Three numbers out of a camber line

Thin-aerofoil theory takes a camber line — a whole function — and returns two numbers. This collection has an essay about where that lift starts, and it establishes the machinery: the camber line is replaced by a vortex sheet, the sheet’s strength is found from the condition that the flow be tangent to the line, and the lift and moment follow.

What it does not say is how much of the function survives.

The series, and where it stops being read #

Write the chordwise station as $x = \tfrac{c}{2}(1 - \cos\theta)$ and expand the camber line’s slope as a cosine series in $\theta$. Then the theory’s two answers are

$$C_l = 2\pi(\alpha - A_0) + \pi A_1, \qquad C_{m,c/4} = \tfrac{\pi}{4}(A_2 - A_1),$$

and nothing else in the series appears anywhere.

Three coefficients. $A_0$ is the mean slope, which acts exactly like an incidence and is why a cambered section has a zero-lift angle at all. $A_1$ is in both formulas. $A_2$ is in the moment only. Everything from $A_3$ upwards changes the shape of the aerofoil, changes the pressure distribution over it, changes where it separates and at what incidence it stalls — and changes neither the lift nor the moment, at any angle of attack, by any amount whatever.

That is a strong statement, so it is worth building the counter-example rather than asserting it.

The series itself deserves a word, because the angular coordinate is doing something specific. The substitution stretches the leading edge and compresses the trailing edge, so a Fourier series in $\theta$ resolves the region where the load is largest and varies fastest with far fewer terms than a series in $x$ would. That is not a convenience: the load has an inverse-square-root singularity at the leading edge, and the $\theta$ coordinate is the one in which that singularity becomes the single term $(1 + \cos\theta)/\sin\theta$ rather than an infinite tail.

It also means the coefficients are not the intuitive ones. $A_3$ is not “the third bump along the chord”; it is the third harmonic of the slope in the stretched coordinate, which in physical terms is a wiggle concentrated towards the leading edge. A designer thinking in chordwise thirds is not thinking in these coefficients, which is part of why the invisibility is not obvious from a drawing.

Two aerofoils, one lift, one moment #

Take a four-digit mean line with two per cent camber at four tenths of the chord and add a fourth harmonic to its slope. The camber line that results is eight tenths of a per cent of the chord away from the original — which is forty per cent of the section’s own camber, a difference no designer would call small and no manufacturing tolerance would permit.

There is one curve on that plot. The second is drawn over the first and the largest difference anywhere on it is four parts in 10¹⁶, which is double precision rather than aerodynamics. The moment about the quarter chord behaves the same way.

Their loads differ by thirty-seven per cent over the middle of the chord. That is not a small difference either, and it is the quantity that decides the things a lift coefficient does not: where the adverse gradient is, how much pressure recovery the aft section has to manage, how much uphill the boundary layer can take before it lets go, and therefore what the maximum lift is and how the stall begins.

So the two sections are the same aerofoil to a balance and different aerofoils to everything else.

The comparison is deliberately made away from the leading edge. The load there is infinite in this theory for any section at any incidence other than the ideal one, so a percentage of the peak is a percentage of a number that depends on where the sampling stopped — a fact that showed up while this figure was being made, when the two loads read as 0.85 per cent apart because the comparison was being normalised by a singularity. Over the middle eighty per cent of the chord, where the load is finite and where a pressure tapping would be, the difference is thirty-seven per cent.

The leading edge is not being avoided because it is inconvenient. It is genuinely outside the theory’s reach — the singular term is the price of replacing a rounded nose by a sheet, and this collection has an essay on the finite force that comes out of that infinite speed — so a comparison anchored on it would be comparing two idealisations rather than two aerofoils.

It is not one coincidence #

The third harmonic, the fourth, the fifth, the sixth and the eighth, each added at one amplitude. Each moves the camber line by between four tenths and one and a half per cent of the chord. Each moves the lift and the moment by a few parts in 10¹⁵.

And the boundary is sharp rather than gradual. The second harmonic moves the moment by 0.047 — a substantial pitching moment, the difference between a conventional section and a reflexed one — and the lift by four parts in 10¹⁶. Above it, nothing; below it, everything.

Why it happens #

The reason is not deep and is worth stating, because it makes the result predictable rather than surprising.

The load on a thin aerofoil is a sum: a leading-edge term proportional to $(\alpha - A_0)(1 + \cos\theta)/\sin\theta$, and $\sin n\theta$ for each $A_n$. The lift is the integral of the load along the chord, and $\int_0^\pi \sin n\theta\,\sin\theta\,d\theta$ vanishes for every $n$ except one. The moment about the quarter chord weights the load by an extra $\cos\theta$, and that picks out $n = 1$ and $n = 2$ and nothing else.

So the two forces are two particular moments of the load, and the higher harmonics are orthogonal to both of the weights involved. It is exactly the structure this collection has met in general form: an exact constraint reaches whatever lies in its own span and nothing else, and here the span is two-dimensional in a space of functions.

The same argument says which quantities are determined. Anything expressible as those two weighted integrals of the load is fixed by the lift and moment: the normal force, the centre of pressure, the aerodynamic centre. Anything else is not.

There is a second way to see it that some readers will find more convincing than the orthogonality argument, because it needs no integrals.

The circulation round the section is the integral of the sheet’s strength, and Kutta and Joukowski turn that into the lift — one formula that does not ask what the shape is. So the lift is a single number extracted from a distribution. There is no more information in it than in any other single number, and expecting it to constrain a function is expecting one equation to determine infinitely many unknowns.

The moment adds a second number, weighted differently along the chord, and therefore a second equation. Two equations, one function. The only surprise in the result is how cleanly the two equations pick out coefficients — a consequence of the angular coordinate making the weights orthogonal to the higher harmonics exactly rather than approximately.

And the same reasoning says what a third measurement would buy. A hinge moment, an aft-loading, a pressure at one station: each is another weighted integral of the load, each adds one equation, and each reaches whichever coefficients its weight is not orthogonal to. Recovering the whole camber line takes a measurement that is a function rather than a number.

What a balance measurement can and cannot recover #

A wind tunnel returns a lift and a moment at each incidence. Thin-aerofoil theory turns them into $A_0$, $A_1$ and $A_2$ — three numbers, recoverable exactly, with a lift curve at two incidences giving the first two and the moment giving the third.

Everything above $A_2$ is not under-determined by that measurement. It is untouched by it. No amount of accuracy in the balance reaches it, no number of repeat runs reduces the uncertainty, and averaging over a sweep of incidence adds nothing, because every incidence gives the same three numbers.

That is a different situation from an ordinary shortage of data and it calls for a different response. More of the same measurement does not help; a measurement of a different kind does, and the obvious one is the pressure distribution, which is the load itself and therefore determines the whole series.

The cases with closed forms #

Two special camber lines anchor the arithmetic, and both are worth having.

A circular arc of camber $m$ has a slope of exactly $4m\cos\theta$, so $A_1 = 4m$ and every other coefficient is exactly zero. Its zero-lift angle is exactly $-2m$ in radians. The quadrature returns those to ten figures, which is what makes the coefficients of every other line on this page believable.

A flat plate has every coefficient identically zero, a lift of $2\pi\alpha$ with no arithmetic in it, and no moment about the quarter chord. It is the fixed point the whole series is measured from, and it is the reason the quarter chord is the point everything is quoted about — the moment there is independent of incidence for every camber line, which is what an aerodynamic centre is.

Three lines a designer would actually draw #

A four-digit mean line, a circular arc and a reflexed line, all with nearly the same first coefficient. Their lifts are close and their moments are not, and the difference is entirely in $A_2$: the reflexed line’s second coefficient is what makes its pitching moment nearly zero, which is what makes it a flying-wing section.

That is the design freedom the theory hands over, and it is exactly three-dimensional. Choose the lift at a given incidence, choose the zero-lift angle, choose the pitching moment — and the shape is still free, in an infinite-dimensional way, for everything else.

Which is the useful reading of all of this. The invisibility of the higher harmonics is not a defect of the theory; it is the reason aerofoil design is possible at all. A designer who has met the force requirements still has the whole of $A_3$ upwards to spend on the pressure distribution, and that is what asking for a pressure and getting a shape is spending.

The freedom is what a section is designed in #

It is worth turning the result the right way up, because stated as a limitation it sounds like bad news and it is mostly the opposite.

Aerofoil families are built by fixing the force characteristics and then spending the rest of the shape on something else. A laminar-flow section spends it on holding a favourable pressure gradient back to sixty per cent of the chord. A high-lift section spends it on keeping the suction peak blunt enough that the boundary layer survives it. A supercritical section spends it on flattening the upper surface so the shock sits far aft and weak. All three can be built at the same lift, the same zero-lift angle and the same pitching moment — which is why they are alternatives rather than different operating points.

The loading nobody used is the same freedom exercised across the span rather than along the chord, and with the same structure: the total lift and the total induced drag are two functionals of the span loading, and everything else about the loading is available for other purposes.

So the practical form of this essay’s result is a design rule rather than a warning. Fix the three numbers first, because they are the ones a balance and a trim calculation care about; then design the load distribution, because that is what the boundary layer cares about; and expect the two steps not to interfere, which is exactly what the orthogonality above guarantees.

Where the theory’s silence becomes a real limit #

Three consequences worth separating, in increasing order of how much trouble they cause.

A lift curve does not identify a section. Two aerofoils with the same lift-curve slope, the same zero-lift angle and the same quarter-chord moment can have different pressure distributions, different suction peaks and different stall behaviour. Matching an existing wing’s forces does not reproduce its aerodynamics.

A thin-aerofoil inverse design is under-specified and has to be closed. Asking for lift and moment gives three equations for a function, so every inverse method adds a smoothness assumption, a family, or a target pressure distribution — and which one it adds decides the answer completely.

And the theory is silent about exactly what breaks the theory. Its assumptions fail when the flow separates, separation is decided by the load distribution, and the load distribution is what the higher harmonics control. So the coefficients thin-aerofoil theory cannot see are the ones that decide where thin-aerofoil theory stops applying, which is a circularity worth being conscious of rather than a paradox.

What the flap and the slot are doing to the series #

Two familiar devices are best understood as ways of moving particular coefficients, and saying so makes their behaviour follow rather than have to be remembered.

A plain flap hinges the aft part of the camber line, which adds a slope discontinuity at the hinge. A discontinuity has a slowly decaying Fourier series, so a flap deflection changes $A_0$, $A_1$ and $A_2$ together and leaves a long tail behind them. That is why a flap moves the lift and the moment at once and why the moment change is the part that sizes the tail: the deflection cannot reach one coefficient without the others.

A slot is not in this theory at all, because the theory has one sheet and a slotted section has two. What it does is change the boundary condition rather than the camber line, and the result is a circulation the single sheet could not have carried — a condition that can be bought rather than satisfied.

And the hinge position is a choice inside the freedom. Two flaps of different chord deflected to give the same lift increment give different moment increments, because they change $A_2$ by different amounts while changing $A_0$ and $A_1$ by the same. That is a design trade the three-number picture predicts directly and a lift-curve picture does not.

The general habit is the one worth taking away. Ask which coefficients a change touches, and the force consequences follow without a calculation: anything symmetric about mid-chord in $\theta$ leaves the moment alone, anything above the second harmonic leaves both forces alone, and anything with a corner in it touches everything.

Where the series is actually used in reverse #

The three-number result has a constructive use as well as a cautionary one, and it is worth ending on because it is how sections are designed rather than analysed.

An inverse design starts from a wanted pressure distribution rather than a wanted lift. That is a function, so it determines the whole series, and there is no freedom left over — which is precisely why inverse methods work and why they are the tool of choice for anything where the boundary layer matters. Specifying the load is specifying every coefficient at once.

Going the other way, a designer who has only force requirements has three equations and needs a rule to close them. Every aerofoil family is such a rule: the four-digit series is a two-parameter shape prescription, the five-digit series moves the maximum camber forward, the six-series prescribes the pressure distribution instead. Each closes the same gap differently, and none of them is more correct than another — they are choices within the freedom this page measures.

That reframes the whole result. The invisibility of the higher harmonics is not a limitation of thin-aerofoil theory; it is the statement that force requirements under-determine a section, which is what makes aerofoil design a design activity rather than a calculation. The three numbers are the brief, and everything above them is the work.

And a last note on what the three numbers are, physically. The mean slope is an incidence in disguise; the first coefficient is the camber’s contribution to the circulation; the second is the fore-and-aft asymmetry of the loading, which is what a pitching moment is. So the three are not an arbitrary truncation of a series — they are the incidence, the lift and the moment, one apiece, and the theory is saying that a camber line has exactly three degrees of freedom that a force measurement can reach because there are exactly three forces to reach them with.

And a note on why the quarter chord. It is the point about which the moment is independent of incidence for every camber line, which follows from the load’s incidence-dependent part being symmetric about it. That is why every table quotes moments there, and it is a property of the theory’s structure rather than a convention adopted for tidiness.

What is not claimed #

Thickness is a separate matter and is not in the series at all. Thin-aerofoil theory reads only the mean line, so the half of a section that carries nothing is invisible to it for a different reason: it does not appear rather than appearing and cancelling.

The forces are the theory’s forces. A real section at the same incidence has a slightly higher lift-curve slope from thickness, a slightly lower one from the boundary layer’s displacement, and a zero-lift angle a fraction of a degree from the theory’s. The statement being made is about what the theory’s own answer depends on, and the invisibility is exact within it.

The higher harmonics are not invisible to everything. They change the load, the suction peak, the adverse gradient and the drag — all of which are computable, and none of which is a lift or a moment. The claim is about two particular functionals, not about observability in general.

The two matched lines are a construction. One was made from the other by adding a harmonic chosen for the purpose, and neither is a section anybody would build. What the construction establishes is that the freedom exists and how large it is; it is not a claim that two aerofoils in a catalogue differ this way.

And a real camber line is not a truncated series. The four-digit mean line used here has a slope discontinuity at its maximum-camber point, so its coefficients decay slowly and it has significant content well above $A_2$ to begin with. The freedom demonstrated by adding a harmonic is a freedom that a real section is already using.