Studies on measurement time reduction techniques for insulation diagnostics in high voltage equipment

Abstract

Over the past few decades electrical power utilities all around the globe are primarily interested in asset management and proper utilization of resources. Among the different electrical equipment transformer is the most important asset to an electrical utility. This is because not only transformers are most expensive of all electrical equipment but also their failure leads to severe interruption of power supply, incurring huge financial loss to an electrical utility. One of the primary reasons behind the failure of a transformer is the degradation of oil-paper insulation. Therefore condition assessment of transformer insulation at periodic intervals is necessary to prevent premature failure of transformers. In this context, dielectric spectroscopy measurement in frequency domain also known as frequency domain spectroscopy (FDS) is a major non-invasive technique that is widely used to assess the condition of oil-paper insulation. In conventional FDS test, series of sine waves at distinct measurement frequency points starting from 0.1mHz to 1kHz are applied on the insulation under test, which makes conventional FDS test lengthy and very time consuming. The overall measurement time can go up to 10-11 hours which is practically problematic. This practical problem has been addressed in this thesis work and several methods to reduce FDS measurement time have been proposed which were not used in practice till date. Chapter 1 of this thesis provides a comprehensive overview of existing techniques for condition monitoring of transformer oil-paper insulation and also introduces the research problem. Chapter 2 presents a method to accelerate FDS measurement using regression approach. For this purpose both sinusoidal and non-sinusoidal waveforms are used for the purpose of experiment and reduction in FDS test time is determined using both sinusoidal and triangular excitation voltages. Chapter 3 presents a method to reduce FDS measurement time using chirp signals. For this work, broadband excitation signals in the form of chirp waveforms with linear and logarithmic frequency sweeps are used for the purpose of experiment. The overall reduction in FDS measurement time is computed for both types of chirp signals. Chapter 4 presents a non-linear model of oil-paper insulation using logarithmic chirp signal. Different parameters of non-linear model are found to be sensitive to change in moisture content. A comparative study of the proposed model with the linear model is also presented. Chapter 5 concludes the thesis by summarizing the main findings of the present work and also gives some idea of possible extension of the research work.