What Is Dissolved Gas Analysis?

Dissolved gas analysis (DGA) is a diagnostic technique used to assess the condition of oil-filled electrical equipment, most commonly power transformers, by identifying and quantifying gases that accumulate in the insulating oil as a result of thermal or electrical stress. When insulating materials decompose under fault conditions, they release characteristic hydrocarbon and non-hydrocarbon gases that dissolve in the oil, providing a chemical record of internal activity that is otherwise invisible without opening the equipment. DGA is one of the most widely adopted predictive maintenance tools in the power industry, covered under IEEE Std C57.104, which establishes permissible gas concentration limits and condition categories for oil-immersed transformers.

The technique draws from analytical chemistry, electrical engineering, and materials science. Oil sampling is performed while equipment remains energized, making DGA non-invasive and suitable for routine monitoring programs without causing service interruptions.

Diagnostic Gases and Fault Signatures

The gases detected in DGA fall into two broad classes: hydrocarbon gases generated from the decomposition of mineral oil, and gases produced from the breakdown of cellulosic insulation. The principal hydrocarbon gases are hydrogen (H2), methane (CH4), ethylene (C2H4), ethane (C2H6), and acetylene (C2H2). Carbon monoxide (CO) and carbon dioxide (CO2) indicate cellulose degradation. Each fault type produces a characteristic gas profile: partial discharge generates primarily hydrogen and small quantities of methane; overheating produces methane, ethane, and ethylene; and arcing, the most severe fault mode, produces acetylene as a signature compound alongside high levels of hydrogen. The relative concentrations and ratios of these gases, rather than any single gas in isolation, allow diagnosticians to distinguish among fault types and assess severity.

Interpretation Methods

Several standardized interpretation methods translate raw gas concentrations into actionable diagnoses. The Doernenburg Ratio Method and the Roger Ratio Method, both developed in the 1970s, compute ratios between pairs of key gases and compare them against fault-type lookup tables. The IEC Ratio Method, codified in IEC 60599, extends this approach with an updated three-ratio scheme. The Duval Triangle Method, introduced by Michel Duval, plots the relative percentages of methane, ethylene, and acetylene on a ternary diagram and identifies fault zones geometrically. More recently, machine learning classifiers have been applied to improve diagnostic consistency; research published in PMC by the National Institutes of Health demonstrated that a random forest ensemble integrating five conventional methods achieved 96 percent accuracy on three-fault classification, substantially outperforming any single ratio method.

Condition Monitoring Practice

In a typical condition monitoring program, oil samples are extracted from the transformer drain valve at scheduled intervals and analyzed in a laboratory using gas chromatography. Key diagnostic outputs include the total dissolved combustible gas (TDCG) concentration, the rate of gas generation between successive samples, and the pattern of individual gas ratios. IEEE Std C57.104 organizes TDCG levels into four condition categories: Condition 1 indicates normal operation; Condition 4 indicates that immediate action may be required. Rate of change is often more informative than absolute concentration, as a rapidly increasing TDCG suggests a developing fault even when absolute levels remain within normal bounds. Online DGA monitors, which sample and analyze gases continuously, extend the technique to real-time surveillance of critical high-value transformers. The ScienceDirect overview of transformer DGA research documents the breadth of methods and instrumentation now applied across the industry.

Applications

Dissolved gas analysis has applications across a range of domains, including:

• Power utility asset management, for scheduling maintenance and avoiding unplanned outages
• Substation condition monitoring programs covering generator step-up and autotransformers
• Industrial facilities where large oil-filled transformers support manufacturing processes
• Forensic investigation of transformer failures after fault events