Вы находитесь на странице: 1из 50

1

September 2009
FLUIDS II
(305-430)

SUPERSONIC JETS AND DIFFUSERS

1. Supersonic Jets

The discharge of a jet into the atmosphere is an interesting phenomenon, particularly for the case of a
supersonic jet. For a subsonic jet, the exit pressure of the jet p
e
must necessarily be adjusted to match the
atmospheric pressure p
a
. If p
e
> p
a
, the jet is underexpanded and hence the jet continues to expand. In a
subsonic flow, the pressure increases as the flow expands as the streamline diverges (i.e. a diffuser). Thus the
jet pressure increases further and will not be able to adjust to the environment. If the jet is overexpanded and
p
e
< p
a
, then upon exit, the jet streamlines will converge. In a subsonic flow, the pressure decreases in a
converging flow (i.e. a nozzle). Thus upon exit, the jet pressure continues to decrease and again cannot match
the environment. Hence, in a subsonic jet, the exit pressure must necessarily be the same as the environment,
i.e. p
e
= p
a
. This condition is achieved by internal adjustments of the flow during the transient starting process.
Compression or expansion waves can propagate upstream in a subsonic flow and adjust the flow field prior to
the jet exit in order to satisfy the condition of p
e
= p
a
.

For a supersonic jet, the phenomenon is more complex. Consider a deLaval nozzle connected to a reservoir of
fixed stagnation conditions. Let us connect the nozzle exit to a large vacuum tank in which the pressure can be
varied. If the tank pressure is above the value required to cause choking at the throat, then the flow in the
nozzle will be subsonic throughout. The pressure variation in the nozzle would correspond to the area variation,
i.e. the pressure decreases first in the converging section (i.e. nozzle) and then increases in the diverging section
(i.e. diffuser). The pressure variations for subsonic flow in the nozzle are shown by curves c in the figure
below.
























2
When the tank pressure is reduced to some critical value, the flow will be choked at the throat. The pressure at
the throat is p* and the ratio 528 . 0
*
0
=
p
p
(for 4 . 1 = ) when choking occurs. Downstream of the throat, the
flow decelerates to M < 1 and the pressure then increases in the subsonic diffuser to match the tank pressure at
the nozzle exit (curve b).

If the tank pressure is further reduced after choking had occurred at the throat, no adjustment of the flow can be
made upstream of the throat because M = 1 and the expansion waves are trapped there. The flow then
continues to expand to supersonic speeds downstream of the throat. However, in order to match the jet exit
pressure to the tank pressure, a shock is now formed inside the diverging section. The location of the shock is
such that the subsonic flow downstream of the shock can decelerate in the remaining diverging section of the
nozzle so that the jet exit pressure can match the tank pressure (curve a). With further decrease in the outside
tank pressure, the normal shock wave moves downstream. It is interesting to note that the strength of the shock
wave increases at first as it moves downstream from the throat, reaches a maximum and then decreases as it
nears the nozzle exit. When the tank pressure is reduced to the value corresponding to the jet exit pressure for
supersonic flow throughout the nozzle, we have the perfect operating condition of p
e
= p
a
and the supersonic jet
discharges parallel from the nozzle into the tank.

It is of interest to outline the steps in determining the shock location to match a given back pressure. Assuming
the upstream stagnation conditions are given and the area ratio of the nozzle is known, the first step is to assume
a certain shock location in the diverging section, hence the area A. The Mach number at the assumed shock
location is then given by the equation

( ) 1 2
1
2
1
1
2
1
1
1
2 1
*

+
(

|
.
|

\
|
+
+
=

M
M A
A


With the upstream shock Mach number M
1
determined, the downstream flow Mach number M
2
behind the
normal shock can be obtained from the equation

( )
( ) 1 2
1 2
2
1
2
1
2

+
=


M
M
M

The stagnation pressure downstream of the shock is given by

( )
( ) ( )
1
1
2
1
1
2
1
2
1
0
0
1 2
1
1
1 2
1
2

|
|
.
|

\
|

+

|
|
.
|

\
|
+
+
=


M M
M
p
p


Knowing the area ratio
e
A
A
where A is the area where the shock is located and A
e
is the nozzle exit area, the
Mach number (downstream of the shock at A), the exit Mach no. can now be found from

3
( ) 1 2
1
2
2
2
2
2
1
1
2
1
1

+

+
=

e
e
e
M
M
M
M
A
A


With the exit Mach number M
e
and the stagnation pressure downstream of the shock
2
0
p determined, the static
pressure of the jet at the nozzle exit can be found from the equation

1 2
0
2
1
1
2

|
.
|

\
|
+
=

e
e
M
p
p

We now check if the computed pressure p
e
is equal to the tank pressure. If not, we change the shock location
and go through the computations again and we iterate for the correct shock location. In practice, it is more
convenient to use the normal shock and isentropic flow tables to carry out the iteration.

Consider now the adjustment of the flow outside the nozzle exit if the tank pressure is further reduced past the
ideal operating condition when the jet exit pressure matches the outside tank pressure, i.e.
a e
p p = . When the
tank pressure is reduced below the jet exit pressure, the jet is said to be underexpanded and further expansion
must occur downstream of the nozzle exit. For an underexpanded subsonic jet, expansion waves can propagate
upstream into the nozzle and readjust the flow so that the exit pressure equal to the outside pressure. For an
underexpanded supersonic jet, the further expansion of the jet to match the outside pressure must take place
outside the jet exit because the expansion waves cannot propagate upstream. Thus expansion waves originate
from the jet boundary and penetrate into the flow causing the jet to diverge downstream of the nozzle exit with a
reduction of the jet pressure as it expands further. Due to inertia, the jet expands past its equilibrium position
and the jet pressure drops below the outside pressure resulting in the jet to converge again. The jet boundary
then becomes periodic and undergoes a few cycles before turbulent mixing in the shear layer of the jet boundary
diffuses the jet momentum as outside air is entrained into the jet. Schlieren photographs illustrating the
supersonic jet downstream of the nozzle exit for various ratios of the jet exit pressure p
1
to the environment
presure
E
p are illustrated below.
4









































Now reverse the condition and start to increase the tank pressure after the supersonic jet has been established.
For the outside tank pressure, slightly higher than the jet exit pressure (the jet is now said to be overexpanded)
readjustment of the pressures must take place outside the exit plane of the nozzle. For overexpanded supersonic
jets, oblique shocks are formed to raise the jet pressure to match the tank pressure. The oblique shock turns the
jet inwards to form a converging boundary (i.e. diffuser) and the jet pressure rises. The oblique shocks reflect
from the jet axis as oblique shocks which, when reflected off the jet boundary, becomes expansion waves and
result in a periodic phenomenon until the growth of the turbulent jet boundary shear layer and an entrainment
diffuses the jet momentum sufficiently to form a subsonic stream. If the back pressure is increased further, the
external oblique shock then moves into the nozzle. Under ideal consideration, we assume a normal shock to
form inside the nozzle with the location of the shock governed by the matching of the pressure at the exit plane.
5
In practice, the phenomenon is quite complex as the oblique shocks start to move into the nozzle. There is
always a subsonic boundary layer at the nozzle wall, thus pressure disturbances can propagate upstream of a
supersonic flow through the subsonic boundary layer. The sharp increase in pressure due to the oblique shock
interacting with the boundary layer causes the boundary layer to separate. The jet boundary then detaches from
the wall and the complex flow pattern is illustrated below.




















If the shock angle is large, the intersection of the oblique shocks can result in a Mach disc formed near the jet
axis. As the back pressure increases, this oblique shock-Mach disc complex moves upstream. The flow
downstream is highly complex and three dimensional.


2. SUPERSONIC WIND TUNNEL

For a channel with two throats (e.g. a supersonic wind tunnel) where a nozzle first expands the flow to
supersonic speeds followed by a diffuser to decelerate the flow to recover the pressure, the stagnation pressure
losses associated with the shock waves that are present during the transient starting process must be accounted
for by the appropriate choice of the throat areas. A typical continuous flow supersonic wind tunnel is illustrated
in the schematic sketch below











6
The mass flow rate depends on the stagnation pressure. If there is a stagnation pressure loss upstream of the
diffuser, then the throat area of the diffuser, i.e.
*
D
A must be larger than the throat area
*
N
A of the nozzle to
accommodate the same mass flow rate. Thus

sN
sD
D
N
p
p
A
A
=
*
*


where
sN
p and
sD
p the stagnation pressure for the nozzle and the diffuser respectively. Since shock waves in
the nozzle and test section result in a stagnation pressure loss, i.e.
sN sD
p p < , we see that
* *
D N
A A < . In other
words the throat area for the diffuser must be larger than that of the nozzle otherwise the tunnel cannot be
started. On the other hand, if
*
D
A is too large, the diffuser flow will be highly supersonic and a shock wave will
be formed downstream of the throat of the diffuser. This shock wave will cause a large stagnation pressure loss
which would have to be compensated for by the compressor resulting in a higher power requirement to run the
compressor.

Let us consider the starting process of the wind tunnel. When the compressor is turned on to increase the
upstream stagnation pressure of the nozzle and decrease the pressure downstream of the diffuser, the flow will
progressively accelerate from subsonic to supersonic flow in the nozzle, subsequently to choking at the nozzle
throat, a shock will be formed which then moves progressively downstream of the diverging section of the
nozzle towards the exit into the test section. This is illustrated by curves a, b, c, d, e of the figure below.



























7
As a result of this shock wave, there is a loss in stagnation pressure and hence the throat area of the diffuser
must be greater than that of the nozzle in order to accommodate the same mass flow rate. The worst case is
when the shock reaches the test section (curve e) when the shock Mach no. is the highest. Thus, the minimum
throat area of the diffuser must be such that it can account for the stagnation pressure loss of a shock wave
corresponding to the Mach number at the test section. Note that the flow in the test section and diffuser is
subsonic for curves a to e. Further decrease in the back pressure downstream of the diffuser now causes the
shock wave to move rapidly through the test section and get swallowed by the diffuser and stabilizing
somewhere downstream of the diffuser throat (curve f). Note that the flow in the nozzle, the test section, and in
the converging part of the diffuser is now shock free. Since
* *
N D
A A > , the diffuser throat is not choked and the
flow is supersonic there. Since the Mach no. increases further downstream of the diffuser throat, it would
minimize the stagnation pressure loss if we could stabilize the shock wave just downstream of the diffuser
throat where the Mach number is smallest. Thus, after the tunnel has been started and the shock wave has been
swallowed, we can now adjust the compressor pressure ratio to move the shock back upstream to the diffuser
throat where the Mach no. is a minimum (curves a, b, c of figure below).




































8


This will reduce the stagnation pressure loss across the shock and save power in running the compressor.
Alternately, we can vary the diffuser throat area after the shock has been swallowed to drive the shock closer to
the throat to minimize the stagnation pressure loss. Note that the shock is not stable in the converging part of
the diffuser. If the shock is at the converging part of the diffuser and some perturbation causes it to move
slightly upstream, then the Mach no. will be higher and the stagnation pressure loss across the shock will be
greater. The mass flow through the diffuser throat will be decreased as a result of the stagnation pressure loss.

This causes a transient mass accumulation between the shock and throat which will drive the shock further
upstream to an even higher Mach no. Thus the shock will eventually be driven out to the test section and into
the nozzle causing an unstart of the wind tunnel. On the other hand, a small displacement of the shock
towards the throat results in a smaller stagnation pressure loss and a higher mass flow rate which then sucks
the shock in further until it stabilizes in the downstream diverging section of the diffuser. The optimum
operating conditions will have be to have the shock stabilized slightly downstream of the diffuser throat where
the Mach number is near the minimum hence where the stagnation pressure loss is near the minimum also.

To start the wind tunnel, the diffuser throat area must be sufficiently large to compensate for the stagnation
pressure loss corresponding to a shock wave of the Mach number of the test section. For example, a M = 3
shock wave will have a stagnation pressure ratio of 328 . 0
1
2
=
s
s
p
p
. Thus the ratio of the diffuser to nozzle throat
area 04 . 3
328 . 0
1
*
*
= =
N
D
A
A


In other words, the diffuser throat area is three times larger than the nozzle throat area in order to swallow the
starting shock wave. When the shock wave has been swallowed and stabilizes downstream of the diffuser
throat, the Mach number at the diffuser throat can be found from the area ratio, i.e. 3
*
*
=
N
D
A
A
and the Mach
number. at the diffuser throat is 2.635. For a normal shock where
M
1
= 2.635, the stagnation pressure ratio is about 0.44. Thus we have a loss of about 56% which the
compressor has to compensate for. To overcome this large stagnation pressure loss, a variable geometry tunnel
where the throat area of the diffuser (or nozzle) could be adjusted to should be used. For example, we could
open up the diffuser throat to swallow the shock during starting, then reduce the throat area afterwards to reduce
the Mach no. at the throat toward sonic condition. We could also simultaneously raise the back pressure of the
diffuser. This is illustrated in the series of curves a, b and c of the figure below.












9

































In the ideal limiting case, the Mach number at the diffuser throat should be unity and the entire flow in the wind
tunnel is then isentropic. However, because of friction, there will always be some stagnation pressure loss in
the nozzle and test section so that the diffuser throat must always be larger than the nozzle throat even in the
absence of shock waves.



3. SUPERSONIC INLETS (DIFFUSERS)

For air-breathing engines, the air flow must be decelerated from the free stream Mach no. (i.e. flight Mach no.)
to less than about M = 0.4 prior to entering the turbocompressor which then brings the Mach number further
down to about 2 . 0 ~ M before the flow enters the combustor. Whether the free stream is subsonic or
supersonic, it must be decelerated in the intake diffuser. The design of the intake diffuser must be such that the
flow is uniform and the stagnation pressure losses are kept to a minimum. Furthermore, the inlet should not
produce a large drag due to the air spillage flow over the external surface of the inlet. Consider first subsonic
diffusers, i.e. the free stream is subsonic. There are two types of subsonic diffusers, the external compression
inlet and the internal compression inlet. For the external diffusion inlet, the diffuser is just a straight duct and
the flow diffusion (i.e. compression) occurs ahead of the inlet plane as seen in the figure below.
10



















The flow entering the diffuser inlet A
1
corresponds to the flow through the so-called capture area A
0

upstream. The flow area divergence from A
0
to A
1
occurs prior to the entrance to the inlet and for subsonic
flow, the flow divergence results in a pressure rise. Since there is no solid surface to generate shear, the
external diffusion is isentropic. The flow between A
0
and A
1
will spill over the lip of the diffuser inlet
creating a large drag. Thus the external diffusion inlet is not satisfactory for high subsonic flight Mach
numbers. For the internal compression subsonic inlet, the capture area is the same as the inlet area and
diffusion is achieved internally by area increase in the duct itself.

























11
For isentropic flow, the pressure increase due to area increase is given by

p
dp
M
M
A
dA
2
2
1

=

For internal diffusion, the most severe problem is due to boundary layer separation. Because of the adverse
pressure gradient in the diffuser, the boundary layer tends to separate. Thus, one must maintain a small pressure
gradient, hence a gentle area divergence (of the order of 5 to 8 degrees). A gentle area divergence results in a
long diffuser length, thus increasing the friction losses. Boundary layer separation will result in a loss of the
internal diffusion due to area change.

For supersonic inlets, we have the simple normal shock diffuser, the convergent-divergent diffuser, and the
spike diffuser. In the normal shock diffuser, a normal shock is stabilized either in front of the inlet, at the
inlet or gets swallowed inside the inlet and thus becomes a series of oblique shock waves. The operating
characteristic of a normal shock diffuser for design condition, for increasing and decreasing back pressure are
illustrated in the sketch below.


































12

When the normal shock stabilizes ahead of the inlet, the capture area
1 0
A A < , external diffusion of the
subsonic flow downstream of the shock occurs prior to entering the inlet followed by further internal diffusion
if there is an area divergence in the duct itself. Since
1 0
A A < , flow spillage occurs and results in large external
drag. When the normal shock stabilizes at the inlet plane, further internal diffusion can occur downstream of
the normal shock if the duct area diverges. The capture area is now the same as the inlet area and no flow
spillage occurs. In a normal shock diffuser, there is a large stagnation pressure loss associated with the normal
shock. Thus the diffuser is inefficient at high free stream or flight Mach numbers (e.g. 8 . 1 ~ M ). For higher
supersonic flows, the normal shock may be swallowed inside the diffuser and turned into a series of oblique
shock waves and Mach stems. The stagnation pressure loss in oblique shocks is less than that of a normal
shock. However, the reflection of the oblique shocks off the wall causes complex shock boundary layer
interactions and flow separations. Normal shock diffusers are simple and give acceptable performance for flight
Mach numbers less than about M = 1.8.

For higher supersonic speeds, the more efficient converging-diverging diffuser should be used. The ideal
operating condition would be for a normal shock to be stabilized slightly downstream of the diffuser throat as in
the case of the diffuser of a supersonic wind tunnel. The Mach number at the diffuser throat should be slightly
supersonic. If the free stream Mach number is decreased while the back pressure is increased, one can achieve
isentropic flow throughout the throat choked and subsonic flow downstream of it (If we assume no friction in
the duct).

The ideal operating condition must be approached from some transient starting conditions where the flow Mach
number accelerates to the operating Mach number. As the free stream Mach number increases, the detached
normal shock (ahead of the inlet) will eventually reach the inlet and the capture area equals the inlet area A
1
. If
the throat is choked, then mass flow rate is a maximum for the value of the stagnation pressure downstream of
the normal shock. For the shock to be swallowed and move downstream of the throat, the throat area must be
increased. If the throat area is slightly decreased, the throat cannot accommodate the mass flow rate and mass
will accumulate and the shock will be expelled and detached from the inlet. The detachment distance will be
such to allow for the accumulated mass to be spilled out of the inlet and flow externally over the lip of the inlet.
So for each free stream Mach number M, we can look up the normal shock table for the downstream Mach
number and then the isentropic flow tables for
1
*
A
A
as a function of this downstream Mach number behind the
shock and plot a curve
1
*
A
A
(M) as shown in figure below.















13




































For any Mach number, we see that for an area ratio
1 1
*
A
A
A
A
t
> where the shock will be swallowed and stabilizes
downstream of the throat. For
1 1
*
A
A
A
A
t
< , the shock will be detached. The
1
*
A
A
(M) curve marks the boundary
between condition of the shock being swallowed or expelled and detached from the inlet.

Consider next the case when the diffuser has already been started and supersonic flow now occurs in the
converging section. Again there corresponds a certain area ratio
1
*
A
A
for a given free stream Mach number M
when the throat becomes choked. For
A
A
A
A
t
*
1
> , the throat becomes unchoked the flow becomes supersonic
14
and the shock is swept downstream. However, for
1 1
*
A
A
A
A
t
< the throat can no longer handle the captured mass
flow for the given supersonic Mach number. Thus the normal shock will be expelled and detached from the
inlet. Looking up the isentropic flow tables for
1
*
A
A
for various Mach numbers, we can plot ( ) M
A
A
1
*
as shown
in the figure below.






































The difference between this curve and the curve in the previous figure is that in the previous case the isentropic
flow is based on the downstream Mach number of the normal shock at the free stream Mach number. The curve
in the figure above is just based on the free stream Mach number since no shock wave is involved.

15
The starting process for a supersonic diffuser can now be clearly illustrated by plotting these two curves
together. These two curves define the boundaries where the shock is swallowed or expelled with and without a
normal shock at the inlet entrance.








































Consider a fixed area ratio
1
A
A
t
inlet. As the free stream Mach M number increases, we move along the
horizontal line a, c, b. The shock will be detached until we reach point b on the boundary curve. At b, the
throat is just choked, and further increase in Mach number will cause the shock to be swallowed. Once the
shock is swallowed, we have isentropic flow throughout the converging section including the throat. The shock
is now stabilized somewhere downstream of the throat. If we now reduce the throat area keeping the Mach
number constant, we move down the vertical line from b to d and the shock remains swallowed and the flow
16
supersonic throughout (including the throat) up to the shock location downstream of the throat. However, if we
decrease the throat area past point c (which is the limit boundary where the throat is now choked) for the free
stream Mach number, the normal shock will be expelled out of the diffuser and we have a so-called unstart
phenomenon. Hence, there is a sort of hysterisis where the shock, once swallowed, will remain downstream of
the throat until the area ratio goes way down past the value when the shock was first swallowed before it gets
expelled when the second boundary curve is reached. Thus, for a fixed geometry diffuser, we can overspeed
to first swallow the shock to start the diffuser, then slow down and operate at any free stream Mach number
between point b and c. The shock will not be expelled until the Mach number has decreased past point c.
Because of the high stagnation pressure loss across a normal shock with increasing Mach number, we see that
we must overspeed considerably to be able to swallow the shock as the area ratio
1
A
A
t
decreases. In fact, we
reach a limit when 6 . 0
1

A
A
t
. For this area ratio, the operating free stream Mach number M = 2 before the
shock gets expelled if the diffuser can be started. However, to start the diffuser for 6 . 0
1
=
A
A
t
, we have to
overspeed to M . Thus, overspeeding is not an efficient way to start the diffuser when the area ratio is
small.

The alternative is to use a variable throat area diffuser. For example, if we want to operate at a flight Mach
number M = 3.2, we first open up the throat area to 7 . 0
1
=
A
A
t
, i.e. point b. The shock will then be swallowed,
and once swallowed, we can close up the throat area and move vertically downwards from point b to point d.
We can operate at M = 3.2 for 2 . 0
1
~
A
A
t
without causing the diffuser to unstart.

We can also open up the throat area larger, say to 8 . 0
1
=
A
A
t
. In that case, the shock will be swallowed when
2 . 2 ~ M (point e in the above figure). Once the shock has been swallowed, we can close the throat area and
simultaneously increase the free stream Mach number. We will follow the dotted curve to the operating point d
in this case.

The variable throat diffuser is a much more efficient way to start the diffuser. However, for a three-dimensional
axisymmetrical inlet, it is difficult to design the mechanism to vary the throat area. The spike inlet offers a nice
solution to the variable area design as well as reducing the stagnation pressure loss by replacing the normal
shock by a series of oblique shocks. The spike inlet was suggested by the famous German aerodynamicist,
Klans Oswatitsch. A sketch is shown below.











17
























As can be observed, a conical oblique shock is now attached to the spike instead of a normal shock and that
reduces the free stream Mach number somewhat (depends on cone angle), but the flow downstream of the
oblique shock is still supersonic. A second normal shock now reduces the flow to subsonic. However, the
normal shock is now weaker and the loss in stagnation pressure is reduced. Further subsonic diffusion can also
be achieved by area divergence between the central spike body and the external housing. With the spike
central body configuration, the mechanical design to vary the throat area is also simpler. The spike inlet is used
in the Lockheed SR-71.




MECH 430 FLUIDS 2



Flow of a compressible fluid in a constant area duct with friction


For an incompressible fluid where the density remains constant, friction results in
a pressure drop in the flow direction. For a compressible flow, viscous dissipation heats
up the gas resulting in a density change. Thus, the flow velocity and pressure can
increase or decrease depending on the rate of expansion due to density decrease and the
convective mass flux through the cross section of the pipe. As in the case of area
change, friction has opposite effect if the flow is subsonic or supersonic.

Let us first review the steady flow of an incompressible fluid in constant area pipe
with friction (which results in a wall shear stress
w
). Referring to the sketch below:



p + dp

w
L
p



The conservation of mass for steady flow gives u=constant and since = constant, the
flow velocity is constant throughout the length of the duct. The momentum equation
gives:

( ) pA p dp A A
w w
+ = 0

where A is the area of the duct ( A
D
=

2
4
for a circular duct) and A
w
is the wetted area
( A D
w
= L for a circular pipe). Rearranging the momentum equation gives:

= = dp
A
A
L
D
w
w w


4

where we have assumed a circular pipe for convenience. The wall shear stress is given
by:


w
r R
u
r
=

=



where is the coefficient of viscosity and

u
r
r R

=
is the velocity gradient at the wall (if
we assume a laminar flow). The velocity profile for steady laminar incompressible flow
in a circular pipe is given by:


u
u
r
R
=

2 1
2


where u is the mean flow velocity (i.e. the classical Hagen-Poiseuille Flow.) This
parabolic velocity profile gives the wall shear stress as:




w
r R
u
r
u
R
u
D
=

= =

=
4 8


The negative sign denotes the direction of
w
that gives the parabolic profile and we have
already considered the appropriate sign (i.e. direction) of
w
when we write the
momentum equation. Thus, we write:

=

= = =

dp
L
p p
L D
u
D
u
D
w
D
1 2
2
2
4 32 64
2
1
Re


where Re
D
uD
=

is the Reynolds number based on the pipe diameter. In general, we


define a friction factor f as:

f
u
w
=
4
2
2



where is determined from experiments. (Moody diagram where f is given for a
wide range of Reynolds number and wall roughness). For the present case of laminar
Poiseuille Flow:
( f
D
Re )

2
f
D
=
64
Re


We may write the pressure drop as:

2
1
2 Re
64
2 2
2 1
u
D
f
D
u
L
p p
D

=



For incompressible flow, we need not worry about the energy equation since = constant
and the problem is a dynamic problem with pressure forces balancing the friction forces.


Compressible Flow (Fanno Flow)

For steady compressible flow in a constant area, adiabatic duct with friction as illustrated
below, i.e.:


dx
P+dp
+ d
T+dT
u+du

w

u
p
T


The conservation of mass gives:

m uA = = constant


d du
u

+ = 0 1.


since dA=0 for a constant area duct.


The conservation of momentum can be written as:
3

Dsx dpA mdu
w
= 2.

where S and A are the perimeter and the area of the cross-section of the tube, respectively.
Defining a friction factor:

f
u
w
=

2
2


and a hydraulic diameter D
H
by:

D
area A
S
H
= =
4 4 *
wetted perimeter


Equation 2 becomes:



udu dp
u fdx
D
H
=
2
2
4
3.

If the flow is adiabatic, the energy equation is given by:

h
u
h
o
+ =
2
2
= constant

Note that there is no work done by the viscous stress
w
since the velocity at the wall
vanishes. The no slip (i.e. u=0) condition at the wall generates a velocity profile resulting
in viscous dissipation. The viscous heating is at the expense of the kinetic energy
decrease in the flow but since there is no heat transfer, the viscous heating remains in the
flow. In the present assumption of a quasi one dimensional flow, the flow velocity is
uniform across the cross section of the duct. We should consider the flow velocity as an
averaged value to permit the presence of a velocity gradient for the viscous stress to
occur. In an inviscid flow, there will be no shear stress at the wall. For adiabatic (not
isentropic) flow where

t cons H
u
h tan
2
0
2
= = +

we get

dh+udu=0

4
Since h=c
p
T, and c
R
p
=

1
and c , the energy equation can be written as: R
2
= T

( )
dT
T
M
du
u
+ = 1
2
0 4.
The equation of state for a perfect gas is:

p RT =

and hence


dp
p
d dT
T
= +

5.

Rewriting Equation 3 as:

udu
p
dp
p
u
p
fdx
D
H
=

2
2
4


and noting that

p
c =
2
, we obtain an expression for
dp
p
as

dp
p
M
du
u
M fdx
D
H
=



2
2
2
4
6.

From Equations 1, 4, 5 and 6, we can get an expression for
du
u
as

( )
du
u
M
M
fdx
D
H
=

2
2
2 1
4
7.

From Equation 1, we see that:

( )
d du
u
M
M
fdx
D
H


= =

2
2
2 1
4
8.

From Equation 4, we get

( )
dT
T
M
du
u
= 1
2



5
and substituting Equation 7 into the above yields


( )
( )
4
2
- -1
4
2 1-

=
H
M
dT fdx
T D M
9.
Substituting Equations 8 and 9 into Equation 5, gives:


( ) ( )
( )
dp
p
M M
M
fdx
D
H
=
+


2 2
2
1 1
2 1
4
10.

To get an expression for the Mach number, we note M
u
c
= , thus
dM
M
du
u
dc
c
=

and since c RT
dc
c
dT
T
2
1
2
= = ,

Hence,

dM
M
du
u
dT
T
=
1
2


And substituting Equations 7 and 9 into the above, gives:


( )
dM
M
M M
M
fdx
D
H
=
+


2 2
2
1
1
2
2 1
4
11.


For the variation of the stagnation pressure along the pipe, we note that the definition of
the stagnation pressure is the pressure one would obtain if we decelerate a flow
isentropically to zero velocity. Thus,

p
p
M
o
= +


1
1
2
2
1



and differentiating, we get:

dp
p
dp
p
M
M
dM
M
o
o
= +
+

2
2
1
1
2

6

The variation of p and M along the pipe is given by Equations 10 and 11, thus, we get:


dp
p
M fd
D
o
o h
=

2
2
4 x
12.

The variation of entropy along the pipe can be found from:


Tds = dh-vdp


with h c T
RT
p
= =

1
, pv=RT, the above becomes:

ds
c
dT
T
dp
p
p
=

1



Equations 9 and 10 give the variation of
dT
T
and
dp
p
, thus we obtain for the change in
entropy along the pipe the following expression:


ds
c
M fd
D
p H
=

1
2
4
2
x



or
ds
R
M f
D
dx
dp
p
H
o
o
= =

2
2
4
13.


since
( ) c
R
p


=
1
. As in the case of the normal shock, the entropy increase can be
correlated to the stagnation pressure loss due to viscous dissipation.


Equations 7-13 give the variation of u, , T, P, M, p
o
and s along the pipe
respectively. Except for
dp
p
o
o
and
ds
R
, all the equations contain the term (1-M
2
) in the
denominator. Thus, depending on whether the flow is subsonic M<1, and (1-M
2
)>0 or
supersonic, M>1 and (1-M
2
) <0, the various thermodynamic states can either increase or
7
decrease due to friction. The table below summarizes the effect of friction on the fluid
and thermodynamic states. The sign + or denotes whether the variable increases or
decreases along the pipe (i.e. increased value of x or positive dx).




Variable Subsonic Supersonic

Velocity u + -
Mach Number M + -
Pressure p - +
Temperature T - +
Density - +
Stagnation pressure p
o
- -
Entropy s + +

The difference between subsonic and supersonic flow is a result of the
competition between the rate of expansion due to viscous heating of the flow and the rate
of mass convected through the cross section of the pipe. Thus, for subsonic flow, the rate
of expansion (characterized by the sound speed) dominates, and the density decreases
resulting in an increase in the flow velocity, a decrease in temperature and pressure due to
the expansion. The reverse happens for supersonic flow when the rate of convection
(characterized by the flow velocity) dominates over the rate of expansion. However, for
the stagnation pressure and entropy, both subsonic and supersonic flows indicate the
same behavior of loss in stagnation pressure and increase in entropy from irreversible
viscous dissipation.


Integration of the equations

To get the variation of the various state variables along the duct, the differential
equations have to be integrated. The important one to start is the variation of the Mach
number with distance given by Equation 11. Rewriting it in the following form:

( )
2 2
M dM 1


we can integrate the above using the method of partial fraction. The above can be
expanded as:

4 fdx
=

1
1
2
4 2
D

M M
H
+

8
1
2
dM 2
dM
2
4 fdx + + dM 1 1
2


Since friction always drives the flow towards M=1, we integrate the above from
0<x<L
*
and 0<M
2
<1 (i.e. we assume the Mach number at x=0 is M and at x=L
*
, M=1. In
general, the friction factor is a function of the Reynolds number and varies along the duct
as the fluid properties change. However, we shall assume an average value for the
friction factor from 0<x<L
*
. We then obtain:

14.

The above equation says that if we have a duct where at x=0, the Mach number is
M, then at x=L
*
, M=1 and the flow is choked. Equation 14 is computed for a range of
Mach numbers and tabulated (Fanno flow tables). So, if we have a pipe of length L and
inlet Mach number M
1
, and if we want to know the Mach number at L, we do the
following:




Referring to the sketch above:

L L L
M M
=
* *
1 2


or

2 2
1
1
2
4

2
2
D
H

= +
M M
+


M
4 1
1
2
1
1
2
2
2
2
2
1 +

+
2
M
fL
*
M

= + ln

D M

M
H
+

M
2
L
L
M1
L
*
M2
M
1
9

4 4 4
1 2
fL
D
fL
D
fL
D
H
M
H
M
H
=
* *


We know L and M
1
, we can look up the table for such that
*
1
L
4
1
fL
D
M
H
*
corresponds
to 1 = M . Then, we can find the value for:

4 4
2 2
fL
D
fL
D
fL
D
M
H
M
H H
* *
=
4


We can then find the value of M
2
corresponding to the value of
4
2
fL
D
M
H
*
.

Equation 14 gives M as a function of L
*
. If we want to know the other flow variables
like p, T, , etc. we can integrate the other equations for
dp
p
dT
T
, , etc. However, it is
more convenient to express them in terms of the Mach number so that once M at a
section is known, we can find the other variables. From Equations 10 and 11, we write:

( ) ( )
( )
( )
dp
p
dM
M
M M
M
fdx
D
M
M
M
fdx
D
H
H
=
+

1 1
2 1
4
2
1
1
2
1
4
2 2
2
2
2
2




and simplifying the above yields:


( ) ( ) dp
p
M
M
dM
M
=
+
+

1 1
2 1
1
2
2
2
2
2



Using partial fraction, the above can be written as:


2
2
2
2
1
2
1 2
2 1
2
dM
dp dM
p M
M

=

+



10
which integrates to yield:


p
p M
M
*
=
+
+

1 1
2 1
1
2
2

15.

The above equation gives the pressure p at M and p = p
*
when M=1. Similarly, we
can find expressions for the other variables in terms of the Mach number as follows:



T
T
c
c
M
* *
=

=
+
+

2
2
1
2 1
1
2

16.




( ) u
u
M
M
*
=
+
+

1
2 1
1
2
2
17.




*
*
= =
+

+
u
u M
M
1
2 1
1
2
1
2
18.




( )
( )
p
p M
M
o
o
*
=
+

1
2 1
1
2
1
2
1
2 1

19.




( ) s s
c
M
M M
p

=
+
+

+
*
ln
2
2 2
1
2
1
2 1
1
2

20.
11


Equations 15 to 20 are also computed for various Mach numbers and tabulated together
with
4 fL
D
H
*
in the Fanno tables for convenience in solving problems of compressible flow
in ducts with friction.



The Fanno Line

The adiabatic flow of a compressible fluid in a constant area duct with friction
satisfies the conservation of mass and energy i.e.

constant = u

0 = +
u
du d



2
constant
2
o
u
h h + = =

0 = + udu dh

From these two equations, we can obtain an expression for the variation of the enthalpy
(or equivalently the temperature) and the entropy along the duct. From the Tds equation,
i.e.:

vdp dh Tds =

we get:

p
dp
h
dh
c
ds
p

1
=

and from the equation of state RT p = , we get:

h
dh d
T
dT d
p
dp
+ = + =



Hence,

12

d
h
dh
h
dh d
h
dh
c
ds
p
1 1 1
=

=

From the conservation of mass, we write:

2
2
2u
du
u
du d
= =



From the energy equation, we note that:

h h
u
o
=
2
2


and ( h h d dh
du
o
= =
2
2
) since constant. =
o
h

Thus, we may write:

( )
( ) h h
h h d d
o
o

=
2



and the equation for the entropy becomes:

( )
( ) h h
h h d
h
dh
c
ds
o
o
p


+ =

2
1 1


We can integrate the above as following:

( )
( )
* * *
1 1
2
o
o
s h h h
o
s h h h
p o
d h h
ds dh
c h h

= +




and obtain:



Since
*
2
1
h h
o
+
=

, we can eliminate h
o
in the above equation and obtain the equation for
the Fanno Line as:

*
h h s s h 1 1
= +


*
h
o
ln ln
*
h c 2
p o
13

2
1
*
2
1
1
*
*
2
1
1
2
ln
h
h
h
h
c
s s
p
21.


The Fanno line gives the variation of the enthalpy as a function of the entropy for
a given mass flow rate in and stagnation enthalpy h
o
(or equivalently h
*
). The lower
branch is for supersonic flow and the upper branch is for subsonic flow. For any given
value of M, friction will drive the flow towards sonic conditions M=1 where the entropy
is a maximum. To prove that entropy is a maximum when the flow becomes sonic, one
could differentiate Equation 20 and equate 0 =
dM
ds
and solve for M. Alternately, we do
the following: from the Tds equation, we write:

1
Tds dh dp

=

and from the energy equation , udu dh = we get:

dp udu Tds

1
=

From the conservation of mass constant = u , we get :

0 = + ud du


If we express ( p s, ) , we can write:

( ) dp
c
ds
s
dp
p
ds
s
p s d
p
s
p
2
1
, +

=



since the sound speed c is defined by:

2
s
p
c




Replacing d in the continuity equation by the above expression yields :

0
2
=

+ + = + ds
s
u dp
c
u
du ud du
p

22.

14
From the Tds equation, we get:

Tds udu dp =

and replacing dp in Equation 22 by the above and rearranging gives:

( ) 0 1
2
2
=


p
s c
T
uds M du

23.

Since
2
c
T
is always positive and
p
s

is always negative (since density decreases


when heat is added at constant pressure and heat added means entropy increases, i.e.
0 < as ds>0 at constant p). The bracketed term is always positive, i.e.:

0
2
>

p
s c
pT



Thus, as , ds must vanish, thus s corresponds to an extremum (maximum) value
when M=1.
1 M

Another alternate expression for the Fanno Line that involves the Mach number can be
obtained as follows: From the Tds equation, we write:

p
Rdp
T
dh
ds =

and rearranging, we write:

dh
dp
p
R
T
ds
dh

=
1
1


From the energy equation, we get:

0
2
2
= +
du
dh

And the conservation of mass gives:

2
2
1
2
d du
du
u u

= =

15
we can express dh as:

d
u
du
dh
2
2
2
= =

And using the equation of state , RT p = we can express:

T
dT
p
dp d
=



Hence,

= =
T
dT
p
dp
u
d
u dh
2 2



And solving for the slope
dh
dp
, we get:

( ) ( )
2
2
1 1 M
u
p
dh
dp
+ =

We can now write
dh
ds
as:

( )

+
=
2
2
2
1
1
1
u
p
M
u
p
p
R
T
ds
dh




or
1
2
2

=
M
T M
ds
dh


The above is an alternate form of the equation for the Fanno Line and it is clear that for
subsonic flow where M<1, 0 <
ds
dh
which corresponds to the upper branch of the Fanno
Line. For M>1, 0 >
ds
dh
, the locus of states is represented by the lower branch of the
Fanno Line. When M=1,
dh
ds
and . max s

To plot a Fanno Line, we first note that the Fanno Line involves the conservation of mass
and energy, i.e.
16

constant = = m u

constant
2
2
= = +
o
h
u
h

Thus, a Fanno Line corresponds to a value for and . Combining conservation of
energy and mass, we get:
m
o
h

constant
2
2 2
= = +
o
h
v m
h

24.

Where

1
= v is the specific value. So, given and , we pick different values for m
o
h
and compute h from Equation 24. And Equation 21 gives s as a function of h. Note
that
*
2
1
o
h

=
+
h , thus the value of h
*
can be found for a given stagnation enthalpy h
o
.


Effect of duct length and back pressure


The flow inside a duct with friction is influenced by the duct length and the back
pressure of the exit of the duct. Consider first the case of subsonic flow in the duct. We
assume the inlet to the duct is from a converging nozzle connected to a large plenum
chamber where the stagnation pressure and temperature is p
o
and T
o
. Consider first
just the nozzle. For a subsonic jet, the exit pressure must be the same as the back
pressure. Thus, for a given exit pressure, the Mach number at the jet exit is given by the
isentropic relationship

1
2
2
1
1


+ =

M
p
p
o


If we now connect a pipe to the exit of the converging nozzle, then the exit of the nozzle
corresponds to the entrance to the duct. Let the end of the duct be connected to a large
plenum chamber where the pressure can be controlled. The system is shown
schematically in the sketch below.
17
stagnation
conditions

P
o
, T
o
plenum
chamber

P
B
P
e

a
b
c
P
e
= P
B

P
e
= P
B

P
e
= P* = P
B

P
e
= P*
P
e
P
B
< P*
decreasing m

a
c
b

M
e

M
e
M
e
= 1
h
o


For curve a, the flow is subsonic throughout the duct, the exit pressure p
e
=p
B
. The
value of
H
D
fL 4
is less than the value
H
D
fL
*
4
corresponding to the inlet Mach number M
1
.

When we lower the back pressure p
B
, the flow is similar but the inlet Mach number to the
duct (hence the mass flow rate) is increased. When the back pressure is reduced to a
value where the flow is choked at the exit, then p
e
=p
*
=p
B
and M
e
=1. For this case, the
18
duct length corresponds to the critical value L
*
for the inlet Mach number (curve c). If
the back pressure is further decreased, i.e. p
B
<p
*
, no change occurs within the duct and
further expansion occurs downstream of the duct exit to permit the jet pressure to adjust
to the ambient pressure. The three cases of a b and c are also illustrated by the
Fanno Lines on the h-s diagram. After choking occurs, (i.e. curve c,
h H
D
fL
D
fL
*
4 4
= ) if
the length of the pipe is increased beyond this critical length, then the inlet Mach number
will have to decrease to permit choking to occur in the longer length of pipe. Thus, for
subsonic flow, the inlet conditions to the pipe can adjust to the different values of the
back pressure (or exit pressure) and the length of the duct.
For supersonic flow, the inlet of the duct must be connected to the stagnation tank
via a converging diverging nozzle. Once the area ratio
*
1
A
A
of the nozzle is fixed, then
the exit Mach number, pressure, flow rate, etc. are all determined from the isentropic
flow relationships (look up the tables). Mach number at the duct inlet is supersonic and
conditions downstream in the duct cannot influence the inlet condition unless a shock
wave is driven back into the nozzle. Consider first a duct length less the critical value for
the given inlet Mach number M
1
. For a sub critical duct length, the exit Mach number is
still supersonic and the flow in the duct is supersonic throughout. This is the case
provided the exit pressure of the duct is greater than the back pressure p
B
in the plenum
chamber, i.e. p
e
<p
*
since the flow is not choked at the exit and p
e
<p
B
. The under
expanded jet will adjust to ambient pressure via expansion waves. When the back
pressure is increased to the exit pressure, i.e. p
e=
p
B
, then the jet exit pressure matches the
ambient pressure and the jet is parallel. When p
B
>p
e
, but not significantly greater, then
conditions inside the duct remains unchanged and adjustment to ambient pressure occurs
downstream of the jet exit via oblique shock waves. When p
B
increased further, the
oblique shock becomes stronger by increasing its angle to the flow. The limiting
condition will be when a normal shock occurs just at the duct exit so that the pressure
downstream of the normal shock matches the back pressure p
B
. When p
B
increases
beyond this limiting value, the shock moves into the duct and becomes stronger since the
Mach number is higher. Downstream of the shock is subsonic, thus the exit pressure will
have to be equaled to the ambient pressure.
The location of the shock is such that this condition is satisfied. The above
discussion is for the case when the duct length is less than the critical value
corresponding to the inlet Mach number.
If the duct length is the critical value, then p
e=
p
*
and M
e
=1. Again, when
p
e=
p
*
>p
B
, the under expanded sonic jet will undergo further expansion downstream of the
duct exit to adjust to the ambient pressure. If p
B=
p
*
, the same kind of events as described
earlier for the case of L<L
*
occurs. For high p
B
, a normal shock will eventually be
formed inside the duct. Once a normal shock occurs inside the duct, the flow is subsonic
downstream of the duct and for a subsonic flow; the exit pressure must match the ambient
pressure. Similarly, if the duct length L>L
*
for the inlet Mach number, a normal shock
will be formed inside the duct. The position of the shock is such that the exit pressure
matches the ambient pressure. If the duct is very long, then the maximum Mach number
the subsonic flow downstream of the shock inside the duct can accelerate to is M=1.
Thus, the flow is chocked at the exit, p
e=
p
*
. If p
B=
p
*
, further adjustment occurs
19
downstream of the duct exit. For p
e=
p
*
, there is only one location where the shock inside
the duct is located.
For supersonic inlet flow and a very high p
B
so that a shock is driven inside the
duct, the process is illustrated in the h-s diagram below.






h
o
normal shock
transition
M
e
, P
e
= P
B
M

> 1
M

< 1
isentropic flow
in nozzle
20
Mech 430A Fluid Mechanics II:
Eect of heat addition on the compressible
ow in a constant area duct
For incompressible ows, the density stays constant and heat addition has
no eect on the dynamics of the uid motion (except if one considers the tem-
perature dependence of the viscosity). For a compressible uid, heat addition
changes the density and depending on whether the ow is subsonic or super-
sonic, competition between expansion (due to density change) and convection
results in dierent behaviors of the ow. We shall consider frictionless ow in
a constant area duct and we are interested to know the variation of the uid
properties along the duct as a result of heat addition (or extraction). Again, we
start from the conservation laws and refer to the simple sketch below, i.e.
The conservation of mass gives:
m = uA = constant
d m = 0 dA = 0
hence
d

=
du
u
(1)
The conservation of momentum gives:
mdu = pA(p +dp) A
Fig. 1: Sketch of the control volume
1
2
Adp + mdu = 0, dp +udu = 0 (2)
The conservation of energy gives:
md
_
h +
u
2
_
2
_
= dq
dh +d
_
u
2
_
2
_
= dq (3)
where dq is the heat per unit mass added to the control volume Adx. We can
write dq = dh
o
where h
o
is the stagnation enthalpy and heat addition essentially
increases the stagnation enthalpy of the ow. Hence, we can also write Equation
3 as:
d
_
h +
u
2
_
2
_
= dh
o
The equation relates p, , and T, i.e.:
dp
p
=
d

+
dT
T
(4)
From Equations 1 to 4, we can solve for the variation of the ow properties as a
function of the heat addition. For example, consider the velocity variation rst.
Equation 2 can be written as:
dp = udu
dp
p
=
u
p
du = M
2
du
u
(5)
Equation 3 can be written as:
dh +udu = dh
o
dT
T
+ ( 1) M
2
du
u
=
dT
o
T
(6)
since
h = c
p
T =
c
2
1
Combining Equations 4, 5 and 6, i.e.:
dp
p
= M
2
du
u
=
d

+
dT
T
=
du
u
( 1) M
2
du
u
+
dT
o
T
or
M
2
du
u
=
du
u
( 1) M
2
du
u
+
dT
o
T
3
we obtain:
du
u
=
1
(1 M
2
)
dT
o
T
Since
T
o
T
=
_
1 +
1
2
M
2
_
we get:
du
u
=
_
1 +
1
2
M
2
_
(1 M
2
)
dT
o
T
o
(7)
Hence,
d

=
_
1 +
1
2
M
2
_
(1 M
2
)
dT
o
T
o
(8)
We can get
dp
p
and
dT
T
from the equation of state and the energy equation
as:
dT
T
=
_
1 +
1
2
M
2
_ _
1 M
2
_
(1 M
2
)
dT
o
T
o
(9)
dp
p
=
M
2
_
1 +
1
2
M
2
_
(1 M
2
)
dT
o
T
o
(10)
Since the Mach number is dened as:
M =
u
c
hence
dM
M
=
du
u

dc
c
=
du
u

1
2
dT
T
substitution of Equations 7 and 9 into the above, yields:
dM
M
=
_
1 +M
2
_ _
1 +
1
2
M
2
_
2 (1 M
2
)
dT
o
T
o
(11)
and since the stagnation pressure is dened as:
p
o
p
=
_
1 +
1
2
M
2
_

1
hence
dp
o
p
o
=
dp
p
+

1
_
1
2
_
2MdM
1 +
1
2
M
2
and rearranging, we get:
4
Fig. 2: Table showing the change in ow properties with heat addition
dp
o
p
o
=
M
2
2
dT
o
T
o
(12)
For the entropy change, we use the Tds equation, i.e.:
Tds = dh vdp
or
ds
c
p
=
dT
T

1

dp
p
and substituting Equations 9 and 10 into the above, yields:
ds
c
p
=
_
1 +
1
2
M
2
_
dT
o
T
o
(13)
Thus, Equations 7 to 13 give the variation of
du
u
,
d

,
dT
T
,
dp
p
,
dM
M
,
dpo
po
and
ds
cp
respectively. Note that, except for Equations 12 and 13, for the stagnation
pressure and entropy, all the other expressions have the term 1M
2
. Thus, the
behavior diers for subsonic and supersonic ows. Table 1 indicates whether
the ow properties increases (+) or decreases (-) with heat addition +dT
o
.
For heat extraction dT
o
< 0 and hence the behavior is reversed for subsonic
and supersonic ows. We can also integrate these equations with reference to
some reference state (choose the critical state where M = 1). We rst integrate
Equation 11 for the Mach number variation with heat addition and we then
express the other ow quantities in terms of the Mach number. Equation 11
can be rewritten as:
dT
o
T
o
=
_
1 M
2
_
dM
2
(1 +M
2
)
_
1 +
1
2
M
2
_
M
2
5
and using the method of partial fraction, we can split up the right hand side of
the above expression and obtain:
dT
o
T
o
=
2d
_
M
2
_
1 +M
2

dM
2
M
2
+
_
1
2
_
dM
2
1 +
1
2
M
2
The above can readily be integrated and we obtain:
T
o
T

o
=
2 ( + 1) M
2
_
1 +
1
2
M
2
_
(1 +M
2
)
2
(14)
The other ow properties can readily be obtained as a function of the Mach
number M as:
u
u

=
( + 1) M
2
1 +M
2
(15)
p
p

=
+ 1
1 +M
2
(16)

=
1 +M
2
( + 1) M
2
(17)
T
T

=
( + 1)
2
M
2
(1 +M
2
)
2
(18)
p
o
p

o
=
+ 1
(1 +M
2
)
_
2
_
1 +
1
2
M
2
_
+ 1
_

1
(19)
s s

c
p
= ln M
2
_
+ 1
1 +M
2
_
+1

(20)
Equations 14 to 20 have been computed for a chosen value of and a range
of values of the Mach number for both subsonic and supersonic ows. These
are tabulated in the Rayleigh ow tables. A graph showing the variations of the
ow parameters with the Mach number is illustrated in Figure 3.
The results of our analysis indicate that heat addition always drives a ow
towards sonic condition. Thus, for a given initial value of M (subsonic or super-
sonic), there corresponds a maximum value of the heat addition (i.e. stagnation
temperature increases) when the ow becomes sonic. If more heat is added
beyond this maximum value that chokes the ow, upstream conditions have to
be modied to accommodate for the additional heat addition.
6
Fig. 3: Variations of ow parameters with Mach number for Rayleigh ow
7
The Rayleigh Line
Heat addition to a compressible ow can be represented graphically on the
Rayleigh Line on the h s diagram. The Rayleigh Line is based on the conser-
vation of mass and momentum, i.e.
m = u = constant
p+u
2
= F = constant
Using the conservation of mass, the momentum equation can be written as:
p +
m
2

= F = p
_
1 +M
2
_
(21)
Thus, given initial conditions where m and F are found, corresponding values of
p for chosen values of can be obtained. Since h =
p
(1)
and the entropy
is given by:
s s
1
c
v
= ln
_
p
p
1
__

values for h and s corresponding to given values of p and can be


obtained and the locus of states along the Rayleigh Line can be plotted on
the h s diagram. The shape of the Rayleigh Line is illustrated in Figure 4.
An analytical expression for the Rayleigh Line can also be obtained. The Tds
equation gives:
Tds = dh vdp
or
ds
c
p
=
dh
h

1

dp
p
which intergrates to yield:
s s

c
p
= ln
h
h

ln
p
p

(22)
where we have used the critical condition M = 1 as the reference state. The
next task is to eliminate
p
p

in terms of
h
h

. From the equation of state, we


write:
p
p

T
T

and using the conservation of mass, we get:


p
p

=
u

u
T
T

In terms of the Mach number, the above can be written as:


8
Fig. 4: Sketch showing the shape of the Rayleigh line
9
p
p

=
u

c
u
c

c
T
T

=
M

M
_
T
T

_1
2
=
1
M
_
h
h

_1
2
where we have used c
2
= RT and M

= 1, h = c
p
T. Squaring the above
equation gives:
_
p
p

_
2
=
1
M
2
_
h
h

_
(23)
From the Equation 16, i.e.:
p
p

=
+ 1
1 +M
2
we can solve for M
2
and obtain:
M
2
=
( + 1)
p

p
1

Replacing M
2
in Equation 23 by the above, we get the following quadratic
equation for
p
p

, i.e.:
_
p
p

_
2
( + 1)
p
p

+
_
h
h

_
= 0
and the solution can be written as:
p
p

=
( + 1)
_
( + 1)
2
4
_
h
h

_
2
the negative root recovers the solution that when
h
h

= 1,
p
p

= 1. Substituting
the above expression for
p
p

into the entropy equation (Equation 22) gives the


desired equation for the Rayleigh Line:
s s

c
p
= ln
_
h
h

2
( + 1)
_
( + 1)
2
4
_
h
h

(24)
The above equation is a bit dicult to illustrate the qualitative features of the
Rayleigh Line. An alternate expression can be derived. From the Tds equation,
we write:
Tds = dh vdp = dh
1

dp
From the momentum equation, we write:
1

dp = udu
10
thus
Tds = dh +udu
from which we obtain:
h
ds
c
p
= dh +udu
and re-arranging, gives:
dh
ds
=
h
c
p
_
1 +u
du
dh
_ (25)
From the momentum equation and the equation of state, i.e.
dp
p
= M
2 du
u
dp
p
=
d

+
dT
T
=
du
u
+
dh
h
where we have used the continuity equation and h = c
p
T in the above. Com-
bining the two expressions, gives:
u
du
dh
=
( 1) M
2
(1 M
2
)
Substituting the above expression for u
du
dh
in Equation 25 leads to the following
result:
dh
ds
=
h
_
1 M
2
_
c
p
(1 M
2
)
(26)
Note that for subsonic ow where M < 1, 1 M
2
> 0, then
dh
ds
> 0 if M
2
< 1
or M <
_
1

. For M < 1 and


_
1

< M < 1,
dh
ds
is negative, thus h is a
maximum
_
dh
ds
= 0
_
at M =
_
1

. When M = 1,
dh
ds
= and for supersonic
ow where M > 1, 1M
2
< 0 but 1M
2
is also negative. Thus
dh
ds
> 0. Thus
heat addition tends to always drive the ow towards M = 1. This is referred
to as thermal choking. For reversible heat transfer, the heat addition can be
equated to Tds. Thus, the area under the Rayleigh Line represents the heat
added. If we write:
dq = Tds =
hds
c
p
= dh
o
Then, the maximum heat that can be added to a ow with initial Mach number
M is:
q
max
=
s

_
s1
h(s) ds
c
p
=
h

o
_
ho
1
dh
o
= h

o
h
o
1
= c
p
T
o1
_
T

o
T
o1
1
_
11
or
q
max
c
p
T
1
=
T
o1
T
1
_
T

o
T
1
1
_
Since:
T
o
T
= 1 +
1
2
M
2
and
T

o
To
is given by Equation 14 as:
T
o
T

o
=
2 ( + 1) M
2
_
1 +
1
2
M
2
_
(1 +M
2
)
2
the maximum heat that can be added for an initial ow Mach number M is
thus:
q
max
c
p
T
=
_
1 +
1
2
M
2
_
_
_
1 +M
2
_
2
2 ( + 1) M
2
_
1 +
1
2
M
2
_ 1
_
which simplies to:
q
max
c
p
T
=
_
M
2
1
_
2
2 ( + 1) M
2
(27)
A plot of
qmax
cpT
as a function of the Mach number M is shown in gure 4.
For a given initial Mach number M, if the heat added is greater than
the maximum value (corresponding to the initial value of M), then the initial
conditions must be modied and the mass ow rate reduced to permit the addi-
tional heat addition. For subsonic ow, information can propagate upstream to
eect this change of mass ow rate. For supersonic ow, a shock wave is formed
and propagates upstream. However, since there is no change in the stagnation
enthalpy across a normal shock, having a shock upstream in the constant area
duct would not permit more heat to be added. Thus, the shock has to advance
all the way back to the nozzle which supplies the supersonic inlet ow to the
duct in order to lower the inlet Mach number. Note that the mass ow rate
remains the same since the ow is chocked at the throat of the nozzle and the
mass ow rate remains invariant irrespective of the downstream ow conditions.
However, the inlet to the duct is subsonic and permits a larger amount of heat
addition before choking occurs. If the amount of heat addition in the duct is
still too large, then the shock propagates past the nozzle throat upstream and
the whole ow is subsonic. The mass ow can now be reduced to accommodate
the heat addition to give choking condition at the exit of the pipe.
The analysis we have carried out can be applied to the study of combustion
waves. Flames and detonations are, in general, very thin compared to the
dimension of the duct and hence can be treated as a discontinuity like a normal
shock. Heat addition occurs inside this thin front in the form of energy release
by chemical reactions. Applying the conservation laws across the combustion
12
Fig. 5: A plot of
qmax
cpT
as a function of the Mach number M
13
wave and following the same procedure as the analysis of normal shock waves,
we can determine the downstream conditions in terms of the upstream state
and the energy addition. The study of the gasdynamics of combustion waves
will be discussed in a course on combustion.

Вам также может понравиться