FLUID MECHANICS 3 - LECTURE 4

ONE-DIMENSIONAL UNSTEADY GAS

Consider an unsteady 1-dimensional ideal gas flow. We assume that this flow is spatially
continuous and thermally isolated, hence, it is isentropic.

The governing equations are

Mass conservation: 
t   u x    x u  0
Equation of motion (Euler): 
t u  u x u  1 
x p0

dp
Since the flow is assumed isentropic we also have d
 a 2 and
s  const

 
x
dp
p  d x   a 2 x 
s  const

We introduce the function of gas density by the formula


ˆ a ˆ a
P(  )   d   P(  ) 
 
The corresponding function of time and spatial coordinate can be defined

P(t , x)  Pˆ[  (t , x)]

We will need derivatives:

t P(t , x)  Pˆ [  (t , x)] t  (t , x)  a t  (t , x)

x P(t , x)  Pˆ [  (t , x)]   (t , x)  a   (t , x)
x  x

Next, we have
   1 
x p  a2 x   x  p
a2 x
and
  1  
x P  a x   a x p  1 x p  a x P

Using the above relation, the equation of motion can be written as

 u u  u a  P  0 (*)
t x x
The mass conservation equation can be multiplied by a 

a 
 t   u a x   a  x u  0
P P
t x

which leads to its equivalent form


t P  u x P  a x u  0 (**)

Addition and subtraction of the equations (*) and (**) yield the system of equations

 (u  P)  (u  a)  (u  P)  0
t x
 (u  P)  (u  a)  (u  P)  0
t x

What is a physical meaning of these equations?

Consider the function f  f (t , x) and the line l : x  X (t ) . One can define the function

F (t ) : f l (t , x)  f [t , X (t )]

Let’s calculate the (ordinary) derivative of the function F. We have

F (t )  t f [t , X (t )]  x f [t , X (t )] X (t )

We see that if t f [t , X (t )]  x f [t , X (t )] X (t )  0 then the derivative F (t ) vanishes
identically meaning that the function f  f (t , x) is constant (or invariant) along the line l !

Looking at the obtained system of the differential equations we conclude that:

r  u  P is constant along each line C such that X (t )  (u  a)[t , X (t )]

s  u  P is constant along each line C such that X (t )  (u  a)[t , X (t )]

It remains to calculate the explicit form of the function P …

 
cp
Since the flow is isentropic we have the relation p  K ,  .
cv
 1
p
Thus a2     K   1  a   K  2

and
   1  1
a 1 2a
Pˆ    d   K 2
d  2
 1 K 2

a
 1
Hence, the explicit formulae for the conserved quantities are

r  u  21 a , s  u  21 a
Summarizing - in a 1D unsteady flow of the Clapeyron gas:
 The quantity r  u  21 a (the 1st Riemann invariant) is constant along each line such
that X (t )  (u  a)[t , X (t )]. We call these lines the C characteristics.
 The quantity s  u  21 a (the 2nd Riemann invariant) is constant along each line such
that X (t )  (u  a)[t , X (t )]. We call these lines the C characteristics.
What is the gain of knowing the shape of characteristic lines? Assume we know that the
characteristics, one C+ and the other C-, intersecting at a given point S (see figure) go through
the points A and B , respectively. We assume also that both velocity and the speed of sound is
known for the points A and B. Then, we can easily determine the velocity and the speed of
sound in the point S:

uS  k21 aS  u A  k21 a A uS  12 (u A  uB )  k11 (a A  aB )

  
uS  k 1 aS  uB  k 1 aB aS  2 (a A  aB )  4 (u A  uB )
2 2 1 k 1

 Note that the above solution is possible only when

the shape of the characteristic lines is a priori known,
which – in general – is not the case, because the slope
of these lines depends on unknown velocity and the
speed of sound!
 There exist a few special cases when the flow
problem can be solved in the closed form, though.
These cases concern the flows with simple waves.
 Conceptually, it is possible to device an
approximate method based on the use of
characteristics – see next figure.
For t  0 both velocity and the speed of sound are known at every point of the flow domain
(the initial conditions). We divide the spatial domain into short segments. Using the initial
conditions, the initial slopes of the characteristic lines of both kind can be found. Then, these
lines are advanced forward in time (as straight lines) until first intersections. Note that – in
general – the intersections do not correspond to te same time instant. The flow parameters at the
intersections can be computed using Riemann invariants from the “root” points.
This procedure can be continued to produce next layer of intersection. Having more layers, the
shape of the characteristics can be approximated by higher-order extrapolation method. In order
to avoid to large “deformations” of the “intersection grid”, the re-meshing procedure should be
used (interpolation of the flow parameter to some t  const layer).
Theorem: If in some region in the (x,t) plane one of the Riemann invariant is constant, then
the characteristics corresponding to the other invariant are the straight lines.

Proof:
Assume, without scarifying generality, that in some region the first Riemann invariant is
constant. It means that at every point in this region

r  u  21 a  const
Consider any characteristics C traversing the region of constant r . Then, at each point
belonging to this characteristics both Riemann invariant are constant, hence, the velocity u and
the speed of sound a are also constant. This, in turn, means that the slope of the characteristics
C , which is equal to u  a is constant, hence, this characteristics is the straight line.

Remarks:
 Any (x,t)-region such that one of the Riemann invariants is constant is called the simple
wave.
 If in a certain (x,t)-region both Riemann invariants are constant, then both velocity and the
speed of sound are constant in this region. In other words, such region corresponds to the
uniform steady flow.
Example 1: Expansion wave in a pipe.

C - red lines, C - blue lines

Calculation of the parameter in the region near the piston (behind the expansion wave)

Note that the second Riemann invariant s is globally constant and equal

s  (u  k21 a) t 0   k21 a0

Then, we can write

(u  k21 a) piston  U P  k21 aP   k21 a0


aP  a0  k21U P

Note that the sign of U P is negative, hence, aP  a0 .

When the acceleration phase shrinks to the point in time, the expansion wave take the focused
form …

The point ( x, t )  (0,0) is a singular point – the gas goes through all intermediate states in no
time.
For the focused wave we can solve the following problem: calculate the flow velocity and the
speed of sound at the duct section x  X and time t  T .

The point ( x, t )  ( X , T ) can be located:

X
 In front of the wave if  a0
T
X
 Inside the wave region if U P  aP   a0
T
X
 Between the piston and the rear of the wave if U P   U P  aP
T
X
 Outside the flow domain if  UP
T

The second option is the most interesting. In order to evaluate gas parameters we note that:

 u X  k21 aX   k21 a0 - the second Riemann invariant is global

X
 u X  aX  - the (constant!) slope of the C characteristics passing through the
T
( X , T ) can be expressed in two ways.
The solution is …

u X  k21 ( TX  a0 ) , aX  kk 11 TX  k21 a0

Finally, let us remind that the speed of the gas at the piston’s face is equal

aP  a0  k21U P (U P  0)

It follows that if U P  U max  k21 a0 then aP  0 which is physical nonsense. We conclude

that if the velocity of the piston exceeds the value of U max then the gas cannot catch up with the
piston and the vacuum mast appear between the piston and the front of the gas stream.

Note also that the maximal gas velocity in this (unsteady) motion is larger than the
maximal velocity of gas in the steady adiabatic motion, where U max
steady
 2 a
k 1 0
Representation of the expansion wave in the (u, a) plane (the hodograph plane).
Example 2: Expansion wave due to the initial pressure jump

Riemann invariant r is global.

The gas parameters at the
given point ( x, t )  ( X , T )
can be determined from the
following system of equations:

u  a  X
 T

u  k 1 a  k 1 a0
2 2

This solution is

u  k21 (a0  TX )


a  k 1 a0  k 1 T
2 k 1 X
Modification: outflow from the semi-infinite pipe.
In the hodograph plane …

If pext  p , the outflow is subsonic and pout  pext . From the isentropic relation (derive !):

aout  a0 ( pext p0 )( k 1) 2k

The velocity uout stems from the first Riemann invariant:

uout  k21 aout  k21 a0  uout  k21 a0[1  ( pext p0 )( k 1) 2k ]

Example 3: Compression wave generated by a moving piston – emergence of a
discontinuity.
Trajectory of the piston: x  X P (t )

In the region of continuous motion:

s  u  k21 a  const

Hence

X P (t )  k21 aP   k21 a0

aP  k21 X P  a0

The C line emerging from the piston’s trajectory point (T , X p (T )) is

x  X P (T )  [ a0  k21 X P (T )](t  T )
X P (T )aP (T )
Concept of an envelope line

Assume that a family of lines in the plane is defined by an implicit formula  (c, x, t )  0 ,
where c  C is the parameter identifying each line in the family. The envelope of this family is
the line with the parametric description t  Te (c), x  X e (c) , such that for each c  C :
 Each point in the envelope belong to the line from the family “labeled” with this c , i.e.,
[c, X e (c), Te (c)]  0
 The envelope is tangent to this line at their common point (see figure).

Tangency condition:

 (c)  [c, X e (c), Te (c)]  0


 (c)  c [..]  x [..]  X e (c)  t [..]  Te(c)
0!  tangency condition


  [c, X (c ), T (c )]  0
c e e
In our problem, c  T and
 (T , x, t )  x  X P (T )  [a0  k21 X P (T )](t  T )  0
  (T , x, t )   X  (T )  k 1
X P (T )(t  T )  [a0  k21 X P (T )] 
T P 2

 a0  k21 X P (T )  k21 X P (T )(t  T )  0

Hence, the parametric equations for the envelope line are:

 a0  k21 X P (T )
t (T )  T  k 1 X  (T )
 2 P

[
 x(T )  X (T )  0 2 Pa  k 1
X  (T )][ a0  k 1 
2 X P (T )]
 P k 1 
2 X P (T )

A discontinuity begins to developed at the intersection of envelope line and the front of the
compression wave, i.e., the C characteristics x  a0t . This point corresponds to T  0 .
Hence,
2 a0 2 ao2
tb  , xb  a0tb 
k  1 X P (0) k  1 X P (0)

Note that the instant tb is determined by the initial acceleration of the piston!
Consider the case when X P (0)   meaning that the piston attains the finite speed U P is no
time. Assume also that this speed is a finale one, i.e., it keeps steady. According to the above
formulae, the shock wave appears immediately – see the figure below.

Determination of U NSW :
U NSW (U NSW  U P )  a2
1 2 1 a 2  k 1 a 2
U 
 2 NSW k 1 0 2( k 1) 

Solution: U NSW  k41 U P  a02  ( k41 U P )2

Exercise: Calculate the slope of C characteristics in the region behind the NSW. Any
interesting observations?