Identification on bodies and properties of areas where solids overlap

[asdf]

Hi,
I'm trying to simulate capacitance of copper plane on PCB. I've already got some results which looks ok but I'm not sure I do it right.
Below is description of what I did.
1. First step was to export gerber file to svg, then do some cleanup in inkscape (including saving as png and converting back to svg), then the svg file could be imported to FreeCAD.
2. In FreeCAD I created 3D structure which consist of:
- the piece of copper imported as above and extruded
- a second piece of copper - simplified ground plane below the described above (I want to know capacitance between these two)
- substrate - layer of dielectric between these two
- simulation box - I want the whole structure (2 copper parts separated by dielectric) to be simulated in some volume of air although so I placed a cube including the parts
3. I prepared geometry in Salome. First, I used function "Explode" to have all the parts separated. Then it was necessary to use "Shape Processing" on the more complicated copper plane with adjusted parameters to avoid further errors. Next I created "Partition" to have all the parts together to create mesh. Also I used "Create Group" to create boundary conditions on two inner surfaces of copper parts.
4. I created mesh in Salome from the partition using gmsh. With "Create Groups from Geometry" I added the boundary conditions.
5. When I opened unv file in ElmerGUI i noticed it's displayed incorrectly. I assigned materials (copper Er=1, Air from library, dielectric Er=4.5), boundary conditions (1V and 0V) and solver (electrostatics). Simulation seems to work ok.

My questions are:
1. As the bodies are not displayed correctly and names from Salome are not imported it's not clear how to assign materials. I opened the mesh in Salome to figure out what is what by looking at the numbers (assumed that Solid_1 is Body Property 1) but I'm not sure it's correct.
2. I'm not sure if I defined the air around my structure correctly. I've got the three parts (2 copper parts separated by dielectric) and the box which includes the structure. The problem is that the box is a solid which is overlapping the other parts and I'm not sure how to make sure the properties of the box have lower priority than properties of the structure. The dielectric has higher dielectric constant (4.5) and I'm not sure if/how elmer can determine if the area between copper parts has Er like air or like the dielectric. A solution maybe would be to add cavity inside the cube representing air around (by using boolean subtraction) but meshing tool failed to mesh that (unless the geometry is simple).

[asdf]

attachments

[Rich_B]

Hello,

In ElmerGUI, if you click on 'View' and check the 'Volume Mesh', do you see the missing bodies? If yes, then it means that not all of the surfaces were created properly.

Also in ElmerGUI, under 'Model', 'Setup', look for the box 'Desired Mesh Pieces' and enter a one (1). If the mesh is non-conforming, then the test will fail, halting the simulation.

Have you been following the Salome to unv file instructions in GetStartedElmer.pdf?
https://www.nic.funet.fi/pub/sci/physics/elmer/doc/

Rich.

[Rich_B]

Also, it always helps if you can post your files, geometry and sif, and maybe a solver log.

Rich.

[asdf]

I can't see the dielectric between copper parts.
Regarding the instruction in GetStartedElmer, I didn't use "Face groups separated by sharp edges" in salome and I've got only boundary conditions on surface of copper parts. I've just tried making the Face groups but there were hundreds of them and ElmerGUI seems to be overloaded by them. Still, the simulation results seem not bad (close to expectations) but maybe just coincidence. Setting Desired Mesh Pieces to 1 doesn't cause issues.

The files:
https://drive.google.com/drive/folders/ ... sp=sharing

[kevinarden]

Several issues on this model.

1. ElmerGUi displays bodies by displaying the boundary surfaces attached to it, however each body can have two boundary surfaces attached. ElmerGUI only uses the parent surface. For more complicated cases viewing "bodies" in GUI can become confusing if you don't understand the boundary surfaces and bodies included. Group in Salome helps, but what you see in GUI may not be what you get. This is a graphical limitation of ElmerGUI that does not impact the solver.

2. The meshes are not conformed to each other. In Elmer each body has to share nodes with other bodies, or other mortar conditions are required to enforce conformance at the boundaries. You can not just create multiple part in space with non-confirming nodes. Elmer communicates data between bodies at shared nodes, or through defined boundary conditions, that define the interaction of the bodies.

3. There seems to be several instances of both 1 an 2 above in this model.

[asdf]

1. If I understand correctly, the solver knows how the bodies look like but the hard part is to assign correct materials to the bodies as it's not clear in GUI which is what.
2. How could it happen and how to determine if the problem exist? The mesh was created in salome with gmsh from one "partition". When I open the mesh in salome it looks to me that the meshes at boundaries fit ok, e.g. top surface of dielectric part looks the same as bottom surface of the piece of copper on top, similarly the bottom copper mesh looks like bottom part of dielectric part.

[kevinarden]

In salome you can use groups to create bodies and boundaries. If 3D groups of volume meshes are bodies, and groups of face elements are boundaries.
Once the groups are created then you can export the unv file.
If you make the elmer mesh with
ElmerGrid 8 2 filename.unv -autoclean
Then the elmer mesh will be written with a mesh.names file tat names the bodies and boundaries and what number they are.
You can refer to them in the sif file by name or number.

[asdf]

It works! Now I can easily identify the parts and boundaries by checking numbers assigned to names which I can see in salome. My first problem (issues with displaying) is solved.