← Back to team overview

yade-users team mailing list archive

Re: [Question #699428]: Cannot simulate periodic box with walls

 

Question #699428 on Yade changed:
https://answers.launchpad.net/yade/+question/699428

    Status: Open => Answered

Jan Stránský proposed the following answer:
> Some sort of wall is essential

depends on what "essential" is.
Also depends on what "some sort" is, because periodic boundary conditions essentially are "some sort of wall"..

> when the box is changed using PeriTriaxController it seems that all
particle coordinates are scaled.

basically yes, but it can be controlled and changed, see [3].
If you set homoDeform=0, you should not get the "scaling"

> This means that stresses will be distributed uniformly.

rather strain is distributed uniformly, stress is then result of stiffness and packing structure.
If the packing is "uniform", then also the resulting stress would be "uniform".
However, if the packing is "non-uniform", also the resulting stress would be "non-uniform".

The "scaling" is only one component of the particles motion, and
therefore only one component to the resulting stress. There are of
course other components (interparticle interactions mainly) influencing
the result.

> This is not the case in real uniaxial compaction, where the stress
varies depending on the height of the slice you are looking at.

What is the reason of stress variation? Dynamics? Gravity? Something else?
For "fast" loading yes, the stress waves should not be neglected and "wall" loading is more realistic.
Contrary, for "slow" loading, this "periodic approach" is actually more realistic than the loading induced only by planes (which induce stress waves not present in reality).

> Using a box as given in [2] did not work

is it still freezing, or just the behavior and results are not as
expected?

> PeriTriaxController ...

I suggest either:
- use PeriTriaxController with no walls
or:
- use "enlarged" periodic cell with only two periodic (transverse) directions and the axial direction enlarged (essentially making it non-periodic axially). Then you can push the sample by walls and somehow control stress/strain in the transverse directions.
or:
- probably both approaches can be used together. The plates are moving "independently", the axial direction is strain-controlled with zero change, the transverse directions are controlled by PeriTriaxController. Have a try :-)

> If I do not fix the plates and give them a velocity then the plates move away.
> How can I make the box-plates move at a constant speed (irrespective of the other forces on them)?

b.state.blockedDOFs = "xyzXYZ # [4]
b.state.vel = someConstantVelocity

> ... dynCell

basically yes, you can compute stress assuming static equilibrium or you
can somehow take into account inertia of the particles.

> Please let me know if I should add a bug report on GitLab about the
problem with walls and periodic cells.

yes, it seems like a bug

> if there is a way that I can help improve the documentation on the
YADE website, please let me know.

Yes. the documentation is part of the source code. You "just" download
the source code, modify it, and commit the changes to the gitlab. I
don't remember exactly the process, but it is described in documentation
and is out of scope of this question anyway (of course you are free to
open a new question aiming on this).

I plan on implementing new features in YADE and would like to get
involved with the development.

Cool, good luck :-)

Cheers
Jan

[3] https://yade-dem.org/doc/yade.wrapper.html#yade.wrapper.Cell.homoDeform
[4] https://yade-dem.org/doc/yade.wrapper.html#yade.wrapper.State.blockedDOFs

-- 
You received this question notification because your team yade-users is
an answer contact for Yade.