VectorHemholtz Solver applicable for 2D

[bplinux]

Hi there,

thank your very much for elmer and the time you developers spend in the current webinar series. They are awesome, thanks!

My question here is if the vectorhelmholtz solver is applicable for 2d. I actually tried to get a simple 2d task but my mission failed due to zero-division error.

I successfully adapt the emwave solver for my problem in the transient space, but i like to see it in steady state / harmonic style.

I am actually just starting with FEM and do not have that experience. I think my BCs are correct, according to the models tutorial.

Is it possible for the solver to work in 2d space? Is it possible for somebody to provide a simple .sif?

Thank you very much in advance
Bastian

Edit:
This was the result in emwave solver

2.5GHz TEM hits a 2mm copper plane

summary.png (215.12 KiB) Viewed 870 times

[raback]

Hi,

No. While all nodal finite element solver in Elmer work similarly in 2D and 3D, the solvers that utilize edge elements are a totally different ballgame. This is also the reason why going 2D->3D in electromagnetics is often a much bigger step than for the other PDEs.

For the AV solver (A for vector potential living an edge elements, V for scalar potential living on nodes) the equations reduce to simple Poisson equation for the A_z component in 2D. Therefore we have written a totally seperate solver for 2D for that. I guess similarly could be done for electromagnetic wave equations. However, this has not been done.

Why are edge elements needed? Nodal elements would require too much continuity over element interfaces spoiling the solution when material changes. For edge elements the degree of freedom is the amplitude on the element edge. Normal component need not be continuous over element interfaces.

-Peter

[bplinux]

Thank you Peter, i think this is perfectly answered