Transformer Winding Resistance: Correct Temperature Before Comparing Phases and Factory Data

Normalize winding resistance to a stated temperature, define the phase-spread calculation, and distinguish thermal or connection effects from genuine contact de

1. A percentage difference needs a common basis

Do not compare a warm field winding directly with a cool factory winding and call the difference a contact defect. Resistance rises with conductor temperature, while the measured terminal path may include leads, tap contacts and parallel winding branches. Establish the same tap and connection path, stable measurement and justified temperature before comparing phases or history. Megger explains that a reference-temperature correction allows measurements made under different field temperatures to be compared. Field measurement challenges.

This guide concerns settled DC winding resistance, not the transient signature of an operating tap change. It applies to diagnostic comparison; it does not replace the specified load-loss temperature correction or a complete factory acceptance procedure.

2. Define the measured path and thermal state

Record vector group, accessible terminals, neutral availability, tap position, conductor material and whether the measurement is phase-to-neutral or line-to-line. A delta terminal measurement contains parallel paths; it is not automatically the resistance of one physical phase winding. Preserve the arrangement used by the factory and ask the manufacturer how to interpret a non-equivalent connection.

Use four-terminal measurement with separate current and potential connections under the approved offline procedure. Record test current, stability history, elapsed time, instrument uncertainty and discharge completion. At non-steady current, inductive voltage remains in the measured result. Megger's application note discusses stabilization, delta paths, test heating and residual magnetization. Measurement principles.

3. Apply an identified correction convention

Use Rref = Rm × (k + Tref) / (k + Tm), with temperatures in °C, Rm the measured resistance and Rref the reference-temperature value. Megger's transformer case study uses k = 234.5 for copper and k = 225 for aluminium. Use the convention required by the governing test procedure and keep it consistent with the reference report; other procedures may use a rounded constant. Temperature correction equation.

The difficult input is often winding temperature, not arithmetic. Ambient or top-oil temperature immediately after shutdown may not represent the conductor. Record how thermal equilibrium was established or how temperature was determined. Do not hide a thermal assumption behind a software correction checkbox. Retain the raw values, temperature and correction convention so another engineer can reproduce the result.

4. Illustrative correction and uncertainty

For copper, assume 0.100 Ω at 20°C and a reference of 75°C. The calculation is 0.100 × (234.5 + 75) / (234.5 + 20) = 0.12161 Ω. The roughly 21.61% increase is the expected modelled temperature effect, not evidence of deterioration.

For a later reading of 0.106 Ω at 35°C, correcting back to 20°C gives 0.106 × (234.5 + 20) / (234.5 + 35) = 0.10010 Ω. Compared with 0.100 Ω, the residual difference is about 0.10%, rather than the uncorrected 6%. These values are illustrative, not field measurements or acceptance limits.

A 5°C temperature-input error around 20°C introduces approximately 5 / 254.5 × 100 = 1.96% correction sensitivity. This first-order estimate shows why an uncertain thermal state can dominate a small resistance difference. It does not include instrument uncertainty or real temperature gradients.

5. Name the phase-comparison metric

Suppose comparable corrected readings are 0.1216 Ω, 0.1220 Ω and 0.1240 Ω. Their mean is 0.12253 Ω. A range-to-mean metric is (0.1240 − 0.1216) / 0.12253 × 100 = 1.96%. The maximum individual deviation from the mean is a different metric, approximately 1.20%. Never report only phase imbalance without defining its denominator.

Comparison	Required common basis	What to investigate
Same phase against factory	Tap, path, material and reference temperature	Persistent change after thermal correction
Phase against phase	Equivalent physical paths and thermal conditions	Repeatable asymmetry, including known construction differences
Resistance against tap position	Stable values across relevant taps	Local contact anomaly or unexpected curve change

No percentage in this example is a universal pass limit. Obtain manufacturer expectations and project criteria, especially for very low resistance where resolution matters.

6. Purchase evidence, not a green status icon

Require raw and corrected values for each relevant tap, terminal diagrams, temperature method, material constant, current stability and instrument uncertainty. Include historic comparison and explicit metric definitions. The report should distinguish repeatable anomalies from unresolved measurement conditions and assign follow-up ownership. Controlled discharge and the required demagnetization closeout must be documented before returning the transformer to service.

Use DRM transition diagnostics when static readings are normal but switching is suspect; use the SFRA transport baseline guide for mechanical fingerprint questions. These complementary methods do not make one resistance number conclusive.

7. Frequently asked questions

Can top-oil temperature always stand for winding temperature?

No. Justify the thermal state and the approved method, particularly after recent loading.

Should all phases have exactly identical resistance?

Not necessarily. Connection paths and construction matter; compare with manufacturer data and equivalent historical readings.

Does a corrected normal result rule out OLTC problems?

No. Settled resistance does not capture every transition defect; a suitable dynamic test may be needed.