IEC 60599

IEC 60599 establishes 90th percentile typical values for dissolved gas concentrations in healthy transformers, providing a statistical baseline for anomaly detection. These limits are not absolute pass/fail criteria but serve as attention flags.

• Hydrogen (H₂): 50-150 ppm depending on transformer type
• Methane (CH₄): 30-130 ppm
• Acetylene (C₂H₂): 2-20 ppm (any presence in main tank signals concern)
• Ethylene (C₂H₄): 60-280 ppm
• Ethane (C₂H₆): 20-90 ppm
• Carbon Monoxide (CO): 500-600 ppm

Exceeding these values triggers further investigation using ratio-based diagnostics.

The standard defines three diagnostic ratios using five key hydrocarbons to classify faults into six distinct categories. This method eliminates the influence of oil volume, providing a normalized diagnostic approach.

The Three Ratios:

• C₂H₂/C₂H₄: Indicates electrical fault intensity (discharge vs. thermal)
• CH₄/H₂: Distinguishes between partial discharge and thermal faults
• C₂H₄/C₂H₆: Reflects the temperature of thermal faults

Each ratio is coded as 0, 1, or 2 based on defined ranges, and the resulting three-digit code maps to a specific fault type in the standard's interpretation table.

IEC 60599 categorizes incipient faults into six distinct types based on the gas ratio code combinations. Each fault type has a characteristic gas signature and energy profile.

This classification enables asset managers to prioritize maintenance based on fault severity.

Beyond absolute concentrations, IEC 60599 emphasizes that the rate of gas increase (ppm/day or ppm/month) is often more diagnostically significant than static values. A rapidly rising trend indicates an active, evolving fault.

Key considerations:

• A stable high concentration may represent historical fault residue
• A low but rapidly increasing concentration demands immediate attention
• The standard recommends establishing baseline trends through periodic sampling
• Critical rate thresholds vary by equipment type and operating conditions

This temporal dimension transforms DGA from a snapshot diagnostic into a continuous monitoring discipline.

IEC 60599 specifically addresses the interpretation of carbon monoxide (CO) and carbon dioxide (CO₂) as primary indicators of solid cellulose insulation degradation, distinct from oil-only faults.

Diagnostic indicators:

• CO₂/CO ratio < 3: Suggests severe paper overheating with charring
• CO₂/CO ratio 3-10: Indicates normal aging or moderate thermal stress
• CO₂/CO ratio > 10: May indicate oxidation without significant paper damage
• Elevated CO with normal hydrocarbon ratios points to insulation involvement

This distinction is critical because paper insulation damage is irreversible and directly impacts transformer remaining useful life.

The standard explicitly warns against mechanical application of ratios without considering operational context. Several factors can produce misleading gas patterns.

Known limitations:

• Multiple simultaneous faults can produce ambiguous ratio codes falling outside defined categories
• On-load tap changer (OLTC) compartments may leak gases into the main tank, mimicking main tank faults
• Stray gassing of certain oils can generate H₂ and CH₄ without any fault present
• Gas migration between compartments requires careful sampling point selection
• The standard applies specifically to mineral oil; natural ester fluids require modified interpretation

Expert judgment and trend analysis remain essential complements to ratio-based diagnostics.

The foundational diagnostic technique that IEC 60599 interprets. DGA measures the concentration of specific fault gases—hydrogen (H₂) , methane (CH₄) , acetylene (C₂H₂) , ethylene (C₂H₄) , and ethane (C₂H₆) —dissolved in mineral oil.

• Mechanism: Oil and cellulose decompose under thermal and electrical stress, generating distinct gas signatures.
• IEC 60599 Role: Provides the normal limit values and gas ratios (e.g., C₂H₂/C₂H₄, CH₄/H₂) to classify faults as partial discharge, thermal faults (T1, T2, T3), or electrical faults (D1, D2).
• Sampling: Requires airtight syringes and prompt lab analysis to avoid gas loss.

A permanently installed multi-gas sensor system that provides continuous, real-time dissolved gas readings for trending and immediate alarming.

• Technology: Uses gas chromatography, photoacoustic spectroscopy, or non-dispersive infrared sensors.
• IEC 60599 Integration: Monitors apply the standard's rate-of-change thresholds (e.g., ppm/day) to trigger alerts before absolute limits are exceeded.
• Key Benefit: Enables condition-based maintenance by detecting incipient faults weeks or months before catastrophic failure.

Methods applied to transformer fault models to provide asset managers with interpretable feature attributions justifying specific maintenance alerts.

• SHAP Values: Quantify how much each gas (e.g., acetylene) contributed to a specific fault classification.
• LIME: Generates local surrogate models to explain individual predictions in human-readable terms.
• IEC 60599 Alignment: XAI bridges the gap between black-box ML predictions and the standard's established physical heuristics, building trust with reliability engineers.

The application of deep learning models to predict future gas trajectories based on historical DGA trends, enabling proactive intervention.

• Models: LSTM networks and Temporal Fusion Transformers capture long-term dependencies in gas evolution.
• Inputs: Sequential DGA readings, load profiles, ambient temperature, and cooling system status.
• Output: Forecasted gas concentrations with confidence intervals, allowing operators to schedule maintenance before IEC 60599 alarm thresholds are breached.