Here is my program's analysis of the NACA 4412 at 10 degrees angle of attack:
And here is some real data, extracted from work done at Baylor University:
Qualitatively, I'm on to something. The difference in peak pressure is easily explained -- I've made no attempt yet to compensate for boundary layer effects. And here's the kicker -- my algorithm's execution time scales linearly with increasing number of points -- twice the accuracy costs only twice the computer time!
The source currently posted at https://github.com/SlowThought/SlowFlight is up to date, and sufficient to reproduce the first figure above, with an installation of Racket. I need to write some documentation, a lift & moment routine, and a boundary layer routine, before I can justify pushing a stand-alone executable, but what I've got is what I aimed for so long ago -- a method that doesn't require vortex panels, that doesn't require singularities within the solution space -- solutions where you're allowed to look anywhere within the space!
