Do not apply a blanket temperature correction to the real parts of fitted zero-sequence T-model constants Z1, Z2, and Z3. Correct only the loss components that can be traced to conductor resistance, then recalculate or refit the T-model from consistently corrected test quantities. If a branch combines copper loss with tank, core, connection, or return-path loss, scaling its complete R value as winding copper will corrupt the model.
What do the wrong-temperature symptoms reveal?
Look at the trend first. Compare the original test sheet, the converted model, and the study output on the same impedance base and at the same stated temperature. A uniform resistance shift may indicate an intentional conductor-temperature conversion; unequal or physically implausible shifts usually point to a conversion applied after the T-model was fitted.
| Signal | Source to inspect | Wrong-value symptom |
|---|---|---|
| Zero-sequence impedance magnitude | Measured zero-sequence test point and test connection | Recalculated T-model no longer reproduces the measured terminal voltage-to-current relationship |
| Zero-sequence real power loss | Test loss measurement at the recorded current and temperature | Calculated input watts disagree even though impedance magnitude appears reasonable |
Real part of Z1, Z2, or Z3
|
T-model fitting calculation | Branch R values change uniformly despite representing different physical loss paths |
| Reactive part of each constant | Original fitted model and impedance base | X changes during an R-only correction or becomes inconsistent with the original complex impedance |
| Neutral or return-path response | Test connection, grounding arrangement, and zero-sequence current path | Model matches one test condition but fails another connection or terminal condition |
The terminal mismatch identifies where to work. If measured loss is wrong after conversion, inspect the resistive decomposition. If only current distribution inside the model changes, inspect the fitting equations, base conversion, and branch placement before changing temperature.
Why is a blanket R correction invalid?
A T-model constant is a mathematical branch impedance, not automatically a measured winding DC resistance. Its real part can contain several contributions collapsed into one fitted value: winding conductor loss, stray loss caused by leakage flux, conductive structural loss, and loss in the zero-sequence return path. These components do not necessarily share one temperature law.
Section 9.5.3 of IEEE Std C57.12.90-2006 describes Z1, Z2, and Z3 as constants in the zero-sequence equivalent circuit. “Constant” identifies their role in that fitted circuit; it does not, by itself, prove that every real component is temperature invariant or that the winding-resistance correction equation applies to the entire component. The governing test calculation must define how the constants were obtained.
The signal chain matters. The test measures terminal voltage, current, real power, connection state, and temperature. The fitting calculation turns those measurements into complex branch constants. A system study then applies those constants to a network model. Changing a fitted branch after that calculation alters what the final network sees without preserving the original terminal measurements.
When is direct temperature correction defensible?
Directly correct a branch R only when the model documentation identifies that R as conductor resistance, gives its conductor material or approved correction constant, and states the reference and target temperatures. The common conductor form is:
R_target = R_test × (T_target + K) / (T_test + K)Use K from the governing test method or transformer documentation for the identified conductor material. Do not infer the material from a typical transformer design. Record whether each temperature is winding temperature, corrected average winding temperature, ambient temperature, or another test-defined value; these are not interchangeable.
If the supplied R is derived from measured watts rather than an isolated conductor measurement, separate the conductor-dependent loss from the other real-loss terms first. Where that decomposition is unavailable, preserve the provided constants at their stated test condition and request either corrected zero-sequence data or the fitting worksheet.
How should the T-model be corrected?
- Establish the data basis. Record the zero-sequence test connection, energized terminals, grounded or open terminals, frequency, impedance base, test current, measured loss, test temperature, and intended reference temperature. Missing connection information blocks a reliable reconstruction because zero-sequence current depends on its return path.
-
Identify measured and fitted quantities. Mark terminal voltage, current, and watts as measurements. Mark
Z1,Z2, andZ3as derived constants unless the report explicitly defines a branch as a physical resistance or leakage reactance. - Decompose real loss. Use the test report calculation to separate winding conductor loss from stray and structural loss. Do not assign every measured watt to copper simply because it appears in the real part of an impedance.
- Correct the conductor term. Apply the approved temperature equation only to the identified conductor-resistance contribution. Keep temperature-independent terms unchanged and handle any other temperature-dependent terms according to their documented model.
-
Reconstruct terminal loss. On the test connection’s defined basis, calculate resistance from loss as
R_eq = P_loss / (m × I²), wheremis the number of identical current paths represented by the recorded total power. Obtainmfrom the connection and calculation sheet rather than assuming a phase count. -
Refit the complex model. Recalculate
Z1,Z2, andZ3using the same equations, constraints, terminal cases, and base used for the original fit. Preserve reactance unless the fitting process couples R and X and therefore requires a complete refit. - Label the result. Store the test temperature, target temperature, correction method, loss decomposition, impedance base, and model revision with the constants.
How do you verify the corrected constants?
Back-calculate the terminal quantities for every test condition used in the fit. The corrected circuit must reproduce the intended zero-sequence impedance magnitude, real power, and phase relationship on the selected reference-temperature basis. Check complex values, not resistance alone.
Next, compare the original and corrected models. The change in real input power should equal the change introduced by the corrected loss components; an unexplained change in reactive power flags an altered X term, a base error, or a branch-fitting error. Also verify that all ohmic and per-unit conversions use the same voltage, power, and winding reference.
Finally, run the network case that consumes the data and inspect zero-sequence current and terminal voltage. A numerically converged study is not proof of a valid correction. The model must remain traceable to the test configuration and corrected loss calculation.
Which pitfalls recur with zero-sequence models?
The most common error is treating the real part of every complex branch as winding DC resistance. Another is correcting already corrected data a second time because the report does not clearly distinguish test-temperature and reference-temperature columns. Preserve the unmodified report values and make the conversion lineage explicit.
Connection errors can dominate temperature errors. Zero-sequence flux and current require a return path, so grounding, winding connection, open terminals, and conductive structural paths affect the measured result. Tuning does not fix wiring: confirm the physical and modeled connection before adjusting constants.
Other recurring problems include mixing line and phase quantities, applying one base to R and another to X, reconstructing impedance from rounded values, and forcing negative or implausible branch resistance to zero. A nonphysical fit calls for review of equations, signs, constraints, and measurement precision—not an undocumented clamp.
FAQ
How do I temperature-correct zero-sequence T-model resistance?
Separate the conductor-resistance contribution, correct that term with the approved conductor equation, and refit Z1, Z2, and Z3. Do not scale each complete branch R unless its physical meaning and temperature dependence are documented.
How do I know whether Z1, Z2, and Z3 contain copper loss?
Inspect the test calculation or fitting worksheet that maps measured volts, amps, watts, and connection states into the three constants. A complex constant’s real part alone does not identify which physical loss produced it.
How do I verify a corrected transformer zero-sequence model?
Back-calculate terminal impedance and watts for every fitted test condition, then confirm that only the intentionally corrected loss terms changed. Check the connection and impedance base before accepting the study result.
When do I stop correcting the T-model and escalate?
Stop when the test temperature, conductor material, zero-sequence connection, loss decomposition, impedance base, or fitting equations cannot be identified, or when the refitted model cannot reproduce the test quantities. Send the original report, model constants, conversion worksheet, and failed back-calculation to the transformer manufacturer’s official support channel and request corrected zero-sequence data or the approved fitting method.