Winding turn-to-turn short-circuit diagnosis using FRA method: Sensitivity of measurement configuration

Anurag A. Devadiga, Noureddine Harid, Huw Griffiths, Naji Al Sayari, Braham Barkat, Sheshakamal Jayaram, Hisatoshi Ikeda, Tadashi Koshizuka, Yasuhiko Taniguchi

Research output: Contribution to journalArticlepeer-review

22 Scopus citations


Frequency response analysis (FRA) is increasingly being accepted as an effective technique to diagnose transformer faults. Transformer electric parameters are affected by such faults in a complex manner, and there is yet no standard approach for interpretation of FRA results. Most studies have focused on diagnosing winding and core deformations, but subtle defects in the winding insulation and turn-to-turn short circuits can develop into a more serious fault, and their early diagnosis is equally important. Furthermore, there are several test configurations which have different sensitivities to different faults. This study reports a study where the winding input impedance is measured to diagnose turn-to-turn short circuits using different measurement configurations. A comparison is made between the sensitivities of each measurement configuration to faults of increasing severity. It is found that this fault is detected in the low- and mid-frequency regions as significant reduction in impedance and a shift in resonance peaks towards high frequencies. The results, analysed using different statistical parameters, indicate differences in sensitivities with different levels of short circuits. Marginal variations were found between the sensitivities of statistical parameters in different frequency regions. The study provides useful contribution into interpretation of FRA signatures for turn-to-turn short-circuit diagnosis of transformers.

Original languageBritish English
Pages (from-to)17-24
Number of pages8
JournalIET Science, Measurement and Technology
Issue number1
StatePublished - 1 Jan 2019


Dive into the research topics of 'Winding turn-to-turn short-circuit diagnosis using FRA method: Sensitivity of measurement configuration'. Together they form a unique fingerprint.

Cite this