Two lessons ago a parcel crossing a gentle speed bump at m/s accelerated at m/s, about g, while every fixed probe read a constant. Last lesson explained the speed-up: streamtubes narrow over a body, so the air in them goes faster, on top of a cylinder. Neither named a force, yet something pushes that parcel, and air touching only air can be pushed only by air. The push is pressure, and Newton’s law with pressure in it is the Euler equations.
They are aerodynamics without friction. They give hydrostatics, Bernoulli’s equation (the Venturi tube, your airspeed indicator), the pressure on a body, and one honest failure, d’Alembert’s paradox, the cleanest route to what lift is and what the “equal transit time” story gets wrong. Tools: the gradient (2.3 Gradient & Directional Derivatives), the material derivative (11.1 Continuum Hypothesis & Flow Kinematics), continuity and the stream function (11.2 Conservation of Mass: The Continuity Equation).
Transcript
- Two lessons ago a parcel crossing a speed bump at fifty metres per second accelerated at seven hundred sixty-four point seven metres per second squared, while every fixed probe read a constant. Continuity explained why the air speeds up, but not what pushes it. Along the centreline only pressure acts, so the pressure gradient must be minus nine hundred thirty-six point eight pascals per metre: the pressure falls toward the crest, by one thousand fifty-six point six pascals in all.
- A one centimetre cube of air has one hundred one thousand three hundred twenty-five pascals on its left face and one hundred one thousand three hundred twenty on its right. Each face feels ten point one three newtons, and the net is five times ten to the minus four. With a mass of one point two two five times ten to the minus six kilograms, it accelerates at four hundred eight point two metres per second squared: forty-one point six g. Per unit volume the force is minus grad p, and Newton's law with it is the Euler equations. With continuity, three unknowns, three equations. Pressure is a Lagrange multiplier.
- At rest, Euler says the pressure gradient balances gravity. In water the pressure grows nine thousand eight hundred ten pascals per metre: nineteen thousand six hundred twenty at two metres. A cube's top and bottom faces differ by one thousand two hundred twenty-six point two five newtons, the weight of the water it displaces, at any depth. That is buoyancy. In the atmosphere, density falls with height. At five kilometres, constant density gives forty-one thousand two hundred fifty-nine pascals, isothermal fifty-six thousand ten, and the standard atmosphere fifty-four thousand twenty, with density zero point seven three six one.
- Integrate Euler along a streamline and you get Bernoulli: pressure plus half rho q squared plus rho g z is constant. It needs steady, inviscid, incompressible flow, on one streamline. Take a Venturi tube: air at twenty metres per second enters a throat of half the area. Continuity gives forty. Bernoulli says the pressure drops by one half times one point two two five times sixteen hundred minus four hundred: seven hundred thirty-five pascals. Read backwards, that is a flowmeter. At fifty to sixty-five, it is the speed bump's one thousand fifty-six point five six pascals.
- A pitot tube points into the wind. The air stops at its mouth, so the tube fills with total pressure, while side holes read the static pressure. Solving Bernoulli, V is the square root of two times the pressure difference over rho. At fifty metres per second at sea level, that is fifteen thirty-one point two five pascals. At five kilometres the density is zero point seven three six one, so the same reading means a true airspeed of sixty-four point five zero metres per second. The dial is marked for sea level, so it shows indicated airspeed.
- Around a body, the pressure coefficient is one minus speed over U, squared. On a cylinder the surface speed is two U sine theta, so Cp is one minus four sine squared theta: plus one at the nose and tail, minus three on top and bottom. Nose and tail push equally, so there is no drag. Top and bottom match, so no lift. The upper half alone pulls three point three three three three, and the lower half cancels it: the sum is about ten to the minus sixteen. That is d'Alembert's paradox. Euler gives no drag, because Euler threw away friction.
- The up-down symmetry can be broken without viscosity. Add a clockwise swirl of circulation: the surface speed gains a constant, and Cp becomes one minus the quantity two sine theta plus G, squared. At G equal to one, Cp reaches minus eight on top, the stagnation points slide to two hundred ten and three hundred thirty degrees, and the lift is twelve point five six six, four pi G, which is rho U Gamma: the Kutta Joukowski theorem. The upper half gives eleven point six one seven, the lower half zero point nine five zero, both upward. Drag stays at round-off.
- Now the popular story: air split at the leading edge must rejoin at the trailing edge, so the longer top path is faster. Nothing enforces that. Release two parcels with RK4, half a unit above and below the dividing streamline. With no circulation both arrive together, at eight point four three five. With G equal to one, the top parcel arrives at eight point eight zero seven and the bottom at ten point seven one one: the top wins by one point nine zero four, on a path of nine point four four against eight point one one. Parcels keep no appointments.
- Euler lets air slide along a surface, but real air sticks to the wing: zero velocity at the skin. Why? That needs viscosity. Next lesson adds it to Newton's law and gets the Navier-Stokes equations, which explain drag, boundary layers and, eventually, stall.
Forces on a fluid parcel
Two kinds of force act on a parcel. Surface forces come from the neighbouring fluid; without friction that is pressure, pushing along the inward normal of each face. Body forces act on every molecule inside: gravity, per unit volume.
Pressure on a box, with numbers
A cm cube of sea-level air has Pa on its left face and Pa on its right. Each m face is pushed with about N; the net is N to the right. The cube holds kg of air, so it accelerates at m/s, g, from a pressure difference your hand could not feel. Uniform pressure, however large, moves nothing.
The general box
With sides , the -faces sit at and have area . By Taylor,
and likewise in and . Per unit volume the pressure force is
pointing from high pressure to low. For a whole body, by the divergence theorem (2.10 Green, Stokes & the Divergence Theorem) applied to each component, so a constant pressure exerts no net force and may be subtracted from everything.
Back to the speed bump
On the centreline m/s, the acceleration peaked at m/s at m. Nothing but pressure acts along , so
The air behind the parcel is at higher pressure than the air ahead, by almost a kilopascal per metre: the missing force.
The Euler equations
Mass times acceleration equals force, per unit volume, with the material derivative as the acceleration:
These are the Euler equations (published in 1757, from memory). In 2D with up and incompressible continuity:
Count. Three unknowns, , , , three equations: closed. If density varies, the gas law brings in temperature and an energy equation (compressible flow, 11.12).
Notice that and have time derivatives and has none: pressure is whatever it must be, instant by instant, to keep . The divergence of the momentum equation (using , ) gives
a Poisson equation over the whole domain: pressure is a Lagrange multiplier enforcing the constraint, like the rod tension that keeps a pendulum bob on its circle.
Hydrostatics
With the Euler equations say , that is, .
The tank
Water, kg/m, , atmospheric Pa at the surface: at depth , . At m that is Pa; at m, Pa, nearly a second atmosphere (one atmosphere is m of water).
Buoyancy is this gradient. A cube of side with its top at depth is pushed down with and up with ; the sides cancel. The net, , is the weight of displaced water, Archimedes’ principle with its mechanism: N for m, at any depth.
The atmosphere
Air’s density falls with height. With the gas law , J/(kg K),
a separable ODE. The International Standard Atmosphere values are quoted from memory: Pa, K, kg/m, , cooling K/km up to km.
- Constant density: , which hits zero at m. At km: Pa.
- Isothermal: , scale height m (the same number, since ). At km: Pa.
- ISA, : , exponent . At km: K, Pa, kg/m.
The probe gains Pa per metre; the cube’s upthrust stays N. Push the altitude past km and the constant-density atmosphere goes negative: it has run out of air.
Bernoulli’s equation
Along a streamline of a steady flow, Euler integrates by hand. Let be arc length and with . Then , whose component along is (the rest is the centripetal turning). With and , the tangential component of steady Euler at constant is
That is Bernoulli’s equation. is the dynamic pressure, and the total pressure, what the air would reach if stopped smoothly. On the speed bump ( to m/s) the crest sits Pa below upstream, and Pa/m is this curve’s slope at m.
The assumptions, and what breaks each.
- Steady. In a gust or a pitching wing, a term joins in.
- Inviscid. Friction drains total pressure, so Bernoulli fails in boundary layers and wakes; a pitot tube in a wing’s wake reads low.
- Incompressible. Above Mach (102 m/s at sea level, with sound at m/s) density changes exceed (11.2 Conservation of Mass: The Continuity Equation).
- One streamline, no machine. The constant may differ between streamlines unless the flow is irrotational (vorticity lesson, 11.6), and a propeller adds total pressure.
Worked example: a Venturi tube
Air at m/s enters a tube whose throat has half the area. Continuity gives m/s, and Bernoulli
cm of water on a manometer. Backwards it is a flowmeter: m/s. Older light aircraft drove gyro instruments from a fuselage Venturi (from memory).
Total pressure stays flat to round-off. Set the ratio to and : m/s in the throat and a Pa drop, the speed bump again. From m/s at ratio the throat exceeds Mach and the readout warns you.
Worked example: the pitot tube
Point a tube into the wind. Air stops at its mouth, so the tube fills with total pressure, while side holes feel the static :
and at m/s at sea level Pa (the Venturi panel at ).
The pressure coefficient
Around a body every streamline starts in the same free stream, so outside the boundary layer and wake one Bernoulli constant, , serves everywhere. Dropping (it only adds the body’s buoyancy) gives the pressure coefficient
at a stagnation point, at free-stream speed, negative (suction) where the air is faster.
Worked example: the cylinder
Last lesson’s stream function in polar form is , so the surface speed is at . With from the downstream axis (nose at ),
at nose and tail, at and (and mirror images), at top and bottom.
Subtract ; with surface element and outward normal the force per unit span, in units of , is
is equal at and , where flips sign: nose and tail push equally, so the drag is zero. is equal at and , where flips: the lift is zero. Yet the upper half alone feels
a pull of upward, cancelled exactly by the lower half.
At the 720-cell midpoint sum gives drag and lift near and halves of .
d’Alembert’s paradox
Inviscid flow past a cylinder exerts no drag, and a general theorem (quoted) extends this to any body in steady, irrotational, inviscid flow. d’Alembert called it a paradox in 1752 (from memory): real bodies are dragged. Euler threw away friction. A real cylinder’s boundary layer cannot climb the rear pressure rise ( from back to ), so it separates and leaves a low-pressure wake, breaking the fore-aft symmetry. Viscosity comes next (11.4), separation in 11.9.
Lift needs circulation
The up-down symmetry can be broken without viscosity. Add a clockwise swirl of circulation , the term in , which adds the constant speed on the surface. With ,
At , reaches on top, the stagnation points slide to and , and the lift is (upper half , lower half , also upward), and drag stays at round-off. Dimensionally that is , the Kutta–Joukowski theorem. In 12.1 a sharp trailing edge fixes the circulation (the Kutta condition): that is where a wing’s lift comes from.
The equal-transit-time myth
The popular story: air split at the leading edge must rejoin at the trailing edge together; the upper path is longer, so the upper air is faster, its pressure lower, and the wing rises. The conclusion is right; the reason is wrong.
Nothing enforces the meeting. Euler and continuity are the whole law of the flow, and neither ties parcels together.
The race. Release a line of parcels at and follow them with RK4. Two highlighted ones start above and below the dividing streamline (ending on the front stagnation point): one goes over, one under. At both reach at (with ), by symmetry, and there is no lift. At the upper parcel arrives at and the lower at : the top wins by . At it wins by . Smoke pulses over real airfoils do the same (from memory).
Path length is not the cause. The body is the same circle at every ; only the circulation changes, and the lift with it, from to . Shape alone does not fix the lift, so a rule reading lift off the shape’s path lengths cannot be right. With circulation the split point slides to , so the surface route over the top ( of arc) is twice the one underneath (); the racing parcels’ paths are over and under, and the longer one finishes first. A flat plate at an angle lifts with equal paths, and aerobatic aircraft fly inverted.
What produces lift is the computation above: pressure, lower on top than below, summed over the surface. Pressure and velocity come out of Euler and continuity together, with the circulation selected by the trailing edge. Equivalently, that pressure field turns the air downward and the air pushes back. No parcel keeps an appointment.
Computing it
For a smooth periodic integrand the midpoint rule over a full period converges faster than any power of the step: the solution prints drag and lift near and to round-off (graded to and ). The half circle is not periodic, the error is , about (it prints ), so is graded to .
Concept checks
Problems
(a) Find the absolute pressure at m in fresh water. (b) A cube of side m has its top face at m. Find the gauge forces on its top and bottom faces and their difference.
Show worked solution
(a) Pa (the tank panel reads gauge at m).
(b) Top: N down. Bottom: N up. Net N up , at any depth (tank panel).
(a) Solve . (b) Evaluate at km with the ISA values and find . (c) A sea-level-calibrated airspeed indicator reads m/s there. Find the true airspeed.
Show worked solution
(a) integrates to ; with , .
(b) K, exponent , Pa, kg/m.
(c) Pa, so m/s (tank panel at ).
With : (a) where is when ? (b) Show the upper half feels upward when . (c) Show the lift is and the drag zero for every .
Show worked solution
(a) : .
(b) .
(c) . Over a period and , so . Each term of integrates to zero: . With , . The sim prints at .
Key Takeaways
- Pressure pushes with per unit volume: Pa across a centimetre gives g; the speed bump’s g needed Pa/m.
- Euler, , plus continuity closes for ; incompressible pressure is a Lagrange multiplier (Stam’s projection).
- Hydrostatics: , buoyancy . At km the ISA gives Pa and kg/m.
- Bernoulli (steady, inviscid, incompressible, one streamline): Venturi Pa at m/s; pitot Pa at m/s, a true m/s at km.
- reaches on a cylinder yet integrates to zero drag and lift: d’Alembert’s paradox.
- Lift needs circulation, . Equal transit time is false: with lift, the top parcel arrives earlier.
- Next, 11.4: Euler lets air slide along a surface, yet real air sticks to the wing, zero velocity at the skin. Why? Viscosity and the Navier–Stokes equations, which explain drag, boundary layers and, eventually, stall.