A non-invasive measurement-based modeling for assessing power transformer aging and reliability

Abstract

A non-invasive and quantitative methodology for assessing the health of oil–paper insulation in power transformers is presented in this paper. Paper moisture (moisture in cellulosic part of insulation) (pm) and dissipation factor (tan δ) are widely recognized as critical indicators of insulation aging and overall system reliability, yet their accurate determination often requires intrusive or time-consuming procedures. To overcome these limitations, this work introduces an Aging Factor (A_F), computed directly from the complete depolarization current profile by estimating the branch parameters of a Debye-based insulation model and evaluating the ratio of charge contributions associated with the two branches having the largest time constants. This approach utilizes the full dielectric relaxation behaviour of the insulation without relying on extensive curve fitting or lengthy frequency-domain measurements. The method was validated using ten laboratory-prepared oil-impregnated insulation samples and subsequently applied to ten in-service power transformers. The AF showed strong correlation with both %pm and %tan δ, enabling their prediction with accuracies of 2.47% and 5.0%, respectively, with all deviations confined within ± 5%. By providing a robust, non-invasive, and time-efficient means to estimate key insulation-health indicators, the proposed AF-based technique supports condition-based maintenance and enhances diagnostic reliability for power transformer insulation systems.