developing and analyzing mathematical models to support decision making. We study the effect of uncertainty in the lifetime distribution on the optimal time-based maintenance strategy, and show that the consequences of ignoring this uncertainty can be substantial. We also analyze under what circumstances it is beneficial to initially postpone preventive maintenance actions in order to reduce the parameter uncertainty faster. Furthermore, we investigate how the benefits of condition-based maintenance over time-based maintenance depend on various practical factors. The obtained insights are useful for companies to assess whether it is beneficial to switch from time-based to condition-based maintenance. Finally, we consider condition-based maintenance optimization for multi-unit systems. We provide general insights into when and how to cluster maintenance operations, and show that clustering is beneficial under a broad range of circumstances