Re: [ESPResSo-users] Inability to replicate a specific type of fluid flow.

[Ulf Schiller]

Hi,

this is possible if you simulate at the given Reynolds numbers. I
suspect you used a smaller Reynolds number. Note that you need to keep
the Mach number small when increasing the Reynolds number (as a rule of
thumb, the velocity in lattice units should stay smaller than 0.01). You
don't need to change the density since LB recovers incompressible
hydrodynamics. You could reduce the viscosity, but most likely you will
have to reduce the resolution and timestep as well. High Reynolds number
simulations with LB can become quite computationally expensive.

Kind regards,
Ulf

On 03/14/2015 11:17 AM, Harmanjit Singh wrote:
> Dear All,
> I can not replicate a fluid flow where in after a cylindrical obstacle,
> secondary vortices form (something similar to
> http://fluidsengineering.asmedigitalcollection.asme.org/data/Journals/JFEGA4/926053/091102_1_f5.jpeg
> ). The Stream lines (visualized in paraview) all flow unbroken from
> source to sink, i.e. there is no vortex after obstacle and i am inclined
> to believe there should be (I only get the fluid flow of the first kind
> as shown here
> http://media-1.web.britannica.com/eb-media/22/2522-004-E94C5246.jpg but
> i don't get fluid flow of second kind). I have played around with the
> density, viscosity and velocity values but to no avail.
> I am curious is this type of behavior possible in Lb-fluid Simulations
> in Espresso3.3.0? If it is, please guide me how can I simulate it.
> Thanks and Regards
> Harmanjit
> Script (ran on version 3.3.0):
> 
> setmd time_step 0.1; setmd skin 0.5
> set boxX 30; set boxY 30; set boxZ 120
> setmd box_l $boxX $boxY $boxZ
> lbfluid gpu agrid 1.0 dens 1.0 visc 8 tau 0.1 friction 0.8
> source boundaries.tcl
> lbboundary rhomboid velocity 0 0 0.05 corner 1 [expr {7*$boxY/8}] 1 a
> [expr $boxX-1] 0 0 b 0 [expr $boxY/8] 0 c 0 0 1 direction 1
> set steps 200; set counter 0
> while { $counter<50} {
>     set cycle [expr $counter*$steps]
>     puts "cycle $cycle"
>     lbfluid print vtk velocity "output/fluid$cycle.vtk"
>     integrate $steps
>     incr counter
> }

-- 
Ulf D. Schiller
Centre for Computational Science
University College London
20 Gordon Street
London WC1H 0AJ
United Kingdom