[nskelly]

Hi,

I am performing some LES of BFS flow and I am computing the primary recirculation zone length located behind the step.

I am comparing my data with experimental data of Armaly, with difficulty. According to Armaly at Re=2000, the RZL (x/S - length of zone divided by step height) should be roughly 13.5, whereas I am getting quite large value of 20. For greater Re this also increases (linearly) whereas shows decrease.

Would anyone know why this would happen? Would wall boundary conditions be the problem? I am using Smagorinsky SGS models, with a mesh of y+ < 1, and no wall functions.

Any help would be appreciated

[FMDenaro]

Originally Posted by nskelly

Hi,

I am performing some LES of BFS flow and I am computing the primary recirculation zone length located behind the step.

I am comparing my data with experimental data of Armaly, with difficulty. According to Armaly at Re=2000, the RZL (x/S - length of zone divided by step height) should be roughly 13.5, whereas I am getting quite large value of 20. For greater Re this also increases (linearly) whereas shows decrease.

Would anyone know why this would happen? Would wall boundary conditions be the problem? I am using Smagorinsky SGS models, with a mesh of y+ < 1, and no wall functions.

Any help would be appreciated

Without more details about your BC.s it is difficult to help you. However, how do you compute the reattachment lenght, after performing a long time averaging? How do you set the inflow profile? The static Smagorinsky model can produce too dissipation, have you tried the dynamic model?
Have you checked the statistics ?

You can read these articles


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


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

[nskelly]

Hi Filippo,

Yes I should have put my BCs. I am using OpenFOAM.

U:
The inflow velocity is a mapped flow condition.
Neumann BC on outlet
No slip on the walls
Cyclic on spanwise faces

p:
Neumann inlet
Dirichlet outlet
Neumann walls
Cyclic on spanwise faces

nut:
inlet and outlet - calculated 0
Neuman walls
Cyclic on spanwise faces

I compute the RZL by time averaging the simulations and looking at wall shear stress along the wall.

I have read about Smagorinsky being too dissipative and not predicting transition well, so that may well be the case. I will consider dynamic Smag and see what results I get.

[FMDenaro]

Originally Posted by nskelly

Hi Filippo,

Yes I should have put my BCs. I am using OpenFOAM.

U:
The inflow velocity is a mapped flow condition.
Neumann BC on outlet
No slip on the walls
Cyclic on spanwise faces

p:
Neumann inlet
Dirichlet outlet
Neumann walls
Cyclic on spanwise faces

nut:
inlet and outlet - calculated 0
Neuman walls
Cyclic on spanwise faces

I compute the RZL by time averaging the simulations and looking at wall shear stress along the wall.

I have read about Smagorinsky being too dissipative and not predicting transition well, so that may well be the case. I will consider dynamic Smag and see what results I get.

Not only the effect of the Smagorinsky model but also the spanwise extension of the domain (of course the experimental apparatus is confined) and, mainly, the inflow profile affect the results. If you see the papers I linked you could find more explanations.

[nskelly]

Thanks, will have a read of the papers.