Hi Steve,
I have a question on this very old post about calculating the viscosity using Muller-Plate Method using fix viscosity. Here is the relevant post:# Either this (both are the same):
http://ehc.ac/p/lammps/mailman/lammps-users/thread/6B126C65D016E2439B381C6CDB783CF4021679A6@…775…/
or
http://lammps.sandia.gov/threads/msg09610.html
Here are the comments from you and the previous post:
Previous post: I also take the velocity profile from the end of the run reported by fix 2. I plot the x-direction velocity as a function of the z-height and take the slope of the linear fit. This is then dvx/dz, which has a value of 0.0098.
You: If you look in the MP paper (cited on the LAMMPS doc page), there is a factor of 2 in this calculation, which you may be missing.
My concern:
As I understand, the original paper has discussed on the PBC and that factor has already been included in the formula:
j= (f_3)/(2tLx*Ly)
Suppose, for simplicity though we can always do regression, I calculate the slope like this:
Slope = [vx(lx/2)-vx(0)]/[lz/2] i.e velocity difference at the middle and bottom layer divided by half the box size.
Initially, I thought that the total flux is through that face of the box. However, as written in the lamps website: “The scalar is the cumulative momentum transferred between the bottom and middle of the simulation box (in the pdim direction) is stored as a scalar quantity by this fix”. Do I still need the extra factor as you mentioned in that post if I calculate slope as mentioned above?
Thanks in advance for you comment.
Subas Dhakal
Syracuse University