Influence of contamination distribution in characterizing the flashover phenomenon on outdoor insulator

• Published in: Ain Shams Engineering Journal Vol. 14; no. 12; p. 102249
• Main Authors: Ali Ahmed Salem, Rahisham Abd-Rahman, Mohd Taufiq Bin Ishak, Kwan Yiew Lau, Zulkurnain Abdul-Malek, Salem Al-ameri, Samir A. Al-Gailani, Sherif S.M. Ghoneim
• Format: Journal Article
• Language: English
• Published: Elsevier 01-12-2023
• Subjects: Artificial neural networks; Distribution; Flashover voltage; Polluted insulators; Pollution scenario
• ISSN: 2090-4479
• DOI: 10.1016/j.asej.2023.102249

The aim of this work is to model the influence of uneven contamination distribution under various humidity on the pollution flashover voltage of 11 kV porcelain insulator disc. Four scenarios of contamination distribution were proposed to test the sample under various severities of contamination simulated by salt deposit density (SDD). Series flashover experiments on contaminated insulators were performed under various conditions. The voltage of flashover under clean condition was appointed as a reference value for analyzing the effect of pollution. Based on the percentage value of breakdown voltage of the contaminated insulator to the clean insulator, the conditions of the tested sample are classified into three categories namely normal (55–60%), caution (45–54 %) and severe (35–44%). In the experimental tests, the uneven contamination area dimension was taken into consideration. An artificial neural network (ANN), derived from experiment results was used as a tool to predict the flashover voltage. The ANN method is built with five inputs related to the geometry of the sample and pollution factors while the flashover voltage was set as the model's output. The results showed that the distribution of pollutants according to the presented scenario has a significant impact on the performance of the flashover voltage. In addition, the error value between the experiment outcomes and the prediction system appeared to be less than 6%. This suggests that the proposed ANN model can be an effective tool in forecasting the insulators’ flashover voltage under test.