Output forces on walls in 2d

Submitted by Silias on Fri, 05/20/2011 - 11:29

Hello everybody,

Because of high simulation effort I'd like to simulate particles in a container in 2d.
I already simulated a small model in 3d where the ouput of forces exerting on container walls works perfectly fine with the fix cad mesh/gran/stressanalysis.

- Now I have problems with outputing forces in 2d (fix mesh/gran/stressanalysis works just with meshes on 3d-CAD, right?)
- So, instead of importing a CAD in 3d, I use the fix wall/gran/history command to import walls in my 2d simulation. How is it possible to output forces on these walls?

Thanks very much in advance,
kind regards

Sebastian

Forums:

ckloss's picture

ckloss | Mon, 05/23/2011 - 14:43

Hi Sebastian,

you have a couple of options:

+ use a mesh wall anyway (there should nothing that would prevent you from using it in 2d)
+ use two fixes of type store/force before and after the wall fix, subtract and reduce them to a scalar

Best
Christoph

Silias | Tue, 05/31/2011 - 12:14

Hi Christoph,

thanks, the second one was a very good hint! But I have another question:

With implementing two fixes of type store/force before and after the wall fix I just can visualize forces on particles which are in contact with a wall (okay, initially that was what I wanted). Furthermore it would be very nice to see kind of "force bridges" (german: Kraftbrücken) within the bulk.

so far I'm not able to output these forces, because LIGGGHTS assumes every particle as forceless which doesn't move (because the netto force vector is 0) - even though the particle is under big pressure...

Do you have an idea, to resolve that problem?

best regards,
Sebastian

ckloss's picture

ckloss | Tue, 05/31/2011 - 12:44

>>because LIGGGHTS assumes every particle as forceless which doesn't move
This is not an assumption, the sum of the forces on a particle is of course 0 if it does not move, you need the per-contact force or stress state instead

I you want to output force chains, you should use the compute pair/gran/local command in combination with a dump local - it outputs you per-contact properties. Currently, there is no viz for that yet. Maybe you can contact JF, he already did something in that direction for another purpose

Or alternatively, you can viz the stress state via compute stress/atom, see 1 or 2 forum threads before

Christoph