Electromagnetic Workbench - again..

About the development of the FEM module/workbench.

Moderator: bernd

ediloren
Posts: 209
Joined: Wed May 08, 2013 9:23 pm
Location: Italy
Contact:

Re: Electromagnetic Workbench - again..

Postby ediloren » Fri Nov 16, 2018 1:41 pm

We plan to be using the materials sooner or later, but for the time being it is an overshoot probably.

FasterCap format is the same of FastCap, but this is pretty specific to those solvers I'm afraid. Also, it is based on pure surface panels geometrical descriptions (akin to what the Mesh workbench does). As I said, these solvers are based on BEM methods, not FEM, so only boundary descriptions are necessary, not volumes - this dramatically reduces the number N of input elements, at the cost of a dense matrix to be inverted, that is tackled with by some fast O(N) or O(NlogN) algorithm.

Best Regards,
Enrico
User avatar
bernd
Posts: 8244
Joined: Sun Sep 08, 2013 8:07 pm
Location: Zürich, Switzerland

Re: Electromagnetic Workbench - again..

Postby bernd » Fri Nov 16, 2018 1:58 pm

99 % of FEM I do are based on face meshes too.
ediloren
Posts: 209
Joined: Wed May 08, 2013 9:23 pm
Location: Italy
Contact:

Re: Electromagnetic Workbench - again..

Postby ediloren » Mon Nov 19, 2018 3:33 pm

Hi Bernd,

I'm not sure we are talking about the same thing here - I have not deeply tested the FEM workbench in itself, but at least from the tutorials and demos I see that the basic FEM_MeshNetgenFromShape creates a volume mesh, as I expect. There could be other simulations where you just use surfaces then, with no volumes at all - this would be perfect for integrating later on some other solver interface. However FastHenry (the inductance solver we are interfacing with the current E.M. workbench development) is based on segments, i.e. non-regular hexahedrons. Still, you do NOT mesh them into tetrahedrons.

I prepared a short video showing the current capabilities of the beta E.M. workbench. You may also note I borrowed some icons.. not using them the same way, but at least they may pave the way for a possible integration :)

Here's the link:

FreeCAD's ElectroMagnetic workbench for FastHenry beta tutorial

Cheers,
Enrico
User avatar
bernd
Posts: 8244
Joined: Sun Sep 08, 2013 8:07 pm
Location: Zürich, Switzerland

Re: Electromagnetic Workbench - again..

Postby bernd » Mon Nov 19, 2018 4:22 pm

ediloren wrote:
Mon Nov 19, 2018 3:33 pm
but at least from the tutorials and demos I see that the basic FEM_MeshNetgenFromShape creates a volume mesh, as I expect.
If you pass a volume it creates a volume mesh, if you pass a face or a shell it creates a faces mesh. Give it a try ... You do not even need an analysis ...

shell-mesh.FCStd
(47.46 KiB) Downloaded 5 times

screen.jpg
screen.jpg (241.92 KiB) Viewed 712 times
ediloren
Posts: 209
Joined: Wed May 08, 2013 9:23 pm
Location: Italy
Contact:

Re: Electromagnetic Workbench - again..

Postby ediloren » Tue Nov 20, 2018 7:16 pm

Hi Bernd - surely this is good!

But what's different from using the Mesh workbench for meshing a shell? (I usually do that)

Enrico
User avatar
bernd
Posts: 8244
Joined: Sun Sep 08, 2013 8:07 pm
Location: Zürich, Switzerland

Re: Electromagnetic Workbench - again..

Postby bernd » Tue Nov 20, 2018 9:58 pm

Mesh workbench only meshes three node triangle. In mesh workbench the netgen mesher is available too. In background it is the same call. But you have your mesh in mesh workbench not in FEM mesh object, means for FreeCAD FEM you would need to convert the mesh to a FEM mesh first.

BUT Netgen FEM mesher is able to mesh 6 node triangle too. Furthermore GMSH mesh in FreeCAD is able to mesh quad meshes too. Furthermore FEM mesher meshes boolean fragments means nonmanifolds. They are meshed to be connected. Means the edges are not meshed twice. There might be even more advanteges. Ahh Gmsh mesh object supports mesh groups and mesh regions for mesh with different size. All these tools are great already but could be improved too.

On the contrary mesh wb has some nice tools to find errors in a three node triangle mesh, or to close holes, ...
ediloren
Posts: 209
Joined: Wed May 08, 2013 9:23 pm
Location: Italy
Contact:

Re: Electromagnetic Workbench - again..

Postby ediloren » Thu Nov 22, 2018 2:31 pm

I see your point; but the Mesh workbench already offered Netgen as an option as mesher (as well as Mephisto) but the exposed options are limited, and also reflecting the fact that the mesh workbench is designed with triangular meshes in mind. So I understand you enlarged the access to more complex mesh options (I have not checked the C/C+ code, only the Python part for the time being). In my case I am mainly interested in triangular meshes actually - and quadrilateral only in very special cases; this is not a limit (as triangles are a sub-set of the capabilities you provide), still I would like to be able to access the Mesh workbench tools for mesh checks.

This is anyway for the next steps. FastHenry uses segments, as explained (i.e. solid parallelepipeds); we have then triangular surface patches for capacitance extraction (FastCap, FasterCap), for Full-Wave extraction (FastImp) and voxels for the newest inductance extractor (VoxHenry; still being worked on, at present available as MatLab/Octave code only).

Cheers,
Enrico
ediloren
Posts: 209
Joined: Wed May 08, 2013 9:23 pm
Location: Italy
Contact:

Re: Electromagnetic Workbench - again..

Postby ediloren » Wed Dec 12, 2018 7:35 pm

Hi Everybody,

for those interested, we published a new version of the EM Workbench, and there are two new tutorial videos explaining how to use the latest enhancements, centered around:

1) define a conductive plane.

2) parametrize the position of the nodes

Enjoy, and any feedback is welcome as usual!
Enrico
ediloren
Posts: 209
Joined: Wed May 08, 2013 9:23 pm
Location: Italy
Contact:

Re: Electromagnetic Workbench - again..

Postby ediloren » Sun Dec 30, 2018 10:21 pm

Hi All,

new commit for the EM Workbench, and new videos:

1) Using the new path object

2) Using the new path object, part 2: changing segment cross-section orientation

3) Conductive plane part 3 - add and remove nodes/holes

I will soon start writing the relevant documentation in the Wiki pages for the add-on, as well as preparing for adding the workbench to the external workbenches.

and happy New Year!
Enrico