Offline to online mechanical deformation diagnosis for power transformers

Internal winding deformations of power transformers can be detected using the conventional offline frequency response analysis (FRA) which is a well-known and widely accepted tool for the detection of winding and core deformations. In addition of being offline technique, interpretation of FRA signature is based on graphical analysis that requires skilled personnel as there is no reliable standard code for FRA signature identification and quantification. This paper presents the possibility of using an alternative online technique based on construction a voltage-current (ΔV-I) locus of the operating transformer and considering it as a reference signature. In order to fully explore the performance and reliability of the new proposed approach particularly for real-life distribution transformers, the paper investigates and compares the performance of the proposed and the FRA approaches for disk space variation and axial displacement faults. The transformer distributed parameter model is used to simulate FRA signatures while a detailed three-dimensional finite element model is used to generate the ΔV-I louses for healthy and faulty operating conditions. Simulation results are compared to highlight the advantages and limitations of the two internal fault detection strategies.