2D airfoil simulation, turbulence, boundary layer, unstructured grid
 

Dear all,

I'm doing flow simulations on an airfoil in order to resolve the aerodynamic loads on this airfoil. I use an unstructured grid since geometries might change a lot for following planned simulations. The airfoil has 1m chord, 68m/s airspeed (mach 0.2), Re=4.5*10^6, Turbulence intensity 0.5%, Turbulence Length Scale 0.0075m

As you can see on this graph, there is a big discrepancy between the simulated lift coefficient and the reference data.

http://www.student.kuleuven.be/~s0160912/CFD/C_l.jpg

I worked with 2 different grids, the first with unstructured boundary layer, resulting in a scattered Y+ chart. The second with a more stable Y+ chart, nevertheless showing the same trends:
http://www.student.kuleuven.be/~s016...lus_refA10.jpg
http://www.student.kuleuven.be/~s016...lus_refD10.jpg

As you can see, the Y+ values jump under 30 and over 300 over a short range.

The methods result in bad Cp distributions:
http://www.student.kuleuven.be/~s0160912/CFD/refA10.jpg

After integration, the lift coefficients are bad.

I'm dealing with a list of questions:
1. What is the effect of 'some points' out of Y+ range 30-300?
2. How can I keep the Y+ chart more stable? Since I need to check grid independency, I will have to refine. Refinement now will result in Y+ values further below 30.
3. Is an unstructured boundary layer problematic, or is it feasible?
4. Are the inlet turbulence entities reasonable.
5. Is the domain big enough?
6. My integration method to resolve the loads, is done based purely on pressure distribution. Mustn't speed be included since viscosity will affect drag coefficient?

A view of the pathlines:
http://www.student.kuleuven.be/~s016..._pathlines.jpg

Most sincerely,

Louis (Belgium)