Transformer oil analysis — breakdown voltage (BDV), gas analysis (DGA), and maintenance limits (IEC 60422)

The guide on the Buchholz relay for oil-filled transformers briefly mentions gas analysis (DGA) as a follow-up test after a Buchholz relay has alarmed on a gas accumulation. Oil testing is, however, a much broader, periodic maintenance discipline in its own right, one that does not wait for a relay alarm but is specifically intended to signal an incipient degradation or fault at an early stage. This guide covers the three main components: the breakdown voltage test (BDV), gas analysis (DGA), and the maintenance limits for oil properties.

Breakdown voltage (BDV) per IEC 60156

The breakdown voltage test (BDV) measures the voltage at which electrical breakdown occurs between two electrodes with a standardized gap distance, under standardized conditions, in an oil sample. IEC 60156 describes the test setup and procedure (electrode shape, electrode gap, rate of rise of the applied voltage, number of repetitions, and how the results are averaged).

Breakdown voltage is essentially a measure of the presence of moisture, particles, and fibers in the oil: an oil free of these contaminants has a considerably higher breakdown voltage than a contaminated oil, even though the chemical composition of the oil itself is otherwise unchanged. A low BDV value therefore primarily points to a need for filtering, drying, or degassing the oil, and only secondarily to a problem with the oil itself.

Note: BDV is sensitive to sampling errors. A sample that has picked up moisture or air during sampling (for example due to an improperly dried sample bottle, or sampling during rainy weather without adequate shielding) produces an unwarrantedly low BDV value that does not reflect the actual condition of the oil in the transformer.

Gas analysis (DGA) per IEC 60599: Duval triangle and Rogers ratios

Thermal or electrical faults in an oil-filled transformer produce specific gases that dissolve in the oil: among others hydrogen (H2), methane (CH4), ethane (C2H6), ethylene (C2H4), acetylene (C2H2), carbon monoxide (CO), and carbon dioxide (CO2). Dissolved Gas Analysis (DGA) measures the concentration of each of these gases, and the ratios between the concentrations indicate the type and severity of a possible fault. IEC 60599 describes the interpretation methods, of which the two best known are:

The Rogers ratio method, which calculates three ratios between specific gas pairs (for example CH4/H2, C2H4/C2H6, C2H2/C2H4) and links this combination of ratios to a fault type: a low-temperature localized overheating, a high-temperature overheating, a partial discharge, or an arcing discharge (high-energy electrical fault).
The Duval triangle, a graphical method that plots the relative shares of methane, ethylene, and acetylene on a triangular diagram, where different zones correspond to different fault types. The Duval triangle is in practice often used as a complementary or alternative tool alongside the Rogers ratios, because it is less sensitive to borderline cases where the Rogers method does not produce an unambiguous classification.

Both methods are intended to distinguish, based on a single oil analysis, between normal, slow aging of the oil (which produces low concentrations of mainly carbon monoxide and carbon dioxide from normal cellulose aging of the solid insulation) and an actively developing fault that requires further investigation or intervention.

Trend analysis: development over time matters more than a single measurement

A single DGA measurement with concentrations below the usual alarm levels says little about whether an actively developing fault is present. More important than the absolute level of each gas is the trend: a gradual, steadily increasing concentration of a specific gas (or gas pair) between consecutive measurements points to an active process, whereas a stable level over multiple measurements more likely indicates an already-completed, historical event or normal, slow aging. For this reason, DGA is typically performed periodically (for example annually for larger power transformers, more frequently for critical units), so that a trend line can be built up rather than relying on a single snapshot.

Maintenance limits for oil properties per IEC 60422

In addition to BDV and DGA, a periodic oil analysis typically also measures other properties: moisture content, the acidity (neutralization) number (a measure of oxidation products), the dissipation factor (tan δ, a measure of dielectric losses), and the interfacial tension (a measure of the presence of polar oxidation products). IEC 60422 provides guideline values and maintenance limits for each of these properties, depending on the voltage class of the transformer and on the function of the oil in question (new oil, in-service oil). When a measured value exceeds a limit from IEC 60422, this gives rise to a specific maintenance action: filtering and drying for excessive moisture, regeneration or replacement for an excessive acidity number, and so on.

Practical relevance

For power transformers subject to a periodic maintenance program (as also covered in the guide on the Buchholz relay and in the guide on the oil/winding temperature indicator (OTI/WTI)), a combined oil analysis (BDV + DGA + IEC 60422 properties) provides a much more complete picture of the transformer's condition than any of these tests taken individually. A low BDV with normal DGA values points to a treatable moisture problem; a normal BDV with a rising acetylene concentration instead points to a developing electrical fault that BDV alone would not have signaled.

Common mistakes

Relying on a single DGA measurement without trend analysis — a snapshot below the alarm levels does not rule out an actively developing fault if the trend is not tracked.
Interpreting BDV as a direct measure of the chemical quality of the oil, whereas in practice a low BDV value often first points to moisture or particle contamination, and only after filtration does any underlying chemical deterioration become visible.
Not recognizing sampling errors as a possible cause of an abnormal BDV value, resulting in an unnecessary oil-treatment recommendation or a missed actual fault.
Applying only the Rogers ratio method or only the Duval triangle without using the other method as a cross-check in borderline cases where one method does not produce an unambiguous result.