Evaluating Success Conditions for Predictive Maintenance

• Published in: Proceedings. Annual Reliability and Maintainability Symposium pp. 1 - 7
• Main Authors: Fischer, Paula, Dazer, Martin
• Format: Conference Proceeding
• Language: English
• Published: IEEE 27.01.2025
• Subjects: Costs; Decision making; Maintenance; Maintenance strategies; Mathematical models; Parameters for PHM; Predictive maintenance; Preventive maintenance; Prognostics and health management; Random access memory; Reliability; RUL; Timing; Uncertainty
• ISSN: 2577-0993
• DOI: 10.1109/RAMS48127.2025.10935273

The simulation study effectively demonstrates the feasibility of different maintenance strategies. The derivation of regression equations for all three maintenance strategies - reactive maintenance (RM), predictive maintenance (PM), and preventive maintenance (IM)-[1] provides valuable insights into the cost implications of each strategy across a wide range of parameter combinations. These regression equations enable calculating costs for each operational strategy based on various input parameters. It is, therefore, possible to calculate which operational strategy is likely to be the most cost-effective under the given conditions, providing clarity on whether the implementation of a predictive maintenance strategy can be successful. A case study shows that when using predictive maintenance with RUL (Remaining Useful Life) prediction, the prediction quantile that determines the maintenance timing should be chosen at a low level. It was also shown that, for certain cost ratios between predictive costs and reactive costs, the range within which predictive maintenance can work under given circumstances can be very narrow (or even nonexistent).