Transformer Temperature Rise Testing — IEC 60076-2 Methods, Limits, and What the Test Report Actually Tells You

The temperature rise test is the most expensive, most time-consuming, and most revealing test you can perform on a power transformer. It's also the one most likely to reveal a design deficiency that escaped the thermal calculation. Here's what happens inside the test bay — and what the numbers on your test certificate mean.

Why the Temperature Rise Test Exists

A transformer's nameplate kVA rating is meaningless without a corresponding temperature rise limit. A 1000 kVA transformer with 65 K winding rise and a 1000 kVA transformer with 55 K rise are physically different machines — the latter has more cooling surface, more oil, possibly lower loss density, and costs more. The temperature rise test is the only way to verify that the thermal design actually works.

Two things the test proves:

• At rated kVA, under rated cooling conditions, none of the thermal limits defined in IEC 60076-2 are exceeded.
• The calculated winding hot-spot temperature (which is not directly measurable but is inferred from the measured average temperature rise) stays within the insulation system's rated capability.

IEC 60076-2: Test Methods

IEC 60076-2:2011 defines three methods for oil-immersed transformers and references IEC 60076-11 for dry-type. The choice of method depends on available test power, transformer size, and the accuracy required.

Method 1: Direct Loading (Back-to-Back / Opposition Method)

Principle: Two identical transformers are connected in a back-to-back configuration. One transformer's primary is connected to the supply; its secondary is connected in parallel opposition to the second transformer's secondary. By adjusting the voltage ratio between the two units, controlled load current circulates through both windings while the supply provides only the total losses (iron loss + copper loss of both units).

Advantages:

• Most accurate — the transformer sees actual rated voltage and actual rated current simultaneously
• Realistic temperature distribution — core heating and winding heating occur together, matching service conditions
• Allows continuous measurement without interruption

Disadvantages:

• Requires two identical transformers (or one test unit + one auxiliary unit of similar rating and voltage)
• Requires sophisticated test setup with variable voltage supply and phase-shifting transformer
• Impractical for large power transformers (>50 MVA) because the auxiliary transformer and supply become prohibitively large

This is the gold-standard method. When the IEC specification says "the temperature rise test shall be performed," Method 1 is the preferred reference method.

Method 2: Short-Circuit Method (Equivalent Loading)

Principle: One winding is short-circuited. The supply voltage is raised until rated current flows. Under these conditions, the transformer experiences full load losses (I²R + eddy + stray) in the windings, but the core operates at a fraction of rated voltage — typically 5-10% of rated, producing negligible iron loss.

Because iron losses are not present, the total heat input to the oil is less than in service. To compensate, the test standard requires:

• Run at rated current until the top-oil temperature rise stabilizes.
• Immediately after reaching steady state (or after a specified time that the top-oil temperature rise has been stable), reduce the current to zero and measure the winding resistance to determine the average winding temperature.

Advantages:

• Requires only one transformer
• Much lower supply kVA requirement than Method 1 (only impedance kVA, approximately 5-12% of rated kVA)
• Practical for factory testing of transformers up to several hundred MVA

Disadvantages:

• Iron loss heating is absent — the oil temperature distribution may differ from service conditions
• The correction for missing iron loss is approximate
• The shutdown process for resistance measurement introduces timing uncertainty that must be corrected mathematically

Correction for Missing Iron Loss:

The measured top-oil temperature rise (Δθ_o_measured) must be corrected upward to account for the missing iron loss contribution:

Δθ_o_corrected = Δθ_o_measured × [(P_total / P_load_loss)^0.8]

Where P_total = P₀ + P_k and P_load_loss = P_k. The exponent 0.8 is an empirical value derived from convective heat transfer models.

For a transformer with P₀ = 1100 W and P_k = 10500 W: correction factor = (11600/10500)^0.8 = 1.08. If the measured oil rise is 50 K, the corrected rise is 54 K.

This correction is the largest source of uncertainty in Method 2. For high-loss designs (older units, high-impedance designs with large stray loss), the correction factor can be 1.10-1.15, and the uncertainty in the exponent translates to ±2-3 K uncertainty in the final result.

Method 3: Simulated Load Method (Separate Sources)

Principle: The core is excited at rated voltage and frequency from one supply (producing iron losses but negligible current in the short-circuited winding), and a separate supply injects current into the windings to produce load losses. The two supplies may operate at different voltages and even different frequencies, but must be decoupled to prevent circulating power between them.

Advantages:

• Most flexible — any combination of iron loss and copper loss can be simulated
• Can test a single transformer without requiring an identical pair
• Lower total supply kVA than Method 1 (typically 30-50% of rated kVA)

Disadvantages:

• The most complex setup — requires at least two independent power supplies with phase synchronization or frequency separation
• Risk of unintended interaction between the two excitation sources
• Less common in routine factory testing, more common in research and development

Temperature Rise Test for Dry-Type Transformers (IEC 60076-11)

Dry-type transformers follow a similar framework but with key differences:

• No oil thermal buffer — the time to reach thermal equilibrium is shorter (typically 4-12 hours vs 8-24 hours for oil-immersed).
• The "shutdown resistance measurement" is even more critical because the winding thermal time constant for cast-resin windings is 5-15 minutes — you have less time to get an accurate hot resistance reading.
• Partial discharge must be monitored during the temperature rise test for cast-resin transformers because thermal expansion of the copper relative to the epoxy can open micro-cracks that manifest as elevated PD at elevated temperature.

Measurement Points

Winding Temperature: The Resistance Method

The fundamental principle: the DC resistance of copper increases linearly with temperature.

R_hot / R_cold = (234.5 + θ_hot) / (234.5 + θ_cold)

Where 234.5 is the temperature coefficient constant for copper. (For aluminum, use 228.1.)

The procedure:

• Before the test, measure the cold winding resistance R_cold at known ambient temperature θ_cold (the transformer must have been at thermal equilibrium for at least 8 hours).
• Run the temperature rise test until thermal equilibrium.
• Shut down the supply and immediately begin measuring the winding resistance as a function of time (typically at 30-second intervals for the first 2 minutes, then every 1 minute for 5 minutes).
• Extrapolate the cooling curve back to the instant of shutdown (t = 0) to determine R_hot at the moment of de-energization.
• Calculate θ_hot from the resistance ratio, then subtract the ambient temperature to get the winding temperature rise.

The extrapolation is the critical step. The cooling curve immediately after shutdown follows an exponential decay with a time constant of 5-15 minutes (winding only). By fitting the measured points to an exponential function and extrapolating to t = 0, the test engineer determines the hot resistance. If the first measurement point is delayed (e.g., due to discharge time of the test circuit, manual switching), the extrapolation error increases. Test labs with fully automated switching and high-speed data acquisition can achieve the first measurement within 30-60 seconds of shutdown. Manual measurement may take 2-3 minutes — a significant source of uncertainty.

Top-Oil Temperature

Measured by an RTD (resistance temperature detector, typically PT100) in a thermometer pocket at the top of the tank, immersed in the hottest oil. For transformers with forced oil circulation, the measurement point is in the oil outlet pipe from the tank to the cooler.

The thermal equilibrium criterion per IEC 60076-2: the top-oil temperature rise shall not have changed by more than 1 K per hour over a 3-hour period.

Ambient Temperature

Three measurement methods, in order of preference:

• Immersed-thermometer method: The sensing element is in a container of oil with a thermal time constant similar to the transformer under test (approximately 2 hours). This smooths out short-term ambient fluctuations.
• Moving-average method: Ambient temperature sensors are read at regular intervals and the average over a period similar to the transformer thermal time constant is used.
• Direct reading: Ambient thermometers located around the transformer, with the average used. Least preferred because it doesn't account for the transformer's thermal inertia.

The ambient must be measured as the average of the air temperature at a height approximately half the transformer height, at a distance of 1-2 m from the tank, on at least three sides.

Temperature Rise Limits

IEC 60076-2 Limits for Oil-Immersed Transformers

Measurement	Limit (K)	Notes
Top-oil temperature rise	60 K	ONAN/ONAF cooling
Top-oil temperature rise	55 K	For transformers with OF or OD cooling
Average winding temperature rise	65 K	Measured by resistance method
Winding hot-spot rise (calculated)	78 K	Assumes H_g_r = 13 K for ONAN
Core surface, structural parts	80 K	Must not damage adjacent materials

These limits assume:

• Maximum ambient temperature: 40°C
• Average ambient over any 24-hour period: 30°C
• Average ambient over any 12-month period: 20°C

The 65 K winding rise limit plus the assumed 40°C maximum ambient gives an absolute maximum winding temperature of 105°C — the limit for Class A (cellulose) insulation.

If the user specifies a different ambient (e.g., 50°C for tropical installation), the temperature rise limits are reduced by the same amount: 65 - 10 = 55 K winding rise, 60 - 10 = 50 K oil rise.

IEC 60076-11 Limits for Dry-Type Transformers

Insulation Class	Average Winding Rise (K)	Maximum Hot-Spot (°C)
Class B (130°C)	80 K	130°C
Class F (155°C)	100 K	155°C
Class H (180°C)	125 K	180°C

Thermal Time Constants

The thermal time constant determines how long the test must run:

Transformer Type	Oil Time Constant	Winding Time Constant	Typical Test Duration
Distribution (≤ 2500 kVA), ONAN	1-3 hours	5-10 minutes	6-10 hours
Power (5-20 MVA), ONAN	2-4 hours	8-15 minutes	8-14 hours
Power (20-60 MVA), ONAF	2-5 hours	10-20 minutes	10-18 hours
Large power (>60 MVA), ODAF	2-4 hours	15-25 minutes	10-16 hours
Dry-type, cast-resin, AN	N/A	10-25 minutes	4-8 hours

A temperature rise test cannot be rushed. The test is complete only when the top-oil temperature rise has stabilized (change < 1 K/hour over 3 hours) AND at least one thermal time constant has elapsed since the last load change. If the test lab tries to declare equilibrium after 4 hours on a 20 MVA unit, question the result.

FAQ

Q: Can a temperature rise test be combined with other tests?

A: Not in the standard sequence, but the loss measurement is typically integrated. The cold resistance measurement (before the rise test) is also the basis for the load loss calculation. After the rise test, the hot resistance measurement provides the data for the load loss corrected to the reference temperature (75°C or 85°C). However, dielectric tests (applied voltage, impulse) must be performed separately and typically either before or after the temperature rise test, not during, because the thermal cycling can affect insulation condition and vice versa.

Q: What happens if the test shows a temperature rise higher than guaranteed?

A: If the measured temperature rise exceeds the guaranteed value by more than the measurement uncertainty (±2 K is typical for the resistance method), the transformer fails acceptance. The manufacturer must either: (1) identify and fix the thermal design issue (increase cooling surface, reduce loss density, improve oil flow), (2) de-rate the transformer to a lower kVA rating where the temperature rise satisfies the limit, or (3) renegotiate the contract with a different temperature rise guarantee. This is why factory test witness inspection is valuable — catching a thermal issue at the factory avoids the cost and schedule impact of a field failure.

Q: How accurate is the resistance measurement extrapolation method?

A: The dominant uncertainty sources are: (1) timing of the first measurement after shutdown — a 10-second delay translates to approximately 0.5-1.5 K uncertainty depending on the winding time constant, (2) accuracy of the cold resistance measurement — if the transformer wasn't at uniform ambient temperature for the full 8 hours before the cold measurement, the baseline is shifted, and (3) the extrapolation fit — exponential fitting to noisy data with a limited number of points. Combined standard uncertainty for a well-executed test (automated data acquisition, tight timings) is approximately ±1-2 K. For a manually executed test with first measurement 2 minutes post-shutdown, uncertainty increases to ±3-5 K. IEC 60076-2 provides the full uncertainty analysis framework.

Q: Do I need to witness the temperature rise test at the factory?

A: For transformers above 5 MVA, or any transformer that is process-critical, yes. The temperature rise test is the one test where the factory's process — how they time the shutdown, how they measure the cooling curve, how they extrapolate — directly determines the result. The routine loss tests and dielectric tests are largely automated and hard to manipulate. The temperature rise test requires judgement. Having your own engineer or a third-party inspector witness and record the shutdown sequence is cheap insurance against a pass result that doesn't reflect reality. Budget $2,000-5,000 for travel and witness time — well under 1% of the transformer cost.

Q: What's the relationship between temperature rise test and the thermal model used for overload analysis?

A: The temperature rise test determines the base parameters that feed into the IEC 60076-7 thermal model: top-oil rise at rated load (Δθ_or) and the hot-spot-to-top-oil gradient at rated load (H_g_r). If the rise test is performed at rated load (Method 1), Δθ_or is measured directly. If it's a short-circuit method (Method 2), Δθ_or is corrected. H_g_r is typically not measured directly — it's calculated from the manufacturer's winding thermal design using computational fluid dynamics (CFD) or empirical correlations validated by fiber-optic temperature sensor measurements during design verification tests. A transformer whose rise test was carefully executed provides reliable parameters for the overload model. A sloppy test produces input data that makes the overload prediction unreliable.

Q: Can a temperature rise test damage the transformer?

A: Not if executed per the standard. The test runs at rated current or close to it, and the temperatures are within the design limits. If the manufacturer made a thermal design error, the temperatures may exceed limits during the test, but the test should be terminated before damage occurs. However, a temperature rise test IS a thermal stress. The insulation ages at an accelerated rate during the test — approximately 10-20 hours at the rated hot-spot temperature. This is equivalent to 10-20 hours of normal service life and is negligible in the context of the transformer's 30+ year design life. The only scenario where a rise test could be harmful is a repeated rise test on the same transformer — which accumulates unnecessary thermal aging and is not recommended.

Q: My transformer nameplate says "ONAN/ONAF with 65°C rise." Does the temperature rise test check both cooling modes?

A: Per IEC 60076-2, if the manufacturer guarantees temperature rise limits in both ONAN and ONAF cooling modes, both must be tested — or only ONAF must be tested if the design is such that ONAN automatically satisfies the limit when ONAF does (which is typically the case, since ONAN cooling is less efficient). The common practice for a 10/12.5 MVA ONAN/ONAF transformer: test at 12.5 MVA with fans running. The ONAN rating (10 MVA) is verified by calculation from the ONAN cooling surface characteristics, not by a separate 10 MVA test. This is acceptable under IEC 60076-2 if stated in the contract. If you want both modes tested, specify it and expect to pay for the additional test time.

References

• IEC 60076-1:2011 — Power transformers — Part 1: General
• IEC 60076-2:2011 — Power transformers — Part 2: Temperature rise for liquid-immersed transformers
• IEC 60076-3:2018 — Power transformers — Part 3: Insulation levels, dielectric tests and external clearances in air
• IEC 60076-7:2018 — Power transformers — Part 7: Loading guide for mineral-oil-immersed power transformers
• IEC 60076-11:2018 — Power transformers — Part 11: Dry-type transformers
• IEEE C57.12.90-2015 — Standard Test Code for Liquid-Immersed Distribution, Power, and Regulating Transformers
• IEEE C57.12.91 — Standard Test Code for Dry-Type Distribution and Power Transformers

*Written from the factory floor. Test what you design, measure what you test. The temperature rise test is where thermal theory meets copper and oil — and sometimes loses.*