Transformer oil test results: DGA, moisture and dielectric strength tables

What do transformer oil test results actually tell you?

Transformer oil testing combines dissolved gas analysis (DGA) for developing internal faults, water measurement for moisture, and dielectric breakdown testing for the sampled liquid's resistance to electrical failure. Start routine in-service DGA annually, the interval suggested by ALS's Transformer Fluid Testing flyer, revision TRI.2022.V4, and shorten the interval when confirmed gas generation accelerates. Read moisture with the oil temperature and fluid type, and read every breakdown voltage with its test method and electrode gap. Combine those results with sample quality, load and alarm history.

The main system here is an in-service mineral-oil transformer with cellulose insulation, the scope of IEC 60599:2022, edition 4. Natural ester, synthetic ester and silicone require their own interpretation. Identify the fluid and retrofill history using the equipment record and the transformer types and insulating fluids guide.

Identify the compartment before comparing gases. A separate load tap changer generates switching gases that have a different meaning from newly increasing main-tank acetylene; oil communication between compartments also matters, as Reclamation's April 2005 guide explains. Keep separate histories for each compartment.

Which tests belong in the routine package?

Order DGA, Karl Fischer water and a named breakdown method together, then add condition tests that answer the asset's aging and contamination questions. ASTM D3612-02(2026) measures dissolved gases by gas chromatography; ASTM D1533-20 measures water by coulometric Karl Fischer titration. Acidity, interfacial tension and liquid dissipation factor describe other aspects of oil condition.

Gas ppm commonly means microliters per liter, or volume concentration; water ppm means mg/kg, or mass concentration. Record those units explicitly. Water ppm and gas ppm measure different quantities.

A strong liquid result does not clear an equipment fault. ASTM D1816-12(2019), section 3.1 states that high breakdown voltage does not establish freedom from contaminants. Keep a confirmed DGA trend active even when breakdown passes its applicable criterion.

Oil dissipation factor measures the liquid; winding insulation power factor measures the equipment insulation system. Keep those records alongside the transformer acceptance baseline and insulation-resistance testing. See insulation resistance for that measurement.

Sampling-quality checklist: how do you avoid false alarms?

A usable trend starts with a representative sample. ASTM D923-15(2023) addresses insulating-liquid sampling, including special techniques for gas, water and particles. Agree the container, total package volume and approved sampling point with the laboratory before the visit.

OBRE instruction 7.3 P 1-3, revision A dated January 3, 2022 specifies about 2 to 4 L of flushing, three syringe rinses and a fourth 50 mL collection, filled by tank pressure rather than pulling the piston. These quantities belong to OBRE's setup; other tests need separately agreed sample volumes.

Reclamation's April 2005 guide, section 7.7.1, prohibits sampling with negative pressure at the sample valve because opening it can draw air into the transformer. Follow the equipment's pressure, level, access and withdrawal procedure; a tank gauge alone does not establish pressure at the port. Stop when the approved sampling conditions cannot be established.

For work covered by OSHA 29 CFR 1910.333(a)(1), exposed live parts must be deenergized before work on or near them unless the employer establishes a permitted exception. Paragraph (c)(2) limits energized electrical work to qualified persons. Establish the electrical isolation procedure and work boundary from the installation, including its arc-flash study and labels; an accessible sampling valve does not authorize opening energized compartments.

Anvilfield's closeout checks supplement the cited sampling guidance. Capture flushed oil and account for total withdrawal, especially on small sealed units.

Recollect when collection introduces air or wet components contaminate the specimen. If a bubble appears after hot oil cools during shipment, preserve the sample and tell the laboratory; venting it changes the gas inventory.

A water spike from a container with wet threads is suspect, but the transformer can also have a moisture problem. Correct the collection defect and obtain a fresh sample before assigning the cause. Protection operation, abnormal pressure or rapidly changing gases with operating symptoms require the operator's response while confirmation proceeds.

A repeat test on the same bottle checks analytical repeatability; a fresh field sample also checks collection quality. Reclamation section 7.7.1 describes stagnant valve oil as a possible source of misleading hydrogen. When hydrogen rises alone, compare the flushing record and preservation-system history before assigning an internal fault mechanism.

Which gases matter in DGA trend interpretation?

Interpret the gas pattern, its change and its compartment together. Reclamation's April 2005 Transformers: Basics, Maintenance, and Diagnostics describes associations between gases and fault activity; each gas needs context.

Total dissolved combustible gas, TDCG, conventionally sums hydrogen, methane, ethane, ethylene, acetylene and carbon monoxide in ppm by volume. Reclamation's 2005 section 6.1.2 and table 8 provide that legacy reporting framework; carbon dioxide, oxygen and nitrogen are outside the sum. Preserve individual gases beside the total because a stable sum can conceal increasing acetylene.

Reclamation's four-condition chart is attributed to IEEE C57.104-1991. For a current program, identify the adopted interpretation edition: the IEEE publisher project page lists C57.104-2019 and a revision project, and IEEE supplies an official 2019 errata sheet. Use the adopted published edition for alarms.

How do you calculate a gas trend without misreading it?

Subtract the earlier concentration from the later concentration, then divide by elapsed days. The result is a net concentration change per day. Gas loss, oil additions, degassing and changed oil inventory separate that quantity from total volumetric gas production.

Gas concentration rate = (later ppm - earlier ppm) / elapsed days.

Label calendar days and loaded days separately. Reclamation's 2005 table 9 note 3 uses days loaded and excludes down days. A calendar-day rate and a loaded-day rate require matching denominators before comparison.

Method: Anvilfield estimates divide Reclamation's published 86 ppm ethylene increase by declared inputs of 7, 30 and 90 calendar days, rounding the outputs to two decimals. Use asset timestamps and alarm criteria.

Compare small changes with reporting resolution, detection limits and laboratory repeatability. The public case's acetylene rises from 4 to 7 ppm, an absolute increase of 3 ppm and a calculated 75% rise. Retain absolute changes alongside percentages and analytical confidence.

Retain negative changes and nondetect qualifiers. A negative delta needs investigation for variation, gas loss, treatment or sample mismatch. Do not normalize negative changes into a Duval mixture or silently replace them with zero. For a nondetect, retain the laboratory qualifier: a ratio with an unquantified denominator is not a measured ratio.

Establish abnormal activity before applying ratios or a Duval classification. Reclamation's 2005 low-level ratio discussion advises particular caution below ten times the detection limit; retain that as a dated confidence screen. A Duval triangle returns a fault category even for an insignificant gas mixture. Select the diagram for the fluid and compartment; retain the gas concentrations and analytical confidence beside the classification.

Use the same laboratory and method where practical, retaining extraction procedure and reporting limits across changes. ASTM D3612-02(2026) covers three procedures and liquids at or below 20 cSt at 40°C. Mark oil processing as a break in the history and establish a post-treatment baseline.

What does moisture mean at the oil's actual temperature?

Relative saturation is the water present divided by the water the liquid can dissolve at that temperature, expressed as a percentage. Water activity equals relative saturation divided by 100, according to Vaisala MMT310 user guide M210474EN-D, September 2015. Cooling reduces solubility, so the same water concentration occupies more of the available capacity.

Use the named model to calculate saturation water content at the oil temperature, then divide measured water by that value and multiply by 100. Vaisala MMT310 M210474EN-D, pages 77 to 78, gives the average mineral-oil model and coefficients; expressed as saturation water content, it uses Ws = 10^[A/(T + 273.16) + B], with A = -1662.7 and B = 7.37; RS = 100 × W/Ws, with T in °C and W and Ws in mg/kg.

Method: Anvilfield estimates hold water at 30 mg/kg and apply Vaisala's average mineral-oil coefficients at each temperature, rounding only the results. Use oil-specific coefficients when available, particularly for aged oil; use a fluid-specific model for FR3 or silicone.

Record temperature at the moisture measurement location. Vaisala's OPT100 guide M211858EN-J, section 2.3.2 distinguishes monitor-chamber RS from tank RS because measurement temperatures differ. Record the location with each comparison.

Vaisala's 2015 guide identifies above 90% RS, or water activity above 0.9, as a free-water-risk alarm point. Cargill G2300's June 2017 comparison describes breakdown performance beginning to decline around 40% RS. These describe different effects: waiting for a free-water alarm can miss an earlier equipment maintenance limit.

Cooling at fixed mg/kg raises RS without adding water. Operating transformers also exchange moisture between oil and cellulose, with a temperature-dependent time lag. Reclamation's section 6.1.11 explains why one oil bottle cannot establish a measured paper-moisture percentage. Drying the liquid can be followed by moisture returning from paper, so distinguish oil processing from a core-and-coil dry-out and verify the subsequent trend.

Record both the sampling temperature and the temperature used by the laboratory or monitor to calculate RS. Room-temperature RS describes the sample at room temperature. Compare transformer conditions using a consistent location and fluid model, and investigate an unexpected rise in water mass concentration separately from the saturation change caused by cooling.

How do ASTM D1816, D877 and IEC 60156 compare?

Breakdown kV belongs to a test geometry, gap and procedure. ASTM D1816-12(2019) uses VDE electrodes and is sensitive to moisture, especially with fibers; ASTM D877/D877M-19e1 uses disk electrodes and states lower contamination sensitivity. A conversion factor cannot make different methods interchangeable.

ASTM D877/D877M-19e1 says water sensitivity falls below 60% saturation and contamination sensitivity declines above approximately 25 kV. Its section 1.2 excludes new insulating liquids upon receipt in favor of D1816; its note 1 recommends migrating in-service testing too. Both ASTM public scopes require their method criteria, including room temperature of 20 to 30°C, for comparable results.

The setup figures above come from Megger's OTS_tg_en_V02 guide, copyright 2013. Specify the adopted procedure and edition on the order; the older setup comparison does not establish the full 2025 IEC procedure. Record the actual program edition, individual breakdowns, mean, electrode gap and temperatures, then apply that edition's repeatability rules. IEC 60156:2025, edition 4 covers liquid breakdown testing up to nominal viscosity of 350 mm²/s at 40°C; its current full procedure governs an IEC test.

A 30 kV D1816 result can exceed a 1 mm benchmark and fail the corresponding 2 mm benchmark. Use the manufacturer comparison below with the equipment's adopted limits before assigning pass or fail.

How do the fluid and manufacturer instructions change the limits?

Fluid identity changes moisture interpretation and analytical preparation. Cargill FR3 guide G2300, June 2017 compares about 40% RS at room temperature with roughly 25 mg/kg in mineral oil and 400 mg/kg in FR3, making a mineral-oil ppm alarm unsuitable for FR3.

Cargill recommends a 30-minute stand after pouring room-temperature FR3 before breakdown testing, allowing entrapped air to escape. Identify that product preparation in the laboratory procedure. The transformer's operation and maintenance manual governs equipment work.

Prolec GE's archived limits retain their printed D1816-97 and IEEE C57.106-2002 basis. Cargill G2300 table 6 calls its investigation values provisional and gives the trend greater weight. Preserve voltage boundaries: its above-69-through-230 kV class includes 230 kV, whereas Prolec GE places 230 kV in the highest mineral-oil class. Apply the equipment's governing limits.

IEC 63585:2026, edition 1 addresses DGA interpretation in natural and synthetic esters. Specify fluid and retrofill history so extraction, calibration and interpretation suit the liquid. Cargill G2300 also documents a chromatographic peak near acetylene that can be misidentified in FR3. A low acetylene result conflicting with the wider evidence calls for laboratory peak-identification review alongside any independent operating response.

Put the applicable acceptance document on the report, including fluid, voltage class, service-aged or new-oil status, method and gap. Investigating a provisional trigger is different from declaring an automatic trip. When a contract and an OEM manual specify different criteria, resolve that difference with the equipment owner and OEM before assigning pass or fail.

How often should you sample or move to online monitoring?

Start stable routine programs with annual DGA under ALS TRI.2022.V4, which separately suggests general and physical testing twice yearly. Reclamation's April 2005 program combines annual oil-condition tests with DGA; Cargill G2300 gives a 12-month fluid-testing fallback when the customer's procedure provides no interval. Select and record the applicable program rather than averaging their schedules.

Shorten the interval when a valid change develops and name who receives the next report. Set the next sampling date, escalation trigger and operating decision together; the annual calendar must not carry forward unchanged after an abnormal result.

Online monitoring supplies evidence between laboratory samples. Specify measured gases, moisture location, alarms, confirmation sampling and the person receiving alerts. A single-gas monitor lacks the pattern supplied by a multi-gas analysis; choose its role from the trend question and the asset's consequence of failure.

Compare monitor readings with laboratory samples, reconciling units, collection times and measured gases. Continuous observations can capture activity missed by periodic bottles; include instrument health and data gaps in the alert review.

Worked example: old gas and new gas give different mixtures

Reclamation's April 2005 guide publishes two sample sets for an unidentified transformer in section 6.1.9.2, printed pages 129 to 130. The worksheet compares the accumulated concentrations with their net changes between collections. Subtract each earlier value, total the three triangle-gas increases, then divide each increase by that total.

Method: Anvilfield estimates calculate differences and proportions from Reclamation's published inputs. The denominator is the combined methane, ethylene and acetylene increase: 50 + 86 + 3 = 139 ppm. Carbon monoxide and carbon dioxide remain in the equipment assessment but outside this three-gas denominator.

In the Bureau of Reclamation's April 2005 transformer case, the ethylene increase is 86 of 139 ppm of the three-gas increase, so its share is 61.87%, correcting the printed 46% on page 130.

The net increase is richer in ethylene than the accumulated mixture, supporting investigation of thermal activity. Obtain the missing dates, load and treatment history, validate the small acetylene change with the laboratory, and assign an operating review. A gas-derived thermal category describes a fault mechanism; it is not a measured winding temperature.

Follow-up action matrix: resample, treat oil or address a fault?

Choose the action for the confirmed problem. Recollection restores valid evidence, physical oil treatment removes contaminants, and equipment repair addresses the fault source. Processing the liquid does not repair a hot internal connection.

IEC 60422:2024, edition 5 addresses mineral-oil supervision and maintenance, including reconditioning and reclamation; equipment diagnosis by DGA and furans lies outside its scope. Reconditioning uses physical treatment to remove particles, water or gas, while reclamation additionally addresses oil degradation products. Select treatment from the full condition evidence and fluid instructions.

The operator applies the asset's operating response and arranges any required outage. The laboratory assesses the submitted samples; the testing contractor provides representative collection and complete test records. Involve the OEM for an equipment-specific limit, design question or repair method.

Correlate thermal gases with load and cooling operation. Electrical thermography adds accessible connection, radiator and surface patterns, while DGA addresses gases within the liquid system. Coordinate an outage investigation with medium-voltage maintenance when connected gear belongs in the scope.

How long does testing take, and what belongs in the scope?

Emerald Transformer publishes typical laboratory analysis of 5 to 7 days and 24-hour rush service. Agree receipt-to-report timing, test scope and transport separately. For an abnormal result, name the operator who receives the telephone notification and the laboratory contact responsible for interpretation.

Quote the same scope across laboratories: compartments, fluid type, DGA extraction procedure, water method, breakdown method and gap, condition tests, interpretation and rush arrangements. Separate field access, collection, kits, shipping and any outage from laboratory work. A basic oil-quality package can omit DGA, while a DGA syringe alone does not supply the volume needed for the complete package.

Build a multi-site scope from the operator's inventory, grouping access, compartments and shipping by route. Route planners use PorchRocket Asset Search to locate visible outdoor power sites and their owners of record across the lower 48 when grouping transformer sampling work by territory. Use substations as a site class and keep distribution-transformer replacement planning connected to the confirmed equipment inventory.

Include previous reports, applicable limits and a replacement-sample plan with the order. Schedule the field visit around approved access and shipping availability as well as laboratory turnaround.

What record makes the next oil test useful?

Keep one history per asset and compartment, with fluid identity, retrofills, oil additions and treatment breaks. Record the sampling point, date and time, oil temperature, ambient conditions, load, operating state, alarms and receipt condition. Retain complete numerical reports with detection limits, extraction procedure, method edition, breakdown gap and individual test readings.

Attach limits and the previous report to make the decision reproducible. Link the asset to its electrical one-line and preserve the acceptance baseline and first post-treatment sample. Finish with an action owner, due date and completion evidence through maintenance work-order closeout. A report closes when its required response has been verified.

Sources checked and how to cite this guide

Anvilfield. Reviewed November 5, 2026. Public standard abstracts establish scope and edition; numerical comparisons retain their named manufacturer and date.

Update log: 2026-11-05: Final technical and editorial review completed; sampling safety, standard editions, moisture calculations and manufacturer limits checked; FR3 voltage classes completed; DGA follow-up and test scope clarified.

• S1: Bureau of Reclamation, Transformers: Basics, Maintenance, and Diagnostics, April 2005
• S2: ASTM D1816-12(2019), scope and significance
• S3: ASTM D3612-02(2026), scope and significance
• S4: ASTM D1533-20, scope
• S5: ASTM D877/D877M-19e1, scope and significance
• S6: ASTM D923-15(2023), scope and significance
• S7: IEC 60599:2022, edition 4, mineral-oil DGA scope
• S8: IEC 60156:2025, edition 4, breakdown-testing scope
• S9: IEEE C57.104 publisher revision-project page, listing C57.104-2019
• S10: ALS Transformer Fluid Testing, TRI.2022.V4
• S11: OBRE DGA sampling instruction, 7.3 P 1-3, revision A, January 3, 2022
• S12: Vaisala MMT310 user guide M210474EN-D, September 2015
• S13: Megger insulating-oil breakdown guide, OTS_tg_en_V02, 2013
• S14: Cargill FR3 operations and maintenance guide G2300, June 2017
• S15: IEEE C57.104-2019 errata, December 5, 2019
• S16: IEC 60422:2024, edition 5, mineral-oil maintenance scope
• S19: Vaisala OPT100 guide M211858EN-J, section 2.3.2, monitor and tank saturation
• S20: Prolec GE Services, undated archived brochure, page 7
• S21: Emerald Transformer Lab Services, provider turnaround offer
• S22: IEC 63585:2026, edition 1, ester DGA scope
• S23: OSHA 29 CFR 1910.333(a)(1) and (c)(2), electrical work practices

How to cite this page: Anvilfield, Transformer oil test results: DGA, moisture and dielectric strength tables, 2026, https://anvilfield.com/field-guides/electrical/transformer-oil-testing-dga-moisture/

Related tools

FAQ

People also ask

• What is the difference between a dry-type and liquid-filled transformer?
• What is a TTR test?
• What does an infrared inspection find?
• What is insulation resistance testing used for?
• Why does medium-voltage switchgear need maintenance if it runs fine?