Talador, thanks for the info about ncrit, that is a parameter I have not fully understood yet. I've left it to the default value so far.
But your email also confuses me. I don't see how a wing (smooth or rough, in turbulent or steady air) will avoid the wing tip drag if it has a finite length. Maybe it's possible to avoid wing tip drag at the sea end of the wing/sail, if there is no gap between the boat and the wing/sail. But the top will get that vortex/drag. So a sail might have half the induced drag of an airplane wing with the same aspect ratio. Not sure though if the endplate will remove the drag entirely, or just decrease it.
Did you look at the picture?
The polars of the foil section shows Cd=0.006 at Cl=1 (L/D=~160). The induced drag is 0.06 at Cl=1 with AR=5. This should mean that the L/D on the real 3d-wing is about 1/10 of the foil section, at Cl=1, right?
In this paper i linked in a previous email, page 23, figure 21, you can see an induced drag of 0.2 at Cl=1.6, while the profile drag is much less. This should mean L/D is 8. It wouldn't matter much if the rig is a wing section with Cd=0.005 or a sail with 0.02, since that drag is so small compared to the induced drag.
http://www.hiswasymposium.com/assets/files/pdf/2006/Fossati@hiswasymposium-2006.pdfLow section drag will help at low lift coefficients, since there is less induced drag. My simulations show that a sail will have higher drag than a wing at low lift. So the wing is better in that case.