[NF] Symetry for Bore notches in FEMM

[BonneelP]

Hello all,

This PR is a follow-up of the issue #24. I have updated the build_geometry of lamination to handle the symmetry on the notches by using the split_line method and here is the result (there are small notches on the rotor too):

The next step was to be able to have such topology in FEMM but I had again some boundary condition issues:

• The airgap/sliding band side lines starts at Rbo (and not bellow)
• The shaft and Frame external radius have a particular BC and I don't know if it is compatible with something else than an Arc.

So for now the yoke notches are not available for symmetry (by waiting further investigation) and for bore notches I have added dedicated surfaces to "close the void in the bore":


An advantage of using this method is that I was able to define bigger elements in the notches than in the Airgap. In the future, I think that we will add such surfaces to "close the slot" again to reduce the number of elements.

To carry-on the work on issue #24 the next question are yoke notches and uneven Bore for LamHole.

Best regards,
Pierre

[SebGue]

That looks great!
What is the actual issue with symmetry and notches in FEMM?

[SebGue]

... for the "close the slot" we could introduce slot wedges (that could also be soft magnetic).
What happens if notches/holes overlap?

[BonneelP]

In the current version, we assumed that the notches weren't colliding with a sym line and if it was the case, the notches was fully drawn (instead of just half of it) which creates an error.
build_geometry now cuts correclty the notches as shown in the plot but for FEMM I had an issue to set the following line BC:

This is why I have added the surfaces (to have a line to set the BC on)

[BonneelP]

... for the "close the slot" we could introduce slot wedges (that could also be soft magnetic).

Yes, for slot with wedges (like SlotW10) we can expand the build_geometry_active to return the wedge surface and the air surface to set the correct materials.

What happens if notches/holes overlap?

An error when drawing the machine in FEMM ^^' The topology checks are not that great (for now) but a machine with such "issue" would not be real. Such case could appear in a parameter sweep in extreme cases for instance but then the error should be handled as an impossible design.

[SebGue]

Ah okay, so I guess I have to look how it's done in code.
BTW. In my former FEMM simulations (pre sliding band) I had a triple airgap devision. So the mesh was coarser in the slot opening and in the pole gaps and fine in the middle for e.g. torque ripple calculations.

[BonneelP]

BTW. In my former FEMM simulations (pre sliding band) I had a triple airgap devision. So the mesh was coarser in the slot opening and in the pole gaps and fine in the middle for e.g. torque ripple calculations.

Yes I think this is something that we can investigate. In particular, I can be interesting for this machine and maybe for LamHole with uneven bore shape.

[BonneelP]

Hello,

There was an hidden issue in Tests/Validation/Magnetics/test_FEMM_periodicity.py, the test was passing but the flux was 0 (bc_ag2 BC was removed from the sliding band due to an arc definition error).
In fact the issue was linked to SlotCirc.build_geometry that returned an Arc1 with angle > 180. I fixed it but we may have the issue again if we encounter again such Arc in FEMM.
The time to figure out what was the issue, I tried several solution including one that we can take inspiration for further work: f538c2e. For is_simplified, instead of removing the lines, I set a property is_draw: False on the line. This remove the need to define open surfaces by removing some lines (which is important to compute surface, point ref...) and can be a first step toward having a build_geometry without colliding lines (which is important for GMSH for instance).

Best regards,
Pierre