Calculation of stress/atom

Submitted by tapsab on Thu, 08/30/2012 - 10:34

Hi,
For the calculation of stress/atom it seems only the neighboring particles are accounted for but not the contact forces that occur between a particle and a wall. Is this by design or something that might have been overlooked.

Thanks,
Tapan

Forums:

ckloss's picture

ckloss | Thu, 08/30/2012 - 10:59

Correct - wall forces are missing, so this part is missing in the first layer of particles. However, including this would only help for (rare) academic cases where you use box walls which are aligned with the simulation box. Otherwise, near-wall stress calculation would be incorrect anyway because the averaging volume is unknown. And calculating this averaging volume will be quite tricky - it would need some clever Monte Carlo and/or convex hull algorithm. For this reason, we haven't put too much effort into including the wall forces

Christoph

tapsab | Thu, 08/30/2012 - 18:21

Christoph,
Thanks for responding. I have an additional question regarding the stress/atom calculation. It seems the equation for virial stress (with just the pair contribution) is geared towards a system of mono-sized particles wherein the stress per atom is just the sum over all the contacts of the contact force multiplied by half of the distance between the two particles (basically the distance between the particle center and the contact point). However if you have say two particles of different radii in contact, this would not hold since the contact point will be closer to the center of the smaller particle than the larger particle and you cannot just divide the distance between the two particles by two. Do you agree?

ckloss's picture

ckloss | Sun, 09/16/2012 - 23:50

Hi Tapan,

to be honest - I think the current implementation should be fine - but I am not 100% sure. I will discuss with the LAMMPS guys and post the results..

Cheers, Christoph