Chapter 3

INTRODUCTORY CONCEPTS TO COMPRESSIBLE FLOW

3.1. The Velocity of Sound

The term compressible flow implies variations in density throughout
the field of flow. These variations are, in many cases, the result princi-
pally of pressure changes from one point to a.nother. The rate of change
of density with respect to pressure is, therefore, an important parameter
in the analysis of compressible flow, and, as we shall see, it is closely
connected with the velocity of propagation of small pressure disturb-
ances, i.e., with the velocity of sound.

NOMENCLATURE
mean molecular velocity T absolute temperature
area V velocity
velocity of sound w mass rate of flow
specific heat at constant pressure W molecular weight
ratio of specific heats
mean free molecular path a Mach angle
length P bulk modulus of compression
Mach Number 6 boundary layer thickness
pressure X coefficientof thermal conductiv-
gas constant ity
universal gas constant P coefficient of viscosity
entropy per unit mass P density

Velocity of Propagation of a Plane Pressure Pulse. Let us calculate

the velocity of sound for a plane, infinitesimal pressure wave proceeding
along a pipe of uniform cross section. Such a wave might have been
initiated, for example, by a slight inward motion of a piston a t the left-
hand end of the pipe.
In Fig. 3.la the wave front is assumed t o propagate steadily to the
right with a velocity c . The fluid through which the wave front has
passed is a t a pressure p + d p , has a density p +
dp, and moves to the
right with a velocity d V . The fluid on the right, into which the wave
front is moving, has a pressure p and a density p, and is motionless.
To simplify the analysis, we reduce the process to one of steady motion
by imagining that the observer travels with the steady speed of the wave
45
46 INTRODUCTORY CONCEPTS T O COMPRESSIBLE FLOW Ch. 3
front. Fig. 3.lb shows the appearance of the process to such an obsemer
who is moving to the right with the velocity c. Fluid flows steadily
from right to left, and, as i t passes through the wave front, the velocity
is reduced from c to ( c - d V ) . At the same time, the pressure rises
+
from p to p d p , and the density from
(0) (b)
Wove Front Moving Stotionory Control
With Velocity c Wove Front

- dV
Stotionory C

FIG.3.1. Propagation of an infinitesimal pressure pulse.

(a) Observer a t rest.
(b) Observer moving with wave front, which is equivalent to superposing a left-
ward velocity c on the flow of part (a).

MOMENTUM EQUATION. Considering a stationary control surface

surrounding the stationary wave front of Fig. 3.1b1 the shear forces on
this control volume are negligibly small compared with the pressure
forces. The momentum equation may be written, +.herefore,as

where A is the cross-sectional area. Simplifying, and noting that from

the continuity equation,
w = pAc
we get
d p = pcdv (3.1)
This equation, it might be noted, is Euler's equation for a steady motion
(note that d V is the decrease in velocity in the direction of motion).
CONTINUITY EQUATION. The continuity equation is then written for
the fluid on both d e s of the wave front. Noting that the area is un-
changed, we get,
PC = ( P +
dp)(c - d V )
uhich reduces tc
Art. 3.1 THE VELOCITY OF SOUND 47

Combination of Eq. 3.1 with Eq. 3.2 then yields

The ratio dp/dp is written in Eq. 3.3 as a partial derivative a t constant

entropy because the variations in pressure and temperature are vanish-
ingly small, and consequently, the process is nearly reversible. More-
over, the comparative rapidity of the process, together with the small-
ness of the temperature variations, makes the process nearly adiabatic.
I n the limit, for an infinitesimal wave, the process may be considered
both reversible and adiabatic, and, therefore, isentropic.
Identical results for the velocity of propagation of a small disturbance
are obtained for a cylindrical wave spreading from a line source and for
a spherical wave spreading from a point source.
Velocity of Sound in a Perfect Gas. For a perfect gas the relation
between pressure and density in an isentropic process is given by
p/pk = constant (3.4)
Putting this into logarithmic form, differentiating, and noting that
p = pRT for a perfect gas, we obtain
In p - k In p = constant

Thus, we get for the velocity of sound in a perfect gas,

where W is the molecular weight.

EFFECT OF MOLECULAR WEIGHT. Since k varies only between narrow
limits, we see that, in general, gases with small molecular weights have
large sound velocities and vice versa. At normal atmospheric tempera-
ture, the speed of sound in air is of the order of 1100 ft/sec; in hydrogen,
of the order of 4200 ftjsec; and in freon refrigerating gases or uranium
hexafluoride, of the order of 300 ft/sec.
An interesting conclusion concerning turbomachinery can be drawn
from these figures. To avoid excessive stresses, the tip speeds of rotors
in turbomachines must not exceed a figure in the neighborhood of 1000
to 1500 ft/sec. It is also a matter of experience that losses in efficiency
mount rapidly as the flow velocity relative to the blades approaches the
sonic velocity. Thus, in air compressors, the limiting design factor on
rotative speed may be either stress considerations or compressibility con-
siderations. In hydrogen compressors, on the other hand, the compress-
48 INTRODUCTORY CONCEPTS TO COMPRESSIBLE FLOW Ch. 3

ibility limitation is never a factor. At the opposite end of the scale,

compressibility effects represent the principal limiting factor on the rota-
tive speed for freon and uranium hexafluoride compressors.
Velocity of Sound in Air. In the case of air at normal pressure and
temperature, Eq. 3.5 becomes
c = 49.02fi (3.6)
where c is in ft/sec and T is in degrees R, or degrees F abs.
VELOCITY OF SHOCK WAVE. A wave of finite strength, usually called
a shock wave, always propagates with a greater velocity than that indi-
cated in Eq. 3.3, the velocity of propagation increasing with the wave
strength. Shock waves are discussed in Chapter 5 and in Volume 11,
Chapter 25.
Velocity of Sound in Incompressible Fluid. Since an incompressible
fluid is one which, by definition, cannot experience changes in density,
Eq. 3.3 shows that the speed of sound is infinite in such a fluid. Pressure
pulses emitted anywhere in the fluid are thus felt simultaneously a t all
other points.
No fluids are truly incompressible, although liquids show little change
in density. The speed of sound in common liquids is of the order of
5000 ft/sec, which is enormously in excess of fluid velocities which can
be produced in a liquid.
Often we speak of the "incompressible flow of a gas," in the sense that
the fractional changes in density are so small as to be negligible.
Realms of Fluid Motion. Some of the different fields of fluid motion
may now be classified as follows:
ACOUSTICS.The fluid velocities are extremely small compared with
the velocity of sound, but the fractional variations in pressure, tempera-
ture, and density are of significant magnitude.
INCOMPRESSIBLE FLUID MECHANICS. The fluid velocities are small
compared with the velocity of sound, and the fractional variation in
density is insignificant; however, the fractional variations in pressure
and temperature may be very large. It will be shown later that the
error produced by neglecting compressibility in the computation of pres-
sure variations is of the order of one-fourth the square of the ratio of
the stream velocity to the sound velocity. Consequently, this ratio may
be as great as 0.2 (corresponding to a velocity of about 200 ft/sec for
air a t normal atmospheric temperature) before the computed error in
the pressure variation exceeds one per cent. For many problems in the
flow of gases, therefore, the flow may with little error be treated as
incompressible.
Art. 3.2 INCOMPRESSIBLE, SUBSONIC, AND SUPERSONIC FLOWS 49

COWPRESSIBLE FLUID MOTION. The velocities are appreciable com-

ge,rd with the velocity of sound, and the fractional variations in pres-
sure, temperature, and density are all of significant magnitude. This
realm, often called Gas Dynamics, is the principal subject of this book.
The science of Ballistics is a special branch of Gas Dynamics, pertain-
ing to the free flight of projectiles a t velocities greater than that of sound.
Although the term Gas Dynamics is used, it is well to remember that
the science of thermodynamics plays a role equally important to that of
the science of dynamics in the motion of compressible fluids.

3.2. Physical Differences Between Incompressible,

Subsonic, and Supersonic Flows

The influence of compressibility of the medium can lead to marked

qualitative differences between the physical natures of incompressible
and compressible flows. And, in compressible flows, there are aston-
ishing differences in flow pattern and general behavior, depending on
whether the fluid velocity is greater or less than the local speed of
sound. Indeed, it is safe to say that one's "instincts" or past experience
with subsonic flows are completely useless when dealing with supersonic
flows and are apt to lead to false conclusions.
Pressure Field Created by a Moving Point Disturbance. To illustrate
this in a simple way, i t is useful to adopt momentarily a point of view
which is somewhat oversimplified yet a t the same time brings out the
salient physical phenomena. When a body moves through a fluid, or
when a fluid flows past a body or within the walls of a duct, each element
of solid surface tends to divert the fluid from the course which i t might
otherwise take. For example, in the case of a projectile moving through
air, each element of the projectile's surface area pushes the neighboring
air out of the way, and this local disturbance creates a pressure pulse
which propagates into the exterior air, much as the impulsive motion of
the piston in Fig. 3.1 results in the propagation of a pressure wave down
the duct. At any instant of time, therefore, the pressure perturbation
a t a point away from the projectile will be some sort of sum (not neces-
sarily linear) of those pressure pulses which were initiated earlier a t each
element of projectile surface and which have arrived a t the point in ques-
tion after spreading spherically from the individual points of emission.
Following this line of thought, let us consider the pressure field cre-
ated by the most elementary type of a moving disturbance-namely, a
point source of disturbance moving a t uniform linear speed through a
compressible medium. At each instant of time the point source may be
imagined to emit an infinitesimal pressure wave which spreads spheri-
cally from the point of emission with the speed of sound relative to the
fluid. The pressure pattern which exists at any instant is then found by
superposition of all the pressure pulses which were previously emittod.
Fig. 3.2 shows several pressure pulse patterns for different values of
che speed of the source compared with the speed of sound in the fluid.

F I ~3.2.
. Pressure field produced by a point source of disturbance moving a t uni-
farm speed leftwards.
(a) Incompressible fluid (V/c = 0).
(b) Subsonic motion (V/c = x).
(c) Transonic motion (V/c = 1).
(d) Supersonic motion, illustrating Karman's three rules of supersonic flow
(Vlc = 2).

In each pattern the point 0 represents the present location of the dis-
turbance, the point - 1 represents the location one unit of time pre-
viously, and so on. For each of these previous locations there is drawn
a concentric circle showing the distance to which the corresponding wave
has spread. For example, to find the present location of the wave which
was emitted a t time -3, a circle is drawn with - 3 as a center and with
a radius 3 4 where t is the unit of time. The distance between point -3
and point 0 is then given by 3Vt where V is the velocity of the point
disturbance with respect t o the medium.
Art. 3.2 INCOMPRESSIBLE, SUBSONIC, AND SUPERSONIC FLOWS 51

INCOMPRESSIBLE FLOW. When the medium is incompressible (Fig.

3.2a), or when the speed of the moving point disturbance is small com-
pared with the speed of sound, the pressure pulse spreads uniformly in
all directions.
SUBSONIC FLOW. When the source moves at subsonic speeds, Fig.
3.2b, the pressure disturbance is felt in all directions and a t all points
in space (neglecting dissipation due to viscosity), but the pressure pat-
tern i,s no longer symmetrical.
SUPERSONIC F LOW. For supersonic speeds Fig. 3.2d indicates that
the phenomena are entirely different from those a t subsonic speeds. All
the pressure disturbances are included in a cone which has the point
source a t its apex, and the effect of the disturbance is not felt upstream
of the source of disturbance. The cone within which the disturbances
are confined is called the Mach cone. Fig. 3 . 2 shows
~ the pressure pat-
tern a t the boundary between subsonic and supersonic flow, that is, for
the case where the stream velocity is identical with the sonic velocity;
here the wave front is a plane.
Karman's Rules of Supersonic Flow. Fig. 3.2d illustrates the three
rules of supersonic flow proposed by von Karman.'') These rules apply
exactly only for small disturbances, but are usually qualitatively appli-
cable for large disturbances.
T HE RULE OF FORBIDDEN SIGNALS. The effect of pressure changes
produced by a body moving a t a speed faster than sound cannot reach
points ahead of the body.
T HE ZONE OF ACTION AND THE ZONE OF SILENCE. A stationary point
source in a supersonic stream produces effects only on points that lie on
or inside the Mach cone extending downstream from the point source.
Conversely, the pressure and velocity a t an arbitrary point of the stream
can be influenced only by disturbances acting a t points that lie on or
inside a cone extending upstream from the point considered and having
the same vertex angle as the Mach cone.
T HE R ULE OF CONCENTRATED ACTION. The proximity of the circles
representing the different pressure impulses in Fig. 3.2 is a measure of
the intensity of the pressure disturbance a t each point in the field of
flow. Thus, for the stationary source, the intensity of the disturbance
is symmetrical about the source. For the subsonic source, the intensity
is unsymmetrical. In the case of the supersonic source, we have the
rule of concentrated action: the pressure disturbance is largely concen-
trated in the neighborhood of the Mach cone that forms the outer limit
of the zone of action.
52 INTRODUCTORY CONCEPTS TO COMPRESSIBLE PLOW Ch. 3

These rules explain why a projectile or rocket moving a t supersonic

speed cannot be heard until the wave attached to the nose of the body

FIG.3.3. Shadowgraph of supersonic bullet passing through cylindrical tube, show-

ing Mach cone and spherically spreading wave fronts (after Aclceret).
(a) Single tube.
(b) Several short tubes in series, showing individual wavelets which passed
through the open space between the tubes.

passes over the ear of the observer; and why, when the latter does occur,
the noise is concentrated in a "crack."
Configurations like those of Fig. 3.2 may easily be observed in the
form of gravity waves on a free water surface when a sharp-pointed
-
object is drawn through the water a t varvine
speeds. The bow wave of a surface ship re-
" -
sembles the Mach cone of Fig. 3.2d.
Patterns like those of Fig. 3.2 can be made
visible in gas flows by means of shadow,
schlieren, or interferometer techniques, which
are optical methods for demonstrating density
variations in a gas. Fig. 3.3a, for example, is
a shadowgraph of a bullet which has just
passed through a cylindrical tube while travel-
ing a t supersonic speed. Fig. 3.3b is a schlie-
ren photograph of a similar arrangement, ex-
cept that slots are cut in the tube to allow
F I G. 3.4. Schlieren pho- only a selected number of wavelets to pass out
tograph of bullet traveling
at near the speed of sound of the tube into the field of view.
(after Aclceret). Fig. 3.4 shows a bullet traveling a t nearly
the speed of sound and having a wave front
similar to that of Fig. 3 . 2 ~ . At low subsonic speeds, corresponding to
Fig. 3.2b, no wave front appears.
Art. 3.2 INCOMPRESSIBLE, SUBSONIC, AND SUPERSONIC FLOWS 53

(b)
FIG.3.5. Cone-cylinder projectile traveling a t supersonic speed (Ballistic Research
Laboratory, Aberdeen).
(a) Schlieren photograph.
(b) Interferometer photograph.
54 INTRODUCTORY CONCEPTS TO COMPRESSIBLE FLOW Ch. 3

The Mach cones produced in a supersonic wind tunnel by the nose,

fins, and tail of a model of a guided missile are shown in Fig. 3.6.

FIG. 3.6. Schlieren photograph of a model of a guided missile in a supersonic wind

tunnel. Mach cones from nose, fins, and tail are visible (NACA).

3.3. The Mach Number and Mach Angle

In the preceding article it was shown that the nature of the flow pat-
tern depends on the comparative magnitudes of the stream velocity and
the sonic velocity. The ratio of these two velocities is called the Mach
Number. Thus,

The speed of sound in Eq. 3.7 is to be taken a t the local temperature

and pressure of the stream, and, of course, it varies from point to point
in the flow field.
The semi-angle of the Mach cone (Fig. 3.2d) is related to the Mach
Number by
sin a = 1/M (3.8)
Note that the Mach angle is imaginary for subsonic flow.
We have found the Mach Number to be a criterion of the type of
flow pattern. Later i t will be shown that it is also a parameter which
almost always appears in the equations of motion. I n the next article
we shall see that it is a dimensionless parameter important for model
testing.
3.4. Similarity Parameters
Consider a prototype flow pattern and a model which is geometrically
similar (as regards solid boundaries) to the prototype. We inquire as to
the conditions which must be met in order that the flow pattern for the
model be similar to that for the prototype.
Art. 3.4 SIMILARITY PARAMETERS 55

Let L be a characteristic length for the prototype, and let jrJ. be the
corresponding characteristic length for the model, where fL is the scale
factor for length. Similarly, we can summarize the other properties of
the prototype flow and model flow as follows:

Prototype Model
fvV
fop
~ P P

~ L L
fcc
fMM

fkk

Ignoring viscosity for the present, Euler's equation of motion must be

satisfied for both the prototype and model. Thus:

Since the scale factors are constant, both of these equations can be satis-
fied only if
f p = fPfv2 (3.9a)
From Eq. 3.5 we obtain in similar fashion

c = dkp/p
and

so that
fcc = 4iix s
fc = m
Likewise, from Eq. 3.7, we have:

and

so that

Combining Eqs. 3.9a, 3.9b, and 3.9c, we finally get:

56 INTRODUCTORY CONCEPTS TO COMPRESSIBLE FLOW Ch. 3

By examining the energy equation, taking into account the further

effects of viscous work and heat conductivity, it may be shown that the
ratio of specific heats, k , must be the same for both model and proto-
type, i.e., fk = 1. Thus, we conclude that

or the Mach Number must be the same for the model and prototype if
the flows are to be similar.
When viscosity is present, a similar analysis applied to the inertia and
viscous terms in the Navier-Stokes equation leads to the criterion that
the Reynolds Number must be the same for similarity in flow patterns.
Dimensionless Groups Governing Compressible Flows. By consider-
ing all the physical equations which govern the flow, namely, the Navier-
Stokes equation, the continuity equation, the energy equation, and the
equation of state, i t is possible to arrive a t four dimensionless param-
eters which must be the same in order for two flow patterns to be sim-
ilar. These are
(i) The Mach Number
(ii) The Reynolds Number
(iii) The ratio of specific heats, k
(iv) The Prandtl Number, cPw/X,where c, denotes the specific heat
a t constant pressure, rc, the coefficient of viscosity, and X the
thermal conductivity.
POTENTIAL FLOW OUTSIDE BOUNDARY LAYER. For those regions of
the flow outside the boundary layer, where viscous effects and heat con-
duction effects are relatively unimportant, it is usually necessary that
only M and k be alike in order to have similarity. Of the two, similarity
in M is by far the more important, sincek has a relatively weak influence
on the flow pattern.
BOUNDARY-LAYER FLOW. In the boundary layer or in the interior of
shock waves, viscous and heat conduction effects are all-important, and
so the Reynolds Number and Prandtl Number must be included in the
similarity conditions. Fortunately, the Prandtl Number is nearly the
same for all gases and varies only slowly with temperature.

3.5. Domain of the Continuum

The concept of the continuum has already been discussed in Chapter

1, where it was pointed out that the criterion for continuum mechanics
is that the mean free molecular path be small compared with the small-
est significant body dimension.
Art. 3.5 DOMAIN OF THE CONTINUUM 57

In order to determine when this condition is valid, let us consider a

steady flow and make some approximate calculations of an order-of-
magnitude nature. ,-

Now kinetic theory (z)shows that, in respect to orders of magnitude,

and

where a is the mean molecular velocity and I is the mean free molecular
path.
Using the foregoing relations, we now express the Reynolds Number

where L is a characteristic body dimension on which Rey is based. This

relation may be rearranged to give
L -
1- M
Rey

thus showing that the ratio of Reynolds Number to Mach Number is a

dimensionless parameter indicative of whether or not a given problem
i s amenable to the continuum hypothesis. It is seen that the contin-
uum concept is likely to fail either a t very high Mach Numbers or a t
extremely low Reynolds Numbers.
Rules for determining the validity of the continuum concept in terms
of Rey and M cannot be stated generally since they depend on the par-
ticular type of problem. However, we may illustrate the approach to
this problem by supposing that, in a given problem, the smallest sig-
nificant body dimension is of the order of the boundary-layer thickness,
6. If the Reynolds Number is large compared with unity, and if the
boundary-layer flow is also laminar, then the boundary-layer relations
for a flat plate show that
6 1

which, when combined with Eq. 3.10, yields

For this case Tsien (3) suggests that the realm of continuum gas dynam-
ics be limited to instances where the boundary-layer thickness is a t least
100 times the mean free path. That is,
58 INTRODUCTORY CONCEPTS TO COMPRESSIBLE FLOW Ch. 3

Fig. 3.7 shows the Reynolds Number per unit length as a function of
flight Mach Number for various altitudes, based on the standard atmos-

0 1.0 2.0 3.0 4.0

Mach number,M
FIG.3.7. Reynolds Number per foot versus Mach Number, based on standard at-
mosphere (NACA Tech. Note, No. 1428).

3.6. Classification of Compressible Flows

When dealing with problems of steady motion, it is sometimes con-
venient to subdivide the subsonic and supersonic categories in classify-
ing different types of flow.
The Adiabatic, Steady-Flow Ellipse. Consider a stream tube in which
the flow does not exchange heat with the fluid in neighboring stream
tubes. The steady-flow energy equation for the flow in such a tube is
then
v 2
- + h = constant
2
Now, for a perfect gas, Ah = c, AT, and T = c 2 /kR. Introducing these,
together with the relations between the gas constants, the energy equa-
tion may be written as
2 2
V2 +-
k-1
c2 = constant = -co2 = V
k-1
2
max (3.11)

where co is the speed of sound a t the stagnation condition (where V is

zero) and Vma,is the maximum possible velocity in the fluid (where the
absolute temperature is zero).
Art. 3.7 OPTICAL METHODS OF INVESTIGATION 5!4

According to Eq. 3.11, the possible st,a,t.~sin the stream tube are rep-
resented in a diagram of c versus V by the steady-flow ellipse (Fig. 3.8).
Different parts of this ellipse represent schematically different realms of
compressible flow having significantly different physical characteristics.
Incompressible Flow. The velocity is small compared with the sonic
speed. Changes in c are very small compared with changes in V.
Subsonic Compressible Flow. The velocity and sonic speed a w of
comparable magnitudes, but the former is less than ithe latter. Changes
in Mach Number occur primarily Subsonic
because of changes in V, and only
secondarily through changes in c. Incorn-
press~ble '.,
C V.C Transonic
Transonic Flow. The di$erence
between V and c is small compared V>C
with either V or c. Changes in V v
"m,.
and c are of comparable magnitude.
FIG.3.8. Steady-flow adiabatic ellipse,
Supersonic Flow- The velocity showing realms of compressible flow.
and sonic speed are of comparable
magnitude,-but the former is larger than the latter. Changes in Mach
Number take place through substantial variations in both V and c.
Hypersonic Flow. The velocity is very large compared with the sonic
speed. Changes in velocity are very small, and thus variations in Mach
Number are almost exclusively the result of changes in c.

3.7. Optical Methods of Investigation

Apart from the conventional methods of experimentally investigating
Row patterns by means of pressure and velocity surveys, compressible
gas flows lend themselves particularly well to optical methods of inves-
tigation.
Comparison of Methods. Fundamentally, the optical methods in
common use (the interferometer, the schlieren, and the shadowgraph)
depend on one of two physical phenomena: (i) the speed of light depends
on the index of refraction of the medium through which it passes, and
the index of refraction of a gas in turn depends upon its density; and,
as a consequence of this first phenomenon, (ii) light passing through a
density gradient in a gas (and therefore through a gradient in index of
refraction) is deflected in the same manner as though i t were passing
through a prism. In a high-speed gas flow the density changes are
sufficiently large to make these phenomena sizable enough for optical
observation.
60 INTRODUCTORY CONCEPTS T O COMPRESSIBLE FLOW Ch. 3
The interferometer, based on phenomenon (i), measures directly
changes of density, and is primarily suited for quantitative determina-
tion of the density field.
The schlieren method, based on phenomenon (ii), measures density
gradients. Although it is theoretically adaptable to quantitative use,
it is inferior to the interferometer in this respect, and its greatest utility
is in giving an easily interpretable picture of the flow field together with
a rough picture of the density variations in the flow.
The shadowgraph method, also based on phenomenon (ii), measures
the second derivative of the density (i.e., the first derivative of the
density gradient). Therefore it makes visible only those parts of the
flow where the density gradients change very rapidly, and i t has found
its greatest utility in the study of shock waves.
Of the three methods mentioned, the interferometer yields the most
information, and the shadowgraph the least. On the other hand, the
interferometer is the most costly and the most difficult to operate,
whereas the shadowgraph is the least costly and the easiest to operate.
Each method therefore has its own useful niche in experimental work,
and the choice of method depends on the nature of the investigation.
In order to assist the reader in the interpretation of photographs taken
by the three methods, the basic principles of operation will be outlined,
without attention to the numerous optical and mechanical refinements
required. Details of these refinements may be found in References 4
to 12.
The Interferometer. Fig. 3.9 shows the essential parts of the Mach-
Zehnder interferometer, the type which seems to be the most useful for
gas-flow investigations.
FORMATION OF INTERFERENCE PATTERNS. Referring to Fig. 3.9, light
from the source first passes through a collimating lens which renders
the light parallel, and then passes through a monochromatic filter. I t
then passes through the first "splitting plate," which is a plane half-
silvered mirror that passes half the light and reflects half the light. The
light which is reflected is changed back to its original direction by the
lower mirror and then passes through the test section. The light which
passes through the first splitting plate passes through a pair of glass
compensating plates which duplicate in thickness and quality the glass
side walls of the test section, and is then directed to the lower splitting
plate by means of the upper mirror. The two beams which divided at
the upper splitting plate are recombined a t the lower splitting plate and
are then focussed by a lens system on either a ground glass screen or on
a photographic plate.
We shall assume for this discussion that the flow in the test section is
normal to the paper and that the flow is two-dimensional in planes nor-
Art. 3.7 OPTICAL METHODS OF INVESTIGATION 61

ma1 t o the paper and normal to the light beam. Thus, except for slight
refraction effects, each ray of light passing through the test section tra-
verses gas of constant density.
The chief difference between the two beams of light in Fig. 3.9 is that
the lower beam has passed through the test section. Since the different
rays of the lower beam are retarded by different amounts as they pass
through portions of the test section of different density, these rays will

COMPENSATOR
CHAMBER
!loUNT
TRANSLATOR

ROTATION MOUNT

COMPENSATING

HALF SILVERED
INTERFEROMETER

SCHEMATIC DIAGRAM - I NTERFEROMETER

FIG.3.9. Mach-Zehnder type of interferometer (M.I.T. Gas Turbine Laboratory).

have various phase differences with the corresponding rays of the upper
beam when they are recombined, and thus an interference pattern will
be formed a t the screen.

INFINITE-FRINGE INTERFEROGRAM. Suppose to begin with that all

the mirrors and splitting plates are exactly parallel and in line, that there
is no flow, and that the air in the test section has exactly the same den-
sity as the reference air of the upper light beam. Then the two light
beams will be exactly in phase when they recombine, and the screen
will appear uniformly bright. Now imagine that the pressure in the
test section is uniformly and gradually increased. Since the entire lower
light beam is thus retarded by the same amount, the screen will remain
uniform in light intensity, but will gradually darken as the test-section
density is increased. After the density has reached the point where the
lower beam is out of pha.se with the upper by one-half wave length, the
02 INTRODUCTORY CONCEPTS TO COMPRESSIBLE FLOW Ch. 3
screen will be uniformly dark. Further increases of density will produce
a lightening of the screen until, with a phase difference of one wave
length, the screen will again be a t maximum brightness. As the density
continues to increase, the cycle of darkening and brightening will be
repeated.
With the same mirror geometry as before, consider now the situation
with flow through the test section. The optics are the same as above,
except that the screen is no longer uniform in brightness, but has on it
-
contours of brightness. These contours represent contours of constant
index of refraction, and therefore
represent contours of constant
density in the flow field. Fig. 3.10
shows such an "infinite-fringe" in-
terferogram of flow through a
supersonic nozzle, t h e t e r m
"infinite-fringe"
- signifying
- . -that the
FIG. 3.10. Infinite-fringe interfero- light field is uniform in the ab-
gram of flow through supersonic nozzle
(M.I.T. Gas Turbine Laboratory). sence of flow through the test sec-
tion. The dark bands in Fig. 3.10
are loci of points where there is complete interference, and represent
contours of constant density.
QUANTITATIVE EVALUATION. The change in density from one dark
band to the next in an interferogram like that of Fig. 3.10 may be eval-
uated in the following manner. On the dark bands the light waves
passing through the test section are out of phase with those which pass
through the room air in the compensating chamber by s,lx,254, . . -
room-air wave lengths. Therefore, the light beams passing through ad-
jacent dark bands of the test section are out of phase by one room-air
wave length. Hence, if a represents the fluid lying in one dark band,
and b the fluid in an adjacent dark band, the difference in time for a
light beam to pass through a as compared with that to pass through b
is given by

where X and V denote respectively the wave length of the light and the
speed of light. This difference in traversal time is also rclatcd to the
difference in speed of light in the test section. Thus,

where L ik the length of test section along the light direction, and V b
and Va denote respectively the speeds of lisht in fluid b and a.
Art. 3.7 OPTICAL METHODS OF INVESTIGATION 63

Now the frequency j of a given monochromatic light is constant.

The velocity of light in a given medium is related to the velocity of

light in vacuo through the index of refraction n, defined by

-
Eliminatinp. t- "h - t.. from
-u -- ---- Ens.
-1-
3.12 and 3.13, and introducing Eqs. 3.14
and 3.15, we may obtain for t hie difference in index of refraction between
adjacent 1
na = X v m u o I L (3.16)
It remains to connect the index of refraction with the gas density.
This is done through the empirical Gladstone-Dale equation, which
states that
n-1
- --KG-D (3.17)
P

where KG-D,the Gladstone-Dale constant, is constant for a given gas.

Eliminating n in Eq. 3.16 in favor of p by means of Eq. 3.17, we obtain
finally

The right-hand side of this equation is easily computed from the di-
mension of the test section, the color of the monochromatic light em-
ployed, and the value of KG.D for air. Referring again t o Fig. 3.10,
the density in the low-speed flow upstream of t8hethroat is found by
measuring the temperature and pressure in this low-speed region. With
this as a reference, the density on each dark band in the nozzle may be
computed from Eq. 3.18, although the accuracy of this procedure using
the infinite-fringe interferogram is not high unless the optical compo-
nents are extraordinarily accurate.
FRINGE-DISPLACEMENT METHOD. For a more accurate quantitative
evaluation, the method described above is modified as now described.
Let us return to the situation where room air is in the test section of
Fig. 3.9. If the second splitting plate is rotated through a small angle
with respect to the first splitting plate, two "coherent" beams of light
which were in phase a t the first splitter will, through the change in
lengths of light paths, be out of phase a t the screen. If the splitters
are rotated about axes normal t o the paper, there will be formed on the
screen successive light and dark fringes, uniformly spaced, with each
fringe lying parallel to the axis of rotation (see, for example, Fig. 3.11a).
64 INTRODUCTORY CONCEPTS TO COMPRESSIBLE FLOW Ch. 3

The spacing of successive dark fringes may be shown to be equal to

2/6,X, where 6 is the difference in angles of rotation between the two
splitters, and thus the number of fringes in the flow field is controlled
by adjusting the geometry of the optical elements.
Now suppose that the air density in the test section is uniformly in-
creased. This will produce a uniform displacement of all the wave fronts
passing through the test section. This displacement in turn will cause

FIG.3.11. Fringe-shift interferograms for supersonic nozzle (M.I.T. Gas Turbine

Laboratory).
(a) No flow.
(b) Supersonic flow in test section.
(c) Superposition of (a) and (b).

the interference bands on the screen to shift in a direction normal to

the bands, although the bands will remain parallel and uniformly spaced.
The fringe shift is a measure of the change in density in the test section.
By an analysis similar to that given previously, it may be shown that

where pz is the density in the test section, p l is the density a t the initial
reference condition, d is the distance between dark fringes in the refer-
ence condition, and I is the distance shifted by a dark fringe in passing
from condition 1 to condition 2.
When there is flow in the test section, with corresponding nonuniform
density changes, similar fringe shifts will occur, but, as they are no longer
uniform, the resultant fringes will be curved (Fig. 3.11b). Eq. 3.19 may
Art. 3.7 OPTICAL METHODS O F INVESTIGATION 65
then be applied a t each point in the flow. If both a flow and a no-flow
photograph are taken, Eq. 3.19 may be used to determine a t each point
the density change referred to the no-flow density. Or, from the flow
photograph alone, Eq. 3.19 may be used for finding the density differ-
ences between two points of the flow.
SUPERPOSITION METHOD. If the no-flow negative (Fig. 3.11a) is super-
imposed on the flow negative (Fig. 3.11b), the light passing through the
combination produces the light pattern of Fig. 3.11~. This is a particu-
larly convenient way of determining the contours of constant density,
for a consideration of the geometry of the two fringe patterns shows
that the fuzzy dark bands of Fig. 3 . 1 1 ~ are lines of constant fringe shift,
and, therefore, that they are lines of constant density. Indeed, the pat-
tern of Fig. 3 . 1 1 ~is like that of Fig. 3.10, and Eq. 3.18 may be applied
to either.
The Schlieren Method. Fig. 3.12 shows one of the numerous types of
schlieren (striae) arrangements. Light from a uniformly illuminated line
FL-I
Two- Dim.

f
Finite
Line Lens
source Screen or
Knife Photographic
Gloss Edge Plote
W0llS
z
Fra. 3.12. Schlieren system.

source of small but finite width is collimated by the first lens and then
passes through the test section. It is then brought to a focus by the
second lens and projected on the screen. At the focal point, where there
exists an image of the source, there is introduced a knife-edge which
cuts off part of the light. With no flow in the test section the knife-
edge is usually adjusted so as to intercept about half the light, and the
screen is uniformly illuminated by the portion of the light escaping the
knife-edge. When the flow is established in the test section (assumed
here, for simplicity, to be two-dimensional, with each light ray passing
through a path of constant air density) any light ray passing through a
region in which there is a density gradient normal to the light direction
will be deflected as though it had passed through a prism. Therefore,
depending on the orientation of the knife-edge with respect to the den-
sity gradirnt, and on t,he sign of the density gradient, more or less of
66 INTRODUCTORY CONCEPTS TO COMPRESSIBLE FLOW Ch. 3

the light passing through each part of the test section d l escape the
knifeedge and illuminate -the screen.
Thus the schlieren system makes density gradients visible in terms of
intensity cf illumination. A photographic plate at the viewing screen
records density gradients in the test section as different shades of gray.
THEORY OF LIGHT DEFLECTION. Let us assume that the flow in the
test section is parallel and in the x,y-plane, and that the light passes
through the test section in the z-direction.
Since the speed of a wave front of light varies inversely with the index
of refraction n of the medium through which the light travels, it follows
that a given wave front will rotate as it passes through a gradient in n.
Accordingly, the normal to the wave front will follow a curved path;
since this normal is what we mean by the light ray, the latter is refracted
as it passes through the density gradient. Noting that n is nearly unity
for gases, i t may be shown that

where R is the radius of curvature of the light ray. The total angular
deflection E of the ray in passing through the test section of width L is
therefore given by 7

Resolving this into Cartesian components, and taking note of Eq. 3.17,
we get for the angular deflections of the light in the x- and y-directions

If the knife-edge is aligned in the y-direction (i.e., normal to the flow),

as in Fig. 3.12, only deflections E , will influence the light passing the
knife-edge. Hence density gradients in the x-direction will be made
visible, but gradients in the y-direction will not be visible. Similarly,
if the knife-edge is parallel to x, only gradients in the y-direction will
become visible.
In interpreting a schlieren photograph of a two-dimensional flow, it
is convenient to imagine that the photograph represents a plan view of
a relief map of the flow field in which the vertical elevation is the gas
density. If the map is illuminated by glancing side light nearly parallel
to x, the pattern of shadows and brightness will correspond to a schlieren
Art. 3.7 OPTICAL METHODS OF INVESTIGATION 67
photograph with knifeedge in the y-direction. Similarly, illumination
in the y-direction corresponds to placing the schlieren knife-edge parallel
to the x-direction.
Figs. 3.13a and 3.13b are schlieren photographs of the supersonic
nozzle of Figs. 3.10 and 3.11, and show the different effects obtainable

FIG.3.13. Sehlieren photographs of flow through nozzle of Figs. 3.10 and 3.11
(M.I.T. Gas Turbine Laboratory).
(a) Knife-edge horizontal.
(b) Knife-edge vertical.

with different orientations of the knife-edge. In Fig. 3.13a (knife-edge

horizontal) light areas on the upper half correspond to dark areas on
the lower half, and vice versa; the boundary layers on the walls are
clearly visible, since with this knife-edge orientation the density gra-
dients normal to the flow are made visible. I n Fig. 3.13b (knife-edge

vln
vertical) the upper and lower halves look alike; the boundary layers are
not visible except on the inclined walls of the nozzle.
The Shadowgraph. A shadow system (Fig. 3.14) comprises simply a
small, bright source, a collimating lens, and a viewing screen or photo-
graphic plate. If the source is far
from the test section the collimat-
ing lens is unnecessary.
4
Assume a t first that the test sec- w l n t
Source
tion has stagnant air in it and that
Viewing
the intensity of illumination on the Flow Screen or
Photographic
screen is uniform. When flow is Piate
established in the test section the F IG. 3.14. Shadowgraph system.
light beam will be refracted wher-
ever there is a density gradient. However, if thc- density gradient were
constant, every light ray would be deflected by the same amount, and
there would be no change in illumination on the screen. Only if there
is a gradient in density gradient will there be any tendency for the light
rays to diverge or converge. From this it is evident that variations in
illumination of the screen are proportional to the second derivative of
the density gradient, i.e., to the term
aZP aZp
-+-
ax2 ay2
assuming two-dimensional flow in the x,y-plane.
68 INTRODUCTORY CONCEPTS TO COMPRESSIBLE FLOW Ch. 3

The shadowgraph is therefore suited to flows with rapidly varying den-

sity gradients, and is insensitive to flows with gently varying gradients.
It is especially convenient and simple for making shock waves visible.
Fig. 3.15a is a typical shadowgraph of a detached shock wave. Across
such a shock the density increases from one side to the other in an
S-shaped curve. Hence the r8te of change of density gradient is posi-
tive on the upstream side of the shock and is negative on the downstream

(a) (b)
FIG.3.15. Detached shock in front of blunt body (Ordnance Aerophysics Labora-
tory).
(a) Shadowgraph.
(b) Schlieren, knife-edge horizontal.

side. In the shadow picture the shock therefore shows as a dark line
followed by a bright line. For comparison, Fig. 3.15b shows a schlieren
photograph of a similar shock.

R E F E R E N C E S A N D S E L E C T E D BIBLIOGRAPW
1. VON KLRMLN,T H. Supersonic Aer~dy~nlics-l'rinciplcs and Applications,
Jour. Aero. Sci., Val. 14, No. 7 (1947), p. 373.
2. JEANS, SIR JAMES. An Introduction to the Kinetic Theory of Gases. Cambridge:
Cambridge University Press, 1946.
3. TSIEN, H. S. Superaerodynamics, Mechanics of Rarefied Gases, Jour. Aero. Sci.,
Vol. 13, No. 2 (1946), p. 653.
4. SCHARDIN, H. Dm Toeplersche Schlierenverfahren, Forschungsheft V.D.I., 367,
Ausgabe B, Band 5 (.July-August, 1934).
5. ZOBEL, TH. Development and Construction of an Interferometer for Optical
Measurements of Density Fields, NACA Tech. Memo., No. 1184 (1947).
6. B ARNES, N. F., and B ELLINGER , S. L. Schlieren and Shadowgraph Equipment
for Air Flow Analysis, Jour. Opt. Soc. Amer., Val. 35, No. 8 (1945), p. 497.
7. L ADENBURG , R., W INCKLER, J., and V AN VOORHIS,C. C. Interferametric
Studies of Faster Than Sound Phenomena, Phys. Rev., Vol. 73, No. 11 (1948),
p. 1359.
8. ASHKENAS, H. I., and BRYSON, A. E. Design and Performance of a Simple Inter-
ferometer for Wind-Tunnel Measurements, Jour. Aero. Sci., Val. 18, No. 2
(1951). n 82.
Art. 3.7 PROBLEMS 69
9. DEFRATE, L. A., BARRY, F. W., and BAILEY, D. Z. A Portable Mach-Zehnder
Interferometer, Meteor Report, No. 51, M . I. T., Cambridge, Mass. (1950).
10. SCHARDIN, H. Theorie und Anwendung des Mach-Zehnderschen Interferenz-
Refraktometers, Zeilschrift jur Instrumentenkunde, Vol. 53 (1933), pp. 396,
424. Also see R.A.E. Translation, No. 79.
11. LIEPMANN, H. W.,and PUCKEIT, A. E. Introduction to Aerodynamics of a Com-
pressible Fluid. New York: John Wiley & Sons, Inc., 1947.
12. PANKHURST, R. C., and HOLDER, D. W. Wind- Tunnel Technique. London:
Sir Isaac Pitman & Sons, 1952.

PROBLEMS
3.1. The compressibility of a liquid is usually expressed in terms of the bulk
modulus of compression,
8 = P-d p
dp
Show that

3.2. Calculate the velocity of sound a t 70°F in the following media:

(a) Air, (b) hydrogen, (c) uranium hexafluoride, (d) mercury vapor, (e)
water vapor, (f) liquid water a t 14.7 psia.
3.3. To what pressure must liquid water be compressed in order that it
leave a nozzle a t atmospheric pressure with a jet velocity equal to the sonic
velocity? Assume a constant bulk modulus of compression of 300,000 psi.
3.4. Plot and compare curves of the sonic velocity in air versus absolute
temperature (a semi-log plot is suggested) using
(a) a value of k of 1.4 for all temperatures, and (b) the actual value of k
for each temperature.
3.6. A projectile in flight carries with i t a more or less conical-shaped shock
front. From physical reasoning it appears that a t great distances from the
projectile this shock wave becomes truly conical and changes in velocity and
density across the shock become vanishingly small.
Photographs of a bullet in flight show that a t a great distance from the bullet
the total included angle of the cone is 50.3". The pressure and temperature of
the undisturbed air are 14.62 psia and 73"F, respectively.
Calculate the velocity of the bullet, in ft/sec, and the Mach Number of the
bullet relative to the undisturbed air.
3.6. Show that, for a perfect gas, the fractional change in pressure across a
small pressure pulse is given by
d p - dV
- - k-
P c
and that the fractional change in absolute temperature is given by
70 INTRODUCTORY CONCEPTS TO COMPRESSIBLE FLOW Ch. 3
3.7. (a) Demonstrate that a compression wave (i.e., a pressure pulse which
increases the density of the fluid over which it passes) which moves rightward
imparts a rightward velocity to the fluid.
(b) Derive similar rules for a rightward-moving rarefaction wave and for
leftward-moving compression and rarefaction waves.
(c) The rightward-moving compression wave of Fig. 3.la strikes a stationary
wall closing the right-hand end of the duct. Demonstrate that it is necessary
for a reflected wave to travel leftward, and determine whether the reflected
wave is a compression or rarefaction. Compare the pressure change across the
incident wave with that across the reflected wave.
3.8. A compression pulse changes the velocity of the fluid over which it
passes by 10 ft/sec. Calculate the pressure rise (psi) across the pulse for (a)
water, and (b) air a t 14.7 psia and 70°F.