CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Large eddy simulation (http://www.cfd-online.com/Forums/fluent/43043-large-eddy-simulation.html)

David December 6, 2006 09:28

Large eddy simulation
 
Hey,

I am currently performing LES for turbulent channel flow at a reynolds number of 9000. I have 3 questions,

1. Should i perform a RANS calculation first until the flow has developed

2. Which is the best SGS model to use

3. At what point in the simulation should i begin data sampling for time statistics.

I have performed several simulations laready but the results are not what i was expecting as the rms velocity plots are not similar to results obtainbed from DNS calculations. Any help would be greatly appreciated

Thanks in advance

hugo December 11, 2006 13:18

Re: Large eddy simulation
 
Hi David, I'm currently performing a LES for turbulent channel flow but a Reynolds number of 200000. First I always perform a RANS calculation (this is advisable for the first calculation). Second i don't know what is the best SGS model but some peoble use an Smagosinsky model with a diferent value of the constant Cs. Third, i begin data sampling for time statistics after 1000 time steps then, Fluent user's manual recomend between 1000 and 10000 time steps with data sampling.

Are you performing a wall function with LES? because i am trying to model a full development flow with LES in a circular straight channel flow but with a very fine mesh near de wall (y+ aprox 1) without wall function and the results in the boundary layer with LES are different than teorical and RANS. Can somebody help me?. i'm using a periodic condition in a duct with 3 hidraulic diameters.

i also have problem to select the correct size mesh and the correct time step for LES.



All times are GMT -4. The time now is 05:44.