[lammps-users] Ensemble selection of Poisseuille flow

Hi, all!
I have a question about Poisseuille flow with Lammps.
I wants to analyse the flowrate and shearing stress between the wall and the water molecules.
The water temperature should be steady.

Should I use fix NVE with fix temp/berendsen or just use fix NVT?
Which one is better for my issue?

I’m really confused by this ensemble selection issue.
Thanks a lot!

Any time you want to thermostat flowing atoms, you need
to subtract off the flow velocity. There are several ways
to do this in LAMMPS, using several thermostats. See
the discussion on thermostatting in sec 4.16 of the manual.


You can also apply the thermostat only to the degrees of freedom
perpendicular to the flow direction, using compute temp/partial with a
given thermostat.


Thanks a lot!
I used compute temp/com and fix_modify to subtract off the flow velocity.
But my question is about the Ensemble selection,

Because of the parabolic profile of a Poiseuille flow, it is not
completely satisfying to simply remove the center of mass velocity, as
it does not correspond to the local hydrodynamic velocity everywhere
in the channel. At least you should use compute temp/profile. Anyway
thermostating a confined flowing liquid is a very tricky business,
there are lots of papers discussing this issue, see for instance a
recent one:

Bernardi, Todd and Debra "Thermostating highly confined fluids", THE

An efficient way to avoid the problem of removing the velocity profile
is to simply apply the thermostat only in the directions perpendicular
to the flow (as the kinetic energy will redistribute spontaneously
along all the directions), using compute temp/partial.

Regarding the choice of the thermostating method, usually Nosé-Hoover
thermostating (fix nvt) gives the best results, for instance the
system remains longer in the linear response regime when using very
large forcing, e.g. shear stress (which is always the case in MD).