A Novel Offline to Online Approach to Detect Transformer Interturn Fault

Abstract

This paper presents an approach based on the use of transformer no-load and light load current harmonic analysis to detect the presence of an interturn fault at the incipient stage. The presence of fault is detected by performing a trend analysis of specific harmonic components and primary current magnitude. Transformer testing methods suffer from a common drawback of offline nature and require an expert opinion for accurate condition assessment. This paper presents a method that transforms no-load current harmonic analysis, which is a special test for transformer testing, into an online method at light load conditions. Thus, the necessity of complete shutdown and dependency on expert opinion is eliminated since it only requires one reference signature per phase. The proposed approach is verified for different vector groups of the three-phase core-type transformer and three independent units connected in star-star and for a fault on the low-voltage or high-voltage winding of the transformer. The presented experimentation shows that the proposed approach can detect the presence of a fault at an incipient stage involving less than 2% of turns and fault current up to 1 p.u. A case study is presented that validates the method for the transformer of a higher power rating.