The answer first: DGA turns your transformer oil into a blood test — and a rising gas level is the earliest, cheapest warning you will ever get
Dissolved gas analysis (DGA) measures the gases dissolved in a transformer’s insulating oil to detect faults that are developing inside the tank — overheating, partial discharge, and arcing — typically months or years before they become a failure. The seven gases that matter are hydrogen (H₂), methane (CH₄), ethane (C₂H₆), ethylene (C₂H₄), acetylene (C₂H₂), carbon monoxide (CO) and carbon dioxide (CO₂); each fault type produces a signature mix, so one oil sample tells you what is happening, not just that something is wrong. The reading rules are simple: acetylene means arcing, ethylene means hot metal (a thermal fault), hydrogen means partial discharge, and CO/CO₂ mean the paper insulation is overheating. Most importantly, the trend beats any single number — a gas level that is rising fast is serious even while it is still below a “limit”, and it should trigger inspection rather than panic.
Key takeaways
- DGA is the highest-value oil test on an in-service transformer. One sample, run through a gas chromatograph, reveals arcing, overheating and partial discharge before protection relays ever trip.
- Each fault has a chemical signature. Acetylene (C₂H₂) = arcing; ethylene (C₂H₄) = thermal fault above ~500 °C; hydrogen (H₂) = partial discharge; CO/CO₂ = overheated paper insulation.
- Trend beats threshold. A gas value climbing quickly is serious even below a “limit”; a high but stable value on an old unit may be benign. Rate of change (ppm per day) is a primary indicator.
- Two standards govern the field. IEC 60599 and IEEE C57.104 define gas limits, total dissolved combustible gas (TDCG) condition levels, and interpretation methods (Rogers Ratio, Duval Triangle).
- Sampling quality is half the result. An air-free, syringe-drawn sample analyzed within days is the difference between a real diagnosis and a false alarm.
Why oil is the perfect sensor: how a transformer “talks” through its oil
A mineral-oil transformer generates gases for one reason — thermal and electrical stress breaks down the oil and the paper insulation. Heat decomposes the oil’s hydrocarbon molecules into simpler gases; an electric arc rips molecules apart at thousands of degrees; partial discharge (a tiny, repeated spark) slowly cracks molecules over months. The gases dissolve into the oil and stay there, so the oil becomes a running record of every stress event the unit has ever seen. That is why DGA — pulling a small oil sample and separating those gases in a laboratory gas chromatograph — can reconstruct what is happening inside the tank without ever opening it. It is standardized in IEC 60599 (interpretation of dissolved and free gas analysis) and IEEE C57.104 (interpretation of gases in mineral-oil-immersed transformers), which is why the same report means the same thing whether your unit is a QDTB oil-immersed transformer or one from another factory.
How to read the seven key gases
The seven gases fall into three families, and each family points at a different failure mode. Learn these three rules and you can read most DGA reports at a glance.
Acetylene (C₂H₂): the arcing fingerprint
Acetylene is the one gas to memorize. It forms only at very high temperature — roughly above 700–1000 °C — which in practice means an electric arc: a bad tap-changer contact, a broken conductor, or an internal flashover. Healthy oil contains essentially zero acetylene, so any measurable C₂H₂ is a strong warning and should be investigated. A transformer showing acetylene usually also trips its protection devices — but not always: a low-energy arc can persist below relay thresholds, which is exactly what DGA is for.
Ethylene and methane: the thermal-fault ladder
Ethylene (C₂H₄) forms in thermal faults — overheating above roughly 500 °C from a hot joint, circulating-current eddy losses, or a stray-flux hot spot. Methane (CH₄) and ethane (C₂H₆) form at lower temperatures (roughly 150–300 °C) and point at milder, slower overheating or an oil hot-spot. Think of it as a temperature ladder: methane/ethane first, then ethylene as the metal gets genuinely hot, then acetylene if an arc develops. The ratio of ethylene to ethane is one of the three ratios the Rogers method uses to place the fault temperature.
Hydrogen (H₂): the partial-discharge flag
Hydrogen is produced by partial discharge (PD) — small, repeated internal sparks in gas bubbles or voids that degrade insulation slowly — and it is also a by-product of corona and of some low-energy arcing. A high H₂ reading with little else points at PD or a local hot spot; because hydrogen is also the most mobile gas (it can migrate and even escape a loose gasket), it is worth re-testing before over-reacting to a hydrogen-only rise.
CO and CO₂: the paper-insulation alarm
Carbon monoxide and carbon dioxide come from cellulose — the paper insulation — not from the oil itself. A climbing CO/CO₂ ratio means the solid insulation (paper) is overheating or aging, which is the most serious signal of all, because paper does not regenerate: once it carbonizes, the transformer’s mechanical integrity is gone. This is the gas signature you want to catch early, since it is the same mechanism behind the insulation-life limits that govern transformer aging.
How the standards turn gas numbers into a diagnosis
Two interpretation frameworks do most of the work in the field. IEEE C57.104 classifies a unit into Condition 1 to 4 (normal, caution, concern, severe) using individual gas concentrations and the TDCG total; IEC 60599 supplies the Rogers Ratio and the Duval Triangle for pinning down the fault type. A good report gives you both: the condition level tells you how urgent it is, and the ratio/triangle tells you what kind of fault it is.
TDCG: one number that summarizes severity
TDCG (total dissolved combustible gas) is the sum of the six combustible gases — H₂ + CH₄ + C₂H₂ + C₂H₄ + C₂H₆ + CO. IEEE C57.104 sets the four condition levels against TDCG for oil-immersed units, on the order of Condition 1 (normal) up to ~720 ppm, Condition 2 ~721–1920 ppm, Condition 3 ~1921–4630 ppm, and Condition 4 above ~4630 ppm (verify against the edition your laboratory reports against). The condition level is your triage number: Condition 1–2 means keep sampling on schedule; Condition 3 means shorten the interval and plan inspection; Condition 4 means investigate now and consider taking the unit out of service.
The Rogers Ratio and the Duval Triangle
The Rogers Ratio codes three gas ratios — CH₄/H₂, C₂H₂/C₂H₄ and C₂H₄/C₂H₆ — each into bands (0, 1, 2) and maps the combination to a fault type such as “thermal fault <300 °C” or “discharge of low energy”. The Duval Triangle is the more visual method: it plots the relative percentages of CH₄, C₂H₄ and C₂H₂ on a triangle divided into zones — PD (partial discharge), T1/T2/T3 (thermal faults of increasing temperature), D1/D2 (low/high-energy discharge) and DT (mixed thermal and electrical). Both methods agree on the core idea that the presence and proportion of acetylene is the single most decisive datum in separating arcing from overheating.
Sampling and testing: how to get a result you can trust
DGA is only as good as the sample. The correct procedure is a syringe-drawn, air-free oil sample taken from the drain/sampling valve into a clean glass syringe (or gas-tight bottle), labelled and shipped so it is analyzed within days. Air leaking into the syringe or a contaminated bottle adds oxygen and nitrogen and skews the gas ratios; a sample left sitting for weeks can lose the most volatile gas (hydrogen) and under-report PD. Take the sample the same way every time — same valve, same technique — so that the trend from one test to the next is meaningful. The best baseline is a DGA run at commissioning, so every later reading can be compared against the unit’s own starting point rather than against a generic table — the same reason the commissioning tests should always log an oil sample. For larger or critical units, online dissolved-gas monitors can stream hydrogen and other gases continuously, catching a fast-developing fault between scheduled samples.
When to test, and how to act on the result
- Routine schedule. Critical or large units: DGA at least annually; new units get a baseline at commissioning and a re-test after the first weeks in service. Smaller distribution units can stretch to every 2–3 years unless readings change.
- After any abnormal event. Re-test after a through-fault, a protection trip, a suspected overload, or any operation of the Buchholz relay or pressure-relief device.
- Act on the trend, not just the value. A gas concentration rising by more than a few ppm per day, or doubling between tests, is grounds to shorten the interval and schedule an inspection — even if the absolute number is still inside Condition 1.
- Cross-check before you shut down. A single high reading can be a sampling or laboratory error; confirm with an immediate re-sample, and correlate with dissolved-gas ratios and electrical tests before deciding to de-energize.
- Document everything. Keep the full DGA history with each unit, because the trend line — not any one report — is what a buyer, insurer or warranty claim will ultimately be judged on.
Sources
Source: IEC 60599 — Mineral oil-impregnated electrical equipment in service — Guide to the interpretation of dissolved and free gases analysis
Source: IEEE C57.104 — IEEE Guide for the Interpretation of Gases Generated in Mineral Oil-Immersed Transformers
Source: IEC 60422 — Mineral insulating oils in electrical equipment — Supervision and maintenance guidance
Source: Industry experience — QDTB Engineering (sampling technique, trending thresholds, and testing-frequency guidance)