IEC 60599 DGA: Transformer Fault Diagnosis and Interpretation

Author: Leon Zhang — Power Monitoring and Testing Equipment Team, Baoding Houfeng Trading Co., Ltd., specifying and commissioning transformer DGA monitoring systems for utility and industrial fleets
Technical review: reviewed by the in-house power monitoring team against IEC 60599:2022 (edition 4.0) and IEEE C57.104-2019, covering the full IEC 60599 DGA interpretation workflow described below
Last updated: 19 October 2026

Key takeaways

• IEC 60599:2022 is an interpretation framework, not a measurement method: it tells you what dissolved gas data means, while IEC 60567 covers how to sample and analyze.
• The three-ratio fault table (Table 1) uses C₂H₂/C₂H₄, CH₄/H₂ and C₂H₄/C₂H₆ to point to PD, D1, D2, T1, T2 or T3 — a result that fits no row is not a diagnosis, it is a prompt to resample.
• The 90% typical concentration values in the standard are population statistics from units in service, not pass/fail limits; alarm thresholds are set by the owner, the manufacturer and the diagnostic expert together.
• Rate of change matters as much as absolute level: a moderate gas level rising fast outranks a higher level that has been flat for four years.
• Ratio results must be corroborated by electrical tests, furan analysis and operating history before any maintenance decision.
• Every standard number cited on this page carries its version year, because edition changes move the numbers: IEC 60599:2022 replaced the 2015 edition, and IEC 60567:2023 replaced the 2011 edition.

Introduction

IEC 60599 DGA is the interpretation framework that IEC 60599:2022 defines for turning dissolved gas measurements from mineral-oil-filled transformers in service into defensible fault categories, and this guide explains how working engineers apply that framework gate by gate in daily fleet practice. The standard describes how gas concentrations can be evaluated to diagnose the condition of oil-filled equipment and to suggest future action. It is not a measurement method and it is not a pass/fail checklist; it is a structured way to move from a table of ppm values toward a defensible statement about what is happening inside a transformer. In our commissioning work, the most common failure we see is not a wrong ratio calculation — it is a team that jumps straight to ratios without first checking whether the sample itself can be trusted and whether any gas is above the typical band at all.

Dissolved gas analysis (DGA)is the laboratory or online measurement of gases dissolved in transformer oil, expressed in ppm, that serves as the raw input for the entire framework. The standard addresses a practical problem: dissolved gas data alone does not indicate whether a transformer is healthy or faulty. Someone has to define what gas levels, ratios and patterns typically mean, and that is precisely the role IEC 60599:2022 plays alongside methods such as the Duval Triangle, Rogers Ratios and the Key Gas Method. It does not replace them — it provides the standard framework within which they are applied.

What is IEC 60599:2022 for DGA?

IEC 60599:2022 is an international standard published by IEC TC 10 on 25 May 2022 as edition 4.0, and it guides the interpretation of dissolved and free gas analysis for oil-filled electrical equipment in service by relating gas patterns to possible thermal and electrical fault conditions. It deliberately stops short of predicting when a transformer will fail, and that restraint is a design decision rather than an omission. The current edition replaced IEC 60599:2015, and the changes between editions matter in daily work: the 2022 edition refined the typical concentration values, added equipment-specific application notes, and sharpened the guidance on cases where ratio combinations do not fit any diagnostic row.

The table below summarizes the standard’s role:

Item	Description
Standard	IEC 60599:2022 (edition 4.0, published 2022-05-25, 80 pages)
Application	Dissolved and free gas analysis of equipment in service
Main purpose	Interpretation of dissolved gases, not measurement
Equipment	Mineral-oil-filled electrical equipment with cellulosic solid insulation
Diagnostic focus	Thermal and electrical fault conditions
Output	Diagnostic indications produced by IEC 60599 DGA interpretation, not a standalone failure prediction

DGA interpretationis the process of converting gas concentration measurements into a condition statement, and that is the exact boundary of the standard: it supports interpretation, it does not by itself provide a complete failure prediction. Within the broader diagnostic landscape it serves as a foundational reference, with the Duval Triangle and Rogers Ratios operating inside that framework rather than beside it.

Scope: which equipment and which data

IEC 60599:2022 applies to electrical equipment filled with mineral insulating oil and insulated with cellulosic paper or pressboard, and its scope is set by equipment type, by the data it covers, and by the boundaries of its diagnostic role in everyday maintenance work. The standard provides application notes for power transformers, instrument transformers, industrial and railway transformers, distribution transformers, reactors, bushings, switchgear and oil-filled cables. Those equipment-specific notes are given only as an indication, while the core framework applies broadly to mineral-oil-filled equipment. It may also be applied, but only with caution, to other liquid-solid insulating systems, with any resulting action taken only under proper engineering judgment.

On data coverage, the standard includes interpretation using gas concentrations and gas generation rates, diagnostic approaches for thermal and electrical fault conditions, typical concentration values and typical rates of gas increase. Sampling and laboratory analysis are explicitly outside this document — they belong to IEC 60567:2023, which replaced IEC 60567:2011 (the 2011 edition was withdrawn on 8 December 2023). If your laboratory report still cites the 2011 edition of the sampling standard, the numbers may be fine, but the procedure should be re-checked against the current edition.

What the standard does not do is replace engineering judgment, laboratory confirmation or electrical testing. The indications obtained are given only as guidance, and this limitation is repeated deliberately throughout the document.

How the interpretation sequence works

Interpretation under IEC 60599:2022 runs as a gate sequence rather than a single lookup table, and each gate can end the process before the ratio stage is ever reached, which is why most routine results stop early by design rather than by neglect. The overall logic runs: DGA measurement → gas concentration check → gas generation rate → gas pattern → diagnostic method → possible fault type → engineering assessment. Most routine results stop at the second or third gate, which is exactly how the standard intends it to work.

1. Reliable measurement first. Interpretation starts with trustworthy dissolved and free gas measurements from mineral-oil-filled equipment in service. Sample date, valve condition, syringe handling, air ingress and unit consistency all gate everything downstream.
2. Compare concentrations against typical values. The standard publishes typical 90% concentration values observed in service per equipment type, and alarm values that trigger investigation.
3. Evaluate rates of increase. A single DGA result is a snapshot; the standard provides typical gas increase rates (µl/l/year) for power transformers so that a rising level can be told apart from a stable one.
4. Apply ratio methods when an abnormality is confirmed. Only when levels or rates indicate an abnormality do the three key ratios — C₂H₂/C₂H₄, CH₄/H₂ and C₂H₄/C₂H₆ — come into play.
5. Close with engineering assessment. The final step combines the diagnostic indication with historical trends, operating conditions and additional testing before any maintenance decision.

The gate structure is also the best defense against the most expensive interpretation error we encounter: treating a ratio label as a finding. A ratio label computed on an untrustworthy sample, or on gas levels that sit inside the normal band, is not a finding — it is arithmetic applied to noise.

Table 1: the three-ratio fault table

When concentrations or generation rates indicate an abnormality, IEC 60599:2022 applies gas ratio analysis to classify the fault type, and Table 1 of the standard is where that classification happens for the six fault cases that service engineers encounter most often. The three ratios — C₂H₂/C₂H₄, CH₄/H₂ and C₂H₄/C₂H₆ — are compared against characteristic ranges for six fault cases. A combination that fits a row earns that row’s label; a combination that fits no row has no diagnostic under this method, and the correct response is to resample and trend, not to force a label onto the result.

Fault case	C₂H₂/C₂H₄	CH₄/H₂	C₂H₄/C₂H₆
PD — partial discharge	NS	< 0.1	< 0.2
D1 — low-energy discharge	> 1	0.1 – 0.5	> 1
D2 — high-energy discharge	0.6 – 2.5	0.1 – 1	> 2
T1 — thermal fault < 300 °C	NS	> 1 but NS	< 1
T2 — thermal fault 300–700 °C	< 0.1	> 1	1 – 4
T3 — thermal fault > 700 °C	< 0.2	> 1	> 4

Three details in this table change how it should be read, and they are the kind of thing that separates a copied table from a used one:

• NS means “not significant”. Where the table prints NS, the ratio does not gate the case whatever its value. Note the odd-looking T1 entry: CH₄/H₂ must be above 1, but beyond that its exact value is not significant for T1.
• PD thresholds shift by equipment. For partial discharge in instrument transformers the standard applies CH₄/H₂ < 0.2, and in bushings < 0.07, instead of the < 0.1 printed for power transformers. Copying the transformer row onto a bushing result is a wrong answer, not a conservative one.
• D1 and D2 overlap on purpose. A single result can legitimately fit both rows; the distinction is kept because discharge energy changes the corrective action, and the overlap is resolved by trending and corroboration rather than by arithmetic.

A rising C₂H₂ value in the T3 row additionally points to a hotspot above 1000 °C. For ratio combinations that fall outside every row, the standard’s own guidance is to treat the case as non-diagnostic and move to a graphical method such as the Duval Triangle. This is also where IEC 60599 DGA practice differs most from textbook practice: the table is only consulted after the concentration and rate gates have confirmed an abnormality worth classifying.

Typical concentration values and what they are not

The 90% typical values published in IEC 60599:2022 describe where nine out of ten healthy in-service units sit for each gas, and they are misread as acceptance limits more often than any other table in the standard, so this section fixes that reading first. The standard publishes these bands per equipment type precisely because a threshold copied from a power transformer table and applied to a bushing is a wrong answer. Below are the published 90% bands for power transformers without a communicating OLTC, in ppm:

Gas	90% typical band (ppm)
H₂	50 – 150
CH₄	30 – 130
C₂H₆	20 – 90
C₂H₄	60 – 280
C₂H₂	2 – 20
CO	400 – 600
CO₂	3,800 – 14,000

A typical valueis a population statistic from fleets in service, not a pass/fail criterion for your specific unit. The standard is explicit that alarm concentration values are decided by the user, the manufacturer and a diagnostic expert together, because oil volume, OLTC design, load history and breathing system all move the numbers. For instrument transformers (CT/VT), the published 90% bands sit far lower — H₂ 6–300, CH₄ 11–120, C₂H₄ 3–40, C₂H₂ 1–5, CO 250–1,100 ppm — and for capacitive bushings the CO band runs from 484 to 11,578 ppm, a range that looks absurd until you remember a bushing holds a few litres of oil, not tens of thousands.

One caution worth writing down: acetylene deserves special handling. Any detectable C₂H₂ above roughly 1–2 ppm in a transformer without a communicating OLTC already merits attention, because the published 90% band tops out at 20 ppm and acetylene only appears with discharge-type energy. Concentration checks are the second gate in the sequence — see DGA Fault Detection for how rate-of-change data extends this gate.

Fault diagnosis: PD, D1, D2, T1, T2, T3

IEC 60599:2022 groups internal faults into diagnostic categories that correspond to damage patterns observed after real failures, and each category carries both a gas signature that shows up in the oil and a physical signature that inspection can confirm afterwards. The standard categorizes faults that can be reliably identified by visual inspection after the event, which is how the categories were built in the first place:

• PD — partial discharges of the cold plasma (corona) type, with possible X-wax formation on paper.
• D1 — low-energy discharges, evidenced by small carbonized punctures in paper.
• D2 — high-energy discharges, with extensive carbonization, metal fusion and possible tripping.
• T1 — thermal faults below 300 °C, where paper turns brownish.
• T2 — thermal faults between 300 °C and 700 °C, where paper carbonizes.
• T3 — thermal faults above 700 °C, evidenced by oil carbonization, metal coloration or fusion.

Thermal fault classificationis the assignment of an overheating condition to one of the three temperature ranges T1 through T3, and it matters because the corrective action differs: a T1 finding can often be managed by load management and re-checking, while a T3 finding usually means metal is involved. When thermal stress affects cellulose, CO and CO₂ rise alongside the hydrocarbon gases, and the CO₂/CO relationship helps decide whether the fault involves solid insulation — a question with direct consequences for remaining-life assessment.

The diagnosis workflow can be visualized as level and rate read together:

The Duval triangle within IEC 60599:2022

The Duval Triangle operates inside the IEC 60599:2022 framework as a graphical cross-check, and the standard references it for specific equipment types including Triangle 2 for on-load tap changers and Triangle 3 for non-mineral insulating oils. The relationship is complementary rather than competitive: the standard provides concentration limits, generation rates and the ratio table, while the triangle converts the percentages of CH₄, C₂H₄ and C₂H₂ into a single plotted point that falls into one of seven zones — PD, D1, D2, T1, T2, T3, or the mixed D+T zone.

The worked example above is the point the triangle earns its keep on: the ratio table gives you a thermal indication, but the triangle pins the same three gases to T2 specifically, which changes the urgency conversation. The triangle always returns a zone — including for points made mostly of background gas — which is exactly why it needs the concentration and rate gates applied first. For the full zone boundaries and calculation, see the dedicated Duval Triangle DGA guide.

IEC 60599 DGA compared with other diagnostic methods

IEC 60599 DGA interpretation provides the framework within which the other diagnostic methods are applied, and the comparison in this section is about roles and handoffs rather than a ranking of which method deserves to win. The goal is not to decide which method is best but to apply the appropriate method — or combination — inside a consistent interpretation framework. IEEE C57.104-2019, the American counterpart guide, focuses on gas limit grading and risk classification with its own norm set, so the two documents answer slightly different questions for the same transformer.

• Duval Triangle — a graphical method using CH₄, C₂H₄ and C₂H₂ percentages to classify fault zones.
• Rogers Ratios — a ratio-based method assigning fault codes from fixed ratio windows.
• Key Gas Method — a simpler approach mapping characteristic gases to fault types.
• Other ratio methods — additional approaches for specific equipment types or fault categories.

The Key Gas Methodis a diagnostic approach that associates each fault type with a dominant characteristic gas, and it remains useful as a sanity check even where ratio methods disagree. In fleet practice, we run the ratio table first and the triangle second on any result that passes the concentration and rate gates; when the two agree, the label is usually solid, and when they disagree, that disagreement is itself diagnostic information pointing to a mixed or evolving condition.

Online DGA monitoring and the standard

Online DGA monitoring and IEC 60599:2022 occupy different and complementary roles, because the monitor supplies frequent or continuous gas measurements while the standard supplies the interpretive framework that turns the resulting data stream into an assessment an engineer can act on. The sequence runs: online DGA monitoring → continuous gas data → trend analysis → IEC 60599-based interpretation → fault assessment → alarm or engineering action. The standard is not a monitoring technology, and an online monitor is not a diagnostic — the value comes from wiring the two together.

Continuous data changes the rate-of-change gate more than any other part of the framework. A monthly laboratory sample gives you one point per month; a multi-gas online monitor gives you daily or hourly points, which means a “stable” verdict can be reached in days instead of quarters, and an accelerating trend can trigger a resample while the evidence is still fresh. For hardware that feeds this workflow, see our transformer DGA online monitoring system and the broader DGA monitoring range.

Limitations and boundaries

IEC 60599:2022 states its own limits explicitly, and three boundaries decide in practice whether the framework produces a usable answer or an overconfident label that a review meeting will later tear apart under load and budget pressure. Each boundary below is written as an operating rule rather than a disclaimer:

1. A ratio label cannot confirm a fault by itself. The indications are guidance only, and any resulting action must be taken under proper engineering judgment; an IEC 60599:2022 fault label that is not corroborated by electrical tests, furan analysis or inspection evidence is a hypothesis, not a finding.
2. A single result cannot establish a trend. One measurement is a snapshot that cannot show whether the condition is stable, developing or accelerating; only repeated measurements over time reveal gas generation behavior, so a first abnormal result should trigger a resample on a defined interval rather than an outage decision.
3. Typical values cannot be transplanted across equipment. The 90% bands are published per equipment type because oil volume, OLTC design and insulation structure change the numbers; applying a power transformer band to a bushing or an instrument transformer produces false alarms or false comfort, so the equipment-specific tables must be used as printed.

Interpretation workflow

The full interpretation process compresses into six gates, and knowing where fleet results usually stop is as useful in daily work as knowing the gates themselves, because each early stop is a decision the standard intends you to make deliberately:

1. Obtain reliable DGA data — date, valve, syringe, air ingress, units; otherwise resample.
2. Review gas concentrations — compare against the per-equipment 90% bands; nothing above the band means record as baseline and stop.
3. Evaluate generation trends — assess rates to distinguish stable, developing and accelerating conditions.
4. Identify gas patterns and ratios — apply Table 1; a combination fitting no row is reported as inconclusive, not forced into a label.
5. Check paper involvement — CO₂/CO and furans decide whether solid insulation is at stake.
6. Corroborate with electrical testing — winding resistance, tan delta and PD measurements; without agreement, the label remains a guess.

In fleet data, most routine samples stop at gate 2 with normal gas levels, older units with accumulated gas often stop at gate 3 with a high but flat level, and the small percentage reaching gate 4 with ratios that fit no row should be resampled on a shortened interval — an inconclusive result is not a reason to keep running without a second sample.

Frequently asked questions about IEC 60599 DGA

What is IEC 60599:2022 used for?

It is the international standard for IEC 60599 DGA interpretation of oil-immersed electrical equipment in service, used for transformer fault classification, condition evaluation and maintenance guidance. It converts a table of ppm measurements into structured diagnostic indications — PD, D1, D2, T1, T2, T3 — that engineering teams can act on after corroboration. It is not a measurement method and not a pass/fail standard.

Which gases does the standard cover?

Seven core gases: H₂, CH₄, C₂H₆, C₂H₄, C₂H₂, CO and CO₂, covering all characteristic gases of electrical and thermal faults in mineral-oil-filled equipment. The three-ratio method uses C₂H₂/C₂H₄, CH₄/H₂ and C₂H₄/C₂H₆, while CO₂/CO supports the paper-involvement check and C₂H₂/H₂ is suggested for specific fault cases. Each gas plays a defined role in the gate sequence rather than being read in isolation.

How does the standard classify faults?

It applies multi-dimensional judgment: gas pattern matching, three-ratio analysis, concentration thresholds and historical trend, rather than any single data point. The ratio table assigns results to partial discharge, low-energy discharge, high-energy discharge, or thermal faults in three temperature grades. A ratio combination that fits no row has no diagnostic under the method and should be resampled or moved to a graphical method such as the Duval Triangle.

Can it detect all transformer faults?

It detects partial discharge, low- and high-energy discharge, and thermal faults across three temperature grades, covering most internal insulation faults of oil-filled equipment. It does not cover conditions that generate no characteristic gases — winding deformation, core clamping loosening, or mechanical failures without thermal or electrical signatures. Those require electrical testing, which is why the standard positions itself as one input to the maintenance decision.

Is IEC 60599:2022 the same as IEEE C57.104?

No. IEC 60599:2022 is the international standard focused on fault pattern interpretation, while IEEE C57.104-2019 is the American guide focused on gas limit grading and risk classification. The two documents use different norms and conditions tables, and a fleet operator working across both frameworks should not mix alarm values from one with ratio labels from the other. Always verify which framework your alarm thresholds were derived from.

Does the standard define the Duval Triangle?

The standard does not define the triangle, but it recognizes graphical diagnostic methods as effective auxiliary verification tools and references the Duval Triangle for specific equipment types, including Triangle 2 for OLTCs and Triangle 3 for non-mineral oils. In practice the triangle converts three gas percentages into a plotted zone while the standard supplies the concentration and rate gates around it. Used together they produce more repeatable diagnoses than either alone.

Can DGA predict when a transformer will fail?

No. DGA is a condition indicator, and final fault confirmation requires electrical tests, visual inspection and other means. The standard supports interpretation toward a fault category, not a failure date; remaining-life statements need furan analysis, insulation power factor and load history in addition. Any supplier promising failure prediction from gas data alone is overselling the method.

How often should DGA sampling follow the standard?

Routine monitoring typically follows annual or quarterly cycles, with intensive monitoring required for abnormal trends and post-fault scenarios. The sampling procedure itself belongs to IEC 60567, whose 2011 edition was withdrawn in December 2023 and replaced by the 2023 edition, so laboratories should confirm which edition their procedures cite. Once an abnormality is confirmed, the sampling interval should shorten until the trend direction is established.

Conclusion

IEC 60599 DGA interpretation provides a unified and repeatable specification for turning dissolved gas data into maintenance decisions, and its core value is the gate sequence that keeps arithmetic away from noise until the evidence can carry a label. Used in that order, the framework avoids the two most expensive interpretation errors: ratio labels computed on noise, and inconclusive results that never get a second sample.

For EPC projects, utility operation and maintenance teams and plant equipment management, standardized diagnosis combined with the Duval Triangle for cross-checking and online monitoring for trend density is the practical recipe for avoiding diagnostic errors and catching latent faults early. If you need a customized online DGA monitoring system for your transformer fleet, or support applying IEC 60599:2022 interpretation to your existing laboratory reports, contact our technical team for a targeted scheme consultation.

References and technical sources

• IEC 60599:2022, Mineral oil-filled electrical equipment in service – Guidance on the interpretation of dissolved and free gases analysis, edition 4.0, published 2022-05-25, TC 10, 80 pages.
• IEC 60567:2011, Oil-filled electrical equipment – Sampling of gases and analysis of free and dissolved gases – Guidance, edition 4.0, published 2011-10-20; withdrawn 2023-12-08 and replaced by IEC 60567:2023.
• IEEE C57.104-2019, IEEE Guide for the Interpretation of Gases Generated in Mineral Oil-Immersed Transformers, published 2019-11-01.
• IEEE C57.104-2019 on IEEE Xplore, DOI 10.1109/IEEESTD.2019.8890040, electronic ISBN 978-1-5044-5973-0.
• CIGRE Working Group 13.09, transformer diagnostics and monitoring techniques, as cited in the IEC 60599 bibliography.

Related reading

• Duval Triangle 1 plotter
• online DGA monitoring guide
• transformer bushing monitoring system
• All DGA & Oil Analysis articles