| dc.description.abstract | An appropriate maintenance policy is essential to reduce expenses and risks related to repairable systems failures. The usual assumptions of minimal or perfect repair at failures are not suitable for many real systems, requiring the application of Imperfect Repair models. In this work, the classes Arithmetic Reduction of Age and Arithmetic Reduction of Intensity, proposed by Doyen and Gaudoin (2004) are explored. Likelihood functions for such models are derived, and the parameters are estimated, allowing to compute reliability indicators to forecast the future behavior of the failure process. Under the classic Imperfect Repair virtual age model presented by Kijima et al. (1988) (particular case of Aithmetic Reduction of Age class), a periodic Preventive Maintenance policy is proposed, which estimates optimal time intervals for Preventive Maintenance, in order to minimize (preventive and corrective) maintenance costs. Under a dynamic perspective, it is showed how this policy can be improved, using each failure observation in order to recalculate the optimal time to Preventive Maintenance for a particular system, considering the effect of the repair action. These policies are applicable to any Imperfect Repair model. Monte Carlo simulation studies are implemented in order to evaluate the performance of the proposed methods. Those methods are applied to a real situation regarding the maintenance of engines of off-road trucks used in a mining company. These results bring valuable information to support decision making regarding Preventive Maintenance policy. | |
