A dissolved gas analysis result becomes useful only when the gas pattern, its rate of change, the transformer’s loading history, and the sample quality are considered together. A single elevated value can be an early warning, a remnant of an old event, or an artifact introduced during sampling. A rising concentration or a sudden shift in the relative gas pattern deserves more attention than an isolated result without a baseline.

Inside an oil immersed medium voltage transformer, mineral oil and cellulose insulation decompose differently under heat, electrical discharge, and oxidation. The resulting gases dissolve in the oil and provide indirect evidence of conditions that cannot be seen from the outside. Hydrogen, methane, ethane, ethylene, acetylene, carbon monoxide, carbon dioxide, oxygen, and nitrogen are commonly reviewed together because their combination is more informative than any one gas alone.

Read the trend before assigning a fault

Start by comparing the latest result with earlier samples taken from the same unit. The sampling interval should be known, along with whether the transformer was energized, its approximate load, oil temperature, recent switching events, and any maintenance performed before the sample was taken. A result obtained after oil processing, topping up, transport, or a period of shutdown may not be directly comparable with an earlier energized sample.

The rate of gas generation often changes the maintenance decision. A moderate concentration that has remained essentially stable through several test cycles can call for continued observation. A lower concentration that is rising sharply within a short interval can indicate an active defect. Concentration establishes the condition to investigate; the trend indicates whether the condition is developing.

Do not calculate a generation rate from samples with uncertain collection dates, different test methods, or inconsistent oil volumes. A poor comparison can create a false alarm or conceal a real escalation. If a result conflicts with the transformer’s observed condition, obtain a carefully controlled confirmation sample before planning intrusive work, unless protection operation, abnormal sound, pressure activity, or other evidence points to an urgent internal fault.

What the main gases suggest

Hydrogen with little or no acetylene often suggests low-energy discharge rather than severe arcing. When acetylene appears alongside hydrogen and other hydrocarbon gases, the concern changes because the energy involved may be higher. That distinction affects the urgency of further tests, loading restrictions, and outage planning.

Methane, ethane, and ethylene need similar care. Methane and ethane are commonly associated with lower-temperature thermal faults in oil. A higher proportion of ethylene can indicate a hotter oil-contacting surface. However, a thermal pattern does not identify the exact component. Loose lead connections, degraded tap changer contacts, blocked cooling paths, magnetic flux-related heating, and circulating currents can all produce heat-driven gas generation. External inspection and operational records are needed to narrow the location.

Acetylene changes the level of attention

Acetylene deserves immediate scrutiny because it is generated at temperatures associated with arcing. Its presence should not automatically trigger an internal inspection, but it should prevent a casual interpretation of the sample. First establish whether acetylene was absent in prior tests, whether it is increasing, and whether a recent through-fault, tap changer operation, or switching event could explain the change.

The main tank and the on-load tap changer must be considered separately when their oil systems are separate. Tap changer operations can generate combustible gases in their own oil compartment. Mixing the diagnostic history of the tap changer with the main tank can wrongly suggest a winding or core fault. Confirm the sample point, oil circuit, and equipment configuration before comparing results.

Where acetylene is increasing in the main tank, review relay records, load events, bushing condition, lead connections, and any indication of intermittent internal discharge. Repeat DGA using a representative sample from the same location. If the repeat result confirms growth, further electrical and condition tests should be selected according to the suspected fault and outage feasibility. The objective is to determine whether the transformer can remain in service under observation or whether internal damage is continuing.

Carbon gases reveal insulation stress, but slowly

Carbon monoxide and carbon dioxide originate largely from cellulose insulation degradation. Their interpretation differs from hydrocarbon gas interpretation because paper and pressboard age over a long period, and the gases may be released gradually. A high carbon dioxide value in an older transformer can reflect accumulated insulation aging rather than a current hot spot.

A rising carbon monoxide level, especially when accompanied by a change in the carbon monoxide to carbon dioxide relationship, merits closer examination. The result becomes more concerning when oil temperature is normal, cooling equipment is functioning, and the trend still accelerates. Moisture data, furan analysis where available, dielectric tests, and thermal inspection records can help distinguish general insulation aging from localized cellulose overheating.

Do not use a carbon dioxide-to-carbon monoxide ratio as a stand-alone pass or fail rule. The ratio changes with oil preservation practices, gas loss through breathing, oil replacement, nitrogen blanketing, and the age of the paper. It is a supporting diagnostic indicator, not a substitute for a trend review.

Sampling errors that distort the diagnosis

A technically correct laboratory method cannot compensate for an unrepresentative oil sample. Air entering the syringe or bottle changes oxygen and nitrogen readings and can dilute dissolved gases. A valve that has not been flushed may contain stagnant oil from the sampling connection rather than oil from the tank. Leaking sample containers, extended transport in high heat, incomplete labeling, and residual cleaning solvent can also undermine the result.

• Use a clean, gas-tight sampling device appropriate for dissolved gas work, and verify that fittings and tubing do not draw in air.
• Flush the sampling connection sufficiently to remove stagnant oil while avoiding unnecessary oil loss or spills.
• Record the sample point, date, transformer load condition, oil temperature when available, and any unusual operating event before dispatching the sample.
• Keep repeat samples consistent: the same sampling valve and comparable operating state make the trend far more reliable.

Oxygen and nitrogen are particularly useful for judging sample integrity and tank breathing behavior. Unexpectedly high values can indicate air ingress during sampling, a leaking conservator system, or recent exposure of oil to air. They do not by themselves diagnose an internal electrical fault, yet they can explain why gas ratios appear inconsistent.

Use operating context to locate the likely source

DGA indicates fault energy and decomposition products; it rarely identifies a precise physical location on its own. A thermal gas pattern that rises only during high-load periods directs attention toward connections, cooling capacity, contact resistance, and current-carrying paths. A pattern that changes after tap operations points toward the tap changer circuit or related leads. Gases that increase after a short circuit event require review of mechanical displacement risk as well as electrical stress.

Backup or temporary supply arrangements can also alter loading patterns. When a site transfers load through an Open Type Diesel Generator Set, confirm whether the transformer experienced unusual load steps, frequent energization, voltage adjustment, or a changed duty cycle around the sampling date. These events do not explain every gas increase, but they are relevant when distinguishing a persistent internal defect from a response to exceptional operating conditions.

Thermal imaging of accessible terminations, cooling-system inspection, review of fan and pump operation, infrared scans of bushings and cable boxes, and tap changer maintenance records should be used to test the DGA hypothesis. The selected follow-up test should answer a specific question. Broad testing without a fault hypothesis often produces unrelated readings and delays the decision.

Turning a laboratory report into an action decision

Classify the result by urgency rather than treating every abnormal gas as an outage condition. Stable historical gases with no adverse operational evidence can remain under a shortened sampling interval. A clear increase in a thermal pattern calls for investigation of cooling, load, contacts, and localized heating sources. Rapid combustible gas growth, newly detected acetylene, protective relay activity, unexplained pressure changes, or abnormal noise justify prompt engineering review and a decision on load reduction or removal from service.

Keep the original laboratory reports, chromatograms when supplied, sample details, loading records, and corrective actions in one transformer history file. This prevents a recurring problem from being treated as a new event after personnel or contractors change. It also makes the next DGA result easier to interpret because the maintenance history explains whether a gas trend continued, stabilized, or responded after a repair.

DGA is most effective when it is treated as evidence that directs the next inspection, not as a diagnosis detached from the equipment’s operating history. A clean confirmation sample, a defensible trend, and targeted follow-up checks provide a stronger basis for maintenance action than any isolated gas value.