SPECIAL FEATURE: FLIGHT

www.iop.org/journals/physed

How do wings work?

Holger Babinsky
Department of Engineering, University of Cambridge, Cambridge CB2 1PZ, UK
E-mail: hb@eng.cam.ac.uk

Abstract
The popular explanation of lift is common, quick, sounds logical and gives
the correct answer, yet also introduces misconceptions, uses a nonsensical
physical argument and misleadingly invokes Bernoulli’s equation. A simple
analysis of pressure gradients and the curvature of streamlines is presented
here to give a more correct explanation of lift.

M This article features online multimedia enhancements

The science behind aeronautics continues to

fascinate and many students are attracted to
engineering as a result of an early interest in
aircraft. The most commonly asked question is
how a wing can produce lift. Unfortunately the S
most widely used explanation of lift is wrong in a
number of key points. Not only is this confusing T
for students, but in the worst case it can lead
to a fundamental misunderstanding of some of
the most important aerodynamic principles. In
this article I will demonstrate why the popular
explanation for lift is wrong and then propose an
alternative explanation.
Figure 1. Streamlines around an aerofoil section
The popular explanation visualized with smoke.
Figure 1 shows a typical aerofoil—the cross- Blow along upper surface of paper
sectional shape of a wing—immersed in a flow
where the streamlines have been visualized with
smoke particles. At the front is the stagnation point
(S), which is the location where the oncoming flow
divides into that moving above and that moving Paper
below the wing. The argument revolves around
the observation that the distance from this point
S to the trailing edge (T) is greater along the Figure 2. Paper lifts when air is blown along its upper
upper surface than along the lower surface. If it surface.
is assumed that two neighbouring fluid particles
which ‘split’ at S should meet again at T then and thus a net upwards pressure force is generated.
this requires that the average velocity on the upper Bernoulli’s equation is often demonstrated by
surface is greater than that on the lower surface. blowing over a piece of paper held between both
Now Bernoulli’s equation is quoted, which hands as demonstrated in figure 2. As air is blown
states that larger velocities imply lower pressures along the upper surface of the sheet of paper it rises

0031-9120/03/060497+07$30.00 © 2003 IOP Publishing Ltd PHYSICS EDUCATION 38 (6) 497

H Babinsky

Mast The ‘equal time’ argument

It is often asked why fluid particles should meet
up again at the trailing edge. Or, in other words,
Sail
why should two particles on either side of the wing
take the same time to travel from S to T? There is
Figure 3. Flow along the cross section of a sail. no obvious explanation and real-life observations
prove that this is wrong. Figure 4 shows a selection
of still frames from a video recording of a smoke-
and, it is said, this is because the average velocity
wind-tunnel experiment (the video clip is available
on the upper surface is greater (caused by blowing)
in the online journal). Here, smoke particles
than on the lower surface (where the air is more
are injected simultaneously upstream of a lifting
or less at rest). According to Bernoulli’s equation
aerofoil section, generating a line of smoke as seen
this should mean that the pressure must be lower
in figure 4(a). This line of smoke moves with
above the paper, causing lift.
the flow, dividing into particles travelling above
The above explanation is extremely wide-
and below the aerofoil. By the time the smoke
spread. It can be found in many textbooks and,
to my knowledge, it is also used in the RAF’s has passed the aerofoil (figure 4(c)) the particles
instruction manuals. The problem is that, while moving along the upper surface are clearly ahead
it does contain a grain of truth, it is incorrect in a of those travelling along the lower surface. They
number of key places. do not meet up at the trailing edge. When lift
is generated, fluid particles travelling along the
upper surface reach the trailing edge before those
What’s wrong with the ‘popular’ travelling along the lower surface1 .
explanation?
The distance argument
The Bernoulli demonstration
While the aerofoil of figure 1 does indeed exhibit a
Blowing over a piece of paper does not
greater distance between S and T along the upper
demonstrate Bernoulli’s equation. While it is true
surface, this is not a necessary condition for lift
that a curved paper lifts when flow is applied
production. For example, consider a sail that is
on one side, this is not because air is moving at
nothing but a vertical wing (generating side-force
different speeds on the two sides. This can easily
to propel a yacht). Figure 3 shows the cross
be demonstrated by blowing along one side of a
section of a sail schematically and it is obvious
straight piece of paper as sketched in figure 5. In
that the distance between the stagnation point and
this case the paper does not experience a force
the trailing edge is more or less the same on both
towards the side subjected to the faster moving
sides. This becomes exactly true in the absence of
a mast—and clearly the presence of the mast is of air. The pressure on both sides of the paper is the
no consequence in the generation of lift. Thus, 1 In fact, it can be proved theoretically that if the particles on

the generation of lift does not require different the upper surface reach the trailing edge at the same time as
distances around the upper and lower surfaces. those travelling along the lower surface no lift is produced.

(a) (b) (c)

Figure 4. Smoke particles flowing along a lifting aerofoil section.
M An MPEG movie of this figure is available from stacks.iop.org/physed/38/497

498 PHYSICS EDUCATION November 2003

 How do wings work?

Blow air along one side Fluid particle

Streamline
v

l
Paper (hanging vertically)
Figure 6. Fluid particle travelling along a straight line.

Figure 5. A straight piece of paper hanging vertically shall see, a force acting in the flow direction causes
doesn’t move when air is blown along one side.
fluid particles to change their speed whereas a
force acting normal to the flow direction causes
same, despite the obvious difference in velocity. streamline curvature (by ‘particle’ we refer to a
It is false to make a connection between the flow very small but finite volume (or element) of the
on the two sides of the paper using Bernoulli’s fluid, not individual molecules).
equation.
The ‘truth’ about Bernoulli
An alternative explanation for lift Imagine a fluid particle travelling along a straight
The above argument has remained popular because line (but not at constant velocity) as shown
it is quick, sounds logical and gives the correct schematically in figure 6. Let the x-direction
answer. However, my concern about using this be in the direction of motion. If the particle is
explanation is that it introduces misconceptions in a region of varying pressure (a non-vanishing
about why aerofoil shapes generate lift, it pressure gradient in the x-direction) and if the
uses a nonsensical physical argument and it particle has a finite size l, then the front of the
often includes an erroneous ‘demonstration’ of particle will be ‘seeing’ a different pressure from
Bernoulli’s equation. the rear. More precisely, if the pressure drops in
the x-direction (dp/dx < 0) the pressure at the
rear is higher than at the front and the particle
Before we begin—some basic assumptions experiences a (positive) net force. According
The key to understanding fluid flow around an to Newton’s second law, this force causes an
object is to examine the forces acting on individual acceleration and the particle’s velocity increases
fluid particles and apply Newton’s laws of motion. as it moves along the streamline. Conversely, if
While there are many different types of forces the pressure increases in the direction of the flow,
acting on a fluid particle it is possible to neglect the particle decelerates. This means that if the
most of these, such as surface tension and gravity. pressure drops along a streamline, the velocity
In fact, for most practical flows the only relevant increases and vice versa. Bernoulli’s equation
forces are due to pressure and friction. As a first describes this mathematically (see the complete
step, we can also assume that there are no friction derivation in the appendix).
forces at work either. This is because in most However, the fact is often overlooked that
flows friction is only significant in a very small Bernoulli’s equation applies only along a stream-
region close to solid surfaces (the boundary layer). line. There is no explicit relationship between the
Elsewhere, friction forces are negligible. pressure and velocity of neighbouring streamlines.
We shall also assume the flow to be steady. Sometimes, all streamlines in a flow originate
In practice this means that we only consider from a region where there is uniform velocity
situations where the overall flowfield does not and pressure (such as a reservoir or a uniform
change very quickly with time. free-stream) and in such a case it is possible to
With these assumption we can now derive the apply Bernoulli’s equation throughout the flow.
rules governing fluid motion by considering the But in the ‘demonstration’ of Bernoulli’s equation
resultant pressure force acting on an individual shown in figure 2 the air moving along the upper
fluid particle and applying Newton’s second law, surface of the paper originates from the mouth
which states that force causes acceleration. As we of the person performing the experiment and the

November 2003 PHYSICS EDUCATION 499

H Babinsky

Poutside Concentric
Curved streamline streamlines
in a vortex Pressure drops
v
(or tornado) towards centre

Pinside

Pressure force
Poutside > Pinside
Figure 8. Pressure gradient across streamlines in a
vortex.
Figure 7. Fluid particle travelling along a curved
streamline.

streamlines can be traced right back into this

person’s lungs. There is no connection with the
‘streamlines’ underneath the paper and Bernoulli’s
equation cannot be applied to compare the pressure
in the two regions2 . In fact, the pressure in the
air blown out of the lungs is equal to that of the
surrounding air (and this is proved when blowing
over a straight sheet of paper—it doesn’t deflect
towards the moving air).

Flow along curved streamlines

Next, examine a particle moving along a curved
streamline as shown in figure 7. For simplicity we
can assume that the particle’s speed is constant3 . Figure 9. Streamlines around a lifting curved plate
Because the particle is changing direction there ‘aerofoil’.
must exist a centripetal force acting normal to
the direction of motion. This force can only be
is a tornado (or any vortex, such as that seen in
generated by pressure differences (all other forces
a bathtub). As sketched in figure 8, an idealized
are ignored), which implies that the pressure on
vortex consists of concentric circles of streamlines.
one side of the particle is greater than that on
The above relationship implies that there is a
the other. In other words, if a streamline is
curved, there must be a pressure gradient across pressure gradient across these streamlines, with
the streamline, with the pressure increasing in the the pressure dropping as we approach the core.
direction away from the centre of curvature. This explains why there are such low pressures in
This relationship (derived mathematically in the centre of vortices (and why tornados ‘suck’
the appendix) between pressure fields and flow objects into the sky). In a real, three-dimensional,
curvature is very useful for the understanding of tornado, the streamlines are not circles but spirals
fluid dynamics (although it doesn’t have a name). which originate from somewhere far away where
Together with Bernoulli’s equation, it describes the air is at rest and at atmospheric pressure.
the relationship between the pressure field and the Applying Bernoulli’s equation to each of these
flow velocity field. A good demonstration of this streamlines shows that the velocities increase the
2 One might argue that the air blown over the top surface
closer we get to the vortex core (which is what we
does eventually come to rest in the room and at that stage
observe in nature)4 .
a connecting ‘streamline’ might be drawn towards the lower
surface. However, in this case the flow is clearly affected by 4 At the very centre, the assumption of frictionless flow is no

friction—this is what brings the flow to rest—and Bernoulli’s longer justified (because streamlines in opposite directions get
equation only describes frictionless flows. very close to each other, creating a strong shear flow) and the
3 This in turn implies, according to Bernoulli’s equation, that above arguments no longer hold (the ‘eye’ of the storm has low
the pressure along the streamline is constant (dp/dx = 0). flow velocities and low pressure).

500 PHYSICS EDUCATION November 2003

 How do wings work?

(a)

Figure 10. Simulated streamlines around thin and (b)

thick aerofoils.

Lift on aerofoils
Now we can return to the original problem.
Figure 9 shows a schematic sketch of the
streamlines around the simplest lifting aerofoil—
a curved plate. Far away the air is undisturbed
by the presence of the wing, the pressure is
atmospheric (= patm ) and the streamlines are
straight and horizontal. Now consider moving
along a line from point A towards the surface,
keeping on a path that is always perpendicular to (c)
the local streamline direction. Starting at A we Figure 11. Streamlines around a symmetrical aerofoil
note that the streamlines are straight and parallel at various angles of attack. (a) Positive angle of attack:
and therefore there is no pressure gradient in the lift points up. (b) Zero angle of attack: no lift.
direction of the dashed line. However, closer (c) Negative angle of attack: lift points down.
to the aerofoil streamlines become increasingly
curved and there must now be a pressure gradient pressure force on the aerofoil, acting upwards, i.e.
across the streamlines. From the direction of lift.
curvature we note that the pressure drops as we From the above we learn that any shape
move downwards. By the time we reach the that introduces curvature into the flowfield can
aerofoil surface at B the pressure is noticeably generate lift. Aerofoils work because the flow
lower than that at A (pB < patm ). In the same follows the local surface curvature on the upper
way we can imagine moving from C to D. Again, and lower surfaces. It is not necessary to consider
as we approach the aerofoil streamlines exhibit frictional forces to explain lift, however; it is only
more and more curvature but this time the pressure due to the action of friction that streamlines take
increases towards the surface. At D the pressure up the pattern we would intuitively expect, so
is therefore greater than that at C (pD > patm ). strictly speaking lift would not be possible without
Hence pB < pD and this generates a resultant friction.

November 2003 PHYSICS EDUCATION 501

H Babinsky

(a) low angle of attack (b) high angle of attack (c) stalled flow

Figure 12. Streamlines around an aerofoil at increasing angle of attack.

Some observations resulting from this generate negative lift, but they do require more
explanation severe negative angles of attack (because they still
Following this line of argument it is possible to produce positive lift at zero angle of attack) which
make some interesting observations. For example, makes them less suitable for flying upside down.
consider the difference between the streamlines This also demonstrates the significance of
over a thin and a thick aerofoil as shown angle of attack, as seen again in figure 12. As
schematically in figure 10 (determined from a the angle of attack of a wing increases, more flow
computer simulation). Despite the difference in curvature is introduced above the wing—compare
thickness, both have similar flow patterns above (b) with (a)—and more lift is generated. However,
the upper surface. However, there is considerable at some point the flow is no longer capable of
difference in the flow underneath. On the thin following the sharp curvature near the nose and
aerofoil the amount of flow curvature below the it ‘separates’ from the surface. As a result the
wing is comparable to that above it and we might amount of streamline curvature above the wing
conclude that the overpressure on the underside is has reduced considerably (see figure 12(c)), which
just as large as the suction on the upper surface— causes a sharp drop-off in lift force. Unfortunately
the two sides contribute almost equally to the the process of flow detaching from the surface
lift. In the case of the thick aerofoil, however, often happens instantaneously when the angle of
there are regions of different senses of curvature attack is increased, making the loss of lift rather
below the lower surface. This suggests that there sudden and dangerous—this is called stall.
will be areas with suction as well as areas with
overpressure. In this case the lower surface does Conclusion
not contribute much resultant force and we can In this article I have attempted to give a ‘hands-on’
conclude that thin aerofoils are better at generating and correct explanation for Bernoulli’s equation,
lift. This is generally true, and birds tend to have the relationship between streamline curvature and
thin curved wings. Aircraft do not, because of the pressure, and lift. To explain lift it is not
structural difficulties of making thin wings, and necessary to go through all of the above steps
also because the volume contained in the wing is in the argument. Most students will be happy
useful, e.g. for fuel storage. with the streamline pattern around a lifting wing
A frequent question is how aircraft manage to (figure 1)—because it intuitively looks right—
fly upside down. To demonstrate this, figure 11 and this should be exploited. A short, but
shows the streamlines over a symmetrical aerofoil correct, explanation might start by discussing
at positive, zero and negative angles of attack. the existence of transverse pressure gradients in
Just by judging the degree of flow curvature curved streamlines and applying this knowledge to
above and below the wing it can be seen that the flowfield around an aerofoil in a similar manner
this aerofoil produces positive, zero and negative to that shown in figure 9. This should explain
lift respectively. Negative lift (which would why pressures on the two sides of an aerofoil are
be required for flying upside down) is simply different. There is no need even to introduce
a question of the angle of attack at which the Bernoulli’s equation or discuss the rather subtle
aircraft flies. Even non-symmetrical aerofoils can significance of friction.

502 PHYSICS EDUCATION November 2003

 How do wings work?

Acknowledgments difference between these points to the velocity

I would like to thank my colleagues, Rex difference:

2
2
Britter, Peter Davidson, Will Graham, John
Harvey, Harriet Holden, Tim Nickels and Len dp = − ρv dv
1 1
Squire, for many useful discussions and helpful  
comments. Will Graham also provided the v22 v2
p2 − p1 = − ρ −ρ 1
computer simulations of aerofoil streamlines. 2 2
which can be rearranged as
Appendix ρ ρ
p1 + v12 = p2 + v22 .
Pressure gradient along streamlines—Bernoulli’s 2 2
equation Because points 1 and 2 are arbitrary locations
along the streamline, the above equation can
Fluid particle be used to connect any two locations along
Surface area A
a streamline—this is in essence Bernoulli’s
Streamline
v equation. Note that the streamline need not be
straight.
x
p p+dp
l Pressure gradient across curved streamlines

In the above figure the cubic fluid particle Poutside

Surface area A
experiences a pressure p behind and a slightly Curved streamline
different pressure p + dp in front. This causes an h
v
acceleration according to Newton’s second law:
dv
F = ma = m . Pinside
dt
Here the resultant pressure force is F = −dp A, Pressure force
which is negative because it points in the –x Poutside > Pinside
direction. The mass of the fluid particle can be
determined from its volume and the fluid density The centripetal force is F = mv 2 /R and we
ρ: m = lAρ. can define pinside = p and poutside = p + dp.
The magnitude of the pressure change Similar to before we note that m = ρAh and
between front and back can be determined from dp = h(dp/dn), where n is the coordinate in the
the pressure gradient in the streamwise direction direction normal to the streamline (pointing away
(dp/dx) and the size of the particle: from the centre of curvature).
Combining all of the above yields
dp
dp = l . dp v2
dx F = A dp = Ah = ρAh
Combining all of the above gives dn R
which can be simplified to
dp dv
−l A = lAρ dp v2
dx dt =ρ
dn R
which simplifies to
which expresses the pressure gradient across
dv streamlines in terms of the local radius of curvature
dp = −ρ dx.
dt R and the flow velocity v. If a streamline is
Noting that dx/dt = v, this can be rewritten as straight, R → ∞ and dp/dn = 0. Therefore,
there is no pressure gradient across straight
dp = −ρv dv. streamlines.
Now we can integrate between two points (1 Received 9 September 2003
and 2) along a streamline to relate the pressure PII: S0031-9120(03)68660-0

November 2003 PHYSICS EDUCATION 503