294 C H A P T E R 5 Airfoils, Wings, and Other Aerodynamic Shapes

independent of the angle of attack. The location of the aerodynamic center for
real aerodynamic shapes can be found from experiment. For low-speed sub-
sonic airfoils, the aerodynamic center is generally very close to the quarter-
chord point.
Returning to Fig. 5.4a, we recall that the resultant aerodynamic force R can
be resolved into components perpendicular and parallel to the relative wind—
the lift and drag, respectively. An alternative to this system is to resolve R into
components perpendicular and parallel to the chord line, as shown in Fig. 5.4b.
These components are called the normal force and axial force and are denoted by
N and A, respectively, in Fig. 5.4b, shown by the heavy solid arrows. Also shown
in Fig. 5.4b are the lift and drag, L and D, respectively, represented by the heavy
dashed arrows. Lift and drag are easily expressed in terms of N and A from the
geometry shown in Fig. 5.4b:
L N cos α − A sin
i α (5.1)
D N sin α + A c os α (5.2)
For airfoils and wings, the use of N and A to describe the aerodynamic force dates
back as early as the work of Otto Lilienthal in 1889, as published in his book
Bird Flight as the Basis of Aviation (see Sec. 1.5). Indeed, the famous “Lilienthal
tables,” which were used by the Wright brothers to design their early gliders
(see Sec. 1.8), were tables dealing with normal and axial forces. The Wrights
preferred to think in terms of lift and drag, and they converted Lilienthal’s results
by using Eqs. (5.1) and (5.2). Today the use of N and A to describe the aerody-
namic force on airfoils and wings is generally passé; L and D are almost always
the system used by choice. However, N and A are still frequently used to denote
the aerodynamic force on bodies of revolution, such as missiles and projectiles.
Thus, it is useful to be familiar with both systems of expressing the aerodynamic
force on a body.

5.3 LIFT, DRAG, AND MOMENT COEFFICIENTS

Again appealing to intuition, we note that it makes sense that for an airplane in
flight, the actual magnitudes of L, D, and M depend not only on α, but also on
velocity and altitude. In fact, we can expect that the variations of L, D, and M
depend at least on
1. Free-stream velocity V∞.
2. Free-stream density ρ∞ (that is, altitude).
3. Size of the aerodynamic surface. For airplanes, we will use the wing area S
to indicate size.
4. Angle of attack α.
5. Shape of the airfoil.
6. Viscosity coefficient μ∞ (because the aerodynamic forces are generated in
part from skin friction distributions).
 5.3 Lift, Drag, and Moment Coefficients 295

7. Compressibility of the airflow. In Ch. 4 we demonstrated that

compressibility effects are governed by the value of the free-stream Mach
number M∞ = V∞/a∞. Because V∞ is already listed, we can designate a∞ as
our index for compressibility.
Hence, we can write that for a given shape of airfoil at a given angle of attack,

L f (V∞ , ρ∞ , S , μ ∞ , a∞ ) (5.3)

and D and M are similar functions.

In principle, for a given airfoil at a given angle of attack, we could find the
variation of L by performing myriad wind tunnel experiments wherein V∞, ρ∞,
S, μ∞, and a∞ are individually varied, and then we could try to make sense out of
the resulting huge collection of data. This is the hard way. Instead we ask: Are
there groupings of the quantities V∞, ρ∞, S, μ∞, a∞, and L such that Eq. (5.3) can
be written in terms of fewer parameters? The answer is yes. In the process of
developing this answer, we will gain some insight into the beauty of nature as
applied to aerodynamics.
The technique we will apply is a simple example of a more general theoreti-
cal approach called dimensional analysis. Let us assume that Eq. (5.3) is of the
functional form
L Z V∞a ρ∞b S d a∞e μ ∞f (5.4)
where Z, a, b, d, e, and f are dimensionless constants. However, no matter what
the values of these constants may be, it is a physical fact that the dimensions
of the left and right sides of Eq. (5.4) must match; that is, if L is a force (say
in newtons), then the net result of all the exponents and multiplication on the
right side must also produce a result with the dimensions of a force. This con-
straint will ultimately give us information about the values of a, b, and so on.
If we designate the basic dimensions of mass, length, and time by m, l, and t,
respectively, then the dimensions of various physical quantities are as given in
the following:

Physical Quantity Dimensions

L ml / t 2 (from Newton’s second law)
V∞ l/t
ρ∞ m/l 3
S l2
a∞ l/t
μ∞ m / (lt)

Thus equating the dimensions of the left and right sides of Eq. (5.4), we obtain

ml ⎛ l ⎞ ⎛ m ⎞ 2 d ⎛ l ⎞ ⎛ m ⎞
a b e f
= ⎜ ⎟ ⎜ ⎟ (l ) ⎜ ⎟ ⎜ ⎟ (5.5)
t2 ⎝ t ⎠ ⎝ l3 ⎠ ⎝ t ⎠ ⎝ lt ⎠
296 C H A P T E R 5 Airfoils, Wings, and Other Aerodynamic Shapes

Consider mass m. The exponent of m on the left side is 1, so the exponents of

m on the right must add to 1. Hence
1= b+ f (5.6)
Similarly, for time t we have
− 2 = −a − e − f (5.7)
and for length l,
1 a 3b + 2d + e − f (5.8)
Solving Eqs. (5.6) to (5.8) for a, b, and d in terms of e and f yields
b f (5.9)
a 2 e− f (5.10)
f
d = 1− (5.11)
2
Substituting Eqs. (5.9) to (5.11) into (5.4) gives
L Z (V∞ )2 e f
ρ∞
1 f
S a∞μ∞f
f //2
2 e
(5.12)
Rearranging Eq. (5.12), we find
⎛ a ⎞ ⎛ μ∞ ⎞
e f

L Z ρ∞V∞2 S ⎜ ∞ ⎟ ⎜ 1/ 2 ⎟
(5.13)
⎝ V∞ ⎠ ⎝ ρ∞V∞ S ⎠

Note that a∞ / V∞ = 1/M∞, where M∞ is the free-stream Mach number. Also note
that the dimensions of S are l2; hence the dimension of S1/2 is l, purely a length.
Let us choose this length to be the chord c by convention. Hence, μ∞ /(ρ∞ V∞ S1/2)
can be replaced in our consideration by the equivalent quantity
μ∞
ρ∞V∞c
However, μ∞ /(ρ∞ V∞ c) ≡ 1/Re, where Re is based on the chord length c.
Equation (5.13) thus becomes
⎛ 1 ⎞ ⎛ 1 ⎞f
e

L Z ρ∞V∞2 S ⎜ ⎟ ⎜ ⎟ (5.14)
⎝ M∞ ⎠ ⎝ Re ⎠
We now define a new quantity, called the lift coefficient cl, as
⎛ 1 ⎞ ⎛ 1 ⎞f
e
cl
≡Z⎜ ⎟ ⎜ ⎟ (5.15)
2 ⎝ M∞ ⎠ ⎝ Re ⎠
Then Eq. (5.14) becomes
L 1
2 ρ∞V∞2 Scl (5.16)
 5.3 Lift, Drag, and Moment Coefficients 297

Recalling from Ch. 4 that the dynamic pressure is q 1

2 ρ V∞2 , we trans-
form Eq. (5.16) into

L q∞ × S cl
↑ ↑ ↑ ↑
(5.17)
Dynamic Wing Li ft
L
Lift
pressure area coefficient

Look what has happened! Equation (5.3), written from intuition but not very
useful, has cascaded to the simple, direct form of Eq. (5.17), which contains a tre-
mendous amount of information. In fact, Eq. (5.17) is one of the most important
relations in applied aerodynamics. It says that the lift is directly proportional to
the dynamic pressure (and hence to the square of the velocity). It is also directly
proportional to the wing area S and to the lift coefficient cl. In fact, Eq. (5.17) can
be turned around and used as a definition for the lift coefficient:
L
cl ≡ (5.18)
q S

That is, the lift coefficient is always defined as the aerodynamic lift divided by
the dynamic pressure and some reference area (for wings, the convenient refer-
ence area S, as we have been using).
The lift coefficient is a function of M∞ and Re as reflected in Eq. (5.15).
Moreover, because M∞ and Re are dimensionless and because Z was assumed
initially as a dimensionless constant, from Eq. (5.15) cl is dimensionless. This is
also consistent with Eqs. (5.17) and (5.18). Also recall that our derivation was
carried out for an airfoil of given shape and at a given angle of attack α. If α were
to vary, then cl would also vary. Hence, for a given airfoil,
c f ( , M∞ ,Re) (5.19)
This relation is important. Fix in your mind that lift coefficient is a function of
angle of attack, Mach number, and Reynolds number.
To appreciate the value of the relationship expressed by Eq. (5.19), let us as-
sume that we are given a particular aerodynamic shape, and we wish to measure
the lift and how it varies with the different parameters. So we go to the laboratory
and set up a series of wind tunnel tests to measure the lift on our given shape.
Reflecting on Eq. (5.3), we know that the lift of the given shape at a given orien-
tation (angle of attack) to the flow depends on the free-stream velocity, density,
reference area, viscosity coefficient, and speed of sound; but we do not know
precisely how L varies with a change in these parameters. We wish to find out
how. We begin by running a set of wind tunnel tests, making measurements of
L where V∞ is varied but S, m∞, and a∞ are held fixed. This gives us a stack of
wind tunnel data from which we can obtain a correlation of the variation of L
with V∞. Next we run another set of wind tunnel tests in which r∞ is varied but
V∞, S, m∞, and a∞ are held fixed. This gives us a second stack of wind tunnel
298 C H A P T E R 5 Airfoils, Wings, and Other Aerodynamic Shapes

data from which we can obtain a correlation of the variation of L with r∞. Then
we run a third set of wind tunnel tests in which S is varied, holding everything
else constant. This gives us a third stack of wind tunnel data from which we can
obtain a correlation of the variation of L with S. We repeat this process two more
times, alternately holding m∞ constant and then a∞ constant. When we are fin-
ished, we end up with five individual stacks of wind tunnel data from which we
can (in principle) obtain the precise variation of L with V∞, r∞, S, m∞, and a∞, as
represented by the functional relation in Eq. (5.3). As you can probably already
appreciate, this represents a lot of personal effort and a lot of wind tunnel testing
at great financial expense. However, if we use our knowledge obtained from our
dimensional analysis—namely Eq. (5.19)—we can realize a great savings of ef-
fort, time, and expense. Instead of measuring L in five sets of wind tunnel tests
as previously described, let us measure the variation of lift coefficient [obtained
from cl = L/(q∞ S)]. Keying on Eq. (5.19) for a given shape at a given angle of
attack, we run a set of wind tunnel tests in which cl is measured, with M∞ varied
but Re held constant. This gives us one stack of wind tunnel data from which
we can obtain a correlation of the variation of cl with M∞. Then we run a second
set of wind tunnel tests, varying Re and keeping M∞ constant. This gives us a
second stack of data from which we can obtain a correlation of the variation of cl
with Re. And this is all we need; we now know how cl varies with M∞ and Re
for the given shape at the given angle of attack. With cl we can obtain the lift
from Eq. (5.17). By dealing with the lift coefficient instead of the lift itself, and
with M∞ and Re instead of r∞, V∞, S, m∞, and a∞, we have ended up with only two
stacks of wind tunnel data rather than the five we had earlier. Clearly, by using
the dimensionless quantities cl, M∞, and Re, we have achieved a great economy
of effort and wind tunnel time.
But the moral to this story is deeper yet. Dimensional analysis shows that cl is a
function of Mach number and Reynolds number, as stated in Eq. (5.19), rather than
just individually of ρ∞, V∞, μ∞, a∞, and the size of the body. It is the combination of
these physical variables in the form of M∞ and Re that counts. The Mach number
and the Reynolds number are powerful quantities in aerodynamics. They are called
similarity parameters for reasons that are discussed at the end of this section. We
have already witnessed, in Ch. 4, the power of M∞ in governing compressible flows.
For example, just look at Eqs. (4.73) through (4.75) and (4.79); only the Mach num-
ber and the ratio of specific heats appear on the right sides of these equations.
Performing a similar dimensional analysis on drag and moments, beginning
with relations analogous to Eq. (5.3), we find that
D q∞ Scd (5.20)
where cd is a dimensionless drag coefficient and
M q∞ Sccm (5.21)
where cm is a dimensionless moment coefficient. Note that Eq. (5.21) differs
slightly from Eqs. (5.17) and (5.20) by the inclusion of the chord length c. This
is because L and D have dimensions of a force, whereas M has dimensions of a
force–length product.
 5.3 Lift, Drag, and Moment Coefficients 299

The importance of Eqs. (5.17) to (5.21) cannot be overemphasized. They

are fundamental to all applied aerodynamics. They are readily obtained from
dimensional analysis, which essentially takes us from loosely defined functional
relationships [such as Eq. (5.3)] to well-defined relations between dimensionless
quantities [Eqs. (5.17) to (5.21)]. In summary, for an airfoil of given shape, the
dimensionless lift, drag, and moment coefficients have been defined as
L D M
cl = cd = cm = (5.22)
q S q S q Sc

where
cl f( M R ) cd = f M∞ ,Re) c f3 ( , M∞ ,Re) (5.23)
Reflecting for an instant, we find that there may be a conflict in our aerody-
namic philosophy. On the one hand, Chs. 2 and 4 emphasized that lift, drag, and
moments on an aerodynamic shape stem from the detailed pressure and shear
stress distributions on the surface and that measurements and/or calculations of
these distributions, especially for complex configurations, are not trivial under-
takings. On the other hand, the equations in Eq. (5.22) indicate that lift, drag, and
moments can be quickly obtained from simple formulas. The bridge between
these two outlooks is, of course, the lift, drag, and moment coefficients. All the
physical complexity of the flow field around an aerodynamic body is implicitly
buried in cl, cd, and cm. Before the simple equations in Eq. (5.22) can be used to
calculate lift, drag, and moments for an airfoil, wing, and body, the appropriate
aerodynamic coefficients must be known. From this point of view, the simplicity
of Eq. (5.22) is a bit deceptive. These equations simply shift the forces of aero-
dynamic rigor from the forces and moments themselves to the appropriate coef-
ficients instead. So we are now led to these questions: How do we obtain values
of cl, cd, and cm for given configurations, and how do they vary with α, M∞, and
Re? The answers are introduced in the following sections.
However, before we leave our discussion of dimensional analysis, it is
important to elaborate on why M∞ and Re are called similarity parameters.
Consider that we have two different flows (say a red flow and a green flow) over
two bodies that are geometrically similar but are different sizes for the red and
green flows. The red and green flows have different values of V∞, ρ∞, μ∞, and a∞,
but they both have the same M∞ and Re. If M∞ is the same for the red and green
flows and if Re is the same for the red and green flows, then from Eq. (5.23),
cl, cd, and cm measured in the red flow will be the same values as the cl, cd, and cm
measured in the green flow, even though the red and green flows are different
flows. In this case the red and green flows are called dynamically similar flows;
hence M∞ and Re are called similarity parameters. The concept of dynamic flow
similarity is elegant, and it goes well beyond the scope of this book. But it is
mentioned here because of its importance in aerodynamics. The concept of dy-
namic similarity allows measurements obtained in wind tunnel tests of a small-
scale model of an airplane to be applied to the real airplane in free flight. If in the
wind tunnel test (say the red flow) the values of M∞ and Re are the same as those
300 C H A P T E R 5 Airfoils, Wings, and Other Aerodynamic Shapes

for the real airplane in free flight (say the green flow), then cl, cd, and cm measured
in the wind tunnel will be precisely the same as those values in free flight. The
concept of dynamic similarity is essential to wind tunnel testing.
In most wind tunnel tests of small-scale models of real airplanes, every ef-
fort is made to simulate the values of M∞ and Re encountered by the real airplane
in free flight. Unfortunately, due to the realities of wind tunnel design and opera-
tion, this is frequently not possible. In such cases the wind tunnel data must be
“extrapolated” to the conditions of free flight. Such extrapolations are usually ap-
proximations, and they introduce a degree of error when the wind tunnel data are
used to describe the conditions of full-scale free flight. The problem of not being
able to simultaneously simulate free-flight values of M∞ and Re in the same wind
tunnel is still pressing today, in spite of the fact that wind tunnel testing has been
going on for almost 150 years. Among other reasons, this is why there are so
many different wind tunnels at different laboratories around the world.

5.4 AIRFOIL DATA

A goal of theoretical aerodynamics is to predict values of cl, cd, and cm from the
basic equations and concepts of physical science, some of which were discussed
in previous chapters. However, simplifying assumptions is usually necessary to
make the mathematics tractable. Therefore, when theoretical results are obtained,
they are generally not exact. The use of high-speed digital computers to solve the
governing flow equations is now bringing us much closer to the accurate calcu-
lation of aerodynamic characteristics; however, limitations are still imposed by
the numerical methods themselves, and the storage and speed capacity of current
computers are still not sufficient to solve many complex aerodynamic flows. As
a result, the practical aerodynamicist has to rely on direct experimental measure-
ments of cl, cd, and cm for specific bodies of interest.
A large bulk of experimental airfoil data was compiled over the years by the
National Advisory Committee for Aeronautics (NACA), which was absorbed in the
creation of the National Aeronautics and Space Administration (NASA) in 1958.
Lift, drag, and moment coefficients were systematically measured for many airfoil
shapes in low-speed subsonic wind tunnels. These measurements were carried out on
straight, constant-chord wings that completely spanned the tunnel test section from
one side wall to the other. In this fashion, the flow essentially “saw” a wing with no
wingtips, and the experimental airfoil data were thus obtained for “infinite wings.”
(The distinction between infinite and finite wings will be made in subsequent sec-
tions.) Some results of these airfoil measurements are given in App. D. The first page
of App. D gives data for cl and cm, c/4 versus angle of attack for the NACA 1408 airfoil.
The second page gives cd and cm, ac versus cl for the same airfoil. Because cl is known
as a function of α from the first page, the data from both pages can be cross-plotted
to obtain the variations of cd and cm, ac versus α. The remaining pages of App. D give
the same type of data for different standard NACA airfoil shapes.
Let us examine the variation of cl with α more closely. This variation is
sketched in Fig. 5.6. The experimental data indicate that cl varies linearly with