CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM Running, Solving & CFD (
-   -   kOmegaSST underpredicts cl compared to xfoil (

klausb March 23, 2013 14:53

kOmegaSST underpredicts cl compared to xfoil
2 Attachment(s)

I am working on a case setup for airfoil optimization (cl/cd).

The case calculates cl and cd for a S809 airfoil at Re = 1.000.000 with alpha = 0, steady state and transient, using the kOmegaSST model.

xfoil predics a cl of 0.14 but OpenFoam underpredicts the cl =0.04.

kOmegaSST should produce good results for small angles of attac.

How can I improve the results?


haakon March 23, 2013 18:17

As far as I can see, you use inlet velocity of 1 m/s, but in the forceCoeffs file, the value of freestream velocity, magUInf, is specified as 2.0 m/s.

This means that your coefficients of lift and drag is calculated to be one fourth of their correct value. If we take your example of 0.04 and multiply that by 4, we get 0.12. This is probably not that bad, considering that you most likely don't have done any mesh convergence studies or tuning of the case.

Good luck with your simulations.

klausb March 24, 2013 06:43

How to conduct a mesh convergence study?
Thank you for the feedback!

How to conduct a mesh convergence study?

I am planning to add another turbulence model to the case which requires y+<1 hence the mesh gets more important.


haakon March 24, 2013 14:12

I don't like to be rude, but that is really something you should be able to find out on your own. Online discussion fora is (generally) not a place where other do the work for you or read the books that you should have read.

Anyways, I Googled, and this was the topmost hit: I think it illustrates the point pretty good, even tough it is written for solid mechanics FEM and Ansys, the general concept is the same for CFD.

If you are interested in lift and drag, start with a coarse mesh (too coarse), and gradually refine it. As you refine it, you note or plot the lift and drag coefficients. When they do not change substantially (say 1-3% from the last mesh) one usually say that one has found a mesh-independent solution, or that the mesh has converged or similar. If the converged result is significantly different from references, one might want to look at other aspects of the simulation, for example the numerical schemes or solution algorithm.

All times are GMT -4. The time now is 00:27.