On April 5, 2023, at the Faculty of Chemical Engineering and Technology, University of Zagreb, our colleague Vedran Đurina, a collaborator in the Laboratory for Physical and Chemical Testing at our Laboratory Center, successfully defended his doctoral thesis on the topic: Assessment of physico-chemical properties of insulating oils by molecular spectroscopy techniques with use of chemometric methods and artificial neural networks.

Abstract

The goal of this thesis was to develop highly reliable computational models to assess the physico-chemical properties of insulating oils from their molecular spectra, providing data on oil aging and the presence of chemical contaminants. In addition to model development, the thesis explored the interrelationships between the physico-chemical properties of oils to enhance the understanding of the information obtained through testing. Significant spectral regions in the molecular spectra were analyzed, identifying signals that vary with changes in physico-chemical properties and determining the responsible chemical species.

Correlation curves were created based on the measured physico-chemical properties of oils, exploring the relationships between indicators of chemical composition, aging indicators, and the interdependence between these properties.

The findings suggest that spectroscopic assessment methods can effectively supplement existing laboratory techniques and contribute to the development of simple measurement devices for monitoring the physico-chemical properties of insulating oils in the field or during equipment operation.

Scientific Contributions

This research offers new insights into how oil aging affects molecular spectra, leading to a better understanding of spectral changes based on oil composition. Correlations between the physico-chemical properties of insulating oils and their molecular spectra were established. The developed computational models can predict key physico-chemical properties from molecular spectra, enabling non-destructive oil analysis.