AE 6020

Compressible Flow
Spring 2008
Instructor:
Dr. Narayanan Komerath
Professor
Daniel Guggenheim School of Aerospace Engineering
Georgia Institute of Technology
Atlanta, GA 30332-0150
[email protected]
Lecture 1. Introduction
Welcome to AE 6020
The course website is at http://www.adl.gatech.edu/classes/ae6020/
The required textbook is
Anderson, J.D., Modern Compressible Flow With Historical Perspective
McGraw-Hill. 2nd edition or later
Panicked over pre-requisites?
Steady Aerodynamics for all speed regimes at the undergraduate level.
Engineering Calculus at the undergraduate level
Some computer programming skills
See the course website for notes on pre-requisite material.
SYLLABUS
Transonic Aerodynamics
1. Introduction
2. Compressible Potential Flow: Review
3. Transonic flow: physical features and approaches
4. Transonic Small Disturbance (TSD) Equations
5. Full Potential Equations (FPE) and solution
6. Supercritical Airfoil Design
Discussion: Laminar Flow Wing and Fuselage Design
Supersonic flow problems
1. Conical Flow
2. Blunt Body Solution
Discussion: Aerodynamic design for reduced sonic boom footprint
Hypersonic Aerodynamics
1. Introduction
2. Hypersonic Shock and Expansion Relations
3. Local Surface Inclination Methods
4. Small Disturbance Theory and Applications
5. CFD Methods for Inviscid Flow
6. Boundary Layer Theory
7. CFD Methods for Viscous Flow
8. High Temperature Effects
Advanced Supersonic Aerodynamics Topics of interest to supersonic flight.
GRADING
Assignments 30% including significant weightage for class participation.
2 Quizzes 5% each; 2 Midterm Tests 15% each
Final 30%
General Policy on Assignments: Ask anyone about methods, but do all work yourself
(or within your assigned team). Acknowledge sources of help.
Quiz 1: Covers review of compressible flow, and introduction to transonics.
Closed book and notes.
Loads on an Airfoil
C
p

p p

1
2

p
p

2
M

_
,

2
M

2
C
N
C
p,lower
C
p, upper
( )
x0
xc

d
x
c


_
,
C
X
C
p,upper
dY
dx


_
,
upper
+ C
p, lower
dY
dx


_
,
lower



1
]
1
x0
xc

d
x
c


_
,
C
l
C
N
cos C
X
sin
C
d
C
X
cos + C
N
sin
C
m,c/ 4
C
p,upper
C
p, lower [ ]
x
c

1
4


_
,
x0
xc

d
x
c


_
,
Normal
Force, N
Lift Drag
Chordwise Force, X
where Y(x) is the airfoil shape and a is the angle of attack.
Continuity Equation
Here, as we recall
Steady:
(?)
t
0
u - component
v - component
Momentum Conservation in Differential Form
Energy equation
Energy equation for steady, inviscid flows may be written as
(uh
0
)
x
+ (vh
0
)
y
+ (wh
0
)
z
= 0
When continuity is used, the above equation becomes
u h
0x
+ v h0
y
+ w h0
z
= 0
In terms of substantial derivative D/Dt,
Dh
0
Dt
= 0
where D/Dt = ?/?t + u ?/?x + v ?/?y + w ?/?z
i.e., specific total enthalpy h0 is constant along a streamline, assuming
adiabatic flows (no heat addition), and no body forces doing work on the fluid.
h
0
= Cp T + (u
2
+ v
2
+ w
2
)/ 2 = Constant
Compressible Potential Flow:
The Full Potential Equation
Introduction
Recall that for incompressible flow conditions, velocity is not large enough to cause
density changes, so density is known. Thus the unknowns are velocity and pressure. We
need two equations. The Continuity equation is enough to solve for velocity as a function of
time and space.
Pressure can be obtained from the Bernoulli equation, which comes from Momentum
Conservation.
If the flow is irrotational as well, we can define a potential, and reduce the continuity equation
to the form of the Laplace equation.
Here is defined such that

r
U
u

x
v

y
w

x

2
0
High-speed flows can also be irrotational.
Objectives
1.Derive the potential equation for compressible flow.
2.Reduce to linearized form for small-perturbation analysis.
3.Apply results to thin airfoils and slender axisymmetric bodies.
Note
Generally, flight vehicles are designed to create as little "perturbation" as possible, because large
disturbances cause large drag.
We will derive the potential equation for 2-D flows. At the end, it will be obvious how to extend it
to 3-D flow, so we will just write down the 3-D equation.
Assume:
1.Isentropic flow.
2.Steady flow.
Speed of sound
i.e.
Substitute (3a) in (2),
For homework,
1.Multiply (4a) by u, and (4b) by v and add. Call this eqn (5)
2.Expand eqn (1) and substitute in eqn (5)
Solution:
This equation contains derivatives of u and v. If we could define a potential, we could reduce the
number of variables.
Assume Irrotational Flow
If this is true, then a potential f can be defined such that

r
U

r
U
v
x

u
y
0
u

x
v

y

r
U
2
u
2
+v
2
(7)
Substituting in eqn. (6) or using the notation
What is 'a', the local speed of sound? To see this, go to the energy equation for steady adiabatic
flow.
or
a
2
+
1
2
U
2
const a

2
+
1
2
U
2
a
0
2
Thus,
or,
Extension to 3-D flow
By inspection, equations (8) and (11) can be written, for 3-D potential flow, as:
where,
Eq. 12 is the 3D Steady Full Potential Equation
a
2
a
0
2

1
2
U
2
Note:
1. This is still an exact equation. We have not made any approximations. In other
words, if the flow satisfies our assumptions (steady, irrotational, isentropic), the
solution of this will still give exact results. Note the very important distinctions
between this and the approximate equations that we will presently derive based on
"engineering estimates".
2. We have not made the assumption that disturbances are small; although we have
implicitly assumed that the shocks are absent (or quite weak). Thus, this equation
can be, and is used to calculate transonic flows over very complex configuration,
such as fighter aircraft and rotor blades.
3. This equation is highly non-linear in .
4. Superposition of solutions will not work.
5. Equally valid for subsonic and supersonic flows.
6. We have used , and
These can be easily modified, and the resulting equations can be used to analyze high-
temperature flows. (rocket exhaust nozzles)
Classic Linear Supersonic Aerodynamic Design Process
Prior Knowledge
Design Mission Objectives
Configuration Features
Initial Component Sizes
Linear Theory Cp
Calculation
Apply Real Flow
Design Constraints
Linear Theory:
Design integration
Design Optimization
LINEAR DESIGN
NO
Satisfy
Design
Constraints
?
Validate
Performance
?
NO
Wind Tunnel Test Programs
Validate Supersonic Performance
Off Design Flap Optimization
YES
Full Scale Performance
Model to Airplane Geometry Differences
Excrescence / Miscellaneous Drag
Power Effects
Scale Viscous Drag to Full Scale Conditions
Using Flat Plate Skin Friction Theory
YES
US SST and Pre 1990
Linear Design Loop
Courtesy Dr. B. Kulfan, BOEING COMPANY, 2004
Prior Knowledge
Design Mission Objectives
Configuration Features
Initial Component Sizes
Wing Leading Edge Design
Linear Theory:
Design integration
Design Optimization
CFD Analyses
Inviscid / Viscous
Apply Real Flow Design
Constraints ?
Off design Flap Parametric
Optimization
Wind Tunnel Test Programs
Validate Supersonic Performance
Off Design Flap Optimization
LINEAR DESIGN
Validate
Performance
?
NO
Satisfy
Design
Constraints
?
NO
YES
YES
Full Scale Performance
Model to Airplane Geometry Differences
Excrescence / Miscellaneous Drag
Power Effects
Scale Viscous Drag to Full Scale
Conditions Using Flat Plate Skin Friction
Theory or Navier Stokes Predictions.
Refined Linear Supersonic Aerodynamic Design Process
Initial HSCT Technology 1990
Courtesy Dr. B. Kulfan, BOEING COMPANY, 2004
Prior Knowledge
Design Mission Objectives
Configuration Features
Initial Component Sizes
Wing Leading Edge Design
Linear Theory:
Design integration
Design Optimization
CFD Analyses
Inviscid Analyses
Viscous Analyses
Viscous Determination of
the flow characteristics
Wind Tunnel Test Programs
Validate Supersonic Performance
Off Design Flap Optimization
LINEAR DESIGN
Validate
Performance
?
NO
Satisfy
Design
Constraints
?
NO
Non-Linear Design Optimization
Point Design
Off Design Flap Optimization
Multi-Point Optimization
YES
YES
Full Scale Performance
Model to Airplane Geometry Differences
Excrescence / Miscellaneous Drag
Power Effects
Scale Viscous Drag
Scale Pressure Drag ?
Current Non-Linear Supersonic Aerodynamic Design Process
HSR Technology - 1998
Courtesy Dr. B. Kulfan, BOEING COMPANY, 2004
Viscous Flows: Navier-Stokes Equation
q
t
+
F
x
+
G
y

R
x
+
S
y
q = Flow properties vector, given by
{ , u , v , E}
E = Total (Internal + Kinetic) energy per unit volume
= C
v
T + (u
2
+ v
2
) / 2
F = { u , u
2
+ p , uv , uh
0
}
G = { v, uv , v
2
+ p , vh
0
}
R = { 0,
xx
,
xy
, u
xx
+ v
xy
+ k T
x
}
S = { o ,
xy
,
yy
, u
xy
+ v
yy
+ k T
y
}
h
0
= Specific total enthalpy, C
p
T + (u
2
+ v
2
) / 2
k = Conductivity (molecular plus turbulent)
Euler Equation
Application: inviscid region of compressible flow over airfoils (and
wings)
Problems with weak shock waves & small angle of attack allow
modeling inviscid region separately from boundary layer, then
doing a viscous-inviscid interaction
q
t
+
F
x
+
G
y
0
Linearized Potential Equation
Preview:
The full potential equation is non-linear and requires numerical solution. We can however, obtain
some nice, simple engineering results by restricting our consideration to small changes in flow
variables. In this case, we can neglect terms involving products of small fractions, and simplify the
potential equation.
Applications
Analyze performance of high-speed airfoils and slender, pointed body shapes at small angles of
attack in the subsonic and supersonic regimes.
Perturbation Potentials
Let us define the potential such that,
, where
Equation (8) for 2-D flows becomes:
u

x
U

x
U

+u'
v

y
v' w

y
w'
.. (14)
Note:
This is still an exact equation, no different from equation (8). Using (14), (11) becomes:
a
2
a

2
1+
1
2
U

2
a

U
2
a

2

'




)



a
2
a

2
1+
1
2
U

2
a

2
+2u'U

2
+u'
2
+v'
2
a

2

'


)

Linearization
Suppose that
if in addition,
This is true if is not too large, i.e. excluding hypersonic flow. Eq (15) with 'a' replaced by
' ' :
Let us look at each of the terms. The second derivatives of f are of the same order of magnitude
(no reason to expect otherwise).
1.The coefficient of
xx
can be written as
if
This is true if we exclude the transonic regime, where .
The coefficient of
yy
can be written as
(exclude hypersonic flow. But then, we already did that)
if
is not too large.
The coefficient of
xy
is
we have to throw this out, or we won't get a linear equation. So, let us assume that
the other coefficients. This means that the airfoil must be very thin if is high, or
else must be very low if the airfoil is not so thin.
The linearized equation then is,
(for > 1)
(for <1)
(for <1)
These equations can be used to describe small perturbations in subsonic (but not transonic) and
supersonic (but not transonic or hypersonic0 flow.
Similarly for 3D,
Note the difference between equations (18) and (19). The sign of the first term of each
depends on whether < or > 1. If we choose the equations so that the first term is
positive, then there is a vast difference between equations (18) and (19). The character of
the solutions to these equations is vastly changed when the sign of the second term changes
from + to -.
Equation (19) describes subsonic flow, where the solution is "elliptic". i.e. changes at any
point affect the solution at every other point.
Equation (18) describes supersonic flow, where the solution is "hyperbolic". It is also called
the "wave equation", since it can be used to describe the propagation of waves (e.g. sound).
One feature of the solution is that changes cannot be felt upstream.
Pressure Coefficient for small perturbations
Recall the definition of pressure coefficient for incompressible flows
When the flow is compressible, it makes more sense to use as the normalizing
parameter. Thus c
p
can be written as,
(show this for homework, and also define limiting values, and the corresponding flow
conditions)
Let us derive a linearized form of c
p
, to use along with out small-perturbation equations.
From the isentropic flow relations,
From the energy equation,
where
or, where
h+
U
2
2
h

+
U

2
2
h
0
U
2
U

+u'
( )
2
+v'
2
The second term on the RHS is << 1. Thus,
is of the form where
This can be expanded using the Binomial theorem:
as:
where
or, for
Thus,
This is valid for a 2-D or 3-D flow over flat wings.
smaller terms
Boundary Conditions for use with small-perturbation theory
We have reduced the conservation equations, under several assumptions to a linearized potential
equation which is mathematically much simpler to solve. As in most engineering analyses the
physics of the problem are preserved by specifying the proper boundary conditions. If a solution of
the equation is to make sense, it must satisfy the boundary conditions that we specify.
: angle of attack
? : unit vector normal to the (upper) surface at a given point
: slope of the surface at a given point is tan
The linearized potential eqn. is a 2
nd
order partial differential equation: 2 boundary conditions are
required.
The perturbation must die away as we move far from the source (from the airfoil).
There can be no flow through the surface of the airfoil. The normal
component of the velocity is zero at the surface.
at the surface

r
U
r
n 0
Another way of saying this is that
or,

r
U
r
n +

n
0

r
U
r
n +
( )

r
n 0
Thus, the local flow velocity at the surface must be tangential to the surface.
is the equation of the airfoil surface.
Thus,
Approximately, since is very small.
Planar Wing Approximation
For thin wings and airfoils with small camber, at small angles of attack, it does not make much
difference whether the boundary surface is taken at the airfoil surface or at the y=0 plane.
[When you calculate you will find that the values of are small and you
will wonder why you went through the trouble of calculating it].
Thus, the boundary condition becomes,
for small (in radians) or,
Note that we have only applied the b.c. at y=0. We have not assumed that the slope of the
surface is zero.

y



_
,

y0
U

Sin U

v'
y0

y



_
,

y0
U

dy
dx
"Similarity Relations for Subsonic Compressible Flow
Airfoils(2-D flow)
Objective:Find the pressure distribution around an airfoil in subsonic compressible flow.
Method: Transform the linearized potential equation into an equation describing a
"related"incompressible flow. This will be the Laplace equation. Its solution will give the pressure
distribution around the "related" airfoil in incompressible flow. Alternately, the pressure
distribution in incompressible flow can be obtained from the extensive wind tunnel data available
in the literature(incompressible flow wind tunnels are far cheaper to run than high-Mach number
tunnels). Transform this solution back to the compressible-flow case.
Linearized potential equation for 2-D subsonic flow
(x and y subscripts denote derivatives)
or where ,
The Laplace equation describing incompressible flow is
i.e.
Equation (1) can then be transformed to(2) by setting
and
and where 'm' is some constant.
Note: The text by Anderson refers to as and .
Thus, equation (1) becomes,
or,
Note that 'm' is still undetermined. We can decide later what we want it to be.
Boundary Conditions
At infinity ,
Surface tangency
Subsonic compressible flow
where y = f (x) is the surface
Incompressible flow
where describes the surface
Note that the same is used in both flows. If we used different ones, everything cancels
out eventually:it makes no difference.
Now,
or
However, the boundary condition in the incompressible flow is
i.e.
or
Relation between surface slopes of the original airfoil and the related incompressible flow airfoil.
Note that the slope is a function of chordwise location on the airfoil. Different choices of will
give different relations between the slopes.
Pressure Coefficients
Incompressible flow:
Depending on the problem to be solved,we can make various choices of m
Consider the choice m=1
This implies that ;i.e. the chordwise pressure distributions in the incompressible and
compressible flows are the same. However, we see also that it implies that
.
(Note 0<b <1). The compressible flow airfoil must have a lower surface slope than the
incompressible-flow airfoil, in order for the pressure distributions to be the same.
The other usual question is: "How does the pressure distribution over a given airfoil change
as the Mach # is increased?"
To answer this, make a choice of 'm' that will give .This is m = b .
Prandtl-Glauert Transformation
c
p
increases with Mach #.
We have so far seen the general idea of transforming the linearized potential equation and the
boundary conditions for our problem, to an equivalent Laplace equation with appropriate boundary
conditions in another coordinate system. We saw that by leaving an "undetermined constant" in
the transformation,we have the flexibility to pick useful transformations. Below, we extend this
basic concept to the problems of a thin wing, and then to the case of an axisymmetric body (a
body of "revolution"). This is in fact a 2-D problem, because things look the same everywhere
around a given section.
Critical Mach Number
Consider the following situation
As subsonic air flow over an airfoil (or wing), it accelerates, reaches a maximum speed and then
decelerates toward the trailing edge.
Since T
0
= constant,
Thus, the Mach number of the flow increases and then decreases. The magnitude of this
change depends on the airfoil shape and the angle of attack. Thus, it is evident that, as you
increase , the highest local M on the surface may exceed 1 long
before reaches 1.
The value of at which the highest M on the airfoil first reaches 1 is called the critical
Mach number M
cr
.
Note
1.M
cr
is a value of the freestream Mach number.
2.M
cr
is less than 1 for anything with thickness at all.
Calculation of M
cr
We know that
(Isentropic flow relation)
or,
At , M reaches 1 somewhere. The value of c
p
at this point can be
found by setting M=1.
Note:
1.For any given airfoil, the value of can be found from the value of the
minimum c
p
on the airfoil.
2.If you have only low-speed data on the airfoil, the c
p
values can be converted to
values at a given subsonic Mach number using the Prandtl-Glauert relations.
Airfoils in Transonic Flow
What happens when M

> M
cr
?
The flow accelerates over the front part of the airfoil. Before the point of minimum pressure
is reached, it goes supersonic. Once the point of minimum pressure is passed, the flow
experiences an adverse pressure gradient (pressure increases downstream). Several
things happen.
The boundary layer begins to get thicker. Note that information can move upstream
through the boundary layer, because the velocity is subsonic (has to reach zero at the
wall)
The flow is forced to turn because of the thickening boundary layer. Compression
waves are formed. These merge into one (or more) oblique shocks (or a normal
shock depending on the Mach number and surface curvature, Reynolds number of
the boundary layer etc.)
The pressure rises suddenly across the shock(s). The boundary layer thickens much more. It may
separate!
The supersonic region ends in a normal shock and the flow becomes subsonic and decelerates.
Drag increases greatly because of the shocks. The wing may stall because of the boundary layer
separation.
()
Supercritical Airfoils
The preceding discussion shows that if > M
cr
, drag rises greatly. How can M
cr
be
increased?
1.Use a very thin airfoil with a sharp leading edge. This is impractical for airliners
(where would fuel be stored?).
2.Reduce the curvature on the upper surface. This reduces the acceleration and
deceleration of the flow, so that any shocks formed will be relatively weak.
Such airfoils were among the first to be designed using detailed mathematical computation.
Comparison of transonic flow over a "usual" (NACA 64A series) airfoil with transonic flow over a
supercritical airfoil.
Transonic Drag Rise
We have seen that if , shocks form above the airfoil. This causes a
large increase in drag.
In addition, the expressions for the supersonic wave drag coefficient have in the
denominator. Thus, the supersonic wave drag coefficient also becomes very high in the
transonic range. Some people believe that linear theory could be close to the truth at ,
so that . Thus, they hypothesized the existence of the "sound barrier". In
practice, does become quite high at , but fortunately stays finite, so that
powerful engines can still accelerate the aircraft through without needing infinite
thrust.
Conclusions:
1.Good aerodynamic design dictates that the surface slopes should be smooth and
gentle to avoid sharp peaks in the pressure distribution.
2.The value of must be kept high for transonic flight.
3.Flight at should be kept to a short duration.
Variation of C
l
with
M

Sweep
Obviously, it is desirable to reduce the Mach number of the flow over the airfoil section. A
long time ago, it was discovered that the flow could be "fooled" by simply yawing the wing.
It was discovered that the characteristics of the yawed wing at were similar to
those of the straight wing at .
Supersonic and Subsonic leading edges
Consider the flying-wing shown. If , the compression waves from the apex will be
felt within the angle m (mach angle). However, in the figure, every point on the leading edge of
the wing is in the undisturbed supersonic flow, and cannot feel the compression and flow
deceleration due to the other points. This is a wing with a "supersonic leading edge".
Now consider the arrow-wing shown below. Here, every point on the leading edge is within
the region of disturbed (decelerated) flow caused by the apex. As a result, the Mach number
at the leading edge is slightly smaller than the free-stream Mach number. This is a "subsonic
leading edge". Note that the flow is still supersonic.
Transonic Area Rule
Within the limitations of small perturbation theory, at a given transonic Mach number, aircraft
with the same longitudinal distribution of cross-sectional area, including fuselage, wings and all
appendages will, at zero lift, have the same wave drag.
Why: Mach waves under transonic conditions are perpendicular to flow.
Implication:
Keep area distribution smooth, constant if possible. Else, strong shocks and hence drag result.
Wing-body interaction leading to shock formation:
Observed: c
p
distributions are such that maximum velocity is reached far aft at root and far
forward at tip.
Hence, streamlines curves in at the root, compress, shock propagates out.