[solved] steel-frame: hard to mesh

About the development of the FEM module/workbench.

Moderator: bernd

thschrader
Posts: 1416
Joined: Sat May 20, 2017 12:06 pm
Location: Germany

[solved] steel-frame: hard to mesh

Postby thschrader » Wed Jan 10, 2018 2:29 pm

This is a steel-frame (3x3,5 m, thin U-Profiles) which is
generated by a macro developed here:
https://forum.freecadweb.org/viewtopic.php?f=23&t=26092
Gmsh/CCX runs without error, the deformation and the calculated
values seems plausible. But from an "optical" point of view I would
say the mesh is invalid.
what do you think?
Thomas
Steel_Frame_For_FEM.fcstd
(270.68 KiB) Downloaded 17 times
EDIT: forgot
OS: Windows 10
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Version: 0.17.12940 (Git)
Build type: Release
Branch: master
Hash: 22ed0bce514174558bf09f6d4f0237f2ac382fc6
Python version: 2.7.8
Qt version: 4.8.7
Coin version: 4.0.0a
OCC version: 7.1.0
Locale: German/Germany (de_DE)
steel_frame.JPG
steel_frame.JPG (148.53 KiB) Viewed 683 times
meshGmshFc.JPG
meshGmshFc.JPG (111.28 KiB) Viewed 683 times
Last edited by thschrader on Thu Jan 11, 2018 5:21 pm, edited 1 time in total.
User avatar
microelly2
Posts: 4361
Joined: Tue Nov 12, 2013 4:06 pm
Contact:

Re: steel-frame: hard to mesh

Postby microelly2 » Wed Jan 10, 2018 2:50 pm

Soory I do not find a non empty mesh inside the file :?
thschrader
Posts: 1416
Joined: Sat May 20, 2017 12:06 pm
Location: Germany

Re: steel-frame: hard to mesh

Postby thschrader » Wed Jan 10, 2018 3:03 pm

microelly2 wrote:
Wed Jan 10, 2018 2:50 pm
Soory I do not find a non empty mesh inside the file :?
The file is posted with a so called "cleared mesh", otherwise the file-size is too big.
Doubleclick on analysis in model-tree ==> FEM-wb activates
Doubleclick on mesh in model-tree, hit apply, remesh starts.
run analysis.
Thomas
User avatar
microelly2
Posts: 4361
Joined: Tue Nov 12, 2013 4:06 pm
Contact:

Re: steel-frame: hard to mesh

Postby microelly2 » Wed Jan 10, 2018 3:37 pm

I get this ... :roll:

We are going to start Gmsh FEM mesh run!
Part to mesh: Name --> Fusion001, Label --> Fusion_Refined, ShapeType --> Compound
CharacteristicLengthMax: 100.0
CharacteristicLengthMin: 0.0
ElementOrder: 2
ElementDimension: 3
/tmp/Fusion001_Geometry.brep
/tmp/Fusion001_Mesh_TmpGmsh.unv
/tmp/shape2mesh.geo
gmsh
Group meshing for analysis.
Empty reference: SolidMaterial
One material with no reference shapes. No need to make a group for materials.
{'FemConstraintFixed': ['Face322'], 'FemConstraintForce': ['Face288'], 'FemConstraintForce001': ['Face323']}
('Unexpected error when creating mesh: ', <type 'exceptions.AttributeError'>)
thschrader
Posts: 1416
Joined: Sat May 20, 2017 12:06 pm
Location: Germany

Re: steel-frame: hard to mesh

Postby thschrader » Wed Jan 10, 2018 3:51 pm

Strange!
When I do a remeshing, I get this:
remesh.JPG
remesh.JPG (55.48 KiB) Viewed 650 times
report.JPG
report.JPG (62.28 KiB) Viewed 650 times
fandaL
Posts: 338
Joined: Thu Jul 24, 2014 8:29 am

Re: steel-frame: hard to mesh

Postby fandaL » Wed Jan 10, 2018 5:10 pm

thschrader wrote:
Wed Jan 10, 2018 2:29 pm
This is a steel-frame (3x3,5 m, thin U-Profiles) which is
generated by a macro developed here:
https://forum.freecadweb.org/viewtopic.php?f=23&t=26092
Gmsh/CCX runs without error, the deformation and the calculated
values seems plausible. But from an "optical" point of view I would
say the mesh is invalid.
what do you think?
Thomas
Steel_Frame_For_FEM.fcstd
I would recommend to work with shell meshes in case of parts like this.
Starting with model as is – it is possible to downgrade solids to shells and delete inner shells to obtain only shell geometry which I expect to be easy to mesh and easy to analyze.
I didn’t look at the macro, but it might be useful to add an option to generate directly shell geometry instead of volumes.
User avatar
hhassey
Posts: 134
Joined: Thu Jun 04, 2015 8:01 pm
Location: Ensenada, Mexico

Re: steel-frame: hard to mesh

Postby hhassey » Wed Jan 10, 2018 5:45 pm

thschrader wrote:
Wed Jan 10, 2018 2:29 pm
This is a steel-frame (3x3,5 m, thin U-Profiles) which is
generated by a macro developed here:
https://forum.freecadweb.org/viewtopic.php?f=23&t=26092
Gmsh/CCX runs without error, the deformation and the calculated
values seems plausible. But from an "optical" point of view I would
say the mesh is invalid.
what do you think?
Thomas
Maybe you could take the original geometry, downgrade it to shells, and compare the results of your applied forces. I have been working hard to make that FEM Switch work with 3D geometry, I am almost done with it. I would really appreciate if you could let me know if my 3D path is wrong and start exploring how hard would it be to create the 2D profiles. (this probably would call for an entire rewrite of the whole macro in order to create the 2D geometry) As I think creating 2D geometry to start with would be easier than transforming 3D studs into 2D.

I still want to make a test where I create the frame with the FEM switch off, then proceed as I did with the one I sent you:
1.- Downgrade the Compound to 20 solids
2.- Fuse the solids into one solid
3.- Refine this solid to take away unwanted edges.

maybe since the studs and tracks are "touching" they might mesh fine and we would not need to overlap them.

What are your thoughts/

Thanks!

Humberto
thschrader
Posts: 1416
Joined: Sat May 20, 2017 12:06 pm
Location: Germany

Re: steel-frame: hard to mesh

Postby thschrader » Wed Jan 10, 2018 6:07 pm

hhassey wrote:
Wed Jan 10, 2018 5:45 pm
What are your thoughts/
Thanks!
Humberto
My thoughts are:
This is a simulation problem in FEM, not a problem of your macro or of
the steel frame. I have calculated the frame with beam elements, as i posted
before (pic below). The calculation was done according to "Theory of second order".
That means: the static equilibrium is computed at the deformed system, you automatically
get a global stability-check. Your construction is ok. Keep in mind, that the maximum
system utilization occurs at the openings (windows/doors).
From a FEM-simulation point of view I think fandal is right. It is better to use 2D-shell
elements instead of 3D. A 3d-approach might work when the profile-thickness is bigger.
In the end I would calculate such a frame structure with beam elements (my daily work)
Thomas
structure_OK.JPG
structure_OK.JPG (72.18 KiB) Viewed 625 times
User avatar
bernd
Posts: 8332
Joined: Sun Sep 08, 2013 8:07 pm
Location: Zürich, Switzerland

Re: steel-frame: hard to mesh

Postby bernd » Wed Jan 10, 2018 9:11 pm

Simplest analysis would be with beams like Thomas did. But shell analysis would be the smart way because all buckling could be done too.

Solid analysis could make problems in meshing because of the thin thickness.

The mesh looks fine to me. What you could do thomas is, use a freuquency analysis instead of a static analysis. If you have a look at the first 5 or 10 eigenmodes you wll see if a member is not connected. They all look good to me.

bernd
thschrader
Posts: 1416
Joined: Sat May 20, 2017 12:06 pm
Location: Germany

Re: steel-frame: hard to mesh

Postby thschrader » Thu Jan 11, 2018 1:25 pm

The first four eigenmodes of the frame at 11,6/15,2/17,7/34,1 Hz.
I can see no gaps. Generated frame with FEM=False, downgrading,fusing,refining
as Humberto described above.
@Humberto: macro runs fine
eigenmodes_frame.JPG
eigenmodes_frame.JPG (129.99 KiB) Viewed 570 times