CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   STAR-CCM+ (http://www.cfd-online.com/Forums/star-ccm/)
-   -   Turbulence model (http://www.cfd-online.com/Forums/star-ccm/63521-turbulence-model.html)

filippo.aero April 11, 2009 02:31

Turbulence model
 
I'm making a simple analysis of a NACA 0015 airfoil with a gourney flap. The experiment to which I refer has U=15.4 m/s and Reynolds=2x10^5: which turbulence model is more appropriate?

A_Lyaskin April 13, 2009 04:11

You can start with Spalart-Allmaras - it is a simple model optimized for such kind of flows

nvtrieu June 13, 2010 08:15

Hello everyone,

I've been doing simulation on NACA 0012 but still haven't obtained the good results on it. I've already finished the tutorial in Star-CCM+. Can anyone help me?

Trieu

A_Lyaskin June 13, 2010 08:58

I think I was wrong about S-A model. For Gourney flap it is important to resolve boundary layer directly, so you can try some of the low-re models

nvtrieu June 13, 2010 09:08

Hello Anton Lyaskin,

Thanks you very much for your reply quickly! I'm learning about Turbulent models and Boundary layer now. It's would be very precious for any help!

Email: Trieuckgt@gmail.com
MSN: Trieu.dut@live.com
YM: Trieu_tme@yahoo.com

abdul099 July 6, 2010 12:33

Try k-epsilon low-Re, all y+ wall treatment, i made good experience with that turbulence model on airfoils.
Make sure, your near wall mesh is suitable for that simulation. You will need a lot of prism layers, maybe about 20. Check the y+ values during the simulation (after some time, when the flow field looks a little bit like it should look like) and make sure, they are below 1 for best results. Adjust near wall prism layer thickness to adjust the y+ values. It is also recommended, that the first cell outside the prism layers is not too much bigger than the last prism layer cell, it should be a smooth transition in cell size.

And make sure, the mesh is fine enough in the wake of the airfoil and gurney flap. Maybe you have to refine the trailing edge a little bit due to the higher gradients with the gurney flap.

Best regards

nvtrieu July 7, 2010 13:30

Hello Abdul,

Thanks you for your information! However, I would ask you a question a bout the Reynolds number. What values of Re are called low-Re? I'm going to simulation at Re=3mil ~ 6mil. Is this high or low Re?

About airfoil meshing in Star-CCM, I've already tried to use prism layer for boundary layer. But it is difficult to get the y plus value less than 1. Thereby, I'm using GAMBIT to generate the mesh, and GAMBIT is very easy to control the y plus value less than 1.

As I known that, the k-w SST can give a good prediction on lift coefficient as long as drag coefficient, and the stall angle than k-e. Therefore, I'm trying to use k-w SST. Unfortunately, there is no information talking about how to set the parameters for the turbulence model (any turbulence model) correctly with the individual case. Could you give me any help on it?

best regards,

Trieu.

abdul099 September 12, 2010 15:41

Hello nvtrieu,

sorry for the late response, I've been off for a while...

Whether you should use a low-Re or a high-Re turb model doesn't depend on the Reynolds number of your flow, it depends on the use of wall functions or if you want to resolve the boundary layer - with all of the common disatvantages. (More accuracy with the low-Re model, but way more computational effort)

It is possible to generate a mesh with a Y+ below 1 in star-ccm+. I'm absolutly confident, because I did it several times... It is easier if you switch to "First cell prism layer thickness" (or what the hell ever it is called) in the mesh continuum reference values and adjust the thickness of the first cell. But keep an eye on the aspect ratio of the cells, otherwise you will not get happy...

What parameters do you want to set? There is not very much to adjust, except you want to use the transition model. But first I would run the case without the transition model and have a look on the results...

Regards

PS. k-w SST is a good choice for an airfoil. But even though it promises better results near the stall angle than the k-e, you shouldn't expect too much, it's just a turbulence model. You would get better results with a LES/DES, but with an incredible computational effort.

nvtrieu September 16, 2010 00:54

Hello abdul099,

Thanks you so much! You've let me understand about low-Re and high-Re turbulence models. Yes, I've learnt how to force the mesh with y+ below 1 for the wall bc. Importantly, as you said the AR of cells should be considered as well.
The parameter that I want to set for simulation is inlet velocity for inlet boundary condition in order to obtain the expected Reynolds number. I'm not going to simulate transition model now.
I've got a big error on lift and drag coefficient at any attack angle. I'm thinking about the boundaries and the scale of model (size of wind tunnel). Could you tell me some experiences or any reference about these problems?
I think the small error is acceptable when using k-w SST or k-e turbulence models.

I would ask you more, so could I contact with you via email or MSN?
My email: trieuckgt@gmail.com
MSN: trieu.dut@live.com

Best regards,

Trieu

zelito September 16, 2010 10:13

Quote:

Originally Posted by nvtrieu (Post 275376)
Hello abdul099,

Thanks you so much! You've let me understand about low-Re and high-Re turbulence models. Yes, I've learnt how to force the mesh with y+ below 1 for the wall bc. Importantly, as you said the AR of cells should be considered as well.
The parameter that I want to set for simulation is inlet velocity for inlet boundary condition in order to obtain the expected Reynolds number. I'm not going to simulate transition model now.
I've got a big error on lift and drag coefficient at any attack angle. I'm thinking about the boundaries and the scale of model (size of wind tunnel). Could you tell me some experiences or any reference about these problems?
I think the small error is acceptable when using k-w SST or k-e turbulence models.

I would ask you more, so could I contact with you via email or MSN?
My email: trieuckgt@gmail.com
MSN: trieu.dut@live.com

Best regards,

Trieu

Hi, a very important question is: are you running 2D or full 3D sim's? What about the results you're comparing with?were these obtained in bi- or tri-dimensional windtunnel tests?

nvtrieu December 21, 2010 03:07

Hi Zelito,

I'm running 3D. And compared my simulation results with experimental data which was done by other member in our lab using tri-dimensional wind tunnel.


All times are GMT -4. The time now is 19:44.