[EddieRaben]

Hi!

I am calibrating a new calculation for use in wing optimization, but I am having problems with high errors in drag values. I am comparing my results to those of the Summary of Low-Speed Airfoil Data, mainly for the S1223 airfoil, as it will probably be the most useful in the optimization (I am aiming for maximum lift force). I am working at Re=200 000 and assuming M=0, medium is normal ISA air.

What I did:
1. made the geometry as close to the real life wind tunnel they used
2. mesh was generated to keep y+ under 1, mesh check was ok
3. using k-omega SST for 3D yielded small CL error, CD error was comparable to 2D
4. as the optimization will be in 2D, I tried the same parameters on a 2D wing (same geometry with the wind tunnel)
5. as was expected, SST was not sufficient (flow separated too soon), I used Spallart-Allmaras instead
6. Tweaking the values and using different mesh sizes, I was able to bring CL to somewhere around 1% error after correcting the values with equations give in the publication
7. CD error stayed over 100% in some cases

Next are a couple of graph from the simulation. 2D_1 to 4 is for different mesh sizes, going roughly from 40 000 to 500 000 cells. "Real" are the values from the publication, not 100% accurate because I used the mk.1 eyeball to get them, but fairly close. (these graphs are without the correction calculations, but the results are not very different with them)



As you can see in the results, my CL is acceptable, but CD has a big error, even over 100% at some AoA. Even though these error values for my geometry sizes are probably around half a Newton at worst, I want to use the model for a flapped wing optimization and thus want to have error % under 5 if possible, under 10 if not.
Anyone has any ideas what am I doing wrong? (I am definitely using the correct reference values)
Thanks.