how to choose turbulence model?
 

Hi,

I am wondering if anyone knows references in which explains how to choose the proper turbulence model based on the model description. for example, I am interested to know when I should change from k-epsilon to LES or maybe for a problem I have LES is an overkill. I usually start with k-epsilon realizable but I do not know how I can know before hand which method is the most proper one based on the characteristics of my case. I appreciate if anyone can send me references if there is any since I couldnt find any myself. most of the links I find are describing the turbulence model details rather when to use a specific model.

tanx

the answer you are looking for depends on the flow problem and on what is the goal of the simulation...

(U)RANS -> DES-> LES -> DNS

from left to right is the formulation you can adopt and the more at left you stay, the lesser are the resolved details in the solution and greatest is the effect of the turbulence model.

You can have a look to some textbooks, for example Wilcox as well as the book of Pope

I got some tips at this forum, But I forget which thread it is.

• k-omega: region near wall
• k-epsion: dissipation is weak
• S-A: for jet
• For different turbulence models, dependencies on mesh are different.

​

Below links may be helpful to you

http://jullio.pe.kr/fluent6.1/help/html/ug/node402.htm

http://www.innovative-cfd.com/turbulence-models.html

It may be difficult to choose between models within a class (say k-epsilon versus k-omega) but the choice should never be difficult between a RANS model compared to LES. A better question is how much turbulence modelling do you want to do?

RANS is highly modelled turbulence and DNS is no turbulence modelling. DES & LES are intermediates.

The choice is similar to asking whether you want to invest in penny stocks or NASDAQ/DOW.