Improvement and Evaluation of a DGA Based Rectangular Method for Enhanced Fault Classification in Power Transformers

This paper studied the rectangular method for classifying faults in power transformers based on dissolved gas analysis(DGA) data collected from Korea. Conventional international approaches, such as the Duval triangle and pentagon, have shown limited accuracy for Korean transformers due to environmental and operational differences. To address this issue, the proposed model normalizes the concentrations of four key gases and determines a representative location using the centroid of a constructed quadrilateral. Two model variations were evaluated against detailed inspection results, with the C H₂ 2;-based model reducing overlapping classifications by 11% and improving diagnostic reliability. Compared to traditional methods, the proposed approach demonstrates better alignment with inspection outcomes and offers enhanced applicability for transformer diagnostics in Korea. This method contributes to more accurate condition monitoring and asset management in power systems.