Condition monitoring of power transformers using frequency response analysis and partial discharge evaluation

verfasst von

Sahand Seifi

betreut von

Peter Werle

Abstract

The rapid increase of electricity demand due to the population increase, emerging new electrical devices in daily life, and breakthrough in the electromobility as well as considering the electricity as an alternative solution for the house heating resulted in a necessity for quick expansion of the electrical network worldwide. Moreover, international agreement about carbon neutrality, reduction of greenhouse gases, and energy independence will make the renewable energy the only source of energy in the future and leads to a huge investment in this area. This investment and the cost of the network can be reduced by increasing the reliability and lifetime of the network equipment. Therefore, the reliability and availability of transformers, as one of the most expensive and complex equipment of power system, are of special importance. Any serious damage as well as an unplanned outage of transformers could result in a huge expenditure due to the interruption in the continuous transmission of electricity as well as repairing costs. Therefore, their failure should be detected at early stages to avoid operating costs or reduce the repairing time. A major source of transformer failure is a fault in the windings due to the mechanical tensions or insulation deterioration due to aging. Frequency response analysis (FRA) and partial discharge (PD) measurements are two common methods to assess the winding condition. FRA was frequently utilized to detect winding mechanical failures like winding movement and deformation as well as winding inter-turn and turn to ground faults. PD measurements were exploited to detect insulating system damages as well as to localize them with different methods of PD measurements. This study has focused on the improvement of frequency response analysis. Firstly, the FRA method was adapted to not only detect all types of internal winding faults but also determine their types and location. Then, FRA measurement was mixed with the PD measurement in order to localize PD and estimate the original charge of PD in transformer windings. In the first part of this study, the sweep reflection coefficient (SRC) was introduced as a new frequency response for transformer applications. The study shows that the sensitivity of the SRC method to fault detection is as good as the FRA method; however, it has the advantage to categorize different types of internal winding faults and to localize them. The method was verified by various types of faults in a model winding and a real distribution transformer. It reveals that this method is able to localize winding turn to ground fault with resistive components with errors less than 5 %, the capacitive turn to ground faults and inter-turn faults as well as modeled mechanical faults with an error less than 15 %. The second part of this study focuses on the PD analysis. First, the performance of PD localization using the electrical method of PD measurement was improved by proposing a new index to compare the reconstructed signal of PD. Then, the original charge of the PD induced in the closest metallic part of the winding in the vicinity of the PD origin was estimated by recording the PD signal at the ends of the winding as well as utilizing sectional winding transfer function and input impedance of the winding. The estimation of the original charge of PD helps to define a more robust PD threshold value for the acceptance test as well as to have a better assessment of fault severity.

Organisationseinheit(en)

Fachgebiet Hochspannungstechnik und Asset Management (Schering-Institut)

Typ

Dissertation

Anzahl der Seiten

191

Publikationsdatum

12.11.2024

Publikationsstatus

Veröffentlicht

Elektronische Version(en)

https://doi.org/10.15488/18115 (Zugang: Offen)