Fast means low pressure

Fluid that speeds up has its pressure fall, and fluid that slows down has it rise. That single trade runs through everything in this subject: it is why wings lift, why carburettors work, why a shower curtain is drawn inwards, and why the pressure over the top of an aeroplane wing is below the surrounding atmosphere almost everywhere.

Where the trade comes from #

It is not a separate law. It is the momentum equation, integrated along a streamline.

A parcel of fluid accelerates only if something pushes it, and the only thing available to push it — in a fluid with no viscosity and no gravity to worry about — is a difference in pressure between its front and its back. So a parcel that is speeding up must be running down a pressure gradient, from higher to lower.

Integrate that along a streamline and the result is the statement that the sum of the pressure and the kinetic term is constant:

$$p + \tfrac12 \rho U^2 = \text{constant}$$

Nothing has been assumed except that the flow is steady, inviscid and incompressible, and the hypotheses matter enormously.

Stagnation, and the highest pressure there is #

The maximum possible pressure in a flow occurs where the fluid has been brought completely to rest.

At such a stagnation point the whole kinetic term has been converted, so the pressure is the free-stream value plus $\tfrac12 \rho U^2$ exactly. That extra amount is the dynamic pressure, and it is the natural unit for everything else in the flow.

Every body in a stream has at least one stagnation point, on its nose. It is the point that gets wet first in rain, the point where an insect strikes, and the point a pitot tube is placed to measure.

The dynamic pressure is also the reason forces scale with the square of speed. Double the speed and the dynamic pressure quadruples, so every pressure difference in the flow quadruples, so every force does.

The coefficient #

Since every pressure in the flow scales with the dynamic pressure, the useful quantity is the ratio.

$$C_p = \frac{p - p_\infty}{\tfrac12 \rho U^2}$$

The pressure coefficient: how far the local pressure is from the free stream, in units of the dynamic pressure. It has no units and no dependence on speed, density or altitude, which is why every pressure distribution in aerodynamics is published this way.

Two values anchor it. At a stagnation point $C_p = 1$ exactly, since the pressure is up by precisely the dynamic pressure. In the undisturbed free stream $C_p = 0$.

And it can go negative without limit. Wherever the flow is faster than the free stream, the pressure is below the free stream and $C_p < 0$. Round the shoulders of a cylinder the flow reaches twice the free stream speed and $C_p = -3$.

Reading a pressure distribution #

Aerodynamic data is usually plotted rather than contoured, and the convention catches people out the first time.

$C_p$ is plotted with the axis inverted — negative upwards. So the upper surface of a lifting aerofoil appears as a curve above the lower surface, and the area enclosed between them is proportional to the lift.

That looks perverse until the reason is clear: with the axis that way round, the picture resembles the aerofoil, with suction above and pressure below. Every aerodynamicist reads them this way and the convention is universal.

The features to look for are three. The suction peak near the leading edge, whose height indicates how hard the upper surface is working. The recovery from that peak back towards the trailing edge, whose steepness decides whether the boundary layer will survive. And the trailing-edge value, which should be close on both surfaces if the Kutta condition is satisfied.

A section with a gentle recovery stalls gently. One with a sharp peak and a steep recovery gives more lift and stalls abruptly, which is a trade every aerofoil designer makes deliberately.

The number that keeps appearing #

The dynamic pressure $\tfrac12 \rho U^2$ turns up so persistently that it is worth naming as the subject’s natural unit.

Lift is a coefficient times dynamic pressure times area. So is drag. Pressure differences are coefficients times dynamic pressure. The load on a structure is dynamic pressure times an area times a factor.

That is why airspeed indicators actually measure dynamic pressure rather than speed, and why aircraft are placarded with limits in indicated airspeed rather than true. An aircraft at altitude is moving faster through the air than its instrument reads, and the instrument is reading the quantity the structure cares about.

The same reasoning explains why aerodynamic limits are quoted the way they are. A wing does not care how fast it is going; it cares how hard the air is pushing, and that is one number.

Suction does the work #

That negative range is where lift comes from, and it is the fact the popular account of lift gets backwards.

On a lifting wing, most of the pressure difference is not high pressure pushing up from below. It is low pressure pulling up from above. At moderate incidence the suction peak near the leading edge can reach $C_p$ of $-3$ or below, while the underside is barely above ambient.

So a wing is mostly hanging from the air above it rather than resting on the air below. That is why the upper surface is the one designers worry about, why the boundary layer on the upper surface is what stalls, and why damage to the top of a wing matters more than the same damage underneath.

What the solver computed #

Every pressure figure on this site was produced the same way: solve for the velocity field, then apply the speed–pressure relation point by point, then contour the result.

That is legitimate here because the fields are steady, inviscid, incompressible and irrotational by construction — all four hypotheses hold, and the last of them means the constant is the same on every streamline so the whole field can be compared against one reference.

The contours are marching-squares bands rather than a bitmap, which means they can be counted and the levels are exact. The band colours ramp from high pressure through the paper colour to suction, so ambient is nearly invisible and the extremes carry the eye.

For the cylinder the computed minimum is $C_p = -3$ at the shoulders, which matches the analytic value $1 - 4\sin^2\theta$ exactly at $\theta = \pi/2$. That agreement is not a coincidence to be noted but a check: the figure would be wrong if it did not hold.

Why the shower curtain moves #

The everyday cases are worth collecting, because the relation is more visible than it seems.

A shower curtain drawn inwards. The spray entrains air and sets up a flow inside the enclosure; the moving air has lower pressure than the still air outside, and the curtain moves towards the low.

Two ships passing closely are drawn together, for the same reason: the water between them is forced to move faster through the gap.

A carburettor relies on it directly, using a constriction to create a suction that draws fuel in.

Windows blown outwards in a gale, rather than inwards, because the flow over the outside of a building accelerates round it and the pressure there drops below the still air inside.

A spinning ball curving, which is the same relation applied to a flow with circulation.

In each case the moving fluid is at lower pressure, and something unconstrained moves towards it.

The misuse worth naming #

The relation is so useful that it gets applied where it does not hold, and the commonest case is worth stating plainly.

Comparing pressures at two points requires them to be connected by a streamline — or, if not, requires the flow to be irrotational so that the constant is shared. Neither holds inside a boundary layer, in a wake, across a fan, or between two flows from different reservoirs.

The paper-lifting demonstration, in which air blown across the top of a sheet supposedly lowers the pressure there, fails on exactly this: the blown air and the still air underneath were never on the same streamline and never had the same constant. The demonstration works and the explanation does not.

The lift argument survives because the outer flow round a wing genuinely is irrotational, so the extra hypothesis is satisfied. That it is almost never stated is a separate problem.

How much suction is available #

There is an upper bound on how negative $C_p$ can go, and it is worth knowing because it caps what a wing can do.

For incompressible flow there is no mathematical limit — the coefficient can be arbitrarily negative if the flow is fast enough locally. Practically there are two limits, and they come from different directions.

In a gas, the constraint is compressibility. A sufficiently negative $C_p$ implies a local speed approaching the speed of sound, and once it does, the incompressible relation used here stops applying and a shock forms. For an airliner in cruise that limit is reached long before anything else.

In a liquid, the constraint is absolute. Pressure cannot fall below the vapour pressure without the liquid boiling, and the resulting cavitation damages propellers, pumps and hydrofoils by collapsing bubbles against solid surfaces with remarkable violence.

So the same relation has quite different practical ceilings in air and water, and neither ceiling appears anywhere in the equation. Both are imposed by physics the equation left out.

Gravity, and where it was dropped #

The relation as written here has no gravity term, and the omission is worth justifying rather than leaving silent.

The full form carries $\rho g z$ alongside the pressure and kinetic terms, and for water it is usually the largest of the three — a metre of depth is worth about ten kilopascals, which dwarfs the dynamic pressure of any ordinary flow.

For air it is negligible over the scale of a body. The pressure change across a two-metre wing due to gravity alone is about twenty-five pascals, against dynamic pressures of tens of kilopascals in flight. So the term is dropped, and it is dropped because of a scale comparison rather than because gravity has been forgotten.

Where it comes back is where the vertical extent is large — atmospheric flows over mountains, thermals, weather — or where there is a free surface, which is what makes ship hydrodynamics a different subject governed by its own dimensionless number.

Where the model stops #

Four hypotheses, and all of them fail somewhere on any real body.

No pressure recovery in reality. The symmetric distribution these figures show over the rear of a cylinder is exactly what a real flow refuses to produce, and the refusal is the drag.

Compressible flow changes the relation. Above about Mach 0.3 the density term matters and the incompressible form is inadequate.

Suction has a limit in a liquid. Water cannot sustain pressures much below vapour pressure without boiling — cavitation — which caps how negative $C_p$ can usefully go on a propeller or a hydrofoil. Air has no such limit.

The trade is not an exchange of stuff #

A conceptual point worth making, because the usual phrasing invites a wrong picture.

Saying that pressure “converts into” speed suggests that some quantity is being handed from one form to another, as energy is in a pendulum. That picture is close enough to be useful and it is not what is happening locally.

What is happening is that a parcel of fluid finds itself in a region where the pressure behind it exceeds the pressure in front, so it accelerates. The pressure field is not a store being drawn down; it is a set of forces acting on parcels, and the relation between speed and pressure is the integrated consequence of those forces along a path.

The distinction matters when the hypotheses fail. In a boundary layer, a parcel is being decelerated by friction as well as by pressure, so speed and pressure no longer trade cleanly — the sum falls, which is exactly the total pressure loss that makes drag measurable.

Holding the force picture rather than the exchange picture makes those failures intelligible instead of surprising.

Where the highest and lowest pressures sit #

A last observation about the figures, since the pattern is the same on every body.

The highest pressure is always at the front stagnation point, and it is $C_p = 1$ exactly regardless of shape. Nothing can exceed it, because nothing can be slower than stopped.

The lowest pressure is wherever the flow is fastest, which is near the point of maximum thickness for a symmetric body and near the leading edge for a section at incidence. Its value depends strongly on the shape and on the angle, and it is the number that most constrains a design.

Between them the pressure falls monotonically and then rises again — and it is that rise, over the rear half, which the ideal theory completes perfectly and a real boundary layer often cannot.

So the pressure distribution contains, in its shape, both the lift and the warning about separation. Reading one distribution well says more about a section than any single coefficient does.

Who wrote it down, and when #

Daniel Bernoulli’s Hydrodynamica appeared in 1738, and the relation in its familiar algebraic form is closer to Euler’s rendering a few years later. The pressure coefficient as a dimensionless presentation is twentieth-century and comes from the same impulse that produced the Reynolds and Mach numbers: remove the units, and the result describes a shape rather than an occasion.

That step is easy to overlook and it is what makes aerodynamic data transferable. A pressure distribution published as $C_p$ against chord position applies to the model and the aircraft, at sea level and at altitude, in a tunnel and in flight.

What the contours are actually showing #

A word about the figures on this site specifically, since pressure is the quantity they most often carry.

The bands are contours of the coefficient, not of pressure, and they are computed by sampling the solved velocity field on a grid and applying the relation at every point. The levels are exact values rather than a colour ramp, which means a reader can count bands and know what each is worth.

That choice was deliberate and it has a cost: a smooth heat map looks better and hides the numbers. Bands are less pretty and can be read. Since the whole argument of this site is that a figure should support a claim rather than illustrate one, bands win.

The ramp runs from the high-pressure side through the paper colour to suction, so a region at ambient pressure is nearly invisible. That is intentional too — what matters about a pressure field is where it departs from ambient, and the eye should go there.

The ladder from here #

Nearby: the pressure coefficient plotted along a surface rather than contoured; stagnation pressure and the pitot tube; cavitation; and the compressible form of the relation.

Then across to where the relation may and may not be applied, and to the lift that the suction produces.