ElmerSolver seems to ingore second body

Numerical methods and mathematical models of Elmer
Post Reply
posch
Posts: 2
Joined: 29 Apr 2019, 00:11
Antispam: Yes

ElmerSolver seems to ingore second body

Post by posch »

Hi,

I'm trying to do some basic calculations using the StatCurrent solver.

I've got a simple geometry consisting of one smaller cubiod within one big cuboid. The two cuboids are different bodies and have a different conductivity. The boundary conditions are two different potentials at two opposing surfaces of the big cubiod.

The problem is: the result isn't influenced by the smaller cuboid at all. Regardless of what the conductivity of the inner cuboid is, the result is always the same. The inner cuboid always has a constant potential of 0 and no volume current.

I have attached my mesh and .sif files.

Does anyone have any idea where the problem might be?

Best regards
Stefan
Attachments
mesh.zip
(18.29 KiB) Downloaded 229 times
case.sif
(2.11 KiB) Downloaded 218 times
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: ElmerSolver seems to ingore second body

Post by mzenker »

Hi,

it seems that the two bodies are not connected, i.e. the nodes at the interfaces are not the same for both bodies. How did you build your geometry and mesh? Which software did you use?

Matthias
posch
Posts: 2
Joined: 29 Apr 2019, 00:11
Antispam: Yes

Re: ElmerSolver seems to ingore second body

Post by posch »

Hi,

I built my mesh using 2 .stl files, converting them to the Elmer Mesh format with netgen and merging them together with ElmerGrid.

Like this:

Code: Select all

netgen -batchmode -geofile='block_big.stl' -meshfile='block_big' -veryfine -meshfiletype="Elmer Format"
netgen -batchmode -geofile='block_small.stl' -meshfile='block_small' -veryfine -meshfiletype="Elmer Format"
ElmerGrid 2 2 block_small -boundtype 1 1 7
ElmerGrid 2 2 block_small -boundtype 2 2 8
ElmerGrid 2 2 block_small -boundtype 3 3 9
ElmerGrid 2 2 block_small -boundtype 4 4 10
ElmerGrid 2 2 block_small -boundtype 5 5 11
ElmerGrid 2 2 block_small -boundtype 6 6 12
ElmerGrid 2 2 block_big -bulktype 1 1 1  ! make bodies [1,1] to 1
ElmerGrid 2 2 block_small -bulktype 1 1 2 ! make bodies [1,1] to 2
ElmerGrid 2 2 block_small -in block_big -out merged_blocks -unite -merge 1.0e-10
Best regards,
Stefan
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: ElmerSolver seems to ingore second body

Post by raback »

Hi Stefan

The "merge" option can only merge meshes that are conforming i.e. have nodes at the same location at the interface of interest. If you have no special strategy to ensure that then probably only a few corner nodes will be merged.

-Peter
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: ElmerSolver seems to ingore second body

Post by kevinarden »

merge 1.0e-10

the value is the tolerance on merge. That is a fairly tight tolerance on a mesh that has nodal locations defined in the range of +/- 0.5

That value would be lost in the accuracy of the computer's ability to track significant digits.

If the smallest distance intended between nodes is 1.E-2 then the merge tolerance could be 1.E-4
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: ElmerSolver seems to ingore second body

Post by kevinarden »

I imported the mesh into GMesh and the nodes between the bodies are far apart. The two bodies have to share the same nodes at the interface. The inner body is much finer than the outer body. Merge is not going to work for this mesh.
nodes.png
(196.97 KiB) Not downloaded yet
Post Reply