INTER-TURN FAULT LOCALIZATION IN TRANSFORMER WINDINGS BY ANALYZING FREQUENCY REGIONS OF SFRA

Abstract

Power system reliability is contingent upon the effective performance of a power transformer. To ensure optimal operation of the power system, it is crucial that a transformer must operate within its designated operating limits and adhere to established standards. Power system transients can impact its core, insulation, and winding turns. A fault internal to the transformer quite often involves only few turns of the winding. While the current in the shorted turns is large in magnitude, change of the currents at the terminals of the transformer is low compared to its rating. As the repair due to fault is a critical, tedious and costly process, recognition of fault location is extremely important. A prior knowledge of fault location helps in reducing the maintenance cost and vulnerability towards further damage. Therefore, identification of fault location within winding necessitates the diagnosis of the transformer itself. In the existing literature, various techniques have been proposed for fault detection in transformers. Commonly employed methods for fault detection encompass chemical analysis, terminal measurement, flux measurement, and frequency response analysis. The chemical analysis is usually done by taking samples of oil and paper insulation of a transformer under investigation. The most common chemical analysis-based method is Dissolved Gas Analysis (DGA). The terminal measurement-based methods make use of current and voltage signals at the terminals of the transformer. Further, utilizing the signals, various approaches based on harmonics analysis, transformation theory have been adopted for fault diagnosis. The flux-based method traces changes in patterns of leakage and mutual fluxes in the transformer. The alteration in flux pattern due to fault changes the induced voltage across the sensor, facilitating fault diagnosis. The analysis of frequency response curve enables detection of electrical and mechanical faults inside the transformer. Moreover, the frequency response can be drawn with the application of either low voltage impulse or sweep frequency response to the transformer windings. In the above-mentioned methods, the sweep frequency response analysis (SFRA) provides high sensitivity and repeatable frequency response curves for analyzing fault in transformers. The analysis of frequency response curve is not only used to detect electrical and mechanical faults but also to locate them. It is based on a graphical comparison, in which SFRA plot of the transformer is compared with its fingerprint to detect and locate the iii faulty section inside the winding. However, significant experience and proficiency in frequency response curve analysis is required to diagnose the fault. Hence, the prime objective of the thesis is to develop methods/algorithms for the localization and severity assessment of faults in transformer windings using SFRA. It is evident from the literature that faults within transformers impact parameters like capacitances and inductances leading to variations in the SFRA curve. These variations, influenced by the deviation in winding parameters, produce alterations in different frequency regions of the SFRA curve. For instance, change in capacitance tend to significantly affect the higher frequency region of SFRA, while variations in an inductance are more likely to impact the lower frequency region. Considering this characteristic of SFRA variation, the work presented in the thesis introduces fault diagnosis and fault location identification algorithms by considering the impact of variation in winding parameters. In the first approach, a new inter-turn fault localization methodology utilizing higher frequency region of SFRA has been developed. The proposed method determines pre-fault Inter-Turn Fault Factor (ITFFpre) for all possible fault locations by estimating the values of series and shunt capacitance of the windings during healthy conditions and storing them in a lookup table. For a faulty transformer, post-fault Inter-Turn Fault Factor (ITFFpost) is calculated by determining the value of the equivalent capacitance using SFRA. The exact fault location in the faulty transformer is obtained by comparing the calculated value of ITFFpost of the faulty transformer with the values stored in the lookup table. While the proposed scheme effectively identifies fault locations, it specifically addresses full disk failure scenarios, relying on significant variations in equivalent capacitance. Hence, in the second approach, an attempt has been made for fault location identification at an incipient level. The efficacy of the suggested approach is verified by emulating faults at various locations on a transformer model developed in the laboratory. Its authenticity is also verified on an existing 400 kVA, 11 kV/440 V distribution transformer installed in the real field. The results indicate that the suggested method is capable to identify inter-turn fault location. In a subsequent chapter, a new method for diagnosing inter-turn faults at the incipient stage within power transformer windings is introduced using SFRA. This proposed technique focuses on the lower frequency region of FRA during pre-fault and post-fault conditions of transformer windings. It determines the deviation in equivalent inductances and calculates a location factor based on the deviation, which indicates the fault location. Utilizing the deviation in equivalent inductance, an accuracy factor (x) is calculated for iv each section of the winding. Subsequently, the extent of turn-to-turn fault and the number of faulted turns is determined based on the accuracy factor of the faulty disk. The effectiveness of the proposed technique is validated through a simulation model, two different laboratory-developed winding models, and a 400 kVA real-field transformer. The results demonstrate that the developed technique is capable of estimating the location of turn-toturn faults even when the number of faulted turns is less than 5% of the total turns. While its response remains robust against various winding configurations, the requirement of an additional terminal becomes a limiting factor. To address the limitation of an additional terminal requirement in the previously suggested method, a new approach for identifying incipient fault location is introduced based on the Fault Location Factor (FLF). The FLF is derived using pre-fault and post-fault winding impedances at a reference frequency (fref ) corresponding to the existing terminals. Using the winding parameters, the FLF value is determined at the reference frequency for each section and stored in a lookup table called FLFcal. During a fault event in the transformer, the post-fault FLF is computed using the measured winding impedance at the reference frequency. By comparing the values of FLFmeas and FLFcal, the precise fault location in the transformer is determined. The effectiveness of this approach is assessed through a comprehensive simulation study to ensure its reliability. Additionally, two different laboratory-developed winding models of the transformer, each with distinct winding configurations, and a 3-phase field transformer of 400 kVA are utilized to validate the proposed approach. The presented work is likely to contribute significantly to the area of inter–turn fault localization. The distinct approach considering different frequency regions of SFRA can be useful for inter-turn fault diagnosis irrespective of the winding configuration. Some suggestions, based on the observations, simulations and analytical studies in this area, are proposed at the end of the thesis for the benefit of potential researchers.