What Kills a Transformer: Failure Physics Every Procurement Manager Should Understand

The phone call no procurement manager wants to receive

Eighteen months after commissioning. The transformer that passed factory acceptance testing, arrived without visible damage, and cleared commissioning is now offline. The project is losing revenue by the hour. The utility is demanding answers. And somewhere in a folder of documents you signed off on, there is a number—a single measurement—that predicted this failure.

You just didn't know which number to look for.

Transformers don't fail randomly. They fail because of physical processes that begin long before the first alarm trips. Most of these processes leave traces in the factory test report. The problem isn't that the information is unavailable. The problem is that procurement managers are rarely trained to read it.

This article is the first in a series on what actually kills transformers—and what you, as a buyer, can do about it before the purchase order is signed.

The bathtub curve: why "it passed FAT" is not the end of the story

Power equipment follows a well-documented reliability pattern known as the bathtub curve. It has three phases:

Phase 1 — Infant mortality (0–2 years). Manufacturing defects that survived factory testing. Poor winding tension. Incomplete drying. Contaminated oil. A transformer that enters this phase typically fails within the first two years of service.

Phase 2 — Useful life (2–20+ years). The transformer operates with low failure probability. How long this phase lasts depends almost entirely on decisions made during design and procurement—winding temperature margin, moisture control during manufacturing, conductor mechanical properties.

Phase 3 — Wear-out (15–30+ years). Insulation materials reach end of life. The cellulose paper that separates conductors loses mechanical strength. The oil oxidizes. Partial discharge activity increases. A transformer that reaches this phase late—after 25 or 30 years—has returned the investment it was designed for. One that enters this phase at year 8 is a procurement failure.

The difference between a transformer that operates for 25 years and one that fails at year 8 is not luck. It is physics. And the physics is visible in the test data.

The five killers: what actually destroys a transformer from the inside

Killer 1: Dissolved gases — the earliest warning system

A transformer in operation generates gases. Normal operation produces trace amounts. Abnormal conditions produce specific gases in specific ratios. A thermal hot spot generates ethylene (C₂H₄). Overheated cellulose generates carbon monoxide (CO). Partial discharge generates hydrogen (H₂). An internal arc generates acetylene (C₂H₂)—the gas that should never appear in a healthy transformer.

Here is the critical point: dissolved gas analysis (DGA) performed at the factory, before shipment, establishes the baseline. If the baseline shows elevated hydrogen or—far worse—any acetylene, the transformer has a problem that routine electrical tests may not detect. Energizing it doesn't fix the problem. It guarantees the problem will grow.

IEC 60599 and IEEE C57.104 provide interpretation frameworks for DGA results. A procurement manager doesn't need to memorize gas ratios. They need to know one thing: ask for the DGA in the FAT report, and if acetylene is present, require investigation and written engineering disposition before accepting the unit.

Read the deep dive: DGA: The Blood Test Your Transformer Should Pass Before Shipment

Killer 2: Partial discharge — the invisible destroyer

Partial discharge (PD) is exactly what it sounds like: an electrical discharge that does not completely bridge the insulation between conductors. It is measured in picocoulombs (pC)—trillionths of a coulomb. A single PD event is microscopic. A PD event repeating thousands of times per second, every day, for years, carves channels through solid insulation like water through limestone.

IEC 60076-3 sets a limit of 100 pC for distribution transformers. But here is what the standard doesn't say: a transformer measuring 85 pC at the factory is legally compliant. It is also statistically more likely to fail within five years than one measuring 30 pC.

PD testing is not required by every procurement specification. Some buyers skip it because the equipment is expensive or the test adds days to the production schedule. That decision saves time and money at the factory. It transfers risk to the project site.

Read the deep dive: Partial Discharge: The Silent Defect Hiding in Your FAT Report

Killer 3: Moisture — the accelerator

The cellulose paper that insulates transformer windings is approximately 90% cellulose. Cellulose is hygroscopic—it attracts and holds water molecules from the surrounding environment.

During manufacturing, if the core and windings are assembled and oil-filled without complete removal of residual moisture, that water stays trapped inside the sealed system for the life of the transformer. It doesn't evaporate. It doesn't drain. It catalyzes the chemical reactions that break down cellulose molecules, one polymer chain at a time.

The effect is quantified in the technical literature: moisture content of 2% in solid insulation reduces expected life by approximately 40%. At 4%, the reduction reaches roughly 70%. These are not marketing numbers. They are consequences of the Arrhenius equation—the same rate law that governs chemical reaction kinetics across all of materials science.

The moisture content of oil at the factory, measured by Karl Fischer titration according to IEC 60814, should be no higher than 20 ppm for new oil. A value below 15 ppm indicates a controlled drying and filling process. A value above 30 ppm should raise questions.

Read the deep dive: Moisture: Why Water Destroys Insulation Faster Than Overload

Killer 4: Short-circuit forces — the structural test

When a fault occurs on the network, the transformer windings experience electromagnetic forces proportional to the square of the fault current. These forces can reach several tons per meter of conductor length. They compress, stretch, and bend the windings in milliseconds.

A winding that survives one fault may not survive the tenth. Each event produces microscopic deformation. Over years of service—and distribution networks experience faults regularly—the cumulative deformation creates spots where insulation is compressed thinner, where the electric field concentrates, and where partial discharge eventually initiates.

The transformer's ability to withstand these forces depends on design choices: conductor material properties, winding tension control during manufacturing, and the mechanical bracing of the coil assembly. These are not tested by routine electrical measurements. They are verified by type tests—specifically, the short-circuit withstand test—performed on a representative design.

If you are buying a transformer design that has never passed a short-circuit type test from an independent laboratory, you are accepting an unknown risk on behalf of your project.

Read the deep dive: Short-Circuit Withstand: The Test Most RFQs Forget to Ask For

Killer 5: Thermal aging — the cumulative enemy

The Arrhenius equation tells us that the rate of a chemical reaction approximately doubles for every 8–10°C increase in temperature. The chemical reaction that matters inside a transformer is the thermal decomposition of cellulose insulation. Every hour the winding operates above its design temperature, the insulation loses a fraction of its remaining life. The loss is cumulative and irreversible.

IEC 60076-2 defines temperature rise limits for transformer windings. But the standard defines the maximum allowed, not the optimal design target. A transformer designed to operate at 55°C winding rise will age significantly slower than one designed for 63°C—even though both are compliant.

The procurement implication: two transformers with identical kVA ratings, identical nameplate data, and identical price tags can have fundamentally different life expectancies because of a design temperature difference that never appears on any specification sheet you compare. The only way to know is to ask for the winding temperature rise from the FAT report—as a number, not a pass/fail.

Read the deep dive: Insulation Aging: Why Two Identical Transformers Age at Different Speeds

The common thread: everything that kills a transformer leaves a trace before shipment

Every failure mechanism listed above—gas generation, partial discharge, moisture, structural weakness, thermal margins—can be detected before the transformer leaves the factory. The tests exist. The standards define them. The equipment to perform them is available.

The gap is not in testing capability. The gap is in procurement specifications.

Most RFQs ask for routine tests: winding resistance, ratio, impedance, applied voltage, induced voltage. These tests confirm that the transformer is not already broken. They do not predict whether it will fail in service.

The tests that predict service life—DGA, partial discharge, moisture content, temperature rise curves at actual site conditions, short-circuit type test certification—are often optional. They cost more. They take more time. They are the first items removed when a buyer is comparing prices and needs to close a gap.

Removing them doesn't save money. It defers costs to the project site, where they compound with downtime, liquidated damages, and reputational damage.

What to do next

If you are evaluating transformer suppliers, here are three actions you can take before your next purchase order:

Request the DGA baseline. Ask for dissolved gas analysis as part of the routine test package. If acetylene (C₂H₂) is present in any concentration, ask the manufacturer to explain why—in writing.

Ask for PD values, not just PD pass/fail. A "pass" at 95 pC and a "pass" at 25 pC are not the same transformer. Ask for the actual measurement.

Compare winding temperature rise, not just nameplate kVA. Two 150 kVA transformers are not equal if one operates 8°C hotter than the other for the next 25 years.

This series continues with five deep dives into the physics of transformer failure—and the procurement actions that prevent it.

→ Contact us to discuss your transformer requirements with our technical team

References: IEC 60076-2 (Temperature rise limits, Arrhenius aging rate), IEC 60076-3 (Insulation levels, partial discharge ≤100 pC), IEC 60156 (Oil dielectric breakdown voltage), IEC 60814 (Water in oil by Karl Fischer), IEC 60599 / IEEE C57.104 (DGA interpretation). Bathtub curve reliability model for power equipment. Arrhenius equation applied to cellulose insulation degradation kinetics. Cellulose paper composition (~90% cellulose) and hygroscopic properties.

Related Transformer Failure and Procurement Guides

Procurement Action

If you are comparing transformer suppliers, send voltage, kVA, application, country, utility requirements and any FAT or test requirements. TransformerGrid can help review the procurement risk before the order is placed.

What Kills a Transformer: The Physics of Failure Every Procurement Manager Should Understand