Saramsha

No. 041  ·  27 October 2026

The Trailing Edge Decides

A wing doesn't lift because split air "reunites" behind it — that popular explanation is a myth. What actually happens: a sharp trailing edge won't tolerate air whipping around it, so the flow is forced to leave smoothly, and that single condition fixes exactly how much the air above has to curve. Drag the angle slider and watch the flow bend, the lift arrow grow — then keep going past where a real wing would already have stalled.

right now idealised (this model) a real wing (measured)

About this piece

The myth, and what's actually true

The explanation most people learn in school goes like this: the top of a wing is more curved, so air travelling over it has farther to go, so it must speed up to "meet its partner" from underneath at the trailing edge, and the faster air on top is lower pressure — lift. It sounds tidy and it is wrong: there is no rule that the two parcels of air have to arrive together, and in reality the air on top arrives well ahead of the air that went underneath. The speed-up on top is real. The reason for it is not.

The real reason starts at the back of the wing, not the front. Picture a wing with a flow that hasn't fully set up yet, so the air underneath has to swing around the sharp trailing edge to reach the top — a real fluid can't do that; it takes infinite speed to turn an infinitely sharp corner. So the flow doesn't try. Within a fraction of a second it rearranges itself into exactly the pattern that lets it leave the trailing edge smoothly, travelling in the same direction off both the top and bottom surfaces. That requirement is called the Kutta condition, and it is the one fact this whole simulation is built on. Everything else — how fast the air moves over the top, how low the pressure gets, how much lift you get — falls out of satisfying that one condition.

Circulation is the missing piece

Satisfying the Kutta condition turns out to require adding a specific amount of swirl to the flow around the wing — not a visible whirlpool, just a mathematical ingredient called circulation, written Γ. Tilt the wing up and the swirl needed to keep the trailing-edge flow smooth grows with it; that swirl adds to the airspeed on top and subtracts from it underneath, which is exactly the asymmetry you can see in the moving dots above: hotter, faster streaks on top, cooler and slower underneath. The Kutta–Joukowski theorem then gives the lift directly from that one number — lift per unit span equals air density times airspeed times Γ — with no separate calculation needed for the pressure field at all. The arrow drawn across the wing is exactly that force, and it always points perpendicular to the oncoming air, not to the wing itself, which is why it leans back slightly at higher angles.

A cambered wing — more curved on top than below, like almost every real wing — needs no tilt at all to start this process: its own shape already forces circulation at zero angle of attack, which is why the "cambered" preset above still lifts when the slider reads 0°. A symmetric wing has no such head start; tilt it back to the angle shown in the readout as "zero-lift" and the circulation, and the lift, genuinely vanish.

Why the model never stalls

The flow drawn here is a textbook idealisation — potential flow, built with no viscosity and no turbulence, using a shape produced by a 1910 piece of complex-number mathematics called the Joukowski transform (take a circle, apply z = ζ + 1/ζ, and a plain circle turns into a wing-shaped cross-section with a sharp trailing edge for free). It is honest about lift and, remarkably, also honest that there should be no drag at all — a genuine 18th-century result called d'Alembert's paradox, which says a body in a perfect, frictionless fluid feels no resistance whatsoever. Real wings obviously do have drag, and they also do something this model structurally cannot: past roughly 15–20°, the air can no longer hug the top surface, the smooth flow breaks away into a churning wake, and lift collapses rather than keeps climbing. That is a stall, and it needs the very viscosity this model leaves out, so it is shown here only as the separately measured "real wing" line on the chart — not something the slider can trigger, however far you push it.

Flow field: exact potential flow around a Joukowski aerofoil, free-stream speed U and angle of attack α set by the slider, circulation Γ solved in closed form from the Kutta condition at the trailing edge (Γ = 4πRU·sin(α − β), where β is the camber-fixed zero-lift angle). Particles are advected in the circle plane, where the velocity field is simplest, and mapped through the same transform used to draw the wing. The "real wing" curve is a stylised composite of typical low-speed aerofoil data, included for contrast — it is not measured from any one specific wing. ?warm=1 jumps straight to a settled frame.