[gmx-users] acc-grps and acceleration

[Kamps, M.]

Dear GMX-users,

I have a question relating the acceleration of certain atoms in my system.
I am trying to simulate a flow of a liquid between two gold plates.
I've created a small, narrow and long box filled with this liquid.

In order to create the flow, a acceleration is applied to the liquid.
However I'm having difficulty understanding the behaviour of this
acceleration. I understand that, due to computational reasons, the
applied acceleration is very high (0.1 nm/ps^2 = 1x10^14 m/s^2).
However, due to the small time-scale, this acceleration still leads to
usable results (literature indicates this).

I'm having difficulty with the development of the flow. I would
suspect to see the flow velocity develop over time. I would suspect
that at t=0 the average velocity is zero, while at t=a the flow would
have a velocity which is non-zero. At t=b the applied acceleration
(=constant force) would be overcome by the fluid/wall friction, and a
steady-state flow would emerge. However, the behaviour that I'm seeing
does not show this 'development' in flow velocity. At t=0 the velocity
is already at a higher velocity and this velocity does not increase
over time.

I created my system and thoroughly applied EM and equilibrated it. In
my final equilibration step I generated a velocity through gen_vel,
while no acceleration was applied. For my final MD run no velocity was
generated and the final simulation was continued from this
equilibration step, with an acceleration.

I've tried to apply different velocities (1 0.1 0.01 0.001, all in
correct direction), and I've varied with the length of the simulation
(500ps, 50ps, 5ps and 0.5ps) and with the dt value (0.001 0.0005 and
even 0.00005).

Any ideas as to what is happening? Or is this behaviour normal and my
interpretation wrong?