[F. Bhuiyan]

Hi,

I am doing open channel flow simulation (hydraulically rough boundary) with obstructions on the channel bed. Seperation of flow and reverse circulatiion zones were observed in laboratory. I have tested several turbulence models in CFX. With a sufficiently fine mesh k-e models results such flow seperation. However, with k-omega based models (SST, BSL EARSM) and RSM models dont show any seperation.

Any comment/experience on that with CFX would be appreciated.

Regards

Faruk

[ghorrocks]

Hi,

Well, you should have a detailed look at the qualities o the various turbulence models and of the flow itself and try to work out why the difference. k-e is known to fail in rotating flows but I don't think that's an issue here. SST would be my default option for turbulence so I can't explain why it did not capture the separation - are you sure you have all the accuracy issues sorted out, eg mesh, timestep, convergence? The RSM models in my experience tend to show earlier separation than the others so this result is puzzling.

If the separation is transient with large vorticies like a bluff body then you may have to consider DES/SAS/LES approaches.

Glenn Horrocks

[pratikmehta]

Hi,

I think there is seriously something wrong. What is the Re number of your system . Did you check your y plus values . Kindly check your .ccl file for boundary step up . I am sure you would get some hints from it .

Best of luck

Pratik

[flyingd]

Hi:
1\there will be boundry layers in your calculation by mesh size.
2\You should choose good turbulent model which can calculate fluctuating pressure.
I suggest you can choose LES or SAS.