Key takeaway: Reading dissolved gas analysis through energy — the Gibbs free energy of gas formation, or a "fault-energy" index — is not a field diagnostic TriboTech uses, and there is a good reason. The equilibrium branch is the historical ancestor of the IEC 60599 and Rogers gas ratios we already run, not a better alternative to them. The one branch with genuine merit, a fault-energy severity index, is a trending complement for online-monitored mineral-oil assets, not a fault-type classifier — and it is not standardised.

The question, answered up front#

Every so often a client or a paper asks whether dissolved gas analysis (DGA) should be interpreted through thermodynamics — the Gibbs free energy of gas formation, or the "total energy" contained in a fault — rather than through the Duval triangles and IEC 60599 gas ratios we actually run. It is a fair question, and the honest answer is no, not as a field diagnostic. What follows is the method, where it came from, and where it does have real value, with the reasoning laid out so the position is auditable rather than asserted.

Two branches that get conflated#

"Thermodynamic DGA" is used loosely for two quite different things, and keeping them apart is the whole point.

The first is Gibbs-equilibrium fault-typing. It computes the equilibrium proportions of the fault gases as a function of temperature and outputs a fault type and a fault temperature. Its family line runs from Halstead's 1973 model, through the Rogers ratios, to IEC 60599.

The second is a fault-energy or severity index. It weights the measured gas concentrations by the gases' heats of formation to produce an energy-like severity number, and outputs how serious a fault is and whether it is getting worse.

The first answers the same question as Duval and the ratios — what kind of fault, and how hot. The second answers a question the Duval methods structurally cannot, because they discard absolute concentration: how severe, and trending which way.

Where it comes from — and why that matters#

The foundation is a 1973 paper by W. D. Halstead, generally described as modelling how insulating oil decomposes into hydrogen and light hydrocarbons that reach chemical equilibrium, then computing the equilibrium gas mixture as a function of temperature. The historically important point is that the empirical methods we use descend from this reasoning. The Rogers ratio method was developed on the basis of Halstead's model (Nanfak et al., 2024), and the IEC 60599 gas ratios — CH₄/H₂, C₂H₂/C₂H₄, C₂H₄/C₂H₆ — with their temperature bands below 300 °C, 300–700 °C and above 700 °C encode the same temperature-versus-composition logic.

So "thermodynamic versus empirical" is a false choice for the ratio methods: the ratios are the field-calibrated distillation of the thermodynamics. That lineage runs to the ratios, not to the Duval triangles — the triangles are database classifiers, built deliberately to avoid the equilibrium assumptions.

Why it fails as a field diagnostic#

The central objection is simple: real faults are not at equilibrium. A fault is a localised hot-spot with steep temperature gradients in a heterogeneous oil, not a single reactant at a single temperature. The governing standard says so directly — it states that thermodynamic models drawn from the petroleum industry cannot accurately predict the gas compositions formed, because they assume ideal-gas or temperature equilibria that do not exist in actual faults (IEC 60599:2022, Clause 5.2). Laboratory work simulating thermal faults reaches the same conclusion (Wang et al., 2018).

Two further limits bite in practice. Every published energy weighting was derived for mineral oil, so it mis-ranks severity on esters and silicone, which decompose by different chemistry. And a cumulative-energy metric integrates gas production over time, which makes it sensitive to sampling interval and to stray gassing.

Where it genuinely has value#

There is one niche the energy approach earns: severity trending on online-monitored mineral-oil transformers. A cumulative combustible-gas-energy figure is easy to derive from gases an online monitor already reports, adds a physically motivated severity axis the Duval methods lack, and is naturally a trend signal — which is exactly where the published evidence is strongest (Dukarm et al., 2020; Mharakurwa et al., 2019). Used that way, it complements the standard toolkit. It does not replace it.

The perspective for advisory work#

The decisive fact is that the thermodynamic and energy approaches are not standardised as a diagnostic in any IEC, IEEE, CIGRE or ASTM document. That means neither can anchor a client-facing diagnosis — it could only appear as engineering judgement backed by journal papers, which is not how we work. And we do not need it. The durable part of the thermodynamic insight already lives in the IEC 60599 and Rogers ratios; fault type is handled by the Duval triangles and pentagons; and severity is handled by the IEEE C57.104 status tables and gas-generation rates — the very question a fault-energy index reaches for, by an established route instead of an experimental one.

Recommendations#

1. For fault typing, stay on the standardised toolkit — the Duval triangles and pentagons first, the IEC 60599 gas ratios as the field-calibrated cross-check.
2. For fault severity, use the IEEE C57.104 status tables and gas-generation rates.
3. Treat a fault-energy index as a watch-item, not a method to adopt — and only where you have high-cadence online DGA on a mineral-oil asset.
4. Never carry a mineral-oil energy weighting over to an ester or silicone fill.
5. If a vendor markets "thermodynamic" or "Gibbs-energy" DGA as a headline diagnostic, ask which branch it is and whether it is validated against a fault-case database. The honest answer is usually "trend signal," not "classifier."

Try it yourself: Classify your gas pattern with the Duval Triangle tool, then cross-check the fault type with the IEC 60599 ratio tool. If the two agree, you have the durable part of the thermodynamic insight — with the equilibrium assumptions already replaced by field data.

If you want a second opinion on an ambiguous gas pattern, get in touch — reading the result correctly is the work we do every week.

Sources#

• Halstead, W. D. (1973). A thermodynamic assessment of the formation of gaseous hydrocarbons in faulty transformers. Journal of the Institute of Petroleum, 59(569), 239–241.
• Nanfak, A., Samuel, E., Fofana, I., Meghnefi, F., et al. (2024). Traditional fault diagnosis methods for mineral oil-immersed power transformer based on dissolved gas analysis: Past, present and future. IET Nanodielectrics. doi:10.1049/nde2.12082
• Wang, X., Wang, Z. D., Liu, Q., et al. (2018). Dissolved gas analysis of thermal faults in transformer liquids simulated using immersed heating method. IEEE Transactions on Dielectrics and Electrical Insulation, 25(5). doi:10.1109/TDEI.2018.007158
• Dukarm, J., Draper, Z., & Piotrowski, T. (2020). Diagnostic simplexes for dissolved-gas analysis. Energies, 13(23), 6459. doi:10.3390/en13236459
• Mharakurwa, E. T., Nyakoe, G. N., & Akumu, A. O. (2019). Power transformer fault severity estimation based on dissolved gas analysis and energy of fault formation technique. Journal of Electrical and Computer Engineering, 2019, 9674054. doi:10.1155/2019/9674054
• IEC 60599:2022. Mineral oil-filled electrical equipment in service — Guidance on the interpretation of dissolved and free gases analysis.