compute_stress/atom

Submitted by strathclyde on Fri, 10/09/2015 - 12:59

Hello all,

I am using compute_stress/atom class. I see the per atom stress is the pair_wise atom stress, not including the stress between the wall and the atom. Is there a simple method to achieve this function?

Thanks in advance.

Leo

Forums:

strathclyde | Fri, 10/09/2015 - 13:12

how about to use this equation to compute the stress with wall:

sigmaij=(xi-xj)*Force/v, xi xj is the position of the particles. of course for the wall, xi-xj is the radius of the particle if it is a sphere .

ckloss's picture

ckloss | Sun, 10/11/2015 - 17:12

Hi Leo,

correct - particle-wall is not included. But please note that this is just a problem for the first particle layer near the wall, because they transmit the p-w force to the next layer via p-p contacts

Best wishes
Christoph

strathclyde | Mon, 10/12/2015 - 19:10

Hi Christoph,

Thanks for your reply. I modified the stress/atom cpp according to the reference 'Construction of an averaged stress tensor for a granular medium' by J. Fortin which explained that the stress includes the pair stress and the inertia stress. The cpp works.

I found that the inertia stress is much smaller than the pair stress. However in the liggghts help document, the stress is composed of pair stress and Ke mainly. Which method should I use in a dynamic analysis considering the interaction between all and particle.

Best regards,

Leo

ckloss's picture

ckloss | Thu, 12/10/2015 - 22:50

Hi Leo,

as mentioned p-w stress contributions are not implemented at the moment and not straight-forward to do - unfortunately!
But if you measure within the bulk (away from the wall), there are no such contributions obviously

best wishes
Christoph