Why Your Motor Simulation Is Underestimating Eddy Current Losses by 50%

The Problem: A Simulation That Looked Fine — Until It Wasn't

I was working on an efficiency optimization project for a PMSM, targeting an improvement from 94% to 95.5% at rated load.

My electromagnetic simulation showed a clean loss breakdown:

• Copper losses: 800W

• Iron losses: 400W

Copper losses dominated, so I focused there — redesigned the windings, got copper losses down to 650W. Solid work, on paper.

Then we built the prototype and tested it. Measured efficiency: 94.3%. Nowhere near the target.

When we disassembled the motor, the permanent magnets were running 30°C hotter than the simulation predicted. That was the tell.

After digging through the literature, the answer was obvious in hindsight: I had set the permanent magnets as non-conductive in the simulation. Magnet eddy current losses were never calculated at all. In actual operation, those losses came in at around 150W — a significant chunk of the total loss budget.

Where Do Magnet Eddy Current Losses Come From?

NdFeB magnets aren't insulators. They have real, measurable electrical conductivity — around 0.6–0.7 × 10⁶ S/m. That's enough to support meaningful eddy currents whenever the magnets are exposed to time-varying magnetic fields.

There are three main sources of flux pulsation that drive these losses:

1. Slot harmonics (cogging effect)
Stator slots create uneven air gap permeance. As the rotor turns, the magnets see a pulsating flux rather than a smooth field.

2. PWM harmonics
Inverter-fed drives inject high-frequency current harmonics that generate high-frequency flux components — exactly the kind that induce strong eddy currents in the magnets.

3. Armature reaction harmonics
Stator current doesn't produce a perfectly sinusoidal field. The distortion adds harmonic content that the magnets pick up.

All of these drive eddy currents through the magnet material, converting electrical energy into heat — heat that has nowhere easy to go since magnets typically have no direct cooling path.

Getting It Right in Simulation

Setting 1: Assign the correct magnet conductivity

In Ansys Maxwell (and most FEA tools), the default conductivity for magnet material may be set to zero. This is wrong for NdFeB.

Use these values:

Setting 2: Model magnet segmentation correctly

A solid, unsegmented magnet block gives eddy currents the maximum possible loop area — which means maximum losses. But in real motors, magnets are typically bonded in segments with insulating gaps between them. Those gaps break up the current paths and dramatically reduce losses.

In simulation, replicate this:

• Divide magnets circumferentially into individual segments

• Keep each segment ≤ 5–10° arc

• Set inter-segment conductivity to near zero

One project example: unsegmented magnet losses were 200W. After modeling 8 segments, losses dropped to 40W — an 80% reduction.

Setting 3: Run transient analysis, not magnetostatic

Magnet eddy currents are inherently a time-domain phenomenon. Steady-state or magnetostatic solvers cannot capture them.

You need a full transient electromagnetic simulation. Set your time step fine enough to resolve slot-passing frequency:

Δt ≤ 1 / (20 × Z × n/60)

Where Z = number of stator slots, n = rotor speed in RPM.

Engineering Approaches to Reduce Magnet Eddy Current Losses

Once you've got an accurate loss picture, here's how to actually bring the number down:

Magnet segmentation — Splitting into 8–12 segments typically cuts eddy current losses by 60–80%. This is usually the highest-impact, lowest-cost intervention.

Skewing — Either skew the stator slots by one slot pitch, or skew the rotor poles by one pole pitch. This smears out the spatial harmonics and reduces pulsation amplitude at the magnet surface.

Pole arc ratio optimization — The ratio of magnet arc to pole pitch directly affects harmonic content in the air gap flux. Tuning this can meaningfully reduce eddy current excitation.

SPM vs IPM topology — Surface-mounted magnets (SPM) sit directly in the high-harmonic air gap field, making them more susceptible to eddy current losses than interior permanent magnet (IPM) designs, where the rotor iron provides some shielding.

Other Details People Miss

Retaining sleeve losses — Carbon fiber sleeves are non-conductive, but titanium alloy sleeves are not. At high speeds (>10,000 RPM), sleeve eddy current losses can be significant and are often ignored entirely.

Temperature-dependent conductivity — NdFeB conductivity increases with temperature. Higher magnet temperature → higher eddy current losses → even higher temperature. This feedback loop can accelerate thermal risk if you're not accounting for it.

Magnet thermal path — Magnets have no direct cooling. Heat generated by eddy currents must conduct through the rotor iron and across the air gap. If the thermal resistance is high, magnet temperature can reach demagnetization threshold even at moderate loss levels. Always pair electromagnetic loss analysis with a thermal model.

A Reliable Simulation Workflow

Here's the process I now follow on every PMSM loss analysis:

1. Set correct NdFeB conductivity in material properties

2. Run transient simulation with appropriate time step

3. Model actual magnet segmentation geometry

4. Extract and validate eddy current losses against any available test data

5. Run coupled thermal analysis to assess magnet temperature rise

6. Optimize segmentation count for the best loss/cost/manufacturability tradeoff

The Takeaway

In high-speed or high-power PMSMs, magnet eddy current losses can account for 10–20% of total losses. Miss them in simulation, and your entire efficiency optimization is built on bad numbers.

The fix isn't complicated — it's a material property setting and a solver choice — but it has to be deliberately made. Default simulation setups won't catch it for you.

Set the conductivity. Run transient. Model your segments. The gap between simulation and reality closes fast once you do.

Technical content provided by Shanghai Shangdian Motor Technology Co., Ltd.