Hi guys,
I have a question.
I am running simulations that needs to be constrained in the z axis and periodic in x and y.
Since I am working with high volume fractions (higher than 0.7), I use the particletemplate/sphere command with an initially low volume fraction, then I compress the box to higher it.
This compression has to preferrably be isotropic (https://www.comsol.nl/model/isotropic-compression-with-cam-clay-material... - for more info).
#1 - What's in my simulation box?
In my simulation box, I have two materials and four layers. In poor words, the two central layers are constituting the "real" sample, while the two external layers are very thin layers that are needed to create a wall made of particles (I generally call this, as "bumpy wall"), since I am not using a mesh?
(for more details see my code, please)
#2 - My approach
After reviewing some commands (see end post) that could compress the REV by reducing its volume, I opted for starting with pp pp ff conditions, inserting particles and after this, define the two "bumpy walls" as rigid bodies (fix rigid), made of particles with addforce (addforce command).
#3 - My problem
With my approach, I see that when I am compressing the box, I am loosing particles, therefore Liggghts cant create a wall since it does not find the particles to make it
Does anyone has ideas or suggestions about how to:
reduce the volume of a box and then tell liggghts that two regions of the reduced volume, should become rigid bodies?
Cheers!
Luca
OTHER INFO
* CODE ATTACHED
** Main adopted commands
http://www.cfdem.com/media/DEM/docu/change_box.html
http://www.cfdem.com/media/DEM/docu/displace_atoms.html
http://www.cfdem.com/media/DEM/docu/fix_move.html
http://www.cfdem.com/media/DEM/docu/fix_deform.html
***(LIGGGHTS VER 3.3.0)
| Attachment | Size |
|---|---|
 forum_12_3_2015.tar_.gz | 6.97 KB |
robin | Fri, 12/04/2015 - 09:18
Boundary
Hi Luca,
If you use f boundary and then do an isotropic fix deform, you may indeed loose particles:
"For style f, the position of the face is fixed. If an atom moves outside the face it may be lost. "
(http://www.cfdem.com/media/DEM/docu/boundary.html)
A classical thing would be to run two distinct simulations one after the other:
- 1) Build an (homogeneous) sample of particles
- 2) Re-use the result, applying the different properties you need in different region (commands: region, group...)
Cheers,
Robin