[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ESPResSo-users] Lattice Boltzmann External Force

From: Ulf Schiller
Subject: Re: [ESPResSo-users] Lattice Boltzmann External Force
Date: Thu, 9 Feb 2012 22:11:55 +0100
User-agent: Mozilla/5.0 (X11; Linux i686; rv:9.0) Gecko/20111222 Thunderbird/9.0

On 02/09/2012 08:52 PM, Georg Rempfer wrote:
Hello Wolfgang,

thank you for your effort. Owen Hickey and I were actually working on
tracking down and fixing this bug yesterday. We agree with you on the
fact that it is not a simple discretisation artifact since it does not
occur in the gpu implementation.

Just a suggestion: there is an exact expression for the discretization error for the case of simple shear flows, e.g. Couette or Poiseuille. It boils down to a finite slip length in the analytic solution of the LB equation. Taking that into account, the profiles should match to within machine precision. This would make a good test case. Note that the hydrodynamic velocity has to be evaluated with a half time step leap-ahead of the force for second order accuracy.


Also as far as I understand the LB populations are accessible from the
TCL level via the "lbnode x y z print pop" command in the CPU
implementation. Unfortunately the GPU implementation does not contain
this feature.

Dominic (if you are on this mailing list): I wanted to ask you anyways
whether you could implement that for us. Is this possible or would it
take too much time?

Georg Rempfer

Am 09.02.2012 14:28, schrieb Wolfgang Riefler:
Hello everyone.

First of all thanks for all your quick answers.

I probably should have said before that the offset I was getting was in
the range of many times the maximum velocity (a maximum velocity of 0.04
had an offset of 0.5), so I don't think its because of the discrete
boundaries. Just to be sure I checked the simulation again for a very
small viscosity and I still get the offset.

At the moment I am working with the version 3.0.2.

Coming back to the original problem, it was actually the PhD thesis of
Ulf "Thermal fluctuations and boundary conditions in the lattice
Boltzmann method" that caught my eye and made me change the source code,
since it says in equation (4.65):

                                j = j' + 0.5 * g * tau

where j is the hydrodynamic momentum density, g is the volumetric force
and tau the lattice boltzmann time step.

Implementing this line, I get a much better result, with only a little
offset, which is probably due to the discrete boundary.

Thanks again for your help,


Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

reply via email to

[Prev in Thread] Current Thread [Next in Thread]