ÄûÃʵ¼º½

Title of the thesis: Decision models for preventive maintenance of technical systems

Thesis defender: Jussi Leppinen
Opponent: Professor Jutta Geldermann, University of Duisburg-Essen, Germany
Custos: Professor Ahti Salo, Aalto University School of Science

Technical systems, such as cars, consist of numerous components that require maintenance to ensure reliable operation of the system. One common maintenance action is to replace worn components with new ones. Components can be replaced either after failure or preventively to avoid breakdowns. Maintenance costs increase if components are replaced too frequently. On the other hand, insufficient maintenance may lead to component failures, resulting in costs due to system downtime. In this setting, the optimization of maintenance schedules can lead to significant cost savings.

Different wearing rates and dependencies between components complicate optimization. These characteristics impact the decisions to replace components simultaneously. Simultaneous replacement may not be justified for components that wear at different rates. However, if a component must be disassembled from the system to replace another due to a structural dependency, it may be cost-effective to replace the disassembled component simultaneously.

This dissertation presents a mathematical model for designing a cost-effective replacement schedule for system components. The key novelty of the model lies in its ability to account for differences in component wear and the structural dependencies between components in the planning of replacements. Structural dependencies affect the maintenance costs and the reliability of the system. The model helps determine which components should be replaced together and which should be replaced later to minimize the total costs resulting from maintenance and expected system downtime. The decision recommendations are based on information about each component’s previous replacement times and current condition.

In addition, the dissertation introduces a decision process to guide the adoption of AI-based solutions for monitoring system condition and predicting their remaining useful life. These solutions can provide more accurate information about the current condition of systems, supporting failure prediction and informed maintenance decisions. Early consideration of organizational and practical requirements, alongside the system's technical performance, increases the likelihood that the developed solutions align with expectations.

Keywords: maintenance scheduling, multi-component system

Contact information: jussi.leppinen@aalto.fi

Thesis available for public display 7 days prior to the defence at .