Correlation between dissolved gases and oil spectral response

Spectroscopy is a powerful non-destructive technique that utilizes electromagnetic radiation interaction effect to determine the energy level and structure of atomic or molecular substance. Spectroscopy technique has been successfully used to quantify some power transformer condition monitoring parameters such as Furan and oil interfacial tension value. This paper investigates the feasibility of extending the applications of this technique to identify and quantify various dissolved gases in the transformer oil. In this context, several transformer oil samples with various gases concentrations have been prepared. Dissolved gases in each sample were first quantified using gas chromatography-mass spectrometry (GC/MS) laboratory measurement technique. The spectral response of each sample was then measured using UV-NIR and FTIR spectroscopy. Correlation between oil spectral response and the concentration of each gas is realized using adaptive neuro fuzzy inference system (ANFIS).