Hello Sorry for my english. I am a french student and I'm working on stator-rotor interactions in a radial turbine.

I use Tascflow3D and I have to specify two constants for turbulence k-epsilon model solving :

- the eddy lenght scale L such as epsilon=k^(3/2)/L

- the intensity factor Tu such as k=3/2*Tu^2 u*u

I haven't found any informations about turbulence in turbines, and I don't know exactly what is the signification of these contants. I would like to know if someone can help me in their calculation, or indicate me where I can find more informations on this subject. Thanks for all.

(1). It is a good idea to learn something about the turbulence modeling before running the calculation. (2). You must read the book "Mathematical Models of Turbulence" by B.E. Launder and D.B. Spalding,1972. They are the inventors of the model. (3). You don't find turbulence in turbines, you find turbulence in "turbulence modeling". You can also search papers under the name of B.E. Launder, D.B.Spalding, or W. Rodi.(4). Next time, don't mention the name of the code you use, it makes people sick. ( like a kid using a gun, trying to figure out how to use it. ) (5). You do have the courage to seek help, so, start reading the book. ( the answer to the question is very simple, but that is not going to solve your problem in the long run.)

Ask someone who has been working with CFD or has been doing measurements in a similar turbine. Depending on how the application looks you should be able to make an intelligent guess yourself - look at what you have upstream of the turbine, how close it is etc. If you have a combustor close by you can get turbulence levels up to 15 or 20%. If you have a bit of settling length etc. the levels can go down to 4% or so. It is all very dependent on how the actual thing looks. A middle guess is 8% - that would be my first shot. You can also guess the length scale based on the size of the things you have upstream. The length scale is usually omething like 1 to 10% if the size of the geometry. If you use the standard k-epsilon model the sensitivity to these values is quite low (mainly for the reason that the k-epsilon model in it self often over-produces k and thereby reduces the effect of the free-stream k). k-omega is a bit more sensitive.

Another good thing to compute is what eddy-viscosity you chosen k-epsilon values gives. It should be a realistic value!

Dear Yann Everyhting depends on the choice of model. All the definitions would be different for different type of models. For k-epsilon model also there are various definitions used for 'epsilon'. My opinion is that you should use the model by Jones and Launder(1972). It is easy to converge and requires less rigid initial conditions. But the trouble is that it is a low-Reynolds type model, so the mesh near the wall should be very fine. But it would not be very hard to deal with, if you have an efficient code and good computer. I hope you do not care if you have to work hard to make your concepts clear. So please go through the AIAA journal issues and you would be able to find many papers on related topics.

Dear sir

It's my great pleasure to write you this email to ask you if you can help me I'm preparing a master in stability of the combustion in gas turbines if you some document in this domain please send it to me, because I need it urgently.

sincerly

Karim