Question regarding StructuredMeshMapper

Extension of Elmer in computational glaciology
Post Reply
norbert
Posts: 9
Joined: 25 Sep 2015, 09:49
Antispam: Yes
Location: Hannover, Germany

Question regarding StructuredMeshMapper

Post by norbert » 18 Nov 2019, 00:22

Hi all,

I have tried to extrude a mesh following the steps in the Tete Rousse Glacier (thanks go to Olivier GAGLIARDINI for the excellent example). I am interpolating measured values and extruding a mesh with rectangular base created using ElmerGrid.

My problem is, that if I open the mesh in ElmerGUI I see that it is not a slab with the top (and bottom) corresponding to my measurements. The sides are open and only one of the rectangle's sides have been extruded upward (see attached picture). I have also tried assigning the boundaries 2, 3, 4 with a do nothing condition. This however then causes the extrusion to fail and the grid to stay 2D. Can someone with more experience in the use of the structured mesh mapper in combination with the "Grid2DInterpolator" maybe give me a hint why the sides are not closed?

Thank you,

Norbert
Attachments
ExtrudedMesh.PNG
ExtrudedMesh.PNG (13.69 KiB) Viewed 244 times

gagliar
Posts: 67
Joined: 04 Sep 2009, 16:34
Location: LGGE - Grenoble
Contact:

Re: Question regarding StructuredMeshMapper

Post by gagliar » 18 Nov 2019, 09:49

Dear Nordert,

Can you post your sif and 2d footprint mesh so that I have a look? Did you try to visualize the resulting 3d mesh in paraview instead of ElmerGUI?

Regards,
Olivier

norbert
Posts: 9
Joined: 25 Sep 2015, 09:49
Antispam: Yes
Location: Hannover, Germany

Re: Question regarding StructuredMeshMapper

Post by norbert » 18 Nov 2019, 12:04

Hi Olivier,

I have visualised the mesh in Paraview. It looks correct there, however, when I use the geometryIDs to filter for my Boundary Indices I only have the three boundaries visible in ElmerGUI. The other Ids are empty.

I have attached my sif file. It is in principal your example with most solvers set to Exec Solver = "never". I added Mapped Mesh Name = String my_mapped_mesh to the StructuredMeshMapper.

The other BCs are at the moment commented out.

I attach the footprint mesh as well.

Thanks,

Norbert

Edit: Attached the forgotten DEM files as well
Attachments
DEM.zip
(183.23 KiB) Downloaded 9 times
frmteterousse1.sif
(6.61 KiB) Downloaded 10 times
dimple2d.grd
(639 Bytes) Downloaded 10 times

gagliar
Posts: 67
Joined: 04 Sep 2009, 16:34
Location: LGGE - Grenoble
Contact:

Re: Question regarding StructuredMeshMapper

Post by gagliar » 18 Nov 2019, 13:41

Hello,

I can already see one problem:

Variable 1 lx = Real 6512.347200
Variable 1 ly = Real 2022.023200

are not the correct lengths (lx = 6.5013280e+03 and ly = 2.0110040e+03 in the DEM files).

Does it improve results to have the correct values?

Regards,
Olivier

norbert
Posts: 9
Joined: 25 Sep 2015, 09:49
Antispam: Yes
Location: Hannover, Germany

Re: Question regarding StructuredMeshMapper

Post by norbert » 18 Nov 2019, 13:49

Hi Olivier,

this does not change the result.

I think I left the lx, ly values from a different surface topology in the sif file. :oops:
I tried several of my DEM files to understand if has to do with my input data.

Regards,

Norbert

norbert
Posts: 9
Joined: 25 Sep 2015, 09:49
Antispam: Yes
Location: Hannover, Germany

Re: Question regarding StructuredMeshMapper

Post by norbert » 18 Nov 2019, 16:10

Hi all,

I have also looked to extrude the original Tete Rousse data with a rectangular footprint. Gives the same result. As I have used the StructuredMeshMapper before on a rectangular footprint without problems but a MATC expression for the heights it seems to be related to the Grid2DInterpolator. Or at least this is a guess. Maybe someone more familiar with the two solvers can help here.

I have looked the Tete Rousse original msh contains only one Boundary Number and thus results in a closed geometry.

Best Regards,

Norbert

gagliar
Posts: 67
Joined: 04 Sep 2009, 16:34
Location: LGGE - Grenoble
Contact:

Re: Question regarding StructuredMeshMapper

Post by gagliar » 18 Nov 2019, 16:59

Hi Norbert,

By convention, the boundary conditions in your sif file should write:

Boundary Condition 1
Target Boundaries = 1
...
End

Boundary Condition 2
Target Boundaries = 2
...
End

Boundary Condition 3
Target Boundaries = 3
...
End

Boundary Condition 4
Target Boundaries = 4
...
End

! cavity roof and Bedrock
Boundary Condition 5
! Bottom Surface = 0
Bottom Surface = Equals BedDEM
...
End

! Upper Surface
Boundary Condition 6
Top Surface = Equals ZsDEM
End

The bottom surface is the first BC after the targeted boundaries from the mesh, the top one comes after. So, if you have N BC numbered in your 2d footprint mesh, the bottom surface BC is N+1 and the upper surface BC will be N+2.

That should solve your issue?
Regards,
Olivier

norbert
Posts: 9
Joined: 25 Sep 2015, 09:49
Antispam: Yes
Location: Hannover, Germany

Re: Question regarding StructuredMeshMapper

Post by norbert » 18 Nov 2019, 17:34

Hi Olivier,

great that was my problem. Thanks a lot for your time and effort.

I admit that I didn't think too much about the order of the BCs :oops:

Best Regards,
Norbert

Post Reply