Duval Triangle

The method uses a ternary plot where each side represents 0% to 100% of a single hydrocarbon gas. The relative percentage of %CH₄, %C₂H₄, and %C₂H₂ is calculated from their total sum, forcing the data point to fall within the triangle. This normalization eliminates the influence of absolute gas concentration, making the diagnosis independent of oil volume and fault severity. The triangle is divided into six fault zones (plus a stray gas zone) corresponding to specific thermal and electrical failure mechanisms.

The triangle partitions into distinct diagnostic regions:

• PD: Partial discharge (corona in gas bubbles)
• T1: Thermal fault below 300°C
• T2: Thermal fault between 300°C and 700°C
• T3: Thermal fault above 700°C
• D1: Low-energy electrical discharge
• D2: High-energy electrical discharge (arcing)
• DT: Mixed thermal and electrical faults

Each zone maps to specific physical degradation mechanisms, from cellulose pyrolysis in T2/T3 zones to copper vaporization in D2 arcing events.

The Duval Triangle implicitly encodes hydrocarbon ratio logic without requiring manual ratio calculation. Key diagnostic indicators include:

• High C₂H₂ (>30%): Strongly indicates arcing (D1/D2 zones), as acetylene forms only at temperatures exceeding 700°C
• High C₂H₄ (>50%): Points toward thermal faults (T2/T3), since ethylene dominates during oil overheating
• High CH₄ (>80%): Suggests low-temperature thermal faults (T1) or partial discharge
• Balanced distribution: Often indicates mixed thermal-electrical faults (DT zone)

This visual approach eliminates the ambiguity of fixed ratio codes like IEC 60599 Rogers ratios.

By plotting sequential DGA samples on the triangle over time, engineers can visualize fault trajectory as a moving point. A point migrating from the T1 zone toward T2/T3 indicates escalating thermal severity, while movement toward D1/D2 signals developing electrical discharge activity. This temporal tracking enables condition-based maintenance decisions—a fault drifting toward the D2 arcing zone demands immediate intervention, while a stable T1 reading may warrant continued monitoring. The trajectory path itself becomes a diagnostic signature.

The Duval Triangle has specific constraints:

• Excludes hydrogen (H₂) and ethane (C₂H₆), which are critical for detecting partial discharge and low-temperature thermal faults respectively
• Cannot distinguish between oil and paper involvement in thermal faults—requires CO and CO₂ analysis
• Single-point ambiguity can occur near zone boundaries
• Not applicable to non-mineral oils or silicone fluids

For comprehensive diagnosis, the triangle is typically combined with Duval Pentagon (adding H₂ and C₂H₆) and IEC 60599 basic gas ratios for cross-validation.

Modern online DGA monitors embed Duval Triangle logic directly into firmware for real-time fault classification. The algorithm:

1. Extracts CH₄, C₂H₄, and C₂H₂ concentrations from multi-gas sensors
2. Calculates relative percentages
3. Maps coordinates to predefined zone polygons
4. Triggers SCADA alarms with fault type and recommended actions

This automation enables edge AI deployment where classification runs locally on substation gateways without cloud dependency, providing sub-second fault alerts to asset managers.

The foundational laboratory technique that measures concentrations of hydrogen (H₂), methane (CH₄), acetylene (C₂H₂), ethylene (C₂H₄), and ethane (C₂H₆) dissolved in transformer oil. DGA provides the raw gas values that the Duval Triangle plots to classify faults. Without accurate DGA sampling following IEC 60567 or ASTM D3612 procedures, triangle interpretation becomes unreliable.

• Key gases correlate to specific fault types
• Requires oil sampling from transformer drain valve
• Results expressed in parts per million (ppm)

The international standard that defines normal gas concentration limits, gas ratio methods, and diagnostic zones for mineral oil-filled equipment. While the Duval Triangle is a popular graphical tool, IEC 60599 provides the authoritative framework for interpreting whether gas levels indicate normal aging, possible fault, or active fault conditions.

• Defines 90th percentile typical gas values
• Specifies sampling and analysis requirements
• Complements Duval Triangle with Rogers Ratio method

A permanently installed multi-gas sensor that provides continuous real-time dissolved gas readings without manual oil sampling. Modern monitors use photo-acoustic spectroscopy or gas chromatography to feed automated Duval Triangle analysis every few hours, enabling immediate alarming when fault zones shift.

• Eliminates sampling lag between lab tests
• Enables trend analysis and rate-of-change alerts
• Critical for unattended substations

The supervised machine learning task that maps DGA patterns to specific fault categories: partial discharge (PD), thermal faults below 300°C (T1), thermal faults 300-700°C (T2), thermal faults above 700°C (T3), and electrical arcing (D1/D2). The Duval Triangle provides the labeled training data structure for these models.

• Random Forest and XGBoost commonly applied
• Requires historical fault confirmation for labels
• Extends triangle logic with additional gas ratios

A complementary diagnostic test that measures furanic compounds (2-FAL, 5-HMF, 2-FOL, 2-ACF, 5-MEF) dissolved in oil as chemical markers of solid paper insulation degradation. While the Duval Triangle identifies active faults in oil, furan analysis reveals the aging state of cellulose insulation, which cannot be detected by gas analysis alone.

• Directly correlates to Degree of Polymerization (DP)
• 2-FAL is the primary degradation marker
• Combined with DGA for complete health assessment

Methods like SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) applied to machine learning models that automate Duval Triangle classification. XAI provides asset managers with feature attribution scores showing exactly which gas ratios drove a fault classification, bridging the gap between black-box AI and traditional graphical methods.

• SHAP values quantify each gas contribution
• Builds trust with reliability engineers
• Required for regulatory compliance in some jurisdictions