DGA and Partial Discharge: Correlating Dissolved Gas Analysis with PD Activity in Transformers

You have two different windows into a transformer’s health. One is the DGA report — a lab analysis of gases dissolved in the insulating oil. The other is a PD test — a direct measurement of partial discharge activity in the windings. Both diagnose insulation problems. Neither alone tells the full story.

When they agree, you have high confidence in what’s happening inside the tank. When they disagree, one of them is wrong — and figuring out which is wrong is one of the highest-value skills in transformer diagnostics.

This article explains how DGA and PD testing relate physically, how to interpret them together, and how to use the combined evidence to make defensible maintenance decisions. Standards referenced are IEC 60599:2022 for DGA interpretation, IEC 60076-3 for transformer dielectric testing, and IEC 60270 for PD measurement.

Two Diagnostic Methods, One Transformer

Before combining the methods, understand why you’d use either one.

DGA (Dissolved Gas Analysis) — An oil sample is drawn from the transformer and analyzed in a lab. Gases dissolved in the oil are extracted and measured. Concentrations (usually in ppm) of H₂, CH₄, C₂H₆, C₂H₄, C₂H₂, CO, and CO₂ are the core outputs.

• Advantages: Non-intrusive (just a sample port), inexpensive per test (~$100–300), historical method with decades of interpretation data, covers the entire transformer (oil circulates throughout).
• Limitations: Detects consequences of faults (the gases produced), not the faults themselves. Gas concentrations depend on fault duration, oil volume, degassing (if any), and gas solubility. Can’t localize where the fault is.

PD (Partial Discharge) Testing — Detection of the actual electrical discharges happening inside the insulation, typically using electrical measurement per IEC 60270 or UHF/acoustic detection.

• Advantages: Detects faults in real time as they occur. Can localize the source (at least roughly). Shows pattern characteristics that distinguish fault types.
• Limitations: Typically requires taking the transformer out of service or installing specialized online monitoring. More expensive and complex than DGA. High noise on in-service measurements.

Both are diagnostic tools. They answer the same fundamental question — is there a fault developing inside the insulation? — but through different mechanisms.

The Physics: What Makes Each Gas Form

Dissolved gases appear in transformer oil because the oil and paper insulation decompose under stress. Different stresses produce different gas signatures.

Per Vahidi & Teymouri, mineral insulating oil is a mixture of hydrocarbon molecules. When the oil breaks down, specific gas molecules are produced: “hydrogen (H–H), methane (CH₃–H), ethane (CH₃–CH₃), ethylene (CH₂=CH₂) and acetylene (CH≡CH).”

The key insight from Halstead’s 1973 work (still the foundation of modern DGA interpretation):

The maximum gases produced at different temperatures with increasing the temperatures are methane, ethane, ethylene and acetylene.

This temperature-gas relationship is what makes DGA diagnostic. Different fault types produce different temperatures, which produce different gases.

Gas-to-fault mapping

Here’s the physical correlation between gas species and fault types:

Gas	Dominant fault type
Hydrogen (H₂)	Low-energy electrical discharges including partial discharge and corona
Methane (CH₄)	Low-intensity electrical stress; initial PD; low-temperature thermal (<300°C)
Ethane (C₂H₆)	Medium-temperature thermal faults (150–300°C); low-energy PD
Ethylene (C₂H₄)	High-temperature thermal faults (>500°C); conductor overheating
Acetylene (C₂H₂)	Very high energy events — arcing, extreme thermal (>1000°C)
Carbon monoxide (CO)	Paper insulation degradation
Carbon dioxide (CO₂)	Paper insulation degradation; normal aging

The key implication for PD correlation

Partial discharge activity — particularly low-energy PD in void cavities — primarily generates H₂ and CH₄, with some C₂H₆. High-energy PD or arcing generates C₂H₂. Paper-insulation-related PD adds CO and CO₂ to the mix.

This means DGA can detect PD activity indirectly — you see the gas fingerprint even when you can’t directly measure the discharges. This is particularly valuable because DGA can be performed without taking the transformer out of service.

Gas Concentration Baselines by Transformer Age

Gas concentrations aren’t interpreted in isolation. They’re interpreted against what’s typical for the transformer’s age.

The Vahidi & Teymouri textbook provides statistical ranges for healthy and failed transformers:

Typical gas concentrations by transformer age (ppm)

Gas	<4 years operation	4–10 years operation	>10 years operation
H₂	100–150	200–300	200–300
CH₄	50–70	100–150	200–300
C₂H₆	30–50	100–150	800–1000
C₂H₄	100–150	150–200	200–400
C₂H₂	20–30	30–50	100–150
CO	200–300	400–500	600–700
CO₂	3000–3500	4000–5000	9000–12,000

How to use these baselines

These are the ranges in which healthy-to-slightly-aged transformers sit. A young transformer (under 4 years) with 800 ppm of ethane is well above normal and indicates an active thermal fault. An older transformer (20+ years) with the same 800 ppm ethane is within normal aging.

Context matters enormously. Acceptance test gas levels on a new transformer (per IEC 60076-3) are typically far lower than these operational values — often single-digit ppm for most gases. Operating transformers accumulate gases gradually through normal aging plus any faults.

The total dissolved combustible gas (TDCG) rule of thumb

As a quick screen, the total concentration of flammable dissolved gases (H₂ + CH₄ + C₂H₆ + C₂H₄ + C₂H₂ + CO) gives an overall health indication:

Total flammable dissolved gas (ppm)	Interpretation
0–500	Satisfactory
500–1,000	Oil degradation; needs monitoring
>1,000	Significant oil decomposition; close and accurate monitoring
>2,500	Very severe oil decomposition; identify faults urgently

This screening number is useful but not a substitute for looking at individual gas patterns and ratios.

The Four Classical DGA Interpretation Methods

Four methods dominate transformer DGA interpretation. Each uses gas ratios to diagnose specific fault types. Modern interpretation typically applies all four and looks for consensus.

Method 1: Dürrenberg Method

The oldest approach, still widely used as a baseline. It first checks whether any gas exceeds a threshold called L1.

Dürrenberg L1 thresholds (ppm):

Gas	L1 limit
CO	350
CH₄	120
H₂	100
C₂H₆	65
C₂H₄	50
C₂H₂	35

Fault detection rule: If at least one of H₂, CH₄, C₂H₄, or C₂H₂ exceeds twice the L1 limit AND at least one other of the three exceeds L1, a fault is present.

Fault type by four ratios:

Fault type	CH₄/H₂	C₂H₂/C₂H₄	C₂H₂/CH₄	C₂H₆/C₂H₂
Thermal decomposition	<1	<0.75	<0.3	>0.4
Corona / partial discharge	<0.1	Insignificant	<0.3	>0.4
Arc	0.1 < R < 1	>0.75	>0.3	<0.4

Strength: Simple, transparent, good for screening. Weakness: Relatively coarse — corona and partial discharge grouped together, and some edge cases don’t fit.

Method 2: Rogers Ratio Method

An improvement on Dürrenberg using four different ratios with coded results. Achieves ~80% accuracy on typical soluble gases.

Uses ratios of C₂H₆/CH₄, C₂H₄/C₂H₆, C₂H₂/C₂H₄, and CH₄/H₂. Each ratio is encoded (0, 1, 2, or 5) based on its value, and the combination of codes identifies the fault.

Rogers distinguishes 12 different fault categories, including:

• Normal decomposition
• Partial discharge
• Mild heating (<150°C, 150–200°C, 200–300°C)
• Conductor heating
• Winding circulating current
• Tank and core circulating current
• Arc (and arc variations)

Strength: More granular fault classification than Dürrenberg; distinguishes between thermal temperature ranges. Weakness: Codes can become ambiguous for some gas combinations.

Method 3: IEC Ratio Method (per IEC 60599)

The current international standard method. Drops the C₂H₆/CH₄ ratio (which only covers a narrow temperature range) and uses three ratios: C₂H₂/C₂H₄, CH₄/H₂, and C₂H₄/C₂H₆.

IEC ratio codes:

Gas ratio	<0.1	0.1 ≤ x < 1.0	1.0 ≤ x < 3.0	≥3.0
C₂H₂/C₂H₄	0	—	1	2
CH₄/H₂	1	0	—	2
C₂H₄/C₂H₆	0	—	1	2

IEC 60599 fault categories (9 types):

Fault type	C₂H₂/C₂H₄	CH₄/H₂	C₂H₄/C₂H₆
Normal, no fault	0	0	0
Partial discharge, low energy	Insignificant	1	0
Partial discharge, high energy	1	1	0
Discharge, low energy	1	0	2
Discharge, high energy	1–2	0	1–2
Thermal fault T ≤ 150°C	0	0	1
Thermal fault 150 < T ≤ 300°C	0	2	0
Thermal fault 300 < T ≤ 700°C	0	2	1
Thermal fault T > 700°C	0	2	2

Strength: International standard, widely adopted, good thermal discrimination. Weakness: Still has ambiguous regions where measured ratios don’t fit any category.

Special notes from IEC 60599:

• If C₂H₂/H₂ > 3, contamination from the on-load tap changer is likely (tap changer oil leaking into main tank)
• If O₂/N₂ < 0.3, unusual heating or oxidation is present

Method 4: Duval Triangle Method

The most widely used method today for graphical interpretation. Proposed by Michel Duval in 1974, revised and extended since.

Uses only three gases — CH₄, C₂H₄, and C₂H₂ — expressed as relative percentages plotted on a triangular diagram. Each point on the triangle falls in a zone corresponding to a specific fault type.

Relative concentrations:

Given A = CH₄, B = C₂H₄, C = C₂H₂ (all in ppm):

• x = 100 × A / (A + B + C)
• y = 100 × B / (A + B + C)
• z = 100 × C / (A + B + C)

Duval Triangle zones (modified version, 7 fault types):

Code	Fault type
PD	Partial discharge
D1	Low energy discharge
D2	High energy discharge
DT	Combination of electrical and thermal faults
T1	Thermal fault T < 300°C
T2	Thermal fault 300 < T < 700°C
T3	Thermal fault T > 700°C

Strength: Graphical, intuitive, always gives an answer (no “indeterminate” zones). Handles cases where ratio methods fail. Weakness: Based only on three gases; ignores H₂, C₂H₆, CO, CO₂ which contain additional information.

Which method to use

Modern practice applies all four methods and looks for consensus:

• If all four methods identify the same fault type → high confidence
• If three of four agree → moderate confidence; the outlier may indicate a complex fault
• If methods disagree significantly → complex fault, repeat sampling, consider PD testing

The Duval Triangle is the most reliable single method. IEC 60599 is the current standard. Rogers provides good thermal discrimination. Dürrenberg is useful for quick screening.

Partial Discharge Gas Signatures

Here’s where DGA and PD testing converge — the specific gases that signify PD activity.

Low-energy PD (void discharges, corona)

Primary gases:

• H₂ (hydrogen) — dominant gas for low-energy discharges
• CH₄ (methane) — secondary
• Some C₂H₆ (ethane)

Key ratios:

• CH₄/H₂ < 0.1 (low-energy PD classical signature per IEC 60599)
• C₂H₂/C₂H₄ negligible (no arcing)

What it means: Low-energy partial discharges, typically from voids or corona in the insulation system. Progressive rather than catastrophic. Correlates with void-type PRPD patterns.

High-energy PD (approaching breakdown)

Primary gases:

• H₂, CH₄ (still present)
• C₂H₂ (acetylene) — now appearing in measurable amounts
• C₂H₄ (ethylene) — increasing

Key ratios:

• C₂H₂/C₂H₄ between 0.1 and 1.0
• CH₄/H₂ around 0.1–1.0

What it means: Higher-energy discharge events — potentially arc-like PD or tracking. This is a more serious condition that often indicates imminent failure. Correlates with surface tracking or high-magnitude void PRPD patterns.

Arcing (catastrophic failure mode)

Primary gases:

• C₂H₂ (acetylene) — high concentrations
• H₂, C₂H₄ also high
• CO/CO₂ if arcing involves paper insulation

Key ratios:

• C₂H₂/C₂H₄ > 1
• C₂H₄/C₂H₆ > 3

What it means: High-energy electrical discharge, possibly an incipient or actual arc. The transformer is likely approaching failure. This is a shut-down-and-investigate condition.

The acetylene threshold

The appearance of acetylene in transformer oil is always significant. Per Halstead’s original work: “at very high temperatures, more than 1000°C in arcs, acetylene is produced.”

Any detectable acetylene (typically >5 ppm, with allowance for some background) indicates high-energy electrical activity. Acetylene above 35 ppm (Dürrenberg L1) typically requires investigation. Acetylene above 100 ppm in a transformer under 10 years old often indicates serious fault development.

When DGA and PD Agree — And When They Don’t

The real diagnostic power comes from combining both methods.

Scenario 1: DGA positive, PD positive (both detect a problem)

Example: DGA shows elevated H₂ (250 ppm) and CH₄ (140 ppm), CH₄/H₂ ratio ≈ 0.56, low acetylene. PD test shows void-type PRPD pattern at 150 pC on the tertiary winding.

Interpretation: Low-energy PD activity, most likely void discharge in the tertiary winding insulation. High-confidence diagnosis. Both methods are detecting the same fault through different mechanisms.

Action: Trend both measurements. Moderate PD (150 pC) with mild gas evolution is usually acceptable for continued operation with increased monitoring frequency.

Scenario 2: DGA positive, PD negative

Example: DGA shows significantly elevated C₂H₄ (300 ppm) and CH₄ (180 ppm), but current PD test shows no significant activity (background <10 pC).

Possible interpretations:

• Thermal fault, not electrical — C₂H₄ dominance suggests a high-temperature thermal fault (conductor overheating, bad connection, hot spot). Thermal faults don’t produce PD because there’s no discharge activity.
• Past fault, now quiet — an event occurred previously (generating the gases), but the underlying cause was cleared or is currently dormant.
• PD in a location the current test can’t detect — online vs offline sensitivity differences, or discharge at a location far from sensors.

Action: Follow up with thermographic survey (for hot spots), examine load history, and consider repeat PD testing at elevated voltage to excite hidden PD sites.

Scenario 3: DGA negative, PD positive

Example: DGA shows all gases within normal age-appropriate baselines. Recent PD test shows 400 pC void-type pattern.

Possible interpretations:

• Early-stage PD — the PD has started but hasn’t yet produced enough gas to register above baseline. DGA is a lagging indicator.
• PD without oil contact — discharge is happening in a region where the produced gases aren’t dissolving in the main oil volume (e.g., inside a sealed bushing).
• False positive PD — noise or measurement artifact in the PD test.

Action: Repeat PD measurement to confirm. If confirmed, treat as early-stage developing defect. Follow up DGA at shorter interval (3 months) to verify gas evolution.

This scenario is actually where combined testing adds the most value. DGA is a lagging indicator for PD — the gases accumulate over time after the discharge activity starts. PD testing catches the activity earlier.

Scenario 4: DGA negative, PD negative

Interpretation: Transformer is in good condition. Both methods agree, high confidence in health.

Action: Continue routine monitoring intervals.

The importance of trends over snapshots

A single DGA measurement tells you the current state. A single PD test tells you current PD activity. Neither is as valuable as trended data.

A transformer with H₂ at 250 ppm that has been stable at 250 ppm for five years is not the same as a transformer whose H₂ increased from 50 ppm to 250 ppm in six months. The trended data identifies the developing problem; the snapshot does not.

Standard practice for critical transformers:

• DGA: Every 6 months routine; monthly if elevated values present; weekly if critical fault suspected
• PD testing: Annually offline or continuous online monitoring for critical units
• Both combined: Establish correlated baselines; trend over time; investigate divergence

Case Patterns: Reading Combined Evidence

Some common scenarios you’ll encounter:

Pattern A: The textbook void discharge

• DGA: H₂ elevated (300+ ppm), CH₄ moderately elevated, CH₄/H₂ ≈ 0.3, no acetylene
• Duval Triangle: PD zone
• IEC 60599 ratio: PD, low energy
• PRPD: Symmetric clusters on rising slopes of voltage waveform, 100–500 pC
• Diagnosis: Void discharges in solid insulation. Classic early-to-mid stage PD in aged insulation.
• Action: Annual PD testing, 6-month DGA, maintain under continued operation.

Pattern B: Developing arc

• DGA: H₂ high, acetylene >50 ppm and increasing, ethylene elevated
• Duval Triangle: D2 zone (high energy discharge)
• IEC 60599 ratio: High energy discharge
• PRPD: Large-magnitude discrete events, irregular pattern, high repetition
• Diagnosis: Arcing or near-arc discharge activity. Critical condition.
• Action: Urgent — consider de-energization, detailed investigation, potential repair.

Pattern C: Hot spot, no PD

• DGA: C₂H₄ dominant, C₂H₆ elevated, H₂ moderate, no acetylene
• Duval Triangle: T2 or T3 zone (thermal fault)
• IEC 60599 ratio: Thermal fault 300–700°C
• PRPD: No significant PD activity
• Diagnosis: Thermal fault — likely bad connection, overheated lead, or stray loss heating. Not an electrical defect.
• Action: Investigate thermal sources (infrared survey, load analysis, inspect connections at next outage). PD testing unnecessary for this fault type.

Pattern D: Paper degradation

• DGA: CO elevated, CO₂ very elevated, CO₂/CO ratio around 8–10
• Furans: 2-furfural >0.1 ppm (indicates paper degradation)
• PRPD: Variable — may show surface discharge patterns if paper damage involves surface tracking
• Diagnosis: Paper insulation degradation. Per Vahidi & Teymouri, the CO₂/CO ratio corresponds to the paper’s degree of polymerization (DP): “A ratio <3 indicates a fault involving paper insulation” while ratios 7.4–8.7 correspond to varying paper health states.
• Action: Assess remaining paper life; consider DP estimation via 2-furfural measurement; plan long-term replacement.

Pattern E: Tap changer contamination

• DGA: Elevated hydrocarbons, especially acetylene; other gases “normal”
• C₂H₂/H₂ ratio >3
• PRPD: Normal (if tap changer is the only source)
• Diagnosis: Per IEC 60599: contamination from on-load tap changer oil leaking into main tank.
• Action: Investigate tap changer seals, not insulation. Electrical diagnosis shouldn’t conclude “arcing in main tank” when the real source is OLTC contamination.

CO/CO₂ Ratio and Paper Degradation

A special case worth understanding: the CO₂/CO ratio as a paper health indicator.

Per Vahidi & Teymouri (Chapter 3, quoting IEC 60599): the normal range for paper aging is 3 < CO₂/CO < 10. Values outside this range indicate fault conditions:

CO₂/CO ratio	Interpretation
<3	Fault involving paper insulation; overheating above 200°C
3–7.4	Healthy paper
7.4–8.0	Normal (stable aging)
8.0–8.7	Weakening paper
>8.7	Near end-of-life paper

The advantages of using CO₂/CO ratio over other paper indicators (like 2-furfural):

1. More stable — CO and CO₂ don’t degrade over time like furans
2. Robust to gas leaks — if some gas leaks out, the ratio is preserved even though absolute concentrations drop
3. Directly measured — no need for separate furfural analysis

This ratio complements PD testing because paper degradation can occur without producing significant PD (particularly thermal aging of paper at hot spots). DGA with CO/CO₂ analysis catches this type of fault; PD testing typically doesn’t.

Acceptance Testing and Trending

Acceptance testing

For new transformers, acceptance testing per IEC 60076-3 requires PD measurement at elevated voltage (typically 1.3× to 1.7× rated, depending on standard and manufacturer) with PD limits usually around 100–300 pC.

DGA baseline should also be established at commissioning:

• Complete gas analysis before energization (should show very low levels — typically <10 ppm for most gases)
• Follow-up DGA at 1 week, 1 month, and 3 months after commissioning to establish operational baseline
• This initial data is invaluable for later trending — without it, you’re comparing to generic “typical” values rather than the specific transformer’s behavior

Trending best practices

Establish a testing schedule based on criticality:

Transformer criticality	DGA interval	PD testing
Critical (main, GSU)	3–6 months	Annual offline, consider online
Important (substation)	6–12 months	2–3 year offline
Standard (distribution)	Annual	Optional
Non-critical	1–2 years	Not typically done

When to shorten intervals:

• Any single gas exceeds age-appropriate baseline → quarterly
• Any gas increasing >30% between tests → quarterly
• Acetylene appears at any concentration in younger transformers → monthly for 3 months
• PD test shows elevated activity → correlate with DGA quarterly

What to track:

• Absolute gas concentrations
• Gas generation rates (ppm/month)
• Ratio evolution (particularly CH₄/H₂ and CO₂/CO)
• PD magnitudes and patterns (if doing PD testing)
• Correlation between DGA changes and PD changes

FAQ

If I can only afford one diagnostic method for my transformers, which should I choose?

DGA, without question. It’s dramatically cheaper, can be done without taking the transformer out of service, and covers the full range of fault types (electrical and thermal). PD testing is a valuable addition for critical units but DGA is the foundation of transformer condition monitoring.

How long does a transformer with active PD take to fail?

Variable — from months to decades depending on operating conditions, fault location, and fault type. Low-energy PD (void discharges, corona) often continues for years without catastrophic failure. High-energy PD (arc-type discharges) can lead to failure within weeks to months. The trend matters more than the absolute level — worsening PD with rising acetylene in DGA is a clear warning sign.

Can DGA results be wrong?

Yes. Common sources of error:

• Sampling errors — air contamination of samples can inflate apparent hydrogen
• Lab accuracy — older labs may have ±30% accuracy; modern labs ±10%
• Transport and storage — samples degrading before analysis
• Dissolution issues — gases taking time to redistribute in the oil after being generated

Always verify unusual results with repeat sampling before acting on them.

What’s the difference between online DGA monitors and periodic lab DGA?

Online monitors (installed on the transformer) give continuous trending of key gases (usually H₂, CH₄, C₂H₂, plus moisture). Periodic lab DGA gives a complete gas analysis but only at sampling intervals. Best practice for critical transformers combines both: online for early warning of rapid gas changes, periodic lab DGA for comprehensive analysis and verification.

Why doesn’t Duval Triangle use hydrogen?

The original Duval Triangle was optimized for fault discrimination rather than fault detection. Using only CH₄, C₂H₄, and C₂H₂ provides clean separation between PD, discharge, and thermal faults. H₂ and C₂H₆ are still useful for other methods (Rogers, IEC) and for confirming Duval Triangle results.

How do I handle an “ambiguous” DGA result?

If two methods say PD and two say thermal, the fault is likely complex (combined electrical and thermal activity) or transitioning between fault types. The Duval Triangle has a DT (“combination”) zone specifically for this. Follow-up actions:

• Repeat DGA to confirm
• Add PD testing if not already done
• Consider additional diagnostics (thermography, frequency response analysis, tan delta)
• Trend carefully — the fault type often clarifies over time

Do I need to worry about gases from the on-load tap changer (OLTC)?

Yes, if OLTC oil can mix with main tank oil (not all transformers have this risk). OLTC operation naturally produces acetylene from arcing during tap changes. This is normal OLTC operation, not a main tank fault. Per IEC 60599: “it is recommended that C₂H₂/H₂ value should be greater than 3 to detect the contamination due to the function of the tap changer.” If C₂H₂/H₂ >3 and other indicators don’t support arcing, suspect OLTC contamination before diagnosing main tank failure.

Can I use DGA on synthetic ester or silicone oils?

Interpretation rules differ. Mineral oil DGA is well-established; silicone and natural/synthetic ester oils have different gas generation signatures. Check IEC 60422 and manufacturer guidelines for non-mineral fluids. Don’t directly apply mineral oil thresholds to other fluids.

Key Takeaways

• DGA detects the consequences of faults (gases produced); PD testing detects the fault activity itself. They’re complementary, not redundant.
• Gas-to-fault mapping: H₂/CH₄ → low-energy PD and corona; C₂H₄ → high-temperature thermal; C₂H₂ → arcing; CO/CO₂ → paper degradation.
• Four classical methods — Dürrenberg, Rogers, IEC 60599, Duval Triangle — should be applied together for consensus interpretation. The Duval Triangle is the most reliable single method; IEC 60599 is the current international standard.
• Gas concentrations are age-dependent. A healthy transformer at 20 years has dramatically higher baseline gas levels than a new transformer. Use age-appropriate baselines.
• Acetylene is the alarm gas. Any measurable acetylene in an electrically-insulated transformer (excluding OLTC contamination) warrants investigation.
• PD activity and DGA results usually agree on electrical faults. When they disagree, the cause is usually a thermal fault (DGA positive, PD negative) or early-stage PD (PD positive, DGA negative).
• CO₂/CO ratio indicates paper health. Values 3–10 are normal; <3 or >10 suggests paper-related faults.
• Trends matter more than snapshots. Stable moderate gas levels are very different from rapidly-increasing moderate levels. Establish baselines early and trend consistently.
• OLTC contamination can mimic arcing. C₂H₂/H₂ > 3 often indicates tap changer oil leakage, not main tank arcing.
• DGA is cheap and non-intrusive. PD testing is more expensive and specialized. For most transformers, DGA is primary; PD testing supplements for critical units.

Standards and References

Standard / Reference	Content
IEC 60599:2022	Mineral oil-filled electrical equipment in service — Guidance on the interpretation of dissolved and free gases analysis
IEC 60076-3	Power transformers — Insulation levels, dielectric tests and external clearances in air
IEC 60270:2000+AMD1:2015	High-voltage test techniques — Partial discharge measurements
IEC 60422:2013	Mineral insulating oils in electrical equipment — Supervision and maintenance guidance
IEEE C57.104	IEEE Guide for the Interpretation of Gases Generated in Mineral Oil-Immersed Transformers
IEEE C57.106	IEEE Guide for Acceptance and Maintenance of Insulating Mineral Oil in Electrical Equipment
CIGRE TB 296	Recent developments on the interpretation of dissolved gas analysis in transformers
Vahidi & Teymouri (2019)	Quality Confirmation Tests for Power Transformer Insulation Systems — Springer (Chapters 4 and 5)