This dissertation presents an information-based maintenance optimization methodology for physical assets; with focus on, but not limited to, predictive maintenance (PdM). The overall concept of information-based maintenance is that of updating maintenance decisions based on evolving knowledge of operation history and anticipated usage of the machinery, as well as the physics and dynamics of material degradation in critical machinery components. Within this concept, predictive maintenance is a maintenance policy that specifically uses predictions of component remaining useful life (RUL) to dynamically schedule maintenance activities. Analysis of the available information-based maintenance methodologies and e-maintenance standards identified the development of advanced maintenance policies like predictive maintenance as the most important challenge. Generally speaking, within e-maintenance the sensor module, the signal-processing module, the condition monitoring module and the diagnostic model can all be (partially) developed using standard means and models. However, this is currently not the case for the decision support modules. Moreover, the evolution of maintenance is not solely based on technical but rather on techno-economic considerations. The right maintenance decision making structure should be in place to fully exploit the potential of these new emerging technologies. Therefore, decision support models and tools for predictive maintenance performance evaluation and optimization are developed in this thesis. Hence, a detailed study on the business economics related to the implementation of an information-based/predictive maintenance policy is performed. Predictive maintenance models for long-term performance evaluation, real-time and dynamic decision making and a combination of both are developed. As such contributions are made towards (i) the development of an imperfect condition monitoring system (CMS) model, (ii) predictive maintenance models incorporating product quality and production capacity and (iii) a dynamic predictive maintenance policy for complex dependent multi-component systems. These models provide maintenance decision support in order to take cost-effective decisions based on predictive maintenance information. Moreover, they provide sound business insight for the justification of PdM and as such assist to determine the cases in which PdM is expected to be very beneficial, beneficial, neutral or possibly too expensive. Furthermore, the effect of predictive maintenance information on inventory management decisions is studied. The major contribution of this dissertation lies within the development of predictive maintenance models. However, contributions to other problems within maintenance management, like (i) the urge for more application based maintenance optimization, (ii) the limited scope with regard to maintenance objectives and criteria and (iii) the availability of maintenance data, are made. As such most of the developed models are applied to real-life case studies to illustrate their applicability in an industrial setting. A methodology, based on the analytic network process (ANP), is developed to select and prioritize business specific maintenance objectives and criteria. And finally, the developed models possess the capability to solve the data problem by providing the maintenance decision maker the right information at the right time to make the right maintenance decision.
展开▼
