Investigation of partial discharge propagation and location in multiple-a and single-a transformer windings using optimized wavelet analysis

Abstract

Partial discharges (PD) are recognized as the main cause of the inner insulation deterioration process in power transformers. Therefore, the optimum inner insulation design is one of the challenges a transformer designer is faced with. Transformer strength, especially during transient conditions, is a criterion for transformer insulation designers. This challenge has made designers initiate and employ other types of winding, for example, rather than ordinary layer and disc windings employ the multiple-α windings. Multiple-α windings have a more complicated structure and are comprised of various parts with different physical structures and electrical characteristics. Typical partial discharge signals cover a wide frequency range from DC up to hundreds of MHz and different frequency components propagate through the winding depending upon the winding structure in different modes. Partial discharge propagation in single-α winding is more predictable compared to multiple-α winding. A 66 kV / 25 MVA interleaved winding, which has 19 fully interleaved discs, plays the role of a single-α winding. When this main winding is connected to the tap winding with a different structure and magnitude response, a multiple-α winding is constructed. Two terminal current signals are detected by the application of two home-made high frequency current transformers (HF-CT). The signals were amplified and fed into a 500 MHz digital storage oscilloscope. Home-made sensors are designed to provide maximum sensitivity in the desired frequency range. In order to evaluate the partial discharge signal accurately, a method for selecting the optimal wavelet is introduced to reduce the noise effects. This method is based on the capability of the chosen mother wavelet for generating coefficients with maximal values. The wavelet based de-noising method proposed can be employed in extracting the PD pulses from the measured signal successfully to provide enhanced information and further infer the original site of the PD pulse through the capacitive ratio method..