6.4 Oil Dielectric (ASTM) & DGA Sampling

Oil Dielectric Breakdown Tests — ASTM D877 and D1816

A dielectric breakdown voltage test measures the kV at which a sample of insulating oil breaks down between two electrodes. It is the most common field screening test for oil quality — low breakdown voltage indicates water, particles, or other contamination that compromises the oil's insulating ability.

ASTM D877 — flat disk electrodes

• Electrodes: flat brass disks, 25.4 mm diameter, with square edges.
• Gap: 2.54 mm (0.1 in), fixed.
• Voltage rise: 3 kV/s until breakdown.
• Characteristics: less sensitive to moisture and particles because the flat disks do not concentrate stress at contaminants. Primarily used for new oil acceptance.
• Typical minimum: 30 kV for new oil; 26 kV for in-service per NETA MTS.

ASTM D1816 — VDE mushroom electrodes

• Electrodes: VDE mushroom-shaped (spherical cap), 36 mm diameter.
• Gap: 1 mm or 2 mm, selectable.
• Stirring: mandatory; oil is stirred during the test so particles are driven through the high-stress region.
• Characteristics: significantly more sensitive to moisture, particles, and dissolved gases than D877. Preferred for in-service oil evaluation.
• The 2 mm gap values are higher than the 1 mm gap values because a longer gap requires more voltage to break down the same oil.

NETA MTS Table 100.4.1 — minimum dielectric breakdown (in-service)

D877 was removed from IEEE C57.106 in-service evaluation criteria in 2002 but is still referenced in NETA MTS with a flat 26 kV minimum across all voltage classes. For in-service oil, prefer D1816.

Dissolved Gas Analysis (DGA) Sampling Procedure

DGA is the single most diagnostic test for an oil-filled transformer. It detects incipient faults — overheating, arcing, partial discharge — by quantifying the gases dissolved in the oil before the transformer fails. The value of DGA depends entirely on sampling technique; a poor sample produces false conclusions and unnecessary (or missed) maintenance.

Per ASTM D3613 and standard utility practice:

1. Use a clean, dry, sealed syringe (typically 30–50 mL) — never an open bottle for DGA. The syringe preserves dissolved gases at their in-service concentration.
2. Flush the sampling port — drain several liters of oil through the valve first to remove stagnant oil and any water or debris in the port. Discard the flush oil.
3. Fill without trapping air — purge all air from the syringe and sampling line before filling; a single air bubble contaminates the sample with atmospheric nitrogen and oxygen and dilutes the fault gases.
4. Sample from the correct tap — usually the lower-main-tank drain valve or a dedicated sampling valve, not a radiator or top fill port. The lower tank sample reflects the bulk oil the transformer is bathed in.
5. Label accurately — transformer ID, date, time, sample point, oil temperature, and weather conditions. Shipping same-day does not exempt you from labeling.
6. Ship promptly in a light-proof, shock-protected container.

Mixing oil from multiple transformers in one bottle, sampling from an open hatch, or skipping the flush are all disqualifying errors.

The Seven Key Fault Gases

Each fault gas is produced by a specific decomposition process. The gas signature points to the fault type.

Temperature-driven formation

• ~150°C: H2 and CH4 begin forming.
• ~250°C: C2H6 starts forming.
• ~350°C: C2H4 starts forming.
• 500–700°C+: C2H2 forms (peaks near 800°C) — significant C2H2 almost always means arcing.
• CO and CO2 indicate the paper insulation is involved, not just the oil.

A key rule of thumb: significant acetylene (C2H2) in a DGA is a serious alarm — it indicates arcing inside the tank and usually requires taking the transformer out of service for internal inspection.

Duval Triangle Basics

The Duval Triangle (Triangle 1), developed by Michel Duval in 1974, is the most widely used graphical DGA interpretation method. It plots the relative percentages of three gases — CH4, C2H4, and C2H2 — on a ternary diagram:

• %CH4 = 100 × CH4 / (CH4 + C2H4 + C2H2)
• %C2H4 = 100 × C2H4 / (CH4 + C2H4 + C2H2)
• %C2H2 = 100 × C2H2 / (CH4 + C2H4 + C2H2)

The triangle is divided into fault zones:

Duval Triangle 1 uses CH4/C2H4/C2H2; Triangle 4 uses H2/CH4/C2H6 for PD and low-thermal refinement; Triangle 5 uses CH4/C2H4/C2H6 for T2/T3 refinement. The method always produces a diagnosis (no 'unidentified' bucket), which is its strength and its weakness — boundary cases between zones still require engineering judgment.

The IEEE Std C57.104-2019 guide and IEC 60599 provide the normative interpretation framework; Duval is a complementary graphical tool.

Why Sampling Technique Matters

A DGA result is meaningless if the sample is bad. The most common errors and their effects:

• Trapped air bubble — inflates N2 and O2, dilutes fault gases, can mask a real fault or trigger a false alarm on hydrogen.
• Unflushed port — stagnant oil in the valve may have different gas concentrations than the bulk tank.
• Wrong sample point — a radiator tap can show cooler, less-mixed oil that misses a hot-spot signature.
• Open container — dissolved gases escape into the atmosphere before the lab can measure them.

The lab can detect some sampling errors (abnormally high O2, low gas totals), but it cannot correct for them. Repeat sampling is the only remedy when a sampling error is suspected.