The transfer function (TF) method is presently a well-known method used to detect various types of winding damage in power transformers. Although abundant research has been done on this subject using laboratory windings as test objects, it is hard to find one, whose test objects are actual large-power transformer windings. Hence, a 400 kV disc winding consisting of 86 discs is used in this paper to study turn-to-turn short circuit with the help of the TF method. To evaluate the effects of this type of fault on TF curves, some mathematical comparison algorithms are used in this research.
As it is found in the related published literatures, the transfer function (TF) evaluation method is the most feasible method for detection of winding mechanical faults in transformers. Therefore, investigation of an accurate method for evaluation of the TFs is very important. This paper presents three new indices to compare the transformer TFs and consequently to detect the winding mechanical faults. These indices are based on estimated rational functions. To develop the method, the necessary measurements are carried out on a 1.3 MVA transformer winding, under intact condition, as well as different fault conditions (axial displacement of winding). The obtained results demonstrate the high potential of proposed method in comparison with two other well-known indices. Additionally, two important methods for describing TFs by rational functions are studied and compared in this paper.