A slot is not a nozzle

A large aircraft on approach is carrying two or three times the lift coefficient its clean wing can make, and it is doing it with a high-lift system whose essential feature is a gap. Not a bigger flap, not a smoother one: a flap with a slot in front of it, and then often a second slot, and a slat with another slot at the front of the wing.

The explanation that goes with it is one of the most repeated in the subject. Air from the high-pressure region under the wing is ducted up through the gap, arrives on top of the flap moving fast, and is blown into the boundary layer there, re-energising it so that the flow stays attached at deflections that would otherwise separate.

Every clause of that is a statement about viscosity. So there is an unusually clean test available: compute the same configuration in a flow that has no viscosity in it whatsoever, and see how much of the lift increment survives.

Almost all of it survives. And the interesting part is where it appears.

The solver, and what it deliberately lacks #

The machinery is the panel method with the counting done twice. Each body gets its own shared vortex strength and its own Kutta row, so N panels across two elements give N tangency conditions plus two Kutta conditions against N + 2 unknowns. The count works out for any number of elements, and the interference between them is carried entirely by the off-diagonal blocks of the influence matrix — panel i on the main element feels every panel on the flap, and the other way about.

What that machinery contains is a velocity field and two walls. What it does not contain: viscosity, a boundary layer, a wake, a shear layer, a mixing region, an energy budget, separation, or any distinction whatever between fast air and slow air except the one the potential flow computes for itself.

The check that it is behaving is the limit. Take the flap forty chords away and the two circulations must return to their isolated values; the gate requires the main element’s to be within two per cent, and it is.

The element that did not move #

The main element’s circulation goes up by a factor of 3.98.

That is the whole finding and it is not what the ducting story predicts. The ducting story is about the flap: it explains why the flap can be deflected further before its own flow separates. It has nothing at all to say about the element in front, which has not moved and whose flow was never in danger.

The mechanism is the one Smith named the circulation effect. The flap sits behind and below the main element’s trailing edge, and its own circulation induces an upwash there. The main element’s Kutta condition is a statement about the flow leaving its trailing edge, and if the local flow at that edge has been tilted upwards, the circulation that satisfies the condition is larger. So the front element responds to the back one by carrying more circulation, and by Kutta–Joukowski that is more lift.

Nothing in that chain mentions a boundary layer. It is a statement about two Kutta conditions in one velocity field.

And the element that did move carries less #

The second finding is the one that makes the ducting story hard to rescue.

The flap, in the system, carries less circulation than the same flap carries alone at the same attitude. The gate requires it: assertSlot fails if the downstream element is not unloaded.

That is Smith’s dumping effect seen from the other side. The flap operates in the main element’s downwash — the wake of a lifting surface deflects the flow it sits in — so its effective incidence is lower than its geometric one, and it produces less. A designer setting a flap at thirty degrees is not getting thirty degrees’ worth of flap; a fraction of the deflection is being spent restoring the incidence the downwash took away.

Two consequences follow, and they are why the arrangement is a good one rather than a wasteful one.

The flap is easier to keep attached than it looks. Its own pressure recovery is milder than its deflection suggests, precisely because it is unloaded. That is a viscous benefit and it arrives out of an inviscid mechanism.

And the trailing edge of the main element is not at rest. Because the main element is discharging into a region where the flap has raised the velocity, its own boundary layer does not have to decelerate all the way to the free-stream value before it leaves the surface. Smith called that dumping: the upstream element dumps its layer into a moving stream rather than into stagnant air, and its pressure recovery is thereby made shallower. That, too, is a viscous benefit produced by an inviscid arrangement.

Where the energy story goes wrong, precisely #

It is worth being exact about the failure, because the ducting account is not a random error and dismissing it as one teaches nothing.

Start with what is true. There is a pressure difference across the gap — the underside of the main element is at a higher pressure than the top of the flap. Flow does go through the slot from below to above. It is moving fast when it arrives, faster than the free stream. And the flap’s boundary layer is the thing that limits how far the flap can be deflected. Four true statements.

The error is in treating the fast air as a source of energy delivered to the layer. In a steady incompressible flow, Bernoulli’s constant is the same on every streamline that came from the same free stream — the flow through the slot is not carrying any more total head than the flow over the top of the wing, because both came from the same undisturbed air. There is no reservoir of high-energy fluid anywhere near a wing. There is only a velocity field.

So the picture of a jet being aimed at a tired boundary layer has no energy to aim. What actually helps the flap’s layer is that the flap starts a new layer at its own leading edge — thin, and therefore able to survive a steeper gradient — rather than continuing the thick one that came down the main element. That is a real and viscous mechanism, it is one of Smith’s five, and it has nothing to do with the slot as a duct: it is a consequence of the layer being interrupted, which any gap would achieve.

The slot is not delivering energy; it is interrupting a boundary layer and arranging two circulations. Both of those are true and neither is the story.

The rigging, which is a real number #

Gap and overlap are the two numbers in a rigging table, and a mechanic setting a flap sets them to a thousandth of an inch. This calculation produces an optimum gap of a few per cent of chord, which is where real rigging tables put it — and the agreement should be read carefully, because the inviscid optimum and the real one are optima of different things.

The inviscid optimum is where the interference is strongest. Too close and the flap sits inside the main element’s own accelerated flow, which reduces its effective incidence further and drowns it. Too far and the elements stop knowing about each other.

The real optimum has a second constraint on top: the gap has to pass enough flow to keep the flap’s own boundary layer attached, and if the slot is too tight the layer on the flap separates whatever the potential flow says. That constraint is the one the ducting story is a garbled version of, and it is genuinely there. It is a constraint on the design, not the mechanism of the lift.

So the honest summary is a division of labour. The lift increment is inviscid. The reason the arrangement does not fall apart at those deflections is viscous. A story that puts both under one heading gets the second one roughly right and the first one entirely wrong.

The slat, which goes the other way #

The clearest evidence that this is a circulation effect rather than an energy effect is what happens when the extra element is put at the front.

A slat also has a slot. On the ducting account it should do the same kind of thing: pass fast air over the nose of the wing and re-energise the layer there.

What it actually does is take the suction peak down, by forty-four per cent, and take the main element’s circulation down with it, by twelve per cent. At a fixed incidence, deploying a slat loses lift — the system makes 1.34 where the clean section made 1.45.

That is not a defect. It is the entire point of a slat and it is the opposite of what the ducting story implies. The slat’s own circulation induces a downwash over the main element’s nose, taking the local incidence off it and flattening the pressure peak. A flattened peak means a gentler pressure recovery behind it, and a gentler recovery is one a boundary layer can survive further. So the wing can be taken to a much higher angle before it separates, and there it makes far more lift than the clean section could.

Everything a slat is worth is in the incidence it allows, and none of it is in the lift at a given incidence. The solver states that as a refusal: assertSlot fails if the slat does not reduce both the peak and the circulation.

Smith’s five effects, and which two this cannot see #

A. M. O. Smith’s 1975 paper High-lift aerodynamics is the standard account, and it lists five things a slot does. Three of them are in the arithmetic above and two are not, and the division is worth stating because it is the honest form of the whole argument.

The slat effect — an element ahead reduces the peak on the one behind. Computed above.

The circulation effect — an element behind raises the circulation on the one in front. Computed above.

The dumping effect — the upstream element discharges into a moving stream, easing its pressure recovery. Half computed: the velocity at the trailing edge is in the solve, and what it does to a boundary layer is not.

Off-the-surface pressure recovery — the upstream element’s wake decelerates in the free stream rather than against a wall, which is a far more efficient way of recovering pressure because a free shear layer can take a gradient a wall layer cannot. Not computed. There is no wake here.

Fresh boundary layer — each element starts a new layer at its own leading edge instead of continuing a thick one, and a thin layer survives a steeper gradient. Not computed. There are no layers here.

So two of the five are genuinely about viscosity, and both are about robustness rather than about the lift itself. Neither is the ducting story either: neither involves transferring energy from one stream to another. Smith is explicit about that, and the sentence is worth quoting for what it concedes and what it denies — the slot is not a nozzle, and the air coming through it is not being used as a jet.

What a designer takes from this #

Three things follow that a rigging table already knows and that the arithmetic explains.

Deflect the flap and re-trim the wing, not the flap. The lift arrives largely on the main element, so the pressure distribution that has to be checked for separation is the main element’s, not the flap’s. That is the opposite of the intuition the ducting story gives.

A slat is an angle-of-attack device and a flap is a lift device, and they should be judged by different numbers. A slat that loses lift at cruise incidence is working correctly; a flap that loses lift at any incidence is not.

And gap and overlap are worth setting precisely, because the interference falls off with distance and the optimum has curvature. A slot rigged half a per cent of chord wide of its optimum gives away a measurable fraction of the increment, which is why the tolerance in the manual is what it is.

What the model does not contain #

No boundary layer, and that is the point. Every number here is the answer a flow with no viscosity gives. The value of the calculation is entirely that the increment survives in that flow; it says nothing about what a real high-lift system achieves, because a real one is limited by separation and this one cannot separate.

The lift keeps climbing with deflection, without limit. Deflect the flap to sixty degrees and this solver reports more lift still. A real system peaks somewhere in the forties and falls away. The falling away is the whole viscous half of the subject and none of it is here.

Two elements, not four. Real systems are triple-slotted with a slat as well, and the interference is between every pair. The machinery handles any number; the essay uses two because the argument is about a mechanism rather than about a configuration.

The geometry is a scaled and rotated copy of the main section, which is not how a flap is shaped. A real flap has its own section, and the numbers here would move if it had one. What would not move is the direction of any of the effects, because all of them come from where the circulation sits rather than from the shape carrying it.

And nothing here is compressible or three-dimensional. A real slotted flap has ends, and the ends leak.

Who found it, and when #

The slotted wing was patented independently by Handley Page in Britain and Lachmann in Germany around 1918 — Lachmann had been trying to understand a stall he had crashed in — and the two ended up sharing the patent. It worked immediately and was understood slowly.

The ducting explanation belongs to that early period, and it is not a foolish guess: the flow does go through the gap, it does arrive on top of the flap, and it is faster there. The mistake is in the causal direction. The flow through the gap is fast because the two elements’ circulations have arranged a velocity field in which it is fast, and pointing at the fast air as the cause is pointing at part of the answer.

Smith’s 1975 paper is where the account above was assembled, and it is unusual in the literature for saying plainly what the mechanisms are not. It came fifty-seven years after the patent, which is a long time for a device that was in production the whole while — and the reason it took that long is that the mechanism only becomes visible once a two-element inviscid flow can be computed, which needed exactly the panel methods the previous essay is about.

The device was in service for half a century before anybody could do the arithmetic that says how it works. That ordering is not unusual in this subject and it is worth noticing when it happens.

Where the ladder goes next #

Everything so far has been about a section — a slice of a wing, with the flow in one plane. A real wing has a span, and the moment it does, the questions change: which part of it runs out of lift first, what the twist is for, and why a swept wing behaves as though it were meeting a slower wind than it is. The last of those turns out to be a theorem rather than an approximation.