Motor Thermal Management
Theory: losses and temperature feed each other
Overview
What sets a motor's continuous output?
Usually winding temperature. Winding insulation has a temperature limit (180 °C for class H, for example), beyond which life shortens sharply. Most winding heat is copper loss, proportional to current squared and resistance. But copper's resistance rises with temperature, so the hotter it gets the more it heats, heating further. Without this effect, temperature is badly underestimated at high currents. Briefly, winding heat capacity slows the rise, allowing currents above the continuous rating; how long is also found by coupling current and temperature.
Steady winding temperature
R₂₀ is phase resistance at 20 °C (20 mΩ), α copper's temperature coefficient (0.00393 /K), R_th winding-to-coolant thermal resistance (0.025 K/W), T_c coolant temperature (65 °C) and I rms phase current. Where the denominator reaches zero, no equilibrium temperature exists.
Short-term overload
C is winding heat capacity (5 kg copper × 385 J/kgK = 1,925 J/K); the time constant is 48 s.
Insulation classes and life
Winding insulation for motors and transformers is classed by withstand temperature (class A 105 °C, B 130 °C, F 155 °C, H 180 °C and so on). Insulation life shortens sharply with temperature; a long-used rule of thumb halves it for every 10 °C rise. EV motors must deliver high power from small packages, so they often run close to insulation limits, and the accuracy of cooling and temperature prediction drives performance.
Worked examples
Example 1: current and steady winding temperature
| Phase current (rms) | Coupled temperature | Constant-R estimate | Copper loss |
|---|---|---|---|
| 100 A | 83.8 ℃ | 80.0 ℃ | 750 W |
| 150 A | 110.8 ℃ | 98.8 ℃ | 1,832 W |
| 200 A | 157.4 ℃ | 125.0 ℃ | 3,696 W |
| 250 A | 239.7 ℃ | 158.8 ℃ | 6,988 W |
Continuous current within the class H 180 °C limit is 217 A; above 412 A no equilibrium temperature exists (thermal runaway).
Example 2: short-term overload
| Condition | Value |
|---|---|
| Steady at 150 A | 110.8 ℃ |
| Time to 180 °C after stepping to 400 A | 12.0 s |
| Winding time constant | 48 s |
At 250 A it's 80 K above the constant-resistance estimate.
The larger the current, the stronger the effect of temperature raising resistance, which raises losses further. At 250 A resistance is about 1.86 times its 20 °C value, and loss grows by the same factor. A constant-resistance estimate of 159 °C looks within limits, but the real 240 °C would destroy the insulation. Motor analysis therefore feeds winding resistance as a function of temperature into the electromagnetic calculation and iterates with the thermal one to equilibrium. Within the winding, the deepest slot regions and end turns, where heat escapes poorly, are hottest, so evaluate maximum rather than average temperature. Rising magnet temperature risks demagnetisation, so rotor temperature is checked too.
Analysis workflow
- Compute copper, iron and magnet losses per operating point with electromagnetic analysis.
- Build a thermal network or 3D thermal model of windings, core, magnets and cooling.
- Make resistance and magnet properties temperature-dependent and iterate electromagnetics and heat.
- Compute temperature histories over drive cycles (acceleration, cruise, climbing).
- Calibrate thermal resistances against winding thermocouples and rotor temperature measurements.
“There was margin in the test”
In one motor design, a constant-resistance thermal analysis gave 140 °C windings at continuous rating, 40 K below the limit. Yet prototype tests reached 175 °C at the winding's hottest point under the same conditions. Coupling temperature-dependent resistance raised the average about 20 K, and slot bottoms and end turns ran about 15 K above average. Revising coolant channels and improving end-turn cooling cut the peak at continuous rating to 155 °C.
Common mistakes
Mistakes and fixes
| Mistake | Effect | Fix |
|---|---|---|
| Constant resistance | Temperature underestimated | Couple with temperature |
| Evaluating average temperature | Local hot spots missed | Use maximum temperature |
| Steady-state only | Peak output underrated | Compute transients |
| Ignoring magnet temperature | Demagnetisation | Analyse the rotor |
| Omitting harmonic losses | Losses underestimated | Use inverter waveforms |
I'd like to learn related topics.
Related pages include Joule heating, eddy-current heating, induction heating, transformer hot-spot analysis and eddy-current brake heating.
Related Topics
Experience the theory firsthand with the interactive simulator for this field
All Simulators