A practical guide to turning an abnormal oil result into a defensible transformer-management decision

The laboratory report arrives. A gas concentration is highlighted. The diagnostic tool indicates a possible fault. Now comes the question that matters: “What should we do?”
Take another sample? Reduce the load? Investigate the cooling system? Plan an outage? Or remove the transformer from service?
Dissolved gas analysis can reveal developing thermal and electrical problems inside a transformer. But identifying a possible fault is not the same as understanding its cause, establishing its urgency, or selecting the right intervention. Gas patterns, trends, and operating history must be considered together. [1]
A DGA result is not a maintenance instruction. It is evidence that must be turned into a decision.
From Diagnostic Evidence to the Right Decision
When dissolved gas analysis identifies a potential problem, the challenge is not simply interpreting the gases. It is deciding what action is justified by the available evidence.
The following decision-making flow chart provides a structured approach, guiding asset managers and engineers through six essential stages:
- Verify: Confirm that the intervention has controlled the problem.
- Check: Establish whether immediate danger exists.
- Validate: Confirm that the DGA results are reliable.
- Analyse: Examine trends and operating conditions.
- Diagnose: Identify the likely fault mechanism.
- Act: Select an appropriate risk-based intervention.

An important feature of this framework is that the process is not always linear.
An uncertain laboratory result may require confirmation sampling. A suspected serious internal fault may demand immediate action before further diagnostic investigation. An intervention that fails to stabilise the condition requires reassessment rather than simply closing the maintenance task.
The objective is to ensure that every significant DGA warning leads to a technically defensible decision, supported by evidence and appropriate operational risk controls.
1. First ask: Is there an immediate safety concern?
Before spending time interpreting a diagnostic plot, establish what is happening at the transformer now.
As an initial response checklist, ask:
- Has any protection operated, or has a gas, pressure or temperature alarm appeared?
- Are there reports of abnormal noise, smoke, significant leakage or other distress?
- Has the cooling system failed or become unavailable?
- Is the DGA showing a new or rapidly increasing gas pattern associated with electrical discharge?
- Has the change followed a system disturbance or another significant operating event?
Acetylene is an important warning indicator associated with arcing, while the rate of gas generation and the wider operating context help establish the significance of the result. A gas name alone is not a complete diagnosis. [1, 5]
Recommended response: Where immediate danger or a potentially serious internal fault is suspected, escalate through the site’s emergency and engineering arrangements. Do not delay protective action simply to obtain another laboratory result. Decisions on continued operation, load reduction or isolation belong with the authorised operating and engineering personnel.
Equally, do not treat every highlighted laboratory value as an automatic shutdown instruction. The response must be proportionate to the evidence and consequences.
2. Validate the result, without dismissing the warning
An unexpected result deserves investigation, not the assumption that either the transformer or the laboratory is wrong.
DGA depends on obtaining a representative oil sample, retaining the dissolved gases and analysing them correctly. Sample integrity therefore matters before interpretation begins. [6]
Use the following verification checklist:
Confirm the asset and compartment
Check the transformer identification, sample date and sampling point. Establish whether the result belongs to the main tank or a separate tap-changer compartment.
Review sample quality
Ask whether the approved gas-tight sampling method was followed and whether leakage, air bubbles, handling or transport problems were recorded.
Speak to the laboratory
Clarify unusual results, reporting limits and any analytical qualifications. A result reported as “not detected” should not automatically be treated as an exact zero.
Check recent interventions
Record oil replacement, topping-up, processing, degassing and maintenance that could affect comparisons with earlier samples.
Recommended response: If the result is questionable and operating conditions permit, arrange a prompt confirmation sample. Set its urgency according to the suspected fault and current risk, rather than automatically returning to the routine sampling interval.
Verify the evidence, but do not use verification as a reason to postpone an urgent response.
3. Read the trend, not just the latest number
Your transformer’s history is part of the diagnosis.
A single DGA result provides a snapshot. Successive samples show whether the condition is stable, developing, or changing character. Gas patterns and generation rates can be more informative than individual concentrations. [1]
When reviewing the history, ask:
- Which gases have changed?
- Over what actual time interval?
- Is the increase sustained or based on one unexpected sample?
- Has a new gas appeared?
- Has the dominant gas pattern changed?
- Were loading, temperature, and cooling conditions comparable?
- Has oil treatment interrupted the historical trend?
The practical distinction
An elevated concentration that has remained broadly stable presents a different diagnostic question from a lower concentration that is increasing rapidly. Neither should be assessed from concentration alone. The trend must be interpreted alongside operating conditions and the transformer’s history. [1]
Recommended practice: Plot individual gases against actual sample dates and review them alongside load, temperature, alarms, and maintenance events. Where sample intervals differ, avoid comparing the size of each increase without considering the elapsed time.
What changed in the transformer’s operation before the gases changed?
4. Diagnose the likely mechanism, then look for supporting evidence
Gas patterns help narrow the investigation. They do not normally identify the exact defective component on their own.
The Key Gas Method associates hydrogen with partial-discharge patterns, methane and ethane with lower-temperature oil overheating, ethylene with higher-temperature oil overheating, and hydrogen with acetylene with arcing. Real samples may not fit ideal profiles, and multiple mechanisms can complicate interpretation. [5]
Use those patterns to guide questions, not to jump straight to a repair.
If the pattern suggests a thermal problem
Start by reviewing the heat-removal system and operating conditions:
- Are the required fans and pumps available and operating correctly?
- Are cooling controls responding as intended?
- Is there evidence of restricted oil circulation?
- Are radiators or heat exchangers obstructed or fouled?
- Has loading or ambient temperature changed?
- Could localised heating, circulating currents or an internal connection problem be involved?
Cooling-system problems, overloading, circulating currents and stray-flux heating are relevant investigation areas. [2]
Important distinction: Finding a cooling defect gives you a credible lead. It does not, by itself, prove that every abnormal gas result has been explained.
If the pattern suggests electrical discharge
Recommended investigation: Review protection and disturbance records, assess whether the gas pattern is new or progressing, and obtain specialist advice on appropriate discharge investigations or outage testing.
Do not assume that a normal external appearance excludes an internal problem. DGA is valuable precisely because it provides evidence of processes occurring within the insulation system. [1]
If solid-insulation involvement is suspected
Carbon monoxide and carbon dioxide are usually associated with cellulose degradation, but their interpretation should form part of a wider assessment rather than a stand-alone verdict on paper condition. [1, 5]
Recommended investigation: Consider moisture, loading and temperature history, relevant oil-quality results and suitable paper-ageing indicators with specialist input.
Use diagnostic tools as supporting evidence
Power Transformer Health provides several interpretation approaches, including gas-ratio methods, Duval methods and LEDT-related case studies. These can help structure the diagnosis, but agreement between plots does not establish the physical fault location or replace an engineering assessment. [2–4]
A diagnostic zone suggests a fault mechanism. It does not tell you which bolt to tighten.
5. Turn the diagnosis into a risk-based action plan
The next step is not simply to write “monitor closely”.
A useful action plan must explain what will be monitored, how often, by whom and what would trigger a different response.
The following is a proposed decision framework, not a set of standard alarm categories:
| Assessment Outcome | Practical Response to Consider |
|---|---|
| Result uncertain, with no corroborating signs of distress | Verify the data and obtain a confirmation sample at an agreed interval. |
| Abnormal condition confirmed, with no evidence of rapid progression | Investigate the cause and establish enhanced monitoring with explicit escalation criteria. |
| Condition progressing or supported by other concerning evidence | Urgent engineering review, operational risk controls and assessment of outage requirements. |
| Immediate danger or suspected serious internal fault | Follow emergency procedures and authorised isolation arrangements. |
For each case, document:
- The evidence: What changed, and how reliable is the information?
- The working diagnosis: What is suspected, and what remains uncertain?
- The operating decision: Under what conditions may the transformer remain in service?
- The investigation: Which inspection or test will answer the next important question?
- The trigger: What result, alarm or operating change requires escalation?
- The ownership: Who is responsible, and when is the next review?
Do not prescribe the same resampling interval for every abnormal result. The suspected mechanism, progression, monitoring capability, and consequences of failure should inform the interval.
6. Verify that the intervention worked
The job is not complete when a fan is repaired, a connection is attended to, or an oil-processing plant is disconnected.
The job is complete when the follow-up evidence supports the conclusion that the condition is controlled.
Recommended follow-up questions include:
- Have the relevant gas trends stabilised?
- Are temperature and cooling performance consistent with the intended operating condition?
- Has the suspected cause actually been corrected?
- Are further tests or restrictions still required?
- Has a new post-intervention baseline been recorded?
- Is there enough evidence to return to routine monitoring?
Be particularly careful after oil processing or replacement. A lower gas concentration may reflect removal or dilution of the gases, rather than removal of the mechanism that produced them. Record the intervention and assess subsequent gas behaviour before declaring success.
Removing the gases is not the same as removing the fault.
Remember: The decision-making process does not end when maintenance is completed. It ends when sufficient evidence demonstrates that the transformer condition is controlled, or when an appropriate revised action plan has been established.
A practical example: From a thermal warning to a controlled response
A transformer’s DGA indicates an increasing thermal-fault pattern. The investigation also finds higher operating temperatures and an unavailable cooling bank.
A weak response would be:
“Cooling fault found. Repair completed. Close the DGA issue.”
A stronger response would be:
- Assess whether continued operation is acceptable under the available cooling capacity.
- Restore the cooling equipment under approved procedures.
- Confirm that the equipment operates correctly and temperatures respond as expected.
- Obtain follow-up DGA at a risk-based interval.
- Investigate further if the gas trend continues despite restored cooling.
The cooling defect is a plausible explanation. The follow-up evidence determines whether it was a sufficient explanation.
Five mistakes to avoid
- Treating one gas as a complete diagnosis – Interpret the gas pattern, history, and operating context together. [1, 5]
- Treating a diagnostic plot as a shutdown instruction – Use the diagnosis to inform a documented engineering and operational decision.
- Waiting for the next routine sample despite evidence of deterioration – Set the next review according to the risk.
- Processing the oil before preserving the diagnostic record – Where safe and practicable, retain the pre-intervention results and relevant operating evidence.
- Closing the issue without checking the response – A completed maintenance task is not, on its own, proof of a resolved condition.
The bottom line
When DGA identifies a possible problem, the objective is neither to panic nor to explain the warning away.
It is to: Check immediate danger. Validate the evidence. Understand the trend. Investigate the mechanism. Act on the risk. Verify the outcome.
The value of DGA is not the warning it gives. It is the quality of the decision that follows.
Need help turning an abnormal DGA result into an action plan?
Use the Free DGA Tool to explore your gas results and trends, or request a transformer health review through Power Transformer Health. For a meaningful assessment, include historical DGA results, sample dates, transformer details, operating conditions, recent maintenance and any relevant alarms. The tool supports interpretation; it does not authorise continued operation. [7]
Request a Transformer Health Review
References and Further Reading
- Power Transformer Health (2025). The Transformer Speaks in Dissolved Gases — The Wise Asset Manager is Fluent in Its Silent Language. Background on fault gases, diagnostic methods, trend interpretation and correlation with operating conditions.
- Power Transformer Health (2024). Mitigation of Thermal Faults in Power Transformers. Practical background on overheating, cooling-system defects, circulating currents, stray-flux heating and complementary diagnostics.
- Power Transformer Health (2024). Dissolved Gas Analysis in Transformers: A Case Study Approach. Case-study discussion comparing diagnostic interpretations using LEDT and Duval methods.
- Power Transformer Health (2023). Duval’s Pentagon Method for Transformer Fault Diagnosis. Explanation of the five-gas diagnostic approach and its fault-classification regions.
- Power Transformer Health (2017). Dissolved Gas Analysis: Key Gas Method Overview and Application. Overview of characteristic gas patterns and the limitations of individual gases and idealised profiles.
- National Institutes of Health, Office of Research Facilities (2017). Dissolved Gas Analysis of Power Transformer. Technical News Bulletin, Issue 65, June. Introductory technical background on transformer oil, gas generation, sampling and laboratory analysis.
- Power Transformer Health (undated). Dissolved Gas Analysis. Practical tool for entering successive oil-sample results and reviewing diagnostic outputs and trends.
Technical note
The six-step response framework, investigation checklists and illustrative scenario in this article are practical guidance developed for this post. They are not reproduced requirements from a particular standard. Operational decisions must follow applicable site procedures, manufacturer guidance and an assessment by appropriately qualified personnel.
The references support the article’s technical background without implying that standards not directly consulted were used to verify every recommendation.