Elmer Discussion Forum

Hello,
Attached is an ElmerGrid file that is similar to your geometry. If one reverses the order of the boundary condition input entries, then the order of parent1 and parent2 are also reversed.

Rich.

That is good, keeps from having to manually edit the mesh. Goo dto know that you can control the parent. I had to make one modification to make the simulation work. The middle boundary only uses 1 parent to calculate electrical force density. So two boundaries on the interior is necessary one pointing to each parent.

The grid file creates the six boundaries, and they do exist in the mesh files and simulation, but the do not appear in ElmerGUI.
ElmerGrid 1 2 layers3
ElmerSolver case.sif

Hi Kevin,

Attached is a third version of layers3.grd.

In the Boundary Definitions, changing the fourth entry called 'double of the boundary' from 1 to 2 tells ElmerGUI to include the extra boundaries. Now all six boundaries appear in ElmerGUI in the Tree, while the graphics window shows one of the two doubled boundaries.

Also, some element ratios were set to better distribute the elements near the boundaries.

Rich.
Edit: I also reordered the boundaries, with 1 at the bottom and 6 at the top, so the case.sif should be updated, or change the numbering back.

It looks good in ElmerGUI bI adjust the sif to match the boundary numbers, but the output in paraview came out wrong

I'm traveling for the next few days and don't have Paraview installed on this net book, so I can't check the results. ElmerVTK won't load the vtu either, just crashes. Do the results look good in Paraview, if the boundaries are not set to double?

Rich.

Yes they did look good without the double

Hi,

The "double" is an old way of creating jumps on boundary conditions. It requires that the solvers know what to do with the jump i.e. have some kind of boundary condition for that. The initial problem was poor thermal contact which is one of the implemented ones. The if you have other field in your multiphysics case there should be other models as well. So things become overly complex. The only test case for this is "HeatGap". The feature could perhaps be removed as there is a more elegant way to deal with the jumps which need not be present in the mesh.

-Peter