What Are the Risks of Transformer Failure — and How Are They Assessed?

This article is Part 1 of a three-part series on transformer risk, reliability, and long-term asset management. Here we address the most common and preventable sources of transformer failure — including condition-based risk, paper insulation degradation, and dissolved gas analysis — and why a complete condition assessment requires more than routine oil testing. In Part 2, we will cover end-of-life planning, application risk, and how to develop a capital-ready reaction plan for aging or custom transformer assets.

Not all transformer failures are alike — and not all of them are unavoidable. While incidents like lightning strikes or downstream faults can trigger failures regardless of unit condition, the more insidious risks are the ones that develop slowly and are frequently overlooked: deteriorating paper insulation, moisture accumulation, acid buildup, and dissolved fault gases. These condition-based risks are also the most preventable. Yet across industrial facilities, transformer maintenance is too often driven by insurance requirements rather than a genuine commitment to system reliability. For any operation that depends on uninterrupted power — and most do — that distinction carries significant consequences. The financial exposure from a transformer failure extends well beyond the cost of the unit itself; business interruption losses routinely dwarf replacement costs. Understanding the full risk picture starts with understanding what actually causes transformers to fail.

Not All Transformer Failures Are the Same

Transformer failures generally fall into two categories: those triggered by external anomalies and those driven by deteriorating unit condition. External events — lightning strikes, downstream faults, electrical surges — can cause failures independent of how well a transformer has been maintained. These are root-cause failures in the traditional sense. But unit condition is a significant contributing factor that is too often excluded from the risk equation.

An overloaded transformer with well-maintained dielectric fluid and sound paper insulation is far more likely to survive that overload event than one that has been neglected. Condition does not cause the incident, but it determines whether the unit survives it. Treating external anomalies and condition-based deterioration as separate, unrelated risk categories leads to incomplete risk assessments and under-resourced maintenance programs.

Why Business Interruption Risk Is Underestimated

The insurance industry typically quantifies transformer failure risk as the cost to replace the failed unit. That figure, while significant, rarely reflects the true financial exposure. Business interruption losses — the cost of lost productive capacity during an unplanned outage — are frequently far higher than the transformer replacement cost itself.

For facilities running custom or long-lead-time units, this gap widens further. A rectifier with a 26-week lead time and no planned spare represents an entirely different risk profile than a standard unit with readily available replacements. Yet both are often evaluated against the same replacement cost metric. Any credible risk quantification must account for the full cost of downtime, not just the cost of the asset.

Condition-Based Risk: The Most Preventable Failure Mode

Of the risk categories that apply to transformer assets, condition-based risk is the most actionable. It is also the most commonly mismanaged. Facilities that conduct regular oil testing often conflate testing with maintenance — these are not the same thing. Testing reveals condition; maintenance addresses it. A program that tests diligently but acts inconsistently is not a maintenance program; it is a documentation program.

A complete condition assessment requires more than trended oil results. Electrical test data, mechanical inspection findings, infrared scanning, and dissolved gas analysis each contribute to a picture that oil chemistry alone cannot provide. In practice, many facilities have years of oil test history on a given unit but no electrical test baseline — a gap that significantly limits the diagnostic value of the chemical data.

The Role of Oil Processing vs. Paper Condition

Oil processing is a common maintenance response to declining oil quality, but it addresses only part of the problem. Moisture is a useful illustration: in most transformer insulation systems, significantly more moisture resides in the paper than in the oil. Processing the oil to reduce moisture content provides a temporary improvement, but within days or weeks, moisture migrates back from the paper into the oil. The same dynamic applies to acid accumulation.

This means that oil processing without addressing the paper condition is, at best, a deferral strategy. A maintenance program that measures success by oil cleanliness without evaluating paper condition is missing the asset’s primary failure mechanism.

Paper Insulation Is the Limiting Factor

Transformer experts disagree on many things. On this point, there is near-universal consensus: the condition of the paper insulation is the primary limiting factor in transformer life. Oil can be processed and reconditioned; paper cannot be replaced in service. Once paper insulation degrades beyond a recoverable threshold, the useful life of the unit is effectively determined.

Accelerated paper degradation is driven by four primary stressors: moisture, acid, overheating, and oxidation. Each of these weakens the mechanical and dielectric integrity of the insulation, reducing the transformer’s ability to withstand electrical stress. Critically, these stressors are not independent — moisture accelerates acid formation, acid accelerates thermal degradation, and overheating accelerates oxidation. A unit exposed to multiple stressors simultaneously deteriorates far faster than linear models would predict.

Why Older Transformers Carried More Margin

Transformers manufactured during the infrastructure buildout of the 1960s and 1970s were frequently overbuilt relative to their design specifications. That margin provided a significant buffer against suboptimal maintenance practices and periodic overloading. Many of those units are still in service — still humming along, as the saying goes — which has created a false baseline for what transformer reliability looks like under normal operating conditions.

Modern transformer design relies on computer modeling to meet specifications precisely, with far less overbuilt margin. Applying the same maintenance posture to a 2010s-era transformer that worked adequately for a 1970s-era unit is likely to produce worse outcomes. Facilities that have not recalibrated their maintenance expectations to account for this shift are operating with a reliability assumption that is no longer valid.

Furan Analysis as a Measure of Paper Aging

The most established method for assessing paper insulation condition is furan analysis. As cellulose insulation degrades, it releases furanic compounds into the oil. The concentration and composition of these compounds — and critically, the rate of change between successive tests — provide a reliable indicator of how aggressively the paper is aging.

Individual furan readings have diagnostic value, but trend data is more informative. A sharp increase in furanic compounds between tests indicates accelerating degradation and warrants immediate follow-up, including a review of loading history, cooling system performance, and oil chemistry. Furan analysis does not tell you how much life remains, but it tells you how fast that life is being consumed.

Dissolved Gas Analysis: Reading the Transformer’s Internal Condition

Dissolved Gas Analysis (DGA) is one of the most powerful diagnostic tools available for in-service transformer assessment. Gases form in transformer oil as a byproduct of normal aging, electrical stress, and thermal and mechanical anomalies. Because different fault types produce distinct gas signatures, analyzing the volume, composition, proportion, and rate of production of dissolved gases allows for a detailed diagnosis of what is occurring — or has occurred — inside the unit.

DGA is not a pass/fail test. It is a trend-based diagnostic that becomes increasingly valuable over time as a baseline is established and deviations from that baseline can be identified. A single DGA result, while useful, is far less informative than a series of results showing how gas concentrations are changing.

Fault Gases and What They Indicate

Gases associated with specific fault conditions — commonly referred to as fault gases — are produced by partial discharge, electrolysis, localized and general overheating, and arcing. Hydrogen is typically present across fault types, making it a useful broad-spectrum indicator. Other gases, including methane, ethylene, acetylene, and carbon oxides, are more diagnostically specific and help differentiate between thermal faults, electrical discharge, and paper degradation.

The relationship between gases is often as informative as individual concentrations. Established interpretation methods such as the Duval Triangle and Rogers Ratio use gas ratios to classify fault type and severity, providing a structured framework for moving from data to diagnosis.

Online Monitoring vs. Periodic Oil Sampling

Online DGA monitors are increasingly common on high-priority units — primary substation transformers and, in steel production environments, furnace transformers. These devices provide continuous or near-continuous gas detection and can trigger alarms when gas levels rise unexpectedly. However, most monitors are optimized for detection, not full characterization. They can confirm that a gas or gases are present; they cannot independently determine the rate of change across multiple gas types with the same precision as a laboratory oil sample.

A robust DGA program typically combines periodic laboratory oil sampling — which provides precise multi-gas quantification and rate-of-change analysis — with online monitoring on critical units for between-sample anomaly detection. Neither approach alone provides the full picture that the combination delivers.

Building a Complete Condition Assessment

A complete transformer condition assessment integrates data from multiple sources: oil chemistry, electrical testing, mechanical inspection, infrared scanning, and gas analysis. Each data stream addresses a different aspect of unit condition, and gaps in any one area limit the diagnostic value of the others.

Electrical test data — including power factor, insulation resistance, and sweep frequency response analysis (SFRA) — establishes a baseline against which future changes can be measured. SFRA in particular is valuable because a factory or installation baseline provides a reference for detecting winding movement or deformation that would otherwise be invisible to oil testing. Without that baseline, the comparison is made against population norms rather than the unit’s own history — a significantly less sensitive diagnostic approach.

The practical implication is straightforward: condition assessment programs should be designed to collect baseline data early and maintain consistent testing intervals so that trend analysis is possible. The value of the data compounds over time. A seven-year oil test history is more useful than a single result; a seven-year history combined with electrical test data and a documented SFRA baseline is more useful still.

Summary:

• Not all transformer failures share the same root cause. External anomalies and condition-based deterioration are distinct risk categories — but unit condition determines whether a transformer survives an external event.
• Business interruption losses typically exceed replacement costs. Risk quantification that accounts only for asset replacement underestimates total financial exposure, particularly for custom or long-lead-time units.
• Testing and maintenance are not the same activity. A program that tests consistently but acts inconsistently on results is not a preventive maintenance program.
• Paper insulation is the primary life-limiting factor. Unlike oil, paper cannot be reconditioned in service. Moisture, acid, overheating, and oxidation accelerate degradation — and interact with one another to compound the rate of deterioration.
• Modern transformers carry less design margin than legacy units. Maintenance strategies calibrated against the reliability history of overbuilt legacy transformers are not appropriate for current-generation units.
• Furan analysis tracks paper aging over time. Rate of change between tests is more diagnostically significant than any individual reading.
• DGA identifies fault type and severity through gas signature analysis. Trend data is essential; a single result provides limited diagnostic value without a baseline for comparison.
• Online DGA monitoring and periodic oil sampling are complementary, not interchangeable. A complete program uses both.
• A complete condition assessment requires data from multiple sources. Oil chemistry, electrical testing, mechanical inspection, infrared scanning, and DGA each address different failure mechanisms — gaps in any one area reduce overall diagnostic accuracy.

In This Series:

• Part 1: What Are the Risks of Transformer Failure — and How Are They Assessed?
• Part 2: How Do You Plan for Transformer End-of-Life Before a Crisis Forces Your Hand?
• Part 3: How Do You Build a Transformer Preventive Maintenance Program That Actually Works?