Condition-based maintenance (CBM) has received increasing attention in the literature over the past years. The application of CBM in practice, however, is lagging behind. This is, at least in part, explained by the complexity of real-life systems as opposed to the stylized ones studied most often. To overcome this issue, research is focusing more and more on complex systems, with multiple components subject to various dependencies. Existing classifications of these dependencies in the literature are no longer sufficient. Therefore, we provide an extended classification scheme. Besides the types of dependencies identified in the past (economic, structural, and stochastic), we add resource dependence, where multiple components are connected through, e.g., shared spares, tools, or maintenance workers. Furthermore, we extend the existing notion of structural dependence by distinguishing between structural dependence from a technical point of view and structural dependence from a performance point of view (e.g., through a series or parallel setting). We review the advances made with respect to CBM. Our main focus is on the implications of dependencies on the structure of the optimal CBM policy. We link our review to practice by providing real-life examples, thereby stressing current gaps in the literature.

Efficient (condition-based) maintenance planning and inventory control of spares for critical components jointly determine the effectiveness of a maintenance strategy and, thereby, balance system uptime and maintenance costs. Duplicating an optimal policy for a single-component system to a multi-component system is not necessarily optimal, while a separate or sequential optimization of the maintenance and inventory decisions is also not guaranteed to yield the lowest costs. We therefore consider the joint optimization of condition-based maintenance and spares planning for multi-component systems. We formulate our model as a Markov Decision Process, and minimize the long-run average cost per time unit. A key insight from our numerical results is that the ( s , S ) inventory policy, popular in theory as well as practice, can be far from optimal for systems consisting of few components. Significant savings can be obtained by basing both the maintenance decisions and the timing of ordering spare components on the system's condition.

a b s t r a c tSystems that require maintenance typically consist of multiple components. In case of economic dependencies, maintaining several of these components simultaneously can be more cost efficient than performing maintenance on each component separately, while in case of redundancy, postponing maintenance on some failed components is possible without reducing the availability of the system. Condition-based maintenance (CBM) is known as a cost-minimizing strategy in which the maintenance actions are based on the actual condition of the different components. No research has been performed yet on clustering CBM tasks for systems with both economic dependencies and redundancy. We develop a dynamic programming model to find the optimal maintenance strategy for such systems, and show numerically that it can indeed considerably outperform previously considered policies (failure-based, age-based, block replacement, and more restricted (opportunistic) CBM policies). Moreover, our numerical investigation provides insights into the optimal policy structure.

Redundancy is often essential for achieving high system availability. An additional benefit of installing redundant components is that the total system load can be shared among components, thus preventing fast deterioration. On the one hand, this provides an incentive to replace failed components as soon as possible, as a component failure increases the load on the remaining components. On the other hand, however, redundancy gives rise to maintenance clustering and postponement opportunities, to reduce the maintenance frequency and thereby lower downtime and maintenance set-up costs. To date, this trade-off has not been investigated under a condition-based maintenance regime. To address this, we consider a parallel system that is subject to both failure dependence through load sharing and economic dependence through maintenance set-up costs. We formulate our system as a Markov Decision Process, and obtain the optimal replacement decisions that minimize the long-run average cost per time unit. Through a numerical investigation and a sensitivity analysis, in which we vary the degree of load sharing, and the maintenance set-up cost, and the degradation process, we obtain key insights into the optimal policy structure. Standard threshold policies, that replace a component as soon as its deterioration exceeds a certain threshold, can be far from optimal, while ignoring or misinterpreting the load sharing effects between components can also lead to a significantly more expensive maintenance policy.

No abstract