< Back to previous page

Publication

Condition-based maintenance and inventory management in aftermarket services

Book - Dissertation

Capital assets, such as manufacturing equipment, require maintenance to remain functioning. Maintenance can be performed when a component breaks down and needs replacement (i.e., corrective maintenance, CM), or the maintenance and part replacement can be performed preventively. Preventive maintenance can be planned on a periodic basis (i.e., periodic maintenance, PM), or it can be triggered by a certain monitored condition (i.e., condition-based maintenance, CBM).CBM makes use of the actual condition of the component to decide when to maintain and/or replace the component, thereby maximising the lifetime of the machine, while minimising the number of service interventions. In this dissertation we combine CBM on one (monitored) component, with CM and PM on the other components of the same machine. We implement two thresholds on the degradation level to decide when to service the monitored component: when the degradation level of the monitored component surpasses a first opportunistic threshold, the monitored component will be serviced together with other components, for instance with a (planned) PM intervention, or upon breakdown of another component, requiring CM. In case none of these opportunities have taken place, and the degradation level surpasses a second intervention threshold, an additional maintenance intervention is planned for the monitored component in order to prevent a failure. Both thresholds are optimised to minimise the total expected maintenance costs of the monitored component, or to minimise the downtime of the machine due to the monitored component. In this dissertation we perform an extensive numerical experiment to demonstrate the potential gains of this hybrid policy with two thresholds compared to using a traditional PM policy, and we identify its key drivers of performance. We also benchmark our results when only one threshold is implemented.We conclude that our hybrid CBM policy can reduce maintenance costs significantly. Therefore, we additionally determine how much of these gains remain, in case monitoring of the failure modes of a component is incomplete. We implement the same hybrid CBM policy with an opportunistic and intervention threshold on the degradation level to decide when to service the monitored component. In contrast to the initial model, only a percentage of failures are now anticipated by the monitored degradation data. We optimise both thresholds, taking into account the level of incomplete monitoring. A numerical experiment demonstrates the impact of such incomplete monitoring.Spare parts are vital to perform maintenance. For many companies it is unclear which impact the different maintenance policies have on the resulting inventory requirements for the spare parts that are used for the maintenance interventions. We study the impact of the maintenance policy on the inventory requirements and the corresponding costs. Preventive policies increase the total demand for spare parts compared to corrective maintenance, since the former do not exploit the entire useful life of the components. This leads to higher inventory requirements. At the same time, the preventive policies contain advance demand information, as the interventions, and correspondingly the spare parts demands, are planned in advance. Using a simulation study, we show that by using this advance demand information in managing the spare parts inventory, the increase in inventory requirements of preventive maintenance policies can to a large extent be offset; for CBM, we find that inventories can even be lower compared to CM, provided that the advance demand information is used correctly when managing inventories. Our analysis sheds light on the behaviour of the inventory related costs under various maintenance policies. All models proposed in this dissertation have been validated and used at an original equipment manufacturer in the compressed air and generator industry.
Publication year:2019