hi, i am confused with dpd thermostat and how to control temperature and get correct velpcity of a poiseuille flow.

Hi,I am simulating poiseuille flow in a 200*104*1 domain (2D),fluid particles are in 200*100*1 region and the rest region is filled with wall particles.An external body force of 0.02 is added to each fluid particles. I used 'compute temp/partial' to exclude the velocity of X component, and use 'dpd pair style', NVT ensemble is for all particles.'fix_modify temp'is also used to control the temperature. However, these command cannot control the temperature and the velocity profile, shear stress distribution profile are not correct.
    I attach my input file and wish you could point out my mistakes.

2013-07-07

Xiaojun Li (李晓军)
Biomechanics and Biomaterials Laboratory
Department of Applied Mechanics
School of Aerospace Engineering
Beijing Institute of Technology(北京理工大学-宇航学院)
Beijing 100081, P. R. China
Email: [email protected]

in.temp_partial010_2D (3.26 KB)

Hi,I am simulating poiseuille flow in a 200*104*1 domain (2D),fluid
particles are in 200*100*1 region and the rest region is filled with wall
particles.An external body force of 0.02 is added to each fluid particles. I
used 'compute temp/partial' to exclude the velocity of X component, and use
'dpd pair style', NVT ensemble is for all particles.'fix_modify temp'is also
used to control the temperature. However, these command cannot control the
temperature and the velocity profile, shear stress distribution profile are
not correct.

you cannot use pair style dpd and fix nvt at the same time. you will
thermalize your system twice. for dpd the thermostat is part of the
pair style. thermostatting the same system multiple times is not
correct.

also, i don't think you can remove the x-velocity bias from the dpd
pair style. this would have to be programmed.

axel.

Hi,

As Axel said, you can't use fix_nvt with pair style DPD. DPD in itself
maintains the thermostat, while fix_nve does the time integration.

DPD is local and Galilean invariant, while NH thermostat (fix_nvt) is
not. This removes the need to correct for the shear-induced bias.

I can't find your input file, so I can't comment on more detailed issues.

For the simulation of shear flow using DPD, see:
Thomas Soddemann, Burkhard Dünweg, and Kurt Kremer, Physical Review E
68, 046702 (2003).
Jing Cao and Alexei E. Likhtman, Physical Review Letters 108 (2012).

Samuel Palato

Hi,I am simulating poiseuille flow in a 200*104*1 domain
(2D),fluid particles are in 200*100*1 region and the rest region is
filled with wall particles.An external body force of 0.02 is added
to each fluid particles. I used 'compute temp/partial' to exclude
the velocity of X component, and use 'dpd pair style', NVT ensemble
is for all particles.'fix_modify temp'is also used to control the
temperature. However, these command cannot control the temperature
and the velocity profile, shear stress distribution profile are not
correct.

you cannot use pair style dpd and fix nvt at the same time. you will
thermalize your system twice. for dpd the thermostat is part of the
pair style. thermostatting the same system multiple times is not
correct.

also, i don't think you can remove the x-velocity bias from the dpd
pair style. this would have to be programmed.

axel.

I attach my input file and wish you could point out my mistakes.

2013-07-07 ________________________________

Xiaojun Li (李晓军) Biomechanics and Biomaterials Laboratory
Department of Applied Mechanics School of Aerospace Engineering
Beijing Institute of Technology(北京理工大学-宇航学院) Beijing 100081, P. R.
China Email: [email protected]

------------------------------------------------------------------------------

This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev
_______________________________________________ lammps-users
mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/lammps-users

-- Dr. Axel Kohlmeyer [email protected] http://goo.gl/1wk0
International Centre for Theoretical Physics, Trieste. Italy.

------------------------------------------------------------------------------

This SF.net email is sponsored by Windows:

I've found the input file you linked. The mailbox was confused.

You are using atom_style angle, but there seems to be no bonds in your
system. Although this doesn't give bad results, atom_style atomic would
be more appropriate. If you plan on using bonds, make sure to set the
special_bonds parameters properly: the default values are set for
molecular dynamics and are generally inappropriate for DPD simulation.

Samuel Palato

DPD is local and Galilean invariant, while NH thermostat (fix_nvt) is
not. This removes the need to correct for the shear-induced bias.

What this means is that the thermostatting happens on a pair-wise
basis and depends on the relative velocity of the 2 atoms in
the pair. Thus even if they are flowing (together) in the x-direction,
their relative velocity is essentially the same as if the system were not flowing.
Hence the thermostatting works regardless of flow.

Steve