[alinik]

Hi folks,

I have two issues. I want to force a separated flow to re-attach and I know that I have two options in CFX: 1) re-attachment modifier and 2) curvature correction.I have tested many different combinations and no specific trend was observed. I am not sure whether changing these coefficients will increase the size of separation region or suppress it. Documentation is really vague in this regard.

Also is there any way to manipulate the boundary layer thickness in CFX?

Thanks,

Ali

[ghorrocks]

CFX attempts to model the real world, not a magic wand. If you want the flow to stay attached then you have to specify the flow such that it would not separate. Likewise for the boundary layer thickness.

Having said that, there are some things you can do to encourage attachment:
* Increase turbulence levels
* Higher surface roughness
* Guide vanes
* momentum source term to push the flow back onto the surface
* Grossly coarsen the mesh in the separating area

But they are all pretty brutal and will only affect marginal separations. If the separation is strong then none of these will work.

[alinik]

Glenn,

I already know that. What you said is general info and nothing new.
I am trying to have a separation followed by reattachment at certain Re number and Tu intensity. I want to have best practice CFD for my experimental data.
In my experiments I have separation and reattachment at specific condition which was reproduced in CFX.

Was hoping someone might have some real suggestion.

[ghorrocks]

You do not appear to have noticed the most important part of my previous post:

CFX attempts to model the real world, not a magic wand. If you want the flow to stay attached then you have to specify the flow such that it would not separate. Likewise for the boundary layer thickness.

It sounds like you are trying to match experimental results. In that case this question is more about CFD accuracy than artificially affecting flow separation. Have you read the FAQ on accuracy? http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

I am hoping that one day I can make a "real" suggestion

[alinik]

Thanks. I already know those deficiencies and have two publication solely on the deficiencies of numerical solvers. Thought someone might have some suggestions to overcome them.

P.S. I do not appreciate your demeaning tone, you talk as if you know everything and others know nothing. I have witnessed your tone not only in here but also other threads as well. It is obvious that not CFX nor any other numerical code can do magic. But we have to make sure that this is the absolute best results we can get from the code. Numerous options that are available in advanced turbulence modeling section only mean that we can and are allowed to tweak the model in such a way that the results are more realistic.

[ghorrocks]

I am sorry if my post appeared demeaning, it was not intentional.

The curvature correction option is intended for strongly rotating flows such as industrial cyclones. It is not intended for affecting separations so its effects here would be unpredictable. I would not recommend using this option.

I am not familiar with the details of reattachment modification options so cannot comment on that.

There is no way to modify the boundary layer besides to do something to the flow which would modify it. You can use a source term to give the boundary layer profile you want but can be tricky to apply in many cases.

[alinik]

Thank you Glenn. I also went too far and for that I apologize.
Adding source term is a good suggestion. I will try that

Thanks,

Ali