Multi-Spectral Analysis for Accurate Furfural Quantification in Transformer Oil as a Diagnostic Indicator for Aging Oil-Paper Insulation

Abstract

The precise determination of furfural content in transformer oil is pivotal for evaluating the aging state of oil-paper insulation and, consequently, facilitating effective diagnosis of power transformer health. In this study, a comprehensive approach was employed to advance the spectral detection options, integrating Raman, infrared, and ultraviolet spectroscopy. To establish a robust molecular simulation model for furfural, the Gaussian 09 W program was utilized, incorporating wave function and dispersion functions based on the 6-311G basis group. By comparing the absorption peak attributions from density-functional simulation calculations, we provide a theoretical foundation for selecting optimal detection spectra. Accelerated thermal aging tests were conducted on oil samples from oil-paper insulation, and Raman, infrared, and ultraviolet spectra were subsequently acquired. Spectral data underwent preprocessing using the polynomial least squares method. Considering sensitivity, detection limit, repeatability, and stability, Raman spectroscopy emerged as the optimal method for furfural detection, a prominent aging product in oil-paper insulation. Expanding on this, a quantitative analysis model for furfural detection in transformer oil was developed based on the linear relationship between Raman characteristic peak area and furfural concentration. The goodness of fit for this model was exceptionally high at 0.997, demonstrating its reliability and accuracy.