[Moreza7]

Hi,

I simulated a duct flow using Dynamic Smagorinsky model.
The simulation is completely converged.
The <w'w'> component of Reynlods stress is about half of DNS (you can see it in the attached image) while the mean streamwise velocity profile has a good agreement with DNS.

Anyone knows why?

Thank you.

[FMDenaro]

Originally Posted by Moreza7

Hi,

I simulated a duct flow using Dynamic Smagorinsky model.
The simulation is completely converged.
The <w'w'> and <v'v'> components of Reynlods stress are about half of DNS (you can see it in the attached image) while the mean streamwise velocity profile has a good agreement with DNS.

Anyone knows why?

Thank you.

How did you compute the RMS?

[Moreza7]

Originally Posted by FMDenaro

How did you compute the RMS?

I used Code Saturne for simulation, calculated <w>^2 and <w^2> and used the following equation:

( is mean wall shear stress over 4 walls)
The other Reynolds stress components are pretty good.

[FMDenaro]

I suggest to check further your RMS. The comparisons we did showed Code_Saturne overestimates the DNS peaks for the plane channel flow, see

https://www.researchgate.net/publica...mulation_codes


maybe the statistically averaged velocity is not well computed

[gnwt4a]

Give some bleeding info about your simulation and that u are comparing with.
Document yr flow before you ask a question.

what Re? what resolution? whose dns? what is the exact geometry, channel or
rec/sq duct? which is the wall-normal and wall parallel component? if duct, where
in the cross section are the profiles from? are u including an estimate of the sgs
intensity in the stats? show a comparison with all three intensities.
--