[Top][All Lists]

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

Re: [ESPResSo-devel] [ESPResSo-users] Simulation of fluid using LB-metho

From: Stefan Kesselheim
Subject: Re: [ESPResSo-devel] [ESPResSo-users] Simulation of fluid using LB-method
Date: Wed, 29 Jan 2014 10:59:47 +0100

Hi everybody,

On Jan 27, 2014, at 6:35 PM, Ulf Schiller <address@hidden> wrote:

> Hi,
> can anybody explain the rationale of the following lines to me? I'd have 
> expected an exponent of either -1 or 2 for the grid spacing, depending on 
> whether ext_force is a force or a force density. In lattice units there is no 
> difference.
>    // unit conversion: force density
>      lbfields[index].force[0] = lbpar.ext_force[0]*pow(lbpar.agrid,4)*tau*tau;
>      lbfields[index].force[1] = lbpar.ext_force[1]*pow(lbpar.agrid,4)*tau*tau;
>      lbfields[index].force[2] = lbpar.ext_force[2]*pow(lbpar.agrid,4)*tau*tau;
> #else

I'm also surprised. I however checked that part relatively carefully a couple 
of years ago. The unit of the force density is energy/length^4, or 
mass*time^2/length. It somehow looks like the two were mixed up in these lines. 

We have a test case which is in principle sensitive to that. The pressure 
tensor test case checks the diagonal elements of the pressure tensor by 
applying a force density normal to a wall (=hydrostatic pressure). In principle 
it is possible that several unit conversion problems are present which entirely 
cancel out as long as Stokes flow is concerned. This can be possible because 
the flow field is linear in this case, and a factor in the force density might 
well be compensated for by an inverse factor e.g. in the output of the flow 
field. Problems like these are quite difficult to debug, because one would in 
principle need a nonlinear flow field case for which reference data is somewhat 
difficult to get.

I'll be looking deeper into that in the next couple of days.


reply via email to

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