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

[Georg Rempfer]

Hello Markus,

for your simulation not containing particles, the skin is irrelevant. Espresso originally being an MD code insists on one. We shifted the LB grid by half an agrid, so that one can use "composite" objects using virtual sites. With an unshifted grid you could have a situation in which the virtual sites of an object are located in an LB cell which is treated by another MPI rank (using the CPU LB with cores), which wouldn't work properly.

Greetings,

Georg

2014-01-30 Stefan Kesselheim <address@hidden>:

Hi,

On Jan 30, 2014 6:20 PM, "Wink, Markus" <address@hidden> wrote:
>
> Hi everybody,
>
> I am quite sure, that the solution has converged. Attached you will find a figure showing the maximum velocity as a function of the integration steps. For me, it looks at is has converged. What do you say.
>
> Just to be clear: if I want a channel width of 50 with a_grid=1, I put the walls at positions 1 and 51 having a box length of 52?

Absolutely. Whatever is written in Georgs thesis.

Cheers
Stefan

>
> Greetings Markus

> ________________________________________
> Von: espressomd-users-bounces+markus.wink=address@hidden [espressomd-users-bounces+markus.wink=address@hidden]&quot; im Auftrag von &quot;Stefan Kesselheim [address@hidden]
> Gesendet: Donnerstag, 30. Januar 2014 14:57
> An: address@hidden List
> Betreff: Re: [ESPResSo-users] Simulation of fluid using LB-method
>
> Hi everybody,
> please forgive me, but is it clear that the solution is converged? Markus, have you tried running for twice (or half) as many steps, and checked that you get exactly the same results?
> Because near the boundary the curvature looks perfect and closer to the center I'm not sure.
>
> Let me make one thing clear. With our staggered grid the position of the node (m,n,o) is at m+0.5, n+0.5, o+0.5. This means if you lbboundary is located at z=1, it means there is a layer of wall nodes at x=*, y=*, z=0.5. For not too high viscosities (should be in the He formula as well) no measurable slip should appear and the hydrodynamic boundary should be located mid-grid, thus at z=1.
> If you place a boundary at - say - 0.8, the hydrodynamic boundary of course is still at 1.
>
> I'm not sure, Markus, if you put these parameters correctly into you theoretical curve. We introduced the staggered grid some time after Georg's bachelor thesis to allow for using a nonzero skin with LB.
>
> Cheers
> Stefan
>
> On Jan 30, 2014, at 2:47 PM, Ulf Schiller <address@hidden> wrote:
>
> > On 01/30/2014 02:33 PM, Georg Rempfer wrote:
> >> Hey all,
> >>
> >> I remember that we fixed the issue with the missing 0.5*h*F some years
> >> ago. For a flat bounce-back wall aligned with the lattice axis, the
> >> actual hydrodynamic no-slip wall is positioned in the middle between the
> >> last fluid node and the first wall node.
> >
> > There is still a viscosity dependent slip velocity, cf. Eq. (20) in [He et al. (1997), J. Stat. Phys. 87, 115].
> >
> > Cheers,
> > Ulf
> >
> > --
> > Dr. Ulf D. Schiller                        Building 04.16, Room 3006
> > Institute of Complex Systems (ICS-2)       Phone:   +49 2461 61-6144
> > Forschungszentrum Jülich, Germany          Fax:     +49 2461 61-3180
> >