Detection of arc faults in transformer windings by transient signal analysis

Abstract In power and distribution transformers, lightning strikes, switching operations, short-circuits and other reasons can cause deterioration of the winding insulation, resulting in arcing between the windings. The arcing event is very important in transformers and if it is detected late, it can cause a rigid short circuit and burn the transformer and even cause the transformer to explode due to heating and pressure. In this study, the effects of arcing between transformer windings on transformer end current and voltage waves are analysed and arcing is determined. In the study, a 15 MVA power transformer was modelled in three dimensions in the Ansys@Maxwell environment and the arc model created in Matlab@Simulink was adapted to the Ansys magnetic model and the arc in the transformer windings as a result of synchronous operation was analysed. The transient regime effects of arcing at 5 different points in the same high voltage winding of the transformer on the current and voltage waves at the input terminals of the transformer are analysed to obtain data that allow the fault to be detected in a short time without the use of sensors. The data obtained from the transformer input was transformed into the frequency domain using Fast Fourier Transform (FFT), and the high-frequency transient regime signals occurring in the transformer under arcing fault conditions were detected from these signals. Using the method proposed in the study, it is possible to detect the occurrence of arcing in the transformer.