Dynamic reliability degradation based models and maintenance optimization

This paper focuses on the reliability estimation of equipment based on degradation data and on a economic maintenance model that takes into account corrective and preventive replacements as well as inspections costs. We generate the degradation data with a semi-Markov model with 3 states. Each state has a constant degradation rate that contributes to the damage. The degradation paths are simulated with the Markov chain until they reach a fixed known threshold. The reliability is then calculated based on the collection of the hitting times. We propose a dynamic model for the reliability estimation that is able to take into account new data. The results showed that the closer the threshold the more accurate the reliability will be. The economic maintenance model was tested without and with inspection costs. For the both case, the application of the proposed methodolgy allows to decrease the mean cost per hour compared to the application of the systematic calendar preventive maintenance. Keywords— reliability, degradation, maintenance optimization, semi-Markov, mean residual life