stress-free packing

Submitted by SergeiD on Sat, 04/16/2011 - 13:24

Hello.
How to generate a stress-free random dense packing with prescribed porosity and radius distribution? Is there some hints?

Forums:

JF's picture

JF | Sat, 04/16/2011 - 20:17

Hi Sergei,

What do you mean by stress-free ? you don't want some overlaps between the particles and the wall.

JF

SergeiD | Sat, 04/16/2011 - 23:41

I want to generate a random dense packing with negligible overlaps.
I need to control porosity and sphere's size distribution.
I suppose to use sphere's growth method as in examples/packing.
But for big solid volume fraction it give big stresses in resulting packing. So, need to expand a packing in order to lower it.
Is there some hints how to use it more clever?
May be to use a periodic boundary or some manipulations with friction.
I ask because maybe someone has already solved this problem?

raguelmoon's picture

raguelmoon | Sun, 04/17/2011 - 08:15

Hi,
I generated such packings by two methods: In one method, I poured particles from top and settled down due to gravity. I poured particles in small chunks with low volume fraction as though they were filled with hopper slowly. In the other method, you can use technique given in example by which we pour particles in once in whole simulation box with lower volume fraction and then increase their diameters to make packing denser up-to 6.5 or so.
Best,
Ram

Ram

JF's picture

JF | Sun, 04/17/2011 - 13:53

Hi,

To generate with LIGGGHTS your sphere packing whitout overlaps between sphere-sphere and sphere-wall, you need to develop a function to control the stress applied on the walls. Like this, you can use in a same time the growing algorithm in order to reach your distribution size and the movement of the walls to decrease the overlaps.

I don't know, if the LIGGGHTS users who work on triaxial test have finished to develop this function. However, the control of the porosity will be very tricky.

Regards
JF

ckloss's picture

ckloss | Sun, 04/17/2011 - 15:29

I would do it like this: Take the packing example, and by changing the scripting you can control the final particle diameter (just some easy changes to the script). Make all the particles, say 1% larger (in the order of the expected overlap) that you need them.
And then, after the packing is generated, apply a final fix adapt and shrink them by 1%, this removes the overlap. You can check if there is any more particle overlap by applying a compute pair/gran/local command

Christoph

renekaiser | Sat, 04/16/2011 - 20:30

The porosity is a result of the particle-size-distribution and the parameters how to mix.
So i think you search away to discripe the psd of a powder.
You can split the psd in some discrete ranges an fill them with the pour/dev command.

If it shoud be denser, put some pressure on top of it. Or if it should have more porosity "wiggle" it.