Introduction To Compressible Fluid Flows
Introduction To Compressible Fluid Flows
Introduction To Compressible Fluid Flows
Our topic is the fluid dynamics of compressible flows. By this we will mean the flow of perfect
gases, and not, say, the compressibility effects that can be seen, for instance, in extremely high-
speed water flows. A perfect gas, you will recall, is an ideal gas with constant specific heats.
For the most part we will also mean by this inviscid compressible flows, that is, we will neglect
viscosity, though in one notable case (flow on the so-called “Fanno line”) we will find that it is
useful to employ a formalism that depends on the friction factor.
For the most part, we will also neglect gravity, because generally speaking we are interested in
systems for which the vertical extent is small compared to the changes in the pressure head
P ρg .
Recall the relevant dimensionless group for dealing with compressible flows is the Mach
V V
number, defined by M = = .
a γ
RT
Now the thing that distinguishes the study of compressible flows from the study of
incompressible flows is that instead of the simple assumption ρ = constant , we have an
equation of state
P
= RT
ρ
that ties together the density with the pressure and a new variable, the temperature. Note R is
the gas constant for the gas in question, which is related to the universal (molar) gas constant R
through the molecular weight w , that is, R = R w . The gas constant can also be related to
Boltzmann’s constant k and the molecular mass m , i.e. R = k m .
We will find that for compressible flows, the temperature, and thus the internal energy, are bound
up closely with the dynamics of the fluid. So much more than for liquid flows, the flow of gases
must be regarded as a thermodynamic process.
mv 2
32
4 m 2
p MB (m, T , v )∆v = v exp
− 2kT
∆v
π 2kT
pMB
0.5
0
0 1
v v~ 2 3
The mean, rms, and most probable velocities are all proportional to the square root of
temperature, as is the speed of sound.
We can rewrite the MB velocity distribution in terms of the most probable speed v~ = 2 RT :
4 v
2
v 2 ∆v
p MB (m, T , v )∆v = exp − ~ ~ ,
π v~
v v
plotted above.
Now of course this means that temperature is proportional to v2 and hence to the kinetic energy
of the gas molecules.
du dh
cv ≡ , cp ≡
dT dT
where u is the specific internal energy and h is the specific enthalpy. Note
c p − cv = R
which follows from
When a gas is well approximated by a model in which cp and cv are constant, that is,
independent of temperature, we say it is a perfect gas.
Now we will also find that it is necessary to introduce the entropy, in order to talk about the
second law. For a perfect gas it is necessary to speak strictly in terms of differences in specific
entropy, since the zero of entropy is only defined at the absolute zero of temperature.
du = dq − dw
that is, the change in the internal energy is equal to the heat in minus the work out. For a
reversible process, dq = Tds . If we restrict the discussion to pressure work, dw = Pdv
(remember lower case v is the specific volume, which is 1/density). So the first law can be
written
Tds = du + Pdv
du P dT dv
⇒ ds = + dv = c v + R
T T T v
⇒ s 2 − s1 = cv ln (T2 T1 )+ R ln (v 2 v1 )
where 1 and 2 label states at the beginning and end of a process. If we apply the ideal gas law,
we can express this same relation in terms of T and P:
ln (v 2 v1 ) = ln (T2 T1 )− ln (P2 P1 )
⇒ s 2 − s1 = c p ln (T2 T1 )− R ln (P2 P1 )
Now for an isentropic process, that is, one which is adiabatic and reversible, ∆s = 0 . These
relationships thus imply
P1v1γ= P2 v 2γ
Pv γ = constant
T γP 1− γ = constant
∫∫ ( )
∂ r r
∫∫∫
dN
RTT: = ηρd 3 x + ηρ V ⋅dA
dt system ∂t
CV CS
dM ∂ r r
dt system
= ∫∫
∂t CV∫ρd 3 x + ∫∫ρV ⋅dA = 0
CS
1 2 r r
( )
dE ∂ 1 r r
dt system
= ∫∫∫ρu + V 2 d 3 x +
∂t CV 2 ∫∫ρu +
CS
V V ⋅dA = Q& −
2 ∫∫PV ⋅dA − W&
CS
other
dS
dt
=
∂
∫∫
∂t CV∫ρsd 3 x + ∫∫ ρs V(
r r
⋅d A ≥ ∑
Q&i
)
system CS i Ti
r r r
To these laws we also add the conservation of angular momentum, where N = L = r ×(mv )
N r r
and η = = r ×v , and the time derivative of the angular momentum is the sum of the
m
external torques on the system (here we ignore gravitational torques).
r
dL ∂ r r
= ∫∫∫ρ r ×V d 3 x +
dt system ∂t CV
( ) (
r r r r r r
)( r
)
∫∫ ×V V ⋅dA = r ×Fshear + τshaft
ρ r
CS
Note that in the First law the extensive property N that appears in the RTT is the sum of the
internal energy U and the kinetic energy V 2 2 . The first law for the system is that the rate of
change of this total energy is equal to the rate at which heat is added minus the rate at which
work is extracted.
Two things about the first law deserve comment. The first is that it is convenient to isolate the
pressure-surface-work contribution to the total work extraction. Other forms of work extraction
are of great importance as well, notably the extraction of work by a turbine and the negative
extraction = input of work by a compressor. In general one also has to account for shear work at
the boundary.
For the second law, note the RHS of the inequality is zero for a closed (i.e. adiabatic) system.
CS CS
1 2 r r
( )
r r
∫∫ρu +
CS
V V ⋅dA = Q& −
2 ∫∫PV ⋅dA − W&
CS
other
&
∫∫ ( ) ∑ QT
r r
ρs V ⋅dA ≥ i
CS i i
ρ r (
r r r r r r
)( r
)
∫∫ ×V V ⋅dA = r ×Fshear + τshaft
CS
It is possible to make drastic simplifications to our laws of conservation of mass and momentum
by restricting our view to uniform flows. That is, we can turn our mass flux integral into a
prescription that m& = ρVA and our momentum flux integral into a stream thrust F = ρV 2 A + PA .
Let’s make explicit the kind of flow situation we’re talking about:
Basically this means that we are going to assume that the flow properties are only a function of
streamwise distance “x”. We will let the flow area change with x, but the variation is assumed to
be gradual enough that no significant departures from 1-D flow are observed.
A=constant A=A(x)
P=P(x) P=P(x)
T=T(x) T=T(x)
rho=rho(x) rho=rho(x)
etc. etc.
Continuity
The definition of volumetric flow rate applies equally well to incompressible and compressible
flows, though the volumetric flow rate Q is far less useful for the latter simply because it is no
longer constant as gas flows through a system. The total mass is always conserved, however,
which means that for steady flows, the mass flow rate into a system must equal the mass flow
out.
Momentum
∫∫ ( ) ∫∫PdA = (ρV
r r r r
)
r
FS ≡ ρV V ⋅dA + 2
A + PA n$
S S
Here the differential area is oriented so that the surface normal points downstream, i.e. so that
r r
V ⋅dA > 0 . If our control volume has one entrance and one exit, stations 1 and 2, we can find
the stream thrusts at those stations and find the equation of momentum conservation (ignoring
Let’s try to make the same flavor simplifications to the First and Second Laws.
Energy
Now for the energy equation, it is convenient to pull the pressure term over to the other side of
the equation.
∫∫
P 1 r r
( )
ρu + + V 2 V ⋅dA = Q& − W&other
ρ 2
CS
Notice how the enthalpy h = u + P ρ enters into our formulation. Because the expression in
parentheses shows up so often, we give it a special name, the “total (specific) enthalpy”and
denote it with a subscript T. The total enthalpy is the sum of the so-called “sensible enthalpy”(or
“static enthalpy”) and the bulk kinetic energy.
P
hS = u +
ρ
V2
hT = hS +
2
Q& − W&other
⇔ hT2 − hT1 = = q − wother
m&
In words, the increase in total enthalpy must equal the heat-in minus the (non-pressure-area)
work-out. (Notation! Upper case Q and W, with a dot to indicate “time derivative”, are
energy/time, i.e. J/s or ft-lbf/s. Lower case q and w, without the dot, are specific heat and work,
i.e. J/kg or ft-lbf/lbm).
Entropy
Q&i
∑ ∑
qi
&2 s2 − m
m &1s1 ≥ ⇔ s2 − s1 ≥
i
Ti i
Ti
Angular momentum
We postpone a discussion of the angular momentum equation until we need to deal with
turbomachine stages.
Now let’s discuss a central concept in the analysis of compressible flows. Because the kinetic
energy is such a large part of the total energy of a moving compressible fluid, we must have
some way to work with it systematically in our formalism. The way do this is as follows: consider
an element of gas moving with some velocity. That element of gas is at some temperature T and
some pressure P (and all the other thermodynamic properties can be deduced from T and P),
and also the element is moving at some velocity V so it has a kinetic energy. Now consider what
would happen if that element of gas were brought to rest adiabatically and reversibly (i.e.
isentropically). That gas element would be hotter and at a higher pressure, as the kinetic energy
was converted to sensible enthalpy. We call this the stagnation condition because among the
streamlines in gas approaching a blunt body is one that is brought to rest in very much this
manner; the air that is thus brought to a rest is “stagnant”and that streamline is called the
stagnation streamline.
Stagnation point
Happily, we know exactly how to calculate the properties of that compressed state. This is just
our old friend, the isentropic compression, a Pv γ= constant process! In terms of P and T, the
P
relation is γ
= constant .
T γ− 1
For the rest of our discussion of compressible flow we will adhere to the following convention:
properties of the gas “as is”, that is, measured as we move along with it, will be called “static”
properties and given a subscript S. Properties of the stagnation state associated with a given
state of motion will be called “total”and given a subscript T.
Now the relation between the total temperature and the static temperature is determined by the
relation between the total enthalpy and the static enthalpy:
V2 V2 V2
hT = hS + ⇒ href + c p TT = href + c p TS + ⇒ TT = TS +
2 2 2c p
[h_ref is the enthalpy at T=0, or more properly, the intercept of the h-vs-T relation in the region
where h-vs-T is linear. This lets us write
dh
cp = ⇒ dh = c p dT ⇒ h = href + c p T
dT
]
Here is our first opportunity to implement the Mach number agenda. We substitute V 2 = γ
RTS
and c p = Rγ( γ− 1) ; we find
γ− 1 2
TT = TS 1 + M
2
TT γ− 1 2
= 1+ M ≡ z( M )
TS 2
“z”is the first “flow function”that we encounter. It is a function of Mach number and gamma that
is the ratio of total temperature to static temperature. Note for air, gamma is 1.4 so
z air = 1 + 0.2 M 2
2.0
1.5
1.0
z
0.5
0.0
0.00 0.50 1.00 1.50 2.00
M
Now it should be apparent that the relationship between the total and static temperature encodes
precisely the information that the enthalpy relation does. In fact, the first law can be expressed in
terms of the total temperature as follows:
q − w other
TT2 − TT1 =
cp
This is an important consideration: for the most part, energy conservation will be expressed in
terms of the total temperature.
Applying the isentropic relation, we find that the ratio of total to static pressure is
γ γ γ
PT TT γ− 1 γ− 1 2 γ− 1
= = 1 + M = z γ− 1
PS TS 2
γ
Note for air, = 35
. .
γ− 1
∆s γ T2 P
= ln − ln 2
R γ− 1 T1 P1
Thus, when gas is flowing adiabatically through a duct, so that no heat or work interactions take
place, the total temperature stays constant, though friction can cause the total pressure to drop.
We will have more to say about this fact in subsequent lectures. For now just start thinking of
temperature as the “energy variable”and pressure as the “entropy variable”.
We found expressions for the ratio of total to static for the temperature and pressure, and
introduced the flow function z.
TT γ− 1 2
= 1+ M ≡z
TS 2
γ
γ
PT TT γ− 1
= = z γ− 1
PS TS
Now let’s work over the continuity equation. Our goal is to have an expression that involves the
mass flow rate, the total temperature and pressure, the flow area, gamma, and the Mach
number.
We start with
m& = ρVA
The rho in this equation is the static density, that is, the density you get by inserting the static
temperature and pressure into the ideal gas law. The velocity is the Mach number times the
speed of sound. So
PS γ PS A
m& = M γ
RTS A = M
RTS R TS
Now we use the equations for the total temperature and pressure in terms of the static properties
and z. Pull the flow properties over to the LHS to leave a function of Mach number on the RHS.
1 γ+ 1
& TT
m γ − 2 γ− 1
= Mz
APT R
{Warning: using conventional English units requires a g_c inside the radical on the RHS:
[m&]= lbm s
[TT ]= R 1 γ+ 1
& TT
m γ
gc −
[A]= ft 2
⇒ = Mz 2 γ− 1
APT R
[PT ]= psf = lbf ft 2
[R]= ft lbf lbm R
}
Now the RHS is just a function of Mach number and gamma, and this is the second of the flow
functions we will encounter. We will call this flow function “D”. So
& TT
m
= D( M )
APT
1 γ+ 1
−
γ γ− 1 2 2 γ− 1
D( M ) ≡ M 1 + M
R 2
The flow function D encodes everything we need to know about the continuity equation, just as z
encodes information about energy conservation.
Note that, as defined, the function D is dimensional, because of the presence of the gas
constant. This means that D will have different numerical values in different systems of units.
This is what the flow function D looks like (the vertical scale here is for English units):
0.3
0.2
0.1
0
0.00 0.50 1.00 1.50 2.00
Mach
Now note a very interesting fact about D: it has a maximum value. It is not hard to show that the
maximum occurs at M=1. Note D goes to zero as M goes to infinity.
Quite generally, there is a standard notation that property and flow function values at the
condition where the flow velocity is equal to the speed of sound (i.e. when M=1), are denoted
with a star (*). Plugging M=1 into the definition of D, we find that for gamma=1.4:
D*=0.5317
0.5
0.4
0.3
0.2
0.1
0
0.00 0.50 1.00 1.50 2.00 2.50 3.00
Mach
1 γ+ 1
−
γ γ+ 1 2 γ− 1 lbm s R
D* = = 0.5317
R 2 lbf
If you work with compressible flow very much, this 0.5317 number becomes burned into your
brain.
kg s K
In SI units we find D = 0.04042
*
.
N
There is a wealth of information in this flow function. The fact that D has a global maximum
value is one of the things that makes compressible flow so intriguing.
What this means is that, given a supply of air at a given total temperature and total pressure,
there is a maximum mass flow rate which can pass through a given flow area.
In other words, if we hold TT, PT, and A fixed, then the mass flow rate is just a function of the
Mach number. If we start from zero, the mass flow rate will increase as the Mach number
increases, until the sonic condition (M=1) is reached. This is called choked flow. If Mach
number is increased beyond unity, the mass flow rate must decrease accordingly.
This may seem like a counter-intuitive result, until you consider that what is happening is that the
density is decreasing for constant TT, PT, as the Mach number increases. After M=1, the density
in decreasing so rapidly that the mass flow rate cannot increase, even though the velocity is
increasing.
Note that if the total temperature and total pressure can vary, the choking condition implies
that mass flow scales as PT and inversely with TT . In other words, a higher upstream pressure
will push a larger mass flow through a choked nozzle. A higher upstream temperature will reduce
the mass flow through a choked nozzle.
With D, we can also determine what happens when as area varies in our quasi-1D model.
Suppose we have a given mass flow rate, so and so many lbm per second, flowing in a duct.
Suppose that we know the total temperature and total pressure, for example, if the duct is fed
from a large tank of air at some temperature and pressure with negligible losses.
mdot, Tt, Pt
Then at any given station, the duct area will determine the Mach number, because we can
simply compute
and this is equal to D(M) for some subsonic Mach number. (The function D(M) must be inverted
numerically – there is no closed-form inverse solution). In fact D(M) tells us how the Mach
number will vary as the duct area is varied, because if losses due to friction and heat transfer are
small, there will be no change in the total temperature and pressure (or mdot!) along the duct.
Thus we see small Mach numbers when the duct area is large and high Mach numbers when the
duct area is small (remember, we’re assuming that the flow is subsonic).
The minimum duct area for the given conditions occurs when M=1, i.e. when D=D*. It is
m& TT
A* =
D* PT
1 γ+ 1
−
γ+ 1 2 γ− 1
A D* 2
= = 1 γ+ 1
A* D −
γ− 1 2 2 γ− 1
M 1 + M
2
but this seems redundant… all you need to know to find the flow area is D(M).]
So, lengthwise along the duct, we can sketch what happens to various flow parameters. See the
accompanying figure. The mass flow, total temperature, and total pressure are assumed to be
constant. We see that the static temperature and static pressure drop as the Mach number
increases.
You may ask the question, “What happens if we decrease the flow area below the choking flow
area?”In other words, suppose we had some mechanism for gradual adjusting the throat area.
What keeps us from cranking it down below the level that theory says can pass the mass flow
that is given?
The answer is that, as is often the case with fluid flow, it depends entirely on the boundary
conditions. Suppose we had a pump upstream that is delivering our mass flow at a particular
temperature and pressure. If we close down the throat area, something will shift… perhaps the
output pressure of the pump will rise, or perhaps the mass flow rate will drop (some combination
of the two occurs in a jet engine). In any case, the system will adjust somehow so that the global
maximum D* is never exceeded. Precisely what happens in a given situation depends on the off-
design characteristics of whatever is supplying the flow.
Now consider the possibility that the flow in some portion of the duct is supersonic. This is the
case with a “converging-diverging”nozzle. (Also called a “de Laval”nozzle by the effete).
The fact that the flow function D has two branches, a subsonic and a supersonic, means that
there are two Mach numbers that are consistent with a given mass flow, temperature, pressure,
and area.
What determines whether the flow will continue to accelerate through a nozzle or just speed up
and slow down as in our previous example? Answer: boundary conditions. If the exit pressure is
low enough, the flow will continue to accelerate through the nozzle. This is depicted in the
following chart.
For a given area ratio, there is exactly one downstream pressure that is perfectly matched with a
given upstream pressure. What happens at the off-design pressures? Generally speaking, there
will be an internal or external shock structure that provides the pressure matching. We will
discuss shocks next week and in particular examine the pressure ratio across a shock. Until then
just absorb the fact that the shocks can move around until the pressure BC’s are satisfied.
2.00
A/A*
1.50
Ts/Tt
1.00 Ps/Pt
0.50 Mach
0.00
-5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00
x
Flow properties with varying duct area
for throat Mach number = 0.60
2.00
1.80
1.60
A/A*
1.40
1.20
1.00
Ts/Tt
0.80 Ps/Pt
0.60
0.40 Mach
0.20
0.00
-5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00
x
Flow properties with varying duct area
for throat Mach number = 0.80
2.00
1.80
1.60
A/A*
1.40
1.20
1.00
Ts/Tt
0.80 Ps/Pt
0.60
0.40 Mach
0.20
0.00
-5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00
x
Flow properties with varying duct area
for throat Mach number = 1.00
2.00
1.80
1.60
A/A*
1.40
1.20
1.00
Ts/Tt
0.80 Ps/Pt
0.60
0.40 Mach
0.20
0.00
-5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00
x
Nozzles and diffusers
Finally, let’s examine the design of nozzles and diffusers for compressible flow.
A nozzle is a device that converts flow with high (static) pressure and low kinetic energy into flow
with low (static) pressure and high kinetic energy. In other words, it speeds up the flow by
converting pressure energy to kinetic energy. Because the density drops in this process, we say
that a nozzle expands the flow.
A diffuser does just the opposite: it converts flow at low (static) pressure and high kinetic energy
into flow with high (static) pressure and low kinetic energy. In other words, it slows down the flow
and converts the kinetic energy into pressure energy. Because the density rises in this process,
we say that a diffuser compresses the flow.
Now to a zeroth approximation, both nozzles and diffusers are adiabatic, reversible devices. This
means that the total temperature and total pressure are approximately constant. (To a better
approximation, total pressure is lost due to the generation of irreversibilities, but ignore that
complication for now).
The flow function D tells us exactly how the tradeoff between Mach number and area is made,
once we decide to hold mdot, PT, and TT fixed. In particular, D(M) tells us that the Mach
number:area relation is exactly opposite for subsonic and supersonic flow.
This means that the area change is opposite for subsonic and supersonic nozzles and diffusers:
Subsonic Supersonic
Nozzle
Diffuser
1 γ+ 1
−
& TT
m γ γ− 1 2 2 γ− 1
Mass: m& = ρVA ó = M 1 + M ≡ D( M )
APT R 2
Momentum: F = ρV 2 A + PA
q − wother
Energy: hT2 − hT1 = q − wother ó TT2 − TT1 =
cp
γ TT2 PT2 q
≥∑ i
qi
Entropy: s2 − s1 ≥ ∑ ó R
ln − ln
i Ti γ− 1 TT1 PT1 i Ti
Expressing the continuity equation in terms of Mach number led us to the extraordinary
revelation that there is a maximum flow rate that can be pushed through a given area at a given
temperature and pressure, or conversely, that there is a minimum flow area that will pass a given
mass flow at a given temperature and pressure. This condition occurs when the throat Mach
number is unity, and we say that the flow is choked. We also discussed the design of nozzles
and diffusers in subsonic and supersonic flow.
Now, the missing piece is the momentum equation. We will find that we can express the stream
& , PT, TT, and area just as easily as we could express continuity.
thrust in terms of m
& = ρVA , we write the individual factors in the stream thrust in terms of
Just as we did with m
pressure and temperature, and then write it as a function of Mach number.
F = ρSV 2 A + PS A
1 + γ
M2
⇒ F=
PS
RTS
( M γRTS )A + PS A = PT A z γγ− 1
2
F = PT AG
1+ γM2 1+ γ
M2
G( M ) ≡ = γγ− 1
z γγ− 1 γ− 1 2
1 + M
2
1.4
1.2
1
G(M)
0.8
0.6
0.4
0.2
0
0.00 1.00 2.00 3.00
M
Note that, unlike D(M), this flow function is non-dimensional (this should be apparent because
pressure times area is already a force). It starts out at unity, with zero slope, and rises to a
maximum at M=1, then drops to zero as Mach number goes to infinity.
1+ γ
G* = γγ− 1 = 12679
. for air
γ+ 1
2
Now what does this mean? This means that we can determine the stream thrust with as little
information as the Mach number, the total pressure, and the area. Obviously, for a given area
and total pressure, the maximum stream thrust occurs at M=1.
m& TT
Now it is useful to combine this result with the continuity equation. Since PT A = , we get
D
G
F = PT AG = m& TT
D
and it is useful to define one more flow function, usually called N(M), that is equal to D/G. Then
& TT
m
F=
N
D( M ) γ M z
N( M)≡ =
G( M ) R 1+ γ
M2
G*=1.2679
1.2
0.8
0.6
0.4
0.2
0
0.00 0.50 1.00 1.50 2.00 2.50 3.00
Mach
The Flow Function N(M)
0.5
0.4
0.3
N(M)
0.2
0.1
0
0.00 1.00 2.00 3.00
M
Note N has the same dimensions as D, and also requires a hidden g_c inside the radical when
using English units.
N(M) starts out at zero, just like D(M), then rises to a maximum value
γ
N* = = 0.4194 for air
2 R( γ+ 1)
γ− 1
N∞ = = 0.2935 for air
2 Rγ
So given a mass flow, total temperature, and Mach number, we can find the stream thrust.
Notice that, for a given mass flow and total temperature, the minimum stream thrust
occurs at M=1.
r r r
Remember, the stream thrust gives us the force on a duct according to R = F2 − F1 , with R
being the force that the duct exerts on the flow and minus R being the opposite.
N*=0.4194
0.4
0.3
0.2
0.1
0
0.00 0.50 1.00 1.50 2.00 2.50 3.00
Mach
Flow with heat transfer: The Rayleigh Line
Now there are two very important situations we can analyze with these flow functions. The first is
flow in a constant area duct with heat transfer, and the second is flow in a constant area duct
with friction. These situations, which are closely related analytically, go by special names:
For both Rayleigh flow and Fanno flow, it is possible to specify the set of properties which are
unknown in many different ways. To start the discussion, it is best to think in the following terms.
We shall suppose that all the upstream properties are known, e.g. the mass flow rate, PT1, TT1,
M1, etc. We shall assume we know something about the process in the duct, i.e. the amount of
heat transfer and the friction force, respectively, for Rayleigh and Fanno flow. Then we shall
solve for the downstream properties, using continuity, momentum, energy, etc.
Let’s examine the Rayleigh line first. Let’s label the entrance and exit stations 1 and 2.
Mdot,Tt1 Mdot,Tt2
Heat xfer q
The situation is simple: because no force acts on the fluid, the stream thrust is conserved so
F2 = F1 .
q TT q
TT2 − TT1 = ⇒ 2 = 1+
cp TT1 c p TT1
so given a TT1 and a q, we can find the total temperature ratio. In the spirit of using TT as our
“energy”variable, we will assume we will put everything in terms of total temperatures.
& TT1
m m& TT2 N2 TT2
= ⇒ =
N1 N2 N1 TT1
This is the fundamental relation that defines the Rayleigh line. It says that the ratio of exit N to
entrance N is equal to the square root of the ratio of the exit total temperature to the entrance
total temperature. There are essentially 4 variables in this equation: the two total temperatures
th
and the two Mach numbers. Given any 3, we can solve for the 4 .
For example: given an upstream Mach number, an upstream total temperature, and a heat
addition q (which determines the downstream total temperature), we find the downstream Mach
number by first computing the temperature ratio, taking the square root, and then multiplying by
Now of course, adding heat (q>0) raises the total temperature. Assuming that we start at a
subsonic M1 (i.e. M1<1), it follows from the form of N(M) that the downstream Mach M2 will be
higher. Conversely, if M1>1, then M2<M1.
In other words, heat addition has the effect of raising a subsonic Mach number and
lowering a supersonic Mach number.
This situation of comparing two Mach numbers to see which is closer to Mach 1 comes up so
frequently that it is useful to have a special symbol: we shall write M 2 > M1 when we mean that
M2 is closer to, but on the same side of, Mach 1, compared to M1. In other words, M 2 > M1 if
(M1 is subsonic, M2 is subsonic, and M2>1), or if (M1 is supersonic, M2 is supersonic, and
M2<M1). The opposite circumstance (M2 further than M1 from M=1) will be denoted M 2 < M1 .
There is another important observation to be made: given M1 and TT1 there is a maximum
amount of heat that can be added before the downstream Mach M2 equals 1! After that point,
something has to change in our boundary conditions because no heat addition can raise the
downstream N=N*. This phenomenon is called thermal choking.
More subtly, heat addition lowers the total pressure. How do we see this? Consider the other
form of the stream thrust: again F2 = F1 for heat addition without friction so
PT2 G1
PT2 AG2 = PT1 AG1 ⇒ =
PT1 G2
Since we know M 2 > M1 , it follows from the form of G(M) that G2>G1. This equation tells us
precisely the amount of total pressure lost due to heat addition.
Since heat addition convects entropy into the control volume, and since total pressure is our
“entropy”variable, it should come as no surprise that heat addition lowers the total pressure.
[Conversely, extracting heat raises the total pressure!]
Once we know the Mach numbers and the upstream properties, we can get any other properties
we desire, for example
TS 2 TT2 z 2 TT2 z1
= =
TS1 TT1 z1 TT1 z 2
and
PS 2 PT2 z 2γγ− 1 PT2 z1γγ− 1
= =
PS1 PT1 z1γγ− 1 PT1 z 2γγ− 1
Why do I call this the “Rayleigh line?”Where is the line? The line is a curve on the T-s diagram
for our gas.
[Overhead]
If we plot T/Ts* versus entropy in units of R, we see the locus on this plot. The lower line is the
static temperature Ts/Ts* and the upper line is the total temperature Tt/Ts*. We see that
1.2
Tt/Ts*
Supersonic 0.8
T/Ts*
0.6
Ts/Ts*
0.4
Subsonic
Supersonic
0.2
0
-3 -2.5 -2 -1.5 -1 -0.5 0
(s-s*)/R
increasing Tt always drives us toward the sonic condition, which is also the condition of
maximum entropy.
I have not plotted lines of constant total pressure on this chart, but if I did you could see
graphically how adding heat lowers the total pressure.
Fanno flow is very similar to Rayleigh flow. Instead of heat transfer without friction, we have
friction without heat transfer. Since friction is just another way of introducing irreversibilities into
a system, we should not be surprised that the consequences are similar.
F1 F2
Friction force R
The key observation is that friction has the effect of reducing the stream thrust. (You can
verify that if F1 and F2 are in the +x direction, then R in the –x direction implies that F2<F1.)
Now since there are no heat or work interactions, the total temperature is constant (i.e.
TT2 = TT1 ). Hence
& TT1
m
F1 =
N1 N 2 F1
⇒ =
& TT2
m N1 F2
F2 =
N2
This is the fundamental relation that defines the Fanno line. It says that the ratio of exit N to
entrance N is equal to the ratio of entrance stream thrust to exit stream thrust. Note F1 F2 > 1
so again M 2 > M1 .
In other words, friction has the effect of raising a subsonic Mach number and lowering a
supersonic Mach number.
Just as we saw the possibility of thermal choking with Rayleigh flow, we see the possibility of
friction choking with Fanno flow. That is, for given upstream boundary conditions, there is a
maximum amount of stream thrust loss (due to high friction factor and/or a long duct) before the
flow reaches M=1 and something has to change (generally, the mass flow rate must drop).
Just as in Rayleigh flow, friction has the effect of lowering the total pressure. Continuity tells us
& TT
m
=D
APT
so DPT = constant for Fanno flow. Hence we can find the total pressure ratio from
PT2 D1
=
PT1 D2
Since M 2 > M1 , it follows that D2 > D1 and thus PT2 < PT1 . Again we can plot flow properties
of the Fanno line on a T-s diagram.
1.2
Tt/Ts*
Subsonic
Ts/Ts*
0.8
T/Ts*
0.6
Supersonic 0.4
0.2
0
-1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0
(s-s*)/R
Comparison of Fanno and Rayleigh lines
Rayleigh Fanno
Heat Transfer w/o friction Friction w/o heat transfer
F2 = F1 F2 < F1
TT2 ≠ TT1 TT2 = TT1
N2 TT2 N 2 F1
= =
N1 TT1 N1 F2
thermal choking friction choking
as TT2 gets larger as F2 gets smaller
PT2 G1 PT2 D1
= =
PT1 G2 PT1 D2
(rest of analysis identical)
N 2 F1 TT2
(Note: Flow with heat transfer AND friction: F2 < F1 and TT2 ≠ TT1 . Then = etc.)
N1 F2 TT1
There is one possibility for flow in streamtubes that we have not yet explored.
Examining the assumptions behind Rayleigh and Fanno flow, we notice that it is possible to have
a situation with zero heat transfer and zero friction loss, yet which offers two possible flow Mach
numbers.
F1 = F2
⇒ N1 = N 2
TT1 = TT2
m& TT
since F = .
N
x: supersonic y: subsonic
Normal shock
0.5
0.4
0.3
N(M)
0.2
0.1
0
0.00 1.00 2.00 3.00
M
Since the flow function N has two branches, one subsonic and one supersonic, this means that
every supersonic Mach number has a subsonic dual state with exactly the same stream thrust.
Although a continuous process clearly must pass through the sonic state to get from one to the
other, there is nothing forbidding a discontinuous process connecting the dual states.
In fact, this is what we see in nature. We call this discontinuity a “shock”. We observe that a
supersonic flow can change abruptly to a subsonic flow in the space of a few mean free paths,
i.e. in a fraction of a micron. As we shall see in a few minutes, the reverse process, whereby a
subsonic flow “unshocks”to a supersonic flow, is forbidden by the second law.
This shock layer forms a sheet or surface. The properties of the flow downstream of the shock
depend on the orientation of the shock layer with respect to the upstream velocity vector. We will
begin the discussion with shocks which are perpendicular to the incoming streamlines; these are
thus called “normal shocks”.
A shock in a perfect gas bears some analogy to a hydraulic jump. We haven’t discussed these
yet, but you may have some experience with them nonetheless. Have you ever observed a high-
velocity jet of water emerging from a faucet and impinging on a flat sink bottom? It forms a high-
velocity region surrounding the impact point, but at some radius, it makes a sudden transition to
a low-velocity flow. The shallow, fast-moving layer is like supersonic flow. The deep, slow layer
is subsonic. The jump or shock is allowed because there are two solutions to the conservation
equations at that flow rate.
The conventional station designation for shocks is to label the properties immediately upstream
of the shock with a subscript “x”, and the properties downstream with a subscript “y”. The
governing condition of the shock is
Nx = Ny
0.5
0.4
0.3
N(M)
0.2
0.1
My Mx
0
0.00 1.00 2.00 3.00
M
Now, if you recall the aside from the last lecture, you will recall that the flow function N is the
only one of the 3 principal flow functions which can be inverted in closed form. So it shouldn’t
surprise you that it is possible to solve for My in terms of Mx explicitly. The algebra is a little
tedious but hardly sophisticated; the bottom line is the formula
γ− 1 2
1+ Mx
My = 2
γ− 1
M x2 −
γ
2
1.0
0.8
0.6
My
0.4
0.2 Asymptote: 0.3780 for air
0.0
1.0 1.5 2.0 2.5 3.0
Mx
Note because N has a supersonic asymptote, there is a minimum Mach number that can be seen
downstream of a normal shock.
Fanno Line
1.2
Rayleigh Line
0.8
T/Ts*
0.6
0.4
0.2
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4
(s-s*)/R
Now the flow velocity is not the only thing to change across a shock. Consider what is conserved:
we know that the mass flow rate m & must be constant. Because there are no energy interactions,
the total temperature is conserved. Because there is no friction, the stream thrust is conserved.
The conservation of these three items allows us to compute the relation between the upstream
and downstream Mach numbers using Nx=Ny.
But what about the other properties? Consider the static temperature. Upstream, the flow has
some temperature and some kinetic energy. Downstream, the flow is subsonic, so the kinetic
energy is lower, so therefore the static temperature must rise, since the total energy is
conserved. We can use the fact that the total temperature is conserved to find the static
temperature rise:
TS y zx
TT y = TTx ⇒ TS y z y = TS x z x ⇒ = >1
TS x zy
since x is supersonic, y is subsonic. Thus we see that the static temperature rises across a
shock as the kinetic energy is transformed into internal energy.
& TT
m PT y Dx
= DPT ⇒ =
A PTx Dy
We can get the same result from the stream thrust equation:
PT y Gx
PT y AG y = PTx AGx ⇒ =
PTx Gy
Dx Dy
This gives the same result since N y = N x ⇒ = .
Gx G y
1
N/N*
0.5
0
0 0.5 1
D/D*
0.9
Supersonic
0.8
0.7
0.6
N/N*
0.5 Subsonic
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
D/D*
The upper branch is supersonic, the lower, subsonic. The significance of the chart is this: The
normal shock process can be represented as a horizontal line on this chart, proceeding from the
supersonic branch to the subsonic. We see that Dy is always greater than Dx. Comparing this to
our previous result, we see that the total pressure drops across the shock. This means that the
entropy rises across the shock.
Of course, this logic is a bit inverted: The fact of the matter is that the “x”and “y”states have
different entropies, by an amount that is precisely calculable. Remember our formulation of the
second law:
γ TT2 PT2 qi
s2 − s1 = R
γ− 1 ln T − ln P ≥ ∑ T
T1 T1 i i
In the case of a normal shock with 1=x and 2=y, the total temperatures are the same, and the log
of the ratio 1 disappears; also there is no heat transfer, so we are left with
s y − sx PT y
= − ln ≥0
R PTx
The second law thus tells us that in a flow situation with constant total temperature and zero
heat addition, the total pressure must drop. With m & , Tt, and area constant, this in turn
implies that the flow function D must rise. Together with our N/N* chart, the second law thus
guarantees that shocks will always take supersonic flow to subsonic flow, and not the reverse.
To reiterate: the total pressure drops across a normal shock. A shock is therefore a
dissipative process.
PS y PT y z yγγ− 1 Gx z xγγ− 1 1 + γM x2
= = =
PS x PTx z xγγ− 1 G y z yγγ− 1 1 + γ
M y2
Since Mx>My, we see that the static pressure rises across a shock.
1.0 12
0.8 10
Pty/Ptx
My 8
0.6
6
0.4
4
0.2 Psy/Psx
2
Tsy/Tsx
0.0 0
1.00 1.50 2.00 2.50 3.00
Mx
Finally, since the normal shock process has zero heat transfer and zero friction, the endpoints lie
on both a Rayleigh line and a Fanno line, albeit for different sonic states. This is visualized in the
following plot:
0.8
Rayleigh
0.6
0.4
Line
0.2
0
-1 -0.5 0 0.5
(s-s*)/R