[gmx-users] couette flow

[Berk Hess]

> Date: Wed, 23 Apr 2008 06:56:32 -0700
> From: schneemilch at yahoo.com
> Subject: RE: [gmx-users] couette flow
> To: gmx-users at gromacs.org
> 
> 
> 
> 
> --- On Wed, 4/23/08, Berk Hess <gmx3 at hotmail.com> wrote:
> 
> > From: Berk Hess <gmx3 at hotmail.com>
> > Subject: RE: [gmx-users] couette flow
> > To: "Discussion list for GROMACS users" <gmx-users at gromacs.org>
> > Date: Wednesday, April 23, 2008, 10:25 PM
> > > Date: Wed, 23 Apr 2008 05:29:26 -0700
> > > From: schneemilch at yahoo.com
> > > Subject: RE: [gmx-users] couette flow
> > > To: gmx3 at hotmail.com
> > > 
> > > 
> > > 
> > > 
> > > --- On Wed, 4/23/08, Berk Hess
> > <gmx3 at hotmail.com> wrote:
> > > 
> > > > From: Berk Hess <gmx3 at hotmail.com>
> > > > Subject: RE: [gmx-users] couette flow
> > > > To: schneemilch at yahoo.com
> > > > Date: Wednesday, April 23, 2008, 7:27 PM
> > > > > Date: Wed, 23 Apr 2008 02:34:49 -0700
> > > > > From: schneemilch at yahoo.com
> > > > > To: gmx-users at gromacs.org
> > > > > Subject: [gmx-users] couette flow
> > > > > 
> > > > > Hi,
> > > > > Has anyone here used gromacs to simulate
> > couette flows
> > > > in slit pores? I can see that it would be
> > possible to use
> > > > the pull code but I would like to be able to run
> > in
> > > > parallel. Any tips before I begin modifying
> > source code
> > > > would be greatly appreciated.
> > > > > 
> > > > > Cheers, Matt
> > > > 
> > > > I don't see what this has to do with the pull
> > code. The
> > > > pull code is made for forces between
> > > > centers of mass of groups.
> > > > 
> > > > There are already some external force options in
> > the mdp
> > > > file.
> > > > 
> > > > But I don't see how you would set up an
> > unbiased
> > > > couette flow.
> > > > In reality the flow is generated by a pressure
> > difference.
> > > > In a simulation you can apply an external force
> > or impose a
> > > > velocity profile.
> > > > But both these options make presumptions about
> > the
> > > > force/flow profile.
> > > > 
> > > > Berk.
> > > > 
> > > 
> > > Couette flow is shear driven, all you need is to have
> > one pore wall moving relative to the other. I figure this
> > could be done using the pull code by assigning each wall to
> > a different group and using the pull code to increase the
> > lateral distance between them at a constant rate. But it
> > seems that the pull code does not work in parallel. Is
> > there another way to prescribe the motion of one of the
> > walls?
> > > Cheers, Matt
> > 
> > Ah, sorry, I confused it with Poiseuille.
> > 
> > The deform option does shear boundary conditions.
> > 
> > But if you want to move real walls, there is currently no
> > simple option that gives
> > a group a constant velocity. The pull code should indeed do
> > it. In the CVS version
> > the pull code also works in parallel.
> > 
> > BTW the reply address in your mail is your own address,
> > instead of gmx-users,
> > I don't you if you caused this, or if something is
> > wrong on the server.
> > 
> > Berk.
> > 
> > 
> In that case I guess it would be easiest to switch to the CVS version. Thanks for the tip.
> Cheers, Matt

I just realized that this will probably not work.
I assume you have periodic walls. In that case one can not define a center of mass.

The easiest way is probably editing the code and use the accelerate group option in the mdp file.
It is not difficult to modify the update routine(s) in src/mdlib/update.c to apply a velocity
instead of an acceleration.

Another option would be to actually apply an acceleration to one of the walls and this will
then result in a certain shear rate.

Berk.


_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE!
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20080423/027bebdd/attachment.html>