Why Power Transformers Fail — A Field Guide to Failure Modes and the Diagnostics That Catch Them
A field guide to power transformer failure modes — windings, tap changers, bushings, insulation and moisture — mapped to the DGA, PRPD, SFRA, oil and thermal diagnostics that catch each one early, anchored to IEEE, IEC and CIGRE standards.
On this page
A large power transformer is one of the most consequential assets on the grid and one of the hardest to replace. The U.S. Department of Energy estimates that roughly 90 percent of all electricity consumed in the United States passes through a large power transformer at some point between generation and end use. When one fails unexpectedly, the replacement is not on a shelf — lead times for large units now run to several years, as the 2026 Power Transformer Fleet Intelligence Report documents. That combination, high consequence and slow recovery, is why understanding how transformers fail, and detecting it early, is worth far more than the diagnostics themselves cost.
The good news for asset managers is that transformers fail in a small number of well-understood ways. Each mechanism leaves a signature, and each signature has a diagnostic designed to catch it. This guide maps the failure modes to the tests — and explains why serious condition monitoring uses a battery of complementary diagnostics rather than any single verdict.
How failures distribute across the transformer
CIGRE’s international transformer reliability survey, Technical Brochure 642, analysed 964 major failures over 167,459 transformer-years reported by 56 utilities across 21 countries, finding an average major-failure rate of approximately 0.53 percent per year for substation transformers and about 0.95 percent per year for generator step-up units. Those rates are low, but the survey’s more useful contribution is where failures originate. Failures concentrate in a handful of subsystems:
- Windings and core — mechanical, dielectric and thermal degradation of the active part.
- On-load tap changers (OLTC) — the one part with moving contacts, and a persistent leading failure location.
- Bushings — a comparatively small component whose failures are strongly associated with the most severe, often energetic, external consequences.
- Insulation and oil system — the slow degraders that set the asset’s true remaining life.
Knowing this distribution is what makes diagnostics purposeful: you are not testing at random, you are watching the subsystems that actually fail.
The five failure mechanisms — and their signatures
1. Winding movement and deformation
Through-faults and short-circuit forces impose large mechanical stresses on windings. Over time, clamping pressure can be lost and conductors can shift, deform, or buckle. Winding movement rarely announces itself in the oil; it is a mechanical change. Its signature is a shift in the transformer’s frequency-response “fingerprint,” which is exactly what sweep frequency response analysis is built to reveal — see SFRA and winding movement.
2. Partial discharge and dielectric breakdown
Small, localised electrical discharges in insulation voids, at surfaces, or around contamination erode dielectric strength long before a full breakdown. Partial discharge (PD) is both a symptom and a cause. Phase-resolved partial discharge (PRPD) analysis maps discharge magnitude and pulse count against phase angle over the AC cycle, and the resulting patterns distinguish corona, internal void, surface, and slot discharges from one another — a mapping validated in the peer-reviewed pattern-recognition literature. Measurement quality matters here as much as interpretation; see reading PRPD signatures.
3. Thermal aging of insulation
Cellulose insulation ages chemically with temperature. The IEEE C57.91 loading guide’s thermal model puts a number on it: aging rate approximately doubles for every roughly 7 °C rise in winding hot-spot temperature above the 110 °C reference for 65 °C-rise thermally upgraded insulation. Thermal aging is the strongest argument for continuous monitoring, because sustained overloading or degraded cooling consumes life silently — there is no oil “event,” just a shorter future. We unpack the mechanism in thermal loading and insulation aging.
4. Moisture and oil contamination
Moisture ingress and oil degradation lower dielectric strength, accelerate aging, and mask other signals. IEC 60422 governs the supervision and maintenance of insulating oil — moisture, breakdown voltage, acidity and contamination — and is the natural complement to gas analysis for tracking slow degradation and moisture markers.
5. Tap-changer wear and coking
As the only subsystem with moving contacts under load, the OLTC accumulates wear, contact deterioration, and coking. Its signatures show up in dedicated gas patterns, contact-resistance and motor-current trends, and dissolved-gas behaviour distinct from the main tank — a reminder that “the transformer” is really several assets sharing a nameplate.
Matching the diagnostic to the failure mode
Dissolved gas analysis is the sentinel test because thermal and electrical faults in oil-filled equipment generate characteristic gases before they become failures. IEEE C57.104 and IEC 60599 frame how those gases are interpreted, and the Duval triangle classifies fault type from gas ratios — walked through in how to interpret DGA results, which you can try on a sample with the interactive DGA Analyzer. An important honesty check: the Duval triangle classifies a fault, it does not by itself establish that one exists — you first determine that gassing is abnormal, which is a trend question, not a single-sample one. That is the whole argument for DGA trend analysis and, where the asset warrants it, continuous online DGA monitoring.
| Failure mechanism | Leading indicator | Primary diagnostic | Standard anchor |
|---|---|---|---|
| Winding movement / deformation | Change in frequency-response fingerprint | SFRA | IEEE C57.149 / IEC 60076-18 |
| Partial discharge / dielectric | PD activity, characteristic gases | PRPD, DGA | IEC 60270, IEC 60599 |
| Thermal aging of insulation | Hot-spot temperature, loss-of-life | Thermal / loading monitoring | IEEE C57.91 |
| Moisture / oil degradation | Moisture, breakdown voltage, acidity | Oil quality testing | IEC 60422 |
| Thermal / electrical faults (general) | Dissolved fault gases and rates | DGA (trend + online) | IEEE C57.104, IEC 60599 |
| Tap-changer wear | Contact resistance, dedicated gas pattern | OLTC diagnostics | IEEE C57.152 |
IEEE C57.152 is the useful umbrella here, because it frames diagnostic field testing as an integrated practice rather than a one-test verdict. No single measurement is decisive; the diagnosis lives in how the streams corroborate — or contradict — one another.
From periodic tests to condition-based decisions
The failure modes above are why the field has moved from calendar-based testing toward condition assessment. CIGRE TB 761 frames condition and health-index methods that connect diagnostic movement to asset decisions, rather than leaving each test as an isolated score. The maturity arc is clear: sample and test on a schedule to establish a baseline, then let the evidence — a rising gas trend, a shifted SFRA trace, a hot-spot excursion, a moisture step — decide what deserves attention now. That shift is the subject of condition-based versus time-based maintenance, and it depends on being able to explain why a health index moved, which is the point of an explainable health index.
This is exactly where GridAPM is designed to help. It does not replace transformer engineers or promise that every failure is foreseen. It assembles the DGA, oil, PD, SFRA, thermal, inspection and history evidence into a source-linked case — showing which streams moved, how fast, whether the measurement was reliable, and what is missing — so a qualified reviewer can decide whether to monitor, retest, inspect, plan an outage, or intervene, with every recommendation traceable to the measurement behind it.
Understanding failure modes is the first half of reliability. Turning their signatures into timely, well-documented decisions is the second. Request a GridAPM pilot to evaluate a focused condition-monitoring workflow against your own fleet’s evidence.
References
- CIGRE TB 642 CIGRE TB 642: Transformer Reliability Survey (WG A2.37)
- CIGRE TB 761 CIGRE TB 761: Condition Assessment of Power Transformers
- CIGRE TB 537 CIGRE TB 537: Guide for Transformer Fire Safety Practices
- IEEE C57.104 IEEE C57.104: Interpretation of Gases Generated in Mineral Oil-Immersed Transformers
- IEEE C57.152 IEEE C57.152: Diagnostic Field Testing of Fluid-Filled Power Transformers
- IEEE C57.91 IEEE C57.91: Guide for Loading Mineral-Oil-Immersed Transformers
- IEC 60599 IEC 60599: Interpretation of dissolved and free gases analysis
- IEC 60422 IEC 60422: Mineral insulating oils — supervision and maintenance guidance
- Phase-Resolved Partial Discharge pattern recognition, Sensors (2024)
Questions engineers ask
What is the most common way a power transformer fails?
There is no single cause. Failures cluster around a few subsystems — windings, on-load tap changers, bushings, and the insulation and oil system. Windings and tap changers are consistently among the leading failure locations, while bushings, though a small component, are a leading contributor to the most severe events. Because the mechanisms differ, utilities run a battery of complementary diagnostics rather than relying on any single test.
Which diagnostic detects which failure mode?
Dissolved gas analysis (DGA) and oil testing catch developing thermal and electrical faults and slow degradation through the oil; partial discharge and PRPD analysis characterise active insulation defects; sweep frequency response analysis (SFRA) checks the mechanical integrity of windings and core after through-faults or transport; and thermal or hot-spot monitoring tracks insulation loss-of-life. They answer different questions and are strongest used together.
How does heat shorten a transformer's life?
Cellulose insulation ages chemically as temperature rises. Under the IEEE C57.91 loading guide's thermal model, the insulation aging rate approximately doubles for every roughly 7 °C increase in winding hot-spot temperature above the 110 °C reference for 65 °C-rise thermally upgraded insulation. Sustained overloading or degraded cooling therefore quietly consumes remaining life.
Can condition monitoring predict a transformer failure?
Condition monitoring does not guarantee prediction. It detects developing stress earlier and with more context so qualified engineers can decide whether to monitor, retest, inspect, plan an outage, or intervene. The value is a faster, better-documented decision — not a promise that every failure is foreseen.
