Module 11 ended on a promise: a wing makes lift per metre of span, the Kutta condition decides how much circulation the sharp trailing edge allows, and the starting vortex of 11.6 Vorticity Dynamics & Circulation balances it. For the light aircraft (chord m, m/s, ) that bound circulation was m/s, lifting N over an m span. But 11.7 Potential Flow & Stream Functions‘s lifting cylinder was a streamline for every : theory alone could not say how much lift there is. A real wing does say. Here one formula turns that cylinder into wing shapes, the shape picks its circulation, and lift, the lift curve and the skin pressure follow, up to where the ideal answer fails: stall.
Tools: 11.7 Potential Flow & Stream Functions (complex potential, ), 11.6 Vorticity Dynamics & Circulation (Kelvin, the starting vortex), 11.8 Boundary Layers (separation), 11.3 Momentum: Pressure & the Euler Equations (Bernoulli) and 2.9 Line Integrals, Circulation & Flux (circulation as a line integral).
Transcript
- Module 11 left a promise. A wing makes lift, rho V gamma per metre of span, and the starting vortex it sheds is the price. Our light aircraft needed a circulation of 18.75, lifting 12 633 newtons over an 11 metre span. But a cylinder is a streamline for every circulation. So what sets gamma? The answer starts with a map.
- Solve the flow past a circle once, then move the circle with a conformal map. Locally it only scales and rotates, so streamlines stay streamlines and circulation is carried along. Joukowski's map is z equals zeta plus b squared over zeta. Shift the circle left by epsilon, 0.10, and the front swells into a teardrop. Lift it by kappa, 0.04, and the plate bends into a cambered airfoil, beta 2.0826 degrees. At zeta equals b the map squares the distance, doubling the angle: 180 degrees become 360, a cusp.
- With zero circulation at 4 degrees, air from the lower surface must whip round the sharp edge: 0.45 micrometres from the cusp the speed is 4768 metres per second, and the rear stagnation point sits 5.03 millimetres ahead of the edge. Real air will not do that. The boundary layer separates and sheds the starting vortex, leaving circulation on the wing, until the flow leaves smoothly. Then the speed is a finite 45.17.
- Put the stagnation point on the trailing edge and the circle gives gamma equals 4 pi U a sine of alpha plus beta. At 4 degrees that is 27.254 square metres per second, a lift coefficient of 0.7268. Watch the circulation grow from zero: the streamlines bend, and the rear stagnation point slides back until it sits exactly on the cusp.
- Now close the aircraft's story: chord 1.5 metres, 50 metres per second, lift coefficient 0.5. A flat plate needs 4.5643 degrees. Our cambered design needs only 2.0979: camber saves two degrees, thickness the rest. Either way the circulation is 18.75, because lift coefficient, speed and chord fix it. Lift per metre is 1.225 times 50 times 18.75, which is 1148.4375 newtons, and over 11 metres, 12 633 newtons.
- The lift curve is a sine. For the flat plate the slope is exactly 1.000000 times 2 pi; the thick design's is 1.090897 times 2 pi, because thickness raises the ideal slope. Camber slides the whole curve up: lift is zero not at zero degrees but at alpha minus beta, minus 2.0826 degrees.
- Lift is the area between the pressure curves. Plot minus Cp on the upper surface. The suction peak deepens far faster than the lift, and slides onto the nose: minus 0.599 at zero degrees, minus 1.505 at 4, minus 3.885 at 8, minus 7.863 at 12, and minus 11.846 at 15. The stagnation point slides under the nose.
- Behind the peak the air must slow again. At 15 degrees the pressure climbs from minus 11.8 to plus 0.25 within 1.5 metres: an adverse gradient. The boundary layer, slowed by no-slip, eventually separates, the Kutta condition fails, circulation falls, and the wing stalls. Typical sections stall near 15 degrees with lift coefficients 1.4 to 1.6, from memory, while the ideal curve keeps rising to 2.0148.
- Honest limits. Joukowski profiles are not real airfoils: the cusp is too thin to build. This is two-dimensional, while real wings have tip vortices. It is inviscid, so no drag and no stall. And it is incompressible, fine at Mach 0.15. Next, 12.2: general shapes have no neat map, so we replace the airfoil by a sheet of vortices, demand a streamline at every panel plus Kutta, and solve a linear system.
Airfoil vocabulary
An airfoil is a wing’s cross-section: sharp trailing edge at the back, rounded leading edge in front, the chord ( m) joining them. The camber line runs midway between the surfaces; its greatest height is the camber, the greatest gap the thickness , both as fractions of . The angle of attack is between chord and oncoming air; lift per span is perpendicular to that air, and .
Conformal maps: potential flow’s superpower
In 11.7 every formula in one complex variable was a potential flow, with . Now chain two formulas. Know the flow past a circle in a “circle plane” , and take a formula that moves every point somewhere else. Then , written through , is still a formula in one complex variable: still a potential flow, now in the plane.
What does the map do to a picture? Near a point , Taylor gives , and multiplying by the complex number scales by and rotates by (11.7). So locally the map only scales and rotates, equally in every direction: tiny squares stay tiny squares, right angles stay right. That is conformal. Each streamline goes to a curve with the same , so the body stays a solid wall. Three more facts:
- If far away, the free stream is untouched: speed at angle in both planes.
- Circulation is the jump of once round the body, and is just carried along by the map, so the circulation is the same in both planes. So is the lift, , since by Blasius (11.7) it depends only on the far field.
- Velocities follow the chain rule: . Where the map is not conformal, and the division warns of trouble.
Strategy: solve the circle once, by hand, then map it to the shape you want.
The Joukowski map
The map is
with a constant (often written ; here is the chord). Far away . Its derivative vanishes at the two critical points . You met it in 11.7’s callout: the circle goes to , a flat plate of chord , with landing on its two edges.
Now move the circle but keep it passing through . Put its centre at
The radius is the distance from to , and seen from the centre the point sits at angle : .
- Shift left (): the circle swells past , so the front fattens into a rounded nose while the back still touches : a symmetric teardrop. At the sim measures at , and its chord is (Problem 1).
- Shift up (): the plate bends into an arc. With the top of the circle, , maps to (Problem 2), so the camber is .
- Both: a cambered airfoil. Our design, and , has , camber at mid-chord and ; scaled to a m chord, m and m.
Why the tail is a cusp. Exactly, . Near the critical point the map squares the distance from , and squaring doubles angles. The circle passes as a smooth curve, an angle of ; doubled, : upper and lower surfaces meet tangent to each other, a cusp. Everywhere else the map is conformal and the circle’s smoothness survives, which is why the nose is round.
Flow past the tilted circle
In the circle plane, a stream at angle past a circle of radius centred at , with clockwise circulation (positive means lift up, as in 11.7):
It is 11.7’s cylinder with the stream turned by (the factor rotates the picture) and moved to . Differentiate:
On the circle, . Substitute and multiply by :
using . The result is purely imaginary, so the velocity is purely tangential: the circle is a streamline, for every . The surface speed is , zero at two stagnation points.
The Kutta condition
On the airfoil the speed is , and at the trailing edge, , the denominator is zero. Unless the circle-plane velocity vanishes there too, the airfoil speed is infinite: the air whips round the cusp from the lower surface to the upper. Near the edge, with , while the distance from the cusp is , so the speed grows like (Problem 3).
In the sim, untick the Kutta condition with at : the surface speed micrometres from the cusp reads m/s, and the rear stagnation point sits mm ahead of the edge on the upper surface, so the lower-surface air must turn the cusp. Real air will not do that. Past the corner the speed would have to fall from huge back to normal, an enormously adverse pressure gradient, so the boundary layer (11.8) separates right at the edge. What leaves is vorticity, rolled up: the starting vortex of 11.6, counterclockwise. By Kelvin the loop around wing plus starting vortex keeps zero circulation, so each bit shed leaves an equal clockwise circulation bound to the wing, which pushes the rear stagnation point back toward the edge. Shedding stops once the stagnation point sits on the trailing edge and the flow leaves smoothly. That end state is the
Kutta condition: the flow leaves a sharp trailing edge smoothly, with finite velocity.
Not a new law: the steady outcome of viscosity, valid while the boundary layer stays attached.
The algebra. The circle-plane velocity must vanish at the point that maps to the cusp, :
For our design at , m/s, as the sim prints; with Kutta ticked the speed micrometres from the cusp is m/s, finite (at a cusp it need not be zero; the flat plate’s is ).
Both planes draw the same levels, so each streamline on the left is carried to one on the right.
Lift and the lift curve
Kutta–Joukowski (11.7), , survives the map. Divide by :
Thin-airfoil form. For small and , and , so
Flat plate. With these are exact (, , ): , and for small , , a slope of per radian ( per degree). At the two differ by half a percent ( against ).
Zero-lift angle. at ( for the design: click zero-lift alpha). For a pure arc ; the design’s camber gives rad , close. Thin airfoil theory (12.2) finds again.
The slope, numerically. At the sim’s central difference reads for the flat plate and for the design. Exactly, there, with : thickness raises the ideal slope by about . The exercise verifies both.
Worked example: the light aircraft, closed
Same wing: m, m/s, kg/m, .
- Flat plate (11.7’s preview): , .
- Our cambered Joukowski airfoil: , so and . Camber () saves two degrees, thickness () the rest.
- Circulation: m/s, 11.6’s number: , and fix whatever the shape.
- Lift: N/m; over m, as a 2D strip, N, the N of 11.6. The story closes.
The lab also sums the surface pressure: N/m for the design, drag . For the flat plate the same sum gives N/m and a spurious drag of N/m, because a sharp leading edge concentrates infinite suction at one point (leading-edge suction), which no finite sum catches; a round nose spreads it out. At the sim also prints the approximations: , while dropping the thickness factor, , is low.
The pressure on the skin
Bernoulli (irrotational, so between any two points) gives , with the surface speed from the map. Plotted as (suction up), the upper surface lies above the lower; the area between them is the force coefficient normal to the chord, essentially at small angles. The lab’s sweep for the design:
| peak suction | at | ||
|---|---|---|---|
grows steadily, but the suction peak grows much faster and moves onto the nose. Why: the stagnation point slides under the nose as rises, and the air from it must race round the tight leading-edge curve to the upper surface. Behind the peak the air has to slow down again, pressure climbing back to near free-stream at the trailing edge. At the sim’s trailing-edge speed is m/s, so there: the upper surface climbs from to within m. That climb is the adverse pressure gradient of 11.8 Boundary Layers.
Ideal and real lift curves: stall
Potential flow’s keeps climbing along a sine, at . Real wings do not, and the reason is the table above read through 11.8: the upper boundary layer must climb that pressure hill with air already slowed by no-slip. A laminar layer tolerates very little adverse gradient (Falkner–Skan), a turbulent one more (11.9), but past some angle it too stops, reverses and separates. Once the flow leaves the surface before the trailing edge, the Kutta condition no longer holds, the circulation falls, lift drops and drag jumps: stall. For typical sections at a few million Reynolds number, is around to , reached near (from memory); the sim shades that region, an indication, not a computation.
Below stall, viscosity still nibbles: the layer thickens toward the trailing edge, so the air “sees” a slightly less cambered shape, and measured slopes are about to times (from memory), below the Joukowski . Thickness raises the slope, viscosity lowers it, and survives as a rule of thumb.
Computing it
Graded: the flat plate’s m/s, the Kutta zero, the cambered airfoil’s at , and the slopes and times (tolerances to ).
Concept checks
Problems
For the symmetric airfoil (), show that the ideal lift slope is and evaluate it for .
Show worked solution
with , so at . The chord runs from to . Dividing by gives the formula. With : , as the sim and the exercise find.
Take : centre , radius . (a) Show the image is an arc from to with height . (b) Find and for camber.
Show worked solution
(a) are on the circle (distance from the centre), mapping to . The top point is with ; since , and . So and .
(b) , so . For , : , close to rad . (In the sim: thickness , camber .)
Without the Kutta condition the circle-plane velocity at is some . Show that the airfoil speed a distance from the cusp is about , and check it against the sim’s m/s at m (design, , , where m/s).
Show worked solution
Put . Then and , so . The speed is . With m: m/s, within of the sim’s (the approximations drop higher powers of ). It is unbounded as .
Key Takeaways
- A conformal map only scales and rotates locally: it carries flows to flows and walls to walls, keeping the free stream and the circulation. Solve the circle, then map it.
- Joukowski, , maps a circle through to an airfoil: centre left for thickness, up for camber; doubles the tail’s angle into a cusp.
- Every but one sends the air round the cusp at unbounded speed; viscosity sheds the starting vortex until the flow leaves smoothly: Kutta, .
- and ; the flat plate’s ; the zero-lift angle is . Slopes: (plate), ().
- The light aircraft: at (cambered) instead of (plate); m/s either way, N/m, N over m.
- : the leading-edge suction peak deepens from at to at ; the climb behind it separates the boundary layer and stalls real wings near (from memory), while the ideal line keeps rising.
- Next, 12.2: a general airfoil, with a finite trailing-edge angle and any shape you like, has no neat map to a circle. So we stop mapping and replace the airfoil by a sheet of vortices on its surface, demand that the surface be a streamline at every panel plus Kutta at the trailing edge, and solve the resulting linear system, your numerical-methods home turf, recovering thin airfoil theory’s and the camber’s zero-lift angle along the way.