Dissolved gas analysis (DGA) is the primary diagnostic tool for detecting developing faults inside oil-filled transformers before they progress to catastrophic failure. This guide covers how to read gas patterns, apply ratio methods, assign action tiers, collect valid samples, and embed DGA into a structured maintenance program. It also addresses procurement decisions that determine whether a transformer can be monitored cost-effectively throughout its service life.<\/p>\n
\n
Quick Diagnosis: Gas Pattern vs. Fault Type<\/h2>\n
Before working through ratio methods or action thresholds, match the dominant gas pattern to a probable fault type using the table below. This is the first filter applied when a new lab result arrives.<\/p>\n
\n\n
\n
Symptom (Dominant Gas Pattern)<\/th>\n
First Test<\/th>\n
Likely Root Cause<\/th>\n
Next Action<\/th>\n<\/tr>\n<\/thead>\n
\n
\n
H2 only or H2 + CH4 (small); low C2H2<\/td>\n
Check moisture in oil; review breather condition<\/td>\n
Partial discharge (PD) in moisture-contaminated oil<\/td>\n
Schedule offline PD testing; shorten sampling interval to monthly<\/td>\n<\/tr>\n
\n
CH4 + C2H6 elevated; negligible C2H2<\/td>\n
Review load history and cooling system logs<\/td>\n
Normal aging or overloaded paper insulation<\/td>\n
Trend review; no immediate electrical action required<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
Key principle:<\/strong> Acetylene above 1-2 ppm in a sealed transformer with no recent through-fault record is the single most reliable indicator demanding expedited response, regardless of other gas levels. Rate of change is often the earlier warning signal; a methane reading that doubled in four weeks carries more urgency than the same reading stable for six months.<\/p>\n
Tools and Acceptance Sources<\/h2>\n
Before interpreting any DGA result, verify that sampling equipment, laboratory methods, and reference standards meet the criteria below. A flawed sample cannot be corrected analytically downstream.<\/p>\n
Calibrated and functionally tested before dispatch; gas volume and color recorded on trip<\/td>\n<\/tr>\n
\n
IEC 60599<\/td>\n
Ratio method reference and fault zone boundaries<\/td>\n
Current edition; apply for regulatory reporting and boundary-case ratio interpretation<\/td>\n<\/tr>\n
\n
IEEE C57.104<\/td>\n
Action threshold levels 1-4; TDCG limits<\/td>\n
Current edition; apply for individual gas and TDCG threshold decisions<\/td>\n<\/tr>\n
\n
OEM transformer manual<\/td>\n
Equipment-specific baseline and cooling data<\/td>\n
Factory acceptance DGA result required as first reference point in trend history<\/td>\n<\/tr>\n
\n
Project specification<\/td>\n
Site-specific alarm levels and response obligations<\/td>\n
Contractual action tiers must match or exceed IEEE C57.104 minimums<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
Core DGA sampling tools, laboratory equipment, and standards references used before interpretation begins.<\/figcaption><\/figure>\n\n
How to Map Fault Types from DGA Gas Ratios<\/h2>\n
Raw gas concentrations tell you what is present. Gas ratios tell you why it is there. Use the methods below as cross-checks rather than isolated pass\/fail rules.<\/p>\n
\n
Core ratio logic:<\/strong> Compare combustible-gas relationships to separate partial discharge, thermal faults, and arcing signatures.<\/li>\n
Rogers ratio method:<\/strong> Useful for clear single-mechanism faults, but it can return undefined results when key gases are near zero.<\/li>\n
IEC 60599 ratio method:<\/strong> Provides a standards-based classification path for reporting and comparison across laboratories.<\/li>\n
Duval Triangle method:<\/strong> Better suited for mixed or evolving faults because it plots the gas pattern instead of relying only on discrete ratio bands.<\/li>\n<\/ul>\n
Ratio Method Comparison<\/h3>\n
\n\n
\n
Criterion<\/th>\n
Rogers Ratios<\/th>\n
IEC 60599<\/th>\n
Duval Triangle<\/th>\n<\/tr>\n<\/thead>\n
\n
\n
Can return undefined result<\/td>\n
Yes<\/td>\n
Yes<\/td>\n
No<\/td>\n<\/tr>\n
\n
Handles mixed faults<\/td>\n
Poor<\/td>\n
Moderate<\/td>\n
Better<\/td>\n<\/tr>\n
\n
Requires C2H2 > 0 for full accuracy<\/td>\n
Yes (R2 fails)<\/td>\n
Yes (ratio 1 fails)<\/td>\n
No<\/td>\n<\/tr>\n
\n
Uses H2 in fault mapping<\/td>\n
Yes (R1)<\/td>\n
Yes<\/td>\n
No<\/td>\n<\/tr>\n
\n
Standards reference<\/td>\n
IEC 60599 \/ IEEE C57.104<\/td>\n
IEC 60599<\/td>\n
IEC 60599<\/td>\n<\/tr>\n
\n
Best use case<\/td>\n
Single-mechanism, clear fault<\/td>\n
Regulatory reporting<\/td>\n
Trending, mixed faults<\/td>\n<\/tr>\n
\n
Risk at low gas concentrations<\/td>\n
High<\/td>\n
High<\/td>\n
Moderate<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
Applying the Ratios to a Field Decision<\/h3>\n
Ratio methods and Duval plotting are used together to classify transformer fault mechanisms with higher confidence.<\/figcaption><\/figure>\n\n
Action Thresholds: When Do DGA Results Require Immediate Response?<\/h2>\n
DGA gas pattern interpretation produces value only when it connects to a clear decision. The tiered action system below reflects IEC 60599 guidance and IEEE C57.104 limits for power transformers rated 69 kV and above.<\/p>\n
Tiered Action Table<\/h3>\n
\n\n
\n
Tier<\/th>\n
Label<\/th>\n
Trigger Conditions<\/th>\n
Required Action<\/th>\n
Time Frame<\/th>\n<\/tr>\n<\/thead>\n
\n
\n
1<\/td>\n
Normal – Continue Monitoring<\/td>\n
All gases below Level 1 limits; ROC stable; no fault ratio flag<\/td>\n
Maintain standard sampling interval<\/td>\n
No urgency<\/td>\n<\/tr>\n
\n
2<\/td>\n
Caution – Increase Sampling<\/td>\n
Any gas between Level 1 and Level 2; ROC >10% per month on any key gas; single ratio flag without corroborating gas rise<\/td>\n
Shorten sampling to monthly; review load history; inspect cooling<\/td>\n
Within 30 days<\/td>\n<\/tr>\n
\n
3<\/td>\n
Warning – Load Reduction and Investigation<\/td>\n
Any gas exceeds Level 2; C2H2 >3 ppm with rising trend; multiple gases rising simultaneously; two or more ratio flags consistent with same fault type<\/td>\n
Reduce load to nameplate; schedule offline inspection within 60 days; increase sampling to weekly<\/td>\n
Within 7 days<\/td>\n<\/tr>\n
\n
4<\/td>\n
Critical – Immediate De-energization<\/td>\n
C2H2 >35 ppm with rapid ROC; H2 >1,800 ppm; CO >1,500 ppm combined with acetylene; any gas doubling in <30 days; Duval plotting in D2 zone<\/td>\n
Remove from service; do not re-energize without internal inspection and engineering sign-off<\/td>\n
Immediate<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
IEEE C57.104 Level 1 and Level 2 Reference Values (ppm dissolved in mineral oil)<\/h3>\n
Step 1 – Screen absolute concentrations.<\/strong> If any gas exceeds Level 2, assign Tier 3 before proceeding. If acetylene exceeds 35 ppm or any gas has doubled since the last sample, assign Tier 4 and stop further analysis pending shutdown. \nStep 2 – Calculate rate of change.<\/strong> A ROC exceeding 1 ppm\/day for acetylene or 10 ppm\/day for hydrogen warrants a minimum Tier 3 assignment regardless of absolute concentration.<\/p>\n
Conditions That Override Standard Thresholds<\/h3>\n
Override the normal tier table when acetylene appears suddenly, any key gas doubles between samples, relay events coincide with gas growth, or an OLTC shares oil with the main tank. In those cases, trend velocity and operating context carry more weight than a single absolute ppm value.<\/p>\n
\n
Field Case: LTC Thermal Fault in a 132\/33 kV Autotransformer<\/h2>\n
This anonymized field case shows how a DGA pattern can move from trend monitoring to outage planning when ethylene and acetylene rise together.<\/p>\n
Field context:<\/strong> A 63 MVA autotransformer commissioned in 2007 averages 28 tap change operations per day. A DGA sample was taken six weeks after an unscheduled sample triggered by a protection relay transient. Oil temperature runs 5-8 deg C above the unit’s benchmark due to increased throughput. \nMeasured gas concentrations (ppm):<\/strong><\/p>\n
Diagnosis:<\/strong> C2H4 is dominant and increasing at approximately 28 ppm per week. The C2H4\/C2H6 ratio of 2.53 is consistent with localized oil temperatures above 500 deg C. The C2H2\/C2H4 ratio of 0.029 indicates low-energy arcing at tap changer contacts as a plausible contributor given the high operation count. Duval Triangle coordinates place the sample in the T2-T3 zone trending toward T3. CO trend is relatively flat, indicating cellulose is not the primary fault material at this stage. This result is Tier 3<\/strong>: load reduction and investigation required within seven days.<\/p>\n
Example field case showing a thermal fault at the LTC interface trending toward a higher-severity condition.<\/figcaption><\/figure>\n\n
Sampling Procedures and Measurement Acceptance Criteria<\/h2>\n
Accurate DGA gas pattern interpretation depends entirely on the quality of the oil sample entering the laboratory. A flawed sample introduces measurement error that no analytical method can correct downstream.<\/p>\n
Sampling Point Selection and Pre-Sample Purge<\/h3>\n
Collect oil from the designated bottom sampling valve after purging stagnant oil from the fitting and hose. Avoid conservator-only samples for fault diagnosis because they can underrepresent heavier gases and distort the gas pattern.<\/p>\n
Procedural Controls That Directly Affect DGA Results<\/h3>\n
\n\n
\n
Control Point<\/th>\n
Required Action<\/th>\n
Consequence of Omission<\/th>\n<\/tr>\n<\/thead>\n
\n
\n
Air exclusion during syringe fill<\/td>\n
Fill syringe while submerged in oil flow; no air bubbles<\/td>\n
Oxygen and nitrogen dilution; artificially lowered fault gas ratios<\/td>\n<\/tr>\n
\n
Syringe over-pressurization prevention<\/td>\n
Back-plunge slightly after fill to seat at 5-10 mL headspace<\/td>\n
Dissolved gas escapes if syringe barrel pressure drops below saturation<\/td>\n<\/tr>\n
\n
Label and chain of custody<\/td>\n
Record transformer ID, MVA rating, voltage class, load at sampling, oil temperature, date and time<\/td>\n
Progressive hydrogen loss from glass after 72 hours documented in CIGRE TB 771<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
Measurement Acceptance Criteria Before Result Use<\/h3>\n
1. Oxygen-to-nitrogen ratio check.<\/strong> In sealed-tank transformers, O2\/N2 should be approximately 0.3-0.5. A ratio above 0.5 indicates air contamination; reject the sample and resample. \n2. Moisture correlation.<\/strong> Verify that dissolved water (ppm by Karl Fischer) is plausible for the insulation class and temperature history. A value above oil saturation at the measured temperature suggests a gross sampling error or seal breach.<\/p>\n\n
Integrating DGA into a Transformer Maintenance Program<\/h2>\n
A DGA result without a defined response path has limited maintenance value. The interpretation becomes actionable only when embedded in a program that specifies who reviews results, at what frequency, against which thresholds, and with what escalation authority.<\/p>\n
Program Structure: Four Functional Layers<\/h3>\n
\n\n
\n
Layer<\/th>\n
Function<\/th>\n
Typical Owner<\/th>\n<\/tr>\n<\/thead>\n
\n
\n
Sampling<\/td>\n
Collect oil samples at defined intervals<\/td>\n
Field technician<\/td>\n<\/tr>\n
\n
Analysis<\/td>\n
Run chromatographic tests, generate gas concentrations<\/td>\n
Laboratory or on-site monitor<\/td>\n<\/tr>\n
\n
Interpretation<\/td>\n
Apply ratio methods, compare to thresholds, classify fault type<\/td>\n
Engineer or diagnostic specialist<\/td>\n<\/tr>\n
\n
Action<\/td>\n
Authorize load reduction, inspection, or outage<\/td>\n
Asset manager or operations lead<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
Sampling Frequency Decision Logic<\/h3>\n
Set the next sampling interval from both TDCG level and rate of change. Stable Level 1 results can stay on routine intervals, while rising Level 2 or Level 3 results need shorter intervals and named escalation owners.<\/p>\n
TDCG Levels and Program Responses<\/h3>\n
\n\n
\n
TDCG Level<\/th>\n
Concentration Range (ppm)<\/th>\n
Program Response<\/th>\n<\/tr>\n<\/thead>\n
\n
\n
Level 1<\/td>\n
Below 720<\/td>\n
Continue normal sampling interval; no action required<\/td>\n<\/tr>\n
\n
Level 2<\/td>\n
720-1,920<\/td>\n
Increase sampling frequency; review individual gas trends; apply Duval Triangle<\/td>\n<\/tr>\n
\n
Level 3<\/td>\n
1,921-4,630<\/td>\n
Sample every 1-4 weeks; prepare contingency plan; consider load reduction if trend is rising<\/td>\n<\/tr>\n
\n
Level 4<\/td>\n
Above 4,630<\/td>\n
Consider immediate de-energization; consult engineering before next energization<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
Where Programs Break Down<\/h3>\n
Interpretation without context.<\/strong> A gas concentration reviewed without previous sample history, load profile, or transformer age produces unreliable conclusions; interpreters must have access to the full DGA history. \nAction authority gaps.<\/strong> If the engineer who interprets the result cannot authorize a load reduction or outage, and the person who can does not receive the interpretation, the program stalls. Define the escalation path explicitly, including who receives the report and within what timeframe.<\/p>\n
A structured DGA program depends on clear handoff from sampling and analysis to interpretation and action authority.<\/figcaption><\/figure>\n\n
Specifying DGA-Ready Transformers and Requesting Supplier Diagnostic Support<\/h2>\n
Effective DGA gas pattern interpretation begins before a transformer is energized. Procurement decisions at the specification stage determine whether a unit can be monitored cost-effectively throughout its service life.<\/p>\n
What to Specify at the Procurement Stage<\/h3>\n
Oil sampling valve placement.<\/strong> Require at minimum one bottom-mounted sampling valve and one top-oil valve, both rated for syringe or vacuum-bottle extraction without de-energizing. Reject sampling valves located above oil level for critical assets due to air ingress risk. \nBuchholz relay and gas collection.<\/strong> For units above 1 MVA, specify a Buchholz relay with a gas collection chamber that permits syringe extraction, positioned on the pipe run between tank and conservator.<\/p>\n
ISO\/IEC 17025 accreditation with DGA listed in scope<\/td>\n<\/tr>\n
\n
Sampling personnel qualification<\/td>\n
Technicians certified to IEC 60567 sampling procedure or equivalent documented training<\/td>\n<\/tr>\n
\n
Turnaround time<\/td>\n
Written commitment: routine results within 5 business days; urgent flag results within 24 hours<\/td>\n<\/tr>\n
\n
Interpretation service<\/td>\n
Named process for escalation beyond raw numbers: ratios applied, trend context reviewed<\/td>\n<\/tr>\n
\n
Reporting format<\/td>\n
Structured report including trend comparison, ratio analysis, and recommended action tier<\/td>\n<\/tr>\n
\n
Equipment traceability<\/td>\n
Calibration records for gas chromatograph available on request<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
Retrofitting DGA Capability on Existing Units<\/h3>\n
For existing transformers, retrofit value depends on access to a reliable oil sampling point, safe isolation practices, and whether the asset is critical enough to justify online monitoring. At minimum, add a documented sampling procedure and baseline DGA before relying on trend interpretation.<\/p>\n
Measures micro-ohm contact condition under controlled current<\/td>\n
Multimeter readings cannot support alarm-limit decisions<\/td>\n<\/tr>\n
\n
Factory acceptance test report<\/td>\n
Provides serial-number baseline and test condition<\/td>\n
No valid comparison point for site trending<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
Field Example<\/h2>\n
Field example: during a service inspection, one phase measured outside its commissioning baseline while the other two phases remained stable. The team repeated the measurement with verified leads, checked timing and contact travel, and used the measured divergence to separate a contact-pressure problem from a generic surface-cleaning issue. The corrective action was documented in the troubleshooting chart so the next DGA sample, inspection note, and maintenance record could be compared against the same fault map.<\/p>\n
Frequently Asked Questions<\/h2>\n
What is the most important single gas to monitor in a transformer DGA result?<\/h3>\n
Acetylene (C2H2) carries the highest diagnostic weight because it is only produced in significant quantities by high-energy electrical discharges. Any confirmed detection above 1-2 ppm in a sealed transformer with no recent through-fault history warrants investigation.<\/p>\n
How often should DGA sampling be performed on a critical power transformer?<\/h3>\n
Sampling frequency should be risk-stratified rather than uniform. A new transformer with no fault history typically requires annual sampling.<\/p>\n
Can DGA be performed on dry-type transformers?<\/h3>\n
No. DGA applies specifically to oil-filled transformers because the diagnostic gases are produced by thermal and electrical decomposition of insulating oil and oil-impregnated cellulose.<\/p>\n
What does a high CO\/CO2 ratio indicate, and when is it alarming?<\/h3>\n
A CO\/CO2 ratio below 0.1 is consistent with normal background paper aging. Ratios above 0.3 indicate active cellulose degradation involving thermal mechanisms.<\/p>\n
Why do ratio methods sometimes give contradictory or undefined results?<\/h3>\n
Ratio methods fail at low absolute gas concentrations because small laboratory measurement uncertainties produce large swings in computed ratio values. They also fail when multiple fault types are active simultaneously.<\/p>\n
What sampling errors most commonly invalidate a DGA result?<\/h3>\n
The three most common invalidating errors are: air contamination during syringe filling, which depresses fault gas concentrations and raises the O2\/N2 ratio above 0.5; exceeding the 72-hour holding time for glass syringes, which causes measurable hydrogen loss; and drawing the sample from the top of the tank or the conservator rather than the bottom drain valve, which underreports heavier fault gases. Any result showing an O2\/N2 ratio above 0.5 in a sealed-tank transformer should be rejected and a fresh sample collected before any maintenance decision is made.<\/p>\n
How should DGA be interpreted when the OLTC shares oil with the main tank?<\/h3>\n
An OLTC sharing oil with the main tank introduces a persistent C2H2 background from normal contact arcing during tap changes. This background must be established as a unit-specific baseline rather than compared directly to standard IEEE or IEC tables.<\/p>\n
![\"Transformer](\"https:\/\/xbrele.com\/wp-content\/uploads\/2026\/05\/dga-tools-and-acceptance-sources-1.webp\")
![\"Comparison](\"https:\/\/xbrele.com\/wp-content\/uploads\/2026\/05\/dga-ratio-methods-fault-mapping-1.webp\")
![\"Autotransformer](\"https:\/\/xbrele.com\/wp-content\/uploads\/2026\/05\/ltc-interface-thermal-fault-dga-scenario-1.webp\")
![\"Workflow](\"https:\/\/xbrele.com\/wp-content\/uploads\/2026\/05\/transformer-dga-maintenance-program-workflow-1.webp\")