Inconsistent results in the calculation of strain energies

About the development of the FEM module/workbench.

Moderator: bernd

Post Reply
arag12
Posts: 2
Joined: Thu Mar 30, 2023 12:42 pm

Inconsistent results in the calculation of strain energies

Post by arag12 »

Good day everyone,

I am still new here in the forum and in the use of FreeCAD, so I apologize if I do not get everything right.

I am working on new topology optimization methods as part of my research topic. For this I need, among other things, the strain energy distribution within a discretized component. Fortunately, with the help of FreeCAD I can create complex design spaces by simple operation. Also, through FreeCAD I get all the data such as stiffness matrices, stress and strain distribution. Thus, it is possible for me to calculate the strain energy for each element. In another post in this forum, someone had also already posted the same formula viewtopic.php?t=42425:

Code: Select all

1.0/2.0 * (sxx*exx + syy*eyy + szz*ezz + 2*sxy*exy + 2*sxz*exz + 2*syz*eyz)
The colleague had the problem that he loses the units when he does a User Defined Equation. Basically, I wouldn't care if I could at least get consistent results. However, after a comparison with ANSYS, I found that completely different maximum results are calculated here. However, the distribution looks relatively the same at first glance.

Image
Image

The cantilever was constructed in the XY plane and extruded in the Z direction. The dimensions are 20x20x200mm.
On the left surface the cantilever is fixed and on the right surface a surface load pulls in negative Y-direction with 10kN.

According to Ansys this results in a maximum strain energy of about 30mJ. After entering the custom equation from above, I get the same distribution of strain energy, but only a maximum strain energy of about 10mJ.

What could be the reason for this.

For completeness, I am working with the latest compiled version of FreeCAD 0.20.2

Thank you very much for your quick answers. :)

Patrick
User avatar
NewJoker
Veteran
Posts: 3017
Joined: Sun Oct 11, 2020 7:49 pm

Re: Inconsistent results in the calculation of strain energies

Post by NewJoker »

CalculiX offers output variable for that. It’s named ENER. You can add it manually to the input file and compare with your other results (you may have to do the postprocessing in other software like GraphiX though). Comparisons with different FEA software are tricky because it’s easy to make a mistake and there can be significant differences in features/formulations. It would be easier to compare the results with Abaqus.
arag12
Posts: 2
Joined: Thu Mar 30, 2023 12:42 pm

Re: Inconsistent results in the calculation of strain energies

Post by arag12 »

NewJoker wrote: Thu Mar 30, 2023 5:10 pm CalculiX offers output variable for that. It’s named ENER. You can add it manually to the input file and compare with your other results (you may have to do the postprocessing in other software like GraphiX though). Comparisons with different FEA software are tricky because it’s easy to make a mistake and there can be significant differences in features/formulations. It would be easier to compare the results with Abaqus.
Thanks a million for your quick hint!
I will try this workaround tomorrow. To the comparison: If I just compare von-mises or the deformation, I get almost the same result with Ansys. It happens only with that formula from above. So maybe the problem lies with that.

Another question I have is: What will be the proper variation of vtk file for postprocessing, considering that there is a possibility to change the kind of vtk for exporting.

Patrick
User avatar
NewJoker
Veteran
Posts: 3017
Joined: Sun Oct 11, 2020 7:49 pm

Re: Inconsistent results in the calculation of strain energies

Post by NewJoker »

arag12 wrote: Thu Mar 30, 2023 9:55 pm What will be the proper variation of vtk file for postprocessing, considering that there is a possibility to change the kind of vtk for exporting.
To convert .frd files to ParaView format, I use this tool: https://github.com/calculix/ccx2paraview
I choose vtk format and that's it.
Post Reply