HI All,

I am trying to model Backward facing step using LES, and I am currently facing problems with the Inlet conditions, I am planning to give a fluctuating inlet based on the Random Numbers, although it is not so accurate, to start with it should be fine.

I am specifying the free stream velocity (u)(alone) during the initialization, and updating the inlet with the fluctuation velocity (which varies along Normal direction) at the begining of each time step,

the question is should I also specify the fluctuations along the Normal and spanwise direction, the 'v' and 'w' velocity at the begining of each time step too, along with the streamwise 'u'.

Can any one also suggest me the best formulation, to impose random fluctutation velocity at the Inlet,based on Random Numbers.

Thanks in advance,

Jean.

Hi Jean,

If you are trying to specify turbulent flow inlet conditions, there are some things which you might want to keep a note of:

1) Turbulence is not just random. There are spatial and temporal correlations that exist in a turbulent flow. You might want to try applying correlated (in space and time) boundary conditions if you are trying to specify turbulent Inlet Conditions.

2) Based on the complexity you want for your problem, there are many approaches to specify turbulent inlet BCs. One approach which I found very simple and effective in implementation and accuracy is Klein's approach of correlating the velocity fields at the inlet. ("A digital filter based generation of inflow data for spatially developing DNS or LES", Klein, Sadiki, Janicka).

3) There are more advanced approaches for specifying anisotropic turbulence ICs, which you can find from Smirnov's paper (2001). Glaze and Frankel have discussed turbulent ICs for a jet in unconfined flow in there 2001 AIAA paper.

4) If you are restarting your LES results from RANS, you have more information to play with and you should try and use it.