CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)
-   -   turbulence model? (https://www.cfd-online.com/Forums/cfx/87630-turbulence-model.html)

jai April 25, 2011 09:00

turbulence model?
 
hi,

i'm simulating centrifugal compressor, which's rotating at 100,000rpm.
which turbulence model is better for this?
can i use k-E model (with y+ is 10-20) for this? some people recommended to use sst but for sst-model is y+ < 2. it difficult to maintain this y+ on all region.
please help me to understand and choose.
thanks.

juliom April 25, 2011 09:51

you should not use K-E model, since you will have a complex fluid flow distribution, if you read more about K-E model you will find that this model does not suit for this application since you will have rotation and swirl. I suggest you to use RSM

ghorrocks April 26, 2011 22:59

Quote:

I suggest you to use RSM
This is unlikely to be any more accurate than the 2 equation models, and will definitely be a lot harder to converge.

The best default option is the SST model. Only change away from this model when you have a good reason to do so. SST works fine with y+ in the wall function region (ie >11), so SST does not demand a y+<2 mesh.

juliom April 27, 2011 09:53

I agree with you about the two model equation, they are the most efficient when the computational resource is low. However we have to take into consideration several condition when they are used.
One of the most important is when there is swirl, like is this case, due to the fact that is a turbumachine, you have to make some proves with the turbulence models, starting with the two equation until RMS. No doubt RMS is the most complex and you need a lot of computational resource, this is the worst part.
FLUENT has several information about it, and from my point of view they explain in more details the topics.
Yours sincerely
JM

stumpy April 27, 2011 10:35

If swirl is important then use SST and enable the Curvature Correction option. This should give something closer to a RSM solution in swirl regions, but retaining the benefits of the wall treatment with the SST model.

ghorrocks April 27, 2011 20:29

In my experience for many flows RSM does not give results any more accurate than a two equation model, even though many two equation models do not include curvature effects.

That is why I recommend SST as a starting point, and enable curvature correction as Stumpy suggests. I would be surprised if an RSM model proves to be more accurate than this, even though RSM models contain "more physics".

jai April 28, 2011 00:50

RE:
 
As you have suggested, tried RSM model and the convergence is poor.
So decided to go about with the SST with curvature correction option.

Thanks for you suggestions.


All times are GMT -4. The time now is 08:28.