Abstract
The presence of polychlorinated biphenyls (PCBs) in transformer oils can influence the aging and deterioration process, posing a significant challenge in the electrical industry based on IEEE STD C57.104-2019. The standard provides guidelines for the interpretation of gases generated in transformer oils, specifically the presence of PCBs. An innovative methodology is proposed that integrates the detection of PCBs with the analysis of gases generated in transformer oils. The approach combines advanced gas chromatography and mass spectrometry techniques, along with a predictive analytics model, to improve accuracy in detecting and analyzing transformer aging. Dielectric oil samples from a 1250 kVA transformer were analyzed, obtaining concentrations of PCBs (Aroclor 1242, 1254 and 1260) below 2 ppm, below the permitted limit of 50 mg/kg. The results show that the concentration of methanol is an indicator of aging, ranging from 0.01 to 0.5 ppm, correlated with oil degradation. The predictive model achieved a sensitivity of 96.5% and a specificity of 94.8% in the detection of incipient faults, surpassing traditional methods such as the Dornenburg and Rogers rules (sensitivity: 85-90%). This study presents an innovative solution with significant implications for the scientific community and industry, improving the accuracy of PCB detection.
