FEM CONSTRAINTFIXED has not yet been defined

[Melk]

Hi Melk and welcome to FreeCAD FEM

- as stated before use FreeCAD 0.18.1 if you would like to use FEM
- if you send a file go to the mesh right click on the mesh clear mesh will clear the mesh but KEEP the mesh parameter
- if you are new to FreeCAD FEM I would stongly recomend to use the CalculiXToolsSolver.
- this is the only solver which has lots of pre checks to get around wrong user input
- if this solver gives you a message no volum elements, than your mesh has no volume elements.
- You have a shell mesh only than I would assume. I assume in your case the mesh looks like volume, but the surface is meshed only. This can happen for various reasons. To check if a mesh has volume elements select the mesh right click display mesh info.

may be more later on ...

cheers bernd

Hi Bernd and thank you for your answers

Yes, as said at the beginning I'm new in FreeCAD and I cannot spend many time in this task, so you can assume I'm not doing everything in the most ortodoxale way.

I gonna follow your instructions so I go on with CalcuiXTools.

Well, there is a little misundertanding about AppImage links in this post https://forum.freecadweb.org/viewtopic.php?f=4&t=35546

Linux AppImages are not Linux supported but Windows ! After several seeks, I assumed I should go on with Win 0.18 version...

[Melk]

If I remesh your mesh I get the following error message from gmsh:

Code: Select all

0.0: We are going to start ...
0.0: Start Gmsh ...
22.3: Gmsh had warnings ...
22.3: Error : No tetrahedra in region 1 0 Warning : Volume 1 consists of no elements Error : OCC Project Point on Surface FAIL Warning : Converged for i=0 j=0 (err=3.2513e-06 iter=2) BUT xyz error = 210.066 in point (2.380000e+02,0.000000e+00,1.000000e+01) on surface 1 Error : OCC Project Point on Surface FAIL Warning : Converged for i=4 OCC Project Point on Surface FAIL Warning : Converged for i=0 j=0 (err=2.59332e-06 iter=2) BUT xyz error = 38 in point 
...
hundreds of errors
...
iter=2) BUT xyz error = 73.2353 in point (2.900000e+02,-7.500000e+00,2.000000e+01) on surface 12 Warning : OCC Project Point 23 is not on surface 13 Error : Reparam on face failed: curve 23 is not on surface 13 Warning : OCC Project Point on Surface FAIL Error : Reparam on face failed: curve 25 is not on surface 13 Error : Reparam on face failed: curve 25 is not on surface 13 Error : Reparam on face failed: curve 23 is not on surface 13 Error : Reparam on face failed: curve 25 is not on surface 13 Error : Reparam on face failed: curve 25 is not on surface 13 Error : Reparam on face failed: curve 34 is not on surface 13 Error : Reparam on face failed: curve 25 is not on surface 13 Error : Reparam on face failed: curve 25 is not on surface 13 Error : Reparam on face failed: curve 34 is not on surface 13 Error : Reparam on face failed: curve 34 is not on surface 13 Error : Reparam on face failed: curve 35 is not on surface 13 Error : Reparam on face failed: curve 35 is not on surface 13 Error : Reparam on face failed: curve 33 is not on surface 14 Error : Reparam on face failed: curve 33 is not on surface 14 Error : Reparam on face failed: curve 25 is not on surface 14 Error : Reparam on face failed: curve 25 is not on surface 14 Error : Reparam on face failed: curve 38 is not on surface 14 Error : Reparam on face failed: curve 38 is not on surface 14 Error : Reparam on face failed: curve 25 is not on surface 14 Error : Reparam on face failed: curve 33 is not on surface 14 Error : Reparam on face failed: curve 33 is not on surface 14 Error : Reparam on face failed: curve 39 is not on surface 14 Error : Reparam on face failed: curve 39 is not on surface 14 Error : Reparam on face failed: curve 38 is not on surface 14 Error : Reparam on face failed: curve 38 is not on surface 14 Error : Reparam on face failed: curve 39 is not on surface 14 Warning : Surface mesh: worst distortion = -41.0595 (avg = -1.04882, 256 elements with jac. 
22.3: Gmsh done!

The resulting mesh looks like this ...

screen.jpg


looks like art but not like a valid finite element mesh ... Even if you would make a fixed constraint it would fail anyway ...

As explained before my mesh looks like this

https://forum.freecadweb.org/viewtopic.php?f=4&t=35546

[Melk]

aha, you passed pad and pad001 into the BooleanFragment, but this are two shape similar. This can not work. You should either use pad or pad001 for meshing and applying loads.

Why do you use contact constraints on the faces you apply the load too? What is this useful for? See attached screen.


screen.jpg

Contact constraints on the faces I applied the loads ??? No, the contact constraints are located on the other side, just in the contact interfaces
I'm checking the file I sent right now, and I don't see any contact constraints on the upper face.

[Melk]

If I have problems with an FEM, I start to simplify the analysis (BTW: no matter if it is FreeCAD FEM or any other FEM software I work with), if I have problems I simplify the model as long as I have problems. Afterwards if it runs I start to add new stuff again. Attached a simplified modell which I can make analysises with.



Barre GRP_Mod8_bernd1.FCStd



screen.jpg

Yes as explained to our friend jmaustpc, that's what I'm doing. I try to simplify my model in order to take some results. Nevertheless, I prefered to send you a file with every detail I gotta deal with so you understand completely the thing.

[Melk]

If I have problems with an FEM, I start to simplify the analysis (BTW: no matter if it is FreeCAD FEM or any other FEM software I work with), if I have problems I simplify the model as long as I have problems. Afterwards if it runs I start to add new stuff again. Attached a simplified modell which I can make analysises with.



Barre GRP_Mod8_bernd1.FCStd



screen.jpg

Yes as explained to our friend jmaustpc, that's what I'm doing, I try to simplify my model in order to take some results. In general this strategy tends to work nicely. Nevertheless, I prefered to send you a file with every detail I gotta deal with so you can understand the hole thing.

I'm really sorry to say that your mesh have 19 elements with jac. So CalulXccXTools brings this :

*ERROR in e_c3d: nonpositive jacobian
determinant in element 10884
*ERROR in e_c3d: nonpositive jacobian
determinant in element 10884
*ERROR in e_c3d: nonpositive jacobian
determinant in element 10884
*ERROR in e_c3d: nonpositive jacobian
determinant in element 10884
*ERROR in e_c3d: nonpositive jacobian
determinant in element 10885
*ERROR in e_c3d: nonpositive jacobian
determinant in element 10885
*ERROR in e_c3d: nonpositive jacobian
determinant in element 10885
*ERROR in e_c3d: nonpositive jacobian
determinant in element 10885
*ERROR in e_c3d: nonpositive jacobian
determinant in element 10886
*ERROR in e_c3d: nonpositive jacobian
determinant in element 10886
*ERROR in e_c3d: nonpositive jacobian
determinant in element 10886
*ERROR in e_c3d: nonpositive jacobian
determinant in element 10886
1.0: CalculiX execute error:

A mesh result is produced but no results are given indeed. Are you sure you did analyses with your model ? I may skip something...

[bernd]

Well, there is a little misundertanding about AppImage links in this post https://forum.freecadweb.org/viewtopic.php?f=4&t=35546

Linux AppImages are not Linux supported but Windows ! After several seeks, I assumed I should go on with Win 0.18 version...

This is surely a misstake of the post. There are linux app images available, but any way it does not matter if you gone use Windows or Linux. I use both too.

[bernd]

aha, you passed pad and pad001 into the BooleanFragment, but this are two shape similar. This can not work. You should either use pad or pad001 for meshing and applying loads.

Why do you use contact constraints on the faces you apply the load too? What is this useful for? See attached screen.


screen.jpg

Contact constraints on the faces I applied the loads ??? No, the contact constraints are located on the other side, just in the contact interfaces
I'm checking the file I sent right now, and I don't see any contact constraints on the upper face.

ok, this one was my misstake. I have been playing to fix your broken geometry, and it moved element names around ...

[bernd]

If I have problems with an FEM, I start to simplify the analysis (BTW: no matter if it is FreeCAD FEM or any other FEM software I work with), if I have problems I simplify the model as long as I have problems. Afterwards if it runs I start to add new stuff again. Attached a simplified modell which I can make analysises with.

Yes as explained to our friend jmaustpc, that's what I'm doing, I try to simplify my model in order to take some results. In general this strategy tends to work nicely. Nevertheless, I prefered to send you a file with every detail I gotta deal with so you can understand the hole thing.
A mesh result is produced but no results are given indeed. Are you sure you did analyses with your model ? I may skip something...

I only had one nonpositive jacobian and just gave it a try and it worked. You should get rid of the non positive jacobians in your mesh to get calculix running. Since ccx does not complain for me I can not say where your nonpositive jacobian problems are, but if you use CalculiXToolsSolver and activate the mesh after calculix gave the errors the mesh nodes shoud be green. At leas you know where the problem is. Change the geometry or change the mesh parameter to get rid of nonpositive jacobians.

cheers bernd

[bernd]

just checked again

- download my file
- open the file
- mesh
- 3 elements with non positive jacobians
- try ccx anyway
-

If a max mesh size element is reduced to 3 mm I get no non positive jacobian. (63312 nodes is a lot but ok, ccx takes less about 20 seconds to run) Try this ...