2D solving with a 3D mesh

Numerical methods and mathematical models of Elmer
spacedout
Posts: 177
Joined: 30 Mar 2020, 23:27
Antispam: Yes

Re: 2D solving with a 3D mesh

Post by spacedout »

Sometimes one needs a break. The Cuba sun indeed gave me fresh new ideas. Here is a minimalistic test case: perform StatElecSolver for a single 2D domain extruded in the 3rd dimension with only one cell. And no coupling to OpenFOAM. This is a pure Elmer problem. Now let

Charge Density = 1.0e30

I know this astronomical value is nonsensical but I am just debugging. The result is that dF/dz is not only non-zero but astronomical as well.
On the other hand with

Charge Density = 0.0

dF/dz remains at 0

I think this goes to show that the linear solver Ax = b for StatElecSolver becomes numerically unstable when one gives it astronomical input values.

So I believe my problem started with the computation of plasma ion densities by OpenFOAM. These became astronomical and since the charge density that is sent to Elmer depends on them, it also became ridiculously large.

I suppose that if I am not able to discover what is wrong with my OpenFOAM code, I can always go back to my Elmer code for solving these ion densities although I recall I also had stability issues with their computation.

I am still grateful for all previous replies to this post.
Best to all
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: 2D solving with a 3D mesh

Post by raback »

Hi

Elmer does use scaling by default making the linear solvers rather robust. But of course this has it limits and I guess with 1e30 you have surpassed them. Maybe use some other unit system. Should be ok as long as you're consistent. With unit charge and speed of light set to 1 maybe things look better...

-Peter
Post Reply