code_saturne User's Forum

Hello everyone,

I set a simulation for a flow over a plan plate (geometry 15*5*5) for the object of observing the turbulence in the boundary flow in using LES. The Re is 4.5e5, generally I should observe the turbulence in the boundary layer. But no matter I set the y+ as 1 (mesh 150*66*50) or 30 (150*50*50), there was always laminar flow in the result. I've tried Smargorinsky and WALE, nothing seems changed. So I'm thinking if it's the problem of mesh or parameter setting. As I'm really new to Saturne, there is one thing that I don't understand well, it's the 'time average' in the 'calculation control'. Could anyone please tell me what's the use and result of this parameter and how to generally set this part for a common LES case ? If there is a simple tutorial for LES, it would be perfect for me. My new case is detached below. Thank you very much.

Best regards,

Wei

Hello,

I am not a specialist in LES, so I am not 100%sure, but the time averages should be usable for LES (they be be useful for any fluctuating setup). Using user subroutines, you can have a vector average for vectors, and also tensor (variance + covariance) time variances. The Doxygen user examples (in the "user parameters, C file" section) has some examples. With the GUI, you get only averages (no variances), and only component-by-component, but it is otherwise the same functionality. For each average you define, you choose the matching variable, and the start time step.

Back to the main point: your calculation is laminar. Did you use LES-specific inputs ? Check the Doxygen documentation (the one installed with the code, not the one on this website, which might not have been updated recently due to a missing script) for LES turbulence inlet conditions. If you do not use a synthetic eddy method or equivalent (or possibly a "mapped inlet", also in the Doxygen boundary condition examples), unless your domain is very long, turbulence will probably not develop very fast.

A "too coarse" mesh could also be the issue.

I don't have time to look at your case right now, so I hope this suggestion is enough, or someone else will help, but otherwise, I can try to check in a few days.

Regards,

Yvan

Regards,

Yvan

Thank you Yvan. You are right, it's the problem of the mesh and LES turbulence inlet which I didn't set. Now I have refined the mesh and added the synthetic eddy method as inlet for the boundary, it works. Thanks so much !

Regards,

Wei

Yvan Fournier wrote:Hello,

I am not a specialist in LES, so I am not 100%sure, but the time averages should be usable for LES (they be be useful for any fluctuating setup). Using user subroutines, you can have a vector average for vectors, and also tensor (variance + covariance) time variances. The Doxygen user examples (in the "user parameters, C file" section) has some examples. With the GUI, you get only averages (no variances), and only component-by-component, but it is otherwise the same functionality. For each average you define, you choose the matching variable, and the start time step.

Back to the main point: your calculation is laminar. Did you use LES-specific inputs ? Check the Doxygen documentation (the one installed with the code, not the one on this website, which might not have been updated recently due to a missing script) for LES turbulence inlet conditions. If you do not use a synthetic eddy method or equivalent (or possibly a "mapped inlet", also in the Doxygen boundary condition examples), unless your domain is very long, turbulence will probably not develop very fast.

A "too coarse" mesh could also be the issue.

I don't have time to look at your case right now, so I hope this suggestion is enough, or someone else will help, but otherwise, I can try to check in a few days.

Regards,

Yvan

Regards,

Yvan