Evaluation of Numerical Indices for Objective Interpretation of Frequency Response to Detect Mechanical Faults in Power Transformers

For assessing the mechanical integrity of power transformers using the well-known frequency response analysis (FRA) method, proper interpretation of FRA results is crucial. In spite of many research efforts in developing a systematic interpretation methodology, there is still no universally accepted and definitive algorithm for interpretation of FRA results. The purpose of interpretation of the FRA results is to detect the extent and type of a mechanical fault followed by a judgement about the condition of the transformer, whether it needs a repair or it can continue its normal operation. In literature, numerical indices are one of the mathematical algorithms, which have been proposed for the interpretation, where the amount of an index extracted from two FRA signatures can show the degree of a mechanical fault. To ease the FRA based condition monitoring, this paper evaluates different numerical indices, presenting a number of new indices and collecting them on a single platform, as an interpretation tool for FRA results. To discuss the sensitivity of the indices against different faults, axial displacement (AD) and radial deformation (RD) as two common transformer-winding faults, with various extents, were practically applied to the windings of a 1 MVA distribution transformer. For each case, numerical indices are evaluated in different frequency bands to provide a deeper understanding to their characteristics. To discuss the sensitivity of different connection schemes, indices are evaluated and compared for four connection schemes; end-to-end open circuit (EE-OC), end-to-end short circuit (EE-SC), capacitive inter-winding (CIW) and inductive inter-winding (IIW) are recommended by IEC standard. The results show the advantage of the proposed indices over the others in the FRA interpretation, however further studies from the field are needed to settle the new indices. The research can be beneficial for the standardization process of the numerical indices.