[gfoam]

Hello
I'm simulating a wing with FLUENT with high AoA (16, 17, 18 y 19°). I'm trying to find the maximum lift coefficient, that's why I'm using these differents AoA. I'm using a CutCell mesh type, generated on Ansys Meshing. The Orth. Quality is 0.357 mimnimum, and Aspect Ratio is of about 89, maximum. The turbulence model utiliced is SST-kw with no Low Re Correction nor Curvature Correction. I've created a inflation around the wing with a First Layer Thickness (FLTH) of 0.3 mm with 6 layers. The RE=4E6. MAC=1.446m. Incompressible flow with rho=1.225 kg/m3 and mu= 1.7894e-05 kg/ms.
The solution methods utiliced are: Scheme= SIMPLE;
Spatial Discretization:
-Gradient: Least Squares Cell Based;
-Pressure: Second Order;
-Momentum: Second Oreder Upwind;
-k= Second Oreder Upwind;
-w= Second Oreder Upwind.
The y+, after the simulation is about 65 on the Leading Edge where the velocity is greater, and aroud 30 in the other parts, except on the zones of low velocity where the y+ goes down to 0.
When i refine my mesh, with a inflation in wich the FLTH= 0.8 mm with 8 layers and a finner mesh near the wing (not too much) and with a Orthogonal Quality of 0.382 and an Aspect Ratio of 138 the value of CL for 17º chages from 1.38 to 1.49, and in the first case this is the maximum lift coeff. obtainable whit this mesh while for the second mesh I I think for the flow visualizations the Cl can grow a little more. So, the questions are:
1) Are the CutCell meshes adecuates for high AoA flow simulations with separated flow?
2) Is the SST-kw model appropiate for predicting BL separation?
3) If the second question's answer is yes, wich are the requisites for the mesh to predict correctly the BL separation point?
4) Must I do a transient (non steady) simulation to predict correctly the flow pattern on the wing at high AoA?

PD: In all cases the residuals converge very well, and always go down, and the force and moment coefficient converge too.
PD2: The y+ contour doesn't change much with respect to the previous simulation

[husker]

Answers,

I am very surprised that force and moment coefficients converge. According to my experience, they have to oscillate at high angles of attack. You can double check your convergence history.

1) I am not sure if cutcell mesh type is capable of predicting separation but C-Type Hyperbolic Extrusion (by Gridgen) mesh type is capable.
2) Yes, you can predict stall with SST k-w turbulence model.
3) Y+ should be around 1 or less with proper near-wall threatment.
4) Stall phenomenea is unsteady so you will need transient runs. However you can perform some steady simulations to check if your mesh and turbulence approach works.

Regards

[gfoam]

Originally Posted by husker

Answers,

I am very surprised that force and moment coefficients converge. According to my experience, they have to oscillate at high angles of attack. You can double check your convergence history.

1) I am not sure if cutcell mesh type is capable of predicting separation but C-Type Hyperbolic Extrusion (by Gridgen) mesh type is capable.
2) Yes, you can predict stall with SST k-w turbulence model.
3) Y+ should be around 1 or less with proper near-wall threatment.
4) Stall phenomenea is unsteady so you will need transient runs. However you can perform some steady simulations to check if your mesh and turbulence approach works.

Regards

Thanks for your reply husker. For the mesh, I can not do that type of meshes because i don't have that soft (Gridgen) and also I do not have ICEM CFD. But the quality of my mesh is very well, I think, althoug the quality of the surface mesh (or the refinement) is not enough to allow a y+=1 mesh for that Reynolds number, so I should try a Patch Conforming Mesh so I can get a High quality surface mesh that may allow a more refined mesh on the surface, but I am a little limited with the memory of my machine.
Relative to the convergence of the coefficients, I think that if there is no vortex sheedding the force and moment coefficients should converge to a steady value without any oscilation. But, I know, the stalll phenomenon is a unsteady one, but may be one can do steady simulations to make an approach (what do you think?).
One thing, do you know a subsonic (M<0.3) CFD validation case of a wing at high AoA? Thanks again.
Gonzalo

[husker]

Steady simulations will give you an idea about the stall angle of attack but I think you will need transient simulations to validate or improve it.

There are many NACA reports that you can find easily for various test data.

Regards