Learning Forum

[kjh0703]

Hi,

 

I'm obtaining a motor's dq inductance matrix.

Well, overall matrix trend seems good. But I need a fine data.

Is there any way to reduce numerical noise? For example, changing solver.

I tried to set smaller mesh at airgap surface between stator and rotor as in the piture below but it didn't improve significantly.

[NA]

Dear @kjh0703,

There are a couple of things that could influence the error on the magnetic field especially in the airgap and hence the inductances of the windings. First comment would be to refine the mesh inside the airgap and not just on surfaces. The second hint would be to always include an inner band and outer band when simulating a motion setup. Third point, but very important for low noise results is to match the motion speed with the band mesh segments and timestep. Basically, you achieve minimum noise when for each timestep you have a rotation of an integer number of segments of shift in the band.

I hope this helps!