Power transformer component reliability using CIGRE large-scale data surveys

Abstract

The probability of failure of a power transformer is difficult to quantify within a single utility because major failures are rare and operating histories are often incomplete. This paper uses large-scale CIGRE surveys (50 utilities, 26,533 transformers, 331,379 operating years) to estimate age-dependent component reliability by voltage class. In total, 1358 major failures and 991 retirements were reported for a reference period of up to 34 years. The data were treated as left-truncated and right-censored. Hazard rates were calculated for active parts, bushings, and on-load tap changers, retirements were assessed using the Kaplan-Meier estimator, and Weibull distribution models were fitted to 100-199 kV and 200-700 kV populations. The overall major failure rate was 0.41% per year. For the 100-199 kV transformers, the component hazard rates were close to constant with age (β ≈ 1), and the late-life pattern was influenced by increasing retirements. For the 200-700 kV transformers, bushing hazard showed a stronger age dependency and exceeded active-part hazard at around 50 years. The results highlight the value of component-focused risk management and show that fleet reliability should be interpreted alongside retirement and condition-management practices. Key limitations include data truncation, censoring, and the lack of categorisation of failures by technology type.