Transformer Oil Analysis and DGA Interpretation
We interpret analyses of transformer oil and the other insulating liquids: synthetic ester, natural ester and silicone liquid. You learn what the results mean for the transformer and what to do next.
Classic transformer oil is a mineral oil. Esters and silicone liquid behave differently and are interpreted against their own standards. The interpretation basis is chosen by the fluid, not by habit.
TriboTech is a consulting engineering company, not a laboratory. Ask us for an analysis programme, or send us your latest report for a second opinion.
Consultants, not a laboratory
We plan the analysis programme and supply the sampling kit and instructions. The analyses are carried out by our partner laboratories using documented laboratory procedures. We then interpret the results and advise on the next step.
The laboratory measures. The interpretation and the recommendation are what you buy from us.
Mineral-oil limits do not carry over to ester or silicone transformers. Applied unchanged, they can flag a healthy unit or miss a real fault. The background is in When the Fluid Changes the Diagnosis.
The analysis programme
We agree the programme per transformer, based on the fluid, the equipment type and its history. The standard analyses are DGA and fluid properties. Specialty analyses are ordered as add-ons.
DGA: standard for all four fluids
Dissolved gas analysis (DGA) covers ten gases: hydrogen, methane, ethane, ethylene, acetylene, propane, carbon monoxide, carbon dioxide, oxygen and nitrogen. The laboratory determines them by headspace gas chromatography.
Fluid properties: standard for mineral oil and esters
| Property | Mineral oil | Synthetic ester | Natural ester |
|---|---|---|---|
| Breakdown voltage | ✓ | ✓ | ✓ |
| Water content (Karl Fischer) | ✓ | ✓ | ✓ |
| Acidity | ✓ | ✓ | ✓ |
| Interfacial tension (IFT) | ✓ | ✓ | ✓ |
| Dielectric dissipation factor at 90 °C | ✓ | ✓, also at 25 °C | ✓, also at 25 °C |
| Resistivity | – | ✓ | – |
| Inhibitor content | ✓ | – | – |
| Viscosity at 40 °C | – | – | ✓ |
| Fire point | – | ✓ | ✓ |
Inhibitor content is measured on all mineral oil, inhibited or not. Colour and appearance can be added as a visual check. For silicone liquid, we agree the property programme with you.
Add-ons: specialty analyses
- Furans: markers of paper ageing in mineral oil.
- PCB: total content and indicator congeners.
- Potentially corrosive sulphur: the IEC 62535 test.
- Extended moisture assessment: estimated water content of the paper, derived from the oil's water content and temperature. Mineral oil only.
- Inhibitor content only: as a separate job. Mineral oil only.
How we interpret
DGA: four questions, in this order
- Can the result support a diagnosis? We check sampling quality and air contamination. At very low gas levels we describe the pattern rather than force a diagnosis. We also check whether the tap-changer compartment communicates with the main tank.
- Are gas levels and rates of increase normal for this kind of equipment? Under IEC 60599, gas formation below typical concentrations and typical rates of increase is normal gas formation, not a fault. IEEE C57.104 places the result in DGA Status 1, 2 or 3, based on gas levels, the change since the last sample and the rate of increase.
- What kind of fault? For mineral oil we use the IEC 60599 ratio method, Duval Triangle 1 for transformers and Triangle 2 for tap-changers, plus the Duval Pentagons from CIGRE TB 771. Esters are interpreted with IEC 63585, silicone liquid with the Duval Triangle in IEEE C57.146.
- Is the paper insulation involved? In mineral oil we look at carbon monoxide and carbon dioxide, and at furans where ordered. IEC 60599 warns that a localised paper fault need not produce significant CO, CO₂ or furans, so we never judge paper involvement on CO and CO₂ alone, and a low furan result does not clear the paper. In esters, IEC 63585 notes that there is no evidence linking CO and CO₂ to cellulose degradation, and the ester itself can form CO₂ as it ages, so we do not read these gases as a paper marker. In silicone, the liquid itself can form large amounts of CO and CO₂ when oxygen is present. Here we interpret on the basis of IEEE C57.146, which uses the absence of hydrocarbon gases to help tell overheated paper from overheated liquid, with furans as evidence of paper degradation.
IEEE C57.104 says explicitly that DGA status is not transformer condition. Our reports keep the two separate. Method choice by fluid is covered in Which Duval Method Should You Use?
Calculate it yourself: enter your gas values in the Duval Triangle, the Duval Pentagon or the IEC 60599 calculator.
Fluid condition
For mineral oil, IEC 60422 classifies oil condition by equipment category as "Good", "Fair" or "Poor", with a recommended action per property. Synthetic ester is assessed against IEC 61203 and natural ester against IEC 62975. For silicone liquid we use IEC 60944 and IEEE C57.166. We never apply mineral-oil limits to any of the three.
IEC 60422, IEC 61203 and IEC 62975 say the same thing: don't act on one result or one property; follow the trend. For silicone liquid, the same advice is in IEEE C57.166. Which limit table applies when is covered in Which Table Applies?
What is different about esters
Some natural esters produce ethane without any fault, depending on how unsaturated the ester is. That makes a single ethylene/ethane ratio hard to read, so IEC 63585 stresses following gas generation over time: a steadily rising ratio can point to an incipient thermal fault even before the ratio reaches the level the standard uses to indicate one.
Stray gassing can also affect ester DGA. It is gas produced by the fluid itself, its additives or construction materials, with no damaging fault. Where it is suspected, IEC 63585 recommends a laboratory stray-gassing test. It is not an ester-only phenomenon: IEC 60599 describes it for mineral oil too.
We use Duval Triangle 3 and Pentagon 3 with the caveat IEC 63585 itself states: there is not yet enough consolidated evidence from real cases that these fault types can be identified in esters.
The scopes are narrow. IEC 63585 excludes switching equipment such as ester-filled tap-changers. IEEE C57.146 applies only where silicone liquid was the original factory fill.
How it works
- Agree the programme. Before we deliver the first analysis or interpretation, we and you sign a framework agreement, and each task is agreed separately. The programme is set per transformer.
- Sampling kit and instructions. We agree the sampling kit and send sampling instructions. By agreement, we can also sample on site.
- Sampling. You take the sample, unless we have agreed to do it. IEC 60475 describes how the sample is taken. See also our oil-sampling guide.
- Shipment. You send the sample to the laboratory. Collection can be arranged by agreement.
- Analysis. Our partner laboratories analyse the sample.
- Interpretation and report. We interpret every sample in the light of the fluid, the equipment and its history, and issue the TriboTech report. It states the recommended action and the date for the next sampling.
Turnaround is typically 10–15 working days from receipt of the sample at the laboratory. Urgent transformer oil (mineral oil) analysis – DGA and fluid properties – can be delivered in 48 hours by agreement, subject to laboratory capacity. The binding deadline is agreed in the task agreement (ARP).
What you receive
The TriboTech report, per transformer and sample:
- status of the sample
- interpretation of DGA and fluid condition
- recommended action
- recommended next sampling date
- trend against previous results that we hold or that you send us
The report keeps the DGA status separate from our overall assessment. That is the difference between a traffic light and an engineering opinion.
Second opinion on another laboratory's report
We interpret analysis reports from other laboratories, whichever laboratory did the measuring.
It helps if you can send:
- all gas values in the units reported, including oxygen and nitrogen where measured
- sampling date, sampling point (main tank, tap-changer or gas relay) and the temperature at sampling, if recorded
- fluid type and product name, for example from the nameplate: the interpretation basis depends on the fluid, so this item carries the most weight
- transformer data: type, rating and voltage, year, sealed or free-breathing, and whether the tap-changer communicates with the main tank
- previous results and events such as trips, fluid treatment, degassing, top-ups or refills
If something is missing, we interpret what there is. The limits are these:
- We can only interpret what was measured. We cannot check the other laboratory's sampling or analysis. IEC 60475 stresses that even the most sophisticated analysis and diagnosis cannot make up for a faulty sample.
- A single result is a snapshot, not a trend. Rates of increase need at least two results.
- Changing laboratory breaks the trend. Part of any difference between two results is analytical scatter, and each laboratory has its own.
- If a parameter is missing, for example furans when paper involvement is suspected in mineral oil, we say what would close the question. You still get our interpretation of what exists.
Send us your report and we will come back with a proposal for interpreting it.
Who it is for
We work for clients in Danish electricity distribution, offshore and onshore wind, solar power and hyperscale data centres. IEC 63585 lists wind and solar transformers among the applications of ester fluids.
Independent interpretation
We don't sell insulating liquids, transformers or other equipment. Our recommendation rests on your data, the standards and our engineering judgement – not on equipment we have to sell.
Frequently asked questions
How often should transformer oil be analysed?
It depends on how critical the transformer is, its voltage class and what the last results showed. For mineral oil, IEC 60422 recommends intervals of 1 to 12 years for routine oil-quality tests, depending on equipment category. It shortens them when the oil is "Fair" or "Poor" or trends abnormally, and calls for the tests after filling, before energising. We set the interval per transformer in your programme.
What is DGA and what does it detect?
Faults in a transformer break down the insulating liquid and the paper into characteristic gases that dissolve in the liquid. DGA measures them. The gas pattern points to the fault type – partial discharges, low- and high-energy electrical discharges, or thermal faults in three temperature bands – and the rate of change shows whether it is active. A separate check looks at whether the paper insulation is involved. DGA does not see everything: as IEEE C57.104 notes, some pre-failure conditions generate no gas.
What does acetylene in transformer oil mean?
IEC 60599 explains that acetylene forms in significant quantities mainly in electrical arcs; much smaller amounts can also form in very hot thermal faults. But acetylene in the main tank is not automatically an internal fault. It can come from a tap-changer compartment that communicates with the main tank, or from non-fault sources such as new stainless steel or welding. The next step is usually a confirmation sample and a look at the trend and the tap-changer arrangement, not an immediate outage.
Can you interpret a report from another laboratory?
Yes. Send us the report with all gas values, sampling date, sampling point and fluid type, and previous results if you have them. We interpret what was measured, but we cannot check the other laboratory's sampling or analysis.
Do you analyse ester and silicone fluids?
Yes, through our partner laboratories. We deliver DGA for all four fluid families. Esters are interpreted with IEC 63585, silicone liquid with IEEE C57.146. Property programmes are defined for mineral oil and both ester types; for silicone liquid, we agree the programme with you.
How is an oil sample taken correctly?
Take it from a representative point, normally the bottom drain or the sampling valve. First flush the valve with enough fluid to clear it – typically 2–5 litres, while making sure that on small-volume equipment the total volume drawn does not endanger operation. Avoid letting air in, and make sure the fluid in energised equipment is not under negative pressure when you sample. Use a clean, gas-tight container, record the temperature at sampling, and have trained personnel do it. These are the elements IEC 60475 sets out; we send sampling instructions and, by agreement, can sample on site.
What does an analysis cost?
It depends on the scope and the analysis programme: DGA only, DGA plus fluid properties, or specialty analyses, and the number of transformers. Ask us for a quote.
How long does it take?
Turnaround is typically 10–15 working days from receipt of the sample at the laboratory. Urgent transformer oil (mineral oil) analysis – DGA and fluid properties – can be delivered in 48 hours by agreement, subject to laboratory capacity. The binding deadline is agreed in the task agreement (ARP).
Contact us
Tell us about your transformers and fluids, and we will propose an analysis programme. If you already have a report, send it to us and we will come back with a proposal for interpreting it.