One-dimensional gas dynamics

The flow in a channel or tube which may be described by specifying the variation of the cross-
sectional area (A) along its axis (x), A  A x  and in which the flow properties are uniform over

each cross-section  p  p  x  ,     x  , etc. can be considered as one-dimensional flow. The

flow quantities can also be time dependent [ u  u  x, t , p  p  x, t  ], i.e., the flow can be unsteady.
If there are sections over which the flow conditions are not uniform it is still possible to apply the
results between sections where they are uniform, i.e., one-dimensional. At non-uniform stations,
the results applied to suitable mean values. Furthermore, the one-dimensional results are
applicable to the individual stream tubes of a general 3-D motion; x being along the stream tube.

For an incompressible flow, complete information about a one-dimensional flow is obtained from
the kinematic relation: “u is inversely proportional to A”. The pressure is then obtained from the
Bernoulli’s equation. For a compressible flow the relation between velocity and area also depends
on density variation since the governing equations are interdependent.

1-D continuity equation

If the flow is unsteady then the mass contained between sections 1 and 2, x distance apart,

changes at the rate Ax  where Ax is the mass contained. The rate of change must be
t
equal to the flow through 1 minus flow through 2, i.e., the net inflow
 
  Ax      uA x
t x

or A   uA  0
t x

If the flow in the tube is steady, the continuity equation is

d
  uA  0
dx
This implies that the mass of fluid that passes a given section must pass all the other sections
downstream. At any two sections where conditions are uniform 1u1 A1   2u2 A2 .
This equation is general, since it holds even if the conditions between the sections are not
uniform. If the flow is uniform at every section, the equation can be written as
uA  m  Constant

Euler’s equation or Momentum equation

u u 1 p
u 
t x  x
for steady flow, the first term is zero, and the derivatives become total derivatives.
dp
u du  0

1 2 dp
or u   constant
2 

It is often convenient to express Euler’s equation in an alternative form that describes the changes
in momentum of the fluid within a fixed control space.
Multiplying Euler’s equation by A and the continuity equation by u

u u p
A  uA   A
t x x

u A  u  uA  0
t x
Adding the two yields the one-dimensional momentum equation

uA   u 2 A   A p     pA  p A
 
t x x x x
Integrating this between any two sections gives
 2

t 1
  
uAdx   2 u 2 2 A2  1u12 A1   p1 A1  p2 A2    pdA
1
2

The first integral is the momentum of the fluid enclosed between 1 and 2 and the last integral may
be evaluated by defining a mean pressure p m .


 2
t 1
 
uAdx   2u 2 2 A2  1u12 A1   p1 A1  p2 A2   pm  A2  A1 

Unsteady Transport or flux of momentum Force in the x direction due to

change in space into the space through end pressures on the end sections
sections and on the walls

The integral form of the momentum equations is more general, since it is valid even when there
are dissipative processes within the control space, provided that the reference sections are
equilibrium states. During the integration of the differential momentum equation the forces on
adjacent internal faces cancel as they are equal and opposite and only the forces and the fluxes
at the boundaries of the control space are left out. If there is a non-equilibrium region inside this
space, it does not affect the integrated result.

For steady flow in a duct of constant area, the momentum equation becomes

 2 u 2 2  1u1 2  p1  p 2

Energy equation

For a fluid flow problem the basic thermodynamic quantity is the enthalpy, rather than internal
energy due to the presence of flow work. In adiabatic flow through a resistance the total enthalpy
per unit mass upstream and downstream of the resistance is the same.

Let’s select a definite portion of the flowing fluid, between sections 1 and 2 for the ‘system’.

During a small time interval in which the fluid is displaced to a

region bounded by sections 1´ and 2´, a quantity of heat, q, is
added. According to the first law,
q + work done = increase in energy

Assuming the volume displaced at 1 is the specific volume 1 corresponding to a unit mass, then

for steady conditions the displacement at 2 is also for unit mass with specific volume 2 . The work

done on the system during this displacement is p1 1  p 2 2 . The local energy of the system is

1
e  u 2 per unit mass. Comparing the energy of the system after the displacement with that
2
before, the net increase in energy is

 1 2  1 2
 e2  u 2    e1  u1 
 2   2 
Hence, the steady flow energy equation is

 1 2  1 2
q  p1 1  p2 2   e2  u 2    e1  u1 
 2   2 
1 2 1 2
or q  h2  h1  u 2  u 1
2 2
And, the adiabatic flow energy equation becomes
1 2 1 2
h2  u 2  h1  u1
2 2
These equations relate conditions at two equilibrium states. They are valid even if there are
viscous stresses, heat transfer, or other non-equilibrium conditions between the two sections
provided sections1 and 2 are equilibrium states.

If equilibrium exists all along, the equilibrium equation is valid everywhere and may be written as
1
h  u 2  constant
2
or dh  udu  0
For a thermally perfect gas this becomes C p dT  udu  0

1 2
And for a thermally and calorically perfect gas the equation becomes C pT  u  constant
2
At a place where u = 0 and the fluid is in equilibrium
1
h  u 2  h0  constant
2
h0 is called reservoir or stagnation enthalpy, the enthalpy of the fluid in a large reservoir where
velocity is practically zero.

If there is no heat addition to the flow between two reservoirs, then the enthalpy of both the
reservoirs is same, h0 . Since h0  C pT0 for a perfect gas, the stagnation temperatures in the two

reservoirs are also same.

 h0  h0 , T0  T0

From second law of thermodynamics s0  s0

T0
s0  s0  R ln
p0
Since for a perfect gas  C p ln 0
p 0/ T0

po
 1
po
 Downstream total pressure must be less than upstream total pressure
1
This is true for any gas, since it follows from the definition of entropy, Tds  dh  dp ,

 An increase in entropy, at constant stagnation enthalpy, will be associated with a decrease of
stagnation pressure.

The increase of entropy, and the corresponding decrease of stagnation pressure, represents an
irreversible process. Entropy is being produced in the flow between the reservoirs. The flow is not
in equilibrium throughout. Only if the flow is in equilibrium throughout, entropy will not be produced

and the flow will be isentropic. Only in such isentropic flow s0  s0 , p0  p0

The reservoir conditions or stagnation conditions are also called total conditions. The terms are
used to define conditions at any point in the flow. The total conditions at any point in the flow are
the conditions that would be attained if the flow there were brought to rest isentropically. For
stagnation conditions to exist it is not enough that the velocity be zero, it is also necessary that
equilibrium conditions exist.

Since the imaginary local stagnation process is isentropic, the total entropy at any point is by

definition equal to local static entropy  so  s . Since To  To , then the local entropy for a perfect

gas is related to total pressure by

po
s  s   R ln
po

A flow which is in equilibrium and adiabatic is isentropic. For adiabatic, non-conducting flow the
energy equation
dh  udu  0 applies all along the flow

Similarly, in the absence of friction forces, Euler’s equation

udu  dp   0 is applicable everywhere

 dh  dp   0

1 dp 
 ds   dh    0 or s = constant along the flow
T  
Thus an adiabatic, non-conducting, inviscid flow is isentropic. In this case, either the momentum
or the energy equation can be replaced by the equation
s = constant
dp
For a perfect gas, using dh   0 , this condition may be written as

 

  T 
 1
p
po   o   To 

The conditions of equilibrium cannot be strictly attained in a real, non-uniform flow, since a fluid
particle must adjust itself continuously to the new conditions that it encounters. Thus, entropy
production is never strictly zero.
Speed of sound & Mach number

Speed of sound is the speed at which small disturbances or waves are propagated through a
compressible fluid or an elastic medium in general. Its relation to the compressibility of the fluid is
given by

 p  1   
a2    , isentropic compressibility  s 
   s   p  s
1 Ks
  , K s is isentropic bulk modulus
 s 

The disturbances (the temperature and velocity gradients) produced in a fluid by a sound wave
are so small that each fluid particle undergoes a nearly isentropic process. In a perfect gas
p  const.  
p
 a2   RT

In a flowing fluid, the speed of sound is a significant measure of the effects of compressibility
when it is compared to the speed of the flow. This introduces the dimensionless parameter called
u
Mach number M 
a
M will vary from point to point in a flow because of change in u and a . In an adiabatic flow an
increase in u always corresponds to an increase in M. A flow is called subsonic if M < 1 and it is
called supersonic if M > 1.

Area-velocity Relation

For a steady adiabatic flow in a stream tube of varying area the continuity equation is
d du dA
  0
 u A
For incompressible flow d  0 and this gives the simple result that increase or decrease of
velocity is proportional to decrease or increase of area. The change in density modifies this simple
relation. Using Euler’s equation for steady flow
dp dp d
u du   
 d 

dp  dp 
Since, adiabatic, inviscid flow is isentropic  
d  d  s
 d
 u du  a 2

d du
or  M 2
 u
 At very low Mach numbers the density changes are so small compared to the velocity
changes, that they may be neglected in flow computation and it may be considered that  =

constant. Hence, equivalent definitions of incompressible flow are a   or M  0 .

The continuity equation now becomes
du du dA
M2   0
u u A
du  dA A
or 
u 1 M 2
(1) At M = 0, a decrease in area gives a proportional increase in velocity
(2) For 0 < M < 1 (subsonic speeds), the relation is qualitatively the same as for incompressible
flow, a decrease of area giving an increase in velocity, but the effect on the velocity is
relatively greater.
(3) At supersonic speeds, the right hand side becomes positive implying an increase of speed is
associated with an increase of area. This is due to the fact that at supersonic speeds the
density decreases faster than the velocity increases, so that the area must increase to
maintain the continuity of mass.

du dA
Form the equation, at M = 1, can be finite only if  0 . Considering a tube in which the
u A
velocity increases continuously from zero to supersonic speeds, it is obvious from above that the
tube must converge in the subsonic portion and diverge in the supersonic portion. Hence, the
area must be minimum at M = 1, – sonic condition is reached only at the ‘throat’. The same logic
holds when the velocity decreases continuously from supersonic to subsonic. Hence M = 1 can be
attained only at a throat of the tube. The inverse is not true, that is, M is not necessarily 1 at a
throat. Since the equation shows that a throat corresponds to du  0 ; the velocity attains a
maximum or a minimum there, depending on whether the flow is subsonic or supersonic. Near M
= 1, the flow is very sensitive to changes in the area, since the denominator (1 – M2) is very small.

Some Important Relations

In adiabatic flow the energy equation for a thermally and calorically perfect gas is
1 2
u  C pT  C pTo
2
  1
Now, a 2  RT  C p 1  T    1C pT
  
a2
or C pT 
 1
2
u2 a2 a
   o
2  1  1
 1
Multiplying by
a2
2
 1 ao T
M 2 1  2
 o
2 a T
To  1 2
 1 M
T 2
Hence, the isentropic relations become
  1
po   1 2 
 1
p 
M 
2 

1  1
o   1 2 
 1  M 
  2 

In the above equations, the values of To and ao are constant throughout the flow and can be taken

as the actual reservoir value. The values of po and  o are the local reservoir values. They are

constant throughout only if the flow is isentropic.

Instead of the reservoir, any other point in the flow can be used for evaluating the constant in the
energy equation. The throat, where M = 1, is a very useful point. The flow variables at the throat
are called sonic and are denoted by superscript *. The flow speed and sound speed are u 

and a  . At sonic condition M = 1 and u   a  . The energy equation then gives

1   1 2 a o
2
u2 a2 u 2 a 2
    a 
2  1 2  1 2  1  1
2
a 2 T
  
ao
2
  1 To
Thus for a given fluid the sonic and the reservoir temperatures are in a fixed ratio, so that T* is
constant throughout in an adiabatic flow.
 T a
For air  0.833,  0.913
To ao
Using the isentropic relations with M = 1,

p   2   1
   0.528
po    1 

  2   1

   0.643
 o    1 
It is not necessary that a throat actually exists in the flow for sonic values to be used as reference.

The speed ratio u  M  is a convenient quantity in many situations.

a
u2 a2 1   1 2
Using   a
2  1 2  1
1 1 1  1 1
Or,  
2   1M 2
2   1 M 2

2    1M 2  1
 
  1M 2
  1M 2
 1
M2
or M 2

  1M 2
 2
2    1M 2 1
 1
M2
2
 1

2
  1
M2

Alternatively

M 2 2
M  2

  1   1 2   1
 M    1
2 2 M 2
 M* < 1 for M < 1, and M* > 1 for M > 1

Using p  RT and eliminating T from the energy equation for adiabatic steady flow,

u2  p  po
 
2   1    1 o

p po
For isentropic conditions, 
 o
  1
 1
p    po  p  
 p  o    
 o o o  po 
 
Hence, the energy equation becomes
  1 
u2  po  p      po
   
2   1  o  po   1 o
This is the steady state Bernoulli’s equation for an adiabatic compressible flow.
1 2
In a compressible flow, the dynamic pressure u (used for normalizing pressure and forces) is
2
not simply the difference between stagnation and static pressure. It depends on Mach number as
well as static pressure.
1 2 1 1  p 2 1
u   M 2 a2   M   pM 2
2 2 2  2

p  p p  p 2  p 
Hence, C p    2 
 1
1
U 2
1
 p M  2  M   p 
2 2
For isentropic flow, this becomes

 
Cp 
2   2     1 M 
2
  1

 1
 2 
 M 2   2     1 M  
 
U 2 u2
M  and U  are reference quantities, M  2  , M 2

a 2 a2
M or a can be eliminated from the above definition using the energy equation in the form
u2 a2 U2 a2
   
2  1 2  1

 
2    1 2  u 2
   1

 Cp  2 
1 M  1  2    1
 M   2  U   
 

Flow through a Constant Area Duct and Normal Shock

The inviscid incompressible flow through a uniform duct has only one possible solution, which is
uniform flow. However, there are two possible solutions when the flow is compressible. If there is
no change in entropy anywhere, the only possible solution is the uniform flow. An alternative
solution which contains a ‘jump’ in the parameters is also possible when there is a change in
entropy or a non-equilibrium region between the two stations.
 (1) (2) (1) (2)
Non-equilibrium region Non-equilibrium region (shock)

Considering two sections (1) and (2) where equilibrium exists but which may contain non-
equilibrium region between them, all the conservation laws apply, and hence
1u1   2 u 2
p1  1u1  p2   2 u 2
2 2

1 2 1 2
h1  u1  h2  u 2
2 2
There is no restriction on the size or details of the dissipative region as long as the reference
sections are outside it. The non-equilibrium region may be idealized by a vanishingly thin region,
across which the flow parameters may ‘jump’. The control sections (1) & (2) may be brought
arbitrarily close to it. Such a discontinuity is called ‘shock wave’. A real fluid cannot have an actual
discontinuity and this is just an idealization of the very high gradients that actually occur in a
shock wave. These severe gradients produce viscous stress and heat transfer (non-equilibrium
conditions) inside the shock.

The three conservation statements are the general equations for a steady, inviscid adiabatic flow
and hence, for a normal shock. For a thermally and calorically perfect gas, the equations can be
solved explicitly in terms of the Mach number M1. The results always apply locally to the
conditions on either side of a shock, provided it is normal to the streamline.

Both sides of the momentum equation divided by the appropriate side of the continuity equation

p1  1u1 p   2u2
2 2
 2
1u1  2u2
2 2
p2 p a a
or u1  u 2   1  2  1
 2 u 2 1u1  u 2  u1
Using the energy equation for a perfect gas, in terms of sound speed

u12 a12 u2 a2 1   1 2
  2  2  a
2  1 2  1 2  1
  u1u 2  a 2
This is known as Prandtl or Meyer relation.

The equation can be stated, in terms of the speed ratio, as M 2  1



M1

M    1 corresponds to M   1 . Thus, the velocity change across a normal shock must be from
supersonic to subsonic or vice-versa. There is nothing in the relation to exclude possibility.
However, from the physical ground, it can be anticipated that accelerating jump is not possible,
since increase in velocity is unlikely when dissipative processes are in action.

Using M 2

  1M 2
the above relation can be written as
2    1M 2

  1 M 2 2 
2     1 M 12
2     1 M 2 2   1 M12

 1 2
1 M1
2  M 2  f M 1 
2
 M2 
 1
 M 12 
2
If M 1  1, M 2  1  infinitely weak normal shock or a March wave
M1  1  M 2  1
As M 1 increases above 1, the normal shock wave becomes stronger and M 2 becomes
progressively smaller.
 1 2
1 M1
2 2  1
As M 1  , M 2  Lim 
M 1   1 2
 M 12 
2
M 2  0.378 for air with   1.4

The velocity ratio is simply

u1 u12 u2
  12  M 12
u2 u1u2 a
Using the continuity equation

 2 u1

2
 M1 
  1M 1 2
1 u 2 2    1M 1
2

p 2  p1  1u1   2 u 2  1u1   2 u1u 2  1u1 u1  u 2 

2 2 2
p 2  p1 1u1  u2 
2
 1  
p1 p1  u1 
2
p1 a1 u2
Using  , and relation for
1  u1
p 2  p1  p1
p1

p1

2
 1
2
M1 1  
p1 is generally used to define the shock strength
p1
2
p2
p1
 1
 1
 M 12  1

T2 p 2 1
The temperature ratio 
T1 p1  2

2  1  M 1  1
 
2 2
T2 a 2 2 h
Also,  2  1 M1 1  2
T1 a1   1 M 1
2 2
h1

The limiting values for air with   1.4

 1
Lim M 2   0.378
M 1  2

2  1
Lim   6
M 1  1  1

p2 T2
Lim  Lim  
M 1  p1 M 1  T
1

All the ratios are functions of upstream Mach number for a calorically perfect gas.
The change in entropy across the shock is given by

 1

T2 p2   p2   1  
 2 1 

s2  s1  C p ln  R ln  R ln    
T1 p1  p1   1  
 
1 

s2  s1  2   1     1 M 12   1
or  ln 1   M 12  1   
  1     1 M 1  2 
2
R 
Since s 2  s1 in an adiabatic process, the condition can only be satisfied if M 1  1 . If M 1  1 , then

s 2  s1  0 , which is impossible according to second law of thermodynamics. Hence, supersonic

to subsonic jump is the only possibility across a normal shock.

For a weak shock, M 12  1 is small and the change in entropy can be approximated as
 2  M 1  1
3 3
2
s2  s1   1   p1 
   
  1 12 2  p1 
2
R 3

 p1
Hence, the increase in entropy is third order in shock strength M 12  1 or   p1
. A small change

in pressure, that is associated with first order change in velocity, density and temperature,
produces third order change in entropy. A weak shock produces nearly isentropic change of state.
Since the flow across a shock is adiabatic, T02  T01 . Hence,

s2  s1 p  po1    po1 
 ln o1  ln     ln 1  
R po 2  po1  po1   po1 

The expression for the total pressure ratio is obtained from that of s
R
1 
    1 M 12   1

po 2  2   1
 1  
M 12  1    
po1    1     1 M 1  2 
2

 po1
For a weak shock  1 hence
po1

2  M 1  1
3
2
s2  s1  po1
 
  1
2
R po1 3

Change in entropy is directly proportional to change in total pressure. Hence, change in total
pressure is also of 3rd order in shock strength.

The shock jump relations (Rankine-Hugoniot relations) are often expressed in terms of pressure
jump instead of upstream Mach number.

Moving Shock Wave Problem

The normal shock wave relations are established above assuming the shock to be stationary and
the fluid flows through it with speed u1. Alternatively, the shock may be taken as propagating
through the fluid with speed u1.

ub us (2) (1)
(a) u2 u1
u(b)
p us u2 u1
Consider the fluid in front of the shock wave is at rest and the wave is moving into it with the
speed us . The shock causes the fluid behind the wave to move with speed ub . The problem is

solved by using a transformation of the reference system. With respect to an observer moving
with the shock the fluid ahead of it is moving with speed u1 (in opposite direction) and the fluid

behind is moving with speed u2 , where

u1  us , and

u2  us  ub  u1  ub , or ub  u1  u2

The static densities, pressures, and temperatures on either side of the shock are not affected by
the transformation. The shock jump relations for these quantities obtained above apply here also.
Hence, a shock wave propagating into a stationary fluid sets it into motion and raises its pressure,
density and temperature.

The ‘jump’ relations across the shock may be rewritten in terms of us and ub by using the

transformation. The Mach number of the shock is M 1 

us
a1
, 
a12  p
d 
1
.

For practical usefulness, all other quantities are expressed in terms of the pressure ratio  2 
p
 p1 
The shock velocity for a perfect gas is
1
   1   1 p2  2
us  M 1a1  a1    
 2 2  p1 
The density ratio and temperature ratio are given by the Rankine-Hugoniot relations.
  1 p2
1
2   1 p1
u
  1
1   1 p2 u2

  1 p1

  1 p2

T2 p 2   1 p1

T1 p1   1 p2
1
  1 p1

The fluid velocity behind the shock is ub  u1  u2  us  1  2 

u
 u1 
 1
 2  2

a p     1 

 ub  1  2  1  
  p1   p2    1 
 p1   1 
The total quantities in the two systems are not the same. It should be noted, for example, that
1
h0 a  ha  h01 , h0b  hb  ub2  h02 and h0b  h0 a .
2
1 2 1
However, h02  h2  u2  h1  u12  h01.
2 2

A weak shock is the one for which the normalized pressure jump is very small,

 p p2  p1
  1
p1 p1

The other disturbances are then also small and can be obtained by expanding the above

equations in series and retaining only the first order terms in

p .
p1
 1  p ub
This gives  
1  p1 a1

 T  1  p    1 p 
 , us  a1  1  
T1  p1  4  p1 

 The speed of very weak shocks is nearly equal to a1 .

For a very strong shock the pressure ratio

p2 is very large. In this case,
p1
2  1

1  1

T2   1 p2

T1   1 p1
1
   1 p2  2
us  a1  
 2 p1 

2 p2
ub  a1
    1 p1
One-dimensional wave motion

Disturbances created in a fluid by a moving body are propagated or communicated to other parts
of the fluid. The motion of the disturbances relative to the fluid is called wave motion, and the
speed of propagation is called the wave speed. Through this mechanism, various parts of a
moving body interact with the fluid and with each other and the forces on the body are
established. The problem is unsteady and appropriate equations need to be solved. Considering
adiabatic, non-viscous motion in a constant area duct,
 u 
Continuity equation:  u 0
t x x
u u 1 p
Euler/Momentum equation: u  0
t x  x

Since friction is neglected and external heat addition is excluded, the isentropic conditions exist

p  
 p  p   , for a perfect gas  
p1  1 

p p  
   a2
x  x x
The disturbances or perturbations are defined relative to the fluid that is at rest,
with u  0,   1 . Assume the perturbed values are u and  . The perturbations are not
necessarily small.
Define   1 1  s 

The dimensionless quantity s 

   1 
1 is called condensation.
With these definitions, the equations of motion may be written
s  u u  s
1  1   s   1u  0
t  x x  x

u u a 2 s
u  0
t x 1  s x


    1  s 
p 
For a perfect gas
 1 
p1

 1  s 
T  1
T1
These equations are exact for frictionless, non-conducting motion, but not easily integrable as
they are non-linear. The equations can be linearized through a small disturbance assumption.
 u u u
Assuming s  1, s  s can be neglected
x x x
s u
The terms u and u are of the same order and may be neglected.
x x
dp  p    2 p   d2 p 
a2   
         a 2
 1 
s 2 
d    1   2 1  d  1
1 1

a 2 s 2 s
Hence, can be approximated to a1
1  s x x
The approximate linearized equations are then
s u
 0
t x

u s
 a12 0
t x
These equations are called acoustic equations because the disturbances due to sound wave are
very small.
The isentropic relations for a perfect gas can be approximated as

 1    1s
p T
 1  s
p1 T1
Differentiating the linearized continuity and momentum equations with respect to t and x
respectively

2s  2u 
 0 
t 2
 xt  2s 2 s
   a1 0
 u
2
 s
2
t 2
x 2
 a1 2  0
 xt x 

 2u 2  u
2
Similarly,  a1 0
t 2 x 2
Both the disturbances u, s  satisfy the wave equation. The disturbance propagates with a definite

signal velocity or wave velocity a1 .

The general solutions of the equations are
s  F  x  a1t   G  x  a1t 

u  f  x  a1t   g  x  a1t  , f  a1 F

g   a1G
To analyze the characters of the solution, let’s take G  0 .
So, the density distribution at time t is
s  F  x  a1t 
This represents a disturbance or wave which at time t = 0 had the shape
s  F x 
and which at time t has exactly the same shape, but with corresponding points displaced a
distance a1t to the right. The velocity of each point in the wave and hence of the wave is a1 .
t dx
 a1
dt

t2

t1

x
s  F  x

A wave in which the propagation velocity is in one direction is called a simple wave. Similarly, the
wave described by s  G  x  a1t  is a simple wave propagating to the left with speed a1 .

dx t
 a1
dt

x
s  G  x

The lines in the x  t plane which trace the progress of the waves, i.e., the lines with

slope dx   a1 , are called the characteristics of the wave equation.

dt

The disturbances propagate through the fluid with the speed a 

dp and the quantity is
d
called the speed of sound or acoustic speed. The result is applicable to disturbances in which

velocity and temperature gradients are very small and u  1 so that dissipative forces have no
a1
considerable effects. It implies that motion in a sound wave is isentropic, and

 p 
a 2   
   s
The amplitude of ordinary audible sound is small and the local production of entropy is negligible.
Friction and local entropy production is negligible for computing the speed of ordinary sound, but
the cumulative effect on the amplitude is not negligible.
The quantity a 2 provides a pressure-density relation and eliminates pressure from the momentum

 p  
equation   a2  . When non-isentropic processes are present pressure depends on
 x x 
entropy also, and
p   p  s
 a2  
x x  s   x

p
a 2 can be evaluated from the equation of state. For a perfect gas a 2    RT


Pressure and particle velocity in a sound wave

The pressure disturbance accompanying the density wave for a perfect gas is
p p  p1
 1  s   s
p1 p1
 Pressure wave has the same shape as the density wave, differing by a constant factor  .

As the wave progresses through the fluid, the pressure disturbance sets the fluid in motion, giving
it a velocity u which is called the particle velocity and, in general, is much smaller than the wave
speed a1 .

A simple wave s  F  x  a1t  , propagating to the right, produces a velocity disturbance

u  a1 F  x  a1t   a1 s

In a left ward propagating wave u  a1G  x  a1t    a1 s . The various parts of the wave are called

condensation and rarefaction depending on whether the density is higher or lower than the
undisturbed density 1 .

The effect that the wave produces on the fluid depends on the gradient of this density and
pressure distribution and on the direction of motion of the wave. Thus the portion of the wave that
increases the density as it passes is called a compression and that which decreases the density is
called an expansion. The corresponding distributions of particle velocity are
u   a1 s
for left and right propagating waves respectively. It may be seen that a compression accelerates
the fluid in the direction of wave motion, whereas an expansion decelerates it. The non-simple
wave is a superposition of two simple waves and the relation between particle velocity and density
u F G
is  .
a1s F  G
In the limit of vanishingly small disturbances the perturbation quantities may be written in

differential form with u and s replaced by du and d

1 . The relations then become
d 
du   a1
1   p1
  dp  d    1a1du
dp d  1

p1 1 

‘Linearized’ Shock Tube

Shock tube is a simple device consisting of a tube that is divided by a membrane or diaphragm
into two chambers in which pressures are different. A wave motion is set up when the diaphragm
is suddenly removed or broken. If the pressure difference is very small, the wave motion may be
approximated as isentropic and can be described by the acoustic equations. The shock tube in
such a case is called acoustic or linearized.

At t = 0, immediately after the membrane is removed the wave has a shape of ‘step’ distribution.
The particle velocity at this instant is ‘zero’ everywhere. The wave at t = 0 is described by

 s  x  0
s  x, 0   F  x   G  x   s0  x    4
0 x0

u  x,0   a1 F  x   a1G  x   0

1
1  s x0
Simultaneous solution gives F  x   G  x   s0  x   2 4
2 0 x0
Hence, the motion at any subsequent time is given by

 s4 x  a1t
1
1 1 
s  x, t   s0  x  a1t   s0  x  a1t    s4  a1t  x  a1t
2 2 2
0 x   a1t
 0 x  a1t
 1
1 1 
u  x, t   a1s0  x  a1t   a1s0  x  a1t    a1s4  a1t  x  a1t
2 2  2
0 x  a1t

s  s4
1
s2  s4
2
s 0

a1 a1

t = t1

A compression wave is propagated to low pressure side and an expansion wave of equal strength
to the high-pressure side.

Finite Amplitude Isentropic waves

The simple acoustic waves have constant wave velocity and permanent shape. These properties
are due to the linearization of the equations achieved through the assumption of infinitesimal
amplitudes and gradients. When such an assumption is not possible, the conditions at a given
point in the wave cannot be approximated by those in the undisturbed fluid. The wave velocity
varies from point to point and the simple wave becomes distorted as it propagates. To describe
such finite disturbance, the complete non-linear equations need to be solved.

Consider a right moving plane compression wave. At two adjacent points, as shown in the figure
below, the fluid properties differ in magnitude by du , dp, d  , da, etc. The respective parts of the

wave passing through these same points differ in wave speed by the amount dc . Since a finite
wave may be thought of as a succession of infinitesimal pressure pulses, each element of the
wave may be analyzed as an acoustic wave. As long as the velocity and temperature gradients
are moderate, the viscous and heat conduction effects are negligible. Hence, each part of the
wave travels at the local speed of sound with respect to the fluid in which it is propagating. The
propagation velocity of a part of the wave with respect to fixed coordinates is then,
c ua
The propagation velocity of an adjacent part of the wave is
c  dc  u  du  a  da
 dc  du  da

dc du da
  
dp dp dp

du 1 dc da 1
Now for a right ward wave as seen above  , hence  
dp  a dp dp  a

Since the entire fluid was originally at rest with uniform pressure and temperature and each
particle of fluid undergoes isentropic changes, the increments in pressure and density between
adjacent fluid-particles obey

d  dp 
dp da d  dp  d  d  dp  da 1 d   d  
a 
2
 2a      
d dp dp  d   dp d   d   dp 2a dp
d
  d  dp  
 
dc 1   d   d   
  1
dp  a  2 dp 
 d 
 s
 1
Replacing density by specific volume   
  
d d d 1 d d
   2   2
d d d  d d

d2 p 
 2
d  s
dc   2
 
dp 2a dp
 
d s

For a thermodynamically stable fluid,  dp d  s

must be negative. Hence, the isentrope must have

a negative slope on the p  diagram. Consequently, the sign of dc depends only on the sign
dp

d2 p 
of  2  , i.e., on whether the isentrope on the p  diagram is concave upwards or
 d s
concave downwards. The higher-pressure parts of the wave overtake the lower-pressure parts
dc
when is positive. Consequently, a compression wave steepens as it progresses and an
dp
dc
expansion wave flattens as it progresses. Opposite happens if is negative.
dp

d2p
(i) Compression waves steepen and expansion waves flatten when    0 , or the
2 
 d  s
isentrope is concave upward. This is the usual case for all real fluids.

 d2 p 
(ii) Compression waves flatten and expansion waves steepen if  2 
 0, or the
 d  s
isentrope is concave downward.

For a gas with straight line isentrope waves of finite amplitude propagate through the gas with
unchanged shape.

 d 2 p      1 p
In a perfect gas, p  is constant and  2 
  0 . Hence, compression waves
 d  s 2
steepen and expansion waves flatten in a perfect gas.
From the point of view of an observer moving with the local particle velocity the acoustic theory
applies locally. Relative to an observer moving with the local fluid velocity, the wave at that point

 
1

propagates with the local acoustic speed a 

dp 2
whereas relative to the fixed frame of
d
reference in the undisturbed fluid, it propagates with the speed c  a  u . Considering both left
and right moving waves, the local wave speed at any point is given by
c  a  u
The wave speed is no longer constant since a is a variable and u may no longer be neglected.
To evaluate these in terms of the density, the acoustic theory is applied locally. Using the

isentropic relation for a perfect gas

p 
p1
to eliminate p from a 2 
 p , gives
  
1 
 1

a  a1   
2

 1 
The particle velocity is evaluated in terms of the density by applying locally
 d 
du   a 
  
Substitute a by the above relation
 1
 
2a1    2
d 2
   1  

 u   a  a  a1 
1    1  1    1
 
 1
or, a  a1  u
2
 1
 c   a1  u
2
   1

   1    2 
or c  a1  1     1 
   1  1  
   

where a1 is the speed of sound in the undisturbed fluid and 1 is the density of undisturbed fluid.

Propagation of Finite waves

Consider the propagation of a simple wave of finite amplitude. Assume that the initial density
distribution is as shown in the figure t  t 0  0 
    1


  1   2  
For the right ward propagating wave c  a1 1     1 
 
   1  1   
 
Hence, the wave speed is higher than a1 in regions of condensation   1  and lower than a1
in regions of rarefaction. Thus the wave distorts as it propagates, the regions of higher
condensation tending to overtake those in regions of lower condensation. In regions of higher
condensation, the characteristic lines are inclined more, since the slope is inversely proportional
to the wave speed.

t3

t2

t1

x
t0 = 0 s  x ,0 
~

In terms of the compression and expansion regions the net effect is to steepen compression
regions and to flatten expansion regions in which the characteristic lines converge and diverge
respectively. In a compression region, the characteristics lines would eventually cross leading to
the situation t  t3 . But this is physically impossible since it implies three values of density at a

given point. Actually, well before this happens, the velocity and temperature gradients in the
compression regions become so large that friction and heat transfer effects become important.
These have a diffusive action which counteracts the steepening tendency. The two opposing
effects achieve a balance and the compression portion of the wave become ‘stationary’, in the
sense that it propagates without further distortion. It is then a shock wave.

In compression regions, the isentropic relations are valid until friction and heat transfer become
important. When a stationary balance between the diffusive and steepening (non-linear) terms
has been reached, the conditions across the wave front are given by the shock wave relations.
The intermediate unsteady, non-isentropic states can be treated only with the full unsteady
equations including viscous and heat transfer terms.

An expansion wave always remains isentropic as it tends to flatten and so reduce the velocity and
temperature gradients further. It never achieves stationary condition, corresponding to the fact
that there are no expansion shocks.
Centered Expansion wave

Consider a duct containing fluid enclosed by a piston. If the piston is withdrawn an expansion

wave is produced. If the piston starts impulsively, with speed u p , the distribution of particle

velocity in the first instant is a step. However, the expansion wave begins to flatten as soon as the
wave starts propagating. At some later time t1 the particle velocity has a linear distribution and

the pressure has a corresponding distribution.

p4

p3
u=0
u3 = - |up|

up a4
(3) (4)

t
  1 
Piston path x  c3 t   a 4  u p t
 2 
x = -|up| t

t1
(3)

x = a4 t
(4) Expansion path
x

The front of any isentropic wave propagates at the speed of sound of the undisturbed fluid. Thus,
the front of the wave propagates with speed a 4 into the undisturbed fluid that is in the opposite
direction of the piston and fluid motion. The wave speed in the portions of the wave behind the
front is given by
 1
c  a4  u
2
Wave speed c decreases continuously through the wave since u  0 . The fan of straight lines are
lines of constant c and thus of constant u and  . These lines are the characteristics. With
increasing time, the fan becomes wider and the wave becomes flatter and the gradients of
velocity, density, temperature become smaller. Thus the wave remains isentropic. The terminating

characteristic is given by x  c  a    1 u and slopes to right or left depending on

t 3 4 p
2
 1
whether a4  u p . Between the terminating characteristics and the piston, the fluid properties
2
have the uniform values  3 , p3 , T3 , etc. For a perfect gas they are given in terms of the isentropic

relations

2
 3    1 u p 
 1

 1
4  2 a4 
 
2

p3    1 u p 
 1

 1  
p4  2 a4 
 
This pressure ratio defines the strength of the expansion wave. The maximum expansion is
2a4
obtained when u p  giving  3  T3  0 . This implies that all the fluid energy is converted into
 1
kinetic energy of flow. If the piston velocity is higher than this limiting value, it produces no further
effect on the flow. The wave produced by an impulsive withdrawal of piston is called a centered
expansion wave after the fan like set of characteristic lines in the x  t plane.

Shock Tube:

p4
p1
The diaphragm at x  0 separates the low pressure chamber (1) from the high pressure chamber
(4). The chambers are also called respectively, the expansion and compression chamber. The
p4
characteristic parameter of a shock tube is the diaphragm pressure ratio . The two
p1

chambers may be at different temperature T1 and T4 and may contain different gases with gas

constants R1 and R4 .

When the diaphragm is burst  t  0  , the pressure distribution is ideally a step. This ‘splits’ into a

shock wave which propagates into the expansion chamber with speed us and an expansion wave

which propagates into the compression chamber with the speed a4 at its front. The condition of
the fluid which is traversed by the shock wave is denoted by (2) and that of the fluid traversed by
the expansion wave is denoted by (3). The interface between these two regions is called the
contact surface. It marks the boundary between the fluids which were initially on either side of the
diaphragm. The region behind the shock is different from the one behind the expansion and they
have different values of entropy. Neglecting diffusion, the fluids in the two regions do not mix, but
are permanently separated by the contact surface.

t = t1

us u2
p2

p1
T2
T4

T1 T3
On either side of the contact surface, the temperature and density may be different, but it is
necessary that the pressure and fluid velocity be the same,
p 2  p3 , u 2  u3

Thus u 2  u 3 must be the velocity of the contact surface. Using shock wave and expansion wave

relations

p  2 1
u2  a1  2  1
 p1   1  1 p2 p    1  1
1

 4 1
 2 4 
2a 4   p3  
u3  1  
 4  1   p 4  
 
Since u 2  u 3 and p 2  p3 , the basic shock tube relation is
2 4
   4 1
   4  1  a1 a   p2 p  1 
p4 p2 
 1  4  1  
p1 p1  
2 1 2 1    1  1  2  1 
p
  p 
 1

p2 p4
This gives the shock strength implicitly as a function of the diaphragm ratio . The
p1 p1
expansion strength is obtained from
p2
p3 p3 p1 p1
  since p3  p2
p4 p1 p4 p4
p1
The values of u and p calculated across the shock and expansion are identical. However,

 and T are usually different. The temperature T3 behind the expansion wave is given by the
isentropic relation
 4 1
 4 1  p2  4

T3  p3  4  p1 
   
T4  p 4   p 4 
 p1 

The temperature T2 behind the shock is given by the Rankine-Hugoniot relation

 1 1 p2
1
T2  1 1 p1

T1  1 p1
1 1
 1 1 p2
The duration of the flow is limited by the lengths of the expansion and compression chambers,
since the shock wave and expansion wave reflect from the ends of the chambers and eventually
interact with each other.

Practice Problems

1. Show that the local and reference free stream conditions are related by
  1 2  u  
2
T
 1 M 1    1 .
T1 2  U  
p 
Obtain corresponding relations for , . Obtain simple approximate relations for the case of
p1 1
small perturbations, i.e., .

2. Show that in flow from reservoir the maximum velocity that may be reached is given by
um2  2h0
2 2
 a for a perfect gas.
 1 0
What are the corresponding values of T and M? Interpret the results.

3. Show that the one-dimensional equations of motion may be written as

2  a a  u
 u a 0
  1  t x  x
u u 2a a
u  0
t x   1 x
Hence, or otherwise show that
   2a 
 t   u  a  x   u    1   0
 
   2a 
      0
  1 
 t u a u
x  
2a 2a
The quantities R1  u  and R2  u  are called Riemann invariants. Show that they
 1  1
dx
are constant on curves, which have slopes  u  a and u  a respectively.
dt

4. An airplane flies at an altitude of 13 Km where the atmospheric pressure is 124.4 mm Hg and

temperature is –56.50C. The flight speed is 650 Km/h. Calculate the critical velocity and
maximum possible velocity of air relative to the aircraft.

5. An airplane speed indicator calibrated without taking compressibility into account registers a
speed of 750 Km/h at an altitude of 6 Km. Determine the correct speed. Also find the
stagnation temperature and density at the nose of the Pitot tube.
6. Show that if two points in the flow have the same pressures and same stagnation temperatures,
but different stagnation pressures, the velocities are governed by the
 1
2  
 V2       1 2
 1
p
relation    1   01
 M1 .
 V1   p02   2
 

p2  p1 p  p2
7. Verify that for a normal shock  1 . This is a Rankine-Hugoniot relation. Show
 2  1 1   2
that the relation can be expressed as

1  1 1  1  p1 p2 
 p1  p2          0.
2 
 1  2    1  1  2 

8. A blast wave, initiated by an atomic bomb explosion, is traveling through air at standard
atmospheric conditions with a speed of 65 km/s. Estimate the changes in pressure, temperature,
velocity, stagnation pressure and temperature produced by the wave with respect to an observer
who is stationary with respect to the undisturbed air.

9. Consider a normal shock in a fluid, not a perfect gas, whose density depends only on the
dp
pressure according to the relation    , where  is a positive constant. Show that the
d
M2
upstream and downstream Mach numbers are related by M 12  M 22  ln 12 . Find expressions
M2
for  p2  p1  and u2 u1 in terms of M 1 and  .

10. Consider a rightward moving wave in a perfect gas. Show that, as one follows a part of the
wave corresponding to a particular fluid velocity and pressure, the time rate of change of wave
2
d  u    1  u 
steepness is given by     . Hence show that the time required for a given
dt  x  2  x 
2 u u
part of the wave to acquire infinite slope is given by t   , where is the initial
  1 x x
wave steepness.

11. A shock tube may be used as a short duration wind tunnel by utilizing the flow behind the
c 
shock wave. Show that in terms of the shock speed M s  s , the density ratio   2 and
a1 1
the conditions at the expansion chamber (1), the flow properties behind the shock in region (2)
are given by
 p2  1
 1   1M s2 1  
p1  
h2  1  1 
 1  1 M s2 1  2 
h1 2   
u2  1
 M s 1  
a1  
h02 cu  1
 1  s 2  1   1  1 M s2 1  
h1 h1  

u2
If shocks are weak enough so that the specific heat remains constant compute M 2  and
a2
show that M 2 has the limiting value 2 .
 1   1  1

12. Assuming the flow in a shock-tube to be one-dimensional, show that the shock speed Mach
number M s , the Mach number behind the shock M 2 and the Mach number behind the contact
surface M 3 are given respectively by the relations
1
c    1   1 p2  2
Ms  s   1  1 
a1  2 1 2 1 p1 
1  4 1 2 4
1p   p    1  1  1 p2   2  p4 p1  
2
M 2   2  1  2  1   ; M3     1
 1  p1   p1  2 1 2 1 p1    4  1  p2 p1  


13. Show that when a plane acoustic wave of strength p is reflected from a closed end of a tube
the pressure there becomes 2 p ; at an open end it is zero and so the reflected wave must be an
expansion wave.