Abstract
A new method for systematically estimating health index, probability of breakdown and remaining life for power transformers is presented. The method combines three basic models; a physical winding degradation model, a health index model based on condition monitoring data combined with expert judgement, and a statistics-based end-of-life model. The statistics-based model uses data from a database of scrapped transformers under development in Norway. Combining the first two models with the statistics-based model, an individual and condition-dependent probability of breakdown is obtained. From this, the expected remaining life is calculated. Finally, the stochasticity of the method is utilized for optimization of maintenance and replacement. The method hence provides key decision support for transformer managers, enabling them to identify transformers in poor condition, and to follow-up and prioritize transformers for maintenance and replacement. The proposed method has been implemented in a transformer asset management tool for Norwegian utilities. The usefulness of the method is illustrated by applying it to selected transformers from one of these utilities. Finally, important limitations, uncertainties and further improvements are discussed.
