What it does
The Key Gas method classifies a fault by which combustible gas dominates the dissolved gas profile. Each of the four conventional fault types has a characteristic signature gas whose presence — as a fraction of total combustible gas — gives the fault away. The shape of the bar chart, not the ratios, drives the diagnosis.
The method is described in IEEE C57.104 and is the simplest of the standard interpretation methods — one glance at a profile chart often gets you to the right answer.
Total combustible gas (TCG)
Every threshold below is expressed as a percent of TCG, the sum of the six combustible species:
TCG = H₂ + CH₄ + C₂H₆ + C₂H₄ + C₂H₂ + CO
Below 20 ppm TCG the percent figures swing wildly on a few-ppm change, so the method abstains rather than highlight a tall bar that’s really just lab noise.
The four signatures
The classifier walks these in priority order — arcing first because acetylene is the most diagnostic and most urgent signal — and returns the first match. A flat distribution where no gas exceeds its threshold is reported as inconclusive.
| Signature | Fault | Key gas | Threshold |
|---|---|---|---|
| Corona | Partial discharge | H₂ | H₂ ≥ 60% of TCG |
| Overheated oil | Thermal fault in oil | C₂H₄ | C₂H₄ ≥ 40% of TCG |
| Overheated cellulose | Thermal fault in paper | CO | CO ≥ 60% of TCG |
| Arcing | High-energy electrical discharge | C₂H₂ | C₂H₂ ≥ 20% of TCG |
Each donut shows the minimum profile that matches a signature — the key gas at its threshold share of TCG, in the severity color used elsewhere in the app. Real samples usually push further past the threshold; these are the floor.
Corona
Partial discharge
Overheated oil
Thermal in oil
Overheated cellulose
Thermal in paper
Arcing
High-energy discharge
When to reach for it
Key Gas is the right tool for a fast read on what kind of fault is generating the gas. It answers “oil thermal vs. paper thermal vs. corona vs. arcing” without algebra — the dominant bar in the combustible profile is the answer. It also does work the ratio methods don’t: separating overheated cellulose (CO-led) from overheated oil (C₂H₄-led).
Where it earns its keep is in the cellulose call. Rogers and the Duval Triangle don’t use CO at all; Key Gas is often the first method to flag paper involvement when the carbon-oxide fraction of TCG climbs.
Strengths and limits
Strengths: intuitive, hard to misread once you’ve seen a few profile charts, and the only standard method that directly addresses paper degradation through CO. It works on absolute concentrations, not ratios, so it stays stable when one gas is at trace levels.
Limits: the thresholds are blunt — a sample with 35% C₂H₄ and 25% H₂ may be a real overheated-oil fault but won’t cross the 40% line. Mixed faults, where two mechanisms are running simultaneously, regularly produce flat distributions that Key Gas can’t resolve. And it can’t separate the thermal severities the way the Triangle’s T1/T2/T3 split or Rogers’ R5 ratio can. Use it alongside the Duval Triangle and Rogers Ratio, not in isolation.
References
- IEEE Std C57.104-2019, Guide for the Interpretation of Gases Generated in Mineral Oil-Immersed Transformers (Key Gas method).
- IEEE Std C57.104-1991, the earlier edition where the method was first formalised with the percentage signatures still in common use.