CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Gamma-Retheta Model Issue (

Rathma May 2, 2013 05:40

Gamma-Retheta Model Issue
3 Attachment(s)
Hi there,

I am using SST with Gamma-Retheta model at default settings on RAE2822 airfoil at alpha=3.9deg, Ma=0.73 and Re=6.5e6.

I got convergence problem in my case. The RMS converges at first, but starts fluctuating at around 1e-5. The lift and drag force also fluctuate in a relatively small range.

The distribution of Cp near the airfoil is quite well compared with the experiment data (see the first attachment). However, the result of upper wall Cf is very poor. There are strange values after the shock wave near 0.6c, and the location of transition point becomes more and more backward as the timestep accumulates (see the second attachment).

I have about 200 grid point on the upper airfoil, y1=2e-5 in order to make y+ approximately equals 1. The boundary layer has 30 mesh layers with a growth rate of 1.1 (see the third attachment).

I would like to improve the result, solve the convergence problem and acquire a more precise Cf distribution. Is there any suggestions?

ghorrocks May 2, 2013 06:24

You have done a nice job on the pressure distribution, but you are correct in that your friction has a little way to go yet.

But your comments boil down to 2 FAQs:

With the main point being mesh sensitivity - have you checked it? Also, are you running this transient or steady state? It is often very difficult to converge the transition model tightly as the laminar separation bubble jiggles about, so a transient model is required if you want tight convergence - if the position of the transition bubble is important.

Rathma May 4, 2013 02:19

Thanks for your reply

I do agree with you that mesh sensitivity is the main point, but I am really not familiar with how to check it. Since the result is poor at the shock wave region, I tried to add additional grid points at 0.5c~0.6c on the upper surface of the airfoil, yet I found no significant improvement so far. On the other hand, I am running the case with steady state, and I think transient state may not be that necessary in my case.

Far May 4, 2013 05:03

Try transient solution. We are also working on transition model (s) and found that transient simulation is necessary for transition model.

All times are GMT -4. The time now is 17:01.