Aerodynamics · Demo
A real vortex-panel solver on a wing you can reshape — or draw yourself.
Forces are per 1 m chord × 1 m span at sea level. Drag is an engineering estimate — the inviscid solver itself predicts zero. Tap any ⓘ row to learn what it means.
The lift curve, swept live
Measured slope for this section: 6.85 per radian — thin-aerofoil theory predicts 2π ≈ 6.28, slightly more for thick sections. Past the shaded line a real wing stalls; this inviscid solver keeps climbing, and that gap is the whole story of stall.
In plain terms
A wing section in a wind tunnel, solved live: the flow around the shape is recomputed dozens of times a second as you play. Tilt it, thicken it, curve it — or draw your own — and watch lift appear as the air over the top speeds up and its pressure drops. Everything on screen, from the streamlines to the numbers, comes from that live solution; nothing is pre-baked.
The flow is solved with a 2D linear-strength vortex panel method. The wing surface is broken into many short panels, each carrying a sheet of vorticity. We solve a linear system so that the flow is tangent to every panel and leaves the sharp trailing edge smoothly (the Kutta condition). From the resulting surface speeds we get the pressure coefficient , and integrating pressure around the section gives the lift.
This matches classical theory well: a symmetric section makes no lift at zero angle of attack, and the lift-curve slope sits near the thin-aerofoil value of per radian (a touch higher for thick sections).
Honest caveat: this is an idealized, inviscid model — there is no viscosity, no boundary layer, and therefore no stall or flow separation. Lift here keeps growing with angle of attack forever; a real wing stalls and loses lift past . The drag figure is a separate empirical estimate (skin friction + a thickness form factor), because an inviscid flow has exactly zero drag — d'Alembert's paradox. Treat the numbers as a clean potential-flow picture, not a wind-tunnel result.
Validation: the solver is pinned to the classical results — zero lift for a symmetric section at zero angle, a lift-curve slope within a few percent of per radian, and NACA 2412's zero-lift angle near . The lift-curve panel on the rail recomputes that slope live for whatever section you build, next to the shaded region where the model honestly ends.