FEM analysis issue

[XtyM]

Hi;

I am trying replicate this model from youtube https://www.youtube.com/watch?v=Y9l1zYnMj4g

I did the model, (it is attached). During the mesh generation I received the error that I have previously read in this forum about (gmsh 4.X) Gmsh error: Unknown element type for UNV export (MSH type 15)
https://forum.freecadweb.org/viewtopic.php?t=33946

The mesh is generated with this warning, (if I reduce the resolution of the mesh, this warning disappears and a new one appears indicating the distortion of the average values in the avg mesh) in any case I have said situations I have been able to start the calculation with other models And has worked.

The problem arises when I start the calculation with calculix which never ends up entering in exorbitant times for what this is (unless there is an error that does not find a solution during the iterations .... I do not think this is the case because the model is identical to the one made in the video whose link I have attached).


OS: "Fedora release 30 (Thirty)"
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Version: 0.18.Unknown
Build type: Unknown
Python version: 3.7.3
Qt version: 4.8.7
Coin version: 3.1.3
OCC version: 6.9.1.oce-0.18
Locale: Swedish/Sweden (sv_SE)

Om Calculix i did compile binary accord this post https://forum.freecadweb.org/viewtopic.php?f=18&t=34024, I did check it with another model and this did work.

Regards.

[XtyM]

Can someone load the file and execute an analysis with FEM calculix to check if it is a issue only with my freecad?? In my pc calculix is running but never stop. the model was created like in the video attached, the "solver" should take less than 1 min to this model. Is there something what i am missing up? or it is just a fail into the gmsh mesh and calculix solver?.

btw my gmsh version is 4.1.5

Thanks in advice.

[UR_]

Some remarks regarding contact analyses:


- Better use linear elements otherwise calculations are very time consuming

Screenshot 005.png (15.06 KiB) Viewed 2346 times

Please obey rules regarding how to assign master an slave faces


- master/slave assignment:
The larger of the two surfaces should act as the master surface.
If the surfaces are of comparable size, the surface on the stiffer body should act as the master surface.
If the surfaces are of comparable size and stiffness, the surface with the coarser mesh should act as the master surface.



Now ccx_2.15_pardiso needs 4 sec to solve

[Kunda1]

Moved to FEM subforum

[XtyM]

Hi thank you. I changed the value of element order to 1 like you did indicate, but calculix stay running and i need close the process via terminal. I dont know what is going wrong.

After i did complie calculix i checked libraries and it was result

[jorge@fedora31 Calculix_2.15]$ ldd ./ccx_2.15
linux-vdso.so.1 (0x00007ffeeb3cb000)
libarpack.so.2 => /lib64/libarpack.so.2 (0x00007f258f8b8000)
liblapack.so.3 => /lib64/liblapack.so.3 (0x00007f258f228000)
libblas.so.3 => /lib64/libblas.so.3 (0x00007f258f1d0000)
libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f258f1a8000)
libgfortran.so.5 => /lib64/libgfortran.so.5 (0x00007f258ef00000)
libm.so.6 => /lib64/libm.so.6 (0x00007f258edb8000)
libmvec.so.1 => /lib64/libmvec.so.1 (0x00007f258ed88000)
libc.so.6 => /lib64/libc.so.6 (0x00007f258ebb8000)
libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f258eb98000)
libquadmath.so.0 => /lib64/libquadmath.so.0 (0x00007f258eb48000)
libopenblasp.so.0 => /lib64/libopenblasp.so.0 (0x00007f258c6c8000)
/lib64/ld-linux-x86-64.so.2 (0x00007f258fb30000)

[bernd]

- Better use linear elements otherwise calculations are very time consuming

- master/slave assignment:
The larger of the two surfaces should act as the master surface.
If the surfaces are of comparable size, the surface on the stiffer body should act as the master surface.
If the surfaces are of comparable size and stiffness, the surface with the coarser mesh should act as the master surface.

addend here: FEM_ConstraintContact

[UR_]

After i did complie calculix i checked libraries ...

Perhaps you should solve some reference inp files before to ensure it's a proper build.


Please run from shell (without using FreeCAD)

Official example inp file from test suite and result file for reference:

From ccx Test Suite.7z

(1.32 KiB) Downloaded 47 times

BTW, for using FEM WB it's recommend to use latest FreeCAD build, ATM version 0.19

[XtyM]

I did run the text with the file you did posted, It was ok except for what is taking only 1 CPU.

I have load the output from terminal and the check file what coincide exact with the file downloaded.

About FEM_ConstraintContact i think i have tested all combinations possibles.

Thank you for your support.

Regards.!

[bernd]

Official example inp file from test suite and result file for reference:
From ccx Test Suite.7z

just as a hint for others the ccx test suite is meant not FreeCADs one.

[bernd]

Official example inp file from test suite and result file for reference:
From ccx Test Suite.7z

if you make a FreeCAD file (or even better immediately code it in python) for it I will add it to the FreeCAD FEM examples. https://github.com/FreeCAD/FreeCAD/tree ... emexamples