K-Omega sst unable to predict wing stall

Project Support
1

Hi, I’ve been using the K-omega SST model to predict stall on a wing. But there seems to be a problem and the results overpredict the stall angle.
So I’ve been doing tests at different angles of attack.(15-20deg)

Specifications:

Airfoil: S1223
Windtunnel tested stall angle: 16 deg ( Re = 175,000 and above)
Mesh Algorithm: Standard
Mesh Y-Plus: Less than 1

The only reason why this over-predicting phenomenon could occur is:

“It is believed that these over-predictions were due to the fact that the turbulence levels (hence, turbulent eddy viscosities) in the boundary layers were too high, thus enhancing a momentum transfer
to the near-wall regions which helped the boundary layer to push through the adverse
pressure gradient regions more easily than otherwise. The resulting delayed separation
then caused the low pressure on the suction side to spread over more area than normal which thus increased the lift. The turbulence levels that were too high, in turn, were caused by the fully turbulent assumptions that were employed in the turbulence model.”

I’ve taken this ^^ from this paper, have a look:

https://www.researchgate.net/publication/264883437_Adjustment_of_k-w_SST_Turbulence_Model_for_an_Improved_Prediction_of_Stalls_on_Wind_Turbine_Blades

Does anyone have a way to solve this ? Any help is appreciated :slightly_smiling_face:

2

I will tag the @cfd_squad here who helped users in project like these in the past, would be surprised if the k-omega on our side is “wrong” :wink: Did you already do a mesh convergence study?

Also please share the link to your project.

Best,

Jousef

3

This is my project:

Have a look at the simulation named as:

  1. 15 AOA
  2. 16 AOA
  3. 18 AOA
  4. 20 AOA

And under each of these simulations, the runs have been named according to their inlet velocity.

I did do a convergence study but for me it was more like figuring out which simulation will converge after running it (using a particular mesh) as the ‘30000s simulation runtime rule’ is applicable for me. You won’t find the convergence study in this project as I had used another account to figure the mesh independency.

I actually also tried to reduce my Reynolds number(inlet velocity) so that during stall the turbulence could potentially be lesser but that did not turn out too good/useful.

4

Also one more thing: According to the fact that I stated in the first post; if I use another airfoil but one with lesser turbulent flow separation then I might be able to visualise stall.

So I ran another test with specifications:

Airfoil: Naca 4415
Re : 250,000
Angle of attack: 20 deg
Stall angle (wind tunnel): 21 deg

Post-processor Screenshot
Post-processor Screenshot1440×1080 104 KB

So it is kinda true that predicting stall for some airfoils is not possible like S1223, but I need a way out :slightly_smiling_face:
Any help is appreciated !