Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
NA sA- CR., - 173, 'ifd.'i
NASA-CR- i7382919840021782
152 PAGES84/08/00
(Dept. of Mechanical EngineeringCSS:
SAP:CIO:MAJS:
DISPLAY 01/2/184N29851*# ISSUE 20 PAGE 3125 CATEGORY 2RPT#: NASA-CR-173829 NAS 1.26:173829 CNT#: NCCl-65UNCLASSIFIED DOCUMENT
UTTL: Implementation of uniform perturbation method for potential flow pastaxisymmetric and two-dimensional bodies TLSP: Progress Report, periodending 31 May 1984
AUTH: A/WONG, T. C.; B/TIWARI, S. N.CORP: Old Dominion Univ., Norfolk, VA.
and Mechanics.) AVAIL.CASIAvail: CASI HC A08/MF A02UNITED STATES/*AERODYNAMIC CHARACTERISTICS/*AIRFOIL PROFILES/*AXISYMMETRIC BODIES/*PERTURBATION THEORY/*POTENTIAL FLOW
MINS: / INTEGRAL EQUATIONS/ LINEAR EQUATIONS/ PANEL METHOD (FLUID DYNAMICS)/PRESSURE DISTRIBUTION/ SUPERPOSITION (MATHEMATICS)
ABA: AuthorABS: The aerodynamic characteristics of potential flow past an axisymmetric
slender body·and a thin airfoil are calculated using a uniformperturbation analysis method. The method is based on the superposition ofpotentials of point singularities distributed inside the body. Thestrength distribution satisfies a linear integral equation by enforcingthe flow tangency condition on the surface of the body. The complete
ENTER: Mnpp
LIBRARY COpy
LANeL[ y :;C~f ARCH CENTERI ' ' 'J \' ':;A:\,~l \,::"-1, \f;':',ul[\JiA-.....-........_., .._'"--""_...~..'-.~------
ABSTRACT
IMPLEMENTATION OF UNIFORM PERTURBATION METHOD FOR POTENTIALFLOW PAST AXISYMMETRIC AND TWO-DIMENSIONAL BODIES
Tin Chee WongOld Dominion University, 1984
Director: Dr. Surendra N. TiwariDr. Chen-Huei Liu
The aerodynamic characteristics of potential flow past an
axisymmetric slender body and a thin airfoil are calculated by using a
uniform perturbation analysis method. The method is based on the
. superposition of potentials of point singularities distributed inside
the body. The strength distribution satisfies a linear integral
equation by enforcing the flow tangency condition on the surface of the
body. The complete uniform asymptotic expansion of its solution is
obtained with respect to the slenderness ratio by modifying and adapting
an existing technique. Results calculated by the perturbation analysis
method are compared with the existing surface-singularity-panel method
and some available analytical solutions for a number of cases under
identical conditions. From these comparisons, it is found that the
perturbation analysis method can provide quite accurate results for
bodies with small slenderness ratio. The present method is much simpler
and requires less memory and computation time than an existing surface-
singularity-panel method of comparable accuracy.
IMPLEMENTATION OF UNIFORM PERTURBATION METHOD FOR POTENTIALFLOW PAST AXISYMMETRIC AND TWO-DIMENSIONAL BODIES
by
Tin Chee WongB.S. (Mechanical Engineering) June 1982, National Taiwan
University, Taiwan, R.O.C.
A Thesis Submitted to the Faculty ofOld Dominion University in Partial Fulfillment
of the Requirements for the Degree of
MASTER OF ENGINEERINGMECHANICAL ENGINEERING
OLD DOMINION UNIVERSITYAugust 1984
Approved by:
~~,(~.Surendra N. Tiwari (Director)
Chen-Huei Liu (Co-Director)
Acknowledgments
I am deeply indebted to the co-directors of my masters program,
Dr. Chen-Huei Liu (NASA Langley Research Center) and Dr. Surendra N.
Tiwari (Old Dominion University) in directing this d;-ss~ertat;bn.Thanks
are also extended to Dr. John M. Kuhlman and Dr. Jeffrey J. Kelly, who
served as the members of the thesis committee.
My sincere thanks are extended to Dr. James F. Geer (State
University of New York, Binghamton) for offering technical assistance
during the course of this research. My sincere appreciation is due to
Mr. Scott O. Kjelgaard, Analytical Methods Branch at the NASA Langley
Research Center for assisting in the solutions of the panel method used
in this study; Mr. Richard J. Margason and his branch staif for their
encouragement and support of this research.
This work was supported by the NASA Langley Research Center under
Cooperative Agreement NCCl-65.
I also wish to thank Ms. Jacqueline Smith for her diligent and
timely typing of the manuscript.
Last but not least, it is an honor and a privilege to extend my
deepest gratitude to my parents and brother for their never-ending moral
and financial support during my entire academic career.
i i
TABLE OF CONTENTS
PAGE
LIST OF TABLES •••••••••••••••••••••••••••••••••••••••••••••••••••••••••v
LIST OF FIGURES •••••••••••••••••••••••••••••••••••••••••••••••••••••••vi
LIST OF SYMBOLS •••••••••••••• ~ •••••••••••••••••••••••••••••••• ••••••••xi
Chapter1. INTRODUCTION ••••••••••••••••••••••••••••••••••••••••••••••••• 1
2. THEORETICAL FORMULATION ••••••••••••••••••••••••••••••••••••••• 6
2.1 Perturbation Analysis Technique ••••••••••••••••••••••••• 7
2.1.1
2.1.2
2.1.3
2.2 Other
2.2.12.2.2
Axially Distributed Singularity Method foran Axisymmetric Body •••••••••••••••••••••••••••••8Line Distribution of Singularities Insidea Two-dimensional Airfoil •••••••••••••••••••••••19Second Order Joukowski Airfoil •••••••••••••••••• 32
Solution Techniques •••••••••••••••••••••••••••••• 38
Some Exact Solutions •••••••••••••••••••••••••••• 38Panel Method •••••••••••••••••••••••••••••••••••• 40
3. APPLICATION OF PERTURBATION METHOD ••••••••••• ~ •••••••••••••• 44
3.1 Perturbation Technique for SlenderBodies of Revolution •••••••••••••••••••••••••••••••••••• 44
3.2 Perturbation Technique for Thin Airfoils •••••• ~ •••••••••46
4. RESULTS AND DISCUSSION ••••••••••••••••••••••••••••••••••••••• 55
4.1 Axisymmetric Bodies •••••••••••••••••••••••••••••••••••• 554.2 Two-dimensional Airfoils •••••••••••••••••••••••••••••••• 56
5. CONCLUDING REMARKS ••••••••••••••••••••••••••••••••••••••••••115
REFERENCES ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• 117
iii
APPENDICES
A. Determination of the Coefficients of a(€) and a(€) forAxisymmetric Bodies •••••.•••••••••••••••••••••••••••••.•••••• 119
B. Boundary Conditons in Complex Plane ••••••••••••••••••••••••••123
C. Expressions for Two-dimensional airfoils ••••••••••••••••••••• 125
1. Linear Operators of the Integral Equation •••••••••••••••• 1252. Determination of the Coefficients of a(€) and a(€) •••••• 125
D. Elliptical Coordinate System ••••••••••••••••••••••••••••••••• 128
E. Simplified Expressions for Axisymmetric Bodies and Two-dimensional Airfoils ••••••••••••••••••••••••••••••••••••••••• 132
1. Axisymmetric Bodies ••••••••••••••••••••••••••••••••••••••132
1.1 Ellipsoidal Body 1321.2 Dumbbell Shaped Body ••••••••••••••••••••••••••••••• 133
2. Two-dimensional Airfoils ••••••••••••••••••••••••••••••••• 133
2.1 Elliptic Airfoil •••••••••••••••••••••••••••••••• ~ •• 1332.2 General Joukowski Airfoil •••••••••••••••••••••••••• 134
iv
TABLE
LIST OF TABLES
PAGE
4.1 Comparison of lift coefficients of the perturbation,panel and exact solutions for the symmetricalJoukowski airfoils ••••••••••••••••••••••••••••••••••••••••••• l01
4.2 Comparison of lift coefficients of the perturbation,panel and exact solutions for the camberedJoukowski airfoi1s ••••••••••••••••••••••••••••••••••••••••••• 114
v
LIST OF FIGURES
FIGURE PAGE
2.1 Axially symmetric body immersed in a uniform stream•••••••••• 9
2.2 Uniform stream past a two-dimensional airfoil withan arbitrary angle of attack •••••••••••••••••••••••••••••••• 20
2.3 Illustrating the derivation of an arbitraryJoukowski airfoil ••••••••••••••••••••••••••••••••••••••••••• 33
2.4 Body immersed in a bounded flowfield •••••••••••••••••••••••• 42
3.1 Ellipsoidal body geometry with € = 0.05 •••••••••••••••••••• 47
3.2 Dumbbell shaped body geometry with e = 0.05; b =3 ••••••••• 48
3.3 Elliptic airfoil of 10% thickness ••••••••••••••••••••••••••• 52
3.4 Symmetrical Joukowski airfoil of 6.5% thickness ••••••••••••• 53
3.5 Cambered Joukowski airfoil of 6.5% thickness •••••••••••••••• 54
4.1 Comparison of pressure distributions from theperturbation analysis method with the analyticalsolution for ellipsoidal body
(a) 0.1; cjI expanded to 2 • •••••.••••••••••.•••••• •59e = up e
(b) 0.1; cjI expanded up to 4 • ••.•••••••••..•••••••.. •60€ = €
(c) 0.1 ; cjI expanded up to 6 •••••..•..••.••..•...•..•61€ = e
4.2 Comparison of pressure distributions from theperturbation analysis method with the analyticalsolution for ellipsoidal body
(a) 0.2; cjI expanded to 2 • •••••••••••••••••••••• •• 62€ = up e
(b) 0.2; cjI expanded to 4 •••..•.••••••.•...•••••• •63e = up e
(c) € = 0.2; 4> expanded up to e.6 ••......•.............. .. 64
4.3 Comparison of the perturbation and exact solutions withthe panel method for ellipsoidal body
vi
(a) e = 0.1; ~ expanded up to e6; 60 panels with cosinespacing •••••••••••••••••••••••••••••••••••••••••••••••• 65
(b) e = 0.1; ~ expanded up to e6; 180 panels with cosinespacing •••••••••••••••••••••••••••••••••••••••••••••••• 66
(c) e = 0.1; ~ expanded up to e6; 300 panels with cosinespacing •••••••••••••••••••••••••••••••••••••••••••••••• 67
4.4 Pressure distributions from the perturbation analysisand the panel methods are compared with the exact solutionfor ellipsoidal body
(a) e = 0.05; ~ expanded up to e6; 300 panels with cosinespacing •••••••••••••••••••••••••••••••••••••••••••••••• 68
(b) e = 0.10; ~ expanded up to e6; 300 panels with cosine
spacing ••••••••••••••••••••••••••••••••••••••••••••••••69
(c) e =0.20; ~ expanded up to e6; 300 panels with cosinespacing •••••••••••••••••••••••••••••••••••••••••••••••• 70
(d) e = 0.25; ~ expanded up to e6; 300 panels with cosinespacing •••••••••••••••••••••••••••••••••••••••••••••••• 71
I
4.5 Comparison of pressure distributions from the .perturbation analysis method with the panel method fordumbbell shaped body
(a) e = 0.03; b = 2; ~ expanded up to e4; 300 panelswith cosine spacing •••••••••••••••••••••••••••••••••••• 72
(b) e = 0.07; b = 2; ~ expanded up to e4 ; 300 panels
with cosine spacing •••••••••••••••••••••••••••••••••••• 73
4.6 Comparison of pressure distributions from theperturbation analysis method with the panel method fordumbbell shaped body
(a)
(b)
4e =0.03; b =3; ~ expand~d up to e ; 300 panels withcosine spacing ••••••••••••••••••••••••••••••••••••••••• 74
e = 0.07; b = 3; ~ expanded up to e4; 300 panels with
cosine spacing ••••••••••••••••••••••••••••••••••••••••• 75
4.7 Comparison of pressure distributions from theperturbation analysis method with the exact solutionfor elliptic airfoil
(a) e = 0.1; ~ expanded up to e2; tIC = 20% •••••••••••••••• 76
(b) e = 0.1; ~ expanded up to e3; tIC = 20% •••••••••••••••• 77
vii
4.8 Comparison of pressure distributions from the perturbationanalysis method with the exact solution for ellipticairfoil
(a)
(b)
g =0.15; ~ expanded up to g2; tiC =30% •••••••••••••• 78
3g = 0.15; ~ expanded up to g ; tiC = 30% ••••••••••••• 79
4.9 Comparison of pressure distributions from theperturbation analysis method with the exact solution forelliptic airfoil
(a) g = 0.05; , expanded up to g3; tiC = 10% ••••••••••••••80
(b) g = 0.10; , expanded up to 3; tiC = 20%••••••••••••••81g
(c) g =0.15; , expanded up to g3; tiC = 30% ••••••••••••••82
4.10 Comparison of pressure distributions from the perturbationanalysis method with the panel method for symmetricalJoukowski airfoil
(a) g = 0.03248; y = 00 ; , expanded up to g2;tiC = 6.5%; 10 panels with cosine spacing ••••••••••••••83
(b) g = 0.03248; y = 00 ; , expanded up to g2;tiC =6.5%; 30 panels with cosine spacing ••••••••••••••84
(c) g = 0.03248; y = 00 ; , expanded up to g2;tiC = 6.5%; 50 panels with cosine spacing •••••••••••••• 85
4.11 Comparison of pressure distributions from theperturbation analysis method with the panel method forsymmetrical Joukowski airfoil
(a) g = 0.03248; y = 60 ; , expanded up to g2;tiC = 6.5%; 10 panels with cosine spacing ••••••••••••••86
2(b) g = 0.03248; y = 60 ; , expanded up to g ;tiC =6.5%; 30 panels with cosine spacing ••••••••••••••87
(c) g = 0.03248; y= 60 ; , expanded up to g2;tiC =6.5%; 50 panels with cosine spacing ••••••••••••••88
4.12 Comparison of pressure distributions from the perturbationanalysis method with the panel method for symmetrical Joukowskiairfoil
(a)
(b)
g = 0.01624;tiC = 3.25%;
€ = 0.01624;tiC = 3.25%;
2y = 00 ; , expanded up to g ;
50 panels with cosine spacing •••••••••••••89
y = 60 ; , expanded up to g2;50 panels with cosine spacing •••••••••••••90
viii
(c) 2£ = 0.01624; y = 120; ~ expanded up to £ ;tIC = 3.25%; 50 panels with cosine spacing ••••••••••••• 91
{a) £ = 0.04872;tIC =9.75%;
(b) £ = 0.04872;tIC = 9.75%;
(c) £ = 0.04872;tIC = 9.75%;
4.13 Comparison of pressure distributions from the perturbationanalysis method with the panel method for symmetrical Joukowskiairfoil
(a) £ = 0.03248; y = 00; ~ expanded up to £2;tIC = 6.5%; 50 panels with cosine spacing •••••••••••••• 92
(b) £ = 0.03248; y = 60; ~ expanded up to £2;tIC = 6.5%; 50 panels with cosine spacing•••••••••••••• 93
(c) £ = 0.03248; y = 120; ~ expanded up to £2;tIC = 6.5%; 50 panels with cosine spacing •••••••••••••• 94
4.14 Comparison of pressure distributions from the perturbationanalysis method with the panel method for symmetrical Joukowskiairfoil
y = 00; • expanded up to £2;50 panels with cosine spacing ••••••••••••• 95
y = 60; ~ expanded up to £2;50 panels with cosine spacing ••••••••••••• 96
y = 120; ~ expanded up to £2;50 panels with cosine spacing ••••••••••••• 97
4.15 Comparison of pressure distributions from the perturbationanalysis method with the panel method for symmetricalJoukowski airfoil
(a) £ = 0.06496; y = 00; ~ expanded up to £2;tIC = 13%; 50 panels with cosine spacing••••••••••••••• 98
2(b) £ = 0.06496; y = 60; ~ expanded up to € ;
tIC = 13%; 50 panels with cosine spacing ••••••••••••••• 99
(c) £ = 0.06496; y = 120; ~ expanded up to £2;tIC = 13%; 50 panels with cosine spacing •••••••••••••• 100
4.16 Comparison of pressure distributions from the perturbationanalysis method with the panel method for camberedJoukowski airfoil
(a) £ =0.02688;tIC = 3.25%;
(b) £ = 0.02688;tIC = 3.25%;
(c) £ = 0.02688;tIC =3.25%;
2y = 00; ~ expanded up to £ ;
50 panels with cosine spacing•••••••••••• l022
y = 60; ~ expanded up to £ ;50 panels with cosine spacing •••••••••••• 103
y = 120; ~ expanded up to £2;50 panels with cosine spacing •••••••••••• l04
ix
(a) £ = 0.08064;tIC = 9.75%;
(b) e = 0.08064;tIC = 9.75%;
(c) e = 0.08064;tIC = 9.75%;
4.17 Comparison of pressure distributions from the perturbationanalysis method with the panel method for camberedJoukowski airfoil
{a} £ = 0.05376; r = 00; ~ expanded up to £2;tiC =6.5%; 50 panels with cosine spacing ••••••••••••• 105
(b) £ = 0.05376; r = 60; $ expanded up to £2;tIC = 6.5%; 50 panels with cosine spacing ••••••••••••• 106
(c) e = 0.05376; r = 120; $ expanded up to e2;tIC = 6.5%; 50 panels with cosine spacing ••••••••••••• 107
4.18 Comparison of pressure distributions from the perturbationanalysis method with the panel method for camberedJoukowski airfoil
r = 00; $ expanded up to e2;50 panels with cosine spacing •••••••••••• 108
r = 60; $ expanded up to e2;50 panels with cosine spacing •••••••••••• 109
r = 120; $ expanded up to e2;50 panels with cosine spacing ••••••••••••110
4.19 Comparison of pressure distributions from the perturbationanalysis method with the panel method for camberedJoukowski airfoil
(a) e = 0.10752; r = 00; $ expanded up to e2; •••••••••••• 111tIC = 13%; 50 panels with cosine spacing
(b) e = 0.10752; r = 60; ~ expanded up to e2;tIC = 13%; 50 panels with cosine spacing •••••••••••••• 112
(c) e = 0.10752; r = 120; $ expanded up to e2;tIC = 13%; 50 panels with cosine spacing •••••••••••••• 113
D. Two families of elliptical and hyperbolic curves withdifferent values of ~ and n •••••••••••••••••• ~ •••••••••••• 129
x
a
a .J
A
b
cC(x)
C(n)(x)
e~, en' :13f(x,e), f(x,€)
f n(x), fmn (x)PGj , L
j, L
j
(h,k)
I(x,€), IP(x,€)
1m [F(x)]
LIST OF SYMBOLS
radius of a circle in ~-p1ane; major semi-axislength of an ellipse
coefficients of a binomial expansion
constant determines the curvature of a camberline
minor semi-axis length of an ellipse; profileparameter of a dumbbell shaped body
coefficients of power series expansions
chord
camber distribution
nth derivative of C(x) with respect to x
lift coefficients from the exact, panel andperturbation analysis methods respectively
pressure coefficient
diameter of an axisymmetric body
differential surface area
unit vectors in elliptical coordinate system
singularity strength
functions in singularity strength expansions
linear operators, (p = 0,1)
coordinates of a point in ~-p1ane
scale factors of the elliptical coordinatesystem
integral operators, (p = 0,1)
imaginary part of a function F(x)
xi
k
L
L(e:)
n
p
p00
(r,a,x)
Re [F(x)]
S(x)
s(n)(x)
t
v
(xI,x2,x3)
Z(l;)
a( e:), B( e:)
ak' 13k
r
y
6
€
e: l
II
(~,ThB)
P
ratio of b to a
body 1ength
1i ft force
unit outward normal vector
pressure
static pressure
cylindrical coordinate system
real part of a function F(x)
thickness distribution
nth derivative of S(x) with respect to x
maximum thickness of an airfoil
surface velocity
uniform free-stream velocity
flow velocity
Cartesian coordinate system
transformed point in Z-plane
constants determine the extent of thesingularity distribution
coefficients of expansionsfor a(e:) and B(e:) respectively
strength of circulation
angle of attack
camber line parameter
slenderness ratio
perturbation parameter
doublet strength per unit area
elliptical coordinate system
fl ui d dens ity
xii
1/1
1/1.J
1/10
SUBSCRIPTS
s
T
source strength per unit area
velocity potential
velocity potential in the imaginary internalflowfield
perturbation velocity potential
incident velocity potential
stream functi on
coefficients of 1/10 expansion
incident stream function
boundary surface
trailing edge
free-stream condition
xiii
1
Chapter 1
INTRODUCTION
In recent years, applications of powerful and sophisticated
computers are being made extensively to analyze and solve important
aerodynamic problems. Computer codes are now being developed to
simulate the .flowfield over entire aircraft configurations by
considering the boundary-layer and full-potential formulations. For
realistic physical applications, it would be desirable to work with the
full Navler-Stokes equations. However, this requires large computer
resources (memory and execution time) which makes the computational
costs prohibitive for many applications. For this reason, the former
method is highly desirable for solving many important aerodynamic
problems. Thus, potential solutions become essential in analyzing many
flow problems. Although it would be desirable to obtain potential
solutions for the complete aerodynamic configurations, it is usually
convenient to analyze different basic components of the system
separately. In analyzing the flowfield over an airplane, it is
customary to divide the entire problem into the three separate problems
of fuselage, wing, and tail. Thus, separate programs are developed to
analyze the flow over the fuselage, wing and tail. The analyses of
axisymmetric bodies and two-dimensional airfoils are considered as the
first step for fuselage and wing designs respectively. The two
dimensional airfoil results are modified usually to account for the
2
effect of three-dimensional wing flow.
The incompressible potential flow over axisymmetric bodies and
wings is essentially governed by the Laplace equation. Although the
Laplace equation is one of the simplest and well known of all partial
differential equations, the number of known closed-form solutions is
quite small. This is due to the non-homogeneous boundary conditions for
the partial differential equation and the requirement for the body
surface to be a coordinate surface of the one of the orthogonal
coordinate system for which the Laplace equation can be solved by
separation of variables. As a consequence, approximate solutions become
more useful in practical applications. One typical example 1s the
slender-body theory where the body boundary condition is simplified for
small slenderness r~tio and angle of attack. In many important fields
successful approximations are found and some of these are discussed in
[1]*. The present study is an application of the slender body th·eory.
Many previous investigators have attempted to solve the potential
flow problem around a slender body. The idea of using a distribution of
singularities inside the body was first treated by von Karman [2]. For
the axisymmetric motion, he used a continuous distribution of sources
along the axis of the given body and gave a method for computing the
distri"bution strength. For the lateral flow past a given body he used a
continuous distribution of doublets along the axis with the doublet axes
normal to the body axis. It is difficult to obtain a closed-form
solution for the exact problem; however, an approximate closed-form
solution may be obtained, if the shape of the body satisfies certain
*The numbers in brackets indicate references.
3
conditions. The body should be slender and should not have any
discontinuities in its surface slope. Zedan and Dalton [3,4] and
Kuhlman and Shu [5,6] extended the method by dividing the singularity
distribution into elements and employing piecewise linear or polynomial
functions to represent the variations of the intensity of the source
distribution over each line element. The effects of the order of the
distribution, the number of elements, the normalization of the body
coordinates, the slenderness ratio and the geometry of the profile on
the performance of the method were studied in detail. It was concluded
that each of these factors should be chosen appropriately in order to
have accurate results.
Another approach to obtain approximate soluti~ns to the slender
body problem is to use the perturbation method. In 1967 Handelsman land
Keller [7] introduced a method to give a uniformly-valid approximate
solution of axially symmetric potential flow containing a small
parameter, known as slenderness ratio £. The source strength distri
bution satisfied a linear integral equation. A uniform asymptotic
solution to the integral equation was obtained. The left-hand side of
the integral equation was expanded in the power series of e2 and the
right-hand side in the power series of e2 and a power series in e2
multiplied by e2(log e2) without taking account of the dependence of the
source strength on £. It was assumed also that the source strength
vanishes in an interval near each end of the body and that the profile
curve of the bodY is analytic. It was possible to give a complete
uniformly-valid solution to any order of approximation in this way. The
method used to determine the axially symmetric potential flow about a
slender body of revolution 1s presented in [7].
4
The axial singularity method can be applied to a two-dimensional
airfoil in a similar manner to solve the potential flow problem.
Similarly, one can have an alternative approach that will simplify the
mathematical conditions of the problem and enable the construction of an
approximate closed-form solution. Such an approach is called the
classical thin airfoil theory. The thin airfoil theory breaks down near
stagnation points and near the edges of the airfoil. This is because
the assumption of small perturbation is not valid in the neighborhood of
such points; the perturbation velocity 1s of the same order of magnitude
as the undisturbed velocity. For a survey.of these difficulties and the
methods that are developed to relieve them one should make reference to
Van Dyke [1]. The first approach is the method of matched asymptotic
expansions and the second is the method of strained coordinates [8].
The method of strained coordinates has been modified by Hoogstraten
[9]. However, these methods are so complicated that only the first two
terms of a uniform expansion are generally obtained. Geer and Keller
[10] presented the method to obtain a uniform asymptotic solution for
potential flow around a thin symmetric airfoil by adopting the method of
Handelsman and Keller [7]. This idea was extended and modified by Geer
[11] to obtain the complete uniform asymptotic expansion for the non
symmetric airfoil. This pertur~ation method 1s presented in Chap. 2.
Since the method to obtain the complete uniform asymptotic expansion for
potential flow past axisymmetric and two-dimensional bodies is known,
the only interest here 1s the assessment of the perturbation method.
Such questions as: how large can the slenderness ratio be or what are
the advantages and disadvantages of this method over other existing
methods are addressed. In the light of these considerations, some
5
programs have been developed to obtain the pressure distributions for.
slender axisymmetric bodies and thin airfoils. In order to make
comparisons, some exact solutions and solutions obtained by using an
existing surface-singularity-panel method have been formulated and
presented in the latter part of Chap. 2. Much effort has been devoted
to obtain the expressions of the asymptotic expansions for the potential
function, with the emphasis on some axisymmetric slender bodies and thin
airfoils, and these are presented in Chap. 3. The pressure distrbutions
on axisymmetric bodies and airfoils obtained by the perturbation
analysis method are compared with the results generated by the panel
method as well as exact solutions whenever available. These results are
presented in Chap. 4. The perturbation analysis method can be applied
to solve the electrostatic and magnetostatic potentials around a slender
conducting body [10-13]. Recently, Homentcovschi eve~ modified this
method to solve two-dimensional elasticity problems [14].
6
Chapter 2
THEORETICAL FORMULATION
Basic potential flow theory is reviewed in this chapter. The
problem under consideration in this study is that of the steady
irrotational flow of .incompressible, inviscid fluid, for which the
Nav1er-Stokes eqaut10ns can be reduced to the classical Laplace
equation. Consider a body immersed 1n an ideal fluid and a space-fixed
reference frame. Then the surface of the body is described by
(2.1)
The variables xl ,x2 and x3 are the Cartesian coordinate system used in
the analysis. Because the fluid is incompressible, the continuity
equation implies that the divergence of velocity must be zero
v.v = 0
Helmholtz's theorem states that the vorticity of an initially
1rrotat1onal, 1nv1scid fluid is zero. Mathematically, one has
vxV = 0
and it follows that
(2.2)
(2.3)
v =v,
7
(2.4)
Here, is the velocity potential. The substitution of Eq. (2.4) into
Eq. (2.2), gives the governing Laplace differential equation for
incompressible potential flow, i.e.,
In order to solve for the potential" Eq. (2.5) should satisfy the
boundary conditions
v+. vF = 0 (2.6)
Also, the components of the velocity should vanish in a certain way with
distance from the body as that distance tends to infinity [15]_, The
Bernoulli equation relat1ng pressure and velocity is given by
(2.7)
Once the velocity on the surface of the body, Us' is determined, the
pressure distribution can be computed through Eq. (2.7).
2.1 Perturbation Analysis Technique
A large number of perturbation techniques to obtain approximate
solutions to physical potential flow problems is available in the
literature [1,7-12]. In this chapter, the special perturbation
8
technique developed by Handelsman and Keller [7,10-12] is reviewed in
detail.
2.1.1 Axially Distributed Singularity Method for Axisymmetric Body
Consider the steady state of incompressible, inviscid and
irrotational fluid past an axially symmetric rigid body which is at zero
angle of attack to the uniform stream, see Fig. 2.1. It is convenient
to analyze the problem by introducing cylindrical coordinates (r,e,x)
with origin at the body·s nose and the x-axis along its line of
synmetry. The surface of the body 1s described by an equation r
= £ R(x), where £ is the slenderness ratio, i.e., the ratio of the
maximum radius of the body to its length. If one uses the length of the
body as a unit of length, then the body intercepts the x-axis at x~O and
x=1. Then, the maximum value of R(x) in absolute value is one. In this
chapter, the slender body of revolution is considered, and e isI
considered to be a small value.
Since it is assumed that the flow is irrotational, the velocity
potential surrounding an axially symmetric body is given as
(2.8)
where .0 is the potential of the incident flow while .b is the
perturbation potential. The perturbation potential .b is due to the
presence of the body, and this is to be determined. This is represented
by a superposition of point sources distributed along the x-axis inside
the body. The boundary conditions for this problem are the no
penetration (or tangent-flow) condition:
Y/L
u .~
Z/L L: BOuY LEUGTH
r = E Rex) =€~S(X)
/.. _-- ..... - .......... ..... --- -- -- .......
{3(E) °1.0 xlL
Fig. 2.1 Axially symmetric body immersed in a uniform stream.
V<I» • ii = 0
and the infinity condition:
on r = eR(x) (2.9)
10
as r + infinity (2.10)
where nis the unit outward normal vector to the surface of the body.. b 0Due to the axial symmetry of the flow, both <I» and <I» are independent
of a and the equation of the profile curve depends on e, so that .b is a
function of x, r2 and e. In this study, our aim is to obtain an
asymptotic expan.sion of .b with respect to e2 around e = O. We define
S(x)=R2(x) and then the cross-sectional area of the body at x is
we2S(x). Assume that Sex) is analytic on 0 < x < 1, with S(O) =0 =S(1) and it can be expanded in power series about endpoints as follows:
•Sex) =r cnxn
n=1
•Sex) =r d (l_x)n
n=1 n
(2.11)
(2.12)
where cn = S(n)(O) }n! and dn = (_l)n S(n)(l) In! • It is assumed that
cl and d1 are non-zero, i.e., the radius of curvature at each end of the
bodY is not zero. The perturbation potential .b is used to represent
the superposit10n of point sources distributed along a segment of the x
axis 1nslije the body with unknown source strength density f{x,e) per
11
unit length [7]. Thus,
(2.13)
where aCe) and a(e) are constants which determine the extent of the
source distribution and must be found in addition to f(x,e), and they
satisfy the inequalities 0 < a < a < 1.
The potential function +can be related to stream function ~
by 'x = -r'r and 'r =r.x• Thus, one can rewrite Eq. (2.13) as
( 2) o( 2) _ 1 fB(E) (x-t) f(ttE) dt, x,r;e =. x,r ~ 2 2 1/2~T aCe) [(x-~) + r ]
(2.14)
The boundary condition, Eq. (2.9), implies that there is no flow of
flu~d through the surface of the body, i.e.,
a(e)J f(~,e}d~ =0aCe)
(2.15)
Since the body is a continuation of the axial streamline for the total
flow, one can obtain
2,[x,e S(x};e ] =0
Using Eqs. (2.15) and (2.16) 1n Eq. (2.14), it follows that
(2.16)
(2.17)
Equation (2.17) is a linear integral equation from which f(t,e}, aCe)
12
and 8(£) need to be determined. The uniform asymptotic expansion of
the solution of this equation is obtained by adopting the method of
Handelsman and Keller [7]. Since the left side of Eq. (2.17) is
analytic in x for 0 < x < 1, f(x,£) must be analytic in its domain of
definition a , X < 8, and the coefficients in the expansion of f(x,e)
with respect to e2 are assumed to be continuous in the interval
o < x < 1. Both ~ and B are found to be power series in e2 of the form
B(e) =1 - IB e2n
n=1 n
(2.18a)
(2.l8b)
In order to obtain an asymptotic expansion of the source strength:
density f(x,e) of Eq. (2.17) with respect to e2, one first expands each2 .
side of Eq. (2.17) with respect to e without taking account of the
dependency of f on e2• The left side can be expanded in a power series
in e2 because .0 is analytic in r2• The right side can be expanded
asymptotically in powers of e2 and powers of e2 multiplied by
10g(e2) with coefficients which are linear expressions in f. When both
sides are expanded, Eq. (2.17) becomes
aD j . x S(e)4'1 L"'j(x)S (x)e2j = I f d~ - I f d~
j=l (1(£) x
(2.l9a)
where
13
(2.19b)
In Eq. (2.19a) Lj and Gj are l1near operators defined by Eqs. (4.13)
(4.15) in [7], with e replaced by e2• Upon differentiating Eq. (2.19a)
with respect to x, and by noting that
(2.20)
there is obtained
CD d - 2j4w.L dx [.J-(x)sJ{x)] £ = 2f(x,e)
J=1
CD 2j d 2+ Led (L.+ log e G.)f
j=l x J J(2.21)
To solve Eq. (2.21), the asymptotic solution for f 1s supposed to be of
the form
CD n-1 2f(x,e) = L Len (log e2)m fnm(x)
n=1 m=O(2.22)
where fnm{x) are functions of x and must be determined by substituting
Eq. (2.22) into Eq. (2.21), one obtains
(2.23)
14
For each nand m , the coefficients of &2n(10g &2)m on both sides of Eq.
(2.23) can be obtained as follows:
f 1m = 0 m > 1
n ) 2
(2.24a)
(2.24b)
(2.24c)
(2.24d)
The fnm(x) can be determined recursively by starting with n=O, m=O.
Thus, once the Lj and Gj are evaluated, Eqs. (2.22)-(2.24d) will yield
the desired asymptotic solution for f(x,&) in terms of coefficients 'n.
Inserting Eq. (2.22) into Eq. (2.13), the asymptotic expansion of + is
obtained as
( 2) o( 2·, x,r ;e = + x,r)
1 • n-1 2n 2 S(e) fnm(~) d~- 4;" r r &. (log & )m f
T n=l m=O m(&) [(x-t)2 + r 2]172 (2.25)
The method to obtain the asymptotic expansion with respect to £
about e = a of the integral operator in Eq. (2.17) is available in
[7].
Let the integral operator applied to a function F(x) which is
independent of e be
15
(2.26)
where a(e) and B(e) ,are given in Eqs. (2.18a) and (2.18b). By adding
and subtracting integrals to Eq. (2.26), one finds
I (x, e:) = / F{ t)dt - l F{ t)dt + / F(t) [ 2{x-t~ 1/2 -11dt~ x ~ [(x-t) + e: Sex)] J
(2.27)
On the right-hand side of Eq. (2.27), one sets x-~=v and ~-x=v to the
third and fourth integral respectively. Then, Eq. (2.27) becomes
x a(e) ,I(x,e) = J F(~)d~ - J F(~)d~ + W(x,e) + V(x,e)
a(e) x
Here, W{x,e:) = fX-~ F{x-v) [v{v2 + e:2S)-1/2 - 1] dvo
a-xV{x,e:) = - f F{x+v) [v{v2 + e:2S)-1/2 - 1] dv
o
(2.28)
(2.28a)
(2.28b)
To find the asymptotic expansion of Wand V, the binomial expansion is
considered
~ :(2.28c)
where aj = {-l)j (.l-){l) {~ + j - l)/j! • ThiS expansion is not
valid throughout the domains of integration because these domains extend
to v=O. Therefore, special treatment s~ould be applied. To F(x+v) in
the integrand of Eq. (2.28b), one adds and subtracts the two leading
16
terms in its Taylor series at v=O. To [v(v 2 + e2S)-1/2 - 1], one again
adds and subtracts the leading terms of its binomial expansion. Thus,
Eq. (2.28b) can be written as
B-x 1 B-X-V(x,e) = F(x)! [v(v 2+e2S)- /2 -1]dv + F'(x)! v[v(v 2+e2S)-1/2 -1]dv
o 0
s-x 1 2 j 1 (j ) j+ ! [v (v 2+e2S)-1/2 _ r a. (~) ][F(J(+v) - r F (x ~v /j! ]dv
o . j =0 J v2 j =0 - 1.1 .
s-x 1 (j) j+ a1e2S! [F(x+v)- r F (~)v ]v-2dv (2.29)
o j=O J.
An analysis of the order of accuracy is described in Appendix A of
[7]. It shows that the first two integrals are O(e) and 0(e210g e2),
respectively. The third term is 0(e2), but the fourth integral
is 0(e3), so that it is asymptotically negligible compared to the three
terms. To obtain the asymptotic expansion of the fourth integral in Eq.';
(2.29)~ one can apply the ~ame procedure. To the first factor of the
integrand, one adds and subtracts the second term in the binomial
expansion" given by Eq •. (2.28c). To the second factor one can add and
subtract the next two terms of the Taylor expansion of F{x+v) about
v=O. In this manner, a sum of integrals 1s obtained; some of which can
be evaluated explicitly, others can be expanded as power series
in e2, and the remaining term is asymptotically small compared to the
other terms. The same procedure can be applied to the remainder, and
this can bel done repeatedly. Thus, V{x,e) can be expressed as a series
of integrals which are successively smaller in e2 as e tends to zero.
Furthermore, W(x,e) can be treated in an exactly analogous way, so one
obtains
17
•W(x,e) + V(x,e) = L anSn(x)e2n[Hn(x,e) - H (x,e)]
n=l n
• F(2n)(x)+ n~1 (2n)1 [Pn(x,e) - Pn(x,e)]
• F(2n+l)(x)- n~o (2n+l)1 [Kn(x,e) + Kn(x,e)]
where, Hn 'Pn ,Kn are defined by
(2.30)
X-CI 2n-l (j) j= f v-2n[F(x_y) - L F ~~)Y] dv
o j =1 J ·(2.30a)
X-CI 2 2 2 1/2 n 2S jPn(x,e)=f vn[v(Y+eS)- - L a·(T)]dv
o j=O J v
x n 2 j= f y2n+1[v(y 2+e2S)-1/2 - L a.(e ~)]elv (aO=I)
o j=O J v
(2.30b)
(2.30c)
'V 'V
The ifunctions Hn, Pn, and Kn are defined by Eqs. (2.30a), (2.30b) and
(2.3Oc) respectively, with (x-a) replaced by (B-x). To determine the
asymptotic expansion of I(x,e), one must expand asymptotically ,the
integrals Eqs. (2.3Qa) and (2.3Ob). Hn(x,e) and Hn(x,e) can be expanded
directly as Taylor series in e2 about e = O. The integrals Eqs. (2.30b)
and (2.30c) have been analyzed in detail 1n Appendix B of [7]. Appe;ndi x
A in [7] shows that P = 0(e2n+1), K = 0(e2n+210g e2) and the samen n'V 'V ~
holds true for P and K. However, in the difference P - P , whichn n n n
occurs in Eq. (2.30), the odd powers of £ will be cancelled. By using
the results of Appendix B in [7] to Eq. (2.30), one finds that I(x,e)
has the asymptotic expansion
I{x,e) = W{x,e) + V{x,e) = re2j[L.+{109 e2)6.] F{x)j=1 J J
(2.31)
18
Here, Lj and Gj are linear operators defined by Eqs. (4.13)-{4.15) in. 2 rw rw rw
[7], with e replaced bye. The functions 9k' 9k,t ij , t ij , hk, hk in
Eq. (4.14) are defined by Eqs. (BI0)-(BI7) in Appendix B of [7]
when e is replaced by e2•
We now return to determine the coefficients an and Bn in the
expansions of a(e) and 8(e), respectively. From Eq. (2.24a) it follows
that fl0(x) is analytic on 0 < x < 1, because '1 is analytic and 5(x) is
analytic with 5(0) = 0 = 5(1) and 5 1 (0) # 0 # 5 1 (1). It then follows
that from the recursive nature of Eqs. (2.24a)-(2.24d), the function
fnm{x) will be analytic for 0 < x < 1 if both GjF{x) and LjF{x) are
analytic, provided that F(x) is analytic. This will be true if all of
these functions 9k' gk hk, hk are analytic. The functions 9k(x)
and 9k{x) are defined in Appendix 8 of [7] as
g(x , e)m 2k 2 1/2= r gk{x)e = {[x-~{e)] + e Sex)}
k=O(2.32a)
(2.32b)
where a(e) and 8(£) are defined in Eq. (2.18a) and (2.18b) respectively.
The functions gk{x) depends on the coefficients ~ in the expansion
of ~(e). They are singular at x=O except for certain values of ~, so
the appropriate ak should be chosen such that 9k(x) can be regular at
x=O. To relate gk{x) with ~, one can square both sides of Eq. (2.32a)
and equate the coefficients of the same power of e. Similarly, 9k{x)
can be related to Bk at x=l. Detailed derivation of ak and Bk is
19
available in Appendix A. Using the results Eqs. (A9) and (All), the
expressions for a(e) and S(e) are given by
a( e) (2.33)
(2.34)
The functions hk(x) are defined in terms of the square root of gk(x).
Once the ak are chosen appropriately to make gk(x) analytic, then it#v
implies that hk(x} are also regular. Similarly, the hk(x) are also
regular at x=l, since 9k(x) are made analytic. Therefore, LjF(x); Gj(x)
and fnm(x) are analytic provided that F(x) is analytic on 0 < x , 1.
The asymptotic expansion of +(x,r2;e) can be obtained by using the
appropriate fnm(x), a(e) and S(e) which are determined by Eqs. (2.24a)
(2.24d), (2.33) and (2.34) respectively in Eq. (2.25). Once the
potential of the flowfield is found, the aerodynamic properties of any
gi"ven axisynmetric slender body can be evaluat"ed.
2.1.2 Line Distribution of Singularities Inside a Two-dimensional
Airfoil
In Sec. 2.1.1, the potential flow past an axially symmetric slender
body has been described and the same technique can be applied to the
two-dimensional airfoil. This idea was presented earlier in [10,11].
In this section, the problem will be analyzed in terms of the functions
of a complex variable. Similar assumptions are made here as in Sec.
2.1.1. Consider the two-dimensional potential motion of an ideal fluid
past an arbitrary airfoil which is shown in Fig. 2.2. The problem is
. j
Y/L-,: Z -plane
.'
.. i
G
~--~-------.......-~ X/Lo 1.0
L:BODY LENGTH
Fig. 2.2 Uniform stream past a two-dimensional airfoil withan arbitrary angle of attack.
No
21
analyzed in Cartesian coordinate (x,y) with the origin at the leading
edge of the airfoil and the trailing edge is some distance intercepts on
the x-axis. Now, one should recall the relationship between the
potential, the stream function and the velocity components in two
dimensional motion. As before, the potential is , and the stream
function is ,. The combination of +and , into an analytic function as
a complex potential +(z), is given by
+(z) = +(x,y) + i,{x,y) (2.35)
and the x-y plane refers to the z-p1ane (z = x + iy).
For this study, it is convenient to think of +as either the real
or imaginary part of a function
(2.36)
which is analytic in the z = x + iy plane outside the profile curve Gof
the airfoil. The potential ,0 is the potential of the incident flow on
the airfoil, while .b is due to the presence of the body. The
potential .b is represented by the superposition of complex potentials
of point sources and vortices distributed along an appropriate arc
inside the airfoil. The boundary conditions of this problem are.
no penetration condition: v,. n= 0 on G
infinity condition: .b =k tan-1(y/x) + O(1/r) as r + infinity
22
where n is the unit outward normal vector to the surface and r is the
total circulation about the airfoil.
The equation of the profile G of the thin airfoil is given by
y = e[C(x)±/S(x)] on 0 , x , 1, and e is defined as slenderness
ratio. This is assumed to be small for a thin airfoil. The leading and
trailing edges intercept x-axis at x=O and x=l respectively, and the
max.IC(x)±S(x)1 =1. It 1s assumed that both C(x) and Sex) are analytic
on 0 < x , 1, with C{x) and S{x) vanishing at x=O and x=l. These
functions can be expanded 1n a Taylor series at the endpoints as
C{x)•
= L c xn
n=l n
•, C(x) = L C (1_x)n
n=l n
•Sex) = L dnXn
n=1
• IV
, Sex) = L dn(1-x)nn=l
(2.37)
_ C(n){O)where cn - In.
_ S(n){O)dn - In.
__ (_1)nC(n)(1), cn - In.
d = (_1)nS(n}{1), n n! (2.38)
In order to solve for " it is convenient to find a function, Eq.
(2.35), which is analytic in z = x + iy outside G and sati'sfies the
condition
Re {[1-e(C ' tS ' )/2 IS] M-} = 0 on G (2.39)
23
Equation (2.39) is verified and explained in Appendix 8. Here, .b
represents the superposition of the potential due to point sources and
vortices distributed along an arc inside the slender airfoil. Thus,
(2.40)
where f(~,e) is the unknown source strength density, which is a complex
expression. The constants a(e) and B(e) determine the extent of the
source distribution and they can be determined after f(~,e) is found and
they must be inside G.
Using Eqs. (2.39) and (2.40) in Eq. (2.36), one obtains
Re {[i-e(C'tS'/2/S)] ~:o [x+ie(Ct/S)]}
= Re {~w [i-e(C'tS'/2/S)] (2.41)
Equation (2.41) is a pair of equations, corresponding to where the upper
(plus sign) and lower (minus sign) surfaces of G, is obtained. By
adding and subtracting these two equations, a pair of coupled linear
integral equations can be obtained. Upon denoting x+ie(C+/S) and
x+1e(C-/S) by xpu and xpl respectively, one has
2w Re {i(l+ieC')[ d.o(xPU)+ d.O(XPl)] eS' d,o(xpu) d.o(xpl)dz dz - 2/S [dZ - dZ ]}
(2.42)
24
and
2'1 Re {i(1+i C')/S[ d.O(xpu)_ d.O(XP1)] _ ~ S,[d!o(xPU) + ~o(xPl)]}e dz dz 2 dz dz
= Re {e /' 2S{1+i eCI
) - r(X+1eC-t) f("t,e) dtla (x+ieC-~) + e S
(2.43)
In Eq. (2.42) and (2.43), f(~,e), a(e) and B(e) need to be determined.
Both a{e) and a(e) are found to be power series in e of the form [11]
•(I( e) = L (In en
n=l,
00
B(e) =1 L Bnen
n=l(2.44)
Using the results of [10] as an instructive example, the function
f{x,e) shall be of the form
f(x,e) = f{x,e)I[s{e) - x][x - a{e)]
(2.45)
where f{x,e) is singular in x at x =o{e) and x = B{e). The function
f(x,e) is be assumed to be analytic before finding its asymptotic
expansion about e = O. Following the procedures given in [10,11], both
sides of Eqs. (2.42) and (2.43) can be expanded with respect to e,
without taking account of the dependency of f on £. Since ,0 is an
analytic function of x + ie[C{x)±/S{x)], the left-hand sides ofiEqs.
(2.42) and (2.43) can be expanded in a power series in e. Then, Eqs.
(2.42) and (2.43) become
• • 00 • 0L eJ J. (x) = Re {i L eJ L~. [f (x , e:) ]}j=O J j=O
(2.46)
• · • · 1L eJ KJ"(X) = Re {L eJ Lj [f(x,e)]}j=O j=O
Here, Jj{x) and Kj{x) are defined as
25
(2.47)
(2.48a)
(2.48b )\ ",
where (2.48c )
In Eqs. (2.48a) and (2.48b), the notation (~) = nl/(jl){n-j)l has beenJ
used and en] ;s the greatest integer not exceeding n. Applying the same
idea of expanding integral operators asymptotically as in Sec. 2.1.1,
one can expand the right sides of (2.42) and (2.43). To do this,
consider the integral operators IP(x,e), p= 0,1 applied to a function
F(,)/[(B_,)(,_«)]1/2 and F(,) is independent of e. The integral
operators IP are defined as
IO(x,e)=fB 2[x+ieC(x)-,]~1+1eC'(x)] +e2S' (x)F(,) d,o {[x+ieC(x)-~] +e S(x)} I{B - ~)(~ - a)
(2.49a)
I1( )_ f8 2S(x)[I+ieC 1 (x)] - SI(x)[x+ieC(x)-,] F(,) dX,£ - £ 2 2 ~
a {[x+i£C(x)-~] + £ S(x)} 1(8 - ~)(~ - a)
26
(2.49b)
The constants a(£) and 8(£) are defined in Eq. (2.44), where the
coefficients an and 8n need to be determined. The operator IP(x,e) can
be expanded by following the similar method as given in [7] and by using
the results derived in [11], one obtains
(2.50)
The linear operators L~, p = 0,1 are defined in Appendix C. From the
definitions given in Appendix C, one has two simple operators
1 / f(x)LO [F(x)] = 2w 'x I-x' F(x)
(2.51)
(2.52)
To solve Eqs. (2.46) and (2.47), an asymptotic solution for f(x,e) is
assumed as
aD
f(x,e) = L fn(x)enn=O
(2.53)
Here, the functions fn(x) are to be determined. By substituting Eq.
(2.53) into Eqs. (2.46) and (2.47), one obtains
(2.54)
r£j K. (x) = Re [r r £j+n L~ [f (x)]lj=O J j=O n=O J n J
27
(2.55)
By equating coefficients of the like powers of £ in the corresponding
equations, Eqs. (2.54) and (2.55), finds
1 [n 0 ~1m {La [f (x)]} = -In(x) - 1m L L. [fn .(x)]n j=l J -J
Re {Lo1 [f (x)]} = K (x) - Re [I L~ [f . (x)]l
n n j=l J n-J J
Using the results of Eqs.(2.S1) and (2.52), one finds that
0001m {LO[fn(x)]} = 1m { LO [fn(x)]} = LO [ 1m fn(x)]
21m [fn(51n 29)] - 1m [fn(x)]= -4/ de
o (51n29 - x)
(2.56)
(2.S7)
(2.58)
(2.59)
Applying the results of inversion of operators 1n [10] with Eqs. (2.58)
and (2.59) in Eqs. (2.56) and (2.S7) respectively, one has
1m f (x) = f (x) + 1m f (O)n n nwhere
; (x) = x2
{J (x) + 1mr L?[f .(x)] ln 'I n j=l J n-J . j
(2.60)
28
2- 2 [ n ] t = s1 n2e
- L I cos e J (t) + 1m L L~[f oCt)] de (2.61)112 0 (s1 n2e - x) n j=1 J n-J t = x
and
(2.62)
~=a
In Eq. (2.61), the notation [G(~)] = G(a)-G(b). From Eqs. (2.61) and~=b
(2.62), the- funct10n fn(x) can be determi ned recursi vely, once the
constants 1m f n(0) are found. In order to dete·rmi ned 1m f n(0), one has
to use the condition that
B ,.,.1m f f(~,£) d~ = r
(I
where•
r = L r 0 e:jj=O J
(2.63)
Here, each rj 1s a prescribed constant independent of e:. By us1ng Eqs.
(2.45), (2.53) and Eq. (2.61) in Eq. (2.63) can be expressed as
r = i e:j IS f j (~) d~ + 11 f ~ 1m fJ.(O)j=O (I 1(8 - 1;)((; - (I) j=O
(2.64)
By setting ~ = (S-«) sin2e +« and expanding each integrand in Eq.
(2.64) in Taylor series about £ = 0, one obtains
'I .
•. j "2" nr • 2 L e
J 1m L ~I I H:e:) i fJo_n[(s - «) sin2e + «]} de
j=O n=O· 0 £=0
•+ 1r L
j=O1m f. (0)
J(2.65)
29
Equating coefficients of like powers of £ on each side of Eq. (2.63) and
(2.65), yields
11'
rn 2 n 1 '2" a k'" . 21m fn(O)= - - - L IT f {(-a) f n k[(B-a)s1" e + a]} de
11' W k=O • 0 £ - £=0(2.66)
Since fO(x) depends only on JO(x), the 1m fn(O) can be determined
recursively from Eq. (2.66). Thus, functions fn(x) can be found through
Eqs. (2.61), (2.62) and (2.66) in a recursive manner.
Determination of the constants a(£) and B(e) is discussed now.
From Eqs. (2.61) and (2.62), one knows that Re fO(x) and 1m fO(x) depend
on +o(z) which is analytic. Thus, it fa1.lows from the recursive nature
of Eqs. (2.61), (2.62) and (2.66) that all the Re fO(x) and 1m fO(x)
must be analytic for 0 , x , 1 if the LP[F(x)] is analytic. This willq
be true if and only if each gj(x) and hj(x) in Eqs. (C2) and (C3) is
analytic on the interval 0 , x '1. In the manner of Appendix A, one
can determine ~ uniquely by demanding that g(x,e) is regular at x=O.
Similarly, all o·f the Bk can be determined by requiring that the
function h(x,e) is regular at x=l. By squaring both sides of Eqs. (C2)
and (C3) and equating the coefficients of the same power in £, one can
find out all the appropriate ~ and Sk' such that singularities at x=O
and x=l of functions gj(x) and hj(x) are eliminated. The similar
technique for the determination of a(e) and 8(e) is given in Appendix A,
and the general expressions for ~ and Sk are available in Appendix C.
Using the results in Appendix C and the definitions of Eqs. (2.37) and
(2.38), the leading terms for a(e) and S(e) are given by
(2.67)
IV IV IV IV IV IV 2
dl 2 i c1d l £3 + d1(d 2 +164c1 ) .,.4 + O{~5)See) = 1 - ~ e - 4 ~ ~
30
(2.68)
Once f{x,£) has been obtained, the expression for ,(z) is obtained by
using Eqs. (2.45), (2.53) in (2.40) and (2.36) is given by
(2.69)
Equation (2.69) is the desired uniform asymptotic exp.ansion for ,.
Suppose now that the thin airfoil has a sharp trailing edge such
that near x=l, S(x) has the expansion
CD IV •
S(x) = L d.(I-x)Jj=2n+l J
n ) 1 (2.70)
where dj = (-l)j S(j)(l)/j. From Eq. (2.70), it follows that one can
take a{e)=l, the asymptotic expansion obtained above remains valid when
Eq. (2.70) holds. The mathematical proof is available in Sec. 7 of
[10,11]. Physically, the velocity of the fluid near the sharp trailing
edge should be finite. As the velocity of the fluid with the potential
given by Eq. (2.69) near the trailing edge 1s examined, the first term
in Eq. (2.69), .o(z) obviously gives a finite ve~ocity at the trailing
edge, because ,o{z) is analytic at z=l. The remaining terms in Eq.
(2.69) contribute velocity components of the form
1f 1 F(~) d~
a(e) z - ~ I[~ - a(e)] (l-~)
(2.70a)
Equation (2.70a) remains bounded as z+l, if and only if F(l)=O.
31
Therefore, one shall make each fk{x), k > 0 vanishes at x=l by properly
choosing the arbitrary constants of 1m fk{O). The mathematical proof of
choosing 1m fk{O), so that each fk{x), k > 0 vanishes at x=l, is shown
in Sec. 7 of [10], i.e.,
(2.71)
If the flow described by ~o is a uniform stream, the circulation
can be determined by Eq. (2.65), such that the lift L{e) of the airfoil
is gi ven by
L{e) = p U r{e)GO (2.72)
where p is the density of the fluid and U is the speed of the uniformGO
stream at infinity. By using Eqs. (2.65) and (2.71), one finds that Eq.
(2.72) becomes
11'
-. j 1'2" k- 2L(e) = pU r eJ 2 1m r ~ f {(aa~) 1 fJ-_k[{B-a)sin e + a]} de
GOj=O k=O If. 1 0 ~ e=O
(2.73)
The uniform asymptotic expansion for ,(zje) can be determined by finding
fn(x), aCe), and S(e). Then, the velocity, pressure distributions on
the surface, and also the lift coefficient of an arbitrary airfoil can
be determined.
32
2.1.3 Second Order Joukowski Airfoil
Certain assumptions have been made in analyzing the problem in Sec.
2.1.2. The upper and lower surface of the thin airfoil have the
equations y = e[C(x)+/S(x)] and y= e[C(x)-/S(x)] for 0 < x < 1; with
C(O)=S(O)=C(I)=S(l)=O. The equations of the thickness and the camber
line of the airfoil are y = e/S(x) and y = eC(x), respectively.
Here e 1s the slenderness ratio, while S(x) and C(x) give the thickness
and camber distributions. Varying e produces a family of affinely
related profiles. More examples are given by Van Dyke in [1]. Now one
expects to seek an expression for a thin Joukowski airfoil by using the
perturbation method. The parametric form of the Joukowski airfoil is
easily obtained by means of conformal mapping [15]. One employs a
transformation x(~,n), Y(~,n), where the ~-n plane refers to as
the ~-plane which will map the region and its boundaries in
the ~-plane into another desired configuration in the x-y plane, which
is referred to as the z-plane. Fig. 2.3 shows the notation of the
Joukowski transformation of a circle into an airfoil. The center of theio
circle with radius a at any arbitrary point is given by p =me. The
symbol Po denotes the intersection of the vector a e-iS with the y
axis. A circle with center at Po and radius of magnitude Ipo - tTl
transforms into a circular arc, which is considered as the skeleton of
the airfoil. The arbitrary circle, the points of which are described by
i9~ =a e + p (2.74)
transforms into an airfoil. The angle B determines the mean curvature
33
7]
z-plane
--r-----r~---~----~'----~x
.°L-2C --JT
Fig. 2.3 Illustrating the derivation of an arbitrary Joukowskiairfoil.
34
of the profile and the magnitude Ip - pol determines the thickness of
the profile.
The circle is chosen such that it passes through ~ = ~T = C, where
C is real positive number. The corresponding point on the airfoil is
gi ven by
(2.• 75)
It should recall that the center ~ of the circle, its radius a, and the
angle 8 are connected by
-i 8 i 6C=ae +me (2.76)
Since C 1s a real positive number, one can separate the real and
imaginary parts of Eq. (2.76) as follows:
C =a case + m cos~
a sins = m sin~
(2.77a)
(2.77b)
Using Eq. (2.74) in Eq. (2.75), with the expressions given by Eqs.
(2.77a) and (2.77b), Z(~) can be separated into real and imaginary parts
as x and y respectively. After further simplification, the parametric
form of the airfoil is obtained as follows:
x = a(coss + h) [1 + C2 J (2.78a)a2(1 + h2 + k2 + 2h case + 2k sine)
35
y = a(sine + k) [1 - C2 ] (2.78b)a2(1 + h,2 + k2 + 2h cose + 2k sine)
where h = (m/a) COSo k = (m/a) si~o (2.79)
The trailing-edge point of the airfoil is denoted by ZT with e = -6 in
Eq. (2.75), one finds x = Re(ZT) = 2C and y = Im(ZT) = o. Thus, the
trailing edge of the airfoil intercepts the x-axis at the point T with
OT=2C. In order to have an asymptotic expression for a thin Joukowski
airfoil a small parametric, £1 should be introduced to expand the
available equations in power series. Generally the quantities h,
k, S, and m should be small. If we choose m=elC, then Eqs. (2.77a),
(2.77b) and (2.79) can be expanded to the second order of accuracy as
follows:
2h = (coss cOS6) e1 + O(e1 )
2k = (coss s1n6) el + O(e1 )
(2.80a)
(2.80b)
(2.80c)
Using Eqs. (2.aOa)-(2.aOc) in Eqs. (2.78a) and (2.78b), and retaining
the terms with order of el , one obtains
+ 2 cos2s cose(cos6 - coss cos6 cose - coss sin6 sine)]}
2+ 0(e1 )
y = Csine cos~ e1[2cos2S case - (coss + c~ss)J
36
(2.81a)
(2.81b)
By examining carefully the order of magnitude of sins, cosS, sin2s
and cos2s, one finds that both coss and cos2s are 0(1) and sins and
sin2 S are 0(e1) and O(e~) respectively. Neglecting the terms of
0(e1), Eq. (2.81a) can be further simplified as
x = 2Ccose + 0(e1) (2.82)
Furthermore, Eq. (2.81b) can be simplified by comparing the order of
-t d f - - 2 d 2 t e d T t e tmagnl u e 0 slnB, Sln S, coss an cos S as men lone. 0 re aln erms
of 0(e1), Eq. (2.81b) becomes
(2.83)
Introducing the nondimensional quantities x= x/2C, and y = y/2C, Eqs.
(2.82) and (2.83) become
(2.84a)
(2.84b)
Upon eliminating sine and cose in terms of i, Eq. (2.84b) becomes
37
'" tv / ",2 · (1 1V2 )Y = £1(-COS6)(1 - x) 1 - x + £1 s1n6 - x
-1 < x < 1 (2.85)
Obviously, if ~ = w, Eq. (2.85) will yield a symmetrical second order
Joukowski airfoil, while the second term vanishes. It is consistent to
the second order Joukowski airfoil shown in Sec. 4.6 of [1]. It also
implies that the second term represents the equation of the camber
line. Therefore, Eq. (2.85) can be rewritten as
y = y + ys c (2.86)
h '" ( ) ( tv) / tv2 IV • ( "'2)were Ys = £1 -cos~ 1 - x 1 - x and Yc = &1 Sln~ 1 - x •
To make Eq. (2.86) consistent with the required form of equation, simple
translation and scaling of the coordinates are done, such that Eq.
(2.86) becomes
YC = 2 s1n6 £1 x(1 - x) o < x < 1
(2.87b)
(2.87b)
By def1ning £ = 2(-COS6)£I"' with £ is the slenderness ratio, Eq. (2.87b)
yields Ys = £ IS(x) and Yc = £ C(x)
where
Sex) = x(1 - x)3 (2.88a)
C(x) = Ax(1 - x) , A = -tan6 (2.88b)
38
S(x) and C(x) give the thickness and camber distributions of an airfoil
respectively, and A is related to the airfoil camber.
2.2 Other Solution Techniques
2.2.1 Some Exact Solutions
The analytical solutions of potential flow problems associated with
the motion of a solid body through an ideal fluid are considered here.
The final aim is to determine the pressure distribution over the surface
of the body and the lift force acting on the body whenever available.
The analytical solution for elliposida1 body of revolution and two
dimensional elliptic airfoil are obtained in this section.
1. Analytical Solution for Ellipsoidal Body in Uniform Stream
Consider a rigid ellipsoidal body moving through a uniform ideal
fluid. The flow is assumed to be steady, incompressible and irrota
tiona1. To simplify the problem, a body-fixed reference frame and
elliptical coordinate system (~,~,B) are introduced. The derivation is
described in Appendix D. Using the results of Eq. (D5a), one is able to
obtain 0 = U e + U e where U~= ~ :' and U = - h\ *.~ ~ ~ n ~ ~ S ~ n ~ S ~
Here, Ut
and Un are velocity components in t and n directions
respectively, and
2 2a, a, a, a, _~ + -:2 - coth~ - - cotn a - 0at an at n
(2.89)
Consider the Neumann boundary condition on ~ = ~o and the infinity
39
condition as ~ tend to infinity. Making reference to Appendix D, the
boundary conditions can be simplified as follows:
on E: = ~o (2. 90a)
as E: + infinity (2. gOb)
Equation (2.89) along with Eqs. (2.90a)-(2.90b) can be solved by
separation of variables. One finds out that the velocity component
in et direction on the body surface is zero (due to the no penetration
on the surface of the body), and the velocity in e direction is givenTl
by
(2.91)U sinn..
Us= -[ cosht - 1,11[oshto+ sinh2toln{ sin~to )J'COSh2t
o- cos2~
With application of Bernoulli equation, the pressure distribution on the
surface of the elliposidal body is given by
(2.92)
2. Analytical Solution for El11ptic Airfoil in Uniform Stream
In Sec. 2.1.3 the idea of conformal mapping has been introduced to
get the shape of an airfoil. Here, the pressure distribution over the
airfoil, by means of Joukowsk1 transformation is discussed. This method
can be used directly to the actual problem of by means of mapping the
flow past an arbitrary circle, for which the solution is already
known. To define the arbitrary circle, we describe the points by
ie~ = r eo
40
(2.93)
(2.94)
where ro = (a+b)/2 with a and b are the major and minor semi-axis
lengths of the ellipse respectively. Making reference to [15], the
complex velocity on the airfoil surface is given by
-iei 2U sine e•W(z) = dz
~
After simplification,
the velocity distribution along the contour of the ellipse is given by
(2.95)
where k =b/a. The pressure distribution on the elliptic airfoil can be
computed through the Bernoulli equation.
2.2.2 Panel Method
A considerable number of panel methods has been developed in the
past two decades to analyze the steady, inviscid and irrotational flow
fields past a body of arbitrary geometry. The incompressible f10wfield
is governed by the Laplace equation which can be transformed to an
integral equation relating the perturbation potential to the source and
doublet singularities distributed over the surface. There are different
discretization techniques among panel methods for solving the integral
equation. Basically, all the methods approximate the surface by
elemental panels of prescribed ge9metric shape and singularity variation
41
and the integral equation is solved by enforcing a boundary condition at
a control point or points on the panel. There is a summarized table in
[16], showing several useful panel methods with panel geometry,
singularity variation and boundary condition specification. In this
section the governing equation and the basic idea of discretization are
reviewed.
Consider steady, inviscld, irrotational-and incompressible fluid
flow in a bounded domain D, Fig. 2.4, governed by the Laplace equation
(2.96)
where V =0 + v.. The perturbation potential ,(P) at any point P canGO
be expressed as the potential induced by a combination of source (a),
and doublet (~) singularities distribution over a bounded surface S .
1 a 1.(P) = If a( - 41rr) dS + ~f J.l an (41rr) dS (2.97)
where r is the distance from the point P to the bounded surface
and n represents the normal to the surface S directed into the fluid
domain. The boundary condition 1n this study is the flow tangency
condition or
v . n = 0 (2.98)
Since there are two unknowns a and p, while there is one boundary
condition, there exists an infinite number of source and doublet
combinations which satisfy Eq. (2.97) and one can specify the doublet
42
Bounded
Surfaces
Fig. 2.4 Body immersed in a bounded flowfield.
43
strength and solve for the source strength or vice-versa. Using the
idea of real external and imaginary internal fields introduced in [17],
the condition for the perturbation potential interior to the region is
constant and set equal to zero, i.e.,
or 4» = l.l
Thus, it follows that
-a =V; • n (2.100)
For Neumann problems, the source density distribution is given by
Eq. (2.100), the doublet distribution is to be determined. For
Dirichlet problems, the doublet distribution is known from the
prescribed potential on the boundaries, and the source distribution is
to be determined.
In order to introduce lift, wake networks extending from the
trailing edge to downstream infinity are considered. As the free stream
Reynolds number increases, the thickness of the wake region diminishes
and 1s often assumed to be zero. Since fluid properties are continuous
across the wake, the wake region is usually modelled as a stream surface
having zero pressure loading imposed by the ·surrounding flow. The wakes
can be represented by a doublet sheet with zero thickness and streamwise
doublet gradient, 'emanating from the trailing edge and parallel to the
free stream. Determination of the strength of the wake constitutes the
Kutta conditi~n which is available in [16]. Some typical numerical
schemes were already presented in the literature [16-21].
44
Chapter 3
APPLICATION OF PERTURBATION METHOD
3.1 Perturbation Technique for Slender Bodies of Revolution
The formulations of Sec. 2.1.1 are applied to the case of a uniform
ideal fluid flow past an axisymmetric slender body. The coordinate
system is fixed to the body and the velocity of the stream at infinity
is U. The incident stream function is given byCD
(3.1)
The stream:function ,0 can be expanded in powers of e2 and the
coefficients ,.(x) can be obtained by using Eq. (3.1) in Eq. (2.19b);J
this yields
'I (x) = U./2 , ,. (x) = 0J
for j ) 2 (3.2)
By inserting these coefficients into Eqs. (2.24a) - (2.24d) together
with the definition of operators Lj and Gj' the first six leading terms
of fnm(x) are given as follows:
(3. 3a)
45
= _.!. U L { sst _ sst _ SS" + Sus log [4X(t-j}]4 CD dx x I-x 5 x
I-x+ S f [S'(x+v) - S'(x) - vSU(x)] v-2 dv
o
x- S f [S'(x-v) - S'(x) + vSU(x)] v-2 dv}
o
- ~ [fa {Slog [4x(1-x}] - S} + (1 __5_)f ]dx 21 S x I-x 21
2f 32 (x) = - i k (G1f 21) = 10 UCD k [S :x2 (SS")]
(3.3b)
(3.3c)
(3.3d)
(3.3e)
(3.3f)
Equations (3.3a) - (3.3f), together with (2.18a) and (2.18b) depend upon
the profile curve. The expressions for fnm(x), a(e) and e(e) can be
evaluated if S(x) is specified. The complete asymptotic expansion
for t(x,r2 ;e) can be obtained from Eq.· (2.25) with to = U x. The flowCD
46
velocity components over the body surface can be determined by taking
appropriate derivatives of +. Next the Bernoulli equation 1s used and
the pressure coefficient on the body surface is evaluated as
U 2C = 1 - (r!-)p GO
(3.4)
, b ) 2
where Us is the velocity distribution on the surface of the body. The
profile curve is' defined by r = e IS(x) on 0 , x , 1.
In this study, two different axisymmetric bodies are analyzed, and
their profiles are given as follows:
1. El"11psoidal Body
S(x) = 4x (I - x)
2. Dumbbell Shaped Body
S(x) = 4bx{l - x)[1 - bx{l - x)]
With the profile shape r = e IS(x) on 0 , x , 1 mentioned, one can
obtain explicit expressions for fnm(x), a(e) and a(e) from Eqs. (3.3a)
(3.3f), (2.18a) and (2.18b) respectively for each of the above ~wo cases
(Figs. 3.1 and 3.2). Lists of the required expressions are presented in
Appendix E.
3.2 Perturbation Technique for Thin Airfoils
The fonmulation given in Sec. 2.1.2 1s applied to the flow problem
when the airfoil is at rest in an ideal fluid which has unit velocity at
infinity. The airfoil 1s at an arbitrary angle of attack y to the x
axis. Thus, the complex potential for the incident stream is given by
(3.5)
.5-
.3
.2
.1
-.2
-.3
-.~
RC Xl = .OSC-SCRTe S( Xl)
S( Xl =qX( I-XI
LENGTH/DIRMETER= 10.00
47
-.5 ----10 .1 .2 .] .'i .5 .s .1 .e .~ I.~
X
Fig.3.1 Ellioosidal body qeometrv with E = 0.05
48
.s
••
R(X): .030xSORTCSeX)) B= 3.00
S( Xl =8X( J-Xl ( i-8Xe I-XI)
.2
.1
ct: 0
-.1
-.2
-.3
-."1
lENGTH/Ol~EnER= 16.67
--------------_._----~
-.5 '---__1--__.L..-__.L--__"---__.r..-__L.-_-..I~_ ___" __
o .1 .2 .3 .'1 .5 .6 ., .e .~ 1.:)X
Fig. 3.2 Dumbbell shaped body geometrY with E = 0.03, b = 3
49
Introducing Eq. (3.5) in Eq. (2.48c), one obtains
-1y,.(x) =} eJ to
if j = 0
if j ) 1(3.6)
Using Eq. (3.6) in Eqs. (2.48a) and (2.48b) yields
Jo(x) = 41f siny
J1(x) = -4. C1(x) COSy
Jj(x) = 0
K1(x) = -2. SI(X) COSy
Kj (x) = 0
if j ) 2
if j = 0,2, •••
(3.7)
The functions fn(x) are obtained by inserting these coefficients
together with the expressions in Appendix C defining the operator
L~t into Eqs. (2.61) and (2.62). The three leading terms fn(x) are
computed as follows:
Re fO(x) = 0
fO(x) = 2x s1ny
(3.8a)
(3.8b)
(3.8c)
(3.9a)
(3. 9b)
(3.9c)
w . 2~ 2 [ ~~=S1n e- ~ f .C~S e 1m Lr [fO(t)] - 4w C1(t) COSy de
T 0 S1n e - x ~=x
11'
r l 2 "2" - • 21m f 1(O) =- - - I f 1(S1" a) da
11' 'II' 0
50
(3.10a)Re f2(x) = - ~w Ix~t;») [Re L~ [flex)] + Re L~ [fo(X)]]
f 2(x) =~ {1m L~ [flex)] + 1m L~ [fO(X)]} (3.l0b)
f 2 [ ~t=sin2e- ;. f ~o~ 8 1m L~ [fl(t)] + 1m L~ [fO(t)] dey 0 Sln a-x. ~=x
11'
r2 2 ~ - 2 2 21m f 2(O) =.- - y f
o[f2(Sin e)+2 siny(a2 cos e - 62 sin e)]de (3.lOc)
o I 0 I .The operators LI , LI , L2, and L2 are g1ven in Appendix C. When Eqs.
(3.8a)-(3.1Oc) are used in Eqs. (2.53) and (2.45) they give the
asymptotic expansion for f(x,e) up to O(e3). Once f(x,e) has been
found, one can obtain an expression of ,(z;£) by using Eq. (2.69) and
the pressure coefficient can be evaluated from Eq. (3.4). Since Eqs.
(3.8a)-(3.IOc), (2.67) and (2.68) are functions of S(x) and C(x), once
they are specified, the complete expansion for +(z;e) up to O(e3) can be
determi ned.
In this study, elliptic airfoils and second order Joukowski
airfoils have been analyzed using the above method. For a symmetric
airfoil the function C(x) is set to zero; only the thickness solution
has to be solved. If the airfoil 1s symmetric and is set at zero angle
of attack, then there is no circulation about the airfoil and all rn are
set equal to zero in determining the constants 1m fn(O). Moreover, if
51
the airfoi~ has a sharp trailing edge, special treatment is applied to
obtain each rn, which is given in Eq. (2.71). The profile curve of the
thin airfoil is defined by y. = e[C(x)t/S(x)] on 0 < x < 1, with small
slenderness ratio e. Three different airfoils are used as test cases,
Figs. 3.3, 3.4 and 3.5, and their geometries are as follows:
1. Elliptic Airfoil
Sex) = 4x{1 - x)
2. Symmetrical Second Order Joukowski Airfoil
S(x) = x(1 - x)3
3. Cambered Second Order Joukowsk1 Airfoil
C(x) = Ax{l - x)
S(x) = x(1 - x)3
Here, A is related to the airfoil camber and is defined in Eq.
(2.88b). Using the above information, expressions for
fn(x), rn, ~(e) and S(e) can be derived in each case. A brief
derivation and the lists of the resulting expressions are included in
Appendix E.
.3
.2
.1
vex)= .050~SCRT(S(XJ )
T/C= 10.00~
52
Fig. 3.3 Elliptic airfoil of 10% thickness.
53
.5
.3 Y( Xl = .100J'( C( Xl ~/ - SORT( S( XJ ) )
C( Xl = .QOO-X( l-Xl Sl XI =X( i-Xl .-3
.2 T/C= 6.50Z DELTR=lBO.OO
.1
>- a
-.1
-.2
-.3
-.11
.5X
.'t-.S~__",,:-__~__.a.-__...L--__.L--__..L-__..L..-__L-.-_--Jl-_----I
a
Fig. 3.4 Symmetrical Joukowski airfoil of 6.5% thickness.
54
.5
.'1
.2
Y( Xl = .100x( CL Xl ./- SQRT( S( Xl )]
C( Xl =1 .OOO-X( l-Xl S( Xl =X( l-Xl-1I3
T/C= 6.5cr~ DELTA=!35.00
.1
>- a
-.1
-.2
-.3
-.'4
1.0.e.1.aX
.'1.2.1
-.s....-..__~ ~ ~ _a
Fig. 3.5 Cambered Joukowski airfoil of 6.5% thickness.
55
Chapter 4
RESULTS AND DISCUSSION
In this chapter an assessment of the merits of the perturbation
analysis method is made. The pressure distribution on axisymmetric
bodies and two-dimensional airfoils obtained by the above method are
compared with those of panel method as well as exact solutions whenever
avaliable. The results of the panel method are obtained by using the
low-order panel method code (program VSAERO, developed by Maskew
[18]). The analytical expressions for the ellipsoidal body and elliptic
airfoil are avaliable 1n the literature presented in Sec. 2.2.1.
4.1 Axisymmetric Bodies
Since the analytical expression for an ellipsoidal body is
available, this body is selected to demonstrate the accuracy of the
perturbation analysis method with the variations of the slenderness
ratio and also the number of terms in the expansion of ,. Figs. 4.1a
4.2c illustrate a comparison of the pressure distributions, as calcu
lated by the perturbation analysis method with different order of
accuracy in the expansion of ., with analytical solutions for the two
values £ = 0.1 and e =0.2, corresponding to LId = 5 and 2.5, respec
tively. The results show that the perturbation solutions agree well
with the exact solutions to a high order of accuracy, i.e., up to
0(e8). With the same order of accuracy in • up to 0(e8), the pressure
56
distributions obtained from the perturbation analysis method and the
panel method ·are compared with the exact sol uti ons, with di fferent
numbers of panels in the chordwise direction for £ = 0.10. The
calculations are presented in Figs. 4.3a-4.3c. It is seen that the
results using 30x6 panels and 50x6 panels yield better agreement with
the exact solutions than using 10x6 panels. Again for the same order of
accuracy in • up to O{e8), and using 50x6 panels, three sets of pressure
distributions are compared with four different values of £. The results
are presented in Figs. 4.4a-4.4c. The perturbation solutions agree well
with the exact solution with small slenderness ratio up to 0.1, and the
panel solutions compare well with exact solutions for all values of £.
The dumbbell shaped body is also analyzed by the perturbation analysis
method wi th the expansi on of • up to 0(e4), as we11 as by the pane1·
method. A comparsion of pressure distributions obtained by the above
method and the panel method, using SOx6 pane1~, with different values
of e and profile parameter b are presented in Figs. 4.5a-4.6b. For
small £ and b, the perturbation solutions agree well with that generated
by the panel method. In the present examples, each perturbation case
required only 2 seconds computing time versus 5 seconds (30x6 panels) or
14 seconds (50x6 panels) for the panel method. This computation time
comparison is based on the CDC Cyber 173.
4.2 Two-Dimensional Airfoils
The analytical pressure distribution on the surface of the elliptic
airfoil is avaliable 1n the literature discussed in Sec. 2.2.1. It is
convenient to do numerical experiments for the perturbation analysis
method using this airfoil. Perturbation analysis solutions for an
57
elliptic airfoil, set at zero angle of attack, are compared with exact
solutions, with different order of accuracy in , expansion for £ = 0.10
and £ = 0.15, corresponding to the tIC ratio = 20% and 40% respectively,
as presented in Figs. 4.7a-4.8b. It is apparent that the perturbation
analysis solution gives quite good agreement with the analytical
solution for high order of accuracy in • up to 0(e4). The pressure
distributions calculated from the perturbation analysis method are
compared with the exact solutions, with different values of slenderness
ratio and these are shown in Figs. 4.9a-4.9c. The results show that the
perturbation solutions agree closely with exact solutions up to 20
percent thick elliptic airfoil.
Results of the symmetrical second order Joukowski airfoil obtained
by the perturbation analysis method with the order of accuracy of +expansion up to 0(e3) are compared with the panel method in the
following section. Here, 3.25%, 6.5%, 10% and 13% thick airfoils are
chosen as the test cases. For the calculation presented, the thickness
of the airfoil, angle of attack, and the number of panels are given for
each of the calculation. The calculations presented in Figs. 4.10a
4.1Oc and 4.11a-4.11c are obtained using different numbers of panels in
the chordw1se direction for symmetrical Joukowski airfoil of 6.5%
thickness at 0° and 6° angles of attack, respectively. The results do
not give a smooth curve near some regions of the leading edge until 50
panels are used. A comparsion of the perturbation analysis solutions
for the symmetrical second order Joukowski airfoil at 0°, 6° and 12°
angles of attack with panel method are presented in Figs. 4.12a-4.15c.
The perturbation solutions agree with the panel solutions except in some
region on the leading edge of the airfoil surface. Table 4.1 illus-
58
trates the appreciable difference between the lift coefficients of the
two methods from the exact solutions.
The perturbation analysis method has been applied to the cambered
second order Joukowski airfoil with the order of accuracy in the
expansion for. up to O(e3). Calculations are presented in Figs. 4.16a
4.19c. The results generated by the panel method with 0°, 6° and 12°
angles of attack are used for comparison. Both pressure distributions
agree closely for small slenderness ratio and angle of attack up to € =
0.03248 and y = 6° respectively, except at some regions near the leading
edge. The lift coefficients obtained from the perturbation analysis and
the panel method compared with the exact solutions are shown in Table
4.2. In the present example, the computation time for each pertur
bation case required 8 seconds, while 0.5 seconds (10 panels), 2 seconds
(30 panels) and 5 seconds (50 panels) were required for each panel-
method solution.
59
~r Xl = .1 C01f:C:~T( -iX( 1-Xl: c;;I,,
.8 ~LPH~= .10CCE-Ol 8ETH= .~~8CE~aQ
.6 SQU~RE--- PERTUR8~rrCN LINE--- E~T
I;,,
I
Cl.U
.2
a
"t1-~
!
-.2
-.'1
-.8
-.8
1.0.8.1.sX
.2-1.0 ....--.__""--__""---__...a..-.__-....- --'-__-..L.__---J.__--J.__--J
a
(a) € = 0.1; ~ expanded UD to E 2 •
Fig. 4.1 Comparison of oressure distributions from the perturbationanalysis method with the analytical solution for -ellipsoidal body.
.8
.6
.ll
.2
-.2
-.Il
-.8
-.8
L~t\GTH/DlR:1E!ER = 5.GO
~LPh~= .101CE-Ol BET~= .~8~~aJ
saU~RE--- PERTURB~rrON LINE--- E~T
60
-1.00 .1 .2 .3 .'1 .1 •8 .7 .a . .~ 1.0
X
O( b.} E = 0.1 <t> expanded up to e: lt•
Fig. 4.1 Co·nti nued
a.u
.8
.6
.q
.2
o
-.11
-.8
-.8
L~~GTH/OIR~E!ER = 5.GO
~LPh~= .101CE-Cl BET~= .~8~~~OJ
SOU~RE--- PERTUR8~rrON LINE~-- EXACT
61
crI
!\ "
-1.00 .1 .2 .] •• .8 .1 .8 .~ 1.0
X
(C) e: = 0.1 <P expanded up to e: 6•
Fig. 4.1 Concluded.
62
.2
-.2
R( Xl = .20D.SCRT( '-:X( I-Xl)
LENGTH/OIMi~EiER = 2.50
saU~RE--- PERTUR8~r.ICN LINE--- EXI=CT
.J......
1r
-.'.1
-.8
-.8
.e.1X
.2.1-l.O'----__I...-__.L..--__"--__L..-__L-__'--__L.-__.L....--........J~----'
o
(a) € = 0.2; $ expanded uo to €2 •
Fig. 4.2 Comoarison of pressure distributions from the perturbationanalysis method with the analytical solution forellipsoidal body.
0..U
t.cqr-
.J\
.&
.'1
-.2
-.8
-.8
R( x: = .2JQ.S:RT: l1X( l-X~ 1
lE~~7r/QIFr£rE~ = 2·50
saU~RE--- PERTURBhrICN LINE--- E~Ci
63
-1.0a .1 .2 .3 ., .5 .8 .7 .s .~ 1.0
X
(b) E = 0.2 ; <p expanded up to £1+.
Fig. 4.2 Continu~d.
0U
.9
.6
.2
a
-.'1
-.6
-.8
R( Xl = .20C.S!:RTC 4X~ l-XI 1
LENGTH/DIAMEfER = 2.50
. SaU~RE--- PERTURSHTION LINE--- EXACT
64
-l.aa .1 .2 .\ .1 .e .7 .8 .~ 1.0
X
(c) e: = 0.2 ; <p expanded up to e: 6 •
Fig. 4.2· Concluded.
65
R( Xl = .1 OO:.SCRT( 4X( I-XlI
LENGTH/OIRMEnER= 5.00 NPRN= 60
ALPH~= .JOI0E-Ol BETR= .98~+ao
.6 CROSS----PERTURBRTION LINE----EX~CT SOU~R~----P~~EL
a..u
.2
o
-.z
-...
-.8
-.8
1.0.~.,.1.1X
-1.0~-_Ioo..-_--"'---",":,-------'"'--__..L..-.__...I..-__.L...-_--J~_-Io
(a). E = 0.1 ; ~ expanded up to £6; 60 panels with cosine spacinq.
Fig. 4.3 Comoar;son of the perturbation and exact solutions withthe oanel method for elliosoidal body.
au
lorW.8
.6
.'1
.2
a
-.2
-.11
-.8
-.8 .
\
R( Xl = .1OO.SbRTC 4X( I-XI]
LENGTH/OIR~Ensq= 5.00 NPA~ 160
~LPHA= .1010E-at 8ET~= .~89SE+aa
CROSS----PERTUR8~TION L[NE----EX~CT sau~RE----PhNEL
66
-l.aa .1 .2 .] .'1: .5 .8 ., .8 .~ 1.0
X
(b) e: = 0.1 ep expanded up to e: 6 ; 180 panels 'IIi th cosine spacing.
Fig.4.3 Continued.
a.u
I.
.9
.6
.2
o
-.2
-.'1
-.8
-.8
RC Xl = .1 GC.S~T( 4XC I-Xl)
LENGTH/Oln~ErER= 5.00 NPhN= 300
RLPHH= .101CE-al BETR= .9a99E~aa
CROSS----PERTUR8~TION LI~----EXHCT SOJ~RE----P~NEL
67
-1.0a .1 .2 .~ .'1 .8 .e ., .1 .~ 1.0
X
(c) e: = 0.1 <p expanded up to e: 6; 300 pane1s with cosine scacing.
Fig. 4.3 Concluded.
.6
68
R( Xl = .OSC~S;~T( 4X( I-Xl)
lENGTH/OIRMEfER= 10.00 NPAN= 3CO
F1LPH~= .2S0SE~2 8ET~= .9975C·ca
CROSS----PERTURBRTION l[NE----EX~CT SOUHRE----P~NEl
0U
.q
.2
o
-.2
-.11
-.8
-.&
(a) E = 0.05 ~ expanded up to E 6 ; 300 panels·with cosine spacing.
Fig. 4.4 Pressure distributions from the perturbation analysisand the panel methods are compared with the exactsolution for ellipsoidal body.
au
I.
o
-.2
-.'1
-.8
-.8
RC Xl = .1 DC»SeRre 4X( I-X) )
LENGTH/O I.=i:1ETER= 5. 00 ~~PF~~= 300
ALPH~= .101CE-Ot BET~= .9899E~ao
CROSS----PERTURB~TION LIN:----EXHCT SOJhR£----P~~~EL
69
-1.0a .1 .2 ., .'1 . .1 .8 .1 .8 .~ 1.0
X
(b) £ = 0.10 <p expanded up to £6; 300 ·panels with cosine spacin~.
Fig. 4.4 Continued.
70
:·t\
.8
.6
.li
.2
~ ,0
-.2
R( Xl = .2Ca.SCRT( 4X( I-XI)
LE~GTH/OIRrEiER= 2.50 NPR~~ 3eD
CROSS----PERiURSATION L[NE----EX~CT· SOL~R~----P~~EL
-.8
-.8
1.0.~.5X
.2.1-1.O~ --,-__-....__-.-.I'--__-----'__..4- ---'-__~__~
D
(c) E = 0.20 ; ~ expanded up to £6; 300 panels with cosine spac;nq.
Fig. 4.4 Continued.
71
,..1.~
.8
.6
.li
.2
-.2
-.'1
~~ Xl = .2S0.SC.~i( 4XC I-Xl ~
LE~GTH/OJAMETER= 2·00 NPRN= 300
CROSS----P£~TuRa~TION Lrt\E----EX~CT SOJHRE----P~NEL
+...
-.8
-.8
.~.,.1.aISX
.•2-l.O~__...Io.-__""""'__---'--_..Ao-..__-&..-__--L.__--J~__..L--__....L-__--1
o
(d) E = 0.25 ; ~ excanded UP to £6; 300 panels with cosine soacing.
Fig. 4.4 Concluded.
72
.6
.\1
R( Xl: .03C~aX( l-Xl (1-8X( I-Xl) 8= 2.CO
LENGiH/OIF~ErER= 16.67 NPRN= JGO
~LPHA= .1810E-02 BETR= .99~~~CO
saU~RE ---P~NEL LI ~~E----PERTURBrlr[C~
-.2
-.'1
-.6
-.8
-1.O ......__.1--_-1~_---.1-----1_____r._____'_____L____L____L____J
a
(a) e: = 0.• 03 b = 2; $ expanded up to e:~;
300 panels with cosine spaci"ng.
Fig. 4.5 Comparison of pressure distributions from the oerturbationanalysis method with the panel method for dumbbell shapedbod.y.
73
au
.6
.li •
.2
o
-.2
-.Ii
-.8
-.8
R( Xl = .070.aX( t -X) ( 1-8Xr I-Xl) 8= 2 ·CO
LENGTH/O I R:1E fER= 7•1Lt N?F~~= 300
RLPHA= .100~E-ol BET~= .98~9E.ca
SQU~RE----P~NEL LINE----PERTURSRfION
-1.0~-_........__..L_.___.Ioo...--------_~__..L.--__..J..____.L___ ___I"_____ ___'
a
(b) E = 0.07; b = 2 ~ expanded up to £4;
300 panels with cosine soacing.
Fig. 4.5 Concluded.
74
ALPHA= .2729E-02 BET~= .9973E+OC
RC Xl = .030.8XC L-Xl ( 1-8X( I -Xl ] 8= 3. 00
LE~GTH/OI~~EnER= 16.57 ~PM~~ 3eD
saU~RE~--P~NEL LINE----PERTuR8~rrON
.2
.8
..".6
n.. au
\'\
-.2
-.q
-.8
(a) E = 0.03; b = 3; ~ expanded up to £4;
300 panels with cosine spacing.
Fig. 4.6 Comparison of pressure distributions from the perturbationanalysis method with the panel method for dumbbell shapedbody.
75
au
I.@-I
t.e.~
.2
o
-.2
R( Xl = .070)l8XC L-Xl ( 1-8X( I-X) ) 8= 3. 00
LENGTH/DIR~ETE~= 7.1li NPR~J= 300
saU~RE----P~NEl lINE----PERTURB~i[a~
-.8
\ -.8
-l.a.......__':--__~__L--._--JL......-----J-----J--.........J------L----.l.--.-J
o
(b) E = 0.07; b = 3 ~ expanded UD to €~;
300 panels with cosine spacing.
Fig. 4.6 Concluded.
76
.8
TIC = 20.0J%
:3I
Ii
.6 SQUHRE--- PERTURaArrON LINE--- EXQCT
.Ii
.2 ,
a- Du
-.2
-.'6
-.S
-.8
.~.e.,.&X
-1.O~_--':-----';:---~--~-_---L__---1__---Jl.-.__L-__L.._---Jo .,
(a) e: = 0.1; ep expanded up to e: 2 , tiC = 20% .
Fig. 4.7 Comparison of pressure distributions from the perturbationanalysis method with the exact solution for ellipticairfoil.
.'1
Y( Xl = .] DC. S:'~T( 4X( I-Xl)
TIC = 2)·OJ~
SCU~~E--- PERIUR~rra~ LINE--- EX=(i
77
o
T
.2
itLrt;
0~
u
I-.2
-.6
-.8
-1.0a .1 .2 ., ., .& .5 ., .8 ., 1.0
X
(b) e: = 0.1 . <p expanded up to e: 3 , tiC = 20~~,
Fiq.4.7 Concluded.
78
l."~y( x; = .150.S~Tr t:X( I-Xll
TIC = 30.00~
.8
.6 SOU~RE--- PERTUR8~ila~ LINE--- E~T
au
.2
o
-.2
-.q
-.8
...l
J
-.8
1.0.a.,.5X
.'t.2.1-1.0 -----. ""'---__..I.--.__~__.L...___L___L.___.L____L-------J
a
(a) E = 0.15; $ expanded up to £2~ tiC = 30% .
Fig. 4.8 Comparison of pressure distributions from the perturbationanalysis method with the exact solution for ellipticairfoil.
a..u
1.
.8
.8
.2
o
-.8
-.8
Y( Xl = •150:aSrRT( 4X( I-Xl J
TIe =30· ocr..!
RLPHA= •2=O:3E-01 BETR= •97XC +00
SCU~E--- PERTU~r[ON LINE--- E~T
79
-1.O ....... ---'----......----'-__--L.__-L-__-'--__.L-_---lL.-..---J
a
(b) E = 0.15; ~ expanded up to £3
Fig. 4.8 Concluded.
tiC = 30%
80
I.~y( x: = .D3G.S:RT~ L1xr 1-x: i
TIC = la.OC~
.8 AL?HA= .2505E-02 8~TA= .9975=-CC
.5 SOU~~E--- PERTUR8~rION LINE--- EX~:i
...
•2
!L.U a
-.2
-.Ii
-.6
-.8
-1.0 L--__..1.-__...L.-__--L..__~___.L.______L_____":......___"'____~_ __...
a .1 .2 •., ,Ii: ,5 .e ,7 ,a .~ 1.0X
(a) E = O~05; ~ expanded up to £3; tIC = 10%
Fig. 4.9 Comparison of pressure distributions from the perturbationanalysis method with the exact solution for elliptica i rfoi 1 •
lot.8
.6
.'1
-.s
-.8
YC Xl = .] DC. 5G~T( 4X( l-XJ J
TIC = 2J.OJZ
SCUq~E--- P£~iU~3rlrrON LINE--- EX=CT
81
(E
I
3I
-1.0a .1 .2 .3 .t .& .5 .1 .e ., 1.0
X
lb) ~ = 0.10 <p expanded up to Eo 3; tiC = 20% .
Fiq.4.9 Continued.
82
y£ X~ = .1S0.SCRT( YX( l-Xl ]
TIC = 30.00~
.8 RLPHA= .2303E-O1 BETR= .977OE+OO
.6 SOU~RE--- PERTURBRrrON LINE--- E~T
a..,u
.2
-.2
-.11
-.6
J
-.8
.5X
.Il.2.1-1 .0 r..-.._----.j"-_---'__--'-__--'-__--'-__~___L___...L_..__...L..__ ___J
o
(cJ e: = 0.15; <p expanded up to E 3 ; tiC = 30% I.
Figure 4.9 - Concluded.
-1.0
-.9
R.3
vex]: .100)l(QX1+1- SCRT(SlXJ)) NPHN= toC(Xl: .OOO:lX(l-Xl S(Xl:X(1-XJlI:s3 ON: 500.0
T/C= 6.S0~ RNGlE OF RTTRCK= 0.00 DEGREE
RLPHR= .2519E-D2. a.
-.6
-.'1
CROSS----PERM8RTION L]NE---P~NEL
-.2
tJ 0
\ \
.8
.8
/f+....
+.,. +
+ .. .. .... ..
1.0-, I 0 .1 .2 ., .'1 .s .8 .7 .e .~ 1.0
X
(a) € = 0.03248 y = o0 ; ep expanded up to e: 2 tiC = 6.5%
10 panels wi th cos i n.e spacing.
Fig. 4.10 Comcar;son of pressure distributions from the perturbationanalysis method with the panel method for symmetricalJoukowski airfoil.
-1.0
-.8
Y( Xl = .10C:J( CC Xl .. /- SCRT( 51 Xl}] NF~~l= 30
C( Xl = •OGO:,X( i-Xl S( Xl =X( L-Xl :- ~3 C~~= SCO.O
TIC: 6.S0~ ~~LE OF RTTRCK= 0.00 CEGREE.
84
~LPH~= .2515E-02. O. - 8ET~=1.0000
a..u
-.6
-.~
-.2
o
•2
.'i
.8
.8
cRass----PERrLR8~TrON LI NE--- -P~~l::L
+ ..
I.a-.1 a .1 .2 .3 .tt .5 .8 .1 .8 .~ 1.0
X
(b)0
expanded to e;2; tIC = 6.5%e: = 0.03248 . y = o ; <P up,
30 panels with cos; ne spac ;.ng.
F; g. 4. 10 Continued.
-1.0
-.8
Y( Xl = .10C:J( C( Xl +/- SeRif Sf Xl)] ~.P~l'~= SO
CC Xl = .OOOj1X( l-Xl S( Xl =XI i-X) ~·.3 c~~= 500.0
T/C= 6.5a~ ~\LLE OF ATT~CK= 0.00 DEGREE
85
RLPH~= .2S19E-Q2. o. EET~=l .0000
-.6
-.\1
-.2
.'1
.8
.8
cRass----PERrLR8~T ION LINE--- -P~~~EL
I.G......__.......-.__.L--__..L.-__""---__.a....-.--~--'___ __.....ll....__ _J____.l____.J
-.1 ax
(c) E = 0.03248 ; y = 0°; $ expanded UD to £2; tiC = 6.5% ;
50 panels with cosine spacing.
Fig. 4.10 Concluded.
86
-9
-8
Y( Xl = .lCC2( C( Xl +/ - SC~T( 51 Xl]) ~,pq'J= toC( Xl = .OCO:sX( l-Xl S( XJ =X( L-Xl ~.3 C~J= 500.0
T/C= 6.50% ~~LE OF ~TTRCK= 6.0J DEGREE
~LPH~= .2519E-02. O. EETH=1.0000
-7 CROSS----PERrLfi8RT ION LINE----?~~~EL
-6
x
+
1L---~:.--_L___-1-__-L..__--'---~:__---':__-~:_----I:__-~:__-__:_-.1
a
-2
·1
-s+
~ -li....
+
-J
(a) e: = 0.03248 y = 6°; ~ expanded up to e: 2 tIC = 6.5%
10 panels with cosine spacing.
Fig. 4.11 Comparison of oressure distributions from the perturbation.analysis method with the panel method for symmetricalJoukowski airfoil.
87
-9
C( Xl = .000-:([ l-X) 51 X} =X( l-X] ~)f3 O,'J= 5eo.o
-9 T/C= S.SOl ~\GLE OF ATT~CK= S.CO OEGR~E
~LPH~= .2519E-02. O. aEr~=1.0CCO
-7 cRass----PERrL.~8~TION LI NE---?H~,EL
-6
-5
x1...-.---..LIIIII---~-_..1..-__..1...---...L.---"'----.L..___.........----"'-_:___-.-...--~-.1
a
-]
-1 .
-2
(b) e: = 0.03248o
; y = 6 ; ~ expanded up to £2; tiC = 6.5%
30 panels with cosine spacinq.
Fig. 4.11 Continued.
88
-9Y( Xl = .100:.w( a Xl .. / - SORTf Sf Xl 1] NPH~I= 50
cex): .OOO:sX(l-X) S[X)=X[l-X)~~3 CN= 500·0
-8 T/C= 6.50% ~NGLE OF RTT~CK=· 6.00 DEGREE
~LPH~= .2519E-Q2. O. 6ET~=l.0000
-7 CROSS----PERfUR8RTION LINE----P~NEL
-8
-5
xl.....-__....._-...l'--- ~_........__-...__-...Io____..____.&..____'_____"____~
-.1
a
-]
-1
-2
(c) E = 0.03248 y = 6°; $ expanded up to £2; tiC = 6.5% ;
50 panels with cosine spacing.
Fig. 4.11 Concluded.
-1.OiI
-.8
y(X~ = .CSQ.( a X~ ~/- seRf; S{Xl]l ~?~~= SO
C( x: = ·OJC~:<: t -Xl S( x: =x( t -x: lIlt3 CN= SC-J.O
T/C= 3.25% ~GLE OF ~iT~~{= 0.00 OEG~~E
89
~LPHP= .S262E-03. O. 8ETP=1.CCOO
-.5
-.'1
-.2'--
.8
.8
cRas·s----PERrLR8~TION LJ NE----Ph~:E:L
J1.0L----I----.L..---L.---L.----'----L.----'-:------'::---~:__-__:=__-___;l.Q-.1 a .L .2 .~ ." .5
x
(a) E = 0.01624 ; y = 0° ~ expanded up to 82
; tiC = 3.25%
50 panels with cosine spacing.
Fig. 4.12 Comparison of oressure distributions from the perturbationanalysis method with the panel method for symmetricalJoukowski airfoil.
90
..9V(X]= .OSO~(aX]+/- SORT( S(X))) N?~N= 50
cex): .ooo~xr l-Xl S( Xl =X( t-X)~.3 ON= SCO.O
..8 T/C= 3.2S~ h'~LE OF ATTHCK= 6.00 DEGREE
~LPH~= .62S2E-03. O. 8ET~=1 .0000
..' CROSS----PERn..:R8~TION LINE----PRNEL
-8
-5
-3
-2
-I
xI .......__~_-..I......-.._-"-_-..__-...__-.l._-----_.---...__-L..__~---'
-.1
o
(b) € = 0.01624 ; Y = 6° ~ expanded up to £2 tiC = 3.25% ;
50 panels with cosine spacing.
Fig. 4.12 Continued.
91
-9YC Xl = .OSC:~d cr Xl f-/- SC~T( Sf Xl}) ~?~:'l= 5J
C( Xl = .OOO~X( t -Xl S( Xl =X{ i-Xl • ..3 C,'~= 5oo.J
-8 T/C= 3.25Z ~~LE OF ATTRCK= 12.00 OEGRE~
~LPHR= .6262E-G3. O. EETH=1.0000
-7 CROSS----PERfLR8RT I ON LINE----?~~:EL
-6
-s
-2
-1
a +
.7.1al.....-.-------.......---- ...a.-.__.J--__~___L_____a_____Jo.___....J..___~----'
-.1x
(c) e: = 0.01624 ; y = 12° $ exoanded up to E 2; tiC = 3.25% ;
50 panels with cosine spac.ing.
Fig. 4.12 Concluded.
-1.0
-.8
-.6
-.2
.'1
.8
.8
92
Y( Xl = .1 OC)l~ C( Xl .. / - SeRTf S( Xl ); NP~~= SJ
C( Xl = .oaoJtX( l-X) Si Xl =X( L-Xl .·3 O~= 5CO.0
T/C= 6 .50% ~N;LE Of ATTriCK= a.00 CE~,~EE
~LP~~= .2519E-02. O. 8ET~=1 .cooacRass----PERr~E~T reN LINE----P~~~EL
1.0-,I a .t .2 .3 ... .5 .8 .7 .8 ., 1.Q
X
(a) E = 0.03248 y = 0° cf> exoanded up to £2 tIC =6.5%
~O panels with cosine spacing.
Fig. 4.13 Comparison of pressure distributiqns from the perturbationanalysis method with the panel method for symmetricalJoukowski airfoil.
-9L"--
-8
V( Xl = .1 JC.; :~ Xl ... / - SCR r: S( Xl )) NP~~'~= SJ
CC Xl = ·OCC·"X[ L-Xl S~ Xi =X, l-Xl ~ -3 c~.,= 500.0
T/C= 6.50% ~~~LE Or ~TTrlCK= 6.00 CE~~EE
93
~LPh~= .25I9E-Q2. O. 8ET~= I .CoOO
-7
-5
-5
-3
-2
-I
a
CRass~-- -PERrL~e~T ION LINE--- -P~~~~L
1-.1 a .1 .2 .] .'t .5 .8 .7 .e .~ 1.0
X
(b) e: = 0.03248 ; y = 6° <P expanded up to e: 2 • tiC = 6.5% ., ,
50 panels with cosine spacing.
Fig. 4.13 Continued.
y( X) = .10Q)I( C{ Xl +-/ - SCRT( Sf X] ) 1 N?~N;: SO
C( Xl = ·OOG~:<' L-Xl S{ XI =X( l-X: :..3 OtJ= 5CJ.0
T/C= 6.50% ~GLE Of RTTPC:~ 12.CO OEGR~E
94
~LPH~= .2519E-02. O. 8ET~= 1•DeOD
-7
-6
cRass----PE~ruR8~T ION LI NE----Ph~~EL
x
(c) E = 0.03248 ; Y = 12° ; ~ expanded up to E2 ; tiC = 6.5%
50 panels·with cosine spacing.
Fig. 4.13 Concluded.
-1.0
-.8
YC Xl = .ISQ:s( cr Xl pi - seRif S( Xl }) NP~l'l= sa
C( Xl = .OOC~X( l-XJ S( Xl =X( I-Xl ~ -3 CN= 500.0
T/C= 9.75Z h~~LE Of RTTHCK= 0.00 CEGREE
95
~LPH~= .572CE-a2. o. EETf.l=! .0000
-.6
-.11
-.2
.2
.Ii
.8
.8
cRass----PERM8~T ION LINE----PR~iEL
.. to
I.O'--- """--__-'---_--------....-.-__..A...-__....J--__-L..__--L-__--J.-__~
-,I 0x·
(a) e: = 0.04872 ; y = 0° ; <p expanded up to e: 2; tiC = 9.75% ;
50 panels with cosine spacing.
Fig. 4.14 Comparison of oressure distributions from the perturbationanal.ysis method with the panel method for s.ymmetricalJoukowski airfoil.
96
-3.: Y( Xl = .15Q.( C( Xl ... / - SOR T( SI XJ J) ~~pq'l: SQ
+ C( Xl = .0005X( l-Xl S( Xl =X( l-Xl ·.Jl3 CN= 500.8
8ETR=1.00GO~LPHH= .5720E-02. O.
T/C= ·9.7S~ ~~LE OF ~TTHCK= 6.00 CEGREE
cRass----PERfLRB~TION LINE----?~NEL
I.O"----......---.Ao...---""--__...I---__.........__~___.l...____'____'_____L__ ___'
-.1
-.2
.5
-2.2
-2.&
.-1.8
fj -1.0
x
(b) € = 0.04872 ; y = 6° ~ expanded up to E 2 ; tiC = 9.75% ;
. 50 panels with cosine spacing.
Fig. 4.14 Continued.
97
TIC= 9.75% ~~~LE OF qT7~CK= l2.00 DEGREE
CC Xl = .CCJ-:«( t -XJ S( XI =X~ i-Xl ~ -3 O~~= 5:C·J
~LRH~= .S72CE-02. O.
\11-
-8
-7- '"
CROSS-- --PERflR8RT rG~ L1~E-- - -P~~~EL
-6
-5
-3
-2
-1
a
I ......__...-..-........_~__...a...-__.........--.,...---~__---'-__~__---'--__-.A..-__-
-.1x
(c) e: = 0.04872 ; y = 12° ; <p expanded up to E 2 ; tiC = 9.75%
50 panels with cosine spacing.
Fig. 4.14 Concluded.
98
Y( Xl = .200:a( a Xl +/ - SORi( S{ Xl ]] ~,?~N= 50
CC Xl = .OOO~X( t -Xl S( Xl =X( L-X) »~3 DN= 5eo.o
T/C= 13.00% ~~LE OF RTTPCK= o.co OEGR~E
~LPH~= .1030E-O l. O. 8ET~.: 1.0000
-.6
-.2
~ 0
.2
.8
.8
cRass----PERnR8~TION LI NE---?h~~EL
+
1.0'---__.....__..L.-__-&-.__..L-__-'------'-------------L----lo-----"------.1 G
x
(a) E = 0.06496 ; y = 0°; ~ expanded up to E 2 ; tiC = 13% ;
50 panels with cosine spacing.
Fig. 4.15 Comparison of pressure distributions from the perturbationanalysis method with the panel method for symmetricalJoukowski airfoil.
99
-3.0Y( Xl = .200:.d a Xl .. /- SORT( S{ XJ)) ~l?~N= 50
C( Xl = .QOO:Jxr L-Xl S( XI =X( l-XJ ~ -3 ON= seQ.o
-2.8 T/C= 13.ao~ ~GLE OF RTT~CK= 6.00 DEGREE
~LPH~= 11 030E-Q l. aI BET~=l .occa
CROSS----PERfCHBHTION LJNE---P~NEL
.i .2 ., .'1 .5 .8 .7 .8 ., 1.0X
(b) e: = 0.06496 ; y = 6° 4> expanded up to e: 2 tiC = 13%
50 panels with cosine spacing.
Fig. 4.15 Continued.
.8
-.6
-.2
1.0~-_........__........__--a...-__-.a.._-~-_---.l_-_.L-..__.l-__.J..-__...L-----'
-.1
-1.8
-2.2
~-l.O
100
-9-
C( Xl = .COO.:«( l-X~ 5{ XI ::X( l-X1 :. - 3 O~j= 5:0.0
-9 TIC=" 13.00% ~'~L~ OF qTT~CK~ L2.CO OcGRE~
~LPHM= .103CE-Ol. O. 8ETH=1.ODeo
-7 cRass----PERrLR8~iIC~ LINE----?A~EL
+
x.,a
l~__~:11111'~....a...-__----__~_--.....----.....-_._..___....___'_____...____J
-.1
o
-8
-s .
-1
-2
" -3
(c) e; = 0.06496 , <P expanded up to e;2 tiC = 13%
50 panels with cosine spacing.
Fig. 4.15 Concluded.
Table 4.1 Comparison of 11ft coefficients of the perturbation, panel and exactsolutions for the symmetrical Joukowskl airfoils
. i
Slenderness Exact Perturbat1on Panelratio C - C C' - C
C.. ,ex C.. ,pt C.. ,P"!,pt !,ex x 1001 "'8" ",ex x 100%
£ C.. ,ex t,ex
y = 6°0.06496 0.71588 0.72245 0.67684 0.92 -5.45
0.04872 0.70234 0.70603 0.66472 0.53 -5.360.03248 0.68797 0.68961 0.65377 0.24 -4.970.01624 0.67278 0.67319 0.63080 0.06 -6.24
y = 12°0.06496 1.42391 1.43698 1.34485 0.92 -5.550.04872 1.39698 1.40432 1.32128 0.53 -5.420.03248 1.36840 0.37167 1.29823 0.24 -5.13
0.01624 1.33819 1.33901 1.23750 0.06 -7.52
-.8
-.5
r
-'2~I
:J D~II
I
.2~
.8
.8
102
Y( Xl = .050.' a Xl ~/ - SORT( Sf Xl )) 'L:)~~;.: SC
C( Xl =l·OOC:tX[ L-X) S( Xl =X( l-Xl "»3 ON= 5CJ.O
T/C= 3.25Z A~LE OF RrT~CK= O.CO QEGR~~
~LP~~= .6216E-Q3. -.3137E-04 3cT~=1 .OCCO
cRass----PERrLJ~8~ r ION II NE --- -P~~E'-
1.0-,t .L .2 .] .'1 .5 .8 .7 .8 .~ 1.0
X
(a) e: = 0.02688 y = 0° <p expanded up to e: 2 tIC = 3 .. 257~
50 panels with cosine spacing.
Fig. 4.16 Comparison of pressure distributions from the perturbationanalysis method with the panel method for camberedJoukowskf airfoil.
-9
-8
-7
-8
-s
-3
-2
-I
Y( Xl = .OSO:t( a Xl +/- SDRTI Sf Xl)) N?~N= 50
C( Xl =1•OOOllX( L-Xl S( Xl =X( l-XJ ll1.3 O~J= 5eo.o
T/C= 3.25% ~LE OF RTT~CK=6.CO DEGREE
~LPH~= .62~6E-03. -.3137E-a! 8ETR=1 .OGeo
CROSS----PERrLRSHT ION II NE----?~~~EL
103
o
1~__......-.__~__"",,--_~ ......... -L..__....L---~-_--L-._-""
-.1x
(b) e: = 0.02688 ~ exoanded up to £2; tiC = 3.25% ;
50 panels with cosine soacinq.
Fig. 4.16 Continued.
-9-
-8
-7
-s
-s
veXl= .05Q~raX] ~/- SCRT( S(Xl I 1 ~?~~= SO
C( Xl =1·OOC.X! L-Xl Sf Xl =X{ ~ -X; ~»3 ON= SCJ.O
T/C= 3.25% R\GlE QF AiT~CK= l2.00 DEG~EE
~LPHR= .62~6E-a3. -.3137E-04 3E i h:l .CCOO
CROSS----PERfLREfiT ION LI NE--- -P~~~EL
104
-3
·2
-I
a
1-,I a .1 .2 .] .i .8 .8 .7 .e .s 1·0
X
(c) e: = 0.02688 y = 12° <P expanded UD to £2 . tiC = 3.25%,
50 panels with cosine spacing.
Fig. 4.16 Conc 1uded.
-l.C-
-.8
-.6
vex}: .loO-(aX} ~/- SO~T(S{Xl J) ~?qN= SO
C( Xl =1·000]1 X( l -X1 S( X) =x(. l -Xl ~ t 3 O~~= 500 •a
T/C= 6.50Z ~GLE OF RTT~CK= 0.00 OEGRE~
~LPH~= .249~E-02, -.2538E-03 8ET~=1 .ODeD
cRass----PERfLqS~T ION LI NE----?~NEL
105
-.'1
-.2
.8
.8
x
1.a........__~ -"-__--L.__-..I. L...-__..L.-.__..L-__-l-__-I-__-J
-.1 a
(a) e: = 0.05376: '; Y = 0° $ expanded up to E 2 ; tiC = 6.5% ;
50 panels with cosine spacihg.
Fig. 4.17 Comparison of pressure distributions from the perturbationanalysis method with the panel method for camberedJoukowski airfoil.
106
YC Xl = .1 QO:s( CC Xl +-/ - SORT( Si Xl )) ~P:~~'J= 50-9
CeX)=1.0CO:sX(L-X) S(XJ=X(l-X)"I .. 3 ON= 5CC.O
-8 T/C= 6.SQZ h~LE Of RTTRCK= 6.CO DEGREE
~LPH~= .2~9qE-02. -.2538E-03 8ET~=1 .ODCO
-7 CROSS----PERrLR8~TICN LINE----P~NEL
-6
-s
-3
-2
-I
a
1.0.~.1x
,tt.2a1......--__.c;p---....----~ --.a.....---.&....---"""'---.a--_ _.JL...__._ ___J~_~
-.1
(b) e: = 0.05376 ; Y = 6° ~ expanded up to £2; tiC = 6.5% ;
50 pa ne1s wi th cos i ne spac.i ng.
Fig. 4.17 Continued.
107
Y( Xl = .1 OO~( a Xl +/- seRTf Sl Xl )] ,'J?~~~:: SO-9
C( Xl =1·OOC·sX{ l-Xl S( XI =X( t -Xj 11-3 ON= 5CJ.C
-8
-7
-6
-5
-3
-2
·1
a
T/C= 6.50% ~GLE OF RTT~C~= l2.CO OEGR:E
~LPH~= .2~9~E-a2. -.2538E-03 8ET~=1 .0000
CROSS----PERrL.~8~T ION lINE----P~~~~L
+
1~-~--.-~-------'_---'--__..L...----J.--....L---L-----1---..J...--J-.1 .2,3.II .5 .8 •1 .8 •3 I •0
x
(c) e:"= 0.053761 ; y = 12° ~ exoanded UD to £2 tiC = 6.5%
50 panels with cosine spacinq.
Fig. 4.17 Concluded.
0U
-I.!)
-.8
-.6
-.q
-.2
o
.2
.Ii
.8
.8
108
.,
+ f' .~
Y( Xl = .150.{ C( Xl +/- SORTf 51 Xl )) NP~N= 50
CeX)=1.000-X{1-Xl SlXl=X(1-X) ••3 ON= SCO.O
T/C= 9.75% ~~LE OF ATTRCK= 0.00 DEGREE
RLPHA: .5593E-C2. -.8722E-Q3 BET~=l.coao
CROSS----PER~~8~TION LINE----PRNEL
+
1.0-.1 a .2 .3 ., .8 .e .7 .8 ., 1.0
X
(a) € = 0.08064 y = 0° <f> expanded up to £2. tiC = 9.75%,
50 panels with cosine spacing.
Fig. 4.18 Comparison of pressure distributions from the perturbationanalysis method with the panel method for camberedJoukowski airfoil.
109
-~.aY( Xl = .1SCJl( CC Xl +/ - SC~ 'i( 51 Xll] ~~pq.'~= 50
C( Xl =1.OCO~X( l-Xl S( Xl =X[ l-XJ :It -3 C~~= 500·0
-3.5
l'
-3.0~
++...
-2.5
-2.0
~-l.S
-1.0
-.5
a
T/C= 9.7S~ ~'~LE OF HTTHCK= 6.00 DEGREE
RLPHH= .5593E-02. -.8722E-03 EET~=l .0000
CROSS----PERM8~TION LINE----PRNEL
...+
x
.5 . , .
I.a-....-.__.... .---~__.........__......_.____'_ ____"___.llo...o__~
-.1
(b) e: = 0.08064 ; Y = 60~ expanded UD to E
2; tiC = 9.75%
50 panels with cosine spacinq.
Fig. 4.18 Continued.
110
-~
I
II. ·9...-.
·7J
I~,I
: If --:
Y( Xl = .1 sa.( a Xl "./ - s:.~ ~( Sf Xi ~; t\P.~'·~= 50
TIC= 9.75Z ~~LE CF ~Tr~CK= t2.00 C:GRE~
~LPH~= .SS93E-C2. -.8722E-03 eET~=!.OOGO
cRass----PERfL~a~i I eN LINE-- --?~~~EL
-6
-3
-2
-1
o•
.~
x••.1a1~-_.....-.-~~......-.--....a--__....L.-___I_____'______L..__........____""____..___~
-.1
(c) e: = 0.08064 ., $ expanded up to £2 tiC =9.75%
50 panels with cosine spacing.
Fig. 4.18 Concluded.
111
-1.0
..
Y( Xl = .200.( a Xl +/- SQRT( SI Xl )1 NPf1N= 50
C(X)=1.000-X( l-Xl SI XJ=X[ 1-~ ••3 ON= 500.0
T/C= 13.0rrl ~GLE OF RTTRCK= 0.00 DEGREE.8
••
-.6
-.8
-.'1
+-+
,.-.2
+..
+
a.. au
.1
~LPHR= •9900E-{]2. -.2 I 20E-(2 BETR:1.0000
CRCSS----PERTURBATION LINE----PRNEL
xo
I.OL-__..L__--L.__--1 L-__~--~--___'::__-~::_---'=_--_:_-__:_
-.1
(a) e: = 0.107521 Y = 0° ; ~ expanded up to £2 tiC = 13%
50 panels with cosine sp~c;ng.
Fig. 4.19 Comparison of pressure distributions from the perturbationanalysis method with the panel method for camberedJoukowski airfoil.
112
-3.C.-·'(( x: = .2JJlI( a x: .. / - S~R ~( S~ Xi 1) ~.;)~n= 50
:( X1 =1 .CJD.:<! l-Xl S(:<: =X( l-X: '-.3 O~L: 5CG.O
-2.5- TIe = ! 3 . JO% ~,~~= CF Mi i~C.<':: 5.CO OEGR::::
-2.2- CRass ----PE.~ fl,R3~T [ O~l LI~~E.- - - -?~~Ef_
..
-1.9
-l.ll
~ -1.0
-.6
-.2
1.0
... + .. • .. • ~ .,.
x••.2.tal.a'--__--'-..----..I~--~__.._._I'___ __' __'_ ...._.__.....---
-.1
.8
(b) e: = 0.10752" ~ expanded up to £2 tiC = 13%
50 panels with cosine spacing.
Fig. 4.19 Continued.
113
-9Y( Xl = .20(h( a Xl .;./- SORTr S( Xl }) N?~N= 50
(eX)=1 .OOO~X{ l-Xl S( XJ =X( L-~~~3 ON= SCO.O
-8 T/C= 13.00% ~~LE OF RTT~CK= 12.00 DEGREE
-7 +
~LPr~= .9900E-Q2. -.2120£-02 8ET~=1 .oeoacRass----PER~~8RTION LINE----P~NEl
-6 +
•
x1.......__..........-....~.....-.__..-...-__.....-.__---.__--..._--"'"--_-.lI.-.-__L....-_-"-------.&
-s
a
-]
-I
-2
(c) E = 0.10752 ; Y = 12° $ expanded up to £2 tiC = 13%
50 oanels with cosine spaci"nq.
Fig. 4.19 Concluded.
Table 4.2 Compar1son of 11ft coeff1cfents of the perturbat1on. panel and exactsolutions for the cambered Joukowski airfoils
Slenderness Exact Perturbation Panel C - C C - Cratio C C.. ,pt C t.pt t.ex x 100% t,pn t.ex x 100%i,ex -t,pn C.. ,ex C
£ t,ex
y = 00
0.10752 0.61721 0.69115 0.63346 11.98 2.630.08064 0.46641 0.50658 0.46610 8.61 -0.070.05376 0.31268 0.32987 0.30390 5.50 -2.810.02688 0.15689 0.16101 0.14812 2.62 -5.59
y = 6°0.10752 1.32350 1.40981 1.29947 6.52 -1.820.08064 1.16264 1.20984 . 1.12378 4.06 -3.340.05376 1.99734 1.01767 0.95359 2.04 -4.390.02688 1.82841 0.83332 0.77878 0.59 -5.99
y = 120
0.10752 2.01528 2.11303 1.94816 4.85 -3.330.08064 1.84614 1.89984 1.76670 2.91 -4.300.05376 1.67107 1.69432 1.58984 1.39 -4.860.02688 1.49084 1.49649 1.38619 0.38 -7.02
115
Chapter 5
CONCLUDING REMARKS
The perturbation analysis method has been used to analyze the
potential flow, due to a uniform stream, past an axisymmetric body or a .
two-dimensional airfoil. The theoretical formulations have been
described in Chap. 2. Results obtained from the above method for
certain classes of either axisymmetric slender bodies or two-dimensional
thin airfoils have been compared with those generated by the panel
method, as well as with exact solutions whenever available.
Numerical experiments for ellipsoidal bodies and elliptic airfoils
have shown that the higher order expansion of +in £ gives better
results, as compared with exact solutions than lower order repre
sentations. The perturbation analysis method for axisymmetric slender
bodies give pressure distributions· which compare well with the panel and
exact solutions for small slenderness ratio up to £ = 0.1, corresponding
to a LId ratio of 5. The perturbation analysis method is quite
inexpensive, because little computer memory is required; but it 1s only
applicable to slender bodies. It is suggested to be an applicable
method as a preliminary step in modelling some simple fuselages as part
of the analysis of a wing mounted on'a slender axisymmetric fuselage.
The utility of the perturbation analysis method for the two
dimensional airfoils is also confined to the thin airfoils with small
slenderness ratios up to £ = 0.03248, corresponding to tIC = 6.5%. The
116
perturbation analysis method seems to relieve the numerical instability
of the panel method at some regions near the leading edge of the airfoil
surface. The perturbation analysis method can also provide more
accurate results for thin airfoils (i.e., £ tends to a small value) as
compared with the panel method. Moreover, the computing time and
computer memory required are relatively small as compared with an
existing panel method. The perturbation analysis method is recommended
as a method which will provide reliable solutions for thin airfoils at
low cost. Using the thin airfoil analysis as an illustrative example,
the perturbation analysis method can be extended to analyze the three
dimensional wing as the next task. If the extension to the three
dimensional wing can be completed, the next step will be the development
of a program for the wing design problem with the axisymmetric fuselage
effects on a pair of wings. This study has been only a preliminary step
for the wing-fuselage design problem.
In addition, the perturbation analysis method can be applied to
solve the electrostatic and magnetostatic potentials around a slender
conducting body in a similar manner.
117
REFERENCES
1. Van Dyke, M., Perturbation Methods in Fluid Mechanics, TheParabolic Press, 1975.
2. von Karman, T., ·Calculation of the Flowfield Around Airships,"NASA TM -574, July 1930.
3. Zedan, M. F. and Dalton, e., ·Potential Flow Around AxisymmetricBodies: Direct and Inverse Problems,· AlAA Journal, Vol. 16, No.3, March 1978, pp. 242-250.
4. Zedan, M. F. and Dalton, e., -Higher-Order Axial SingularityDistributions for Potential Flow About Bodies of Revolution,"Com uter Methods 1n lied Mechanics and En ineerin , Vol. 21, No., arc , pp. •
5. Shu, J.-Y. and Kuhlman, J. M., ··Calculation of Potential Flow PastNon-Lifting Bodies at Angle of Attack Using Axial and SurfaceSingularity Method,U NASA CR-166058, February 1983.
6. Kuhlman, J. M. and Shu, J.-Y., ·Potential Flow Past AxisymmetricBodies at Angle of Attack," Journal of Aircraft, Vol. 21, No.3,March 1984, pp. 218-220.
7. Handelsman, R. A. and Keller, J. B., -Axially Symmetric PotentialFlow Around a Slender Body,· Journal of Fluid Mechanics, Vol. 28,part I, Apr1l 1967, pp. 131-147.
8. L1ghth1ll, M. J., "A New Approach to Thin Airfoil Theory," T:leAeronautical Quarterly, Vol. 13, November 1951, pp. 193-210--.--
9. Hoogstraten, H. W., ·Uniform Valid Approximations in Twodimensional Subsonic Thin Airfoil Theory,· .Journal of EngineeringMathematics, Vol. 1, No.1, January 1967, pp. 51-66.
10. Geer, J. F. and Keller, J. B., "Uniform Asymptotic Solutions forPotential Flow Around a Thin Airfoil and the ElectrostaticPotential About a Thin Conductor,I' SIAM Journal on AppliedMathematics, Vol 16, No.1, January 1968, pp. 15-101.
11. Geer, J. F., "Uniform Asymptotic Solutions for the Two-dimensionalPotent1 a1 F1 e1d About a 51 ender Body, II SIAM Journal on Appli edMathematics, Vol 26, No.3, May 1974, pp. 539-553.
12. Handelsman, R. A. and Keller, J. B., "The Electrostatic FieldAround a Slender Conducting Body of Revolution," SIAM Journal on
118
Applied Mathematics, Vol. 15, No.4, July 1967, pp. 824-841.
13. Homentcovschi, D., "Uniform Asymptotic Solutions for the Twodimensional Potential Field Problem with Joining Relations on theSurface of a Slender Body," International Journal of EngineeringScience, Vol. 20, No.6, 1982, pp. 153-767.
14. Homentcovschi, D., '·Uni form Asymptoti c Sol uti ons of Two-dimensi analProblems of Elasticity for the Domain Exterior to a Thin Region,"SIAM Journal on Applied Mathematics, Vol. 44, No.1, February 1984,pp. 1-10.
15. Karamcheti, K., Principles of Ideal-Fluid Aerodynamics, Wiley,August 1966.
16. Thomas, J. L., Luckring, J. M•. , and Sellers, W. L., III,"Evaluation of Factors Determining the Accuracy of LinearizedSubsonic Panel Methods," Presented at the 1st AIAA AppliedAerodynamics Conference, Paper No. AIAA-83-1826, July 1983.
17. Bristow, D. R. and "Grose, G. G., "Modification of the DouglasNeumman Program to Improve the Efficiency of Predicting ComponentInterference and High Lift Characteristics," NASA CR 3020, August1978.
18. Haskew, B., "Predicting of Subsonic Aerodynamic Characteristics--aCase for Low-order Panel Methods, II Presented at the AIM 19thAerospace Sciences Meeting, Paper No. AIAA 81-0252, January 1981.
19. Bristow, D. R. and Hawk, J. D., "Subsonic Panel Method for theEfficient Analysis of Multiple Geometry Perturbations," NASA CR3528, March 1982.
20. Haskew, B., Rao, B. M., and Dvorak, F. A., ·Prediction ofAerodynamic Characteristics for Wings with Extensive Separations, ..Paper No. 31 in Computation of Vlscous-inv1scid Interactions,AGARD-CPP-291, September 1980.
21. Maskew, B., ·Predicting Aerodynamics of Vortical Flows on Threedimensional Configurations Using a Surface-Singularity-PanelMethod,· Paper No. 13 1n Aerodynamics of Vortical Type Flows inThree Dimensions, AGARD-CP-342, April 1983.
22. Glauert, H., IIA Generalized Type of Jouko~ski Aerofoil," Reportsand Memoranda of the Aeronautical Research Committee No. 911,January 1924.
23. Currie, I. G., Fundamental Mechanics of Fluids, McGraw-Hill BookCompany, February 1979.
APPENDICES
119
APPENDIX A
DETERMINATION OF THE COEFFICIENTS OF aCe) and e(e) FORAXISYMMETRIC BODIES
Appropriate constants ak need to be determined such that 9k(x) is
regular at x=O. One can proceed in the following way. The
functions 9k(x) are defined by
g(x,e) • 2k 2 2 1/2= L 9 (x)e = {[x - a(e)] + e S(x)}k=O k
(AI)
• 2where «(e) = r «n en. Since S(O) =0 is assumed, as x is set equaln=l
to zero 1n Eq. (AI), one obtains
, k ) 1 (A2)
To determine «t explicitly, one squares both sides of Eq. (AI) and has
2 2 · "[k ] 2k[x - «(e)] + e S(x) = r .r gJ.(x) gk_j(x) ek=O J=O
Here, «0 =- x is defined, such that «(e) can be written as
•a( e) = L
n=O
Using Eq. (A4) in Eq. (A3), the left side of Eq. (A3) becomes
(A3)
(A4)
120
aD[k ]2k 2r .r ClJe ~_Je e: + e: S(x)k=O J=O
Equating the coefficients of the same power of £2 on both sides of the
above equation, one obtains
kr Cl.~_. + ~k 1 Sex) =
j =0 J J ,
k.r gJe (x )9k_Je(x)J=O
(A5)
where 6mn is the Kronecker delta, i.e., 6mn = 1 if m=n, 6mn =0 if m ~ n.
For k = 0
For k = 1
(A6)
. (A7)
Using Eq. (A2) and with the Taylor series expansion of Sex), Eq. (A7)
yields a1= 51 (0)/4 •
For k > 2, Eq. (A5) becomes
(ABa)
Using Eq. (A2) in Eq. (ASa), one has
(ABb)
One can determine ot recursively by using Eqs. (ABa) and (A8b) and has
the result as follows:
121
_ 51 (0) 2 51 (0)5 11 (0) 4a(e) - 4 e - . 32 e
1 [[SI(O)]2SIII(O) SI(O)[SII(O)]2] 6 + 0(£8)+02r 6 + 2 £ (A9)
Similiarly, 9k(x) can be made regular at x=1 and the coefficients Bk are
determined by following the same procedures as above. The resulting
expressions are as follows:
g(x,e)• #v 2k 2 2 1/2= I gk(x)e ={[x - B(e)] + e S(x)}
k=O(AIO)
• 2n .where B(e) =1 - I Bne with BO = 0 and Bl = - SI(I)/4 •
n=l
-(All)
k kI BJo B + ~k 1 S(x) = I 9J
o(X)9k_J- (x)j=O k-j' j=O
90(x) =1 - x
For k ) 2
k ) 0 (AI2)
(A13)
(A14)
(A15)
(A16)
122
See) =1 + SO(l) e2 _ SO(l)SOO(l) e4-,- 32
(All)1 [[so(l)]2sooo(l) SO(l)[SOO(1)]2] 6 8
+ 10 6 + 2 e + O(e )
By using the above resulting equations, a(e) and B(.e) can be evaluated.
123
APPENDIX B
BOUNDARY CONDITIONS IN COMPLEX PLANE
The complex potential ,(z) is defined in Eq. (2.35) as
+(z)= +(x,y) + i,(x,y) which is analytic in the z-plane outside G. The
derivative of the complex potential with respect to z is denoted by W{z)
and is known as the complex velocity, i.e.,
w(z) =~ =.!t + ..!tdz ax ' ax
=!! ,.!tay - ay
=u{x,y) - iv{x,y)
where u and v are the velocity components in x and y direction
respectively.
From Fig. 2.2, on the surface of the airfoil, one finds that
~ =!.ax u
The equation of the profile G is given by
(Bl)
(B2)
y = e [C(x) t IS(x)] on a < x < 1
*= € [C'(X) ±~]
Equating Eqs. (82) and (83), yields
v - € U [C' (x) ± ~~~{~)] = 0
Using Eq. (B1) in Eq. (2.39), and considering the real part of the
124
(83)
(84)
resulting expression, one has the same expression as Eq. (84).
To satisfy the boundary condition at infinity in Eq. (2.39), one
needs
8 SRe f f(~,£)d~ =0 and 1m f f(~,£)d~ = r
« B
where r is the total circulation about the body. Physically, the first
condition says that there is no fluid flow through the surface of the
body.
125
APPENDIX C
EXPRESSIONS FOR TWO-DIMENSIONAL AIRFOILS
1. Linear Operators of the Integral Equation
Linear operators L:,p=O, 1 in Eq. (2.50) are defined 1n Sec. 4 of
[11], and some of the related operators are given by
g(x,e:)
nbn(x) = &n,o - j~l aj(x)bn_j(x)
• n 2 2 1/2= L 9n(x)e ={[m - x - 1eC(x)] + e Sex)}n=O
(Cl)
(C2)
h(x,e) = r hn(x)en = {[s - x - 1C(x)]2+ e2S(x)}1/2 (C3)n=O
hn,j (x,F) = tr I~t~) j [ (T-x-i eC) -( 2k+l) [ F(T) (C4)
- 2~ F(P) (x+i eC) (T-x-i eC) p]l} deP=O pI e=O
where T = [m(e)-S(e)] cos2e + S(e) and C =C(x).
In particular, with n = j = 0 in Eq. (C4), yields
2. Determination of the Coeffi.cients of a(e) and 8(e)
-The functions g(x,e) and h(x,e:) are defined in Eqs. (C2) and (C3)
126
~ .where aCe) = L am em and B{e) =1 - L amem •
m=2 m=2
Following the similar procedures mentioned in Appendix A, the general
expressions for the determination of <1(£) and a{£) can be found as
below:
9 (O) = a , m) 2m m
91 (x) = iC{x)
For m ) 3
1 m-2 I
«- = - iC'(O) ~ - ~ ~ (I 9 (0) (C6)m in-I c. j~2 j m-j
9 (x: ) = -(I _ 1C(x) [ ()] 1 m~2[ () ()] (C7)m m x· am-I+ 90.-1 x + 2xj~2 aj a.n-j. - gj x gm-j x
Applying Eqs. (C6) and (C7) with m=3 and m=4, one obtains
8 = 0 = 8o 1
h (x) = 1 - xo
h (1) = Bm m
, m ) 2
h (x) = 8 + S(x)2 - 2 2(1-x)
For m ) 3
m-2 .8m = - ; Bm.,_,'1,. CI ( I) +~ LB. hi. ( I) .
, j =2 J rn-J
Using Eqs. (C8) and (C9) with m=3 and m=4, this yields
= C(x}S'(x)- 63 ~ i. 2
2(1-x)
h (x) = - B + B S(x) _ [C(x)]2S(x) _ TS(x)J~4 4 2 2(1-x)2 2(1-x}3 8(l-x)
(C8)
(C9)
By using the resulting expressions, a(E) and 8(£) "can be determined.
128
APPENDIX 0
ELLIPTICAL COORDINATE SYSTEM
Consider the equations
x =Acosh~ COSn
y =A s1nh~ sinn cose
z =A sinh~ sinn sine
(01)
With appropriate combination of x,y,z in Eq. (01), one has two families
of elliptical (~ is constant) and hyperbolic (n is constant) curves as
shown in figure D. The fineness ratio is defined by
cotht =length of major axislength of minor axis
The system of the reference unit vectors at point P are denoted
by e t , en' and es' corresponding to the coordinate t, n, s. Making
reference to [15], the scale factors can be determined
(D2)
2 2 1/2ht =hn =A(cosh t - cos n)
hS =A sinht sinn
(D3a)
(D3b)
129
! RIA
~o
X/A----t---t---t------+---+--~~-~--.~---I---\---+----~
Fig. 0 Two families of elliptical and hvoerbolic curveswith different values of ~ and n .
l::SU
The unit vectors of the desired coordinate system in terms of the
~artesian coordinate system are given by
et
=~ (f sinht CoSn + J cosht sinn coss + ~ cosht sinn sins)~
e = -}- (i cosht sinn - J sinht COSn coss - I( sinht COSn sins) (04)n n
es = - J sins + k coss
Making reference to Chap. 2 of [15], one obtains
(D5a)
ht
et
h e hs
esn n
vxv 1 a a a (D5b)= hthnhs at all as
htVth v 0
Tl n
Using Eqs. (03a) and (D3b) in Eqs. (05a) and (D5b), yields
2 2a • + a • _ coth t .!! - cotn .!! = 0~ ~ at an
To find out the boundary conditions, one uses the requirements of no
penetration and infinity conditions on Eq. -(D5a), and has
(06)
,(~,n) = constant = C· on (07)
as ~ + infinity
131
(08)
By using Eqs. (D7) and "(08), Eq. (06) can be solv"ed by separation of
variables method.
132
APPENDIX E
SIMPLIFIED EXPRESSIONS FOR AXISYMMETRIC BODIES ANDTWO-DIMENSIONAL AIRFOILS
1. AXISYMMETRIC BODIES
Using the Eqs. (3.3a)-(3.3f) with the specified Sex), one can find
out some leading terms of fnm(x), «(e) and See) as follows:
1.1 Ellipsoidal Body
Sex) = 4x(1 - x)i
246 8«(£) = £ + £ + 2£ + D{e )
246 8B(e) = 1 - e - e - 2e + O(e )
1.2 Dumbbell Shaped Body
133
S(x) = 4bx(1 - x)[l - bx(l - x)] , b ) 2
f10(x) = 4bw U~(2x - 1)[2bx(1 - x) - 1]
3 2 2 432+ b(28x -42x +16x-1)+2x-1] + b x(168x -42Ox +356x -114x+10)
224 3! 2f 21 (x) = 8b • U.[b x(36x - 90x + 76x -24x+2)
3 2+ b(28x -42x +16x-l)+2x-l]
2 2 4 6B(e) =1 - be - b (1. + b)e + O(e )
2. TWO-DIMENSIONAL AIRFOILS
2.1 Elliptic Airfoil
If the airfoil is ~ymmetric and is set at zero angle of attack,
134
then y =0, f n =0 and C(x) =O. Notice that the imaginary parts of the
functions fn(x) all vanish, which simplifies the derivation. With all
these assumptions, one can find out f(x,e) up to 0(e3) as shown below:
f 1(x) = - 2{1 - 2x)
f 2(x) = - 4(1 - 2x)
f 3(x) =- 8(1 - 2x)
B{e) = 1 2 4- e - e -
2.2 General Joukowski Airfoil
With the specified S(x) and C(x), all the operators can be
evaluated, through Eqs. (3.8a)-(3.10c) can be evaluated and the
constants 1m fn(O) can be computed from Eq. (2.71). With this
information, the following three fn(x) are obtained as follows:
fO(x) = i(2 siny)(x - 1)
Re f1(x) = (cosy + Asiny)(x - 1)(1 - 4x)
1m f 1(x) = 2{x - 1)[2(Acosy - siny)x + s1ny]
135
Re f 2(x) = O.S(x - 1){3 (COSy + Asiny)(8x2+ ax + 1)
+ 2A [6(Acosy - siny)x(1 - 2x) + siny(1 - 4x)]}
2O.S(x - 1)[A(cosy + Asiny)(24x - 16x + 1)
+ (Acosy - siny)(24x2 - 20x + 1) + 4siny(2x - 1)]
Notice that fn(x) are complex-valued functions, so each fn(x) is
represented by
, n = 0, 1, 2
Using the above results in Eq. (2.73), with the definition of lift
coeff~c1ent Ct,pt one obtains
Ct,pt =2. [s1ny + O.S(Acosy + s1ny)e + 0.2SA(cosy)e2] + 0(e3)
The leading terms for a(e) is given in Eq. (2.67). Since this airfoil
has a sharp trailing edge, S(e) should be set to one. Thus, by using
the resulting expre~sions, the asymptotic expansion for ,(z;e) is given
by
· 1 2 k 1 In(z-~) fk(~) dl; 3,(z;£) = e-1y z - -- ~ ~ fL'- -----.-.0lIl..-..-. + O( £ )
2. k=O «(e) [(1 _ ~)(~ _ «)]1/2
In particular, if A = 0 or C(x) = 0, the real part of the results are
136
suitable for the symmetrical Joukowski airfoil. For this case, the
first two terms of +(z;e) above agree exactly with Eq. (12.9) of [10],
when his results are expanded up to 0{e3).
The third order lift coefficient for the general Joukowski airfoil
is formulated to compare with the results obtained from the perturbation
analy~ls and the panel methods. Making reference to [15, 22, 23], the
exact ] i. ft coeffi ci ent. ·for the Joukowski ai rfoi 1 is gi yen by
Ct ex= 8. a si~{y~J/t, where y represents the angle of attack, .-,
and! is the chord of the airfoil. The radius a, the angles Band 6 are
connected by the Eq. (2.76). By applying the similar technique in Sec.
2.1.3, a perturbation parameter £1 1s chosen, ~hen
a = C{1-e1cos~)/{1-0.5 e~Sin2~) and t = 4C{1-e1cos~)2/{1-2e1cos~).
By using the resulting expressions and retaining terms up to order··
of e12, the lift coefficient can be.expressed as
Ct,ex =.21rSi,n{Y+S){1 - e1cos~ - e12cos2~ - 0.5 e12sin2~) + 0{e13)
where S =t~n-1{e1sin&/{1-e1cos~» and the lift coefficient for the
Joukowski airfoil can be computed up to 0{e13) if the angle of
attack y, the parameters ~and e1 are specified.