The grid used is the following :

The first runs done with this model did not converge at all :

and for 12 and 18 degrees :

To reduce this problem a numerical diffusion was included. This made
the simulation converge for 7 degrees. The results obtained were the following
:

A solution to have results for this problem without introducing non-physical data in the equations solved is to restart the simulation using the converged result obtained with the diffusive simulation. There was not enough time to do so.

The simulations done on the airfoil with angle of attack of 12 and 18 degrees did not converge. This is probably due to the fact that the airflow is very unstationary at high angle of attack. The simulations should be done with a unstationary model.

The residuals given for the simulation at 18 degrees are very small, but the results look unstationar.

a visualisationaof the velovity vecrors confirm the problem, but also show another problem : There is no turbulence model, and this causes huge recirculations far behind the wing.