Summary
Ensuring the reliability and longevity of power transformers requires accurate fault diagnostics and targeted maintenance strategies. Conventional diagnostic approaches often fail to detect early-stage faults effectively, necessitating a transition toward condition-based monitoring and data-driven analysis. This study presents a comprehensive diagnostic investigation of two generator step-up transformer failures at Samra Electric Power Company, Jordan, a multi-stage diagnostic framework was implemented to systematically investigate transformer failure case histories. Traditional diagnostic methods, including electrical measurements, chemical analysis, and mechanical inspection were applied to identify underlying failure mechanisms and guide structured testing and troubleshooting procedures.
Read more Read lessThe first case study involves a 132/14.5 kV, 139/185 MVA generator step-up transformer that exhibited signs of localized heating and progressive gas accumulation in the insulating oil.
Dissolved Gas Analysis (DGA) revealed elevated concentrations of hydrogen (H₂), methane
(CH₄), ethane (C₂H₆), and ethylene (C₂H₄). Duval’s Triangle diagnostics indicated a T3 fault signature, representing thermal faults occurring above 300°C. Subsequent internal inspection focused on mechanical and electrical interfaces, including the tank shunt, high-voltage wall structure, winding lead exits, and bushing terminals. Electrical diagnostics included winding resistance, tan δ, insulation resistance, and excitation current testing. Results confirmed insulation damage due to localized magnetic flux concentration, traced to a bent tank shunt and sharp metallic edges. These mechanical defects caused thermal stress accumulation, degrading the lead shield insulation and initiating partial discharge activity.
The second case study investigates a 420/13.5 kV, 135/150 MVA generator transformer affected by cellulose insulation degradation. The transformer’s Degree of Polymerization (DP) was estimated at 466, corresponding to approximately 61% of the paper’s remaining life based on results obtained from the oil laboratory analysis report issued by Morgan Schaffer - Doble company. Both DGA and furfural analysis revealed a 2-Furaldehyde (2-FAL) concentration of 119 ppb, signifying moderate to advanced paper degradation likely caused by overheating.
Diagnostic findings attributed this degradation to the absence of a nitrogen sealing system and conservator airbag, which allowed ingress of atmospheric oxygen and moisture. These environmental stressors, compounded by load cycling and thermal fluctuations, accelerated cellulose hydrolysis and chemical aging. This process increased the risk of gas bubble formation and dielectric failure under high-stress operational conditions.
Based on these findings, practical recommendations were formulated to mitigate identified failure risks. For magnetic flux leakage, the study advocates for secure, gap-free tank shunt assembly, rounded tank contours on the high-voltage side, and enhanced shielding with multilayer insulation. To control insulation aging, it recommends installing moisture barriers such as conservator bags, continuous monitoring of cellulose aging indicators (e.g., 2-FAL levels and
DP), and periodic vacuum dehydration to remove dissolved moisture. Additionally, the installation of online DGA systems, improvements in cooling system efficiency, and load management are advised to prevent thermal overstress. Implementing these measures is expected to markedly enhance transformer reliability and extend operational life.
The result of this study offers valuable insights into developing effective maintenance, inspection programs, and techniques to detect hidden failures and mitigating defects responsible for magnetic flux leakage and cellulose insulation degradation.
Additional informations
| Publication type | Session Materials |
|---|---|
| Reference | A2_11893_2026 |
| Publication year | |
| Publisher | CIGRE |
| Country | Jordan, Hashemite Kingdom of |
| Study committees | |
| File size | 1 MB |
| Price for non member | 30 € |
| Price for member | 30 € |
Authors
MASHAGBEH Yousef - SEPCO-Samra Electric Power Generating Co; ALBATTAT Saleh - SEPCO-Samra Electric Power Generating Co
Keywords
Deep-Reinforcement-Learning; Digital Twin; Fuel-Optimal Control