Fault Gas Evolution and Laboratory-Calibrated Duval Triangle Adaptation for Dielectric Nanofluids

The reliability of power transformers is strongly influenced by the behavior of insulating liquids under fault conditions. While mineral oil (MO) remains the industry standard, the introduction of dielectric nanofluids raises important challenges for conventional dissolved gas analysis (DGA)-based diagnostics. In this study, MO and its nanofluid formulations containing Al2O3, TiO2, and SiO2nanoparticles at concentrations of 0.02 g/L and 0.04 g/L were subjected to controlled thermal (T1, T2) and electrical discharge (D1) fault conditions. Fault-induced chemical responses were evaluated using DGA. The results demonstrate that nanofluids significantly intensify gas generation, producing up to four times higher total dissolved combustible gas (TDCG) levels compared to MO. Under thermal stress, nanofluids exhibited elevated CH4and C2H4concentrations and generated detectable C2H2, which was absent in MO, indicating accelerated degradation pathways even under mild overheating. During D1 discharges, nanofluids produced higher TDCG levels with markedly increased C2H2ratios. Application of the conventional Duval Triangle led to systematic misclassification of nanofluid faults, with shifting T1 cases to T2, 66% of T2 cases to T3, and 33% of D1 cases to D2, due to nanoparticle-induced gas amplification. To address this limitation, a modified Duval Triangle with adjusted %C2H4and %C2H2boundaries was proposed, enabling improved fault discrimination for nanofluid-based transformer insulation systems.