Dissolved gas analysis (DGA) is a common technique to identify transformer faults. The transformer faults are identified based on the concentration of the combustible gases such as Hydrogen (H₂), Methane (CH₄), Ethan (C₂H₆), Ethylene (C₂H₄), Acetylene (C₂H₂), and Carbon mono-oxide (CO) in addition to the ratios between these gases. On the other hand, the DGA technique identifies only the transformer fault type not the severity of this fault. In this paper, three objectives should be achieved, the first one is determining the transformer fault type based on the Duval triangle rules using a Fuzzy Logic model and secondly, determining the partial discharge severity based on a thermodynamic approach based on the starting decomposing materials (n-Octane (C₈H₁₈) and Eicosane (C₂₀H₄₂)), thirdly, comparing the severity of the partial discharge based on the type of the starting decomposing material. The results indicate that the severity of the partial discharge influenced by the type of starting decomposing material.