The Duval Triangle for Transformer Fault Diagnosis

Introduction

Dissolved Gas Analysis (DGA) is a vital diagnostic technique for assessing the condition of mineral oil-immersed transformers. By examining the gases dissolved in the transformer's insulating oil, DGA helps detect potential faults before they escalate into catastrophic failures. The concentrations of specific gases reveal the nature of electrical or thermal stresses within the transformer. Accurate interpretation of DGA results is critical, and among the various methods available, the Duval Triangle stands out as a widely adopted and reliable approach. Specifically, Duval Triangle 1 is a graphical tool designed to classify transformer faults based on the concentrations of three key hydrocarbon gases: methane (CH₄), ethylene (C₂H₄), and acetylene (C₂H₂). This article explores the concept of the Duval Triangle 1, the procedure for calculating and plotting DGA points, and the six basic fault types it identifies, as outlined in IEEE Std C57.104-2019.

The Duval Triangle Concept

The Duval Triangle 1 is a triangular coordinate system used to interpret DGA results by mapping the relative percentages of CH₄, C₂H₄, and C₂H₂. Each corner of the triangle represents 100% of one gas and 0% of the other two: CH₄ at the top vertex, C₂H₄ at the bottom left, and C₂H₂ at the bottom right. The interior of the triangle is divided into distinct zones, each corresponding to a specific type of fault that may occur within a transformer. By plotting a point based on the gas concentrations from a DGA sample, operators can visually identify the likely fault type.

This method, as described in section 6.2.3 of IEEE Std C57.104-2019, was developed to enhance the accuracy and reliability of fault diagnosis compared to earlier techniques. The Duval Triangle leverages the ratios of the three gases to provide a more nuanced interpretation, reflecting the thermal and electrical conditions inside the transformer. The zones within the triangle (figure 1) are empirically derived from extensive research and validated against real-world fault cases, making it a robust tool for transformer maintenance.

The fault zones of the Duval Triangle 1 are:

• Partial Discharge (PD): Low-energy discharges often linked to high hydrogen and methane levels.

• Discharges of Low Energy (D1): Surface or tracking discharges, more energetic than PD but less than arcing.

• Discharges of High Energy (D2): High-energy events like arcing, typically marked by elevated acetylene.

• Thermal Faults < 300°C (T1): Low-temperature overheating, producing methane and ethane.

• Thermal Faults 300°C to 700°C (T2): Moderate-temperature faults with increased ethylene.

• Thermal Faults > 700°C (T3): High-temperature faults generating significant ethylene and acetylene.

• Mixtures of Electrical and Thermal Faults (DT).

Figure 1 - Duval Triangle 1 Method

Table 1 - Fault zone boundaries for figure 1

Gas % / Fault	%CH₄	%C₂H₄	%C₂H₂
PD	≥ 98	-	-
T1	< 98	< 20	< 4
T2	-	≥ 20 and < 50	< 4
T3	-	≥ 50	< 15
DT	-	< 50	≥ 4 and < 13
	-	≥ 40 and < 50	≥ 13 and < 29
	-	≥ 50	≥ 15 and < 29
D1	-	< 23	≥ 13
D2	-	≥ 23	≥ 29
	-	≥ 23 and < 40	≥ 13 and < 29

Calculating and Displaying DGA Points

To utilize the Duval Triangle 1, one must calculate the relative percentages of CH₄, C₂H₄, and C₂H₂ from a DGA sample and plot them on the triangle. Annex D.4 of IEEE Std C57.104-2019 outlines the following procedure:

1. Obtain Gas Concentrations: Get the concentrations of CH₄, C₂H₄, and C₂H₂ in parts per million (ppm) from the DGA report.

2. Calculate Total Concentration: Sum the concentrations of the three gases: Total = CH₄ + C₂H₄ + C₂H₂.

3. Compute Relative Percentages: Determine the percentage of each gas relative to the total:

   • %CH₄ = (CH₄ / Total) × 100

   • %C₂H₄ = (C₂H₄ / Total) × 100

   • %C₂H₂ = (C₂H₂ / Total) × 100

     These percentages always sum to 100%, aligning with the triangular coordinate system.

4. Plot the Point: Using the calculated percentages, locate the corresponding point on the Duval Triangle. The position within a specific zone indicates the fault type.

Example Calculation

Consider a DGA sample with the following concentrations:

CH₄ = 25 ppm

C₂H₄ = 15 ppm

C₂H₂ = 10 ppm

Total Concentration: 25 + 15 + 10 = 50 ppm

Relative Percentages:

%CH₄ = (25 / 50) × 100 = 50%

%C₂H₄ = (15 / 50) × 100 = 30%

%C₂H₂ = (10 / 50) × 100 = 20%

Using these values (50%, 30%, 20%), the point is plotted on the Duval Triangle. This can be done manually on a printed triangle or via software tools that automate the plotting and interpretation process. The resulting position within a fault zone provides diagnostic insight.

Figure 2 - Example of Duval Triangle 1 Plot

Note: It is not recommended to attempt fault identification using the method described above if all gas levels are below the values of IEEE Std C57.104-2019 Table 1.

Types of Faults Identified by the Duval Triangle

Annex C.1 of IEEE Std C57.104-2019 defines the six basic types of faults that the Duval Triangle 1 can identify. These faults reflect different thermal and electrical stresses within the transformer:

1. Partial Discharges (PD): Partial discharges of the cold plasma (corona) type, resulting in possible X-wax deposition on paper insulation.

2. Discharges of Low Energy (D1): Discharges in mineral oil and/or paper, evidenced by larger carbonized perforations through paper (punctures), carbonization of the paper surface (tracking), carbon particles in mineral oil (as in tap changer diverter operation), or partial discharges of the sparking type, inducing pinhole or carbonized perforations (punctures) in paper.

3. Discharges of High Energy (D2): High-energy electrical discharges in mineral oil and/or paper, with power follow-through, evidenced by extensive destruction and carbonization of paper, metal fusion at the discharge extremities, extensive carbonization in mineral oil and, in some cases, tripping of the equipment, confirming the large current follow-through.

4. Thermal Faults Less Than 300°C (T1): Low-temperature overheating, typically due to minor overloads or inadequate cooling. These faults primarily affect the oil and/or paper, with possible evidence of brownish paper.

5. Thermal Faults Between 300°C and 700°C (T2): Moderate-temperature thermal faults, affecting both oil and solid insulation, often caused by localized hotspots in windings or the core.

6. Thermal Faults Greater Than 700°C (T3): Severe high-temperature faults that rapidly degrade both oil and solid insulation, with strong evidence of carbonization of the mineral oil (>700 ºC), metal discoloration (800 ºC) or metal fusion (>1000 ºC).

Understanding these fault types enables operators to prioritize maintenance actions, such as inspections, load adjustments, or repairs, based on the severity and nature of the detected issue.

Conclusion

The Duval Triangle 1 is a cornerstone of DGA interpretation for mineral oil-immersed transformers, offering a clear, visual method to classify faults based on CH₄, C₂H₄, and C₂H₂ concentrations. Its structured approach—rooted in empirical data and standardized in IEEE Std C57.104-2019—enhances the early detection and diagnosis of transformer issues, helping prevent failures and extend equipment lifespan. By following the calculation and plotting procedure, operators can pinpoint fault types and assess their implications using the six basic fault categories. When integrated with other diagnostic tools and expert analysis, the Duval Triangle 1 significantly bolsters the reliability and effectiveness of transformer maintenance programs.

Our Solutions

The HV Assets Care Platform is a complete solution for data analysis and diagnostics. It includes all the recommended methods from the IEEE standard, including the Duval Triangle and the advanced Combined Duval Pentagon, integrated in an Asset Management Dashboard. In addition, provides a Health Index with individual scores to create an Asset Ranking. For more information, click here.

References

"IEEE Guide for the Interpretation of Gases Generated in Mineral Oil-Immersed Transformers," in IEEE Std C57.104-2019 (Revision of IEEE Std C57.104-2008) , vol., no., pp.1-98, 1 Nov. 2019, doi: 10.1109/IEEESTD.2019.8890040.