Non Linear Spring-back calculation imposing displacements

[freman4]

Hello engineers and developers and Z88 supporters,

first of all, congratulations for this useful FEA tool. Please continue developing keeping it free

Looking for previous similar questions in this formum I didn't find any useful answer so, I hope this will be useful also for other in the future.

Few days ago, working on an university project, I came across a problem of plastic deformation and I would like to understand if Z88 Non linear environment is capable to calculate the final shape of a plastically deformed object activating the spring-back option.
What is special about this problem is the fact that the known boundary conditions are not the forces, but the displacements, ence I need to calculate the spring back of a plastically deformed object imposing the displacements.
I already performed many simulations on a simple geometry and studied the really useful Z88 manuals, but it seems to me that the software is not capable to perform such an analysis and my conclusion is that the NON-Linear Spring Back calculation in Z88 is only possible when imposing external forces as boudary conditions, not displacements.

Is my conclusion right?
Is it possible to calculate the spring-back in Z88 imposing displacements as a boundary condition?
Is there any workaround you may suggest?

Thank you in advance for any support you may give

[DGoller]

Hello freman4,

thank you very much, we are pleased to hear from people who enjoy our work

like you already figured out, in Z88 the nonlinear calculation is unfortunatelly not possible if you impose only displacements as boundary conditions.
I suggest you try to substitube the displacements with forces, by imposing force boundary conditions and try to find the correct values by means of several simulations.

Good luck,
DGoller

[freman4]

Dear DGoller,

thank you for your reply, that is the confirmation I needed.
Actually, in the meantime I realized additional sample simulations using forces as a boundary condition instead of displacemens and Z88 calculation of spring-back gave good results.

I anyway need to solve an additional problem regarding the same application: two of the faces of the object have to remain parallel during the deformation.
In order to obtain this, I tried to perform some simulations with a mixed boundary conditions setup: imposition of forces to obtain the main deformation, plus imposition of desired displacement to obtain that a faces will remain parallel to another.
The result was not encouraging, seems Aurora doesn't like such a mixed setup for large plastic deformations and the solver stopped during one of the last iterations for excessive deformation.

I attached two example pictures for a better understanding: one of an undeformed sample and another of the desired final deformation I would like to simulate.

Thank you again for the support you may give.

freman4

[DGoller]

Hello freman4,

is it possible to attach an archive containing the project files? This would enable us to check the imposed forces and displacements in detail to determine the difficulties.

DGoller

[freman4]

Hello DGoller,

of course it is.
Please have a look at the attached archive.
I removed the output files as they are too eavy, but the setup you will find is ready to be calculated (it may took ~ 1 hour or ~ 1 hour and half to be executed).

The problem blocking the calculation is guided by the "Y" axial displacement imposed to obtain that the two extreme faces of the geometry remain parallel:
- for very small "Y" axial negative imposed displacements (e.g.: ~ -0,7mm) the calculation terminates all the load steps (including the springback calc.), but unfortunately such a displacement doesn't corresponds to reality;
- for more realistic, but higher "Y" axial negative displacements (e.g.: for values bigger in norm than -1mm) the calculation fault to terminate.

...and however, in both the cases a realistic deformed shape is obtained for only the first percentages of the imposed deformation (e.g.: up to the load steps of 30 - 40 %); at the following load steps, is clear that the resulting calculated deformed shape is corrupted and non realistic.

My personal hypothesis :
1) The stress-strain curve of the material used is too flat in the plastic zone (even if this is the a realistic curve for such a material) and the convergence is difficult for that reason.
2) The mesh become too much deformed and the Von Mises non linear model used does not correctly represents the reality of the problem.

Other ideas?

Thanks,

freman4

[DGoller]

Ok thank you for the project folder. I'll run it and will give you feedback as soon as possible.

[DGoller]

Hello freman4,

i have made several simulations but did not yet get to the bottom of the problem. I already altered the stress-strain curve and the number of load steps, but unfortunately no improvement.
To your points:
1) yes this is a general problem and our algorithm is not meant for materials with such flat stress-strain curves
2) This could be a Problem and leads to inaccurate results because it should be calculated with consideration of geometric non-linearity. However unfortunately in z88 material nl can not be combined with geometric nl at current state.

sincerely,
DGoller

[freman4]

Hello DGoller,

thank you for the support. It is however useful having had a confirmation that there weren't errors in the setup and that my hypothesis to explain the non-convergence of the problem were realistics.

Aurora Z88 remains a great free software (better if it would also opensource, to benefit of the open development) and it would be fantastic if the NON-Linear simulation part would be further developed and the multibody meshing capability would be implemented.

If you have any additional suggestion about how to workaroud the problem to simulate such a non-linear behaviour, will be welcome.



Have a nice time.
Regards,

freman4