[pda]

Hey all,

I am trying to validate the Naca-0012 test case with settings: M = 0.16 and Re = 3Mil. I use simpleFoam in combination with the standard k-epsilon model. My average yplus is ~30 and with upper and lower boundaries 19 - 40, hence quite good I would say. Inlet conditions are Uinf = 55m/s, k and epsilon based on 0.05% turbulence intensity at the leading edge of the airfoil.

I use the liftDrag tool as mentioned on the forum to find the drag (Cd) and lift (Cl) coefficient. I performed calculations for 0 and 5 degrees. The Cl value is comparable to literature, the Cd value is over estimated by a factor 1.5-2!

I know that using this approach, the flow is considered to be turbulent right from the leading edge, which is incorrect since in reality there will be a transition from laminar to turbulent. I am wondering, though, if this can account for the discrepancy in the drag coefficient? If someone could comment on this, it would be really appreciated!

Another issue is the transition. As fas as I know, this cannot be modeled with the k-epsilon model. In the current implementation, is there a way of specifying a transition location on the airfoil? Personally, I think this is nonsense for the k-epsilon model since the flow is all ready turbulent right from the start, so how can it go from laminar to turbulent?

Any comments on this matter are greatly appreciated!

Greetings Pieterjan

[gregorv]

The transitions are indeed something that can not be captured by using the turbulence models alone. Using the wall function approach I would therefore expect that the results should be compared to rough surface experiments and not smooth airfoils.

That said, when I was trying a similar thing not too long ago, it was suggested that one should use a higher order scheme especially for the divergence of phi-v term. You might wish to switch this on only after the problem has converged with a more stable scheme, such as upwind. Using higher order schemes reduced the pressure drag calculation by a few 10%.

I am not exactly sure to what accuracy the theory behind the models used in simpleFoam allows one to predict drag around streamlined bodies, where the skin friction drag is usually quite important. If anybody has more insight I would appreciate the answer greatly as well.