A Simulation-based Evaluation of Dynamic Task Prioritization in Maintenance Management

Reliability and availability of machines are crucial for an efficient manufacturer and great maintenance operations are the key. Due to the advancement of condition monitoring and predictive analytics more information is available and the field shifts from corrective to preventive task fulfillment. Nevertheless, well trained experts are rare and responsible for different machines at the same time. Most experts have already identified that the sequence of task execution could heavily influence the production output. However, in practice only simple heuristics are used, yet they still improve the performance significantly. The question arise if a prioritization of work orders could be further enhance using bottleneck detection methods. To show the potential of these techniques, a simulation based study is conducted. First, the reliabilities of selected bottleneck detection methods are tested on a simplified line model. Second, the performances of all prioritization policies are evaluated and compared using a real industrial use case. Contrary to the expectations, a simple heuristics has shown a good performance. Nevertheless, applying the Active Period Method has significantly outperformed the other approaches and the bottleneck ranking has proven to be a very good indicator to prioritize work orders.