A New Six-Gas-Based Fault Prediction Graphical Technique in Dissolved Gas Analysis for Oil-Immersed Power Transformer

Abstract

Transformers are among the key equipment of electrical power systems for maintaining the continuity of transmission and distribution of power. Dissolved gas analysis (DGA) methods are accepted by utilities for condition monitoring and detecting incipient faults in transformers. The graphical DGA methods have gained popularity in the recent past. The three-gas and five-gas-based graphical methods are regularly used, and recently, seven-gas and four-gas methods have been proposed. This work is based on six gases, and a hexagon method (HexM) is proposed. The HexM is tested for the IEC TC 10 database for the detection of six types of faults. None of the graphical methods proposed so far have a region defined for normal conditions. In this method, a region in the hexagon is earmarked for normal aging, and samples from the published literature are used for testing it. The results of HexM are compared with generally used Duval triangles and pentagons and the recently proposed heptagon method (HepM) for fault-type interpretation. The suggested technique is tested and compared with graphical methods. Based on the test samples, the hexagon figure gives 93.6% accuracy, and the heptagon gives 84.9% accuracy.