Are you sure?
This action might not be possible to undo. Are you sure you want to continue?
Linear Elastic Waves
Wave propagation and scattering are among the most fundamental processes
that we use to comprehend the world around us. Whereas these processes are
often very complex, one way to begin to understand themis to study wave prop
agation in the linear approximation. This is a book describing such propagation
using, as a context, the equations of elasticity. Two unifying themes are used.
The ﬁrst is that an understanding of plane wave interactions is fundamental to
understanding more complex wave interactions. The second is that waves are
best understood in an asymptotic approximation where they are free of the com
plications of their excitation and are governed primarily by their propagation
environments. The topics covered include reﬂection, refraction, propagation of
interfacial waves, integral representations, radiation and diffraction, and prop
agation in closed and open waveguides.
Linear Elastic Waves is an advanced level textbook directed at applied mathe
maticians, seismologists, and engineers.
John G. Harris received an undergraduate degree (honors) in 1971 fromMcGill
University, where he studied electrical engineering and gained a lasting interest
in wave propagation. In 1979, he received a doctoral degree in applied mathe
matics from Northwestern University for a thesis on elasticwave diffraction
problems. Following this, he joined the Theoretical and Applied Mechan
ics Department at the University of Illinois (UrbanaChampaign), where he
teaches dynamics and mathematical methods and carries out research on elastic
wave problems at microwave frequencies. He has used asymptotic methods of
analysis to explore diffraction and imaging, propagation and scattering of
interfacial waves, and waveguiding.
Cambridge Texts in Applied Mathematics
Maximum and Minimum Principles
M. J. SEWELL
Solitons
P. G. DRAZIN AND R. S. JOHNSON
The Kinematics of Mixing
J. M. OTTINO
Introduction to Numerical Linear Algebra and Optimisation
PHILIPPE G. CIARLET
Integral Equations
DAVID PORTER AND DAVID S. G. STIRLING
Perturbation Methods
E. J. HINCH
The Thermomechanics of Plasticity and Fracture
GERARD A. MAUGIN
Boundary Integral and Singularity Methods for Linearized Viscous Flow
C. POZRIKIDIS
Nonlinear Wave Processes in Acoustics
K. NAUGOLNYKH AND L. OSTROVSKY
Nonlinear Systems
P. G. DRAZIN
Stability, Instability and Chaos
PAUL GLENDINNING
Applied Analysis of the Navier–Stokes Equations
C. R. DOERING AND J. D. GIBBON
Viscous Flow
H. OCKENDON AND J. R. OCKENDON
Scaling, SelfSimilarity, and Intermediate Asymptotics
G. I. BARENBLATT
A First Course in the Numerical Analysis of Differential Equations
ARIEH ISERLES
Complex Variables: Introduction and Applications
MARK J. ABLOWITZ AND ATHANASSIOS S. FOKAS
Mathematical Models in the Applied Sciences
A. C. FOWLER
Thinking About Ordinary Differential Equations
ROBERT E. O’MALLEY
A Modern Introduction to the Mathematical Theory of Water Waves
R. S. JOHNSON
Rareﬁed Gas Dynamics
CARLO CERCIGNANI
Symmetry Methods for Differential Equations
PETER E. HYDON
High Speed Flow
C. J. CHAPMAN
Wave Motion
J. BILLINGHAM AND A. C. KING
An Introduction to Magnetohydrodynamics
P. A. DAVIDSON
Linear Elastic Waves
JOHN G. HARRIS
Linear Elastic Waves
JOHN G. HARRIS
University of Illinois at UrbanaChampaign
iuniisuio n\ rui iiiss s\xoicari oi rui uxiviisir\ oi caxniioci
The Pitt Building, Trumpington Street, Cambridge, United Kingdom
caxniioci uxiviisir\ iiiss
The Edinburgh Building, Cambridge CB2 2RU, UK
40 West 20th Street, New York, NY 100114211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
Ruiz de Alarcón 13, 28014 Madrid, Spain
Dock House, The Waterfront, Cape Town 8001, South Africa
http://www.cambridge.org
First published in printed format
ISBN 0521643686 hardback
ISBN 052164383X paperback
ISBN 0511040415 eBook
Cambridge University Press 2004
2001
(netLibrary)
©
To Beatriz
Contents
Preface page xiii
1 Simple Wave Solutions 1
1.1 Model Equations 1
1.1.1 OneDimensional Models 3
1.1.2 TwoDimensional Models 4
1.1.3 Displacement Potentials 5
1.1.4 Energy Relations 5
1.2 The Fourier and Laplace Transforms 6
1.3 A Wave Is Not a Vibration 11
1.4 Dispersive Propagation 13
1.4.1 An Isolated Interaction 13
1.4.2 Periodic Structures 15
References 18
2 Kinematical Descriptions of Waves 20
2.1 TimeDependent Plane Waves 20
2.2 TimeHarmonic Plane Waves 22
2.3 PlaneWave or AngularSpectrum Representations 24
2.3.1 A Gaussian Beam 24
2.3.2 An AngularSpectrum Representation of a
Spherical Wave 26
2.3.3 An AngularSpectrum Representation of a
Cylindrical Wave 28
2.4 Asymptotic Ray Expansion 28
2.4.1 Compressional Wave 29
2.4.2 Shear Wave 33
ix
x Contents
Appendix: Spherical and Cylindrical Waves 34
References 35
3 Reﬂection, Refraction, and Interfacial Waves 37
3.1 Reﬂection of a Compressional Plane Wave 37
3.1.1 Phase Matching 39
3.1.2 Reﬂection Coefﬁcients 40
3.2 Reﬂection and Refraction 41
3.3 Critical Refraction and Interfacial Waves 44
3.4 The Rayleigh Wave 48
3.4.1 The TimeHarmonic Wave 49
3.4.2 Transient Wave 50
3.4.3 The Rayleigh Function 51
3.4.4 Branch Cuts 52
References 55
4 Green’s Tensor and Integral Representations 56
4.1 Introduction 56
4.2 Reciprocity 57
4.3 Green’s Tensor 58
4.3.1 Notes 61
4.4 Principle of Limiting Absorption 62
4.5 Integral Representation: A Source Problem 64
4.5.1 Notes 65
4.6 Integral Representation: A Scattering Problem 65
4.6.1 Notes 66
4.7 Uniqueness in an Unbounded Region 68
4.7.1 No Edges 68
4.7.2 Edge Conditions 69
4.7.3 An Inner Expansion 71
4.8 Scattering from an Elastic Inclusion in a Fluid 72
References 76
5 Radiation and Diffraction 77
5.1 Antiplane Radiation into a HalfSpace 77
5.1.1 The Transforms 78
5.1.2 Inversion 79
5.2 Buried Harmonic Line of Compression I 82
5.3 Asymptotic Approximation of Integrals 86
Contents xi
5.3.1 Watson’s Lemma 87
5.3.2 Method of Steepest Descents 90
5.3.3 Stationary Phase Approximation 94
5.4 Buried Harmonic Line of Compression II 96
5.4.1 The Complex Plane 96
5.4.2 The Scattered Compressional Wave 98
5.4.3 The Scattered Shear Wave 99
5.5 Diffraction of an Antiplane Shear Wave at an Edge 101
5.5.1 Formulation 102
5.5.2 Wiener–Hopf Solution 104
5.5.3 Description of the Scattered Waveﬁeld 108
5.6 Matched Asymptotic Expansion Study 112
Appendix: The Fresnel Integral 116
References 119
6 Guided Waves and Dispersion 121
6.1 Harmonic Waves in a Closed Waveguide 121
6.1.1 Partial Waves and the Transverse Resonance Principle 123
6.1.2 Dispersion Relation: A Closed Waveguide 124
6.2 Harmonic Waves in an Open Waveguide 128
6.2.1 Partial Wave Analysis 129
6.2.2 Dispersion Relation: An Open Waveguide 131
6.3 Excitation of a Closed Waveguide 134
6.3.1 Harmonic Excitation 134
6.3.2 Transient Excitation 135
6.4 Harmonically Excited Waves in an Open Waveguide 139
6.4.1 The Waveﬁeld in the Layer 140
6.4.2 The Waveﬁeld in the HalfSpace 145
6.4.3 Leaky Waves 146
6.5 A Laterally Inhomogeneous, Closed Waveguide 146
6.6 Dispersion and Group Velocity 150
6.6.1 Causes of Dispersion 150
6.6.2 The Propagation of Information 151
6.6.3 The Propagation of Angular Frequencies 153
References 157
Index 159
Preface
A wave
builds up
perhaps it says its name, I don’t understand,
mutters, humps its load
of movement and foam
and withdraws. Who
can I ask what it said to me?
1
Wave propagation and scattering are among the most fundamental processes
that we use to comprehend the world around us. While these processes are
often very complex, one way to begin to understand them is to study linear
wave propagation. This is a book describing such propagation.
I use the equations of linear elasticity to form a context for my description of
wave propagation. However, the reader’s knowledge of elasticity need not be
very great, and experience with a related ﬁeld theory, such as ﬂuid mechanics or
electromagnetic theory, is sufﬁcient to understand what is written here. In many
places I treat only the antiplane shear problem because I do not believe that the
extra work needed to do the analogous inplane problem adds anything of signi
ﬁcance to understanding the underlying wave processes. Nevertheless, where
an inplane elastic problem introduces a unique feature, such as the presence of
a nondispersive surface wave, that problem is treated.
This is also a book describing the parts of applied mathematics that de
scribe the propagation and scattering of linear elastic waves. It assumes that the
reader has a good background in calculus, differential equations, and complex
analysis. By this I mean that the reader should have studied most of the topics in
1
Neruda, Pablo, Soliloquy in the Waves, pp. 185–186. In Isla Negra, a Notebook, translated by
A. Reid. New York: Noonday Press. 1981.
xiii
xiv Preface
Courant and John, Introduction to Calculus and Analysis, Vols. 1 and 2 (1989)
and in Boyce and DiPrima, Elementary Differential Equations and Boundary
Value Problems (1992). Moreover, the reader should be able to look things
up in Carrier, Krook, and Pearson, Functions of a Complex Variable (1983) or
Ablowitz and Fokas Complex Variables, Introduction and Applications (1997)
and Zauderer Partial Differential Equations of Applied Mathematics (1998)
and not feel lost. None of the mathematical analyses exceed, in sophistica
tion or difﬁculty, those found in Courant and John (1989). I work almost
entirely in Cartesian coordinates so that no knowledge of special functions or
the transforms associated with them is needed. Some previous experience with
asymptotic analysis would be helpful, but is not essential. All the asymptotic
analysis needed is explained in the book.
I have used two unifying themes throughout. The ﬁrst is that an understand
ing of planewave interactions is fundamental to understanding more complex
wave interactions. The second is that waves are best understood in an asymp
totic approximation where they are free of the complications of their excitation
and are governed primarily by their propagation environment. Therefore plane
wave spectral analyses and asymptotic approximations are the main techniques
used to study the more complicated problems.
The selection of problems for the reader is small and directed at engaging
him or her in the development of the subject. The problems are an integral part
of the book and most should be attempted.
I have tried to avoid a menagerie of symbols. In general I use Cartesian ten
sors such as τ
i j
, where the indices i, j = 1, 2, 3, or a boldface notation τ. The
symbol ∂
i
is used to represent the partial derivative with respect to the i th coor
dinate. Similarly, sometimes I use d
x
f to represent d f /dx. Repeated indices
are summed over 1, 2, 3 unless otherwise indicated. For problems engaging
only two coordinates, subscripts using Greek letters such as α, β = 1, 2 are
used so that a vector component would be written as u
β
and a partial deriva
tive as ∂
α
. When these subscripts are repeated they are summed over 1, 2. At
times I use symbols such as c
L
or c
T
when there is need to distinguish between
parameters that relate to compressional or shear disturbances, but when that
distinction is not important I drop the subscript. Constants such as A are used
over and over again and have no special meaning.
Professor Jan Achenbach was my research advisor. He taught me the subject.
His inﬂuence is everywhere is this book. I thank him. I have been supported
over the years primarily by the National Science Foundation, though recently
also by the Air Force Ofﬁce of Scientiﬁc Research. I am very grateful for
their support. Elaine Wilson typed the early parts of this manuscript and Mike
Preface xv
Greenberg provided the illustrations. Don Carlson and Bill Phillips tolerated
my complaints with humor; Eduardo Velasco prevented me from publishing a
discussion of group velocity that was at best muddled. To everyone, thank you.
The book undoubtedly contains errors, for which I alone am responsible. I
anticipate that few if any are serious.
Books Cited
Ablowitz, M. J. and Fokas, A. S. 1997. Complex Variables, Introduction and
Applications. New York: Cambridge.
Boyce, W. E. and DiPrima, R. C. 1992. Elementary Differential Equations and
Boundary Value Problems, 5th ed. New York: Wiley.
Carrier, G. F., Krook, M., and Pearson, C. E. 1983. Functions of a Complex Variable.
Ithaca, NY: Hod Books.
Courant, R. and John, F. 1989. Introduction to Calculus and Analysis, Vols. 1 and 2.
New York: Springer.
Zauderer, E. 1983. Partial Differential Equations of Applied Mathematics. New York:
WileyInterscience.
Evanston and Urbana, Summer 2000.
1
Simple Wave Solutions
Synopsis
Chapter 1 provides the background, both the model equations and some of
the mathematical transformations, needed to understand linear elastic waves.
Only the basic equations are summarized, without derivation. Both Fourier
and Laplace transforms and their inverses are introduced and important sign
conventions settled. The Poisson summation formula is also introduced and
used to distinguish between a propagating wave and a vibration of a bounded
body.
A linear wave carries information at a particular velocity, the group velocity,
which is characteristic of the propagation structure or environment. It is this
transmitting of information that gives linear waves their special importance.
In order to introduce this aspect of wave propagation, propagation in one
dimensional periodic structures is discussed. Such structures are dispersive and
therefore transmit information at a speed different from the wavespeed of their
individual components.
1.1 Model Equations
The equations of linear elasticity consist of (1) the conservation of linear and
angular momentum, and (2) a constitutive relation relating force and deforma
tion. In the linear approximation density ρ is constant. The conservation of
mechanical energy follows from (1) and (2). The most important feature of the
model is that the force exerted across a surface, oriented by the unit normal n
j
,
by one part of a material on the other is given by the traction t
i
=τ
j i
n
j
, where
τ
j i
is the stress tensor. The conservation of angular momentummakes the stress
tensor symmetric; that is, τ
i j
=τ
j i
. The conservation of linear momentum, in
1
2 1 Simple Wave Solutions
differential form, is expressed as
∂
k
τ
ki
+ρf
i
=ρ∂
t
∂
t
u
i
. (1.1)
The term f is a force per unit mass.
Deformation is described by using a strain tensor,
i j
=(∂
i
u
j
+∂
j
u
i
)/2, (1.2)
where u
i
is ith component of particle displacement. The symmetrical deﬁnition
of the deformation ensures that no rigidbody rotations are included. However,
the underlying dependence of the deformation is upon the ∇u. For a homoge
neous, isotropic, linearly elastic solid, stress and strain are related by
τ
i j
=λ
kk
δ
i j
+2µ
i j
, (1.3)
where λ and µare Lam´ e’s elastic constants. Substituting (1.2) in (1.3), followed
by substituting the outcome into (1.1), gives one formof the equation of motion,
namely
(λ +µ)∂
i
∂
k
u
k
+µ∂
j
∂
j
u
i
+ρf
i
=ρ∂
t
∂
t
u
i
. (1.4)
Written in vector notation, the equation becomes
(λ +µ)∇(∇ · u) +µ∇
2
u+ρ f =ρ∂
t
∂
t
u. (1.5)
When the identity ∇
2
u =∇(∇ · u) −∇ ∧∇ ∧u is used, the equation can also
be written in the form
(λ +2µ)∇(∇ · u) −µ∇ ∧∇ ∧u+ρ f =ρ∂
t
∂
t
u. (1.6)
This last equation indicates that elastic waves have both dilitational ∇ · u and
rotational ∇ ∧u deformations.
If ∂R is the boundary of a region R occupied by a solid, then commonly
t and u are prescribed on ∂R. When t is given over part of ∂R and u over
another part, the boundary conditions are said to be mixed. One very common
boundary condition is to ask that t =0 everywhere on ∂R. This models the
case in which a solid surface is adjacent to a gas of such small density and
compressibility that it is almost a vacuum. When R is inﬁnite in one or more
dimensions, special conditions are imposed such that a disturbance decays to
zero at inﬁnity or radiates outward toward inﬁnity from any sources contained
within R.
1.1 Model Equations 3
Another common situation is that in which ∂R
12
is the boundary between
two regions, 1 and 2, occupied by solids having different properties. Contact
between solid bodies is quite complicated, but in many cases it is usual to
assume that the traction and displacement, t and u, are continuous. This models
welded contact. One other simple continuity condition that commonly arises
is that between a solid and an ideal ﬂuid. Because the viscosity is ignored, the
tangential component of t is set to zero, while the normal component of traction
and the normal component of displacement are made continuous.
These are only models and are often inadequate. To brieﬂy indicate some
of the possible complications, consider two solid bodies pressed together. A
(linear) wave incident on such a boundary would experience continuity of trac
tion and displacement when the solids press together, but would experience a
tractionfree boundary condition when they pull apart (Comninou and Dundurs,
1977). This produces a complex nonlinear interaction.
The reader may consult Hudson (1980) for a succinct discussion of linear
elasticity or Atkin and Fox (1980) for a somewhat more general view.
1.1.1 OneDimensional Models
We assume that the various waveﬁeld quantities depend only on the variables
x
1
and t . For longitudinal strain, u
1
is ﬁnite, while u
2
and u
3
are assumed to be
zero, so that (1.2) combined with (1.3) becomes
τ
11
=(λ +2µ)∂
1
u
1
, τ
22
=τ
33
=λ∂
1
u
1
, (1.7)
and (1.1) becomes
(λ +2µ)∂
1
∂
1
u
1
+ρf
1
=ρ∂
t
∂
t
u
1
. (1.8)
For longitudinal stress, all the stress components except τ
11
are assumed to be
zero. Now (1.3) becomes
τ
11
= E∂
1
u
1
, E =µ
3λ +2µ
λ +µ
, (1.9)
and
∂
2
u
2
=∂
3
u
3
=−ν∂
1
u
1
, ν =
λ
2(λ +µ)
. (1.10)
Now (1.1) becomes
E∂
1
∂
1
u
1
+ρf
1
=ρ∂
t
∂
t
u
1
. (1.11)
4 1 Simple Wave Solutions
Note that (1.8) and (1.11) are essentially the same, though they have somewhat
different physical meanings. The longitudinal stress model is useful for rods
having a small cross section and a tractionfree surface. Stress components that
vanish at the surface are assumed to be negligible in the interior.
1.1.2 TwoDimensional Models
Let us assume that the various waveﬁeld quantities are independent of x
3
. As a
consequence, (1.1) breaks into two separate equations, namely
∂
β
τ
β3
+ρf
3
= ρ∂
t
∂
t
u
3
, (1.12)
∂
β
τ
βα
+ρf
α
= ρ∂
t
∂
t
u
α
. (1.13)
Greek subscripts α, β =1, 2 are used to indicate that the independent spatial
variables are x
1
and x
2
. The case for which the only nonzero displacement
component is u
3
(x
1
, x
2
, t ), namely (1.12), is called antiplane shear motion, or
SH motion for shear horizontal.
τ
3β
=µ∂
β
u
3
, (1.14)
giving, from (1.12),
µ∂
β
∂
β
u
3
+ρf
3
=ρ∂
t
∂
t
u
3
. (1.15)
Note that this is a twodimensional scalar equation, similar to (1.8) or (1.11).
The case for which u
3
=0, while the other two displacement components
are generally nonzero, (1.13), is called inplane motion. The initials P and SV
are used to describe the two types of inplane motion, namely compressional
and shear vertical, respectively. For this case (1.3) becomes
τ
αβ
=λ∂
γ
u
γ
δ
αβ
+µ(∂
α
u
β
+∂
β
u
α
), (1.16)
and
τ
33
=λ∂
γ
u
γ
. (1.17)
The equation of motion remains (1.4); that is,
(λ +µ)∂
α
∂
β
u
β
+µ∂
β
∂
β
u
α
+ρf
α
=ρ∂
t
∂
t
u
α
. (1.18)
This last equation is a vector equation and contains two wave types, compres
sional and shear, whose character we explore shortly. It leads to problems of
some complexity.
1.1 Model Equations 5
These twodimensional equations are the principal models used. The scalar
model, (1.14), allows us to solve complicated problems in detail without being
overwhelmed by the size and length of the calculations needed, while the vector
model, (1.18), allows us enough structure to indicate the complexity found in
elasticwave propagation.
1.1.3 Displacement Potentials
When (1.4)–(1.6) are used, a boundary condition, such as t =0, is relatively
easy to implement. However, in problems in which there are fewboundary con
ditions, it is often easier to cast the equations of motion into a simpler form and
allow the boundary condition to become more complicated. One way to do this
is to use the Helmholtz theorem (Phillips, 1933; Gregory, 1996) to express the
particle displacement u as the sumof a scalar ϕ and a vector potential ψ; that is,
u =∇ϕ +∇ ∧ψ, ∇ · ψ =0. (1.19)
The second condition is needed because u has only three components (the parti
cular condition selected is not the only possibility). Assume f =0. Substituting
these expressions into (1.6) gives
(λ +2µ)∇
∇
2
ϕ −
1/c
2
L
∂
t
∂
t
ϕ
+µ∇ ∧
∇
2
ψ−
1/c
2
T
∂
t
∂
t
ψ
=0.
(1.20)
The equation can be satisﬁed if
∇
2
ϕ =
1/c
2
L
∂
t
∂
t
ϕ, c
2
L
=(λ +2µ)/ρ, (1.21)
∇
2
ψ =
1/c
2
T
∂
t
∂
t
ψ, c
2
T
=µ/ρ. (1.22)
The terms c
L
and c
T
are the compressional or longitudinal wavespeed, and shear
or transverse wavespeed, respectively. The scalar potential describes a wave of
compressional motion, which in the planewave case is longitudinal, while the
vector potential describes a wave of shear motion, which in the planewave
case is transverse. Knowing ϕ and ψ, do we know u completely? Yes we do.
Proofs of completeness, along with related references, are given in Achenbach
(1973).
1.1.4 Energy Relations
The remaining conservation law of importance is the conservation of mecha
nical energy. Again assume f =0. This law can be derived directly from
6 1 Simple Wave Solutions
(1.1)–(1.3) by taking the dot product of ∂
t
u with (1.1). This gives, initially,
∂
j
τ
j i
∂
t
u
i
−ρ(∂
t
∂
t
u
i
)∂
t
u
i
=0. (1.23)
Forming the product τ
kl
kl
, using (1.3), and making use of the decomposition
∂
j
u
i
=
j i
+ω
j i
, where ω
j i
=(∂
j
u
i
−∂
i
u
j
)/2, allows us to write (1.23) as
1
2
∂
t
(ρ∂
t
u
i
∂
t
u
i
+τ
ki
ki
) +∂
k
(−τ
ki
∂
t
u
i
) =0. (1.24)
The ﬁrst two terms become the time rates of change of
K =
1
2
ρ∂
t
u
k
∂
t
u
k
, U =
1
2
τ
i j
i j
. (1.25)
These are the kinetic and internal energy density, respectively. The remaining
term is the divergence of the energy ﬂux, F, where F is given by
F
j
=−τ
j i
∂
t
u
i
. (1.26)
Then (1.24) can be written as
∂E/∂t +∇ · F=0, (1.27)
where E =K + U and is the energy density. This is the differential statement
of the conservation of mechanical energy. To better understand that the energy
ﬂux or power density is given by (1.26), consider an arbitrary region R, with
surface ∂R. Integrating (1.27) over Rand using Gauss’ theorem gives
d
dt
R
E(x, t ) dV =−
∂R
F· ˆ n dS. (1.28)
Therefore, as the mechanical energy decreases within R, it radiates outward
across the surface ∂Rat a rate F · ˆ n.
1.2 The Fourier and Laplace Transforms
All waves are transient in time. One useful representation of a transient wave
form is its Fourier one. This representation imagines the transient signal de
composed into an inﬁnite number of timeharmonic or frequency components.
One important reason for the usefulness of this representation is that the trans
mitter, receiver, and the propagation structure usually respond differently to the
different frequency components. The linearity of the problem ensures that we
can work out the net propagation outcomes for each frequency component and
then combine the outcomes to recreate the received signal.
1.2 The Fourier and Laplace Transforms 7
The Fourier transform is deﬁned as
¯ u(x, ω) =
∞
−∞
e
i ωt
u(x, t ) dt. (1.29)
The variable ω is complex. Its domain is such as to make the above integral
convergent. Moreover, ¯ u is an analytic function within the domain of conver
gence, and once known, can be analytically continued beyond it.
1
The inverse
transform is deﬁned as
u(x, t ) =
1
2π
∞
−∞
e
−i ωt
¯ u(x, ω) dω. (1.30)
Thus we have represented u as a sum of harmonic waves e
−i ωt
¯ u(x, ω). Note
that there is a speciﬁc sign convention for the exponential term that we shall
adhere to throughout the book.
A closely related transform is the Laplace one. It is usually used for initial
value problems so that we imagine that for t <0, u(x, t ) is zero. This is not es
sential and its deﬁnition can be extended to include functions whose
t dependence extends to negative values. This transform is deﬁned as
¯ u(x, p) =
∞
0
e
−pt
u(x, t ) dt. (1.31)
As with ω, p is a complex variable and its domain is such as to make ¯ u(x, p)
an analytic function of p. The domain is initially deﬁned as ( p) >0, but
the function can be analytically continued beyond this. Note that p =−i ω so
that, when t ∈ [0, ∞), (ω) >0 gives the initial domain of analyticity for
¯ u(x, ω). The inverse transform is given by
u(x, t ) =
1
2πi
+i ∞
−i ∞
e
pt
¯ u(x, p) dp, (1.32)
where ≥0. The expressions for the inverse transforms, (1.30) and (1.32), are
misleading. In practice we deﬁne the inverse transforms on contours that are
designed to capture the physical features of the solution. A large part of this
book will deal with just how those contours are selected. But, for the present,
we shall work with these deﬁnitions.
1
Analytic functions deﬁned by contour integrals, including the case in which the contour extends
to inﬁnity, are discussed in Titchmarsh (1939) in a general way and in more detail by Noble
(1988).
8 1 Simple Wave Solutions
Consider the case of longitudinal strain. Imagine that at t =−∞a disturbance
started with zero amplitude. Taking the Fourier transform of (1.8) gives
d
2
¯ u
1
/dx
2
1
+k
2
¯ u
1
=0, (1.33)
where k =ω/c
L
and c
L
is the compressional wavespeed deﬁned in (1.21). The
parameter k is called the wavenumber. Here (1.33) has solutions of the form
¯ u
1
(x
1
, ω) = A(ω)e
±i kx
1
. (1.34)
If we had sought a timeharmonic solution of the form
u
1
(x
1
, t ) = ¯ u
1
(x
1
, ω)e
−i ωt
, (1.35)
we should have gotten the same answer except that e
−i ωt
would be present. In
other words, taking the Fourier transform of an equation over time or seeking
solutions that are time harmonic are two slightly different ways of doing the
same operation.
For (1.35), it is understood that the real part can always be taken to obtain
a real disturbance. Much the same happens in using (1.30). In writing (1.30)
we implicitly assumed that u(x, t ) was real. That being the case, ¯ u(x, ω) =
¯ u
∗
(x, −ω), where the superscript asterisk to the right of the symbol indicates
the complex conjugate. From this it follows that
u(x, t ) =
1
π
∞
0
e
−i ωt
¯ u(x, ω) dω. (1.36)
The advantage of this formulation of the inverse transform is that we may
proceed with all our calculations by using an implied e
−i ωt
and assuming ω is
positive. The importance of this will become apparent in subsequent chapters.
Now (1.36) can be regarded as a generalization of the taking of the real part of
a timeharmonic wave (1.35).
Problem 1.1 Transform Properties
Check that ¯ u(x, ω) = ¯ u
∗
(x, −ω) and derive (1.36) from (1.30). The reader
may want to consult a book on the Fourier integral such as that by Papoulis
(1962).
When the plus sign is taken, (1.35) is a timeharmonic, plane wave propagat
ing in the positive x
1
direction. We assume that the wavenumber k is positive,
unless otherwise stated. The wave propagates in the positive x
1
direction be
cause the term (kx
1
−ωt ) remains constant, and hence u
1
remains constant,
1.2 The Fourier and Laplace Transforms 9
only if x
1
increases as t increases. The speed with which the wave propagates
is c
L
. The term ω is the angular frequency or 2π f , where f is the frequency.
That is, at a ﬁxed position, 1/f is the length of a temporal oscillation. Similarly,
k, the wavenumber, is 2π/λ, where λ, the wavelength, is the length of a spatial
oscillation.
Note that if we combine two of these waves, labeled u
+
1
and u
−
1
, each going
in opposite directions, namely
u
+
1
= Ae
i (kx
1
−ωt )
, u
−
1
= Ae
−i (kx
1
+ωt )
, (1.37)
we get
u
1
= Ae
−i ωt
2 cos(kx
1
). (1.38)
Taking the real part gives
u
1
=2 A cos(ωt +α) cos(kx
1
). (1.39)
We have taken A as complex so that α is its argument. This disturbance does
not propagate. At a ﬁxed x
1
the disturbance simply oscillates in time, and at a
ﬁxed t it oscillates in x
1
. The wave is said to stand or is called a standing wave.
Problem 1.2 Fourier Transform
Continue with the case of longitudinal strain and consider the following
boundary, initialvalue problem. Unlike the previous discussion in which the
disturbance began, with zero amplitude, at −∞, here we shall introduce a distur
bance that starts up at t =0
+
. Consider an elastic halfspace, occupying x
1
≥0,
subjected to a nonzero traction at its surface. The problem is one dimensional,
and it is invariant in the other two so that (1.8), the equation for longitudinal
strain, is the equation of motion. At x
1
=0 we impose the boundary condition
τ
11
=−P
0
e
−ηt
H(t ), where H(t ) is the Heaviside step function and P
0
is a con
stant. As x
1
→∞we impose the condition that any wave propagate outward in
the positive x
1
direction. Why? Moreover, we ask that, for t < 0, u
1
(x
1
, t ) =0
and ∂
t
u
1
(x
1
, t ) =0. Note that, using integration by parts, the Fourier transform,
indicated by F, of the second time derivative is
F[∂
t
∂
t
u
1
] =−ω
2
¯ u
1
(x
1
, ω) +i ωu
1
(x
1
, 0
−
) −∂
t
u
1
(x
1
, 0
−
). (1.40)
In deriving this expression we have integrated fromt =0
−
to ∞so as to include
any discontinuous behavior at t =0. Taking the Fourier transformof (1.8) gives
10 1 Simple Wave Solutions
(1.33). Show that the inverse transform of the stress component τ
11
is given by
τ
11
(x
1
, t ) =
P
0
2πi
∞
−∞
e
i (kx
1
−ωt )
ω+i η
dω. (1.41)
In the course of making this step you will need to chose between the solutions to
the transformed equation, (1.33). Why is the solution leading to (1.41) selected?
Note that, if the disturbance is to decay with time, η must be positive. Next show
that
τ
11
(x
1
, t ) =−P
0
H(t −x
1
/c
L
)e
−η(t −x
1
/c
L
)
. (1.42)
Explain how the conditions for convergence of the integral, as its contour is
closed at inﬁnity, give rise to the Heaviside function. Note howthe sign conven
tions for the transformpair, by affecting where the inverse transformconverges,
give a solution that is causal.
Problem 1.3 Laplace Transform
Solve Problem 1.2 by using the Laplace transform over time. Why select the
solution e
−px
1
/c
L
? How does this relate to the demand that waves be outgoing
at ∞?
The solution of Problem 1.2 suggests how we shall deﬁne the Fourier trans
form over the spatial variable x. Suppose we have taken the temporal transform
obtaining ¯ u(x, ω). Then its Fourier transform over x is deﬁned as
∗
¯ u(k, ω) =
∞
−∞
e
−i kx
u(x, t ) dx, (1.43)
and its inverse transform is
¯ u(x, ω) =
1
2π
∞
−∞
e
i kx ∗
¯ u(k, ω) dk . (1.44)
Again note the sign conventions for the transform pair. This will remain the
convention throughout the book. Moreover, note that
u(x, t ) =
1
4π
2
∞
−∞
∞
−∞
e
i (kx−ωt )∗
¯ u(k, ω) dωdk . (1.45)
This shows that quite arbitrary disturbances can be decomposed into a sum of
timeharmonic, plane waves and thereby indicates that the study of such waves
is very central to the understanding of linear waves.
1.3 A Wave Is Not a Vibration 11
1.3 A Wave Is Not a Vibration
A continuous body vibrates when a system of standing waves is established
within it. Vibration and wave propagation can be explicitly linked by means
of the Poisson summation formula. This formula might better be termed a
transformandis quite useful, especiallyfor asymptoticallyapproximatingsums.
Proposition 1.1. Consider a function f (t ) and its Fourier transform
¯
f (ω).
Restrictions on f (t ) are given below. The Poisson summation formula states
that
∞
m=−∞
f (mλ) =
1
λ
∞
k =−∞
¯
f
2πk
λ
, (1.46)
where λ is a parameter.
This formula relates a series to one comprising its transformed terms. If we want
to approximate the lefthand side of (1.46) for λ that is small, then knowing the
Fourier transform of each term enables us to use an approximation based on the
fact that λ
−1
is large. The lefthand side of (1.46) may converge only slowly
for a small λ.
Proof.
2
This proof follows that of de Bruijn (1970). Consider the function
ϕ(x) given by
ϕ(x) =
∞
m =−∞
f [(m +x)λ], (1.47)
where the sum converges uniformly on x ∈ [0, 1]. The function ϕ(x) has
period 1. We assume that f (t ) is such that ϕ(x) has a Fourier series, ϕ =
∞
k=−∞
c
k
e
i k2πx
. Its kth Fourier coefﬁcient equals
1
0
e
−i k2πx
ϕ(x) dx =
1
0
∞
m =−∞
e
−i k2πx
f [(m +x)λ] dx
=
∞
m =−∞
(m+1)
m
e
−i k2πx
f (xλ) dx
=
1
λ
∞
−∞
e
−i k2π(x/λ)
f (x) dx. (1.48)
2
A minimum amount of analysis is used, both here and elsewhere, and no attempt at rigorous
proofs is made. The conclusions are usually valid under more general conditions than those
cited.
12 1 Simple Wave Solutions
Note that the integral
(m+t )
m
e
−i kx
f (xλ)dx →0 as m →±∞, uniformly in
x ∈ [0, 1], as the sum (1.47) converges uniformly.
These conditions are more restrictive than needed. Lighthill (1978), among
others, indicates that the Poisson summation formula holds for a far more
general class of functions than assumed here.
Consider the impulsive excitation of a rod of ﬁnite length 1. The governing
equation is (1.11). Assume f
1
=0, set c
b
and ρ =1 (c
2
b
= E/ρ), and assume
that for t < 0, u
1
(x
1
, t ) =∂
t
u
1
(x
1
, t ) =0. The boundary conditions are
τ
11
(0, t ) =−Tδ(t ), τ
11
(1, t ) =0. (1.49)
By using a Fourier transform over t and solving the boundaryvalue problem
in x
1
, we get
3
τ
11
=iTH(t )
∞
n =−∞
e
−i nπt
sin[nπ(1 −x
1
)]
cos(nπ)
. (1.50)
Thus the rod is ﬁlled with standing waves. One usually considers a solution in
this form as a vibration. This is a very useful way to express the answer, assum
ing the pulse has reverberated within the rod for a time long with respect to that
needed for one echo fromthe far end to return to x =0. But the individual inter
actions with the ends have been obscured. To ﬁnd these interactions we apply
the Poisson summation formula to (1.50). Break up the sin[nπ(1 −x
1
)] term
into exponential ones and apply (1.46) to each term. The crucial intermediate
step is the following, where we have taken one of these terms.
∞
m =−∞
1
cos mπ
e
−i mπ[t −(1−x
1
)]
= (πt +x
1
−2)
−1
×
∞
n =−∞
∞
−∞
e
−i ω
1−
2n
t +x
1
−2
dω
= 2
∞
n =−∞
δ(t +x
1
−2 −2n). (1.51)
The outcome to our calculation is
τ
11
= T
∞
k=1
δ(t +x
1
−2k) −T
∞
k=0
δ(t −x
1
−2k). (1.52)
3
The reader should check that this is the solution.
1.4 Dispersive Propagation 13
This is a wave representation. It is very useful for times of the order of the echo
time. For example, if t ∈ (1, 2), then τ
11
= Tδ[(t −1) +(x
1
−1)]. We have thus
isolated the pulse returning from its ﬁrst reﬂection at the end x
1
=1. However,
the representation is not very useful for t that is large because it becomes tedious
to work out exactly at what reﬂection you are tracking the pulse. Moreover, the
representation (1.52) would have been awkward to work with if, instead of delta
function pulses, we had had pulses of sufﬁcient length that they overlapped one
another. Nevertheless, the representation captures quite accurately the physical
phenomena of a pulse bouncing back and forth in a rod struck impulsively at
one end.
Avibration therefore is deﬁned and conﬁned by its environment. It is the out
come of waves reverberating in a bounded body. A period of time, sometimes a
long one, is needed for the environment to settle into a steady oscillatory motion.
In contrast, a wave is a disturbance that propagates freely outward, returning
to its source perhaps only once, experiencing only a ﬁnite number of interac
tions. Understanding howan individual wave interacts with its environment and
tracking it through each of its interactions constitute the principal problems of
wave propagation. Moreover, while one works frequently with timeharmonic
propagating waves, one usually assumes that at some stage a Fourier synthesis
will be carried out, mapping the unending oscillatory motion into a disturbance
conﬁned both temporally and spatially.
1.4 Dispersive Propagation
1.4.1 An Isolated Interaction
A basic interaction of a wave with its environment is scattering from a discon
tinuity. Continue to consider waves in a rod by using the longitudinal stress
approximation. Consider timeharmonic disturbances of the form
u
1
= ¯ u
1
(x
1
)e
−i ωt
, τ
11
=ρc
2
b
∂
1
u
1
. (1.53)
We shall not indicate the possible dependence on ω of ¯ u
1
unless this is needed.
The equation of motion (1.11) becomes
d
2
¯ u
1
/dx
2
1
+k
2
¯ u
1
=0, k =ω/c
b
. (1.54)
Assume there is a region of inhomogeneity within x
1
∈ (−L, L). Incident on
this inhomogeneity is the waveﬁeld
¯ u
i
1
(x
1
) =
A
1
e
i kx
1
, x
1
<−L,
A
2
e
−i kx
1
, x
1
> L,
(1.55)
14 1 Simple Wave Solutions
where we have allowed waves to be incident fromboth directions. Upon striking
the inhomogeneity, the scattered waveﬁeld
¯ u
s
1
(x
1
) =
B
1
e
−i kx
1
, x
1
<−L,
B
2
e
i kx
1
, x
1
> L
(1.56)
is excited. Note that the scattered waves have been constructed so that they are
outgoing from the scatterer. The linearity of the problem suggests that we can
write the scattered amplitudes in terms of the incident ones as
B
1
B
2
=
S
11
S
12
S
21
S
22
A
1
A
2
, (1.57)
or, more compactly, as
B =SA. (1.58)
The matrix S is called a scattering matrix and characterizes the inhomogeneity.
We next consider a speciﬁc example. Imagine that the rod has a cross
sectional area 1 and that the inhomogeneity is a point mass M, at x
1
=0.
The lefthand ﬁgure in Fig. 1.1 indicates the geometry. The conditions across
the discontinuity are
u
1
(0
−
, t ) =u
1
(0
+
, t ), M∂
t
∂
t
u
1
=−τ
11
(0
−
, t ) +τ
11
(0
+
, t ). (1.59)
That is, the rod does not break, but the acceleration of the mass causes the
traction acting on the cross section to be discontinuous. Setting tan θ =kM/2ρ,
with θ ∈ (0, π/2), we calculate the matrix S to be
S =
sin θe
i (θ+π/2)
cos θe
i θ
cos θe
i θ
sin θe
i (θ+π/2)
. (1.60)
x
1
x
1
0
0 L 2L
1 2
Fig. 1.1. One or more point masses M are embedded in a rod of crosssectional area
1. The lefthand ﬁgure shows a single, embedded, point mass. The righthand ﬁgure
shows an endless periodic arrangement of embedded point masses, each separated by a
distance L. The masses are labeled 0, 1, 2, . . . , with the zeroth mass at x
1
=0.
1.4 Dispersive Propagation 15
It is also of interest to relate the wave amplitudes on the right to those on
the left. This matrix T, called the transmission matrix, gives R =TL, where
L =[A
1
, B
1
]
T
and R =[B
2
, A
2
]
T
. The matrix T is readily calculated from S
and is given by
T =
1 +i tan θ i tan θ
−i tan θ 1 −i tan θ
. (1.61)
Note that the amplitudes A
1
and B
2
are those of rightpropagating waves, while
B
1
and A
2
are those of leftpropagating ones.
Problem 1.4 Scattering From a Layer
Imagine a layer of width d and properties ( ¯ ρ, ¯ c
L
) embedded in a material
with properties (ρ, c
L
). Let the x
1
axis be perpendicular to the face of the
layer and place the origin at the left face, so that the layer occupies x
1
∈ (0, d).
A wave of longitudinal strain Ae
i kx
1
is incident from x
1
< 0. Calculate the
reﬂection coefﬁcient B/A and the transmission coefﬁcient C/A, where the
reﬂected wave is Be
−i kx
1
and the transmitted one is Ce
i k(x
1
−d)
. Treat the waves
in the layer as a sum of a single right and left propagating wave, each with
different amplitudes, and imagine that the amplitudes contain the effects of all
the multiple reﬂections from each side of the layer. Similarly B and C contain
the effects of all the reﬂections and transmissions from the layer into the outer
material. The wavenumber k =ω/c
L
. For what values of d is the reﬂection
minimized? Can you identify any of the components of the S matrix for the
layer in terms of these coefﬁcients?
1.4.2 Periodic Structures
One of the more interesting aspects of wavebearing structures is that they often
contain several length scales. Propagation in such a structure often can only take
place if the angular frequency ω is linked to the wavenumber – a term we must
deﬁne a bit more carefully here – in a nonlinear way. To consider this possibility
we use the matrix T, (1.61), to analyze propagation in a periodic structure. We
imagine an inﬁnite rod, crosssectional area 1, in which equal point masses,
M, are periodically embedded. The righthand ﬁgure of Fig. 1.1 indicates the
geometry and the how the masses are labeled. One such mass has the nominal
position x
1
=0 and is labeled n =0. Each mass is separated from its neighbors
by a distance L. A cell of length L is thereby formed and is labeled n if the
nth mass occupies its left end. There are thus two length scales, the wavelength
16 1 Simple Wave Solutions
λ =2π/k and the cell length L. In this problem we do not concern ourselves
with howthe waves are excited, but only with the simpler question, What waves
does this structure support? Consider the zeroth cell, where x
1
∈ (0, L). Within
that cell the solution to (1.54) is
¯ u
1
(x
1
) = R
0
e
i kx
1
+L
0
e
−i kx
1
. (1.62)
At x
1
= L
−
the waveﬁeld is [R
0
e
i kL
, L
0
e
−i kL
]
T
. This can be written as L
1
=
PR
0
, with R
0
=[R
0
, L
0
]
T
. The matrix P is called the propagator or the prop
agation matrix and is given by
P =
e
i kL
0
0 e
−i kL
. (1.63)
At x
1
= L
+
, within the 1th cell, the waveﬁeld amplitudes R
1
=[R
1
, L
1
]
T
are
R
1
=TPR
0
. (1.64)
This relation is readily generalized. If R
n
=[R
n
, L
n
]
T
, then
R
n+1
=TPR
n
. (1.65)
The central feature of the propagation structure is that it has translational
symmetry. The central feature of the disturbance we seek is that its phase
changes from cell to cell in a way that represents propagation. Speciﬁcally con
sider propagation to the right. To capture these two features, the waveﬁeld at
a point within the (n +1)th cell can differ from that at a point within the nth
cell, where the two points are separated by a distance L, by at most a multi
plicative phase factor. This kinematic constraint is expressed by the relation
R
n+1
=e
i κL
R
n
, (1.66)
where κ is unknown. κ is positive, if real, and such as to cause decay, if complex.
Combining (1.65) and (1.66) gives a 2 ×2 system of algebraic equations that
has a nontrivial solution if and only if
det(TP−e
i κL
I) =0. (1.67)
Recalling our previous deﬁnition of tan θ =kM/2ρ, we can write this equation
compactly as
cos κL =cos(kL +θ)/cos θ. (1.68)
1.4 Dispersive Propagation 17
This is a nonlinear relationship between the angular frequency ω =c
b
k and the
effective wavenumber κ, though it may not be apparent, as yet, that κ (and not
k) is the wavenumber of interest.
Note that if κL ∈ [−π, π] is a solution, then κL ±2nπ, for n =1, 2, . . . is
also a solution. Accordingly, we need only consider κL ∈ [−π, π]. The term
κL is real provided  cos κL ≤1. Therefore the boundaries between real and
complex κL are given by
cos(kL +θ)/ cos θ =±1. (1.69)
When +1 is taken, the solutions are sin(kL/2) =0 or tan(kL/2) =−tan(θ).
That is, kL =2nπ or kL +2θ =2mπ, where n and m are integers. When
−1 is taken, the solutions are cos(kL/2) =0 or cot(kL/2) =tan(θ). That is,
kL =(2n −1)π or kL +2θ =(2m −1)π, where, again, n and m are inte
gers. All these cases are covered by kL =nπ or kL +2θ =mπ. For kL ∈
[(n −1)π, (nπ −2θ)], κL is real. These intervals are called passbands. Else
where κL is complex, causing the disturbance to decay as it pro
pagates, and the intervals are called stopbands. At the lower boundary of a
passband, L is an integer number of halfwavelengths. If T were real, then
all the reﬂected waves add constructively and little or nothing is transmit
ted. The actual situation is complicated by the complex T, but the con
structive interference of the reﬂected waves is the basic physical mecha
nism giving rise to the stopbands. This phenomenon is referred to as Bragg
scattering.
Consider the interval x
1
∈ (nL, (n +1)L). Then
¯ u
1
(x
1
) = R
n
e
i k(x
1
−nL)
+L
n
e
−i k(x
1
−nL)
= e
i κL
R
n−1
e
i k(x
1
−nL)
+L
n−1
e
−i k(x
1
−nL)
= e
i κL
¯ u
1
(x
1
−L) (1.70)
This equation is a restatement of (1.66). Further, it indicates that ¯ u
1
(x
1
) must
satisfy the functional equation ¯ u
1
(x
1
+L) =e
i κL
¯ u
1
(x
1
) if the kinematic con
straint is to be enforced. Within each cell there are nominally two waves, as
indicated in (1.70), that we call partial waves. However, we seek a solution for
the wave globally propagating to the right along the structure, as distinguished
from the right and leftpropagating partial waves in each cell. With this in
mind, the solution to the functional equation is
¯ u
1
(x
1
) =e
i κx
1
ϕ(x
1
), (1.71)
18 1 Simple Wave Solutions
where ϕ(x
1
+L) =ϕ(x
1
)
4
. That is, ϕ(x
1
) is a periodic function and can be
represented by a Fourier series, whose coefﬁcients are c
n
. Therefore, ¯ u
1
(x
1
)
becomes
¯ u
1
(x
1
) =
∞
−∞
c
n
e
i x
1
(κ−2πn/L)
. (1.72)
The timeharmonic waveﬁeld ¯ u
1
(x
1
)e
−i ωt
is thus a consequence of an inﬁnite
number of space harmonics. Note that shifting κL by ±2mπ would not change
this expression.
More importantly, it is clear that it is κ, through the term e
i (κx
1
−ωt )
, that is
the wavenumber. Equation (1.68) indicates that ω is a function of κ, or κ a
function of ω. A relation such as this is called a dispersion relation. Writing κ
as ω/c(ω), we see that ¯ u
1
(x
1
, ω) propagates at a different speed for each ω. If
we excited the structure with a pulse, then the pulse would be composed of an
inﬁnite number of such components, as indicated by (1.36). Each component
would then propagate at its own speed and the pulse would become dispersed. A
pulse is information, whereas a sinusoid is not. Hence what we have inferred is
that dispersion can cause the distortion of or loss of information from a signal.
We shall explore this topic further in Chapter 6.
There are many fascinating aspects to propagation in periodic structures. The
discussion here has followed parts of Levine (1978), and a reader seeking to
learn more may wish to read this work further.
References
Achenbach, J.D. 1973. Wave Propagation in Elastic Solids. Amsterdam:
NorthHolland.
Atkin, R.J. and Fox, N. 1980. An Introduction to the Theory of Elasticity. London:
Longman.
de Bruijn, N.G. 1970. Asymptotic Methods in Analysis, 3rd ed., pp. 52–56.
Amsterdam: NorthHolland.
Comninou, M. and Dundurs, J. 1977. Reﬂexion and refraction of elastic waves in the
presence of separation. Proc. R. Soc. Lond., A, 356: 509–528.
Friedman, B. 1956. Principles and Techniques of Applied Mathematics, pp. 65–67.
New York: Wiley.
Gregory, R.D. 1996. Helmholtz’s theorem when the domain is infinite and when the
ﬁeld has singular points. Quart. J. Mech. Appl. Math. 49:439–450.
Hudson, J.A. 1980. The Excitation and Propagation of Elastic Waves. Cambridge:
University Press.
Levine, H. 1978. Unidirectional Wave Motions, pp. 273–308, 339–345, and elsewhere.
Amsterdam: NorthHolland.
4
This is a partial statement of Floquet’s theorem (Friedman, 1956).
References 19
Lighthill, M.J. 1978. Fourier Analysis and Generalized Functions, pp. 67–71.
Cambridge: University Press.
Noble, B. 1988. Methods Based on the WienerHopf Technique, pp. 11–27. New York:
Chelsea.
Papoulis, A. 1962. The Fourier Integral and its Applications. New York: McGrawHill.
Phillips, H.B. 1933. Vector Analysis, pp. 182–196. New York: Wiley.
Titchmarsh, E.C. 1939. The Theory of Functions, 2nd ed., pp. 85–86. Oxford:
Clarendon Press.
2
Kinematical Descriptions of Waves
Synopsis
A wave equation is an equation of motion, so that a study of its solutions is
perhaps more than just the study of the kinematics. However, when one writes
little about how a waveﬁeld is excited and concentrates primarily upon the
geometrical descriptionof a waveﬁeld’s propagation, it is reasonable tolabel this
as such. The present chapter describes the kinematics of waves in a propagation
environment without boundaries. Descriptions of a plane wave are emphasized
because they forma canonical kinematical object fromwhich more complicated
solutions can be constructed.
2.1 TimeDependent Plane Waves
A timedependent plane wave is one whose form is
u(x, t ) =
ˆ
d u(t −s · x), (2.1)
where
ˆ
d is a constant unit vector
1
called the polarization and s is a constant
vector called the slowness. The time is t and the position vector is x. The
equation s · x = t −C, where C is a constant, is a parametric representation of
a family of planes, where each value of parameter t gives a different member of
the family. The normal to the plane is s. Figure 2.1 sketches this relationship.
As t is incremented, then x must be incremented along s if the argument
t −s · x is to equal the constant C. Accordingly, u retains a constant value over
each plane deﬁned by t provided this plane advances along the normal s as t
increases. The wave’s name then arises from this fact and the plane is called a
wave front. Setting s = s ˆ p so that ˆ p is the unit normal to each plane, it is then
1
The circumﬂex indicates a unit vector.
20
2.1 TimeDependent Plane Waves 21
x
2
x
3
x
1
x
s
Fig. 2.1. Asketch of the plane wavefront s · x = t −C, where t is a parameter identify
ing a particular plane. The position vector x identiﬁes a point on the plane; the slowness
vector s is normal to the plane and hence gives its orientation.
straightforward to ﬁnd that
ˆ p · dx/dt = 1/s, (2.2)
so that s
−1
is the speed of advance of the plane along ˆ p.
Substituting (2.1) into the equation of motion, (1.6), gives
c
2
L
s(s · ∂
t
∂
t
u) −c
2
T
s ∧ s ∧ ∂
t
∂
t
u = ∂
t
∂
t
u. (2.3)
Because, for any vector A, s(s · A) −s ∧s ∧ A = s
2
A, (2.3) implies that either
s
2
= 1/c
2
L
, s ∧ u = 0, (2.4)
or
s
2
= 1/c
2
T
s · u = 0. (2.5)
Thus either
u =
ˆ
du(t − ˆ p · x/c
L
), ˆ p ∧
ˆ
d = 0, (2.6)
or
u =
ˆ
du(t − ˆ p · x/c
T
), ˆ p ·
ˆ
d = 0. (2.7)
22 2 Kinematical Descriptions of Waves
The ﬁrst is a longitudinal wave with its polarization parallel to its direction
of propagation, while the second is a transverse wave with its polarization
perpendicular to this direction.
The ﬂux of energy Fis given by (1.26). Calculating Ffor (2.6) or (2.7) gives
F
I
= ρc
I
(∂
t
u)
2
ˆ p, (2.8)
where I = L or T and F
I
is the corresponding ﬂux. The energy density is
E = K+U, where K, the kinetic energy, and U, the internal energy, are given
by (1.25). For these two plane waves the velocity of energy transport, C, is
C
I
= F
I
/E
I
= c
I
ˆ p, (2.9)
where I = L or T, and F
I
and E
I
are the corresponding ﬂux and energy density.
2.2 TimeHarmonic Plane Waves
In one sense, timeharmonic plane waves are only a particular case of a general
plane wave, but because we are working with the temporal transform, we have
the additional ﬂexibility of allowing ˆ p to be complex. This has some interesting
implications for the form of the corresponding timedependent plane wave that
we shall explore in Section 3.3. For the present, assume that the time depen
dence is e
−i ωt
or that the waveﬁeld has been transformed over time. The e
−i ωt
is not given, unless explicitly needed, and ω is assumed real and positive, for
the present.
A timeharmonic plane wave has the form
2
u = A
ˆ
de
i k ˆ p·x
, (2.10)
where the amplitude A =  Ae
i θ
and the wavenumber k = ω/c. The combi
nation k ˆ p = ωs is the wavevector. To recover a real, timeharmonic wave,
multiply by e
−i ωt
and take the real part of the expression.
As indicated previously, we may take ˆ p = p
r
+ i p
i
, with p
r
and p
i
both
real. Because ˆ p · ˆ p = 1, p
r
· p
i
= 0. The real and imaginary components are
perpendicular to one another. If p
i
= 0, the plane wave is an inhomogeneous or
evanescent one, while if p
i
= 0, it is homogeneous. Writing(2.10) indetail gives
u = A
ˆ
de
−kx· p
i
e
i kx· p
r
. (2.11)
2
In the previous chapter we distinguished a timeharmonic wave from a timedependent one by
using an overbar. The overbar is no longer used unless explicitly needed.
2.2 TimeHarmonic Plane Waves 23
An inhomogeneous plane wave propagates in the p
r
direction, but decays in
the p
i
direction. That such waves arise in practice will soon become apparent.
Substituting (2.10) into the equation of motion, (1.6), gives, just as it did in (2.3),
c
2
L
/c
2
ˆ p( ˆ p ·
ˆ
d) −
c
2
T
/c
2
ˆ p ∧ ˆ p ∧
ˆ
d =
ˆ
d. (2.12)
Thus, either ˆ p ∧
ˆ
d = 0 and c = c
L
, or ˆ p ·
ˆ
d = 0 and c = c
T
. Note that these
statements imply that
ˆ
d may also be complex.
The symbol ˆ p
i
is the i th component of ˆ p. Assume that ˆ p
3
= 0 and ˆ p
1
is real.
Then ˆ p
2
may be real or imaginary. That is,
ˆ p
2
=
1 − ˆ p
2
1
1/2
,  ˆ p
1
 < 1,
i
ˆ p
2
1
− 1
1/2
,  ˆ p
1
 > 1.
(2.13)
In the ﬁrst case the wave is homogeneous:
u = A
ˆ
de
i k(x
1
ˆ p
1
+x
2
ˆ p
2
)
. (2.14)
In the second it is inhomogeneous:
u = A
ˆ
de
−k ˆ αx
2
e
i kx
1
ˆ p
1
. (2.15)
Here α = −i ˆ p
2
. The implications of allowing the vectors ˆ p and
ˆ
d to be
complex lead to many interesting results, many of which are discussed by
Boulanger and Hayes (1993).
For the remainder of the section, we shall assume that ˆ p is real. Calculating
the instantaneous energy ﬂux gives
F
I
= ρc
I
ω
2
(i Ae
i η
I
)(i Ae
i η
I
) ˆ p, (2.16)
where η
I
= k
I
(x · ˆ p − c
I
t ), and I = L or T. The instantaneous value is of
less interest than the average value over a period T(= 2π/ω). This average is
deﬁned as
G =
1
T
t +T
t
G(τ) dτ, (2.17)
where G(t ) is a real periodic function of t with period T. It can be shown by
direct calculation that
(Ae
i η
)(B
i η
) =
1
2
(AB
∗
), (2.18)
where A and B are the complex amplitudes and η has the formgiven previously.
24 2 Kinematical Descriptions of Waves
Problem 2.1
Carry out the calculation leading to (2.18).
Applying (2.18) to (2.16) gives
F
I
=
1
2
ρc
I
ω
2
AA
∗
ˆ p. (2.19)
Problem 2.2
Show that F
I
= c
I
E
I
ˆ p and F
I
and E
I
are the corresponding ﬂux and
energy density.
2.3 PlaneWave or AngularSpectrum Representations
Spherical and cylindrical waves can be constructed directly from solutions to
the equations of motion in the corresponding coordinate systems. A summary
of this approach is given in an Appendix. However, these waves can also be
constructed fromcollections of homogeneous and inhomogeneous plane waves.
These representations are called planewave or angularspectrum representa
tions. The advantage of these representations is that, upon constructing the
outcome of an interaction of a plane wave with an obstacle, linearity allows
us to use it to construct the outcome of an interaction with the same obstacle
by a more general waveﬁeld. Here, we construct an antiplane shear, Gaussian
beam and a compressional spherical wave, and give the result for an antiplane
shear, cylindrical wave. This last construction is taken up in Problem 4.1, be
cause it is more easily motivated by starting from a radiation problem. The
clarity of the planewave spectral technique is well demonstrated by Clemmow
(1966) through the description and solution of several electromagnetic wave
propagation and scattering problems.
2.3.1 A Gaussian Beam
Consider an antiplane shear wave whose equation of motion is (1.15). We shall
continue to use u
3
for the particle displacement, but write c rather than the
awkward c
T
. When a timeharmonic disturbance is assumed, (1.15) gives, in a
region where there are no sources of excitation,
∂
α
∂
α
u
3
+ k
2
u
3
= 0. (2.20)
We consider a signaling problem. This is a problem wherein the source is given
as a function of time over an initial plane – in the present case it is given as a
2.3 PlaneWave or AngularSpectrum Representations 25
timeharmonic function along the x
2
axis – and the excited waveﬁeld is allowed
to propagate outward, usually in only one direction, to inﬁnity. At x
1
= 0,
u
3
(0, x
2
) = Ae
−(x
2
/b)
2
, (2.21)
where A is a complex amplitude and b is a parameter that measures the initial
width of the waveﬁeld. While it will not be clear from what we do here, this
waveﬁeld will remain Gaussian in cross section but spreads as it propagates
outward. As x
1
→∞, we ask that the wave be outgoing (there are no sources
at inﬁnity).
When the spatial Fourier transform over x
2
is used, the equation of motion
becomes
d
2∗
u
3
/dx
2
1
+k
2
1
∗
u
3
= 0, (2.22)
where
k
1
=
k
2
−k
2
2
1/2
, (k
1
) ≥ 0, (k
1
) ≥ 0. (2.23)
Deﬁning the branches of the radical k
1
is very important, but for the present
we do not need to know in detail where the branch cuts are placed. This will
be discussed in Section 3.4.4. The solution to (2.22) that satisﬁes the outgoing
condition is
∗
u
3
= U(k
2
)e
i k
1
x
1
. Therefore
u
3
(x
1
, x
2
) =
1
2π
∞
−∞
U(k
2
)e
i (k
1
x
1
+k
2
x
2
)
dk
2
. (2.24)
Note that we could simply have asserted that this was a solution to (2.20) and
hence begun the discussion at this point.
To ﬁnd U, we use (2.21) to write
U(k
2
) = A
∞
−∞
e
−(x
2
/b)
2
e
−i k
2
x
2
dx
2
,
= π
1/2
Abe
−(k
2
b/2)
2
. (2.25)
Therefore,
u
3
(x
1
, x
2
) =
bA
2π
1/2
∞
−∞
e
−(k
2
b/2)
2
e
i (k
1
x
1
+k
2
x
2
)
dk
2
. (2.26)
We have succeeded in taking a rather general waveﬁeld and expressing it as an
integral over a set of plane waves; hence the name planewave representation
and the descriptive term planewave spectra. Note that k
1
takes both real and
imaginary values as k
2
ranges over (−∞, ∞). That is, the integral is one over
both homogeneous and inhomogeneous, time harmonic, plane waves.
26 2 Kinematical Descriptions of Waves
2.3.2 An AngularSpectrum Representation of a Spherical Wave
Consider a spherically symmetric compressional wave such that the particle
displacement u is described by ∇ϕ. The equation governing ϕ is (1.21). The
only spatial dependence is upon r = (x
2
1
+ x
2
2
+ x
2
3
)
1/2
. Again we assume that
the disturbance is time harmonic. For r > 0, (1.21) becomes
1
r
2
∂
∂r
r
2
∂ϕ
∂r
+k
2
ϕ = 0, (2.27)
where we have written c to replace c
L
and k = ω/c to replace k
L
. For r = 0 a
solution of (2.27) is
ϕ = A(e
i kr
/kr), (2.28)
where Ais an amplitude that may be complex. Clearly its wavefront is spherical.
The energy ﬂux is proportional to r
−2
balancing the increase in surface area
of the spherical wavefront as the wave propagates outward. In Section 4.3.1,
Note 2, we suggest a possible source for such a spherical, compressional wave.
Rather than consider the wave excited by a compact source, we pose a sig
naling problem. Set ρ = (x
2
1
+ x
2
2
)
1/2
. The potential ϕ satisﬁes (2.27), while,
at x
3
= 0,
ϕ = A(e
i kρ
/kρ). (2.29)
As x
3
→ ±∞ we ask that the wave be outgoing. For the moment consider
x
3
≥ 0. The potential ϕ can be represented as
ϕ =
1
(2π)
2
∞
−∞
∞
−∞
∗
ϕ(k
1
, k
2
)e
i (k
1
x
1
+k
2
x
2
+k
3
x
3
)
dk
1
dk
2
, (2.30)
with
k
3
=
k
2
−k
2
1
−k
2
2
1/2
, (k
3
) ≥ 0, (k
3
) ≥ 0. (2.31)
This is a solution to (2.27). The termk
3
has been deﬁned so that the waveﬁeld is
outgoing. Note that we have moved directly to the step represented previously
by (2.24). Applying the condition at x
3
= 0, (2.29), gives
∗
ϕ(k
1
, k
2
) =
A
k
∞
−∞
∞
−∞
e
i kρ
ρ
e
−i (k
1
x
1
+k
2
x
2
)
dx
1
dx
2
. (2.32)
Integrals of this form are most readily evaluated by transforming the coordi
nates of the integrand to polar ones, both in the physical and in the transform
2.3 PlaneWave or AngularSpectrum Representations 27
space. The transformations are
k
1
= κ cos ψ, k
2
= κ sin ψ, x
1
= ρ cos θ, x
2
= ρ sin θ, (2.33)
where κ = (k
2
1
+k
2
2
)
1/2
and ρ was given previously. The integral in (2.32) can
now be written as
∗
ϕ =
A
k
2π
0
dθ
∞
0
e
iρ[k−κ cos(ψ−θ)]
dρ. (2.34)
Performing the integration gives
∗
ϕ = 2πi A/kk
3
, (2.35)
where k
3
is given by (2.31). Therefore, the spherical wave, (2.28), can be repre
sented as an integral, taken over both homogeneous and inhomogeneous waves,
given by
ϕ =
i A
2πk
∞
−∞
∞
−∞
e
i (k
1
x
1
+k
2
x
2
+k
3
x
3
)
dk
1
dk
2
k
3
. (2.36)
This then is the planewave representation of a spherical wave. It serves to
indicate again the centrality of plane waves as kinematical objects. Note that
x
3
has been replaced with its absolute value so as to give a spherical wave over
all space. Had we assumed x
3
negative, then the deﬁnition of the branch of the
function k
3
, (2.31), would have to be changed. The outcome of assuming both
x
3
negative and changing the branch of k
3
that is taken is equivalent to retaining
the deﬁnition of (2.31) and replacing x
3
with its absolute value.
3
It is possible to do still more with this representation. Each plane wave in
the integrand has the form e
i k·x
, where k is the complex wavevector and x the
position vector. Written this way, it is possible to imagine that a spherical wave
could be constructed froma bursting of wavevectors fromsome compact source
region, with each wavevector piercing a spherical surface. It is therefore enough
to identify each wavevector with two angles.
Again let us momentarily assume that x
3
> 0. Consider the following trans
formation, sometimes called the Sommerfeld transformation.
κ = k sin ξ, k
3
= k cos ξ, (2.37)
3
There are two points in this calculation that have not been explained clearly. The ﬁrst, how the
branch of a radical such as k
3
is determined, is explained in Section 3.4.4. The second, why we
can assume that e
i kρ
→0 as ρ →∞, is explained in Section 4.4.
28 2 Kinematical Descriptions of Waves
where κ was given previously. The variables k
1
and k
2
both vary from −∞
to +∞, so that κ varies from 0 to ∞. Moreover, k
3
must satisfy (2.31). The
integration with respect to ψ goes from 0 to 2π. But the integration over ξ is
more complicated because ξ must vary over both real and complex values to
capture the full range of κ. Let ξ = ξ
r
+ i ξ
i
, with ξ
r
and ξ
i
both real. Then
sin(ξ
r
+i ξ
i
) = sin ξ
r
cosh ξ
i
+i cos ξ
r
sinh ξ
i
and cos(ξ
r
+i ξ
i
) = cos ξ
r
cosh ξ
i
−
i sin ξ
r
sinh ξ
i
. Let ξ
r
vary from0 to π/2 and keep ξ
i
= 0, so that κ varies from0
to k and (k
3
) ≥ 0. To have κ continue to ∞, we hold ξ
r
= π/2 and let ξ
i
vary to
either +∞or −∞. To keep (k
3
) ≥ 0 we must have ξ
i
vary from 0 to −∞. We
are integrating in the complex ξ plane so that a strict adherence to this contour
is not essential, but the contour must begin at zero and end somewhere near
π/2−i ∞. Further, it must be such that the integral is convergent and represents
an outgoing waveﬁeld. Carefully changing the variables of integration in (2.36)
gives
ϕ =
i A
2πk
2π
0
dψ
π/2−i ∞
0
e
i k·x
sin ξ dξ. (2.38)
Note that k
3
has has been removed by this transformation. The expression (2.38)
is called an angularspectrum representation and succinctly captures the image
of a spherical wave as a burst of wavevectors. Note that to fully achieve the
spherical wavefront, inhomogeneous waves are also needed.
2.3.3 An AngularSpectrum Representation of a Cylindrical Wave
To achieve a planewave or angularspectrum representation of a cylindrical
wave, it is easier to work from the equation of motion with a line force than
to try to pose a signaling problem. This is done in Problem 4.1. However, as
indicated in the Appendix, an outgoing cylindrical wave is described by the
Hankel function, H
(1)
0
(kρ). Sommerfeld (1964) describes how to represent this
function as an angular spectrum and gives the result
H
(1)
0
(kρ) =
1
π
C
e
i kρ cos ξ
dξ. (2.39)
The contour C begins at −η +i ∞and ends at η −i ∞, where η ∈ (0, π).
2.4 Asymptotic Ray Expansion
A much older way to represent waves is by using rays. Born and Wolf
(1986) give a concise introduction to ray theory for electromagnetic waves,
while Achenbach et al. (1982) provide a detailed asymptotic development of
2.4 Asymptotic Ray Expansion 29
elasticwave ray theory. There are, at least, two ways to approach ray descrip
tions. The one presented next builds upon the timeharmonic plane wave, with
the assumption, lurking in the background, that we can synthesize the time
dependent case from this construction, should we need to. However, it is also
possible to use the fact that hyperbolic equations permit discontinuities to be
transported along their characteristics. A wave is permitted to have a discon
tinuity at its wavefront and the kinematics of the discontinuity is developed.
This is the approach taken by Hudson (1980) for elastic waves, and in a more
general context by Whitham (1974). The two approaches are connected by
the fact that a discontinuity in a wavefront engages predominantly the high
frequency components of the temporal Fourier transform (Lighthill, 1978). We
expect ray descriptions to be useful only when the wavelengths are small with
respect to any other characteristic length in the propagation environment.
2.4.1 Compressional Wave
We begin by working with the scalar potential ϕ. Knowing the representation
for ϕ, it is straightforward to construct that for the vector potential ψ and hence
that for the particle displacement u. We are, then, left examining the equation
∇
2
ϕ +k
2
ϕ = 0, (2.40)
where we have again written c for c
L
, so that k = ω/c.
As a generalization of a plane wave, we assume that ϕ can be expanded as
ϕ(x) ∼ e
i k[S(x)−ct ]
(i k)
−α−1
∞
n=0
(i k)
−n
A
n
(x), (2.41)
where α < 1 is a positive number (the term raised to the power α is added
for generality). We have added the e
−i ωt
, where t is time, to the exponential to
aid in our subsequent discussions. In writing this expression, we are using two
ideas. First, we note that any surface can be approximated by its tangent planes
and hence locally any waveﬁeld will have a propagating part or phase
4
that
behaves as does that of a plane wave. Second, by examining the representations
(2.38) and (2.39), we should expect that the amplitude of a wave with a curved
wavefront depends insome wayonk or, equivalently, the wavelengthλ = 2π/k.
4
The word phase has a rather confused meaning in wave propagation. Looking back at (2.10), we
note the argument of the amplitude A makes a contribution to the exponential term of a plane
wave. This contribution is sometimes called the phase. However, referring to (2.41), we note that
the exponential, kS(x), is also called the phase. When the word phase is used in this book, it
refers to this latter term, that controlled by the wavenumber k.
30 2 Kinematical Descriptions of Waves
Therefore, it is reasonable that the amplitude can be expanded as an asymptotic
power series
5
in the principal parameter λ, or equivalently k
−1
, and that the
approximation becomes increasingly accurate as λ → 0 or k → ∞. We begin
the expansionwithk
−α−1
sothat the approximationfor the particle displacement
will start as k
−α
. Lastly, it is not good workmanship to expand in a parameter
with a dimension because the measure of large or small remains too vague. In
the present case the reader should imagine that k is multiplied by a distance
representing that between interactions, call it L, and that it is kL that is very
large. Rather than introduce an extra parameter, however, we suppress the L
unless explicitly needed.
Substituting (2.41) into (2.40) gives
∞
n=0
(∇S · ∇S −1)A
n
+ [2(∇S · ∇)A
n−1
+∇
2
SA
n−1
] (2.42)
+∇
2
A
n−2
(i k)
−(n−1)
= 0,
where the A
n
with negative subscripts are zero. The set {(i k)
−n
} is linearly
independent. Accordingly, for A
0
= 0,
∇S · ∇S = 1. (2.43)
Note that ∇S = 1. This equation is called the eikonal equation. At n = 1 we
have
2(∇S · ∇)A
0
+(∇
2
S)A
0
= 0, (2.44)
and, for n > 1,
2(∇S · ∇)A
n−1
+(∇
2
S)A
n−1
= −∇
2
A
n−2
. (2.45)
These last two equations are called the transport equations. Collectively, (2.43)–
(2.45) provide a recursive scheme whereby the solution to one equation gives
the missing information needed for the solution to the next. The solution of the
eikonal equation gives a family of curves along which the amplitudes A
n
are
transported.
We deﬁne a ray as the vector k∇S. It is tangent to one of the curves found
from solving (2.43). We deﬁne a wavefront, just as we did following (2.1),
5
Holmes (1995) and Hinch (1991) provide useful starting points for the reader to learn more about
asymptotic and perturbation methods.
2.4 Asymptotic Ray Expansion 31
as the family of surfaces S(x) = ct + C, where t is a parameter giving each
member of the family and C is a constant.
To solve (2.43), let x(s, q) deﬁne a family of curves that are orthogonal to each
surface S(x) = ct +C. The variable s deﬁnes the arclength along the curve,
while q indicates each member of the family. For the moment, we imagine that
x does not depend on q and suppress any reference to it until we need it. By
deﬁnition,
dx/ds = ∇S. (2.46)
Setting ˆ p = ∇S, we see that the lefthand side of (2.43) is ˆ p · ∇S, which is
nothing more than the directional derivative dS/ds. Thus S(x) = s(x) + ct
0
,
where t
0
is an initial time. We have assumed that the equation for each curve
x(s) can be inverted to give s as a function of x. We do not at present know
x(s).
Direct differentiation and using (2.43) indicate that d ˆ p/ds = 0 along a curve
x(s, q), where we have reinserted the q to indicate that we are now considering
a family (a pencil) of such curves. Therefore the curves are straight lines whose
equations are
x(s, q) = x
0
(q) +s ˆ p(q). (2.47)
In Fig. 2.2 we have sketched some aspects of the construction. The vector
x
0
(q) is the starting point for a ray identiﬁed by q on the initial wavefront
S(x
0
) = ct
0
. Boththese equations are assumedknownor given. The components
of the vector q = (q
1
, q
2
) constitute a coordinate systemon the initial wavefront
and deﬁne the point at which the ray is launched at t = t
0
. The rays propagate
along and are tangent to x(s, q) and pierce each wavefront perpendicularly.
Inverting (2.47) to obtain s(x) and q(x) gives the equation for a wavefront as
S(x) = s(x) +S(x
0
), where s(x) = c(t −t
0
)
6
. Recall that (2.47), by the implicit
functiontheorem, is locallyinvertible if the Jacobiandoes not vanish. Surfaces at
which this Jacobian vanishes are called caustics. In the neighborhood of these
surfaces, the expansion (2.41) becomes disordered. Choi and Harris (1989,
1990) give an interesting example of caustic formation in elastic materials,
though their asymptotic analysis starts from integral representations, rather
than ray expansions.
6
Note that this description of the rays, while qualitatively correct for anisotropic or inhomogeneous
materials, requires some substantial modiﬁcation in these cases.
32 2 Kinematical Descriptions of Waves
q
S(x
0
)
x(s, q) ϭ x
0
(q) ϩ sp
S(x)
ˆ
Fig. 2.2. A ray is launched in a direction normal to an initial wavefront S(x
0
) from the
point q on S(x
0
). It is a straight line. After propagating a distance s along this line, it
pierces the wavefront S(x), again in a normal direction. The starting point of the ray is
x
0
(q), and its present position is x(s, q) = x
0
(q) +s ˆ p, where ˆ p is a unit vector pointing
along (tangent to) the ray
Weatherburn (1939)
7
shows that
∇
2
S = 1/ρ
1
+1/ρ
2
, (2.48)
where ρ
1
and ρ
2
are the principal radii of curvature of the surface S(x) =
s(x) + S(x
0
). These radii take a sign. If a wavefront is expanding, its radii of
curvature are positive, and the normal to the surface points in the direction of
expansion. If the wavefront is contracting, the radii of curvature are negative,
but the normal continues to point in the direction of propagation, namely that
of contraction. Moreover, because s is the distance along the straight lines
orthogonal to each surface, ρ
1
= ρ
01
+ s and ρ
2
= ρ
02
+ s, where the ρ
0i
are
the radii of curvature of the initial surface S(x
0
) = ct
0
. Accordingly, the ﬁrst
transport equation, (2.49), can be written as
dA
0
ds
+
1
2
1
ρ
01
+s
+
1
ρ
02
+s
A
0
= 0. (2.49)
This is a differential equation along each ray q. Its solution is
A
0
=
A(q)
[(ρ
01
+s)(ρ
02
+s)]
1/2
, (2.50)
where A(q) is assumed to be given.
7
The difference in sign from that in Weatherburn arises because of how we have deﬁned the signs
of the radii of curvature.
2.4 Asymptotic Ray Expansion 33
Collecting the various pieces and noting that the compressional component
of the particle displacement u
L
= ∇ϕ, gives, to leading order,
u
L
∼
ˆ pA
L
(q)
(i k
L
)
α
[(ρ
01
+s)(ρ
02
+s)]
1/2
e
i k
L
[s(x)−c
L
(t −t
0
)]
, (2.51)
where we have put back the subscript L to indicate that we have calculated the
compressional component. The power α must also be given, typically as part
of the initial data for the problem.
This is a remarkable expression. It captures our intuitive notions of a wave as
an object whose phase steadily increases as we move along a curve, the tangent
vector to the curve being a ray, and whose amplitude decays geometrically
such as to conserve energy (Problem 2.4). Note also that the denominator may
vanish, in which case our asymptotic approximation is no longer accurate. This
happens on the caustic surfaces mentioned previously. Lastly, note that the
polarization of the leadingorder term (2.51) is that of a longitudinal wave, but
this does not preclude the possibility that the higherorder terms could have
different polarizations.
Problem 2.4 Conservation of Energy in a Ray Tube
Show, by multiplying through by A
0
, that the ﬁrst transport equation, (2.49),
can be written as
∇ ·
∇SA
2
0
= 0. (2.52)
Integrate this over a tube formed from rays and, for an inﬁnitesimal cross
section, deduce the result, (2.50). Interestingly, this equation establishes more
than just the conservation of timeaveraged energy. The A
0
can be complex and
the solution to (2.52) gives both its magnitude and its argument. If conservation
of energy were our only goal, we should consider ∇ · (∇SA
0
A
∗
0
) = 0. The
reader is encouraged to do so.
2.4.2 Shear Wave
Next we consider a shear wave. The vector potential ψ is written as
ψ(x) ∼ e
i k
T
[S(x)−c
T
t ]
(i k
T
)
−α−1
∞
n=0
(i k
T
)
−n
A
n
(x), (2.53)
where k
T
= ω/c
T
. Recall that ∇ · ψ = 0. Provided we do not seek terms
beyond the leading one, this implies that
A
0
· ∇S = 0. (2.54)
34 2 Kinematical Descriptions of Waves
The kinematic analysis here will be identical to that of the previous section so
that ˆ p = ∇S gives the direction of the rays, which propagate along straight
lines. Hence, A
0
will be orthogonal to ˆ p and have two linearly independent
components, A
T V
0
ˆ
d
T H
and A
T H
0
ˆ
d
T V
. We select the unit vectors
ˆ
d
I
so that
ˆ p ∧
ˆ
d
T V
=
ˆ
d
T H
. (2.55)
At this point the analysis giveninthe previous sectionapplies toeachcomponent
of A
0
. We ﬁnd that, to leading order, the shear particle displacement is
u
I
∼
χ
I
ˆ
d
I
A
I
(q)
(i k
T
)
α
[(ρ
01
+s)(ρ
02
+s)]
1/2
e
i k
T
[s(x)−c
T
(t −t
0
)]
, (2.56)
where χ
I
= ∓1 as I = T H, T V. Note that the radii of curvature ρ
0i
are
different for the two wave types (2.51) and (2.56). It takes the presence of a
source or boundary to couple the waves together. As with the compressional
wave, the polarization can change as one proceeds to the higherorder terms.
Appendix: Spherical and Cylindrical Waves
The following is a summary of information about spherical and cylindrical
waves.
To begin, we consider a waveﬁeld whose only dependence is upon the radial
coordinate r and whose only component of particle displacement u is in the
radial direction. The equation of motion becomes
∂
2
u
∂r
2
+
2
r
∂u
∂r
−
2u
r
2
=
1
c
2
L
∂
2
u
∂t
2
. (2.57)
Such a waveﬁeld could be excited by a uniformly pressurized spherical cavity.
Setting u = ∂ϕ/∂r, we ﬁnd that this equation reduces to
∂
2
(rϕ)
∂r
2
=
1
c
2
L
∂
2
(rϕ)
∂t
2
, (2.58)
and its solution is
ϕ(r, t ) =
1
r
f
t −
r
c
L
+
1
r
g
t +
r
c
L
, (2.59)
where f and g are arbitrary functions. The ﬁrst term represents an outgoing
wave and the second an incoming one. Let us assume that there is only an
References 35
outgoing wave and take its temporal transform. Then we get
¯ ϕ =
¯
f (ω)
r
e
i k
L
r
, (2.60)
in agreement with (2.28). Note that the particle displacement u = ∂ϕ/∂r so
that there are both r
−1
and r
−2
terms.
Next we consider inplane, rotary shear motion. We use a cylindrical coordi
nate system to describe it. The only component of particle displacement is v, a
displacement in the angular, θ direction. The only dependence is upon the polar,
radial coordinate ρ. There is no dependence on x
3
. The equation of motion is
∂
2
v
∂ρ
2
+
1
ρ
∂v
∂ρ
−
v
ρ
2
=
1
c
2
T
∂
2
v
∂t
2
. (2.61)
Such a waveﬁeld could be excited by a traction, solely in the θ direction, on the
walls of a cylindrical cavity. Setting v = −∂ψ
3
/∂ρ, we ﬁnd that this equation
reduces to
∂
2
ψ
3
∂ρ
2
+
1
ρ
∂ψ
3
∂ρ
=
1
c
2
T
∂
2
ψ
3
∂t
2
. (2.62)
Taking the temporal transform, we get
∂
2
¯
ψ
3
∂ρ
2
+
1
ρ
∂
¯
ψ
3
∂ρ
+k
2
T
¯
ψ
3
= 0. (2.63)
This is Bessel’s equation of order zero and has, as two of its linearly independent
solutions, the Hankel functions H
(1,2)
0
(k
T
ρ). The asymptotic behavior of these
functions is
H
(1,2)
0
(k
T
ρ) ∼ (2/πk
T
ρ)
1/2
e
±i (k
T
ρ−π/4)
, k
T
ρ →∞, (2.64)
where the plus sign goes with the 1 and the minus sign with the 2. From this
behavior, we see that H
(1)
0
(k
T
ρ) represents an outgoing wave and H
(2)
0
(k
T
ρ) an
incoming one. Recall that the particle displacement is v = −∂ψ
3
/∂ρ.
References
Achenbach, J.D., Gautesen, A.K., and McMaken, H. 1982. Ray Methods for Waves in
Elastic Solids, pp. 77–88. Boston: Pitman.
Born, M. and Wolf, E. 1986. Principles of Optics, 6th (corrected) ed., pp. 109–127.
Oxford: Pergamon.
Boulanger, Ph. and Hayes, M. 1993. Bivectors and Waves in Mechanics and Optics.
New York: Chapman and Hall.
36 2 Kinematical Descriptions of Waves
Choi, H.C. and Harris, J.G. 1989. Scattering of an ultrasonic beam from a curved
interface. Wave Motion 11: 383–406.
Choi, H.C. and Harris, J.G. 1990. Focusing of an ultrasonic beam by a curved
interface. Wave Motion 12: 497–511.
Clemmow, P.C. 1966. The Plane Wave Spectrum Representation of Electromagnetic
Fields. Oxford: Pergamon.
Hinch, E.J. 1991. Perturbation Methods. Cambridge: University Press.
Holmes, M.H. 1995. Introduction to Perturbation Methods. New York: Springer.
Hudson, J.A. 1980. The Excitation and Propagation of Elastic Waves. New York:
Cambridge.
Lighthill, M.J. 1978. Fourier Analysis and Generalized Functions, pp. 46–57. New
York: Cambridge
Sommerfeld, A. 1964. Partial Differential Equations in Physics, Lectures on
Theoretical Physics, Vol. VI, pp. 84–101. Translated by E.G. Straus. New York:
Academic.
Weatherburn, C.E. 1939. Differential Geometry of Three Dimensions, Vol. 1,
pp. 225–227. Cambridge: University Press.
Whitham, G.B. 1974. Linear and Nonlinear Waves. New York: WileyInterscience.
3
Reﬂection, Refraction, and Interfacial Waves
Synopsis
Chapter 3 describes reﬂection and refraction at an interface between two mate
rials having different densities and wavespeeds. Moreover, the chapter describes
waves that propagate along an interface, while decaying perpendicularly away
from it. These waves, in many cases, must be continuously renewed from a
waveﬁeld in the interior of one or both the materials to sustain their propagation.
However, a tractionfree elastic surface is special in that a wave can be excited
on such a surface and not require an interior waveﬁeld to sustain it.
There is an intimate relation between the topography of the complex waven
umber plane and the physical manifestation of the propagating waveﬁeld, as we
have previously noted. Extending this connection, we discuss how branches of
the function (k
2
−z
2
)
1/2
, where z is the complex variable and k a parameter, are
selected and how these selections manifest themselves in the physical domain.
3.1 Reﬂection of a Compressional Plane Wave
We consider a longitudinal or compressional plane wave incident to a traction
free surface. Figure 3.1 indicates the geometry of the problem along with a
schematic representation of the interaction. Further, we assume that the distur
bance is time harmonic, but suppress the e
−i ωt
unless it is explicitly needed.
We do so knowing that using (1.36) will carry the resulting expressions into
corresponding timedependent ones. The incident wave u
0
is described by
u
0
= A
0
ˆ p
0
e
i k
L
ˆ p
0
·x
, (3.1)
where the unit vector ˆ p
0
, which describes both the polarization of the wave and
its direction of propagation, is given by
ˆ p
0
= sin θ
0
ˆ e
1
+cos θ
0
ˆ e
2
. (3.2)
37
38 3 Reﬂection, Refraction, and Interfacial Waves
Fig. 3.1. The elastic solid ﬁlls the x
2
< 0 halfspace, while x
2
> 0 is a vacuum. The
surface is free of tractions. A compressional plane wave is incident to the surface in the
direction ˆ p
0
.
The angle θ
0
is indicated in Fig. 3.1, and the ˆ e
i
are the unit vectors along the
x
i
axes (i = 1, 2). The amplitude A
0
is real and positive. As in previous work,
k
L
= ω/c
L
. The incident polarization has no antiplane component. Therefore
any waves excited at the boundary must also be inplane. Further, the incident
wave is plane so that it imposes a projection of its phase everywhere on the
surface. We therefore expect both compressional and shear plane waves to
be reﬂected
1
from the boundary, because both wave types can have matching
phase components along the surface.
The reﬂected compressional wave is described by
u
1
= A
1
ˆ p
1
e
i k
L
ˆ p
1
·x
, (3.3)
where the vector ˆ p
1
is given by
ˆ p
1
= sin θ
1
ˆ e
1
−cos θ
1
ˆ e
2
. (3.4)
The reﬂected shear wave is described by
u
2
= A
2
ˆ
d
2
e
i k
T
ˆ p
2
·x
, (3.5)
where the propagation direction is given by
ˆ p
2
= sin θ
2
ˆ e
1
−cos θ
2
ˆ e
2
, (3.6)
1
I use the phrase scattered wave or scattered waveﬁeld to describe almost any disturbance that
is returned from or perturbed by an obstacle struck by an incident wave. When the obstacle is
impenetrable, the scattered waveﬁelds generally break into reﬂected and diffracted ones. Any
waveﬁeld that penetrates a shadowed region is a diffracted one; otherwise, it is reﬂected.
3.1 Reﬂection of a Compressional Plane Wave 39
but the polarization direction by
ˆ
d
2
= ˆ e
3
∧ ˆ p
2
. (3.7)
Figure 3.1 deﬁnes the angles θ
1
and θ
2
. The amplitudes A
1
and A
2
may be
complex. And as in previous work, k
T
= ω/c
T
. Note that both reﬂected waves
propagate away fromthe surface. Also note that the polarization of the reﬂected
shear wave is chosen so that ˆ p
2
∧
ˆ
d
2
= ˆ e
3
.
3.1.1 Phase Matching
The boundary conditions at x
2
= 0 are that τ
22
= τ
21
= 0. Direct application
of these boundary conditions can be tedious. However, a little thought indicates
that they must take the form
(· · ·)A
0
e
i k
L
sin θ
0
x
1
+(· · ·)A
1
e
i k
L
sin θ
1
x
1
+(· · ·)A
2
e
i k
T
sin θ
2
x
1
= 0. (3.8)
This condition must hold for all x
1
. This can happen only if θ
0
= θ
1
and
s
L
sin θ
0
= s
T
sin θ
2
, where s
I
= c
−1
I
is the slowness. These two conditions
are called the phasematching condition. The phrase phasematching is often
used as a verb in the sense that “the reﬂected waves phase match to the incident
one.” Phase matching is nothing more than a demand that the projections, on the
surface x
2
= 0, of the various wavelengths be equal or, equivalently, because
ω is common to all the k
I
, that the wavespeeds along the surface be identical.
In fact it is a kinematical condition, rather than a kinetic one. The surface is
translationally invariant and the disturbance at one point differs from that at
another only by an exponential phase term that indicates propagation, in this
case to the right. This is a similar condition to that used to derive (1.66). Phase
matching is a fundamental principle of linear wave propagation, and, whenever
it occurs, something physically interesting will happen.
There is a very useful diagram that geometrically represents the phase
matching condition. We deﬁne the slowness vectors
s
0
= ˆ p
0
/c
L
, s
1
= ˆ p
1
/c
L
, s
2
= ˆ p
2
/c
T
. (3.9)
The tip of each slowness vector describes a circle, a slowness surface,
2
as the
parameter θ
0
is variedthrough2π. This is showninFig. 3.2. The phasematching
condition is stated as the condition that the s
1
components of each slowness
vector must be equal. We have indicated this by the vertical dashed line. In the
2
Slowness surfaces for isotropic materials are always spheres (circles really, because only two
dimensions are involved), but for anisotropic materials they can become very elaborate indeed.
Auld (1990) gives an account of these surfaces and of their use in determining phasematching.
40 3 Reﬂection, Refraction, and Interfacial Waves
s
1
s
2
s
0
s
1
s
2
Fig. 3.2. The slowness surfaces for an isotropic elastic solid. The phasematching
condition is indicated geometrically by demanding that the horizontal components (pro
jections on the surface x
2
= 0) of the slowness vectors be equal.
case we are examining here, for real θ
0
, θ
2
must be real and less than π/2, even
in the limit that θ
0
= π/2.
3.1.2 Reﬂection Coefﬁcients
Problem 3.1 asks the reader to calculate the longitudinal or compressional
reﬂection coefﬁcient R
L
(θ
0
) = A
1
/A
0
and the transverse or shear reﬂection
coefﬁcient R
T
(θ
0
) = A
2
/A
0
. When this is done the reader will ﬁnd that
R
L
(θ
0
) = A
−
(θ
0
)/A
+
(θ
0
), (3.10)
R
T
(θ
0
) = 2κ sin 2θ
0
cos 2θ
2
/A
+
(θ
0
). (3.11)
The auxiliary functions are
A
∓
= sin 2θ
0
sin 2θ
2
∓κ
2
cos
2
2θ
2
. (3.12)
These terms must be supplemented with the condition s
L
sin θ
0
= s
T
sin θ
2
. The
term κ = c
L
/c
T
is given by
κ = [2(1 −ν)/(1 −2ν)]
1/2
, (3.13)
where ν is Poisson’s ratio. For many materials, κ is close to 3
1/2
.
3.2 Reﬂection and Refraction 41
Note that, provided θ
0
is real, these reﬂection coefﬁcients have no frequency
dependence and hence can be used both for timeharmonic and timedependent
plane waves.
Problem 3.1 Reﬂection Coefﬁcients
Show that applying the boundary condition τ
22
= 0 leads to
λ +2µcos
2
θ
0
(A
1
/A
0
) −κµsin 2θ
2
(A
2
/A
0
) = −
λ +2µcos
2
θ
0
,
(3.14)
and that applying the condition τ
21
= 0 leads to
−µsin 2θ
0
(A
1
/A
0
) −κµcos 2θ
2
(A
2
/A
0
) = −µsin 2θ
0
. (3.15)
Hence deduce (3.10)–(3.12).
Problem 3.2 Conservation of Energy
Show that the power ﬂux into the surface equals that away from it. That is,
show that
R
L

2
+R
T

2
cos θ
2
κ cos θ
0
= 1. (3.16)
Hint. The calculation becomes straightforward once the reader realizes that
he or she must be careful to use the same element of area at the surface to
calculate both the ﬂux into the surface and that away from it.
While there are several other problems of this general kind that could be
considered, we treat only one other, namely the reﬂection and refraction of an
antiplane shear wave at the interface between two contrasting materials.
3.2 Reﬂection and Refraction
We consider a plane, antiplane shear wave incident to an interface between two
materials that are in welded contact. In this case both the traction and the particle
displacement are continuous across the interface. To simplify the notation, we
replace c
T
, k
T
and similarly labeled symbols by c, k, and so on. It is clear that
only an antiplane particle displacement can be excited at the interface by an
42 3 Reﬂection, Refraction, and Interfacial Waves
Fig. 3.3. In the x
2
< 0 halfspace the density is ρ, the elastic shear modulus µ, and
the wavespeed c = (µ/ρ)
1/2
, while in the x
2
> 0 halfspace the density is ¯ ρ, the elastic
shear modulus ¯ µ, and the wavespeed ¯ c = ( ¯ µ/¯ ρ)
1/2
. A plane, antiplane shear wave is
incident to the surface in the direction ˆ p
0
.
incident wave with an antiplane polarization, so that the scattered waves must
also be similarly polarized.
Figure 3.3 indicates the various angles. The incident wave is described by
u
30
= A
0
e
i k ˆ p·x
, (3.17)
where the subscript 3 indicates the component and therefore the polarization,
and the 0 that the wave is incident. The incident direction is given by (3.2). The
reﬂected wave, indicated with the additional subscript 1, is described by
u
31
= A
1
e
i k ˆ p·x
, (3.18)
with its propagation direction given by (3.4). The wave that is transmitted
across the interface is said to be refracted. The refracted wave, indicated by the
additional subscript 2, is described by
u
32
= A
2
e
i
¯
k ˆ p
2
·x
, (3.19)
with its propagation direction given by
ˆ p
2
= sin θ
2
ˆ e
1
+cos θ
2
ˆ e
2
. (3.20)
3.2 Reﬂection and Refraction 43
s
1
s
1
s
1
s
1
s
2
s
2
s
2
s
2
Fig. 3.4. The upper slowness surface is that for the material of the upper halfspace
in Fig. 3.3 and the lower for the material of the lower halfspace. The lefthand side
indicates the case ¯ c > c, and the right the opposite one. The phasematching condition
is indicated by the dashed vertical lines.
In Fig. 3.4 the slowness surfaces for the two materials are arranged vertically
in a pattern corresponding with the positions of the halfspaces in Fig. 3.3.
The two possible cases are shown. The lefthand side indicates that when the
upper material is faster and the righthand side that when it is slower. Phase
matching demands that the s
1
components of the slowness vectors be equal; that
is, θ
0
= θ
1
and ¯ s sin θ
2
= s sin θ
0
. More importantly, the slowness diagrams
indicate the occurrence of critical refraction. This occurs when a wave incident
to the interface from the slower material excites a wave skimming along the
interface in the faster material. In principle the reverse can also occur. For
the case ¯ c > c or, equivalently ¯ s < s, the critical angle of incidence is θ
0c
,
where s sin θ
0c
= ¯ s. Of course θ
0
can be greater than θ
0c
, in which case θ
2
must be complex to satisfy the phasematching condition. This gives rise to an
inhomogeneous plane wave in the upper material but to no timeaverage ﬂux
of energy across the interface, as the solution to Problem 3.4 indicates. This
wave, clinging to the interface, is propelled along it by the waves in the lower
material.
44 3 Reﬂection, Refraction, and Interfacial Waves
Applying the continuity of traction and particle displacement at the interface
gives the antiplane shear, reﬂection coefﬁcient R(θ
0
) and the antiplane shear,
refraction or transmission coefﬁcient T(θ
0
), namely
R(θ
0
) = C
−
(θ
0
)/C
+
(θ
0
), (3.21)
T(θ
0
) = 2 cos θ
0
/C
+
(θ
0
). (3.22)
The auxiliary functions are
C
∓
= cos θ
0
∓( ¯ µc/µ¯ c) cos θ
2
. (3.23)
Problem 3.3 Slowness Surfaces
Figure 3.4 shows how we can place slowness surfaces above one another to
work out the phasematching conditions and the various critical angles. Draw
diagrams, similar to Fig. 3.4, of the slowness surfaces for inplane motion.
Consider both an incident compressional plane wave and inplane shear one.
Consider all the possibilities for critical reﬂection and critical refraction. Critical
reﬂection, analogouslytocritical refraction, arises whena wave skimmingalong
the interface or surface of a material is excited by a slower wave incident to
the interface from the same material. The reader may be surprised by the large
number of occurrences of critical reﬂection and refraction.
3.3 Critical Refraction and Interfacial Waves
Our discussion of critical refraction indicates that we do not need to consider
the angle of refraction as real. Moreover, as Problem 3.3 has indicated, critical
phenomena occur frequently in elasticwave scattering. We must then regard
reﬂection and transmission coefﬁcients as functions of a complex variable.
In fact, we shall ﬁnd in Chapters 5 and 6 that where the complex plane is
punctured by the poles or cut by the branches of these coefﬁcients, interesting
wave processes occur.
Let us continue to consider the problem of Section 3.2 and work with the
case of ¯ c > c. When θ
0
> θ
0c
, θ
2
must be complex to permit sin θ
2
> 1. To
ﬁnd its form we set θ
2
= π/2 ± iβ, where β is real and positive. In this case
sin θ
2
= cosh β and cos θ
2
= ∓i sinh β. Next we look back at (3.21)–(3.23) and
note that R(θ
0
) = 1, implying that R(θ
0
) = e
−i 2ϕ
and T(θ
0
) = T(θ
0
)e
−i ϕ
.
Here ϕ is the argument of C
+
(θ
0
). Setting A
0
= 1, we ﬁnd that the particle
displacement in x
2
> 0 becomes
u
32
= T(θ
0
)e
−i ϕ
e
±
¯
k sinh βx
2
e
i
¯
k cosh βx
1
, (3.24)
3.3 Critical Refraction and Interfacial Waves 45
where ˆ p
2
has been written in full as ˆ p
2
= cosh β ˆ e
1
∓ i sinh β ˆ e
2
. Assuming
that
¯
k is positive, we must select θ
2
= π/2 − iβ to ensure that the waveﬁeld
decays as x
2
→∞. Thus (3.24) is an example of both an inhomogeneous plane
wave and of a wave that clings to the interface. We call the wave an interfacial
one. However, note that the wave would not exist unless a plane wave were
continuously incident to the interface from x
2
< 0 sustaining it.
There remains one subtle point. In choosing the minus sign we assumed that
¯
k
and thus ω was positive. If ω were negative then we must choose θ
2
= π/2+iβ
to ensure decay. Therefore, when θ
0
≥ θ
0c
, we must write
R(θ
0
) = e
−i 2ϕ sgnω
, T(θ
0
) = T(θ
0
)e
−i ϕ sgnω
, (3.25)
where
sgn ω =
1, ω > 0,
−1, ω < 0,
(3.26)
to take account of this ω dependence. The argument ϕ is determined when ω is
positive. It was to handle such situations that we rewrote the inverse temporal
Fourier transform in the form of (1.36).
Problem 3.4 Flux of Energy in the Upper Material
For the case of critical refraction, show that the timeaverage ﬂux of energy
in x
2
> 0 is given by
F =
1
2
¯ µω
¯
kT(θ
0
)
2
e
−2
¯
k sgnω sinh β x
2
cosh β ˆ e
1
. (3.27)
Thus there is no ﬂux of energy to x
2
→∞and energy is continually returned
to the slower material.
When θ
0
< θ
0c
the reﬂection and transmission coefﬁcients have no frequency
dependence and therefore a plane pulse reﬂects and refracts exactly as a time
harmonic, plane wave. When critical refraction takes place there is a frequency
dependence that distorts howa pulse reﬂects and refracts. Moreover, this partic
ular frequency dependence is rather interesting because it depends only on the
sign of ω. To explore this further
3
we construct an incident planewave pulse,
namely
u
30
(t − ˆ p
0
· x/c) =
1
π
∞
0
A
0
(ω) e
−i ω(t − ˆ p
0
·x/c)
dω, (3.28)
3
This argument has been taken from Friedlander (1947).
46 3 Reﬂection, Refraction, and Interfacial Waves
where we have used the fact that A
0
(ω) = A
∗
0
(−ω) because u
30
is real. A
0
(ω) is
no longer simply a positive real constant, but can nowbe an arbitrary amplitude
with a frequency dependence. From the linearity of the problem it follows that
the reﬂected pulse is given by
u
31
(t − ˆ p
1
· x/c) =
1
π
∞
0
A
0
(ω)e
−i 2ϕ
e
−i ω(t − ˆ p
1
·x/c)
dω, (3.29)
while the refracted pulse is given by
u
32
(x, t ) = T(θ
0
)
1
π
∞
0
A
0
(ω)e
−i ϕ
e
−ωx
2
sinh β/¯ c
e
−i ω(t −x
1
cosh β/¯ c)
dω.
(3.30)
The reﬂected pulse retains many of the features of the incident pulse and most
importantly the argument indicating that the wave is plane. The refracted pulse
is rather more complicated.
We rewrite the reﬂected pulse in the form
u
31
(t − ˆ p
1
· x/c) = cos 2ϕ u
30
(t − ˆ p
1
· x/c) +sin 2ϕ v(t − ˆ p
1
· x/c),
(3.31)
where
v(t − ˆ p
1
· x/c) =
1
π
∞
0
A
0
(ω) e
−i ω(t − ˆ p
1
·x/c)
dω. (3.32)
The incident pulse is reﬂected with a diminished amplitude and a newpulse has
been excited. The function v is called the allied function (Titchmarsh, 1948).
To make this a little more concrete we examine an incident pulse of the form
u
30
(t ) = H(t )−H(t −a), where H(t ) is the Heaviside function and a measures
the length of the pulse. Its allied function is
v(t ) =
1
π
ln
a −t
t
. (3.33)
The reﬂected pulse then becomes
u
31
(t − ˆ p
1
· x/c) = cos 2ϕ u
30
(t − ˆ p
1
· x/c)
+ sin 2ϕ
1
π
ln
t −a − ˆ p
1
· x/c
t − ˆ p
1
· x/c
. (3.34)
Note that the second pulse is present for all time and appears to arrive before
the incident pulse. This is an example of a twosided wave. It cannot exist alone
and must be part of a more extended disturbance.
3.3 Critical Refraction and Interfacial Waves 47
To explain this anomalous behavior we must consider the transmitted pulse.
Carrying out the necessary integrations gives
u
32
(x, t ) = T(θ
0
)
1
π
cos ϕ
tan
−1
τ
λ
−tan
−1
τ −a
λ
+
1
2
sin ϕ ln
(τ −a)
2
+λ
2
τ
2
+λ
2
, (3.35)
where
τ = t − x
1
cosh β/¯ c, λ = x
2
sinh β/¯ c. (3.36)
Clearly the refracted pulse is also two sided and present for all time.
This is not as anomalous as it might at ﬁrst appear, once we recall that the
upper material is faster than the lower and that the incident plane pulse has been
exciting the upper material from t →−∞, at x
1
→−∞(θ
0
> 0) in Fig. 3.3.
At the far left, the incident pulse excites a disturbance in the faster material.
This disturbance propagates along the interface at ¯ c and ﬁlls the whole upper
material, or, at least, the regionnear the interface. However, the maindisturbance
moves along the interface at ¯ c/ cosh β = c/ sin θ
0
. Therefore the precursor in
the upper material gets ahead of the trace of the wavefront, of the incident
pulse, at the interface and must somehow satisfy the boundary condition. It
does so by shedding a reﬂected pulse into the slower material that appears
before the incident one arrives, the second term in (3.34). Moreover, no energy
is permanently carried into the upper material. The refracted pulse continually
reradiates into the slower, lower material. It is also interesting to note that,
as the distance from the interface increases, u
32
decays as 1/x
2
rather than
as e
−
¯
kx
2
sinh β
, so that the pulse decays algebraically, while the timeharmonic
disturbance decays exponentially.
The origin of this phenomenon is the cos θ
2
in the expressions for C
∓
, (3.23).
But cos θ
2
= [1−(¯ c/c)
2
sin
2
θ
0
]
1/2
. If we regard (ω/c) sin θ
0
as a complex vari
able z, then in fact the phenomenon arises fromthe deﬁnition of the branch of the
function (
¯
k
2
−z
2
)
1/2
(=
¯
k cos θ
2
) that appears in the reﬂection and transmission
coefﬁcients. Critical refraction, and reﬂection, therefore, manifest themselves
as branch cuts in the complex plane.
Problem 3.5 Reﬂection of an Acoustic Wave from a Plate
Consider the reﬂection of a time harmonic, plane acoustic wave from a thin
elastic plate. We assume that the ﬂuid has no viscosity so that the incident
acoustic wave excites predominantly ﬂexural motions in the thin plate. More
over, we assume that the ﬂuid belowthe plate is sufﬁciently dense that the elastic
48 3 Reﬂection, Refraction, and Interfacial Waves
Fig. 3.5. A thin elastic plate separates a dense ﬂuid such as water from a tenuous
one such as air, which we model as a vacuum. A time harmonic, plane acoustic wave,
incident from the ﬂuid, strikes the plate and is reﬂected.
plate and the ﬂuid can interact, but that the ﬂuid above the plate is sufﬁciently
tenuous that it can be treated as a vacuum. Water and air, respectively, satisfy
these assumptions. Figure 3.5 indicates the geometry of the problem.
In the ﬂuid
∇
2
ϕ +k
2
ϕ = 0 (3.37)
where ϕ is the velocity potential, k = ω/c, and a time dependence of the form
e
−i ωt
is assumed but suppressed. The pressure p = i ωρϕ and the particle
displacement is −i ωu = ∇ϕ. The density of the ﬂuid is ρ.
The one dimensional, ﬂexural motion of the plate is described by
(∂
1
∂
1
)
2
w −k
4
p
w = p(x
1
, 0)/D, (3.38)
where k
4
p
= ω
2
ρ
p
h/D. The terms ρ
p
, h, and D are the density per unit area,
the thickness, and the elastic constant for the plate, respectively. At x
2
= 0, the
particle displacement is continuous so that −i ωw(x
1
, x
3
) = ∂ϕ/∂x
2
. Note that
phase matching must take place so that the x
1
dependence of wis determined by
the incident wave. Equation (3.38) then gives a boundary condition connecting
p and ϕ at x
2
= 0.
The incident and reﬂected waves are ϕ
0
= A
0
e
i k ˆ p
0
·x
and ϕ
1
= A
1
e
i k ˆ p
1
·x
,
respectively. Find the reﬂection coefﬁcient R(θ
0
). Note that R(θ
0
) = 1 so that
R(θ
0
) = −e
i 2α
. Find α. Speculate about the form of a reﬂected pulse.
3.4 The Rayleigh Wave
We have seen that critical refraction produces a wave in the faster material that
clings to the interface. However, this wave could not exist without being con
tinuously sustained by a wave from the interior. In contrast there are waves that
3.4 The Rayleigh Wave 49
cling to a surface or interface that are selfsustaining. Their particle displace
ment decays exponentially with distance away from the surface or interface, in
the timeharmonic approximation. Such waves are also referred to as surface or
interfacial waves. It is one of the distinctive features of linear elasticity that a
tractionfree surface guides such a wave, called a Rayleigh wave. Both here and
in Chapter 5, it is demonstrated that this latter surface wave, and ones similar
to it, arises from a pole rather than a branch point.
3.4.1 The TimeHarmonic Wave
We consider an elastic halfspace with the x
1
coordinate lying along the surface
and the positive x
2
coordinate pointing into the interior (the reverse of that
shown in Figs. 3.1 and 3.3). We seek a timeharmonic wave whose particle
displacement is inplane and that decays in the positive x
2
direction. We use
a scalar potential ϕ and a single component ψ
3
of the vector potential [the
divergence condition (1.19) is then automatically satisﬁed], and assume that
ϕ = Ce
−βγ
L
x
2
e
iβx
1
, ψ
3
= De
−βγ
T
x
2
e
iβx
1
, (3.39)
where β = ω/c and c is the unknown wavespeed along the surface x
2
= 0. To
satisfy (1.21) and (1.22) for the potentials,
γ
L
=
_
1 −
c
2
c
2
L
_
1/2
, γ
T
=
_
1 −
c
2
c
2
T
_
1/2
. (3.40)
The boundary conditions at x
2
= 0 are τ
22
= τ
21
= 0. To satisfy these boundary
conditions,
_
2 −c
2
/c
2
T
2i γ
T
−2i γ
L
2 −c
2
/c
2
T
_ _
C
D
_
=
_
0
0
_
. (3.41)
Setting the determinant of the matrix to zero, the condition for a nontrivial
solution, gives
_
2 −
_
c
2
_
c
2
T
__
2
−4γ
L
γ
T
= 0. (3.42)
This equation, or an insigniﬁcant modiﬁcation of it, is called the Rayleigh
equation. We examine it momentarily. It has a real positive solution c = c
r
,
giving the wavespeed of the surface wave. Note that γ
L
and γ
T
must be real if
we are to have decay into the interior; thus 0 < c
r
< c
T
. Returning to (3.41)
we ﬁnd that
C
D
=
−2i γ
T
_
2 −c
2
r
_
c
2
T
_. (3.43)
50 3 Reﬂection, Refraction, and Interfacial Waves
With the use of (1.19), the particle displacement components are
u
1
= iβ
r
C
2
_
2e
−β
r
γ
L
x
2
−
_
2 −
c
2
r
c
2
T
_
e
−β
r
γ
T
x
2
_
e
iβ
r
x
1
,
u
2
=
−β
r
C
2γ
T
_
2γ
L
γ
T
e
−β
r
γ
L
x
2
−
_
2 −
c
2
r
c
2
T
_
e
−β
r
γ
T
x
2
_
e
iβ
r
x
1
, (3.44)
where β
r
= ω/c
r
. This then is the form of the time harmonic, Rayleigh wave.
3.4.2 Transient Wave
Previously we have seen that a critically refracted wave becomes a twosided
wave in the time domain. Much the same thing happens when a timeharmonic
Rayleigh wave is mapped into the time domain. In this case the frequency
dependence enters through the terms e
−β
r
γ
L
x
2
and e
−β
r
γ
T
x
2
. Proceeding as we
did in (3.28)–(3.30), the transient displacement components are given by
u
1
(t − x
1
/c
r
, x
2
) =
1
π
_
∞
0
A
0
(ω)
_
2e
−(ω/c
r
)x
2
γ
L
−
_
2 −
c
2
r
c
2
T
_
e
−(ω/c
r
)x
2
γ
T
_
e
−i ω(t −x
1
/c
r
)
dω,
u
2
(t − x
1
/c
r
, x
2
) =
1
π
_
∞
0
A
0
(ω)e
i π/2
γ
−1
T
_
2γ
L
γ
T
e
−(ω/c
r
)x
2
γ
L
−
_
2 −
c
2
r
c
2
T
_
e
−(ω/c
r
)x
2
γ
T
_
e
−i ω(t −x
1
/c
r
)
dω, (3.45)
where A
0
(ω) has the same form as in (3.28)–(3.30) and would be set by the
source. For simplicity we take A
0
(ω) = πC, where C is a constant. Note the
e
i π/2
in u
2
. Performing the integrations gives
u
1
(t − x
1
/c
r
, x
2
) =
2C (x
2
/c
r
)γ
L
_
(x
2
/c
r
)
2
γ
2
L
+(t − x
1
/c
r
)
2
_
−
C
_
2 −
_
c
2
r
/c
2
T
__
[(x
2
/c
r
)γ
T
]
_
(x
2
/c
r
)
2
γ
2
T
+(t − x
1
/c
r
)
2
_ ,
u
2
(t − x
1
/c
r
, x
2
) = −
2C (t − x
1
/c
r
)γ
L
_
(x
2
/c
r
)
2
γ
2
L
+(t − x
1
/c
r
)
2
_
+
C
γ
T
_
2 −
_
c
2
r
/c
2
T
__
(t − x
1
/c
r
)
_
(x
2
/c
r
)
2
γ
2
T
+(t − x
1
/c
r
)
2
_ . (3.46)
3.4 The Rayleigh Wave 51
Because of our choice of A
0
, these pulses are more singular than real ones,
but nevertheless they exhibit the basic features of interest. Note that there is no
leading wavefront. We have a twosided wave. Using our approximations, we
have implicitly assumed that the leading wavefront, which is that of a compres
sional pulse, has already propagated everywhere along the surface. The com
pressional pulse was excited at t →−∞and propagates outward at wavespeed
c
L
, where c
L
> c
T
> c
r
. Also note that the exponential decay of the time
harmonic disturbance has become an algebraic decay. Lastly, the reader, having
noted the presence of the e
i π/2
in (3.45), may ﬁnd it interesting to show that the
u
2
component is essentially the allied function of the u
1
component.
3.4.3 The Rayleigh Function
We are thus left with examining the roots of (3.42). The most important feature
of this equation is that the function on the lefthand side contains two radicals
whose branches must be deﬁned with care. To do so it is more convenient to
work with R(s), a slightly modiﬁed version of (3.42). This function, sometimes
called the Rayleigh function, is given by
R(s) =
2s
2
−s
2
T
2
−4s
2
s
2
−s
2
L
1/2
s
2
−s
2
T
1/2
, (3.47)
where s = 1/c and s
I
= 1/c
I
.
Recall that the denominator of the reﬂection coefﬁcient for the inplane re
ﬂection is A
+
(θ
0
), given in (3.12). Setting s = s
L
sin θ
0
= s
T
sin θ
2
, we ﬁnd
that A
+
(θ
0
) = R(s
L
sin θ
0
)/(s
L
s
T
)
2
. This feature is not accidental. A pole in the
reﬂection coefﬁcient implies that a surface wave exists. The question then is
whether the pole is real or complex, and, if complex, is its imaginary part such
that the surface wave is physically meaningful. The slownesses are ordered as
s > s
T
> s
L
for a surface wave to exist, so that sin θ
0
= s/s
L
> 1, and thus θ
0
must be complex. Ahomogeneous plane wave striking a tractionfree boundary
cannot excite a surface wave.
Problem 3.6 A Surface Wave Supported by an Impedance
Consider a time harmonic disturbance u(x
1
, x
2
) that satisﬁes
∂
1
∂
1
u +∂
2
∂
2
u +k
2
u = 0, (3.48)
where k = ω/c. Further, consider the same general geometry as shown in
Fig. 3.5, but with the plate replaced by an impedance. That is, the boundary
52 3 Reﬂection, Refraction, and Interfacial Waves
condition at x
2
= 0 becomes
∂
2
u = i k Zu, Z = R +i X, (3.49)
with Z the impedance. The impedance may depend upon ω, but not upon the
angle of incidence of the wave. That is, the response at a given point on the
boundary does not depend upon the response of the surrounding points. Taking
as the incident wave u
0
one identical to (3.17), and as the reﬂected wave u
1
one
identical to (3.18), calculate the reﬂection coefﬁcient.
Next show that the surface wave
u = Ce
−bx
2

e
i ax
(3.50)
is also a solution. a and b are positive and real. C is a constant. What restrictions
must R and X satisfy for a surface wave to exist? Find a and b in terms of a Z
that permits a surface wave to exist. For this same Z show that the reﬂection
coefﬁcient has a pole at a particular θ
0
, say θ
0s
. Using this θ
0s
, show that the
reﬂected wave takes the form of the surface wave, (3.50).
3.4.4 Branch Cuts
The Rayleigh function contains two radicals of the form
γ = (α
2
−k
2
)
1/2
, (3.51)
where α is the independent variable and k is known. Almost always, k is a
wavenumber. Thus we may, by imagining that the material in which the wave
is propagating is slightly lossy, set k = k
r
+ i k
i
, where k
i
/k
r
 1. By con
sidering a plane wave propagating in the positive x direction with wavenumber
k, namely, e
i k
r
x
e
−k
i
x
, we see that k
r
and k
i
must both be positive. Looking back
at (3.39) and (3.40), we note that for a surface wave (γ
L
) ≥ 0 and (γ
T
) ≥ 0.
Accordingly, we want to deﬁne the branches of the radical, so that (γ ) ≥ 0 ∀α.
The curves (γ ) = 0 therefore deﬁne the branch cuts.
It is hard to work with γ directly, but somewhat easier to begin by working
4
with γ
2
. Setting α = σ +i τ, we can write γ
2
as
γ
2
=
(σ
2
−τ
2
) −
k
2
r
−k
2
i
+2i (στ −k
r
k
i
). (3.52)
We next partition the α plane, as shown in Fig 3.6, using the curves (γ
2
) = 0
and (γ
2
) = 0. Perturbing α slightly, we ﬁnd that (γ
2
) < 0 between the
4
I ﬁrst learned of this argument from Mittra and Lee (1971).
3.4 The Rayleigh Wave 53
Fig. 3.6. The complex α plane with a sketch of the curves (γ
2
) = 0 and (γ
2
) = 0.
From these curves we select the branch cuts indicated by the wavy overlying line on the
curves (γ
2
) = 0.
two curves (γ
2
) = 0, while (γ
2
) < 0 between the two curves (γ
2
) = 0.
Expressing γ as γ e
i θ
, we note that, if (γ ) > 0, then θ < π/2 and thus
2θ < π. The branch cuts, (γ ) = 0, therefore must lie along the curves
2θ = ±π. That is along (γ
2
) = 0, where (γ
2
) < 0. These are the curves
in Fig. 3.6 overlain with the wavy line. In the limit as k
i
→0, we arrive at the
branch cuts shown in Fig. 3.7.
To investigate this branch, express γ as
γ = (r
1
r
2
)
1/2
cos
θ
1
+θ
2
2
+i sin
θ
1
+θ
2
2
, (3.53)
where
(α −k)
1/2
= r
1/2
1
e
i θ
1
/2
, (α +k)
1/2
= r
1/2
2
e
i θ
2
/2
. (3.54)
Figure 3.7 indicates from what reference the angles are measured. Note that
(γ ) ≥ 0 ∀α, while (γ ) > 0 in the ﬁrst and third quadrants and negative in
the other two. As well, note that to go from the ﬁrst to the second quadrant, and
remain on the same Riemann sheet, we must ﬁrst move into the fourth and then
poke through the gap at the origin into the second.
54 3 Reﬂection, Refraction, and Interfacial Waves
Fig. 3.7. The complex α plane as k
i
→0. (γ ) ≥ 0 ∀α on the Riemann sheet exhibited
in the drawing.
Returning to the Rayleigh function R(s), (3.47), we note that there are two
radicals appearing as the product (s
2
− s
2
L
)
1/2
(s
2
− s
2
T
)
1/2
. Both radicals are
deﬁned as just indicated. However, the product is only discontinuous across
a branch cut connecting s
L
and s
T
, and one connecting −s
L
and −s
T
. These
are the appropriate branch cuts for the product. Hence the Rayleigh function
R(s) is deﬁned as that branch occupying the Riemann sheet (sγ
L
) ≥ 0,
(sγ
T
) ≥ 0 ∀s and cut as just indicated. Recall that s = 1/c. We next seek a
root or roots to R(s) = 0 on this Riemann sheet.
First, note that if s
r
is a root then so is −s
r
. Second, at s = s
L
and at
s = s
T
, R(s) > 0 but as s → ∞along the positive real s axis, R(s) becomes
negative. Thus a real root s
r
, with s
T
< s
r
< ∞, exists. By symmetry a
root −s
r
also exists. Third, does R(s) have any other roots on this Riemann
sheet? The answer is no. This can be proven conclusively by using a theorem
sometimes called the argument principle (Ablowitz and Fokas, 1997). This
theorem allows one to systematically calculate the number of poles and zeros
of a function by performing a contour integral of its logarithmic derivative.
Achenbach (1973) does this in some detail, though the original argument may
be found in Cagniard (1962). Lastly, do the roots on the other Reimann sheets
ever manifest themselves? Yes, they do when they lie close to a branch cut.
This point is discussed in Aki and Richards (1980) in their description of the
reﬂection of spherical waves from a tractionfree surface.
While knowing that only the Rayleigh slownesses ±s
r
exist on the Reimann
sheet of interest is important, actuallysolvingfor s
r
is less so. It canbe calculated
References 55
to good accuracy by using the formula
κ
r
= (0.86 +1.12ν)/(1 +ν) (3.55)
The term κ
r
= s
T
/s
r
and ν is Poisson’s ratio. For many materials κ
r
is close to
0.9.
The branches of (2.23) and (2.31) were selected as just indicated. However,
note that k
1
= i γ in (2.23) and k
3
= i γ in (2.31). In particular, in the case
of (2.23), the complex k
2
plane is cut so that (k
1
) > 0 ∀k
1
, corresponding
exactly with the branch of γ just selected. Further, the contour in (2.24) starts
at −∞ in the second quardant, passes above the branch cut, pokes into the
fourth quadrant, passes below the branch cut, and ends at +∞. The radicals we
encounter will often present themselves in the formξ = (k
2
−α
2
)
1/2
, rather than
as that in (3.51). The two are connected by ξ = i γ . Only when considering the
Cagniard–deHoop integration technique in Section 5.1 do we select a different
branch.
References
Ablowitz, M.J. and Fokas, A.S. 1997. Complex Variables, pp. 259–265. New York:
Cambridge.
Achenbach, J.D. 1973. Wave Propagation in Elastic Solids, pp. 189–191. Amsterdam:
NorthHolland.
Aki, K. and Richards, P.G. 1980. Quantitative Seismology, Theory and Methods,
Vol. 1, pp. 319–333. San Francisco: Freeman.
Auld, B.A. 1990. Acoustic Fields and Waves in Solids, 2nd ed., Vol. 2, pp. 1–57.
Malabar, FL: Krieger.
Cagniard, I. 1962. Reﬂection and Refraction of Progressive Seismic Waves,
pp. 42–49. Translated and revised by E.A. Flinn and C.H. Dix. New York:
McGrawHill.
Friedlander, F.G. 1974. On the total reﬂection of plane waves. Quart. J. Mech. Appl.
Math. 1: 379–383.
Mittra, R. and Lee, S.W. 1971. Analytical Technique in the Theory of Guided Waves,
pp. 20–23. New York: Macmillan.
Titchmarsh, E.C. 1948. Introduction to the Theory of Fourier Integrals, 2nd ed.,
pp. 119–121, 147–148, and elsewhere. Oxford: Clarendon Press.
4
Green’s Tensor and Integral Representations
Synopsis
In Chapter 4 we discuss the formulation of integral representations of solutions
to rather general problems in elasticwave propagation. Two constructions are
used: the reciprocity identity and the Green’s tensor for a full space. For these
representations for an inﬁnite domain to be derived, the principle of limiting
absorption is introduced. This is needed for timeharmonic problems because
the disturbance, in a sense, has been going on forever, resulting in no initial
wavefront being present. Moreover, we establish a uniqueness result, indicating
as we do so both the role of the principle of limiting absorption and that of
specifying an edge condition. The chapter closes with an example that uses
these ideas to develop an integral representation for the scattering of an acoustic
wave by an elastic inclusion.
4.1 Introduction
In Chapter 2 we moved away fromdiscussing plane waves to an introduction of
planewave spectral representations in Section 2.3. This allowed us to discuss
more general waveﬁelds and to understand their propagation characteristics
in terms of those of plane waves. We continue with this general theme, but
construct, in this chapter, both far more general representations and ones in
physical space rather than in wavenumber space. Though we make limited use
of it in the chapters that follow, this material is very important because it is the
basis for formulating elasticwave problems in a form suitable to be analyzed
numerically. Ageneral survey of many useful representation results is provided
by deHoop (1995).
When the waveﬁeld is time dependent, it is straightforward (straightfor
ward is not a synonym for easy) to seek its representations (Friedlander, 1958;
Achenbach, 1973; Hudson, 1980). We must work in a fourdimensional space,
56
4.2 Reciprocity 57
the fourth dimension being time. Nevertheless, the waveﬁeld propagates out
ward from its sources at a ﬁnite speed and thus for a ﬁnite time it occupies a
region that is bounded. However, a representation in this form is less useful
than it might seem because both the source and receiver of the waveﬁeld have
their own frequency behaviors, and to incorporate those effects in an overall
propagation model the Fourier component of the waveﬁeld rather than, say,
its timedependent response to a sudden impulse is of greater use. Thus we
consider almost exclusively timeharmonic waveﬁelds.
When working with a timeharmonic waveﬁeld, though we need only work
with regions in threedimensional space, we have imagined that the waveﬁeld
was excited at the time minus inﬁnity, and it has thus, even for a ﬁnite positive
time, ﬁlled all of space. In constructing representations for various waveﬁelds,
we must unambiguously identify waves that are outgoing from their source of
excitation. Moreover, at some stage we must send the outer surfaces, over which
integrals are being taken, to inﬁnity and must be assured that these integrals
either go to zero or a ﬁnite, unambiguously identiﬁed value. In this chapter we
use the principle of limiting absorption as the technical device to achieve these
outcomes.
4.2 Reciprocity
Proposition 4.1. In a bounded region R
x
with surface ∂R
x
, the equations
of motion in the timeharmonic approximation for two reciprocating elastic
waveﬁelds, indicated by the superscripts 1 and 2, are
∂
j
τ
1,2
j i
+ρf
1,2
i
+ρω
2
u
1,2
i
= 0. (4.1)
The tractions on ∂R
x
are t
1,2
i
= τ
1,2
j i
ˆ n
j
. The unit normal ˆ n to ∂R
x
is outward.
Then
R
x
f
2
i
u
1
i
− f
1
i
u
2
i
ρdV =
∂R
x
u
2
i
t
1
i
−u
1
i
t
2
i
dS. (4.2)
Proof. We take the scalar product of each equation in (4.1) with the particle
displacement of its reciprocating waveﬁeld and subtract. This gives
ρ
f
2
i
u
1
i
− f
1
i
u
2
i
= −u
1
i
∂
j
τ
2
j i
+u
2
i
∂
j
τ
1
j i
. (4.3)
Using the relation between stress and strain, (1.3), we note that τ
1
j i
∂
j
u
2
i
=
τ
2
j i
∂
j
u
1
i
so that the righthand side of (4.3) equals ∂
j
(τ
1
i j
u
1
i
−τ
2
i j
u
1
i
); (4.2) follows
after integration.
58 4 Green’s Tensor and Integral Representations
There is one special case of (4.2) of interest. Let us assume that the integral
over ∂ R
x
vanishes and that f
1,2
i
= a
1,2
i
δ(x − x
1,2
), where the a
1,2
i
are constant
vectors. Then the reciprocity statement becomes a
2
i
u
1
i
(x
2
) = a
1
i
u
2
i
(x
1
). This is
the origin of the careless statement “the source and receiver can be interchanged
by reciprocity.”
4.3 Green’s Tensor
Inthe propositiontofollowwe shall use the threedimensional Fourier transform
pair given by
∗
u(k) =
∞
−∞
u(x)e
−i k·x
dx, (4.4)
u(x) =
1
(2π)
3
∞
−∞
∗
u(k)e
i k·x
dk. (4.5)
This is a straightforward generalization of (1.43) and (1.44). The differen
tial in physical space dx =dx
1
dx
2
dx
3
and that in the transform space dk =
dk
1
dk
2
dk
3
.
Proposition 4.2. The particle displacement u
G
i
is the solution to
(λ +µ)∂
i
∂
k
u
G
k
+µ∂
j
∂
j
u
G
i
+ρω
2
u
G
i
= −ρ ˆ a
i
δ(x), (4.6)
subject to the condition that, as x → ∞, the solution represent an outgoing
waveﬁeld. x = x. Here ˆ a is a constant unit vector giving the direction of the
point force at the origin. The Green’s tensor
1
for a full space, u
G
i k
, is related to
u
G
i
by u
G
i
= ˆ a
k
u
G
i k
, with u
G
i k
given by
u
G
i k
=
1
k
2
T
c
2
T
∂
i
∂
k
[−G(k
L
x) + G(k
T
x)] +k
2
T
δ
i k
G(k
T
x)
, (4.7)
where
G(k
I
x) = (1/4πx)e
i k
I
x
, I = L, T. (4.8)
Note that we may replace x with the more general vector x − x
, where x
is
the position vector of the point of action of the delta function.
1
It is possibly better to describe the pair (u
G
i k
, τ
G
i j k
) as a Green’s state.
4.3 Green’s Tensor 59
Proof.
2
Fourier transforming (4.6) in all three spatial variables and solving
the resulting algebraic equations give
∗
u
G
i
= −
1
c
2
T
k
2
L
−k
2
ˆ a
i
+
1 −c
2
T
/c
2
L
k
i
k
k
ˆ a
k
k
2
L
−k
2
k
2
T
−k
2
, (4.9)
where k
2
= k· k. From the deﬁnition of u
G
i j
, and with some rewriting, we arrive
at
∗
u
G
i k
=
1
c
2
T
k
i
k
k
k
2
T
1
k
2
−k
2
L
−
1
k
2
−k
2
T
+
δ
i k
k
2
−k
2
T
. (4.10)
Note that ∂
i
in physical space transforms to i k
i
in transform space and vice
versa. Therefore, to transform (4.10) to physical space, all that we have to do
is evaluate integrals of the form
I
I
=
1
(2π)
3
∞
−∞
e
i k·x
k
2
−k
2
I
dk. (4.11)
To evaluate (4.11) we introduce, in transform space, a spherical coordinate
system (k, ξ, ν), where the azimuthal axis is chosen in the direction x and ξ
is the azimuthal angle. Note that this is identical, apart from the change in
symbols, to the coordinate transformation (2.33) and (2.37) used to construct
an angular spectrum representation of a spherical wave. The transformation is
deﬁned as
k
1
= k sin ξ cos ν, k
2
= k sin ξ sin ν, k
3
= k cos ξ. (4.12)
The integral (4.11) can then be expressed as
I
I
=
1
(2π)
3
∞
0
π
0
2π
0
k
2
sin ξ
k
2
−k
2
I
e
i kx cos ξ
dkdξdν. (4.13)
The integrations over the angles are readily done, and we are left with evaluating
I
I
=
−i
(2π)
2
x
∞
−∞
ke
i kx
k
2
−k
2
I
dk. (4.14)
Recall that e
i (k
I
x−ωt )
is an outgoing wave. Thus to satisfy the condition that
waves be outgoing as x →∞, the contour of the integral must pass above the
2
Contrast how we have used a threedimensional transform in this problem, but have used a two
dimensional one for the very similar problem of constructing a planewave representation of a
spherical wave in (2.36).
60 4 Green’s Tensor and Integral Representations
pole at −k
I
but below that at k
I
. Then closing the contour in the upper half of
the k plane, where the integral is convergent, gives an outgoing wave and hence
gives (4.8) with G(k
I
x) = I
I
. Thus (4.7) follows.
The corresponding Green’s stress τ
G
i j k
is calculated by using
τ
G
i j k
= c
i jlm
∂
l
u
G
mk
, (4.15)
where
c
i jlm
= λδ
i j
δ
lm
+µ(δ
il
δ
j m
+δ
i m
δ
jl
). (4.16)
These last two equations are simply restatements of (1.3).
Problem 4.1 TwoDimensional Green’s Function
In this problem we ask the reader to calculate the time harmonic, antiplane
shear Green’s function, u
G
3
. In fact it will turn out to be identical to the angu
lar spectrum representation of a cylindrical wave, cited in Section 2.3.3. The
equation to be solved
3
is
∂
α
∂
α
u
G
3
+k
2
u
G
3
= −δ(x), (4.17)
where x has components (x
1
, x
2
). At inﬁnity u
G
3
represents an outgoing wave.
Show that u
G
3
can be expressed as
u
G
3
(x) =
i
4π
∞
−∞
e
i (k
1
x
1
+k
2
x
2
)
dk
1
k
2
. (4.18)
How are the branches of the radical k
2
deﬁned? Next introduce the transfor
mations, k
1
= k
T
cos ξ, x
1
= ρ cos θ, x
2
 = ρ sin θ, and show that (4.18)
reduces to
u
G
3
(ρ) = −
i
4π
C
e
i k
T
ρ cos(θ−ξ)
dξ. (4.19)
Most importantly show that C is a contour beginning near π − i ∞ and end
ing near i ∞. The integral representation of the Hankel function H
(1)
0
(kρ)
(Sommerfeld, 1964b) is such that u
G
3
= (i /4)H
(1)
0
(kρ).
3
The righthand side of this equation should be multiplied by c
−2
, if the reader wants its dimen
sional form to be identical to (4.6).
4.3 Green’s Tensor 61
4.3.1 Notes
1. For future work we shall need the farﬁeld results. The farﬁeld is that
region for which k
I
x
 k
I
x and k
I
x 1. Approximating x − x
 by
using the law of cosines gives
x − x
 ∼ x −( ˆ p · x
), ˆ p = x/x. (4.20)
With this approximation, (4.7), with x replaced by x−x
, splits into two parts,
namely
u
G
i k
∼ u
GL
i k
+u
GT
i k
, (4.21)
where
u
GL
i k
=
1
c
2
L
ˆ p
i
ˆ p
k
e
i k
L
x
4πx
e
−i k
L
ˆ p·x
, (4.22)
u
GT
i k
=
1
c
2
T
(δ
i k
− ˆ p
i
ˆ p
k
)
e
i k
T
x
4πx
e
−i k
T
ˆ p·x
. (4.23)
Further, using (4.15) and (4.16), we have
τ
G
j i k
ˆ p
j
∼ τ
GL
j i k
ˆ p
j
+τ
GT
j i k
ˆ p
j
, (4.24)
where
τ
GL
j i k
ˆ p
j
= i k
L
(λ +2µ)u
GL
i k
, (4.25)
τ
GT
j i k
ˆ p
j
= i k
T
µu
GT
i k
. (4.26)
Each of the above expansions is carried out only to O[(k
I
x)
−1
] in the amplitude
while the additional term (i k
I
ˆ p · x
) is retained in the exponential terms. The
character of a waveﬁeld is more sharply determined by its phase than by its
amplitude. As before, I = L or T.
2. There are a number of very localized sources that can be constructed from
point forces and their derivatives (Hudson, 1980). Here we consider one such
construction, the center of compression. Consider a special (threedimensional)
body force, namely
f = −c
2
L
F(t )∂
x
δ(x)
4πx
2
ˆ p, (4.27)
where x
is set to zero, x = x, and ˆ p is deﬁned in (4.20).
62 4 Green’s Tensor and Integral Representations
The scalar potential ϕ satisﬁes
1
x
2
∂
x
(x
2
∂
x
ϕ) −
1
c
2
L
∂
t
∂
t
ϕ = F(t )
δ(x)
4πx
2
(4.28)
and the vector potential ψ = 0. This particular source is useful because it excites
only compressional waves. The solution to (4.28), satisfying the condition that
the wave be outgoing from its source, is
ϕ = −
1
4πx
F
t −
x
c
L
. (4.29)
The center of compression can be thought of as a small uniformly pressurized
cavity or as three dipoles without moment.
Problem 4.2 A Causal Green’s Function
Problem 1. Show that (4.29) is the solution to (4.28), subject to the condi
tion that the wave be outgoing from the source. Here (4.29) is the freespace
causal Green’s function for (4.28) if F(t ) = δ(t ). Begin by showing that xϕ =
f (t − x/c
L
) is a solution, for x = 0. Then integrate the equation of motion,
(4.28) over a small sphere of radius , centered at the origin, and, taking the
limit as →0, identify f (t ).
Problem 2. Equation (4.27) is a threedimensional body force. In two di
mensions, (x
1
, x
2
), the termin brackets becomes δ(x)/(2πx). Find the response
ϕ for a twodimensional center of compression or line of compression.
4.4 Principle of Limiting Absorption
Figure 4.1 shows a typical region R
x
within which we shall work. It is a large
spherical region with radius x. Both the bounded subregions contained within
R
x
, S and R, will be sources of waves. Two questions arise. (1) How do we
determine unambiguously which waves are outgoing fromthese sources? (2) In
the arguments that follow we shall ask that the integrals over ∂R
x
vanish or
approach some ﬁnite known value as x → ∞. How do we ensure that this
happens? As we indicated in Section 4.1, this issue does not arise in the time
dependent case because any outgoing waveﬁeld will propagate toward inﬁnity
at a ﬁnite speed and, therefore, for a ﬁnite t the surface ∂R
x
, as x →∞, will
eventually pass into a region that the waveﬁeld has not reached. However, in
4.4 Principle of Limiting Absorption 63
Fig. 4.1. The large spherical region R
x
, with surface ∂R
x
, has a radius x = x. In
several of the arguments x → ∞. Within are two bounded subregions, S and R. The
ﬁrst contains sources that excite an incident waveﬁeld, while the second is a scatterer,
with surface ∂R. The unit normal ˆ n points out from R
x
\R.
the timeharmonic case, we are imagining that the waveﬁeld started abruptly
at t = −∞and thus it now ﬁlls all space. Stated somewhat differently, there is
no initial wavefront because we have imposed no initial conditions.
We solve this problem by demanding that ω = ω
0
+i , where ω
0
and > 0
are real. Hence, lim
t →−∞
f (x)e
i (kx−ωt )
= 0, for x ﬁxed, so that the waveﬁeld
is forced to zero. This gives the missing initial conditions. Though, as t →∞
the waveﬁeld becomes unbounded, t will always be taken as ﬁnite and, in a
timeharmonic problem, often never appears explicitly in the calculation. It in
no sense indicates a temporal instability. Moreover, note that the wavenumber
k
I
= ω/c
I
. That is, k
I
= k
I 0
+ i /c
I
, where k
I 0
= ω
0
/c
I
. Therefore, for t
ﬁxed, lim
x→∞
f (x)e
i (k
I
x−ωt )
= 0 provided f (x) remains bounded. That is, if
we want the waveﬁeld to go to zero as x → ∞, we must use the exponential
term e
i k
I
x
in combination with e
−i ωt
, and reject e
−i k
I
x
as a possible solution.
In Proposition 4.5 we shall ﬁnd that this choice of ω and hence of k is needed
to ensure a unique solution. Once we have completed our calculation we may
invoke analytic continuation
4
and, by letting → 0, recover the result for the
case of real ω. This device of setting ω = ω
0
+i to determine which waveﬁelds
are outgoing as x → ∞ and hence to select which particular solutions to a
problem give outgoing waves is called the principle of limiting absorption.
Note that the sign conventions leading to the evaluation of the contour integral,
(4.14), and all previous contour integrals, are consistent with this principle. In
particular we have implicitly used this principle when evaluating the contour
integral, (4.14).
4
Here we are invoking the law of permanence of functional equations (Hille, 1973).
64 4 Green’s Tensor and Integral Representations
4.5 Integral Representation: A Source Problem
Proposition 4.3. Consider the region R
x
, shown in Fig. 4.1, with the region
R absent, but S present. S is a bounded subregion containing time harmonic,
body forces per unit mass, f
i
, that excite the waveﬁeld u
i
in R
x
; u
i
is the
solution to
(λ +µ)∂
i
∂
k
u
k
+µ∂
j
∂
j
u
i
+ρω
2
u
i
= −ρf
i
(x), f
i
(x) = 0, x / ∈ S.
(4.30)
The waveﬁeld satisﬁes the principle of limiting absorption as x → ∞. After
∂R
x
is sent to inﬁnity, the solution u
i
can be represented as
u
k
(x) =
S
f
i
(x
)u
G
i k
(x − x
) dV(x
), (4.31)
where u
G
i k
is given by (4.7) and also satisﬁes the principle of limiting absorption.
Proof. Starting with the bounded region R
x
, we select as reciprocating wave
ﬁeld1the triple ( f
i
, u
i
, τ
i j
) andas 2the triple [δ(x−x
) ˆ a
i
, u
G
i k
(x−x
) ˆ a
k
, τ
G
i j k
(x−
x
) ˆ a
k
], where u
G
i k
is given by (4.7). Then (4.2) becomes
R
x
δ(x − x
)u
i
(x) ˆ a
i
− f
i
(x)u
G
i k
(x − x
) ˆ a
k
ρdV(x) = I
k
ˆ a
k
, (4.32)
where
I
k
=
∂R
x
u
G
i k
(x − x
)τ
j i
(x) −u
i
(x)τ
G
j i k
(x − x
)
ˆ n
j
dS(x). (4.33)
We shall subsequently show that lim
x→∞
I
k
= 0. Therefore (4.32) reduces to
u
i
(x
) ˆ a
i
=
S
f
i
(x)u
G
i k
(x − x
) ˆ a
k
dV(x). (4.34)
By inspection, u
G
i k
(x − x
) = u
G
i k
(x
− x). Hence we are lead to (4.31), after a
relabeling of the independent and integration variables.
Returning to the question of how I
k
behaves, we use the asymptotic results
of (4.22)–(4.26) to show that, as x →∞,
I
k
∼
1
c
2
L
e
i k
L
x
4πx
∂R
x
[τ
j i
ˆ n
j
−i k
L
(λ +2µ)u
i
] ˆ n
i
ˆ n
k
e
−i k
L
ˆ n·x
dS(x
)
+
1
c
2
T
e
i k
T
x
4πx
∂R
x
[τ
j i
ˆ n
j
−i k
T
µu
i
](δ
i k
− ˆ n
i
ˆ n
k
)e
−i k
T
ˆ n·x
dS(x
). (4.35)
4.6 Integral Representation: A Scattering Problem 65
Note that we have relabeled the variables from those used in (4.33). When the
principle of limiting absorption is invoked, lim
x→∞
I
k
= 0.
4.5.1 Notes
1. When thinking through the argument just given, note that two issues have
been settled by using the principle of limiting absorption. First, reexamining the
argument leading to the Green’s tensor, recall that we demanded that the waves
be outgoing. In place of the phrase outgoing, we now ask that the waveﬁeld
satisfy the principle of limiting absorption and thus approach zero as x →∞.
Therefore we must choose the solution that behaves as e
i kx
and not e
−i kx
. The
principle thus serves to determine the solution uniquely. Second, through its
use, integrals such as I
k
can be shown to approach zero without having to make
detailed asymptotic estimates of how they behave as x →∞.
2. As an alternative to using the principle of limiting absorption, we could
have demanded that
lim
x→∞
∂R
x
[τ
i j
ˆ n
j
−i k
L
(λ +2µ) u
i
] ˆ n
i

2
dS(x
) = 0,
lim
x→∞
∂R
x
[τ
i j
ˆ n
j
−i k
T
µu
i
](δ
i k
− ˆ n
i
ˆ n
k
)
2
dS(x
) = 0. (4.36)
This is one way of stating the elastodynamic radiation conditions. Note that
they would only be satisﬁed by an outgoing wave. That is one having the form
f (x)e
i k
I
x
. From this it is possible to show that lim
x→∞
I
k
= 0 (Achenbach
et al., 1982).
3. Note that (4.31) asymptotically approaches the form
u
i
∼ A
i
(x)
e
i k
L
x
k
L
x
+ B
i
(x)
e
i k
T
x
k
T
x
, x →∞, (4.37)
provided S is bounded. In this case S is said to be a compact source because
the support for f
i
(x) is compact or bounded. Irrespective of the geometry of
the source, provided it is compact, the waveﬁeld must evolve into two separate
spherical waves with radiation patterns given by A
i
and B
i
.
4.6 Integral Representation: A Scattering Problem
We continue to consider the region R
x
shown in Fig. 4.1. Within it we imagine
that the total waveﬁeld u
t
i
= u
i
i
+ u
i
, where u
i
i
is the waveﬁeld excited by the
sources in S in the absence of the subregion R and u
i
is that scattered by R.
We next perform a slight of hand whereby we imagine that the subregion S
66 4 Green’s Tensor and Integral Representations
is removed from R
x
so that the scattered waveﬁeld u
i
does not interact with
the sources in S. Lastly, we consider R to be an empty cavity and ask that the
traction vanish on ∂R. We now formulate a boundaryvalue problem for u
i
,
with t
i
= −t
i
i
on ∂R acting as a source. It is useful at this point to deﬁne R
as an open region so that it does not contain its boundary points ∂R. Then we
deﬁne the region R
x
\R as that contained in R
x
excluding R, but containing
the boundary points ∂R.
Proposition 4.4. Consider the region R
x
shown in Fig. 4.1 with the region
S absent, but R present. Within R
x
\R the timeharmonic waveﬁeld u
i
is a
solution to (4.30), with the righthand side set to zero. On ∂R, u
i
is unknown,
but τ
j i
ˆ n
j
= −τ
i
j i
ˆ n
j
, where τ
i
j i
is known. Moreover, u
i
satisﬁes the principle of
limiting absorption as x → ∞. It follows then that as the radius x → ∞, u
i
can be represented as
u
k
(x) = −
1
ρ
∂R
u
G
i k
(x − x
)τ
i
j i
(x
) +u
i
(x
)τ
G
j i k
(x − x
)
ˆ n
j
dS(x
). (4.38)
The prime given τ
G
j i k
indicates that the x
is differentiated when τ
G
j i k
is calculated
by using (4.15).
Proof. Select as reciprocating waveﬁeld 1 the triple (0, u
i
, τ
i j
) and as 2 the
triple [ ˆ a
i
δ(x − x
), u
G
i k
(x − x
) ˆ a
k
, τ
G
i j k
(x − x
) ˆ a
k
]. Using (4.2) and proceeding
as we did with Proposition 4.3, we arrive at
ρu
i
(x
) ˆ a
i
=
∂R
u
G
i k
τ
j i
−u
i
τ
G
j i k
ˆ a
k
ˆ n
j
dS(x) + I
k
ˆ a
k
. (4.39)
Using the boundary condition on ∂R, relabeling the independent and integration
variables and noting that again lim
x→∞
I
k
= 0, by the principle of limiting
absorption, we obtain (4.38).
4.6.1 Notes
1. Note, in (4.38), that τ
G
j i k
is calculated from u
G
i k
(x −x
) by differentiating
with respect to x
, the second argument of the combination (x − x
).
2. Note that (4.38) is not a solution to the boundaryvalue problembecause
the integral contains the unknown u
i
evaluated on the boundary. In fact we have
used the incorrect Green’s tensor for this problem. We should have calculated a
Green’s tensor for which τ
G
j i k
ˆ n
j
= 0 on ∂R. This is usually far from easy to do.
Nevertheless, (4.38) forms the starting point for many useful approximations.
4.6 Integral Representation: A Scattering Problem 67
3. The integral representation for u
i
(x) can be used to construct an integral
equation, as Problem4.3 suggests. In forming the integral equation care must be
taken because, as x approaches the surface, the Green’s tensors become singular.
Kevorkian (1993) gives an introduction to how to examine the consequences
of these singularities within his discussion of integral equations in potential
theory. Integral equations are sometimes easier to solve numerically than are
differential ones, and their solution lends insight that direct numerical methods
often do not. One book describing the numerical solution of integral equations
is that by Delves and Mohamed (1985).
Problem 4.3 Integral Equation Problems
These are two dimensional, antiplane shear problems designed to ask the
reader to derive directly, in this simpler context, several of the formulas given
previously. In both the problems that follow, the elastic halfspace is described
by {(x
1
, x
2
) −∞< x
1
< ∞, −∞< x
2
≤ 0}.
Problem1. An elastic halfspace is clamped at the surface x
2
= 0 by a rigid
strip over x
1
 < a. This forces the particle displacement at the surface to go to
zero there. For x
1
 > a the surface is free of traction. An antiplane shear wave
u
i
3
= Ae
i kx
2
(4.40)
is normally incident to the rigid strip and adjacent free surface. Determine an
integral equation for the scattered waveﬁeld.
Problem2. Nowassume that the region x
1
 > a is clamped by rigid sheets,
while within a slot x
1
 < a the surface is free of traction. For the same incident
wave, determine an integral equation for the scattered waveﬁeld.
Outline of the Solution to Problem 1
1. Divide the total particle displacement u
t
3
into the sum of two parts, namely,
u
t
3
= u
i
3
+ u
s
3
, where u
s
3
is the scattered waveﬁeld. Then divide this latter
waveﬁeld again into two parts, namely, u
s
3
= u
r
3
+u
3
, where u
r
3
is the plane
wave reﬂected fromthe surface with no strip present and u
3
is that caused by
the presence of the rigid strip. Showthat u
3
satisﬁes the timeharmonic wave
equation and the following boundary conditions along x
2
= 0: u
3
= −2A
on x
1
 < a, and ∂
2
u
3
= 0 on x
1
 > a. How does u
3
behave at ∞? At ±a?
2. Use the method of images to ﬁnd the antiplane shear, Green’s function
u
G
3
(x
1
, x
2
x
1
, x
2
), for the halfspace satisfying a homogeneous boundary
68 4 Green’s Tensor and Integral Representations
condition ∂
2
u
G
3
= 0 at x
2
= 0, and the principle of limiting absorption as
x → ∞. Note that what is needed is the Green’s function for this prob
lem. In contrast, the Green’s tensor derived in Proposition 4.2 and used in
Propositions 4.3 and 4.4 was that for a region without boundaries and not
the correct Green’s tensor for the scattering problem just discussed.
Hint. The reader may ﬁnd the results of Problem 4.1 useful.
3. Derive the reciprocity identity for two reciprocating, antiplane shear wave
ﬁelds. Taking as one reciprocating waveﬁeld the Green’s function and as the
other u
3
, show that
u
3
(x
1
, x
2
) = −
a
−a
u
G
3
(x
1
, x
2
x
1
, 0)I (x
1
)dx
1
, (4.41)
where I (x
1
) = ∂
2
u
3
(x
1
, 0) and the prime indicates that the derivative is taken
with respect to x
2
.
4. Use the boundary conditions satisﬁed by u
3
to derive the integral equation
2A =
i
2
a
−a
H
(1)
0
(kx
1
− x
1
)I (x
1
)dx
1
. (4.42)
Note that I (x
1
) is the unknown to be solved for. After I (x
1
) is found, the
representation (4.41) is used to calculate u
3
throughout the region of inter
est. Describe the nature of the singularities, in [−a, a], possessed by the
integrand.
Repeat the above steps for the second problem, making appropriate changes
where necessary. In particular use a Green’s function that satisﬁes u
G
3
= 0
at x
2
= 0. Note that on this occasion the integrand is more singular than in
the previous case. While one can work with singular integral equations, this
particular equation can be reduced to one that is no more singular than that of
the ﬁrst problem (Sommerfeld, 1964a).
4.7 Uniqueness in an Unbounded Region
4.7.1 No Edges
Proposition 4.5. Consider the region R
x
shown in Fig. 4.1. Consider two time
harmonic waveﬁelds (u
1
i
, τ
1
i j
) and (u
2
i
, τ
2
i j
) that satisfy
∂
j
τ
1,2
j i
+ρf
i
+ρω
2
u
1,2
i
= 0. (4.43)
in R
x
\R. The force f
i
= 0 for x ∈ S but is zero elsewhere; ρ, λ, and µ are
real parameters. The tractions on ∂R are equal; that is, τ
1
j i
ˆ n
j
= τ
2
j i
ˆ n
j
. The
4.7 Uniqueness in an Unbounded Region 69
waveﬁelds are sufﬁciently continuous on ∂ R that Gauss’ theorem may be used.
In accord with the principle of limiting absorption, ω = ω
0
+ i so that both
waveﬁelds go to zero as x →∞. Then the two waveﬁelds are identically equal
in R
x
\R.
Proof. Set u = u
1
−u
2
and τ
i j
= τ
1
i j
−τ
2
i j
. Then (u
i
, τ
i j
) satisﬁes
∂
k
τ
kl
+ρω
2
u
l
= 0 (4.44)
in R
x
\R. Moreover, (u
∗
i
, τ
∗
i j
) satisﬁes the same equation with ω
2
replaced by
(ω
2
)
∗
. As was done previously, the superscript asterisk indicates the complex
conjugate. Take the scalar product of (4.44) with u
∗
i
and the scalar product of
the complex conjugate of (4.44) with u
i
. Subtracting one from the other and
using (1.3), the stress–strain relation, gives
[(ω
2
)
∗
−ω
2
]ρu
∗
l
u
l
= ∂
j
(τ
j i
u
∗
i
−τ
∗
j i
u
i
). (4.45)
Integrating this result gives
[(ω
2
)
∗
−ω
2
]
R
x
\R
ρu
∗
l
u
l
dV =
∂R∪∂R
x
(τ
j i
u
∗
i
−τ
∗
j i
u
i
) ˆ n
j
dS. (4.46)
The righthand integral is zero because τ
i j
ˆ n
j
= 0 on ∂R and the integral over
∂R
x
goes to zero as x →∞, by the principle of limiting absorption. Therefore
the righthand side equals zero. In addition, ω
∗
= ω, by the principle of limiting
absorption, so that the lefthand integral must equal zero. The integrand is either
everywhere zero or positive. If positive somewhere, then the integral cannot be
zero, giving a contradiction. Therefore u
i
and hence τ
i j
must be zero throughout
R
x
\R
5
.
4.7.2 Edge Conditions
To prove uniqueness (or, in fact, to prove Proposition 4.1 and hence all the
subsequent propositions), it was essential that we be able to use Gauss’ theorem.
Thus the waveﬁelds must have a certain measure of continuity on the surfaces.
In the simpler proofs one asks that the partial derivatives be continuous apart
from ﬁnite jumps (Courant and John, 1989; Kellogg, 1970). In Fig. 4.2 a sketch
of the region R, in cross section, with a lancetshaped singularity whose edge
is perpendicular to the page, is given. The waveﬁeld is singular there. The
5
Note that the analytic continuation of a function is unique. Hence, when the unique solution is
determined by setting ω = ω
0
+i , that uniqueness is not lost by taking the limit →0.
70 4 Green’s Tensor and Integral Representations
Fig. 4.2. The subregion Ris drawn, in cross section, with a lancetshaped singularity
whose edge is perpendicular to the page. At such an edge the waveﬁeld is singular. The
edge is surrounded by a cylindrical region R
with a radius . Both regions are contained
within R
x
.
particle displacement is usually bounded even continuous, but its derivatives
are singular. Thus Gauss’ theorem is inapplicable near this edge without a cer
tain amount of care being taken.
To investigate what happens at this edge, we surround it with a region
R
whose surface is ∂R
. This region does not intrude into (is disjoint from)
R. Again it is useful to deﬁne R and R
as open regions that do not contain
their boundary points. Gauss’ theorem may be applied to the region Q =
R
x
\(R∪R
). The reader should look back at Fig. 4.1 and thereby recall that R
and R
are both contained within R
x
. The surface ∂Q = ∂R
x
∪ ∂(R ∪ R
).
Now (4.46) is written as
[(ω
2
)
∗
−ω
2
]
Q
ρu
∗
l
u
l
dV =
∂Q
(τ
j i
u
∗
i
−τ
∗
j i
u
i
) ˆ n
j
dS. (4.47)
To complete the uniqueness proof we must ask additionally that
lim
→0
∂R
τ
j i
u
∗
i
ˆ n
j
dS
= 0. (4.48)
Then the righthand side of (4.47) goes to zero subject to (4.48) being satisﬁed.
And the uniqueness argument proceeds as before. A condition such as this is
called an edge condition. However, it is more usual to enforce the edge condition
by asserting the nature of the singularity that will be permitted at the edge so
that (4.48) is satisﬁed. It is important to realize that this singularity must be
speciﬁed before the scattering problemis solved and is as important in achieving
a unique solution as is specifying a boundary condition or invoking the principle
of limiting absorption.
Next we shall work through a speciﬁc case. However, for an edge, an ele
ment of surface area S
of ∂R
is O(). Moreover, typically u
i
∼ O(
α
)
and τ
i j
∼ O(
α−1
), where α < 1. Therefore, to satisfy (4.48), we demand
α > 0.
4.7 Uniqueness in an Unbounded Region 71
Fig. 4.3. The subregion Rhas been collapsed to a inﬁnitesimally thin, semiinﬁnite slit
or crack and ∂R becomes its two surfaces, upper and lower. The edge is perpendicular
to the page, extending inward and outward to inﬁnity. It is surrounded by a cylindrical
region R
with a radius that is smaller than a wavelength.
4.7.3 An Inner Expansion
We nowconsider a speciﬁc case of a scatterer with an edge. Figure 4.3 shows the
region R collapsed to a inﬁnitesimally thin, semiinﬁnite slit or crack deﬁned
by {(x
1
, x
3
) − ∞ < x
1
≤ 0, −∞ < x
3
< ∞}. Its surface ∂R, the upper
and lower surfaces of the crack, is free of traction. Though R is no longer
bounded, our previous arguments are readily extended to this case. We wish to
ﬁnd the waveﬁeld in the neighborhood of the edge to determine the nature of
the waveﬁeld’s singularity there.
For simplicity, we take the waveﬁeld to be an antiplane shear one. We seek a
solution to (1.15), in the timeharmonic approximation and expressed in polar
coordinates, namely
(1/r)∂
r
(r∂
r
u
3
) +(1/r
2
)∂
2
θ
u
3
+k
2
u
3
= 0, (4.49)
where k is the wavenumber. At θ = ±π, r = 0, ∂
θ
u
3
= 0. We do not
intend to solve a global problem but merely to explore possible solutions in
the neighborhood of r = 0. Therefore we should use a coordinate expansion in
kr for u
3
beginning with (kr)
α
, 0 < α < 1, with the assumption that kr →0.
However, a bit more insight is gained by introducing the length scale shown
in Fig. 4.3, where k 1. Setting ρ = r/ and w = u
3
/U (U is a max
imum particle displacement), we reexpress (4.49) in terms of (ρ, θ) and w.
Asymptotically expanding w as
w(ρ, θ) ∼ (k)
α
n≥0
w
n
(ρ, θ)(k)
n
, (4.50)
72 4 Green’s Tensor and Integral Representations
we ﬁnd that the lowestorder term satisﬁes
(1/ρ)∂
ρ
(ρ∂
ρ
w
0
) +(1/ρ
2
)∂
2
θ
w
0
= 0. (4.51)
The waveﬁeld near the edge is essentially equivalent to the static ﬁeld. This
further illustrates the principle that a wave has to propagate several wavelengths,
freeing itself of its source, before its propagating character becomes manifest.
An expansion such as (4.50) is called an inner expansion. This expansion will
contain an unknown constant that can be found by matching it to an outer
expansion. The inner expansion connects the source – in this case the crack
tip – with the propagating waveﬁeld, the outer expansion, a few wavelengths
away. This method of analysis is called matching asymptotic expansions (Hinch,
1991; Holmes, 1995). We shall continue with this analysis in Section 5.6.
Subject to the boundary condition stated previously, (4.51), has solutions of
the form
w
0
(ρ, θ) = (Aρ
β
+ Bρ
−β
) sin βθ, (4.52)
where β = (2n+1)/2, n = 0, 1, 2, . . . and Aand B are undeterminedconstants.
To satisfy (4.48), B = 0 and α = β =
1
2
. Larger values of β also give acceptable
solutions, but this is the minimum such value and hence determines the most
singular termallowable. Thus ku
3
∼ (kr)
1/2
A sin(θ/2)+ O[(kr)
3/2
] as kr →0.
Therefore, for a problem with the geometry of Fig. 4.3, we must specify,
when we ﬁrst formulate the problem, that ku
3
= O[(kr)
1/2
] as kr →0 to ensure
that we arrive at a unique solution.
4.8 Scattering From an Elastic Inclusion in a Fluid
To convince the reader of just how powerful and general the foregoing results
can be, in this closing section we extend slightly these results to develop an
expression for the total waveﬁeld present in an ideal ﬂuid that contains an
elastic inclusion. We again take the time dependence as harmonic. The elastic
inclusion, in contrast to the empty cavity of Proposition 4.4, is penetrable.
Waves enter it, reverberate within it, and then reradiate. This calculation is
based on the work of Wickham (1992) and Leppington (1995).
The elastic inclusion occupies a region Rwith surface ∂R. The unit normal
ˆ n points out of R (in contrast with our previous convention). We indicate by
∂R
+
the surface approached fromoutside Rand by ∂R
−
that approached from
within. Figure 4.1 indicates the geometry, if we imagine that the radius x →∞
and the region S is absent. To indicate when a position vector x identiﬁes a
4.8 Scattering From an Elastic Inclusion in a Fluid 73
point in R, we deﬁne the function χ(x) as
χ(x) =
1, x ∈ R,
0, otherwise
(4.53)
We treat the inclusion as a linearly elastic solid embedded in a linearly elastic
(ideal) ﬂuid. The equations of motion for an elastic ﬂuid are given by those of
linear elasticity when the shear coefﬁcient is set to zero. The elastic ﬂuid is
compressible but cannot withstand shearing. The equation of motion for the
particle displacement u is
¯
λ∂
i
∂
k
u
k
+ω
2
¯ ρu
i
= −¯ ρ f
i
, x ∈ R. (4.54)
The stress tensor is given by τ
i j
= −pδ
i j
, where p, the acoustic pressure, equals
−
¯
λ∂
k
u
k
. As before, f is the body force per unit mass. The parameters ¯ ρ and
¯
λ
are the density and bulk modulus, respectively, of the elastic ﬂuid. Lastly, there
is another equation of motion, namely ∇ ∧ u = 0, arising because the motion
is irrotational.
We write the equation of motion for the particle displacement u in the elastic
solid in such a way that the solid’s inertial and elastic properties appear as a
body force within the region R. The equation is thus written as
¯
λ∂
i
∂
k
u
k
+ω
2
¯ ρu
i
= −∂
j
σ
j i
−ωp
i
, x ∈ R. (4.55)
The righthand terms are given by
σ
i j
= (λ −
¯
λ)δ
i j
∂
k
u
k
+µ(∂
i
u
j
+∂
j
u
i
), (4.56)
p
i
= ω(ρ − ¯ ρ)u
i
. (4.57)
Next we imagine that an incident waveﬁeld u
i
scatters from the elastic in
clusion, giving rise to the scattered waveﬁeld u
s
. The total waveﬁeld u(x) =
[1−χ(x)]u
i
(x) +u
s
(x). Across the surface ∂Rthe normal components of par
ticle displacement and traction are continuous and the tangential components
of traction vanish. We express these conditions as
[u
s
j
] ˆ n
j
= −u
i
j
ˆ n
j
,
∂
k
u
s
k
δ
i j
ˆ n
j
= −∂
k
u
i
k
δ
i j
ˆ n
j
+σ
j i
ˆ n
j
/
¯
λ. (4.58)
The symbol [· · ·] indicates the jump in going from ∂R
+
to ∂R
−
.
Taken together, (4.55)–(4.58) capture the concept that the elastic solid is
an inclusion within a host material, the elastic ﬂuid, with the departure of the
74 4 Green’s Tensor and Integral Representations
properties of the solid from those of the ﬂuid expressed as body force terms on
the righthand side of (4.55) and in the conditions across ∂R, (4.58).
Our goal is to use Propositions 4.2–4.4, or a slight variation of them, to give
an integral representation for u as a volume integral over R that contains σ
i j
and p
i
. To do so we need the Green’s tensor for the elastic ﬂuid. This satisﬁes
an equation analogous to (4.6), namely
¯
λ∂
i
∂
k
u
G
km
+ ¯ ρω
2
u
G
i m
= −¯ ρδ
i m
δ(x). (4.59)
It is calculated exactly as outlined in Proposition 4.2 and is given by
u
G
i k
= 1/k
2
c
2
[−∂
i
∂
k
(e
i kx
/4πx) +∂
m
∂
m
(δ
i k
/4πx)]. (4.60)
Note that this is not the usual Green’s function for linear acoustics, which
function is the response to a mass ﬂux at a single point. However, it is the
correct Green’s tensor when one considers the ﬂuid as elastic and seeks its
response to a force acting at a single point. For a point force in the direction ˆ a,
the reader is asked to verify that u
G
i k
ˆ a
k
is irrotational, provided x = 0.
Next we apply arguments identical to those used in Propositions 4.4 and then
in 4.3. The reciprocity relation is applied to the complement of R. The Green’s
tensor is one reciprocating waveﬁeld and u
s
the second. The outcome is
u
s
m
(x)[1 −χ(x)] = −c
2
∂R
+
u
G
j m
(x − x
)∂
k
u
s
k
(x
)
−u
s
j
(x
)∂
k
u
G
km
(x − x
)
ˆ n
j
dS(x
). (4.61)
The wavespeed c = (
¯
λ/¯ ρ)
1/2
. The prime indicates that the derivative is taken
with respect to x
. The unit normal ˆ n points out of R rather than out of its
compliment.
Next consider the region R. Arguments that combine those used in both
Propositions 4.3 and 4.4 are used. Again the reciprocity relation is applied, but
now within R itself. The Green’s tensor is one reciprocating waveﬁeld and u
s
the second. The outcome is
u
s
m
(x)χ(x) =
1
¯ ρ
R
[∂
i
σ
i j
(x
) +ωp
j
(x
)]u
G
j m
(x − x
)dV(x
)
+c
2
∂R
−
u
G
j m
(x − x
)∂
k
u
s
k
(x
)
−u
s
j
(x
)∂
k
u
G
km
(x − x
)
ˆ n
j
dS(x
). (4.62)
Recall that within R, u = u
s
. We add (4.61) and (4.62) to obtain the following
4.8 Scattering From an Elastic Inclusion in a Fluid 75
representation for u
s
.
u
s
m
(x) =
1
¯ ρ
R
[∂
i
σ
i j
(x
) +ωp
j
(x
)]u
G
j m
(x − x
)dV(x
)
− c
2
∂R
u
G
j m
(x − x
)
∂
k
u
s
k
(x
)
−
u
s
j
(x
)∂
k
u
G
km
(x − x
)
ˆ n
j
dS(x
). (4.63)
As it stands, this is a remarkable expression. First note that the departure of the
properties of the elastic solid from the host elastic ﬂuid appears as a body force
term in the integral over R. Second note that the jumps [u
s
j
] ˆ n
j
and [∂
k
u
s
k
]δ
i j
ˆ n
j
,
given by (4.58), are present in the integral over ∂R. The superscripts ± on ∂R
have served their purpose and are omitted from now on.
Now we imagine that the region R contains only the incident waveﬁeld u
i
and use the reciprocity relation along with the Green’s tensor to write
u
i
m
(x)χ(x) = c
2
∂R
u
G
i m
(x − x
)∂
k
u
i
k
(x
)δ
i j
− u
i
j
(x
)∂
k
u
G
km
(x − x
)
ˆ n
j
dS(x
). (4.64)
This in itself is a interesting result. It is called an extinction theorem,
6
and it
indicates that the surface integral vanishes for observation points x outside R.
We now subtract (4.63) from (4.64) to obtain
u
m
(x) = u
i
m
(x) +
1
¯ ρ
R
[∂
i
σ
i j
(x
) +ωp
j
(x
)]u
G
j m
(x − x
)dV(x
)
−
1
¯ ρ
∂R
u
G
i m
(x − x
)σ
i j
(x
)
n
j
dS(x
). (4.65)
We are almost done, but we are still left with a surface integral. Noting the
divergence termin the volume integral, we make one more application of Gauss’
theorem to give
u
m
(x) = u
i
m
(x) −
1
¯ ρ
R
σ
i j
(x
)∂
i
u
G
j m
(x − x
)
− ωp
j
(x
)u
G
j m
(x − x
)
dV(x
). (4.66)
This is a very satisfying outcome. We have, through successive applications of
the Green’s tensor in combination with the reciprocity relation, shown that the
6
The expressions (4.61) and (4.62) are also examples of extinction theorems.
76 4 Green’s Tensor and Integral Representations
total waveﬁeld u outside R equals the incident waveﬁeld u
i
plus a scattered
waveﬁeldu
s
whose source is the departure of the inertial andelastic properties of
the solidinclusionfromthose of its host, the elastic ﬂuid. However, (4.66) is not a
solution to the boundaryvalue problem for u
s
. Just as with (4.38), the integral
contains unknown terms. These are σ
i j
and p
i
, which cannot be calculated
without knowing u within R. Thus this expression is only a representation and
not a solution to the problem. However, it can be made the basis for deriving a
system of integral equations for σ
i j
and p
i
. One uses (4.66) in (4.56) and (4.57)
to enforce selfconsistency (Leppington, 1995).
References
Achenbach, J.D. 1973. Wave Propagation in Elastic Solids, pp. 96–110. Amsterdam:
NorthHolland.
Achenbach, J.D., Gautesen, A.K., and McMaken, H. 1982. Ray Methods for Waves in
Elastic Solids, pp. 22–27 and 34–38. Boston: Pitman.
Courant, R. and John, F. 1989. Introduction to Calculus and Analysis, Vol. II,
pp. 597–602. New York: Springer.
deHoop, A.T. 1995. Handbook of Radiation and Scattering of Waves. London:
Academic.
Delves, L.M. and Mohamed, J.L. 1985. Computational Methods for Integral
Equations. Cambridge: University Press.
Friedlander, F.G. 1958. Sound Pulses. Cambridge: University Press.
Hille, E. 1973. Analytic Function Theory, Vol. II, pp. 31–36. New York: Chelsea.
Hinch, E.J. 1991. Perturbation Methods, pp. 52–101. Cambridge: University Press.
Holmes, M.H. 1995. Introduction to Perturbation Methods, pp. 47–104. New York:
Springer.
Hudson, J.A. 1980. The Excitation and Propagation of Elastic Waves, pp. 106–109.
New York: Cambridge.
Kellogg, O.D. 1970. Foundations of Potential Theory, pp. 84–121. New York:
Frederick Ungar.
Kevorkian, J. 1993. Partial Differential Equations, pp. 64–75 and 94–99. New York:
Chapman and Hall.
Leppington, S.J. 1995. The Scattering of Sound by a FluidLoaded, SemiInﬁnite Thick
Elastic Plate, pp. 75–105. Manchester: Ph.D. Dissertation, The University of
Manchester.
Sommerfeld, A. 1964a. Optics, Lectures on Theoretical Physics, Vol. IV, pp. 273–289.
Translated by O. LaPorte and P.A. Moldauer. New York: Academic.
Sommerfeld, A. 1964b. Partial Differential Equations in Physics, Lectures on
Theoretical Physics, Vol. VI, pp. 84–101. Translated by E.G. Straus. New York:
Academic.
Wickham, G.R. 1992. A polarization theory for scattering of sound at imperfect
interfaces. J. Nondestr. Eval. 11: 199–210.
5
Radiation and Diffraction
Synopsis
Chapter 5 summarizes the basic propagation processes that are encountered
when studying radiation or edge diffraction. Three problems of progressive dif
ﬁculty are studied. We begin by calculating the transient, antiplane radiation
excited by a line source at the surface of a halfspace. The Cagniard–deHoop
method is used to invert the integral transforms. We then return to consider
ing how plane waves and a knowledge of their interactions can be used to
construct more general waveﬁelds. We calculate the time harmonic, inplane ra
diation, from a twodimensional center of compression buried in a halfspace.
Planewave spectral techniques are used and the resulting integrals are ap
proximated by the method of steepest descents. This method is discussed in
detail. Lastly, we extend our knowledge of planewave interactions by calcu
lating the diffraction of a time harmonic, plane, antiplane shear wave by a
semiinﬁnite slit or crack. This problem is solved exactly by using the Wiener–
Hopf method and approximately by using matched asymptotic expansions. An
Appendix describing the reduction of the diffraction integral to Fresnel integrals
is included.
5.1 Antiplane Radiation into a HalfSpace
We consider an elastic halfspace. The x
1
coordinate stretches along its surface
and the positive x
2
coordinate extends into the interior. At the origin a line load
is applied to an otherwise tractionfree surface. The line load is a tangentially
acting force very localized in x
1
and directed from−∞to ∞in the x
3
direction.
The equations of motion are given by (1.15) combined with (1.14). On the
surface x
2
=0,
µ∂
2
u
3
=−µδ(x
1
) f (t ). (5.1)
77
78 5 Radiation and Diffraction
The halfspace is quiescent for t < 0 so that f (t ) ≡0 and u
3
≡0 for t < 0. The
waves are outgoing from the source. As we have done previously, the subscript
T is dropped.
5.1.1 The Transforms
Cagniard (1962) developed a very instructive way to invert the integral trans
forms arising in transient wave problems. DeHoop (1960) popularized
Cagniard’s method by making it more readily understood. The essence of the
technique, now referred to as the Cagniard–deHoop technique, is the map
ping of the phase term, and subsequently the integration contour, of the spatial
inverse transform into a form that allows one to immediately identify the in
verse temporal transform. What is most satisfying about the technique is that
it shifts the burden of inverting the transform to understanding a mapping that
primarily affects the phase term, the amplitude term being adjusted almost as
an afterthought. And the transform merely provides a shell within which the
mapping lives.
Arguably, the simplest way to understand the technique is to use a onesided
Laplace transform over time and a twosided one over space. First, the one
sided transform over time (1.31) is taken, with the restriction that the transform
variable p remain real and positive. This is not a serious restriction because
we shall not need to invert this transform. Moreover, if necessary, we can
analytically continue the transform into the complex p plane. Second, we take
the twosided Laplace transform over the spatial coordinate; that is,
∗
¯ u
3
(α, x
2
, p) =
∞
−∞
e
−pαx
1
¯ u
3
(x
1
, x
2
, p) dx
1
, (5.2)
where the overbar indicates the temporal transform. Note that p is used to scale
the spatial transform variable α; α is complex. Though we have not used this
transform previously, it is only a slight modiﬁcation of the Fourier transform
of (1.43). The inverse of (5.2) is
¯ u
3
(x
1
, x
2
, p) =
p
2πi
+i ∞
−−i ∞
e
pαx
1
∗
¯ u
3
(α, x
2
, p) dα, (5.3)
where ≥0.
Applying the transforms over t and x
1
leads to the ordinary differential
equation
d
2∗
¯ u
3
dx
2
2
− p
2
χ
2 ∗
¯ u
3
=0, χ =(s
2
−α
2
)
1/2
. (5.4)
5.1 Antiplane Radiation into a HalfSpace 79
The slowness s =1/c is used instead of the wavespeed c. The solution must
be selected in such a way that the outgoing condition is enforced. This can
be achieved by asking that
1
(χ) ≥0 ∀α and taking as the solution
∗
¯ u
3
=
A(α, p)e
−pχx
2
. Enforcing the transformed boundary condition at x
2
=0 gives
A(α, p) =
¯
f ( p)/( pχ). Inverting in α by using (5.3) gives
¯ u
3
(x
1
, x
2
, p) =
¯
f ( p)
2πi
+i ∞
−−i ∞
e
pαx
1
e
−pχx
2
χ
dα. (5.5)
Next, inverting in time gives
u
3
(x
1
, x
2
, t ) = f (t ) ∗ I (x
1
, x
2
, t ), (5.6)
where the centered asterisk indicates a convolution in t . Hence, we are left with
inverting the integral
¯
I (x
1
, x
2
, p) =
1
2πi
+i ∞
−−i ∞
e
pαx
1
e
−pχx
2
χ
dα. (5.7)
The has been added to the inversion contour in (5.3), and hence (5.7), so
that should the contour be rotated to give a standard Fourier transform, the
contour will pass above and below the branch cuts in such a way that −pχ,
with p =−i ωand ω > 0, will give an outgoing wave for x
2
> 0 in the integrand
of (5.7).
5.1.2 Inversion
To make further progress we must decide how to cut the α plane. The cuts are
not the same as those indicated previously, because we are now using a Laplace
rather than a Fourier transform. Nevertheless the reasoning is identical to that
given in Section 3.4.4. Note that χ =−i γ , where γ is given by (3.51). Thus,
for (χ) ≥0 ∀ α, (γ ) ≥0 ∀α and the branch cuts shown in Fig. 5.1 follow.
Here χ is given as
χ =(r
1
r
2
)
1/2
e
i θ
1
/2
e
i θ
2
/2
e
−i π/2
, (5.8)
where (r
i
, θ
i
) are deﬁned in Fig. 5.1. The angle θ
1
∈ (0, 2π) and θ
2
∈ (−π, π).
Note that the sign of (χ) varies from quadrant to quadrant, as indicated in the
caption to Fig. 5.1.
1
In the absence of propagation, we are asking that the components of the disturbance decay toward
inﬁnity. For p that is real and positive, this choice will give us the freedom to distort, almost
anywhere, the spatial inversion contour in the complex α plane.
80 5 Radiation and Diffraction
Fig. 5.1. The complexα plane cut sothat (χ) ≥ 0 ∀α. Inquadrants 1and2, (χ) > 0,
and in quadrants 3 and 4, (χ) < 0. The magnitude r
i
and argument θ
i
for each radical
are shown. Contrast this ﬁgure with Fig. 3.7.
We start by mapping the variable of integration fromα to t , without assigning
any physical meaning to t . That is,we set
t (α) =−αx
1
+χx
2
. (5.9)
The variable t traces out a contour in the complex t plane as α ranges from
(− −i ∞) to ( +i ∞). However, rather than look at the t plane, we ﬁnd
that it is simpler to continue examining the α plane. Setting x
1
=r cos θ and
x
2
=r sin θ, where r =(x
2
1
+x
2
2
)
1/2
, we use (5.9) to ﬁnd α as a function of t .
A quadratic equation for α is arrived at, and when solved gives the two roots
α
±
=−
t
r
cos θ ±i sin θ
t
2
r
2
−s
2
1/2
. (5.10)
We next ask, along what contour in the α plane is t real and positive, and can
we distort the current contour to that one? The answer to the latter question is
yes, for x
2
> 0, because, with p real and positive, the integral (5.7) is conver
gent throughout the particular Riemann sheet we are working on. Figure 5.2
indicates the answer to the former question. The original contour is shown by
the heavy, dashed line and the new one by the solid line. The new contour has
two branches evinced by the subscripts plus and minus attached to the α. Elim
inating the parameter t shows that the curve is a hyperbola with asymptotes
(α
±
)/(α
±
) =∓tan θ. The parameter t starts at sr, where α
±
=−s cos θ,
and goes to ∞ along each branch α
±
. This contour is the Cagniard–deHoop
5.1 Antiplane Radiation into a HalfSpace 81
Fig. 5.2. A sketch of the original contour (heavy, dashed line) and the Cagniard–
deHoop contour (solid line) in the complex α plane. Also shown are the asymptotes
(light, dashed lines) to the Cagniard–deHoop contour and the angle θ. Note how the
angle θ is deﬁned.
contour. One last term is needed, namely
dα
±
dt
=−
1
r
cos θ ±i
t sin θ
r
2
[(t
2
/r
2
) −s
2
]
1/2
. (5.11)
Equation (5.7) can now be written as
¯
I (x
1
, x
2
, p) =
1
2πi
∞
sr
e
−pt
(s
2
−α
2
+
)
1/2
dα
+
dt
−
e
−pt
(s
2
−α
2
−
)
1/2
dα
−
dt
dt. (5.12)
Noting that one component of the integrand is the complex conjugate of the
other, we rewrite the integral as
¯
I (x
1
, x
2
, p) =
1
π
∞
sr
e
−pt
dα
+
/dt
(s
2
−α
2
+
)
1/2
dt. (5.13)
Thus, by inspection,
I (x
1
, x
2
, t ) = H(t −sr)
1
π
dα
+
/dt
(s
2
−α
2
+
)
1/2
, (5.14)
where H(x) is the Heaviside function.
The burden of the inversion has rested with the mapping from the α plane
to the t plane by using (5.9) and its inverse (5.10). Recall, however, that a
convolution integral (5.6) must still be evaluated. Moreover, the example just
82 5 Radiation and Diffraction
given is a particularly simple instance of this mapping, and indeed Cagniard
(1962) should be explored for more complicated cases.
It is also of interest to note the form of (5.13). It is precisely of the form
needed to approximate it, for large p, using Watson’s lemma, something we
discuss in Section 5.3.1. With a bit of exploring, we should ﬁnd that in doing
so the leading term is determined by the nature of the singularity in the dα
+
/dt
of the integrand. Using a Tauberian theorem (van der Pol and Bremmer, 1950),
we can construct an approximation to I (x
1
, x
2
, t ) for t near sr. This is called
a wavefront approximation. This technique is described, within the context of
elastic waves, by Knopoff and Gilbert (1959) and is used extensively by Harris
(1980a,1980b). It is also brieﬂy explored in Problem 5.3.
Problem 5.1 HalfPlane and Strip Problems
Problem 1. Consider the halfspace {(x
1
, x
2
) −∞< x
1
<∞, x
2
≥0}. It
is subjected, at x
2
=0, to the antiplane traction
τ
23
=−µ H(x
1
)d f /dt (5.15)
f (t ) ≡0 for t < 0. Determine the particle displacement u
3
(x
1
, x
2
, t ) in the
interior by using the Cagniard–deHoop technique to invert the transforms. The
integrand of the inverse spatial transformwill have a pole and two branch points.
Identify the waves represented by the pole term and by the integral taken along
the Cagniard–deHoop contour.
Problem 2. A simple variation to Problem 1 is to consider exactly the same
problem except that (5.15) is replaced by
τ
23
=−µd f /dt [H(x
1
+a) − H(x
1
−a)], (5.16)
where a is a constant. Determine u
3
(x
1
, x
2
, t ) for this case.
5.2 Buried Harmonic Line of Compression I
Finding the response of an elastic halfspace to a point or line load applied
either at its surface or buried just below it is possibly the most studied problem
in elastic waves. It is usually called Lamb’s problem. Ewing et al. (1957) and
Cagniard (1962) are primary references to these problems, though all the books
on elastic waves cited previously discuss them. We limit our discussion to
calculating the response of an elastic halfspace to a timeharmonic line of
compression (a twodimensional center of compression) by using the plane
wave spectral approach.
5.2 Buried Harmonic Line of Compression I 83
We consider an elastic halfspace, identical to that discussed in Section 5.1,
with a line of compression located at (0, h). A source of this kind was intro
duced in Section 4.3.1 and explored further in Problem 4.2. The outcome of
that problem suggests that the present calculations are most easily begun by
working with potentials. The potentials and their governing equations are given
by (1.19)–(1.22). The vector potential ψ =ψˆ e
3
, so that the divergence condi
tion is automatically satisﬁed. The potentials are divided into two parts, namely
ϕ
t
=ϕ
i
+ϕ and ψ
t
=ψ. The terms ϕ
t
and ψ
t
are the total compressional and
shear potentials, respectively. The term ϕ
i
is that excited by the line of com
pression in the absence of the tractionfree surface and the terms ϕ and ψ are
the compressional and shear potentials scattered from the surface. The source
is time harmonic, and we suppress that dependence.
The term ϕ
i
satisﬁes
∂
α
∂
α
ϕ
i
+k
2
L
ϕ
i
= F
0
δ(x)δ(y −h), (5.17)
an equation almost identical to (4.17). The constant F
0
has the dimensions of
length squared. It is sometimes useful to set F
0
= A/k
2
L
, where A is a dimen
sionless constant. The potentials ϕ and ψ satisfy the same equation, but with
no source term on the righthand side, and in the equation governing ψ, with
k
L
replaced by k
T
. Note that (4.18) gives the solution to (5.17). Written here
using a different notation, the solution is
ϕ
i
=
−i F
0
4π
C
β
e
i (βx
1
+γ
L
x
2
−h)
dβ
γ
L
. (5.18)
The radical γ
L
=(k
2
L
−β
2
)
1/2
. Note that γ
L
=i γ , where γ is the radical deﬁned
by (3.51)
2
. We ask that (γ
L
) ≥0 ∀β. The radical γ
T
=(k
2
T
−β
2
)
1/2
, which
will soon be needed, is deﬁned in the same way. This leads to the complex plane
structured as indicated in Fig. 5.3 (imagine the branch cuts for γ
L
and γ
T
lying
on top of one another). The contour C
β
is also shown.
The total particle displacement u
t
=u
i
+u, the incident plus scattered wave
ﬁelds. For x
2
< h, u
i
is given by
u
i
=
F
0
4π
C
β
(βˆ e
1
−γ
L
ˆ e
2
) e
i [βx
1
+γ
L
(h−x
2
)]
dβ
γ
L
. (5.19)
We next introduce the Sommerfeld transformation β =k
L
cos α, γ
L
=k
L
sin α
ﬁrst introduced in (2.37). Then (5.18) becomes
u
i
=−
k
L
F
0
4π
C
ˆ p
0
(α)e
i k
L
ˆ p
0
·x
e
i k
L
h sin α
dα, (5.20)
2
γ
L
is the radical noted at the end of Section 3.44 and there labeled ξ.
84 5 Radiation and Diffraction
Fig. 5.3. The complex β plane cut so that (γ
I
) ≥0 ∀β, where I = L, T. In quadrants
1 and 2, (γ
I
) > 0, and in quadrants 3 and 4, (γ
I
) < 0. The contour C
β
is shown, as
are the Rayleigh poles at ±k
r
.
where
ˆ p
0
(α) =cos α ˆ e
1
−sin α ˆ e
2
. (5.21)
In writing (5.20), we have used a notation very similar to that of Section 3.1.
Here (5.21) is identical to (3.2) with θ
0
=π/2 +α, though we now use α as
the independent variable. The α plane, with the contour C, is shown in Fig. 5.4.
We discuss its topography further in Section 5.4.
The integrand of (5.20) is a plane, compressional wave incident to the
tractionfree surface x
2
=0. Using as the integrands the reﬂected plane, com
pressional and shear waves calculated in Section 3.1, we can construct the
scattered waveﬁelds such that the boundary condition is satisﬁed, as well as the
condition that the scattered waveﬁelds be outgoing. The total scattered wave
ﬁeld u =u
sL
+u
sT
, where the ﬁrst termrepresents the scattered compressional
waveﬁeld and the second the scattered shear waveﬁeld. These waveﬁelds are
given by
u
sL
= −
k
L
F
0
4π
C
ˆ p
1
(α)R
L
(α)e
i k
L
ˆ p
1
·x
e
i k
L
h sin α
dα, (5.22)
u
sT
= −
k
L
F
0
4π
C
ˆ
d
2
(α)R
T
(α)e
i k
T
ˆ p
2
·x
e
i k
L
h sin α
dα. (5.23)
5.2 Buried Harmonic Line of Compression I 85
Fig. 5.4. The complex α plane. The branch cuts with branch points α
T
and π −α
T
, the
two Rayleigh poles α
r
and π −α
r
, and the contour C are indicated. Moreover, the saddle
point θ
1
, for the compressional wave, and the contour of steepest descents C
s
, with its
asymptotes at θ
1
±π/2, are also indicated. Quadrants 1 and 2 of Fig. 5.3 are mapped
into quadrants 1
and 2
. Note how part of the contour of steepest descents passes onto
the other Riemann sheet. This is indicated by the dashed portion of the contour.
The several unit vectors are given by
ˆ p
1
(α) =cos α ˆ e
1
+sin α ˆ e
2
, (5.24)
ˆ p
2
(α) =cos ¯ α ˆ e
1
+sin ¯ α ˆ e
2
, (5.25)
and
ˆ
d
2
(α) = ˆ e
3
∧ ˆ p
2
(α). (5.26)
These are identical to the unit vectors deﬁned in Section 3.1 by (3.4), (3.6) and
(3.7), when θ
0
=π/2 +α and θ
2
=π/2 + ¯ α. The two reﬂection coefﬁcients
R
L
(α) and R
T
(α) are also given by (3.10)–(3.12). The independent variable is
taken as α, with the understanding that c
−1
L
cos α =c
−1
T
cos ¯ α is used to relate
α to ¯ α. Expressing the reﬂection coefﬁcients in terms of α and ¯ α, we have
R
L
(α) = A
−
(α)/A
+
(α), (5.27)
R
T
(α) = 2κ sin 2α cos 2α/A
+
(α), (5.28)
86 5 Radiation and Diffraction
with
A
∓
=sin 2α sin 2¯ α ∓κ
2
cos
2
2¯ α (5.29)
and κ =c
L
/c
T
.
Deﬁning X = x
1
ˆ e
1
+(x
2
+h)ˆ e
2
, we can write the phase of the integrand of
(5.22) as e
i k
L
ˆ p
1
·X
, implying that the scattered compressional wave appears to
come from a virtual line of compression at (0, −h). The phase of the integrand
of (5.23) is complicated by the term e
i k
L
h sin α
. The scattered shear wave does
not appear to come from a virtual line source. We indicate subsequently that its
virtual source is a caustic.
Problem 5.2 Lamb’s Problem
The problem of a buried line of compression is often solved differently from
the method we have just used. The present problemindicates this more common
method.
Continue to work with the potentials. To apply the boundary conditions, the
reader will need τ
12
and τ
22
expressed in terms of the potentials. Accordingly,
show that
τ
12
= µ(2∂
1
∂
2
ϕ +∂
2
∂
2
ψ −∂
1
∂
1
ψ), (5.30)
τ
22
= λ∂
α
∂
α
ϕ +2µ(∂
2
∂
2
ϕ −∂
2
∂
1
ψ). (5.31)
Take a Fourier transform in x
1
of (5.17) and solve the ordinary differential
equation in x
2
. Note that [d
2
∗
ϕ
i
]
h+
h−
= F
0
. Thus show that
∗
ϕ
i
(β, x
2
) =−
i F
0
2γ
L
e
i γ
L
x
2
−h
. (5.32)
Showthat the transformed, scattered potentials are given by
∗
ϕ =(β)e
i γ
L
x
2
and
∗
ψ =(β)e
i γ
T
x
2
. Are the radicals γ
I
identical to those deﬁned previously?
Expressing the traction terms as τ
t
i j
=τ
i
i j
+τ
i j
, the boundary conditions at
x
2
=0 become τ
12
=−τ
i
12
and τ
22
=−τ
i
22
. Enforce the boundary conditions in
the transform domain to show that
(β) =
i F
0
2γ
L
e
i γ
L
h
k
2
T
−2β
2
2
−4β
2
γ
L
γ
T
R(β)
, (5.33)
and ﬁnd a similar expression for (β). The function R(β) is the Rayleigh
function, (3.47), multiplied by ω
4
to account for the change in scaling.
5.3 Asymptotic Approximation of Integrals 87
Lastly, calculate the x
1
component of the scattered particle displacement,
u
1
=u
sL
1
+u
sT
1
. Show that
u
sT
1
=
F
0
2π
C
β
U
sT
(β)e
i (βx
1
+γ
T
x
2
)
e
i γ
L
h
dβ, (5.34)
where
U
sT
=
2βγ
T
k
2
T
−2β
2
R(β)
. (5.35)
Find the expression for u
sL
1
.
5.3 Asymptotic Approximation of Integrals
The calculations of the previous section, as well as those undertaken in the pre
vious chapters, have indicted how readily one arrives at integrals such as (5.22)
or (5.23). We now consider their asymptotic approximation when k
I
r →∞;
that is, when the distance r is many wavelengths long. Asymptotic approxima
tions give a satisfying interpretation to these integrals as ray ﬁelds. This is a
another way of generating asymptotic approximations such as those discussed
in Section 2.4. We are then concerned with integrals of the form
I (κ) =
C
1
f (z)e
κq(z)
dz, (5.36)
where q(z) =u(x
1
, x
2
) +i v(x
1
, x
2
). Also note that I (κ) is also dependent on the
contour of integration C
1
, though that dependence is seldomexplicitly indicated.
The parameter
3
κ, which is often k
I
r, is assumed to be real, positive and large,
though the approximations can (with care) be analytically continued to sectors
of the complex κ plane.
5.3.1 Watson’s Lemma
To facilitate our work we need to study brieﬂy the gamma function and two of
its friends. We deﬁne the gamma function
4
as
(z)! :=
∞
0
e
−t
t
z
dt, (5.37)
3
The ﬁrst step in asymptotically approximating an integral is to scale the variables so that the
large parameter κ can be clearly identiﬁed. This scaling and identiﬁcation depends a good deal
on the situation of interest. Harris (1987) indicates how these decisions can effect approximating
diffraction from an aperture, while Carrier et al. (1983) indicate how they affect the asymptotic
approximation of a Hankel function.
4
(z)! is also written as (z +1).
88 5 Radiation and Diffraction
the incomplete gamma function as
γ (z, x) :=
x
0
e
−t
t
z−1
dt, (5.38)
and the complement to the incomplete gamma function as
(z, x) : = (z −1)! −γ (z, x),
=
∞
x
e
−t
t
z−1
dt. (5.39)
These deﬁnitions are for (z) > 0, though the functions can be analytically
continued to (z) < 0. We take x as real and positive.
Lemma.
(a, x) ∼e
−x
n≥1
(a −1)!
(a −n)!
x
a−n
, x →∞, (5.40)
where a and x are real and positive.
Proof The following proof is from Copson (1971), but it is repeated here
both for completeness and because it illustrates a very simple but useful way
to generate an asymptotic approximation to an integral, namely integration by
parts. After N such integrations (5.39) becomes
(a, x) =e
−x
N
n=1
(a −1)!
(a −n)!
x
a−n
+
(a −1)!
(a − N −1)!
(a − N, x). (5.41)
To estimate the magnitude of the remainder term, we note that
(a −1)!
(a − N −1)!
∞
x
e
−t
t
a−N−1
dt
<
(a −1)!
(a − N −1)!
x
a−N−1
∞
x
e
−t
dt
=
(a −1)!
(a − N −1)!
x
a−N−1
e
−x
, (5.42)
for N > (a −1). Equation (5.40) follows.
We begin by considering (5.36) with a contour C
1
that begins at a ﬁnite point
z
1
and extends to ∞ in a sector of the complex plane wherein the integral is
convergent. We assume that d
z
q =0 even at the end point z
1
. We look for a
contour C
2
from z
1
to ∞ along which q(z) <q(z
1
) and q(z) =q(z
1
).
5.3 Asymptotic Approximation of Integrals 89
First, we map the z plane into the s plane by using
s =q(z
1
) −q(z). (5.43)
Second, we deform the contour in the s plane to one along the real s axis from
zero to ∞. As with the Cagniard–deHoop contour, one seldom works directly
in the s plane. Rather it is the contour C
1
in the z plane that is deformed to a
new contour C
2
in that plane. Any poles or branch cuts encountered during the
deformation must be appropriately surrounded and their contributions added to
the asymptotic approximation of the integral, though we do not include such
contributions here.
With the change of variables of (5.43) we introduce the function G(s) =
−f (z)/d
z
q and write G(s) as G(s) = g(s)s
λ
, where g(s) is analytic at s =
0 (z = z
1
) with a radius of convergence ¯ r. The term s
λ
includes any singularity
at s =0 either from f (z) itself or as a result of mapping to the s plane. The
integral now assumes the form exp[κq(z
1
)]J(κ), where J(κ) is given by (5.44).
Proposition 5.1 (Watson’s Lemma). Consider the integral
J(κ) =
∞
0
g(s)s
λ
e
−κs
ds, (5.44)
arrived at by the change of variables described in the preceding paragraphs.
λ >−1. Let real constants K and b exist so that g(s) ≤ Ke
bs
as s →∞. Let
g(s) be analytic at s =0 so that for s ∈ [0,r
∗
],
g(s) =a
0
+a
1
s +a
2
s
2
+· · · + R
m+1
(s), (5.45)
with R
m+1
(s) ≤Cs
m+1
; ¯ r is the radius of convergence for g(s) and r
∗
< ¯ r.
Then
J(κ) ∼
n≥0
a
n
(λ +n)!
κ
λ+n+1
, κ →∞. (5.46)
Proof This is proven in Copson (1971), Carrier et al. (1983), Ablowitz and
Fokas (1997) and Wong (1989) with greater generality and weaker assumptions
than those stated here.
The integral J(κ) is written as
J =
r
∗
0
e
−κs
s
λ
(a
0
+a
1
s +a
2
s
2
+· · · +a
m
s
m
) ds
+
r
∗
0
e
−κs
s
λ
R
m+1
(s) ds +
∞
r
∗
g(s)s
λ
e
−κs
ds. (5.47)
90 5 Radiation and Diffraction
The ﬁrst integral is again split into two parts by being written as one from zero
to ∞minus one fromr
∗
to ∞. The integral fromzero to ∞is readily evaluated,
giving
a
0
λ!
κ
λ+1
+a
1
(λ +1)!
κ
λ+2
+· · · +a
m
(λ +m)!
κ
λ+m+1
. (5.48)
Each term of the second integral, say that with index n, is such that
∞
r
∗
e
−κs
a
n
s
λ+n
ds =
a
n
κ
λ+n+1
∞
κr
∗
e
−t
t
λ+n
dt. (5.49)
Noting that the integral on righthand side is (λ +n +1, κr
∗
), we conclude
from the previous Lemma that each term of this second integral is
∞
r
∗
e
−κs
a
n
s
λ+n
ds = O
e
−κr
∗
κ
. (5.50)
With R
m+1
(s) ≤Cs
m+1
, the second integral in (5.47) is such that
r
∗
0
e
−κs
s
λ
R
m+1
(s) ds
≤
C
κ
λ+m+2
κr
∗
0
e
−t
t
λ+m+1
dt
= O
κ
−(λ+m+2)
, (5.51)
where we recognize that the integral on the right is, after a change of variables,
the incomplete gamma function, (5.38). With the use of (5.39) and the previous
Lemma, the ordering follows. Lastly, using g(s) ≤ Ke
bs
, as s →∞, and the
previous Lemma, we ﬁnd that
K
∞
r
∗
e
−s(κ−b)
s
λ
ds = O
e
−r
∗
(κ−b)
(κ −b)
. (5.52)
All the the bounds are taken as κ →∞. The asymptotic approximation (5.46)
follows.
One simple weakening of the assumptions of this proposition follows imme
diately by noting that g(s) need only have the asymptotic behavior exhibited
by (5.45) as s →0. This fact and Watson’s lemma are the starting point for the
various Abelian and Tauberian theorems found in Doetsch (1974) or van der
Pol and Bremmer (1950). It also demonstrates the principle that how a function
behaves at its origin is projected into how its transform behaves at inﬁnity and
vice versa.
5.3 Asymptotic Approximation of Integrals 91
5.3.2 Method of Steepest Descents
We next consider (5.36) with a contour C
1
that stretches across the complex
plane, beginning in one sector of the complex plane at inﬁnity and ending in
another sector at inﬁnity. The contours C
β
and C deﬁning integrals (5.19) and
(5.20) are examples. Assume that (q) has a stationary point at z
s
. We then
deform the contour C
1
into a contour C
s
passing through this point and deﬁned
by q(z) ≤q(z
s
) and q(z) =q(z
s
), assuming, of course, that such a defor
mation is possible. Pole and branch cut contributions arising from deforming
C
1
to C
s
must be added to the contribution made by the new integral over C
s
,
though they are not explicitly included in the present discussion. If f (z) is an
alytic and slowly varying near z
s
, the principal contribution to the new integral
comes from z
s
and I (κ) takes the approximate form
I (κ) ≈ f (z
s
)e
i κv(x
1s
,x
2s
)
C
s
e
κu(x
1
,x
2
)
dz. (5.53)
The Steepest Descents Contour
To ﬁx our ideas more precisely, we examine the topography of the function
q(z) =u(x
1
, x
2
) +i v(x
1
, x
2
) near its stationary point z
s
. Our discussion follows
a similar one given in Felsen and Marcuvitz (1994). We assume that q(z) is
analytic in region Q containing z
s
, and that d
z
q =0 but d
2
z
q =0 at z
s
. The
function q(z) satisﬁes the Cauchy–Riemann equations
∂
1
u =∂
2
v, ∂
2
u =−∂
1
v (5.54)
in Q so that u and v are harmonic functions. Thus, if at (x
1s
, x
2s
) the curvature
of the surface u(x
1
, x
2
) =C, or v(x
1
, x
2
) =C, where C is a constant, is positive
along the x
1
axis, it is negative along the (perpendicular) x
2
axis. The stationary
point z
s
is therefore a saddle point and its neighborhood a col or saddle (Courant
and John, 1974). Figure 5.5 sketches the topography in the neighborhood
of z
s
.
To investigate this neighborhood and the contour C
s
, we parameterize the
contour C
s
with the arclength r so that the directional derivative of u along this
contour is
d
r
u =∂
1
u cos α +∂
2
u sin α, (5.55)
where d
r
x
1
=cos α and d
r
x
2
=sin α. Viewing (5.55) as a function of α, the
direction of maximum change in u is given by
d(d
r
u)/dα =−∂
1
u sin α +∂
2
u cos α =0. (5.56)
92 5 Radiation and Diffraction
Fig. 5.5. A threedimensional sketch of the topography of (q) =u near z
s
, the sta
tionary point. The contour C
s
, its arclength r, and the angle it makes with the x
1
axis as
it passes through z
s
are shown.
With the use of the Cauchy–Riemann equations, this equation becomes d
r
v =0.
Therefore v is constant along the same contour on which u changes most rapidly.
However, there are two such contours and we need to select that one along which
u decreases, as we move away from the saddle point z
s
. The contour C
s
along
which v is constant but along which u decreases most rapidly is called the path
of steepest descents, or the steepest descents contour.
To determine this contour, we examine the behavior of q(z) in the neigh
borhood of z
s
. Expanding q(z) about z
s
gives
exp[κq(z)] ≈exp[κq(z
s
)] exp
(κ/2) d
2
z
q(z
s
)(z −z
s
)
2
, (5.57)
or
exp[κq(z)] ≈ exp[κq(z
s
)]
×exp
(κ/2)
d
2
z
q(z
s
)(z −z
s
)
2
(cos 2ψ +i sin 2ψ)
, (5.58)
where
ψ =arg(z −z
s
) +
1
2
arg
d
2
z
q(z
s
)
. (5.59)
Examining (5.58) indicates that, locally, e
κu
decreases most rapidly and e
i κv
remains constant for ψ =±π/2. This then is the contour C
s
we seek. For
ψ =0 or π, e
κu
increases most rapidly and e
i κv
again remains constant; for
ψ =±π/4 or ±3π/4, e
κu
remains constant and e
i κv
oscillates most rapidly.
These latter contours are paths of stationary phase or stationary phase contours.
5.3 Asymptotic Approximation of Integrals 93
Note that we could shift the steepest descents contour to pass through a point
higher up the ridge. However, this would mean that the exponential in the inte
grand would oscillate and, therefore, there could be cancellations, voiding the
assumption that the maximumcontribution to the integral comes fromthis point.
An Isolated FirstOrder Saddle Point
With the following proposition we put the knowledge just gained into a more
precise form. As with previous propositions, the conditions cited are stronger
than they have to be. Any of the references cited when Watson’s lemma was dis
cussed also discuss the method of steepest descents fromvarious points of view.
Proposition 5.2. Let the function q(z) =u(x
1
, x
2
) +i v(x
1
, x
2
) be analytic in a
region Q containing z
s
. The point z
s
is a saddle point, and it is isolated and
ﬁrst order. That is, d
z
(z
s
) =0, but d
n
z
q(z
s
) =0, n ≥2.
Assume that the contour C
1
has been deformed into the steepest descents
contour C
s
deﬁned previously.
Let the function f (z) be analytic in a region R containing z
s
such that
f (z
s
) =0. Moreover, let there be constants K and b such that
2 f (z)[q(z
s
) −q(z)]
1/2
d
z
q(z)
< Ke
b[q(z)−q(z
s
)]
,
as z →∞in sectors of the complex plane where C
s
begins and ends.
Then
I (κ) ∼
(−2π)
1/2
κ d
2
z
q(z
s
)
1/2
f (z
s
)e
i κq(z
s
)
+O
κ
−3/2
, κ →∞. (5.60)
The argument of [−2/d
2
z
q(z
s
)]
1/2
is deﬁned by the argument of d
s
z at z
s
.
Proof We again use a mapping that captures the essential topographical
features of the phase function q(z) in the neighborhood of the stationary point.
We map from the z plane to the s plane by using
s
2
=q(z
s
) −q(z). (5.61)
We deﬁne the inverse mapping so that, as s varies from −∞to ∞, z traces the
steepest descents contour from beginning to end. Figure 5.5 suggests how the
contour C
s
might look in the z plane. With this change of integration variable,
the integral I (κ) is now given by
I (κ) =e
κq(z
s
)
∞
−∞
G(s)e
−κs
2
ds, (5.62)
94 5 Radiation and Diffraction
where
G(s) = f (z)
dz
ds
,
dz
ds
=
−2s
d
z
q(z)
. (5.63)
Note that at z = z
s
, d
s
z becomes indeﬁnite so that a limit must be taken as
z →z
s
.
We have assumed that f (z) is analytic in a region containing z
s
. Moreover,
d
s
z is also analytic in a region containing s =0 (the singularity is removable).
Thus G(s) is analytic in a region containing s =0 and can be expanded in a
power series with a radius of convergence ¯ r.
G(s) = G(0) +d
s
G(0)s +d
2
s
G(0)(s
2
/2) +· · · + R
2(m+1)
(s), (5.64)
where s ≤r
∗
< ¯ r. This expansion is seldom easy to calculate because (5.61)
expanded in a power series in z −z
s
must be inverted to give z = z(s) as a
power series in s. Copson (1935) describes how to invert a series. However, it
is usually enough to calculate only the ﬁrst term G(0)
5
. Therefore
G(0) = f (z
s
)
dz
ds
s=0
,
dz
ds
s=0
=
−2
d
2
z
q(z
s
)
1/2
. (5.65)
The differential element ds is real and positive along the path of integration, so
that at s =0,
arg
dz
ds
s=0
= arg(dz)
z=z
s
=α. (5.66)
Because G(s) is analytic, R
2(m+1)
(s) < Cs
2(m+1)
, where C is a constant.
Moreover, setting G(s) = g(s)s, we note that by hypothesis g(s) < Ke
bs
2
as
s →∞. At this point we follow a procedure almost identical to that used to
establish Proposition 5.1. Setting I (κ) =exp[κq(z
s
)]J(κ), we write
J =
r
∗
−r
∗
e
−κs
2
G(0) +· · · +d
2m
s
G(0)
s
2m
(2m)!
ds
+ 2
r
∗
0
e
−κs
2
R
2(m+1)
(s) ds +
∞
r
∗
g(s)s e
−κs
2
ds. (5.67)
Estimating the contributions of the various integrals is identical to that used
previously in (5.48)–(5.52). The asymptotic approximation, (5.60), follows.
Note howr
∗
enters the calculation. If r
∗
is small we have not really achieved
very much, so that we want it to be at least O(1). This is what is meant by the
phrase that f (z) be slowly varying near z
s
.
5
If, say, d
2
s
G(0) is the ﬁrst nonzero term, the pattern of the proof is unchanged.
5.3 Asymptotic Approximation of Integrals 95
5.3.3 Stationary Phase Approximation
The stationary phase approximation is usually applied to integrals of the form
I (κ) =
x
2
x
1
f (x)e
i κp(x)
dx, x
1
< x
2
, (5.68)
where p(x) is a real function when x is real, and κ is real, positive, and assumed
large. The contour is thus a stationary phase contour. We assume that there is a
ﬁrstorder stationary point x
s
between x
1
and x
2
.
To asymptotically approximate this integral, we use the method of steepest
descents. We assume i p(z) is analytic in region containing x
s
so that the sta
tionary point becomes a saddle point. The integration contour is then deformed
to a path of steepest descents passing through x
s
. Using (5.60), we ﬁnd the
contribution from x
s
is
I (κ) ∼
2π
κ
d
2
x
p(x
s
)
1/2
f (x
s
)e
i [κp(x
s
)±π/4]
, κ →∞. (5.69)
The plus sign is taken for d
2
x
p(x
s
) > 0 and the minus sign for d
2
x
p(x
s
) < 0.
The end point contributions are estimated by using the method outlined in
Section 5.3.1 and Watson’s lemma. The contributions from the end points x
1
and x
2
are
I (κ) ∼
1
i κ
f (x
2
)
d
x
p(x
2
)
e
i κp(x
2
)
−
f (x
1
)
d
x
p(x
1
)
e
i κp(x
1
)
, κ →∞. (5.70)
The integral (5.68) is asymptotically approximated by the sum of the contribu
tions given by (5.69) and (5.70).
The references cited in connection with Watson’s lemma discuss the sta
tionary phase approximation from various other viewpoints and provide more
detailed discussions.
Problem 5.3 Asymptotic Approximations of Integrals
Problem 1. Consider an integral having the form given by (5.7). Show that
the Cagniard–deHoop contour is one of steepest descents and that α =−s cos θ
is a saddle point. The angle θ is that shown in Fig. 5.2.
Problem 2. Consider the integral given by (5.13). Use Watson’s lemma to
show that the ﬁrst term of an asymptotic expansion in p is
¯
I (x
1
, x
2
, p) ∼
a
0
(−1/2)!
p
1/2
e
−psr
, p →∞. (5.71)
96 5 Radiation and Diffraction
Find a
0
. Invert this to approximate I (x
1
, x
2
, t ). This approximation is accurate
as t →sr (consult Doestch, 1974 or van der Pol and Bremmer, 1950 to learn
why) and is therefore sometimes called a wavefront approximation.
Problem 3. Consider an integral of the form
I (kr) =
C
P(cos α)e
i kr cos(θ−α)
dα, (5.72)
where θ ∈ (0, π). The contour C begins at π −i ∞and ends at i ∞. Sketch the
contour and show that the integral converges provided it begins and ends in
the sectors containing these points. Speculate on the structure that P(cos α)
might be permitted in order that an asymptotic approximation of the integral be
successful. Consider the change of integration variable
τ =2
1/2
e
i π/4
sin[(θ −α)/2]. (5.73)
Show that a deformation of the contour to one along which τ is real gives an
integral along the path of steepest descents C
s
.
5.4 Buried Harmonic Line of Compression II
We nowturn to the asymptotic approximation of the integrals, (5.22) and (5.23),
that describe the waveﬁeld scattered from the tractionfree surface.
5.4.1 The Complex Plane
We begin by recalling the Sommerfeld transformation used to arrive at (5.20),
namely
β =k
L
cos α =k
T
cos ¯ α, γ
L
=k
L
sin α, γ
T
=k
T
sin ¯ α, (5.74)
where c
−1
L
cos α =c
−1
T
cos ¯ α and γ
I
=(k
2
I
−β
2
)
1/2
. Moreover, recall Figs. 5.3
and 5.4 where the features of the β plane and the α plane, respectively, are
described. In particular, the contours C
β
and C are indicated there. Section 3.4.3
points out that A
+
(α) is identical to the Rayleigh function, provided the def
inition of α given in the paragraph preceding (5.27) is used. The integrands
of (5.22) and (5.23) therefore have poles at α
r
and π −α
r
. These are indicated
in Fig. 5.4, as are their βplane counterparts in Fig. 5.3. These poles give rise
to oppositely directed Rayleigh surface waves.
Before continuing, it is useful to ask what Fig. 5.4 and (5.22) and (5.23)
mean. Consider the values of α to the left of the vertical line through π/2.
5.4 Buried Harmonic Line of Compression II 97
Those to the right have a very similar meaning. From π/2 to zero the incident
compressional wave is composed of plane waves, each incident at the angle
α in this range. They are reﬂected into plane compressional and shear waves
propagating away from the surface at α and ¯ α, respectively. As α climbs the
imaginary axis, incident inhomogeneous plane waves are added to the overall
incident waveﬁeld. Theydecaytowardthe surface. Reciprocallytheyscatter into
compressional inhomogeneous plane waves that decay away from the surface.
However, as α takes on imaginary values from zero to a critical angle α
T
,
deﬁned by c
−1
L
cos α
T
=c
−1
T
, shear plane waves are still being reﬂected at the
real angle ¯ α. At the limiting (imaginary) angle α
T
a shear wave grazing the
surface is excited. For values of α beyond α
T
the scattered shear plane waves
are now also inhomogeneous and decay away from the surface. Lastly, note
that as we climb beyond α
T
, we quite quickly encounter the Rayleigh pole at
α
r
. This term must be included in our evaluation of (5.22) and (5.23) when it is
enclosed. Note that the Rayleigh wave could not be excited unless the incident
disturbance contained inhomogeneous waves.
5.4.2 The Scattered Compressional Wave
Setting x
1
=r cos θ
1
and (x
2
+h) =r sin θ
1
, we express (5.22) as
u
sL
=−
k
L
F
0
4π
C
ˆ p
1
(α)R
L
(α)e
i k
L
r cos(α−θ
1
)
dα. (5.75)
To approximate this integral for k
L
r large, we must ﬁrst ascertain where the
contour of steepest descents C
s
begins and ends, and what it looks like near the
stationary point. Expressing α as α =α
1
+i α
2
,
cos(α −θ
1
) =cos(α
1
−θ
1
) cosh α
2
−i sin(α
1
−θ
1
) sinh α
2
. (5.76)
For the integral (5.75) to converge, [cos(θ
1
−α)] > 0 so that C
s
must begin
and end in a region of the α plane where sin(α
1
−θ
1
) sinh α
2
< 0. That is, the
contour must begin in a region α
1
∈ (θ
1
, π +θ
1
), where α
2
< 0, and end in
α
1
∈ (θ
1
−π, θ
1
), where α
2
> 0.
The function q(α) =i cos(α −θ
1
) and the stationary point, a saddle point, is
given by α =θ
1
. The contour C
s
is described by [q(α)] =1, or, from(5.76), by
cos(α
1
−θ
1
) cosh α
2
=1. (5.77)
Near α =θ
1
, C
s
is described by α
2
=±(α
1
−θ
1
). Because we want [q(α)]
to achieve a maximum along C
s
at θ
1
, α
2
=−(α
1
−θ
1
). Further, for α large,
α
1
→±π/2 +θ
1
.
98 5 Radiation and Diffraction
Figure 5.4 indicates the contour C
s
passing through the saddle point θ
1
, as
well as the two limiting lines at α
1
=±π/2 +θ
1
. Note that depending upon
θ
1
, the poles at α
r
and π −α
r
may or may not be enclosed. Problem 5.4 asks
the reader to evaluate the contribution from α
r
. Suppose that θ
1
is such that in
distorting C to C
s
the pole at α
r
is enclosed. From (5.77), the condition for this
ﬁrst to happen is that cos θ
1
cosh(−i α
r
) =1. This is equivalent to the condition
that cos θ
1r
=c
r
/c
L
, where we have added the subscript r to indicate this special
value. Thus for values of θ
1r
< θ
1
< π/2, no Rayleigh wave is present. If we
inverted our result in time we should ﬁnd that this is equivalent to asserting
that the Rayleigh wave cannot be excited before the incident compressional
disturbance strikes the surface.
Note also that C
s
passes onto the other Riemann sheet – the dashed portion
in Fig. 5.4 – over part of its path. One must be careful to ensure that when this
happens that one can reemerge onto the Riemann sheet of physical interest. Oc
casionally one must also be careful that one does not forget to include any pole
contributions that may arise from poles on this other sheet, should they be en
closed by this excursion. When one cannot reemerge onto the Riemann sheet of
physical interest, then the contour C
s
must instead be wrapped around the branch
cut, on the physically meaningful Riemann sheet, and this contribution asymp
totically approximated. Harris and Pott (1985) discuss such contributions.
The remaining issue before using the steepest descents result (5.60) is to settle
what the argument of the square root is. Following the rule set out in Proposition
5.2 , arg[−2/d
2
α
q(θ
1
)]
1/2
=3π/4. Putting the various pieces together, we ﬁnd
that
u
sL
∼
A
4πk
L
2π
k
L
r
1/2
ˆ p
1
(θ
0
)R
L
(θ
0
)e
i k
L
r
e
−i π/4
, k
L
r →∞, (5.78)
where F
0
= A/k
2
L
has been used. Note that this has given us a representation
of the kind investigated in Section 2.4. The overall structure of (5.78) is that of
a farﬁeld expression for a cylindrical wave radiating from a virtual source at
(0, −h), in agreement with what we indicated previously. At the surface x
2
=0,
r =h/ sin θ
1
is the radius of curvature of the reﬂected cylindrical wave as it is
just about to form.
5.4.3 The Scattered Shear Wave
Equation (5.23), which is repeated in the following equation, is a somewhat
harder integral to approximate.
u
sT
=−
k
L
F
0
4π
C
ˆ
d
2
(α)R
T
(α)e
i k
T
ˆ p
2
·x
e
i k
L
h sin α
dα. (5.79)
5.4 Buried Harmonic Line of Compression II 99
The complexity arises from the mixing of wave types: an incident compres
sional wave whose phase lingers in the term e
i k
L
h sin α
and the scattered shear
wave whose phase is e
i k
T
ˆ p
2
·x
. Finding C
s
and investigating it in the neighbor
hood of the stationary point is similar to that for the compressional case. We
therefore know approximately what C
s
looks like and can apply (5.60) without
working out C
s
in detail. Instead, the key to understanding (5.79) can be found
by considering the geometry of the rays. We know from (5.78) that a steepest
descents approximation will give phase terms that are those of a ray theory.
The ﬁrst clue can be found by examining (2.51) and (2.56). These expressions
are singular when the distance along the ray is the negative of the radius of
curvature. That is, the radius of curvature has its origin either at a point on a
caustic surface or at a single point. We have seen that the reﬂected compressional
wave (5.78) appears to originate at the virtual source point (0, −h). In a similar
way we should expect that the reﬂected shear wave will appear to originate,
if not from a virtual source point, then from a point on a virtual caustic. We
have sketched this possibility in Fig. 5.6. The radius of curvature ρ
T
measures
the distance from a point on the virtual caustic to a point on the surface x
2
=0
at which the scattered shear wave is starting to form. Let s
0
be the distance
the compressional ray propagates from (0, h) to this same point on the surface
and s
2
be the distance the reﬂected shear ray propagates from this point to the
Fig. 5.6. Asketch of the shear ray scattered fromthe surface when struck by an incident
compressional ray. The compressional ray originates at (0, h) and propagates a distance
s
0
, striking the surface at an angle θ
0
. The scattered shear ray emerges at an angle θ
2
.
It appears to originate from a point on a virtual caustic located by extending the ray
backward a distance ρ
T
.
100 5 Radiation and Diffraction
observation point (x
1
, x
2
). We do not at present have an analytic expression
for ρ
T
, nor do we have one for the virtual caustic. However, we know that
the asymptotic expansion of (5.79) must contain the term (1 +s
2
/ρ
T
)
1/2
in its
denominator so that we can identify ρ
T
from the ﬁnal asymptotic expression.
Setting (x
1
−s
0
cos θ
0
) =s
2
cos θ
2
and x
2
=s
2
sin θ
2
, we express (5.79) as
u
sT
=−
k
L
F
0
4π
C
s
ˆ
d
2
(α)R
T
(α)e
(k
T
s
2
+k
L
s
0
)q(α)
dα, (5.80)
where
q(α) =i
k
T
s
2
(k
T
s
2
+k
L
s
0
)
cos( ¯ α −θ
2
) +i
k
L
s
0
(k
T
s
2
+k
L
s
0
)
cos(α −θ
0
). (5.81)
The parameter (k
T
s
2
+k
L
s
0
) is large. The stationary point, a saddle point, is
given by ¯ α =θ
2
and α =θ
0
. This is a manifestation of Fermat’s principle that
the ray path is an extremum. The unknowns at this point can be taken as s
0
, s
2
,
θ
0
, and θ
2
, while x
1
, x
2
and h (=s
0
sin θ
0
) can be considered as given. Note, as
well, that the phasematching condition c
−1
L
cos θ
0
=c
−1
T
cos θ
2
gives a fourth
relation. Thus we may solve for the unknowns in terms of what is given.
We have enough information at this point to approximate (5.80) by us
ing (5.60). This gives, after some algebraic manipulation,
u
sT
∼
A
4πk
L
2π
(k
L
s
0
)(1 +s
2
/ρ
T
)
1/2
×
ˆ
d
2
(θ
0
)R
T
(θ
0
)e
i (k
T
s
2
+k
L
s
0
)
e
−i π/4
, k
T
s
2
+k
L
s
0
→∞, (5.82)
where again F
0
= A/k
2
L
has been used. The radius of curvature ρ
T
is given by
ρ
T
=s
0
c
L
c
T
sin
2
θ
2
sin
2
θ
0
, (5.83)
and s
0
, s
2
, θ
0
, and θ
2
are determined in terms of x
1
, x
2
, h, and the ratio κ =
c
L
/c
T
. κ is given in terms of Poisson’s ratio by (3.13).
In closing we note that the contour in the β plane, C
β
, Fig. 5.3, could be
distorted so as to wrap around the Rayleigh pole and the branch cuts. This
would give a representation of the waveﬁelds u
sL
and u
sT
that corresponds to
an eigenfunction expansion. Of particular interest is the fact that the spectrum
has a discrete eigenvalue, the contribution of the Rayleigh pole, along with a
continuous distribution of eigenvalues, the contribution from the branch cuts.
Problem 5.4 The Rayleigh Wave
Calculate the Rayleighpole contributions to u
sL
and u
sT
, and hence the
Rayleigh wave excited by a harmonic line of compression at (0, h). There are a
5.5 Diffraction of an Antiplane Shear Wave at an Edge 101
number of starting points. I should likely begin with (5.22) and (5.23), but this
is not the only starting point.
5.5 Diffraction of an Antiplane Shear Wave at an Edge
The previous problem has considered scattering of a cylindrical wave from an
inﬁnite tractionfree surface. The complications came about not only because
both compressional and shear waveﬁelds were scattered from the surface, but
also because the incident waveﬁeld was composed of a complete spectrum of
plane waves. In this section we consider a related problem: one in which the
scatterer has an edge that excites a full spectrum of plane waves, though the
incident wave is a single plane one. Speciﬁcally, we consider a time harmonic,
antiplane shear wave normally incident to a semiinﬁnite slit or crack. The
crack lies along the positive x
1
axis. On both sides, the traction acting on the
surfaces of the crack is zero. Figure 5.7 indicates the geometry of the problem.
The equations of motion are given by (1.14) and (1.15), with the corresponding
timeharmonic equation being given by (2.20). As we have done previously
when dealing with antiplane shear waves, we simplify the notation by dropping
Crack
1
4
5 2
3
x
1
x
2
Fig. 5.7. Asketch of the crack, the cylindrical diffracted wave emanating fromthe crack
tip, and the reﬂected and transmitted geometrical waves. The rippled arrows indicate
their directions of propagation. Separating the diffracted and geometrical waveﬁelds, in
regions 1,2, and 3, are parabolic shaped, transition, or boundary layers, labeled regions
4 and 5.
102 5 Radiation and Diffraction
the subscript T and use c and k =ω/c as the wavespeed and wavenumber,
respectively.
5.5.1 Formulation
The plane wave u
i
3
, described by
u
i
3
=(A/k)e
i kx
2
, (5.84)
is incident to the crack. The constant A is dimensionless, having been made
so by dividing by k, the wavenumber. The total waveﬁeld u
t
3
=u
i
3
+u
3
, where
u
3
is the scattered waveﬁeld. The boundary and continuity conditions to be
satisﬁed at x
2
=0 are
µ∂
2
u
t
3
(x
1
, 0
±
) =0, x
1
> 0; u
t
3
(x
1
, 0
+
) =u
t
3
(x
1
, 0
−
), x
1
< 0. (5.85)
The superscripts ± indicate that the boundary is approached through positive
or negative values of x
2
, respectively. Also we ask that the scattered wave be
outgoing from its source, the crack, and that it therefore satisfy the principle of
limiting absorption (Section 4.4).
Note that x
2
=0 is a plane of reﬂection symmetry for the problem. As a way
to formulate the problem for the scattered disturbance, the problem is divided
into two, one symmetric and one antisymmetric with respect to this plane.
To begin, we divide the incident waveﬁeld into symmetric and antisymmetric
components as follows.
u
i
3
=(A/2k)(e
i kx
2
+e
−i kx
2
) +(A/2k)(e
i kx
2
−e
−i kx
2
), (5.86)
so that u
i
3
=u
i s
3
+u
i a
3
. We next set the scattered waveﬁeld u
3
=u
s
3
+u
a
3
where
u
s
3
is symmetric and u
a
3
antisymmetric with respect to x
2
=0. The symmet
ric scattered waveﬁeld is excited by the incident symmetric one and the an
tisymmetric scattered waveﬁeld by the incident antisymmetric one. The total
waveﬁeld u
t
3
=u
t s
3
+u
t a
3
, where u
t s
3
and u
t a
3
are the symmetric and antisymmet
ric components, respectively. Lastly, each waveﬁeld, symmetric and antisym
metric, separately satisﬁes the boundary and continuity conditions (5.85). The
problem has therefore been divided into two separate ones. Reﬂecting the sym
metric problemin the plane x
2
=0 leaves the particle displacement unchanged,
while reﬂecting the antisymmetric one multiplies it by −1.
We consider the symmetric problem. By construction, ∂
2
u
i s
3
=0 ∀ x
1
at x
2
=
0. From the symmetry, ∂
2
u
s
3
=0 ∀ x
1
at x
2
=0. This and the condition that
the scattered waveﬁeld propagate outward imply that u
s
3
≡0. Therefore the
symmetric part of the incident waveﬁeld remains unaffected by the presence of
5.5 Diffraction of an Antiplane Shear Wave at an Edge 103
the crack in its path of propagation. The complete problem is then equivalent
to the antisymmetric one and the scattered waveﬁeld is antisymmetric in x
2
.
Reasserting the decomposition u
t
3
=u
i
3
+u
3
, but noting that u
3
=u
a
3
, we
are led to reformulate the problem, in the halfspace x
2
> 0, for the scattered
waveﬁeld u
3
as follows.
∂
α
∂
α
u
3
+(k +i )
2
u
3
=0, (5.87)
subject to the conditions, at x
2
=0,
∂
2
u
3
(x
1
, 0) =−i Ae
−x
1
, x
1
> 0; u
3
(x
1
, 0) =0, x
1
< 0. (5.88)
As well, the condition that u
3
be outgoing and satisfy the principle of limiting
absorption is enforced. It is for this latter reason that the i , 0 < 1, has
been explicitly added to k in (5.87). We set k
=k +i , with k taken as real and
positive. The addition of the e
−x
1
in the boundary condition is an artiﬁce also
neededtoenforce the principle of limitingabsorption. Inthis particular problem,
because the incident wave strikes the crack normally, the scattered waveﬁeld
will contain geometrically reﬂected waves that, as with the incident wave, do
not vanish as kx
1
 →∞unless this artiﬁce is used. However, it is not needed
for other angles of incidence, as we indicate in Problem 5.5. Lastly, we must
also append to these equations an edge condition as explained in Section 4.7.2.
In fact we use the analysis of Section 4.7.3 to demand that
ku
3
=O
(kr)
1/2
, kr →0, (5.89)
where r =(x
2
1
+x
2
2
)
1/2
. This condition is essential to ensure that the solution
be unique. The scattered waveﬁeld for x
2
< 0 is found by using the fact that
u
3
(x
1
, x
2
) =−u
3
(x
1
, −x
2
).
We could solve this problem by formulating an integral equation for the
scattered waveﬁeld following the ideas outlined in Problem4.3. We should then
ﬁnd that the integral equation can be solved by using the Wiener–Hopf method
(Titchmarsh, 1948). However, it is easier, and perhaps just as informative, to
solve the problem directly with the Wiener–Hopf method as used by Jones
(Jones, 1952; Noble, 1988).
5.5.2 Wiener–Hopf Solution
We begin by setting
u
3
(x
1
, 0) =
0, x
1
< 0,
u
3+
, x
1
> 0,
(5.90)
104 5 Radiation and Diffraction
∂
2
u
3
(x
1
, 0) =
τ
−
, x
1
< 0,
τ
+
=−i Ae
−x
1
, x
1
> 0.
(5.91)
Our strategy in solving the problem for the scattered waveﬁeld is to ﬁnd a
functional relation between the two unknowns τ
−
and u
3+
.
We seek a solution to (5.87) by using a representation very similar to that
used previously in Section 2.3.1 (the roles of x
1
and x
2
are interchanged). That
is, we set
u
3
=
1
2π
∞
−∞
∗
u
3+
e
i (βx
1
+γ x
2
)
dβ, (5.92)
where γ =(k
2
−β
2
)
1/2
. The transform
∗
u
3+
remains to be determined. Recall
that x
2
> 0 so that the β plane is cut such that (γ ) ≥0, ∀β. This is the same
choice of Riemannsheet as was takeninSection5.2andﬁrst explicitlydiscussed
in Section 3.4.4. Figure 5.8 shows the branch points with the accompanying
cuts.
Imposing the conditions at x
2
=0 on (5.92) and using (5.90) gives the
following:
∗
u
3+
=
∞
0
u
3+
(x
1
)e
−iβx
1
dx
1
. (5.93)
With some thought, one realizes that, to the right along the crack in Fig. 5.6,
the dominant waveﬁeld is a plane one, propagating normally away from the
Fig. 5.8. The complex β plane showing the branch cuts, pole at i , and strip of common
analyticity 2 wide. The contour for the inverse transform initially lies within this strip.
As →0 the pole will approach a position just above this contour.
5.5 Diffraction of an Antiplane Shear Wave at an Edge 105
crack, but one that is forced to decay as x
1
→∞ because of the e
−x
1
placed
in the ﬁrst of (5.88). Any other wave present will originate from the crack tip.
6
That part grazing the crack face will also decay as e
−x
1
because k
=k +i .
Thus ku
3+
= O(e
−x
1
) as kx
1
→∞, from which it follows that
∗
u
3+
is an
analytic function of β for (β) < . Noble (1988) discusses in some detail
the conditions needed to ensure that a transform is an analytic function of its
independent variable, while Titchmarsh (1939) discuses this somewhat more
generally. In essence, the transform is an analytic function of its variable β
for those regions of the complex β plane wherein the integral is uniformly
convergent. Also note that the condition that u
3
=0 for x
1
< 0, x
2
=0 has been
built into
∗
u
3+
.
Working next with (5.91), we ﬁnd that
∗
τ
+
=
∞
0
τ
+
(x
1
)e
−iβx
1
dx
1
=
−A
(β −i )
. (5.94)
Note the pole at i . Further we deﬁne the transform
∗
τ
−
as
∗
τ
−
=
0
−∞
τ
−
(x
1
)e
−iβx
1
dx
1
. (5.95)
The crack tip is the only possible source for a scattered wave in x
1
< 0. A
cylindrical scattered wave is emitted from the tip. Thus τ
−
= O(e
−x
1

) as
kx
1
→−∞, because k
=k +i , from which it follows that
∗
τ
−
is an analytic
function of β for (β) >−.
We demand that ∂
2
u
3
calculated from (5.92) be consistent with that repre
sented by (5.94) and (5.95) at x
2
=0. This gives
i γ
∗
u
3+
=[−A/(β −i )] +
∗
τ
−
, (5.96)
which is the functional relation we are looking for. Note that different parts of
it are analytic in different parts of the complex β plane, but that all parts are
analytic inthe commonstrip(β) ∈ (−, ). The goal of the next fewparagraphs
will be to rearrange this expression so that one side is analytic in the region
(β) >−, while the other is analytic in the region (β) < . The two sides will
6
We need to guess at the kinematics of the scattered wave to deduce where
∗
u
3+
and
∗
τ
−
are
analytic. Experience with similar problems is how this is done. However, the ﬁnal answer must
be checked to ensure that it is consistent with these assumptions.
106 5 Radiation and Diffraction
be equal in the common strip and hence will be different representations of the
same function, which function must then be entire. This process is the Wiener–
Hopf method. Proceeding to this relation directly from the transforms, rather
than through ﬁrst forming an integral equation, is the simpliﬁcation introduced
by Jones (1952).
We next write (5.96) as
i (k
−β)
1/2 ∗
u
3+
=
−A
(β −i )(k
+β)
1/2
+
∗
τ
−
(k
+β)
1/2
, (5.97)
so that its left side becomes an analytic function for (β) < . The right side
is almost one for (β) >−, but for the presence of the pole at i . To isolate
the pole term we add and subtract the residue multiplied by (β −i )
−1
. The
outcome is
i (k
−β)
1/2 ∗
u
3+
+
A
(β −i )(k
+i )
1/2
=−A
(k
+i )
1/2
−(k
+β)
1/2
(β −i )(k
+β)
1/2
(k
+i )
1/2
+
∗
τ
−
(k
+β)
1/2
. (5.98)
We have succeeded in achieving our goal. The left side is analytic for (β) <
and the right side for (β) >−, while the two sides are equal in the common
strip (β) ∈ (−, ). The two sides are thus the analytic continuations of one
another, and together they represent a function analytic everywhere in the ﬁnite
β plane. The function is entire and its nature is determined by its behavior at
inﬁnity.
We have still not used the edge condition. This is the condition that tells us
how the entire function behaves at inﬁnity. Recall that we asked that ku
3+
=
O[(kx
1
)
1/2
] as kx
1
 →0. From this we may infer that τ
−
= O[(kx
1
)
−1/2
]
as kx
1
 →0. Adapting an Abelian theorem k
2 ∗
u
3+
= O(k
3/2
β
−3/2
) and
that k
∗
τ
−
= O(k
1/2
β
−1/2
) as k
−1
β →∞. Using Liouville’s theorem
(Titchmarsh, 1939) it follows that the entire function must be identically zero.
Therefore
∗
u
3+
=
i A
(β −i )(k
+i )
1/2
(k
−β)
1/2
. (5.99)
Note how the edge condition was essential to the determination of a unique
solution.
5.5 Diffraction of an Antiplane Shear Wave at an Edge 107
At this point has served its purpose and we take the limit →0. We are
thus left with an integral expression for the scattered waveﬁeld, namely
u
3
=
i A
2πk
1/2
∞
−∞
e
i (βx
1
+γ x
2
)
β(k −β)
1/2
dβ, (5.100)
where the contour passes below β =0. Integrals of this form or of the form
achieved after using the Sommerfeld transformation are sometimes referred
to as diffraction integrals. In closing, we recall that this is the solution to the
scattered waveﬁeld only for x
2
> 0.
Problem 5.5 An Arbitrary Angle of Incidence
Problem1. Repeat the calculations leading to (5.100) for an arbitrary angle
of incidence θ
0
. The incident wave is given by
u
i
3
=(A/k)e
i k
(cos θ
0
x
1
+sin θ
0
x
2
)
, (5.101)
where k
=k +i and > 0. Initially assume that θ
0
< π/2. Why might this
be a useful restriction? Once the solution is obtained, can the restriction be
removed? Show that the scattered waveﬁeld u
3
is given by
u
3
=
i A
2πk
1/2
sin θ
0
e
i k(cos θ
0
x
1
)
(1 +cos θ
0
)
1/2
∞
−∞
e
i (βx
1
+γ x
2
)
(β −k cos θ
0
)(k −β)
1/2
dβ. (5.102)
Problem 2. The integral (5.102) has a pole at β =k cos θ
0
. Calculate an
asymptotic approximation to (5.102) that is not uniform. Include the pole term
when it it is needed and note when the asymptotic approximation breaks down.
When we calculate a steepest descents approximation to (5.102), we must
take care that the stationary point does not lie near the pole, because in that case
we can no longer argue that the integrand, aside from the exponential term, is
slowly varying. An asymptotic approximation, calculated when one parameter
is large, that becomes disordered for certain values of a second parameter, in
this case the angle of incidence θ
0
, is said not to be uniform. It is uniform when
the approximation is accurate for all values of the second parameter.
5.5.3 Description of the Scattered Waveﬁeld
We have indicated ﬁve regions in Fig. 5.6, each of which has a somewhat
different waveﬁeld. In region 1 the waveﬁeld u
3
is composed of a residue term,
from the pole in the integrand of (5.100) at β =0, plus the integral itself. The
108 5 Radiation and Diffraction
residue contribution cancels the u
i
3
so that the crack casts the region behind
it into partial silence. The integral represents a diffracted wave that appears
as a cylindrical one radiating from the crack tip. This is the only wave that
penetrates the silence. In region 2 the waveﬁeld u
3
consists solely of the integral
representing the diffracted wave, though the total waveﬁeld would include u
i
3
as well. In region 3 the waveﬁeld u
3
is again composed of a residue term plus
the integral itself. However, note that in using the antisymmetry, namely that
u
3
(x
1
, x
2
) =−u
3
(x
1
, −x
2
) to calculate u
3
for x
2
< 0, the residue contribution
gives the wave reﬂected fromthe crack, while the integral continues to represent
the diffracted wave. The incident wave u
i
3
must be added to this to produce the
total waveﬁeld. Regions 4 and 5 are transition or boundary layers. Moving from
left to right through these layers, the scattered waveﬁeld makes a transition
from solely a diffracted wave to a diffracted plus geometrical wave. These
regions are characterized by Fresnel integrals, as we subsequently demonstrate.
Within these regions the diffracted wave comes close to phase matching to the
geometrical wave so that the two kinds of waves strongly interact, thereby
producing a complicated waveﬁeld, but propagate independently elsewhere.
Note that (5.100) contains both the diffracted wave and geometrical terms.
A uniform asymptotic approximation to (5.100) is calculated in Felsen and
Marcuvitz (1994).
7
However, most of the manipulations undertaken to do so
are nomore difﬁcult thanthose neededtoreduce (5.100) toanexact combination
of Fresnel integrals. This reduction is described in detail in the Appendix. The
Fresnel integral is deﬁned as
F(z) :=
∞
z
e
i ξ
2
dξ. (5.103)
We introduce the polar coordinates (r, θ), where x
1
=r cos θ and x
2
=r sin θ.
For x
2
> 0, θ ∈ (0, π], while for x
2
< 0, θ ∈ (0, −π]. The outcome of the cal
culations described in the Appendix is that
u
3
=
e
−i π/4
A
π
1/2
k
e
−i kr cos(θ−π/2)
F
(2kr)
1/2
cos
θ −π/2
2
−e
−i kr cos(θ+π/2)
F
−(2kr)
1/2
cos
θ +π/2
2
. (5.104)
It is important to note that this expression assumes that x
2
> 0 or, equivalently,
7
The calculation is done by separating a term that contains the pole from the rest of the integrand.
One way to do this is to subtract and add the integrand of (5.125). The integral with no pole term
in its integrand is approximated in the usual way, while the integral with the pole term becomes
a Fresnel integral.
5.5 Diffraction of an Antiplane Shear Wave at an Edge 109
θ ∈ (0, π]. To obtain an expression for x
2
< 0, we use the fact that u
3
(r, θ) =
−u
3
(r, −θ).
To calculate the total waveﬁeld, we use the following property of the Fresnel
integral:
F(z) + F(−z) =π
1/2
e
i π/4
. (5.105)
The total waveﬁeld u
t
3
, for θ ∈ (−π, π], is, after adding u
i
3
to (5.104), given by
u
t
3
=
e
−i π/4
A
π
1/2
k
e
−i kr cos(θ−π/2)
F
(2kr)
1/2
cos
θ −π/2
2
+e
−i kr cos(θ+π/2)
F
(2kr)
1/2
cos
θ +π/2
2
. (5.106)
For kr 1 these integrals can be asymptotically approximated. Problem 5.7
describes the asymptotic approximations to the Fresnel integral. When this is
done we ﬁnd, for θ ∈ (0, π], that
u
t
3
∼
A
k
e
i kx
2
H(−x
1
) + D(θ, π/2)
A
k
e
i kr
(kr)
1/2
, kr →∞, (5.107)
where
D(θ, π/2) =
e
i π/4
2(2π)
1/2
1
cos[(θ −π/2)/2]
+
1
cos[(θ +π/2)/2]
. (5.108)
We ﬁnd a very similar expression for θ ∈ (0, −π]. Note the similarity in struc
ture between the expansion (5.107) and the earlier expansion (2.41). The coef
ﬁcient D(θ, π/2) is called a diffraction coefﬁcient. Its arguments are intended
to suggest that a ray incident to the crack tip at an angle π/2 excites a diffracted
ray that leaves the tip at an angle θ. In fact, there are a whole fan of such rays
as θ is allowed to take on all its values. These coefﬁcients play an important
role in the geometrical theory of diffraction for edges. The interested reader
can pursue this theory further in books by Achenbach et al. (1982), Babiˇ c and
Buldyrev (1991), and Jull (1981). Problem 5.6 introduces some of the ideas
used in this description of diffraction.
The integrals of (5.106) can also be approximated for kr 1, as Problem
5.7 indicates. Any singular behavior as kr →0 is contained entirely within u
3
,
whose approximation in this limit is
ku
3
= O[(kr)
1/2
], kr →0. (5.109)
This reproduces the edge condition we enforced in (5.89).
110 5 Radiation and Diffraction
The approximation (5.107) breaks down near θ =±π/2. Examining (5.106),
we note that one or the other of the arguments of the Fresnel functions changes
sign at these angles. Let us consider the neighborhood of π/2 so that it is the
second Fresnel integral in (5.106) whose argument changes sign. Examining
this argument, we ﬁnd that if θ is close to π/2 the argument of the Fresnel
integral is too small to approximate it asymptotically, despite kr being quite
large. Thus if
(2kr)
1/2
cos
θ +π/2
2
≤1, (5.110)
suchanapproximationis not accurate. Takingthe equalityas formingthe bound
ary of the region outside of which the asymptotic approximation is sufﬁciently
accurate, we ﬁnd that the boundary is described by kr(1 −sin θ) =1, the equa
tion of a parabola in polar form. This then marks the boundary of region 4 with
regions 1 and 2. The boundary layer lies within this boundary, and its scale is
set by the expression on the left in (5.110). The boundary of region 5 is simply
the reﬂection in x
2
=0 of that for region 4.
Problem 5.6 The Geometrical Theory of Diffraction
Consider an antiplane shear wave normally incident to the opening between
two cracks, as shown in Fig. 5.9. Estimate the waveﬁeld diffracted by the strip
by using what has been learned fromthe semiinﬁnite crack problem. Howdoes
this problem, aside from the fact that the time dependence is harmonic, differ
from Problem 5.1?
Fig. 5.9. Asketch of the strip indicated by the dashed line, formed by two semiinﬁnite
cracks, indicated by the solid lines. Each crack tip ±b serves as the origin for one of the
polar coordinate systems (s
i
, θ
i
). Here i =1 at b and 2 at −b.
5.6 Matched Asymptotic Expansion Study 111
As before, express the total waveﬁeld as u
t
3
=u
i
3
+u
3
. The incident wave
is given by (5.84). Are the boundary and continuity conditions at x
2
=0 the
following?
µ∂
2
u
t
3
(x
1
, 0
±
) = 0, x
1
∈ (−∞, −b) ∪(b, ∞),
u
t
3
(x
1
, 0
+
) = u
t
3
(x
1
, 0
−
), x
1
∈ (−b, b). (5.111)
Formulate the problemfor the scattered waveﬁeld u
3
, in x
2
> 0, stating all the
conditions that must be satisﬁed. An exact analytic solution to this problemmay
not be possible. Explain, using what has been learned from the semiinﬁnite
crack problem, why the following expression is a plausible asymptotic solution
to this problem
u
t
3
=
A
k
[H(x
1
+b) − H(x
1
−b)]e
i kx
2
+ D
π −θ
2
,
π
2
A
k
e
i ks
2
(ks
2
)
1/2
+ D
θ
1
,
π
2
A
k
e
i ks
1
(ks
1
)
1/2
, (5.112)
where the coordinates (s
1
, θ
1
) and (s
2
, θ
2
) are deﬁned in Fig. 5.9 and D(θ, π/2)
is given by (5.108). This solution is best for 2kb 1. Why?
Consider the boundary layers surrounding θ
1,2
=π/2. They growin width as
one moves away fromthe edges. At what point do they choke off the geometrical
term given by the ﬁrst term of (5.112)? The earlier parts of Harris (1987) might
be useful, but are not in any way essential, in answering this last question.
5.6 Matched Asymptotic Expansion Study
We have just undertaken a relatively complex solution to what, looking at
Fig. 5.7, might seem a simple wave process. After all, to some leading order
of approximation, the crack merely blocks the incident wave from reaching
the other side. In this closing section we construct an asymptotic solution that
captures this idea mathematically. In writing this I have followed a similar
discussion in Zauderer (1983) and beneﬁted from a very detailed analysis of
edge diffraction by Gautesen (1979).
In Section 2.4 we explored howwaves could be approximated by a ray theory
and provided an asymptotic structure that explained the connection. Using this
structure, we begin our analysis of the diffraction problem by assuming that the
waveﬁeld can be described with the asymptotic approximation
u
3
∼e
i kS(x
1
,x
2
)
n≥0
(−i k)
−n
A
n
(x
1
, x
2
). (5.113)
112 5 Radiation and Diffraction
This assumption leads to the eikonal equation (2.43) for S and the transport
equation (2.44) for A
0
.
First, we consider the region x
1
> 0. To match the phase of the incident
wave u
i
3
at x
2
=0, given by (5.84), we ask that S(x
1
, 0) =0 and that the scat
tered waves be outgoing fromthe crack. The appropriate solution is S(x
1
, x
2
) =
±x
2
H(x
1
), where the plus sign is used for x
2
> 0 and the minus sign for x
2
< 0,
and H(x
1
) is the Heaviside function. Knowing S, the transport equation, (2.44),
indicates that A
0
can have at most an x
1
dependence. Moreover, it must be such
that the traction for the total waveﬁeld vanishes on both sides of the crack. The
appropriate solution is A
0
(x
1
) =∓(A/k) H(x
1
), where the minus sign is used
for x
2
> 0 and the plus sign for x
2
< 0. It also follows that A
n
≡0 for n ≥1.
Second, we consider the region x
1
< 0. There are no boundary conditions to
enforce, leading to the conclusion that A
n
≡0 for n ≥0. Therefore, to leading
order the total waveﬁeld u
t
3
is given by
u
t
3
(x
1
, x
2
) =
(A/k)e
i kx
2
, region 2,
(A/k)(e
i kx
2
+e
−i kx
2
), region 3,
0, region 1,
(5.114)
where the regions are those indicated in Fig. 5.7.
This solution gives the geometrical part of (5.106), but does not capture
the transition in regions 4 and 5. Even in the absence of an exact solution,
we might suspect that our asymptotic ansatz does not have enough struc
ture to capture the effects of the rapid changes in the waveﬁeld near x
1
=0
just from the apparent discontinuity there. We have encountered a bound
ary layer. What we need to do is scale the problem in such a way that the
boundary layer is opened up and an equation governing the waveﬁeld within
it found. The scaling needed to open the layer up is usually not known and
must be found as part of discovering the governing equation. We then con
struct the solution to this equation so that it matches the surrounding waveﬁeld
to some order of approximation. Such a procedure is called matched asymp
totic expansions. Both Hinch (1991) and Holmes (1995) discuss this technique
thoroughly.
Moreover, recall that, from our analysis in Section 4.7.3, we know that very
near the crack tip the waveﬁeld is quasistatic, with part of it behaving as
(kr)
1/2
. In addition to the boundary layer, that we are about to examine, there
is a nearﬁeld or inner region that is not described by the asymptotic anzatz
(5.113). However, for the present we avoid this nearﬁeld.
5.6 Matched Asymptotic Expansion Study 113
We consider the case that x
2
> 0 so that θ is near π/2 and leave the case x
2
< 0
to the reader. Examining (5.106), we note that in this region a propagator term
e
i kx
2
always appears. Accordingly, we set
u
3
=w(x
1
, x
2
)e
i kx
2
(5.115)
and arrive at the following equation for w:
2i k∂
2
w+∂
2
2
w+∂
2
1
w =0. (5.116)
We have stripped away the oscillatory part of the waveﬁeld. At this point we
note that if w varies rapidly in the neighborhood of x
1
=0, then ∂
2
1
w may be
very large and to solve the equation we shall need another term to balance it.
The term 2i k∂
2
w seems a likely candidate because it is multiplied by the large
parameter k.
We have indicated previously that a length scale is very important, as it sets
a gauge to measure large and small. The diffraction problem does not have a
natural length scale, so that we must introduce one somewhat artiﬁcially, just
as we did when examining the local ﬁeld near the crack tip in Section 4.7.3. We
again take the length as a reference length. It might represent a wavelength
at a reference frequency or a distance at which a measurement is made. We
then deﬁne the nearﬁeld as that for which k 1 and the farﬁeld as that for
which k 1. We are examining the waveﬁeld in the region k 1 with θ
near π/2.
We scale the problem by setting ¯ x
i
= x
i
/ and ¯ w =wk/A (recall that A/k is
a magnitude of the incident wave) and reexpress (5.116) in terms of the scaled
variables. Having done this, we now omit the overbar. Equation (5.116) has
become
2i (k)∂
2
w+∂
2
2
w+∂
2
1
w =0. (5.117)
The argument, expressed in terms of the scaled (x
1
, x
2
), of the second Fresnel
function in (5.106), namely −x
1
(k/2x
2
)
1/2
, suggests the scaling needed to
open up the boundary layer. We set y
1
=(k)
β
x
1
with β =1/2. This change
of coordinate is called introducing a stretching transformation. Note, however,
that we do not in general know β and must determine it from the rescaled
equation by balancing the various terms (Hinch, 1991; Holmes, 1995). In this
case it is the ﬁrst and third terms in (5.117) that balance. The coordinates (y
1
, x
2
)
are called the inner coordinates and the (scaled) (x
1
, x
2
) the outer coordinates.
114 5 Radiation and Diffraction
We next introduce an asymptotic expansion of the form
w ∼w
0
(y
1
, x
2
) +(k)
γ
w
1
(y
1
, x
2
) . . . , (5.118)
where γ > 0. The leadingorder term is governed by the parabolic Schrodinger
equation,
2i (∂w
0
/∂x
2
) +
∂
2
w
0
∂y
2
1
=0. (5.119)
This is the equation governing the behavior of the waveﬁeld in the boundary
layer.
Recalling the fundamental or causal Green’s function for the diffusion equa
tion (Zauderer, 1983), we write the solution to (5.119) as
w
0
=
1
x
1/2
2
∞
−∞
f
(k)
1/2
s
e
i (y
1
−s)
2
/(2x
2
)
ds. (5.120)
Toﬁndthe unknown f (x) we express this equationinterms of the outer variables
as
w
0
=
(k)
1/2
x
1/2
2
∞
−∞
f (s)e
i k(x
1
−s)
2
/(2x
2
)
ds. (5.121)
The reader should recall that we are assuming x
2
> 0. This integral can be
asymptotically expanded for large k, provided f (s) does not vary rapidly near
the stationary point s = x
1
. However, the function f (s) must vary rapidly near
x
1
=0 if it is to capture the transition in the boundary layer, so that an asymp
totic approximation there will not be accurate. However, for x
1
 > 0 we expect
that the asymptotic approximation to (5.121) should match our earlier approx
imation to u
3
. That is, w
0
∼0 when x
1
< 0 and w
0
∼−1 for x
1
> 0. Using the
stationary phase approximation, (5.69), or the steepest descents approximation,
we ﬁnd that
w
0
∼ f (x
1
)(2π)
1/2
e
i π/4
. (5.122)
Thus we ﬁnd for x
2
> 0 that
f (x
1
) =−e
−i π/4
/(2π)
1/2
H(x
1
). (5.123)
This process of ﬁnding f (x
1
) is referred to as matching the inner and outer
expansions. At this point we could now match the nearﬁeld expansion (4.52)
to (5.120) to determine unequivocally that β =1/2 and also to determine the
unknown constant A in the nearﬁeld expansion (4.52) (Gautesen, 1979).
Appendix: The Fresnel Integral 115
Removing all the scaling so that we may compare our result with (5.104), we
ﬁnd that
u
3
∼−
Ae
−i π/4
kπ
1/2
e
−i kx
2
F
−x
1
k
2x
2
1/2
, (5.124)
where F(z) is the Fresnel integral deﬁned previously by (5.103). For x
2
> 0 and
x
1
≈0, (5.104) reduces to this expression. If we need a boundary layer solution
for x
2
< 0, region 5, we use the antisymmetry of the scattered waveﬁeld, namely
u
3
(x
1
, x
2
) =−u
3
(x
1
, −x
2
), which is a global property of the solution.
Had we expanded (5.121) to a second term, we should have encountered a
term O[(k)
−1/2
] but found no similar term in the expansion (5.113) to match
it. Why? The ansatz (5.113) contains only the geometrical waveﬁeld and not the
diffracted one. Examining (5.107), we see immediately that the missing term
comes from the diffracted waveﬁeld. To include this possibility we should have
allowed for fractional powers of k in positing (5.113) just as we did with the α in
(2.41). Had we not known the solution nor suspected what was going on, the fail
ure to match at higher order would have indicated that something was missing.
Appendix: The Fresnel Integral
Our purpose is to derive the expression (5.104). The integral to be evaluated is
(5.100), which we rewrite here.
u
3
=
i A
2πk
1/2
C
β
e
i (βx
1
+γ x
2
)
β(k −β)
1/2
dβ. (5.125)
An integral that has a pole and stationary point (usually a saddle), points that
maycoalesce for certainvalues of the physical coordinates, is calleda diffraction
integral. The essence of its evaluation is to reduce it to one on its contour of
steepest descents. I follow the derivation given in Born and Wolf (1986). It is
repeated here so that the calculation begun in Section 5.5 can be completed.
Recall that the diffraction integral is evaluated along the contour C
β
that
passes below the pole at β =0. We use the Sommerfeld transformation β =
k cos α and γ =k sin α and introduce the coordinates (r, θ) by setting x
1
=
r cos θ and x
2
=r sin θ to reduce (5.125) to an integral over the contour C in
the α plane, namely
u
3
=−
i A
2
1/2
πk
C
cos(α/2)
cos α
e
i kr cos(α−θ)
dα. (5.126)
Note that the contour now passes above the pole at α =π/2.
116 5 Radiation and Diffraction
Fig. 5.10. The complex β plane for the diffraction integral, showing the contour C
β
,
is sketched on the left. The complex α plane, showing the contours C and −C
s
(θ), is
sketched on the right. θ is the saddle point in the α plane.
The following two identities are needed for the work to follow:
2
3/2
cos(α/2)
cos α
≡
1
cos[(α −π/2)/2]
+
1
cos[(α +π/2)/2]
,
(5.127)
4 cos(α/2) cos[(θ −π/2)/2]
cos α +cos(θ −π/2)
≡
1
cos[(α +θ −π/2)/2]
+
1
cos[(α −θ +π/2)/2]
. (5.128)
The second identity is a generalization of the ﬁrst.
We now distort the contour C to the steepest descents contour C
s
described
by (5.76) and (5.77). We next reverse the direction of integration, writing
the symbol for the contour as −C
s
(θ). This contour is shown in Fig. 5.10.
The position of the stationary point is indicated as an argument of this con
tour to underscore the structure of the contour. The diffraction integral is now
written as
u
3
=
i A
4πk
−C
s
(θ)
e
i kr cos(α−θ)
×
1
cos[(α −π/2)/2]
+
1
cos[(α +π/2)/2]
dα, (5.129)
where (5.127) has been used.
Setting aside the i A/(4πk), we consider the ﬁrst of the two integrals, which
we label I . We note that the calculation of the second is not different from that
Appendix: The Fresnel Integral 117
of the ﬁrst. We next change variables, giving
I =
1
2
_
−C
s
(0)
e
i kr cos α
×
_
1
cos[(α +θ −π/2)/2]
+
1
cos[(α −θ +π/2)/2]
_
dα. (5.130)
Using (5.128), we can collapse this integral into the more symmetric form
I =2
_
−C
s
(0)
cos(α/2) cos[(θ −π/2)/2]
cos α +cos(θ −π/2)
e
i kr cos α
dα. (5.131)
We now introduce yet another change of variables with the relation τ =
2
1/2
e
i π/4
sin(α/2) (recall Problem 5.3). Thus the integral I becomes
I =−2e
i π/4
e
i kr
η
_
∞
−∞
e
−krτ
2
τ
2
−i η
2
dτ, (5.132)
where η =2
1/2
cos[(θ −π/2)/2].
A third identity, namely
_
∞
kr
e
i η
2
ζ
_
∞
−∞
e
−ζ τ
2
dτdζ ≡π
1/2
_
∞
kr
ζ
−1/2
e
i η
2
ζ
dζ, (5.133)
is needed. Using this (interchange the order of integration on the lefthand side),
we ﬁnd that
η
_
∞
−∞
e
−krτ
2
τ
2
−i η
2
dτ =
2π
1/2
η
η
e
−i krη
2
_
∞
η(kr)
1/2
e
i µ
2
dµ
=
2π
1/2
η
η
e
−i krη
2
F
_
η(kr)
1/2
_
, (5.134)
where the Fresnel function F(z), repeating the deﬁnition (5.103), is
F(z) :=
_
∞
z
e
i ξ
2
dξ. (5.135)
We now return to (5.129). Note that this integral contains both a pole contri
bution and a part composed of Fresnel integrals. Let u
3np
represent the particle
displacement that does not include the pole contribution and u
3p
the pole con
tribution itself. Now u
3np
is given by
u
3np
=
e
−i π/4
A
π
(1/2)
k
_
e
−i kr cos(θ−π/2)
F
_
(2kr)
1/2
cos
_
θ −π/2
2
__
±e
−i kr cos(θ+π/2)
F
_
±(2kr)
1/2
cos
_
θ +π/2
2
___
, (5.136)
118 5 Radiation and Diffraction
where the plus sign is used for θ ∈ (0, π/2) and the minus sign for θ ∈ (π/2, π).
Here u
3p
is given by
u
3p
=−(A/k)e
−i kr cos(θ−π/2)
H(π/2 −θ). (5.137)
Recalling (5.105),
F(z) + F(−z) =π
1/2
e
i π/4
, (5.138)
we use it to combine (5.137) with the second term of (5.136) to give (5.104).
Problem 5.7 Asymptotic Approximations to the Fresnel Integral
Consider the Fresnel integral (5.135) for z = x, where x is real.
Problem 1. Find the ﬁrst term of an asymptotic expansion of the Fresnel
integral for both positive and negative x.
(a) Assume x > 0 and consider the integral
G(x) =e
−i x
2
∞
x
e
i ζ
2
dζ. (5.139)
Show that
G(x) ∼
i
2x
+O(x
−3
), x →∞. (5.140)
One way to do this is to deformthe contour to one such that Watson’s lemma
can be used.
(b) Assume x < 0. The approach of part (a) must be modiﬁed. Why? Write the
integral in (5.139) as one from x to −∞and one from −∞to ∞. Do not
use (5.138). Hence show that
G(x) ∼π
1/2
e
i π/4
e
−i x
2
+
i
2x
+O(x
−3
), x →−∞. (5.141)
That the two approximations differ for x that is positive or negative is an
example of the Stokes’ phenomena.
Problem 2. Find the ﬁrst term of an asymptotic expansion of the Fresnel
integral for x 1. Write the integral as the difference between one from zero
References 119
to ∞and one from zero to x. Expand the integrand of the second integral in a
Taylor expansion. Hence show that
F(x) ∼
π
1/2
e
i π/4
/2 −x +O(x
3
), x →0. (5.142)
References
Ablowitz, M.J. and Fokas, A.S. 1997. Complex Variables, pp. 411–513. New York:
Cambridge.
Achenbach, J.D., Gautesen, A.K., and McMaken, H. 1982. Ray Methods for Waves in
Elastic Solids. Boston: Pitman.
Babiˇ c, V.M. and Buldyrev, V.S. 1991. ShortWavelength Diffraction Theory. Berlin:
Springer.
Born, M. and Wolf, E. 1986. Principles of Optics, 6th (corrected) ed., pp. 565–578.
Oxford: Pergamon.
Cagniard, L. 1962. Reﬂection and Refraction of Progressive Seismic Waves. Translated
and revised by E.A. Flinn and C.H. Dix. New York: McGrawHill.
Carrier, G.F., Krook, M., and Pearson, C.E. 1983. Functions of a Complex Variable,
pp. 249–283. Ithaca, NY: Hod Books.
Copson, E.T. 1935. An Introduction to the Theory of Functions of a Complex Variable,
pp. 121–125. Oxford: Clarendon Press.
Copson, E.T. 1971. Asymptotic Expansions, pp. 13–14 and 48–62. Cambridge:
University Press.
Courant, R. and John, F. 1989. Introduction to Calculus and Analysis, Vol. II,
pp. 345–350. New York: Springer
deHoop, A.T. 1960. A modiﬁcation of Cagniard’s method for solving the seismic pulse
problem. Appl. Sc. Res., B 8: 349–356.
Doetsch, G. 1974. Introduction to the Theory and Application of the Laplace
Transform, pp. 218–230. New York: Springer.
Ewing, W.M., Jardetzky, W.S., and Press, F. 1957. Elastic Waves in Layered Media.
New York: McGrawHill.
Felsen, L.B. and Marcuvitz, N. 1994. Radiation and Scattering of Waves, pp. 370–441.
New York: IEEE and Oxford University Presses.
Gautesen, A.K. 1979. On matched asymptotic expansions for two dimensional
elastodynamic diffraction by cracks. Wave Motion 1: 127–140.
Harris, J.G. 1980a. Diffraction by a crack of a cylindrical longitudinal pulse. Z. Angew.
Math. Phys. 31: 367–383. Errata, Z. Angew. Math. Phys. 34.
Harris, J.G. 1980b. Uniform approximations to pulses diffracted by a crack. Z. Angew.
Math. Phys. 31: 771–775.
Harris, J.G. 1987. Edge diffraction of a compressional beam. J. Acoust. Soc. Am. 82:
635–646.
Harris, J.G. and Pott, J. 1985. Further studies of scattering of a Gaussian beam from a
ﬂuidsolid interface. J. Acoust. Soc. Am. 78: 1072–1080.
Hinch, E.J. 1991. Perturbation Methods, pp. 52–101. Cambridge: University Press.
Holmes, M.H. 1995. Introduction to Perturbation Methods, pp. 47–104. New York:
Springer.
Jones, D.S. 1952. A simplifying technique in the solution of a class of diffraction
problems. Quart. J. Math. 3: 189–196.
Jull, E.V. 1981. Aperture Antennas and Diffraction Theory. London: Peter Peregrinus.
120 5 Radiation and Diffraction
Knopoff, L. and Gilbert, F. 1959. First motion methods in theoretical seismology.
J. Acoust. Soc. Am. 31: 1161–1168.
Noble, B. 1988. Methods Based on the WienerHopf Technique, pp. 11–27 and 48–97.
New York: Chelsea.
Titchmarsh, E.C. 1939. The Theory of Functions, 2nd ed., pp. 85–86 and 99–101.
Oxford: Clarendon Press.
Titchmarsh, E.C. 1948. Introduction to the Theory of Fourier Integrals, 2nd ed.
Oxford: Clarendon Press.
van der Pol, B. and Bremmer, H. 1950. Operational Calculus, pp. 122–132 and
elsewhere. Cambridge: University Press.
Wong, R. 1989. Asymptotic Approximations of Integrals, pp. 20–31 and elsewhere.
Boston: Academic.
Zauderer, E. 1983. Partial Differential Equations of Applied Mathematics,
pp. 653–658 and 403–405. New York: WileyInterscience.
6
Guided Waves and Dispersion
Synopsis
Chapter 6 discusses guided waves and the dispersion they experience. Only the
antiplane shear problem is treated. The guided waves are constructed by using
partial waves and their dispersion calculated by using the transverse resonance
principle. Both harmonic and transient excitations of a closed waveguide are
studied by using an expansion of modes. The harmonic excitation of an open
waveguide by a line source is also studied, though in this case by using both
ray and mode representations. As a last example, we examine propagation in a
closed waveguide with a slowly varying thickness, using an asymptotic expan
sion that combines features of both rays and modes. We close by examining
how information and energy propagate at the group velocity.
6.1 Harmonic Waves in a Closed Waveguide
We consider a layer of inﬁnite extent in the x
1
direction and of ﬁnite thick
ness in the x
2
direction. Within the layer, the coordinate x
2
∈ (−h, h) and the
plane x
2
=0 is a plane of reﬂection symmetry. This structure is a waveguide or
guide because the waves are forced to propagate in the x
1
direction and the
guide is closed because waves are completely trapped within the structure.
We are interested in learning what kinds of antiplane waves propagate in the
guide without, at present, seeking to know how they are excited. Accordingly,
we seek possible solutions to the following antiplane problem. In the layer, u
3
must satisfy (2.20), which, rewritten here, is
∂
a
∂
a
u
3
+k
2
u
3
=0, (6.1)
and at the surfaces
µ∂
2
u
3
(x
1
, ±h) =0. (6.2)
121
122 6 Guided Waves and Dispersion
As in previous chapters the subscript T has been dropped. Problem6.1 indicates
how solutions to this problem can be found by reducing it to an eigenvalue
problemin the transverse coordinate. When this is done we ﬁnd that the possible
waveﬁelds are described by
u
3m
= A
m
cos
sin
(γ
m
x
2
)e
iβ
m
x
1
, m =0, 1, 2, . . . , (6.3)
where u
3m
is the mth (waveguide) mode for this guide. Cosine is used for m
even and sine for m odd. This wave stands in the x
2
direction, but it propagates
in the x
1
direction. Here β
m
is the lateral wavenumber for the mth mode. The
transverse wavenumbers for this mode are ±γ
m
. β
m
=ω/c
m
, where c
m
is the
phase velocity of the mode. β
m
is given by
β
m
=[k
2
−(mπ)
2
/(2h)
2
]
1/2
, (6.4)
where (β
m
) ≥0 or (β
m
) ≥0 for x
1
> 0. In this particular case β
m
is either
real or imaginary, but not complex. For β
m
that is real, the wavenumber and
phase velocity in the propagation direction depend on ω (through k =ω/c).
Thus (6.4) is a dispersion relation and is similar to that found previously for
propagation in periodic structures (there we used κ to represent the effective
wavenumber). The square root is deﬁned so that, for imaginary values, the mode
decays in its direction of propagation (x
1
> 0).
Introducing the term
1
e
iβx
1
will reduce a linear partial differential equation,
or system of such equations, in (x
1
, x
2
) to an eigenvalue problem in x
2
for β,
though that problem may not be easy to solve, or its solution may not be partic
ularly informative. However, by examining the solution, (6.3), we can learn to
reconstruct it in a way that uses only the kinematical features of reﬂection and
transmission of a plane wave at a boundary. Thereafter this method of solution
can be used to solve other guided wave problems without directly considering
the underlying eigenvalue problem.
Problem 6.1 An Eigenvalue Problem: A Closed Waveguide
Equation (6.1), with boundary condition (6.2), can be reduced to an ordinary
differential equation in x
2
by seeking solutions of the form
u
3
= f (x
2
)e
iβx
1
. (6.5)
1
This is equivalent to taking the spatial Fourier transform, transformvariable β, in the x
1
direction.
6.1 Harmonic Waves in a Closed Waveguide 123
Show that the differential equation in x
2
leads to an eigenvalue problem for β
2
.
Usually the β is linked with γ , where
γ =(k
2
−β
2
)
1/2
, (6.6)
and γ
2
or −γ
2
is considered the eigenvalue. In this book β
2
is considered the
eigenvalue. Show that solutions symmetric with respect to the reﬂection plane
are
f
2n
(x
2
) = A
2n
cos γ
2n
x
2
, γ
2n
=nπ/h, (6.7)
and that those antisymmetric to this plane are
f
2n+1
(x
2
) = A
2n+1
sin γ
2n+1
x
2
, γ
2n+1
=(2n +1)π/(2h), (6.8)
where n =0, 1, 2, . . . . The A
m
are constants.
6.1.1 Partial Waves and the Transverse Resonance Principle
The central feature of a waveguide is that the waves phase match in the propa
gation direction and stand in the transverse direction. Equivalently stated, as the
waves reﬂect back and forth within the guide, they must interfere constructively
to reconstruct themselves and form a sustained waveﬁeld. There are two sets of
plane waves in a guide, as indicated in Fig. 6.1 – one propagating downward
and one upward. The members of each set are referred to as partial waves, just
as were the individual plane waves in each cell of the periodic structure exam
ined in Section 1.4. Those that propagate downward are indicated by the solid
lines and those that propagate upward by the dashed ones. The downward and
Fig. 6.1. The dashed lines indicate the upward propagating set of plane waves, while
the solid lines indicate the downward propagating set. These rays are shown in the
slowness diagram to the right. Note that phase matching must occur in the x
1
direction.
124 6 Guided Waves and Dispersion
upward propagating sets are
u
3
= Ae
i (βx
1
−γ x
2
)
, (6.9)
u
3
= Be
i (βx
1
+γ x
2
)
, (6.10)
respectively. The term γ is given by (6.6). We have, at this stage, assumed
only that (γ ) ≥0. From our previous work, given in Section 3.2, we can,
by a slight modiﬁcation of that calculation, show that the reﬂection coefﬁcient
at x
2
=±h is one. Each upward propagating wave must be reﬂected into a
downward propagating one and vice versa. Therefore, at the upper and lower
boundaries,
Ae
−i γ h
Be
i γ h
= 1, (6.11)
Be
−i γ h
Ae
i γ h
= 1, (6.12)
respectively. The terms A and B are constants. For these two equations to hold
simultaneously, the following must be true:
γ h =mπ/2, (6.13)
and
A = B for m =2n, n =0, 1, 2, . . . (6.14)
or
A =−B for m =2n +1, n =0, 1, 2, . . . . (6.15)
Therefore
u
3m
= A
m
cos
sin
(γ
m
x
2
)e
iβ
m
x
1
, m =0, 1, 2, . . . , (6.16)
where β
m
is given by (6.4), γ
m
=mπ/2h, and A
m
is 2A or 2i A. This method of
ﬁnding the dispersion relation, (6.4), wherein partial waves are made to stand
or resonate in the transverse direction, is referred to as the transverse resonance
principle.
6.1.2 Dispersion Relation: A Closed Waveguide
The most intriguing feature of guided waves is that the lateral wavenumber β
m
is a function of ω, or, alternatively ω is a function of β
m
. The dispersion relation
6.1 Harmonic Waves in a Closed Waveguide 125
Fig. 6.2. A sketch of the dispersion relation for antiplane shear modes in a closed
waveguide. The normalized angular frequency is plotted as a function of the normalized
lateral wavenumber.
can be written as
2hβ
m
=[(2hω/c)
2
−(mπ)
2
]
1/2
. (6.17)
Asketch of this relation is shown in Fig. 6.2. While the β
m
may be thought of as
a function of ω, when the frequency is a free variable, it is usual to consider ω as
a function of β
m
. For the mth mode, there is no propagation for (2hω/c) ≤mπ
and the frequency ω
m
=mπc/2h is called the cutoff frequency. The phase
velocity c
m
for the mth mode is
c
m
=ω/β
m
. (6.18)
The velocity of real interest, however, is the group velocity C
m
. Unlike c
m
, C
m
for the mth mode is given by the slope of the mth branch of the dispersion
relation, namely
C
m
=dω/dβ
m
. (6.19)
The group velocity is both the velocity of energy propagation and the velocity
with which information propagates. Here we demonstrate the ﬁrst assertion
and leave to Section 6.6 the demonstration of the second. The time average of a
waveﬁeld quantity G
m
(β
m
x
1
−ωt, x
2
) for the mth waveguide mode is deﬁned as
G
m
=
1
T
t +T
t
h
−h
G
m
(β
m
x
1
−ωτ, x
2
)dx
2
dτ. (6.20)
126 6 Guided Waves and Dispersion
The average G
m
does not depend upon x
1
or t because at a given x
1
, the
argument β
m
x
1
−ωτ is simply a translation of the argument ωτ, a fact noted
previously when deriving (2.18).
The instantaneous energy density E
m
(x
1
, x
2
, t ) in the mth mode is
E
m
=
1
2
ρ ∂
t
u
3m
∂
t
u
3m
+
1
2
µ∂
a
u
3m
∂
a
u
3m
, (6.21)
and, using (6.20) with (2.18), we ﬁnd that its average is
E
m
=
1
2
ρc
2
m
hβ
2
m
A
m
A
∗
m
. (6.22)
For propagation β
m
is real. The instantaneous ﬂux of energy density
F
m
(x
1
, x
2
, t) in the mth mode is
F
m
= −µ∂
1
u
3m
∂
t
u
3m
, (6.23)
and the average ﬂux is
F
m
=
1
2
µc
m
hβ
2
m
A
m
A
∗
m
. (6.24)
The velocity of energy propagation along the axis of the waveguide for the mth
mode is, therefore, the ratio F
m
/E
m
. Direct calculation shows that
F
m
/E
m
= c
2
/c
m
=C
m
. (6.25)
The average (6.20) can also be used to demonstrate the equipartition of
energy. Usingthis principle whencalculatingthe average energydensity, instead
of proceeding directly as we just did, usually reduces the amount of calculation
needed, though care is required because not all equations describing waveﬁelds
exhibit equipartition. The Lagrangian density L for antiplane shear motion is
given by
L =
1
2
ρ(∂
t
u
3
)
2
−
1
2
µ[(∂
1
u
3
)
2
+(∂
2
u
3
)
2
]. (6.26)
Using (6.20) we readily ﬁnd that L = 0, for the mth mode. It follows then
that K = U, where the kinetic and internal energy densities, K and U, were
ﬁrst deﬁned by (1.25). Clearly, the converse is also true.
Dispersion need not arise from a geometrical constraint, as it does with
a waveguide, but can arise from the structure of the equation itself, as the
following problem demonstrates.
6.1 Harmonic Waves in a Closed Waveguide 127
Problem 6.2 Dispersion 1
Problem 1. Consider the following two differential equations, the Klein–
Gordon equation,
∂
2
t
ϕ −a
2
∂
2
x
ϕ +b
2
ϕ =0, (6.27)
and the equation for ﬂexural motion in a rod,
∂
2
t
ϕ +a
2
∂
4
x
ϕ =0. (6.28)
The terms a and b are constants. In both cases ﬁnd a dispersion relation in the
form ω =ω(k) by seeking a wave solution of the form
ϕ = Ae
i (kx−ωt )
, (6.29)
where A is a constant. Assume that the wave propagates to the right. Calculate
the phase and group velocities c and C, where c =ω/k and C =dω/dk.
Demonstrate that the group velocity is the velocity of timeaveraged energy
transport. To do so you will need to knowboth the instantaneous energy density
and instantaneous ﬂux, so that you can calculate their average values. One way
to ﬁnd these is to construct conservation laws directly from (6.27) and (6.28).
Such laws have the form
∂
t
+ ∂
x
= − . (6.30)
The ﬁrst box is the instantaneous energy density, the second the instantaneous
ﬂux, and the third, on the righthand side, a dissipative term. For the equa
tions being studied here, the righthand side is zero. To ﬁnd a conservation
law, multiply each of (6.27) and (6.28) by ∂
t
ϕ and conﬁgure the result to co
incide with (6.30). Thus, show that the instantaneous energy density for the
Klein–Gordon equation is
E =(∂
t
ϕ)
2
/2 +a
2
(∂
x
ϕ)
2
/2 +b
2
ϕ
2
/2 (6.31)
and the instantaneous ﬂux is
F =−a
2
(∂
t
ϕ)(∂
x
ϕ). (6.32)
Also, show that the instantaneous energy density for the equation for ﬂexural
motion is
E =(∂
t
ϕ)
2
/2 +a
2
∂
2
x
ϕ
2
2 (6.33)
128 6 Guided Waves and Dispersion
and the instantaneous ﬂux is
F =a
2
(∂
t
ϕ)
∂
3
x
ϕ
−a
2
(∂
t
∂
x
ϕ)
∂
2
x
ϕ
. (6.34)
Using (2.18), complete the demonstration.
Problem 2. The equation describing acoustic waves in a wind having speed
U (<c, where c is the speed of sound) in the x
1
direction is given by
(∂
t
+U∂
1
)
2
ϕ =c
2
∇
2
ϕ, (6.35)
where (x, t ) is a ﬁxed coordinate system. Using a solution of the form
ϕ = Ae
i (k·x−ωt )
, (6.36)
ﬁnd the dispersion relation. The vector k is needed because the propagation
environment is anisotropic.
6.2 Harmonic Waves in an Open Waveguide
We continue to consider an antiplane problem having the governing equation,
(6.1). We now consider a layer on a halfspace. Figure 6.3 indicates the geome
try. Note that the positive x
2
direction points into the interior. The interior of the
Fig. 6.3. A slowonfast structure can support trapped waves in the layer. Drawing
slowness diagrams for the layer and halfspace indicates the condition for waves to be
trapped in the layer, namely that the vertical dashed line in the diagram to the right not
intersect the lower slowness circle.
6.2 Harmonic Waves in an Open Waveguide 129
layer occupies x
2
∈ (−h, 0) and its wavespeed is c (k =ω/c). The equation of
motion in the layer is (6.1). The interior of the halfspace occupies x
2
∈ (0, ∞).
The equation of motion in the halfspace is also given by (6.1) with a different
wavenumber. The equation, rewriting it once again, is
∂
α
∂
α
u
3
+
¯
k
2
u
3
=0, (6.37)
where
¯
k =ω/¯ c and ¯ c is the wavespeed in the halfspace. At x
2
=−h,
µ∂
2
u
3
(x
1
, −h) =0, (6.38)
and at x
2
=0,
u
3
(x
1
0
−
) =u
3
(x
1
0
+
), µ∂
2
u
3
(x
1
, 0
−
) = ¯ µ∂
2
u
3
(x
1
, 0
+
), (6.39)
where µand ¯ µare the elastic constants for the layer and halfspace, respectively.
The layer is called an open waveguide because there is now the possibility
that waves may not remain trapped, but may radiate into the halfspace. How
ever, waves will remain trapped provided the wavespeed in the layer is less
than that in the halfspace. This conﬁguration is sometimes referred to as a
slowonfast guiding structure. To understand why waves are trapped, consider
the slowness diagrams sketched in Fig. 6.3. The wavespeed in the layer is c
and that in the underlying halfspace ¯ c, so that the corresponding slownesses
are s and ¯ s, respectively. A slowonfast structure is one for which s > ¯ s. If the
waves are to phase match at x
2
=0, s
1
must be the same in both the layer and
halfspace. If the waves are to remain trapped in the layer, unable to radiate
away, ¯ s
2
must be imaginary, with the sign of ¯ s
2
selected so that decay takes
place with depth. Therefore, the waves in the layer able to phase match to waves
decaying into the halfspace are those for which s
1
= ¯ s
1
> ¯ s. When s
1
≤ ¯ s <s,
waves launched in the layer phase match to waves in the halfspace for which
¯ s
2
is real and are therefore not trapped. The slowonfast structure is analyzed
next by using partial waves and the transverse resonance principle. In Problem
6.3 the problemis reduced to solving an eigenvalue problemin the x
2
direction.
A fastonslow structure (s < ¯ s) does not guide waves so that, except in
Problem 6.4, we do not consider it.
6.2.1 Partial Wave Analysis
In the layer, the upward and downward propagating sets of waves are
u
3
= Be
i (βx
1
−γ x
2
)
, (6.40)
u
3
= Ae
i (βx
1
+γ x
2
)
. (6.41)
130 6 Guided Waves and Dispersion
In the halfspace the set of waves is
u
3
=
¯
Ae
i (βx
1
+¯ γ x
2
)
. (6.42)
The waves in the halfspace decay or propagate downward. This statement is
consistent with the principle of limiting absorption, given in Section 4.4. To
satisfy the equations of motion,
γ =(k
2
−β
2
)
1/2
, ¯ γ =(
¯
k
2
−β
2
)
1/2
, (6.43)
where (γ ) ≥0 and ( ¯ γ ) ≥0. We discuss the branches of these radicals more
carefully in Section 6.4. The reﬂection and transmission coefﬁcients at x
2
=0
are, respectively,
R(γ ) = (µγ − ¯ µ ¯ γ )/(µγ + ¯ µ ¯ γ ), (6.44)
T(γ ) = (2µγ )/(µγ + ¯ µ ¯ γ ). (6.45)
The R(γ ) and T(γ ) are gotten from (3.21) and (3.22) by noting that β =
k sin θ
0
=
¯
k sin θ
2
, γ =k cos θ
0
, and ¯ γ =
¯
k cos θ
2
. Exactly as in the case of the
closed guide, at x
2
=−h,
Ae
−i γ h
/Be
i γ h
=1, (6.46)
and at x
2
=0,
B/A =(µγ − ¯ µ ¯ γ )/(µγ + ¯ µ ¯ γ ). (6.47)
Moreover, in the case of the open guide, we also have
¯
A/A =(2µγ )/(µγ + ¯ µ ¯ γ ). (6.48)
These last two equations indicate that plane waves incident on the lower bound
ary must, in addition to being reﬂected, be transmitted into those that decay or
propagate away from the boundary. From (6.46) and (6.47) we ﬁnd that for
nontrivial solutions,
e
−2i γ h
=(µγ − ¯ µ ¯ γ )/(µγ + ¯ µ ¯ γ ). (6.49)
The case of interest is ¯ s < s
1
≤s, where β =ωs
1
, so that ¯ γ =i ¯ α, where ¯ α ≥0.
In this case (6.49) reduces to
tan γ h = ¯ µ¯ α/µγ. (6.50)
Recalling (6.43), we see that (6.49) or (6.50) is the dispersion relation for the
structure. It gives ω =ω(β) or β =β(ω), albeit indirectly. The antiplane waves
6.2 Harmonic Waves in an Open Waveguide 131
guided by the layer are called Love waves. Knowing the dispersion relation,
we can ﬁnd B and
¯
A in terms of A, and thus construct the waveﬁelds in the
layer and halfspace. Note that the absence of a plane of reﬂection symmetry
means that the modes do not divide into symmetric and antisymmetric ones.
Problem 6.3 A Second Eigenvalue Problem: An Open Waveguide
Note that (6.29) and (6.30), with boundary conditions (6.31) and (6.32), can
be reduced to a singular eigenvalue problem (Friedman, 1956) in x
2
by looking
for solutions of the form
u
3
= f (x
2
)e
iβx
1
, x
2
∈ (−h, 0), (6.51)
u
3
= g(x
2
)e
iβx
1
, x
2
∈ (0, ∞). (6.52)
Note that the form of the function in x
1
has been chosen to give a wave prop
agating in the positive x
1
direction and ensure phase matching at x
2
=0. Use
the condition that g(x
2
) →0 as x
2
→∞. Showthat solutions to the differential
equation in x
2
and its boundary conditions can be expressed as
f (x
2
) =C cos[γ (x
2
+h)], g(x
2
) = De
−¯ αx
2
, (6.53)
where ¯ γ =i ¯ α, ¯ α ≥0. The transverse wavenumbers γ and ¯ γ are given by (6.36).
Find the constant D in terms of C. Can you recover the dispersion relationship,
(6.50)?
The problemjust discussed is called a singular eigenvalue problembecause it
has a continuous spectrumas well as a discrete one. The precedinganalysis gives
the discrete eigenvalues and eigenfunctions, but gives one little information
about the continuous spectrum, unless the reader is quite perceptive. Friedman
(1956), mentioned previously, will help the reader understand the underlying
mathematics of singular eigenvalue problems; Tolstoy and Clay (1987) will
give the reader a reasoned physical explanation for the continuous spectrum.
What condition stated previously in this problem must be modiﬁed to ﬁnd the
continuous eigenvalues and eigenfunctions?
6.2.2 Dispersion Relation: An Open Waveguide
The transcendental equation, (6.50), in combination with (6.43), gives either
ω =ω(β) or β =β(ω). To examine this dispersion relation, we begin with
a diagram such as that shown in Fig. 6.4, where three distinct regions are
identiﬁed. First we identify the boundaries between which β is real (assuming
that c < ¯ c). These boundaries are denoted by c and ¯ c, though the slopes are 1
and ¯ c/c.
132 6 Guided Waves and Dispersion
Fig. 6.4. A sketch of the dispersion relation for Love waves. The normalized angular
frequency is plotted as a function of the normalized lateral wavenumber. There are three
regions whose boundaries are formed by the two wavespeeds c < ¯ c.
In region 1, γ =±i α and ¯ γ =i ¯ α, where α and ¯ α are real and positive. It
is readily seen that no solution for real β is possible and therefore no trapped
wave propagates in the x
1
direction. In region 3 both γ and ¯ γ are real. In this
case (6.43) becomes
tan(γ h) =−i ¯ µ ¯ γ /µγ. (6.54)
Again it is clear that there is no solution for β that is real, though there are
solutions for complexβ. We shall brieﬂydiscuss this possibilityinSection6.4.3.
The absence of roots for β real tells us that there are no trapped waves, exactly
what we would expect for γ and ¯ γ both real.
Solutions for real β are possible in region 2, as can be seen from examining
(6.50). Consider the following limits. First, let βh →∞within region 2. Write
(6.50) as
tan
βh[(k
2
/β
2
) −1]
1/2
=
¯ µ[1 −(
¯
k
2
/β
2
)]
1/2
µ[(k
2
/β
2
) −1]
1/2
. (6.55)
The argument of the tangent in (6.55) approaches a ﬁnite limit or zero because
the right side remains positive, and approaches zero or +∞. There is only one
possibility, namely k/β →1. In other words, by allowing the layer to become
an inﬁnite number of wavelengths thick, its response is no longer inﬂuenced
by the presence of the halfspace. Second, let βh move toward zero through
region 2 and reason as before. For the lowest mode
¯
k/β →1 as βh →0 and
6.2 Harmonic Waves in an Open Waveguide 133
ωh →0. However, for the higher modes
¯
k/β →1, while βh and ωh remain
ﬁnite. In this case, for some ω =ω
n
, ¯ α =0 and
[(ω
n
h/βhc)
2
−1]
1/2
βh =nπ, (6.56)
where n is a positive integer. The frequency ω
n
is that at which the waves
move frombeing trapped to radiating into the halfspace. This frequency is also
called a cutoff frequency, though the term transition frequency might be more
appropriate. Unlike the case of a closed waveguide, the group velocity does not
vanish at this point. Keep in mind that we have, so far, restricted β to positive
real values and have not deﬁned the branches of γ and ¯ γ carefully.
The outcomes of the problem just described should be compared with those
of the reﬂection problem that follows. The viewpoint there is rather different.
In the reﬂection problem the layer is viewed from the halfspace, and we are no
longer as concerned as we were here with whether the structure is a slowonfast
or fastonslow one.
Problem 6.4 Reﬂection From a Layer
Referring to the geometry of Fig. 6.3, let the plane wave
¯ u
30
= A
0
e
i (βx
1
−¯ γ x
2
)
(6.57)
be incident to the layer from the halfspace. Make no assumption at present as
to the relative magnitudes of c and ¯ c. If A
1
is the unknown amplitude of the
reﬂected wave, calculate the reﬂection coefﬁcient of the layer, namely R(θ
0
) =
A
1
/A
0
. Show that R(θ
0
) = A
−
(θ
0
)/A
+
(θ
0
), where
A
∓
= cos θ
0
cos(kh cos θ) ∓i
µ¯ c
¯ µc
cos θ sin(kh cos θ), (6.58)
and
sin θ/c =sin θ
0
/¯ c. (6.59)
The wavenumbers β =
¯
k sin θ
0
and ¯ γ =
¯
k cos θ
0
.
The reader should contrast the two possible cases, fastonslow and slowon
fast. In particular, how might a trapped wave be excited by using a disturbance
incident from the halfspace when the structure is slow on fast? Note that this
reﬂection coefﬁcient is frequency dependent.
134 6 Guided Waves and Dispersion
6.3 Excitation of a Closed Waveguide
6.3.1 Harmonic Excitation
We start by considering a closed waveguide that is harmonically excited by
applying an antiplane traction at one end, while the other extends to inﬁnity.
One common way to solve such problems is to expand the waveﬁeld in the guide
in terms of its modes. To do so requires that a mathematical framework be in
place. As a minimum we need a means of calculating the coefﬁcients of the
expansion – an orthogonality relation – and some assurance that the expansion
is complete. While these questions have still not been satisfactorily answered
for inplane elastic waves in a guide with an end (Miklowitz, 1978; Folguera
and Harris, 1999), the antiplane modes are simply the terms of a Fourier series,
so that in this case these questions are settled.
Problem 6.5 A Waveguide Mode Expansion
In part, (6.1) and (6.2) specify the problem. The geometry of the problem is
now described by a layer similar to that shown in Fig. 6.1, but starting at x
1
=0
and extending through positive values of x
1
to inﬁnity. At x
1
=0 the waveguide
is excited with a traction T(x
2
) symmetric in x
2
so that
µ∂
1
u
3
=−T(x
2
), T(x
2
) = T(−x
2
). (6.60)
What condition should be imposed upon the forced disturbance as x
1
→∞?
Noting the symmetry of the excitation, how might the forced waveﬁeld be
expanded? Show that a representation of the solution is
u
3
(x
1
, x
2
) =
N−1
n=0
(−A
n
)
iβ
n
cos
nπ
h
x
2
e
iβ
n
x
1
+
∞
n=N
A
n
α
n
cos
nπ
n
x
2
e
−α
n
x
1
,
(6.61)
where N is the smallest n for which the inequality
nπ/h <ω/c (6.62)
is reversed. The frequency of excitation is ω, β
n
is given by (6.4), and, for
n ≥ N, β
n
=i α
n
. Using the orthogonality of the modes (eigenfunctions), write
down explicit expressions for the coefﬁcients A
n
.
Note that the modes in the ﬁrst sum of (6.61) propagate, while those in the
second are evanescent. Far from the source, only those that propagate make a
signiﬁcant contribution.
6.3 Excitation of a Closed Waveguide 135
6.3.2 Transient Excitation
We next consider a transient excitation. In Problem 6.5, we replace (6.60) with
µ∂
1
u
3
=−T(x
2
)δ(t ). (6.63)
Moreover, we assume that the waveguide is quiescent until the excitation is
applied. That is,
u
3
=∂
t
u
3
=0, t < 0. (6.64)
Using (1.36), we synthesize the transient response as
u
3
(x
1
, x
2
, t ) =
∞
n=0
(−A
n
)
π
cos
nπ
h
x
2
∞
0
e
i t (β
n
x
1
/t −ω)
iβ
n
dω. (6.65)
Using the principle of limiting absorption, replace ω with ω+i , where > 0.
The n =0 term is easily inverted, giving
∞
0
e
i (ω+i )(x
1
/c−t )
i (ω+i )/c
dω =−πc H(t −x
1
/c). (6.66)
The n > 0 terms are less straightforward. The integral to be evaluated is
∞
0
e
i x
1
(β
n
−ωt /x
1
)
iβ
n
dω. (6.67)
The left half of Fig. 6.5 shows the complex ω plane and the branch cuts for
n > 0. Following the arguments in Section 3.4.4, we have cut the plane so that
(β
n
) ≥0 for all ω. Using the principle of limiting absorption for ω > 0, we
ﬁnd the branch point is at c nπ/h −i . By symmetry, for ω < 0, the branch
point is at −cnπ/h +i . The integration contour in the ωplane must pass above
the branch point cnπ/h −i . Note that c nπ/h is a cutoff frequency so that
propagation has ceased when ω < cnπ/h. Note, as well, that (β
n
) > 0 in the
ﬁrst and third quadrants. Once the contour has been determined, the in (6.67)
can be set to zero.
Integrals of the form of (6.67), even when they can be evaluated in closed
form, give the clearest description of the waveﬁeld at points far fromthe source.
Accordingly, we approximate (6.67) by using the method of stationary phase
described in Section 5.3.3. We ﬁx t /x
1
and let x
1
be the large parameter. For a
136 6 Guided Waves and Dispersion
Fig. 6.5. The lefthand ﬁgure shows the complex ω plane. The contour extends from
ω =0 along the real axis. (β
n
) ≥0 ∀ω, (β
n
) >0 in the upper right quadrant of the
ω plane. The righthand ﬁgure shows the dispersion curve for the nth mode with the
stationary point indicated by (β
ns
, ω
ns
).
given point in (x
1
, t ), the stationary point is given by
dβ
n
/dω =t /x
1
. (6.68)
Solving this equation gives the stationary point (β
ns
, ω
ns
) as a function of (x
1
, t )
(more precisely x
1
/t ), namely
ω
ns
c
=
nπ/h
[1 −(x
1
/ct )
2
]
1/2
, x
1
/ct <1, (6.69)
and
β
ns
=
nπ
h
x
1
/ct
[1 −(x
1
/ct )
2
]
1/2
, x
1
/ct <1. (6.70)
No signal can travel faster than c (Problem 6.6). Therefore, x
1
/ct <1 for prop
agating waves and (β
ns
, ω
ns
) are real.
The integral, (6.67), is then approximated as
∞
0
e
i x
1
(β
n
−ωt /x
1
)
iβ
n
dω ∼
c (2π)
1/2
(nπ/h)(β
ns
x
1
)
1/2
e
i x
1
(β
ns
−ω
ns
t /x
1
)
e
−i π/4
,
(6.71)
as x
1
→∞, while t /x
1
is held ﬁxed. After some simpliﬁcation, the particle
6.3 Excitation of a Closed Waveguide 137
displacement of the nth mode is approximated as
u
3n
(x
1
, x
2
, t ) ∼
(−A
n
)
π
c cos
_
nπ
h
x
2
_
_
(2h/n)
x
1
[(ct /x
1
)
2
−1]
1/2
_
1/2
×cos
_
(nπ/h)x
1
[(ct /x
1
)
2
−1)]
1/2
+π/4
_
, (6.72)
and the total particle displacement can be written as
u
3
(x
1
, x
2
, t ) ∼ A
0
c H(t −x
1
/c) +
∞
n=1
u
3n
(x
1
, x
2
, t ). (6.73)
Examining (6.72) shows that the nth mode is O(n
−1/2
). The higherorder modes
therefore make a weaker contribution than do the lowerorder ones.
Note that the stationary phase approximation breaks down for x
1
/t near c.
Dai and Wong (1994) give a correction for this case. Further, note that as x
1
/t
sweeps through its values from zero to c, ω
s
takes on all its values from cnπ/h
to inﬁnity. In fact, for ﬁxed x
1
, (6.68) provides a onetoone mapping from the
t plane to the ω plane.
At (x
1
, t ) the nth mode is approximated by an expression of the form
_
A
n
(x
1
, x
2
, t )e
i (β
ns
x
1
−ω
ns
t )
_
. (6.74)
This is a group in the sense that it arises from a cluster or group of wavenum
bers and frequencies in the neighborhood of (β
ns
, ω
ns
). Note that (6.68) can
be viewed as x
1
/t =C
n
(ω
ns
), where C
n
is the group velocity dω/dβ
n
. We have
already noted in (6.25) that C
n
is the velocity of timeaveraged energy trans
mission. Now we also note that it is the velocity with which the frequency ω
ns
propagates to the point (x
1
, t ). This is not the velocity with which the phase
θ(x
1
, t ) =(β
ns
x
1
−ω
ns
t ) propagates outward. Rather, a constant phase θ prop
agates at a rate c
ns
=ω
ns
/β
ns
.
Problem 6.6 No Signal Travels Faster than the Velocity c
Using the problem just discussed, show that no signal travels faster than c.
Write each integral over ω as
π
_
∞
0
e
−i ω(t −x
1
/c)
e
i (β
n
x
1
−ωx
1
/c)
iβ
n
dω (6.75)
and note that, in the upper right quadrant of Fig. 6.5, with the ω plane cut as
138 6 Guided Waves and Dispersion
indicated,
lim
ω→∞
e
i (β
n
x
1
−ωx
1
/c)
iβ
n
=0. (6.76)
Therefore, for t < x
1
/c, showthat the integration contour may be distorted from
one along the real ω axis to the one along the imaginary axis. Lastly, show that
the integral can be written as
π
∞
0
i e
p(t −x
1
/c)
e
−(α
n
x
1
−px
1
/c)
(−α
n
)
dp. (6.77)
This integral must be zero (why?), indicating that no signal is present for t <
x
1
/c.
While this result is the outcome of the analysis of a speciﬁc problem, it is
true generally. Moreover, it indicates that the group velocity does not exceed
the wavespeed of the medium,
2
a result that is also generally true.
For separable problems, such as the one discussed in Problem 6.5, an expan
sion of the solution in the eigenfunctions of the transverse eigenvalue problem
results, after invoking the orthogonality of the eigenfunctions, in reducing the
partial differential equation to an ordinary one in x
1
, involving each mode
separately. The domain of the transverse eigenvalue problem is ﬁnite and there
fore the eigenvalues are discrete. Expanding the source in these eigenfunctions
causes the equation in x
1
to be forced by a coefﬁcient of this expansion. This is
more or less what we did in solving Problem 6.5, though the modal expansion
already contained the solution to the differential equation in x
1
, so that that step
was leaped over. In Problem 6.7 the reader is asked to solve much the same
problem by using a continuous eigenfunction expansion in x
1
∈ (0, ∞), thus
reducing the partial differential equation to an ordinary one in x
2
. Moreover,
rather than use a Fourier synthesis to construct the transient solution, the reader
is asked to use a Laplace transform.
2
The dispersion is said to be anomalous when the group velocity exceeds the velocity of propa
gation of a harmonic plane wave in the medium (Sommerfeld, 1964a; Brillouin, 1960). In this
case the group velocity is no longer a measure of the speed at which information or energy
propagates. In such cases a speciﬁc signaling problem must be worked through, and veloc
ities that characterize the speed at which information is propagated and at which energy is
propagated must be deﬁned as best one can. Moreover, the anomalous dispersion may only be
apparent and not real. One must take care to distinguish between anomalous dispersion caused
by some approximate derivation of the equation being studied and that caused by the underly
ing physical situation. The presence of frequencydependent attenuation confuses the issue even
further.
6.4 Harmonically Excited Waves in an Open Waveguide 139
Problem 6.7 A Continuous Eigenfunction Expansion
Continue to consider the problem just solved, with the excitation (6.63). Use
the cosine transform
∗
u
3
(ξ, x
2
, t ) =
∞
0
u
3
(x
1
, x
2
, t ) cos(ξ x
1
)dx
1
, (6.78)
followed by the Laplace transform
∗
¯ u
3
(ξ, x
2
, p) =
∞
0
∗
u
3
(ξ, x
2
, t )e
pt
dt, (6.79)
to ﬁnd an ordinary differential equation in x
2
. Find
∗
¯ u
3
and begin to invert the
transforms. It is perhaps easier to set
∗
¯ v
3
(ξ, x
2
, p) = p
∗
¯ u
3
(ξ, x
2
, p) where v
3
is the particle velocity. Find
∗
v
3
(ξ, x
2
, t ) =
1
2πi
+i ∞
−i ∞
∗
¯ v
3
(ξ, x
2
, p)e
pt
dp (6.80)
and then approximate
v
3
(x
1
, x
2
, t ) =
2
π
∞
0
∗
v
3
(ξ, x
2
, t ) cos(ξ, x
1
)dξ (6.81)
for large x
1
, with x
1
/t held ﬁxed, using the method of stationary phase. This
integral can also be inverted exactly.
The reader may be surprised to ﬁnd that his or her expression for v
3
contains
both forward and backward propagating waves. This is because we have chosen
toworkwitha cosine transform. Whywouldone not workwitha sine transform?
At what point has the reader imposed the condition that waves be outgoing from
the source? Can the reader express (6.81) as a sum of waves propagating solely
in the positive x
1
direction?
6.4 Harmonically Excited Waves in an Open Waveguide
We next return to the layered antiplane structure, sketched on the left in Fig. 6.3,
to calculate the waveﬁeld radiated by a harmonic line source placed at (0, x
20
),
where x
20
< 0. Except for the absence of a source, the equations of motion
and the boundary conditions are given by (6.38) and (6.39). This problem
may be solved by constructing the solution by using eigenfunction expansions.
However, as we have done previously, we shall construct the radiated waveﬁeld
by superposing collections of partial waves. This method of construction will
show how ray representations and modal expansions are related to one another.
Our discussion follows a very similar one in Brekhovskikh (1980).
140 6 Guided Waves and Dispersion
6.4.1 The Waveﬁeld in the Layer
We begin by representing the waveﬁeld radiated by a line source as a spectrumof
plane waves, using the result of Problem 4.1. Equation (4.19), with α replacing
ξ, is here rewritten by expanding cos(θ −α) to give
u
i
3
=−
i F
0
4π
C
e
i k(x
1
cos α+x
2
−x
20
 sin α)
dα. (6.82)
F
0
= A/k, where A is a dimensionless constant. The expression (6.82) gives
the free space, particle displacement u
i
3
excited by a line source at (0, x
20
). The
contour C begins near π −i ∞and ends near i ∞. How then does each plane
wave in the integral, (6.82), behave in the layered environment?
Consider an arbitrary observation point (x
1
, x
2
) in the layer, where x
1
> 0 and
x
2
> x
20
. Given that the x
1
dependence must be of the form e
i k cos α x
1
, in both
the layer and halfspace, to ensure phase matching, we ask what partial waves
can reach the point (x
1
, x
2
) and, when added together, satisfy the boundary
conditions. By sketching rays and wavefronts, as suggested in Fig. 6.6, we
ﬁnd that there are four distinct partial waves that reach (x
1
, x
2
) and that all the
subsequent partial waves can be constructed from these four by successively
reﬂecting them in the planes x
2
=0 and x
2
=−h.
e
i k[x
1
cos α+sin α(x
2
−x
20
)]
, R(α)e
i k[x
1
cos α−sin α(x
2
+x
20
)]
,
e
i k[x
1
cos α+sin α(x
2
+2h+x
20
)]
, R(α)e
i k[x
1
cos α−sin α(x
2
−2h−x
20
)]
. (6.83)
The ﬁrst partial wave reaches (x
1
, x
2
) without reﬂection, the second after being
Fig. 6.6. A sketch showing the ﬁrst four partial waves reaching the observation point
(x
1
, x
2
). The dashed lines indicate the propagation paths and the solid ones indicate the
wavefronts passing through (x
1
, x
2
).
6.4 Harmonically Excited Waves in an Open Waveguide 141
reﬂected once at the x
2
=0 boundary, the third after being reﬂected from the
x
2
=−h boundary, and the last after being reﬂected ﬁrst from the x
2
=−h
boundary and second fromthe x
2
=0 boundary. All subsequent reﬂected partial
waves can be constructed from these four. The reﬂection coefﬁcient at x
2
=−h
is 1 and at x
2
=0 is R(α), where R(α) is given by (6.44), with γ =k sin α
and ¯ γ =
¯
k sin ¯ α. Or it can be gotten from (3.21) and (3.23) by noting that
α =π/2 −θ
0
. The angles α and ¯ α are related by the phasematching condition,
namely k cos α =
¯
k cos ¯ α. Note that the argument of the function R is now
given as α (rather than as γ or θ
0
) in keeping with the structure of the present
calculation.
By adding these four partial waves and all the subsequent reﬂections and
invoking superposition, we construct the response of the layer to the excitation.
It is given by
u
3
= −
i F
0
4π
C
e
i k cos αx
1
e
i k sin α(x
2
−x
20
)
+ R(α)e
−i k sin α(x
2
+x
20
)
+e
i k sin α(x
2
+2h+x
20
)
+ R(α)e
−i k sin α(x
2
−2h−x
20
)
×
∞
m=0
[R(α)]
m
e
i k sin α(2mh)
dα (6.84)
To understand this representation further, we approximate one of the terms
by the method of steepest descents. We consider m =0, the fourth term,
namely
u
04
3
=−
i F
0
4π
C
R(α)e
i k[cos α x
1
−sin α (x
2
−2h−x
20
)]
dα. (6.85)
The geometry is shown in Fig. 6.7. By setting x
1
=r
04
cos θ
04
and x
2
−2h −
x
20
=−r
04
sin θ
04
, we put (6.85) in a form previously studied in Section 5.4.
Its asymptotic approximation is therefore
u
04
3
= −
i F
0
4π
C
R(α)e
i kr
04
cos(α−θ
04
)
dα ∼
2π
kr
04
1/2
F
0
4π
R(θ
04
)e
i kr
04
e
i π/4
.
(6.86)
This term then appears as a cylindrical wave radiated by an image source at
(0, 2h +x
20
). The actual path of the ray is indicate by the dashed line within
the layer in Fig. 6.7. Approximating each term in (6.84) by the method of
steepest descents gives an inﬁnite sum of terms such as (6.86). This is a ray
representation of the waveﬁeld in the layer.
142 6 Guided Waves and Dispersion
Fig. 6.7. Asketch of the geometry for the steepest descents approximation of the fourth
partial wave.
Problem 6.8 Ray Representation
Verify the statements just made by asymptotically approximating each term
in (6.84) to give
u
3
∼
∞
l=0
4
j =1
u
l j
3
, (6.87)
where each termis similar to that given by (6.86). Is this a useful representation,
and, if so, in what circumstances?
By summing the series in (6.84), the waveﬁeld can be recast as
u
3
= −
i A
4kπ
C
e
i k cos α x
1
[e
i k sin α x
2
+ R(α)e
−i k sin α x
2
]
×
e
−i k sin α x
20
+e
i k sin α (2h+x
20
)
1 − R(α)e
i k sin α(2h)
dα. (6.88)
F
0
= A/k has been used to indicate the dimensions of u
3
. Recall that we have
assumed x
2
> x
20
. When x
2
< x
20
, the two coordinates in the above expression
are interchanged.
To understand this representation we must examine the complex α plane.
However, the discussion will be clearer if we ﬁrst examine the β plane, where
β =k cos α. Recall that this is the Sommerfeld transformation and that it was
this transformation that lead us to (6.82). Figure 6.8 shows the β plane with
its branch cuts and several poles. The transverse wavenumbers γ and ¯ γ were
given in (6.43). These radicals are deﬁned so that (γ ) ≥0 and ( ¯ γ ) ≥0 ∀β.
6.4 Harmonically Excited Waves in an Open Waveguide 143
Fig. 6.8. A sketch of the complex β plane, showing the branch cuts and poles that have
emerged onto the physical sheet. The physical sheet is the one for which (γ ) ≥0 and
( ¯ γ ) ≥0 ∀ β. In the various quadrants the real parts take values as follows: quadrants
1 and 2, (γ ) >0 and ( ¯ γ ) >0; quadrants 3 and 4, (γ ) <0 and ( ¯ γ ) <0. These
quadrants correspond to their primed counterparts in the α plane in Fig. 6.11.
This is the Riemann sheet, as we have seen in Sections 3.4.4 and 5.4, that
ensures that the waves radiate or decay away from their source (satisfy the
principle of limiting absorption). We call this the physical sheet and the others
the unphysical ones. Figure 6.9 shows the α plane and the contour C. The branch
points are now given by ¯ α and π − ¯ α, where ¯ α is deﬁned by
¯
k =k cos ¯ α.
Fig. 6.9. The α plane. The poles tothe left are α
1
andα
2
; those tothe right are π −α
1
and
π −α
2
. The two branch points are ¯ α and π − ¯ α. Poles on the lower sheet progressively
move to the physical sheet, popping through the branch points as ¯ ω is increased.
144 6 Guided Waves and Dispersion
Fig. 6.10. The dispersion relation and the points on it that correspond to the poles
shown in Figs. 6.8 and 6.9.
Noting that γ =k sin α, we see that the poles of the integrand in (6.88) are
given by the dispersion relation
e
−2i γ h
= R(α), (6.89)
which is a restatement of (6.49). The poles of the integrand therefore give us
the trapped modes of the waveguide. For a given ω = ¯ ω, there is a pair of poles
on the physical sheet of the β or α plane corresponding to each guided mode
that is trapped in the layer. In Figs. 6.8 and 6.9, there are two pairs, ±β
1
and
±β
2
or (α
1
, π −α
1
) and (α
2
, π −α
2
). These points are shown on the dispersion
curves in Fig. 6.10. The modes that could radiate into the interior correspond
to poles that lie on an unphysical sheet. These are indicated by the crosses in
Fig. 6.8. As ¯ ω is increased, these poles move from the unphysical sheet to the
physical one, popping up through the branch point ¯ α in the α plane. The next
problem indicates how the residues from the poles on the physical sheet, in the
integrand of (6.88), become part of a modal sum, similar to that for a closed
waveguide.
Problem 6.9 Modal Representation
Show that the residue contributions of (6.88) give the Love waves of (6.51)
through (6.53). Do this by deforming the contour C to that shown in Fig. 6.11.
Note that there will be a branch cut integral as indicated in the ﬁgure. The sum
of residues gives the waves trapped in the layer or the sum over the discrete
6.4 Harmonically Excited Waves in an Open Waveguide 145
Fig. 6.11. The deformed contour for Problem 6.9. Propagation to the right is being
considered. The quadrants 1
, 2
, and 4
correspond to quadrants 1, 2, and 4 in the β
plane in Fig. 6.8.
eigenfunctions, whereas the branch cut integral gives the waveﬁeld that can ra
diate into the lower halfspace or the sum over the continuous eigenfunctions.
The combination is the modal expansion of the waveﬁeld in the layer.
Ray representations are useful for observation points near the source, whereas
modal representations are most descriptive a considerable distance from it, at
points where the waveﬁeld has lost some sense of how it was excited and
has adapted to its propagation environment. The modal representation can be
derived directly from (6.84) by using the Poisson sum formula, discussed in
Section 1.3. This is true even when we cannot sum the series (6.84).
6.4.2 The Waveﬁeld in the HalfSpace
The waveﬁeld in the halfspace is constructed in much the same way as was
that in the layer. The two fundamental partial waves reaching (x
1
, x
2
) are
T(α)e
i k cos α x
1
e
−i k sin α x
20
e
i
¯
k sin ¯ α x
2
, T(α)e
i k cos α x
1
e
i k sin α (x
20
+2h)
e
i
¯
k sin ¯ α x
2
.
(6.90)
The remainder can be constructed fromthese two. Integrating over all the partial
waves gives
u
3
= −
i F
0
4π
C
T(α)e
i k cos α x
1
e
i
¯
k sin ¯ α x
2
e
−i k sin α x
20
+e
i k sin α (2h+x
20
)
×
∞
m=0
[R(α)]
m
e
i k sin α (2mh)
dα. (6.91)
If each term were approximated by the method of steepest descents, then this
representation would give a ray series for the waveﬁeld in the halfspace. By
146 6 Guided Waves and Dispersion
carrying out the summation, we can recast the representation as
u
3
= −
i A
4kπ
C
T(α)e
i k cos α x
1
e
i
¯
k sin ¯ α x
2
×
e
−i k sin α x
20
+e
i k sin α (2h+x
20
)
1 − R(α)e
i k sin α (2h)
dα. (6.92)
F
0
= A/k has been used to indicate the dimensions of u
3
. The poles and branch
cuts of the integrand also give rise to a modal representation.
6.4.3 Leaky Waves
We continue to consider propagation solely to the right. Figures 6.8, 6.9, and
6.11 indicate how the β plane is structured and how it relates to the α plane.
Moreover, we continue to assume that there are two real roots, β
1
=k cos α
1
and
β
2
=k cos α
2
, present on the physical sheet. To the left of the ω/β = ¯ c line, in
Figs. 6.4 and 6.10, there are also roots to the dispersion relation if β is allowed
to be complex. They lie on the (γ ) ≥0, ( ¯ γ ) ≤0 sheet, an unphysical sheet of
the β plane, and therefore on the unphysical sheet in the α plane. Their positions
are approximately indicated by the crosses in both Figs. 6.8 and 6.11. To ensure
that waves radiate away from their source, we take (β) >0 so that the waves
in the layer radiate or leak into the interior as they propagate to the right. We call
them leaky waves or leaky modes. Note that ( ¯ γ ) >0 for propagation to the
right, but that ( ¯ γ ) <0 so that these waves increase exponentially with depth.
That is why they do not lie on the physical sheet of the Riemann surface. They
can, however, inﬂuence the solution if they come close to the branch cut or if the
branch cut is moved. These waves can be quite important for some observation
points. This issue is discussed at greater length in DeSanto (1992).
6.5 A Laterally Inhomogeneous, Closed Waveguide
We have considered both open and closed waveguides whose geometry has con
formed to a rectangular coordinate system. Moreover we have only considered
a very simple variation in the propagation environment with depth, namely the
layer on a halfspace. This last example of waveguiding considers a propagation
environment that changes slowly in the propagation or lateral direction. We con
sider a slow variation, with respect to wavelength, in the thickness of a closed
guide whose unperturbed geometry is that given in Fig. 6.1. We construct an
asymptotic approximation that makes use of rays to describe the propagation in
the lateral direction and modes to take account of the variation in the transverse
one.
6.5 A Laterally Inhomogeneous, Closed Waveguide 147
The equation of motion remains (6.1). To clarify the scales we introduce
scaled coordinates ¯ x
1
=kx
1
and ¯ x
2
=kx
2
. The slow variable y
1
=δkx
1
is in
troduced to describe the slow lateral variation. Also we set ¯ u
3
=ku
3
. Having
scaled the problem, we omit the overbar and reintroduce the variables x
2
and
u
3
with the understanding that these are now scaled variables. The equation of
motion becomes
δ
2
∂
2
u
3
/∂y
2
1
+
∂
2
u
3
/∂x
2
2
+u
3
=0. (6.93)
The boundary conditions require the vanishing of the normal traction. This is
the second way in which the small parameter δ enters the problem. The top and
bottom surfaces are given by
x
2
=±H
±
(x
1
) =±h
0
+h
±
(y
1
). (6.94)
[H
±
(x
1
+2π) − H
±
(x
1
)]/2π ≈δ[dh
±
/dy
1
], where we assume that dh
±
/
dy
1
= O(1); δ then measures the change in the thickness of the guide over
a wavelength. The outward unit normal vectors are
ˆ n
±
=
−δ
dh
±
dy
1
ˆ e
1
±ˆ e
2
+O(δ
2
)
. (6.95)
We now seek an asymptotic solution in the form
u
3
∼e
i θ(y
1
)/δ
ν≥0
A
ν
3
(y
1
, x
2
)δ
ν
, (6.96)
where
A
ν
3
(y
1
, x
2
) =
n≥0
a
n
ν
(y
1
) u
n
3
(y
1
, x
2
), (6.97)
and u
n
3
(y
1
, x
2
) is the nth mode for a waveguide whose thickness is determined
at y
1
. This asymptotic construction is often called the JWKB technique. Note
that the expansion has the formof a modulated group such as we encountered in
(6.74). For simplicity, assume from now on that the guide is symmetric so that
h
±
=h and H
±
= H. Then the u
n
3
(x
2
, y
1
) are the cosine or sine functions ﬁrst
given in (6.3), with h replaced by H(x
1
) and γ
n
replaced by kγ
n
, γ
n
=nπ/2H.
The boundary conditions at x
2
=±H become
µ
−δ
2
dh
dy
1
∂u
3
∂y
1
±
∂u
3
∂x
2
=0. (6.98)
Note that the boundary conditions are homogeneous, so that this equation is
exact.
148 6 Guided Waves and Dispersion
Now (6.96) is substituted into (6.93) and (6.98), and the coefﬁcient of each
power of δ set to zero. We are then led to the sequence of equations
L−
dθ
dy
1
2
A
ν
3
+i
d
2
θ
dy
2
1
+2
dθ
dy
1
∂ A
ν−1
3
∂y
1
+
∂
2
A
ν−2
3
∂y
2
1
=0, (6.99)
with their accompanying boundary conditions. We have introduced the opera
tor L, where L :=∂
2
/∂x
2
2
+1, to make the structure of the equations clearer.
The terms for which the superscripts are negative are zero. The equation corre
sponding to ν =0 is
L
A
0
3
=β
2
n
A
0
3
, (6.100)
with
∂ A
0
3
∂x
2
(y
1
, ±H) =0. (6.101)
To arrive at (6.100) we have set
(dθ/dy
1
)
2
=β
2
n
. (6.102)
This is an eikonal equation, similar in some respects to (2.43).
Equations (6.100) and (6.101) constitute an eigenvalue problem, identical to
that solved in Problem 6.1. Note that β
n
of (6.4) has become kβ
n
and that now
β
n
=
1 −γ
2
n
(y
1
)
1/2
, (6.103)
where γ
n
=nπ/[2h
0
+2h(y
1
)]. Further,
u
n
3
(y
1
, x
2
) = N
n
cos
sin
(γ
n
x
2
), (6.104)
where N
n
are constants, often determined by a normalization condition. The
solution to (6.100) and (6.101) is then A
0
3
=a
n
0
(y
1
)u
n
3
(y
1
, x
2
), with β
2
n
as the
corresponding eigenvalue. Note that y
1
enters β
n
through h(y
1
). The reader is
asked to recall the earlier comments in Problem 6.1 as to what the eigenvalue
of interest is. The eikonal equation, (6.102), is integrated to give θ(y
1
).
To ﬁnd a
n
0
we need to go to the next order in δ. This equation is
−L
A
1
3
+β
2
n
A
1
3
=i
d
2
θ
dy
2
1
A
0
3
+2i
dθ
dy
1
∂ A
0
3
∂y
1
, (6.105)
6.5 A Laterally Inhomogeneous, Closed Waveguide 149
with boundary conditions, at x
2
=±H,
±
∂ A
1
3
∂x
2
=i
dh
dy
1
dθ
dy
1
A
0
3
. (6.106)
We shall not seek terms of order higher than ν =0 here. Burridge and Weinberg
(1977) indicate how higherorder terms are gotten.
Examining (6.105) and (6.106), we note that to have a bounded solution for
A
1
3
it must not resonate with the A
0
3
forcing term. That is, it must be orthogonal
in some sense to u
n
3
. We use this fact to ﬁnd a
n
0
. First we introduce the inner
product
[a, b] :=
_
H
+
H
−
a(x
2
) b
∗
(x
2
)dx
2
(6.107)
and agree to normalize the eigenfunctions such that [u
n
3
, u
m
3
] =δ
nm
. The N
n
are
then given by
N
n
={
n
[h
0
+h
1
(y
1
)]}
−1/2
(6.108)
with
n
=2 for n =0, and
n
=1, otherwise. Calculating the inner product
[L(A
1
3
), u
n
3
] and using (6.105) and (6.106) gives the following transport equation
for a
n
0
, namely
dθ
dy
1
da
n
0
dy
1
+
1
2
a
n
0
d
2
θ
dy
2
1
+a
n
0
dθ
dy
1
_
_
∂u
n
3
/∂y
1
, u
n
3
_
+
1
2
dh
dy
1
_¸
¸
u
n+
3
¸
¸
2
+
¸
¸
u
n−
3
¸
¸
2
_
_
=0, (6.109)
where the superscript plus and minus signs mean that these terms are evaluated
at x
2
=±H, respectively. Note that
_
u
n
3
,
∂u
n
3
∂y
1
_
=−
_
∂u
n
3
∂y
1
, u
n
3
_
−
dh
dy
1
_¸
¸
u
n+
3
¸
¸
2
+
¸
¸
u
n−
3
¸
¸
2
_
. (6.110)
Taking the complex conjugate of (6.109) and adding the two gives the much
simpler equation
1
2
d
dy
1
ln
_
dθ
dy
1
dθ
∗
dy
1
_
+
d
dy
1
_
a
n
0
a
n∗
0
_
=0. (6.111)
From this it follows immediately that
¸
¸
¸
¸
dθ
dy
1
¸
¸
¸
¸
¸
¸
a
n
0
¸
¸
2
=constant. (6.112)
150 6 Guided Waves and Dispersion
Equation (6.112) is a statement of energy conservation. In other words, to lowest
order, no waves are reﬂected by the slowly changing width. The unknown
constant would be determined from an initial condition at x
1
=0.
To determine the argument θ
n
0
of a
n
0
=a
n
0
e
i θ
n
0
, substitute this into (6.109).
It is readily determined that θ
n
0
is a constant. Thus
a
n
0
=c
n
0
e
i θ
n
0
β
1/2
n
, (6.113)
where c
n
0
is a real constant. Bringing the pieces together, we ﬁnd the approximate
expression for the nth mode is
u
3
∼exp
i
δ
y
1
β
n
(s) ds
a
n
0
(y
1
)N
n
(y
1
)
cos
sin
(γ
n
x
2
). (6.114)
The β
n
is given by (6.103), N
n
by (6.108), and a
n
0
by (6.113).
This particular approach can be extended to inplane elastic waves (Folguera
and Harris, 1999). Among several newer features, arising when this is done is a
reformulation of the inplane eigenvalue problemand the use of an inner product
for the orthogonality condition that is not the norm of the space in which the
problem is set.
6.6 Dispersion and Group Velocity
6.6.1 Causes of Dispersion
We ﬁrst encountered dispersion in the study of a periodic structure and then
later in the study of guided waves. The periodic structure could be imagined to
be a rod possessing a periodic microstructure represented by the concentrated
masses. It is the microstructure, in combination with the additional length scale
introduced by it, that causes the dispersion. In fact, we could have examined
the periodic structure by beginning with the equation
∂
2
1
u
1
=
1 +
M
ρ
∞
−∞
δ(x
1
−nL)
1
c
2
b
∂
2
t
u
1
. (6.115)
In the end we should have solved it much as we did, but writing the equation
in this form exhibits exactly how the microstructure enters the equation.
For guided waves the dispersion is caused by the geometrical or kinematical
constraint that the waves must reﬂect from the boundaries in such a way that
they reinforce one another to form a sustained waveﬁeld. By considering only
6.6 Dispersion and Group Velocity 151
the nth waveguide mode for a closed waveguide, written as
u
3n
= A
n
cos[(nπ/h)x
2
]P
n
(x
1
, t ), (6.116)
we ﬁnd, by substituting this into the equation of motion, (6.1), that the propa
gating term P
n
(x
1
, t ) satisﬁes
c
−2
∂
2
t
P
n
−∂
2
1
P
n
+(nπ/h)
2
P
n
=0. (6.117)
Note that (6.2) is automatically satisﬁed and that this equation captures the
dispersive propagation.
The equation (6.117) is the Klein–Gordon equation, (6.27), and both (6.115)
and (6.117) have the same form as do the Klein–Gordon equation and that for
the ﬂexural motion for a rod, (6.28). In Problem6.2 the reader showed that both
these latter equations led to dispersive propagation.
Continuing, recall that in Section 6.1.2 we showed that the energy in the
nth mode propagates at the group velocity C
n
, and that in Section 6.3.2, using
the method of stationary phase, we showed that the group velocity is that with
which each ω
ns
propagated to a point (x
1
, t ). While it would be misleading to
assert that the energy in all waveﬁelds or that the information encoded in them
always propagates at the group velocity, this does occur for many waveﬁelds.
In this section, we study dispersion by focusing our attention on the meanings
given the group velocity in the context of one dimensional, dispersive equations
having a form such as (6.115) or (6.117).
Dispersive propagation is an extraordinarily rich subject. In addition to the
work of Lighthill (1965) and Whitham (1974), which we introduce here, the
reader is referred to Sommerfeld (1964a) and Brillouin (1960) for extended
discussions of the dispersion of electromagnetic waves in dielectric materials.
6.6.2 The Propagation of Information
We consider a signaling problem. The wave, amplitude u(x, t ), begins at x =0
as a square pulse modulating a harmonic carrier and propagates outward in the
positive x direction. It is described at x =0 by
u(0, t ) =(t ) cos ω
0
t, (6.118)
with the modulation described by
(t ) = H(t +T) − H(t −T). (6.119)
152 6 Guided Waves and Dispersion
As in previous instances, H(t ) is the Heaviside function. The dispersion relation
has been found by seeking a solution in the form Ae
i (kx−ωt )
and solving for
k =k(ω). We construct a solution to the signaling problem as
u(x, t ) =
π
∞
0
∗
u(0, ω)e
i [k(ω)x−ωt ]
dω, (6.120)
where the transform of u(0, t ) is given by
∗
u(0, ω) =
sin[(ω−ω
0
)T]
ω−ω
0
+
sin[(ω+ω
0
)T]
ω+ω
0
. (6.121)
Note that (1.36) has been used to write (6.120). If ω
0
T is large,
∗
u(0, ω) is
concentrated within intervals near ±ω
0
. In particular, the major contribution to
the integral, (6.120), comes from ω ∈ (ω
0
−π/T, ω
0
+π/T). In this interval
we can approximate the dispersion relation as
k(ω) ≈k(ω
0
) +
dk
dω
(ω
0
)(ω−ω
0
), (6.122)
provided the derivative of k(ω) is well behaved and does not vanish. We can
thus approximate (6.120) as
u(x, t ) ≈
π
e
i (k
0
x−ω
0
t )
M
t −
x
C
, (6.123)
where
M
t −
x
C
≈
ω
0
+π/T
ω
0
−π/T
∗
u(0, ω)e
−i (ω−ω
0
)(t −x/C)
dω, (6.124)
provided ω
0
T is large; k
0
=k(ω
0
) and C
−1
=dk/dω(ω
0
). Using the change of
variable =ω−ω
0
, we can write (6.124) as
M(t −x/C) ≈
π/T
−π/T
∗
()e
−i (t −x/C)
d, (6.125)
where
∗
() is the Fourier transform of (t ). Therefore, provided
∗
u(0, ω) is
concentrated near ω
0
, the modulation (t ) ≈ M(t ) propagates at the speed C,
the group velocity.
This is essentially the kinematic argument put forward by Stokes
(Sommerfeld, 1964b) to demonstrate that information propagates at the group
velocity. There are shortcomings to this argument. The principal one is that
∗
u(0, ω) must be concentrated in the neighborhood of the carrier frequency
ω
0
, in order that there be a welldeﬁned modulation. This, however, may not
6.6 Dispersion and Group Velocity 153
always be the case, and that raises the question as to whether or not the concept
of group velocity has a more general signiﬁcance.
6.6.3 The Propagation of Angular Frequencies
Whitham (1974) and Lighthill (1965) give a more general meaning to group
velocity that is motivated by the stationary phase condition. However, they
consider an initial value problem rather than a signaling problem so that their
formulation is given in terms of a local wavenumber. Problem6.10 suggests why
they take this approach. They propose both a kinematic theory describing the
propagation of a local wavenumbers as well as a kinetic theory describing the
evolution of a group. Little more than a dispersion relation and the assumption
that far from from the source a waveﬁeld evolves into a form described by
u(x, t ) ∼[A(x, t )e
i θ(x,t )
] is required.
Recall that following (6.74) we interpreted the group velocity as that with
which a given local angular frequency, one that characterizes a group, propa
gates. We chose to describe the local angular frequency, rather than the local
wavenumber, because we were workingwitha signalingproblemand, therefore,
it was natural to express the solution as a Fourier transform over the angular
frequency ω. Nevertheless, the ideas of Whitham and Lighthill prove just as
adept at describing the evolution of the local angular frequency and the group
it characterizes.
A Signaling Problem
We again consider the signaling problem whose solution is given by (6.121).
However, we now assume that ω
0
T is such that
∗
u(0, ω) is slowly varying. We
approximate (6.120) for large x, assuming that t /x is ﬁxed, using the stationary
phase approximation (5.69). This gives
u(x, t ) ∼
π
(2π)
1/2
[xd
2
k/dω
2
]
1/2
∗
u(0, ω)e
i [k(ω)x−ωt ±π/4]
. (6.126)
The stationary phase condition is dk/dω =t /x. The solution to this equation
3
gives ω =ω(x, t ), and k =k(x, t ) through the dispersion relation k =k(ω). We
noted previously, following (6.73), that the stationary phase condition, written
as
C(ω) = x/t, (6.127)
3
The argument is really x/t rather than the more general (x, t ). However, this latter notation will
permit some generalizations in the subsection that follows.
154 6 Guided Waves and Dispersion
is a mapping, for propagating waves, fromthe t plane to the ω plane for ﬁxed x.
In this calculation we emphasize this aspect of the stationary phase condition
and have dropped the subscript s that was used previously to designate an ω
satisfying this condition. The term d
2
k/dω
2
is given by
d
2
k
dω
2
(ω) =−
1
C
2
(ω)
dC
dω
(ω). (6.128)
We assume that dC/dω > 0 and take the minus sign in (6.126). If dC/dω < 0,
then the reasoning remains the same, though signs change here and there.
Moreover, we are, throughout, assuming that dC/dω =0. Thus, u(x, t ) ∼
[A(x, t )e
i θ(x,t )
], where
A(x, t ) =
1
π
(2π)
1/2
[xd
2
k/dω
2
]
1/2
∗
u(0, ω)e
−i π/4
(6.129)
and
θ(x, t ) =k(ω)x −ω(x, t )t. (6.130)
A Kinematic Theory
Assuming that we know θ(x, t ) and a dispersion relation k =k(ω), we can
deﬁne a local angular frequency and a local wavenumber as
ω(x, t ) =−∂
t
θ, k(x, t ) =∂
x
θ. (6.131)
If these deﬁnitions are to be consistent with one another,
∂
t
k +∂
x
ω =0, (6.132)
or, by using the dispersion relation,
∂
t
ω+C(ω)∂
x
ω =0, (6.133)
where C =dω/dk. This is a partial differential equation for the angular fre
quency ω. Note that (6.132) has the form, deﬁned by (6.30), of a conservation
equation, and that (6.133) captures the idea suggested by the stationary phase
condition that the frequency ω propagates at the group velocity C.
Equation (6.133) is solved by the method of characteristics (Whitham, 1974;
Zauderer 1983). Provided the characteristics do not intersect, the solution can
be gotten quite simply.
1. Assume that there exists a curve S in the (x, t ) plane such that along S,
dt /dx =C
−1
. Then (6.133) becomes dω/dt =0 along this curve.
6.6 Dispersion and Group Velocity 155
2. ω(x, t ) is therefore constant along S and, because C is constant if ω is, S is
given by t = x C
−1
(ω) +t
0
, where t
0
is the intercept with the t axis. Hereafter
we write S(t
0
) to identify that member of the family of curves with inter
cept t
0
.
3. Assume that an initial distribution of frequencies ω(0, t
0
) = f (t
0
) is given.
Along each S(t
0
), ω(0, t
0
) =ω(x, t ). It follows then that ω(x, t ) = f [t −
x C
−1
(ω)] along S(t
0
). Variation of t
0
gives a solution throughout (x, t ).
It is not an explicit solution for ω. Nevertheless, it indicates its evolution. Lastly,
if we assume that f (t
0
) is localized near the origin so that t
0
/x 1 for x large
and t /x ﬁxed, we recover the stationary phase condition, (6.127).
We can also deﬁne a local phase velocity. Assume θ(x, t ) =θ
0
, a constant.
Implicit differentiation gives
c(x, t ) =−∂
t
θ/∂
x
θ, (6.134)
where the phase velocity c =dx/dt . Thus c =ω/k where ω and k are the local
frequency and wavenumber deﬁned by (6.126).
We call lines of constant angular frequency group lines and those of constant
phase phase lines. Reverting to the notation of Section 6.3.2, we again consider
the stationary phase approximation to the transiently excited waves in a closed
waveguide. The local angular frequency determined by the stationary phase
condition is given by (6.129). That expression, rewritten here, is
ω
n
(x
1
, t ) =
(cnπ/h)
[1 −(x
1
/ct )
2
]
1/2
. (6.135)
The phase, given by the argument of the cosine in (6.72), is
θ
n
(x
1
, t ) =−
x
1
nπ
h
ct
x
1
2
−1
1/2
. (6.136)
For both expressions ω
n
∈ (cnπ/h, ∞). The group lines or curves of constant
local angular frequency ω
n
=ω
n0
are given by
x
1
ct
=
1 −
cnπ
ω
n0
h
2
1/2
. (6.137)
The curves of constant phase θ
n
=θ
n0
are given by
c
2
t
2
−x
2
1
=
θ
n0
h
nπ
2
. (6.138)
156 6 Guided Waves and Dispersion
These two families of curves are quite different so that, once again, we note
that a phase and a group evolve differently in a dispersive environment.
A Kinetic Theory
Previously, we explicitly showed that for a closed waveguide the averaged
energy of a particular mode propagates with the group velocity for that mode,
(6.25). Auld (1990) uses a reciprocity relation to prove this result in somewhat
greater generality. Lighthill (1965) gives a general argument showing that, for
any wave structure or environment that can be characterized by a Lagrangian
and for which equipartition of energy occurs, the group velocity is the velocity
with which the average energy propagates. Aki and Richards (1980) give a
restricted version of this argument that is applicable to waveguide modes. Here
we continue to follow Whitham (1974) and Lighthill (1965) and advance an
argument using the stationary phase condition.
We deﬁne a quantity Q as follows.
Q(x) =
t
2
t
1
A(x, t )A
∗
(x, t )ω
2
(x, t )dt, (6.139)
where t
2
>t
1
. This is reminiscent of an energy term. For a time harmonic, plane
wave and wave system for which equipartition of energy occurs, this term is
proportional to the timeaveraged energy density. Recalling our earlier comment
that (6.127) is a mapping from t to ω, we change the variable of integration in
(6.138) to ω. Noting that
dt = −
x
C
2
(ω)
dC
dω
(ω)dω, (6.140)
we see that (6.139) becomes
Q(x) =
2
π
ω
1
ω
2

∗
u(0, ω)
2
ω
2
dω (6.141)
For dC/dω >0, ω
1
>ω
2
. This is a remarkable result. It indicates that for a sig
nal, such as a fragment of speech, the energy contained within a given frequency
band is constant when the frequency band remains between the two group lines
identiﬁed by t
i
= x C
−1
(ω
i
), where i = 1, 2. This then is a generalization of the
idea that timeaveraged energy propagates at the group velocity.
While the kinematic and kinetic descriptions of group velocity put forward by
Whitham (1974) and Lighthill (1965) appear quite complete, they assume that
attenuation is not present. When dispersion arises from material properties,
it is very often accompanied by frequencydependent attenuation. The same
References 157
physical mechanisms are responsible for both dispersion and attenuation so that
they cannot be easily disentangled, unless the attenuation is quite weak. The
propagation of linearly viscoelastic waves (Hudson, 1980) is an example that
indicates the tangling of these two effects. In such cases, attributing a meaning
to group velocity can become much more difﬁcult. An ab initio calculation may
be the only guide (see also footnote 2).
Problem 6.10 Dispersion 2
As we indicated previously, Whitham (1974) and Lighthill (1965) use an
initial value problem to advance their ideas about group velocity. Thus they
write that the wavenumber propagates at the group velocity and that the energy
conﬁned between two group lines deﬁned by two ﬁxed k
i
is constant. The
following two problems indicate why they develop their ideas in this way.
Problem 1. Consider once again the Klein–Gordon equation, (6.27). Solve
the initial value problem for this equation, given the initial conditions
ϕ(x, 0) =δ(x), ∂
t
ϕ(x, 0) =0. (6.142)
For t < 0
−
, ϕ and ∂
t
ϕ are identically zero. Thus show that the appropriate
solution is
ϕ =
1
π
∞
0
cos(kx) cos[ω(k)t ]dk, (6.143)
where ω =ω(k) is the dispersion relation. Note that (6.143) is to be interpreted
as a generalized function or distribution. Note that k, and not ω, is the transform
variable.
Using the method stationary phase, show that this solution for t →∞, with
x/t positive and ﬁxed, can be put into the form
ϕ ∼
A(x, t )e
i θ(x,t )
. (6.144)
Give explicit expressions for A(x, t ) and θ(x, t ).
Problem 2. Repeat Problem 1 for the equation for ﬂexural motion in a rod,
(6.28).
References
Aki, K. and Richards, P.G. 1980. Quantitative Seismology, Theory and Methods,
Vol. 1, pp. 286–292. San Francisco: Freeman.
158 6 Guided Waves and Dispersion
Auld, B.A. 1990. Acoustic Fields and Waves in Solids, 2nd ed., Vol. 2, pp. 203–206.
Malabar, FL: Krieger.
Brekhovskikh, L.M. 1980. Waves in Layered Media, 2nd ed., pp. 299–325. New York:
Academic.
Brillouin, L. 1960. Wave Propagation and Group Velocity, 2nd ed. New York:
Academic.
Burridge, R. and Weinberg, H. 1977. Horizontal rays and vertical modes. In Wave
Propagation and Underwater Acoustics, pp. 86–152, ed. J.B. Keller and
J.S. Papadakis. New York: Springer.
Dai, H.H. and Wong, R. 1994. A uniform asymptotic expansion for the shearwave
front in a layer. Wave Motion 19: 293–308.
deSanto, J.A. 1992. Scalar Wave Theory, pp. 162–179. New York: Springer.
Folguera, A. and Harris, J.G. 1999. Coupled Rayleigh surface waves in a slowly
varying elastic waveguide. Proc. R. Soc. Lond., A 455: 917–931.
Friedman, B. 1956. Principles and Techniques of Applied Mathematics. New York:
Wiley.
Hudson, J.A. 1980. The Excitation and Propagation of Elastic Waves,
pp. 188–222. Cambridge: University Press.
Lighthill, M.J. 1965. Group velocity. J. Inst. Maths. Appl., 1: 1–28.
Miklowitz, J. 1978. The Theory of Elastic Waves and Waveguides, pp. 178–200 and
409–466. New York: NorthHolland.
Sommerfeld, A. 1964a. Optics, Lectures on Theoretical Physics, Vol. IV,
pp. 273–289. Translated by O. LaPorte and P.A. Moldauer. New York: Academic.
Sommerfeld, A. 1964b. Mechanics of Deformable Bodies, Lectures on Theoretical
Physics, Vol. II, pp. 184–191. Translated by G. Keurti. New York: Academic.
Tolstoy, I. and Clay, C.S. 1987. Ocean Acoustics, Theory and Experiment in
Underwater Sound, pp. 76–80 and elsewhere. New York: American Institute of
Physics.
Whitham, G.B. 1974. Linear and Nonlinear Waves, pp. 363–402. New York:
WileyInterscience.
Zauderer, E. 1983. Partial Differential Equations of Applied Mathematics,
pp. 35–77. New York: WileyInterscience.
Index
κ, deﬁned, 40
κ
r
, deﬁned, 55
Abelian theorem, 82, 90, 107
allied function, 46
angular frequency
deﬁned, 9
angularspectrum; see planewave, 24
asymptotic approximation of integrals, 86–96
end point contribution, 95
Fresnel integral, 119
integration by parts, 87
stationary phase, 94–95
contour, 92
waveguide modes, 136
steepest descents, 90–94
branch cut contribution, 99
contour, 91, 98
pole contribution, 99
Stokes’ phenomenon, 119
uniform, 108
Watson’s lemma, 82, 87–90
wavefront approximation, 82, 95
asymptotic power series, 30
asymptotic ray expansion, 28–34, 112
compressional, 29
shear, 33
average
for a closed waveguide mode, 126
Lagrangian density, 126
time average for a plane wave, 23
boundary layer; see matched asymptotic
expansion, 109
branch cuts, deﬁning, 52–55
buried harmonic line of compression, 82–86,
96–101
asymptotic approximation of the scattered
compressional wave, 98
asymptotic approximation of the scattered
shear wave, 99
Cagniard–deHoop technique, 77–82
contour, 81
inversion, 79–82
caustic, 31, 85, 100
center of compression
threedimensional, 61
twodimensional, 62
complex unit vector, 23
compressional wave, deﬁned, 5
critical angle
incidence, 43
reﬂection, 44
refraction, 43
cutoff frequency, 125, 133
cylindrical wave, 34
diffraction at an edge, 101–119
diffracted wave, 108
diffraction coefﬁcient, 110
diffraction integral, 108; see Fresnel
integral
geometrical theory, 110
dispersion; see velocity, group, 13
and stationary phase, 136
anisotropic medium, 128
anomalous, 138
causes of, 150
closed waveguide, 122, 125
from the poles of an integral
representation, 144
geometrical, 150
microstructure, 150
159
160 Index
dispersion (cont.)
open waveguide, 130, 131
periodic structure, 18
eigenvalue problem, 122
eigenfunction
continuous expansion, 139
discrete or mode expansion, 134
eigenvalues, 123
discrete and continuous, 145
eigenvalues and eigenfunctions
discrete, 122
discrete and continuous, 131
eikonal equation, 30, 148
energy relations, 5
kinetic energy density, deﬁned, 6
averaged for a closed waveguide mode, 126
averaged for a plane wave, 23
conservation law, 6, 127
energy ﬂux during critical refraction, 45
energy ﬂux during reﬂection, 41
energy ﬂux, deﬁned, 6
energy in a band of frequencies, 156
equipartition of energy, 126
for a transient plane wave, 22
for the ﬂexural motion equation, 128
for the Klein–Gordon equation, 128
internal energy density, deﬁned, 6
equations of motion, 1–6
boundary conditions, 2
dilitational motion, 2
onedimensional, 3
rotational motion, 2
twodimensional, 4
antiplane motion, 4
inplane motion, 4
extinction theorem, 75
farﬁeld
compact source, 61, 65
of an edge, 114
Fermat’s principle, 101
ﬂexural motion equation, 48, 127, 157
Fourier transform
space, 10
threedimensional, 58
time, 7
frequency
deﬁned, 9
local, 155
Fresnel integral, 109, 116–119
gamma function, 87
Gaussian beam; see planewave
representations, 25
geometrical theory of diffraction;
see diffraction at an edge, 111
Green’s function, 60, 62
method of images, 68
Green’s tensor, 58, 62
correct and incorrect, 66
elastic ﬂuid, 74
stress, 60
threedimensional, 58
guided waves; see waveguide, 121
Helmholtz theorem, 5
initial value problem, 157
inner expansion; see matched asymptotic
expansion, 71
integral equations, 67, 76
integral representation
scattering problem, 65–67
source problem, 64–65
JWKB asymptotic expansion
eigenvalue problem, 148
rays and modes, 147
Klein–Gordon equation, 127, 150, 157
Lagrangian density, 126
Lamb’s problem; see buried harmonic line of
compression, 82
Laplace transform
time, 7
twosided, 78
leaky wave; see waveguide, open, 146
limitingabsorption, principle of, 62–63, 65, 135
and the Wiener–Hopf method, 104–108
outgoing wave, 58, 59, 79
line of compression; see center of compression,
83
longitudinal wave; see compressional, 22
Love wave; see waveguide, open, 131
matched asymptotic expansion, 112–116
boundary layer, 109, 111, 113
inner, outer coordinates, 71, 114
inner, outer expansion, 71, 114
matching, 115
Schrodinger equation, 114
stretching transformation, 114
nearﬁeld of an edge, 72, 104, 110, 113
outer expansion; see matched asymptotic
expansion, 71
outgoing wave; see limiting absorption, 59
partial waves, 17, 123, 140, 145
periodic structure, 15–18
effective wavenumber, 18
Index 161
phase matching, 39–40
phase, its meaning, 29
plane wave
homogeneous, 22
inhomogeneous or evanescent, 22, 43, 96
timedependent, 20
timeharmonic, 22
planewave representations, 24–28
cylindrical wave, 28, 140
Gaussian beam, 24
in an open waveguide, 142, 146
spherical wave, 27, 28
waves scattered from a surface, 84
Poisson summation formula, 11
polarization
change in a ray expansion, 33
deﬁned, 20
potentials, displacement, 5
scalar potential, 5
vector potential, 5
propagation matrix, 16
radiation conditions, 65; see limiting
absorption, 65
ray; see asymptotic ray expansion, 30
deﬁned, 30
fan of diffracted rays, 110
ray tube, 33
rays and modes, 146
Rayleigh wave, 48–52, 96, 101
Rayleigh function, 51
Rayleigh equation, 49
timeharmonic, 49
transient, 50
twosided wave, 51
reciprocity, 57
reﬂection
antiplane shear coefﬁcient, 44
antiplane shear wave, 42
compressional coefﬁcient, 40
compressional wave, 38
from a layer, 133
from a plate in a ﬂuid, 47
inplane shear coefﬁcient, 40
inplane shear wave, 38
refraction
antiplane shear coefﬁcient, 44
antiplane shear wave, 42
interfacial wave, 45
twosided wave, 46
scalar potential; see potentials,
displacement, 5
scattering
Bragg, 17
from a lumped mass, 13
from a strip, 67
from an elastic inclusion, 72
scattering matrix, 14
shear wave, deﬁned, 5
signaling problem, 151–157
deﬁned, 24
slowness
deﬁned, 20
surface, 39, 44
vector, 20, 39
Sommerfeld transformation, 27, 59, 84, 96, 142
spherical wave, 34
standing wave, deﬁned, 9
stationary phase; see asymptotic approximation
of integrals, 94
steepest descents; see asymptotic
approximation of integrals, 90
strain tensor, deﬁned, 2
stress tensor, deﬁned, 1
surface wave; see Rayleigh wave, 51
guided by an impedance boundary, 51
Tauberian theorem; see Abelian theorem, 82
traction, deﬁned, 1
transforms, Fourier and Laplace; see Fourier
transform, Laplace transform, 6
deﬁned, 6–10
transmission coefﬁcient; see refraction, 44
transmission matrix, 15
transport equation, 30, 149
transverse resonance principle, 124
transverse wave; see shear, 22
uniqueness in an unbounded region, 63,
68–72
edge conditions, 69, 104, 107
no edges, 68
vector potential; see potentials, displacement, 5
velocity
energy transport, 22, 24, 125
fastest, 137
group; see dispersion, 18
and stationary phase, 137
group lines, 155
kinematic theory, 154
kinetic theory, 156
periodic structure, 18
propagation of angular frequencies,
153–157
propagation of information, 151
propagation of wavenumbers, 153
waveguide mode, 125
phase
local, 155
phase lines, 155
waveguide mode, 125
vibration, not a wave, 11
162 Index
Watson’s lemma; see asymptotic approximation
of integrals, 87
wave kinematics, deﬁned, 20
wavefront
curvature, 32
deﬁned, 30
plane, spherical, etc., 20
waveguide
closed, 121
evanescent mode, 135
excitation, 134–139
harmonic excitation, 134
mode expansion, 134
transient excitation, 135
laterally inhomogeneous, 146
mode, 122
open, 129
excitation, 139–146
leaky modes, 146
modal representation, 144
planewave representation, 142, 146
ray representation, 141, 145
wavelength, deﬁned, 9
wavenumber
complex, 22
deﬁned, 8
effective, periodic structure, 18
lateral, 122
local, 155
transverse, 122
wavevector
deﬁned, 22
real and complex, 22
Wiener–Hopf method, 104–108
This page intentionally left blank
Linear Elastic Waves
Wave propagation and scattering are among the most fundamental processes that we use to comprehend the world around us. Whereas these processes are often very complex, one way to begin to understand them is to study wave propagation in the linear approximation. This is a book describing such propagation using, as a context, the equations of elasticity. Two unifying themes are used. The ﬁrst is that an understanding of plane wave interactions is fundamental to understanding more complex wave interactions. The second is that waves are best understood in an asymptotic approximation where they are free of the complications of their excitation and are governed primarily by their propagation environments. The topics covered include reﬂection, refraction, propagation of interfacial waves, integral representations, radiation and diffraction, and propagation in closed and open waveguides. Linear Elastic Waves is an advanced level textbook directed at applied mathematicians, seismologists, and engineers. John G. Harris received an undergraduate degree (honors) in 1971 from McGill University, where he studied electrical engineering and gained a lasting interest in wave propagation. In 1979, he received a doctoral degree in applied mathematics from Northwestern University for a thesis on elasticwave diffraction problems. Following this, he joined the Theoretical and Applied Mechanics Department at the University of Illinois (UrbanaChampaign), where he teaches dynamics and mathematical methods and carries out research on elasticwave problems at microwave frequencies. He has used asymptotic methods of analysis to explore diffraction and imaging, propagation and scattering of interfacial waves, and waveguiding.
.
DRAZIN AND R. DRAZIN Stability. C. I. FOWLER Thinking About Ordinary Differential Equations ROBERT E. FOKAS Mathematical Models in the Applied Sciences A. CIARLET Integral Equations DAVID PORTER AND DAVID S. MAUGIN Boundary Integral and Singularity Methods for Linearized Viscous Flow C. D. G. NAUGOLNYKH AND L. C. GIBBON Viscous Flow H. G. J. HINCH The Thermomechanics of Plasticity and Fracture GERARD A. DAVIDSON Linear Elastic Waves JOHN G. DOERING AND J. SEWELL Solitons P. OCKENDON Scaling. SelfSimilarity. KING An Introduction to Magnetohydrodynamics P. OSTROVSKY Nonlinear Systems P. O’MALLEY A Modern Introduction to the Mathematical Theory of Water Waves R. JOHNSON The Kinematics of Mixing J. OCKENDON AND J. JOHNSON Rareﬁed Gas Dynamics CARLO CERCIGNANI Symmetry Methods for Differential Equations PETER E. M. J. STIRLING Perturbation Methods E. POZRIKIDIS Nonlinear Wave Processes in Acoustics K. A. S. BILLINGHAM AND A. OTTINO Introduction to Numerical Linear Algebra and Optimisation PHILIPPE G. R. G. J. S. CHAPMAN Wave Motion J. HYDON High Speed Flow C. ABLOWITZ AND ATHANASSIOS S. HARRIS . Instability and Chaos PAUL GLENDINNING Applied Analysis of the Navier–Stokes Equations C. BARENBLATT A First Course in the Numerical Analysis of Differential Equations ARIEH ISERLES Complex Variables: Introduction and Applications MARK J. and Intermediate Asymptotics G.Cambridge Texts in Applied Mathematics Maximum and Minimum Principles M. R.
.
Linear Elastic Waves JOHN G. HARRIS University of Illinois at UrbanaChampaign .
Cambridge. Spain Dock House. UK 40 West 20th Street.org © Cambridge University Press 2004 First published in printed format 2001 ISBN 0511040415 eBook (netLibrary) ISBN 0521643686 hardback ISBN 052164383X paperback . The Pitt Building. The Waterfront. NY 100114211. USA 477 Williamstown Road. New York. Trumpington Street. Cambridge CB2 2RU. Port Melbourne. Cape Town 8001. 28014 Madrid. United Kingdom The Edinburgh Building. VIC 3207.cambridge. Australia Ruiz de Alarcón 13. South Africa http://www.
To Beatriz .
.
3.3.3 Kinematical Descriptions of Waves TimeDependent Plane Waves TimeHarmonic Plane Waves PlaneWave or AngularSpectrum Representations 2.2 2.1 A Gaussian Beam 2.1 OneDimensional Models 1.3.4.1 An Isolated Interaction 1.3 Displacement Potentials 1.1.3 An AngularSpectrum Representation of a Cylindrical Wave 2.1 2.1 Model Equations 1.1 Compressional Wave 2.4 Energy Relations 1.2 An AngularSpectrum Representation of a Spherical Wave 2.2 The Fourier and Laplace Transforms 1.1.4 Dispersive Propagation 1.4.1.4 Asymptotic Ray Expansion 2.4.3 A Wave Is Not a Vibration 1.4.2 Periodic Structures References 2 2.Contents Preface 1 Simple Wave Solutions 1.2 Shear Wave ix page xiii 1 1 3 4 5 5 6 11 13 13 15 18 20 20 22 24 24 26 28 28 29 33 .1.2 TwoDimensional Models 1.
x Contents 34 35 37 37 39 40 41 44 48 49 50 51 52 55 56 56 57 58 61 62 64 65 65 66 68 68 69 71 72 76 77 77 78 79 82 86 Appendix: Spherical and Cylindrical Waves References 3 Reﬂection.3 Asymptotic Approximation of Integrals .1 No Edges 4.1 The TimeHarmonic Wave 3.2 4.4.1 Antiplane Radiation into a HalfSpace 5.3 Green’s Tensor and Integral Representations Introduction Reciprocity Green’s Tensor 4.1 Notes 4.1.3 The Rayleigh Function 3.5 Integral Representation: A Source Problem 4.1 Reﬂection of a Compressional Plane Wave 3.1 4.6.1 Notes 4.5.7 Uniqueness in an Unbounded Region 4.4 Branch Cuts References 4 4.2 Edge Conditions 4.2 Reﬂection and Refraction 3.1.2 Buried Harmonic Line of Compression I 5.8 Scattering from an Elastic Inclusion in a Fluid References 5 Radiation and Diffraction 5.4.2 Transient Wave 3.7.4.3 An Inner Expansion 4.1 Notes 4.1 The Transforms 5.1.2 Reﬂection Coefﬁcients 3.2 Inversion 5.1.6 Integral Representation: A Scattering Problem 4.4 Principle of Limiting Absorption 4.4 The Rayleigh Wave 3.1 Phase Matching 3.7. and Interfacial Waves 3.3 Critical Refraction and Interfacial Waves 3. Refraction.3.7.4.
1 Harmonic Waves in a Closed Waveguide 6.4.4 Buried Harmonic Line of Compression II 5.3 Leaky Waves 6.3 Description of the Scattered Waveﬁeld 5.1 Harmonic Excitation 6.2 Dispersion Relation: A Closed Waveguide 6.6.1 Formulation 5.2 The Scattered Compressional Wave 5.2 The Propagation of Information 6.3 The Propagation of Angular Frequencies References Index xi 87 90 94 96 96 98 99 101 102 104 108 112 116 119 121 121 123 124 128 129 131 134 134 135 139 140 145 146 146 150 150 151 153 157 159 .5 A Laterally Inhomogeneous.1.3 Stationary Phase Approximation 5.2 Dispersion Relation: An Open Waveguide 6.5.2 Wiener–Hopf Solution 5.2 The Waveﬁeld in the HalfSpace 6.1 Causes of Dispersion 6.4.2 Method of Steepest Descents 5.1 Partial Waves and the Transverse Resonance Principle 6.5 Diffraction of an Antiplane Shear Wave at an Edge 5.1.1 The Complex Plane 5.5.3.3.1 The Waveﬁeld in the Layer 6.2.4.Contents 5.2 Harmonic Waves in an Open Waveguide 6.3 The Scattered Shear Wave 5.1 Watson’s Lemma 5.3.1 Partial Wave Analysis 6.3.3 Excitation of a Closed Waveguide 6.4.4.6.6 Matched Asymptotic Expansion Study Appendix: The Fresnel Integral References 6 Guided Waves and Dispersion 6.4 Harmonically Excited Waves in an Open Waveguide 6. Closed Waveguide 6.2.3.6 Dispersion and Group Velocity 6.5.2 Transient Excitation 6.4.6.
.
Preface
A wave builds up perhaps it says its name, I don’t understand, mutters, humps its load of movement and foam and withdraws. Who can I ask what it said to me?1
Wave propagation and scattering are among the most fundamental processes that we use to comprehend the world around us. While these processes are often very complex, one way to begin to understand them is to study linear wave propagation. This is a book describing such propagation. I use the equations of linear elasticity to form a context for my description of wave propagation. However, the reader’s knowledge of elasticity need not be very great, and experience with a related ﬁeld theory, such as ﬂuid mechanics or electromagnetic theory, is sufﬁcient to understand what is written here. In many places I treat only the antiplane shear problem because I do not believe that the extra work needed to do the analogous inplane problem adds anything of signiﬁcance to understanding the underlying wave processes. Nevertheless, where an inplane elastic problem introduces a unique feature, such as the presence of a nondispersive surface wave, that problem is treated. This is also a book describing the parts of applied mathematics that describe the propagation and scattering of linear elastic waves. It assumes that the reader has a good background in calculus, differential equations, and complex analysis. By this I mean that the reader should have studied most of the topics in
1
Neruda, Pablo, Soliloquy in the Waves, pp. 185–186. In Isla Negra, a Notebook, translated by A. Reid. New York: Noonday Press. 1981.
xiii
xiv
Preface
Courant and John, Introduction to Calculus and Analysis, Vols. 1 and 2 (1989) and in Boyce and DiPrima, Elementary Differential Equations and Boundary Value Problems (1992). Moreover, the reader should be able to look things up in Carrier, Krook, and Pearson, Functions of a Complex Variable (1983) or Ablowitz and Fokas Complex Variables, Introduction and Applications (1997) and Zauderer Partial Differential Equations of Applied Mathematics (1998) and not feel lost. None of the mathematical analyses exceed, in sophistication or difﬁculty, those found in Courant and John (1989). I work almost entirely in Cartesian coordinates so that no knowledge of special functions or the transforms associated with them is needed. Some previous experience with asymptotic analysis would be helpful, but is not essential. All the asymptotic analysis needed is explained in the book. I have used two unifying themes throughout. The ﬁrst is that an understanding of planewave interactions is fundamental to understanding more complex wave interactions. The second is that waves are best understood in an asymptotic approximation where they are free of the complications of their excitation and are governed primarily by their propagation environment. Therefore planewave spectral analyses and asymptotic approximations are the main techniques used to study the more complicated problems. The selection of problems for the reader is small and directed at engaging him or her in the development of the subject. The problems are an integral part of the book and most should be attempted. I have tried to avoid a menagerie of symbols. In general I use Cartesian tensors such as τi j , where the indices i, j = 1, 2, 3, or a boldface notation τ . The symbol ∂i is used to represent the partial derivative with respect to the ith coordinate. Similarly, sometimes I use dx f to represent d f /d x. Repeated indices are summed over 1, 2, 3 unless otherwise indicated. For problems engaging only two coordinates, subscripts using Greek letters such as α, β = 1, 2 are used so that a vector component would be written as u β and a partial derivative as ∂α . When these subscripts are repeated they are summed over 1, 2. At times I use symbols such as c L or cT when there is need to distinguish between parameters that relate to compressional or shear disturbances, but when that distinction is not important I drop the subscript. Constants such as A are used over and over again and have no special meaning. Professor Jan Achenbach was my research advisor. He taught me the subject. His inﬂuence is everywhere is this book. I thank him. I have been supported over the years primarily by the National Science Foundation, though recently also by the Air Force Ofﬁce of Scientiﬁc Research. I am very grateful for their support. Elaine Wilson typed the early parts of this manuscript and Mike
Preface
xv
Greenberg provided the illustrations. Don Carlson and Bill Phillips tolerated my complaints with humor; Eduardo Velasco prevented me from publishing a discussion of group velocity that was at best muddled. To everyone, thank you. The book undoubtedly contains errors, for which I alone am responsible. I anticipate that few if any are serious.
Books Cited
Ablowitz, M. J. and Fokas, A. S. 1997. Complex Variables, Introduction and Applications. New York: Cambridge. Boyce, W. E. and DiPrima, R. C. 1992. Elementary Differential Equations and Boundary Value Problems, 5th ed. New York: Wiley. Carrier, G. F., Krook, M., and Pearson, C. E. 1983. Functions of a Complex Variable. Ithaca, NY: Hod Books. Courant, R. and John, F. 1989. Introduction to Calculus and Analysis, Vols. 1 and 2. New York: Springer. Zauderer, E. 1983. Partial Differential Equations of Applied Mathematics. New York: WileyInterscience. Evanston and Urbana, Summer 2000.
.
The conservation of mechanical energy follows from (1) and (2). τi j = τ ji . where τ ji is the stress tensor. The Poisson summation formula is also introduced and used to distinguish between a propagating wave and a vibration of a bounded body. In order to introduce this aspect of wave propagation. The most important feature of the model is that the force exerted across a surface. and (2) a constitutive relation relating force and deformation. without derivation. A linear wave carries information at a particular velocity. The conservation of angular momentum makes the stress tensor symmetric. Only the basic equations are summarized. It is this transmitting of information that gives linear waves their special importance. both the model equations and some of the mathematical transformations. oriented by the unit normal n j . that is. In the linear approximation density ρ is constant.1 Model Equations The equations of linear elasticity consist of (1) the conservation of linear and angular momentum. Both Fourier and Laplace transforms and their inverses are introduced and important sign conventions settled. by one part of a material on the other is given by the traction ti = τ ji n j . the group velocity. which is characteristic of the propagation structure or environment. The conservation of linear momentum. propagation in onedimensional periodic structures is discussed. 1.1 Simple Wave Solutions Synopsis Chapter 1 provides the background. in 1 . Such structures are dispersive and therefore transmit information at a speed different from the wavespeed of their individual components. needed to understand linear elastic waves.
linearly elastic solid.2) in (1. the equation can also be written in the form (λ + 2µ)∇(∇ · u) − µ∇ ∧ ∇ ∧ u + ρ f = ρ∂t ∂t u. (1.1) = (∂i u j + ∂ j u i )/2. However.2) where u i is ith component of particle displacement. One very common boundary condition is to ask that t = 0 everywhere on ∂R. The term f is a force per unit mass. ij (1. followed e by substituting the outcome into (1. then commonly t and u are prescribed on ∂R. This models the case in which a solid surface is adjacent to a gas of such small density and compressibility that it is almost a vacuum. When t is given over part of ∂R and u over another part.5) (1.6) This last equation indicates that elastic waves have both dilitational ∇ · u and rotational ∇ ∧ u deformations. (1. . Substituting (1.2 1 Simple Wave Solutions differential form. namely (λ + µ)∂i ∂k u k + µ∂ j ∂ j u i + ρ f i = ρ∂t ∂t u i . When R is inﬁnite in one or more dimensions. gives one form of the equation of motion.3) where λ and µ are Lam´ ’s elastic constants. (1. If ∂R is the boundary of a region R occupied by a solid. the underlying dependence of the deformation is upon the ∇u. the equation becomes (λ + µ)∇(∇ · u) + µ∇ 2 u + ρ f = ρ∂t ∂t u. isotropic. For a homogeneous. stress and strain are related by τi j = λ kk δi j + 2µ ij. the boundary conditions are said to be mixed. is expressed as ∂k τki + ρ f i = ρ∂t ∂t u i . Written in vector notation. The symmetrical deﬁnition of the deformation ensures that no rigidbody rotations are included.4) When the identity ∇ 2 u = ∇(∇ · u) − ∇ ∧ ∇ ∧ u is used.1). Deformation is described by using a strain tensor.3). (1. special conditions are imposed such that a disturbance decays to zero at inﬁnity or radiates outward toward inﬁnity from any sources contained within R.
Now (1. A (linear) wave incident on such a boundary would experience continuity of traction and displacement when the solids press together. Now (1.1.1) becomes (λ + 2µ)∂1 ∂1 u 1 + ρ f 1 = ρ∂t ∂t u 1 .1 OneDimensional Models We assume that the various waveﬁeld quantities depend only on the variables x1 and t. and (1.8) τ22 = τ33 = λ∂1 u 1 . The reader may consult Hudson (1980) for a succinct discussion of linear elasticity or Atkin and Fox (1980) for a somewhat more general view. while the normal component of traction and the normal component of displacement are made continuous. To brieﬂy indicate some of the possible complications. 1 and 2. One other simple continuity condition that commonly arises is that between a solid and an ideal ﬂuid. but in many cases it is usual to assume that the traction and displacement.3) becomes τ11 = (λ + 2µ)∂1 u 1 . 1. all the stress components except τ11 are assumed to be zero. but would experience a tractionfree boundary condition when they pull apart (Comninou and Dundurs. consider two solid bodies pressed together.3) becomes τ11 = E∂1 u 1 . while u 2 and u 3 are assumed to be zero. are continuous. the tangential component of t is set to zero. 1977). Contact between solid bodies is quite complicated.10) E =µ 3λ + 2µ .1. Because the viscosity is ignored. These are only models and are often inadequate.1) becomes E∂1 ∂1 u 1 + ρ f 1 = ρ∂t ∂t u 1 .2) combined with (1. For longitudinal strain. (1. (1. so that (1.11) ν= λ . t and u. (1.1 Model Equations 3 Another common situation is that in which ∂R12 is the boundary between two regions. 2(λ + µ) (1. This models welded contact.9) .7) For longitudinal stress. occupied by solids having different properties. u 1 is ﬁnite. and ∂2 u 2 = ∂3 u 3 = −ν∂1 u 1 . λ+µ (1. This produces a complex nonlinear interaction.
1) breaks into two separate equations.12) (1. τ3β = µ∂β u 3 . . though they have somewhat different physical meanings. The longitudinal stress model is useful for rods having a small cross section and a tractionfree surface. similar to (1. compressional and shear.15) (1.4). (1.11) are essentially the same. (1. The initials P and SV are used to describe the two types of inplane motion.13). and τ33 = λ∂γ u γ . (1. namely ∂β τβ3 + ρ f 3 = ρ∂t ∂t u 3 . 1. from (1. µ∂β ∂β u 3 + ρ f 3 = ρ∂t ∂t u 3 . is called antiplane shear motion. It leads to problems of some complexity.18) (1.4 1 Simple Wave Solutions Note that (1. (λ + µ)∂α ∂β u β + µ∂β ∂β u α + ρ f α = ρ∂t ∂t u α . The equation of motion remains (1.2 TwoDimensional Models Let us assume that the various waveﬁeld quantities are independent of x3 . (1. while the other two displacement components are generally nonzero. The case for which the only nonzero displacement component is u 3 (x1 .14) Note that this is a twodimensional scalar equation. β = 1. x2 .11). The case for which u 3 = 0. namely (1. whose character we explore shortly.8) or (1. respectively. (1. t).3) becomes ταβ = λ∂γ u γ δαβ + µ(∂α u β + ∂β u α ).13) Greek subscripts α. As a consequence. For this case (1.16) This last equation is a vector equation and contains two wave types.17) (1.8) and (1.1. ∂β τβα + ρ f α = ρ∂t ∂t u α . Stress components that vanish at the surface are assumed to be negligible in the interior.12).12). is called inplane motion. 2 are used to indicate that the independent spatial variables are x1 and x2 . that is. giving. namely compressional and shear vertical. or SH motion for shear horizontal.
while the vector potential describes a wave of shear motion. do we know u completely? Yes we do.14). u = ∇ϕ + ∇ ∧ ψ. 1.1. (1. respectively. it is often easier to cast the equations of motion into a simpler form and allow the boundary condition to become more complicated. One way to do this is to use the Helmholtz theorem (Phillips.3 Displacement Potentials When (1. The scalar potential describes a wave of compressional motion. Knowing ϕ and ψ. which in the planewave case is longitudinal. (1. Gregory.1. L (1.6) gives 2 (λ + 2µ)∇ ∇ 2 ϕ − 1/c2 ∂t ∂t ϕ + µ∇ ∧ ∇ 2 ψ − 1/cT ∂t ∂t ψ = 0. allows us enough structure to indicate the complexity found in elasticwave propagation. which in the planewave case is transverse. Assume f = 0.1.4)–(1. such as t = 0. along with related references. is relatively easy to implement. L 2 ∇ 2 ψ = 1/cT ∂t ∂t ψ. (1. (1.6) are used. This law can be derived directly from .20) The equation can be satisﬁed if ∇ 2 ϕ = 1/c2 ∂t ∂t ϕ. in problems in which there are few boundary conditions. ∇ · ψ = 0. However. Again assume f = 0.19) The second condition is needed because u has only three components (the particular condition selected is not the only possibility). 1933. c2 = (λ + 2µ)/ρ.4 Energy Relations The remaining conservation law of importance is the conservation of mechanical energy. Substituting these expressions into (1. L 2 cT = µ/ρ. are given in Achenbach (1973). allows us to solve complicated problems in detail without being overwhelmed by the size and length of the calculations needed. and shear or transverse wavespeed. while the vector model.22) The terms c L and cT are the compressional or longitudinal wavespeed. that is. 1.1 Model Equations 5 These twodimensional equations are the principal models used. 1996) to express the particle displacement u as the sum of a scalar ϕ and a vector potential ψ. The scalar model. Proofs of completeness.18).21) (1. a boundary condition.
and making use of the decomposition ∂ j u i = ji + ω ji . One important reason for the usefulness of this representation is that the transmitter. (1.6 1 Simple Wave Solutions (1.2 The Fourier and Laplace Transforms All waves are transient in time. 2 U = 1 τi j 2 ij. One useful representation of a transient waveform is its Fourier one. (1. where F is given by F j = −τ ji ∂t u i . To better understand that the energy ﬂux or power density is given by (1. receiver.26). t) d V = − ˆ F · n d S. This representation imagines the transient signal decomposed into an inﬁnite number of timeharmonic or frequency components.23) Forming the product τkl kl . F. The linearity of the problem ensures that we can work out the net propagation outcomes for each frequency component and then combine the outcomes to recreate the received signal.3) by taking the dot product of ∂t u with (1. where ω ji = (∂ j u i − ∂i u j )/2. (1. 1. initially. ∂ j τ ji ∂t u i − ρ(∂t ∂t u i )∂t u i = 0.23) as 1 ∂ (ρ∂t u i ∂t u i 2 t + τki ki ) + ∂k (−τki ∂t u i ) = 0.3). The remaining term is the divergence of the energy ﬂux. it radiates outward ˆ across the surface ∂R at a rate F · n.25) These are the kinetic and internal energy density. allows us to write (1.1).26) where E = K + U and is the energy density. as the mechanical energy decreases within R.28) R ∂R Therefore. (1.1)–(1. Then (1. with surface ∂R.27) (1. . (1. using (1.24) can be written as ∂E/∂t + ∇ · F = 0.24) The ﬁrst two terms become the time rates of change of K = 1 ρ∂t u k ∂t u k . This is the differential statement of the conservation of mechanical energy. This gives.27) over R and using Gauss’ theorem gives d dt E(x. and the propagation structure usually respond differently to the different frequency components. respectively. consider an arbitrary region R. Integrating (1.
This is not essential and its deﬁnition can be extended to include functions whose tdependence extends to negative values. p is a complex variable and its domain is such as to make u(x.1 The inverse transform is deﬁned as u(x. . but the function can be analytically continued beyond this. ω). t) is zero. for the present. when t ∈ [0. (1. t) dt. are discussed in Titchmarsh (1939) in a general way and in more detail by Noble (1988). It is usually used for initialvalue problems so that we imagine that for t < 0. p) dp. Note that p = −iω so that. (1. t) = 1 2πi +i∞ −i∞ ¯ e pt u(x. Moreover.31) ¯ As with ω. The inverse transform is given by u(x. The expressions for the inverse transforms. Note that there is a speciﬁc sign convention for the exponential term that we shall adhere to throughout the book. t) = 1 2π ∞ −∞ ¯ e−iωt u(x.32). p) an analytic function of p. and once known. u is an analytic function within the domain of convergence. including the case in which the contour extends to inﬁnity.30) ¯ Thus we have represented u as a sum of harmonic waves e−iωt u(x. u(x. can be analytically continued beyond it. In practice we deﬁne the inverse transforms on contours that are designed to capture the physical features of the solution. ω) = ∞ −∞ 7 eiωt u(x. (1. are misleading. ω) dω. ∞). (1. 1 Analytic functions deﬁned by contour integrals. t) dt. Its domain is such as to make the above integral ¯ convergent.32) where ≥ 0. p) = 0 ∞ e− pt u(x.29) The variable ω is complex. ω). A large part of this book will deal with just how those contours are selected.30) and (1. (ω) > 0 gives the initial domain of analyticity for ¯ u(x. But. The domain is initially deﬁned as ( p) > 0. A closely related transform is the Laplace one.2 The Fourier and Laplace Transforms The Fourier transform is deﬁned as ¯ u(x. (1.1. This transform is deﬁned as ¯ u(x. we shall work with these deﬁnitions.
8
1 Simple Wave Solutions
Consider the case of longitudinal strain. Imagine that at t = −∞ a disturbance started with zero amplitude. Taking the Fourier transform of (1.8) gives
2 ¯ ¯ d 2 u 1 /d x1 + k 2 u 1 = 0,
(1.33)
where k = ω/c L and c L is the compressional wavespeed deﬁned in (1.21). The parameter k is called the wavenumber. Here (1.33) has solutions of the form ¯ u 1 (x1 , ω) = A(ω)e±ikx1 . If we had sought a timeharmonic solution of the form ¯ u 1 (x1 , t) = u 1 (x1 , ω)e−iωt , (1.35) (1.34)
we should have gotten the same answer except that e−iωt would be present. In other words, taking the Fourier transform of an equation over time or seeking solutions that are time harmonic are two slightly different ways of doing the same operation. For (1.35), it is understood that the real part can always be taken to obtain a real disturbance. Much the same happens in using (1.30). In writing (1.30) ¯ we implicitly assumed that u(x, t) was real. That being the case, u(x, ω) = ¯ u ∗ (x, −ω), where the superscript asterisk to the right of the symbol indicates the complex conjugate. From this it follows that u(x, t) = 1 π
∞ 0
¯ e−iωt u(x, ω) dω.
(1.36)
The advantage of this formulation of the inverse transform is that we may proceed with all our calculations by using an implied e−iωt and assuming ω is positive. The importance of this will become apparent in subsequent chapters. Now (1.36) can be regarded as a generalization of the taking of the real part of a timeharmonic wave (1.35). Problem 1.1 Transform Properties ¯ ¯ Check that u(x, ω) = u ∗ (x, −ω) and derive (1.36) from (1.30). The reader may want to consult a book on the Fourier integral such as that by Papoulis (1962). When the plus sign is taken, (1.35) is a timeharmonic, plane wave propagating in the positive x1 direction. We assume that the wavenumber k is positive, unless otherwise stated. The wave propagates in the positive x1 direction because the term (kx1 − ωt) remains constant, and hence u 1 remains constant,
1.2 The Fourier and Laplace Transforms
9
only if x1 increases as t increases. The speed with which the wave propagates is c L . The term ω is the angular frequency or 2π f , where f is the frequency. That is, at a ﬁxed position, 1/ f is the length of a temporal oscillation. Similarly, k, the wavenumber, is 2π/λ, where λ, the wavelength, is the length of a spatial oscillation. Note that if we combine two of these waves, labeled u + and u − , each going 1 1 in opposite directions, namely u + = Aei(kx1 −ωt) , 1 we get u 1 = Ae−iωt 2 cos(kx1 ). Taking the real part gives u 1 = 2A cos(ωt + α) cos(kx1 ). (1.39) (1.38) u − = Ae−i(kx1 +ωt) , 1 (1.37)
We have taken A as complex so that α is its argument. This disturbance does not propagate. At a ﬁxed x1 the disturbance simply oscillates in time, and at a ﬁxed t it oscillates in x1 . The wave is said to stand or is called a standing wave. Problem 1.2 Fourier Transform Continue with the case of longitudinal strain and consider the following boundary, initialvalue problem. Unlike the previous discussion in which the disturbance began, with zero amplitude, at −∞, here we shall introduce a disturbance that starts up at t = 0+ . Consider an elastic halfspace, occupying x1 ≥ 0, subjected to a nonzero traction at its surface. The problem is one dimensional, and it is invariant in the other two so that (1.8), the equation for longitudinal strain, is the equation of motion. At x1 = 0 we impose the boundary condition τ11 = −P0 e−ηt H (t), where H (t) is the Heaviside step function and P0 is a constant. As x1 → ∞ we impose the condition that any wave propagate outward in the positive x1 direction. Why? Moreover, we ask that, for t < 0, u 1 (x1 , t) = 0 and ∂t u 1 (x1 , t) = 0. Note that, using integration by parts, the Fourier transform, indicated by F, of the second time derivative is ¯ F [∂t ∂t u 1 ] = −ω2 u 1 (x1 , ω) + iωu 1 (x1 , 0− ) − ∂t u 1 (x1 , 0− ). (1.40)
In deriving this expression we have integrated from t = 0− to ∞ so as to include any discontinuous behavior at t = 0. Taking the Fourier transform of (1.8) gives
10
1 Simple Wave Solutions
(1.33). Show that the inverse transform of the stress component τ11 is given by τ11 (x1 , t) = P0 2πi
∞ −∞
ei(kx1 −ωt) dω. ω + iη
(1.41)
In the course of making this step you will need to chose between the solutions to the transformed equation, (1.33). Why is the solution leading to (1.41) selected? Note that, if the disturbance is to decay with time, η must be positive. Next show that τ11 (x1 , t) = −P0 H (t − x1 /c L )e−η(t−x1 /cL ) . (1.42)
Explain how the conditions for convergence of the integral, as its contour is closed at inﬁnity, give rise to the Heaviside function. Note how the sign conventions for the transform pair, by affecting where the inverse transform converges, give a solution that is causal. Problem 1.3 Laplace Transform Solve Problem 1.2 by using the Laplace transform over time. Why select the solution e− px1 /cL ? How does this relate to the demand that waves be outgoing at ∞? The solution of Problem 1.2 suggests how we shall deﬁne the Fourier transform over the spatial variable x. Suppose we have taken the temporal transform ¯ obtaining u(x, ω). Then its Fourier transform over x is deﬁned as
∗
¯ u(k, ω) =
∞ −∞
e−ikx u(x, t) d x,
(1.43)
and its inverse transform is ¯ u(x, ω) = 1 2π
∞ −∞
¯ eikx ∗ u(k, ω) dk .
(1.44)
Again note the sign conventions for the transform pair. This will remain the convention throughout the book. Moreover, note that u(x, t) = 1 4π 2
∞ −∞ ∞ −∞
¯ ei(kx−ωt)∗ u(k, ω) dω dk .
(1.45)
This shows that quite arbitrary disturbances can be decomposed into a sum of timeharmonic, plane waves and thereby indicates that the study of such waves is very central to the understanding of linear waves.
The lefthand side of (1. ϕ = ∞ ik2π x . Consider a function f (t) and its Fourier transform f¯(ω).48) 1 λ ∞ −∞ e−ik2π (x/λ) f (x) d x. 1]. f λ k =−∞ λ (1.3 A Wave Is Not a Vibration 11 A continuous body vibrates when a system of standing waves is established within it.46) for λ that is small. then knowing the Fourier transform of each term enables us to use an approximation based on the fact that λ−1 is large. . The conclusions are usually valid under more general conditions than those cited. This formula relates a series to one comprising its transformed terms. Proposition 1. A minimum amount of analysis is used.1.46) may converge only slowly for a small λ. The Poisson summation formula states that ∞ m=−∞ f (mλ) = 1 ∞ ¯ 2π k . This formula might better be termed a transform and is quite useful. 2 This proof follows that of de Bruijn (1970). If we want to approximate the lefthand side of (1. Proof. both here and elsewhere. The function ϕ(x) has period 1. Its kth Fourier coefﬁcient equals k=−∞ ck e 1 0 e−ik2π x ϕ(x) d x = 0 1 ∞ m =−∞ e−ik2π x f [(m + x)λ] d x (m+1) m = = 2 ∞ m =−∞ e−ik2π x f (xλ) d x (1.3 A Wave Is Not a Vibration 1.1. Consider the function ϕ(x) given by ϕ(x) = ∞ m =−∞ f [(m + x)λ]. especially for asymptotically approximating sums.47) where the sum converges uniformly on x ∈ [0. We assume that f (t) is such that ϕ(x) has a Fourier series.46) where λ is a parameter. and no attempt at rigorous proofs is made. (1. Restrictions on f (t) are given below. Vibration and wave propagation can be explicitly linked by means of the Poisson summation formula.
The crucial intermediate step is the following. t) = −T δ(t).52) The reader should check that this is the solution.47) converges uniformly. as the sum (1. we get3 τ11 = iTH(t) ∞ n =−∞ e−inπ t sin[nπ(1 − x1 )] . set cb and ρ = 1 (cb = E/ρ). Assume f 1 = 0. (1. cos(nπ) (1. But the individual interactions with the ends have been obscured. where we have taken one of these terms. To ﬁnd these interactions we apply the Poisson summation formula to (1. The governing 2 equation is (1. among others.11). Break up the sin[nπ(1 − x1 )] term into exponential ones and apply (1. τ11 (1. uniformly in Note that the integral m x ∈ [0. One usually considers a solution in this form as a vibration. t) = 0. This is a very useful way to express the answer.50). indicates that the Poisson summation formula holds for a far more general class of functions than assumed here.46) to each term. . and assume that for t < 0. (1. These conditions are more restrictive than needed. t) = ∂t u 1 (x1 . u 1 (x1 .12 1 Simple Wave Solutions (m+t) e−ikx f (xλ)d x → 0 as m → ± ∞. 1]. Consider the impulsive excitation of a rod of ﬁnite length 1.49) By using a Fourier transform over t and solving the boundaryvalue problem in x1 . The boundary conditions are τ11 (0. (1. 1 e−imπ [t−(1−x1 )] = (π t + x1 − 2)−1 cos mπ m =−∞ × =2 The outcome to our calculation is τ11 = T ∞ k=1 3 ∞ ∞ n =−∞ ∞ ∞ e −∞ 2n −iω 1− t+x −2 1 dω δ(t + x1 − 2 − 2n).51) n =−∞ δ(t + x1 − 2k) − T ∞ k=0 δ(t − x1 − 2k).50) Thus the rod is ﬁlled with standing waves. t) = 0. Lighthill (1978). assuming the pulse has reverberated within the rod for a time long with respect to that needed for one echo from the far end to return to x = 0.
The equation of motion (1. It is the outcome of waves reverberating in a bounded body.4 Dispersive Propagation 1. We have thus isolated the pulse returning from its ﬁrst reﬂection at the end x1 = 1. Incident on this inhomogeneity is the waveﬁeld ¯ u i1 (x1 ) = A1 eikx1 . sometimes a long one. the representation captures quite accurately the physical phenomena of a pulse bouncing back and forth in a rod struck impulsively at one end.1 An Isolated Interaction A basic interaction of a wave with its environment is scattering from a discontinuity. k = ω/cb . Moreover. (1.54) Assume there is a region of inhomogeneity within x1 ∈ (−L . Consider timeharmonic disturbances of the form ¯ u 1 = u 1 (x1 )e−iωt . However. It is very useful for times of the order of the echo time.55) . mapping the unending oscillatory motion into a disturbance conﬁned both temporally and spatially. L). 1. In contrast. one usually assumes that at some stage a Fourier synthesis will be carried out. Moreover.4 Dispersive Propagation 13 This is a wave representation.1. is needed for the environment to settle into a steady oscillatory motion. while one works frequently with timeharmonic propagating waves. 2). A period of time. instead of deltafunction pulses. the representation is not very useful for t that is large because it becomes tedious to work out exactly at what reﬂection you are tracking the pulse. experiencing only a ﬁnite number of interactions. A vibration therefore is deﬁned and conﬁned by its environment.4.11) becomes 2 ¯ ¯ d 2 u 1 /d x1 + k 2 u 1 = 0. 2 τ11 = ρcb ∂1 u 1 . For example.53) ¯ We shall not indicate the possible dependence on ω of u 1 unless this is needed. (1. Nevertheless. the representation (1. if t ∈ (1. we had had pulses of sufﬁcient length that they overlapped one another. A2 e −ikx1 x1 < −L . . Continue to consider waves in a rod by using the longitudinal stress approximation. x1 > L . Understanding how an individual wave interacts with its environment and tracking it through each of its interactions constitute the principal problems of wave propagation. then τ11 = T δ[(t − 1) + (x1 − 1)].52) would have been awkward to work with if. a wave is a disturbance that propagates freely outward. (1. returning to its source perhaps only once.
. each separated by a distance L.59) That is. 1. . the scattered waveﬁeld ¯1 u s (x1 ) = B1 e−ikx1 . (1. . We next consider a speciﬁc example. The lefthand ﬁgure shows a single.14 1 Simple Wave Solutions where we have allowed waves to be incident from both directions.56) is excited. The linearity of the problem suggests that we can write the scattered amplitudes in terms of the incident ones as B1 B2 or. more compactly. The righthand ﬁgure shows an endless periodic arrangement of embedded point masses. Note that the scattered waves have been constructed so that they are outgoing from the scatterer. (1. but the acceleration of the mass causes the traction acting on the cross section to be discontinuous. (1. t). . t) + τ11 (0+ . Setting tan θ = k M/2ρ. with the zeroth mass at x 1 = 0. 2. (1. M∂t ∂t u 1 = −τ11 (0− .60) 2 x1 0 L 2L x1 Fig. at x1 = 0. π/2). x1 > L . 1. point mass.1 indicates the geometry. The lefthand ﬁgure in Fig. embedded. t) = u 1 (0+ . we calculate the matrix S to be S= sin θ ei(θ +π/2) cos θ eiθ 0 cos θeiθ sin θ ei(θ +π/2) 1 . One or more point masses M are embedded in a rod of crosssectional area 1. t). (1. The conditions across the discontinuity are u 1 (0− . The masses are labeled 0.1. the rod does not break. 1. as B = SA.57) The matrix S is called a scattering matrix and characterizes the inhomogeneity.58) = S11 S12 S21 S22 A1 A2 . Imagine that the rod has a crosssectional area 1 and that the inhomogeneity is a point mass M. Upon striking the inhomogeneity. . with θ ∈ (0. B2 e ikx1 x1 < −L .
1.4 Dispersive Propagation
15
It is also of interest to relate the wave amplitudes on the right to those on the left. This matrix T, called the transmission matrix, gives R = TL, where L = [A1 , B1 ]T and R = [B2 , A2 ]T . The matrix T is readily calculated from S and is given by T= 1 + i tan θ −i tan θ i tan θ 1 − i tan θ . (1.61)
Note that the amplitudes A1 and B2 are those of rightpropagating waves, while B1 and A2 are those of leftpropagating ones. Problem 1.4 Scattering From a Layer Imagine a layer of width d and properties (ρ, c L ) embedded in a material ¯ ¯ with properties (ρ, c L ). Let the x1 axis be perpendicular to the face of the layer and place the origin at the left face, so that the layer occupies x1 ∈ (0, d). A wave of longitudinal strain Aeikx1 is incident from x1 < 0. Calculate the reﬂection coefﬁcient B/A and the transmission coefﬁcient C/A, where the reﬂected wave is Be−ikx1 and the transmitted one is Ceik(x1 −d) . Treat the waves in the layer as a sum of a single right and left propagating wave, each with different amplitudes, and imagine that the amplitudes contain the effects of all the multiple reﬂections from each side of the layer. Similarly B and C contain the effects of all the reﬂections and transmissions from the layer into the outer material. The wavenumber k = ω/c L . For what values of d is the reﬂection minimized? Can you identify any of the components of the S matrix for the layer in terms of these coefﬁcients?
1.4.2 Periodic Structures One of the more interesting aspects of wavebearing structures is that they often contain several length scales. Propagation in such a structure often can only take place if the angular frequency ω is linked to the wavenumber – a term we must deﬁne a bit more carefully here – in a nonlinear way. To consider this possibility we use the matrix T, (1.61), to analyze propagation in a periodic structure. We imagine an inﬁnite rod, crosssectional area 1, in which equal point masses, M, are periodically embedded. The righthand ﬁgure of Fig. 1.1 indicates the geometry and the how the masses are labeled. One such mass has the nominal position x1 = 0 and is labeled n = 0. Each mass is separated from its neighbors by a distance L. A cell of length L is thereby formed and is labeled n if the nth mass occupies its left end. There are thus two length scales, the wavelength
16
1 Simple Wave Solutions
λ = 2π/k and the cell length L. In this problem we do not concern ourselves with how the waves are excited, but only with the simpler question, What waves does this structure support? Consider the zeroth cell, where x1 ∈ (0, L). Within that cell the solution to (1.54) is ¯ u 1 (x1 ) = R0 eikx1 + L 0 e−ikx1 . (1.62)
At x1 = L − the waveﬁeld is [R0 eik L , L 0 e−ik L ]T . This can be written as L1 = PR0 , with R0 = [R0 , L 0 ]T . The matrix P is called the propagator or the propagation matrix and is given by P= eik L 0 0 . e−ik L (1.63)
At x1 = L + , within the 1th cell, the waveﬁeld amplitudes R1 = [R1 , L 1 ]T are R1 = TPR0 . This relation is readily generalized. If Rn = [Rn , L n ]T , then Rn+1 = TPRn . (1.65) (1.64)
The central feature of the propagation structure is that it has translational symmetry. The central feature of the disturbance we seek is that its phase changes from cell to cell in a way that represents propagation. Speciﬁcally consider propagation to the right. To capture these two features, the waveﬁeld at a point within the (n + 1)th cell can differ from that at a point within the nth cell, where the two points are separated by a distance L, by at most a multiplicative phase factor. This kinematic constraint is expressed by the relation Rn+1 = eiκ L Rn , (1.66)
where κ is unknown. κ is positive, if real, and such as to cause decay, if complex. Combining (1.65) and (1.66) gives a 2 × 2 system of algebraic equations that has a nontrivial solution if and only if det(TP − eiκ L I) = 0. (1.67)
Recalling our previous deﬁnition of tan θ = k M/2ρ, we can write this equation compactly as cos κ L = cos(k L + θ )/cos θ. (1.68)
1.4 Dispersive Propagation
17
This is a nonlinear relationship between the angular frequency ω = cb k and the effective wavenumber κ, though it may not be apparent, as yet, that κ (and not k) is the wavenumber of interest. Note that if κ L ∈ [−π, π ] is a solution, then κ L ± 2nπ, for n = 1, 2, . . . is also a solution. Accordingly, we need only consider κ L ∈ [−π, π]. The term κ L is real provided  cos κ L ≤ 1. Therefore the boundaries between real and complex κ L are given by cos(k L + θ)/ cos θ = ±1. (1.69)
When +1 is taken, the solutions are sin(k L/2) = 0 or tan(k L/2) = − tan(θ). That is, k L = 2nπ or k L + 2θ = 2mπ, where n and m are integers. When −1 is taken, the solutions are cos(k L/2) = 0 or cot(k L/2) = tan(θ). That is, k L = (2n − 1)π or k L + 2θ = (2m − 1)π , where, again, n and m are integers. All these cases are covered by k L = nπ or k L + 2θ = mπ . For k L ∈ [(n − 1)π, (nπ − 2θ)], κ L is real. These intervals are called passbands. Elsewhere κ L is complex, causing the disturbance to decay as it propagates, and the intervals are called stopbands. At the lower boundary of a passband, L is an integer number of halfwavelengths. If T were real, then all the reﬂected waves add constructively and little or nothing is transmitted. The actual situation is complicated by the complex T, but the constructive interference of the reﬂected waves is the basic physical mechanism giving rise to the stopbands. This phenomenon is referred to as Bragg scattering. Consider the interval x1 ∈ (n L , (n + 1)L). Then ¯ u 1 (x1 ) = Rn eik(x1 −n L) + L n e−ik(x1 −n L) = eiκ L Rn−1 eik(x1 −n L) + L n−1 e−ik(x1 −n L) ¯ = eiκ L u 1 (x1 − L) (1.70)
¯ This equation is a restatement of (1.66). Further, it indicates that u 1 (x1 ) must ¯ ¯ satisfy the functional equation u 1 (x1 + L) = eiκ L u 1 (x1 ) if the kinematic constraint is to be enforced. Within each cell there are nominally two waves, as indicated in (1.70), that we call partial waves. However, we seek a solution for the wave globally propagating to the right along the structure, as distinguished from the right and leftpropagating partial waves in each cell. With this in mind, the solution to the functional equation is ¯ u 1 (x1 ) = eiκ x1 ϕ(x1 ), (1.71)
1978. Cambridge: University Press. Comninou. Soc. we see that u 1 (x1 . Proc. R. u 1 (x1 ) becomes ¯ u 1 (x1 ) = ∞ −∞ cn ei x1 (κ−2π n/L) .36). Mech. Wave Propagation in Elastic Solids. If we excited the structure with a pulse. J. ω) propagates at a different speed for each ω. M. de Bruijn.D.. 1973. A. There are many fascinating aspects to propagation in periodic structures. J. N. 1956. Reﬂexion and refraction of elastic waves in the presence of separation. Amsterdam: NorthHolland.G.68) indicates that ω is a function of κ. 52–56. That is. Lond. through the term ei(κ x1 −ωt) . (1. B. Hence what we have inferred is that dispersion can cause the distortion of or loss of information from a signal. Math. Writing κ ¯ as ω/c(ω). N.. Therefore. The Excitation and Propagation of Elastic Waves. Equation (1. 1980. and a reader seeking to learn more may wish to read this work further. 1980. Appl. H. or κ a function of ω.J. 273–308. whose coefﬁcients are cn . 356: 509–528. R. R. Amsterdam: NorthHolland. 1977. The discussion here has followed parts of Levine (1978). A relation such as this is called a dispersion relation.72) ¯ The timeharmonic waveﬁeld u 1 (x1 )e−iωt is thus a consequence of an inﬁnite number of space harmonics. pp. Principles and Techniques of Applied Mathematics. London: Longman. . whereas a sinusoid is not. We shall explore this topic further in Chapter 6. ϕ(x1 ) is a periodic function and can be ¯ represented by a Fourier series. 3rd ed. 1956). 4 This is a partial statement of Floquet’s theorem (Friedman. that is the wavenumber.D. References Achenbach. as indicated by (1. Gregory. J. 1996. J. pp. Asymptotic Methods in Analysis. An Introduction to the Theory of Elasticity. Quart. Each component would then propagate at its own speed and the pulse would become dispersed. and Dundurs. New York: Wiley. Levine. then the pulse would be composed of an inﬁnite number of such components. Friedman. it is clear that it is κ. Amsterdam: NorthHolland. 65–67. 1970. Unidirectional Wave Motions.18 1 Simple Wave Solutions where ϕ(x1 + L) = ϕ(x1 )4 . Hudson. and elsewhere. Helmholtz’s theorem when the domain is infinite and when the ﬁeld has singular points. 49:439–450. Atkin. 339–345. Note that shifting κ L by ±2mπ would not change this expression.A. A pulse is information. pp. and Fox. More importantly.
B. pp. The Fourier Integral and its Applications. 67–71. A. New York: Chelsea.B. New York: Wiley.. 1962. E. 182–196. Titchmarsh. Papoulis.C. 1933. 1939. pp. Noble. M. . 1988. Phillips. Vector Analysis. Fourier Analysis and Generalized Functions. pp. pp. 11–27. Methods Based on the WienerHopf Technique. 1978.J. Cambridge: University Press. The Theory of Functions.References 19 Lighthill. H. Oxford: Clarendon Press. New York: McGrawHill. 2nd ed. 85–86.
2.2 Kinematical Descriptions of Waves Synopsis A wave equation is an equation of motion. The time is t and the position vector is x. The present chapter describes the kinematics of waves in a propagation environment without boundaries. The wave’s name then arises from this fact and the plane is called a ˆ ˆ wave front. The normal to the plane is s. Descriptions of a plane wave are emphasized because they form a canonical kinematical object from which more complicated solutions can be constructed. 20 . u retains a constant value over each plane deﬁned by t provided this plane advances along the normal s as t increases. is a parametric representation of a family of planes. However. it is reasonable to label this as such. As t is incremented. where C is a constant. The equation s · x = t − C.1) ˆ where d is a constant unit vector1 called the polarization and s is a constant vector called the slowness. t) = d u(t − s · x). it is then 1 The circumﬂex indicates a unit vector. Setting s = s p so that p is the unit normal to each plane. Accordingly. where each value of parameter t gives a different member of the family. (2. then x must be incremented along s if the argument t − s · x is to equal the constant C. when one writes little about how a waveﬁeld is excited and concentrates primarily upon the geometrical description of a waveﬁeld’s propagation. so that a study of its solutions is perhaps more than just the study of the kinematics. Figure 2.1 sketches this relationship.1 TimeDependent Plane Waves A timedependent plane wave is one whose form is ˆ u(x.
2. ˆ ˆ p · d = 0. (2. or ˆ ˆ u = du(t − p · x/cT ). Substituting (2. (2. L or 2 s 2 = 1/cT s ∧ u = 0. (2.2) (2. s(s · A) − s ∧ s ∧ A = s 2 A.1) into the equation of motion. ˆ so that s −1 is the speed of advance of the plane along p. (2. the slowness vector s is normal to the plane and hence gives its orientation. (2. The position vector x identiﬁes a point on the plane.6).7) ˆ ˆ p ∧ d = 0.3) Because.6) . (1. for any vector A.2. A sketch of the plane wavefront s · x = t − C. where t is a parameter identifying a particular plane. straightforward to ﬁnd that ˆ p · d x/dt = 1/s.1.5) Thus either ˆ ˆ u = du(t − p · x/c L ).3) implies that either s 2 = 1/c2 .1 TimeDependent Plane Waves x3 21 x s x2 x1 Fig. gives 2 c2 s(s · ∂t ∂t u) − cT s ∧ s ∧ ∂t ∂t u = ∂t ∂t u.4) s · u = 0. L (2.
timeharmonic wave. Because p · p = 1. multiply by e−iωt and take the real part of the expression. A timeharmonic plane wave has the form2 ˆ ˆ u = Adeik p·x .10) where the amplitude A = Aeiθ and the wavenumber k = ω/c. For these two plane waves the velocity of energy transport.25). For the present. and U. we may take p = pr + i pi . 2.26). and F I and E I are the corresponding ﬂux and energy density. C. .2 TimeHarmonic Plane Waves In one sense. (2.9) where I = L or T . The real and imaginary components are perpendicular to one another.11) In the previous chapter we distinguished a timeharmonic wave from a timedependent one by using an overbar. while the second is a transverse wave with its polarization perpendicular to this direction. pr · pi = 0. This has some interesting implications for the form of the corresponding timedependent plane wave that we shall explore in Section 3. the plane wave is an inhomogeneous or evanescent one. ˆ As indicated previously. The e−iωt is not given. assume that the time dependence is e−iωt or that the waveﬁeld has been transformed over time. for the present. The energy density is E = K + U. where K. (2. are given by (1.7) gives ˆ F I = ρc I (∂t u)2 p. Writing (2. If pi = 0. The combiˆ nation k p = ωs is the wavevector. The overbar is no longer used unless explicitly needed. The ﬂux of energy F is given by (1. the internal energy.3. unless explicitly needed. (2. 2 (2. timeharmonic plane waves are only a particular case of a general plane wave. with pr and pi both ˆ ˆ real. the kinetic energy. Calculating F for (2. while if pi = 0. it is homogeneous. To recover a real. and ω is assumed real and positive.8) where I = L or T and F I is the corresponding ﬂux. we have ˆ the additional ﬂexibility of allowing p to be complex.10) in detail gives ˆ u = Ade−kx· pi eikx· pr .6) or (2. but because we are working with the temporal transform. is ˆ C I = F I /E I = c I p.22 2 Kinematical Descriptions of Waves The ﬁrst is a longitudinal wave with its polarization parallel to its direction of propagation.
just as it did in (2.6).  p 1  > 1. (2. t (2. gives.15) ˆ ˆ ˆ Here α = −i p 2 . This average is deﬁned as G = 1 T t+T G(τ ) dτ. and I = L or T . The instantaneous value is of less interest than the average value over a period T (= 2π/ω). ˆ Then p 2 may be real or imaginary. That is. That such waves arise in practice will soon become apparent.3). either p ∧ d = 0 and c = c L . ˆ For the remainder of the section. ˆ .2 TimeHarmonic Plane Waves 23 An inhomogeneous plane wave propagates in the pr direction. . The implications of allowing the vectors p and d to be complex lead to many interesting results. It can be shown by direct calculation that (Aeiη ) (B iη ) = 1 2 (AB ∗ ). L (2. or p · d = 0 and c = cT . Calculating the instantaneous energy ﬂux gives ˆ F I = ρc I ω2 (i Aeiη I ) (i Aeiη I ) p. (1.12) ˆ ˆ ˆ ˆ Thus.13) In the ﬁrst case the wave is homogeneous: ˆ ˆ ˆ u = Adeik(x1 p1 +x2 p2 ) . Note that these ˆ may also be complex.18) where A and B are the complex amplitudes and η has the form given previously.16) ˆ where η I = k I (x · p − c I t).17) where G(t) is a real periodic function of t with period T . we shall assume that p is real. (2. ˆ p2 = ˆ1 1 − p2 ˆ1 i p2 − 1 1/2 . but decays in the pi direction. (2. ˆ  p 1  < 1.2. many of which are discussed by Boulanger and Hayes (1993). 2 ˆ ˆ ˆ ˆ ˆ ˆ ˆ c2 /c2 p( p · d) − cT /c2 p ∧ p ∧ d = d. 1/2 (2. (2. Substituting (2. statements imply that d ˆ ˆ ˆ ˆ The symbol pi is the ith component of p.10) into the equation of motion. Assume that p 3 = 0 and p 1 is real.14) In the second it is inhomogeneous: ˆ ˆ ˆ u = Ade−k αx2 eikx1 p1 .
The advantage of these representations is that. This last construction is taken up in Problem 4.15) gives. However. cylindrical wave. in a region where there are no sources of excitation. linearity allows us to use it to construct the outcome of an interaction with the same obstacle by a more general waveﬁeld. (2.1 Carry out the calculation leading to (2. We shall continue to use u 3 for the particle displacement. 2. The clarity of the planewave spectral technique is well demonstrated by Clemmow (1966) through the description and solution of several electromagnetic wave propagation and scattering problems.20) We consider a signaling problem. upon constructing the outcome of an interaction of a plane wave with an obstacle.24 2 Kinematical Descriptions of Waves Problem 2.18) to (2. because it is more easily motivated by starting from a radiation problem. 2 Problem 2.1. Applying (2.1 A Gaussian Beam Consider an antiplane shear wave whose equation of motion is (1. but write c rather than the awkward cT . (1. A summary of this approach is given in an Appendix.3 PlaneWave or AngularSpectrum Representations Spherical and cylindrical waves can be constructed directly from solutions to the equations of motion in the corresponding coordinate systems.15). these waves can also be constructed from collections of homogeneous and inhomogeneous plane waves.19) ˆ Show that F I = c I E I p and F I and E I are the corresponding ﬂux and energy density. 2.18). When a timeharmonic disturbance is assumed. ∂α ∂α u 3 + k 2 u 3 = 0. Here.16) gives ˆ F I = 1 ρc I ω2 A A∗ p. These representations are called planewave or angularspectrum representations.3. Gaussian beam and a compressional spherical wave. and give the result for an antiplane shear.2 (2. This is a problem wherein the source is given as a function of time over an initial plane – in the present case it is given as a . we construct an antiplane shear.
20) and hence begun the discussion at this point. To ﬁnd U .24) Note that we could simply have asserted that this was a solution to (2. to inﬁnity. (2. .21) where A is a complex amplitude and b is a parameter that measures the initial width of the waveﬁeld. ∞). The solution to (2. While it will not be clear from what we do here.26) We have succeeded in taking a rather general waveﬁeld and expressing it as an integral over a set of plane waves. At x1 = 0. the integral is one over both homogeneous and inhomogeneous. Note that k1 takes both real and imaginary values as k2 ranges over (−∞. hence the name planewave representation and the descriptive term planewave spectra. As x1 → ∞. 2 (2. plane waves. 2 2 1/2 Abe−(k2 b/2) . this waveﬁeld will remain Gaussian in cross section but spreads as it propagates outward. 2 (2.2. x2 ) = bA 2π 1/2 ∞ −∞ ∞ −∞ e−(x2 /b) e−ik2 x2 d x2 . we use (2. x2 ) = Ae−(x2 /b) . (2.21) to write U (k2 ) = A =π Therefore.4. u 3 (0. (2. (2.4. but for the present we do not need to know in detail where the branch cuts are placed.3 PlaneWave or AngularSpectrum Representations 25 timeharmonic function along the x2 axis – and the excited waveﬁeld is allowed to propagate outward.25) e−(k2 b/2) ei(k1 x1 +k2 x2 ) dk2 .22) that satisﬁes the outgoing condition is ∗ u 3 = U (k2 )eik1 x1 . we ask that the wave be outgoing (there are no sources at inﬁnity). (k1 ) ≥ 0. u 3 (x1 . the equation of motion becomes 2 2 d 2∗ u 3 /d x1 + k1 ∗ u 3 = 0. usually in only one direction. This will be discussed in Section 3.22) where 2 k1 = k 2 − k2 1/2 . time harmonic.23) Deﬁning the branches of the radical k1 is very important. When the spatial Fourier transform over x2 is used. That is. Therefore u 3 (x1 . (k1 ) ≥ 0. x2 ) = 1 2π ∞ −∞ U (k2 )ei(k1 x1 +k2 x2 ) dk2 .
26 2 Kinematical Descriptions of Waves 2. Note that we have moved directly to the step represented previously by (2. ϕ = A(eikρ /kρ). we pose a sig2 2 naling problem.27) is ϕ = A(eikr /kr ). Note 2. For the moment consider x3 ≥ 0. (2. Again we assume that the disturbance is time harmonic. The potential ϕ can be represented as ϕ= with 2 2 k3 = k 2 − k1 − k2 1/2 1 (2π)2 ∞ −∞ ∞ −∞ ∗ ϕ(k1 . Rather than consider the wave excited by a compact source. In Section 4. (2.31) This is a solution to (2.3. Clearly its wavefront is spherical.24).21) becomes ∂ϕ 1 ∂ r2 2 ∂r r ∂r + k 2 ϕ = 0. For r = 0 a solution of (2. The term k3 has been deﬁned so that the waveﬁeld is outgoing. k2 )ei(k1 x1 +k2 x2 +k3 x3 ) dk1 dk2 . we suggest a possible source for such a spherical.27). (1. The energy ﬂux is proportional to r −2 balancing the increase in surface area of the spherical wavefront as the wave propagates outward. at x3 = 0. both in the physical and in the transform .1. gives ∗ ϕ(k1 . The 2 2 2 only spatial dependence is upon r = (x1 + x2 + x3 )1/2 .30) . compressional wave.3. The equation governing ϕ is (1. (2. k2 ) = A k ∞ −∞ ∞ −∞ eikρ −i(k1 x1 +k2 x2 ) d x1 d x2 .2 An AngularSpectrum Representation of a Spherical Wave Consider a spherically symmetric compressional wave such that the particle displacement u is described by ∇ϕ. (k3 ) ≥ 0. Applying the condition at x3 = 0. Set ρ = (x1 + x2 )1/2 . while. (k3 ) ≥ 0. (2.27) where we have written c to replace c L and k = ω/c to replace k L .32) Integrals of this form are most readily evaluated by transforming the coordinates of the integrand to polar ones. (2. (2.21).27).29). For r > 0. The potential ϕ satisﬁes (2.29) As x3 → ±∞ we ask that the wave be outgoing.28) where A is an amplitude that may be complex. e ρ (2.
where k is the complex wavevector and x the position vector. why we can assume that eikρ → 0 as ρ → ∞.3 PlaneWave or AngularSpectrum Representations space.31) and replacing x3 with its absolute value. (2. x2 = ρ sin θ.4. x1 = ρ cos θ. Note that x3 has been replaced with its absolute value so as to give a spherical wave over all space. The transformations are k1 = κ cos ψ. (2. Had we assumed x3 negative.28). it is possible to imagine that a spherical wave could be constructed from a bursting of wavevectors from some compact source region. 27 (2. (2. then the deﬁnition of the branch of the function k3 . given by ϕ= iA 2πk ∞ −∞ ∞ −∞ ei(k1 x1 +k2 x2 +k3 x3 ) dk1 dk2 . Each plane wave in the integrand has the form eik·x .35) where k3 is given by (2. sometimes called the Sommerfeld transformation. Written this way.33) 2 2 where κ = (k1 + k2 )1/2 and ρ was given previously. k3 (2.3 It is possible to do still more with this representation.31). the spherical wave. (2.32) can now be written as ∗ ϕ= A k 2π ∞ dθ 0 0 eiρ[k−κ cos(ψ−θ )] dρ. Again let us momentarily assume that x3 > 0. 3 k3 = k cos ξ.31). is explained in Section 4.36) This then is the planewave representation of a spherical wave. Therefore. The integral in (2. . (2. It is therefore enough to identify each wavevector with two angles.4.34) Performing the integration gives ∗ ϕ = 2πi A/kk3 . Consider the following transformation. how the branch of a radical such as k3 is determined. The ﬁrst. κ = k sin ξ.37) There are two points in this calculation that have not been explained clearly. k2 = κ sin ψ. taken over both homogeneous and inhomogeneous waves. The second. The outcome of assuming both x3 negative and changing the branch of k3 that is taken is equivalent to retaining the deﬁnition of (2. It serves to indicate again the centrality of plane waves as kinematical objects. with each wavevector piercing a spherical surface.4. would have to be changed.2. is explained in Section 3. can be represented as an integral.
while Achenbach et al. Moreover. Let ξr vary from 0 to π/2 and keep ξi = 0. π).3 An AngularSpectrum Representation of a Cylindrical Wave To achieve a planewave or angularspectrum representation of a cylindrical wave.31).28 2 Kinematical Descriptions of Waves where κ was given previously. as indicated in the Appendix. so that κ varies from 0 to k and (k3 ) ≥ 0. We are integrating in the complex ξ plane so that a strict adherence to this contour is not essential.38) Note that k3 has has been removed by this transformation. Born and Wolf (1986) give a concise introduction to ray theory for electromagnetic waves.4 Asymptotic Ray Expansion A much older way to represent waves is by using rays. The integration with respect to ψ goes from 0 to 2π.39) The contour C begins at −η + i∞ and ends at η − i∞. However. (2. so that κ varies from 0 to ∞.38) is called an angularspectrum representation and succinctly captures the image of a spherical wave as a burst of wavevectors. it is easier to work from the equation of motion with a line force than to try to pose a signaling problem. The variables k1 and k2 both vary from −∞ to +∞. where η ∈ (0. C (2. Let ξ = ξr + iξi . (1982) provide a detailed asymptotic development of . Sommerfeld (1964) describes how to represent this function as an angular spectrum and gives the result (1) H0 (kρ) = 1 π eikρ cos ξ dξ.3. But the integration over ξ is more complicated because ξ must vary over both real and complex values to capture the full range of κ. Note that to fully achieve the spherical wavefront. it must be such that the integral is convergent and represents an outgoing waveﬁeld. H0 (kρ). inhomogeneous waves are also needed. we hold ξr = π/2 and let ξi vary to either +∞ or −∞.36) gives ϕ= iA 2πk 2π π/2−i∞ dψ 0 0 eik·x sin ξ dξ. Further. 2. an outgoing cylindrical wave is described by the (1) Hankel function. The expression (2. This is done in Problem 4. To keep (k3 ) ≥ 0 we must have ξi vary from 0 to −∞. Carefully changing the variables of integration in (2. Then sin(ξr +iξi ) = sin ξr cosh ξi +i cos ξr sinh ξi and cos(ξr +iξi ) = cos ξr cosh ξi − i sin ξr sinh ξi . but the contour must begin at zero and end somewhere near π/2−i∞. with ξr and ξi both real. k3 must satisfy (2. 2. To have κ continue to ∞.1.
should we need to. We are. by examining the representations (2. Looking back at (2. equivalently. 4 The word phase has a rather confused meaning in wave propagation. left examining the equation ∇ 2 ϕ + k 2 ϕ = 0. referring to (2. We expect ray descriptions to be useful only when the wavelengths are small with respect to any other characteristic length in the propagation environment. we are using two ideas. it refers to this latter term. However. to the exponential to aid in our subsequent discussions. that controlled by the wavenumber k. at least. we note that any surface can be approximated by its tangent planes and hence locally any waveﬁeld will have a propagating part or phase4 that behaves as does that of a plane wave. where t is time.38) and (2. that we can synthesize the timedependent case from this construction.41) where α < 1 is a positive number (the term raised to the power α is added for generality). The two approaches are connected by the fact that a discontinuity in a wavefront engages predominantly the highfrequency components of the temporal Fourier transform (Lighthill. (2. (2. Second. the wavelength λ = 2π/k.41).10). We have added the e−iωt . lurking in the background. A wave is permitted to have a discontinuity at its wavefront and the kinematics of the discontinuity is developed. However.4. then.2. with the assumption. is also called the phase. so that k = ω/c. In writing this expression. we should expect that the amplitude of a wave with a curved wavefront depends in some way on k or. As a generalization of a plane wave. Knowing the representation for ϕ. . it is also possible to use the fact that hyperbolic equations permit discontinuities to be transported along their characteristics.1 Compressional Wave We begin by working with the scalar potential ϕ.39). When the word phase is used in this book. k S(x). and in a more general context by Whitham (1974). The one presented next builds upon the timeharmonic plane wave.4 Asymptotic Ray Expansion 29 elasticwave ray theory. 1978). two ways to approach ray descriptions. This is the approach taken by Hudson (1980) for elastic waves. There are. we note that the exponential.40) where we have again written c for c L . we assume that ϕ can be expanded as ϕ(x) ∼ eik[S(x)−ct] (ik)−α−1 ∞ n=0 (ik)−n An (x). This contribution is sometimes called the phase. we note the argument of the amplitude A makes a contribution to the exponential term of a plane wave. it is straightforward to construct that for the vector potential ψ and hence that for the particle displacement u. 2. First.
and that the approximation becomes increasingly accurate as λ → 0 or k → ∞. 5 Holmes (1995) and Hinch (1991) provide useful starting points for the reader to learn more about asymptotic and perturbation methods. we suppress the L unless explicitly needed. Collectively. and that it is k L that is very large. call it L.42) where the An with negative subscripts are zero.45) (2. (2. . or equivalently k −1 .44) These last two equations are called the transport equations. it is reasonable that the amplitude can be expanded as an asymptotic power series5 in the principal parameter λ. 2(∇ S · ∇)An−1 + (∇ 2 S)An−1 = −∇ 2 An−2 . just as we did following (2.43) Note that ∇ S = 1.43)– (2. Substituting (2. We deﬁne a wavefront. At n = 1 we have 2(∇ S · ∇)A0 + (∇ 2 S)A0 = 0. We begin the expansion with k −α−1 so that the approximation for the particle displacement will start as k −α .1).43). (2. (2.40) gives ∞ n=0 (∇ S · ∇ S − 1)An + [2(∇ S · ∇)An−1 + ∇ 2 S An−1 ] + ∇ 2 An−2 (ik)−(n−1) = 0. for A0 = 0. ∇ S · ∇ S = 1. Accordingly.30 2 Kinematical Descriptions of Waves Therefore.41) into (2. We deﬁne a ray as the vector k∇ S. The set {(ik)−n } is linearly independent. for n > 1. In the present case the reader should imagine that k is multiplied by a distance representing that between interactions. The solution of the eikonal equation gives a family of curves along which the amplitudes An are transported. Lastly. (2. and. Rather than introduce an extra parameter. This equation is called the eikonal equation. It is tangent to one of the curves found from solving (2.45) provide a recursive scheme whereby the solution to one equation gives the missing information needed for the solution to the next. however. it is not good workmanship to expand in a parameter with a dimension because the measure of large or small remains too vague.
The rays propagate along and are tangent to x(s. which is nothing more than the directional derivative d S/ds. we imagine that x does not depend on q and suppress any reference to it until we need it. though their asymptotic analysis starts from integral representations. requires some substantial modiﬁcation in these cases.43) is p · ∇ S. Choi and Harris (1989. we see that the lefthand side of (2. By deﬁnition. . Both these equations are assumed known or given. To solve (2. while qualitatively correct for anisotropic or inhomogeneous materials. q) and pierce each wavefront perpendicularly. q2 ) constitute a coordinate system on the initial wavefront and deﬁne the point at which the ray is launched at t = t0 . (2. Recall that (2. ˆ Direct differentiation and using (2. We do not at present know x(s). q).4 Asymptotic Ray Expansion 31 as the family of surfaces S(x) = ct + C. For the moment. where t is a parameter giving each member of the family and C is a constant. 6 Note that this description of the rays. The variable s deﬁnes the arclength along the curve. is locally invertible if the Jacobian does not vanish. We have assumed that the equation for each curve x(s) can be inverted to give s as a function of x.46) ˆ ˆ Setting p = ∇ S. where t0 is an initial time.43). while q indicates each member of the family.47). by the implicit function theorem.41) becomes disordered. Inverting (2. let x(s.2 we have sketched some aspects of the construction. 1990) give an interesting example of caustic formation in elastic materials. Thus S(x) = s(x) + ct0 . The vector x 0 (q) is the starting point for a ray identiﬁed by q on the initial wavefront S(x 0 ) = ct0 .47) In Fig.2. Therefore the curves are straight lines whose equations are ˆ x(s.47) to obtain s(x) and q(x) gives the equation for a wavefront as S(x) = s(x) + S(x 0 ). q) = x 0 (q) + s p(q).43) indicate that d p/ds = 0 along a curve x(s. where s(x) = c(t − t0 )6 . The components of the vector q = (q1 . where we have reinserted the q to indicate that we are now considering a family (a pencil) of such curves. d x/ds = ∇ S. q) deﬁne a family of curves that are orthogonal to each surface S(x) = ct + C. the expansion (2. rather than ray expansions. (2. In the neighborhood of these surfaces. Surfaces at which this Jacobian vanishes are called caustics. 2.
ρ1 = ρ01 + s and ρ2 = ρ02 + s. where the ρ0i are the radii of curvature of the initial surface S(x 0 ) = ct0 . [(ρ01 + s)(ρ02 + s)]1/2 (2. but the normal continues to point in the direction of propagation. again in a normal direction. q) x0(q) ˆ sp S(x) q S(x0) Fig. where p is a unit vector pointing along (tangent to) the ray Weatherburn (1939)7 shows that ∇ 2 S = 1/ρ1 + 1/ρ2 . A ray is launched in a direction normal to an initial wavefront S(x 0 ) from the point q on S(x 0 ). namely that of contraction. and its present position is x(s. Accordingly. ds 2 ρ01 + s ρ02 + s This is a differential equation along each ray q.32 2 Kinematical Descriptions of Waves x(s. If a wavefront is expanding. 7 The difference in sign from that in Weatherburn arises because of how we have deﬁned the signs of the radii of curvature. These radii take a sign. and the normal to the surface points in the direction of expansion.49). (2. the ﬁrst transport equation. Moreover. After propagating a distance s along this line.49) where A(q) is assumed to be given. The starting point of the ray is ˆ ˆ x 0 (q). Its solution is A0 = A(q) . it pierces the wavefront S(x). 2. (2. the radii of curvature are negative.2. its radii of curvature are positive.48) where ρ1 and ρ2 are the principal radii of curvature of the surface S(x) = s(x) + S(x 0 ). It is a straight line. because s is the distance along the straight lines orthogonal to each surface. If the wavefront is contracting. q) = x 0 (q) + s p.50) (2. can be written as 1 1 1 dA0 + + A0 = 0. .
in which case our asymptotic approximation is no longer accurate. The vector potential ψ is written as ψ(x) ∼ eikT [S(x)−cT t] (ik T )−α−1 ∞ n=0 (ik T )−n An (x). Lastly. note that the polarization of the leadingorder term (2.2 Shear Wave Next we consider a shear wave. can be written as ∇ · ∇ S A2 = 0. Interestingly. It captures our intuitive notions of a wave as an object whose phase steadily increases as we move along a curve. but this does not preclude the possibility that the higherorder terms could have different polarizations.51) is that of a longitudinal wave. If conservation of energy were our only goal. uL ∼ ˆ p A L (q) eik L [s(x)−cL (t−t0 )] .52) Integrate this over a tube formed from rays and. (2. (2. The power α must also be given.53) where k T = ω/cT .4.4).52) gives both its magnitude and its argument. Note also that the denominator may vanish.4 Asymptotic Ray Expansion 33 Collecting the various pieces and noting that the compressional component of the particle displacement u L = ∇ϕ.4 Conservation of Energy in a Ray Tube Show. deduce the result. (2.49). for an inﬁnitesimal cross section. The 0 reader is encouraged to do so.51) where we have put back the subscript L to indicate that we have calculated the compressional component. 0 (2. we should consider ∇ · (∇ S A0 A∗ ) = 0. gives. This is a remarkable expression.50).2. Problem 2. that the ﬁrst transport equation. (ik L )α [(ρ01 + s)(ρ02 + s)]1/2 (2.54) . this implies that A0 · ∇ S = 0. this equation establishes more than just the conservation of timeaveraged energy. the tangent vector to the curve being a ray. typically as part of the initial data for the problem. This happens on the caustic surfaces mentioned previously. The A0 can be complex and the solution to (2. 2. (2. to leading order. by multiplying through by A0 . Recall that ∇ · ψ = 0. Provided we do not seek terms beyond the leading one. and whose amplitude decays geometrically such as to conserve energy (Problem 2.
we ﬁnd that this equation reduces to 1 ∂ 2 (r ϕ) ∂ 2 (r ϕ) = 2 . T V . A0 will be orthogonal to p and have two linearly independent T ˆ T ˆ ˆ components. We select the unit vectors d I so that ˆ ˆ ˆ p ∧ dT V = dT H .34 2 Kinematical Descriptions of Waves The kinematic analysis here will be identical to that of the previous section so ˆ that p = ∇ S gives the direction of the rays.57) Such a waveﬁeld could be excited by a uniformly pressurized spherical cavity. (2. Let us assume that there is only an . Setting u = ∂ϕ/∂r . ∂r 2 c L ∂t 2 and its solution is ϕ(r. to leading order. t) = r 1 f t− r cL r 1 + g t+ .51) and (2.55) At this point the analysis given in the previous section applies to each component of A0 . the shear particle displacement is uI ∼ χ I d I A I (q) α [(ρ + s)(ρ (ik T ) 01 02 ˆ eikT [s(x)−cT (t−t0 )] . The ﬁrst term represents an outgoing wave and the second an incoming one. The equation of motion becomes 2 ∂u 1 ∂ 2u 2u ∂ 2u + − 2 = 2 2. ∂r 2 r ∂r r c L ∂t (2. It takes the presence of a source or boundary to couple the waves together.59) (2. we consider a waveﬁeld whose only dependence is upon the radial coordinate r and whose only component of particle displacement u is in the radial direction. which propagate along straight ˆ lines. A0 V d T H and A0 H d T V .58) where f and g are arbitrary functions. the polarization can change as one proceeds to the higherorder terms. To begin. r cL (2. Appendix: Spherical and Cylindrical Waves The following is a summary of information about spherical and cylindrical waves.56).56) + s)]1/2 where χ I = ∓1 as I = T H. As with the compressional wave. We ﬁnd that. Note that the radii of curvature ρ0i are different for the two wave types (2. (2. Hence.
Oxford: Pergamon. = 2 2 ρ ∂ρ cT ∂t 2 ∂ρ Taking the temporal transform..2) H0 (k T ρ) ∼ (2/πk T ρ)1/2 e±i(kT ρ−π/4) . 6th (corrected) ed.60) in agreement with (2. solely in the θ direction.2) solutions. r 35 (2. pp. k T ρ → ∞.63) (2. Setting v = −∂ψ3 /∂ρ. Bivectors and Waves in Mechanics and Optics. The only component of particle displacement is v. J. 1982. There is no dependence on x3 . Ray Methods for Waves in Elastic Solids. (2. and McMaken.64) where the plus sign goes with the 1 and the minus sign with the 2. we ﬁnd that this equation reduces to 1 ∂ψ3 1 ∂ 2 ψ3 ∂ 2 ψ3 + . We use a cylindrical coordinate system to describe it. the Hankel functions H0 (k T ρ).62) This is Bessel’s equation of order zero and has. The only dependence is upon the polar. Ph. Note that the particle displacement u = ∂ϕ/∂r so that there are both r −1 and r −2 terms. we see that H0 (k T ρ) represents an outgoing wave and H0 (k T ρ) an incoming one. A. 1993. H. Gautesen. on the walls of a cylindrical cavity. 77–88. Principles of Optics.K. + ∂ρ 2 ρ ∂ρ ρ cT ∂t (2. M.. The equation of motion is v 1 ∂v 1 ∂ 2v ∂ 2v − 2 = 2 2. Boston: Pitman. Then we get ϕ= ¯ f¯ (ω) ik L r e . radial coordinate ρ. From this (1) (2) behavior. and Wolf.. pp. Born. 1986. The asymptotic behavior of these functions is (1. References Achenbach. Recall that the particle displacement is v = −∂ψ3 /∂ρ. 109–127.61) Such a waveﬁeld could be excited by a traction. New York: Chapman and Hall. rotary shear motion. and Hayes. Next we consider inplane.References outgoing wave and take its temporal transform. as two of its linearly independent (1. E. .28). Boulanger. θ direction. M. we get ¯ ¯ 1 ∂ ψ3 ∂ 2 ψ3 2 ¯ + + k T ψ 3 = 0. 2 ρ ∂ρ ∂ρ (2.D. a displacement in the angular.
H. 1990. 1974. Focusing of an ultrasonic beam by a curved interface. Vol. pp. 1995. pp. C. E.J.G. J. 1980. Differential Geometry of Three Dimensions. New York: WileyInterscience. Cambridge: University Press. Choi. Hinch. Vol. 1978. and Harris. pp. P. M. G. New York: Cambridge Sommerfeld. Linear and Nonlinear Waves. H.B. Lighthill. 1991.C. Straus. Oxford: Pergamon.E. VI.A. A. New York: Cambridge. 84–101. Fourier Analysis and Generalized Functions. 1.C. Partial Differential Equations in Physics. New York: Springer.C. 225–227. 1964. J. 1939. Weatherburn. Whitham. Wave Motion 11: 383–406.36 2 Kinematical Descriptions of Waves Choi. 46–57. Translated by E.G. New York: Academic. 1989. . Holmes. and Harris. Perturbation Methods.J. Introduction to Perturbation Methods. M.G. Clemmow. The Plane Wave Spectrum Representation of Electromagnetic Fields. Wave Motion 12: 497–511. Lectures on Theoretical Physics. 1966. H. The Excitation and Propagation of Elastic Waves. J. Hudson. Scattering of an ultrasonic beam from a curved interface. Cambridge: University Press.
as we have previously noted. but suppress the e−iωt unless it is explicitly needed.2) . The incident wave u0 is described by ˆ ˆ u0 = A0 p 0 eik L p0 ·x . These waves. must be continuously renewed from a waveﬁeld in the interior of one or both the materials to sustain their propagation. a tractionfree elastic surface is special in that a wave can be excited on such a surface and not require an interior waveﬁeld to sustain it. 37 (3. while decaying perpendicularly away from it. in many cases. are selected and how these selections manifest themselves in the physical domain.1 indicates the geometry of the problem along with a schematic representation of the interaction. (3. which describes both the polarization of the wave and its direction of propagation. is given by ˆ ˆ ˆ p 0 = sin θ0 e1 + cos θ0 e2 .1) ˆ where the unit vector p 0 . where z is the complex variable and k a parameter. the chapter describes waves that propagate along an interface.1 Reﬂection of a Compressional Plane Wave We consider a longitudinal or compressional plane wave incident to a tractionfree surface.3 Reﬂection. Extending this connection. we assume that the disturbance is time harmonic. and Interfacial Waves Synopsis Chapter 3 describes reﬂection and refraction at an interface between two materials having different densities and wavespeeds. Moreover. Refraction. Further. we discuss how branches of the function (k 2 − z 2 )1/2 . Figure 3. However.36) will carry the resulting expressions into corresponding timedependent ones. 3. There is an intimate relation between the topography of the complex wavenumber plane and the physical manifestation of the propagating waveﬁeld. We do so knowing that using (1.
the scattered waveﬁelds generally break into reﬂected and diffracted ones.3) ˆ where the vector p 1 is given by ˆ ˆ ˆ p 1 = sin θ1 e1 − cos θ1 e2 . A compressional plane wave is incident to the surface in the ˆ direction p 0 .38 3 Reﬂection. it is reﬂected. and the ei are the unit vectors along the xi axes (i = 1. the incident wave is plane so that it imposes a projection of its phase everywhere on the surface. 2). while x2 > 0 is a vacuum. otherwise. . Any waveﬁeld that penetrates a shadowed region is a diffracted one. The amplitude A0 is real and positive. The reﬂected shear wave is described by ˆ ˆ u2 = A2 d 2 eikT p2 ·x . 3.4) (3. The surface is free of tractions. The elastic solid ﬁlls the x2 < 0 halfspace. Therefore any waves excited at the boundary must also be inplane.5) where the propagation direction is given by ˆ ˆ ˆ p 2 = sin θ2 e1 − cos θ2 e2 . Refraction. (3. As in previous work. 3. Further. The reﬂected compressional wave is described by ˆ ˆ u1 = A1 p 1 eik L p1 ·x . When the obstacle is impenetrable. 1 (3.1. and Interfacial Waves Fig. because both wave types can have matching phase components along the surface. We therefore expect both compressional and shear plane waves to be reﬂected1 from the boundary.1. (3. The incident polarization has no antiplane component. ˆ The angle θ0 is indicated in Fig.6) I use the phrase scattered wave or scattered waveﬁeld to describe almost any disturbance that is returned from or perturbed by an obstacle struck by an incident wave. k L = ω/c L .
The phrase phasematching is often used as a verb in the sense that “the reﬂected waves phase match to the incident one. a little thought indicates that they must take the form (· · ·)A0 eik L sin θ0 x1 + (· · ·)A1 eik L sin θ1 x1 + (· · ·)A2 eikT sin θ2 x1 = 0.1 deﬁnes the angles θ1 and θ2 . equivalently. a slowness surface. whenever it occurs.1. We deﬁne the slowness vectors ˆ s 0 = p 0 /c L . where s I = c−1 is the slowness. ˆ s 2 = p 2 /cT . Note that both reﬂected waves propagate away from the surface. (3. The amplitudes A1 and A2 may be complex.2 as the parameter θ0 is varied through 2π .2.1 Phase Matching The boundary conditions at x2 = 0 are that τ22 = τ21 = 0. in this case to the right. This can happen only if θ0 = θ1 and s L sin θ0 = sT sin θ2 . And as in previous work. of the various wavelengths be equal or. .66). The phasematching condition is stated as the condition that the s1 components of each slowness vector must be equal. 39 (3. something physically interesting will happen. Direct application of these boundary conditions can be tedious. These two conditions I are called the phasematching condition. This is a similar condition to that used to derive (1. Auld (1990) gives an account of these surfaces and of their use in determining phasematching. 3. k T = ω/cT . This is shown in Fig. but for anisotropic materials they can become very elaborate indeed. ˆ s 1 = p 1 /c L . Also note that the polarization of the reﬂected ˆ ˆ ˆ shear wave is chosen so that p 2 ∧ d 2 = e3 . on the surface x2 = 0.9) The tip of each slowness vector describes a circle. because only two dimensions are involved). We have indicated this by the vertical dashed line. 3. rather than a kinetic one.1 Reﬂection of a Compressional Plane Wave but the polarization direction by ˆ ˆ ˆ d 2 = e3 ∧ p 2 . and. In fact it is a kinematical condition. that the wavespeeds along the surface be identical. In the 2 Slowness surfaces for isotropic materials are always spheres (circles really. Phase matching is a fundamental principle of linear wave propagation. However. because ω is common to all the k I . There is a very useful diagram that geometrically represents the phasematching condition.8) This condition must hold for all x1 .3. (3.” Phase matching is nothing more than a demand that the projections. The surface is translationally invariant and the disturbance at one point differs from that at another only by an exponential phase term that indicates propagation.7) Figure 3.
even in the limit that θ0 = π/2. case we are examining here.1. The phasematching condition is indicated geometrically by demanding that the horizontal components (projections on the surface x2 = 0) of the slowness vectors be equal. For many materials. where ν is Poisson’s ratio.12) (3. and Interfacial Waves s2 s0 s1 s1 s2 Fig. for real θ0 . 3. The slowness surfaces for an isotropic elastic solid. θ2 must be real and less than π/2. 3. When this is done the reader will ﬁnd that R L (θ0 ) = A− (θ0 )/A+ (θ0 ). κ is close to 31/2 .10) (3. (3. (3.11) These terms must be supplemented with the condition s L sin θ0 = sT sin θ2 . The term κ = c L /cT is given by κ = [2(1 − ν)/(1 − 2ν)]1/2 .1 asks the reader to calculate the longitudinal or compressional reﬂection coefﬁcient R L (θ0 ) = A1 /A0 and the transverse or shear reﬂection coefﬁcient RT (θ0 ) = A2 /A0 .13) . The auxiliary functions are A∓ = sin 2θ0 sin 2θ2 ∓ κ 2 cos2 2θ2 . RT (θ0 ) = 2κ sin 2θ0 cos 2θ2 /A+ (θ0 ).40 3 Reﬂection.2 Reﬂection Coefﬁcients Problem 3. Refraction.2.
namely the reﬂection and refraction of an antiplane shear wave at the interface between two contrasting materials. Hence deduce (3. (3. While there are several other problems of this general kind that could be considered. To simplify the notation.2 Conservation of Energy Show that the power ﬂux into the surface equals that away from it. and so on. k T and similarly labeled symbols by c. show that R L 2 + RT 2 cos θ2 = 1. provided θ0 is real.10)–(3. Problem 3. That is. In this case both the traction and the particle displacement are continuous across the interface.3.16) (3.2 Reﬂection and Refraction 41 Note that. 3. κ cos θ0 (3.12). these reﬂection coefﬁcients have no frequency dependence and hence can be used both for timeharmonic and timedependent plane waves. antiplane shear wave incident to an interface between two materials that are in welded contact.15) Hint.1 Reﬂection Coefﬁcients Show that applying the boundary condition τ22 = 0 leads to λ + 2µ cos2 θ0 (A1 /A0 ) − κµ sin 2θ2 (A2 /A0 ) = − λ + 2µ cos2 θ0 . k. we treat only one other. It is clear that only an antiplane particle displacement can be excited at the interface by an . we replace cT . Problem 3.14) and that applying the condition τ21 = 0 leads to −µ sin 2θ0 (A1 /A0 ) − κµ cos 2θ2 (A2 /A0 ) = −µ sin 2θ0 . The calculation becomes straightforward once the reader realizes that he or she must be careful to use the same element of area at the surface to calculate both the ﬂux into the surface and that away from it.2 Reﬂection and Refraction We consider a plane.
The incident wave is described by ˆ u 30 = A0 eik p·x . is described by ˆ u 32 = A2 ei k p2 ·x .20) .3 indicates the various angles. the elastic ¯ ¯ shear modulus µ. indicated by the additional subscript 2.2). and Interfacial Waves Fig. antiplane shear wave is ¯ ¯ ¯ ˆ incident to the surface in the direction p 0 . In the x2 < 0 halfspace the density is ρ. Refraction. and the wavespeed c = (µ/ρ)1/2 . Figure 3. and the wavespeed c = (µ/ρ)1/2 . is described by ˆ u 31 = A1 eik p·x .18) with its propagation direction given by (3. ¯ (3. A plane.4). indicated with the additional subscript 1. the elastic shear modulus µ.17) where the subscript 3 indicates the component and therefore the polarization. The refracted wave. The wave that is transmitted across the interface is said to be refracted.42 3 Reﬂection. (3.19) with its propagation direction given by ˆ ˆ ˆ p 2 = sin θ2 e1 + cos θ2 e2 . The incident direction is given by (3. while in the x2 > 0 halfspace the density is ρ. so that the scattered waves must also be similarly polarized. (3. incident wave with an antiplane polarization. The reﬂected wave. and the 0 that the wave is incident. (3.3. 3.
3. The phasematching condition is indicated by the dashed vertical lines. 3. is propelled along it by the waves in the lower material.3 and the lower for the material of the lower halfspace. the critical angle of incidence is θ0c . as the solution to Problem 3. in which case θ2 must be complex to satisfy the phasematching condition. θ0 = θ1 and s sin θ2 = s sin θ0 . 3. This wave. that ¯ is. Of course θ0 can be greater than θ0c . This occurs when a wave incident to the interface from the slower material excites a wave skimming along the interface in the faster material. The lefthand side ¯ indicates the case c > c. ¯ where s sin θ0c = s . . This gives rise to an inhomogeneous plane wave in the upper material but to no timeaverage ﬂux of energy across the interface. The lefthand side indicates that when the upper material is faster and the righthand side that when it is slower. 3. and the right the opposite one. In principle the reverse can also occur. The upper slowness surface is that for the material of the upper halfspace in Fig.3. The two possible cases are shown. Phase matching demands that the s1 components of the slowness vectors be equal. More importantly.3. clinging to the interface. equivalently s < s. For ¯ ¯ the case c > c or.2 Reﬂection and Refraction s2 s2 43 s1 s1 s2 s2 s1 s1 Fig. In Fig.4. the slowness diagrams indicate the occurrence of critical refraction.4 indicates.4 the slowness surfaces for the two materials are arranged vertically in a pattern corresponding with the positions of the halfspaces in Fig.
We must then regard reﬂection and transmission coefﬁcients as functions of a complex variable. ¯ ¯ (3. Moreover.3 has indicated. analogously to critical refraction. When θ0 > θ0c . Consider all the possibilities for critical reﬂection and critical refraction. and Interfacial Waves Applying the continuity of traction and particle displacement at the interface gives the antiplane shear. T (θ0 ) = 2 cos θ0 /C+ (θ0 ). reﬂection coefﬁcient R(θ0 ) and the antiplane shear. as Problem 3.23) (3. 3. critical phenomena occur frequently in elasticwave scattering. In this case sin θ2 = cosh β and cos θ2 = ∓i sinh β. implying that R(θ0 ) = e−i2ϕ and T (θ0 ) = T (θ0 )e−iϕ . of the slowness surfaces for inplane motion. namely R(θ0 ) = C− (θ0 )/C+ (θ0 ). Let us continue to consider the problem of Section 3.4. θ2 must be complex to permit sin θ2 > 1.3 Critical Refraction and Interfacial Waves Our discussion of critical refraction indicates that we do not need to consider the angle of refraction as real. (3.2 and work with the ¯ case of c > c. The auxiliary functions are ¯ c C∓ = cos θ0 ∓ (µc/µ¯ ) cos θ2 . Critical reﬂection. The reader may be surprised by the large number of occurrences of critical reﬂection and refraction.44 3 Reﬂection. refraction or transmission coefﬁcient T (θ0 ). interesting wave processes occur. In fact.23) and note that R(θ0 ) = 1. Here ϕ is the argument of C+ (θ0 ).22) 3. similar to Fig. To ﬁnd its form we set θ2 = π/2 ± iβ.3 Slowness Surfaces Figure 3. Consider both an incident compressional plane wave and inplane shear one. Problem 3. Draw diagrams. arises when a wave skimming along the interface or surface of a material is excited by a slower wave incident to the interface from the same material.24) . Setting A0 = 1.4 shows how we can place slowness surfaces above one another to work out the phasematching conditions and the various critical angles. Next we look back at (3. we ﬁnd that the particle displacement in x2 > 0 becomes u 32 = T (θ0 )e−iϕ e±k sinh βx2 ei k cosh βx1 .21)–(3. where β is real and positive.21) (3. Refraction. we shall ﬁnd in Chapters 5 and 6 that where the complex plane is punctured by the poles or cut by the branches of these coefﬁcients.
It was to handle such situations that we rewrote the inverse temporal Fourier transform in the form of (1.3 Critical Refraction and Interfacial Waves 45 ˆ ˆ ˆ ˆ where p 2 has been written in full as p 2 = cosh β e1 ∓ i sinh β e2 .3. . namely ˆ u 30 (t − p 0 · x/c) = 3 1 π ∞ 0 ˆ A0 (ω) e−iω(t− p0 ·x/c) dω.25) to take account of this ω dependence.4 Flux of Energy in the Upper Material For the case of critical refraction.28) This argument has been taken from Friedlander (1947). When θ0 < θ0c the reﬂection and transmission coefﬁcients have no frequency dependence and therefore a plane pulse reﬂects and refracts exactly as a timeharmonic. (3. we must write R(θ0 ) = e−i2ϕ sgnω . we must select θ2 = π/2 − iβ to ensure that the waveﬁeld decays as x2 → ∞. ω < 0. In choosing the minus sign we assumed that k and thus ω was positive.36). −1. The argument ϕ is determined when ω is positive. We call the wave an interfacial one. ω > 0. Moreover. Therefore. If ω were negative then we must choose θ2 = π/2 + iβ to ensure decay. Thus (3. note that the wave would not exist unless a plane wave were continuously incident to the interface from x2 < 0 sustaining it. when θ0 ≥ θ0c . 2 ¯ (3. When critical refraction takes place there is a frequency dependence that distorts how a pulse reﬂects and refracts. this particular frequency dependence is rather interesting because it depends only on the sign of ω. (3. ¯ There remains one subtle point. where sgn ω = 1. (3. show that the timeaverage ﬂux of energy in x2 > 0 is given by ˆ ¯ ¯ F = 1 µωkT (θ0 )2 e−2k sgnω sinh β x2 cosh β e1 . Assuming ¯ that k is positive.26) T (θ0 ) = T (θ0 )e−iϕ sgnω . plane wave. However. To explore this further3 we construct an incident planewave pulse. Problem 3.24) is an example of both an inhomogeneous plane wave and of a wave that clings to the interface.27) Thus there is no ﬂux of energy to x2 → ∞ and energy is continually returned to the slower material.
The refracted pulse is rather more complicated. From the linearity of the problem it follows that the reﬂected pulse is given by ˆ u 31 (t − p 1 · x/c) = 1 π ∞ 0 ∞ 0 ˆ A0 (ω)e−i2ϕ e−iω(t− p1 ·x/c) dω. (3. To make this a little more concrete we examine an incident pulse of the form u 30 (t) = H (t)− H (t −a). Refraction.32) The incident pulse is reﬂected with a diminished amplitude and a new pulse has been excited. where H (t) is the Heaviside function and a measures the length of the pulse. π t (3.29) while the refracted pulse is given by u 32 (x. We rewrite the reﬂected pulse in the form ˆ ˆ ˆ u 31 (t − p 1 · x/c) = cos 2ϕ u 30 (t − p 1 · x/c) + sin 2ϕ v(t − p 1 · x/c). Its allied function is v(t) = The reﬂected pulse then becomes ˆ ˆ u 31 (t − p 1 · x/c) = cos 2ϕ u 30 (t − p 1 · x/c) + sin 2ϕ ˆ t − a − p 1 · x/c 1 ln .46 3 Reﬂection. and Interfacial Waves where we have used the fact that A0 (ω) = A∗ (−ω) because u 30 is real. but can now be an arbitrary amplitude with a frequency dependence.31) where ˆ v(t − p 1 · x/c) = 1 π ∞ 0 ˆ A0 (ω) e−iω(t− p1 ·x/c) dω.34) a−t 1 ln . The function v is called the allied function (Titchmarsh.30) The reﬂected pulse retains many of the features of the incident pulse and most importantly the argument indicating that the wave is plane. 1948). A0 (ω) is 0 no longer simply a positive real constant. (3.33) Note that the second pulse is present for all time and appears to arrive before the incident pulse. This is an example of a twosided wave. It cannot exist alone and must be part of a more extended disturbance. ˆ π t − p 1 · x/c (3. (3. . (3. t) = T (θ0 ) 1 π c c A0 (ω)e−iϕ e−ωx2 sinh β/¯ e−iω(t−x1 cosh β/¯ ) dω.
3 Critical Refraction and Interfacial Waves 47 To explain this anomalous behavior we must consider the transmitted pulse. u 32 decays as 1/x2 rather than ¯ as e−kx2 sinh β . plane acoustic wave from a thin elastic plate. t) = T (θ0 ) 1 π cos ϕ tan−1 + where c τ = t − x1 cosh β/¯ . This is not as anomalous as it might at ﬁrst appear. λ = x2 sinh β/¯ . The refracted pulse continually reradiates into the slower. while the timeharmonic disturbance decays exponentially. It does so by shedding a reﬂected pulse into the slower material that appears before the incident one arrives. Carrying out the necessary integrations gives u 32 (x. Moreover. the incident pulse excites a disturbance in the faster material. the main disturbance ¯ moves along the interface at c/ cosh β = c/ sin θ0 . We assume that the ﬂuid has no viscosity so that the incident acoustic wave excites predominantly ﬂexural motions in the thin plate. once we recall that the upper material is faster than the lower and that the incident plane pulse has been exciting the upper material from t → −∞. of the incident pulse. no energy is permanently carried into the upper material. and reﬂection. so that the pulse decays algebraically. ¯ This disturbance propagates along the interface at c and ﬁlls the whole upper material.3. 3. therefore.35) (τ − a)2 + λ2 1 sin ϕ ln 2 τ 2 + λ2 Clearly the refracted pulse is also two sided and present for all time.3. c But cos θ2 = [1 − (¯ /c)2 sin2 θ0 ]1/2 .36) τ λ − tan−1 τ −a λ . the region near the interface. at the interface and must somehow satisfy the boundary condition. lower material. However. then in fact the phenomenon arises from the deﬁnition of the branch of the ¯ ¯ function (k 2 − z 2 )1/2 (= k cos θ2 ) that appears in the reﬂection and transmission coefﬁcients. The origin of this phenomenon is the cos θ2 in the expressions for C∓ . c (3. Problem 3. (3. we assume that the ﬂuid below the plate is sufﬁciently dense that the elastic . At the far left. Therefore the precursor in the upper material gets ahead of the trace of the wavefront.34). at least. at x1 → −∞ (θ0 > 0) in Fig.23). If we regard (ω/c) sin θ0 as a complex variable z. It is also interesting to note that. manifest themselves as branch cuts in the complex plane. Moreover. Critical refraction. (3.5 Reﬂection of an Acoustic Wave from a Plate Consider the reﬂection of a time harmonic. or. the second term in (3. as the distance from the interface increases.
Note that phase matching must take place so that the x1 dependence of w is determined by the incident wave.37) where ϕ is the velocity potential. respectively. The terms ρ p . Water and air. 0)/D. incident from the ﬂuid. h. A thin elastic plate separates a dense ﬂuid such as water from a tenuous one such as air. and a time dependence of the form e−iωt is assumed but suppressed. Refraction. respectively.5. The density of the ﬂuid is ρ. Figure 3. However. and D are the density per unit area. k = ω/c. and Interfacial Waves Fig. 3. ﬂexural motion of the plate is described by (∂1 ∂1 )2 w − k 4 w = p(x1 . ˆ ˆ The incident and reﬂected waves are ϕ0 = A0 eik p0 ·x and ϕ1 = A1 eik p1 ·x . p (3. but that the ﬂuid above the plate is sufﬁciently tenuous that it can be treated as a vacuum. Find α. x3 ) = ∂ϕ/∂ x2 .48 3 Reﬂection. The one dimensional.5 indicates the geometry of the problem. which we model as a vacuum. The pressure p = iωρϕ and the particle displacement is −iωu = ∇ϕ.38) where k 4 = ω2 ρ p h/D.38) then gives a boundary condition connecting p and ϕ at x2 = 0. respectively. A time harmonic. and the elastic constant for the plate. the particle displacement is continuous so that −iωw(x1 . p the thickness. 3. strikes the plate and is reﬂected. Equation (3. plate and the ﬂuid can interact. In the ﬂuid ∇ 2ϕ + k2ϕ = 0 (3. At x2 = 0. Speculate about the form of a reﬂected pulse. In contrast there are waves that . plane acoustic wave.4 The Rayleigh Wave We have seen that critical refraction produces a wave in the faster material that clings to the interface. satisfy these assumptions. Note that R(θ0 ) = 1 so that R(θ0 ) = −ei2α . this wave could not exist without being continuously sustained by a wave from the interior. Find the reﬂection coefﬁcient R(θ0 ).
40) The boundary conditions at x2 = 0 are τ22 = τ21 = 0. γL = 1 − c2 c2 L 1/2 . 2 2 D 2 − cr cT (3.4 The Rayleigh Wave 49 cling to a surface or interface that are selfsustaining.19) is then automatically satisﬁed]. gives 2 2 − c2 cT 2 − 4γ L γT = 0.3. 3.3). Their particle displacement decays exponentially with distance away from the surface or interface.4. It is one of the distinctive features of linear elasticity that a tractionfree surface guides such a wave.41) we ﬁnd that −2iγT C = . 2 2 − c2 /cT −2iγ L 2iγT 2 2 − c2 /cT C 0 = . (3. We seek a timeharmonic wave whose particle displacement is inplane and that decays in the positive x2 direction. Both here and in Chapter 5.43) . is called the Rayleigh equation. it is demonstrated that this latter surface wave. We examine it momentarily. Such waves are also referred to as surface or interfacial waves. (3. or an insigniﬁcant modiﬁcation of it. (3.41) Setting the determinant of the matrix to zero.21) and (1.22) for the potentials. D 0 (3. ψ3 = De−βγT x2 eiβx1 . and assume that ϕ = Ce−βγL x2 eiβx1 . thus 0 < cr < cT . It has a real positive solution c = cr .42) This equation. and ones similar to it. γT = 1 − c2 2 cT 1/2 .1 and 3. To satisfy (1. giving the wavespeed of the surface wave. arises from a pole rather than a branch point. called a Rayleigh wave. 3. in the timeharmonic approximation. To satisfy these boundary conditions. We use a scalar potential ϕ and a single component ψ3 of the vector potential [the divergence condition (1.39) where β = ω/c and c is the unknown wavespeed along the surface x2 = 0. Returning to (3. Note that γ L and γT must be real if we are to have decay into the interior.1 The TimeHarmonic Wave We consider an elastic halfspace with the x1 coordinate lying along the surface and the positive x2 coordinate pointing into the interior (the reverse of that shown in Figs. the condition for a nontrivial solution.
This then is the form of the time harmonic. 2γT cT (3. In this case the frequency dependence enters through the terms e−βr γL x2 and e−βr γT x2 . e e 2 cT −1 A0 (ω)eiπ/2 γT 2γ L γT e−(ω/cr )x2 γL A0 (ω) 2e−(ω/cr )x2 γL − 2− u 2 (t − x1 /cr .50 3 Reﬂection. Refraction. 2 γT (x2 /cr )2 γT + (t − x1 /cr )2 (3. e e 2 cT (3. where C is a constant.4.28)–(3. Proceeding as we did in (3. x2 ) = − + 2 2 C 2 − cr /cT [(x2 /cr )γT ] 2 (x2 /cr )2 γT + (t − x1 /cr )2 . Much the same thing happens when a timeharmonic Rayleigh wave is mapped into the time domain. 3. Performing the integrations gives u 1 (t − x1 /cr . x2 ) = 2C (x2 /cr )γ L 2 (x2 /cr )2 γ L + (t − x1 /cr )2 − u 2 (t − x1 /cr .28)–(3.44) where βr = ω/cr .30) and would be set by the source. For simplicity we take A0 (ω) = πC.46) . Rayleigh wave. Note the eiπ/2 in u 2 . x2 ) = 1 π ∞ 0 2 cr −(ω/cr )x2 γT −iω(t−x1 /cr ) dω.45) where A0 (ω) has the same form as in (3. and Interfacial Waves With the use of (1.30).19). the transient displacement components are given by u 1 (t − x1 /cr .2 Transient Wave Previously we have seen that a critically refracted wave becomes a twosided wave in the time domain. 2C (t − x1 /cr )γ L 2 (x2 /cr )2 γ L + (t − x1 /cr )2 2 2 C 2 − cr /cT (t − x1 /cr ) . 2 cT 2 −βr C cr 2γ L γT e−βr γL x2 − 2 − 2 e−βr γT x2 eiβr x1 . x2 ) = 1 π ∞ 0 − 2− 2 cr −(ω/cr )x2 γT −iω(t−x1 /cr ) dω. the particle displacement components are u 1 = iβr u2 = 2 cr C 2e−βr γL x2 − 2 − 2 e−βr γT x2 eiβr x1 .
and. Recall that the denominator of the reﬂection coefﬁcient for the inplane reﬂection is A+ (θ0 ). is given by 2 R(s) = 2s 2 − sT 2 2 − 4s 2 s 2 − s L 1/2 2 s 2 − sT 1/2 . To do so it is more convenient to work with R(s). Using our approximations.4 The Rayleigh Wave 51 Because of our choice of A0 . but with the plate replaced by an impedance.4. Further. The compressional pulse was excited at t → −∞ and propagates outward at wavespeed c L . The question then is whether the pole is real or complex. A pole in the reﬂection coefﬁcient implies that a surface wave exists.48) where k = ω/c. Note that there is no leading wavefront. these pulses are more singular than real ones.42). but nevertheless they exhibit the basic features of interest. given in (3. has already propagated everywhere along the surface. having noted the presence of the eiπ/2 in (3. consider the same general geometry as shown in Fig. This feature is not accidental. a slightly modiﬁed version of (3.6 A Surface Wave Supported by an Impedance Consider a time harmonic disturbance u(x1 . so that sin θ0 = s/s L > 1. the reader. 3.5. That is. the boundary . x2 ) that satisﬁes ∂1 ∂1 u + ∂2 ∂2 u + k 2 u = 0. if complex. The slownesses are ordered as s > sT > s L for a surface wave to exist. where c L > cT > cr . This function.3. The most important feature of this equation is that the function on the lefthand side contains two radicals whose branches must be deﬁned with care. 3. Lastly. Also note that the exponential decay of the timeharmonic disturbance has become an algebraic decay. Setting s = s L sin θ0 = sT sin θ2 . We have a twosided wave.12). and thus θ0 must be complex. sometimes called the Rayleigh function.47) where s = 1/c and s I = 1/c I . (3. (3.3 The Rayleigh Function We are thus left with examining the roots of (3.42). may ﬁnd it interesting to show that the u 2 component is essentially the allied function of the u 1 component. Problem 3. A homogeneous plane wave striking a tractionfree boundary cannot excite a surface wave.45). is its imaginary part such that the surface wave is physically meaningful. which is that of a compressional pulse. we ﬁnd that A+ (θ0 ) = R(s L sin θ0 )/(s L sT )2 . we have implicitly assumed that the leading wavefront.
Perturbing α slightly. the response at a given point on the boundary does not depend upon the response of the surrounding points. we can write γ 2 as 2 γ 2 = (σ 2 − τ 2 ) − kr − ki2 + 2i(σ τ − kr ki ). eikr x e−ki x .50). a and b are positive and real. Setting α = σ + iτ . so that (γ ) ≥ 0 ∀ α.6. Z = R + i X. k is a wavenumber. namely. 3. and Interfacial Waves condition at x2 = 0 becomes ∂2 u = ik Z u. but somewhat easier to begin by working4 with γ 2 . The curves (γ ) = 0 therefore deﬁne the branch cuts.18). we want to deﬁne the branches of the radical.52) We next partition the α plane. by imagining that the material in which the wave 1. Accordingly.39) and (3. By conis propagating is slightly lossy. (3. where ki /kr  sidering a plane wave propagating in the positive x direction with wavenumber k.49) with Z the impedance. Next show that the surface wave u = Ce−bx2  eiax (3. C is a constant. show that the reﬂected wave takes the form of the surface wave.17). (3. set k = kr + iki . Almost always.51) where α is the independent variable and k is known. That is. For this same Z show that the reﬂection coefﬁcient has a pole at a particular θ0 .52 3 Reﬂection.40). we note that for a surface wave (γ L ) ≥ 0 and (γT ) ≥ 0. Refraction. using the curves (γ 2 ) = 0 and (γ 2 ) = 0. we ﬁnd that (γ 2 ) < 0 between the 4 I ﬁrst learned of this argument from Mittra and Lee (1971). and as the reﬂected wave u 1 one identical to (3. say θ0s . (3.4 Branch Cuts The Rayleigh function contains two radicals of the form γ = (α 2 − k 2 )1/2 . calculate the reﬂection coefﬁcient. but not upon the angle of incidence of the wave. Using this θ0s . It is hard to work with γ directly. .4. we see that kr and ki must both be positive. What restrictions must R and X satisfy for a surface wave to exist? Find a and b in terms of a Z that permits a surface wave to exist. as shown in Fig 3. The impedance may depend upon ω. Thus we may. Looking back at (3. Taking as the incident wave u 0 one identical to (3.50) is also a solution. (3.
3. two curves (γ 2 ) = 0.7 indicates from what reference the angles are measured. . Expressing γ as γ eiθ . express γ as γ = (r1r2 )1/2 cos where (α − k)1/2 = r1 eiθ1 /2 .54) Figure 3. (γ ) = 0. if (γ ) > 0.6 overlain with the wavy line.7. To investigate this branch.6. note that to go from the ﬁrst to the second quadrant. while (γ 2 ) < 0 between the two curves (γ 2 ) = 0. 1/2 θ1 + θ2 2 + i sin θ1 + θ2 2 .4 The Rayleigh Wave 53 Fig. 3. and remain on the same Riemann sheet. 1/2 (3. The complex α plane with a sketch of the curves (γ 2 ) = 0 and (γ 2 ) = 0. From these curves we select the branch cuts indicated by the wavy overlying line on the curves (γ 2 ) = 0. These are the curves in Fig.3. 3. As well. we note that. we arrive at the branch cuts shown in Fig. In the limit as ki → 0. while (γ ) > 0 in the ﬁrst and third quadrants and negative in the other two. we must ﬁrst move into the fourth and then poke through the gap at the origin into the second. The branch cuts. That is along (γ 2 ) = 0.53) (α + k)1/2 = r2 eiθ2 /2 . Note that (γ ) ≥ 0 ∀ α. therefore must lie along the curves 2θ = ±π. then θ < π/2 and thus 2θ < π. where (γ 2 ) < 0. (3.
actually solving for sr is less so. 1997). exists. do the roots on the other Reimann sheets ever manifest themselves? Yes.47). and one connecting −s L and −sT . at s = s L and at s = sT . Hence the Rayleigh function R(s) is deﬁned as that branch occupying the Riemann sheet (sγ L ) ≥ 0. However. (sγT ) ≥ 0 ∀ s and cut as just indicated. This theorem allows one to systematically calculate the number of poles and zeros of a function by performing a contour integral of its logarithmic derivative.7. Thus a real root sr . While knowing that only the Rayleigh slownesses ±sr exist on the Reimann sheet of interest is important. does R(s) have any other roots on this Riemann sheet? The answer is no. Both radicals are deﬁned as just indicated. Second. R(s) > 0 but as s → ∞ along the positive real s axis. We next seek a root or roots to R(s) = 0 on this Riemann sheet. though the original argument may be found in Cagniard (1962). they do when they lie close to a branch cut. Third. R(s) becomes negative. (3. Refraction. Achenbach (1973) does this in some detail. Recall that s = 1/c. with sT < sr < ∞. the product is only discontinuous across a branch cut connecting s L and sT . These are the appropriate branch cuts for the product. Returning to the Rayleigh function R(s). This can be proven conclusively by using a theorem sometimes called the argument principle (Ablowitz and Fokas. and Interfacial Waves Fig. 3. Lastly. First. By symmetry a root −sr also exists.54 3 Reﬂection. note that if sr is a root then so is −sr . This point is discussed in Aki and Richards (1980) in their description of the reﬂection of spherical waves from a tractionfree surface. It can be calculated . The complex α plane as ki → 0. we note that there are two 2 2 radicals appearing as the product (s 2 − s L )1/2 (s 2 − sT )1/2 . (γ ) ≥ 0 ∀ α on the Riemann sheet exhibited in the drawing.
A.G. On the total reﬂection of plane waves. Introduction to the Theory of Fourier Integrals. Theory and Methods.24) starts at −∞ in the second quardant. The two are connected by ξ = iγ .51). Appl. Auld. For many materials κr is close to 0. FL: Krieger. However. 2.23). Quantitative Seismology.C. Translated and revised by E. J. pp.86 + 1. 1980. and ends at +∞. Quart. passes below the branch cut.D. Wave Propagation in Elastic Solids. 2nd ed. pp. pp. pp. S. in the case of (2. 42–49. Math. New York: McGrawHill. A. Further. note that k1 = iγ in (2.23) and k3 = iγ in (2. Vol. I. Malabar. 1962. Dix..H. 2nd ed. K. References Ablowitz. F. and Richards. and elsewhere.G. pokes into the fourth quadrant. Mech. E. Cagniard. 147–148.55) The term κr = sT /sr and ν is Poisson’s ratio. pp. and Lee.References to good accuracy by using the formula κr = (0. R. The radicals we encounter will often present themselves in the form ξ = (k 2 −α 2 )1/2 . rather than as that in (3.12ν)/(1 + ν) 55 (3. Reﬂection and Refraction of Progressive Seismic Waves. In particular.J.23) and (2. 1–57. Friedlander. New York: Macmillan. 119–121. 189–191. Only when considering the Cagniard–deHoop integration technique in Section 5.S.31). M.W. Aki.9. The branches of (2. 1. B. J. San Francisco: Freeman. Analytical Technique in the Theory of Guided Waves. Amsterdam: NorthHolland. Complex Variables. . pp. Mittra. Vol. 259–265.1 do we select a different branch. Achenbach. the complex k2 plane is cut so that (k1 ) > 0 ∀ k1 . and Fokas. P. Flinn and C. 319–333. 1971.31) were selected as just indicated.. 1973. Acoustic Fields and Waves in Solids. New York: Cambridge. pp. 1990. the contour in (2. 1997. 20–23. Oxford: Clarendon Press.A. 1948. passes above the branch cut. 1974. Titchmarsh. 1: 379–383. corresponding exactly with the branch of γ just selected.
resulting in no initial wavefront being present. indicating as we do so both the role of the principle of limiting absorption and that of specifying an edge condition. but construct. Though we make limited use of it in the chapters that follow.4 Green’s Tensor and Integral Representations Synopsis In Chapter 4 we discuss the formulation of integral representations of solutions to rather general problems in elasticwave propagation. 1980).3.1 Introduction In Chapter 2 we moved away from discussing plane waves to an introduction of planewave spectral representations in Section 2. both far more general representations and ones in physical space rather than in wavenumber space. 56 . 1973. in a sense. A general survey of many useful representation results is provided by deHoop (1995). We must work in a fourdimensional space. The chapter closes with an example that uses these ideas to develop an integral representation for the scattering of an acoustic wave by an elastic inclusion. 1958. Hudson. this material is very important because it is the basis for formulating elasticwave problems in a form suitable to be analyzed numerically. This allowed us to discuss more general waveﬁelds and to understand their propagation characteristics in terms of those of plane waves. we establish a uniqueness result. Achenbach. in this chapter. 4. We continue with this general theme. Moreover. has been going on forever. For these representations for an inﬁnite domain to be derived. This is needed for timeharmonic problems because the disturbance. the principle of limiting absorption is introduced. it is straightforward (straightforward is not a synonym for easy) to seek its representations (Friedlander. When the waveﬁeld is time dependent. Two constructions are used: the reciprocity identity and the Green’s tensor for a full space.
4. Moreover. (4.1) 1. at some stage we must send the outer surfaces. In constructing representations for various waveﬁelds. (4.3) 1 Using the relation between stress and strain. and it has thus. and to incorporate those effects in an overall propagation model the Fourier component of the waveﬁeld rather than. we have imagined that the waveﬁeld was excited at the time minus inﬁnity.2 Reciprocity 57 the fourth dimension being time.2 = τ ji n j .1) with the particle displacement of its reciprocating waveﬁeld and subtract. we note that τ ji ∂ j u i2 = 2 τ ji ∂ j u i1 so that the righthand side of (4. though we need only work with regions in threedimensional space. When working with a timeharmonic waveﬁeld. However. indicated by the superscripts 1 and 2.1.2) follows after integration. to inﬁnity and must be assured that these integrals either go to zero or a ﬁnite. ﬁlled all of space. We take the scalar product of each equation in (4. In this chapter we use the principle of limiting absorption as the technical device to achieve these outcomes. the equations of motion in the timeharmonic approximation for two reciprocating elastic waveﬁelds.2 = 0. . (4.2 + ρω2 u i1. a representation in this form is less useful than it might seem because both the source and receiver of the waveﬁeld have their own frequency behaviors. over which integrals are being taken.4. unambiguously identiﬁed value. Nevertheless. are 1.2) Proof. Thus we consider almost exclusively timeharmonic waveﬁelds.2 ˆ ˆ The tractions on ∂Rx are ti1.3). This gives 2 1 ρ f i2 u i1 − f i1 u i2 = −u i1 ∂ j τ ji + u i2 ∂ j τ ji . (1. (4. even for a ﬁnite positive time.2 ∂ j τ ji + ρ f i1. say. Then Rx f i2 u i1 − f i1 u i2 ρd V = ∂Rx u i2 ti1 − u i1 ti2 d S. The unit normal n to ∂Rx is outward.2 Reciprocity Proposition 4. its timedependent response to a sudden impulse is of greater use. In a bounded region Rx with surface ∂Rx . we must unambiguously identify waves that are outgoing from their source of excitation. the waveﬁeld propagates outward from its sources at a ﬁnite speed and thus for a ﬁnite time it occupies a region that is bounded.3) equals ∂ j (τi1j u i1 −τi2j u i1 ).
44).” 4. Then the reciprocity statement becomes ai2 u i1 (x 2 ) = ai1 u i2 (x 1 ). This is the origin of the careless statement “the source and receiver can be interchanged by reciprocity. Proposition 4.2 = ai1. The particle displacement u iG is the solution to G ˆ (λ + µ)∂i ∂k u k + µ∂ j ∂ j u iG + ρω2 u iG = −ρ a i δ(x). the solution represent an outgoing waveﬁeld. x = x.43) and (1. The differential in physical space d x = d x1 d x2 d x3 and that in the transform space dk = dk1 dk2 dk3 .58 4 Green’s Tensor and Integral Representations There is one special case of (4. u ik . Here a is a constant unit vector giving the direction of the ˆ G point force at the origin. The Green’s tensor1 for a full space.2. τiG ) as a Green’s state. ∞ −∞ ∗ (4.6) subject to the condition that. I = L .2) of interest. (4. T. (4.4) u(x) = 1 (2π)3 u(k)eik·x dk. Let us assume that the integral over ∂ Rx vanishes and that f i1. with u ik given by G u ik = 1 2 2 k T cT 2 ∂i ∂k [−G(k L x) + G(k T x)] + k T δik G(k T x) . as x → ∞.3 Green’s Tensor In the proposition to follow we shall use the threedimensional Fourier transform pair given by ∗ u(k) = ∞ −∞ u(x)e−ik·x d x.5) This is a straightforward generalization of (1. where x is the position vector of the point of action of the delta function.8) Note that we may replace x with the more general vector x − x . (4. 1 G It is possibly better to describe the pair (u ik .7) where G(k I x) = (1/4π x)eik I x . where the ai1. (4.2 ).2 δ(x − x 1.2 are constant vectors. is related to G G G G ˆ u i by u i = a k u ik . jk .
where the azimuthal axis is chosen in the direction x and ξ is the azimuthal angle.6) in all three spatial variables and solving the resulting algebraic equations give ∗ G ui =− 2 2 ˆ ˆ 1 k L − k 2 a i + 1 − cT /c2 ki kk a k L . Note that this is identical. and we are left with evaluating II = −i (2π)2 x ∞ −∞ keikx dk.36). 2 Fourier transforming (4. all that we have to do is evaluate integrals of the form II = 1 (2π)3 ∞ −∞ eik·x dk. but have used a twodimensional one for the very similar problem of constructing a planewave representation of a spherical wave in (2.14) Recall that ei(k I x−ωt) is an outgoing wave. Thus to satisfy the condition that waves be outgoing as x → ∞. Therefore. k2 − k2 I (4. to the coordinate transformation (2. to transform (4. in transform space. and with some rewriting. The transformation is deﬁned as k1 = k sin ξ cos ν. ν).10) Note that ∂i in physical space transforms to iki in transform space and vice versa.33) and (2. k3 = k cos ξ. k2 = k sin ξ sin ν. we arrive j at ∗ G u ik = 1 ki k k 2 2 cT k T k2 1 1 − 2 2 2 − kL k − kT + k2 δik . From the deﬁnition of u iG . apart from the change in symbols. (4. the contour of the integral must pass above the 2 Contrast how we have used a threedimensional transform in this problem.11) can then be expressed as II = 1 (2π)3 ∞ 0 0 π 0 2π k 2 sin ξ ikx cos ξ e dkdξ dν.12) The integral (4.9) where k 2 = k · k. 2 − kT (4.37) used to construct an angular spectrum representation of a spherical wave.3 Green’s Tensor 59 Proof. ξ.4. k2 − k2 I (4.11) we introduce. .13) The integrations over the angles are readily done.11) To evaluate (4. a spherical coordinate system (k. k2 − k2 I (4. 2 2 2 k2 − k2 cT kL − k T (4.10) to physical space.
Problem 4. gives an outgoing wave and hence gives (4. The integral representation of the Hankel function H0 (kρ) (1) G (Sommerfeld.60 4 Green’s Tensor and Integral Representations pole at −k I but below that at k I . x2  = ρ sin θ. where the integral is convergent. The equation to be solved3 is G G ∂α ∂α u 3 + k 2 u 3 = −δ(x). . 1964b) is such that u 3 = (i/4)H0 (kρ).3).6).3. x2 ).18) How are the branches of the radical k2 deﬁned? Next introduce the transformations. Then closing the contour in the upper half of the k plane.19) Most importantly show that C is a contour beginning near π − i∞ and end(1) ing near i∞.8) with G(k I x) = I I . k1 = k T cos ξ. and show that (4. x1 = ρ cos θ.16) (4. antiplane G shear Green’s function. Thus (4. (4. 3 The righthand side of this equation should be multiplied by c−2 . if the reader wants its dimensional form to be identical to (4. The corresponding Green’s stress τiG is calculated by using jk G τiG = ci jlm ∂l u mk . u 3 .1 TwoDimensional Green’s Function In this problem we ask the reader to calculate the time harmonic. G Show that u 3 can be expressed as G u 3 (x) = i 4π ∞ −∞ ei(k1 x1 +k2 x2 ) dk1 . k2 (4.3. At inﬁnity u 3 represents an outgoing wave. jk (4.7) follows. These last two equations are simply restatements of (1. cited in Section 2.18) reduces to G u 3 (ρ) = − i 4π eikT ρ cos(θ −ξ ) dξ. In fact it will turn out to be identical to the angular spectrum representation of a cylindrical wave. C (4.17) G where x has components (x1 .15) where ci jlm = λδi j δlm + µ(δil δ jm + δim δ jl ).
4. Approximating x − x  by region for which k I x  using the law of cosines gives ˆ x − x  ∼ x − ( p · x ). the center of compression.26) Each of the above expansions is carried out only to O[(k I x)−1 ] in the amplitude ˆ while the additional term (ik I p · x ) is retained in the exponential terms. For future work we shall need the farﬁeld results. 1980). ˆ (4.1 Notes 61 1.25) (4. e pi pk 4π x c2 L 1 eikT x −ikT p·x ˆ ˆ ˆ e (δik − pi p k ) . (4. using (4. p = x/x. namely f = −c2 F(t)∂x L δ(x) p. with x replaced by x − x . (4.27) ˆ where x is set to zero.20) With this approximation. we have G GL GT ˆ ˆ ˆ τ jik p j ∼ τ jik p j + τ jik p j .3.15) and (4.21) where G u ik L = 1 eik L x −ik L p·x ˆ ˆ ˆ . The farﬁeld is that k I x and k I x 1.24) where GL G ˆ τ jik p j = ik L (λ + 2µ)u ik L . splits into two parts.23) Further. . I = L or T . Consider a special (threedimensional) body force. and p is deﬁned in (4. (4. ˆ 4π x 2 (4. Here we consider one such construction. There are a number of very localized sources that can be constructed from point forces and their derivatives (Hudson. x = x.20). As before. 2.7). (4. The character of a waveﬁeld is more sharply determined by its phase than by its amplitude. namely G G GT u ik ∼ u ik L + u ik .3 Green’s Tensor 4.22) GT u ik = (4.16). GT GT ˆ τ jik p j = ik T µu ik . 2 4π x cT (4.
29) The center of compression can be thought of as a small uniformly pressurized cavity or as three dipoles without moment. (x1 . for a ﬁnite t the surface ∂Rx . x2 ).2 A Causal Green’s Function Problem 1.62 4 Green’s Tensor and Integral Representations The scalar potential ϕ satisﬁes 1 δ(x) 1 ∂x (x 2 ∂x ϕ) − 2 ∂t ∂t ϕ = F(t) x2 4π x 2 cL (4.28). identify f (t). 4.4 Principle of Limiting Absorption Figure 4. (1) How do we determine unambiguously which waves are outgoing from these sources? (2) In the arguments that follow we shall ask that the integrals over ∂Rx vanish or approach some ﬁnite known value as x → ∞. This particular source is useful because it excites only compressional waves. Begin by showing that xϕ = f (t − x/c L ) is a solution. It is a large spherical region with radius x. therefore. subject to the condition that the wave be outgoing from the source. will eventually pass into a region that the waveﬁeld has not reached. this issue does not arise in the timedependent case because any outgoing waveﬁeld will propagate toward inﬁnity at a ﬁnite speed and. However.29) is the freespace causal Green’s function for (4. Problem 2.27) is a threedimensional body force.28) and the vector potential ψ = 0. In two dimensions. for x = 0. (4.1 shows a typical region Rx within which we shall work. Find the response ϕ for a twodimensional center of compression or line of compression. How do we ensure that this happens? As we indicated in Section 4. S and R.28) if F(t) = δ(t). centered at the origin. Show that (4.29) is the solution to (4. and. Two questions arise. the term in brackets becomes δ(x)/(2π x). as x → ∞. is ϕ=− x 1 F t− . taking the limit as → 0. The solution to (4. Then integrate the equation of motion. will be sources of waves.28) over a small sphere of radius . Equation (4. satisfying the condition that the wave be outgoing from its source.28). Here (4. Problem 4. 4π x cL (4. Both the bounded subregions contained within Rx .1. in .
In several of the arguments x → ∞. there is no initial wavefront because we have imposed no initial conditions.1.14). recover the result for the case of real ω. This gives the missing initial conditions. has a radius x = x. We solve this problem by demanding that ω = ω0 + i . (4. (4. are consistent with this principle. so that the waveﬁeld is forced to zero. This device of setting ω = ω0 +i to determine which waveﬁelds are outgoing as x → ∞ and hence to select which particular solutions to a problem give outgoing waves is called the principle of limiting absorption. S and R. where k I 0 = ω0 /c I . in a timeharmonic problem. t will always be taken as ﬁnite and. 4 Here we are invoking the law of permanence of functional equations (Hille. That is. Stated somewhat differently.4. limx→∞ f (x)ei(k I x−ωt) = 0 provided f (x) remains bounded. for t ﬁxed. note that the wavenumber k I = ω/c I . ˆ the timeharmonic case. we must use the exponential term eik I x in combination with e−iωt . Note that the sign conventions leading to the evaluation of the contour integral. while the second is a scatterer. In Proposition 4. where ω0 and > 0 are real. Hence. as t → ∞ the waveﬁeld becomes unbounded.5 we shall ﬁnd that this choice of ω and hence of k is needed to ensure a unique solution. and reject e−ik I x as a possible solution. often never appears explicitly in the calculation. 4. for x ﬁxed. Moreover. The large spherical region Rx . Once we have completed our calculation we may invoke analytic continuation4 and. The ﬁrst contains sources that excite an incident waveﬁeld. with surface ∂Rx . That is. 1973). and all previous contour integrals. It in no sense indicates a temporal instability. In particular we have implicitly used this principle when evaluating the contour integral.14). k I = k I 0 + i /c I .4 Principle of Limiting Absorption 63 Fig. Therefore. we are imagining that the waveﬁeld started abruptly at t = −∞ and thus it now ﬁlls all space. if we want the waveﬁeld to go to zero as x → ∞. lim t→−∞ f (x)ei(kx−ωt) = 0. by letting → 0. Within are two bounded subregions. . The unit normal n points out from Rx\R. with surface ∂R. Though.
u ik (x − x ) = u ik (x − x). we select as reciprocating waveˆ G ˆ ﬁeld 1 the triple ( f i . Hence we are lead to (4.7). where u ik is given by (4. Starting with the bounded region Rx .31) G where u ik is given by (4. the solution u i can be represented as u k (x) = S G f i (x )u ik (x − x ) d V (x ). / (4. u ik (x−x )a k . S is a bounded subregion containing time harmonic. x ∈ S. Proof. as x → ∞. τiG (x− jk G ˆ x )a k ].22)–(4.26) to show that.32) where Ik = ∂Rx G G ˆ u ik (x − x )τ ji (x) − u i (x)τ jik (x − x ) n j d S(x). (4. (4. f i (x) = 0. (4. (4. Consider the region Rx . but S present. Then (4.32) reduces to ˆ u i (x )a i = S G ˆ f i (x)u ik (x − x )a k d V (x).2) becomes Rx G ˆ ˆ ˆ δ(x − x )u i (x)a i − f i (x)u ik (x − x )a k ρd V (x) = Ik a k .3.64 4 Green’s Tensor and Integral Representations 4. 4. that excite the waveﬁeld u i in Rx . (4. f i . u i is the solution to (λ + µ)∂i ∂k u k + µ∂ j ∂ j u i + ρω2 u i = −ρ f i (x). after a relabeling of the independent and integration variables. shown in Fig.31). Ik ∼ 1 eik L x c2 4π x L + ∂Rx ˆ ˆ ˆ ˆ [τ ji n j − ik L (λ + 2µ)u i ]n i n k e−ik L n·x d S(x ) ˆ ˆ ˆ ˆ [τ ji n j − ik T µu i ](δik − n i n k )e−ikT n·x d S(x ). Returning to the question of how Ik behaves. with the region R absent. u i .30) The waveﬁeld satisﬁes the principle of limiting absorption as x → ∞.7) and also satisﬁes the principle of limiting absorption. Therefore (4. we use the asymptotic results of (4.33) We shall subsequently show that limx→∞ Ik = 0. τi j ) and as 2 the triple [δ(x−x )a i .34) G G By inspection. body forces per unit mass.5 Integral Representation: A Source Problem Proposition 4. After ∂Rx is sent to inﬁnity.1.35) 1 eikT x 2 cT 4π x ∂Rx .
1982). The principle thus serves to determine the solution uniquely. 4. integrals such as Ik can be shown to approach zero without having to make detailed asymptotic estimates of how they behave as x → ∞. through its use. 4.5.37) provided S is bounded. In this case S is said to be a compact source because the support for f i (x) is compact or bounded. This is one way of stating the elastodynamic radiation conditions. (4. recall that we demanded that the waves be outgoing.36) x→∞ ∂R x lim ˆ ˆ ˆ [τi j n j − ik T µu i ](δik − n i n k )2 d S(x ) = 0. reexamining the argument leading to the Green’s tensor. Therefore we must choose the solution that behaves as eikx and not e−ikx . We next perform a slight of hand whereby we imagine that the subregion S . Within it we imagine that the total waveﬁeld u it = u ii + u i . 4.31) asymptotically approaches the form u i ∼ Ai (x) eik L x eikT x + Bi (x) . provided it is compact. When the principle of limiting absorption is invoked. When thinking through the argument just given. the waveﬁeld must evolve into two separate spherical waves with radiation patterns given by Ai and Bi . First. Note that they would only be satisﬁed by an outgoing wave.1. where u ii is the waveﬁeld excited by the sources in S in the absence of the subregion R and u i is that scattered by R. we now ask that the waveﬁeld satisfy the principle of limiting absorption and thus approach zero as x → ∞.6 Integral Representation: A Scattering Problem 65 Note that we have relabeled the variables from those used in (4.. In place of the phrase outgoing. 3.4. Second. kL x kT x x → ∞. 2. That is one having the form f (x)eik I x .6 Integral Representation: A Scattering Problem We continue to consider the region Rx shown in Fig.33). Note that (4. From this it is possible to show that limx→∞ Ik = 0 (Achenbach et al. limx→∞ Ik = 0. note that two issues have been settled by using the principle of limiting absorption. we could have demanded that x→∞ ∂R x lim ˆ ˆ [τi j n j − ik L (λ + 2µ) u i ]n i 2 d S(x ) = 0. Irrespective of the geometry of the source.1 Notes 1. (4. As an alternative to using the principle of limiting absorption.
Nevertheless. u i satisﬁes the principle of limiting absorption as x → ∞. Using (4. Moreover. In fact we have used the incorrect Green’s tensor for this problem. Lastly. by the principle of limiting absorption. i i ˆ ˆ but τ ji n j = −τ ji n j . u i can be represented as u k (x) = − 1 ρ G i G ˆ u ik (x − x )τ ji (x ) + u i (x )τ jik (x − x ) n j d S(x ). (4. but containing the boundary points ∂R.6. It follows then that as the radius x → ∞. Proposition 4. We should have calculated a G ˆ Green’s tensor for which τ jik n j = 0 on ∂R.38). u ik (x − x )a k . Then we deﬁne the region Rx\R as that contained in Rx excluding R. Select as reciprocating waveﬁeld 1 the triple (0. G τ jik . where τ ji is known. 2. It is useful at this point to deﬁne R as an open region so that it does not contain its boundary points ∂R. u i is unknown. Note.15). We now formulate a boundaryvalue problem for u i .2) and proceeding jk as we did with Proposition 4. This is usually far from easy to do. the second argument of the combination (x − x ). with ti = −tii on ∂R acting as a source. we consider R to be an empty cavity and ask that the traction vanish on ∂R.38) ∂R G G The prime given τ jik indicates that the x is differentiated when τ jik is calculated by using (4. we obtain (4. in (4. Note that (4. τiG (x − x )a k ].38). (4. that with respect to x .39) ∂R Using the boundary condition on ∂R. with the righthand side set to zero. Within Rx \R the timeharmonic waveﬁeld u i is a solution to (4.30). Consider the region Rx shown in Fig. (4.38) forms the starting point for many useful approximations.3. Proof.1 with the region S absent.1 Notes G is calculated from u ik (x − x ) by differentiating 1.66 4 Green’s Tensor and Integral Representations is removed from Rx so that the scattered waveﬁeld u i does not interact with the sources in S. 4.4. relabeling the independent and integration variables and noting that again limx→∞ Ik = 0. On ∂R.38) is not a solution to the boundaryvalue problem because the integral contains the unknown u i evaluated on the boundary. but R present. 4. u i . τi j ) and as 2 the G ˆ ˆ ˆ triple [a i δ(x − x ). we arrive at ˆ ρu i (x )a i = G G ˆ ˆ ˆ u ik τ ji − u i τ jik a k n j d S(x) + Ik a k .
Kevorkian (1993) gives an introduction to how to examine the consequences of these singularities within his discussion of integral equations in potential theory. In forming the integral equation care must be taken because. the elastic halfspace is described by {(x1 . as Problem 4. Divide the total particle displacement u t3 into the sum of two parts. Outline of the Solution to Problem 1 1. u s = u r + u 3 . namely. Show that u 3 satisﬁes the timeharmonic wave equation and the following boundary conditions along x2 = 0: u 3 = −2A on x1  < a. For x1  > a the surface is free of traction.6 Integral Representation: A Scattering Problem 67 3. and their solution lends insight that direct numerical methods often do not. One book describing the numerical solution of integral equations is that by Delves and Mohamed (1985). x2 x1 . Problem 2. Problem 4. For the same incident wave. while within a slot x1  < a the surface is free of traction. −∞ < x2 ≤ 0}. as x approaches the surface. Now assume that the region x1  > a is clamped by rigid sheets.4.3 Integral Equation Problems These are two dimensional. An antiplane shear wave u i3 = Aeikx2 (4. An elastic halfspace is clamped at the surface x2 = 0 by a rigid strip over x1  < a. In both the problems that follow. in this simpler context. where u r is the plane 3 3 3 wave reﬂected from the surface with no strip present and u 3 is that caused by the presence of the rigid strip. x2 ) − ∞ < x1 < ∞. and ∂2 u 3 = 0 on x1  > a. several of the formulas given previously. x2 ). where u s is the scattered waveﬁeld. namely. determine an integral equation for the scattered waveﬁeld. antiplane shear problems designed to ask the reader to derive directly. The integral representation for u i (x) can be used to construct an integral equation. Problem 1. How does u 3 behave at ∞? At ±a? 2. Integral equations are sometimes easier to solve numerically than are differential ones.40) is normally incident to the rigid strip and adjacent free surface. the Green’s tensors become singular. This forces the particle displacement at the surface to go to zero there. Use the method of images to ﬁnd the antiplane shear. for the halfspace satisfying a homogeneous boundary . Green’s function G u 3 (x1 . Determine an integral equation for the scattered waveﬁeld.3 suggests. u t3 = u i3 + u s . Then divide this latter 3 3 waveﬁeld again into two parts.
x2 ) = − a −a G u 3 (x1 .1. 4.5. In particular use a Green’s function that satisﬁes u 3 = 0 at x2 = 0. making appropriate changes G where necessary. antiplane shear waveﬁelds. (4.43) in Rx \R. ρ. this particular equation can be reduced to one that is no more singular than that of the ﬁrst problem (Sommerfeld.41) where I (x1 ) = ∂2 u 3 (x1 . Consider two timeharmonic waveﬁelds (u i1 . 3. and the principle of limiting absorption as x → ∞. Derive the reciprocity identity for two reciprocating.1 No Edges Proposition 4. Use the boundary conditions satisﬁed by u 3 to derive the integral equation 2A = i 2 a −a (1) H0 (kx1 − x1 )I (x1 )d x1 . 4. (4.7 Uniqueness in an Unbounded Region 4. Consider the region Rx shown in Fig.1 useful. in [−a. The force f i = 0 for x ∈ S but is zero elsewhere. 4. (4.3 and 4.4 was that for a region without boundaries and not the correct Green’s tensor for the scattering problem just discussed. While one can work with singular integral equations. possessed by the integrand.2 and used in Propositions 4. 0)I (x1 )d x1 . The tractions on ∂R are equal. and µ are 1 2 ˆ ˆ real parameters. λ. the Green’s tensor derived in Proposition 4. Note that what is needed is the Green’s function for this problem. The . 0) and the prime indicates that the derivative is taken with respect to x2 . After I (x1 ) is found.42) Note that I (x1 ) is the unknown to be solved for. Note that on this occasion the integrand is more singular than in the previous case. τi1j ) and (u i2 . τ ji n j = τ ji n j . the representation (4. that is. a].2 ∂ j τ ji + ρ f i + ρω2 u i1.7. Repeat the above steps for the second problem. The reader may ﬁnd the results of Problem 4. τi2j ) that satisfy 1. In contrast.68 4 Green’s Tensor and Integral Representations G condition ∂2 u 3 = 0 at x2 = 0. Hint. Taking as one reciprocating waveﬁeld the Green’s function and as the other u 3 . show that u 3 (x1 .2 = 0. 1964a). Describe the nature of the singularities.41) is used to calculate u 3 throughout the region of interest. x2 x1 .
3). Thus the waveﬁelds must have a certain measure of continuity on the surfaces. Take the scalar product of (4. 1989. The 5 Note that the analytic continuation of a function is unique. τi j ) satisﬁes ∂k τkl + ρω2 u l = 0 (4. Set u = u1 − u2 and τi j = τi1j − τi2j . The integrand is either everywhere zero or positive. ω = ω0 + i so that both waveﬁelds go to zero as x → ∞. (u i∗ . when the unique solution is determined by setting ω = ω0 + i . that uniqueness is not lost by taking the limit → 0. In addition.7.2 Edge Conditions To prove uniqueness (or.2 a sketch of the region R.45) Integrating this result gives [(ω2 )∗ − ω2 ] ρu l∗ u l d V = ∗ ˆ (τ ji u i∗ − τ ji u i )n j d S. Then (u i . If positive somewhere. . is given.4. Moreover. gives ∗ [(ω2 )∗ − ω2 ]ρu l∗ u l = ∂ j (τ ji u i∗ − τ ji u i ).1 and hence all the subsequent propositions). to prove Proposition 4. ω∗ = ω. 1970). As was done previously. the superscript asterisk indicates the complex conjugate.44) with u i∗ and the scalar product of the complex conjugate of (4. so that the lefthand integral must equal zero. In accord with the principle of limiting absorption. In the simpler proofs one asks that the partial derivatives be continuous apart from ﬁnite jumps (Courant and John. Therefore the righthand side equals zero. (4. 4. Subtracting one from the other and using (1. it was essential that we be able to use Gauss’ theorem. giving a contradiction. 4.44) in Rx\R. with a lancetshaped singularity whose edge is perpendicular to the page. the stress–strain relation. then the integral cannot be zero.46) Rx\R ∂R∪∂Rx ˆ The righthand integral is zero because τi j n j = 0 on ∂R and the integral over ∂Rx goes to zero as x → ∞. Kellogg. in fact. Therefore u i and hence τi j must be zero throughout Rx\R5 . In Fig. The waveﬁeld is singular there. τi∗j ) satisﬁes the same equation with ω2 replaced by (ω2 )∗ . Then the two waveﬁelds are identically equal in Rx\R.44) with u i . by the principle of limiting absorption. Hence.7 Uniqueness in an Unbounded Region 69 waveﬁelds are sufﬁciently continuous on ∂ R that Gauss’ theorem may be used. by the principle of limiting absorption. Proof. in cross section. (4.
The surface ∂Q = ∂Rx ∪ ∂(R ∪ R ). 4. we surround it with a region R whose surface is ∂R . Moreover. where α < 1.46) is written as [(ω2 )∗ − ω2 ] ρu l∗ u l d V = ∗ ˆ (τ ji u i∗ − τ ji u i )n j d S. in cross section. an element of surface area S of ∂R is O( ). To investigate what happens at this edge. Now (4.48). The subregion R is drawn. Next we shall work through a speciﬁc case. Gauss’ theorem may be applied to the region Q = Rx\(R ∪ R ). it is more usual to enforce the edge condition by asserting the nature of the singularity that will be permitted at the edge so that (4. to satisfy (4.48) being satisﬁed. particle displacement is usually bounded even continuous. A condition such as this is called an edge condition.47) goes to zero subject to (4. we demand α > 0.70 4 Green’s Tensor and Integral Representations Fig. with a lancetshaped singularity whose edge is perpendicular to the page. . 4.48) →0 ∂R Then the righthand side of (4.47) Q ∂Q To complete the uniqueness proof we must ask additionally that lim ˆ τ ji u i∗ n j d S = 0.1 and thereby recall that R and R are both contained within Rx . However. However. At such an edge the waveﬁeld is singular. for an edge. And the uniqueness argument proceeds as before. but its derivatives are singular. (4. (4.2. Thus Gauss’ theorem is inapplicable near this edge without a certain amount of care being taken. The reader should look back at Fig. Therefore. This region does not intrude into (is disjoint from) R.48) is satisﬁed. Again it is useful to deﬁne R and R as open regions that do not contain their boundary points. It is important to realize that this singularity must be speciﬁed before the scattering problem is solved and is as important in achieving a unique solution as is specifying a boundary condition or invoking the principle of limiting absorption. typically u i ∼ O( α ) and τi j ∼ O( α−1 ). The edge is surrounded by a cylindrical region R with a radius . Both regions are contained within Rx .
3 shows the region R collapsed to a inﬁnitesimally thin. x3 ) − ∞ < x1 ≤ 0. upper and lower. our previous arguments are readily extended to this case.3. we take the waveﬁeld to be an antiplane shear one. We wish to ﬁnd the waveﬁeld in the neighborhood of the edge to determine the nature of the waveﬁeld’s singularity there. Asymptotically expanding w as w(ρ. For simplicity. 4. It is surrounded by a cylindrical region R with a radius that is smaller than a wavelength. θ) ∼ (k )α n≥0 wn (ρ.7.3. semiinﬁnite slit or crack deﬁned by {(x1 . 0 < α < 1. The edge is perpendicular to the page.15). the upper and lower surfaces of the crack. namely 2 (1/r )∂r (r ∂r u 3 ) + (1/r 2 )∂θ u 3 + k 2 u 3 = 0. in the timeharmonic approximation and expressed in polar coordinates. r = 0. Therefore we should use a coordinate expansion in kr for u 3 beginning with (kr )α . θ) and w. ∂θ u 3 = 0. −∞ < x3 < ∞}. Setting ρ = r/ and w = u 3 /U (U is a maximum particle displacement). semiinﬁnite slit or crack and ∂R becomes its two surfaces. However. (4. We do not intend to solve a global problem but merely to explore possible solutions in the neighborhood of r = 0. where k 1. θ)(k )n . The subregion R has been collapsed to a inﬁnitesimally thin.7 Uniqueness in an Unbounded Region 71 Fig.50) . We seek a solution to (1. (4. Figure 4.49) in terms of (ρ. At θ = ±π. 4. a bit more insight is gained by introducing the length scale shown in Fig. Though R is no longer bounded.3 An Inner Expansion We now consider a speciﬁc case of a scatterer with an edge. we reexpress (4. with the assumption that kr → 0. 4. extending inward and outward to inﬁnity.49) where k is the wavenumber.4. Its surface ∂R. is free of traction.
(4. We shall continue with this analysis in Section 5. . if we imagine that the radius x → ∞ and the region S is absent. Subject to the boundary condition stated previously. is penetrable. in this closing section we extend slightly these results to develop an expression for the total waveﬁeld present in an ideal ﬂuid that contains an elastic inclusion.4. in contrast to the empty cavity of Proposition 4. 1991. B = 0 and α = β = 1 .51) The waveﬁeld near the edge is essentially equivalent to the static ﬁeld.48). but this is the minimum such value and hence determines the most singular term allowable. . Larger values of β also give acceptable 2 solutions. Waves enter it. θ) = (Aρ β + Bρ −β ) sin βθ. and then reradiate. The elastic inclusion occupies a region R with surface ∂R. for a problem with the geometry of Fig.50) is called an inner expansion. Figure 4.6. To satisfy (4. This further illustrates the principle that a wave has to propagate several wavelengths. The inner expansion connects the source – in this case the crack tip – with the propagating waveﬁeld. This expansion will contain an unknown constant that can be found by matching it to an outer expansion.72 4 Green’s Tensor and Integral Representations we ﬁnd that the lowestorder term satisﬁes 2 (1/ρ)∂ρ (ρ∂ρ w0 ) + (1/ρ 2 )∂θ w0 = 0. n = 0. This calculation is based on the work of Wickham (1992) and Leppington (1995). To indicate when a position vector x identiﬁes a .3. the outer expansion. An expansion such as (4. Thus ku 3 ∼ (kr )1/2 A sin(θ/2)+ O[(kr )3/2 ] as kr → 0. 4. we must specify.51). freeing itself of its source. 1.52) where β = (2n+1)/2. . 1995). and A and B are undetermined constants. before its propagating character becomes manifest. that ku 3 = O[(kr )1/2 ] as kr → 0 to ensure that we arrive at a unique solution.1 indicates the geometry. This method of analysis is called matching asymptotic expansions (Hinch.8 Scattering From an Elastic Inclusion in a Fluid To convince the reader of just how powerful and general the foregoing results can be. Holmes. has solutions of the form w0 (ρ. a few wavelengths away. 4. We again take the time dependence as harmonic. The unit normal n points out of R (in contrast with our previous convention). 2. (4. (4. when we ﬁrst formulate the problem. reverberate within it. We indicate by ˆ ∂R+ the surface approached from outside R and by ∂R− that approached from within. Therefore. The elastic inclusion.
where p. with the departure of the . ˆ ˆ ˆ ¯ ∂k u s δi j n j = −∂k u ik δi j n j + σ ji n j /λ. the elastic ﬂuid. otherwise 73 (4. x ∈ R. The equations of motion for an elastic ﬂuid are given by those of linear elasticity when the shear coefﬁcient is set to zero. f is the body force per unit mass. giving rise to the scattered waveﬁeld us . Lastly.53) We treat the inclusion as a linearly elastic solid embedded in a linearly elastic (ideal) ﬂuid. x ∈ R.55) Next we imagine that an incident waveﬁeld ui scatters from the elastic inclusion. respectively. arising because the motion is irrotational.56) (4.58) capture the concept that the elastic solid is an inclusion within a host material. k (4. we deﬁne the function χ(x) as χ(x) = 1.4. ¯ pi = ω(ρ − ρ)u i . The equation is thus written as ¯ ¯ λ∂i ∂k u k + ω2 ρu i = −∂ j σ ji − ωpi . As before. Taken together. Across the surface ∂R the normal components of particle displacement and traction are continuous and the tangential components of traction vanish.8 Scattering From an Elastic Inclusion in a Fluid point in R. equals ¯ ¯ ¯ −λ∂k u k . of the elastic ﬂuid. 0. (4. The equation of motion for the particle displacement u is ¯ ¯ ¯ λ∂i ∂k u k + ω2 ρu i = −ρ f i . the acoustic pressure.54) The stress tensor is given by τi j = − pδi j . (4. (4. The elastic ﬂuid is compressible but cannot withstand shearing.55)–(4. namely ∇ ∧ u = 0.58) The symbol [· · ·] indicates the jump in going from ∂R+ to ∂R− . We express these conditions as ˆ ˆ [u sj ]n j = −u ij n j .57) x ∈ R. The total waveﬁeld u(x) = [1 − χ (x)]ui (x) + us (x). (4. The righthand terms are given by ¯ σi j = (λ − λ)δi j ∂k u k + µ(∂i u j + ∂ j u i ). there is another equation of motion. We write the equation of motion for the particle displacement u in the elastic solid in such a way that the solid’s inertial and elastic properties appear as a body force within the region R. The parameters ρ and λ are the density and bulk modulus.
However.4 are used. G ˆ the reader is asked to verify that u ik a k is irrotational.62) R + c2 ∂R− G ˆ − u sj (x )∂k u km (x − x ) n j d S(x ). Our goal is to use Propositions 4. or a slight variation of them. it is the correct Green’s tensor when one considers the ﬂuid as elastic and seeks its ˆ response to a force acting at a single point. ¯ ¯ The wavespeed c = (λ/ρ)1/2 . Next consider the region R. Arguments that combine those used in both Propositions 4. The outcome is u s (x)χ(x) = m 1 ρ ¯ [∂i σi j (x ) + ωp j (x )]u G (x − x )d V (x ) jm u G (x − x )∂k u s (x ) jm k (4. (4. The prime indicates that the derivative is taken ˆ with respect to x . u = us .60) Note that this is not the usual Green’s function for linear acoustics. This satisﬁes an equation analogous to (4.4. to give an integral representation for u as a volume integral over R that contains σi j and pi . The Green’s tensor is one reciprocating waveﬁeld and us the second.61) and (4.61) ∂R+ G ˆ − u sj (x )∂k u km (x − x ) n j d S(x ). provided x = 0.59) It is calculated exactly as outlined in Proposition 4.55) and in the conditions across ∂R. To do so we need the Green’s tensor for the elastic ﬂuid.74 4 Green’s Tensor and Integral Representations properties of the solid from those of the ﬂuid expressed as body force terms on the righthand side of (4. Recall that within R. The unit normal n points out of R rather than out of its compliment.4 and then in 4. We add (4.58). The outcome is u s (x)[1 − χ(x)] = −c2 m u G (x − x )∂k u s (x ) jm k (4. (4. (4. Again the reciprocity relation is applied.3.2 and is given by G u ik = 1/k 2 c2 [−∂i ∂k (eikx /4π x) + ∂m ∂m (δik /4π x)].2–4. Next we apply arguments identical to those used in Propositions 4.3 and 4. which function is the response to a mass ﬂux at a single point. The Green’s tensor is one reciprocating waveﬁeld and us the second. but now within R itself.6). The reciprocity relation is applied to the complement of R. namely G G ¯ ¯ ¯ λ∂i ∂k u km + ρω2 u im = −ρδim δ(x). For a point force in the direction a.62) to obtain the following .
First note that the departure of the properties of the elastic solid from the host elastic ﬂuid appears as a body force ˆ ˆ term in the integral over R. are present in the integral over ∂R. (4.64) This in itself is a interesting result. Second note that the jumps [u sj ]n j and [∂k u s ]δi j n j .64) to obtain u m (x) = u im (x) + 1 − ρ ¯ 1 ρ ¯ [∂i σi j (x ) + ωp j (x )]u G (x − x )d V (x ) jm − x )σi j (x ) n j d S(x ). Noting the divergence term in the volume integral. this is a remarkable expression. u s (x) = m 1 ρ ¯ [∂i σi j (x ) + ωp j (x )]u G (x − x )d V (x ) jm u G (x − x ) ∂k u s (x ) jm k 75 R − c2 ∂R G ˆ − u sj (x )∂k u km (x − x ) n j d S(x ).8 Scattering From an Elastic Inclusion in a Fluid representation for us .58). k given by (4. jm This is a very satisfying outcome. we make one more application of Gauss’ theorem to give u m (x) = u im (x) − 1 ρ ¯ σi j (x )∂i u G (x − x ) jm (4.63) from (4.62) are also examples of extinction theorems. (4.6 and it indicates that the surface integral vanishes for observation points x outside R.65) R ∂R G u im (x We are almost done. through successive applications of the Green’s tensor in combination with the reciprocity relation. but we are still left with a surface integral.61) and (4.63) As it stands. We now subtract (4. We have. shown that the 6 The expressions (4. (4. Now we imagine that the region R contains only the incident waveﬁeld ui and use the reciprocity relation along with the Green’s tensor to write u im (x)χ(x) = c2 G u im (x − x )∂k u ik (x )δi j ∂R G ˆ − u ij (x )∂k u km (x − x ) n j d S(x ).66) R − ωp j (x )u G (x − x ) d V (x ).4. It is called an extinction theorem. The superscripts ± on ∂R have served their purpose and are omitted from now on. .
New York: Frederick Ungar. and Mohamed.D.T. Vol. Partial Differential Equations. 1982.66) in (4. Hinch. Handbook of Radiation and Scattering of Waves. Foundations of Potential Theory. New York: Academic. 22–27 and 34–38. 84–101. A. 1958. Ray Methods for Waves in Elastic Solids. New York: Chapman and Hall. Perturbation Methods. 75–105.. One uses (4.A. Leppington. Thus this expression is only a representation and not a solution to the problem. 1964a. Amsterdam: NorthHolland. it can be made the basis for deriving a system of integral equations for σi j and pi . pp. Sommerfeld. A. J. pp. R. 31–36. . Nondestr. pp. II. 1980. Moldauer. Dissertation. VI. the elastic ﬂuid. J. London: Academic. Lectures on Theoretical Physics. 1985. Cambridge: University Press. The Excitation and Propagation of Elastic Waves. 84–121. Gautesen.K. 1989. 1992. 1991. Wave Propagation in Elastic Solids. and McMaken. Computational Methods for Integral Equations.D. 47–104. 1995.76 4 Green’s Tensor and Integral Representations total waveﬁeld u outside R equals the incident waveﬁeld ui plus a scattered waveﬁeld us whose source is the departure of the inertial and elastic properties of the solid inclusion from those of its host. Courant. New York: Cambridge. Kevorkian. pp. Vol. Analytic Function Theory. pp. pp. 1973. pp.G.66) is not a solution to the boundaryvalue problem for us . Wickham. However. New York: Springer. G. L. E.H. 1993. J.38).M. Introduction to Perturbation Methods. 273–289. II. which cannot be calculated without knowing u within R. 1995.. However. pp. J. Optics.56) and (4. 106–109. Achenbach. Friedlander. 96–110. F. Translated by O. H. 1964b. Hudson. Just as with (4. Sound Pulses.G. pp. Delves.D. Cambridge: University Press. LaPorte and P. 64–75 and 94–99. Holmes.R. pp.57) to enforce selfconsistency (Leppington. IV.A. J. Vol. 1973.J. 1970. New York: Academic. 1995. M. J. O. 11: 199–210. Sommerfeld. References Achenbach. Boston: Pitman. 52–101. Partial Differential Equations in Physics. F. E. Manchester: Ph.L. New York: Springer. pp. Introduction to Calculus and Analysis. Vol. Cambridge: University Press. Lectures on Theoretical Physics. New York: Chelsea. (4. deHoop. A. These are σi j and pi . Eval. the integral contains unknown terms. The University of Manchester. Hille.J. Kellogg. S. Translated by E. and John. 1995). pp. A polarization theory for scattering of sound at imperfect interfaces.D. A. Straus. SemiInﬁnite Thick Elastic Plate. 597–602. The Scattering of Sound by a FluidLoaded.
Planewave spectral techniques are used and the resulting integrals are approximated by the method of steepest descents. 77 (5. plane. antiplane shear wave by a semiinﬁnite slit or crack.5 Radiation and Diffraction Synopsis Chapter 5 summarizes the basic propagation processes that are encountered when studying radiation or edge diffraction. The Cagniard–deHoop method is used to invert the integral transforms. This problem is solved exactly by using the Wiener– Hopf method and approximately by using matched asymptotic expansions. An Appendix describing the reduction of the diffraction integral to Fresnel integrals is included. Lastly. On the surface x2 = 0. inplane radiation. At the origin a line load is applied to an otherwise tractionfree surface. The equations of motion are given by (1. The line load is a tangentially acting force very localized in x1 and directed from −∞ to ∞ in the x3 direction. we extend our knowledge of planewave interactions by calculating the diffraction of a time harmonic.1) . This method is discussed in detail. from a twodimensional center of compression buried in a halfspace. antiplane radiation excited by a line source at the surface of a halfspace.1 Antiplane Radiation into a HalfSpace We consider an elastic halfspace. We begin by calculating the transient. 5. The x1 coordinate stretches along its surface and the positive x2 coordinate extends into the interior. We then return to considering how plane waves and a knowledge of their interactions can be used to construct more general waveﬁelds. Three problems of progressive difﬁculty are studied.15) combined with (1. µ∂2 u 3 = −µδ(x1 ) f (t).14). We calculate the time harmonic.
(5. x2 . p) = ∞ −∞ ¯ e− pαx1 u 3 (x1 .43). x2 . Arguably. ∗ ¯ u 3 (α.2) is ¯ u 3 (x1 .1. α is complex. the amplitude term being adjusted almost as an afterthought. Moreover. that is.2) where the overbar indicates the temporal transform. the subscript T is dropped.3) where ≥ 0. of the spatial inverse transform into a form that allows one to immediately identify the inverse temporal transform. and subsequently the integration contour. Second. p) = p 2πi +i∞ − −i∞ ¯ e pαx1 ∗ u 3 (α. This is not a serious restriction because we shall not need to invert this transform. As we have done previously. The waves are outgoing from the source. (5. if necessary. the simplest way to understand the technique is to use a onesided Laplace transform over time and a twosided one over space.4) . What is most satisfying about the technique is that it shifts the burden of inverting the transform to understanding a mapping that primarily affects the phase term. it is only a slight modiﬁcation of the Fourier transform of (1. we take the twosided Laplace transform over the spatial coordinate. The inverse of (5.78 5 Radiation and Diffraction The halfspace is quiescent for t < 0 so that f (t) ≡ 0 and u 3 ≡ 0 for t < 0. we can analytically continue the transform into the complex p plane. is the mapping of the phase term. 5. with the restriction that the transform variable p remain real and positive. And the transform merely provides a shell within which the mapping lives. The essence of the technique. χ = (s 2 − α 2 )1/2 . x2 . x2 . the onesided transform over time (1. p) d x1 . Though we have not used this transform previously. p) dα.1 The Transforms Cagniard (1962) developed a very instructive way to invert the integral transforms arising in transient wave problems. now referred to as the Cagniard–deHoop technique. DeHoop (1960) popularized Cagniard’s method by making it more readily understood. Applying the transforms over t and x1 leads to the ordinary differential equation 2 ¯ ¯ d 2∗ u 3 d x2 − p 2 χ 2 ∗ u 3 = 0. Note that p is used to scale the spatial transform variable α. (5.31) is taken. First.
inverting in time gives u 3 (x1 . For p that is real and positive. x2 .5. The angle θ1 ∈ (0. The solution must be selected in such a way that the outgoing condition is enforced. p) = f¯( p)/( pχ). Thus. π). so that should the contour be rotated to give a standard Fourier transform. Hence. . 5.5) where the centered asterisk indicates a convolution in t. p) = Next.1.3) gives ¯ u 3 (x1 .8) where (ri . 5.1. because we are now using a Laplace rather than a Fourier transform.4. p) = 2πi +i∞ − −i∞ e pαx1 e− pχ x2 dα. x2 .1 follow. 5. (5.6) f¯( p) 2πi +i∞ − −i∞ e pαx1 e− pχ x2 dα. we are asking that the components of the disturbance decay toward inﬁnity. will give an outgoing wave for x2 > 0 in the integrand of (5. Enforcing the transformed boundary condition at x2 = 0 gives A(α.7) The has been added to the inversion contour in (5. Note that the sign of (χ) varies from quadrant to quadrant. Here χ is given as χ = (r1r2 )1/2 eiθ1 /2 eiθ2 /2 e−iπ/2 . (5.51). as indicated in the caption to Fig.4.1. for (χ) ≥ 0 ∀ α. Note that χ = −iγ .2 Inversion To make further progress we must decide how to cut the α plane. θi ) are deﬁned in Fig. 1 In the absence of propagation. χ (5. where γ is given by (3. the spatial inversion contour in the complex α plane. p)e− pχ x2 .7). x2 . Nevertheless the reasoning is identical to that given in Section 3. Inverting in α by using (5. χ (5. t). t) = f (t) ∗ I (x1 . with p = −iω and ω > 0. almost anywhere.3). The cuts are not the same as those indicated previously. (γ ) ≥ 0 ∀ α and the branch cuts shown in Fig. the contour will pass above and below the branch cuts in such a way that − pχ . 5. x2 . this choice will give us the freedom to distort. This can ¯ be achieved by asking that1 (χ) ≥ 0 ∀ α and taking as the solution ∗ u 3 = A(α.7). and hence (5. 2π) and θ2 ∈ (−π. we are left with inverting the integral 1 ¯ I (x1 .1 Antiplane Radiation into a HalfSpace 79 The slowness s = 1/c is used instead of the wavespeed c.
(5.2 indicates the answer to the former question.1. we ﬁnd that it is simpler to continue examining the α plane. We start by mapping the variable of integration from α to t.9) The variable t traces out a contour in the complex t plane as α ranges from (− − i∞) to ( + i∞). The complex α plane cut so that (χ) ≥ 0 ∀ α. (χ) > 0. Eliminating the parameter t shows that the curve is a hyperbola with asymptotes (α ± )/ (α ± ) = ∓ tan θ.10) We next ask. (5. The original contour is shown by the heavy. we use (5. The parameter t starts at sr . where r = (x1 + x2 )1/2 . 3. However. This contour is the Cagniard–deHoop . and when solved gives the two roots t t2 α± = − cos θ ± i sin θ 2 − s 2 r r 1/2 . Figure 5. because.80 5 Radiation and Diffraction Fig. 5. Setting x1 = r cos θ and 2 2 x2 = r sin θ. In quadrants 1 and 2.7) is convergent throughout the particular Riemann sheet we are working on. The magnitude ri and argument θi for each radical are shown. rather than look at the t plane.7. along what contour in the α plane is t real and positive. dashed line and the new one by the solid line.we set t(α) = −αx1 + χ x2 . and in quadrants 3 and 4.9) to ﬁnd α as a function of t. The new contour has two branches evinced by the subscripts plus and minus attached to the α. where α± = −s cos θ. That is. and can we distort the current contour to that one? The answer to the latter question is yes. A quadratic equation for α is arrived at. (χ) < 0. with p real and positive. without assigning any physical meaning to t. and goes to ∞ along each branch α± . the integral (5. Contrast this ﬁgure with Fig. for x2 > 0.
we rewrite the integral as 1 ¯ I (x1 . however. 5.9) and its inverse (5. 2 (s 2 − α+ )1/2 (5. by inspection.10). x2 . Note how the angle θ is deﬁned.12) Noting that one component of the integrand is the complex conjugate of the other. x2 .6) must still be evaluated. t) = H (t − sr ) 1 π dα+ /dt . p) = 2πi ∞ sr (5.13) where H (x) is the Heaviside function. 2 1/2 dt 2 − α 2 )1/2 dt (s (s − α− ) + (5. A sketch of the original contour (heavy. 2 (s 2 − α+ )1/2 (5. the example just . dashed lines) to the Cagniard–deHoop contour and the angle θ.5. The burden of the inversion has rested with the mapping from the α plane to the t plane by using (5.2. dashed line) and the Cagniard– deHoop contour (solid line) in the complex α plane. namely 1 t sin θ dα± = − cos θ ± i 2 2 2 . x2 .14) ∞ sr e− pt dα+ /dt dt.1 Antiplane Radiation into a HalfSpace 81 Fig. Also shown are the asymptotes (light.7) can now be written as 1 ¯ I (x1 .11) e− pt e− pt dα+ dα− − 2 dt. One last term is needed. p) = π Thus. Recall. contour. that a convolution integral (5. Moreover. I (x1 . dt r r [(t /r ) − s 2 ]1/2 Equation (5.
15) f (t) ≡ 0 for t < 0. where a is a constant.3. It is also of interest to note the form of (5. though all the books on elastic waves cited previously discuss them. Ewing et al.15) is replaced by τ23 = −µ d f /dt [H (x1 + a) − H (x1 − a)]. t) for this case. Determine the particle displacement u 3 (x1 . Problem 5. x2 . x2 . and indeed Cagniard (1962) should be explored for more complicated cases. We limit our discussion to calculating the response of an elastic halfspace to a timeharmonic line of compression (a twodimensional center of compression) by using the planewave spectral approach. to the antiplane traction τ23 = −µ H (x1 )d f /dt (5.82 5 Radiation and Diffraction given is a particularly simple instance of this mapping. x2 ≥ 0}. It is also brieﬂy explored in Problem 5. Using a Tauberian theorem (van der Pol and Bremmer. Consider the halfspace {(x1 . A simple variation to Problem 1 is to consider exactly the same problem except that (5. This technique is described. using Watson’s lemma. It is subjected. It is usually called Lamb’s problem.1980b).1. t) in the interior by using the Cagniard–deHoop technique to invert the transforms. t) for t near sr . The integrand of the inverse spatial transform will have a pole and two branch points. With a bit of exploring.1 HalfPlane and Strip Problems Problem 1. Determine u 3 (x1 .2 Buried Harmonic Line of Compression I Finding the response of an elastic halfspace to a point or line load applied either at its surface or buried just below it is possibly the most studied problem in elastic waves.3. at x2 = 0. by Knopoff and Gilbert (1959) and is used extensively by Harris (1980a. Identify the waves represented by the pole term and by the integral taken along the Cagniard–deHoop contour. (1957) and Cagniard (1962) are primary references to these problems. This is called a wavefront approximation.13). 5. something we discuss in Section 5. It is precisely of the form needed to approximate it.16) . 1950). x2 ) − ∞ < x1 < ∞. within the context of elastic waves. for large p. (5. x2 . we should ﬁnd that in doing so the leading term is determined by the nature of the singularity in the dα+ /dt of the integrand. Problem 2. we can construct an approximation to I (x1 .
The term ϕ i is that excited by the line of compression in the absence of the tractionfree surface and the terms ϕ and ψ are the compressional and shear potentials scattered from the surface. (5.17).18) 2 The radical γ L = (k L − β 2 )1/2 . respectively. so that the divergence condiˆ tion is automatically satisﬁed. where A is a dimensionless constant. and in the equation governing ψ. which will soon be needed.5. The total particle displacement ut = ui + u. γL (5. The radical γT = (k T − β 2 )1/2 . This leads to the complex plane structured as indicated in Fig. The term ϕ i satisﬁes 2 ∂α ∂α ϕ i + k L ϕ i = F0 δ(x)δ(y − h). with a line of compression located at (0. The constant F0 has the dimensions of 2 length squared. the solution is ϕi = −i F0 4π ei(βx1 +γL x2 −h) Cβ dβ . Written here using a different notation.44 and there labeled ξ .37). namely ϕ t = ϕ i + ϕ and ψ t = ψ. ui is given by ui = F0 4π Cβ (β e1 − γ L e2 ) ei[βx1 +γL (h−x2 )] ˆ ˆ dβ . The contour Cβ is also shown.1 and explored further in Problem 4. identical to that discussed in Section 5. 5.2 Buried Harmonic Line of Compression I 83 We consider an elastic halfspace.18) gives the solution to (5. Note that γ L = iγ . . The potentials and their governing equations are given by (1. Then (5. For x2 < h. the incident plus scattered waveﬁelds. Note that (4. ˆ (5. The terms ϕ t and ψ t are the total compressional and shear potentials.1. The vector potential ψ = ψ e3 . and we suppress that dependence. h).19) We next introduce the Sommerfeld transformation β = k L cos α.3 (imagine the branch cuts for γ L and γT lying on top of one another).18) becomes ui = − 2 k L F0 4π C ˆ p 0 (α)eik L p0 ·x eik L h sin α dα. The potentials ϕ and ψ satisfy the same equation. It is sometimes useful to set F0 = A/k L . We ask that by (3. with k L replaced by k T . The source is time harmonic.51) (γ L ) ≥ 0 ∀ β. but with no source term on the righthand side.20) γ L is the radical noted at the end of Section 3. The potentials are divided into two parts.17).2.22).19)–(1.3. is deﬁned in the same way.17) an equation almost identical to (4. γL (5. A source of this kind was introduced in Section 4. where γ is the radical deﬁned 2 2 . The outcome of that problem suggests that the present calculations are most easily begun by working with potentials. γ L = k L sin α ﬁrst introduced in (2.
is shown in Fig.23) .1.2) with θ0 = π/2 + α. The integrand of (5. T .4. Using as the integrands the reﬂected plane.20) is a plane. The complex β plane cut so that (γ I ) ≥ 0 ∀ β. where I = L . ˆ (5. The total scattered waveﬁeld u = us L + usT .84 5 Radiation and Diffraction Fig. 5.21) In writing (5. Here (5. (γ I ) > 0.1. we can construct the scattered waveﬁelds such that the boundary condition is satisﬁed. (5. we have used a notation very similar to that of Section 3. where ˆ ˆ p 0 (α) = cos α e1 − sin α e2 .4. In quadrants 1 and 2. These waveﬁelds are given by us L = − usT = − k L F0 4π k L F0 4π C ˆ p 1 (α)R L (α)eik L p1 ·x eik L h sin α dα. The contour Cβ is shown.21) is identical to (3. as well as the condition that the scattered waveﬁelds be outgoing. (γ I ) < 0. compressional wave incident to the tractionfree surface x2 = 0. as are the Rayleigh poles at ±kr . The α plane. though we now use α as the independent variable.3. 5.20). and in quadrants 3 and 4. compressional and shear waves calculated in Section 3. with the contour C. where the ﬁrst term represents the scattered compressional waveﬁeld and the second the scattered shear waveﬁeld. We discuss its topography further in Section 5.22) C ˆ ˆ d 2 (α)RT (α)eikT p2 ·x eik L h sin α dα. ˆ (5.
the two Rayleigh poles αr and π − αr . (5. when θ0 = π/2 + α and θ2 = π/2 + α. (3. The complex α plane. RT (α) = 2κ sin 2α cos 2α/A+ (α). with its asymptotes at θ1 ± π/2.5.28) .12). are also indicated. for the compressional wave.10)–(3. Expressing the reﬂection coefﬁcients in terms of α and α. we have ¯ ¯ R L (α) = A− (α)/A+ (α).26) (5. Quadrants 1 and 2 of Fig. Note how part of the contour of steepest descents passes onto the other Riemann sheet.24) (5.1 by (3. 5. and the contour of steepest descents Cs .7).4. (5. and the contour C are indicated.2 Buried Harmonic Line of Compression I 85 Fig. ˆ ˆ ˆ ¯ˆ ¯ˆ p 2 (α) = cos α e1 + sin α e2 . with the understanding that c−1 cos α = cT cos α is used to relate ¯ L α to α.27) (5. Moreover. The two reﬂection coefﬁcients R L (α) and RT (α) are also given by (3.3 are mapped into quadrants 1 and 2 .25) These are identical to the unit vectors deﬁned in Section 3. ˆ and ˆ ˆ ˆ d 2 (α) = e3 ∧ p 2 (α). The branch cuts with branch points αT and π − αT .4). The independent variable is −1 taken as α. The several unit vectors are given by p 1 (α) = cos α e1 + sin α e2 . 5. the saddle point θ1 . This is indicated by the dashed portion of the contour.6) and ¯ (3.
implying that the scattered compressional wave appears to come from a virtual line of compression at (0. scattered potentials are given by ∗ ϕ = (β)eiγL x2 and ∗ ψ = (β)eiγT x2 . Problem 5.23) is complicated by the term eik L h sin α . 2γ L (5. ˆ e Deﬁning X = x1 e1 + (x2 + h)ˆ 2 . The function R(β) is the Rayleigh function. Note that [d2 ∗ ϕ i ]h+ = F0 . . Accordingly.29) and κ = c L /cT . The scattered shear wave does not appear to come from a virtual line source. multiplied by ω4 to account for the change in scaling.47).22) as eik L p1 ·X . R(β) 2 (5.2 Lamb’s Problem The problem of a buried line of compression is often solved differently from the method we have just used. We indicate subsequently that its virtual source is a caustic. x2 ) = − i F0 iγL x2 −h e .30) (5. Continue to work with the potentials.86 with 5 Radiation and Diffraction ¯ ¯ A∓ = sin 2α sin 2α ∓ κ 2 cos2 2α (5.31) Take a Fourier transform in x1 of (5. To apply the boundary conditions. the boundary conditions at i i x2 = 0 become τ12 = −τ12 and τ22 = −τ22 . the reader will need τ12 and τ22 expressed in terms of the potentials. show that τ12 = µ(2∂1 ∂2 ϕ + ∂2 ∂2 ψ − ∂1 ∂1 ψ). The phase of the integrand of (5. Thus show that h− ∗ i ϕ (β. The present problem indicates this more common method. −h). Enforce the boundary conditions in the transform domain to show that (β) = i F0 iγL h e 2γ L 2 k T − 2β 2 − 4β 2 γ L γT . Are the radicals γ I identical to those deﬁned previously? Expressing the traction terms as τitj = τiij + τi j . (5.33) and ﬁnd a similar expression for (β).32) Show that the transformed. τ22 = λ∂α ∂α ϕ + 2µ(∂2 ∂2 ϕ − ∂2 ∂1 ψ). we can write the phase of the integrand of ˆ (5. (3.17) and solve the ordinary differential equation in x2 .
positive and large. u 1 = u s L + u sT . .4. 1 5.3 Asymptotic Approximation of Integrals 87 Lastly. We are then concerned with integrals of the form I (κ) = f (z)eκq(z) dz. which is often k I r . C1 2 2βγT k T − 2β 2 . R(β) F0 2π U sT (β)ei(βx1 +γT x2 ) eiγL h dβ. We deﬁne the gamma function4 as ∞ (z)! := 0 3 e−t t z dt. calculate the x1 component of the scattered particle displacement. The parameter3 κ. though that dependence is seldom explicitly indicated.5.35) (5. Harris (1987) indicates how these decisions can effect approximating diffraction from an aperture. as well as those undertaken in the previous chapters. that is. when the distance r is many wavelengths long. We now consider their asymptotic approximation when k I r → ∞. Cβ (5. (z)! is also written as (z + 1). is assumed to be real.22) or (5.3.36) where q(z) = u(x1 . This is a another way of generating asymptotic approximations such as those discussed in Section 2.37) 4 The ﬁrst step in asymptotically approximating an integral is to scale the variables so that the large parameter κ can be clearly identiﬁed. though the approximations can (with care) be analytically continued to sectors of the complex κ plane.3 Asymptotic Approximation of Integrals The calculations of the previous section.23). (1983) indicate how they affect the asymptotic approximation of a Hankel function. Asymptotic approximations give a satisfying interpretation to these integrals as ray ﬁelds. Show that 1 1 u sT = 1 where U sT = Find the expression for u s L . have indicted how readily one arrives at integrals such as (5. x2 ) + iv(x1 . x2 ). This scaling and identiﬁcation depends a good deal on the situation of interest. while Carrier et al. (5. 5.34) (5.1 Watson’s Lemma To facilitate our work we need to study brieﬂy the gamma function and two of its friends. Also note that I (κ) is also dependent on the contour of integration C1 .
namely integration by parts.39) becomes (a. (5. . (a. = x ∞ e−t t z−1 dt.88 5 Radiation and Diffraction the incomplete gamma function as γ (z. (a − N − 1)! for N > (a − 1). but it is repeated here both for completeness and because it illustrates a very simple but useful way to generate an asymptotic approximation to an integral. We assume that dz q = 0 even at the end point z 1 . Equation (5. We begin by considering (5. (5. though the functions can be analytically continued to (z) < 0.36) with a contour C1 that begins at a ﬁnite point z 1 and extends to ∞ in a sector of the complex plane wherein the integral is convergent. (a − n)! (a − N − 1)! (5. Proof The following proof is from Copson (1971). x) ∼ e−x n≥1 (a − 1)! a−n x . we note that (a − 1)! (a − N − 1)! < = ∞ x ∞ x e−t t a−N −1 dt e−t dt (5. x). We look for a contour C2 from z 1 to ∞ along which q(z) < q(z 1 ) and q(z) = q(z 1 ).39) These deﬁnitions are for (z) > 0.38) and the complement to the incomplete gamma function as (z. (5.41) To estimate the magnitude of the remainder term. x). x) := 0 x e−t t z−1 dt. x) := (z − 1)! − γ (z. After N such integrations (5. We take x as real and positive. x) = e−x N n=1 (a − 1)! (a − 1)! a−n + x (a − N . (a − n)! x → ∞. Lemma.40) where a and x are real and positive.40) follows.42) (a − 1)! x a−N −1 (a − N − 1)! (a − 1)! x a−N −1 e−x .
46) Proof This is proven in Copson (1971). g(s) = a0 + a1 s + a2 s 2 + · · · + Rm+1 (s).45) ¯ ¯ with Rm+1 (s) ≤ Cs m+1 . λ > −1. Let g(s) be analytic at s = 0 so that for s ∈ [0. we map the z plane into the s plane by using s = q(z 1 ) − q(z). Rather it is the contour C1 in the z plane that is deformed to a new contour C2 in that plane. The term s λ includes any singularity at s = 0 either from f (z) itself or as a result of mapping to the s plane. though we do not include such contributions here. Then J (κ) ∼ n≥0 an (λ + n)! . Proposition 5.43) we introduce the function G(s) = − f (z)/dz q and write G(s) as G(s) = g(s)s λ .43) Second. one seldom works directly in the s plane.3 Asymptotic Approximation of Integrals First. Any poles or branch cuts encountered during the deformation must be appropriately surrounded and their contributions added to the asymptotic approximation of the integral. where g(s) is analytic at s = ¯ 0 (z = z 1 ) with a radius of convergence r . The integral J (κ) is written as J = 0 r∗ e−κs s λ (a0 + a1 s + a2 s 2 + · · · + am s m ) ds r∗ 0 + e−κs s λ Rm+1 (s) ds + ∞ r∗ g(s)s λ e−κs ds. where J (κ) is given by (5. r is the radius of convergence for g(s) and r ∗ < r . Ablowitz and Fokas (1997) and Wong (1989) with greater generality and weaker assumptions than those stated here. (5. With the change of variables of (5.47) .5.r ∗ ]. Consider the integral J (κ) = 0 ∞ g(s)s λ e−κs ds. (5. Let real constants K and b exist so that g(s) ≤ K ebs as s → ∞. (5. (5.44). we deform the contour in the s plane to one along the real s axis from zero to ∞. As with the Cagniard–deHoop contour. Carrier et al. (1983). κ λ+n+1 κ → ∞.1 (Watson’s Lemma). 89 (5.44) arrived at by the change of variables described in the preceding paragraphs. The integral now assumes the form exp[κq(z 1 )]J (κ).
One simple weakening of the assumptions of this proposition follows immediately by noting that g(s) need only have the asymptotic behavior exhibited by (5. With the use of (5. (5. we conclude from the previous Lemma that each term of this second integral is ∞ r∗ e−κs an s λ+n ds = O e−κr κ ∗ . say that with index n. (5. the ordering follows. Lastly.45) as s → 0. λ+2 κ κ an κ λ+n+1 ∞ κr ∗ (5. (5. the incomplete gamma function.50) With Rm+1 (s) ≤ Cs m+1 . .90 5 Radiation and Diffraction The ﬁrst integral is again split into two parts by being written as one from zero to ∞ minus one from r ∗ to ∞. This fact and Watson’s lemma are the starting point for the various Abelian and Tauberian theorems found in Doetsch (1974) or van der Pol and Bremmer (1950). is such that ∞ r∗ e−κs an s λ+n ds = e−t t λ+n dt. the second integral in (5. we ﬁnd that ∞ K r∗ e −s(κ−b) λ e−r (κ−b) s ds = O .48) Each term of the second integral. after a change of variables. as s → ∞. It also demonstrates the principle that how a function behaves at its origin is projected into how its transform behaves at inﬁnity and vice versa. (5. giving a0 λ! κ λ+1 + a1 (λ + 1)! (λ + m)! + · · · + am λ+m+1 .47) is such that r∗ 0 e−κs s λ Rm+1 (s) ds C κr ∗ 0 ≤ κ λ+m+2 e−t t λ+m+1 dt . The integral from zero to ∞ is readily evaluated.51) = Oκ −(λ+m+2) where we recognize that the integral on the right is.39) and the previous Lemma. and the previous Lemma. The asymptotic approximation (5.46) follows. (κ − b) ∗ (5. κr ∗ ). using g(s) ≤ K ebs .38).49) Noting that the integral on righthand side is (λ + n + 1.52) All the the bounds are taken as κ → ∞.
x2 ) = C. where C is a constant. and that dz q = 0 but dz q = 0 at z s . Viewing (5. (5. The function q(z) satisﬁes the Cauchy–Riemann equations ∂1 u = ∂2 v.x2s ) eκu(x1 .3. or v(x1 . (5. 1974). Thus. Our discussion follows a similar one given in Felsen and Marcuvitz (1994). beginning in one sector of the complex plane at inﬁnity and ending in another sector at inﬁnity. x2 ) near its stationary point z s .55) as a function of α.56) . though they are not explicitly included in the present discussion. We assume that q(z) is 2 analytic in region Q containing z s .x2 ) dz.3 Asymptotic Approximation of Integrals 5.2 Method of Steepest Descents 91 We next consider (5.36) with a contour C1 that stretches across the complex plane. Pole and branch cut contributions arising from deforming C1 to Cs must be added to the contribution made by the new integral over Cs . assuming. we parameterize the contour Cs with the arclength r so that the directional derivative of u along this contour is dr u = ∂1 u cos α + ∂2 u sin α. x2s ) the curvature of the surface u(x1 . The stationary point z s is therefore a saddle point and its neighborhood a col or saddle (Courant and John. Cs (5. that such a deformation is possible. x2 ) = C. To investigate this neighborhood and the contour Cs . We then deform the contour C1 into a contour Cs passing through this point and deﬁned by q(z) ≤ q(z s ) and q(z) = q(z s ). ∂2 u = −∂1 v (5. we examine the topography of the function q(z) = u(x1 . the direction of maximum change in u is given by d(dr u)/dα = −∂1 u sin α + ∂2 u cos α = 0. the principal contribution to the new integral comes from z s and I (κ) takes the approximate form I (κ) ≈ f (z s )eiκv(x1s .5. of course.5 sketches the topography in the neighborhood of z s .20) are examples. If f (z) is analytic and slowly varying near z s .19) and (5.53) The Steepest Descents Contour To ﬁx our ideas more precisely. Assume that (q) has a stationary point at z s .54) in Q so that u and v are harmonic functions. if at (x1s . The contours Cβ and C deﬁning integrals (5. is positive along the x1 axis. Figure 5.55) where dr x1 = cos α and dr x2 = sin α. it is negative along the (perpendicular) x2 axis. x2 ) + iv(x1 .
as we move away from the saddle point z s . and the angle it makes with the x1 axis as it passes through z s are shown. eκu increases most rapidly and eiκv again remains constant. eκu decreases most rapidly and eiκv remains constant for ψ = ±π/2.58) indicates that. this equation becomes dr v = 0. 5. The contour Cs . Therefore v is constant along the same contour on which u changes most rapidly. locally. (5. With the use of the Cauchy–Riemann equations. 2 (5.57) or exp[κq(z)] ≈ exp[κq(z s )] 2 × exp (κ/2) dz q(z s )(z − z s )2 (cos 2ψ + i sin 2ψ) . for ψ = ±π/4 or ±3π/4. eκu remains constant and eiκv oscillates most rapidly. The contour Cs along which v is constant but along which u decreases most rapidly is called the path of steepest descents. These latter contours are paths of stationary phase or stationary phase contours. (5. we examine the behavior of q(z) in the neighborhood of z s .92 5 Radiation and Diffraction Fig. Expanding q(z) about z s gives 2 exp[κq(z)] ≈ exp[κq(z s )] exp (κ/2) dz q(z s )(z − z s )2 . However.59) Examining (5. . its arclength r . This then is the contour Cs we seek.58) where 2 ψ = arg(z − z s ) + 1 arg dz q(z s ) . For ψ = 0 or π. or the steepest descents contour.5. To determine this contour. A threedimensional sketch of the topography of (q) = u near z s . the stationary point. there are two such contours and we need to select that one along which u decreases.
Figure 5.61) We deﬁne the inverse mapping so that.5 suggests how the contour Cs might look in the z plane. Proof We again use a mapping that captures the essential topographical features of the phase function q(z) in the neighborhood of the stationary point. (5. as s varies from −∞ to ∞. Any of the references cited when Watson’s lemma was discussed also discuss the method of steepest descents from various points of view.62) .5. κ → ∞. this would mean that the exponential in the integrand would oscillate and. x2 ) be analytic in a region Q containing z s . Then I (κ) ∼ (−2π)1/2 2 κ dz q(z s ) 1/2 f (z s )eiκq(zs ) + O κ −3/2 . (5. We map from the z plane to the s plane by using s 2 = q(z s ) − q(z). z traces the steepest descents contour from beginning to end. there could be cancellations. Assume that the contour C1 has been deformed into the steepest descents contour Cs deﬁned previously. An Isolated FirstOrder Saddle Point With the following proposition we put the knowledge just gained into a more precise form. Moreover. and it is isolated and n ﬁrst order. dz (z s ) = 0. That is. With this change of integration variable. n ≥ 2. The point z s is a saddle point. dz q(z) as z → ∞ in sectors of the complex plane where Cs begins and ends. the integral I (κ) is now given by I (κ) = eκq(zs ) ∞ −∞ G(s)e−κs ds. but dz q(z s ) = 0. voiding the assumption that the maximum contribution to the integral comes from this point.2. Let the function q(z) = u(x1 .60) 2 The argument of [−2/dz q(z s )]1/2 is deﬁned by the argument of ds z at z s . let there be constants K and b such that 2 f (z)[q(z s ) − q(z)]1/2 < K eb[q(z)−q(zs )] . As with previous propositions. therefore. the conditions cited are stronger than they have to be. Let the function f (z) be analytic in a region R containing z s such that f (z s ) = 0. 2 (5. Proposition 5. However. x2 ) + iv(x1 .3 Asymptotic Approximation of Integrals 93 Note that we could shift the steepest descents contour to pass through a point higher up the ridge.
48)–(5.64) ¯ where s ≤ r ∗ < r . 2 +2 e−κs R2(m+1) (s) ds + ∞ r∗ (5. G(s) = G(0) + ds G(0)s + ds2 G(0)(s 2 /2) + · · · + R2(m+1) (s). At this point we follow a procedure almost identical to that used to establish Proposition 5. ds G(0) is the ﬁrst nonzero term.1.66) Because G(s) is analytic. The asymptotic approximation.65) The differential element ds is real and positive along the path of integration. the pattern of the proof is unchanged. we note that by hypothesis g(s) < K ebs as s → ∞. Thus G(s) is analytic in a region containing s = 0 and can be expanded in a ¯ power series with a radius of convergence r . If r ∗ is small we have not really achieved very much.52). ds (5. We have assumed that f (z) is analytic in a region containing z s . (5. Therefore G(0) = f (z s ) dz ds .67) Estimating the contributions of the various integrals is identical to that used previously in (5. Moreover. we write J = r∗ −r ∗ e−κs r∗ 0 2 G(0) + · · · + ds2m G(0) 2 s 2m ds (2m)! g(s)s e−κs ds. Copson (1935) describes how to invert a series. ds z becomes indeﬁnite so that a limit must be taken as z → zs . so that we want it to be at least O(1).63) Note that at z = z s . say. Note how r ∗ enters the calculation. ds dz q(z) G(s) = f (z) dz . However. (5. This is what is meant by the phrase that f (z) be slowly varying near z s . follows. R2(m+1) (s) < Cs 2(m+1) . Setting I (κ) = exp[κq(z s )]J (κ). arg dz ds = arg(dz)z=zs = α. ds z is also analytic in a region containing s = 0 (the singularity is removable). .94 where 5 Radiation and Diffraction dz −2s = . (5.60). it is usually enough to calculate only the ﬁrst term G(0)5 . 5 2 If.61) expanded in a power series in z − z s must be inverted to give z = z(s) as a power series in s. where C is a constant. s=0 (5. so that at s = 0. This expansion is seldom easy to calculate because (5. setting G(s) = g(s)s. 2 Moreover. s=0 dz ds = s=0 −2 2 dz q(z s ) 1/2 .
x2 .7). The contour is thus a stationary phase contour.68) is asymptotically approximated by the sum of the contributions given by (5. The integration contour is then deformed to a path of steepest descents passing through xs . positive. To asymptotically approximate this integral. Consider an integral having the form given by (5. and κ is real. x1 < x2 . We assume i p(z) is analytic in region containing xs so that the stationary point becomes a saddle point.2.3. The end point contributions are estimated by using the method outlined in Section 5. Problem 2.70) The integral (5. 5. (5.3 Asymptotic Approximations of Integrals Problem 1. Using (5. κ → ∞. Show that the Cagniard–deHoop contour is one of steepest descents and that α = −s cos θ is a saddle point. we use the method of steepest descents. (5. The contributions from the end points x1 and x2 are I (κ) ∼ f (x1 ) iκ p(x1 ) f (x2 ) iκ p(x2 ) 1 − e e .69) and (5. Consider the integral given by (5.71) .1 and Watson’s lemma. Problem 5.3 Stationary Phase Approximation 95 The stationary phase approximation is usually applied to integrals of the form I (κ) = x2 x1 f (x)eiκ p(x) d x. The angle θ is that shown in Fig.70).3. I (x1 .68) where p(x) is a real function when x is real. Use Watson’s lemma to show that the ﬁrst term of an asymptotic expansion in p is a0 (−1/2)! − psr ¯ e .3 Asymptotic Approximation of Integrals 5.13). iκ dx p(x2 ) dx p(x1 ) κ → ∞. we ﬁnd the contribution from xs is I (κ) ∼ 2π 2 p(x ) κ dx s 1/2 f (xs )ei[κ p(xs )±π/4] . (5.5. p) ∼ p 1/2 p → ∞. (5.69) 2 2 The plus sign is taken for dx p(xs ) > 0 and the minus sign for dx p(xs ) < 0. and assumed large. The references cited in connection with Watson’s lemma discuss the stationary phase approximation from various other viewpoints and provide more detailed discussions. We assume that there is a ﬁrstorder stationary point xs between x1 and x2 .60).
96 5 Radiation and Diffraction Find a0 . Before continuing.3. In particular. This approximation is accurate as t → sr (consult Doestch.73) Show that a deformation of the contour to one along which τ is real gives an integral along the path of steepest descents Cs . Consider an integral of the form I (kr ) = P(cos α)eikr cos(θ −α) dα.22) and (5. provided the definition of α given in the paragraph preceding (5. (5. 5.22) and (5. The contour C begins at π − i∞ and ends at i∞.4. Problem 3.4 Buried Harmonic Line of Compression II We now turn to the asymptotic approximation of the integrals. recall Figs.72) C where θ ∈ (0.4 where the features of the β plane and the α plane. Sketch the contour and show that the integral converges provided it begins and ends in the sectors containing these points. respectively. 5. x2 . are described.23). 1974 or van der Pol and Bremmer. 5. namely ¯ β = k L cos α = k T cos α.4 and (5. the contours Cβ and C are indicated there.1 The Complex Plane We begin by recalling the Sommerfeld transformation used to arrive at (5.3 L I and 5.4. Speculate on the structure that P(cos α) might be permitted in order that an asymptotic approximation of the integral be successful. Section 3. as are their βplane counterparts in Fig. it is useful to ask what Fig.4.27) is used. (5.23) mean. γT = k T sin α.74) −1 ¯ where c−1 cos α = cT cos α and γ I = (k 2 − β 2 )1/2 .22) and (5. . Invert this to approximate I (x1 . Moreover. γ L = k L sin α. 5. t). The integrands of (5. 1950 to learn why) and is therefore sometimes called a wavefront approximation. π). 5.3 points out that A+ (α) is identical to the Rayleigh function. Consider the change of integration variable τ = 21/2 eiπ/4 sin[(θ − α)/2]. 5. that describe the waveﬁeld scattered from the tractionfree surface. ¯ (5. (5. These are indicated in Fig.20).23) therefore have poles at αr and π − αr . These poles give rise to oppositely directed Rayleigh surface waves. Consider the values of α to the left of the vertical line through π/2.
The function q(α) = i cos(α − θ1 ) and the stationary point. For values of α beyond αT the scattered shear plane waves are now also inhomogeneous and decay away from the surface. (5.22) as us L = − k L F0 4π ˆ p 1 (α)R L (α)eik L r cos(α−θ1 ) dα. each incident at the angle α in this range. a saddle point. incident inhomogeneous plane waves are added to the overall incident waveﬁeld. cos(α − θ1 ) = cos(α1 − θ1 ) cosh α2 − i sin(α1 − θ1 ) sinh α2 . As α climbs the ¯ imaginary axis. Cs is described by α2 = ±(α1 − θ1 ). . for α large.4 Buried Harmonic Line of Compression II 97 Those to the right have a very similar meaning.22) and (5.23) when it is enclosed. (5. [cos(θ1 − α)] > 0 so that Cs must begin and end in a region of the α plane where sin(α1 − θ1 ) sinh α2 < 0. Because we want [q(α)] to achieve a maximum along Cs at θ1 .76). That is. shear plane waves are still being reﬂected at the L real angle α.5. and end in α1 ∈ (θ1 − π.4. θ1 ). we express (5. note that as we climb beyond αT . we quite quickly encounter the Rayleigh pole at αr .76) For the integral (5. or. is given by α = θ1 .75) C To approximate this integral for k L r large. The contour Cs is described by [q(α)] = 1. the contour must begin in a region α1 ∈ (θ1 . and what it looks like near the stationary point. Reciprocally they scatter into compressional inhomogeneous plane waves that decay away from the surface. They are reﬂected into plane compressional and shear waves propagating away from the surface at α and α. where α2 > 0. α1 → ±π/2 + θ1 .75) to converge. However. −1 deﬁned by c−1 cos αT = cT . by cos(α1 − θ1 ) cosh α2 = 1. we must ﬁrst ascertain where the contour of steepest descents Cs begins and ends. π + θ1 ).77) Near α = θ1 . α2 = −(α1 − θ1 ). from (5. From π/2 to zero the incident compressional wave is composed of plane waves. Expressing α as α = α1 + iα2 . This term must be included in our evaluation of (5. At the limiting (imaginary) angle αT a shear wave grazing the ¯ surface is excited. (5. Further. as α takes on imaginary values from zero to a critical angle αT . They decay toward the surface. Lastly. respectively. Note that the Rayleigh wave could not be excited unless the incident disturbance contained inhomogeneous waves.2 The Scattered Compressional Wave Setting x1 = r cos θ1 and (x2 + h) = r sin θ1 . 5. where α2 < 0.
Harris and Pott (1985) discuss such contributions. Note that this has given us a representation of the kind investigated in Section 2. arg[−2/dα q(θ1 )]1/2 = 3π/4. the condition for this ﬁrst to happen is that cos θ1 cosh(−iαr ) = 1. we ﬁnd that us L ∼ A 4πk L 2π kL r 1/2 ˆ p 1 (θ0 )R L (θ0 )eik L r e−iπ/4 .4 – over part of its path. 5. usT = − k L F0 4π ˆ ˆ d 2 (α)RT (α)eikT p2 ·x eik L h sin α dα. k L r → ∞. as well as the two limiting lines at α1 = ±π/2 + θ1 . The remaining issue before using the steepest descents result (5. in agreement with what we indicated previously. 5.4. The overall structure of (5. should they be enclosed by this excursion. Suppose that θ1 is such that in distorting C to Cs the pole at αr is enclosed.2 .78) is that of a farﬁeld expression for a cylindrical wave radiating from a virtual source at (0. on the physically meaningful Riemann sheet. −h). At the surface x2 = 0. This is equivalent to the condition that cos θ1r = cr /c L . Occasionally one must also be careful that one does not forget to include any pole contributions that may arise from poles on this other sheet.77). Note also that Cs passes onto the other Riemann sheet – the dashed portion in Fig.79) C . If we inverted our result in time we should ﬁnd that this is equivalent to asserting that the Rayleigh wave cannot be excited before the incident compressional disturbance strikes the surface. Putting the various pieces together. (5. the poles at αr and π − αr may or may not be enclosed. Following the rule set out in Proposition 2 5. is a somewhat harder integral to approximate.23).78) 2 where F0 = A/k L has been used. (5.98 5 Radiation and Diffraction Figure 5. From (5. Thus for values of θ1r < θ1 < π/2.60) is to settle what the argument of the square root is. then the contour Cs must instead be wrapped around the branch cut. Problem 5. Note that depending upon θ1 .4 indicates the contour Cs passing through the saddle point θ1 . which is repeated in the following equation. When one cannot reemerge onto the Riemann sheet of physical interest.3 The Scattered Shear Wave Equation (5.4.4 asks the reader to evaluate the contribution from αr . no Rayleigh wave is present. and this contribution asymptotically approximated. One must be careful to ensure that when this happens that one can reemerge onto the Riemann sheet of physical interest. r = h/ sin θ1 is the radius of curvature of the reﬂected cylindrical wave as it is just about to form. where we have added the subscript r to indicate this special value.
then from a point on a virtual caustic. Let s0 be the distance the compressional ray propagates from (0. We have seen that the reﬂected compressional wave (5. h) to this same point on the surface and s2 be the distance the reﬂected shear ray propagates from this point to the Fig. In a similar way we should expect that the reﬂected shear wave will appear to originate.4 Buried Harmonic Line of Compression II 99 The complexity arises from the mixing of wave types: an incident compressional wave whose phase lingers in the term eik L h sin α and the scattered shear ˆ wave whose phase is eikT p2 ·x .6. 5. It appears to originate from a point on a virtual caustic located by extending the ray backward a distance ρT . h) and propagates a distance s0 . 5. We therefore know approximately what Cs looks like and can apply (5.78) appears to originate at the virtual source point (0.6. Finding Cs and investigating it in the neighborhood of the stationary point is similar to that for the compressional case.56).51) and (2.60) without working out Cs in detail. The ﬁrst clue can be found by examining (2.78) that a steepest descents approximation will give phase terms that are those of a ray theory. the radius of curvature has its origin either at a point on a caustic surface or at a single point.79) can be found by considering the geometry of the rays. We know from (5. That is. −h). The scattered shear ray emerges at an angle θ2 . the key to understanding (5. The radius of curvature ρT measures the distance from a point on the virtual caustic to a point on the surface x2 = 0 at which the scattered shear wave is starting to form. if not from a virtual source point.5. . striking the surface at an angle θ0 . The compressional ray originates at (0. These expressions are singular when the distance along the ray is the negative of the radius of curvature. Instead. We have sketched this possibility in Fig. A sketch of the shear ray scattered from the surface when struck by an incident compressional ray.
Fig. along with a continuous distribution of eigenvalues. as −1 well. The radius of curvature ρT is given by ρT = s0 c L sin2 θ2 . In closing we note that the contour in the β plane. and hence the Rayleigh wave excited by a harmonic line of compression at (0.79) as usT = − where q(α) = i k L s0 k T s2 cos(α − θ2 ) + i ¯ cos(α − θ0 ). We do not at present have an analytic expression for ρT . There are a . we express (5. h. (k T s2 + k L s0 ) (k T s2 + k L s0 ) (5. nor do we have one for the virtual caustic.80) by using (5. s2 . a saddle point.100 5 Radiation and Diffraction observation point (x1 . Cβ . Of particular interest is the fact that the spectrum has a discrete eigenvalue. This gives. s2 .4 The Rayleigh Wave Calculate the Rayleighpole contributions to us L and usT . x2 ). Setting (x1 − s0 cos θ0 ) = s2 cos θ2 and x2 = s2 sin θ2 . θ0 .60).82) 2 where again F0 = A/k L has been used. k T s2 + k L s0 → ∞. after some algebraic manipulation. Problem 5. that the phasematching condition c−1 cos θ0 = cT cos θ2 gives a fourth L relation. (5. 5. could be distorted so as to wrap around the Rayleigh pole and the branch cuts. usT ∼ 2π A 4πk L (k L s0 )(1 + s2 /ρT ) 1/2 ˆ × d 2 (θ0 )RT (θ0 )ei(kT s2 +k L s0 ) e−iπ/4 . Thus we may solve for the unknowns in terms of what is given. x2 and h (= s0 sin θ0 ) can be considered as given.13). while x1 . θ0 . the contribution from the branch cuts. and the ratio κ = c L /cT . However. and θ2 . and θ2 are determined in terms of x1 .81) k L F0 4π ˆ d 2 (α)RT (α)e(kT s2 +k L s0 )q(α) dα.83) and s0 . (5. is given by α = θ2 and α = θ0 . The unknowns at this point can be taken as s0 . h). The stationary point. cT sin2 θ0 (5.3. This would give a representation of the waveﬁelds us L and usT that corresponds to an eigenfunction expansion. κ is given in terms of Poisson’s ratio by (3.80) Cs The parameter (k T s2 + k L s0 ) is large. This is a manifestation of Fermat’s principle that ¯ the ray path is an extremum. we know that the asymptotic expansion of (5.79) must contain the term (1 + s2 /ρT )1/2 in its denominator so that we can identify ρT from the ﬁnal asymptotic expression. Note. x2 . the contribution of the Rayleigh pole. We have enough information at this point to approximate (5.
23). The crack lies along the positive x1 axis. labeled regions 4 and 5. A sketch of the crack.20). though the incident wave is a single plane one. in regions 1.5 Diffraction of an Antiplane Shear Wave at an Edge The previous problem has considered scattering of a cylindrical wave from an inﬁnite tractionfree surface. Speciﬁcally.7 indicates the geometry of the problem. with the corresponding timeharmonic equation being given by (2.5 Diffraction of an Antiplane Shear Wave at an Edge 101 number of starting points. we simplify the notation by dropping 2 5 3 x1 Crack 4 1 x2 Fig. antiplane shear wave normally incident to a semiinﬁnite slit or crack. 5.2. Figure 5. the traction acting on the surfaces of the crack is zero. Separating the diffracted and geometrical waveﬁelds. On both sides.15). The rippled arrows indicate their directions of propagation. As we have done previously when dealing with antiplane shear waves. but this is not the only starting point. I should likely begin with (5. but also because the incident waveﬁeld was composed of a complete spectrum of plane waves. the cylindrical diffracted wave emanating from the crack tip. or boundary layers.14) and (1. In this section we consider a related problem: one in which the scatterer has an edge that excites a full spectrum of plane waves.22) and (5.7. 5. and 3.5. . The equations of motion are given by (1. transition. and the reﬂected and transmitted geometrical waves. The complications came about not only because both compressional and shear waveﬁelds were scattered from the surface. we consider a time harmonic. are parabolic shaped.
The problem has therefore been divided into two separate ones. The total waveﬁeld u t3 = u i3 + u 3 . From the symmetry. each waveﬁeld. The total waveﬁeld u t3 = u ts + u ta . u i3 = (A/2k)(eikx2 + e−ikx2 ) + (A/2k)(eikx2 − e−ikx2 ). (5. described by u i3 = (A/k)eikx2 . the crack. the wavenumber. We consider the symmetric problem. we divide the incident waveﬁeld into symmetric and antisymmetric components as follows.84) is incident to the crack. (5. where u 3 is the scattered waveﬁeld.102 5 Radiation and Diffraction the subscript T and use c and k = ω/c as the wavespeed and wavenumber. By construction.86) so that u i3 = u is + u ia . The boundary and continuity conditions to be satisﬁed at x2 = 0 are µ∂2 u t3 (x1 .5. 0± ) = 0. Reﬂecting the symmetric problem in the plane x2 = 0 leaves the particle displacement unchanged. Lastly. separately satisﬁes the boundary and continuity conditions (5. u t3 (x1 . The symmet3 3 ric scattered waveﬁeld is excited by the incident symmetric one and the antisymmetric scattered waveﬁeld by the incident antisymmetric one. respectively. having been made so by dividing by k.85). and that it therefore satisfy the principle of limiting absorption (Section 4.1 Formulation The plane wave u i3 . the problem is divided into two. where u ts and u ta are the symmetric and antisymmet3 3 3 3 ric components. To begin.4). 5. x1 < 0. respectively. Also we ask that the scattered wave be outgoing from its source.85) The superscripts ± indicate that the boundary is approached through positive or negative values of x2 . respectively. 0+ ) = u t3 (x1 . (5. x1 > 0. This and the condition that 3 the scattered waveﬁeld propagate outward imply that u s ≡ 0. symmetric and antisymmetric. Therefore the 3 symmetric part of the incident waveﬁeld remains unaffected by the presence of . We next set the scattered waveﬁeld u 3 = u s + u a where 3 3 3 3 u s is symmetric and u a antisymmetric with respect to x2 = 0. As a way to formulate the problem for the scattered disturbance. ∂2 u s = 0 ∀ x1 at x2 = 0. ∂2 u is = 0 ∀ x1 at x2 = 3 0. while reﬂecting the antisymmetric one multiplies it by −1. one symmetric and one antisymmetric with respect to this plane. 0− ). The constant A is dimensionless. Note that x2 = 0 is a plane of reﬂection symmetry for the problem.
0) = −i Ae− x1 (5. it is not needed for other angles of incidence. the condition that u 3 be outgoing and satisfy the principle of limiting absorption is enforced. We could solve this problem by formulating an integral equation for the scattered waveﬁeld following the ideas outlined in Problem 4. with k taken as real and positive.5. The scattered waveﬁeld for x2 < 0 is found by using the fact that u 3 (x1 . x1 < 0. Reasserting the decomposition u t3 = u i3 + u 3 . but noting that u 3 = u a .3 to demand that ku 3 = O (kr )1/2 . The complete problem is then equivalent to the antisymmetric one and the scattered waveﬁeld is antisymmetric in x2 . 1988). do not vanish as kx1  → ∞ unless this artiﬁce is used. (5.7.2 Wiener–Hopf Solution We begin by setting u 3 (x1 .5. 1948). as we indicate in Problem 5. 0) = 0. ∂2 u 3 (x1 . In this particular problem.87). 5. has been explicitly added to k in (5. 0) = 0. x2 ) = −u 3 (x1 . we 3 are led to reformulate the problem. u 3+ . x1 > 0. We set k = k + i . In fact we use the analysis of Section 4. (5. u 3 (x1 .7.88) As well. −x2 ). 1952. We should then ﬁnd that the integral equation can be solved by using the Wiener–Hopf method (Titchmarsh. as with the incident wave. However. ∂α ∂α u 3 + (k + i )2 u 3 = 0. Noble.90) . and perhaps just as informative. in the halfspace x2 > 0. The addition of the e− x1 in the boundary condition is an artiﬁce also needed to enforce the principle of limiting absorption. x1 < 0. This condition is essential to ensure that the solution be unique.87) . (5. x1 > 0. It is for this latter reason that the i . subject to the conditions. we must also append to these equations an edge condition as explained in Section 4. the scattered waveﬁeld will contain geometrically reﬂected waves that.89) 2 2 where r = (x1 + x2 )1/2 .3. 0 < 1. Lastly. for the scattered waveﬁeld u 3 as follows. at x2 = 0.2.5. because the incident wave strikes the crack normally. to solve the problem directly with the Wiener–Hopf method as used by Jones (Jones. However.5 Diffraction of an Antiplane Shear Wave at an Edge 103 the crack in its path of propagation. it is easier. kr → 0.
2 and ﬁrst explicitly discussed in Section 3.8.1 (the roles of x1 and x2 are interchanged).93) With some thought. 5. This is the same choice of Riemann sheet as was taken in Section 5. Figure 5. As → 0 the pole will approach a position just above this contour. 5. one realizes that.91) Our strategy in solving the problem for the scattered waveﬁeld is to ﬁnd a functional relation between the two unknowns τ− and u 3+ . The contour for the inverse transform initially lies within this strip. Imposing the conditions at x2 = 0 on (5. (5.6.90) gives the following: ∗ u 3+ = 0 ∞ u 3+ (x1 )e−iβx1 d x1 . x1 < 0.3. Recall that x2 > 0 so that the β plane is cut such that (γ ) ≥ 0.104 5 Radiation and Diffraction ∂2 u 3 (x1 . . ∀ β. The complex β plane showing the branch cuts. (5.4. 0) = τ− .92) and using (5. pole at i . we set u3 = 1 2π ∞ −∞ ∗ u 3+ ei(βx1 +γ x2 ) dβ. We seek a solution to (5. and strip of common analyticity 2 wide.92) where γ = (k 2 − β 2 )1/2 . the dominant waveﬁeld is a plane one.87) by using a representation very similar to that used previously in Section 2.4. (5.8 shows the branch points with the accompanying cuts. That is. propagating normally away from the Fig. The transform ∗ u 3+ remains to be determined. τ+ = −i Ae− x1 . x1 > 0. to the right along the crack in Fig.
while the other is analytic in the region (β) < . Thus ku 3+ = O(e− x1 ) as kx1 → ∞. but that all parts are analytic in the common strip (β) ∈ (− . Note that different parts of it are analytic in different parts of the complex β plane. from which it follows that ∗ τ− is an analytic function of β for (β) > − .6 That part grazing the crack face will also decay as e− x1 because k = k + i . We demand that ∂2 u 3 calculated from (5. In essence. Further we deﬁne the transform ∗ τ− as ∗ τ− = 0 −∞ τ− (x1 )e−iβx1 d x1 . (5. ).95) at x2 = 0.94) and (5. but one that is forced to decay as x1 → ∞ because of the e− x1 placed in the ﬁrst of (5.5. Experience with similar problems is how this is done. Also note that the condition that u 3 = 0 for x1 < 0. Any other wave present will originate from the crack tip. The goal of the next few paragraphs will be to rearrange this expression so that one side is analytic in the region (β) > − . (5.88). x2 = 0 has been built into ∗ u 3+ .91). Noble (1988) discusses in some detail the conditions needed to ensure that a transform is an analytic function of its independent variable.5 Diffraction of an Antiplane Shear Wave at an Edge 105 crack. the transform is an analytic function of its variable β for those regions of the complex β plane wherein the integral is uniformly convergent. while Titchmarsh (1939) discuses this somewhat more generally. This gives iγ ∗ u 3+ = [−A/(β − i )] +∗ τ− . from which it follows that ∗ u 3+ is an analytic function of β for (β) < . A cylindrical scattered wave is emitted from the tip.92) be consistent with that represented by (5. because k = k + i . we ﬁnd that ∗ τ+ = 0 ∞ τ+ (x1 )e−iβx1 d x1 (5.94) = −A . the ﬁnal answer must be checked to ensure that it is consistent with these assumptions. .96) which is the functional relation we are looking for. Working next with (5. The two sides will 6 We need to guess at the kinematics of the scattered wave to deduce where ∗ u 3+ and ∗ τ− are analytic. Thus τ− = O(e− x1  ) as kx 1 → −∞. (β − i ) Note the pole at i .95) The crack tip is the only possible source for a scattered wave in x1 < 0. However.
The function is entire and its nature is determined by its behavior at inﬁnity. . 1/2 (β − i )(k + β) (k + β)1/2 (5. This is the condition that tells us how the entire function behaves at inﬁnity. while the two sides are equal in the common strip (β) ∈ (− .99) Note how the edge condition was essential to the determination of a unique solution. (β − i )(k + i )1/2 (k − β)1/2 (5. The outcome is i(k − β)1/2 ∗ u 3+ + = −A A (β − i )(k + i )1/2 (5. The left side is analytic for (β) < and the right side for (β) > − . is the simpliﬁcation introduced by Jones (1952). 1939) it follows that the entire function must be identically zero. rather than through ﬁrst forming an integral equation.97) so that its left side becomes an analytic function for (β) < .96) as i(k − β)1/2 ∗ u 3+ = ∗ −A τ− + . Proceeding to this relation directly from the transforms. We next write (5.98) ∗ (k + i )1/2 − (k + β)1/2 τ− . Using Liouville’s theorem (Titchmarsh. ). Recall that we asked that ku 3+ = O[(kx1 )1/2 ] as kx1  → 0. + 1/2 (k + i )1/2 (β − i )(k + β) (k + β)1/2 We have succeeded in achieving our goal. but for the presence of the pole at i . The two sides are thus the analytic continuations of one another. To isolate the pole term we add and subtract the residue multiplied by (β − i )−1 . and together they represent a function analytic everywhere in the ﬁnite β plane. Therefore ∗ u 3+ = iA . From this we may infer that τ− = O[(kx1 )−1/2 ] as kx1  → 0. which function must then be entire. The right side is almost one for (β) > − . This process is the Wiener– Hopf method. We have still not used the edge condition. Adapting an Abelian theorem k 2 ∗ u 3+ = O(k 3/2 β−3/2 ) and that k ∗ τ− = O(k 1/2 β−1/2 ) as k −1 β → ∞.106 5 Radiation and Diffraction be equal in the common strip and hence will be different representations of the same function.
5 Diffraction of an Antiplane Shear Wave at an Edge 107 At this point has served its purpose and we take the limit → 0. In region 1 the waveﬁeld u 3 is composed of a residue term. each of which has a somewhat different waveﬁeld.3 Description of the Scattered Waveﬁeld We have indicated ﬁve regions in Fig. Problem 5.6. from the pole in the integrand of (5.102) Problem 2.102). The .102) has a pole at β = k cos θ0 . namely u3 = iA 2π k 1/2 ∞ −∞ ei(βx1 +γ x2 ) dβ. (β − k cos θ0 )(k − β)1/2 (5. 5. An asymptotic approximation. we must take care that the stationary point does not lie near the pole. We are thus left with an integral expression for the scattered waveﬁeld.100) for an arbitrary angle of incidence θ0 . Repeat the calculations leading to (5.101) where k = k + i and > 0.5.100) where the contour passes below β = 0. It is uniform when the approximation is accurate for all values of the second parameter. is said not to be uniform. The incident wave is given by u i3 = (A/k)eik (cos θ0 x1 +sin θ0 x2 ) . that becomes disordered for certain values of a second parameter. In closing.100) at β = 0. because in that case we can no longer argue that the integrand. Integrals of this form or of the form achieved after using the Sommerfeld transformation are sometimes referred to as diffraction integrals. we recall that this is the solution to the scattered waveﬁeld only for x2 > 0. The integral (5. Include the pole term when it it is needed and note when the asymptotic approximation breaks down. plus the integral itself. (5. can the restriction be removed? Show that the scattered waveﬁeld u 3 is given by u3 = i A sin θ0 eik(cos θ0 x1 ) 2πk 1/2 (1 + cos θ0 )1/2 ∞ −∞ ei(βx1 +γ x2 ) dβ. in this case the angle of incidence θ0 . Calculate an asymptotic approximation to (5. aside from the exponential term. 5. Initially assume that θ0 < π/2. Why might this be a useful restriction? Once the solution is obtained.5 An Arbitrary Angle of Incidence Problem 1. is slowly varying. When we calculate a steepest descents approximation to (5.5.102) that is not uniform. calculated when one parameter is large. β(k − β)1/2 (5.
−x2 ) to calculate u 3 for x2 < 0. Within these regions the diffracted wave comes close to phase matching to the geometrical wave so that the two kinds of waves strongly interact. In region 3 the waveﬁeld u 3 is again composed of a residue term plus the integral itself. note that in using the antisymmetry. However.7 However. equivalently. One way to do this is to subtract and add the integrand of (5. Regions 4 and 5 are transition or boundary layers. Moving from left to right through these layers.125). as we subsequently demonstrate. 7 The calculation is done by separating a term that contains the pole from the rest of the integrand.104) It is important to note that this expression assumes that x2 > 0 or. For x2 > 0. while the integral with the pole term becomes a Fresnel integral. the scattered waveﬁeld makes a transition from solely a diffracted wave to a diffracted plus geometrical wave. .100) contains both the diffracted wave and geometrical terms.100) to an exact combination of Fresnel integrals. These regions are characterized by Fresnel integrals. θ ∈ (0. The integral with no pole term in its integrand is approximated in the usual way. The incident wave u i3 must be added to this to produce the total waveﬁeld. θ ∈ (0.108 5 Radiation and Diffraction residue contribution cancels the u i3 so that the crack casts the region behind it into partial silence. This reduction is described in detail in the Appendix.103) We introduce the polar coordinates (r. The Fresnel integral is deﬁned as ∞ F(z) := z eiξ dξ. In region 2 the waveﬁeld u 3 consists solely of the integral representing the diffracted wave. x2 ) = −u 3 (x1 . This is the only wave that penetrates the silence. (5. where x1 = r cos θ and x2 = r sin θ. θ). though the total waveﬁeld would include u i3 as well. namely that u 3 (x1 . π].100) is calculated in Felsen and Marcuvitz (1994). −π]. The outcome of the calculations described in the Appendix is that u3 = e−iπ/4 A −ikr cos(θ −π/2) θ − π/2 e F (2kr )1/2 cos π 1/2 k 2 − e−ikr cos(θ +π/2) F − (2kr )1/2 cos θ + π/2 2 . Note that (5. The integral represents a diffracted wave that appears as a cylindrical one radiating from the crack tip. while for x2 < 0. the residue contribution gives the wave reﬂected from the crack. 2 (5. A uniform asymptotic approximation to (5. most of the manipulations undertaken to do so are no more difﬁcult than those needed to reduce (5. thereby producing a complicated waveﬁeld. but propagate independently elsewhere. while the integral continues to represent the diffracted wave.
109) This reproduces the edge condition we enforced in (5.89). and Jull (1981). Note the similarity in structure between the expansion (5. given by u t3 = e−iπ/4 A −ikr cos(θ −π/2) θ − π/2 F (2kr )1/2 cos e 1/2 k π 2 + e−ikr cos(θ +π/2) F (2kr )1/2 cos θ + π/2 2 . as Problem 5. Any singular behavior as kr → 0 is contained entirely within u 3 . The interested reader can pursue this theory further in books by Achenbach et al.41).104). −π]. To calculate the total waveﬁeld. (1982). (5. When this is done we ﬁnd. π/2) k k (kr )1/2 kr → ∞. e H (−x1 ) + D(θ.7 describes the asymptotic approximations to the Fresnel integral. Babiˇ and c Buldyrev (1991). Problem 5. π/2) is called a diffraction coefﬁcient.107) We ﬁnd a very similar expression for θ ∈ (0. that u t3 ∼ where D(θ. whose approximation in this limit is ku 3 = O[(kr )1/2 ]. . −θ). In fact. we use the following property of the Fresnel integral: F(z) + F(−z) = π 1/2 eiπ/4 . for θ ∈ (−π. π ]. θ) = −u 3 (r. π].7 indicates. after adding u i3 to (5. (5. we use the fact that u 3 (r.108) 2(2π)1/2 cos[(θ − π/2)/2] cos[(θ + π/2)/2] A ikx2 A eikr .5. kr → 0. is. (5.5 Diffraction of an Antiplane Shear Wave at an Edge 109 θ ∈ (0. (5. (5. The coefﬁcient D(θ.106) can also be approximated for kr 1. there are a whole fan of such rays as θ is allowed to take on all its values. π/2) = 1 1 eiπ/4 + . Its arguments are intended to suggest that a ray incident to the crack tip at an angle π/2 excites a diffracted ray that leaves the tip at an angle θ. The integrals of (5. To obtain an expression for x2 < 0. for θ ∈ (0.107) and the earlier expansion (2.106) For kr 1 these integrals can be asymptotically approximated.105) The total waveﬁeld u t3 .6 introduces some of the ideas used in this description of diffraction. Problem 5. π]. These coefﬁcients play an important role in the geometrical theory of diffraction for edges.
Estimate the waveﬁeld diffracted by the strip by using what has been learned from the semiinﬁnite crack problem. we ﬁnd that if θ is close to π/2 the argument of the Fresnel integral is too small to approximate it asymptotically. θi ). A sketch of the strip indicated by the dashed line. the equation of a parabola in polar form.1? Fig. Problem 5. as shown in Fig. 5. The boundary layer lies within this boundary.106) whose argument changes sign. (5. formed by two semiinﬁnite cracks. Taking the equality as forming the boundary of the region outside of which the asymptotic approximation is sufﬁciently accurate. indicated by the solid lines. we ﬁnd that the boundary is described by kr (1 − sin θ) = 1.110). aside from the fact that the time dependence is harmonic. .9. Let us consider the neighborhood of π/2 so that it is the second Fresnel integral in (5. How does this problem. Examining (5. Examining this argument. differ from Problem 5. we note that one or the other of the arguments of the Fresnel functions changes sign at these angles.106). despite kr being quite large. The boundary of region 5 is simply the reﬂection in x2 = 0 of that for region 4.9. This then marks the boundary of region 4 with regions 1 and 2.110) such an approximation is not accurate. 5. and its scale is set by the expression on the left in (5. Thus if (2kr )1/2 cos θ + π/2 2 ≤ 1. Here i = 1 at b and 2 at −b.107) breaks down near θ = ±π/2.6 The Geometrical Theory of Diffraction Consider an antiplane shear wave normally incident to the opening between two cracks. Each crack tip ±b serves as the origin for one of the polar coordinate systems (si .110 5 Radiation and Diffraction The approximation (5.
π/2) is given by (5. b). express the total waveﬁeld as u t3 = u i3 + u 3 . u t3 (x1 .112) where the coordinates (s1 .113) .2 = π/2.6 Matched Asymptotic Expansion Study We have just undertaken a relatively complex solution to what.7. −b) ∪ (b. Explain. Using this structure. might seem a simple wave process. ∞). we begin our analysis of the diffraction problem by assuming that the waveﬁeld can be described with the asymptotic approximation u 3 ∼ eik S(x1 . At what point do they choke off the geometrical term given by the ﬁrst term of (5. 0± ) = 0. 5. Why? Consider the boundary layers surrounding θ1. x2 ). (5. 5. In this closing section we construct an asymptotic solution that captures this idea mathematically. In writing this I have followed a similar discussion in Zauderer (1983) and beneﬁted from a very detailed analysis of edge diffraction by Gautesen (1979). stating all the conditions that must be satisﬁed. the crack merely blocks the incident wave from reaching the other side.108). looking at Fig.9 and D(θ.111) u t3 (x1 .84). 0− ). 5. in x2 > 0. The incident wave is given by (5. using what has been learned from the semiinﬁnite crack problem.112)? The earlier parts of Harris (1987) might be useful.x2 ) n≥0 (−ik)−n An (x1 . Are the boundary and continuity conditions at x2 = 0 the following? µ∂2 u t3 (x1 . k (ks1 )1/2 (5. x1 ∈ (−b. They grow in width as one moves away from the edges. but are not in any way essential. + D θ1 . (5. to some leading order of approximation. θ2 ) are deﬁned in Fig. This solution is best for 2kb 1.4 we explored how waves could be approximated by a ray theory and provided an asymptotic structure that explained the connection. An exact analytic solution to this problem may not be possible. in answering this last question. In Section 2. 0+ ) = x1 ∈ (−∞.5. After all. 2 k (ks2 )1/2 2 A eiks1 .6 Matched Asymptotic Expansion Study 111 As before. why the following expression is a plausible asymptotic solution to this problem u t3 = A [H (x1 + b) − H (x1 − b)]eikx2 k π A eiks2 π + D π − θ2 . θ1 ) and (s2 . Formulate the problem for the scattered waveﬁeld u 3 .
It also follows that An ≡ 0 for n ≥ 1. What we need to do is scale the problem in such a way that the boundary layer is opened up and an equation governing the waveﬁeld within it found. where the minus sign is used for x2 > 0 and the plus sign for x2 < 0. given by (5. We then construct the solution to this equation so that it matches the surrounding waveﬁeld to some order of approximation. x2 ) = ±x2 H (x1 ).106).7. that we are about to examine.112 5 Radiation and Diffraction This assumption leads to the eikonal equation (2. we ask that S(x1 . for the present we avoid this nearﬁeld. we consider the region x1 > 0. recall that. In addition to the boundary layer. 5. Both Hinch (1991) and Holmes (1995) discuss this technique thoroughly. where the plus sign is used for x2 > 0 and the minus sign for x2 < 0. Knowing S. to leading order the total waveﬁeld u t3 is given by (A/k)eikx2 .43) for S and the transport equation (2.84). Moreover.113). The appropriate solution is S(x1 . but does not capture the transition in regions 4 and 5. it must be such that the traction for the total waveﬁeld vanishes on both sides of the crack. This solution gives the geometrical part of (5.3. we might suspect that our asymptotic ansatz does not have enough structure to capture the effects of the rapid changes in the waveﬁeld near x1 = 0 just from the apparent discontinuity there. 0) = 0 and that the scattered waves be outgoing from the crack. from our analysis in Section 4. we consider the region x1 < 0. 0. there is a nearﬁeld or inner region that is not described by the asymptotic anzatz (5. The appropriate solution is A0 (x1 ) = ∓(A/k) H (x1 ). There are no boundary conditions to enforce. and H (x1 ) is the Heaviside function. leading to the conclusion that An ≡ 0 for n ≥ 0. with part of it behaving as (kr )1/2 . region 2. indicates that A0 can have at most an x1 dependence. The scaling needed to open the layer up is usually not known and must be found as part of discovering the governing equation. (2. Even in the absence of an exact solution. However. We have encountered a boundary layer. region 1. Therefore.114) where the regions are those indicated in Fig. Second. (5.7. t u 3 (x1 . First. . the transport equation. Moreover. we know that very near the crack tip the waveﬁeld is quasistatic. x2 ) = (A/k)(eikx2 + e−ikx2 ). region 3. To match the phase of the incident wave u i3 at x2 = 0.44) for A0 . Such a procedure is called matched asymptotic expansions.44).
so that we must introduce one somewhat artiﬁcially.116) has become 2 2 2i(k )∂2 w + ∂2 w + ∂1 w = 0. as it sets a gauge to measure large and small. Equation (5. then ∂1 w may be very large and to solve the equation we shall need another term to balance it. We then deﬁne the nearﬁeld as that for which k 1 and the farﬁeld as that for which k 1. We again take the length as a reference length.106).117) The argument. suggests the scaling needed to open up the boundary layer. It might represent a wavelength at a reference frequency or a distance at which a measurement is made. We are examining the waveﬁeld in the region k 1 with θ near π/2.116) We have stripped away the oscillatory part of the waveﬁeld. however. 1995). Accordingly. Holmes.5. The coordinates (y1 . (5. of the second Fresnel function in (5.3. Having done this.6 Matched Asymptotic Expansion Study 113 We consider the case that x2 > 0 so that θ is near π/2 and leave the case x2 < 0 to the reader. that we do not in general know β and must determine it from the rescaled equation by balancing the various terms (Hinch. This change of coordinate is called introducing a stretching transformation. At this point we 2 note that if w varies rapidly in the neighborhood of x1 = 0. In this case it is the ﬁrst and third terms in (5.7. we set u 3 = w(x1 . The term 2ik∂2 w seems a likely candidate because it is multiplied by the large parameter k. we now omit the overbar. x2 ).116) in terms of the scaled variables. The diffraction problem does not have a natural length scale.106). ¯ ¯ We scale the problem by setting xi = xi / and w = wk/A (recall that A/k is a magnitude of the incident wave) and reexpress (5. Note. .115) (5. Examining (5. We have indicated previously that a length scale is very important. expressed in terms of the scaled (x1 . we note that in this region a propagator term eikx2 always appears. just as we did when examining the local ﬁeld near the crack tip in Section 4. We set y1 = (k )β x1 with β = 1/2.117) that balance. (5. 1991. namely −x1 (k /2x2 )1/2 . x2 )eikx2 and arrive at the following equation for w: 2 2 2ik∂2 w + ∂2 w + ∂1 w = 0. x2 ) the outer coordinates. x2 ) are called the inner coordinates and the (scaled) (x1 .
(5. so that an asymptotic approximation there will not be accurate.123) (5.119) This is the equation governing the behavior of the waveﬁeld in the boundary layer. 1983). However. That is.69). 2 (5. Recalling the fundamental or causal Green’s function for the diffusion equation (Zauderer.119) as w0 = 1 1/2 x2 ∞ −∞ f (k )1/2 s ei(y1 −s) /(2x2 ) ds.121) should match our earlier approximation to u 3 . for x1  > 0 we expect that the asymptotic approximation to (5. . 1979).120) To ﬁnd the unknown f (x) we express this equation in terms of the outer variables as w0 = (k )1/2 1/2 x2 ∞ −∞ f (s)eik (x1 −s)2 /(2x2 ) ds. we ﬁnd that w0 ∼ f (x1 )(2π )1/2 eiπ/4 . (5. x2 ) + (k )γ w1 (y1 . (5. .121) The reader should recall that we are assuming x2 > 0.120) to determine unequivocally that β = 1/2 and also to determine the unknown constant A in the nearﬁeld expansion (4. provided f (s) does not vary rapidly near the stationary point s = x1 . However. (5. the function f (s) must vary rapidly near x1 = 0 if it is to capture the transition in the boundary layer. x2 ) . w0 ∼ 0 when x1 < 0 and w0 ∼ −1 for x1 > 0.52) (Gautesen. we write the solution to (5. 2 2i(∂w0 /∂ x2 ) + ∂ 2 w0 ∂ y1 = 0. The leadingorder term is governed by the parabolic Schrodinger equation.122) This process of ﬁnding f (x1 ) is referred to as matching the inner and outer expansions. or the steepest descents approximation.52) to (5. . Using the stationary phase approximation. (5. Thus we ﬁnd for x2 > 0 that f (x1 ) = −e−iπ/4 /(2π)1/2 H (x1 ). . This integral can be asymptotically expanded for large k .114 5 Radiation and Diffraction We next introduce an asymptotic expansion of the form w ∼ w0 (y1 .118) where γ > 0. At this point we could now match the nearﬁeld expansion (4.
.126) Note that the contour now passes above the pole at α = π/2. If we need a boundary layer solution for x2 < 0. (5. Appendix: The Fresnel Integral Our purpose is to derive the expression (5. region 5. (5. we ﬁnd that u3 ∼ − Ae−iπ/4 −ikx2 k e F − x1 kπ 1/2 2x2 1/2 . namely u3 = − iA 21/2 πk C cos(α/2) ikr cos(α−θ ) dα. Recall that the diffraction integral is evaluated along the contour Cβ that passes below the pole at β = 0.121) to a second term.104). For x2 > 0 and x1 ≈ 0.41).100). Examining (5.104) reduces to this expression. θ) by setting x1 = r cos θ and x2 = r sin θ to reduce (5.107). The integral to be evaluated is (5. which we rewrite here. To include this possibility we should have allowed for fractional powers of k in positing (5. Had we expanded (5. which is a global property of the solution.113) to match it. Had we not known the solution nor suspected what was going on. −x2 ). The essence of its evaluation is to reduce it to one on its contour of steepest descents.5 can be completed. Why? The ansatz (5.103). x2 ) = −u 3 (x1 . We use the Sommerfeld transformation β = k cos α and γ = k sin α and introduce the coordinates (r. we use the antisymmetry of the scattered waveﬁeld. e cos α (5. points that may coalesce for certain values of the physical coordinates.125) to an integral over the contour C in the α plane.124) where F(z) is the Fresnel integral deﬁned previously by (5. I follow the derivation given in Born and Wolf (1986). we should have encountered a term O[(k )−1/2 ] but found no similar term in the expansion (5. β(k − β)1/2 (5. It is repeated here so that the calculation begun in Section 5.Appendix: The Fresnel Integral 115 Removing all the scaling so that we may compare our result with (5.113) just as we did with the α in (2. namely u 3 (x1 . the failure to match at higher order would have indicated that something was missing.113) contains only the geometrical waveﬁeld and not the diffracted one. u3 = iA 2π k 1/2 ei(βx1 +γ x2 ) dβ. is called a diffraction integral. we see immediately that the missing term comes from the diffracted waveﬁeld.104).125) Cβ An integral that has a pole and stationary point (usually a saddle).
We next reverse the direction of integration. cos[(α − π/2)/2] cos[(α + π/2)/2] (5.127) has been used. we consider the ﬁrst of the two integrals. writing the symbol for the contour as −Cs (θ). The complex α plane. θ is the saddle point in the α plane. The following two identities are needed for the work to follow: 1 1 23/2 cos(α/2) ≡ + . The complex β plane for the diffraction integral. Setting aside the i A/(4πk). We now distort the contour C to the steepest descents contour Cs described by (5. cos[(α − θ + π/2)/2] (5. is sketched on the right.77).76) and (5. 5. showing the contours C and −Cs (θ). The diffraction integral is now written as u3 = iA 4πk × eikr cos(α−θ ) −Cs (θ ) 1 1 + dα.128) The second identity is a generalization of the ﬁrst. cos α cos[(α − π/2)/2] cos[(α + π/2)/2] (5.116 5 Radiation and Diffraction Fig.10.129) where (5. which we label I .127) 1 4 cos(α/2) cos[(θ − π/2)/2] ≡ cos α + cos(θ − π/2) cos[(α + θ − π/2)/2] + 1 . is sketched on the left. 5. We note that the calculation of the second is not different from that .10. This contour is shown in Fig. showing the contour Cβ . The position of the stationary point is indicated as an argument of this contour to underscore the structure of the contour.
(5. we can collapse this integral into the more symmetric form I =2 cos(α/2) cos[(θ − π/2)/2] ikr cos α e dα. 2 (5.132) where η = 21/2 cos[(θ − π/2)/2]. 2 (5.3). Now u 3np is given by u 3np = e−iπ/4 A −ikr cos(θ −π/2) θ − π/2 e F (2kr )1/2 cos (1/2) k π 2 ± e−ikr cos(θ +π/2) F ± (2kr )1/2 cos θ + π/2 2 .130) cos[(α + θ − π/2)/2] cos[(α − θ + π/2)/2] Using (5.103). Using this (interchange the order of integration on the lefthand side). Let u 3np represent the particle displacement that does not include the pole contribution and u 3 p the pole contribution itself. repeating the deﬁnition (5. A third identity. Note that this integral contains both a pole contribution and a part composed of Fresnel integrals.136) .133) is needed. (5.129). cos α + cos(θ − π/2) (5.128).135) We now return to (5. giving I = 1 2 × eikr cos α −Cs (0) 1 1 + dα.134) 2 = 2π η −ikr η2 F η(kr )1/2 . τ 2 − iη2 2 (5.Appendix: The Fresnel Integral 117 of the ﬁrst.131) −Cs (0) We now introduce yet another change of variables with the relation τ = 21/2 eiπ/4 sin(α/2) (recall Problem 5. is F(z) := z eiξ dξ. namely ∞ kr eiη 2 ζ ∞ −∞ e−ζ τ dτ dζ ≡ π 1/2 2 ∞ kr ζ −1/2 eiη ζ dζ. e η 1/2 ∞ where the Fresnel function F(z). we ﬁnd that η ∞ −∞ 2π 1/2 η −ikr η2 e−kr τ dτ = e τ 2 − iη2 η 2 ∞ η(kr )1/2 eiµ dµ (5. We next change variables. Thus the integral I becomes I = −2e iπ/4 ikr e η ∞ −∞ e−kr τ dτ.
138).139) One way to do this is to deform the contour to one such that Watson’s lemma can be used.141) That the two approximations differ for x that is positive or negative is an example of the Stokes’ phenomena. (b) Assume x < 0. Why? Write the integral in (5.104). 2x x → −∞.140) 2 ∞ x eiζ dζ.137) with the second term of (5. where x is real.7 Asymptotic Approximations to the Fresnel Integral Consider the Fresnel integral (5.118 5 Radiation and Diffraction where the plus sign is used for θ ∈ (0. (5. Recalling (5. Hence show that G(x) ∼ π 1/2 eiπ/4 e−i x + 2 i + O(x −3 ). Here u 3 p is given by u 3 p = −(A/k)e−ikr cos(θ −π/2) H (π/2 − θ). Find the ﬁrst term of an asymptotic expansion of the Fresnel integral for both positive and negative x.137) we use it to combine (5.139) as one from x to −∞ and one from −∞ to ∞. 2x x → ∞. The approach of part (a) must be modiﬁed. Problem 1.105). π).136) to give (5. (5. Find the ﬁrst term of an asymptotic expansion of the Fresnel integral for x 1. Do not use (5.135) for z = x. F(z) + F(−z) = π 1/2 eiπ/4 . Problem 5. Problem 2. Write the integral as the difference between one from zero . 2 (5. (a) Assume x > 0 and consider the integral G(x) = e−i x Show that G(x) ∼ i + O(x −3 ). π/2) and the minus sign for θ ∈ (π/2.138) (5. (5.
References
119
to ∞ and one from zero to x. Expand the integrand of the second integral in a Taylor expansion. Hence show that F(x) ∼ π 1/2 eiπ/4 /2 − x + O(x 3 ), x → 0. (5.142)
References
Ablowitz, M.J. and Fokas, A.S. 1997. Complex Variables, pp. 411–513. New York: Cambridge. Achenbach, J.D., Gautesen, A.K., and McMaken, H. 1982. Ray Methods for Waves in Elastic Solids. Boston: Pitman. Babiˇ , V.M. and Buldyrev, V.S. 1991. ShortWavelength Diffraction Theory. Berlin: c Springer. Born, M. and Wolf, E. 1986. Principles of Optics, 6th (corrected) ed., pp. 565–578. Oxford: Pergamon. Cagniard, L. 1962. Reﬂection and Refraction of Progressive Seismic Waves. Translated and revised by E.A. Flinn and C.H. Dix. New York: McGrawHill. Carrier, G.F., Krook, M., and Pearson, C.E. 1983. Functions of a Complex Variable, pp. 249–283. Ithaca, NY: Hod Books. Copson, E.T. 1935. An Introduction to the Theory of Functions of a Complex Variable, pp. 121–125. Oxford: Clarendon Press. Copson, E.T. 1971. Asymptotic Expansions, pp. 13–14 and 48–62. Cambridge: University Press. Courant, R. and John, F. 1989. Introduction to Calculus and Analysis, Vol. II, pp. 345–350. New York: Springer deHoop, A.T. 1960. A modiﬁcation of Cagniard’s method for solving the seismic pulse problem. Appl. Sc. Res., B 8: 349–356. Doetsch, G. 1974. Introduction to the Theory and Application of the Laplace Transform, pp. 218–230. New York: Springer. Ewing, W.M., Jardetzky, W.S., and Press, F. 1957. Elastic Waves in Layered Media. New York: McGrawHill. Felsen, L.B. and Marcuvitz, N. 1994. Radiation and Scattering of Waves, pp. 370–441. New York: IEEE and Oxford University Presses. Gautesen, A.K. 1979. On matched asymptotic expansions for two dimensional elastodynamic diffraction by cracks. Wave Motion 1: 127–140. Harris, J.G. 1980a. Diffraction by a crack of a cylindrical longitudinal pulse. Z. Angew. Math. Phys. 31: 367–383. Errata, Z. Angew. Math. Phys. 34. Harris, J.G. 1980b. Uniform approximations to pulses diffracted by a crack. Z. Angew. Math. Phys. 31: 771–775. Harris, J.G. 1987. Edge diffraction of a compressional beam. J. Acoust. Soc. Am. 82: 635–646. Harris, J.G. and Pott, J. 1985. Further studies of scattering of a Gaussian beam from a ﬂuidsolid interface. J. Acoust. Soc. Am. 78: 1072–1080. Hinch, E.J. 1991. Perturbation Methods, pp. 52–101. Cambridge: University Press. Holmes, M.H. 1995. Introduction to Perturbation Methods, pp. 47–104. New York: Springer. Jones, D.S. 1952. A simplifying technique in the solution of a class of diffraction problems. Quart. J. Math. 3: 189–196. Jull, E.V. 1981. Aperture Antennas and Diffraction Theory. London: Peter Peregrinus.
120
5 Radiation and Diffraction
Knopoff, L. and Gilbert, F. 1959. First motion methods in theoretical seismology. J. Acoust. Soc. Am. 31: 1161–1168. Noble, B. 1988. Methods Based on the WienerHopf Technique, pp. 11–27 and 48–97. New York: Chelsea. Titchmarsh, E.C. 1939. The Theory of Functions, 2nd ed., pp. 85–86 and 99–101. Oxford: Clarendon Press. Titchmarsh, E.C. 1948. Introduction to the Theory of Fourier Integrals, 2nd ed. Oxford: Clarendon Press. van der Pol, B. and Bremmer, H. 1950. Operational Calculus, pp. 122–132 and elsewhere. Cambridge: University Press. Wong, R. 1989. Asymptotic Approximations of Integrals, pp. 20–31 and elsewhere. Boston: Academic. Zauderer, E. 1983. Partial Differential Equations of Applied Mathematics, pp. 653–658 and 403–405. New York: WileyInterscience.
6
Guided Waves and Dispersion
Synopsis Chapter 6 discusses guided waves and the dispersion they experience. Only the antiplane shear problem is treated. The guided waves are constructed by using partial waves and their dispersion calculated by using the transverse resonance principle. Both harmonic and transient excitations of a closed waveguide are studied by using an expansion of modes. The harmonic excitation of an open waveguide by a line source is also studied, though in this case by using both ray and mode representations. As a last example, we examine propagation in a closed waveguide with a slowly varying thickness, using an asymptotic expansion that combines features of both rays and modes. We close by examining how information and energy propagate at the group velocity.
6.1 Harmonic Waves in a Closed Waveguide We consider a layer of inﬁnite extent in the x1 direction and of ﬁnite thickness in the x2 direction. Within the layer, the coordinate x2 ∈ (−h, h) and the plane x2 = 0 is a plane of reﬂection symmetry. This structure is a waveguide or guide because the waves are forced to propagate in the x1 direction and the guide is closed because waves are completely trapped within the structure. We are interested in learning what kinds of antiplane waves propagate in the guide without, at present, seeking to know how they are excited. Accordingly, we seek possible solutions to the following antiplane problem. In the layer, u 3 must satisfy (2.20), which, rewritten here, is ∂a ∂a u 3 + k 2 u 3 = 0, and at the surfaces µ∂2 u 3 (x1 , ±h) = 0. 121 (6.2) (6.1)
However. . . the mode decays in its direction of propagation (x1 > 0). with boundary condition (6. In this particular case βm is either real or imaginary. Here βm is the lateral wavenumber for the mth mode. . in (x1 .4) where (βm ) ≥ 0 or (βm ) ≥ 0 for x1 > 0. Thus (6. The square root is deﬁned so that. though that problem may not be easy to solve. The transverse wavenumbers for this mode are ±γm . 1. For βm that is real. can be reduced to an ordinary differential equation in x2 by seeking solutions of the form u 3 = f (x2 )eiβx1 . transform variable β. (6. for imaginary values. Cosine is used for m even and sine for m odd. When this is done we ﬁnd that the possible waveﬁelds are described by u 3m = Am cos (γm x2 )eiβm x1 . . (6. 1 (6.4) is a dispersion relation and is similar to that found previously for propagation in periodic structures (there we used κ to represent the effective wavenumber). or its solution may not be particularly informative. the wavenumber and phase velocity in the propagation direction depend on ω (through k = ω/c). where cm is the phase velocity of the mode. Problem 6. or system of such equations. in the x1 direction. (6.2). Problem 6. Thereafter this method of solution can be used to solve other guided wave problems without directly considering the underlying eigenvalue problem. 2.1 An Eigenvalue Problem: A Closed Waveguide Equation (6. βm is given by βm = [k 2 − (mπ )2 /(2h)2 ]1/2 .122 6 Guided Waves and Dispersion As in previous chapters the subscript T has been dropped.3).3) where u 3m is the mth (waveguide) mode for this guide.1). but it propagates in the x1 direction. but not complex. This wave stands in the x2 direction. βm = ω/cm .5) This is equivalent to taking the spatial Fourier transform. Introducing the term1 eiβx1 will reduce a linear partial differential equation. sin m = 0.1 indicates how solutions to this problem can be found by reducing it to an eigenvalue problem in the transverse coordinate. . x2 ) to an eigenvalue problem in x2 for β. we can learn to reconstruct it in a way that uses only the kinematical features of reﬂection and transmission of a plane wave at a boundary. by examining the solution.
1.1 Harmonic Waves in a Closed Waveguide 123 Show that the differential equation in x2 leads to an eigenvalue problem for β 2 . In this book β 2 is considered the eigenvalue. γ2n+1 = (2n + 1)π/(2h). 6. Show that solutions symmetric with respect to the reﬂection plane are f 2n (x2 ) = A2n cos γ2n x2 . while the solid lines indicate the downward propagating set. as indicated in Fig.1.4.1 – one propagating downward and one upward. Equivalently stated. . . Note that phase matching must occur in the x 1 direction. 6. .7) where n = 0. The downward and Fig. where γ = (k 2 − β 2 )1/2 . they must interfere constructively to reconstruct themselves and form a sustained waveﬁeld. There are two sets of plane waves in a guide. These rays are shown in the slowness diagram to the right. as the waves reﬂect back and forth within the guide. Usually the β is linked with γ .1. . (6. 2. The Am are constants. and that those antisymmetric to this plane are f 2n+1 (x2 ) = A2n+1 sin γ2n+1 x2 . (6.8) γ2n = nπ/ h. 6. (6. just as were the individual plane waves in each cell of the periodic structure examined in Section 1. Those that propagate downward are indicated by the solid lines and those that propagate upward by the dashed ones.1 Partial Waves and the Transverse Resonance Principle The central feature of a waveguide is that the waves phase match in the propagation direction and stand in the transverse direction. The dashed lines indicate the upward propagating set of plane waves. The members of each set are referred to as partial waves.6) and γ 2 or −γ 2 is considered the eigenvalue.6. .
or. wherein partial waves are made to stand or resonate in the transverse direction. the following must be true: γ h = mπ/2. 1. The terms A and B are constants. . sin m = 0. . alternatively ω is a function of βm . 2. show that the reﬂection coefﬁcient at x2 = ±h is one. and Am is 2A or 2i A. is referred to as the transverse resonance principle. n = 0. For these two equations to hold simultaneously.6).15) for m = 2n.2. and A=B or A = −B Therefore u 3m = Am cos (γm x2 )eiβm x1 . Beiγ h Be−iγ h = 1.13) where βm is given by (6. (6.9) (6. given in Section 3.4). .2 Dispersion Relation: A Closed Waveguide The most intriguing feature of guided waves is that the lateral wavenumber βm is a function of ω. (6. From our previous work. We have. Each upward propagating wave must be reﬂected into a downward propagating one and vice versa. we can. Therefore.4). n = 0. The dispersion relation . at this stage. 6. This method of ﬁnding the dispersion relation. . . . at the upper and lower boundaries. γm = mπ/2h. (6. u 3 = Bei(βx1 +γ x2 ) .124 6 Guided Waves and Dispersion upward propagating sets are u 3 = Aei(βx1 −γ x2 ) .1. assumed only that (γ ) ≥ 0. (6. by a slight modiﬁcation of that calculation. 2.16) for m = 2n + 1. . .14) (6.12) respectively. The term γ is given by (6. (6. . 1.10) respectively.11) (6. . 1. Ae−iγ h = 1. . Aeiγ h (6. 2.
20) . (6. namely Cm = dω/dβm . there is no propagation for (2hω/c) ≤ mπ and the frequency ωm = mπc/2h is called the cutoff frequency. it is usual to consider ω as a function of βm . 6. can be written as 2hβm = [(2hω/c)2 − (mπ )2 ]1/2 . 6.18) The velocity of real interest.17) A sketch of this relation is shown in Fig. The normalized angular frequency is plotted as a function of the normalized lateral wavenumber. Cm for the mth mode is given by the slope of the mth branch of the dispersion relation. Unlike cm .6 the demonstration of the second. (6. The time average of a waveﬁeld quantity Gm (βm x1 − ωt. when the frequency is a free variable. x2 ) for the mth waveguide mode is deﬁned as Gm = 1 T t+T t h −h Gm (βm x1 − ωτ.2. While the βm may be thought of as a function of ω. Here we demonstrate the ﬁrst assertion and leave to Section 6. (6.19) The group velocity is both the velocity of energy propagation and the velocity with which information propagates. The phase velocity cm for the mth mode is cm = ω/βm . is the group velocity Cm . A sketch of the dispersion relation for antiplane shear modes in a closed waveguide. x2 )d x2 dτ. however. (6.2.6. For the mth mode.1 Harmonic Waves in a Closed Waveguide 125 Fig.
K and U.126 6 Guided Waves and Dispersion The average Gm does not depend upon x1 or t because at a given x1 .25) The average (6.22) For propagation βm is real. instead of proceeding directly as we just did. though care is required because not all equations describing waveﬁelds exhibit equipartition. t) in the mth mode is Em = 1 ρ ∂t u 3m ∂t u 3m + 1 µ ∂a u 3m ∂a u 3m .26) Using (6. The Lagrangian density L for antiplane shear motion is given by L= 1 2 ρ(∂t u 3 )2 − 1 2 µ[(∂1 u 3 )2 + (∂2 u 3 )2 ].18). .18). The instantaneous ﬂux of energy density Fm (x1 . a fact noted previously when deriving (2. The instantaneous energy density Em (x1 . but can arise from the structure of the equation itself.25). x2 . where the kinetic and internal energy densities. (6. Using this principle when calculating the average energy density. Direct calculation shows that Fm / Em = c2 /cm = Cm . for the mth mode. therefore. Dispersion need not arise from a geometrical constraint. Clearly. (6.21) (6. m 2 (6. the converse is also true. we ﬁnd that its average is 2 2 Em = 1 ρcm hβm Am A∗ . as it does with a waveguide. and the average ﬂux is 2 Fm = 1 µcm hβm Am A∗ .23) (6. 2 2 and. the ratio Fm / Em . t) in the mth mode is Fm = −µ∂1 u 3m ∂t u 3m . were ﬁrst deﬁned by (1. as the following problem demonstrates. It follows then that K = U .24) The velocity of energy propagation along the axis of the waveguide for the mth mode is. the argument βm x1 − ωτ is simply a translation of the argument ωτ .20) with (2. usually reduces the amount of calculation needed. using (6.20) we readily ﬁnd that L = 0.20) can also be used to demonstrate the equipartition of energy. m 2 (6. x2 .
multiply each of (6. (6. Assume that the wave propagates to the right. where c = ω/k and C = dω/dk.28) by ∂t ϕ and conﬁgure the result to coincide with (6.33) .6.2 Dispersion 1 127 Problem 1. One way to ﬁnd these is to construct conservation laws directly from (6. a dissipative term.1 Harmonic Waves in a Closed Waveguide Problem 6. the second the instantaneous ﬂux. (6. Such laws have the form ∂t + ∂x =− .27) and the equation for ﬂexural motion in a rod. show that the instantaneous energy density for the equation for ﬂexural motion is 2 E = (∂t ϕ)2 /2 + a 2 ∂x ϕ 2 2 (6. (6. the Klein– Gordon equation. 2 ∂t2 ϕ − a 2 ∂x ϕ + b2 ϕ = 0.28). on the righthand side. Demonstrate that the group velocity is the velocity of timeaveraged energy transport.31) Also. the righthand side is zero.30).27) and (6. In both cases ﬁnd a dispersion relation in the form ω = ω(k) by seeking a wave solution of the form ϕ = Aei(kx−ωt) . 4 ∂t2 ϕ + a 2 ∂x ϕ = 0. show that the instantaneous energy density for the Klein–Gordon equation is E = (∂t ϕ)2 /2 + a 2 (∂x ϕ)2 /2 + b2 ϕ 2 /2 and the instantaneous ﬂux is F = −a 2 (∂t ϕ)(∂x ϕ). Consider the following two differential equations.29) where A is a constant. (6. (6. For the equations being studied here.30) The ﬁrst box is the instantaneous energy density. Thus. To ﬁnd a conservation law.28) The terms a and b are constants. and the third.27) and (6. so that you can calculate their average values. To do so you will need to know both the instantaneous energy density and instantaneous ﬂux. Calculate the phase and group velocities c and C.32) (6.
. where c is the speed of sound) in the x1 direction is given by (∂t + U ∂1 )2 ϕ = c2 ∇ 2 ϕ.36) (6.2 Harmonic Waves in an Open Waveguide We continue to consider an antiplane problem having the governing equation.18). The interior of the Fig. where (x. Problem 2. (6. (6. Note that the positive x2 direction points into the interior. Figure 6. The equation describing acoustic waves in a wind having speed U (<c. t) is a ﬁxed coordinate system. namely that the vertical dashed line in the diagram to the right not intersect the lower slowness circle. A slowonfast structure can support trapped waves in the layer.1).3 indicates the geometry.35) ﬁnd the dispersion relation. 6. complete the demonstration. Drawing slowness diagrams for the layer and halfspace indicates the condition for waves to be trapped in the layer. The vector k is needed because the propagation environment is anisotropic.3. 6.34) Using (2. We now consider a layer on a halfspace. (6.128 6 Guided Waves and Dispersion and the instantaneous ﬂux is 3 2 F = a 2 (∂t ϕ) ∂x ϕ − a 2 (∂t ∂x ϕ) ∂x ϕ . Using a solution of the form ϕ = Aei(k·x−ωt) .
40) (6. but may radiate into the halfspace.39) (6. s 2 must be imaginary. consider the slowness diagrams sketched in Fig. If the waves are to remain trapped in the layer.2. u 3 = Aei(βx1 +γ x2 ) . waves launched in the layer phase match to waves in the halfspace for which ¯ s 2 is real and are therefore not trapped. This conﬁguration is sometimes referred to as a slowonfast guiding structure. To understand why waves are trapped. 6. u 3 (x1 0− ) = u 3 (x1 0+ ). the waves in the layer able to phase match to waves ¯ ¯ ¯ decaying into the halfspace are those for which s1 = s1 > s.37) where µ and µ are the elastic constants for the layer and halfspace. ¯ A fastonslow structure (s < s) does not guide waves so that. so that the corresponding slownesses ¯ ¯ are s and s. rewriting it once again. The wavespeed in the layer is c ¯ and that in the underlying halfspace c. with the sign of s 2 selected so that decay takes place with depth. the upward and downward propagating sets of waves are u 3 = Bei(βx1 −γ x2 ) . ¯ (6.2 Harmonic Waves in an Open Waveguide 129 layer occupies x2 ∈ (−h. ∞). s1 must be the same in both the layer and halfspace.41) . The equation. Therefore.1). we do not consider it. 0) and its wavespeed is c (k = ω/c). 0− ) = µ∂2 u 3 (x1 .4.3 the problem is reduced to solving an eigenvalue problem in the x2 direction. except in Problem 6. respectively. 0+ ). A slowonfast structure is one for which s > s.6. At x2 = −h.3. In Problem 6. respectively. The equation of motion in the layer is (6. µ∂2 u 3 (x1 .1 Partial Wave Analysis In the layer. and at x2 = 0. The equation of motion in the halfspace is also given by (6. ¯ The layer is called an open waveguide because there is now the possibility that waves may not remain trapped. 6.1) with a different wavenumber. However. waves will remain trapped provided the wavespeed in the layer is less than that in the halfspace. If the waves are to phase match at x2 = 0. When s1 ≤ s < s. The interior of the halfspace occupies x2 ∈ (0. µ∂2 u 3 (x1 .38) (6. −h) = 0. is ¯2 ∂α ∂α u 3 + k u 3 = 0. unable to radiate ¯ ¯ away. The slowonfast structure is analyzed next by using partial waves and the transverse resonance principle. (6. ¯ ¯ ¯ where k = ω/c and c is the wavespeed in the halfspace.
47) (6. The reﬂection and transmission coefﬁcients at x2 = 0 are.45) The R(γ ) and T (γ ) are gotten from (3. ¯¯ (6.4. in addition to being reﬂected.21) and (3. where α ≥ 0. Ae−iγ h /Beiγ h = 1.46) These last two equations indicate that plane waves incident on the lower boundary must. ¯¯ ¯¯ e−2iγ h = (µγ − µγ )/(µγ + µγ ). In this case (6.130 6 Guided Waves and Dispersion In the halfspace the set of waves is ¯ u 3 = Aei(βx1 +γ¯ x2 ) . in the case of the open guide.43). γ = k cos θ0 . respectively. be transmitted into those that decay or propagate away from the boundary.49) reduces to tan γ h = µα/µγ . (6.44) (6. (6. We discuss the branches of these radicals more ¯ carefully in Section 6. albeit indirectly.49) or (6. Exactly as in the case of the closed guide. ¯¯ (6. This statement is consistent with the principle of limiting absorption. we also have ¯ A/A = (2µγ )/(µγ + µγ ). where β = ωs1 .50) Recalling (6.46) and (6. From (6.50) is the dispersion relation for the structure.4. ¯ γ = (k 2 − β 2 )1/2 . at x2 = −h.48) (6.42) The waves in the halfspace decay or propagate downward.43) where (γ ) ≥ 0 and (γ ) ≥ 0. and γ = k cos θ2 .47) we ﬁnd that for nontrivial solutions. given in Section 4. ¯¯ (6. To satisfy the equations of motion. ¯¯ ¯¯ Moreover. so that γ = i α. we see that (6. R(γ ) = (µγ − µγ )/(µγ + µγ ). ¯ (6. γ = (k 2 − β 2 )1/2 .49) ¯ ¯ ¯ ¯ The case of interest is s < s1 ≤ s. and at x2 = 0. It gives ω = ω(β) or β = β(ω). The antiplane waves . B/A = (µγ − µγ )/(µγ + µγ ). ¯¯ ¯¯ T (γ ) = (2µγ )/(µγ + µγ ).22) by noting that β = ¯ ¯ ¯ k sin θ0 = k sin θ2 .
and thus construct the waveﬁelds in the layer and halfspace. The preceding analysis gives the discrete eigenvalues and eigenfunctions. The transverse wavenumbers γ and γ are given by (6. unless the reader is quite perceptive. ¯ ¯ ¯ ¯ Find the constant D in terms of C. with boundary conditions (6. 6. Show that solutions to the differential equation in x2 and its boundary conditions can be expressed as f (x2 ) = C cos[γ (x2 + h)]. Use the condition that g(x2 ) → 0 as x2 → ∞. .4. (6. Knowing the dispersion relation.50)? The problem just discussed is called a singular eigenvalue problem because it has a continuous spectrum as well as a discrete one. mentioned previously. gives either ω = ω(β) or β = β(ω).29) and (6. x2 ∈ (0.30). What condition stated previously in this problem must be modiﬁed to ﬁnd the continuous eigenvalues and eigenfunctions? 6. though the slopes are 1 ¯ and c/c. First we identify the boundaries between which β is real (assuming ¯ ¯ that c < c). u 3 = g(x2 )eiβx1 . (6.2. 0). we begin with a diagram such as that shown in Fig.53) where γ = i α.2 Harmonic Waves in an Open Waveguide 131 guided by the layer are called Love waves. but gives one little information about the continuous spectrum.32). α ≥ 0. where three distinct regions are identiﬁed. Note that the absence of a plane of reﬂection symmetry means that the modes do not divide into symmetric and antisymmetric ones. These boundaries are denoted by c and c. Friedman (1956).2 Dispersion Relation: An Open Waveguide The transcendental equation. To examine this dispersion relation. 1956) in x2 by looking for solutions of the form u 3 = f (x2 )eiβx1 . can be reduced to a singular eigenvalue problem (Friedman. ∞). x2 ∈ (−h. Problem 6. will help the reader understand the underlying mathematics of singular eigenvalue problems.3 A Second Eigenvalue Problem: An Open Waveguide Note that (6. Tolstoy and Clay (1987) will give the reader a reasoned physical explanation for the continuous spectrum.51) (6.6.52) Note that the form of the function in x1 has been chosen to give a wave propagating in the positive x1 direction and ensure phase matching at x2 = 0. (6. ¯ g(x2 ) = De−αx2 .31) and (6.43).50). ¯ we can ﬁnd B and A in terms of A. in combination with (6. (6.36). Can you recover the dispersion relationship.
55) The argument of the tangent in (6. as can be seen from examining (6. Consider the following limits. It ¯ ¯ ¯ is readily seen that no solution for real β is possible and therefore no trapped ¯ wave propagates in the x1 direction. let βh move toward zero through ¯ region 2 and reason as before.4.55) approaches a ﬁnite limit or zero because the right side remains positive. There is only one possibility. 6. Second.50). Write (6. where α and α are real and positive. ¯ Solutions for real β are possible in region 2. 2 /β 2 ) − 1]1/2 µ[(k 2 (6.54) Again it is clear that there is no solution for β that is real. γ = ±iα and γ = i α. ¯¯ (6. In region 3 both γ and γ are real. In this case (6. by allowing the layer to become an inﬁnite number of wavelengths thick. though there are solutions for complex β. its response is no longer inﬂuenced by the presence of the halfspace. and approaches zero or +∞. exactly what we would expect for γ and γ both real. The absence of roots for β real tells us that there are no trapped waves. In other words. First. There are three ¯ regions whose boundaries are formed by the two wavespeeds c < c. The normalized angular frequency is plotted as a function of the normalized lateral wavenumber.4. For the lowest mode k/β → 1 as βh → 0 and .132 6 Guided Waves and Dispersion Fig.43) becomes tan(γ h) = −i µγ /µγ . In region 1.50) as tan βh[(k 2 /β 2 ) − 1]1/2 = ¯ µ[1 − (k /β 2 )]1/2 ¯ . let βh → ∞ within region 2.3. We shall brieﬂy discuss this possibility in Section 6. A sketch of the dispersion relation for Love waves. namely k/β → 1.
fastonslow and slowonfast. while βh and ωh remain ¯ ﬁnite.4 Reﬂection From a Layer Referring to the geometry of Fig. α = 0 and [(ωn h/βhc)2 − 1]1/2 βh = nπ.58) ¯ ¯ ¯ The wavenumbers β = k sin θ0 and γ = k cos θ0 . where A∓ = cos θ0 cos(kh cos θ) ∓ i and ¯ sin θ/c = sin θ0 /c.59) ¯ µc cos θ sin(kh cos θ). Keep in mind that we have. the group velocity does not vanish at this point. so far. In particular. for some ω = ωn . . In the reﬂection problem the layer is viewed from the halfspace. However.56) where n is a positive integer. If A1 is the unknown amplitude of the reﬂected wave. (6.6. In this case. restricted β to positive real values and have not deﬁned the branches of γ and γ carefully. µc ¯ (6. and we are no longer as concerned as we were here with whether the structure is a slowonfast or fastonslow one. The frequency ωn is that at which the waves move from being trapped to radiating into the halfspace. ¯ The outcomes of the problem just described should be compared with those of the reﬂection problem that follows.57) be incident to the layer from the halfspace. calculate the reﬂection coefﬁcient of the layer. namely R(θ0 ) = A1 /A0 . Show that R(θ0 ) = A− (θ0 )/A+ (θ0 ). (6. for the higher modes k/β → 1. This frequency is also called a cutoff frequency. Unlike the case of a closed waveguide. Make no assumption at present as ¯ to the relative magnitudes of c and c. though the term transition frequency might be more appropriate.3. Problem 6. how might a trapped wave be excited by using a disturbance incident from the halfspace when the structure is slow on fast? Note that this reﬂection coefﬁcient is frequency dependent. 6. let the plane wave ¯ u 30 = A0 ei(βx1 −γ¯ x2 ) (6.2 Harmonic Waves in an Open Waveguide 133 ¯ ωh → 0. The viewpoint there is rather different. The reader should contrast the two possible cases.
but starting at x1 = 0 and extending through positive values of x1 to inﬁnity. for n ≥ N . while those in the second are evanescent. T (x2 ) = T (−x2 ). and.5 A Waveguide Mode Expansion In part.3. (6. x2 ) = N −1 n=0 ∞ nπ nπ (−An ) An x2 eiβn x1 + x2 e−αn x1 . At x1 = 0 the waveguide is excited with a traction T (x2 ) symmetric in x2 so that µ∂1 u 3 = −T (x2 ). One common way to solve such problems is to expand the waveﬁeld in the guide in terms of its modes. Far from the source. 1999). The frequency of excitation is ω. 1978. Folguera and Harris. βn is given by (6.1) and (6. only those that propagate make a signiﬁcant contribution. While these questions have still not been satisfactorily answered for inplane elastic waves in a guide with an end (Miklowitz. while the other extends to inﬁnity. The geometry of the problem is now described by a layer similar to that shown in Fig. write down explicit expressions for the coefﬁcients An .61) propagate. Using the orthogonality of the modes (eigenfunctions). the antiplane modes are simply the terms of a Fourier series.62) is reversed.1.2) specify the problem. . cos cos iβn h αn n n=N (6. how might the forced waveﬁeld be expanded? Show that a representation of the solution is u 3 (x1 .3 Excitation of a Closed Waveguide 6. As a minimum we need a means of calculating the coefﬁcients of the expansion – an orthogonality relation – and some assurance that the expansion is complete. (6.1 Harmonic Excitation We start by considering a closed waveguide that is harmonically excited by applying an antiplane traction at one end. Note that the modes in the ﬁrst sum of (6.134 6 Guided Waves and Dispersion 6.4).60) What condition should be imposed upon the forced disturbance as x1 → ∞? Noting the symmetry of the excitation. βn = iαn . 6. To do so requires that a mathematical framework be in place. so that in this case these questions are settled. Problem 6.61) where N is the smallest n for which the inequality nπ/ h < ω/c (6.
(6.60) with µ∂1 u 3 = −T (x2 )δ(t). (6. we replace (6. the in (6. Integrals of the form of (6. Following the arguments in Section 3. The n = 0 term is easily inverted.3.6. In Problem 6. We ﬁx t/x1 and let x1 be the large parameter.3.36).63) Moreover.2 Transient Excitation 135 We next consider a transient excitation. 6.4.64) Using (1. we approximate (6.4. (6. For a . we ﬁnd the branch point is at c nπ/ h − i . t < 0. t) = ∞ n=0 nπ (−An ) cos x2 π h ∞ 0 eit(βn x1 /t−ω) dω. even when they can be evaluated in closed form. Note. as well.67) by using the method of stationary phase described in Section 5. The integration contour in the ω plane must pass above the branch point cnπ/ h − i . we assume that the waveguide is quiescent until the excitation is applied. i(ω + i )/c (6. we have cut the plane so that (βn ) ≥ 0 for all ω. giving ∞ 0 ei(ω+i )(x1 /c−t) dω = −π c H (t − x1 /c). x2 .5.66) The n > 0 terms are less straightforward. u 3 = ∂t u 3 = 0.67) can be set to zero.65) iβn Using the principle of limiting absorption. we synthesize the transient response as u 3 (x1 . Accordingly. replace ω with ω + i . The integral to be evaluated is ∞ 0 ei x1 (βn −ωt/x1 ) dω.67). for ω < 0. Note that c nπ/ h is a cutoff frequency so that propagation has ceased when ω < cnπ/ h. That is.67) The left half of Fig. Once the contour has been determined. where > 0.3 Excitation of a Closed Waveguide 6. By symmetry.5 shows the complex ω plane and the branch cuts for n > 0. that (βn ) > 0 in the ﬁrst and third quadrants. iβn (6. the branch point is at −cnπ/ h + i . Using the principle of limiting absorption for ω > 0.3. give the clearest description of the waveﬁeld at points far from the source.
After some simpliﬁcation. t). [1 − (x1 /ct)2 ]1/2 x1 /ct < 1. x1 /ct < 1 for propagating waves and (βns .70) x1 /ct < 1. namely nπ/ h ωns = .5. iβn (nπ/ h)(βns x1 )1/2 (6. c [1 − (x1 /ct)2 ]1/2 and βns = nπ h x1 /ct . ωns ) as a function of (x1 . The contour extends from ω = 0 along the real axis. while t/x1 is held ﬁxed. (6. the stationary point is given by dβn /dω = t/x1 . (6.69) No signal can travel faster than c (Problem 6. 6. The righthand ﬁgure shows the dispersion curve for the nth mode with the stationary point indicated by (βns . ωns ). the particle . (βn ) ≥ 0 ∀ ω. t) (more precisely x1 /t). is then approximated as ∞ 0 c (2π)1/2 ei x1 (βn −ωt/x1 ) dω ∼ ei x1 (βns −ωns t/x1 ) e−iπ/4 . The lefthand ﬁgure shows the complex ω plane. (βn ) > 0 in the upper right quadrant of the ω plane. Therefore. given point in (x1 . (6.6). The integral.67).136 6 Guided Waves and Dispersion Fig.68) Solving this equation gives the stationary point (βns .71) as x1 → ∞. (6. ωns ) are real.
Problem 6. At (x1 . This is not the velocity with which the phase θ(x1 . for ﬁxed x1 . We have already noted in (6. Rather.6. t) ∼ π h (2h/n) x1 [(ct/x1 )2 − 1]1/2 1/2 137 × cos (nπ/ h)x1 [(ct/x1 )2 − 1)]1/2 + π/4 . x2 . 6. and the total particle displacement can be written as u 3 (x1 . Further. (6.5. Dai and Wong (1994) give a correction for this case. t) the nth mode is approximated by an expression of the form An (x1 . (6. x2 .6 No Signal Travels Faster than the Velocity c Using the problem just discussed. Now we also note that it is the velocity with which the frequency ωns propagates to the point (x1 . x2 .75) and note that.72) shows that the nth mode is O(n −1/2 ). show that no signal travels faster than c. t)ei(βns x1 −ωns t) . with the ω plane cut as . note that as x1 /t sweeps through its values from zero to c. t). In fact. where Cn is the group velocity dω/dβn .68) provides a onetoone mapping from the t plane to the ω plane.3 Excitation of a Closed Waveguide displacement of the nth mode is approximated as (−An ) nπ c cos x2 u 3n (x1 . ωs takes on all its values from cnπ/ h to inﬁnity. in the upper right quadrant of Fig. t) = (βns x1 − ωns t) propagates outward. t) ∼ A0 c H (t − x1 /c) + ∞ n=1 (6. x2 . t).73) Examining (6.72) u 3n (x1 . a constant phase θ propagates at a rate cns = ωns /βns . Write each integral over ω as ∞ π e−iω(t−x1 /c) 0 ei(βn x1 −ωx1 /c) dω iβn (6. (6. Note that the stationary phase approximation breaks down for x1 /t near c.68) can be viewed as x1 /t = Cn (ωns ). Note that (6. The higherorder modes therefore make a weaker contribution than do the lowerorder ones. ωns ).74) This is a group in the sense that it arises from a cluster or group of wavenumbers and frequencies in the neighborhood of (βns .25) that Cn is the velocity of timeaveraged energy transmission.
for t < x1 /c. This is more or less what we did in solving Problem 6. Expanding the source in these eigenfunctions causes the equation in x1 to be forced by a coefﬁcient of this expansion. 2 The dispersion is said to be anomalous when the group velocity exceeds the velocity of propagation of a harmonic plane wave in the medium (Sommerfeld. Lastly. One must take care to distinguish between anomalous dispersion caused by some approximate derivation of the equation being studied and that caused by the underlying physical situation. The presence of frequencydependent attenuation confuses the issue even further. though the modal expansion already contained the solution to the differential equation in x1 . The domain of the transverse eigenvalue problem is ﬁnite and therefore the eigenvalues are discrete. Moreover. such as the one discussed in Problem 6. the reader is asked to use a Laplace transform. indicating that no signal is present for t < x1 /c. 1964a.5. In this case the group velocity is no longer a measure of the speed at which information or energy propagates. rather than use a Fourier synthesis to construct the transient solution. it is true generally. and velocities that characterize the speed at which information is propagated and at which energy is propagated must be deﬁned as best one can. after invoking the orthogonality of the eigenfunctions.77) This integral must be zero (why?).5. Moreover. While this result is the outcome of the analysis of a speciﬁc problem. involving each mode separately. ω→∞ iβn lim (6. .76) Therefore. In Problem 6. (−αn ) (6.7 the reader is asked to solve much the same problem by using a continuous eigenfunction expansion in x1 ∈ (0.138 indicated. ∞). thus reducing the partial differential equation to an ordinary one in x2 . Brillouin. For separable problems. In such cases a speciﬁc signaling problem must be worked through. it indicates that the group velocity does not exceed the wavespeed of the medium. so that that step was leaped over. 1960). the anomalous dispersion may only be apparent and not real. show that the integral can be written as ∞ π ie p(t−x1 /c) 0 e−(αn x1 − px1 /c) dp. in reducing the partial differential equation to an ordinary one in x1 . 6 Guided Waves and Dispersion ei(βn x1 −ωx1 /c) = 0.2 a result that is also generally true. show that the integration contour may be distorted from one along the real ω axis to the one along the imaginary axis. Moreover. an expansion of the solution in the eigenfunctions of the transverse eigenvalue problem results.
x1 )dξ (6. x2 . as we have done previously. x2 .39). t) = 0 ∞ u 3 (x1 . Why would one not work with a sine transform? At what point has the reader imposed the condition that waves be outgoing from the source? Can the reader express (6. sketched on the left in Fig. Use the cosine transform ∗ u 3 (ξ.3. t) cos(ξ.7 A Continuous Eigenfunction Expansion 139 Continue to consider the problem just solved.4 Harmonically Excited Waves in an Open Waveguide Problem 6. It is perhaps easier to set ∗ v 3 (ξ. This problem may be solved by constructing the solution by using eigenfunction expansions. x2 .80) and then approximate v3 (x1 . p) where v3 ¯ is the particle velocity. This method of construction will show how ray representations and modal expansions are related to one another. t) cos(ξ x1 )d x1 . Find ∗ u 3 and begin to invert the ¯ transforms. This integral can also be inverted exactly. However. 6. Our discussion follows a very similar one in Brekhovskikh (1980). Find ∗ v 3 (ξ. x2 . with the excitation (6. using the method of stationary phase. x2 . x2 . t) = 1 2πi ∞ 0 +i∞ −i∞ ∗ v 3 (ξ. x2 .6. with x1 /t held ﬁxed. x20 ). t) = 2 π ∗ v 3 (ξ. where x20 < 0. x2 . p) = p ∗ u 3 (ξ. x2 . This is because we have chosen to work with a cosine transform. Except for the absence of a source. (6. t)e pt dt.78) followed by the Laplace transform ∗ ¯ u 3 (ξ.63).4 Harmonically Excited Waves in an Open Waveguide We next return to the layered antiplane structure.38) and (6.79) ¯ to ﬁnd an ordinary differential equation in x2 . we shall construct the radiated waveﬁeld by superposing collections of partial waves. to calculate the waveﬁeld radiated by a harmonic line source placed at (0. p)e pt dp ¯ (6. x2 .81) as a sum of waves propagating solely in the positive x1 direction? 6. The reader may be surprised to ﬁnd that his or her expression for v3 contains both forward and backward propagating waves.81) for large x1 . (6. the equations of motion and the boundary conditions are given by (6. . p) = 0 ∞ ∗ u 3 (ξ.
R(α)eik[x1 cos α−sin α(x2 −2h−x20 )] . Equation (4. The contour C begins near π − i∞ and ends near i∞.6. x2 ) without reﬂection. when added together.1. R(α)eik[x1 cos α−sin α(x2 +x20 )] .82). (6. x2 ) in the layer. we ﬁnd that there are four distinct partial waves that reach (x1 . to ensure phase matching. C (6. How then does each plane wave in the integral. is here rewritten by expanding cos(θ − α) to give u i3 = − i F0 4π eik(x1 cos α+x2 −x20  sin α) dα. behave in the layered environment? Consider an arbitrary observation point (x1 . x20 ). A sketch showing the ﬁrst four partial waves reaching the observation point (x1 .19). 6. x2 ) and. Given that the x1 dependence must be of the form eik cos α x1 . particle displacement u i3 excited by a line source at (0. .1 The Waveﬁeld in the Layer We begin by representing the waveﬁeld radiated by a line source as a spectrum of plane waves.83) The ﬁrst partial wave reaches (x1 . (6.6. x2 ) and that all the subsequent partial waves can be constructed from these four by successively reﬂecting them in the planes x2 = 0 and x2 = −h. the second after being Fig. x2 ). with α replacing ξ . eik[x1 cos α+sin α(x2 +2h+x20 )] . 6. The dashed lines indicate the propagation paths and the solid ones indicate the wavefronts passing through (x1 .140 6 Guided Waves and Dispersion 6.82) F0 = A/k. eik[x1 cos α+sin α(x2 −x20 )] . using the result of Problem 4.4. we ask what partial waves can reach the point (x1 . satisfy the boundary conditions. x2 ). By sketching rays and wavefronts. where A is a dimensionless constant. The expression (6.82) gives the free space. as suggested in Fig. in both the layer and halfspace. where x1 > 0 and x2 > x20 .
The actual path of the ray is indicate by the dashed line within the layer in Fig.7. the fourth term. with γ = k sin α and γ = k sin α. we put (6. we approximate one of the terms by the method of steepest descents.84) by the method of steepest descents gives an inﬁnite sum of terms such as (6. Approximating each term in (6. By adding these four partial waves and all the subsequent reﬂections and invoking superposition. Note that the argument of the function R is now ¯ given as α (rather than as γ or θ0 ) in keeping with the structure of the present calculation.21) and (3.85) in a form previously studied in Section 5. namely u 04 = − 3 i F0 4π R(α)eik[cos α x1 −sin α (x2 −2h−x20 )] dα. This is a ray representation of the waveﬁeld in the layer. The angles α and α are related by the phasematching condition. The reﬂection coefﬁcient at x2 = −h is 1 and at x2 = 0 is R(α).7. where R(α) is given by (6. Or it can be gotten from (3. Its asymptotic approximation is therefore u 04 = − 3 i F0 4π R(α)eikr04 cos(α−θ04 ) dα ∼ 2π kr04 1/2 C F0 R(θ04 )eikr04 eiπ/4 .23) by noting that ¯ ¯ ¯ ¯ α = π/2 − θ0 . ¯ namely k cos α = k cos α.44).85) The geometry is shown in Fig.86).4 Harmonically Excited Waves in an Open Waveguide 141 reﬂected once at the x2 = 0 boundary. It is given by u3 = − i F0 4π eik cos αx1 eik sin α(x2 −x20 ) + R(α)e−ik sin α(x2 +x20 ) + R(α)e−ik sin α(x2 −2h−x20 ) (6. C (6. By setting x1 = r04 cos θ04 and x2 − 2h − x20 = −r04 sin θ04 .84) C +e × ik sin α(x2 +2h+x20 ) ∞ [R(α)]m eik sin α(2mh) dα m=0 To understand this representation further. We consider m = 0. All subsequent reﬂected partial waves can be constructed from these four. we construct the response of the layer to the excitation. 2h + x20 ). .4. 4π (6.86) This term then appears as a cylindrical wave radiated by an image source at (0.6. the third after being reﬂected from the x2 = −h boundary. and the last after being reﬂected ﬁrst from the x2 = −h boundary and second from the x2 = 0 boundary. 6. 6.
88) C −ik sin α x20 1 − R(α)eik sin α(2h) F0 = A/k has been used to indicate the dimensions of u 3 . When x2 < x20 . Recall that we have assumed x2 > x20 . where β = k cos α. the two coordinates in the above expression are interchanged.87) l=0 j=1 where each term is similar to that given by (6. Is this a useful representation. in what circumstances? By summing the series in (6.142 6 Guided Waves and Dispersion Fig. 6. lj (6. These radicals are deﬁned so that (γ ) ≥ 0 and (γ ) ≥ 0 ∀ β.43).82). However.8 shows the β plane with its branch cuts and several poles. To understand this representation we must examine the complex α plane.84) to give u3 ∼ ∞ 4 u3 . and. Figure 6.7.86). A sketch of the geometry for the steepest descents approximation of the fourth partial wave.84). Problem 6. ¯ . the waveﬁeld can be recast as u3 = − × iA 4kπ e eik cos α x1 [eik sin α x2 + R(α)e−ik sin α x2 ] + eik sin α (2h+x20 ) dα. the discussion will be clearer if we ﬁrst examine the β plane. if so. (6.8 Ray Representation Verify the statements just made by asymptotically approximating each term in (6. Recall that this is the Sommerfeld transformation and that it was this transformation that lead us to (6. The transverse wavenumbers γ and γ were ¯ given in (6.
11. In the various quadrants the real parts take values as follows: quadrants ¯ 1 and 2.9 shows the α plane and the contour C. A sketch of the complex β plane. 6. 6. The physical sheet is the one for which (γ ) ≥ 0 and (γ ) ≥ 0 ∀ β.4 Harmonically Excited Waves in an Open Waveguide 143 Fig. The two branch points are α and π − α. those to the right are π − α1 and π − α2 . (γ ) < 0 and (γ ) < 0. These ¯ ¯ quadrants correspond to their primed counterparts in the α plane in Fig. ¯ . This is the Riemann sheet. Figure 6. (γ ) > 0 and (γ ) > 0. Poles on the lower sheet progressively ¯ ¯ move to the physical sheet.6. 6. that ensures that the waves radiate or decay away from their source (satisfy the principle of limiting absorption). The branch ¯ points are now given by α and π − α.8. ¯ ¯ ¯ ¯ Fig.4 and 5. showing the branch cuts and poles that have emerged onto the physical sheet. where α is deﬁned by k = k cos α. as we have seen in Sections 3. The α plane.4. The poles to the left are α1 and α2 . We call this the physical sheet and the others the unphysical ones. quadrants 3 and 4. popping through the branch points as ω is increased.9.4.
49). These are indicated by the crosses in Fig.8 and 6.9 Modal Representation Show that the residue contributions of (6. become part of a modal sum. The next ¯ problem indicates how the residues from the poles on the physical sheet.53). in the integrand of (6. there is a pair of poles ¯ on the physical sheet of the β or α plane corresponding to each guided mode that is trapped in the layer. 6. The sum of residues gives the waves trapped in the layer or the sum over the discrete .10.8. popping up through the branch point α in the α plane. Noting that γ = k sin α. π − α1 ) and (α2 . The modes that could radiate into the interior correspond to poles that lie on an unphysical sheet. these poles move from the unphysical sheet to the ¯ physical one.10. For a given ω = ω. In Figs. (6.9. The dispersion relation and the points on it that correspond to the poles shown in Figs.88) are given by the dispersion relation e−2iγ h = R(α). we see that the poles of the integrand in (6. The poles of the integrand therefore give us the trapped modes of the waveguide. 6. 6. π − α2 ). Note that there will be a branch cut integral as indicated in the ﬁgure.51) through (6. similar to that for a closed waveguide.9. Do this by deforming the contour C to that shown in Fig.89) which is a restatement of (6. ±β1 and ±β2 or (α1 . Problem 6.8 and 6. These points are shown on the dispersion curves in Fig. there are two pairs. 6.88). As ω is increased. 6. 6.144 6 Guided Waves and Dispersion Fig.11.88) give the Love waves of (6.
2.6. 6. 2 .90) The remainder can be constructed from these two.84) by using the Poisson sum formula. Integrating over all the partial waves gives u3 = − × i F0 4π ∞ m=0 ¯ T (α)eik cos α x1 ei k sin α x2 e−ik sin α x20 + eik sin α (2h+x20 ) ¯ C [R(α)]m eik sin α (2mh) dα. then this representation would give a ray series for the waveﬁeld in the halfspace.4. The quadrants 1 . The combination is the modal expansion of the waveﬁeld in the layer. whereas modal representations are most descriptive a considerable distance from it. whereas the branch cut integral gives the waveﬁeld that can radiate into the lower halfspace or the sum over the continuous eigenfunctions.11. 6. x2 ) are ¯ T (α)eik cos α x1 e−ik sin α x20 ei k sin α x2 . 6. discussed in Section 1.84). and 4 correspond to quadrants 1.4 Harmonically Excited Waves in an Open Waveguide 145 Fig. ¯ ¯ T (α)eik cos α x1 eik sin α (x20 +2h) ei k sin α x2 . eigenfunctions. ¯ (6. and 4 in the β plane in Fig. The two fundamental partial waves reaching (x1 . The deformed contour for Problem 6. This is true even when we cannot sum the series (6.91) If each term were approximated by the method of steepest descents. By .9.3. Propagation to the right is being considered. Ray representations are useful for observation points near the source. The modal representation can be derived directly from (6.2 The Waveﬁeld in the HalfSpace The waveﬁeld in the halfspace is constructed in much the same way as was that in the layer. (6.8. at points where the waveﬁeld has lost some sense of how it was excited and has adapted to its propagation environment.
and therefore on the unphysical sheet in the α plane. an unphysical sheet of ¯ the β plane. and 6. we continue to assume that there are two real roots.1. we take (β) > 0 so that the waves in the layer radiate or leak into the interior as they propagate to the right. They lie on the (γ ) ≥ 0.8. there are also roots to the dispersion relation if β is allowed to be complex.8 and 6. 6. 6. 6.4 and 6. These waves can be quite important for some observation points. (γ ) ≤ 0 sheet. present on the physical sheet.146 6 Guided Waves and Dispersion carrying out the summation. β1 = k cos α1 and ¯ β2 = k cos α2 . We consider a slow variation. . Moreover. ¯ That is why they do not lie on the physical sheet of the Riemann surface. We call them leaky waves or leaky modes. To ensure that waves radiate away from their source. however. This last example of waveguiding considers a propagation environment that changes slowly in the propagation or lateral direction.11 indicate how the β plane is structured and how it relates to the α plane.5 A Laterally Inhomogeneous. Closed Waveguide We have considered both open and closed waveguides whose geometry has conformed to a rectangular coordinate system.9.4. (6. We construct an asymptotic approximation that makes use of rays to describe the propagation in the lateral direction and modes to take account of the variation in the transverse one. 6. 6. Figures 6. Moreover we have only considered a very simple variation in the propagation environment with depth.3 Leaky Waves We continue to consider propagation solely to the right. This issue is discussed at greater length in DeSanto (1992). inﬂuence the solution if they come close to the branch cut or if the branch cut is moved. Their positions are approximately indicated by the crosses in both Figs.11. we can recast the representation as u3 = − × iA 4kπ e ¯ T (α)eik cos α x1 ei k sin α x2 ¯ C −ik sin α x20 + eik sin α (2h+x20 ) 1 − R(α)eik sin α (2h) dα.10. The poles and branch cuts of the integrand also give rise to a modal representation. with respect to wavelength. Note that (γ ) > 0 for propagation to the ¯ right.92) F0 = A/k has been used to indicate the dimensions of u 3 . in Figs. To the left of the ω/β = c line. 6. They can. in the thickness of a closed guide whose unperturbed geometry is that given in Fig. namely the layer on a halfspace. but that (γ ) < 0 so that these waves increase exponentially with depth.
The equation of motion becomes 2 2 δ 2 ∂ 2 u 3 /∂ y1 + ∂ 2 u 3 /∂ x2 + u 3 = 0. This asymptotic construction is often called the JWKB technique. The top and bottom surfaces are given by x2 = ±H ± (x1 ) = ±h 0 + h ± (y1 ). with h replaced by H (x1 ) and γn replaced by kγn . .5 A Laterally Inhomogeneous. 3 (6. To clarify the scales we introduce ¯ ¯ scaled coordinates x 1 = kx1 and x 2 = kx2 .96) where Aν (y1 . For simplicity. The boundary conditions at x2 = ±H become µ − δ2 dh ∂u 3 ∂u 3 = 0. so that this equation is exact. The outward unit normal vectors are ˆ n± = − δ dh ± ˆ ˆ e1 ± e2 + O(δ 2 ) .97) and u n (y1 . x2 ). ± dy1 ∂ y1 ∂ x2 (6. (6.93) The boundary conditions require the vanishing of the normal traction. we omit the overbar and reintroduce the variables x2 and u 3 with the understanding that these are now scaled variables.74). (6.98) Note that the boundary conditions are homogeneous.6. γn = nπ/2H . Closed Waveguide 147 The equation of motion remains (6. y1 ) are the cosine or sine functions ﬁrst 3 given in (6. x2 ) is the nth mode for a waveguide whose thickness is determined 3 at y1 . assume from now on that the guide is symmetric so that h ± = h and H ± = H . Also we set u 3 = ku 3 . Having scaled the problem. x2 )δ ν . dy1 (6. 3 (6.95) We now seek an asymptotic solution in the form u 3 ∼ eiθ (y1 )/δ ν≥0 Aν (y1 .1).3). where we assume that dh ± / dy1 = O(1). Then the u n (x2 . Note that the expansion has the form of a modulated group such as we encountered in (6.94) [H ± (x1 + 2π) − H ± (x1 )]/2π ≈ δ[dh ± /dy1 ]. The slow variable y1 = δkx1 is in¯ troduced to describe the slow lateral variation. This is the second way in which the small parameter δ enters the problem. x2 ) = 3 n≥0 n aν (y1 ) u n (y1 . δ then measures the change in the thickness of the guide over a wavelength.
43).100) and (6. Further. ∂ x2 To arrive at (6.93) and (6.1.148 6 Guided Waves and Dispersion Now (6. (6. The equation corresponding to ν = 0 is 2 L A0 = βn A0 .103) where Nn are constants. ±H ) = 0. ∂ y1 (6. where L := ∂ 2 /∂ x2 + 1.100) we have set 2 (dθ/dy1 )2 = βn . The n 2 solution to (6.104) 1/2 . x2 ) = Nn 3 cos (γn x2 ).1 as to what the eigenvalue of interest is. with βn as the 3 3 corresponding eigenvalue. We have introduced the opera2 tor L. u n (y1 . and the coefﬁcient of each power of δ set to zero. to make the structure of the equations clearer.100) and (6.101) is then A0 = a0 (y1 )u n (y1 .101) constitute an eigenvalue problem. The reader is asked to recall the earlier comments in Problem 6. 2 2 dy1 ∂ y1 dy1 ∂ y1 (6.105) .102) This is an eikonal equation. Note that y1 enters βn through h(y1 ).99) with their accompanying boundary conditions.96) is substituted into (6.100) with ∂ A0 3 (y1 .101) (6. 3 3 (6. often determined by a normalization condition. This equation is 2 −L A1 + βn A1 = i 3 3 d 2θ 0 dθ A + 2i 2 3 dy1 dy1 ∂ A0 3 . identical to that solved in Problem 6. (6.102).98). (6. is integrated to give θ(y1 ). sin (6. similar in some respects to (2.4) has become kβn and that now βn = 1 − γn2 (y1 ) where γn = nπ/[2h 0 + 2h(y1 )]. The eikonal equation. Note that βn of (6. The terms for which the superscripts are negative are zero. n To ﬁnd a0 we need to go to the next order in δ. We are then led to the sequence of equations L− dθ dy1 2 Aν + i 3 d 2θ dθ ∂ Aν−1 ∂ 2 Aν−2 3 3 +2 + = 0. x2 ). Equations (6.
Calculating the inner product [L(A1 ).5 A Laterally Inhomogeneous. Burridge and Weinberg (1977) indicate how higherorder terms are gotten.u − u n+ + u n− =− 3 ∂ y1 ∂ y1 3 dy1 3 2 . (6.107) and agree to normalize the eigenfunctions such that [u n . 3 ∂u n ∂u n n dh 2 3 3 . at x2 = ±H .109) where the superscript plus and minus signs mean that these terms are evaluated at x2 = ±H . we note that to have a bounded solution for 1 A3 it must not resonate with the A0 forcing term.105) and (6. u n + 3 3 1 dh 2 u n+ + u n− 3 2 dy1 3 2 = 0. u m ] = δnm . Note that un .106).105) and (6.108) with n = 2 for n = 0.106) gives the following transport equation 3 3 n for a0 . otherwise. (6. and n = 1. Closed Waveguide with boundary conditions. The Nn are 3 3 then given by Nn = { n [h 0 + h 1 (y1 )]}−1/2 (6. respectively. ∂ x2 dy1 dy1 3 149 (6.111) = constant. (6. First we introduce the inner 3 product H+ [a. Examining (6.6. it must be orthogonal 3 n in some sense to u n .106) We shall not seek terms of order higher than ν = 0 here. That is. u n ] and using (6.112) . b] := H− a(x2 ) b∗ (x2 )d x2 (6. namely n dθ da0 1 n d 2 θ + a 2 dy1 dy1 2 0 dy1 n + a0 dθ dy1 ∂u n /∂ y1 . We use this fact to ﬁnd a0 .110) Taking the complex conjugate of (6. dy1 0 0 (6.109) and adding the two gives the much simpler equation dθ dθ ∗ 1 d ln 2 dy1 dy1 dy1 From this it follows immediately that dθ dy1 n a0 2 + d a n a n∗ = 0. ± dh dθ 0 ∂ A1 3 =i A .
It is the microstructure. This particular approach can be extended to inplane elastic waves (Folguera and Harris.113). n It is readily determined that θ0 is a constant. 1999). The periodic structure could be imagined to be a rod possessing a periodic microstructure represented by the concentrated masses. In fact.103). to lowest order. n (6. Among several newer features.115) In the end we should have solved it much as we did. and a0 by (6.112) is a statement of energy conservation. that causes the dispersion. Thus n n 1/2 a0 = c0 eiθ0 βn . but writing the equation in this form exhibits exactly how the microstructure enters the equation.1 Causes of Dispersion We ﬁrst encountered dispersion in the study of a periodic structure and then later in the study of guided waves. For guided waves the dispersion is caused by the geometrical or kinematical constraint that the waves must reﬂect from the boundaries in such a way that they reinforce one another to form a sustained waveﬁeld. n n n n To determine the argument θ0 of a0 = a0 eiθ0 . arising when this is done is a reformulation of the inplane eigenvalue problem and the use of an inner product for the orthogonality condition that is not the norm of the space in which the problem is set.114) n The βn is given by (6. Nn by (6.109).6 Dispersion and Group Velocity 6.150 6 Guided Waves and Dispersion Equation (6.6.113) n where c0 is a real constant. 2 cb t (6. no waves are reﬂected by the slowly changing width. 6. substitute this into (6. we ﬁnd the approximate expression for the nth mode is u 3 ∼ exp i δ y1 n βn (s) ds a0 (y1 )Nn (y1 ) cos (γn x2 ).108). in combination with the additional length scale introduced by it. we could have examined the periodic structure by beginning with the equation 2 ∂1 u 1 = 1 + M ρ ∞ −∞ δ(x1 − n L) 1 2 ∂ u1. The unknown constant would be determined from an initial condition at x1 = 0. Bringing the pieces together. In other words. sin (6. By considering only .
151 (6. It is described at x = 0 by u(0.2.119) (t) cos ω0 t. and both (6. t) = with the modulation described by (t) = H (t + T ) − H (t − T ). we showed that the group velocity is that with which each ωns propagated to a point (x1 . Dispersive propagation is an extraordinarily rich subject.3. by substituting this into the equation of motion. (6. In addition to the work of Lighthill (1965) and Whitham (1974). The wave. (6. (6. Continuing. using the method of stationary phase. recall that in Section 6. that the propagating term Pn (x1 .117) Note that (6.2 the reader showed that both these latter equations led to dispersive propagation.117) is the Klein–Gordon equation. and that in Section 6.117) have the same form as do the Klein–Gordon equation and that for the ﬂexural motion for a rod. written as u 3n = An cos[(nπ/ h)x2 ]Pn (x1 .6. While it would be misleading to assert that the energy in all waveﬁelds or that the information encoded in them always propagates at the group velocity.115) or (6.6 Dispersion and Group Velocity the nth waveguide mode for a closed waveguide. The equation (6. we study dispersion by focusing our attention on the meanings given the group velocity in the context of one dimensional. (6. which we introduce here. (6. amplitude u(x.1.1). dispersive equations having a form such as (6.2) is automatically satisﬁed and that this equation captures the dispersive propagation. begins at x = 0 as a square pulse modulating a harmonic carrier and propagates outward in the positive x direction.116) we ﬁnd.2 The Propagation of Information We consider a signaling problem.6. t) satisﬁes 2 c−2 ∂t2 Pn − ∂1 Pn + (nπ/ h)2 Pn = 0.28).115) and (6. the reader is referred to Sommerfeld (1964a) and Brillouin (1960) for extended discussions of the dispersion of electromagnetic waves in dielectric materials.117). this does occur for many waveﬁelds. t). In this section. In Problem 6. t). 6. (6. t).2 we showed that the energy in the nth mode propagates at the group velocity Cn .27).118) .
There are shortcomings to this argument. This.124) as M(t − x/C) ≈ π/T −π/T ∗ ( )e−i (t−x/C) d . The principal one is that ∗ u(0.125) where ∗ ( ) is the Fourier transform of (t). In this interval we can approximate the dispersion relation as k(ω) ≈ k(ω0 ) + dk (ω0 )(ω − ω0 ).120). ω0 + π/T ). we can write (6. comes from ω ∈ (ω0 − π/T. If ω0 T is large. in order that there be a welldeﬁned modulation. may not . ω)e−i(ω−ω0 )(t−x/C) dω.124) provided ω0 T is large. Using the change of variable = ω − ω0 . H (t) is the Heaviside function. Therefore. the modulation (t) ≈ M(t) propagates at the speed C. This is essentially the kinematic argument put forward by Stokes (Sommerfeld. In particular.120) as u(x. ω − ω0 ω + ω0 (6. the major contribution to the integral. provided ∗ u(0. 1964b) to demonstrate that information propagates at the group velocity. (6. We can thus approximate (6. The dispersion relation has been found by seeking a solution in the form Aei(kx−ωt) and solving for k = k(ω).120). the group velocity.121) Note that (1. ω) must be concentrated in the neighborhood of the carrier frequency ω0 . (6. We construct a solution to the signaling problem as u(x. however. ω)ei[k(ω)x−ωt] dω. (6.122) provided the derivative of k(ω) is well behaved and does not vanish. t) is given by ∗ u(0. k0 = k(ω0 ) and C −1 = dk/dω(ω0 ).152 6 Guided Waves and Dispersion As in previous instances. (6.36) has been used to write (6. ω) = sin[(ω − ω0 )T ] sin[(ω + ω0 )T ] + . ω) is concentrated near ω0 . ∗ u(0. (6. t) = ∞ π ∗ u(0.120) 0 where the transform of u(0. dω (6. t) ≈ where M t− x C ≈ ω0 +π/T ω0 −π/T ∗ π ei(k0 x−ω0 t) M t − x C . ω) is concentrated within intervals near ±ω0 .123) u(0.
73). it was natural to express the solution as a Fourier transform over the angular frequency ω.74) we interpreted the group velocity as that with which a given local angular frequency.t) ] is required. they consider an initial value problem rather than a signaling problem so that their formulation is given in terms of a local wavenumber. assuming that t/x is ﬁxed. We approximate (6.6. t). This gives u(x. ω) is slowly varying. However. following (6.126) The stationary phase condition is dk/dω = t/x.6. However. one that characterizes a group. that the stationary phase condition. and k = k(x.127) The argument is really x/t rather than the more general (x. we now assume that ω0 T is such that ∗ u(0. 6.121).10 suggests why they take this approach. t) through the dispersion relation k = k(ω).69). t) ∼ (2π)1/2 π [xd 2 k/dω2 ]1/2 ∗ u(0. and that raises the question as to whether or not the concept of group velocity has a more general signiﬁcance. 3 (6. t)eiθ (x.3 The Propagation of Angular Frequencies Whitham (1974) and Lighthill (1965) give a more general meaning to group velocity that is motivated by the stationary phase condition. . using the stationary phase approximation (5. because we were working with a signaling problem and. therefore.120) for large x. However. t) ∼ [A(x.6 Dispersion and Group Velocity 153 always be the case. rather than the local wavenumber. Problem 6. We noted previously. propagates. ω)ei[k(ω)x−ωt±π/4] . They propose both a kinematic theory describing the propagation of a local wavenumbers as well as a kinetic theory describing the evolution of a group. Little more than a dispersion relation and the assumption that far from from the source a waveﬁeld evolves into a form described by u(x. t). The solution to this equation3 gives ω = ω(x. the ideas of Whitham and Lighthill prove just as adept at describing the evolution of the local angular frequency and the group it characterizes. (6. A Signaling Problem We again consider the signaling problem whose solution is given by (6. Nevertheless. Recall that following (6. this latter notation will permit some generalizations in the subsection that follows. written as C(ω) = x/t. We chose to describe the local angular frequency.
the solution can be gotten quite simply. t)eiθ (x. we are. t) = −∂t θ. u(x.128) We assume that dC/dω > 0 and take the minus sign in (6.129) If these deﬁnitions are to be consistent with one another. . t) plane such that along S. Assume that there exists a curve S in the (x. t) ∼ [A(x. k(x. or.133) captures the idea suggested by the stationary phase condition that the frequency ω propagates at the group velocity C. Equation (6.132) where C = dω/dk. though signs change here and there. 2 dω C (ω) dω (6. The term d 2 k/dω2 is given by 1 dC d 2k (ω) = − 2 (ω). Moreover. dt/d x = C −1 . t) and a dispersion relation k = k(ω). we can deﬁne a local angular frequency and a local wavenumber as ω(x. In this calculation we emphasize this aspect of the stationary phase condition and have dropped the subscript s that was used previously to designate an ω satisfying this condition.133) is solved by the method of characteristics (Whitham. ω)e−iπ/4 (6. t) = k(ω)x − ω(x. ∂t k + ∂x ω = 0. by using the dispersion relation. Provided the characteristics do not intersect. throughout. ∂t ω + C(ω)∂x ω = 0. then the reasoning remains the same. 1. from the t plane to the ω plane for ﬁxed x. where A(x. (6.130) (2π)1/2 1 π [xd 2 k/dω2 ]1/2 ∗ u(0. Then (6.t) ]. (6. 1974. for propagating waves. and that (6. assuming that dC/dω = 0.132) has the form.126).154 6 Guided Waves and Dispersion is a mapping. t) = ∂x θ. Zauderer 1983). of a conservation equation. If dC/dω < 0. This is a partial differential equation for the angular frequency ω. Thus. t) = and θ (x. A Kinematic Theory Assuming that we know θ (x.133) becomes dω/dt = 0 along this curve. t)t. deﬁned by (6. Note that (6.133) (6.30).131) (6.
it indicates its evolution. (6. The group lines or curves of constant local angular frequency ωn = ωn0 are given by cnπ x1 = 1− ct ωn0 h 2 1/2 . Assume θ (x. ∞). is ωn (x1 . S is given by t = x C −1 (ω) + t0 . 3. Along each S(t0 ). a constant. The local angular frequency determined by the stationary phase condition is given by (6. ω(x. t) = f [t − x C −1 (ω)] along S(t0 ). t) = (cnπ/ h) . Assume that an initial distribution of frequencies ω(0. t). rewritten here. Hereafter we write S(t0 ) to identify that member of the family of curves with intercept t0 .126).138) . It follows then that ω(x. Reverting to the notation of Section 6. Lastly.127). t0 ) = f (t0 ) is given. we again consider the stationary phase approximation to the transiently excited waves in a closed waveguide.136) For both expressions ωn ∈ (cnπ/ h. It is not an explicit solution for ω. t) = − ∂t θ/∂x θ. given by the argument of the cosine in (6. where t0 is the intercept with the t axis. We can also deﬁne a local phase velocity. (6. Implicit differentiation gives c(x. We call lines of constant angular frequency group lines and those of constant phase phase lines. Thus c = ω/k where ω and k are the local frequency and wavenumber deﬁned by (6.6.135) The phase. because C is constant if ω is. t) is therefore constant along S and. is θn (x1 . (6. ω(0.6 Dispersion and Group Velocity 155 2.134) where the phase velocity c = d x/dt. t) = θ0 . t).72). (6. we recover the stationary phase condition.137) The curves of constant phase θn = θn0 are given by 2 c2 t 2 − x 1 = θn0 h nπ 2 . That expression. 1 for x large if we assume that f (t0 ) is localized near the origin so that t0 /x and t/x ﬁxed. Variation of t0 gives a solution throughout (x.129). [1 − (x1 /ct)2 ]1/2 (6. t) = − x1 nπ h ct x1 2 1/2 −1 .3. t0 ) = ω(x.2. (6. Nevertheless.
Lighthill (1965) gives a general argument showing that. the group velocity is the velocity with which the average energy propagates. t)dt.139) where t2 > t1 . Aki and Richards (1980) give a restricted version of this argument that is applicable to waveguide modes. 2. For a time harmonic. It indicates that for a signal.156 6 Guided Waves and Dispersion These two families of curves are quite different so that. t)ω2 (x. Q(x) = t1 t2 A(x. dω (6. While the kinematic and kinetic descriptions of group velocity put forward by Whitham (1974) and Lighthill (1965) appear quite complete. This is a remarkable result. once again. where i = 1. When dispersion arises from material properties. Recalling our earlier comment that (6. This then is a generalization of the idea that timeaveraged energy propagates at the group velocity. we change the variable of integration in (6. they assume that attenuation is not present. (6. it is very often accompanied by frequencydependent attenuation. we note that a phase and a group evolve differently in a dispersive environment. for any wave structure or environment that can be characterized by a Lagrangian and for which equipartition of energy occurs.138) to ω.127) is a mapping from t to ω. Auld (1990) uses a reciprocity relation to prove this result in somewhat greater generality.140) ∗ u(0.141) For dC/dω > 0. t)A∗ (x. This is reminiscent of an energy term. (6. Here we continue to follow Whitham (1974) and Lighthill (1965) and advance an argument using the stationary phase condition. We deﬁne a quantity Q as follows. A Kinetic Theory Previously. ω)2 ω2 dω (6. we explicitly showed that for a closed waveguide the averaged energy of a particular mode propagates with the group velocity for that mode.25). Noting that dt = − we see that (6. this term is proportional to the timeaveraged energy density. the energy contained within a given frequency band is constant when the frequency band remains between the two group lines identiﬁed by ti = x C −1 (ωi ).139) becomes Q(x) = 2 π ω1 ω2 x C 2 (ω) dC (ω)dω. The same . such as a fragment of speech. ω1 > ω2 . plane wave and wave system for which equipartition of energy occurs.
Repeat Problem 1 for the equation for ﬂexural motion in a rod. (6. Whitham (1974) and Lighthill (1965) use an initial value problem to advance their ideas about group velocity.143) where ω = ω(k) is the dispersion relation. (6. (6. t)eiθ (x.27). San Francisco: Freeman. and not ω. 286–292. t) and θ (x. . and Richards.References 157 physical mechanisms are responsible for both dispersion and attenuation so that they cannot be easily disentangled. Problem 6. with x/t positive and ﬁxed. Thus show that the appropriate solution is ϕ= 1 π ∞ cos(kx) cos[ω(k)t]dk. can be put into the form ϕ∼ A(x.10 Dispersion 2 As we indicated previously. ∂t ϕ(x. 1980) is an example that indicates the tangling of these two effects. is the transform variable.144) Give explicit expressions for A(x. P. Problem 2. 0 (6.G. (6.28). Solve the initial value problem for this equation. Vol.143) is to be interpreted as a generalized function or distribution. Note that k. The propagation of linearly viscoelastic waves (Hudson. show that this solution for t → ∞. K. 1. Note that (6. attributing a meaning to group velocity can become much more difﬁcult. 0) = δ(x). References Aki. pp. The following two problems indicate why they develop their ideas in this way. given the initial conditions ϕ(x. t).142) For t < 0− . Using the method stationary phase. 1980. unless the attenuation is quite weak. Consider once again the Klein–Gordon equation. Problem 1.t) . Theory and Methods. Thus they write that the wavenumber propagates at the group velocity and that the energy conﬁned between two group lines deﬁned by two ﬁxed ki is constant. An ab initio calculation may be the only guide (see also footnote 2). ϕ and ∂t ϕ are identically zero. In such cases. Quantitative Seismology. 0) = 0.
273–289. M. I.B. Optics. L. 1956.G. Waves in Layered Media. pp. 184–191. and Wong. Hudson. pp. Inst. H. New York: Springer. New York: WileyInterscience. Scalar Wave Theory. ed.J. LaPorte and P. Burridge. Moldauer. Sommerfeld. pp. 1999. R. pp. New York: Academic. Principles and Techniques of Applied Mathematics. 1980. deSanto. Brillouin. Horizontal rays and vertical modes. J. 1983.. IV. 2. J. Coupled Rayleigh surface waves in a slowly varying elastic waveguide.M.. Miklowitz. Acoustic Fields and Waves in Solids.H. Proc. Lond. A uniform asymptotic expansion for the shearwave front in a layer. Whitham. 2nd ed. pp. FL: Krieger. Maths.A. J. The Theory of Elastic Waves and Waveguides. Mechanics of Deformable Bodies. and Harris.A. Soc. Partial Differential Equations of Applied Mathematics.S. 1974. pp. pp. Brekhovskikh.. 1994. R. Ocean Acoustics. A. pp. J. Lighthill. Lectures on Theoretical Physics. In Wave Propagation and Underwater Acoustics. Zauderer.A. J. Keurti. pp. 1987. Keller and J. New York: Springer. New York: American Institute of Physics. 1977.. New York: Academic. 35–77. II. 1992. New York: Academic. . Malabar. Lectures on Theoretical Physics. Wave Motion 19: 293–308. New York: Wiley.S. Friedman. Theory and Experiment in Underwater Sound. Vol. and Weinberg. Vol. Tolstoy. 203–206. B. Wave Propagation and Group Velocity. New York: NorthHolland. Linear and Nonlinear Waves. 1978. G. Translated by G. New York: Academic. 178–200 and 409–466. 76–80 and elsewhere. pp. Appl. L. 86–152. 1964a. A 455: 917–931. A. Group velocity.B. The Excitation and Propagation of Elastic Waves. 1965. 1960. J. 1980. 2nd ed. R. 1964b. Vol. H. New York: WileyInterscience. B. Sommerfeld. Folguera. 363–402. 188–222. Papadakis. 162–179. E. 1990. A. 299–325. C.A. Translated by O. 2nd ed. pp. Dai.158 6 Guided Waves and Dispersion Auld. 1: 1–28. Cambridge: University Press. and Clay.
24 asymptotic approximation of integrals. 23 boundary layer. 90–94 branch cut contribution. 95 Fresnel integral. 96–101 asymptotic approximation of the scattered compressional wave. 108. 81 inversion. 23 compressional wave. 61 twodimensional. see velocity. 33 average for a closed waveguide mode. 109 branch cuts. 108 diffraction coefﬁcient. 31. 30 asymptotic ray expansion. 122. 144 geometrical. 9 angularspectrum. 28–34. 44 refraction. deﬁned. 125. 87 stationary phase. 55 Abelian theorem. 133 cylindrical wave. 99 contour. 91. 100 center of compression threedimensional. 82. deﬁning. 150 microstructure. 13 and stationary phase. 98 pole contribution. 62 complex unit vector. 79–82 caustic. 99 Stokes’ phenomenon. 112 compressional. 86–96 end point contribution. 108 Watson’s lemma. 95 asymptotic power series. 77–82 contour. 82–86. 90. 5 critical angle incidence. 126 time average for a plane wave. 34 diffraction at an edge. 126 Lagrangian density. 92 waveguide modes. 150 159 . 150 closed waveguide. 128 anomalous. 107 allied function. 119 integration by parts. 138 causes of. 101–119 diffracted wave. 85. 46 angular frequency deﬁned. 119 uniform. 98 asymptotic approximation of the scattered shear wave. see planewave. 136 steepest descents. 94–95 contour. 29 shear. 52–55 buried harmonic line of compression. 125 from the poles of an integral representation. deﬁned. 110 dispersion. 136 anisotropic medium. 82. 82. 40 κr . see Fresnel integral geometrical theory. see matched asymptotic expansion. 110 diffraction integral. 99 Cagniard–deHoop technique. 87–90 wavefront approximation. group. deﬁned. 43 cutoff frequency.Index κ. 43 reﬂection.
128 for the Klein–Gordon equation. 61. see matched asymptotic expansion. 60. 126 averaged for a plane wave. 155 Fresnel integral. 6 energy in a band of frequencies. 148 rays and modes. 60 threedimensional.) open waveguide. 59 partial waves. 101 ﬂexural motion equation. 62 method of images. 130. 48. see matched asymptotic expansion. 145 periodic structure. 112–116 boundary layer. 115 Schrodinger equation. 10 threedimensional. 157 Lagrangian density. 7 twosided. 114 nearﬁeld of an edge. 109. 157 Fourier transform space. see diffraction at an edge. 123. outer coordinates. 68 Green’s tensor. 113 outer expansion. 114 matching. 65. outer expansion. 127 energy ﬂux during critical refraction. 2 twodimensional. deﬁned. 30. 122 eigenfunction continuous expansion. 2 dilitational motion. 67. 5 initial value problem. 145 eigenvalues and eigenfunctions discrete. 62 correct and incorrect. see waveguide. 148 energy relations.160 dispersion (cont. 126 Lamb’s problem. 131 periodic structure. 79 line of compression. 17. 58 guided waves. 71 integral equations. 121 Helmholtz theorem. 65 of an edge. 126 for a transient plane wave. 25 geometrical theory of diffraction. 114 inner. 74 stress. 4 inplane motion. 134 eigenvalues. 45 energy ﬂux during reﬂection. 128 internal energy density. 9 local. 18 Index Green’s function. 139 discrete or mode expansion. 156 equipartition of energy. 111. 146 limiting absorption. 58 time. 111 . 114 stretching transformation. open. 131 eikonal equation. 1–6 boundary conditions. 104. see planewave representations. 135 and the Wiener–Hopf method. 62–63. 122 discrete and continuous. 5 kinetic energy density. 109. 66 elastic ﬂuid. 150. 82 Laplace transform time. 71. 83 longitudinal wave. 4 antiplane motion. 147 Klein–Gordon equation. 71. see limiting absorption. 75 farﬁeld compact source. 127. 6. 110. deﬁned. see center of compression. see waveguide. 58. 23 conservation law. 6 equations of motion. deﬁned. see buried harmonic line of compression. 22 for the ﬂexural motion equation. 116–119 gamma function. 6 averaged for a closed waveguide mode. 4 extinction theorem. 41 energy ﬂux. 3 rotational motion. see compressional. 15–18 effective wavenumber. 65–67 source problem. 104–108 outgoing wave. open. 2 onedimensional. 127. 78 leaky wave. 157 inner expansion. 131 matched asymptotic expansion. 72. 59. 22 Love wave. 114 Fermat’s principle. 58. see waveguide. 76 integral representation scattering problem. 18 eigenvalue problem. principle of. 113 inner. 71 outgoing wave. 140. 123 discrete and continuous. 87 Gaussian beam. 7 frequency deﬁned. 64–65 JWKB asymptotic expansion eigenvalue problem.
see asymptotic ray expansion. 155 waveguide mode. 28 waves scattered from a surface. 18 and stationary phase. 51 guided by an impedance boundary. 1 transforms. 110 ray tube. 5 propagation matrix. 137 group. 24 slowness deﬁned. see refraction. 39–40 phase. Laplace transform. 39. 28. 84. 49 timeharmonic. 96. 34 standing wave. 44 vector. 39 Sommerfeld transformation. 154 kinetic theory. 43. 30. 47 inplane shear coefﬁcient. deﬁned. 5 vector potential. 13 from a strip. 38 refraction antiplane shear coefﬁcient. 156 periodic structure. 94 steepest descents. see dispersion. 133 from a plate in a ﬂuid. 5 velocity energy transport. 46 scalar potential. not a wave. 40 compressional wave. 51 reciprocity. 57 reﬂection antiplane shear coefﬁcient. 42 compressional coefﬁcient. 5 signaling problem. 101 Rayleigh function. 125 vibration. see potentials. see Abelian theorem. 124 transverse wave. 68 vector potential. 30 fan of diffracted rays. 45 twosided wave. 24 in an open waveguide. deﬁned. displacement. 82 traction. 14 shear wave. 16 radiation conditions.Index phase matching. 6 deﬁned. 42 interfacial wave. 27. 1 surface wave. 6–10 transmission coefﬁcient. 22 inhomogeneous or evanescent. 24. 11 . 125 fastest. displacement. 65 ray. 24–28 cylindrical wave. 153 waveguide mode. 142 spherical wave. 38 from a layer. 2 stress tensor. 51 Rayleigh equation. 149 transverse resonance principle. 65. 44 transmission matrix. 69. 125 phase local. 44 antiplane shear wave. 22. 20 surface. 50 twosided wave. deﬁned. displacement. see limiting absorption. 20 timeharmonic. 59. 140 Gaussian beam. 17 from a lumped mass. 51 Tauberian theorem. 104. 67 161 from an elastic inclusion. 155 phase lines. deﬁned. see asymptotic approximation of integrals. 48–52. see potentials. 22 uniqueness in an unbounded region. 151 propagation of wavenumbers. 44 antiplane shear wave. 96. 20. 27. 30 deﬁned. 96 timedependent. 151–157 deﬁned. 155 kinematic theory. 146 spherical wave. 20 potentials. see shear. 146 Rayleigh wave. its meaning. 84 Poisson summation formula. 49 transient. 40 inplane shear wave. 18 propagation of angular frequencies. 63. 137 group lines. 33 rays and modes. see Fourier transform. 22 planewave representations. 29 plane wave homogeneous. 5 scalar potential. 11 polarization change in a ray expansion. 68–72 edge conditions. 22. see asymptotic approximation of integrals. 33 deﬁned. 5 scattering Bragg. Fourier and Laplace. 107 no edges. 15 transport equation. 72 scattering matrix. deﬁned. 153–157 propagation of information. 9 stationary phase. 142. see Rayleigh wave. 90 strain tensor.
129 . 142.. 22 Wiener–Hopf method. 139–146 leaky modes. 146 ray representation. 134–139 harmonic excitation. 18 lateral. 135 excitation. 104–108 Watson’s lemma. 22 deﬁned. 9 wavenumber complex.162 Index excitation. 32 deﬁned. spherical. 146 modal representation. deﬁned. 20 wavefront curvature. 87 wave kinematics. 30 plane. 122 wavevector deﬁned. 145 wavelength. 155 transverse. 20 waveguide closed. periodic structure. deﬁned. 144 planewave representation. see asymptotic approximation of integrals. 122 open. 134 mode expansion. 141. 146 mode. 8 effective. 135 laterally inhomogeneous. etc. 121 evanescent mode. 122 local. 134 transient excitation. 22 real and complex.
This action might not be possible to undo. Are you sure you want to continue?
Use one of your book credits to continue reading from where you left off, or restart the preview.