Discussion Forum

Firstly, I would like to say hi to the comsol community, this being my very first post here.

My problem concerns DG and using it in 2D axial and 3D simulations.
I've made a model of a static permanent magnets system, using AC/DC Module Magnetic Fields, No currents Physics.

The image of the geometry is attached to this post.

I'm calculating electromagnetic force that acts on the inside pair of magnetic rings. The magnets are magnetized axially.

At the beginning the forces balance out. In 2D-axial, I can calculate the z component Fz that appears when I move the inner magnets in reference to the outer ones along the z axis (dZ displacement).
However, to calculate the radial component (x or y) Fr that appears while moving the magnets in the radial direction (x or y) I need a full 3D model.

I've created the 2D-axial model, according to the "Axial Magnetic Bearing Using Permanent Magnets" tutorial. As in this guide, I wanted to use a parametric sweep with a geometric parameter. The parameter is taken into account in my geometry as well as in the DG: Prescribed Mesh Displacement. I think that everything works as it should and via parametric sweep I get the same singular results as for the same parameter value without using DG and the sweep.

However, I've made a 3D model with the very same geometry and physics settings. When I use the radial displacement dR both in model geometry and DG I get different results using parametric sweep (2 times greater) than in the single step simulation for a certain value of dR without using DG. In my particular case, the function Fr(dR) is close to linear, so it looks like the distance is taken into account twice and therefore I get 2 times greater results.

On the other hand the magnetic flux density graph in the crossection plane zx (y=0) looks as if the dR is only accounted for once.

I tried refining the mesh, but the effect persists.

Could you please tell me what can be the case here? Do you have any idea why I do get different results? Should I take the displacement dR into account both in the geometry and DG in 3D? And if not, why is it different than in 2D axial?

I would really appreciate any help, because I need the simulation for my thesis.