Learning Forum

[tlcong]

Hi, I read lots of papers on the topic of DNS and LES simulation. Most of the simulations are carried out for the Couette flow between two parallel plates with constant pressure gradient along the flow direction. The size of the computational domain is often determined by several times of half of the channel height between two plates (&delta, such 2πδ*πδ*2δ. However, the absolute value of the δ is not given.  Can anyone give some instruction on how to determine the value of δ.

[raul.raghav]

Usually the non-dimensional friction Reynolds number is provided from which you can evaluate half of the distance between the plates (delta).

Re_tau = (u_tau * delta) / kin-visc

u_tau = sqrt (wall-shear/density)

 

[paguado]

Hi tlcong,

If δ is the channel height why should it be given? The important thing is to keep non-dimensional variables constant so that if you increase the channel height you have to increase the domain size. Of course your integral turbulent structures will be larger as you increase δ.

[tlcong]

Different  δ with same u_tau and and Re_tau will lead to different Reynolds number based on the bulk velocity(Re_bulk). We need to determine the Re_bulk with given  δ. However, for most cases, the  δ is not given. Is there anyway to determine it? Or there is conventional values for the δ in DNS simulation?

[tlcong]

Usually the non-dimensional friction Reynolds number is provided from which you can evaluate half of the distance between the plates (delta).

Re_tau = (u_tau * delta) / kin-visc

u_tau = sqrt (wall-shear/density)

 

However, the wall-shear stress is unknown in the simulation. (only Re_tau is given in the publications)