Learning Forum

[Carlos]

In a transient 2D simulation involving permanent magnets and ferromagnetic cores in Ansys Maxwell, the common recommendation is to assign 0 mA of current to the magnets and enable Eddy Current Losses in the Excitations section. Additionally, Core Losses should be activated when the core material includes loss data, allowing proper post-processing of results.

However, I noticed that the loss curves obtained in 2D—with this configuration—are significantly different from those obtained in a 3D transient simulation, even when following the same setup logic. In 3D simulations, it is not necessary to apply 0 current to the magnets, and doing so doesn't affect the outcome in the same way.

❓ Why is this 0-current setup critical for 2D transient simulations but not for 3D ones?
Why the Eddy Current loss curve is completly different between 2D and 3D analysis? 

Also worth noting: in my case, Eddy Current Losses were only enabled for the magnets, not for the core material, which might lead to confusion when interpreting results—although core losses are indeed accounted for in the simulation.

Any clarification on this discrepancy between 2D and 3D handling of losses in permanent magnets would be highly appreciated.

[Ivonne Marti]

Hi Carlos,

In Maxwell 2D, it is needed to apply zero current in the magnets because it forces the eddy currents to return only through the magnet, rather than through the surrounding background region. This is important for obtaining reliable results in 2D simulations, where the geometry and magnetic field distribution are simplified compared to 3D models.

Regarding the difference in losses. Please check that the model depth in 2D corresponsd to the 3D model one. Also notice that if the magnets are segmented on the axial direction, then you will either need to reduce the loss by a factor, or reduce the material conductivity to account for the 3D segmentation effects.

[Carlos]

Hi Ivonne,

Thanks for your contribution!

Yes, I considered that as well. In fact, the model length, material conductivity, and magnet coercivity seem to have no influence on the curve pattern — and that's precisely the issue I'm trying to resolve by sharing this with experts here.

I re-ran the simulations using different magnet and core materials to check whether the curve pattern would change, but it didn’t.

Both the 2D and 3D models have exactly the same geometry and length. The 3D model was simulated in full (no segmentation) to avoid that variable.

As a result, I observed that the loss curve pattern from the 3D simulation matches the 2D simulation only when the 0 mA current is not applied to the magnets. That’s where the inconsistency lies.