Dear all,
Hope you are doing well. for a simulation with ( p p s ) boundary conditions which is under uni-axial tension in x direction with fix deform and is under NPT equilibration along y direction, what comment we can have for strain ZZ ? I assume because the stress in z wants to keep around 0 due to definition of free surface the simulation box in z direction fluctuates. But, how we can minimize the fluctuation ? Is there any correlation between strain in non-periodic direction and the material’s Poisson ratio ?
Many thanks in advance,
Shargh
The fluctuations in the z-dimension are tied to the x- and y-dimensions changing. If you’re pulling in x with NPT set for y and shrink-wrapped boundary for z, I imagine you would see some kind of Poisson contraction in z. Whether or not that’s realistic/correct, I’m not sure. Is there a particular reason z is shrink-wrapped instead of periodic? When I’m doing a uniaxial tension sim, I have periodic boundary conditions for all three dimensions with NPT set for the other two dimensions that aren’t being deformed.
Will
Hello Will,
Thanks for the response. Yes, there is a reason for using shrink-wrap for z. Basically, I’m working with nanomenbrane and there is no surprise that by adding free surface the story for plastic deformation and off course the failure would be change. I’m interested in that behavior. About your answer, with NPT I agree about the contraction due to poison ratio. But, for the z direction it is not what I see currently in the simulation.
Best,
Hello Shargh and Will,
To add to this interesting discussion, can empty space in Z direction be added so that the periodic image in Z dimension have no interaction and then either PPF or PPP boundary can be used? Does it effectively produce same free surface as using PPS boundary? The idea is that would minimize box dimension fluctuation in z direction.
Best
Shoieb
If you have either fixed or shrinkwrap boundary condition, the physics of your system is as if the box in that non-periodic dimension is infinitely large. The same is true if you have as system with a periodic boundary, no long-range electrostatics, and add to the box length a distance significantly larger than the cutoff distance of the largest cutoff (and atoms do not separate from your system). However, this will usually result in reduced computational efficiency due to load imbalance.
What you actually see as box length with shrinkwrap is just an arbitrary number that has no meaning beyond being the enclosing the min-max of the atom positions, so those fluctuations have no physical meaning as well.
Axel.