[Tsiolkovsky]

Hello FOAMers,

I am a beginner in OpenFOAM: I’ve learned the basics through some tutorials and I’ve been recently trying to obtain some good results from a 2D airfoil case.

I’m using simpleFoam with kOmegaSST turbulence model to simulate the flow around a NACA 2415 airfoil (Reynolds number approximately 3*10^6 and 8 degrees angle of attack) and I’m currently struggling to achieve nice results in terms of drag coefficient.

I can’t really understand where the error lies, so I’m asking for some help.


MESH:
Could this be the problem?: I built it using a Matlab code from https://www.hpc.ntnu.no/display/hpc/...l+Calculations
This script generates the blockMeshdict file you can find at the end of this post.

This is how the mesh looks like in Paraview:





BOUNDARY CONDITIONS:
As in the airfoil2d tutorial, I’m using freestream BC for U both at inlet and outlet, and obviously zero velocity condition on airfoil patch.
For p, freestreampressure at inlet/outlet and zerogradient on airfoil.
For k, freestream BC at inlet/outlet and kqRWallFunction on airfoil.
For omega, freestream BC at inlet/outlet and omegaWallFunction on airfoil.
For nut, calculated BC at inlet/outlet and nutkWallFunction on airfoil.

Boundary values for k and omega were taken from http://www.cfd-online.com/Wiki/Turbu...ary_conditions, using a turbulence intensity I=0.005 and turbulent length scale 0.05 (5% of airfoil chord).

FINALLY, THE RESULTS:
After 2000 iterations, i get the following results (this were the best I could get after many little changes in terms of BC and mesh):
lift coefficient Cl = 1.0307
drag coefficient Cd = 0.01679

I have the following experimental data from the famous Abbott-Von Doenhoff database:
Cl approximately 1
Cd around 0.010

As you can see, lift coefficient is good while drag coefficient is really high.

These values were the best I could get after some mesh/BC tuning: usally Cd values wander between 0.018 and 0.022.

Regarding y+, I’ve run yPlusRAS after the simulation and these are the results for the last timestep:
Patch 2 named airfoil y+ : min: 14.2581 max: 162.504 average: 107.005
Is this bad/good? I can't really say as I haven't found (yet) any advice on good y+ for kOmegaSST.

Thank you for your patience and for your help.
Down here you’ll find the complete case directory with all the files (it's just 120 kB)

https://www.dropbox.com/s/sm99dt8lq8...ACA2415_KW.tar


Thanks in advance,

Marco

[pbachant]

Your BCs at the foil surface seem good for the mesh resolution, but for accurate drag prediction, you may want to get down to a y+ ~ 1. Then I believe you'd use these BCs:

k:

fixedValue;
uniform 1e-12;

nut:

nutUSpaldingWallFunction;
uniform $internalField; // or similar

[Tsiolkovsky]

Originally Posted by pbachant

Your BCs at the foil surface seem good for the mesh resolution, but for accurate drag prediction, you may want to get down to a y+ ~ 1. Then I believe you'd use these BCs:

k:

fixedValue;
uniform 1e-12;

nut:

nutUSpaldingWallFunction;
uniform $internalField; // or similar

Thank you for your reply!
I've just made a (very) quick test changing the mesh expansion rates to reach an average y+ ~ 3 and setting BCs as you suggested, but results don't seem to improve very much...

[whophil]

I don't have an answer about improving Cd values, but in the .tar file you posted, the 0/U file as well as the liftDir and dragDir in the system/forceCoeffs dictionary are pointing in the wrong directions.

The grid generated by blockMesh puts the airfoil in the x-z plane, while the velocity and lift/drag vectors are in the x-y plane.

[sharonyue]

Sorry to hijack this thread. Could you tell me how to insert your pic in your main passage. I just can attach pic. Thanks

[Tsiolkovsky]

Originally Posted by whophil

I don't have an answer about improving Cd values, but in the .tar file you posted, the 0/U file as well as the liftDir and dragDir in the system/forceCoeffs dictionary are pointing in the wrong directions.

The grid generated by blockMesh puts the airfoil in the x-z plane, while the velocity and lift/drag vectors are in the x-y plane.

Sorry, you're right. I forgot to mention that after building the mesh I rotate it using the command:

rotateMesh "(0 0 1)" "(0 1 0)"

After doing this the airfoil lies in the xy plane.

[N.Ganesh]

Hii,
Maybe, the Grid which you are using is not orthogonal to the airfoil surface. So make it orthogonal and give it a try, it may help you.