CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   NUMECA (http://www.cfd-online.com/Forums/numeca/)
-   -   Turbulence models (http://www.cfd-online.com/Forums/numeca/50954-turbulence-models.html)

Galathea May 4, 2005 21:35

Turbulence models
 
Hi, I am new to CFD. May I know under what situation should I use Baldwin-Lomax model over k-epsilon models? This is the first time I actually came across this model as most textbooks mentioned only k-epsilon or k-omega models.

Any advice will be very much appreciated.


Jonas Larsson May 5, 2005 06:41

Re: Turbulence models
 
The Baldwin-Lomax model is a very robust and simple model which can produce surpricingly good results for turbomachinery blading simulations. It is suitable to use for quick design iterations. If you want to predict stall, complex 3D flows or separations it is probably not that suitable.

Galathea May 5, 2005 21:21

Re: Turbulence models
 
I read it somewhere that the Baldwin-Lomax model is more suitable than k-epsilon in the case of transonic flow, but no explanation was given. Is this true?

Jonas Larsson May 6, 2005 07:43

Re: Turbulence models
 
The Baldwin-Lomax model is a much simpler algebraic turbulence model which does not include any transported turbulent quantities. That makes it very robust and it seldom behaves very badly.

K-epsilon on the other hand is a two-equation model which transports two turbulent properties (turbulent energy and dissipation). This makes it possible for k-epsilon models and other two-equation models to account for history effects in a way which Baldwin-Lomax can never do. Hence, k-epsilon is a more advanced model with the potential of predicting more complex phenomena.

However, k-epsilon models often produce completely unphysicial results with unrealisticly viscous regions. This often happens in regions with strong normal strain, ie in regions with strong acceleration or decelleration. In transonic flows where you have shocks the shocks can often make your standard k-epsilon model produce this kind of problematic viscous regions. To handle this you need special versions of the k-epsilon models that, for example, have some extra realizability constraint applied to them.

Hence, there is no simple answer to your question. Which model is best depends on your case (weather or not history effects are important), your own experience in identifying problems with k-epslion models and which k-epsilon model you are using.

Yan May 28, 2005 09:50

Re: Turbulence models
 
What about the Spallars-Almarlas? in case of incompressible simulation like water pumps. i found that in some cases the S-A model behaves better than the B-L, while some case not.

Jonas Larsson May 28, 2005 17:19

Re: Turbulence models
 
The Spalart-Almares model is a one-equation model that it something in between an algebraic model like the Baldwin-Lomax model and a two-equation model like the k-epsilon model. Since it includes one transported turbulent quantity it has the potential to include at least some history effects (transportation of turbulent energy). It is a more modern model than the BL model, but that is of course not a guarantee that it always produces better results.

Tha SA model is very robust and is easy to use. For attached flows it often produces good results. It is popular in aero-space applications and for quick design-iteration simulations in the turbomachinery field. The SA model rarely produces the completely unphysical results that a k-epsilon model can produce sometimes. This has made the SA model quite popular in the last 5 years.

Spalart has also developed a nice DES variant of the SA model, where the large eddies are resolved and the smaller edies are modeled using the SA model. This type of hybrid RANS/LES models have produced very good results for massively separated flows in aerospace applications - there is a very nice example of a SA DES simulation of a stalling F18 which you can probably find on the net if you google a bit.

For heat transfer applications I'd not recommend SA. It often under-predicts heat-transfer.

kali charn nayak August 12, 2005 06:33

Re: Turbulence models
 
What are the better turbulence models to predict the flow pattern and heat transfer in a stator -rotor cavity like in turbo-machinery secondary flow regions where many strong re-circulating zones are present.

Thanks, Kali

sathish kumar.a October 10, 2005 06:42

Re: Turbulence models
 
hi, can u please send me basic materials for reynolds stress modeling and its latest trend in cfd.

thank you

bye, sathish kumar

SOHAIL Muhammad Amjad October 14, 2005 04:55

Re: Turbulence models
 
respected sir i am working on a small ducted propellor, duct diameter is 180 mm and its lenght is .1 m in which the propellor is installed. this whole geometry is moving with speed 10m/sec. and rpm of motor used to rotate the propellor is 10000. now please tell me which model is better to use. baldwin_lomax , Spallars-Almarlas or k-e model. flow is nearly steady and this ducted propellor is used for uav. u r kindly requested to reply as soon as possible. sohail muhammad amjad

PROF. VIVEK. YAKKUNDI November 8, 2005 23:31

Re: Turbulence models
 
respected sir , kindly suggest the suitable model for external aerodynamics of car. thanks vivek

Bedotto June 10, 2010 15:51

Hi,

I was wondering if the Spalart-Allmaras model implemented in Fine/turbo has been implemented to use wall functions when the mesh resolution is not sufficiently fine (like in Fluent). Because in my case I have 100<y+<250 and I'm wondering if results obtained with a so high y+ have any value?

Best regards

Quentin

Jade M August 5, 2010 16:07

Bedotto, I would recommend refining your grid to determine whether your y+ value is sufficient. Essentally, demonstrate grid independence.

guuthi August 6, 2010 08:19

cfd
 
how to calculate the lift drag by using cfd?


All times are GMT -4. The time now is 09:33.