Modeling the gain due to maintenance over transformer lifespan

The aim of this paper is to develop a methodology that makes it possible to take advantage of power transformer operators' knowledge in order to reach a better control of process of preventive maintenance (PM). The main idea is to construct probabilistic models using a Bayesian approach along with an age reduction model in order to compensate for both the lack of failure data on the maintained system and the co mplete absence of failure data on the non-maintained system. The originality of our approach is that we appeal to the operators to give their belief concerning the effectiveness of PM tasks. The results obtained consist of two models, one for the intrinsic behavior of the power transformer without PM, and one for the power transformer with PM. The comparison of these two models allows the quantification of the effectiveness of the PM activity, which can be used for an optimization of the scheme.