Fault severity determination in transformers using dissolved gas analysis(DGA)

Abstract

Severity is the intrinsic property of transformer faults. DGA of gaseous species evolving from power transformers due to electric and thermal stresses distinguishes between types of faults. No information about severity is provided by the DGA based techniques. For the severity study of incipient faults, thermodynamics involved in their formation is considered. Energy weighted DGA (EWDGA) which uses energy differences between fault gases is used for severity prediction. In this paper, severity of faults detected by Duval triangle method has been evaluated. Prediction of fault severity by Duval triangle method limits the concept of EWDGA to EWMEA (Energy Weighted Methane Ethylene and Acetylene). EWMEA requires subset comprising of three of primary fault gases MEA(methane, ethylene and acetylene) to be energy weighed. Duval triangle also classifies the faults using the same set of gases besides showing least percentage of inaccurate results. Henceforth, EWDGA is applied to Duval triangle to further enhance its fault diagnosis capabilities. The paper also compares the unweighted and weighed counterparts of MEA gas combinations for various fault categories of Duval triangle method. Results show the higher sensitivity of these gases for high-energy faults. The concept of energy weighing has been implemented for Duval triangle by using fuzzy inference system (FIS). The FIS uses both unweighted and weighed inputs for detection of faults and their respective severities simultaneously.