Bouncing rigid body

Submitted by tapsab on Fri, 08/07/2015 - 06:20

Hi,
I am trying to simulate a packing of non-spherical particles using the "hybrid granular molecular" style and "fix rigid" options. I ran a test wherein a single clump of particles hits a flat surface and I expected it to bounce a couple of times and settle down. Instead it keeps bouncing around even with a really low coefficient of restitution. Based on previous discussions in the forum I understand that this has to do with the damping not being computed accurately with the "fix rigid" approach. Are there any workarounds for this problem or any pointers on what one might have to change in the code to get a reasonably accurate result.

Thanks,
Tapan

Forums:

ckloss's picture

ckloss | Mon, 08/24/2015 - 11:05

Hi Tapan,

this is a well-known effect for the "standard" straight-forward implementation of clumps/multi-sphere. There is a Powder Technology paper by Harald Kruggel-Emden on that topic. We've resolved that in our commercial support version (premium version)

Best wishes
Christoph

tapsab | Fri, 08/28/2015 - 04:42

Christoph,
Thanks for the response. I did a little digging around the code and I think the problem is that in the "hertz" pair style the effective mass used to compute damping between a rigid body and another particle uses the mass of the constituent particle of the clump rather than the mass of the whole clump. I think a workaround for this is to instead use the "hooke/stiffness" pairstyle with "absolute_damping" set to "on". You can then directly feed in the right damping coefficient that is computed using the clump mass. Let me know if I am missing something obvious here though that makes this approach wrong.

Thanks,
Tapan

cheng1988sjtu | Wed, 09/02/2015 - 17:07

Hi Tapan,

Have you find a solution for this problem? I think I'm facing a similar problem here, and this problem is extremely serious if I design two particles with large overlaps.

C.Z. U of D

tapsab | Thu, 09/03/2015 - 03:49

Hi,
I ended up using the hooke/stiffness pair style with the "absolute_damping on" option and the results seem reasonable. The values of kn and gamman_abs are computed based on material properties as per equations given in the manual. I assigned the same kn for all interaction pairs wherein kn is computed using meff and Reff that correspond to properties of the regular particles. For gamman_abs, I computed a different value for each interaction pair. The term that changes between the different interaction pairs is meff. Thus if your regular particles have a mass m and the clump has a mass M, then meff = m*M/(m+M) for the regular-clump interaction, meff = (m*m)/(m+m) for the regular-regular interaction and so forth.

Best,
Tapan