From: Pavan G (pavan.namd_at_gmail.com)
Date: Fri Mar 29 2013 - 08:39:59 CDT
Turns out the simulation is not fine. I tested for margins of 0.5, 1.0 and
1.5. In all cases water 'leaks' from around the porous media. Some water
does go thorough the media as expected. Can somebody comment on how I can
prevent this leakage? I expected the x and y images of the porous media
will essentially make it an infinite slab preventing any water leakage.
On Wed, Mar 27, 2013 at 1:51 PM, Pavan G <pavan.namd_at_gmail.com> wrote:
> Hello All,
>
> I am trying to run a simulation with 3 components; water, porous media and
> vapor. Without the vapor region the box dimensions are;
>
> cellOrigin 36.914 50.465 10.028
> cellBasisVector1 98.480 0.000 0.000
> cellBasisVector2 0.000 81.957 0.000
> cellBasisVector3 0.000 0.000 83.311
>
> I want to add a 20A vapor region on the +z side of the porous media. This
> obviously means that the simulation has to be 2d periodic (in x and y). To
> achieve this I removed the 'cellBasisVector3' line and since the vapor
> region adds 20A in the z dimension, the center of the box will shift by 10A
> along z. So I now have
>
> cellOrigin 36.914 50.465 20.028
> cellBasisVector1 98.480 0.000 0.000
> cellBasisVector2 0.000 81.957 0.000
> # cellBasisVector3 0.000 0.000 83.311
>
> I had to remove PME since a sample run failed complaining that PME
> requires PBC in all 3 directions (or something on those lines).
>
> So when I ran this simulation w/o PME and w/o z axis PBC, I got few ps of
> data before it died with this error:
>
> FATAL ERROR: Periodic cell has become too small for original patch grid!
> Possible solutions are to restart from a recent checkpoint,
> increase margin, or disable useFlexibleCell for liquid simulation.
>
> I did restart from the last checkpoint and it ran for few more ps and dies
> with the same error. useFlexibleCell is already set to 'no'.
>
> From the forums I found that increasing the 'margin' in the config file
> helps. I increased this from default to 1.5. The simulation is running fine
> and looks fine as well.
>
> I am writing to check if my approach is correct. Is this how you would do
> it?
>
> Thank you
> PG
>
This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:04 CST