Inside a sealed transformer, every overheated joint and internal spark leaves a chemical trace in the oil. Dissolved gas analysis of transformer oil reads those traces, revealing a unit's internal condition without taking it offline. It is the most powerful tool in a transformer oil analysis program, which also tracks moisture, acidity, and furans.
A few parts per million of the right gas is enough to flag a problem months early. What follows covers how DGA works, which gases signal which faults, how to read results against IEEE and IEC standards, and what to do when the numbers turn serious. For units already showing warning signs, those findings can also guide transformer repair decisions.
Dissolved gas analysis (DGA) is a diagnostic test that measures the gases dissolved in transformer insulating oil to detect electrical and thermal faults inside the unit. Each fault type produces a distinct gas signature, so the mix and amount of gas tell technicians what is going wrong and how serious it is.
Think of it as a blood test for a transformer. The oil carries evidence of internal stress, and the lab reads that evidence before the damage turns visible or catastrophic.
What is DGA in everyday maintenance? It is the primary early-warning tool for liquid-filled transformers. It catches developing faults that no external inspection would find, often months or years ahead of failure.
A single test gives a snapshot. A series of tests over time shows direction, which matters far more than any one reading. Stripped down, the meaning of DGA is one call: whether a unit keeps running, gets watched more closely, or comes offline for repair.
Heat and electrical stress break down two materials inside a transformer: the mineral oil and the cellulose paper insulation. As these materials degrade, they release gases that dissolve into the surrounding oil. This is why DGA should be treated as part of a broader transformer maintenance program, not a one-off lab report.
Normal aging produces small, slow amounts of gas. A fault produces gas faster, in different proportions, and at higher temperatures. The type of gas depends on how much energy the fault releases:
Reading those proportions is the whole point of DGA.
Transformer oil dissolved gas analysis follows three steps: pull a representative oil sample, extract and measure the dissolved gases in a lab, then interpret the results against known fault patterns and industry standards. The accuracy of the whole process rides on the quality of that first step.
Samples come from a sampling valve on the transformer, usually drawn into a glass syringe or a sealed metal cylinder that keeps air out. The right oil sample kit keeps the draw clean and the gas readings honest. Exposure to air is the enemy here. Atmospheric oxygen and nitrogen contaminate the sample and skew the gas readings, which sends interpretation in the wrong direction.
Sampling happens with the transformer energized and at normal operating temperature, since gas levels shift with load and heat. A cold or lightly loaded unit may hide a fault that shows clearly under full load. Drawing a live sample is qualified-personnel work under NFPA 70E, not a job to improvise.
Oil Sampling Checklist:
In the lab, the dissolved gases are separated from the oil and measured one at a time. Two extraction methods dominate: headspace and vacuum. Both feed the recovered gas into a gas chromatograph, which separates each component and measures its concentration in parts per million (ppm).
The result is a breakdown of every fault gas in the sample, reported as individual values plus total combustible gas. Those numbers become the raw data for interpretation.
DGA works because different faults burn at different temperatures, and temperature controls which gases form. A loose connection running hot produces a different signature than an arc jumping across failed insulation. The oil records both.
Three qualities give the test its value:
No other single test reveals that much from a sample drawn through a valve.
DGA measures seven key gases, each tied to a specific kind of fault. Five are combustible (hydrogen, methane, ethane, ethylene, acetylene) and two relate to paper insulation (carbon monoxide and carbon dioxide). Together they form the fault signature.
Hydrogen is the most common fault gas and the least specific. Nearly every fault produces some, which makes it a sensitive first alarm but a poor diagnosis on its own. High hydrogen with little else often signals partial discharge, a low-energy fault that erodes insulation over time. Rising hydrogen alongside other gases points to something more serious.
Methane forms during moderate overheating, typically below 300°C. On its own it suggests a thermal fault at lower temperature. Paired with hydrogen and ethane, it strengthens the case for a hot spot somewhere in the windings or connections.
Ethane appears with thermal faults in a slightly higher temperature range than methane. Climbing ethane usually means localized overheating that has not yet reached the severity that produces ethylene. As a middle marker, it helps track how hot a fault is running.
Ethylene signals high-temperature thermal faults, generally above 500°C, and rises further as the fault intensifies. Strong ethylene means a serious hot spot: an overloaded connection, circulating currents, or a failing joint. When ethylene dominates the gas profile, the unit is overheating somewhere it should not be.
Acetylene is the gas no one wants to see. It forms only above roughly 700°C, which means arcing. Any measurable acetylene deserves attention, and a sudden jump signals an active, high-energy electrical fault that fails a transformer quickly. Treat new acetylene as urgent.
Carbon monoxide and carbon dioxide come from the cellulose paper insulation, not the oil. Rising levels mean the paper is aging or overheating, which is harder to reverse than an oil-side fault. The ratio of CO₂ to CO helps separate normal paper aging from active overheating. Because paper insulation sets the limit on transformer life, these two gases carry long-term weight even when combustible levels look calm.
DGA does not measure faults directly. It infers them from gas patterns. The presence, proportion, and rate of specific gases map to recognized fault categories, each carrying different urgency.
The table below links common gas pairings to the fault they usually indicate, along with how urgently each one calls for a response. Treat it as a quick reference: the dominant gases narrow the fault type, while the rate of change decides how seriously to take it.
Thermal faults come from excess heat: overloaded connections, bad joints, circulating currents, or restricted cooling. The gas signature shifts with temperature. Methane and ethane lead at lower temperatures, then ethylene takes over as the hot spot climbs past 500°C. The hotter the fault, the more ethylene and the faster the damage.
Partial discharge is a low-energy electrical fault, small sparks inside voids, bubbles, or weak points in the insulation. Hydrogen is the main product, often with a little methane. On its own, partial discharge is rarely an emergency, but left alone it carves a path that grows into full arcing. Trend matters more than the absolute number here.
Arcing releases enormous energy in a short time, and acetylene is its fingerprint. High acetylene with hydrogen marks an active, dangerous fault: a breakdown between windings, a failing tap changer, or a short across insulation. Arcing damages a transformer fast. Results in this range usually justify pulling the unit for inspection rather than waiting for the next scheduled test.
When carbon monoxide and carbon dioxide climb, the paper insulation is breaking down. Heat, moisture, and oxygen speed the process. Paper damage is permanent and directly shortens transformer life, so a paper-side problem shifts the conversation from repair toward long-term replacement planning.
Not every abnormal result is a fault. Air leaks and moisture push up nitrogen, oxygen, and certain gas readings without any internal electrical problem. A sudden change in oxygen or nitrogen often points to a sealing issue rather than a failing winding. Ruling out contamination before declaring a fault saves a lot of unnecessary alarm.
Two standards anchor DGA interpretation worldwide: IEEE C57.104 in North America and IEC 60599 internationally. Both translate raw gas numbers into condition assessments, and most labs report against one or both.
The IEEE C57.104 DGA guide sets screening limits for individual gases and total dissolved combustible gas, then sorts results into levels that signal how concerned to be. Earlier versions used a four-condition scale tied to combustible gas totals. The 2019 revision moved to a status-based approach built on gas concentration percentiles and rate of generation:
The standard gives a defensible threshold, not a verdict. A reading above a limit means look closer, not condemn the unit.
The IEC 60599 interpretation guidance provides typical gas concentration values and a set of gas ratios that map to fault types: partial discharge, low and high-energy discharge, and thermal faults across defined temperature bands. It leans harder on ratios than on raw limits, which makes it strong for identifying fault type once a problem is flagged.
Standards give maintenance teams a common language and a baseline for comparison. They turn a page of ppm values into a condition code that engineers, managers, and service providers all read the same way. In practice, IEEE C57.104 and IEC 60599 sit alongside ANSI/NETA MTS, which sets the broader maintenance testing specifications for the unit as a whole. A standard will not make the call for you, but it frames the data so the call stays consistent and defensible across a fleet.
Interpreting transformer DGA analysis means reading three things at once: how much gas is present, how fast it is changing, and which gases lead the pattern. No single method answers everything, so experienced analysts combine several.
Absolute levels are the starting point. Each gas has typical and concerning concentration ranges, and a value well above normal flags a possible fault. Levels alone tell an incomplete story, though. A high but stable reading from an old fault matters less than a lower reading that is climbing fast.
Rate of change often tells more than the raw number. A gas that doubles between tests signals an active, growing fault even when it sits below a published limit. Generation rate, measured in ppm per day or per month, is one of the strongest signals in DGA. A flat trend is reassuring; a steep one is not.
One test is a data point. Several tests are a story. Comparing results across months reveals whether a fault is dormant, stable, or accelerating, which drives the maintenance decision more than any single sample. Consistent sampling conditions, same valve, similar load and temperature, keep the comparison honest.
The key gas method identifies the fault by which gas dominates the sample. High hydrogen points to partial discharge, high ethylene to overheating, high acetylene to arcing, high carbon monoxide to paper damage. Quick to apply and easy to read, it suits a first pass, though it struggles when several faults overlap.
Ratio methods compare pairs of gases to pin down fault type. The Rogers ratio uses combinations such as methane to hydrogen and ethylene to ethane to separate thermal faults from electrical ones. The Doernenburg ratio works on similar pairs, with a validity check to confirm there is enough gas to trust the result. Both excel at distinguishing fault types that absolute levels blur together.
The Duval Triangle plots three gases (methane, ethylene, acetylene) as percentages on a triangular chart, landing the result in a zone tied to a specific fault. The newer Duval Pentagon adds two more gases for finer resolution. Both stay popular because they are visual, hard to misread, and almost always return a fault classification, which makes them a reliable cross-check against ratio methods.
Several things distort DGA results: oil temperature at sampling, transformer load history, prior degassing, air or moisture ingress, even the type of oil. A past fault that was already repaired leaves residual gas that looks alarming on a fresh test. Sound interpretation accounts for the unit's history, not the lab sheet alone.
DGA is powerful but not complete. It flags that something is wrong and suggests what, yet it cannot pinpoint a fault's exact location or always gauge its full extent. When results raise concern, pair DGA with other transformer testing methods: furan analysis for paper condition, moisture and oil quality testing, power factor, and electrical diagnostics. The full picture comes from several tests, not one.
A concerning DGA result calls for a graded response, not panic. The right move depends on which gases are elevated, how fast they are rising, and how critical the transformer is. These paths run from closer monitoring to repair to replacement, and each one should line up with the right transformer service options.
Mild elevation with a slow or flat trend usually means watch, not act. Shorten the interval between tests, track the gases that moved, and confirm whether the change holds. Many flagged results settle into a stable pattern that needs nothing more than a closer eye.
A clear fault signature with a rising trend moves the unit toward planned repair. Thermal faults from loose connections, cooling problems, or tap changer issues often respond to maintenance once located. Scheduling a repair on your terms beats an emergency outage on the fault's terms.
Some results point past repair. Heavy acetylene from internal arcing, advanced paper degradation, or repeated faults in an aging unit push the decision toward replacement as the safer economic choice. The math weighs repair cost and downtime risk against lead time, budget, and available transformer inventory.
When results keep moving in the wrong direction despite retesting and load checks, the problem is likely real and physical. Persistent abnormal readings call for inspection, oil treatment, or a repair plan rather than another round of waiting.
Fast increases in acetylene, hydrogen, methane, or ethylene signal an active fault, electrical or thermal, that is progressing now. Quick generation shortens the timeline for action and usually triggers an urgent review.
High acetylene, a large jump in combustible gas, or several fault gases rising together point to serious internal damage. Signatures at this level often justify taking the transformer out of service for testing, repair, or a replacement evaluation.
The final decision balances several factors: transformer age, how critical the load is, repair cost, outage risk, transformer rental options, spare availability, and whether the fault is still progressing. A young unit with a fixable fault leans toward repair. An aging unit with paper damage and a long replacement lead time leans the other way.
DGA earns its place in a maintenance program because it changes outcomes. A small oil sample, read correctly, protects an expensive asset and an uninterrupted operation.
DGA comes in two forms: offline lab testing of manually drawn samples, and online monitors that track key gases continuously on the unit. Each fits different needs, and many fleets use both.
Offline lab testing delivers the fullest gas breakdown at a low per-test cost. It covers the complete set of fault gases and remains the standard for routine monitoring across a fleet. The trade-off is timing: a fault that develops between scheduled tests goes unseen until the next sample.
Online monitors watch key gases around the clock and raise an alert the moment a trend breaks. For a critical transformer where an unexpected failure would halt production or risk safety, continuous coverage closes the gap between scheduled tests. The cost runs higher, so the value depends on what the unit protects.
Match the method to the stakes. Routine, lower-risk units do well with periodic offline testing. Critical units, aging units, or any transformer whose failure would stop operations make the case for online monitoring, often layered on top of regular lab tests. Most strong programs blend the two: continuous watch on the units that matter most, scheduled testing across the rest.
DGA transformer testing follows three triggers: regular preventive schedules, response to abnormal events, and extra attention for aging or critical units. The right frequency depends on a transformer's importance and condition.
Healthy transformers get tested on a fixed schedule, commonly once a year for general units and more often for important ones. Routine testing builds the baseline and trend history that make every later result readable. Without a baseline, a single number means very little.
Any unusual event warrants a test: a through-fault, an overload, a protection trip, odd noise, or a temperature spike. Gas generated during the event reveals whether internal damage occurred. A post-event sample often catches problems that would otherwise stay hidden until the next routine test.
Older units and mission-critical transformers need tighter intervals, sometimes quarterly or continuous. Aging insulation generates more gas and fails less predictably, so the margin for surprise shrinks. The more a transformer matters and the older it gets, the more often it should be tested.
Good DGA depends as much on discipline as on chemistry. Consistent sampling, sensible frequency, and a regular oil-filled transformer maintenance routine separate useful data from noise.
Reliable results start at the valve. Sample under consistent conditions, keep air out, label every sample fully, and get it to the lab quickly. Drawing from the same valve under similar load and temperature each time makes trend comparison meaningful. A sloppy sample produces a misleading report, no matter how good the lab is.
Set frequency by risk. A practical baseline:
Move the interval up whenever results start to climb.
A few errors undercut DGA more than any others:
Each one turns good data into a wrong decision.
The value of DGA shows up in decisions, not data sheets. A consistent program of sampling, trending, and grading results against recognized limits buys the lead time to act before a fault forces an outage. Its accuracy compounds with history: a unit with years of clean baselines makes a new spike easy to read, while one tested for the first time offers a number with nothing to weigh it against. The harder call comes after the data, balancing the fault against the unit's age, its load, the cost of repair, and how fast a replacement could arrive.
Few teams make that call alone. When results point toward repair or replacement, H2LV can help connect the decision to sourcing, rentals, and quick-ship replacements, so a serious DGA finding does not turn into unplanned downtime.
DGA, or dissolved gas analysis, is a test that measures gases dissolved in transformer oil to detect electrical and thermal faults inside the unit before they cause failure.
Transformer DGA analysis is highly reliable for detecting and identifying faults, especially when trended over time. Accuracy depends on clean sampling and on pairing results with other diagnostic tests, like furan and moisture analysis.
Most transformers are tested annually, while critical or aging units need quarterly or continuous monitoring. Test immediately after any fault, overload, or abnormal operating event.
DGA cannot guarantee a prediction, but it detects developing faults early enough to act. Rising or rapidly changing gas levels are strong warning signs of approaching failure.
