CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Random walk models (http://www.cfd-online.com/Forums/main/1158-random-walk-models.html)

Hesham El-Batsh August 12, 1999 10:15

Random walk models
 
I am simulating particle dispersion due to turbulence in the boundary layer using random walk models available in FLUENT. Part of the boundary layer is laminar flow and in this part, I would like to set the turbulent kinetic energy to zero to ignore the effect of turbulent fluctuating velocity in the particle movement. Could someone tell me whether this assumption is correct or not. Do you know some materials concerning the value of turbulent kinetic energy in the laminar boundary layer and its distribution as a function of wall units Y+.

Joern Beilke August 12, 1999 10:50

Re: Random walk models
 
Rüdiger Schwarze (schwarze@iff.tu-freiberg.de) did a lot of work in his field.

Particle tracking in the boundary layer is much more complicated than one expets is at first. So just try to get some answer from him.

http://www.iff.tu-freiberg.de/staff/R_Schwarze.html

John C. Chien August 12, 1999 11:12

Re: Random walk models
 
(1). There is no laminar flow in a turbulent boundary layer. (2). The region next to the wall is called "viscous sub-layer". (3). Most CFD papers on turbulence modeling should have TKE profile plotted,especially in the near wall region.

Sung-Eun Kim August 12, 1999 11:22

Re: Random walk models
 
Hi,

Part of any turbulent boundary layer, i.e., the so-called "viscous sublayer", is always effectively laminar due to near-wall effects, mostly damping. In fact, the wall effects are felt even beyond the viscous sublayer and this "less" turbulent region is called "buffer layer". The question is whether you want to resolve or skip the viscosity-effected near-wall region. Your idea is practical and I understand what's on your mind, considering your situation. But unfortunately, we can't arbitrarily set turbulent kinetic energy to zero there, primarily because near-wall turbulence feels the wall effects gradually, and because we can hardly say we know a priori how to demarcate the viscosity-affected region, especially if the flow in question is complicated.

If your flow is such that you ought to resolve the near-wall, instead of skipping it (i.e., relying upon wall function), I would rather suggest that you try the two-layer model in FLUENT 5. The two-layer near-wall model allows you to resolve the viscosity-affected near-wall region all the way down to walls. FLUENT itself, not you, will figure out where the viscosity effects are dominant over turbulence and adjust turbulence (turbulent viscosity, turbulence length scale, etc.) accordingly.

All you have to do is to throw in lots of mesh so that you can resolve the near-wall region and activate the two-layer model in the "Define"/"Model"/'Viscous" panel. But if you want to know what FLUENT internally does, please take a look at the on-line user's guide. Please find the "help" in the main menu bar inside FLUENT and click on "Turbulence Models", "Near-Wall Treatments for Turbulent Wall-bounded Flows", and "Two-Layer Zonal Model" in the order shown here.


All times are GMT -4. The time now is 18:36.