Influence of water and powder contaminants on dielectric strength of transformer oil

Abstract

Transformer failure statistics reveal that most of the transformer failures are due to insulation failures, where a considerable fraction of that percentage is caused by oil contamination. In this paper, the influence of water and powder contaminants on the electrical performance of transformer oil is investigated. The powder contaminant used is the fibrous dust obtained from the pressboard insulation of high voltage transformers. In a transformer, pressboard powder can easily contaminate the oil during the operation due to electrical and thermal stresses. Levels of powder contaminants selected for the research are 20mg, 40mg, and 80mg and added water levels were 1 ppm, 2 ppm, 3 ppm, and 4 ppm for an initial water level of 26ppm. Effect of contamination on AC breakdown voltage and visual streamer inception voltage of transformer oil was measured, and the results were statistically analyzed. This consists of Pearson correlation analysis, one-way multivariate analysis of variance and one-way analysis of variance. The correlation between the measured parameters of powder and water contamination levels was calculated with the Pearson correlation coefficient. The effect size which is a parameter that indicates how much the variance of the response (dependent) variable is described by the predictor(independent) variable was calculated using one-way multivariate analysis of variance and one-way analysis of variance methods. The results obtained from one-way multivariate analysis of variance showed an 86.1% effect size for the case of powder contamination. For the same case, with univariate one-way analysis of variance results, 88.8% and 96.2% effect sizes were calculated for breakdown voltage strength and visual streamer inception voltage respectively. For the water contamination case, one-way analysis of variance statistical method was utilized, and the effect size was calculated as 99.34% for the response variable (average of breakdown voltage and visual streamer inception voltage). Results of both cases made it evident that there is a significant effect of contamination on the dielectric strength of transformer oil. With a higher number of experimental results, the accuracy of the statistical analysis could be improved and with an increased number of response variables the analysis can be extended.