Transmission voltages worldwide are increasing to accommodate higher power transfer from power generators to load centers. Insulator dimensions cannot increase linearly with the voltage, as supporting structures become too tall and heavy. Therefore, it is necessary to optimize the insulator design considering all operating conditions including dry, wet and contaminated. In order to design insulators suitably, a better understanding of the insulator flashover is required, as it is a serious issue regarding the safe operation of power systems. However, it is not always feasible to conduct field and laboratory studies due to limited time and money.
The desire to accurately predict the performance of insulator flashovers requires mathematical models. Dynamic models are more appropriate than static models in terms of the instantaneous variation of arc parameters. In this dissertation, a dynamic model including conditions for arc dynamics, arc re-ignition and arc motion with AC supply is first developed.
For an AC power source, it is important to consider the equivalent shunt capacitance in addition to the short circuit current when evaluating pollution test results. By including the power source in dynamic models, the effects of source parameters on the leakage current waveform, the voltage drop and the flashover voltage were systematically investigated. It has been observed that for the same insulator under the same pollution level, there is a large difference among these flashover performances in high voltage laboratories and real power systems. Source strength is believed to be responsible for this discrepancy. Investigations of test source strength were conducted in this work in order to study its impact on different types of insulators with a variety of geometries.
Traditional deterministic models which have been developed so far can only predict whether an insulator would flashover or withstand. In practice, insulator flashover is a statistical process, given that both pollution severity and flashover voltage are probabilistic variables. A probability approach to predict the insulator flashover likelihood is presented based on the newly developed dynamic model.