Particle diffusion in icoLagrangianFoam
 

Hello Foamers,

I have had some experience deciphering the icoLagrangianFoam and I had some questions.

1. If we are injecting parcels into a turbulent flow and they travel into/through a boundary layer, while in the laminar region do they "feel" diffusional forces from brownian motion?

2. Imagine sitting a particle next to a no-slip wall in a laminar region or a flow of zero velocity. Do these particles move at all? No they don't, but how can one include this in the simulation?

I have seen a tutorial on icoLagrangianFoam (the old version of the solver) that explains how the solver works. Basically the diffusion of the particle is governed my the self diffusivity of the fluid that the particles are released. I guess the big question I'm asking is if the parcel motion includes diffusion and if not, where should I start to add that?

Thanks in advance.

Dan

Hi Dan,

to 1.) Well it depends on the scales. How big are your particles compared to the scales of brownian motion and what are your velocities in the laminar region compared to the brownian velocities.

Have a look at the Gosman and Ioanides Paper "Aspects of computer simulation of liquid-fueled combustors" there they describe a method to model particle dispersion
in turbulent flows using information from the k-equation. But that model is most commonly used for (U)-RANS, so no resolved boundary layer.

What are you planning to do LES or DNS ??

regards gregor

I have no requirements of the particle size, since I am using the particles as more of a method to query the flow. I would like to determine the turbulent Schmidt number from the flow inasmuch the same way that PIV or LDV is used to obtain an estimate of the turbulent Schmidt number. Basically for PIV:

http://hausheer.osola.com/latex2png/...0/0/result.png

I will look at the velocity scales in my system. In openfoam, the kinematicCloud parcel does have a stochastic dispersion submodel for turbulent flows to take into account the influence of turbulent fluctuations. Currently, Im using a steady state RANS solution, running lagrangian particles separately in the domain (one way coupled), and then extracting some information. the ultimate goal is to determine the turbulent diffusivity without making an assumption of Sc_t=0.7 as most books,codes,documentation suggest. I will look at the paper you suggested.

For the brownian motion in laminar flow, I am thinking that the stochastic dispersion method used for turbulent fluctuations can be emulated for brownian motion if the particles are small enough. Thoughts?

Dan