CIGRE Technical Brochure 939 is the most recent edition of CIGRE's long-running global survey of AC power transformer reliability. It pools service and failure data from utilities worldwide to answer a deceptively simple question: how often do large transformers fail, in what ways, where in the unit, and how are those failures found? The result is a population-level reference that gives meaning to the failures and ageing patterns seen on any individual fleet.
What it covers
The brochure analyses a very large body of operating experience — hundreds of thousands of transformer-years drawn from dozens of utilities across many countries, covering high-voltage power transformers and reactors. It reports overall failure and retirement rates and breaks them down by voltage class, showing how the likelihood of failure climbs toward the highest-voltage units. It then dissects the failures themselves along several dimensions: the failure mode (electrical, dielectric, mechanical, thermal, and so on), the location within the transformer (windings, bushings, tap-changers, and other components), the underlying cause, and the method by which each failure was detected.
A particularly useful contribution is the comparison with the previous survey. Because CIGRE has repeated this exercise across decades, the brochure can show how the fleet has changed — which failure locations have grown in relative importance, how the overall hazard rate has shifted as the industry has invested in reliability, and where root-cause knowledge remains weak. It also notes the still-limited service experience for transformers in newer renewable applications, where the data is not yet sufficient for firm conclusions.
Why it matters in practice
Condition monitoring only earns its keep if it catches problems before they become outages, and TB 939 quantifies how much of that gap remains. The survey shows that the majority of failures are still discovered at the moment of trip, by protection devices, rather than in advance by diagnostic testing — and that advance detection, while modest across the whole population, becomes more significant at the highest voltage classes where monitoring investment is concentrated. That finding is a direct argument for the kind of disciplined oil-analysis and dissolved-gas programmes that shift failures from the trip column into the detected-in-advance column.
The brochure is also a benchmarking tool. When a unit shows an unusual result, the natural question is whether it reflects a common pattern for its age and voltage class or something genuinely anomalous. Population statistics turn that judgement from intuition into something defensible, and they help calibrate the weightings used in asset-health frameworks.
How we use it
We use TB 939 as the current population baseline behind our condition-assessment advice. It lets us frame a client's individual findings against what is typical for comparable equipment, justify the value of a structured oil and dissolved-gas monitoring programme in terms of the detection gap the survey documents, and prioritise attention toward components — bushings prominent among them — whose relative share of failures has been growing. Used this way, the survey grounds our recommendations in industry-wide evidence.