Transformer reliability survey

Introduction

Reliable service experience data for high-voltage transformers is essential for both utilities and manufacturers. It supports product improvement, asset management strategies, procurement decisions, maintenance planning, benchmarking, and overall power system reliability assessment. Statistical analysis of transformer failure data provides valuable insights into future failure behavior and contributes to improving international standards.

CIGRE Working Group A2.37 (Transformer Reliability), established in 2008, launched a comprehensive international survey to harmonize transformer failure data collection and deliver updated reliability statistics. This Technical Brochure presents the methodology, results, statistical analysis, and recommendations derived from the survey

The primary objectives of WG A2.37 were:

• Review existing transformer reliability surveys and data collection practices
• Conduct a new international transformer failure survey
• Compile, analyze, and classify failure data
• Calculate failure rates and hazard curves
• Standardize definitions and methodologies for future benchmarking

Review of transformer reliability practices and existing surveys

The brochure begins with a review of developments in transformer design, manufacturing, and technology, assessing their influence on long-term reliability trends. It also examines reliability terminology, statistical definitions, and recommendations found in the literature.

A key finding from the review of five national surveys and five company-level surveys is that methodologies differ significantly. National surveys typically focus on supply availability and interruption duration but often lack detailed information on:

• Failure location within the transformer
• Root causes and failure modes
• Repair actions and asset management implications

In contrast, internal company surveys provide more detailed equipment-level statistics, offering stronger value for asset management and performance benchmarking.

However, inconsistent definitions and reporting practices across surveys limited the ability to create a coherent international database. This gap motivated WG A2.37 to develop a standardized questionnaire for transformer failure data collection.

Standardized data collection methodology

To ensure comparability and enable meaningful statistical analysis, the Working Group developed a structured questionnaire divided into two major sections.

1. Population data

The first section collects general information about the transformer population under investigation, including:

• Transformer application (substation, generator step-up, etc.)
• Type and construction
• Number of phases
• Rated voltage and power
• Typical loading conditions
• Manufacturing period

This allows normalization of failure statistics based on transformer-years.

2. Failure data

The second section captures detailed failure information across four categories:

• Identification of the unit
• Technical features (cooling system, oil type, tap changer, overvoltage protection, etc.)
• Occurrence details (year of failure, service years to failure, loading before failure)
• Consequences (failure location, mode, cause, external effects, corrective action, detection method)

Definition of major failure

A “major failure” was defined as a failure requiring removal from service for more than seven days and involving significant remedial work, typically requiring:

• Opening of the transformer tank or tap changer
• Replacement of bushings
• Major factory repair
• Extensive oil processing in specific cases

This definition ensures consistency across international reporting.

Failure rate calculation

Failure rate was calculated annually using the standard formula based on:

• Number of failures in year i
• Number of transformers operating in year i
• Reference period (typically one year)

Because many utilities provided population data for only one year, transformer-years were estimated by assuming a constant operating population across the reporting period.

Survey scope and population

The international survey collected 964 major failures occurring between 1996 and 2010. These failures were recorded across:

• 58 utilities
• 21 countries
• 167,459 transformer-years

Manufacturing years of surveyed transformers ranged from the 1950s to 2009.

Substation transformers

For substation transformers, the dataset included:

• 22,181 transformers
• 150,072 transformer-years
• 799 major failures
  A2.37 Electra eng

The overall failure rate was approximately 0.53% per year, with variation by voltage class.

Generator Step-Up (GSU) transformers

For GSU transformers:

• 1,703 transformers
• 17,387 transformer-years
• 165 major failures
  A2.37 Electra eng

The overall failure rate was approximately 0.95% per year, higher than substation transformers, particularly in voltage ranges between 200 kV and 500 kV.

The substation transformer population (150,072 transformer-years) was nearly four times larger than earlier international surveys, providing a robust statistical foundation.

Failure location analysis

Failure data were analyzed according to the primary component where the failure originated. Generator step-up transformers without tap changers were excluded from certain comparisons to avoid distortion.

Voltage class trends

Up to 700 kV, key trends were identified:

• Bushing-related failures increased with voltage level
• Lead exit failures also increased with voltage
• Tap changer failures decreased as voltage increased

Substation vs. GSU transformers

Significant differences were observed between applications:

• Winding failures were the largest contributor in both applications
• GSU transformers showed higher proportions of:
  • Winding failures (48%)
  • Lead exit failures (13%)
• Substation transformers showed higher tap changer failure contributions (31%)
• Bushing failure contributions were similar in both applications

Winding-related failures frequently resulted in scrapping of the transformer due to severity.

Failure modes and consequences

Across applications and manufacturing periods:

• Dielectric failures were the most significant failure mode
• Winding failures were the dominant contributor to major failures
• Bushing failures were often associated with severe external consequences, including fire and explosion

Hazard curves were calculated for selected populations, including treatment of truncated data. No distinct ageing pattern was identified in the analyzed populations, suggesting relatively stable failure probability over service life within the studied range.

Conclusions and recommendations

The CIGRE WG A2.37 Transformer Reliability Survey provides one of the most comprehensive international datasets on transformer major failures.

Key findings

• Overall major failure rate below 1% per year
• Winding failures are the leading cause of major transformer outages
• Bushing failures are associated with the most severe external effects
• Dielectric failure modes dominate across transformer types
• GSU transformers exhibit higher failure rates than substation transformers

Standardization for future benchmarking

To simplify and harmonize future surveys, the Working Group recommends:

• Use of the standardized Excel-based questionnaire developed by WG A2.37
• Immediate recording of failure data following root cause analysis
• Consistent classification of failure location, cause, and mode
• Adoption of uniform definitions for major failures

Standardized data collection enables reliable benchmarking between utilities, supports asset management decision-making, and improves long-term transformer reliability assessment worldwide.