Elmer Discussion Forum

Dear Forum User,

I'm struggling a bit with a basic electrostatic simulation(zip attached).
The geometry and mesh was set up using salome 9.6.0

So, it's a simple T shape 3D structure... as BC, the base of the T is grounded, one side is at 2V and the other one is at 4V. There are also 2 different dielectric constants from the 2 T parts shaping the T (top bar and vertical bar).
Setting up the mesh in Salome, I took care to join all the solids when meshing in order to avoid redundant shared elements or nodes (If needed, the salome file is in geom.zip).

When I'm running the simulation, the results in Paraview shows a potential map ranging from 0 to 1V, and the BC are not applied as expected... at least, from what I can see.

Does anybody know what can be the reason for this?

Thanks a lot for your support.

Regards

Mat

Ok!

So, I did some extra tests and it seems to me that the problem comes from the way I set up the mesh in Salome.

In order to simplify the process of applying BC, I was originally creating groups of faces in Salome... those groups were then appearing in the elmer boundaries, allowing me a faster allocation of BC... not good! for a reason that I ignore, this is confusing Elmer.

So, for good practice, the bodies should be defined in Salome, but not the surfaces of interest.

Now we know...

Sorry... but following a set of extra tests, I would say that my previous post was not correct.
The problem is appearing in the case of capacitance calculation by the solver... when the option is "off", then all goes good.

Is there something that I'm setting up wrong to arrive to that conclusion?
Also, are the results of capacitance calculation correct?

Hello,

Looking at this test case,

elstat_infty

which is under elmerfem/fem/tests, there is this note inside Boundary Condition 1:

! The "Capacitance body" keyword automatically sets the potential.
! Potential = 1.0

so in this test case, the potential in the boundary condition is commented out. Calculating the capacitance matrix therefore uses a potential of 1.0, regardless of other potential settings.

So it seems one should run one case to calculate the capacitance matrix, and run another case to calculate potential. You seem to have found this is to be true, so it's not necessarily a bug, but rather maybe the documentation needs to be updated?

Rich.

Hello Rich,

and thanks for your feedback. It's clarifying the use of capacitance calculation.

All the best.

M

Hi,

Yes, capacitance matrix computation needs N permutations to compute N x N capacitance matrix. This is done automatically overriding the other BCs. If your N is one then you could very well compute the capacitance from any potential BC but not so easy for N>1.

-Peter