Statistical analysis of non-standard overvoltage waveforms measured at 220 kV terminals of a power transformer

Power transformers are designed to withstand switching and lightning overvoltages which can occur in the power system. The knowledge about waveshapes of these events grows rapidly with the introduction of online transient monitoring systems. Consequently, it is important to compare standard test voltage impulses with the ones existing in the power system. In this paper, 123 measured overvoltages in 220 kV air-insulated substation are analyzed. The observed overvoltages are fitted using double exponential mathematical function. Function parameters are calculated using genetic algorithm. To compare measured signals with the standard ones, statistical analysis of their parameters is performed. This approach can help in better dimensioning of the transformer insulation in order to withstand the real overvoltages that can occur in the power network and in avoiding potentially dangerous situation due to the resonances. Additionally, frequency domain severity factor is calculated for the observed signals.