A statistical analysis of metal oxide varistor current under distorted supply voltage conditions

Leakage current monitoring is still relied upon when on-site condition assessment or stability analysis of metal oxide varistors is to be conducted. The non-availability of adequate testing equipment, capable of directly measuring resistive component of varistor current, has paved the way to indirect diagnostic techniques which makes use of third harmonic component of varistor-arrester leakage current. The major shortcoming related to this method consists of high probability of measurement errors and subsequent misinterpretation of actual condition of varistor units should third harmonic-frequencies not be compensated in the supply voltage, prior to measurement. In this study, 90 identical low voltage varistor samples acquired from three leading manufacturers are subjected to distorted and non-distorted ac supply voltage. The magnitude of third harmonic currents observed in both measurement conditions, is deduced from the decomposition of time-domain leakage current waveforms into the frequency-domain using the flattop window algorithm of the Fast Fourier Transform. The large-sample statistical test of Hypothesis methodology is invoked to analyse the difference between the mean third harmonic currents of the varistor samples before and after harmonic injection in the supply. The average percentage of third harmonic injected in the supply voltage is 5%. However, the results obtained suggest over 95 % confidence that the measurement deviation in the third harmonic current-based assessment of varistor-arresters upon 5% third harmonic distortion in the supply voltage lies between 32%and 67%.