Integral Transform Techniques For Green S Function

Lecture Notes in Applied and Computational Mechanics 71
Kazumi Watanabe
Integral Transform
Techniques for
Green’s Function
Lecture Notes in Applied and Computational
Mechanics
Volume 71
Series Editors
F. Pfeiffer, Garching, Germany
P. Wriggers, Hannover, Germany
For further volumes:

http://www.springer.com/series/4623
Kazumi Watanabe
Integral Transform
Techniques for Green’s
Function
123
Kazumi Watanabe
Department of Mechanical Engineering
Yamagata University
Yonezawa
Japan
ISSN 1613-7736 ISSN 1860-0816 (electronic)

ISBN 978-3-319-00878-3 ISBN 978-3-319-00879-0 (eBook)
DOI 10.1007/978-3-319-00879-0
Springer Cham Heidelberg New York Dordrecht London
Library of Congress Control Number: 2013940095
Ó Springer International Publishing Switzerland 2014

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed. Exempted from this legal reservation are brief
excerpts in connection with reviews or scholarly analysis or material supplied specifically for the
purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the
work. Duplication of this publication or parts thereof is permitted only under the provisions of
the Copyright Law of the Publisher’s location, in its current version, and permission for use must
always be obtained from Springer. Permissions for use may be obtained through RightsLink at the
Copyright Clearance Center. Violations are liable to prosecution under the respective Copyright Law.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
While the advice and information in this book are believed to be true and accurate at the date of
publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for
any errors or omissions that may be made. The publisher makes no warranty, express or implied, with
respect to the material contained herein.
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

Dedicated to my teachers,
Dr. Akira Atsumi
(Late Professor, Tohoku University)
and
Dr. Kyujiro Kino
(Late Professor, Osaka Institute of
Technology)
Preface
When I was a senior student, I found a book on the desk of my advisor Professor
and asked him how to get it. His answer was in the negative, saying its content was
too hard, even for a senior student. Some weeks later, I found it again in a
bookstore, the biggest in Osaka. This was my first encounter with ‘‘Fourier
Transforms’’ written by the late Professor I. N. Sneddon. Since then, I have learned
the power of integral transform, i.e. the principle of superposition.
All phenomena, regardless of their fields of event, can be described by differ-
ential equations. The solution of the differential equation contains the crucial
information to understand the essential feature of the phenomena. Unfortunately,
we cannot solve every differential equation, and almost all phenomena are
governed by nonlinear differential equations, of which most are not tractable. The
differential equations which can be solved analytically are limited to a very small
number. But their solutions give us the essence of the event. The typical partial
differential equations which can be solved exactly are the Laplace, the diffusion
and the wave equations. These three partial differential equations, which are
linearized for simplicity, govern many basic phenomena in physical, chemical and
social events. In addition to single differential equations, some coupled linear
partial differential equations, which govern somewhat complicated phenomena,
are also solvable and their solutions give much information about, for example, the
deformation of solid media, propagation of seismic and acoustic waves, and fluid
flows.
In any case where phenomena are described by linear differential equations, the
solutions can be expressed by superposition of basic/fundamental solutions. The
integral transform technique is a typical superposition technique. The integral
transform technique does not require any previous knowledge for solving differ-
ential equations. It simply transforms partial or ordinary differential equations to
reduced ordinary differential equations or to simple algebraic equations. However,
a substantial difficulty is present regarding the inversion process. Many inversion
integrals are tabulated in various formula books, but typically, it is not enough.
If a suitable integration formula cannot be found, the complex integral must be
considered and Cauchy’s integral theorem is applied to the inversion integral.
Thus, integral transform techniques are intrinsically connected with the theory of
complex integrals.
vii
viii Preface
This book intends to show how to apply integral transforms to partial

differential equations and how to invert the transformed solution into the actual
space–time domain. Not only the use of integration formula tabulated in books,
but also the application of Cauchy’s integral theorem for the inversion integrals
are described concisely and in detail. A particular solution for a differential
equation with a nonhomogeneous term of a point source is called the ‘‘Green’s
function’’. The Green’s functions for coupled differential equations are called
‘‘Green’s dyadic’’. The Green’s function and Green’s dyadic are the basic and
fundamental solutions of the differential equation and give the principal features
of the event. Furthermore, these Green’s functions and dyadics have many
applications for numerical computation techniques such as the Boundary
Element Method. However, the Green’s function and Green’s dyadic have been
scattered in many branches of applied mechanics and thus, their solution
methods are not unified. The book intends to present and illustrate a unified
solution method, namely the method of integral transform for the Green’s
function and Green’s dyadic. Thus, the fundamental Green’s function for the
Laplace and wave equations and the Green’s dyadic for the elasticity equations
are gathered in this single book so that the reader can have access to a proper
Green’s function and understand the mathematical process of its derivation.
Chapter 1 describes roughly the definition of the integral transforms and the
distributions to be used throughout the book. Chapter 2 shows how to apply an
integral transform for solving a single partial differential equation such as the
Laplace and wave equations. The basic technique of the integral transform method
is demonstrated. Especially, in the case of the time-harmonic response for the
wave equation, the integration path for the inversion integral is discussed in detail.
At the end of the chapter, the obtained Green’s functions are listed in a table so
that the reader can easily find the difference in the functional form among the
Green’s functions. An evaluation technique for a singular inversion integral which
arises in a 2D static problem of Laplace equation is developed.
The Green’s dyadic for 2D and 3D elastodynamic problems are discussed in
Chap. 3. Three basic responses, impulsive, time-harmonic and static responses, are
obtained by the integral transform method. The time-harmonic response is derived
by the convolution integral of the impulsive response without solving the differ-
ential equations for the time-harmonic source.
Chapter 4 presents the governing equations for acoustic waves in a viscous
fluid. Introducing a small parameter, the nonlinear field equations are linearized
and reduced to a single partial differential equation for velocity potential or
pressure deviation. The Green’s function which gives the acoustic field in a
uniform flow is derived by the method of integral transform. A conversion tech-
nique for the inversion integral is demonstrated. That is, to transform an inversion
integral along the complex line to that along the real axis in the complex plane.
It enabled us to apply the tabulated integration formula.
Chapter 5 presents Green’s functions for beams and plates. The dynamic
response produced by a point load on the surface of a beam and a plate is
discussed. The impulsive and time-harmonic responses are derived by the integral
Preface ix
transform method. In addition to the tabulated integration formulas, the inversion

integrals are evaluated by application of complex integral theory.
Chapter 6 presents a powerful inversion technique for transient problems of
elastodynamics, namely the Cagniard-de Hoop method. Transient response of an
elastic half-space to a point impulsive load is discussed by the integral transform
method. Applying Cauchy’s complex integral theorem, the Fourier inversion
integral is converted to an integral of the Laplace transform and then its Laplace
inversion is carried out by inspection without using any integration formula. The
Green’s function for an SH-wave and Green’s dyadics for P, SV and SH-waves are
obtained.
The last Chap. 7 presents three special Green’s functions/dyadics. The 2D static
Green’s dyadic for an orthotropic elastic solid and that for an inhomogeneous solid
are derived. In the last section, moving boundary problems is discussed. Two
different Laplace transforms are applied for a single problem, and a conversion
formula between two Laplace transforms is developed with use of Cauchy’s
theorem. This conversion enables us to apply the integral transform technique to a
moving boundary problem.
The integral transform technique has been used for many years. The inversion
process inevitably requires a working knowledge of the theory of complex func-
tions. The author finds the challenge of a complex integral amusing, especially the
challenge of choosing the right contour for the inversion integral. He hopes that
young researchers will join the fun and carry on with the inversion techniques.
In this respect it must be mentioned that he feels a lack of mathematical skill in the
recent research activities, since some researchers tend to use numerical techniques
without considering the possibility of an analytical solution. The increased
mathematical techniques expand wider the horizon of the differential equations,
and one can extract more firm knowledge from nature which is described by the
differential equations. The author hopes that the book will give one more technique
to the younger researchers.
Finally, the author wishes to express his sincere thanks to Dr. Mikael
A. Langthjem, Associate Professor of Yamagata University, for his advice and
nice comments.
Yonezawa, Japan, January 2013 Kazumi Watanabe

Contents
1 Definition of Integral Transforms and Distributions ........... 1

1.1 Integral Transforms . . . . . . . . . . . . . . . . . . . . . ........... 1
1.2 Distributions and Their Integration Formulas . . . ........... 4
1.3 Comments on Inversion Techniques
and Integration Formulas . . . . . . . . . . . . . . . . . ........... 9
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........... 10
2 Green’s Functions for Laplace and Wave Equations . . . . . . . . . . . 11

2.1 1D Impulsive Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 1D Time-Harmonic Source . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 2D Static Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.6 3D Static Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3 Green’s Dyadic for an Isotropic Elastic Solid . . . . . . . . . . . . . . . . 43

3.3 2D Static Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.6 3D Static Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4 Acoustic Wave in a Uniform Flow . . . . . . . . . . . . . . . . . . . . . . . . 77

4.1 Compressive Viscous Fluid. . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.2 Linearization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.3 Viscous Acoustic Fluid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
xi
xii Contents
4.4 Wave Radiation in a Uniform Flow . . . . . . . . . . . . . . . . . . . . . 85

4.5 Time-Harmonic Wave in a Uniform Flow . . . . . . . . . . . . . . . . 91
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5 Green’s Functions for Beam and Plate . . . . . . . . . . . . . . . . . . . . . 93

5.1 An Impulsive Load on a Beam . . . . . . . . . . . . . . . . . . . . . . . . 93
5.2 A Moving Time-Harmonic Load on a Beam. . . . . . . . . . . . . . . 95
5.3 An Impulsive Load on a Plate. . . . . . . . . . . . . . . . . . . . . . . . . 99
5.4 A Time-Harmonic Load on a Plate . . . . . . . . . . . . . . . . . . . . . 101
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
6 Cagniard-de Hoop Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

6.1 2D Anti-Plane Deformation . . . . . . . . . . . . . . . . . . . . . . . . . . 108
6.2 2D In-Plane Deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
6.3 3D Dynamic Lamb’s Problem . . . . . . . . . . . . . . . . . . . . . . . . . 130
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
7 Miscellaneous Green’s Functions . . . . . . . . . . . . . . . . . . . . . .... 157

7.1 2D Static Green’s Dyadic for an Orthotropic Elastic Solid . .... 157
7.2 2D static Green’s Dyadic for an Inhomogeneous
Elastic Solid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 164
7.3 Reflection of a Transient SH-Wave at a Moving Boundary .... 173
Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 186
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Chapter 1
Definition of Integral Transforms
and Distributions
This first chapter describes a brief definition of integral transforms, such as the
Laplace and Fourier transforms, and a rough definition of delta and step functions
which are frequently used as the source function. The multiple integral transforms
and their notations are also explained. The last short comment lists some important
formula books which are crucial for the inverse transform, i.e. the evaluation of the
inversion integral.
1.1 Integral Transforms
For a well-defined function f ðxÞ; x 2 ða; bÞ, when the integral with the kernel
function Kðn; xÞ,
Zb
F ð nÞ ¼ K ðn; xÞf ð xÞdx ð1:1:1Þ
a
has its inverse integral with another kernel function K ðn; xÞ,
Z
f ð xÞ ¼ K ðn; xÞF ðnÞdn ð1:1:2Þ
L
we call this integration pair an ‘‘integral transform.’’ The function f ðxÞ is an

original function and the function FðnÞ is the ‘‘image or transformed function’’ in
the transformed domain. If the reciprocal f ðxÞ , FðnÞ holds, we call FðnÞ the
‘‘integral transform’’ of f ðxÞ and the two-variable-functions Kðn; xÞ and K ðn; xÞ
the kernels of the integral transform.
We have already learned many integral transforms which are classified and
named depending on the kernel function and the integration range. A well-known
integral transform is the Laplace transform. The (one-sided) Laplace transform is,
in the present book, defined for a time-variable function f ðtÞ; t 2 ð0; 1 as
K. Watanabe, Integral Transform Techniques for Green’s Function, 1

Lecture Notes in Applied and Computational Mechanics 71,
DOI: 10.1007/978-3-319-00879-0_1, Ó Springer International Publishing Switzerland 2014
2 1 Definition of Integral Transforms and Distributions
Z1

f ðsÞ ¼ f ðtÞ expðstÞdt ð1:1:3Þ
0
where ‘‘s’’ is the transform parameter and the transform kernel is expðstÞ. The
inverse transform is also defined by the integral along the complex line,
Z
cþi1
1
f ðt Þ ¼ f ðsÞ expðstÞds ð1:1:4Þ
2pi
ci1
where the integration path from c i1 to c þ i1 is called the ‘‘Bromwich line.’’

The real constant ‘‘c’’ must be larger than the real part of any singular point of the
transformed function f ðsÞ. Thus this line is placed at far right from all singular
points in the complex s-plane.
In these definitions, the variable s is called the transform parameter and the two
kernels for the transform and the inverse transform are exponential functions:
1
K ðs; tÞ ¼ expðstÞ; K ðs; tÞ ¼ expðstÞ ð1:1:5Þ
2pi
Further, the integration ranges are also different from each other. The transform
integral is carried out along the semi-infinite real line ð0; 1 for the time and the
inverse transform is carried out along an infinite line ½c i1; c þ i1 in the com-
plex s-plane. Since any notation for the transform parameter is available, one should
be aware of the notation of the parameter since some authors use ‘‘p’’ instead of ‘‘s.’’
So far, many integral transform pairs have been found and defined. We choose
and use one suitable integral transform depending on the geometry (integration
range) and the simplicity of its application. The followings are typical integral
transforms which are much used in applications.
(a) Finite Fourier transform (Fourier series); f ðxÞ; x 2 ðp; pÞ
When an original function is defined within a finite region, the Fourier finite
transform, i.e. Fourier series is used and its transform pair can be defined as
follows.
(a1) Complex Fourier series:
Zp
1 X 1
fn ¼ f ð xÞ expðinxÞdx; f ð xÞ ¼ fn expðinxÞ ð1:1:6Þ
2p n¼1
p
where the index ‘‘n’’ is an integer.

(a2) Fourier cosine series:
Zp Zp
1 1X 1
a0 ¼ f ð xÞdx; an ¼ f ð xÞ cosðnxÞdx; f ð xÞ ¼ an cosðnxÞ ð1:1:7Þ
2 p n¼0
p p
1.1 Integral Transforms 3
(a3) Fourier sine series:

Zp
1X 1
bn ¼ f ðxÞ sinðnxÞdx; f ðxÞ ¼ bn sinðnxÞ ð1:1:8Þ
p n¼1
p
(b) Complex Fourier transform: f ð xÞ; x 2 ½1; 1

When the function is defined in an infinite region, the complex Fourier trans-
form pair is defined by
Z1 Z1
f ðnÞ ¼ 1 f ðnÞ expðinxÞdn
f ð xÞ expðinxÞdx; f ð xÞ ¼ ð1:1:9Þ
2p
1 1
(c) Fourier cosine/sine transform: f ðxÞ; x 2 ð0; 1

When the original function is even/odd in the infinite region, or the function is
defined in a semi-infinite region, we employ the Fourier cosine/sine transforms.
They are defined as follows.
(c1) Fourier cosine transform:
Z1 Z1
ðCÞ 2 f ðCÞ ðnÞ cosðnxÞdn
f ð nÞ ¼ f ð xÞ cosðnxÞdx; f ð xÞ ¼ ð1:1:10Þ
p
0 0
(c2) Fourier sine transform:

Z1 Z1
f ðSÞ ðnÞ ¼ 2 f ðSÞ ðnÞ sinðnxÞdn
f ð xÞ sinðnxÞdx; f ð xÞ ¼ ð1:1:11Þ
p
0 0
(d) Hankel transform: f ð xÞ; x 2 ð0; 1

Another semi-infinite integral transform is Hankel transform (Sneddon 1951,
pp. 48) defined by
Z1 Z1
~fn ðnÞ ¼ xf ð xÞJn ðnxÞdx; f ðxÞ ¼ n ~fn ðnÞJn ðnxÞdn ð1:1:12Þ
0 0
where Jn ðzÞ is the n-th order Bessel function of the first kind.
In order to guarantee the application of the integral transform, each integral

must converge. For example, the Fourier transform requires a convergence con-
dition at infinity,
f ð xÞ ) Oðjxjm Þ; m [ 1 ð1:1:13Þ
x!1
for the original function. However, if we employ the Hankel transform, it is

allowable for the original function to be finite since the Bessel function decays
with the order of the inverse square root at the infinity.
In some applications, where we encounter multi-variable functions, or where a
multiple integral transform, a different mark for the transform is defined such as
f ; ~f ; f , and the multiple transform is denoted by piling up the transform marks,
like ~ f . In the present book, we employ the Laplace transform with respect to the
time variable t as defined in Eqs. (1.1.3) and (1.1.4). For the space variables
ðx; y; zÞ, we apply three Fourier transforms with the transform parameters ðn; g; fÞ
respectively. They are defined as
Z1 Z1
f ð xÞ expðþinxÞdx; f ð xÞ ¼ ð1:1:14aÞ
2p
1 1
Z1 Z1
~f ðgÞ ¼ 1 ~f ðgÞ expðigyÞdg
f ð yÞ expðþigyÞdy; f ð yÞ ¼ ð1:1:14bÞ
2p
1 1
Z1 Z1
^f ðfÞ ¼ 1 ^f ðfÞ expðifzÞdf
f ðzÞ expðþifzÞdz; f ðzÞ ¼ ð1:1:14cÞ
2p
1 1
Please remember that the pair of the space variable and transform parameter is
fixed throughout the present book, such as the pair, ðx; nÞ, ðy; gÞ and ðz; fÞ.
1.2 Distributions and Their Integration Formulas
For modeling engineering phenomena, many mathematical functions are used.

Elementary and some special functions are used for continuous phenomena. But
for discontinuous phenomena, distributions such as delta and step functions are
frequently used. This subsection explains briefly the definition of three distribu-
tions: Dirac’s delta function, Heaviside’s unit step function and Heisenberg’s delta
function.
(1) Heaviside’s unit step function: Hðx aÞ
1.2 Distributions and Their Integration Formulas 5
Fig. 1.1 Heaviside’s unit H ( x − a)

step function +1
0
x=a
Heaviside’s unit step function is defined by the graphical form in Fig. 1.1. This
function takes the value 0 in the negative region and +1 in the positive region,

þ1; x [ a
H ðx aÞ ¼ ð1:2:1Þ
0 ; x\a
but it is not defined at the discontinuous point x ¼ a. So, it takes two limiting
values from the positive and negative sides of the discontinuous point,
H ð0þ Þ ¼ þ1; H ð0 Þ ¼ 0 ð1:2:2Þ

Then, we have to understand that the step function is not defined at x ¼ a.
(2) Dirac’s delta function: dðx aÞ
Dirac’s delta function is defined as the limit e ! 0 of a rectangular pulse with
width e and height 1=e as shown in Fig. 1.2. The center of the rectangular dis-
tribution is fixed at x ¼ a in the limiting process and we understand the point as an
application point of the delta function. It is denoted by dðx aÞ. In the limit, the
width of the delta function vanishes and the functional value becomes infinite. But,
its internal invisible area is one due to the definition ð1=eÞ e ¼ 1. This nature
makes the evaluation of the integral very simple, and we have the formula where
the integrand includes the delta function as
Zb (
f ðcÞ; a\c\b
f ð xÞdðx cÞdx ¼ ð1:2:3Þ
0; c \ a or b \ c
a
Using this simple integration formula, we can obtain an integral representation

for the delta function. Applying Fourier transform defined by Eq. (1.1.9) to the
delta function, f ðxÞ ¼ dðx cÞ, we use the integration formula (1.2.3). The
transform integral is evaluated as
Fig. 1.2 Schematic (a) (b) ∞

definition of Dirac’s delta δ ( x − a)
function. a Rectangular pulse. ↑
b Dirac’s delta function
ε 1/ ε ε →0
0 0 0 0
x=a x=a
Z1
f ðnÞ ¼ dðx cÞ expðinxÞdx ¼ expðincÞ ð1:2:4Þ
1
The inverse Fourier transform is also applied to the above f ðnÞ and its integration
range is reduced to the semi-infinite,
Z1 Z1
1 1
f ð x Þ ¼ dð x c Þ ¼ expfinðx cÞgdn ¼ cosfnðx cÞgdn ð1:2:5Þ
2p p
1 0
Then, we have the integral representation for the Dirac’s delta function, i.e.
Z1
1
dðx cÞ ¼ cosfnðx cÞgdn ð1:2:6Þ
p
0
Finally, we would like to add one useful formula between delta and step
functions. That is
d
dð x aÞ ¼ H ð x aÞ ð1:2:7Þ
dx
This relation will be understood from the graphical discussion of delta and step
functions.
(3) Heisenberg’s delta function: d ðxÞ
In many applications, we have to express a semi-infinite distribution of a
physical quantity, such as the uniform load over the surface. In order to treat a
semi-infinite distribution, another delta function is defined. That is Heisenberg’s
delta function which is the Fourier transform of a semi-infinite distribution.
Let us consider Fourier transform of Heaviside’s unit step function,

1; 0 \ x\ þ 1
H ð xÞ ¼ ð1:2:8Þ
0 ; 1 \ x \ 0
Its formal Fourier transform is
Z1 Z1

hð nÞ ¼ H ð xÞ expðinxÞdx ¼ expðinxÞdx ð1:2:9Þ
1 0
This integral cannot be evaluated in the regular sense of calculus. Instead, we shall
look for the transformed image function hðnÞ so that its Fourier inversion integral
results in the step function. Remember the delta function whose Fourier transform
is a constant, and assume that the image function is the sum of delta function and a
newly introduced function h1 ðnÞ,
1.2 Distributions and Their Integration Formulas 7

hðnÞ ¼ pdðnÞ þ
h1 ð n Þ ð1:2:10Þ
Let us apply the Fourier inversion integral to the above function and look for the
suitable form of the unknown function h1 ðnÞ. The inversion integral is processed
as
Z1 Z1
1 1
H ð xÞ ¼
hðnÞ expðinxÞdn ¼ pdðnÞ þ h1 ðnÞ expðinxÞdn
2p 2p
1 1
Z1
1 1 must be 1; x[0
¼ þ
h1 ðnÞ expðinxÞdn )
2 2p 0; x\0
1
ð1:2:11Þ
From the last line in the above equation, we learn that the Fourier transform
(inversion) integral of the unknown function
h1 ðnÞ must be +1/2 in x [ 0 and -1/2
in x\0, i.e.
Z1
1 þ1=2 ; x [ 0
h1 ðnÞ expðinxÞdn ¼ ð1:2:12Þ
2p 1=2 ; x \ 0
1
We remember the integration formula that gives 1=2 in each semi-infinite x-

region (Erdélyi 1954 vol I, p. 64, 3), that is
Z1
1 sinðxnÞ þ1=2 ; x[0
dn ¼ ð1:2:13Þ
p n 1=2 ; x\0
0
Comparing Eq. (1.2.12) with Eq. (1.2.13), we learn that the whole integrand in
Eq. (1.2.12) must be the function
sinðxnÞ
ð1:2:14Þ
n
and thus we can guess that the suitable functional form for h1 ðnÞ is
i
h1 ð n Þ ¼ ð1:2:15Þ
n
We shall now examine whether this form is really suitable or not. Substitute the
above Eq. (1.2.15) into the last line in Eq. (1.2.11) and split the integral into real
and imaginary parts, we get
Z1 Z1 Z1
1 1 i 1 1 1 i 1
þ expðinxÞdx ¼ þ sinðnxÞdx þ cosðnxÞdx
2 2p n 2 2p n 2p n
1 1 1
ð1:2:16Þ
The third term on the right hand side vanishes due to the anti-symmetric nature of
the integrand. The second term is evaluated with aid of formula (1.2.13) as
Z1 Z1
1 1 sinðnxÞ 1 1 sinðnxÞ 1 1 þ1 ; x[0
þ dn ¼ þ dn ¼ þ
2 2p n 2 p n 2 2 1 ; x\0
1 0
1; x [ 0
¼ ð1:2:17Þ
0; x\0
Then, we learn that the supposed function h1 ðnÞ in Eq. (1.2.15) is correct and that
the Fourier transform of the step function is given by
i
hðnÞ ¼ pdðnÞ þ ð1:2:18Þ
n
This function is called ‘‘Heisenberg’s delta function’’ and its transform pair is
Z1 Z1
i 1
dþ ðnÞ ¼ pdðnÞ þ ¼ expðinxÞdx; H ð xÞ ¼ dþ ðnÞ expðinxÞdn
n 2p
0 1
ð1:2:19Þ
Heisenberg’s delta function dþ ðnÞ is the Fourier transform of the step function
HðxÞ. If we have another step function HðxÞ, its transform pair is also given by
Z0 Z1
i 1
d ðnÞ ¼ pdðnÞ ¼ expðinxÞdx; H ðxÞ ¼ d ðnÞ expðinxÞdn
n 2p
1 1
ð1:2:20Þ
Then, Heisenberg’s delta function has two definitions as
Z1
i 1
d ðnÞ ¼ pdðnÞ ; H ðxÞ ¼ d ðnÞ expðinxÞdn ð1:2:21Þ
n 2p
1
As a byproduct, we can obtain a new formula. Two operations, addition and

subtraction of Eqs. (1.2.19) and (1.2.20), give us the formulas as
Z1 Z1
1
cosðnxÞdn ¼ pdð xÞ; sinðnxÞdn ¼ ð1:2:22Þ
x
0 0
The formula on the left is the same as the integral in Eq. (1.2.6), but with c ¼ 0.
1.3 Comments on Inversion Techniques and Integration Formulas 9
1.3 Comments on Inversion Techniques and Integration

Formulas
The integral transform technique is a powerful tool for solving linear differential
equations. However, its success is up to the evaluation of the inversion integral. So
far, many integration formulas have been found and published. The most compre-
hensive formula books are Erdélyi (1954), Gradshteyn and Ryzhik (1980), Magnus
et al. (1966), Watson (1966), Moriguchi et al. (1972) and Titchmarsh (1948).
The books from Erdélyi (1954) to Watson (1966) are well-known and it is not
difficult to obtain access to them. The book Watson (1966) deals solely with Bessel
functions and not with integral transforms, but it gives many integration formulas
for the Hankel transform. The book Moriguchi et al. (1972), written in Japanese, is
very compact and is separated into three small handy books. In spite of its com-
pactness, the principal formulas which are included in Erdélyi (1954), Gradshteyn
and Ryzhik (1980), Magnus et al. (1966) and Watson (1966) are cited. The author
believes that the three handy books are most convenient as a ‘‘first aid.’’ The last
book Titchmarsh (1948) describes the mathematics of the theory of Fourier
transform. When someone needs more detailed mathematics for the integral
transform, this book will give proper answers.
If a desired formula cannot be found in these books, the complex integral is
employed to evaluate the inversion integral. The complex integral based on Cau-
chy’s integral theorem is the most useful evaluation technique. If the complex
integral does not give any compact result, the inversion integral is left in its definition
form or is converted to the numerically tractable form by the complex integral.
Exercises
(1.1) Apply the finite Fourier transform, complex Fourier series defined by
Eq. (1.1.6), to Dirac’s delta function defined in a finite region ðp; þpÞ as
dð x aÞ ; p \ a \ þ p ðaÞ
and then show the series form of the delta function,
" #
1 X1
dð x aÞ ¼ 1þ2 cosfnðx aÞg ðbÞ
2p n¼1
(1.2) If we expand the x-range to ð1; þ1Þ, the Fourier series in the above
equation (b) gives an infinite sequence of delta function, i e.
" #
1 X1 X
þ1
1þ2 cosfnðx aÞg ¼ dðx a 2mpÞ
2p n¼1 m¼1
Explain why the infinite sequence?

References
Erdélyi A (ed) (1954) Tables of integral transforms, vol I and II. McGraw-Hill, New York
Gradshteyn IS, Ryzhik IM (1980) In: Jefferey A (ed) Table of integrals, series, and products, 5th
edn. Academic Press, San Diego
Magnus W, Oberhettinger F, Soni RP (1966) Formulas and theorems for the special functions of
mathematical physics. Springer, New York
Moriguchi S, Udagawa K, Ichimatsu S (1972) Mathematical formulas, vol I, II, III. Iwanami,
Tokyo (in Japanese)
Sneddon IN (1951) Fourier transforms. McGraw-Hill, New York
Titchmarsh EC (1948) Introduction to the theory of Fourier integrals, 2nd edn. Clarendon Press,
Oxford
Watson GN (1966) A treatise on the theory of bessel functions. Cambridge University Press,
Cambridge
Chapter 2
Green’s Functions for Laplace and Wave
Equations
This chapter shows the solution method for Green’s functions of 1, 2 and 3D
Laplace and wave equations. Lengthy and detailed explanations are given in order
to instruct the basic technique of the integral transform. Especially, Fourier
inversion integral for the time-harmonic Green’s function is discussed in detail.
2.1 1D Impulsive Source
Now, we start from the simplest wave equation that has two variables: a single
space variable x and the time t,
o2 / 1 o2 /
¼ PdðxÞdðtÞ ð2:1:1Þ
ox2 c2 ot2
The nonhomogeneous term represents a wave source with magnitude P, and two
Dirac’s delta functions, dðxÞ and dðtÞ, show the location and the impulsive nature
of the source. The Green’s function is a particular solution of the differential
equation corresponding to the impulsive source. The Green’s function is sought
under the quiescent condition at an initial time,

o/
/jt¼0 ¼ ¼ 0 ð2:1:2Þ
ot t¼0
and the convergence condition at infinity,

o/
/jx!1 ¼ ¼0 ð2:1:3Þ
ox x!1
To obtain a particular solution of the wave equation (2.1.1), we apply the
integral transforms and reduce the differential equation to an algebraic equation in
the transformed domain. Since the unknown function /ðx; tÞ has two variables, we

12 2 Green’s Functions for Laplace and Wave Equations
apply the double transform: Laplace transform with respect to the time variable
t : 0 t\ þ 1,
Z1

f ðsÞ ¼ L½f ðtÞ ¼ f ðtÞ expðstÞdt ð2:1:4Þ
0
and Fourier transform with respect to the space variable x : 1\x\ þ 1,

Zþ1 Zþ1
f ðxÞ expðþinxÞdx; f ðxÞ ¼ ð2:1:5Þ
2p
1 1
Firstly, we multiply the kernel of the Laplace transform expðstÞ to both sides
of the differential equation (2.1.1),
Z1 2
o / 1 o2 /
¼ PdðxÞdðtÞ expðstÞdt ð2:1:6Þ
ox2 c2 ot2
0
and perform the Laplace transform integral term by term. The order of integration
and differentiation are interchanged for the first term in the left hand side of the
equation. The first term in the right hand side is left in its order and then integrated
by parts. The last nonhomogeneous term, which has the delta function, is evaluated
by using the formula (1.2.3). Then, Eq. (2.1.6) is rewritten as
Z1 Z1 Z1
d2 1 o2 /
/ expðstÞdt ¼ expðstÞdt PdðxÞ dðtÞ expðstÞdt ð2:1:7Þ
dx2 c2 ot2
0 0 0
We define the Laplace transform of the unknown function as

Z1

/ ¼ / expðstÞdt ð2:1:8Þ
0
The time-derivative term is integrated by parts,

Z1 2 t!1 Z1
o / o/ t!1 2
expðstÞdt ¼ expðstÞ þ ½s/ expðstÞt¼0 þ s / expðstÞdt
ot2 ot t¼0
0 0
¼ s 2 /
ð2:1:9Þ
where the quiescent condition at the initial time is incorporated. To the last
nonhomogeneous term, the simple integration formula for the delta function,
Z1
dðtÞ expðstÞdt ¼ 1 ð2:1:10Þ
0
2.1 1D Impulsive Source 13
is applied. Then, we have an ordinary differential equation for the function / in

the transformed domain,
d2 /
¼ ðs=cÞ2 / PdðxÞ ð2:1:11Þ
dx2
It is possible to obtain the exact solution for the above ordinary differential
equation by the elementary method. However, in order to demonstrate the integral
transform technique, we further apply Fourier transform to the ordinary differential
equation (2.1.11). The Fourier transform defined by Eq. (2.1.5) is applied to
Eq. (2.1.11),
Zþ1
d 2 / 2
¼ ðs=cÞ / PdðxÞ expðþinxÞdx ð2:1:12Þ
dx2
1
The Fourier transform integral is applied to each term as

Zþ1 Zþ1 Zþ1
d2 /
expðþinxÞdx ¼ ðs=cÞ2 / expðþinxÞdx P dðxÞ expðþinxÞdx
dx2
1 1 1
ð2:1:13Þ
The convergence condition is also transformed, as
Z1
o/ d/
/jx!1 ¼ ¼ 0 expðstÞdt ) / jx!1 ¼ ¼0
ox x!1 dx x!1
0
ð2:1:14Þ
Defining the Fourier transform of the Laplace transformed unknown function,
Zþ1

/ ¼ / expðþinxÞdx ð2:1:15Þ
1
the transform of the space-derivative in Eq. (2.1.13) is carried out with aid of the
convergence condition,
Zþ1 2
d /
expðþinxÞdx
dx2
1
x!þ1 Zþ1 ð2:1:16Þ
d/ x!þ1
¼ expðþinxÞ in½/ expðþinxÞx!1 n2
/ expðþinxÞdx
dx x!1
1
~
¼ n2 /
The integration formula for the delta function

Zþ1
dðxÞ expðþinxÞdx ¼ 1 ð2:1:17Þ
1
is also used for evaluating the last term. The Fourier transform of the ordinary
differential equation (2.1.12) yields a simple algebraic equation for the double

transformed unknown function / ,
s2
n2 / ¼ / P ð2:1:18Þ
c2

Then we have the exact expression for the double transformed function / ,
P
/ ¼ ð2:1:19Þ
n þ ðs=cÞ2
2
The unknown function has just been determined explicitly in the transformed
domain. We shall carry out two inverse transforms successively. As the first
inversion, we apply the Fourier inversion integral which is defined by the second
of Eq. (2.1.5). The formal Fourier inversion is given by the integral
Zþ1
1 P
/ ¼ expðinxÞdn ð2:1:20Þ
2p n þ ðs=cÞ2
2
1
Due to the symmetric nature of the integrand, the integral is reduced to the simpler
semi-infinite integral,
Z1
P 1
/ ¼ cosðnxÞdn ð2:1:21Þ
p n þ ðs=cÞ2
2
0
This is a simple integral and the integration formula (Erdélyi 1954, pp. 8, 11),
Z1
1 p
cosðxyÞdx ¼ expðajyjÞ ð2:1:22Þ
x2 þa 2 2a
0
can be applied. Then, Eq. (2.1.21) yields to

P
/ ¼ expðsjxj=cÞ ð2:1:23Þ
2ðs=cÞ
The next step is to carry out the Laplace inversion. The symbolical form for the
Laplace inversion is given by
Fig. 2.1 1D expanding wave

from a source point
cP/ 2
ct

cP 1 1
/ ¼ L1 ½/ ¼ L expðsjxj=cÞ ð2:1:24Þ
2 s
Fortunately, we have the Laplace inversion formula (Erdélyi 1954, pp. 241, 1),

1 0; t\a
L1 expðasÞ ¼ ¼ Hðt aÞ ð2:1:25Þ
s 1; t [ a
where Hð:Þ is Heaviside’s unit step function. Applying this formula to Eq. (2.1.24),
the solution for the non-homogeneous 1D wave equation is obtained as
cP cP
/¼ Hðt jxj=cÞ ¼ Hðct jxjÞ ð2:1:26Þ
2 2
This solution shows an expanding (or out-going) 1D wave with uniform amplitude
cP=2 as shown in Fig. 2.1. Consequently, we get the Green’s function for the 1D
wave equation,
o2 / 1 o2 / cP
¼ PdðxÞdðtÞ ) /ðx; tÞ ¼ Hðct jxjÞ ð2:1:27Þ
ox2 c2 ot2 2
2.2 1D Time-Harmonic Source
When the source is time-harmonic, the nonhomogeneous term in Eq. (2.1.1) is

replaced with a harmonic function, but the source location is unchanged. Thus, the
wave equation with a time-harmonic source is given by
o2 / 1 o2 /
¼ QdðxÞ expðþixtÞ ð2:2:1Þ
ox2 c2 ot2
where Q is the source magnitude and x the frequency of the time-harmonic
vibration. We assume that its solution satisfies the convergence condition at
infinity, i.e.

o/
/jx!1 ¼ ¼0 ð2:2:2Þ
ox x!1
As the first step of the solution method, we assume that the solution is also
time-harmonic,
/ðx; tÞ ¼ /# ðxÞ expðþixtÞ ð2:2:3Þ

where /# is called the ‘‘amplitude function.’’ Due to this assumption, the con-
vergence condition (2.2.2) is rewritten for the amplitude function,

#
d/#
/ x!1 ¼ ¼0 ð2:2:4Þ
dx x!1
Substituting the time-harmonic assumption of Eq. (2.2.3) into the wave equation
(2.2.1), we have the ordinary differential equation for the amplitude function,
d 2 /#
þ ðx=cÞ2 /# ¼ QdðxÞ ð2:2:5Þ
dx2
The exact solution of this ordinary differential equation can be obtained by the
elementary method. However, in order to demonstrate the integral transform
technique, we apply the Fourier transform, which is defined by Eq. (2.1.5), to the
ordinary differential equation (2.2.5),
Zþ1 2 #
d / 2 #
þ ðx=cÞ / ¼ QdðxÞ expðþinxÞdx ð2:2:6Þ
dx2
1
Defining the Fourier transform of the amplitude function as

Zþ1
#
/ ¼ /# expðþinxÞdx ð2:2:7Þ
1
the space derivative term in Eq. (2.2.6) is integrated by parts as

Zþ1
d2 /#
expðþinxÞdx
dx2
1
# x!þ1 Zþ1
d/ # x!þ1 2
¼ expðþinxÞ in / expðþinxÞ x!1 n /# expðþinxÞdx
dx x!1
1
ð2:2:8Þ
We apply the convergence condition of Eq. (2.2.4) and the definition of the Fourier
transform (2.2.7) to the above equation. The Fourier transform of the double
derivative is then reduced to
2.2 1D Time-Harmonic Source 17
Zþ1
d 2 /# #
expðþinxÞdx ¼ n2 / ð2:2:9Þ
dx2
1
The nonhomogeneous term is evaluated as

Zþ1
Q dðxÞ expðþinxÞdx ¼ Q ð2:2:10Þ
1
Then, Eq. (2.2.6) yields a simple algebraic equation for the Fourier transformed
amplitude function,
# þ ðx=cÞ2 /
n2 / # ¼ Q ð2:2:11Þ
The Fourier transformed amplitude is determined completely,
# ¼ Q
/ ð2:2:12Þ
n2 ðx=cÞ2
Our next task is to invert the transformed amplitude. Applying the formal
Fourier inversion integral to Eq. (2.2.12), we get
Zþ1
# 1 P
/ ¼ expðinxÞdn ð2:2:13Þ
2p n2 ðx=cÞ2
1
Inspecting the integrand, we see that it has two simple poles at n ¼ ðx=cÞ, that is
to say, the poles are located on the integration path (real axis in the complex n-
plane). Since the integration cannot be performed through these singular points, we
have to distort the integration path around the poles. There are two ways of
deforming the path. One is through an upper semi-circle, other is through a lower
semi-circle as shown in Fig. 2.2. We have to determine which semi-circle is
suitable. Discussing the nature of the initial wave equation (2.2.1), we learn that
the wave will expand to the outer region from the source point, i. e. wave radiation
Im(ξ )
+ω / c +∞
−∞ ? Re(ξ )
?
−ω / c
Fig. 2.2 Possible deformations of the integration path around the pole
from the source. Therefore, we have to choose the path so that the inversion
integral results in the radiation (out-going) wave from the source.
Still, it is somewhat complicated to explain the path selection. To aid the
understanding, two integrals with complex frequency are considered. After dis-
cussing the wave nature derived from each integral, we will determine and
understand the path distortion.
Let us introduce and add a small imaginary number e to the frequency in
Eq. (2.2.13) so that the poles are shifted from the real axis and are not on the
integration path. The frequency is considered in two ways, positive and negative
imaginary parts, x ! - ie. Employing the complex frequency, we consider the
complex integral,
Z
1 P
U¼ expðinxÞdn ð2:2:14Þ
2p n2 ðx=cÞ2
C
where the integrand is the same as that in the Fourier inversion integral (2.2.13),
but the frequency is complex, x ¼ - ie. The integration loop C for the two cases
of complex frequency, with positive and negative imaginary parts, is discussed
separately.
(1) Small positive imaginary: x ¼ - þ ie
When the frequency has a small positive imaginary part, the poles are shifted
from the real axis. The integration path C is chosen so that the integrand vanishes
on the large semi-circle with infinite radius. We employ the lower closed loop CðÞ
in the case of positive x and the upper loop C ðþÞ in that of negative x as shown in
Fig. 2.3.
Fig. 2.3 Integration path in Im(ξ )

the case of the positive
(+)
imaginary part of the C
frequency
x<0
(ϖ + iε ) / c
Re(ξ )
−(ϖ + iε ) / c
x>0
C (−)
Applying Cauchy’s integral theorem (Jordan’s lemma) to the complex integral

U with the loop C ðÞ in Fig. 2.3, the integral along the real axis is evaluated as the
residue at the lower pole n ¼ x=c ¼ ð- þ ieÞ=c ,
Zþ1 " #
1 P 2pi Pðn þ x=cÞ
expðinxÞdn ¼ expðinxÞ
2p n2 ðx=cÞ2 2p n2 ðx=cÞ2
n¼x=c
1
ð2:2:15Þ
Rewriting the above equation, we have for positive x,
Zþ1
1 P iP
2 2
expðinxÞdn ¼ expðþixx=cÞ; x[0 ð2:2:16Þ
2p n ðx=cÞ 2ðx=cÞ
1
On the other hand, when x\0, we employ the upper loop C ðþÞ for the complex
integral U and have
Zþ1 " #
1 P 2pi Pðn x=cÞ
expðinxÞdn ¼ expðinxÞ ð2:2:17Þ
n¼x=c
1
Then, in the case of negative x, we have

Zþ1
1 P iP
2 2
expðinxÞdn ¼ expðixx=cÞ ; x\0 ð2:2:18Þ
1
Unifying the two Eqs. (2.2.16) and (2.2.18), we have for the Fourier inversion
integral, where the frequency has a positive imaginary part, i.e.
Zþ1
1 P iP
2 2
expðinxÞdn ¼ expðþixjxj=cÞ ; x ¼ - þ ie
1
ð2:2:19Þ
(2) Small negative imaginary x ¼ - ie

When the imaginary part of the frequency is negative, the poles are also shifted
from the real axis as shown in Fig. 2.4. In order to guarantee the convergence at
the large semi-circle, the lower loop C ðÞ is taken in the case of positive x, and the
upper C ðþÞ in the case of negative x.
When x [ 0, we employ the loop C ðÞ and apply Jordan’s lemma to the
complex integral U in Eq. (2.2.14). The integral along the real axis is converted to
the residue at the lower pole, n ¼ xð¼ - ieÞ=c,
Fig. 2.4 Integration path in Re(ξ )

the case of the negative (+)
imaginary part of the C
frequency
x<0
−(ϖ − iε ) / c
Im(ξ )
(ϖ − iε ) / c
x>0
C (−)
Zþ1 " #
1 P 2pi Pðn x=cÞ
n¼x=c
1
ð2:2:20Þ
Rewriting the above, we have in the case of positive x,
Zþ1
1 P iP
2 2
expðinxÞdn ¼ expðixx=cÞ ; x [ 0 ð2:2:21Þ
1
Similarly, when we employ the upper loop C ðþÞ in the case of x\0,
Zþ1 " #
1 P 2pi Pðn þ x=cÞ
n¼x=c
1
ð2:2:22Þ
the Fourier inversion integral is evaluated as
Zþ1
1 P iP
2 2
expðinxÞdn ¼ expðþixx=cÞ ; x\0 ð2:2:23Þ
1
Unifying two Eqs. (2.2.21) and (2.2.23), we have for the frequency with the
negative imaginary part,
Zþ1
1 P iP
2 2
expðinxÞdn ¼ expðixjxj=cÞ ; x ¼ - ie
1
ð2:2:24Þ
Two expressions are obtained for the single Fourier inversion integral. They are
Eqs. (2.2.19) and (2.2.24) and are summarized in a unified expression as
(
iP
2ðx=cÞ expðixjxj=cÞ; x ¼ - ie
/# ¼ iP ð2:2:25Þ
þ 2ðx=cÞ expðþixjxj=cÞ; x ¼ - þ ie
The wave nature of the two expressions is discussed by multiplying the time
factor,
(
iP
2ðx=cÞ expfþixðt jxj=cÞg ; x ¼ - ie
/¼ iP ð2:2:26Þ
þ 2ðx=cÞ expfþixðt þ jxj=cÞg ; x ¼ - þ ie
Inspecting the argument of the exponential function in the above Eq. (2.2.26), the
upper solution for the negative imaginary part gives an out-going (radiation) wave
from the source, but the lower solution for the positive imaginary part gives an
incoming wave, coming from infinity. Since a suitable solution must have the out-
going wave behavior, we employ the upper formula in Eqs. (2.2.25) and (2.2.26).
Thus, the frequency with negative imaginary part is the suitable assumption. Then,
the suitable integration contour for the Fourier inversion integral is that of Fig. 2.4
and the final result for the time-harmonic response is given by
o2 / 1 o2 / iQ
2
¼ 2 2 QdðxÞ expðþixtÞ ) /¼ expfþixðt jxj=cÞg
ox c ot 2ðx=cÞ
ð2:2:27Þ
We have just learned that the proper selection for a complex frequency is
x ¼ - ie, i.e. negative imaginary part; and the suitable integration loop is CðÞ
in Fig. 2.4. If we do not introduce a small imaginary part and keep the integration
path and the location of the poles on the real axis, the integration path around the
pole should be deformed by a small semi-circle shown in Fig. 2.5. This defor-
mation is only valid in the case of a positive time factor, expðþixtÞ. If we use a
negative time factor expðixtÞ, the integration path on the real axis must be
deformed by that shown in Fig. 2.6. Thus, the selection of the deformed path
around the pole depends on the sign of the frequency. Finally, we could answer to
the initial question about the deformation of the integration path for the Fourier
inversion integral.
Im(ξ )
Integration path
exp(+iω t)
+ω / c +∞
−∞
Re(ξ )
−ω / c
Fig. 2.5 Path deformation for the positive frequency
Integration path
Im(ξ )
exp(−iω t)
+ω / c +∞
−∞
Re(ξ )
−ω / c
Fig. 2.6 Path deformation for the negative frequency
2.3 2D Static Source
Let us consider Green’s function for a typical partial differential equation, the so-
called Laplace equation. The Laplace equation with a source S is the nonhomo-
geneous differential equation,
o2 / o2 /
þ ¼ SdðxÞdðyÞ ð2:3:1Þ
ox2 oy2
The product of two delta functions in the nonhomogeneous term shows the
location of the source, i.e. the source S is placed at the coordinate origin ð0; 0Þ in
ðx; yÞ-plane. The convergence condition at infinity,

o/ o/
/jpxffiffiffiffiffiffiffiffiffi ¼ ¼ ¼0 ð2:3:2Þ
2 þy2 !1
ox pxffiffiffiffiffiffiffiffiffi
2 þy2 !1 oy pxffiffiffiffiffiffiffiffiffi
2 þy2 !1
is also imposed.
Now, we apply the integral transforms to Eq. (2.3.1). Since the unknown
function / has two space variables, we apply the double Fourier transform defined
by
2.3 2D Static Source 23
Zþ1 Zþ1
1
/ðnÞ ¼ /ðxÞ expðþinxÞdx; /ðxÞ ¼ /ðnÞ expðinxÞdn ð2:3:3Þ
2p
1 1
Zþ1 Zþ1
~ 1 ~
/ðgÞ ¼ /ðyÞ expðþigyÞdy; /ðyÞ ¼ /ðgÞ expðigyÞdg ð2:3:4Þ
2p
1 1
to the differential equation (2.3.1) successively, as

0 þ1 1
Zþ1 Z 2 2
@ o / o /
þ ¼ SdðxÞdðyÞ expðþinxÞdxA expðþigyÞdy ð2:3:5Þ
ox2 oy2
1 1
With aid of the convergence condition (2.3.2), each term is transformed as follows:
0 þ1 1
Zþ1 Z 2
@ o / ~
expðþinxÞdxA expðþigyÞdy ¼ n2 /
ox2
1 1
0 þ1 1
Zþ1 Z 2
@ o / ~
expðþinxÞdxA expðþigyÞdy ¼ g2 / ð2:3:6Þ
oy2
1 1
0 þ1 1
Zþ1 Z
@ SdðxÞdðyÞ expðþinxÞdxA expðþigyÞdy ¼ S
1 1
~ ¼ S for the double
Then, we have the simple algebraic equation ðn2 þ g2 Þ/
transformed function. and its solution is given by
~
¼ S
/ 2
ð2:3:7Þ
n þ g2
The reader will find that the partial differential equation (2.3.1) is transformed
to a simple algebraic equation. There is thus no need of solving a differential
equation directly. The subsequent inversion process is however crucial for the
solution. The formal Fourier inversion integral with respect to the parameter g,
Zþ1 Z1
¼ 1
/
S
expðigyÞdg ¼
S 1
cosðgyÞdg ð2:3:8Þ
2p 2 2 p 2
n þg n þ g2
1 0
is evaluated with aid of the formula (2.1.22) and yields
¼ S expðjnjjyjÞ
/ ð2:3:9Þ
2jnj
The next inversion is to evaluate the Fourier inversion integral with respect to
the parameter n,
Zþ1
1 S
/¼ expðjnjjyjÞ expðinxÞdn ð2:3:10Þ
2p 2jnj
1
Inspecting the integrand, the singular point at n ¼ 0 lies on the real axis, i.e. on the
integration path. It is impossible to evaluate the integral in the regular sense. So,
we have to deform the integration path around the pole as in the previous section.
But it was somewhat complicated to determine the path deformation. In order to
avoid this troublesome work, we consider a simpler way of inverting the integral.
Since the trouble stems from the singular point at n ¼ 0, in order to avoid the
trouble, we differentiate Eq. (2.3.10) with respect to the space variables, x and y,
respectively,
Zþ1
o/ S in
¼ expðjnjjyjÞ expðinxÞdn ð2:3:11Þ
ox 4p jnj
1
Zþ1
o/ S
¼ sgnðyÞ expðjnjjyjÞ expðinxÞdn ð2:3:12Þ
oy 4p
1
where sgn(.) is the sign function defined by

þ1 ; y[0
sgnðyÞ ¼ ð2:3:13Þ
1 ; y\0
Using the symmetric nature of the integrand in Eqs. (2.3.11) and (2.3.12), the
integrals are reduced to the real valued semi-infinite integrals,
Z1
o/ S
¼ expðnjyjÞ sinðnxÞdn ð2:3:14Þ
ox 2p
0
Z1
o/ S
¼ sgnðyÞ expðnjyjÞ cosðnxÞdn ð2:3:15Þ
oy 2p
0
The two integrals in the above equations are well-known from Calculus and we
have the formulas,
Z1 Z1
x y
expðnyÞ sinðnxÞdn ¼ 2 ; expðnyÞ cosðnxÞdn ¼ ð2:3:16Þ
x þ y2 x2 þ y 2
0 0
Then, Eqs.(2.3.14) and (2.3.15) are evaluated as

o/ S x o/ S y
¼ ; ¼ ð2:3:17Þ
ox 2p x þ y2
2 oy 2p x þ y2
2
The derivative of / is completely determined. We integrate the above two

equations with respect to x and y, respectively. This integration leads to two
expressions for the single function / as
o/ S
) /¼ log x2 þ y2 þ c0 ðyÞ
ox 4p
ð2:3:18Þ
o/ S
) / ¼ log x2 þ y2 þ c00 ðxÞ
oy 4p
Since the above two equations must be equal, two constant terms should be
identical,
c0 ðyÞ ¼ c00 ðxÞ ð2:3:19Þ

This condition is satisfied only when the two terms are pure constant and do not
include any space variable:
c0 ðyÞ ¼ c00 ðxÞ ¼ constant ð2:3:20Þ

Then, we have Green’s function for the Laplace equation,
o2 / o2 / S
þ ¼ SdðxÞdðyÞ ) /¼ log x2 þ y2 þ arbitrary constant
ox2 oy2 4p
ð2:3:21Þ
This Green’s function does not satisfy the convergence condition at infinity. This
is because we could not carry out the Fourier inversion integral of Eq. (2.3.10)
directly.
The 2D wave equation with an impulsive source is given by
o2 / o2 / 1 o2 /
þ ¼ PdðxÞdðyÞdðtÞ ð2:4:1Þ
ox2 oy2 c2 ot2
The nonhomogeneous term states that the source with magnitude P is placed at the
coordinate origin and is impulsive in time. The quiescent condition at an initial time,

o/
/jt¼0 ¼ ¼ 0 ð2:4:2Þ
ot t¼0

p ffiffiffiffiffiffiffiffiffi o/ o/
/j x2 þy2 !1 ¼ pffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffi ¼ 0 ð2:4:3Þ
ox x2 þy2 !1 oy x2 þy2 !1
are also employed. As the wave equation (2.4.1) has two space variables x and y,
and one time variable t, a triple integral transform is applied to the differential
equation (2.4.1): the Laplace transform with respect to the time variable,
Z1

/ ðsÞ ¼ /ðtÞ expðstÞdt ð2:4:4Þ
0
and the double Fourier transform with respect to the space variables,
Zþ1 Zþ1
1
2p
1 1
Zþ1 Zþ1
~ 1 ~
2p
1 1
Applying this triple integral transform with the quiescent and convergence con-
ditions, the original differential equation (2.4.1) is transformed to the simple
algebraic equation for the unknown function / ~
,
~
n2 / ~
g2 / ~
¼ ðs=cÞ2 /
P ð2:4:7Þ
Then, the exact expression for /~
in the transformed domain is given by
~
¼ P
/ ð2:4:8Þ
2
n þ g2 þ ðs=cÞ2
For the inversion, three inversion integrals must be carried out successively.
The first one is the Fourier inversion integral with respect to the parameter g. This
is reduced to a semi-infinite integral, as
Zþ1 Z1
¼ 1
/
P
expðigyÞdg ¼
P cosðgyÞ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2 dg
2p 2 2 2 p
n þ g þ ðs=cÞ
1 0 g2 þ n2 þ ðs=cÞ2
ð2:4:9Þ
The integral on the far right side is easily evaluated by applying the formula
(2.1.22). Then, the first Fourier inversion integral in Eq. (2.4.9) yields
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
/ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jyj n2 þ ðs=cÞ2 ð2:4:10Þ
2 n2 þ ðs=cÞ2
Secondly, we apply the inversion integral with respect to the parameter n,

Zþ1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 P
/ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jyj n2 þ ðs=cÞ2 expðinxÞdn ð2:4:11Þ
2p
1 2 n2 þ ðs=cÞ2
The above integral is also reduced to a semi-infinite integral, as

Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
1
/ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jyj n2 þ ðs=cÞ2 cosðnxÞdn ð2:4:12Þ
2p
0 n2 þ ðs=cÞ2
and we apply the integration formula (Erdélyi 1954, pp. 17, 27) to get
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp c x2 þ a2 cosðbxÞdx ¼ K0 a b2 þ c2 ð2:4:13Þ
x 2 þ a2
0
where K0 ð:Þ is the zeroth order modified Bessel function of the second kind. Then,
Eq. (2.4.12) takes the compact form
P s pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
/ ¼ K0 x2 þ y2 ð2:4:14Þ
2p c
The last inversion is the Laplace inversion. The Laplace inversion is symbol-
ically expressed as
P 1 h s pffiffiffiffiffiffiffiffiffiffiffiffiffiffi i
/¼ L K0 x2 þ y2 ð2:4:15Þ
2p c
We have the suitable inversion formula (Erdélyi 1954, pp. 277, 8),

Hðt aÞ 0 ; t\a
L1 ½K0 ðasÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffi
1 ffi
; t[a ð2:4:16Þ
2
t a 2
t2 a2
Applying this formula to Eq. (2.4.15), we have the simple expression for /,
(
cP H ðct r Þ cP 0 ; ct\r
/¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ p ffiffiffiffiffiffiffiffiffiffiffiffi
1
; ct [ r ð2:4:17Þ
2p 2p
ðctÞ2 r 2 ðctÞ2 r 2
where Hð:Þ is Heaviside’s unit step function and the radial distance r from the
source is defined by
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r ¼ x2 þ y 2 ð2:4:18Þ
Consequently, we have the exact expression for Green’s function of the 2D

wave equation as
o2 / o 2 / 1 o2 / cP Hðct rÞ
þ ¼ PdðxÞdðyÞdðtÞ ) /¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2:4:19Þ
ox2 oy2 c2 ot2 2p
ðctÞ2 r 2
When the source is vibrating harmonically, the nonhomogeneous 2D wave

equation is given by
o2 / o2 / 1 o2 /
þ ¼ QdðxÞdðyÞ expðþixtÞ ð2:5:1Þ
ox2 oy2 c2 ot2
where Q and x are magnitude and frequency of the source, respectively. We also
assume that its Green’s function satisfies the convergence condition at infinity,

p ffiffiffiffiffiffiffiffiffi o/ o/
/j x2 þy2 !1 ¼ pffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffi ¼ 0 ð2:5:2Þ
ox x2 þy2 !1 oy x2 þy2 !1
As the standard technique for the time-harmonic response, the Green’s function
is assumed as the product
/ðx; y; tÞ ¼ /# ðx; yÞ expðþixtÞ ð2:5:3Þ
where /# ðx; yÞ is an amplitude function to be determined. Substituting this
assumption into the nonhomogeneous wave equation (2.5.1), we have the reduced
wave equation (the so-called Helmholtz equation) for the amplitude function
/# ðx; yÞ,
o2 /# o2 /#
þ þ ðx=cÞ2 /# ¼ QdðxÞdðyÞ ð2:5:4Þ
ox2 oy2
The convergence condition (2.5.2) is also rewritten for the amplitude function as

o/# o/#
/# pffiffiffiffiffiffiffiffiffi ¼ ¼ ffi ¼0 ð2:5:5Þ
x2 þy2 !1 ox pxffiffiffiffiffiffiffiffiffi
2 þy2 !1 oy pffiffiffiffiffiffiffiffi
x2 þy2 !1
The double Fourier transform with respect to two space variables as defined by
Eqs. (2.4.5) and (2.4.6) is applied to the nonhomogeneous Helmholtz equation
(2.5.4),
2 3
Zþ1 Zþ1 2 # 2 #
4 o / o /
þ ¼ ðx=cÞ / QdðxÞdðyÞ expðþinxÞdx5 expðþigyÞdy
2 #
ox2 oy2
1 1
ð2:5:6Þ
Defining the double transform of the amplitude function as

2 3
Zþ1 Zþ1
~ #
¼
/ 4 / ðx; yÞ expðþinxÞdx5 expðþigyÞdy
#
ð2:5:7Þ
1 1
Equation (2.5.6) is transformed into the algebraic equation for the unknown
~
#,
amplitude function /
~ # ~
g2 /
n2 /
# ~ Q
¼ ðx=cÞ2 / #
ð2:5:8Þ
Then, the explicit expression for the amplitude function in the transformed domain
is
~
# ¼ Q
/ ð2:5:9Þ
n2 þ g2 ðx=cÞ2
As the first inversion, we apply the Fourier inversion integral with respect to the
parameter g. Its formal inversion integral is simplified as
Zþ1 Z1
# 1 Q Q cosðgyÞ
/ ¼ expðigyÞdg ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2 dg
2p n2 þ g2 ðx=cÞ2 p
1 0 g2 þ n2 ðx=cÞ2
ð2:5:10Þ
The far right integral is evaluated with aid of the formula (2.1.22) and yields
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Q
#
/ ¼ q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jyj n2 ðx=cÞ2 ð2:5:11Þ
2 2
2 n ðx=cÞ
The second Fourier inversion integral with respect to the parameter n is given
by
Zþ1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Q 1
#
/ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jyj n2 ðx=cÞ2 expðinxÞdn ð2:5:12Þ
4p
1 n2 ðx=cÞ2
The integrand in the above inversion integral has two branch points at n ¼ x=c
which are on the real axis in the complex n-plane, i.e. on the integration path.
Thus, the radical n2 ðx=cÞ2 changes its sign around the branch point and we
have to discuss the path deformation around these as was done in the case of the
1D time-harmonic problem in Sect. 2.2. Since a detailed discussion for the
introduction of branch cuts is somewhat complicated, we shall follow the math-
ematical procedure described in the book of Ewing, Jardetzky and Press (Ewing
et al. 1957, pp. 44–51) and explain it concisely.
In order to evaluate the inversion integral of Eq. (2.5.12), we define the integral
I as
1 1
I¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp inx y n2 ðx=cÞ2 dn ð2:5:13Þ
2p
1 n2 ðx=cÞ2
and consider the complex integral having the same integrand as that in the above
equation,
Z qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1
U¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp inx y n2 ðx=cÞ2 dn ð2:5:14Þ
2p 2 2
C n ðx=cÞ
The radical n2 ðx=cÞ2 has two branch points, and two branch cuts are
introduced in order to keep the radical single-valued. If we add a small negative
imaginary number to the frequency, as was done in Sect. 2.2, the branch points are
shifted from the real axis. Two branch cuts are introduced along the hyperbola as
shown in Fig. 2.7. When the imaginary part of the frequency vanishes, the branch
cut lies on the real and imaginary axes as shown by the red line in Fig. 2.8. Then,
the closed loop C for the complex integral is composed of four contours. One is a
straight line ABC along the real axis (green line), the second one is the line DEF
\
and GHI along the branch cut (blue line), a small circle FG around the branch
\ \
point and the last is two quarter circles, CD and AI , with infinite radius (dotted
curve). The integral along the small circle around the branch point vanishes as its
radius tends to zero, and the two integrals along the large arcs also vanish as the
Im(ξ )
branch cut
arg ( β ) = −π / 2
for x < 0
ξ = −(ω / c) Re(ξ )
arg ( β ) = +π / 2
ξ = +(ω / c)
for x > 0 arg ( β ) = −π / 2
branch cut
Fig. 2.7 Branch cuts in the case of the negative imaginary part of the frequency
Im(ξ )
A
Fourier inversion line
F
B
E
C Re(ξ )
H G
arg ( β ) = −π / 2
arg ( β ) = +π / 2
For x > 0
D I
Fig. 2.8 Closed loop C for the complex integral U in Eq. (2.5.14)
radius tends to the infinity. It is enough to consider two dominant integration paths
for the complex integral. One is the line ABC along the real axis and other is the
lines DEF and IHG along the branch cut. The argument and value of the radical
along the branch cut is shown in Table 2.1
Then, the closed loop C for the complex integral is ABCDEFGHIA in Fig. 2.8
and we can apply Cauchy’s theorem to the complex integral U. As no singular
point is included within the loop, and the integrals around all circular paths vanish,
the line integral along the real axis, i.e. the Fourier inversion integral, is converted
to the integrals along the branch line DEF and GHI. For the branch line integrals,
we replace the integration variable with n ¼ g or ig, where g is real and
positive. Then, the line integral is converted to real valued integrals, as
1 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp inx y n2 ðx=cÞ2 dn
2p
1 n2 ðx=cÞ2
Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1
¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cos y g2 þ ðx=cÞ2 expðgxÞdg
p
0 g2 þ ðx=cÞ2
ð2:5:15Þ
Zx=c qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cos y ðx=cÞ2 g2 sinðgxÞdg
p
0 ðx=cÞ2 g2
Zx=c qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
i 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cos y ðx=cÞ2 g2 cosðgxÞdg
p
0 ðx=cÞ2 g2
Table 2.1 Argument of the qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

Line n
radical along the branch line n2 ðx=cÞ2
! qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n ¼ ig
DE þi g2 þ ðx=cÞ2
n ¼ þg
EF þi ðx=cÞ2 g2
n ¼ þg
HG i ðx=cÞ2 g2
n ¼ ig
HI i g2 þ ðx=cÞ2
g is positive real
The Fourier inversion integral has just been converted to three real integrals.
Fortunately, we can evaluate these integral with aid of integration formulas. Two
integration formulas (Gradshteyn and Ryzhik 1980, pp. 755, 6.677, No. 4)*
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Za pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cos b a2 þ x2 expðcxÞdx pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cos b a2 x2 sinðcxÞdx
2
a þx 2 a2 x2
0 0
p pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ Y0 a b2 þ c2
2
ð2:5:16Þ
and (Erdélyi 1954, pp. 28, 42)
Za pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 p pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cos b a2 x2 cosðcxÞdx ¼ J0 a b2 þ c2 ð2:5:17Þ
a2 x 2 2
0
are applied to the right hand side of Eq. (2.5.15) and the Hankel function (Watson
1966, pp. 73) is also introduced. Equation (2.5.15) is simplified as
1 1
2p
1 n2 ðx=cÞ2
1 x pffiffiffiffiffiffiffiffiffiffiffiffiffiffi i x pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2:5:18Þ
¼ Y0 x2 þ y2 J0 x2 þ y2
2 c 2 c
i ð2Þ x pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ H0 x2 þ y2
2 c
where J0 ð:Þ and Y0 ð:Þare Bessel functions of the first and second kind, respectively.
Finally, we have a simple expression for the Fourier inversion integral. That is
1 1
2p
1 n2 ðx=cÞ2
i ð2Þ x pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ H0 x2 þ y2 ð2:5:19Þ
2 c
It will be shown later that the above equation is valid for the case of the positive
frequency, expðþixtÞ.
*Derivation of formula (2.5.16).

For the integration formula (Gradshteyn and Ryzhik 1980, pp. 755, 6.677, No. 4),
8 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi >
< pffiffiffiffiffiffiffiffiffi
1 ffi
sin z a2 b2 ; b\a
a2 b2
2 2
Y0 a x þ z cosðbxÞdx ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
>
: pffiffiffiffiffiffiffiffiffi
1 ffi
exp z b2 a2 ; b [ a
0 2 2 b a
ð2:5:20Þ
if we consider the above as the Fourier cosine transform with the transform
parameter b, its inverse cosine transform is given by
Za qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
Y0 a x2 þ z2 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sin z a2 b2 cosðbxÞdb
2 a2 b2
0
ð2:5:21Þ
Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp z b2 a2 cosðbxÞdb
b2 a 2
a
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Introducing the variable transform, u ¼ b2 a2 for the second integral in the
right hand side, the formula (2.5.16) is obtained as
Za qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
2
Y0 a x þ z ¼ 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sin z a2 b2 cosðbxÞdb
2 a2 b2
0
ð2:5:22Þ
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi expðzuÞ cos x u2 þ a2 du
u2 þ a2
0
The Fourier inversion integral has just been evaluated and is expressed by the
Hankel function of the second kind and the amplitude function of Eq. (2.5.12) is
expressed in a compact form as
iQ ð2Þ x pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
/# ¼ H0 x2 þ y2 ð2:5:23Þ
4 c
Thus the 2D Green’s function for the time-harmonic source is given by
o2 / o2 / 1 o2 /
þ ¼ QdðxÞdðyÞ expðþixtÞ )
ox2 ox2 c2 ot2 ð2:5:24Þ
iQ ð2Þ
/ ¼ H0 ðxr=cÞ expðþixtÞ
4
where r is the 2D radial distance from the source, defined by Eq. (2.4.18).
Let us examine our Green’s function to see if it satisfies the radiation condition
at infinity. Assuming a large distance from the source, i.e. r ! 1, we employ the
asymptotic formula for the Hankel function (Watson 1966, pp. 198),
rffiffiffiffiffi
ð2Þ 2
Hm ðzÞ expfiðz mp=2 p=4Þg ; z ! 1 ð2:5:25Þ
pz
The asymptotic form of the Green’s function (2.5.24) is thus given by
rffiffiffiffiffiffiffiffiffi
iQ 2c
/ expfþixðt r=cÞ þ ip=4g ; r ! 1 ð2:5:26Þ
4 pxr
It is clear that this equation shows an out-going wave from the origin, since the
argument of the exponential function includes only the wave propagation character
t r=c. Furthermore, its amplitude decays with the inverse square root of the
radial distance. Thus, our selection of the closed loop is correct and the Green’s
function of Eq. (2.5.24) is the suitable solution for our wave equation with a time-
harmonic source.
The static 3D Green’s function for the Laplace equation is a particular solution of
the nonhomogeneous differential equation,
o 2 / o2 / o2 /
þ þ 2 ¼ SdðxÞdðyÞdðzÞ ð2:6:1Þ
ox2 oy2 oz
where S is magnitude of the source, which is placed at the coordinate origin
ð0; 0; 0Þ. The convergence condition at infinity,

p ffiffiffiffiffiffiffiffiffiffiffiffiffiffi o/ o/ o/
/j x2 þy2 þz2 !1 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0
ox x2 þy2 þz2 !1 oy x2 þy2 þz2 !1 oz x2 þy2 þz2 !1
ð2:6:2Þ
is applied to the Green’s function /.
The Green’s function is obtained by the method of integral transform. For three
space variables, the triple Fourier transform defined by
Zþ1 Zþ1
1
2p
1 1
Zþ1 Zþ1
~ 1 ~
2p
1 1
Zþ1 Zþ1
^ 1 ^ expðifzÞdf ð2:6:5Þ
/ðfÞ ¼ /ðzÞ expðþifzÞdz; /ðzÞ ¼ /ðfÞ
2p
1 1
is applied to Eq. (2.6.1),

2 3
Zþ1 * Zþ1 Zþ1 2 +
4 o / o2 / o2 / 5
þ þ 2 ¼ SdðxÞdðyÞdðzÞ expðþinxÞdx expðþigyÞdy expðþifzÞdz
ox2 oy2 oz
1 1 1
ð2:6:6Þ
Applying the convergence condition (2.6.2), the above Eq. (2.6.6) is transformed
^~
to the algebraic equation for the triple transformed function /,
^
~
¼ S
ðn2 þ g2 þ f2 Þ/ ð2:6:7Þ
Thus, the transformed function is determined explicitly,
^
~
¼ S
/ ð2:6:8Þ
n þ g 2 þ f2
2
The first Fourier inversion integral with respect to the parameter f is

Zþ1 Z1
¼ 1
~
/
S
expðifzÞdf ¼
S 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2 cosðfzÞdf ð2:6:9Þ
2p 2 2 2 p
n þg þf f þ n2 þ g2
2
1 0
The above integral is easily evaluated with aid of the formula (2.1.22) which yields
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
~ S
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
/ ffi exp jzj n2 þ g2 ð2:6:10Þ
2 n2 þ g 2
The second inversion integral with respect to the parameter g is reduced to the
semi-infinite integral
Zþ1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ 1
/
S
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jzj n2 þ g2 expðigyÞdg
2p 2 n 2 þ g2
1
ð2:6:11Þ
Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
S 1
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jzj n2 þ g2 cosðgyÞdg
2p n2 þ g2
0
The latter semi-infinite integral can be evaluated by using formula (2.4.13). It

follows that
¼ S K0 jnj y2 þ z2
/ ð2:6:12Þ
2p
The last inversion integral with respect to the parameter n is also reduced to a
Zþ1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
S
/¼ K0 jnj y2 þ z2 expðinxÞdn
4p2
1
ð2:6:13Þ
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
S
¼ K0 n y2 þ z2 cosðnxÞdn
2p2
0
Fortunately, we have the integration formula, (Erdélyi 1954, pp. 49, 40)
Z1
p
K0 ðanÞ cosðbnÞdn ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2:6:14Þ
2 a2 þ b2
0
Applying this formula to Eq. (2.6.13), the last inversion integral is exactly eval-
uated as
S
/¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2:6:15Þ
4p x þ y2 þ z2
2
Consequently, we have the static 3D Green’s function for the Laplace equation,
o2 / o2 / o 2 / S
þ þ 2 ¼ SdðxÞdðyÞdðzÞ ) /¼ ð2:6:16Þ
ox2 oy2 oz 4pR
where R is the 3D radial distance from the source,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
R ¼ x2 þ y2 þ z2 ð2:6:17Þ
Green’s function for the 3D wave equation is discussed. The wave equation with
an impulsive point source located at the coordinate origin is given by
o2 / o2 / o2 / 1 o2 /
2
þ 2 þ 2 ¼ 2 2 PdðxÞdðyÞdðzÞdðtÞ ð2:7:1Þ
ox oy oz c ot
where P is the magnitude of the source. The quiescent condition at an initial time,

o/
/jt¼0 ¼ ¼ 0 ð2:7:2Þ
ot t¼0

p ffiffiffiffiffiffiffiffiffiffiffiffiffiffi o/ o/ o/
/j x2 þy2 þz2 !1 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0
ð2:7:3Þ
are also imposed.
Our dynamic Green’s function has four variables: three space and one time
variables. Laplace transform with respect to the time,
Z1
/ ðsÞ ¼ L½/ðtÞ ¼ /ðtÞ expðstÞdt ð2:7:4Þ
0
and the triple Fourier transform with respect to three space variables,
Zþ1 Zþ1
1
2p
1 1
Zþ1 Zþ1
~ 1 ~
2p
1 1
Zþ1 Zþ1
^ 1 ^ expðifzÞdf ð2:7:7Þ
/ðfÞ ¼ /ðzÞ expðþifzÞdz; /ðzÞ ¼ /ðfÞ
2p
1 1
are applied to the nonhomogeneous wave equation (2.7.1). With aid of the qui-
escent and convergence conditions, the wave equation is transformed to the simple
^~
algebraic equation for the multi-transformed unknown function / ,
^
~
¼ P
fn2 þ g2 þ f2 þ ðs=cÞ2 g/ ð2:7:8Þ
The inversion starts from the transformed function,

^
~
¼ P
/ ð2:7:9Þ
2
n þ g2 þ f2 þ ðs=cÞ2
As the first inversion, the Fourier inversion integral with respect to the parameter f
Zþ1
¼ 1
~
/
P
expðifzÞdf
2p 2
n þ g2 þ f2 þ ðs=cÞ2
1
Z1 ð2:7:10Þ
P cosðfzÞ
¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2 df
p
0 f2 þ n2 þ g2 þ ðs=cÞ2
is carried out by applying the formula (2.1.22). It follows

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
~
P
/ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jzj n2 þ g2 þ ðs=cÞ2 ð2:7:11Þ
2 n2 þ g2 þ ðs=cÞ2
The second inversion is the integral with respect to the parameter g,

Zþ1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 P
/ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jzj n2 þ g2 þ ðs=cÞ2 expðigyÞdg
2p 2 2 2
1 2 n þ g þ ðs=cÞ
Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P 1
¼ q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi exp jzj n2 þ g2 þ ðs=cÞ2 cosðgyÞdg
2p 2 2
0 n þ g2 þ ðs=cÞ
ð2:7:12Þ
The latter semi-infinite integral is also evaluated by the formula (2.4.13) and it
yields
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ P K0
/ n2 þ ðs=cÞ2 y2 þ z2 ð2:7:13Þ
2p
The third inversion integral with respect to the parameter n is
Zþ1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P 1
/ ¼ K0 n2 þ ðs=cÞ2 y2 þ z2 expðinxÞdn
2p 2p
1
ð2:7:14Þ
Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
¼ 2 K0 n2 þ ðs=cÞ2 y2 þ z2 cosðnxÞdn
2p
0
The integration formula (Erdélyi 1954, pp. 56, 43)

Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p
K0 a n2 þ b2 cosðcnÞdn ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp b a2 þ c2 ð2:7:15Þ
2 a2 þ c 2
0
is very helpful for our task. Then, applying the above formula (2.7.15) to the last
integral in Eq. (2.7.14), we have
P s pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
/ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp x 2 þ y 2 þ z2 ð2:7:16Þ
4p x2 þ y2 þ z2 c
The last inversion is the Laplace inversion

P h s pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii
/ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi L1 exp x2 þ y2 þ z2 ð2:7:17Þ
4p x2 þ y2 þ z2 c
The transform parameter ‘‘s’’ is included only in the argument of the exponential
function. We remember the simple Laplace inversion formula for the delta
function,
L1 ½expðasÞ ¼ dðt aÞ ð2:7:18Þ

Thus, Eq. (2.7.17) is fully inverted as
P
/¼ dðt R=cÞ ð2:7:19Þ
4pR
where the radial distance R from the source is defined by Eq. (2.6.17). Finally, we
have the 3D Green’s function for the wave equation with the impulsive point
source,
o2 / o2 / o2 / 1 o2 / P
þ þ ¼ PdðxÞdðyÞdðzÞdðtÞ ) /¼ dðt R=cÞ
ox2 oy2 oz2 c2 ot2 4pR
ð2:7:20Þ
This section derives the 3D Green’s function for a time-harmonic source. It is the
convolution integral of the impulsive Green’s function obtained in the previous
section. The wave equation with the time-harmonic source is given by
o2 / o2 / o2 / 1 o2 /
þ þ ¼ QdðxÞdðyÞdðzÞ expðþixtÞ ð2:8:1Þ
ox2 oy2 oz2 c2 ot2
where Q is magnitude of the source and x frequency of the time-harmonic
vibration. The convergence condition at infinity,

o/ o/ o/
/jpffiffiffiffiffiffiffiffiffiffiffiffiffiffi
x2 þy2 þz2 !1
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0
ð2:8:2Þ
is imposed.
The standard multiple integral transform technique is available for getting the
time-harmonic Green’s function. However, as shown in the case of 2D Green’s
function in Sect. 2. 5, a tiresome complex integration must be carried out. In order
to avoid this difficulty, we take a very simple way. That is, the convolution integral
of the Green’s function for the impulsive source.
In the previous section we get the Green’s function for the impulsive source as
o 2 / o2 / o 2 / 1 o2 / P
2
þ 2
þ 2
¼ 2 2
PdðxÞdðyÞdðzÞ ) / ¼ dðt R=cÞ ð2:8:3Þ
ox oy oz c ot 4pR
Replacing the source magnitude P with the time-harmonic function with the fre-
quency x,
P ! Q expfþixðt t0 Þg ð2:8:4Þ
and denote the time variable in the Green’s function (2.8.3) with t0 . The convo-
lution integral yields
Zt
Q
/¼ dðt0 R=cÞ expfþixðt t0 Þgdt0 ð2:8:5Þ
4pR
0
It is very easy to evaluate the above integral, since the integrand includes Dirac’s
delta function and we can apply the simple integration formula (1.2.3) in Sect. 1.2,
Zb
f ðcÞ ; a\c\b
f ðxÞdðx cÞdx ¼ ð2:8:6Þ
0 ; c\a or b\c
a
Then, we can evaluate the integral in Eq. (2.8.5) and have for /
Q
/¼ Hðt R=cÞ expfþixðt R=cÞg ð2:8:7Þ
4pR
The step function ahead of the equation means that Eq. (2.8.7) is the transient
response to the time-harmonic source and the disturbance starts from the wave
arrival t ¼ R=c. When sufficient long time has passed and the response becomes
steady, the step function is meaningless. Then, we have the steady-state time-
harmonic response as
Q
/¼ expfþixðt R=cÞg ð2:8:8Þ
4pR
Table A Green’s function for Laplace and Wave equations
Differential equation Source Green’s function
2
1D o2 / PdðxÞdðtÞ /ðx; tÞ ¼ cP
ox2 ¼ c12 oot/2 Source 2 Hðct jxjÞ
QdðxÞ expðþixtÞ iQ
/ðx; tÞ ¼ 2ðx=cÞ expfþixðt jxj=cÞg
2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2D o2 / SdðxÞdðyÞ S
ox2 þ ooy/2 ¼ Source /ðx; yÞ ¼ 2p logðrÞ þ arbitrary constant ; r ¼ x2 þ y2
2 2
o2 / PdðxÞdðyÞdðtÞ cP
2
ox2 þ oox/2 ¼ c12 oot/2 Source /ðx; y; tÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffi
ffi H ðct rÞ ; r ¼ x2 þ y2
2
2p ðctÞ r
ð2Þ
QdðxÞdðyÞ expðþixtÞ /ðx; y; tÞ ¼ iQ r ¼ x2 þ y 2
4 H0 ðxr=cÞ expðþixtÞ ;
2 2
3D o2 / SdðxÞdðyÞdðzÞ S
ox2 þ ooy/2 þ ooz/2 ¼ Source /ðx; y; zÞ ¼ 4pR ; R¼
x2 þ y2 þ z 2
2 2 2
o2 / PdðxÞdðyÞdðzÞdðtÞ P
ox2 þ ooy/2 þ ooz/2 ¼ c12 oot/2 Source /ðx; y; z; tÞ ¼ 4pR dðt R=cÞ ; R ¼ x2 þ y2 þ z2
Q
QdðxÞdðyÞdðzÞ expðþixtÞ /ðx; y; z; tÞ ¼ 4pR expfþixðt R=cÞg ; R ¼ x2 þ y2 þ z2
41
The 3D Green’s function for the wave equation with the time-harmonic source is
given by
o2 / o2 / o2 / 1 o2 /
2
þ 2 þ 2 ¼ 2 2 QdðxÞdðyÞdðzÞ expðþixtÞ
ox oy oz c ot ð2:8:9Þ
Q
) /¼ expfþixðt R=cÞg
4pR
Exercises
(2.1) Using the impulsive response (2.1.27) for 1D wave equation, derive the
time-harmonic Green’s function through the convolution integral and com-
pare it with the time-harmonic Green’s function (2.2.27).
(2.2) Using the impulsive Green’s function (2.4.19) for 2D wave equation, derive
the time-harmonic Green’s function (2.5.23) through the convolution
integral.
References
Erdélyi A (ed) (1954) Tables of integral transforms, vol I. McGraw-Hill, New York
Ewing WM, Jardetzky WS, Press F (1957) Elastic waves in layered media. McGraw-Hill, New
York
Gradshteyn IS, Ryzhik IM (1980) In: Jefferey A (ed) Table of integrals, series, and products, 5th
edn. Chap.6, Academic Press, San Diego
Cambridge
Chapter 3
Green’s Dyadic for an Isotropic
Elastic Solid
The present chapter shows how to derive an exact closed form solution, the
so-called Green’s dyadic, for elasticity equations. Introducing the Cartesian
coordinate system ðxi Þ ðx; y; zÞ, we employ the notation u ðui Þ for the dis-
placement, e ðeij Þ for the strain, r ðrij Þ for the stress, B ðBi Þ for the body
force and q for the density. The governing equations for the deformation of an
isotropic elastic solid are constituted by the following equations of motion,
orxx oryx orzx o2 ux

þ þ þ qBx ¼ q 2
ox oy oz ot
orxy oryy orzy o2 uy
þ þ þ qBy ¼ q 2 ð3:1Þ
ox oy oz ot
orxz oryz orzz o2 uz
þ þ þ qBz ¼ q 2
ox oy oz ot
and by the stress–strain relation, so-called Hooke’s law,
2 3 2 32 3
rxx k þ 2l k k 0 0 0 exx
6 ryy 7 6 k þ 2l k 0 0 0 7 6 eyy 7
6 7 6 76 7
6 rzz 7 6 þ 0 7 6 7
6 7 6 k 2l 0 0 76 ezz 7
6 rxz 7 ¼ 6 7
0 76 7 ð3:2Þ
6 7 6 2l 0 6 exz 7
4 ryz 5 4 sym: 2l 0 5 4 eyz 5
rxy 2l exy
Here ðk; lÞ are Lame’s constants which are expressed by Young’s modulus E and
Poisson ratio m as
mE E
k¼ ; l¼ ð3:3Þ
ð1 þ mÞð1 2mÞ 2ð1 þ mÞ
The strains in Eq. (3.2) are defined by the displacement gradient,

44 3 Green’s Dyadic for an Isotropic Elastic Solid
oux ouy ouz

exx ¼ ; eyy ¼ ; ezz ¼ ;
ox oy oz
ð3:4Þ
1 oux ouy 1 ouy ouz 1 oux ouz
exy ¼ þ ; eyz ¼ þ ; ezx ¼ þ
2 oy ox 2 oz oy 2 oz ox
It should be noticed that the symmetry relations for the stress and strain compo-
nents are satisfied:
rij ¼ rji ; eij ¼ eji ; i; j ¼ x; y; z ð3:5Þ
The elasticity equations constitute a set of coupled partial differential equations
with 15 unknowns. In order to reduce the differential equations to more compact
forms, the strain in Hooke’s law is replaced with the displacement gradient and
then the stress is substituted into the equations of motion. We then get a set of
displacement equations, the so-called Navier equations, with only three unknown
functions (displacement components),

2 o oux ouy ouz o2 ux o2 ux o2 ux 1 o2 ux 1
ðc 1Þ þ þ þ 2 þ 2 þ 2 ¼ 2 2 2 Bx
ox ox oy oz ox oy oz cs ot cs
2 2 2 2
o oux ouy ouz o uy o uy o uy 1 o uy 1
ðc2 1Þ þ þ þ 2 þ 2 þ 2 ¼ 2 2 2 By ð3:6Þ
oy ox oy oz ox oy oz cs ot cs
2 2 2 2
o oux ouy ouz o uz o uz o uz 1 o uz 1
ðc2 1Þ þ þ þ 2 þ 2 þ 2 ¼ 2 2 2 Bz
oz ox oy oz ox oy oz cs ot cs
where cs and cd are the velocities of shear and dilatational waves, respectively, and
defined by
rffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
l k þ 2l
cs ¼ ; cd ¼ ð3:7Þ
q q
The velocity ratio c is defined and expressed by Poisson ratio m,

sffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cd k þ 2l 2ð1 mÞ
c¼ ¼ ¼ ð3:8Þ
cs l 1 2m
Then, in this chapter we shall show a solution method for the coupled dis-
placement equations (3.6) with nonhomogeneous body force terms Bi . The solu-
tion method that we employ is the integral transforms: Fourier transforms with
respect to the space variables and Laplace transform with respect the time variable.
A particular solution corresponding to a point body force is called a Green’s
function. However, each displacement component produced by one source com-
ponent is called a ‘‘Green’s dyadic.’’ Thus, the present chapter shows the solution
method for the Green’s dyadic.
We assume here the deformation of plane-strain. Take the 2D coordinate system

ðx; yÞ in an infinite elastic solid and neglect the anti-plane deformation produced
by the anti-plane displacement component uz . The in-plane displacement com-
ponents ðux ; uy Þ are functions of two space variables ðx; yÞ and the time t. As a
wave source, we assume an impulsive point body force with magnitude Pi placed
at the coordinate origin. Under these assumptions, the displacement equations (3.6)
is reduced to the simpler form

o oux ouy o2 ux o2 ux 1 o 2 u x Px
ðc2 1Þ þ þ 2 þ 2 ¼ 2 2 2 dðxÞdðyÞdðtÞ
ox ox oy ox oy cs ot cs
2 2 2
ð3:1:1Þ
2 o oux ouy o uy o uy 1 o u y Py
ðc 1Þ þ þ 2 þ 2 ¼ 2 2 2 dðxÞdðyÞdðtÞ
oy ox oy ox oy cs ot cs
Now, we consider these coupled differential equations. The solutions are two
displacement components ðux ; uy Þ corresponding to the nonhomogeneous body
force. Our solution strategy is very simple, namely, to transform the differential
equations into two simultaneous algebraic equations in the transformed domain.
We employ Laplace transform with respect to time, defined by
Z1
uj ¼ uj expðstÞdt ð3:1:2Þ
0
and the double Fourier transform with respect to the two space variables, defined
by
Zþ1 Zþ1
1
uj ¼
uj expðþinxÞdx; uj ¼ uj expðinxÞdn
2p
1 1
ð3:1:3Þ
Zþ1 Zþ1
1
uj ¼
~ uj expðþigyÞdy; uj ¼ ~uj expðigyÞdg
2p
1 1
where the subscript j stands for x and y. The quiescent condition at an initial time,

ouj
uj t¼0 ¼ ¼0 ð3:1:4Þ
ot t¼0
and the convergence condition at infinity

ouj ouj
p ffiffiffiffiffiffiffiffiffi
uj x2 þy2 !1 ¼ ¼ ¼0 ð3:1:5Þ
2 þy2 !1 oy pffiffiffiffiffiffiffiffiffi
x2 þy2 !1
are assumed.
Applying the triple integral transform to the displacement equations (3.1.1), we

have the algebraic equations for the transformed displacement components ~uj ,

inðc2 1Þ in ux ig
~ uy n2 þ g2
~ ux ¼ ðs=cs Þ2 ~ux Px =c2s
~
ð3:1:6Þ
igðc2 1Þ in ux ig
~ uy n2 þ g2
~ uy ¼ ðs=cs Þ2 ~uy Py =c2s
~
Solving for the displacement, we have the exact expressions in the transformed
domain ðn; g; sÞ,

Px 1 1 1 1
ux ¼ 2 2 n2 Px þ ngPy 2 2 2
~
ð3:1:7aÞ
c s as s ad a s

Py 1 1 1 1
uy ¼ 2 2 ngPx þ g2 Py 2 2 2
~
ð3:1:7bÞ
c s as s ad a s
where two radicals are defined by
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ad ¼ n2 þ g2 þ ðs=cd Þ2 ; as ¼ n2 þ g2 þ ðs=cs Þ2 ð3:1:8Þ
Our main task is to invert the transformed displacement to the original space
ðx; y; tÞ. Examining the Eq. (3.1.7), we learn that the inversion of the four trans-
formed functions is enough for evaluating the displacement. These four funda-
mental functions are
I 0 ¼ 1
~ ð3:1:9Þ
c2s a2s
2
I xx ¼ n
~ 1

1
ð3:1:10Þ
s2 a2d a2s

I xy ¼ ng 1 1
~ ð3:1:11Þ
s2 a2d a2s
2

I yy ¼ g
~ 1

1
ð3:1:12Þ
s2 a2d a2s
If we have the four inversions, the displacement can be expressed as
ux ¼ Px ðI0 þ Ixx Þ þ Py Ixy ; uy ¼ Px Ixy þ Py ðI0 þ Iyy Þ ð3:1:13Þ
Then, we shall consider the inversion of the four fundamentals, successively.
(1) Inversion of ~
I 0
Rewriting ~

I 0 as
I 0 ¼ 1
~ 1
ð3:1:14Þ
c2s n2 þ g2 þ ðs=cs Þ2
we can find that this equation is the same as Eq. (2.4.8) for the transformed
~
with the replacement P ! 1=c2 . Then, we apply the same
Green’s function / s
mathematics as in Sect. 2.4 and have the inversion,
1 Hðcs t rÞ
I0 ðx; y; tÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3:1:15Þ
2pcs
ðcs tÞ2 r 2
where H(.) is Heaviside’s unit step function.
(2) Inversion of ~
I xx
As we have no previous result for the second inversion, the inversion integrals
are carried out successively. The Fourier inversion integral with respect to the
parameter g is given by
Z1

Ixx n2 1 1
¼ expðigyÞdg
2ps2 a2d a2s
1
ð3:1:16Þ
2 Z1 ( )
n 1 1
¼ 2 cosðgyÞdg
ps n2 þ g2 þ ðs=cd Þ2 n2 þ g2 þ ðs=cs Þ2
0
The semi-infinite integral is evaluated with use of the formula (2.1.22) and is
arranged so that each term has the same radical,
8 9
>
< qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 >
=
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 ðs=cd Þ

Ixx ¼ 2 n þ ðs=cd Þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jyj n2 þ ðs=cd Þ2
2 2
2s >: >
n2 þ ðs=cd Þ2 ;
8 9
>qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi >
1 < ðs=cs Þ2 =
þ 2 n2 þ ðs=cs Þ2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jyj n2 þ ðs=cs Þ2
2s >: >
n2 þ ðs=cs Þ2 ;
ð3:1:17Þ
The next Fourier inversion integral with respect to the parameter n is reduced to a
semi-infinite integral as
8 9
Z1 >
< qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 >
=
1 2 ðs=cd Þ

Ixx ¼ 2
n þ ðs=cd Þ q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jyj n2 þ ðs=cd Þ2 cosðnxÞdn
2ps2 > : >
0 n þ ðs=cd Þ ;
2 2
8 9
Z1 >
< qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 >
=
1 2 ðs=cs Þ
þ 2
n þ ðs=cs Þ q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jyj n2 þ ðs=cs Þ2 cosðnxÞdn
2ps2 > : >
0 n þ ðs=cs Þ ;
2 2
ð3:1:18Þ
We have already had the integration formula (2.4.13). Here, it is recited again,
1
x 2 þ a2
0
where K0 ð:Þ is the zeroth order modified Bessel function of the second kind. This
formula is applicable only for the second term in the bracket; another formula is
necessary for the first term. Differentiating the formula (3.1.19) twice with respect
to the parameter ‘‘c,’’ we have a new formula for our use,
x2 þ a2 exp c x2 þ a2 cosðbxÞdx
0 ð3:1:20Þ
a2 c 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi a b2 c2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
¼ 2 K a b 2 þ c2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi K a b 2 þ c2
0 1
b þ c2 b2 þ c 2 b 2 þ c 2
Applying these two formulas to the integrals in Eq. (3.1.18), we have
1 x2 1 x2 y2
Ixx ¼þ 2 2
K0 ðrs=cd Þ þ K1 ðrs=cd Þ
2pcd r 2psrcd r2
ð3:1:21Þ
1 x2 1 x2 y2
K0 ð rs=c s Þ K1 ðrs=cs Þ
2pc2s r 2 2psrcs r 2
where the radial distance from the source is defined by
r ¼ x2 þ y 2 ð3:1:22Þ
The last inversion is the Laplace inversion. Its inversion is expressed sym-
bolically as

1 x2 1 1 x2 y2 1 1
Ixx ¼ þ L ½ K 0 ð rs=c d Þ þ L K 1 ð rs=c d Þ
2pc2d r2 2prcd r2 s
ð3:1:23Þ
1 x2 1 1 x2 y2 1 1
L ½ K 0 ð rs=c s Þ L K1 ðrs=c s Þ
2pc2s r 2 2psrcs r 2 s
We have already used the inversion formula (2.4.16),
Hðt aÞ
L1 ½K 0 ðasÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3:1:24Þ
t 2 a2
One more inversion formula that includes the modified Bessel function of the
second kind is that (Erdélyi 1954, p. 277, 11),
1 1 t 2 a2
L K 1 ðasÞ ¼ Hðt aÞ ð3:1:25Þ
s a
Applying these two inversion formulas to Eq. (3.1.23), we have the final form for
Ixx ,
8 9
>
< 2 2 2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi=
>
Hðt r=cd Þ x 1 x y 2
Ixx ðx; y; tÞ ¼ þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ ðcd tÞ r 2
2pcd >
:r
2 r4 >
;
ðcd tÞ2 r 2
8 9
> qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi>
Hðt r=cs Þ <x2 1 x2 y2 2
=
q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ ðc s tÞ r 2
2pcs >
:r
2 r4 >
;
ðcs tÞ2 r 2
ð3:1:26Þ
(3) Inversion of ~
I yy
The same inversion procedure as that for ~ I xx is applied to the inversion of ~I yy ,
and its final form is
8 9
>
< 2 2 2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi=
>
Hðt r=cd Þ y 1 y x 2
Iyy ðx; y; tÞ ¼ þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ ðcd tÞ r 2
2pcd >
:r
2 r4 >
;
ðcd tÞ2 r 2
8 9
> qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi>
Hðt r=cs Þ <y2 1 y2 x2 2
=
q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ ðc s tÞ r 2
2pcs >
:r
2 r4 >
;
ðcs tÞ2 r 2
ð3:1:27Þ
The difference between Ixx in Eq. (3.1.26) and Iyy in Eq. (3.1.27) should be
noticed. The difference is in the space variables only. If we replace x with y in Ixx
of Eq. (3.1.26), we could have Iyy of Eq. (3.1.27). This is easily anticipated from
comparison of the transformed form in Eqs. (3.1.10) and (3.1.12). In these equa-
tions, each transform parameter corresponds to the original space variable. If we
exchange n with g in Eq. (3.1.10), it yields Eq. (3.1.12). Similarly, in the original
space we can exchange the space variables.
(4) Inversion of ~
I xy
The inversion technique is essentially the same as that for the former two cases,
but the integration formulas are slightly different. The Fourier inversion integral
with respect to the parameter g is reduced to the semi-infinite integral
Z1 Z1

Ixy 1 1 1 in g g
¼ ng 2 2 expðigyÞdg ¼ 2 sinðgyÞdg
2ps2 a d as ps a2d a2s
1 0
ð3:1:28Þ
The integration formula (Erdélyi 1954, p. 65, 15)
Z1
x p
sinðbxÞdx ¼ sgnðbÞ expðajbjÞ ð3:1:29Þ
x2 þ a2 2
0
is applied to Eq. (3.1.28). It then follows that

in

Ixy ¼ 2 sgnðyÞ exp jyj n2 þ ðs=cd Þ2 exp jyj n2 þ ðs=cs Þ2
2s
ð3:1:30Þ
where sgnð:Þ is the sign function defined by

þ1 ; x [ 0
sgnðxÞ ¼ ð3:1:31Þ
1 ; x\0
The next Fourier inversion integral with respect to the parameter n
Zþ1
sgnðyÞ 1

Ixy ¼ ðþinÞ exp jyj n2 þ ðs=cd Þ2 exp jyj n2 þ ðs=cs Þ2 einx dn
2s2 2p
1
ð3:1:32Þ
is reduced to the semi-infinite integral
Z1
sgnðyÞ

Ixy ¼ n exp ajyj n2 þ ðs=cd Þ2 exp ajyj n2 þ ðs=cs Þ2 sinðnxÞdn
2ps2
0
ð3:1:33Þ
Applying the integration formula (Erdélyi 1954, p. 75, 35)
a2 bc
x exp b x2 þ a2 sinðcxÞdx ¼ 2 K2 a b2 þ c 2 ð3:1:34Þ
b þ c2
0

to the integral in Eq. (3.1.33), we have for Ixy

xy 1 1
Ixy ¼ K 2 ð rs=c d Þ K 2 ð rs=c s Þ ð3:1:35Þ
2pr 2 c2d c2s
Lastly, the Laplace inversion formula (Erdélyi 1954 p. 277, 12)

1 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
L1 ½K2 ðbsÞ ¼ Hðt bÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ 2 t2 b2 ð3:1:36Þ
t 2 b2 b
is applied to Eq. (3.1.35). The final form of Ixy is given by
8 9
>
< 2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi>
=
xy ðr=cd Þ 2 ðr=c Þ2 Hðt r=c Þ
Ixy ¼ þ q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ 2 t d d
2pr 4 >
: t2 ðr=cd Þ2 >
;
8 9 ð3:1:37Þ
>
< 2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi>
=
xy ðr=cs Þ 2 ðr=c Þ2 Hðt r=c Þ
q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ 2 t s s
2pr 4 >
: t2 ðr=cs Þ2 >
;
We have obtained the exact expressions for the three, Ixx ; Ixy and Iyy , the unified
expression for these is given by
( pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi )
Hðt r=cd Þ xi xj 1 2xi xj ðcd t2 Þ r 2
Iij ðx; y; tÞ ¼ þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ dij
2pcd r2 ðcd t2 Þ r 2 r2 r2
8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi9
>
< >
Hðt r=cs Þ xi xj 1 2xi xj ðcs tÞ2 r 2 =
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ dij
2pcs >
:r
2 r2 r2 >
;
ðcs tÞ2 r 2
ð3:1:38Þ
where the subscripts i and j stand for x and y, and it should be understood that
xx x; xy y. Further, dij is Kronecker’s delta defined by

1; i¼j
dij ¼ ð3:1:39Þ
0 ; i 6¼ j
Thus, we have just obtained the exact expressions for the four fundamental
functions. Substituting Eqs. (3.1.15) and (3.1.38) into Eq. (3.1.13), the displace-
ment in the actual space ðx; y; tÞ is expressed as
2 8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi9
>
< >
X 6Hðt r=cd Þ xi xj 1 2xi xj ðcd tÞ2 r2 =
ui ðx; y; tÞ ¼ Pj 4 q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ d ij
2pcd >
:r
2 r2 r2 >
;
j¼x;y ðcd tÞ2 r2
> >
Hðt r=cs Þ <xi xj 1 2xi xj ðcs tÞ2 r2 =7
d ij q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ d ij 5
2pcs >
: r
2 r2 r2 >
;
ðcs tÞ2 r2
ð3:1:40Þ
Fig. 3.1 Radiation of P and y

SV waves from a point source
cd t
cs t
x
In this equation, the operation of Heaviside’s unit step function determines the
disturbed region. Hðt r=cd Þ shows a circular cylindrical region disturbed by the
dilatational wave which expands with velocity cd , while Hðt r=cs Þ does that by
the shear wave with cs . These two waves are the basic disturbances in the 2D
dynamic deformation and are shown in Fig. 3.1.
Now, we rewrite the above equation (3.1.40) as
X
ui ðx; y; tÞ ¼ Pj Gij ðx; y; tÞ ð3:1:41Þ
j¼x;y
The function Gij ðx; y; tÞ is called ‘‘Green’s dyadic’’ and it expresses the dis-
placement component in the i-axis direction due to a unit body force in the j-axis
direction. The explicit form of the dyadic Gij is given by
> >
Hðt r=cd Þ <xi xj 1 2xi xj ðcd tÞ2 r 2 =
Gij ðx; y; tÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ dij
2pcd >
:r
2 r 2 r 2 >
;
ðcd tÞ2 r2
8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi9 ð3:1:42Þ
> >
Hðt r=cs Þ <xi xj 1 2xi xj ðcs tÞ2 r2 =
dij qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ dij
2pcs >
: r
2 r 2 r 2 >
;
ðcs tÞ2 r2
When the source is harmonically vibrating with frequency x, the non-homoge-

neous body force term in the displacement equation (3.1.1) is replaced with the
time-harmonic source, i.e.

o oux ouy o2 ux o2 ux 1 o2 ux Qx
ðc2 1Þ þ þ 2 þ 2 ¼ 2 2 2 dðxÞdðyÞ expðixtÞ
ox ox oy ox oy cs ot cs
ð3:2:1aÞ

o oux ouy o2 uy o2 uy 1 o2 uy Qy
ðc2 1Þ þ þ 2 þ 2 ¼ 2 2 2 dðxÞdðyÞ expðixtÞ
oy ox oy ox oy cs ot cs
ð3:2:1bÞ
where Qi is magnitude of the source component.
It is possible to apply the integral transform method for obtaining the time-
harmonic Green’s dyadic. However, we do not employ the method here since the
cumbersome complex integration must be discussed as that for the time-harmonic
Green’s function in Sect. 2.5. We employ a simpler way, i.e. the convolution
integral. The time-harmonic response can be obtained by the convolution integral
of the impulsive response as
Zt
ðimpulseÞ
ui ðx; y; tÞ ¼ lim ui ðx; y; t0 Þ expfþixðt t0 Þgdt0 ð3:2:2Þ
t!1
0
ðimpulseÞ
where ui ðx; y; t0 Þ
is the impulsive response given by Eq. (3.1.41) with
(3.1.42). In this convolution integral, the time-harmonic function, expðþixtÞ, is
excluded from the limit, but we keep the upper limit of the integral be infinite,
since the steady-state response takes place long time after the initial disturbance.
Then, the convolution integral for the time-harmonic response takes the form
tZ
!1
ðimpulseÞ
ui ðx; y; tÞ ¼ expðþixtÞ ui ðx; y; t0 Þ expðixt0 Þdt0 ð3:2:3Þ
0
Substituting Eqs. (3.1.41) with (3.1.42) into Eq. (3.2.3), we have

2 8 9
> Z1 >
6
>
> xi x j
> expðixt0 Þ 0
>
>
>
6 >
> qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dt >
>
6 >
> r 2 2
0 Þ r2 >
>
X 6 1 < r=cd ðc d t =
ui ðx; y; tÞ ¼ expðþiÞ Qj 6
62pcd >
6 > Z qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 >
>
j¼x;y
6 >
> 2xi xj 1 >
>
4 >
> þ dij 2 ðcd tÞ2 r 2 expðixt0 Þdt0>
>
>
> r 2 r >
>
: ;
r=cd
8 9 3
>
> x x Z1 expðixt0 Þ >
>
>
> i j
dij qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dt 0 >
> 7
>
> >
> 7
>
> r
2 >
>
1 < r=cs ðcs t0 Þ2 r 2 =77
7
2pcs > Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi > 7
>
> 2xi xj 1 >
> 7
>
> 2 2 0 0>
> 7
>
> þ dij 2 ðcs tÞ r expðixt Þdt > > 5
>
: r 2 r >
;
r=cs
ð3:2:4Þ
where the source magnitude Pi for the impulsive solution is replaced with Qi . Two
integrals in the above equation are the integral representations for Hankel function
of the second kind (Watson 1966, p. 169). They are
Z1
expðixzÞ ip
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dx ¼ H0ð2Þ ðazÞ
x 2 a2 2
a
ð3:2:5Þ
ipa ð2Þ
x2 a2 expðixzÞdx ¼ þ H ðazÞ
2z 1
a
Applying these formulas to the integrals in Eq. (3.2.4), the time-harmonic

response is expressed as
X
ui ðx; y; tÞ ¼ Qj gij ðx; yÞ expðþixtÞ ð3:2:6Þ
j¼x;y
where gij ðx; yÞ is the Green’s dyadic for the time-harmonic response and is given
by

i 1 xi xj ð2Þ 2xi xj cd ð2Þ
gij ðx; yÞ ¼ H ðrx=c d Þ þ d ij H ðrx=c d Þ
4 c2d r 2 0 r 2 rx 1

1 xi x j ð2Þ 2xi xj cs ð2Þ
2 d ij H 0 ðrx=c s Þ þ d ij H ðrx=c s Þ
cs r2 r 2 rx 1
ð3:2:7Þ
Then, Eq. (3.2.6) with Eq. (3.2.7) gives the particular solution of the displacement
equation (3.2.1).
If the frequency is negative, x ! xð¼ xepi Þ, and the time factor is
expðixtÞ, we apply the formula
Hnð2Þ ðxepi Þ ¼ ð1Þnþ1 Hnð1Þ ðxÞ ð3:2:8Þ

to the Hankel function in Eq. (3.2.7), and then the displacement is expressed as
X
ui ðx; y; tÞ ¼ Qj gij ðx; yÞ expðixtÞ ð3:2:9Þ
j¼x;y
and its dyadic gij ðx; yÞ is given by

gij ðx; yÞ ¼ þ H ðrx=cd Þ þ dij 2 H ðrx=cd Þ
4 c2d r 2 0 r rx 1

1 xi xj
ð1Þ 2xi xj cs ð1Þ
2 d ij H 0 ðrx=c s Þ þ d ij H ðrx=c s Þ
cs r2 r 2 rx 1
ð3:2:10Þ
ð1Þ
where Hn ð:Þ is the n-th order Hankel function of the first kind, defined by
Hnð1Þ ðxÞ ¼ Jn ðxÞ þ iYn ðxÞ ð3:2:11Þ
If we consider to derive the static solution from the time-harmonic solution, we

have to take the limit x ! 0 in the dyadic given by Eq. (3.2.7) or (3.2.10).
However, the limit does not exist since the static displacement in the 2D plane
deformation does not vanish at infinity. We have to start from the original dif-
ferential equation and solve it. The static source is a non-time-dependent point
body force and the displacement equation which has no inertia term is given by

o oux ouy o2 ux o2 ux Sx
ðc2 1Þ þ þ 2 þ 2 ¼ 2 dðxÞdðyÞ
ox ox oy ox oy cs
2 2
ð3:3:1Þ
2 o oux ouy o uy o uy Sy
ðc 1Þ þ þ 2 þ 2 ¼ 2 dðxÞdðyÞ
oy ox oy ox oy cs
where the static body force is placed at the coordinate origin and its magnitude is
Si . For these coupled differential equations, the convergence condition at infinity

ouj ouj
p ffiffiffiffiffiffiffiffiffi
uj x2 þy2 !1 ¼ ¼ ¼0 ð3:3:2Þ
2 þy2 !1 oy pffiffiffiffiffiffiffiffiffi
x2 þy2 !1
is assumed in order to apply the Fourier transform (but, the final solution does not
satisfy the first condition in the above equation). We should understand that the
convergence condition guarantees the application of the Fourier transform.
In order to obtain the static Green’s dyadic, we apply the double Fourier
transform with respect to two space variables. Applying the double Fourier
transform defined by Eq.(3.1.3) to the displacement equations (3.3.1), we have the
algebraic equations for the transformed displacement components ~uj ,

inðc2 1Þ in~ uy n2 þ g2 ~ux ¼ Sx =c2s
ux ig~

ð3:3:3Þ
igðc2 1Þ in~ uy n2 þ g2 ~uy ¼ Sy =c2s
ux ig~

and their solutions are

( )
Sx 1 c2 1 n2 S y c2 1 ng
~
ux ¼ 2
2
cs n þ g2 2 2 2
c ðn þ g Þ 2 2 cs c ðn þ g2 Þ2
2 2
( ) ð3:3:4Þ
S x c2 1 ng Sy 1 c2 1 g2
~
uy ¼ 2
þ
cs c2 ðn2 þ g2 Þ2 c2s n2 þ g2 c2 ðn2 þ g2 Þ2
Examining the above equations, we learn that four inversions are needed for the
full Fourier inversion since the displacement is given by

Sx c2 1 ~ 2
I xx Sy c 1 ~I xy
ux ¼ 2 ~
~
I0 2
cs c cs c2
2
ð3:3:5Þ
Sx c2 1 ~
I xy þ Sy ~
2
I 0 c 1 ~I yy
~
uy ¼ 2

cs c2 c2s c2
where the four fundamentals to be inverted are
~
I 0 ¼ 1 ~
I xx ¼ n2 ~
I yy ¼ g2 ~I ¼ ng
2
; 2
; ; xy
n þ g2 2
ðn þ g2 Þ ðn þ2
g2 Þ 2 ðn þ g2 Þ2
2
ð3:3:6Þ
The inversion for each fundamental is carried out in the following subsections.
(1) Inversion of ~
I 0
We have already inverted this function in Sect. 2.3. The result is given by
Eq. (2.3.21), i.e.
~
¼ S S
/ ) /¼ logðx2 þ y2 Þ þ const: ð3:3:7Þ
n2 þ g2 4p
Applying this result, we have the inversion for ~
I 0 as
1
I0 ¼ logðx2 þ y2 Þ þ const: ð3:3:8Þ
4p
(2) Inversion of ~
I xx
The Fourier inversion integral with respect to the parameter g is reduced to the
Zþ1 Z1
Ixx ¼ 1 n2 n2 1
expðigyÞdg ¼ cosðgyÞdg ð3:3:9Þ
2p ðn2 þ g2 Þ2 p ðn þ g2 Þ2
2
1 0
The integration formula (Gradshteyn and Ryzhik 1980, p. 449, 3.729 1)

Z1
1 p
cosðaxÞdx ¼ ð1 þ abÞ expðabÞ ; a [ 0; b[0 ð3:3:10Þ
ðx2 þ b2 Þ 2 4b3
0
is applied to the far right integral in Eq. (3.3.9). We have for Ixx
2 Z1
Ixx ¼ n 1 1 jyj
2
cosðgyÞdg ¼ þ expðjnjjyjÞ ð3:3:11Þ
p ðn2 þ g2 Þ 4jnj 4
0
The last Fourier inversion with respect to the parameter n is given by the sum of
two integrals,
Zþ1
1 1 jyj
Ixx ¼ þ expðjnjjyjÞ expðinxÞdn
2p 4jnj 4
1
ð3:3:12Þ
Z1 Z1
1 1 jyj
¼ expðnjyjÞ cosðnxÞdn þ expðnjyjÞ cosðnxÞdn
4p n 4p
0 0
Inspecting the two integrals in the last equation, the second integral can be
evaluated by applying the formula (2.3.16). The first integral, however, has no
formula since its integrand has a first order singularity at n ¼ 0 and it is impossible
to evaluate the integral in this form. We extract the first integral
Z1
ð1Þ 1 1
Ixx ¼ expðnjyjÞ cosðnxÞdn ð3:3:13Þ
4p n
0
and consider its derivative with respect to each space variable as

ð1Þ Z1
oIxx 1
¼ expðnjyjÞ sinðnxÞdn
ox 4p
0
ð3:3:14Þ
ð1Þ Z1
oIxx 1
¼ expðnjyjÞ cosðnxÞdn
ojyj 4p
0
These two integrals are easily evaluated by the formulas (2.3.16) to yield
ð1Þ ð1Þ
oIxx 1 x oIxx 1 jyj
¼ ; ¼ ð3:3:15Þ
ox 4p x2 þ y2 ojyj 4p x2 þ y2
Then, we return the two derivatives to the original one by the integration with
ð1Þ
respect to each space variable. Two expressions for the single Ixx are obtained as
Z
ð1Þ 1 x 1
Ixx ¼ 2 2
dx ¼ logðx2 þ y2 Þ þ C1 ðyÞ
4p x þ y 8p
Z ð3:3:16Þ
ð1Þ 1 jyj 1 2 2
Ixx ¼ djyj ¼ logðx þ y Þ þ C2 ðxÞ
4p x2 þ y2 8p
These two equations must be equal. This condition is fulfilled by setting C1 ðyÞ ¼
C2 ðxÞ and the two functions must be a pure constant,
C1 ðyÞ ¼ C2 ðxÞ ¼ const: ð3:3:17Þ

ð1Þ
Then, we have for Ixx ,
ð1Þ 1
Ixx ¼ logðx2 þ y2 Þ þ const: ð3:3:18Þ
8p
The second integral in Eq. (3.3.12) is easily evaluated by the formula (2.3.16)
and it yields
Z1
ð2Þ jyj 1 y2
Ixx ¼ expðnjyjÞ cosðnxÞdn ¼ ð3:3:19Þ
4p 4p x2 þ y2
0
Finally, substituting Eqs. (3.3.18) and (3.3.19) into (3.3.12), we have the inversion
1 1 y2
Ixx ¼ logðx2 þ y2 Þ þ þ const: ð3:3:20Þ
8p 4p x2 þ y2
Further, since the constant in the above equation is arbitrary, we can rewrite Ixx as
1 1 x2
Ixx ¼ logðx2 þ y2 Þ þ const: ð3:3:21Þ
8p 4p x2 þ y2
(3) Inversion of ~
I yy
Rewriting the transformed equation in Eq. (3.3.6), we learn that this inversion
can be decomposed into the sum of the former two functions, as
~ 1 n2
I yy ¼ ¼ ~I 0 ~I xx ð3:3:22Þ
n2 þ g2 ðn2 þ g2 Þ2
Then, we can apply the former results, Eqs. (3.3.8) and (3.3.21), to the above
equation,
Iyy ¼ I0 Ixx
1 1 1 y2
¼ logðx2 þ y2 Þ þ const: þ logðx2 þ y2 Þ þ const:
4p 8p 4p x2 þ y2
ð3:3:23Þ
The final form of Iyy is given by
1 1 y2
Iyy ¼ logðx2 þ y2 Þ þ const: ð3:3:24Þ
8p 4p x2 þ y2
This can be derived from Ixx in Eq. (3.3.21) by the exchange of the variables.
Changing x with y in Eq. (3.3.21), we have the inversion for Iyy .
(4) Inversion of ~
I xy
The Fourier inversion integral with respect to the parameter g is reduced to the
Zþ1 Z1
Ixy ¼ 1 ng in g
2
expðigyÞdg ¼ sinðgyÞdg ð3:3:25Þ
2p 2 2
ðn þ g Þ p ðn þ g2 Þ2
2
1 0
The integration formula (Gradshteyn and Ryzhik 1980, p. 449, 3.729 2),
Z1
x pa
sinðaxÞdx ¼ expðabÞ ð3:3:26Þ
ðx2 þ b2 Þ 2 4b
0
is applied to the last integral in Eq. (3.3.25). We have
Ixy ¼ iy sgnðnÞ expðjnjjyjÞ ð3:3:27Þ

4
The next Fourier inversion integral with respect to the parameter n is reduced to
the simple integral,
Zþ1
iy 1
Ixy ¼ sgnðnÞ expðjnjjyjÞ expðinxÞdn
4 2p
1
ð3:3:28Þ
Z1
y
¼ expðnjyjÞ sinðnxÞdn
4p
0
Applying the formula (2.3.16), we have for Ixy ,

1 xy
Ixy ¼ ð3:3:29Þ
4p x2 þ y2
The four inversions in Eq. (3.3.5) have thus been completed. Since the dis-
placement in the original space is given by

Sx c2 1 Sy c2 1
ux ¼ 2 I 0 2
I xx 2 Ixy
cs c c s c2
ð3:3:30Þ
Sx c 2 1 Sy c2 1
uy ¼ 2 Ixy þ 2 I0 Iyy
c s c2 cs c2
we substitute I0 and Iij given by Eqs. (3.3.8), (3.3.21), (3.3.24) and (3.3.29) into the
above Eq. (3.3.30). The final form for the displacement is given by

x2
Sx 2 2 Sy xy
ux ¼ 2
ðc þ 1Þ logðrÞ þ ðc 1Þ þ 2
ðc2 1Þ 2
4pcd r 4pcd r

y2 ð3:3:31Þ
Sx xy S y
uy ¼ ðc2 1Þ 2 þ ðc2 þ 1Þ logðrÞ þ ðc2 1Þ
4pc2d r 4pc2d r
where cd ¼ ccs defined by Eq. (3.5) is used. The constant term is omitted since it
gives a simple rigid motion but no strain. The reader should notice that the con-
stant term in the displacement breaks the applicability of the Fourier transform;
thus the last inversion integral with respect to the parameter n has a singular point.
The expression for the displacement is rewritten in the form of a dyadic,
ux ¼ Sx gðstaticÞ
xx ðx; yÞ þ Sy gðstaticÞ
xy ðx; yÞ
ð3:3:32Þ
uy ¼ Sx gðstaticÞ
xy ðx; yÞ þ Sy gðstaticÞ
yy ðx; yÞ
where the dyadic for the static source is given by

x2
1
gðstaticÞ
xx ðx; yÞ ¼ ðc 2
þ 1Þ logðrÞ þ ðc 2
1Þ
4pc2d r
1 xy
gðstaticÞ
xy ðx; yÞ ¼ ðc2 1Þ 2 ð3:3:33Þ
4pc2d r

y2
1
gðstaticÞ
yy ðx; yÞ ¼ ðc 2
þ 1Þ logðrÞ þ ðc 2
1Þ
4pc2d r
This static Green’s dyadic for the plane deformation is called the ‘‘two dimen-
sional Kelvin’s solution.’’
An impulsive point source placed at the coordinate origin is expressed by the body
force
0 1 0 1
Bx Px
@ By A ¼ @ Py AdðxÞdðyÞdðzÞdðtÞ ð3:4:1Þ
Bz Pz
where Pi is the magnitude in the i-direction. The displacement equation with this
impulsive source is given by

o oux ouy ouz o2 ux o2 ux o2 ux
ðc2 1Þ þ þ þ 2 þ 2 þ 2
ox ox oy oz ox oy oz
1 o2 ux P x
¼ 2 dðxÞdðyÞdðzÞdðtÞ
c2s ot2 cs

o oux ouy ouz o2 uy o2 uy o2 uy
ðc2 1Þ þ þ þ 2 þ 2 þ 2
oy ox oy oz ox oy oz
ð3:4:2Þ
1 o2 uy P y
c2s ot2 cs

o oux ouy ouz o2 uz o2 uz o2 uz
ðc2 1Þ þ þ þ 2 þ 2 þ 2
oz ox oy oz ox oy oz
1 o 2 u z Pz
c2s ot2 cs
where cs is shear wave velocity. The velocity ratio c is defined by Eq. (3.8). For
these displacement equations, we impose the quiescent condition at an initial time,
oui
ui jt¼0 ¼ j ¼0 ð3:4:3Þ
ot t¼0
oui pffiffiffiffiffiffiffiffiffiffiffiffiffiffi oui pffiffiffiffiffiffiffiffiffiffiffiffiffiffi oui pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ui jpxffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 þy2 þz2 !1 ¼ j ¼ j ¼ j ¼0
ð3:4:4Þ
Laplace transform with respect to the time variable,
Z1
uj ¼ uj expðstÞdt ð3:4:5Þ
0
and the triple Fourier transform with respect to three space variables,
Zþ1 Zþ1
1
uj ¼
uj expðþinxÞdx; uj ¼ uj expðinxÞdn
2p
1 1
Zþ1 Zþ1
1
uj ¼
~ uj expðþigyÞdy; uj ¼ ~uj expðigyÞdg ð3:4:6Þ
2p
1 1
Zþ1 Zþ1
1
uj ¼
^ uj expðþifzÞdz; uj ¼ ûj expðifzÞdf
2p
1 1
are applied to the displacement equations (3.4.2). The simple algebraic equations
for the transformed displacement are
2
^ ^
^ 2 ^ s ^ P x
2
ðc 1Þ n2
~ ~ ~ 2 2
nguy nfuz ðn þ g þ f Þux ¼
ux ~
~ux
cs c2s

s 2 ^ P y
ðc2 1Þ ng^ uy gf^
ux g 2 ^ uz ðn2 þ g2 þ f2 Þ^~uy ¼

~
~
~
~uy ð3:4:7Þ
cs c2s

s 2 ^ P z
ðc2 1Þ nf^
ux gf^ uz ðn2 þ g2 þ f2 Þ^~uz ¼
uy f2 ^

~
~
~
~uz
cs c2s
and the displacement components in the transformed domain are given by
!
^ Px 2 1 1 1
~
ux ¼
n Px þ ngPy þ nfPz 2
c2s b2s s b2d b2s
!
^ Py 2
1 1 1
~
uy ¼
ngPx þ g Py þ gfPz 2 ð3:4:8Þ
c2s b2s s b2d b2s
!
^ Pz 2
1 1 1
~
uz ¼
nfPx þ gfPy þ f Pz 2
c2s b2s s b2d b2d
where the radicals are defined by

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
bd ¼ n2 þ g2 þ f2 þ ðs=cd Þ2 ; bs ¼ n2 þ g2 þ f2 þ ðs=cs Þ2 ð3:4:9Þ
In order to explore the convenient way for the inversion of the transformed
displacement, we inspect the expression of Eq. (3.4.8). It is found that seven
inversion formulas are necessary. They are
I 0 ðn; g; f; sÞ ¼ 1
^
~ ð3:4:10Þ
c2s b2s
! !
2 2
I 11 ðn; g; f; sÞ ¼ n
^ 1 1 ^ g 1 1

~ ~
; I 12 ðn; g; f; sÞ ¼ 2 ;
s2 b2d b2s s b2d b2s
! ð3:4:11Þ
2
I 13 ðn; g; f; sÞ ¼ f
^ 1 1

~
s2 b2d b2s
! !
I 21 ðn; g; f; sÞ ¼ ng 1 1 ;
^ I 22 ðn; g; f; sÞ ¼ gf 1 1 ;
^

~ ~
s2 b2d b2s s2 b2d b2s
! ð3:4:12Þ
^
~
I 23 ðn; g; f; sÞ ¼ nf 1 1

s2 b2d b2s
^
The first inversion for ~
I 0 is the same as the inversion of the 3D wave equation in
Sect. 2.7. Applying that result to our inversion, we have
1
I0 ðx; y; z; tÞ ¼ dðt R=cs Þ ð3:4:13Þ
4pc2s R
where the 3D radial distance is defined by
R ¼ x2 þ y2 þ z2 ð3:4:14Þ
^
Subsequently, we consider the inversion of ~I 1j in Eq. (3.4.11). The three Fourier
transforms are same in the form of definition integral and thus the exchange of two
space variables is equivalent to the exchange of two integration parameters. That is
^
I 12 ðn; g; f; sÞ ¼ ^
~
~
I 11 ðg; n; f; sÞ ) I12 ðx; y; z; tÞ ¼ I11 ðy; x; z; tÞ
ð3:4:15Þ
I 13 ðn; g; f; sÞ ¼ ^
^
~
~
I 11 ðf; g; n; sÞ )
I13 ðx; y; z; tÞ ¼ I11 ðz; y; x; tÞ
^
Due to this exchangeability, if we could obtain the inversion for ~I 11 as Fðx; y; z; tÞ,
^
the inversion of ~
I is given by the exchange of the space variables as Fðy; x; z; tÞ
12
^
and the inversion of ~ I 13 is also given by the variable exchange as Fðz; y; x; tÞ. Thus,
we can draw a schematic inversion diagram as
!
2
^
~
I 11 ðn; g; f; sÞ ¼ n 1 1
) I11 ðx; y; z; tÞ ¼ Fðx; y; z; tÞ
s2 b2d b2s
# exchange n with g # exchange x with y ð3:4:16Þ
!
2
I 12 ðn; g; f; sÞ ¼ g
^ 1 1

~ 2 ) I12 ðx; y; z; tÞ ¼ Fðy; x; z; tÞ
s2 2
bd bs
^
Similarly, the third inversion ~ I 13 is also obtained by the exchange of the space
variables as
!
2
^
~
I 11 ðn; g; f; sÞ ¼ n 1 1
) I11 ðx; y; z; tÞ ¼ Fðx; y; z; tÞ
s2 b2d b2s
# exchangen with f # exchange x with z ð3:4:17Þ

!
2
I 13 ðn; g; f; sÞ ¼ f
^ 1 1

~ ) I13 ðx; y; z; tÞ ¼ Fðz; y; x; tÞ
s2 b2d b2s
Then, the necessary inversion formula is reduced to only one, that is

!
2
~I ðn; g; f; sÞ ¼ n
^
1

1
ð3:4:18Þ
11
s2 b2d b2s
As for the second group of the inversion in Eq. (3.4.12), we can learn that the
necessary inversion formula is
!
I 21 ðn; g; f; sÞ ¼ ng 1 1
^
~ ð3:4:19Þ
s2 b2d b2s
If we could get the inversion as Gðx; y; z; tÞ, the other two are given by the
exchange of the space variables. They are
!
^
~
I 21 ðn; g; f; sÞ ¼ ng 1 1
) I21 ðx; y; z; tÞ ¼ Gðx; y; z; tÞ
s2 b2d b2s
# exchange n with f # exchange x with z ð3:4:20Þ
!
I 22 ðn; g; f; sÞ ¼ gf 1 1
^
~ ) I22 ðx; y; z; tÞ ¼ Gðz; y; x; tÞ
s2 b2d b2s
and
!
I 21 ðn; g; f; sÞ ¼ ng 1 1
^
~ ) I21 ðx; y; z; tÞ ¼ Gðx; y; z; tÞ
s2 b2d b2s
# exchange g with f # exchange y with z ð3:4:21Þ
!
I 23 ðn; g; f; sÞ ¼ nf 1 1
^
~ ) I23 ðx; y; z; tÞ ¼ Gðx; z; y; tÞ
s2 b2d b2s
^ ^
Consequently, it is enough to develop the inversion formulas only for ~I 11 and ~I 21 .
^
(1) Inversion of ~
I 11
The formal Fourier inversion integral with respect to the parameter f is given by
Zþ1 !
~
I 11 ðn; g; z; sÞ 1 n2 1 1
¼ expðifzÞdf
2p s2 b2d b2s
1
! ð3:4:22Þ
Z1
n2 1 1
¼ 2 cosðfzÞdf
ps f2 þ a2d f2 þ a2s
0
where
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
aj ¼ n2 þ g2 þ ðs=cj Þ2 ; j ¼ d; s ð3:4:23Þ
Applying the simple integration formula (2.1.22) to the integral in Eq. (3.4.22), we
have
2

I 11 ðn; g; z; sÞ ¼ n
~ 1
expða d jzjÞ
1
expða d jzjÞ ð3:4:24Þ
2s2 ad ad
The second inversion integral with respect to the parameter g is given by

Zþ1

I11 1 n2 1 1
ðn; y; z; sÞ ¼ 2
expðad jzjÞ expðad jzjÞ expðigyÞdg
2p 2s ad ad
1
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi3
Z 1 exp jzj n2 þ g2 þ ðs=c Þ2 exp jzj n2 þ g2 þ ðs=cs Þ2
d
n2 6 7
¼ 6 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 7 cosðgyÞdg
2ps2 4 2 2 þ ðs=c Þ 2 2 2 þ ðs=c Þ 2
5
0 n þ g d n þ g s
ð3:4:25Þ
The integration formula (2.4.13)
1
x 2 þ a2
0
is applied to the integral in Eq. (3.4.25). We have for I11

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n2

I11 ðn; y; z; sÞ ¼ K0 r n þ ðs=cd Þ K0 r n2 þ ðs=cs Þ2
2 2
ð3:4:27Þ
2ps2
where K0 ð:Þ is the zeroth order modified Bessel function of the second kind and
r ¼ y2 þ z2 ð3:4:28Þ
The last Fourier inversion integral with respect to the parameter n is given by
Zþ1
1 n2 2

I11 ðx; y; z; sÞ ¼ K 0
r n 2
þ ðs=c d Þ K r n2 þ ðs=cs Þ2
0 expðinxÞdn
2p s2
1
Z1
1
¼ n2 K0 r n2 þ ðs=cd Þ2 K0 r n2 þ ðs=cs Þ2 cosðnxÞdn
2p2 s2
0
ð3:4:29Þ
The last integral, which includes a modified Bessel function, has not yet been
tabulated in reference books. However, we have a formula which resembles our
integral, i.e.
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p exp c a2 þ b2
I¼ K0 a x2 þ c2 cosðbxÞdx ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3:4:30Þ
2 a 2 þ b2
0
*Notice that the above equation is just the Fourier cosine inversion of
Eq. (3.4.26)!
If we differentiate this integration formula with respect to the parameter ‘‘b’’
twice, it yields to our necessary formula,
Z1 ( pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi )
o2 I pffiffiffiffiffiffiffiffiffiffiffiffiffiffi p o2 exp c a2 þ b2
2
2¼ x K0 a x2 þ c2 cosðbxÞdx ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ob 2 ob2 a2 þ b 2
0
ð3:4:31Þ
That is,
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
x2 K0 a x2 þ c2 cosðbxÞdx
0

p 1 1 3b2 b2 c 2
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ c 1 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp c a2 þ b2
2 a2 þ b2 a þ b2
2 a þ b2 a2 þ b 2
ð3:4:32Þ
Applying this new formula (3.4.32) to the integral in Eq. (3.4.29), we have

1 3x2 1 3x2 1 x2
I11 ðx; y; z; sÞ ¼ 1 2 þ 1 2 expfsðR=cd Þg
4pR2 R Rs2 R cd s c2d R

3x2 1 3x2 1 x2
1 2 þ 1 expfsðR=c s Þg
R Rs2 R2 cs s c2s R
ð3:4:33Þ
where the 3D radial distance R is defined by Eq. (3.4.14).
The last Laplace inversion is carried out by applying the inversion formulas,
L1 ½expðasÞ ¼ dðt aÞ ð3:4:34Þ

1 1 ; t[a
L1 expðasÞ ¼ Hðt aÞ ¼ ð3:4:35Þ
s 0 ; t\a

1 t a ; t[a
L1 2 expðasÞ ¼ Hðt aÞðt aÞ ¼ ð3:4:36Þ
s 0 ; t\a
where dð:Þ and Hð:Þ are Dirac’s delta and Heaviside’s unit step functions defined in
Sect. 1.2. Finally, application of the Laplace inversion formulas leads to an
explicit expression for the inversion:

1 x2 x2
I11 ðx; y; z; tÞ ¼ 2 2 dðt R=cd Þ þ 2 2 dðt R=cs Þ
4pR cd R cs R
2
ð3:4:37Þ
3x t
þ 1 2 Hðt R=c d ÞHðR=c s tÞ
R R2
I 12 and ^
^
The other inversions for ~ ~
I 13 can be obtained by the simple exchange of the
space variables as shown in Eqs. (3.4.16 and 3.4.17). They are
1 y2 y2
4pR cd R cs R
2
ð3:4:38Þ
3y t
R R2

1 z2 z2
4pR cd R cs R
2
ð3:4:39Þ
3z t
R R2
^
(2) Invesion of ~
I 21
The Fourier inversion integral with respect to the parameter f

Zþ1 !
~
I 21 ðn; g; z; sÞ 1 ng 1 1
¼ expðifzÞdf
2p s2 b2d b2s
1
! ð3:4:40Þ
Z1
ng 1 1
¼ cosðfzÞdf
ps2 f2 þ a2d f2 þ a2s
0
is evaluated by applying the formula (2.1.22).

I 21 ðn; g; z; sÞ ¼ ng 1 expðad jzjÞ 1 expðas jzjÞ
~ ð3:4:41Þ
2s2 ad as
The second inversion integral with respect to the parameter g is given by
Zþ1

I21 1 ng 1 1
ðn; y; z; sÞ ¼ expðad jzjÞ expðas jzjÞ expðigyÞdg
2p 2s2 ad as
1
Z1 ( qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
in g
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jzj g2 þ c2d
2ps2 g2 þ c2d
0
)
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
g
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jzj g2 þ c2s sinðgyÞdg
g2 þ c2s
ð3:4:42Þ
where
cj ¼ n2 þ ðs=cj Þ2 ; j ¼ d; s ð3:4:43Þ
We apply the integration formula (Erdélyi 1954, pp. 75, 36),

x ab
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp c x2 þ a2 sinðbxÞdx ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi K1 a b2 þ c2
x 2 þ a2 b2 þ c2
0
ð3:4:44Þ
to the integral in Eq.(3.4.42). It follows that

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
y

I21 ðn; y; z; sÞ ¼ in n2 þ ðs=cd Þ2 K1 r n2 þ ðs=cd Þ2
2pr s2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3:4:45Þ
þin n2 þ ðs=cs Þ2 K1 r n2 þ ðs=cs Þ2
where r has already been defined by Eq. (3.4.28). The last Fourier inversion
integral with respect to the parameter n is given by
Zþ1
y 1

I21 ðx; y; z; sÞ ¼ in n2 þ ðs=cd Þ2 K1 r n2 þ ðs=cd Þ2
2pr s2 2p
1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
þin n2 þ ðs=cs Þ2 K1 r n2 þ ðs=cs Þ2 expðinxÞdn
Z1
y
¼ 2 2 n n2 þ ðs=cd Þ2 K1 r n2 þ ðs=cd Þ2
2p r s
0
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n n2 þ ðs=cs Þ2 K1 r n2 þ ðs=cs Þ2 sinðnxÞdn
ð3:4:46Þ
We have the formula (Erdélyi 1954, pp. 113, 45)
x x2 þ b2 K1 a x2 þ b2 sinðcxÞdx
0 ð3:4:47Þ

p ab2 c 3 3
¼ 1 þ p ffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi þ exp b a2 þ c2
2 ða2 þ c2 Þ3=2 b a2 þ c2 b2 ða2 þ c2 Þ
The integrals in Eq. (3.4.46) are evaluated and we have

xy 1 3cd 3c2d
I21 ðx; y; z; sÞ ¼ 1 þ þ expfsðR=cd Þg
4pR3 c2d sR s2 R2

ð3:4:48Þ
1 3cs 3c2s
2 1þ þ expfsðR=cs Þg
cs sR s2 R2
The Laplace inversion is carried out by applying the formulas (3.4.34–3.4.36) and
we have the final form for I21 ðx; y; z; tÞ,

xy 1 1 3t
I21 ðx; y; z; tÞ ¼ dðt R=cd Þ 2 dðt R=cs Þ þ 2 Hðt R=cd ÞHðR=cs tÞ
4pR3 c2d cs R
ð3:4:49Þ
Other two inversions are obtained by the exchange of the space variables as
shown in Eqs. (3.4.20) and (3.4.21). They are

yz 1 1 3t
I22 ðx; y; z; tÞ ¼ dðt R=c d Þ dðt R=c s Þ þ Hðt R=c d ÞHðR=c s tÞ
4pR3 c2d c2s R2
ð3:4:50Þ

xz 1 1 3t
I23 ðx; y; z; tÞ ¼ dðt R=cd Þ 2 dðt R=cs Þ þ 2 Hðt R=cd ÞHðR=cs tÞ
4pR3 c2d cs R
ð3:4:51Þ
From inspection of Eq. (3.4.8), the displacement components can be expressed in
terms of Iij ðx; y; z; tÞ as
ux ¼ Px fI0 ðx; y; z; tÞ I11 ðx; y; z; tÞg þ Py fI21 ðx; y; z; tÞg þ Pz fI23 ðx; y; z; tÞg
uy ¼ Px fI21 ðx; y; z; tÞg þ Py fI0 ðx; y; z; tÞ I12 ðx; y; z; tÞg þ Pz fI22 ðx; y; z; tÞg
uz ¼ Px fI23 ðx; y; z; tÞg þ Py fI22 ðx; y; z; tÞg þ Pz fI0 ðx; y; z; tÞ I13 ðx; y; z; tÞg
ð3:4:52Þ
Substituting Eqs. (3.4.37–3.4.39) and (3.4.49–3.4.51) into Eq. (3.4.52), we get the
following explicit expressions for the displacement components:
8 2 9
> x 1 x2 >
>
> 2 dðt R=c Þ þ 1 dðt R=c Þ >
>
Px < cd R2 =
d s
c2s R2
ux ¼
4pR >
> t 3x2 >
>
>
: 2 1 2 Hðt R=cd ÞHðR=cs tÞ > ;
R R

Py xy xy 3xyt
þ dðt R=c d Þ dðt R=c s Þ þ Hðt R=c d ÞHðR=c s tÞ
4pR c2d R2 c2s R2 R4

Pz xz xz 3xzt
þ dðt R=c d Þ dðt R=c s Þ þ Hðt R=c d ÞHðR=cs tÞ
ð3:4:53Þ

Px xy xy 3xyt
uy ¼ dðt R=c d Þ dðt R=c s Þ þ Hðt R=c d ÞHðR=c s tÞ
8 2 2
9
> y 1 y >
>
> 2 2 dðt R=cd Þ þ 2 1 2 dðt R=cs Þ >
>
Py < cd R cs R =
þ
4pR >> t 3y2 >
>
>
: 2 1 2 Hðt R=cd ÞHðR=cs tÞ > ;
R R

Pz yz yz 3yzt
þ dðt R=cd Þ dðt R=c s Þ þ Hðt R=c d ÞHðR=c s tÞ
ð3:4:54Þ

Px xz xz 3xzt
uz ¼ dðt R=c d Þ dðt R=c s Þ þ Hðt R=cd ÞHðR=c s tÞ

Py yz yz 3yzt
þ dðt R=c d Þ dðt R=c s Þ þ Hðt R=c d ÞHðR=c s tÞ
8 9
>
< z2
dðt R=c d Þ þ 1
1 z2
dðt R=c s Þ >
=
Pz c2d R2 c2s R2
þ
4pR >
: Rt2 1 3z
2 >
;
R2 Hðt R=cd ÞHðR=cs tÞ
ð3:4:55Þ
In terms of the Green’s dyadic we can write
X
ui ¼ Pj Gij ðx; y; z; tÞ ; i ¼ x; y; z ð3:4:56Þ
j¼x;y;z
where the dyadic Gij ðx; y; z; tÞ is given by

8 9
> xi xj 1 xi xj >
> dðt R=c Þ þ d dðt R=c Þ >
1 < c2d R2 d
c 2
s
ij
R 2 s =
Gij ðx; y; z; tÞ ¼
4pR >
> 3xi xj >
: Rt2 dij 2 Hðt R=cd ÞHðR=cs tÞ > ;
R
ð3:4:57Þ
It is clear that there are two spherical wave fronts, denoted by cd t ¼ R and
cs t ¼ R for dilatational and shear waves respectively, and that the displacement
has the singularity of the delta function (i.e. a first order singularity) at the wave
fronts.
When the source is a time-harmonic vibration, the body force with magnitude Qi is
given by
0 1 0 1
Bx Qx
@ By A ¼ @ Qy AdðxÞdðyÞdðzÞ expðþixtÞ; ð3:5:1Þ
Bz Qz
and the displacement equation is

o oux ouy ouz o2 ux o2 ux o2 ux
ðc2 1Þ þ þ þ 2 þ 2 þ 2
ox ox oy oz ox oy oz
1 o2 ux Qx
¼ 2 dðxÞdðyÞdðzÞ expðixtÞ
c2s ot2 cs

o oux ouy ouz o2 uy o2 uy o2 uy
ðc2 1Þ þ þ þ 2 þ 2 þ 2
ð3:5:2Þ
1 o2 uy Qy
c2s ot2 cs

2 o oux ouy ouz o2 uz o2 uz o 2 uz
ðc 1Þ þ þ þ 2 þ 2 þ 2
oz ox oy oz ox oy oz
1 o2 uz Q z
c2s ot2 cs
In this section we do not employ the method of the integral transform for solving
the displacement equations. Instead, we take a short-cut. That is, we use the con-
volution integral with the impulsive solution obtained in the previous section. When
we express the displacement in terms of the time-harmonic Green’s dyadic,
X
ui ¼ Qj gij ðx; y; z; tÞ ; i ¼ x; y; z ð3:5:3Þ
j¼x;y;z
the dyadic for the time-harmonic response can be derived by the convolution
integral of the impulsive dyadic given by Eq. (3.4.57). The convolution integral is
evaluated as follows
Zt
gij ðx; y; z; tÞ ¼ lim Gij ðx; y; z; t0 Þexpfþixðt t0 Þgdt0
t!1
0
tZ
!1
ð3:5:4Þ
=expðþixtÞ Gij ðx; y; z; t0 Þexpðixt0 Þdt0
0
Substituting Eq. (3.4.57) into the last integral in the above equation, we can easily
perform the integration and obtain the time-harmonic Green’s dyadic gij as
gij ðx; y; z; tÞ
( )
1 xi xj 1 xR 3xi xj
¼ þ 1þi dij 2 expfixðt R=cd Þg
4pR c2d R2 ðxRÞ2 cd R
( )
1 1 xi xj 1 xR 3xi xj
þ dij 2 1þi dij 2 expfixðt R=cs Þg
4pR c2s R ðxRÞ2 cs R
ð3:5:5Þ
When a static point force is placed at the coordinate origin, the body force with
magnitude Si is given by
0 1 0 1
Bx Sx
@ By A ¼ @ Sy AdðxÞdðyÞdðzÞ ð3:6:1Þ
Bz Sz
The static deformation can be considered as a time-harmonic response with zero-
frequency. Taking the limit x ! 0 in Eq. (3.5.5), the displacement can be
expressed in terms of the static dyadic as
X ðKÞ
ui ¼ Sj Gij ðx; y; zÞ ; i ¼ x; y; z ð3:6:2Þ
j¼x;y;z
where the source magnitude Qi is replaced with Si and the static dyadic
ðKÞ
Gij ðx; y; zÞ is derived from the limit and is given by
ðKÞ 1 n 2 2 xi xj o
Gij ðx; y; zÞ ¼ lim gij ðx; y; z; tÞ ¼ ðc þ 1Þdij þ ðc 1Þ ð3:6:3Þ
x!0 8pc2d R R2
Since the velocity ratio can be replaced with the simple function of Poisson ratio m
as
2ð1 mÞ
c2 ¼ ð3:6:4Þ
1 2m
the dyadic can also be rewritten as
1 n xi x j o
ðKÞ
Gij ðx; y; zÞ ¼ ð3 4mÞd ij þ ð3:6:5Þ
8pð1 2mÞc2d R R2
This dyadic is called ‘‘Kelvin’s solution’’ for the static deformation.
Exercises
(3.1) From the unified expression for the Green’s dyadic (3.1.42), derive the
explicit expressions for Gxx ; Gxy ; Gyx ; Gyy and show that Gxy ¼ Gyx .
(3.2) Show that the explicit expression of the Green’s dyadic (3.4.57) is the same
as the corresponding one in the displacement equations (3.4.53–3.4.55).
Table B Green’s dyadic for elastodynamic equations
Displacement Equations Source Bi Green’s dyadic
P
2D 2D inplane deformation (plane strain) Pi dðxÞdðyÞdðtÞ ui ðx; y; tÞ ¼ Pj Gij ðx; y; tÞ ; i; j ¼ x; y
j¼x;y
o oux ouy o2 ux o2 ux 8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi9
ðc2 1Þ þ þ 2 þ 2 > >
ox ox oy ox oy Hðt r=cd Þ <xi xj 1 2xi xj ðcd tÞ2 r2 =
Gij ðx; y; tÞ ¼ 2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ 2
dij 2
1 o2 ux 1 2pcd > r r >
¼ 2 Bx ðx; y; tÞ :r ðcd tÞ2 r2 ;
c2s ot2 cs 8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi9
> >
Hðt r=cs Þ <xi xj 1 2xi xj ðcs tÞ2 r 2 =
2
dij qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ 2
dij 2
o oux ouy o2 uy o2 uy 2pcs > r r >
ðc2 1Þ þ þ 2 þ 2 : r ðcs tÞ2 r2 ;
oy ox oy ox oy
1 o2 uy 1
¼ 2 By ðx; y; tÞ Qi dðxÞdðyÞ expðþixtÞ P
c2s ot2 cs ui ðx; y; tÞ ¼ Qj gij ðx; yÞ expðþixtÞ ; i; j ¼ x; y
j¼x;y

gij ðx; yÞ ¼ 2 2
H0 ðrx=cd Þ þ dij 2 H1 ðrx=cd Þ
4 c r r rx

d
1 xi xj ð2Þ 2xi xj cs ð2Þ
2 2
dij H0 ðrx=cs Þ þ dij 2 H1 ðrx=cs Þ
cs r r rx
Si dðxÞdðyÞ P ðstaticÞ
ui ¼ Sj gijðx; yÞ ; i; j ¼ x; y
j¼x;y

x 2
1
gðstaticÞ
xx ðx; yÞ ¼ ðc2 þ 1Þ logðrÞ þ ðc2 1Þ
4pc2d r
1 2 xy
gðstaticÞ
xy ðx; yÞ ¼ ðc 1Þ 2
4pc2d r

y 2
ðstaticÞ 1 2 2
gyy ðx; yÞ ¼ ðc þ 1Þ logðrÞ þ ðc 1Þ
4pc2d r
(Continued)
73
Table B (continued)
74
Displacement Equations Source Bi Green’s dyadic

P
3D o oux ouy ouz o2 ux o2 ux o2 ux Pi dðxÞdðyÞdðzÞdðtÞ ui ¼ Pj Gij ðx; y; z; tÞ ; i; j ¼ x; y; z
ðc2 1Þ þ þ þ 2 þ 2 þ 2 j¼x;y;z
ox ox oy oz ox oy oz 8 xx 1 9
> i j xi x j >
1 o2 ux 1 >
< c2 R2 dðt R=cd Þ þ c2 dij R2 dðt R=cs Þ >
=
¼ 2 Bx ðx; y; z; tÞ 1 d
c2s ot2 cs Gij ðx; y; z; tÞ ¼ 4pR s
>
> t 3xi xj >
: dij 2 Hðt R=cd ÞHðR=cs tÞ > ;
R2 R
P
o oux ouy ouz o2 uy o2 uy o2 uy Qi dðxÞdðyÞdðzÞ expðþixtÞ ui ¼ Qj gij ðx; y; z; tÞ ; i; j ¼ x; y; z
ðc2 1Þ þ þ þ 2 þ 2 þ 2 j¼x;y;z
gij ðx; y; z; tÞ
1 o2 uy 1 (
¼ 2 By ðx; y; z; tÞ 1 x i xj 1

xR

3xi xj
)
c2s ot2 cs ¼ þ 1 þ i d
ij expfixðt R=cd Þg
4pR c2d R2 ðxRÞ2 cd R2
( )
1 1 xi xj 1 xR 3xi xj
þ d ij 1 þ i d ij expfixðt R=cs Þg
4pR c2s R2 ðxRÞ2 cs R2

o oux ouy ouz o2 uz o2 uz o2 uz Si dðxÞdðyÞdðzÞ P ðKÞ
ðc2 1Þ þ þ þ 2 þ 2 þ 2 ui ¼ Sj Gij ðx; y; zÞ ; i; j ¼ x; y; z
oz ox oy oz ox oy oz j¼x;y;z
2 ðKÞ 1
xi xj
1 o uz 1 Gij ðx; y; zÞ ¼ 8pð12mÞc2d R
ð3 4mÞdij þ R2
¼ 2 Bz ðx; y; z; tÞ
c2s ot2 cs
3 Green’s Dyadic for an Isotropic Elastic Solid
References 75
References
Erdélyi A (ed) (1954) Tables of integral transforms, Vol I. McGraw-Hill, New-York

Cambridge
Gradshteyn IS, Ryzhik IM (1980) Jefferey A (ed) Table of integrals, series, and products, 5th edn.
Academic Press, San Diego
Chapter 4
Acoustic Wave in a Uniform Flow
Traditionally, when we use the word ‘‘wave,’’ it means acoustic waves or water
waves. The acoustic wave is much more familiar to our daily life and we expe-
rience many wave phenomena such as reflection, refraction, diffraction, the
Doppler effects, etc. The governing equations for the acoustic wave are rigorously
derived from the fluid equations and Green’s function for the acoustic wave in a
flowing fluid is discussed by applying the method of integral transform.
4.1 Compressive Viscous Fluid
Motions and disturbances in fluids such as water, oil and gas are governed by four
groups of equations:
(1) Equations of motion,

orxx oryx orzx ovx ovx ovx ovx
þ þ þ qBx ¼ q þ vx þ vy þ vz
ox oy oz ot ox oy oz

orxy oryy orzy ovy ovy ovy ovy
þ þ þ qBy ¼ q þ vx þ vy þ vz ð4:1:1Þ

orxz oryz orzz ovz ovz ovz ovz
þ þ þ qBz ¼ q þ vx þ vy þ vz

78 4 Acoustic Wave in a Uniform Flow
(2) Constitutive equations for the linear Newtonian fluid,

2l omx omy omz omx
rxx ¼ p þ þ þ 2l
3 ox oy oz ox

2l omx omy omz omy
ryy ¼ p þ þ þ 2l
3 ox oy oz oy
ð4:1:2Þ
2l omx omy omz omz
rzz ¼ p þ þ þ 2l
3 ox oy oz oz

omx omy omy omz omx omz
rxy ¼l þ ; ryz ¼ l þ ; rzx ¼ l þ
oy ox oz oy oz ox
(3) Continuity equation

oq oðqvx Þ oðqvy Þ oðqvz Þ
þ þ þ ¼0 ð4:1:3Þ
ot ox oy oz
(4) Equation of state for the acoustic medium (adiabatic change)

j
p q
¼ ð4:1:4Þ
p0 q0
In these equations, the viscosity l is a known/given constant that specifies the
nature of the fluid. The body force Bi is assumed to be a source and its functional
form must be specified. The stress rij , the particle velocity vi , the hydro-static
pressure p, and the density q are unknown functions to be determined. The sub-
script ‘‘0’’ stands for the quantities at a reference state. The constant parameter j is
the ratio of two specific heats. Thus, the governing equations for a linear fluid are a
set of strongly coupled partial differential equations with 11 unknowns.
Substituting the constitutive equations (4.1.2) into the equations of motion
(4.1.1), we obtain the well-known Navier-Stokes equations,
2
op l o ovx ovy ovz o v x o2 v x o2 v x
þ þ þ þl þ 2 þ 2 þ qBx
ox 3 ox ox oy oz ox2 oy oz
ð4:1:5aÞ
ovx ovx ovx ovx
¼ q vx þ vy þ vz þ
ox oy oz ot
2
op l o ovx ovy ovz o vy o2 vy o2 vy
þ þ þ þl þ 2 þ 2 þ qBy
oy 3 oy ox oy oz ox2 oy oz
ð4:1:5bÞ
ovy ovy ovy ovy
ox oy oz ot
4.1 Compressive Viscous Fluid 79
2
op l o ovx ovy ovz o v z o2 v z o2 v z
þ þ þ þl þ þ þ qBz
oz 3 oz ox oy oz ox2 oy2 oz2
ð4:1:5cÞ
ovz ovz ovz ovz
ox oy oz ot
When the fluid is incompressible and no density change takes place, the con-
tinuity equation is simplified and the equation of state is unnecessary. Then, the
incompressible fluid is governed by a simplified form of the Navier-Stokes
equations. However, when we discuss acoustic waves, the two reminder equa-
tions(4.1.3) and (4.1.4) must be fully incorporated since acoustic waves express
the propagation of density changes.
4.2 Linearization
The nonlinearity stems from the acceleration terms in Navier-Stokes equations

(4.1.5), the product ðqvi Þ in the continuity equations (4.1.3) and the power of
density in the equation of state (4.1.4). Thus, the governing equations for acoustic
waves are fully nonlinear, coupled partial differential equations. In order to reduce
the differential equations to tractable ones, we introduce the assumption that the
acoustic waves are infinitesimal small disturbances superimposed on a reference
nonlinear fluid motion. Based on this assumption, we assume that the infinitesimal
small disturbance is a deviation from the nonlinear reference state. We introduce a
small parameter e, and denote the disturbance by ðvi ; p; qÞ. The five unknown
functions, the velocity component vi , the static hydro-static pressure p, and the
density q, are approximated as the power of the small parameter e,
vi ¼ Vi þ evi þ Oðe2 Þ; p þ Oðe2 Þ;

p ¼ p0 þ e q þ Oðe2 Þ
q ¼ q0 þ e ð4:2:1Þ
We also assume that the wave source described by the body force is a small
quantity, of order e1 ,
i þ Oðe2 Þ
Bi ¼ eB ð4:2:2Þ
Substituting Eqs. (4.2.1) and (4.2.2) into Eq. (4.1.5a), and neglecting terms of
higher order than Oðe1 Þ, the equation is reduced to two parts, the zeroth order and
first order terms. The zeroth order terms are
2
op0 l o oVx oVy oVz o Vx o 2 V x o 2 V x
þ þ þ þl þ þ
ox 3 ox ox oy oz ox2 oy2 oz2
ð4:2:3Þ
oVx oVx oVx oVx
¼ þq0 þ Vx þ Vy þ Vz
ot ox oy oz
and the first order terms are
2
o
p l o ovx ovy ovz o vx o2vx o2vx x
e þ þ þ þl þ þ þ qB

oVx oVx oVx
¼ þq Vx þ Vy þ Vz
ox oy oz

ovx ovx oVx ovx oVx ovx oVx
þq0 þ q0 Vx þ vx þ q 0 Vy þ vy þ q0 Vz þ vz
ot ox ox oy oy oz oz
ð4:2:4Þ
Similarly, we substitute Eqs. (4.2.1) and (4.2.2) into the reminder of Eq. (4.1.5).
From Eq. (4.1.5b), we have the zeroth order terms
2
op0 l o oVx oVy oVz o Vy o 2 V y o 2 V y
þ þ þ þl þ 2 þ 2
ð4:2:5Þ
oVy oVy oVy oVy
ot ox oy oz
and the first order terms
2
op l o ovx ovy ovz o vy o2vy o2vy y
e þ þ þ þl þ þ þ qB
oy 3 oy ox oy oz ox2 oy2 oz2

oVy oVy oVy
ox oy oz

ovy ovy oVy ovy oVy ovy oVy
þq0 þ q0 Vx þ vx þ q0 Vy þ vy þ q0 Vz þ vz
ð4:2:6Þ
From Eq. (4.1.5c), we also have for the zeroth order terms
2
op0 l o oVx oVy oVz o V z o 2 Vz o 2 V z
þ þ þ þl þ þ
ð4:2:7Þ
oVz oVz oVz oVz
ot ox oy oz
and for the first order terms
2
op l o ovx ovy ovz o vz o2vz o2vz z
e þ þ þ þl þ 2 þ 2 þ qB
oz 3 oz ox oy oz ox2 oy oz

oVz oVz oVz
ox oy oz

ovz ovz oVz ovz oVz ovz oVz
þq0 þ q0 Vx þ vx þ q 0 Vy þ vy þ q0 Vz þ vz
ð4:2:8Þ
4.2 Linearization 81
Substituting Eq. (4.2.1) into Eq. (4.1.3), the continuity equation is decomposed
into two parts,
oðq0 þ e
qÞ oðq0 þ eqÞðVx þ evx Þ oðq0 þ e
qÞðVy þ evy Þ oðq0 þ e
qÞðVz þ evz Þ
þ þ þ
ot ox oy oz
oq0 oðq0 Vx Þ oðq0 Vy Þ oðq0 Vz Þ
¼ þ þ þ
ot ox oy oz

oq oðq0vx Þ oðq0vy Þ oðq0vz Þ oðVx q Þ oðVy qÞ oðVz qÞ
þe þ þ þ þ þ þ
ot ox oy oz ox oy oz
þ Oðe2 Þ ¼ 0
ð4:2:9Þ
The equation of state for the adiabatic change in Eq. (4.1.4) is approximated as

p0 þ e
p q0 þ eq j
p j
q p j
q
¼ ) 1 þ e ¼ 1 þ e þ Oðe2 Þ ) ¼
p0 q0 p0 q0 p 0 q0
ð4:2:10Þ
Thus, we have a linear relation between the pressure and density deviations.
Consequently, we have the equations for the reference state as the Oðe0 Þ part,
2
op0 l o oVx oVy oVz o Vx o2 Vx o2 Vx
ox 3 ox ox oy oz ox2 oy oz
ð4:2:11aÞ
oVx oVx oVx oVx
ot ox oy oz
2
op0 l o oVx oVy oVz o V y o 2 Vy o 2 V y
ð4:2:11bÞ
oVy oVy oVy oVy
ot ox oy oz
2
op0 l o oVx oVy oVz o Vz o2 Vz o2 Vz
oz 3 oz ox oy oz ox2 oy oz
ð4:2:11cÞ
oVz oVz oVz oVz
ot ox oy oz
oq0 oðq0 Vx Þ oðq0 Vy Þ oðq0 Vz Þ

þ þ þ ¼0 ð4:2:12Þ
ot ox oy oz
The disturbance of Oðe1 Þ is governed by the equations
2
op l o ovx ovy ovz o vx o2vx o2vx x
þ þ þ þl þ þ þ q0 B

oVx oVx oVx
ox oy oz

ovx ovx oVx ovx oVx ovx oVx
þ q0 þ q0 Vx þ vx þ q0 Vy þ vy þ q0 Vz þ vz
ð4:2:13aÞ
2
op l o ovx ovy ovz o vy o2vy o2vy y
þ þ þ þl þ 2 þ 2 þ q0 B

oVy oVy oVy
¼ þ
q Vx þ Vy þ Vz
ox oy oz

ovy ovy oVy ovy oVy ovy oVy
ð4:2:13bÞ
2

oVz oVz oVz
¼ þ
q Vx þ Vy þ Vz
ox oy oz

ovz ovz oVz ovz oVz ovz oVz
ð4:2:13cÞ
q oðq0vx Þ oðq0vy Þ oðq0vz Þ oðVx q

o Þ oðVy q
Þ oðVz q
Þ
þ þ þ þ þ þ ¼0 ð4:2:14Þ
q0
¼
q
p ð4:2:15Þ
jp0
We should understand that the zeroth order equations, which govern the reference
fluid flow, are already satisfied by the reference quantities ðVi ; p0 ; q0 Þ. On the other
hand, the first order equations, which include the change/deviation from the ref-
erence state, are the coupled differential equations for the 5 unknowns ðvi ; p; q Þ.
4.3 Viscous Acoustic Fluid
When the reference state is a uniform flow with viscosity l, we can assume
Vi ; p0 ; q0 ¼ const:, and then the first order equations (4.2.13–14) are reduced to the
simpler forms
4.3 Viscous Acoustic Fluid 83
2
o
p l o ovx ovy ovz o vx o2vx o2vx x
ð4:3:1aÞ
ovx ovx ovx ovx
¼ þq0 þ q0 Vx þ Vy þ Vz
ot ox oy oz
2
o
p l o ovx ovy ovz o vy o2vy o2vy y
oy 3 oy ox oy oz ox2 oy2 oz2
ð4:3:1bÞ
ovy ovy ovy ovy
¼ þq0 þ q0 Vx þ Vy þ Vz
ot ox oy oz
2
ð4:3:1cÞ
ovz ovz ovz ovz
¼ þq0 þ q0 V x þ Vy þ Vz
ot ox oy oz
With use of the linear relation between the pressure and density deviations, the
continuity equation (4.2.14) can be rewritten in terms of the pressure and velocity
gradients as

o
p op o
p o
p ovx ovy ovz
þ Vx þ Vy þ Vz þ jp0 þ þ ¼0 ð4:3:2Þ
The above four Eqs. (4.3.1) and (4.3.2) are the governing equations for acoustic
waves in the uniformly flowing viscous fluid. The coupled differential equations,
which have four unknowns ðvi ; pÞ, can be reduced to a single differential equation
by the introduction of a velocity potential. We assume that the velocity compo-
nents ðvi Þ can be derived from a single velocity potential /ðx; y; z; tÞ as
o/ o/ o/
vx ¼ ; vy ¼ ; vz ¼ ð4:3:3Þ
ox oy oz
As to the body force, we also assume that there exists a body force potential
Bðx; y; z; tÞ and each body force component is derived as,
x ¼ oB ;
B y ¼ oB ;
B z ¼ oB
B ð4:3:4Þ
ox oy oz
Substituting Eqs. (4.3.3) and (4.3.4) into (4.3.1a), we obtain

o
p o o/ o/ o/ o/ 4l o o2 / o2 / o2 / oB
þ q0 þ Vx þ Vy þ Vz 2
þ 2
þ 2
q0
ox ox ot ox oy oz 3 ox ox oy oz ox
¼0
ð4:3:5Þ
Since all terms in the above equation are derivative with respect to the space
variable x, we integrate it with respect to the variable x,

o/ o/ o/ o/ 4l o2 / o2 / o2 /
p ¼ q0
þ Vx þ Vy þ Vz þ þ þ þ q0 B
ot ox oy oz 3 ox2 oy2 oz2
ð4:3:6Þ
where the constant term is neglected since it is one of the reference quantities.
Substitution into Eqs. (4.3.1b and 4.3.1c) leads to the same equation and thus the
equations of motion in all three directions are reduced to a single Eq. (4.3.6). The
velocity potential defined by Eq. (4.3.3) is also substituted into the continuity
equations (4.3.2),
2
o
p o
p o
p o
p o / o2 / o2 /
þ Vx þ Vy þ Vz þ jp0 þ þ 2 ¼0 ð4:3:7Þ
ot ox oy oz ox2 oy2 oz
Then, we have the coupled differential equations (4.3.6) and (4.3.7) in terms of
only two unknowns, the velocity potential and the pressure deviation. Further, if
we substitute Eq. (4.3.6) into Eq. (4.3.7), we have just a single differential equa-
tion for the velocity potential /,

2 4l o o o o
r /þ þ V x þ Vy þ V z r2 /
3jp0 ot ox oy oz

q0 o o o o 2 q0 oB oB oB oB
¼ þ V x þ V y þ Vz / þ Vx þ Vy þ Vz
jp0 ot ox oy oz jp0 ot ox oy oz
ð4:3:8Þ
where r2 is the Laplacian operator defined by
o2 o2 o2
r2 ¼ þ þ ð4:3:9Þ
ox2 oy2 oz2
We can also derive the differential equation for the pressure deviation. We
apply the Laplacian to both sides of Eq. (4.3.6),

o o o o 4l
r2
p ¼ q0 þ V x þ Vy þ V z r2 / þ r2 r2 / þ q0 r2 B
ot ox oy oz 3
ð4:3:10Þ
and derive the Laplacian of the velocity potential from Eq. (4.3.7) as

2 1 o p o
p op op
r /¼ þ Vx þ V y þ V z ð4:3:11Þ
jp0 ot ox oy oz
The Laplacian of the velocity potential in Eq. (4.3.10) is replaced with the above
pressure gradient. We therefore have a single differential equation for the pressure
deviation as
4.3 Viscous Acoustic Fluid 85

4l o o o o
r2
pþ þ V x þ Vy þ V z r2
p
3jp0 ot ox oy oz
2 ð4:3:12Þ
q o o o o 2 o B o2 B o2 B
¼ 0 þ V x þ Vy þ V z p þ q0
þ þ
jp0 ot ox oy oz ox2 oy2 oz2
Two differential equations (4.3.8) and (4.3.12) for the potential and the pressure
are mathematically the same, but the non-homogeneous terms of the body force
are little bit different. We easily see that both equations are wave equations and
that their wave velocity is
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c ¼ ðjp0 Þ=q0 ð4:3:13Þ
If we introduce a reference length l and define the Reynolds number,
Re ¼ ðq0 clÞ=l ð4:3:14Þ
and the Mach numbers,
Mj ¼ Vj =c; j ¼ x; y; z ð4:3:15Þ
the governing equation for the viscous acoustic media is simplified to

4l 1 o o o o
r2 / þ þ Mx þ My þ Mz r2 /
3Re c ot ox oy oz

1o o o o 2 1 1 oB oB oB oB
¼ þ Mx þ My þ Mz / þ Mx þ My þ Mz
c ot ox oy oz c c ot ox oy oz
ð4:3:16Þ
for the velocity potential, and to

4l 1 o o o o
r2
pþ þ Mx þ My þ Mz r2 p
3Re c ot ox oy oz
2 ð4:3:17Þ
1o o o o 2 o B o2 B o2 B
¼ þ Mx þ My þ Mz p þ q0
þ þ
c ot ox oy oz ox2 oy2 oz2
for the pressure deviation. These two differential equations have only two
parameters, the Reynolds number and the Mach number. Thus, the nature of
acoustic waves in a flowing viscous fluid is characterized by these two numbers.
4.4 Wave Radiation in a Uniform Flow
This section discusses a 2D wave propagation problem in a uniform flow. Take the
ðx; yÞ plane and assume that all quantities are independent of z. Assume that the
non-viscous acoustic fluid (l ¼ 0) is flowing along the x-axis with the uniform
Fig. 4.1 A point source in a y

uniformly flowing fluid
δ (t )
Qδ ( x )δ ( y ) iω t
e
Vx
x
velocity Vx (Fig. 4.1). We assume here the Laplacian of the body force potential as
the wave source Q,
2
o B o2 B
q0 þ ¼ QdðxÞdðyÞdðtÞ ð4:4:1Þ
ox2 oy2
Then, we substitute Re ¼ 1; My ¼ Mz ¼ 0 into Eq. (4.3.17). The governing
equation for the pressure deviation is then reduced to the simple form

1o o 2
r2 p ¼ þ Mx p þ QdðxÞdðyÞdðtÞ ð4:4:2Þ
c ot ox
where c is the acoustic wave velocity defined by Eq. (4.3.13). The Mach number
Mx is defined by Eq. (4.3.15). Furthermore, the pressure in this equation is the
deviation from the reference state and the correct notation is p(over bar). However,
we do not use the over bar for the pressure deviation since we will apply the double
Fourier transform and one of the transforms is denoted by the over bar. So, in order
to avoid confusion the over bar is dropped from the pressure deviation.
Since our acoustic field is of infinite extent, it is enough to obtain a particular
solution of Eq. (4.4.2). Laplace transform with respect to the time t and double
Fourier transform with respect to the two space variables x and y are applied. The
triple integral transform is defined as
Laplace transform:
Z1

0
Fourier transform:
Z1 Z1
2p
0 0
Fourier transform:
Z1 Z1
f ðyÞ expðþigyÞdy; f ðyÞ ¼ ð4:4:5Þ
2p
0 0
4.4 Wave Radiation in a Uniform Flow 87
Applying the triple transform to Eq. (4.4.2), a simple algebraic equation for the
transformed pressure is obtained as
p ¼ ðs=c inMx Þ2 ~p þ Q

ðn2 þ g2 Þ~
ð4:4:6Þ
The pressure in the transformed domain is given by
Q Q
p ¼
~
2 2
¼ ð4:4:7Þ
n þ g2 þ ðs=c inMx Þ g2 þ q 2
where q is written as
q2 ¼ n2 þ ðs=c inMx Þ2
( 2 2 )
2 iMx s=c
¼ ð1 Mx Þ n ðs=cÞ þ ð4:4:8Þ
1 Mx2 1 Mx2
n o
¼ ð1 Mx2 Þ ðn iaMx Þ2 þa2
The Laplace transform parameter s is included in the parameter a which is defined

by
s=c
a¼ ð4:4:9Þ
1 Mx2
As to the first inversion, we apply the Fourier inversion with respect to the
parameter g, and reduce it to the semi-infinite integral,
Zþ1 Z1
1 Q Q 1
p ¼
2 2
expðigyÞdg ¼ cosðgyÞdg ð4:4:10Þ
2p g þq p g2 þ q2
1 0
The last integral is evaluated easily by the formula (2.1.22) and yields
Q
p ¼
expðqjyjÞ ð4:4:11Þ
2q
The next step is to apply the Fourier inversion integral with respect to the
parameter n,
Zþ1
Q 1 1
p ¼ expðqjyjÞ expðinxÞdn ð4:4:12Þ
2 2p q
1
Recalling q in Eq. (4.4.8), the integral in the above equation is rewritten in the
explicit form,

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiqffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Zþ1 exp jyj 1 Mx2 ðn iaMx Þ2 þ a2 inx
Q 1
p ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dn
2 1 Mx2 2p ðn iaM Þ 2
þ a 2
1 x
ð4:4:13Þ
Introducing the variable transform, n ! 1, as
1 ¼ n iaMx ð4:4:14Þ
we obtain a simpler integrand, but the integration path is shifted into the complex
plane, not on the real axis, i. e.
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z
þ1iaM x
exp jyj 1 M 2 1 2 þ a2 i1x
Q expðþaMx xÞ 1 x
p ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d1
2 1 Mx 2 2p 1 þ a2
2
1iaMx
ð4:4:15Þ
The integration path for this integral is slightly shifted from the real axis in the
complex 1-plane and is shown by the line CD in Fig. 4.2. In order to transform the
integral to that along the real axis, we consider the complex integral U with the
!
closed loop LðCDBAC Þ in the figure,
Z exp jyj 1 M 2 12 þ a2 i1x

U¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d1 ð4:4:16Þ
2 1 Mx 2 2p 1 þ a2
2
L
The integrand has two branch points, at 1 ¼ ia. Two branch cuts are introduced
along the imaginary axis, as shown in the figure. Fortunately, if we assume sub-
sonic motion of the flow, Mx ð¼ Vx =cÞ\1, these branch cuts do not cross the
integration line CD and no singular point are then included in the closed loop L. In
addition, the two integrals
along the line AC and BD, with an infinite real part,
p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
vanish since Re 12 þ a2 [ 0. Then, applying Cauchy’s integral theorem to the
complex integral U in Eq. (4.4.16), the integral along the complex path CD is
converted to one along the real axis AB in the 1-plane. That is
Fig. 4.2 Transform of Im(ς )

integration path
+iα
Branch cut
A B Re(ς )
C
−iα M x D
−iα
4.4 Wave Radiation in a Uniform Flow 89
Zþ1 exp jyj 1 M 2 12 þ a2 i1x

p ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d1 ð4:4:17Þ
2 1 Mx2 2p 1 2 þ a2
1
The above integral can be further reduced to the semi-infinite integral,

Z1 exp jyj 1 M 2 12 þ a2
p ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cosð1xÞd1 ð4:4:18Þ
2 1 Mx2 p 1 2 þ a2
0
Applying the integration identity, which is the integral representation of the

modified Bessel function (Erdélyi 1954 pp. 17, 27 and Watson 1966 pp. 172),
Z1 exp y 12 þ a2
K0 a x2 þ y2 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cosð1xÞd1
12 þ a2
0 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z1 exp au x2 þ y2
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi du ð4:4:19Þ
u2 1
1
to Eq. (4.4.18), we have

Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Q expðþaMx xÞ 1
p ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi exp au x2 þ ð1 Mx2 Þy2 du ð4:4:20Þ
2p 1 Mx2 u2 1
1
Further, we recall the definition of a which includes the Laplace transform

parameter s. The above Eq. (4.4.20) can be rewritten as
n pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
Z1 exp s=c 2 u x2 þ ð1 M 2 Þy2 Mx x
Q 1 1M x
p ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi x
pffiffiffiffiffiffiffiffiffiffiffiffiffi du ð4:4:21Þ
2p 1 Mx2 u2 1
1
If we introduce the variable transform, u ! t, as defined by

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u x2 þ ð1 Mx2 Þy2 Mx x ð1 Mx2 Þct þ Mx x
t¼ 2
; u ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi;
cð1 Mx Þ x2 þ ð1 Mx2 Þy2
ð4:4:22Þ
cð1 Mx2 Þ
du ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi dt
x2 þ ð1 Mx2 Þy2
Equation (4.4.21) is converted to the form of the Laplace transform integral,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Z1
cQ expðstÞdt
p ¼ 1 Mx2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2p pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi fð1 Mx2 Þct þ Mx xg2 fx2 þ ð1 Mx2 Þy2 g
x2 þð1Mx Þy2 Mx x
cð1Mx2 Þ
ð4:4:23Þ
The above equation is the Laplace transformed pressure, but it is just in the form of
the Laplace transform integral. Thus, the original pressure is its integrand with a
shifted starting time since the lower limit is not zero. From inspection of
Eq. (4.4.23), we can find the original pressure as
cQ2p 1 Mx2
p ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð4:4:24Þ
fð1 Mx2 Þct þ Mx xg2 fx2 þ ð1 Mx2 Þy2 g
where its valid time range is

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
x2 þ ð1 Mx2 Þy2 Mx x
\t\1 ð4:4:25Þ
cð1 Mx2 Þ
In order to obtain a more compact expression, the argument in the radical and the
time range are rewritten as
x2 þ ð1 Mx2 Þy2 \ctð1 Mx2 Þ þ Mx x ) ðx Vx tÞ2 þ y2 \ðctÞ2
ð4:4:26Þ
fð1 Mx2 Þct þnMx xg2 fx2 þ ð1 Mx2o

Þy2 g
¼ ð1 Mx2 Þ ðctÞ2 ðx Vx tÞ2 y2 ð4:4:27Þ
Finally, the pressure deviation in the flowing fluid is expressed in the simple form,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Q 1
p ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi H ct ðx Vx tÞ2 þ y2 ð4:4:28Þ
2p 2
t fðx Vx tÞ2 þ y2 g=c2
We have just arrived at the final form of the solution that gives the pressure
fluctuations in a flowing fluid with a stationary impulsive point source. The dis-
turbed region is easily derived from the argument of the step function: it is a circle
but its center is moving with the flow velocity,
ðx Vx tÞ2 þ y2 ¼ ðctÞ2 ð4:4:29Þ

Figure 4.3 shows the typical wave front and the circular disturbed region in the
flowing fluid.
4.5 Time-Harmonic Wave in a Uniform Flow 91
Fig. 4.3 Disturbed circular y

region in a uniform flow
Vx t
Flow Vx ct
x
source
4.5 Time-Harmonic Wave in a Uniform Flow
The steady-state acoustic response produced by a stationary time-harmonic source

is governed by

2 1o o 2
r p þ Mx p ¼ QdðxÞdðyÞ expðþixtÞ ð4:5:1Þ
c ot ox
A solution to this equation can be obtained by the convolution integral of the
impulsive response as
Zt
p ¼ lim pðimpulseÞ ðx; y; t0 Þ expfixðt t0 Þgdt0 ð4:5:2Þ
t!1
0
Substituting the impulsive response of Eq. (4.4.24), not the final one, we have
cQ 1 Mx2 expðixt0 Þdt0
p¼ expðixtÞ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2p pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi fð1 Mx2 Þct0 þ Mx xg2 fx2 þ ð1 Mx2 Þy2 g
x2 þð1Mx2 Þy2 Mx x
cð1Mx2 Þ
ð4:5:3Þ
For this integral, we introduce the variable transform, t0 ! u, as
u Mx x du
u ¼ ð1 Mx2 Þct0 þ Mx x; t0 ¼ ; dt0 ¼ ð4:5:4Þ
ð1 Mx2 Þc ð1 Mx2 Þc
Equation (4.5.3) is now rewritten as
n o
Z1 exp ixu
Q 1 Mx x ð1Mx Þc 2
p ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp ix t þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi du
2p 1 Mx2 1 Mx2 c pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 2 fx2 þ ð1 M 2 Þy2 g
x
x2 þð1Mx2 Þy2
ð4:5:5Þ
Since the integral in the above equation is just the definition integral of the Hankel
function of the second kind (which is Eq. (3.2.5) in Sect. 3.2), we replace the
integral with the Hankel function. The final form for the time-harmonic response is
then given by
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!
iQ ð2Þ x x2 þ ð1 Mx2 Þy2 Mx x
p ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi H0 exp ix t þ ð4:5:6Þ
4 1 Mx2 c 1 Mx2 1 Mx2 c
Exercises
(4.1) When an wave source is moving with uniform velocity V along the x-axis,
how do you change the nonhomogeneous wave source term in Eq. (4.4.1)?
(4.2) With aid of the asymptotic formula for the Hankel function (Watson 1966
p. 198),
rffiffiffiffiffi
ð2Þ 2
Hm ðzÞ expfiðz mp=2 p=4Þg ðaÞ
pz
show that the time-harmonic response (4.5.6) is the out-going wave.
References
Cambridge
Chapter 5
Green’s Functions for Beam and Plate
This chapter presents the dynamic Green’s functions for the elastic beam and plate.
The Green’s function is the deflection response produced by a point load. The
deflection equation with the nonhomogeneous term of the applied load is discussed
by applying the method of integral transform.
5.1 An Impulsive Load on a Beam
We shall obtain Green’s function for an elastic beam. The deflection of the beam,
wðx; tÞ, is governed by the well-known partial differential equation, so called
beam/deflection equation,
o4 w o2 w
EI þ qA ¼ pðx; tÞ ð5:1:1Þ
ox4 ot2
where EI is bending rigidity, qA mass per unit length and pðx; tÞ the load on the
beam. An impulsive point load with magnitude P is assumed by
pðx; tÞ ¼ PdðxÞdðtÞ ð5:1:2Þ
where dð:Þ is Dirac’s delta function. The nonhomogeneous deflection equation to
be solved is rewritten as
o 4 w o2 w
a4 þ 2 ¼ QdðxÞdðtÞ ð5:1:3Þ
ox4 ot
where
EI P
a4 ¼ ; Q¼ ð5:1:4Þ
qA qA
For solving the deflection equation, we employ the quiescent condition at an initial
time,

94 5 Green’s Functions for Beam and Plate

ow
wjt¼0 ¼ ¼ 0 ð5:1:5Þ
ox t¼0

ow o2 w o3 w
wjjxj!1 ¼ ¼ ¼ ¼0 ð5:1:6Þ
ox jxj!1 ox2 jxj!1 ox3 jxj!1
The higher derivatives in the above equation mean the vanishing of the moment
and shear force at infinity.
In order to solve the nonhomogeneous deflection equation, we employ the
method of integral transform. Since the deflection depends on two variables, we
apply the following double transform: Laplace transform with respect to the time
variable,
Z1
0
and Fourier transform with respect to the space variable,

Zþ1 Zþ1
2p
1 1
Applying the double transform to the deflection equation (5.1.3), we have the
algebraic equation for the transformed deflection and then the deflection in the
transformed domain as
Q
¼
w ð5:1:9Þ
ðanÞ4 þ s2
The Laplace inversion is applied firstly. The Laplace inversion formula (Erdélyi
1954, pp. 150, 1),

1 1 1
L ¼ sinðatÞ ð5:1:10Þ
s2 þ a2 a
is applied to Eq. (5.1.9) and we have for the Fourier transformed deflection,
Q
¼
w sinða2 tn2 Þ ð5:1:11Þ
ðanÞ2
The Fourier inversion integral of the above equation is reduced to a semi-infinite
integral, as
5.1 An Impulsive Load on a Beam 95
Zþ1 Z1
1 Q 2 Q 1
w¼ 2
2
sinða tn Þ expðinxÞdn ¼ 2 2
sinða2 tn2 Þ cosðxnÞdn
2p ðanÞ pa n
1 0
ð5:1:12Þ
Fortunately, we have a convenient formula for the inversion integral. That is the
formula (Erdélyi 1954, pp. 23, 3)
Z1 2
1 2 p y y pffiffiffiffiffiffi y p
sinðax Þ cosðxyÞdx ¼ y S p ffiffiffiffiffiffiffi
ffi C p ffiffiffiffiffiffiffiffi þ pa sin þ
x2 2 2pa 2pa 4a 4
0
ð5:1:13Þ
where CðxÞ and SðxÞ are Fresnel integrals, defined by
Zx
CðxÞ 1 1 sinðuÞ
¼ pffiffiffiffiffiffi pffiffiffi du ð5:1:14Þ
SðxÞ 2p u cosðuÞ
0
Applying this formula to the last integral in Eq. (5.1.12) and rewriting the
expression, we obtain the Green’s function for the dynamic deflection of the beam
as
rffiffiffi np o
Q t pz
wðx; tÞ ¼ pffiffiffi fSðzÞ CðzÞg þ sin ð2z2 þ 1Þ ð5:1:15Þ
a p 2 4
where the dimensionless variable z is defined by
x
z ¼ pffiffiffiffiffiffiffi ð5:1:16Þ
a 2pt
Note that the Green’s function given by Eq. (5.1.15) is an impulsive response;
however, it does not show any wave nature since the deflection wave is dispersive
and its wave velocity depends on the frequency. The impulsive source includes an
infinite frequency, i.e. the initial disturbance spreads all over the beam at once
without showing any wave nature.
5.2 A Moving Time-Harmonic Load on a Beam
When a time-harmonic load with frequency x is moving with the uniform velocity
V, the deflection equation for the beam is given by,
o4 w o2 w
EI þ qA ¼ Pdðx VtÞ expðþixtÞ ð5:2:1Þ
ox4 ot2
where the load location is expressed by the delta function and is at x ¼ Vt. The
convergence condition given by Eq. (5.1.6) is also assumed.
Since the time variable t is included not only in the argument of the exponential
function, but also in that of delta function, it is not good to assume a simple time-
harmonic vibration such as wðx; tÞ ¼ w# ðxÞ expðþixtÞ. We directly apply the
Fourier transform defined by Eq. (5.1.8) to the deflection Eq. (5.2.1),
d2 w

þ ðanÞ4 w
¼ Q expfþiðVn þ xÞtg ð5:2:2Þ
dt2
where a and Q are defined by Eq. (5.1.4). The particular solution corresponding to
the nonhomogeneous loading term is easily obtained as
Q
¼
w expfþiðnV þ xÞtg ð5:2:3Þ
ðanÞ ðnV þ xÞ2
4
Factorizing the denominator, we get
ðanÞ4 ðnV þ xÞ2 ¼ a4 ðn nþ þ

1 Þðn n1 Þðn n2 Þðn n2 Þ ð5:2:4Þ
where the eigenvalues are given by
8 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 9 8 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 9
< 2 = < 2 =
1 V V 1 V V
n
1 ¼ xþ þ ; n
2 ¼ i x ð5:2:5Þ
a: 2a 2a; a: 2a 2a;
the Fourier inversion integral with respect to the parameter n is expressed as

Zþ1
Q expfinðx VtÞg
w¼ expðþixtÞ dn ð5:2:6Þ
2pa4 ðn nþ
1 Þðn n þ
1 Þðn n2 Þðn n2 Þ
1
In order to evaluate the integral in Eq. (5.2.6), we apply the theory of complex
integrals (the residue theorem). Following the discussion in Sect. 2.2, the complex
frequency with a small negative imaginary part is assumed. Due to this assump-
tion, all singular points do not lie on the real axis. Two singular points, n þ
1 and n2 ,
þ
are in the upper complex n-plane and the other two, n1 and n2 , are in the lower
plane. Then, the complex integral
Z
1 expfinðx VtÞg
U¼ dn ð5:2:7Þ
2p ðn n1 Þðn n
þ þ
C
is to be discussed. Two closed loops C are shown in Fig. 5.1. We employ the
circuit C þ in the case of x Vt\0 and C in that of x Vt [ 0 in order to
guarantee the convergence on the large semi-circle. Jordan’s lemma is applied to
the complex integral U and after somewhat lengthy calculation we arrive at the
expressions,
5.2 A Moving Time-Harmonic Load on a Beam 97
Im(ξ )
Im(ξ )
−(c / 2α ) + i ω − (c / 2α ) 2
ξ 2+ =
+ ω + (c / 2α ) 2 + (c / 2α ) α
ξ =+
α
1
C+
Re(ξ )
C−
Re(ξ )
− ω + (c / 2α ) + (c / 2α )
2
ξ1− =
−(c / 2α ) − i ω − (c / 2α ) 2 α
ξ 2− =
α
(a) x − ct > 0 (b) x − ct < 0
Fig. 5.1 Closed loops C for the complex integral U
Zþ1
1 expfinðx VtÞg
dn
2p ðn nþ n
1 Þðn
þ
1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
exp ði=aÞ x þ ðV=2aÞ þ ðV=2aÞ ðx VtÞ
ia3
¼ 2
4 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi
x þ ðV=2aÞ x þ ðV=2aÞ þ ðV=2aÞ
2
exp ð1=aÞ x ðV=2aÞ þ iðV=2aÞ ðx VtÞ
a3
2 ; x Vt [ 0
4 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi
x ðV=2aÞ x ðV=2aÞ iðV=2aÞ
2
exp ði=aÞ x þ ðV=2aÞ ðV=2aÞ ðx VtÞ
ia3
¼ 2
4 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi
x þ ðV=2aÞ x þ ðV=2aÞ ðV=2aÞ
2
exp ð1=aÞ x ðV=2aÞ þ iðV=2aÞ ðx VtÞ
a3
2 ; x Vt\0
4 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2
x ðV=2aÞ x ðV=2aÞ þ iðV=2aÞ
ð5:2:8Þ
Then, the deflection produced by the moving time-harmonic load is given by
n
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
iQ exp ði=aÞ x þ X2 þ X ðx VtÞ þ ixt
wðx; tÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
4a
x þ X2 x X2 þ X
n
2
Q exp ð1=aÞ x X þ iX ðx VtÞ þ ixt
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 ; x Vt [ 0
4a
x X2 x X2 iX
ð5:2:9aÞ
n
2
iQ exp ði=aÞ x þ X X ðx VtÞ þ ixt
wðx; tÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
4a
x þ X2 x X2 X
n
2
Q exp ð1=aÞ x X þ iX ðx VtÞ þ ixt
þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 ; x Vt\0
4a
x X2 x X2 þ iX
ð5:2:9bÞ
where
V
X¼ ð5:2:10Þ
2a
Inspecting the above equations, the radical x X2 in the denominator gives a
critical velocity
pffiffiffiffi
Vcr ¼ 2a x ð5:2:11Þ
at which the deflection divergence takes place.
When the load is stationary at the coordinate origin, the simple time-harmonic
response is obtained by taking the limit X ! 0 ðV ! 0Þ in Eq. (5.2.9). This yields
pffiffiffiffi pffiffiffiffi
Q x x
wðx; tÞ ¼ pffiffiffiffi3 i exp þi xt jxj þ exp jxj þ ixt
4a x a a
ð5:2:12Þ
This is the particular solution of the deflection equation with the time-harmonic
stationary load,
o 4 w o2 w
a4 þ 2 ¼ QdðxÞ expðþixtÞ ð5:2:13Þ
ox4 ot
5.3 An Impulsive Load on a Plate 99
5.3 An Impulsive Load on a Plate
Let us consider an impulsive response of an infinite elastic plate. Taking the ðx; yÞ
coordinate on the neutral plane of the plate, the deflection equation for the plate
with an impulsive point load is given by
4
o w o4 w o4 w o2 w
D 4
þ 2 2 2
þ 4
þ qh 2 ¼ PdðxÞdðyÞdðtÞ ð5:3:1Þ
ox ox oy oy ot
where D is bending rigidity, qh density per unit area and P magnitude of the load.
The initial condition,

ow
wjt¼0 ¼ ¼ 0 ð5:3:2Þ
ot t¼0
and the convergence condition,

p ffiffiffiffiffiffiffiffiffi ow o2 w o3 w
wj x2 þy2 !1 ¼ pffiffiffiffiffiffiffiffiffi ¼ 2 pffiffiffiffiffiffiffiffiffi ¼ 3 pffiffiffiffiffiffiffiffiffi ¼ 0
ox x2 þy2 !1 ox x2 þy2 !1 ox x2 þy2 !1
ð5:3:3Þ
ow o w
2
o w
3
o w
2
¼ 2 pffiffiffiffiffiffiffiffiffi ¼ 3 pffiffiffiffiffiffiffiffiffi ¼ ¼0
oy pxffiffiffiffiffiffiffiffiffi
2 þy2 !1 oy x2 þy2 !1 oy x2 þy2 !1 oxoypffiffiffiffiffiffiffiffiffi
x2 þy2 !1
are applied to the deflection equation.

In order to obtain the particular solution corresponding to the non-homogeneous
loading term, we apply the triple integral transform to Eq. (5.3.1). The Laplace
transform with respect to the time t,
Z1

0
and the double Fourier transform with respect to the space variables ðx; yÞ,
Zþ1 Zþ1
2p
1 1
Zþ1 Zþ1
2p
1 1
are applied to Eq. (5.3.1). The exact solution in the transformed domain is easily
obtained as
Q
¼
~
w ð5:3:7Þ
s2 þ fb ðn2 þ g2 Þg2
2
where
D P
b4 ¼ ; Q¼ ð5:3:8Þ
qh qh
Now, let us consider the inversion. The Laplace inversion of Eq. (5.3.7) is
easily performed with use of the formula (5.1.10). It follows that
Q
~
¼
w sinfb2 ðn2 þ g2 Þtg ð5:3:9Þ
b2 ðn2 þ g2 Þ
The double Fourier inversion is expressed by a double integral as
Zþ1 Zþ1
1 Q
w¼ sinfb2 ðn2 þ g2 Þtg expfiðnx þ gyÞgdndg
ð2pÞ2 b2 ðn2 þ g2 Þ
1 1
ð5:3:10Þ
If we introduce the polar coordinates ðr; hÞ and the polar integration variables
ðu; uÞ as
x ¼ r cos h; y ¼ r sin h ð5:3:11Þ
n ¼ u sin u; g ¼ u cos u; dndg ¼ ududu ð5:3:12Þ

the double integral can be rewritten as
Z1 Z2p
Q 1
w¼ 2 2
sinðb2 tu2 Þ expfiur sinðu þ hÞgdudu ð5:3:13Þ
ð2pÞ b u
0 0
The exponential function in the integrand is the generating function of Bessel

function (Watson 1966, pp. 22),
X
þ1
expfiur sinðu þ hÞg ¼ Jn ðurÞ expfinðu þ hÞg ð5:3:14Þ
1
where Jn ð:Þ is the n-th order Bessel function of the first kind. We substitute the
above into Eq. (5.3.13) and exchange the order of the summation and integration
as
Z1 Z2p X
þ1
Q 1
w¼ sinðb2 tu2 Þ Jn ðurÞ expfinðu þ hÞgdudu
ð2pÞ2 b2 u 1
0 0
ð5:3:15Þ
þ1 Z Z
1 2p
Q X 1 2 2
¼ sinðb tu ÞJ n ðurÞdu expfinðu þ hÞgdu
ð2pÞ2 b2 1 u
0 0
5.3 An Impulsive Load on a Plate 101
The inner integral with respect to the angle variable u is easily evaluated as
Z2p
2p; n ¼ 0
expfinðu þ hÞgdu ¼ ð5:3:16Þ
0 ; n¼6 0
0
Then, we have the single integral for the deflection

Z1
Q 1
w¼ sinðb2 tu2 ÞJ0 ðurÞdu ð5:3:17Þ
2pb2 u
0
Lastly, the integration formula (Erdélyi 1954, pp. 11, 39),

Z1 2
1 2 1 b
sinðax ÞJ0 ðbxÞdx ¼ si ð5:3:18Þ
x 2 4a
0
where si(x) is one of sine integrals, defined by

Z1
sinðuÞ
si(xÞ ¼ du ð5:3:19Þ
u
x
is applied to Eq. (5.3.17). The Green’s function for the dynamic plate deflection is
thus given by
Q r
w¼ 2
si 2
; r ¼ x2 þ y2 ð5:3:20Þ
4pb 4b t
5.4 A Time-Harmonic Load on a Plate
Let us consider an elastic plate on which a time-harmonic point load is applied.

The dynamic deflection of the plate is governed by the deflection equation with the
nonhomogeneous loading term,
4
o w o4 w o4 w o2 w
D þ 2 þ þ qh ¼ PdðxÞdðyÞ expðþixtÞ ð5:4:1Þ
ox4 ox2 oy2 oy4 ot2
where P is the magnitude of the load (but is not of same dimension as that of the
impulsive load in the previous section). The convergence condition at infinity,

ow o2 w o3 w
wjpxffiffiffiffiffiffiffiffiffi
2 þy2 !1 ¼ ¼ ¼ ¼0
ox pffiffiffiffiffiffiffiffiffi
x2 þy2 !1 ox2 pxffiffiffiffiffiffiffiffiffi
2 þy2 !1 ox3 pxffiffiffiffiffiffiffiffiffi
2 þy2 !1
ð5:4:2Þ
ow o2 w o3 w o2 w
¼ 2 pffiffiffiffiffiffiffiffiffi ¼ 3 pffiffiffiffiffiffiffiffiffi ¼ ¼ 0
oy pxffiffiffiffiffiffiffiffiffi
2 þy2 !1 oy x2 þy2 !1 oy x2 þy2 !1 oxoypffiffiffiffiffiffiffiffiffi
x2 þy2 !1
is also assumed. The double Fourier transform defined by Eqs. (5.3.5) and (5.3.6)
is applied to the deflection equation (5.4.1). We then have the simple differential
equation,
~
d2 w

2
þ b4 ðn2 þ g2 Þ2 w
~
¼ Q expðþixtÞ ð5:4:3Þ
dt
and its particular solution is given by
Q
~
¼
w expðþixtÞ ð5:4:4Þ
b ðn þ g2 Þ2 x2
4 2
where b and Q are defined by Eq. (5.3.8).

The formal Fourier inversion is given by the double integral
Zþ1 Zþ1
Q expðinx igyÞ
w¼ 2
expðþixtÞ dndg ð5:4:5Þ
ð2pÞ b4 ðn2 þ g2 Þ2 x2
1 1
Introducing the polar coordinate system defined by Eqs. (5.3.11) and (5.3.12) in
the previous section, the double Fourier inversion integral can be rewritten as
Z1 Z2p
Q expfiur sinðu þ hÞg
w¼ 2
expðþixtÞ ududu ð5:4:6Þ
ð2pÞ ðbuÞ4 x2
0 0
The inner integral with respect to the angular variable u is evaluated with aid of
the formulas (5.3.14) and (5.3.16), and the above Eq. (5.4.6) is reduced to the
single integral,
Z1
Q u
w¼ expðþixtÞ pffiffiffiffi 4 J0 ðurÞdu ð5:4:7Þ
2pb4 u4 ð x=bÞ
0
where the radial distance r is x2 þ y2 .
The last integral with respect to the variable u is evaluated with use of the
complex integral theory. As was done in the former time-harmonic problem in
Sect. 5.2 and earlier too, we introduce a small negative imaginary part to the
frequency, and then consider the complex integral
5.4 A Time-Harmonic Load on a Plate 103
Z
z ð1Þ
U¼ pffiffiffiffi 4 H0 ðzrÞdz ð5:4:8Þ
z4 ð x=bÞ
C
ð1Þ
where H0 ð:Þ is Hankel function of the first kind. The closed loop C is shown in
Fig. 5.2. Since the Hankel function has a logarithmic singularity and a branch
point at the origin, a branch cut along the negative real axis is introduced. The
integrand has four poles, whereof two of them are in the upper z-plane due to the
introduction of the imaginary part to the frequency. They are marked in the figure
and these two poles are specified with argument
pffiffiffiffi pffiffiffiffi pffiffiffiffi pffiffiffiffi
x x x x
z¼ ¼ expðþpiÞ; z ¼ þi ¼ expðþpi=2Þ ð5:4:9Þ
b b b b
It should be understood that the argument of z along the positive real axis is zero,
but that along the negative real axis is þp. Then, we apply Cauchy’s theorem to
the complex integral U in Eq. (5.4.8). The integral along the large semi-circle
vanishes as its radius tends to infinite. That along the small semi-circle also
vanishes as its radius tends to zero. Thus, the integral along the real axis is
converted to the sum of two residues at the poles. That is
Z1 n o
u ð1Þ ð1Þ pi
p ffiffiffi
ffi 4
H0 ðurÞ H 0 ðure Þ du
u4 ð x=bÞ
0 ð5:4:10Þ
b2 ð1Þ r x pi=2 b2 ð1Þ r x pi
¼ 2pi H e þ H e
4x 0 b 4x 0 b
The formulas for the Hankel function (Watson 1966, pp. 75 and 78),
ð1Þ ð2Þ ð1Þ 2i

H0 ðxepi Þ ¼ H0 ðxÞ; H0 ðxepi=2 Þ ¼ K0 ðxÞ ð5:4:11Þ
p
are very useful for arranging Eq. (5.4.10). Then, we have the integration formula
for our use,
Z1 pffiffiffiffi pffiffiffiffi
u b2 pi ð2Þ r x r x
J
pffiffiffiffi 4 0 ðruÞdu ¼ H0 þ K0 ð5:4:12Þ
4
u ð x=bÞ 2x 2 b b
0
Applying this formula to Eq. (5.4.7), the time-harmonic response of the plate, i. e.
the time-harmonic Green’s function, is given by
Q pi ð2Þ r x r x
wðx; y; tÞ ¼ 2
H0 þ K 0 expðþixtÞ ð5:4:13Þ
4pb x 2 b b
where r is the radial distance from the load.
C
arg( z ) = +π
z = +i ω / β
z = − ω /β
F
D E A B
arg( z ) = 0
branch cut
Fig. 5.2 A closed loop C for the complex integral U
The reader will find that the deflection has two components: One is a time-
ð2Þ pffiffiffiffi
harmonic wave which is given by the product H0 ðr x=bÞ expðþixtÞ. The other
pffiffiffiffi
is a simple time-harmonic vibration K0 ðr x=bÞexpðþixtÞ whose amplitude
decays exponentially with the distance. The same nature can be found in the 1D
beam response in Eq. (5.2.12).
Exercises
(1) Derive the wave velocity from the time-harmonic response (5.2.9) and show
that the velocity depends on the frequency.
(2) Using the generating formula for Bessel function (5.3.14), derive the integral
representation for the Bessel function as
Zþp Zp
1 1
Jn ðzÞ ¼ expðiz sin u + inuÞdu ¼ cosðnu z sin uÞdu ðaÞ
2p p
p 0
(3) The asymptotic formula for the Hankel function is given by

rffiffiffiffiffi
ð1Þ 2
H0 ðzÞ expfþiðz p=4Þg; z ! 1 ðbÞ
pz
Explain why the integral along the semi-circle BCD in Fig. 5.2 vanishes for the
complex integral (5.4.8).
Table C Green’s functions for beam and plate
Deflection equation Source pðx; tÞ Green’s function
2
1D beam o4 w PdðxÞdðtÞ pffiffih i
EI ox4þ qA ootw2 ¼ pðx; tÞ wðx; tÞ ¼ Qa pt ppzffiffi2 fSðzÞ CðzÞg þ sin p4 ð2z2 þ 1Þ ; z ¼ apxffiffiffiffiffi

1=4 2pt
EI P
a ¼ qA ; Q ¼ qA PdðxÞ expðþixtÞ Q x x
wðx; tÞ ¼ pffiffiffi3 i exp þi xt jxj þ exp jxj þ ixt
4a x a a
5.4 A Time-Harmonic Load on a Plate
4

Q
2D plate o w o4 w o4 w PdðxÞdðyÞdðtÞ r2
D þ 2 þ wðx; y; tÞ ¼ 4pb2 si
4b2 t
;
r ¼ x2 þ y2
ox4 ox2 oy2 oy4 n
ð2Þ

pffiffiffi
pffiffiffio pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
PdðxÞdðyÞ expðþixtÞ wðx; y; tÞ ¼ 4pbQ2 x pi2 H0 r bx þ K0 r bx expðþixtÞ; r¼ x2 þ y 2
o2 w
þqh 2 ¼ pðx; y; tÞ

1=4 ot
D P
b ¼ qh ; Q ¼ qh
105
References
Cambridge
Erdélyi A (ed) (1954) Tables of integral transforms, vol II, McGraw-Hill, New York
Chapter 6
Cagniard-de Hoop Technique
The success of the integral transform method hangs on the evaluation of inversion
integrals. It is not easy to find a suitable integration formula. If we cannot find any
suitable formula, the inversion is left in its integral form and some numerical
techniques must be applied for the evaluation. However, in the case of double
inversion, such as Laplace and Fourier inversions, if we could convert the first
Fourier inversion integral to a definition integral of Laplace transform, the next
Laplace inversion can be carried out by inspection without evaluating its Laplace
inversion integral. For example, when we have the double transformed function
Fðn; sÞ and its Fourier inversion integral
Zþ1

f ðx; sÞ ¼ Fðn; sÞ expðinxÞdn ð6:1Þ
1
if we could apply some mathematical techniques to the integral and convert the
integral to the definition form of Laplace transform,
Z1

f ðx; sÞ ¼ Gðx; tÞ expðstÞdt ð6:2Þ
a
i.e. the Laplace transform integral, the Laplace inversion is easily carried out by
inspection and its inversion is given by

Gðx; tÞ ; t [ a
f ðx; tÞ ¼ ¼ Hðt aÞGðx; tÞ ð6:3Þ
0 ; t\a
The most important and substantial part of the this technique is to convert the
Fourier inversion integral to the form of the Laplace transform,
Zþ1 Z1
Fðn; sÞ expðinxÞdn ¼ Gðx; tÞ expðstÞdt ð6:4Þ
1 a

108 6 Cagniard-de Hoop Technique
This conversion is carried out with use of the theory of complex integrals and is
called ‘‘Cagniard-de Hoop technique’’. Cagniard (1962) is the author of this
technique, but his original technique is to map the inversion integral onto the
integral in another complex t-plane. de Hoop (1961) has modified the technique to
use the same complex plane without mapping. The present chapter explains the
technique developed by de Hoop, but uses the name ‘‘Cagniard-de Hoop tech-
nique.’’ The essential idea is to convert the first inversion integral to the definition
integral of the second integral transform. This technique is applicable to other
combinations of two integral transforms, not limited to Laplace-Fourier trans-
forms. In addition, it will be worth to cite textbooks (Achenbach 1973; Fung 1970;
Graff 1975; Miklowitz 1978), which treat elastic waves and explain the Cagniard-
de Hoop technique in some depths.
The first section in the present chapter treats a very simple problem in order to
demonstrate the Cagniard-de Hoop technique in details. In the following sections, 2D
and 3D Green’s functions for the coupled elastodynamic equations are discussed and
some additional explanations for the Cagniard-de-Hoop technique are described.
6.1 2D Anti-Plane Deformation
As the first example of the Cagniard-de Hoop technique, the simplest elastody-
namic problem is discussed. We consider an elastic half space and take the
Cartesian coordinate ðx; y; zÞ as shown in Fig. 6.1 where the z-axis is normal to the
paper plane. The surface of the solid is denoted by y ¼ 0 and its interior y [ 0. An
impulsive anti-plane shear load directed to the negative z-direction is applied on
the surface and is expressed by

ryz y¼0 ¼ Pz dð xÞdðtÞ ð6:1:1Þ
where Pz is load magnitude. Since the load, of infinite length, is directed to the
anti-plane z-direction, an anti-plane deformation is induced. The equation of
motion is given by
orxz oryz o2 uz
þ ¼q 2 ð6:1:2Þ
ox oy ot
where q is density. Hooke’s law for two shear stresses is given by
ouz ouz
rxz ¼ l ; ryz ¼ l ð6:1:3Þ
ox oy
where l is shear rigidity. We also employ the quiescent condition at an initial time,

ouz
uz jt¼0 ¼ ¼0 ð6:1:4Þ
ot t¼0
6.1 2D Anti-Plane Deformation 109
Fig. 6.1 An impulsive anti- σ yz y =0

= Pzδ ( x)δ (t )
plane shear load on the
surface of a semi-infinite y=0 x
elastic solid
μ, ρ

p ffiffiffiffiffiffiffiffiffi ouz ouz
uz j x2 þy2 !1 ¼ ¼ ¼0 ð6:1:5Þ
x2 þy2 !1 oy pffiffiffiffiffiffiffiffiffi
x2 þy2 !1
Equations (6.1.1–6.1.5) constitute the present elastodynamic problem.

Substituting Hooke’s law of Eq. (6.1.3) into the equation of motion (6.1.2), we
obtain the typical wave equation for the anti-plane displacement uz ,
o2 uz o2 uz 1 o2 uz
þ ¼ ð6:1:6Þ
ox2 oy2 c2s ot2
where shear wave velocity cs is defined by
pffiffiffiffiffiffiffiffi
cs ¼ l=q ð6:1:7Þ
In order to solve our elastodynamic problem, Laplace and Fourier transforms
are employed; Laplace transform with respect to the time,
Z1

0
and Fourier transform with respect to the space variable x,

Zþ1 Zþ1
2p
1 1
With aid of the quiescent and convergence conditions, the displacement equation
(6.1.6) and Hooke’s law (6.1.3) are transformed to
uz n 2
d2 o
2
n þ ðs=cs Þ2 uz ¼ 0 ð6:1:10Þ
dy
duz
xz ¼ inl
r uz ; yz ¼ l
r ð6:1:11Þ
dy
After solving the transformed displacement equation (6.1.10), the displacement
and stress, which satisfy the convergence condition at infinity, are given by
uz ¼ Aðn; sÞ expðas yÞ;
yz ¼ las Aðn; sÞ expðas yÞ
r ð6:1:12Þ
Here Aðn; sÞ is an unknown coefficient to be determined by the loading condition.

The radical is defined by
as ¼ n2 þ ðs=cs Þ2 ; Reðas Þ [ 0 ð6:1:13Þ
The boundary condition is also transformed, to

yz ¼ Pz
r ð6:1:14Þ
y¼0
Substituting the second equation in Eq. (6.1.12) into the above transformed
boundary condition, the coefficient is determined as
Pz
Aðn; sÞ ¼ ð6:1:15Þ
las
and thus, the displacement in the transformed domain is explicitly determined as
Pz
uz ¼
expðas yÞ ð6:1:16Þ
las
Now, we shall consider the inversion. The formal Fourier inversion for the
displacement is given by the indefinite integral
Zþ1
Pz 1
uz ¼ expðas y inxÞdn ð6:1:17Þ
2pl as
1
The above equation (6.1.17) is the Laplace transformed displacement, but in the
form of Fourier inversion integral. If we could convert the integral into the defi-
nition form of Laplace transform integral, its Laplace inversion can be performed
by inspection (i.e. the Cagniard-de Hoop technique).
Let us start to apply the technique. First of all, in order to eliminate the Laplace
transform parameter s, leaving it only in the argument of the exponential function
in the integrand, a variable transform from n to 1 is introduced as
n ¼ s1 ð6:1:18Þ
Here we assume that the Laplace transform parameter s is a positive real constant.
The integral of Eq. (6.1.17) is rewritten as
Zþ1 n qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
Pz 1
uz ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp s y 12 þ 1=c2s þ i1x d1 ð6:1:19Þ
2pl 12 þ 1=c2s
1
where the radical must satisfy the radiation condition,

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Re 12 þ 1=c2s [ 0 ð6:1:20Þ
Examining the integral in Eq. (6.1.19), the Laplace transform parameter is

included only in the argument of the exponential function as a simple multiplier. If
we could change the argument to a simple product, such as st where t is a new
variable, the integral will have the form of the definition integral of the Laplace
transform. So, the variable transform from 1 to the new variable t is introduced as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
t ¼ y 12 þ 1=c2s þ i1x ð6:1:21Þ
However, its inverse gives multiple values for 1

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðÞ itx y t2 ðx2 þ y2 Þ=c2s
1s ¼ ð6:1:22Þ
x 2 þ y2
Due to the multi-valuedness, we are puzzled as to which one is the suitable
inversion for 1. In order to solve this puzzle, we examine the inverse as a function
ðÞ ðÞ
of the real parameter t. Denote 1s as 1s ¼ X iY, where
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
y t2 ðx2 þ y2 Þ=c2s tx
X¼ ; Y¼ 2 ð6:1:23Þ
x2 þ y2 x þ y2
Eliminating the parameter t, we obtain an equation for a hyperbolic curve in ðX; YÞ
plane,
2 2
X Y 1
¼ 2 ð6:1:24Þ
y x cs
ðÞ
Thus, the inverse 1s has the form of two semi-hyperbolic curves in the complex
ðÞ ðþÞ
1-plane. When x [ 0, 1s is the left half and 1s is the right half of the hyperbola
in the lower 1-plane, as shown in Fig. 6.2. On the other hand, when x\0, the
ðÞ
hyperbolic curves 1s are in the upper 1-plane. The connected two semi-hyper-
bolic curves CAB shown in the figure is called Cagniard’s path.
If we could connect the real line with the edges of two semi-hyperbolas in the
complex plane, a closed loop for the complex integral is formed; and then Cauchy’s
integral theorem can be applied. Thus, we consider the complex integral U, whose
integrand is the same as that of the Fourier inversion integral in Eq. (6.1.19),
Fig. 6.2 Cagniard’s path Im( ς )
C x<0 B
ς s( − ) ς s( + )
A
saddle point
R e( ς )
A
ς s( − ) ς s( + )
C x>0 B
I n qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
1 1
U¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp s y 12 þ 1=c2s þ i1x d1 ð6:1:25Þ
2p 12 þ 1=c2s
C
The integrand has two branch points at

1 ¼ i=cs ð6:1:26Þ
In order to guarantee that the radiation condition of Eq. (6.1.20) is satisfied, two
branch cuts are introduced along the imaginary axis in the complex 1-plane as
shown in Fig. 6.3. When the real parameter t varies from x2 þ y2 =cs to þ1, the
ðþÞ ðÞ
inverse 1s moves on the semi-hyperbola from A to B, and the other inverse 1s
moves on the other branch from A to C. These two semi-hyperbolas are connected
at the saddle point A,
ix
1saddle ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:1:27Þ
cs x2 þ y2
since the saddle point is always smaller in magnitude than the branch point
1 ¼ i=cs . The other edges at infinite t must be connected with the real line. The
edge B is connected to the line at the positive infinity F with the large arc. The
edge C is also connected to the line at the negative infinity D with the large arc, as
shown in Fig. 6.3. Then, the closed loop for the complex integral is formed by the
loop CABFEDC. The closed loop CðÞ in the lower 1-plane is employed in the
case of positive x and the loop CðþÞ in the upper plane is in the case of negative x.
In either case, no singular point is included in the closed loop. The integral along
the large arc vanishes as its radius tends to infinity. Thus, the integral along the real
Fig. 6.3 Closed loop C ðÞ branch cut Im( ς )

for the complex integral U
C B
x<0
R→ ∞
+i / cs
ς (− ) ς s( + )
s
A C (+ )
F
D E
D E F R e( ς )
C (− ) A
ς (s+ )
ς s( − ) −i / cs
C B x>0
axis is converted to those along the two semi-hyperbolas CAB due to Cauchy’s
integral theorem.
The integral along the hyperbola is one of the parametric integrals and its
parameter is the variable t. The identity between two integrals along the real axis
and along the hyperbolas is thus given by
1 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp s y 12 þ 1=c2s þ i1x d1
2p 12 þ 1=c2s
1 8 9
Z1
1 < 1 d1 1 d1 =
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi expðstÞdt
2p pffiffiffiffiffiffiffiffiffi : 12 þ 1=c2s dt ðþÞ 12 þ 1=c2s dt ðÞ
;
1¼1s 1¼1s
x2 þy2 =cs
ð6:1:28Þ
The integrand in the right hand side of Eq. (6.1.28) can be simplified. Explicit
expressions for the radical and the gradient are derived from the definition of
Eq. (6.1.21) and its inverse equation (6.1.22),
1 x2 þ y2
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðÞ
2 yt ix t2 ðx2 þ y2 Þ=c2s
1s þ1=c2s
ð6:1:29Þ
ðÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
d1s yt ix t2 ðx2 þ y2 Þ=c2s
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dt ðx2 þ y2 Þ t2 ðx2 þ y2 Þ=c2s
Substituting these into the integral in the right hand side of Eq. (6.1.28), the
parametric integral along the hyperbola is simplified as
8 9
Z1
1 < 1 d1 1 d1 =
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi expðstÞdt
2p pffiffiffiffiffiffiffiffiffi : 12 þ 1=c2s dt ðþÞ 12 þ 1=c2s dt ðÞ
;
1¼1s 1¼1s
x2 þy2 =cs
Z1
1 1
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi expðstÞdt
p pffiffiffiffiffiffiffiffiffi t ðx2 þ y2 Þ=c2s
2
x2 þy2 =cs
ð6:1:30Þ
Thus, we have converted the Fourier inversion integral to an integral of the
Laplace transform,
1 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp s y 12 þ 1=c2s þ i1x d1
2p 12 þ 1=c2s
1
Z1 ð6:1:31Þ
1 1
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi expðstÞdt
p pffiffiffiffiffiffiffiffiffi t ðx2 þ y2 Þ=c2s
2
x2 þy2 =cs
Substituting the right hand side of the above equation into the Laplace transformed
displacement in Eq. (6.1.19), we have the Laplace transformed displacement in the
form of a Laplace transform integral,
Z1
Pz 1
uz ðx; y; sÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi expðstÞdt ð6:1:32Þ
pl pffiffiffiffiffiffiffiffiffi t ðx2 þ y2 Þ=c2s
2
x2 þy2 =cs
The integral is just the form of Laplace transform, but its lower limit is not zero.
The integral states that the original function is vanishing before t ¼ x2 þ y2 =cs
and after this time the function has the form of the integrand. Then, we can
anticipate the original displacement function before Laplace transform, i.e.
Laplace inversion of Eq. (6.1.32),

uz ðx; y; tÞ ¼ L1 uz ðx; y; sÞ

Pz < pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
; t [ x2 þ y2 =cs ð6:1:33Þ
t2 ðx2 þy2 Þ=c2s
¼ p ffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pl :
0 ; t\ x2 þ y2 =cs
Utilizing Heaviside’s unit step function, the displacement is expressed in the
compact form,
Pz H c s t x2 þ y2
uz ðx; y; tÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:1:34Þ
pl t2 ðx2 þ y2 Þ=c2s
This is our final result for the double inversion. The reader should notice that we
did not use any integration formula for the Laplace inversion but we did the
inversion. This is the Cagniard-de Hoop technique!
The conditional for the step function gives the circular region disturbed by the
transient SH-wave,
ðcs tÞ2 [ x2 þ y2 ð6:1:35Þ

and its edge is a circular (cylindrical in 3D) wave front with the center at the
source point,
ðcs tÞ2 ¼ x2 þ y2 ð6:1:36Þ
6.2 2D In-Plane Deformation
We consider a 2D transient response of an semi-infinite elastic solid. Take the

coordinate system ðx; yÞ so that the surface of the solid is at y ¼ 0 and its interior in
y [ 0 as shown in Fig. 6.4. An impulsive point load is applied on the surface and
is expressed by

ryx y¼0 ¼ Px dðxÞdðtÞ; ryy y¼0 ¼ Py dðxÞdðtÞ ð6:2:1Þ
where Pj ; j ¼ x; y are the components of the load. The in-plane deformation of the
elastic solid is governed by the equations of motion,
orxx oryx o2 u x
þ ¼q 2
ox oy ot
ð6:2:2Þ
orxy oryy o2 u y
þ ¼q 2
ox oy ot
Fig. 6.4 An impulsive load σ yy = Py δ ( x) δ (t )

y=0
on the surface of a semi-
infinite elastic solid
σ yx = Pxδ ( x) δ (t ) x
y=0
y=0
λ , μ, ρ
y
where q is density. Hooke’s law for the plane strain is given by

oux ouy oux
rxx ¼ k þ þ 2l
ox oy ox

oux ouy ouy
ryy ¼ k þ þ 2l ð6:2:3Þ
ox oy oy

oux ouy
rxy ¼ ryx ¼ l þ
oy ox
where k and l are Lame’s constants. We employ the quiescent condition at an
initial time,

oui
ui jt¼0 ¼ ¼ 0 ; i ¼ x; y ð6:2:4Þ
ot t¼0
and the convergence (radiation) condition at infinity

oui oui
ui jpffiffiffiffiffiffiffiffiffi ¼ ¼ ¼0; i ¼ x; y ð6:2:5Þ
x2 þy2 !1 ox pffiffiffiffiffiffiffiffiffi
x2 þy2 !1 oy pxffiffiffiffiffiffiffiffiffi
2 þy2 !1
Equations from (6.2.1) to (6.2.5) constitute the impulsive Lamb’s problem (Fung,
1970) for the 2D semi-infinite elastic solid. Substituting Hooke’s law into the
equations of motion, the displacement equations are obtained as
o2 ux o2 ux o2 uy 1 o2 ux
c2 2
þ 2 þ ðc2 1Þ ¼ 2 2
ox oy oxoy cs ot
ð6:2:6Þ
2 o2 ux o2 u y o2 u y 1 o2 uy
ðc 1Þ þ 2 þ c2 2 ¼ 2 2
oxoy ox oy cs ot
where c is the velocity ratio defined by Eq. (3.8) in Chap. 3. We apply the double
integral transform: Laplace transform with respect to the time,
Z1

0
and Fourier transform with respect to the space variable x,

Zþ1 Zþ1
2p
1 1
With use of the quiescent and radiation conditions, the displacement equations are
transformed to the coupled ordinary differential equations with constant
coefficients,
6.2 2D In-Plane Deformation 117
ux n 2 2
d2 2
o
2
duy
c n þ ðs=c s Þ
u x inðc 1Þ ¼0
dy2 dy
ð6:2:9Þ
du d2 uy n o
inðc2 1Þ x þ c2 2 n2 þ ðs=cs Þ2 uy ¼ 0
dy dy
Hooke’s law and the boundary condition are also transformed to
1 1 du
xy ¼ r
r yx ¼ x inuy
l l dy
ð6:2:10Þ
1 duy
yy ¼ inðc2 2Þ
r ux þ c2
l dy
and

yx
r ¼ Px ; yy
r ¼ Py ð6:2:11Þ
y¼0 y¼0
The solution for the displacement equation (6.2.9) is given by

in
ux ¼
Aðn; sÞ expðad yÞ þ Bðn; sÞ expðas yÞ
ad
ð6:2:12Þ
in
y ¼ Aðn; sÞ expðad yÞ Bðn; sÞ expðas yÞ
u
as
where Aðn; sÞ and Bðn; sÞ are unknown coefficients, and the two radicals are
defined by
aj ¼ n2 þ ðs=cj Þ2 ; j ¼ d; s ð6:2:13Þ
with the radiation condition Reðaj Þ [ 0. Substituting Eq. (6.2.12) into

Eq. (6.2.10), the stress components are obtained as
1 a2 þ n2
yx ¼ 2inAðn; sÞ expðad yÞ s
r Bðn; sÞ expðas yÞ
l as
ð6:2:14Þ
1 a2 þ n2
yy ¼ s
r Aðn; sÞ expðad yÞ þ 2inBðn; sÞ expðas yÞ
l ad
Applying the boundary condition of Eq. (6.2.11) to the above stresses, we obtain
the following simple algebraic equations for the unknown coefficients,
a2s þ n2
2inAðn; sÞ Bðn; sÞ ¼ Px =l
as
ð6:2:15Þ
a2 þ n2
s Aðn; sÞ þ 2inBðn; sÞ ¼ Py =l
ad
Their solutions are given by
2inad as ðPx =lÞ þ ad ða2s þ n2 ÞðPy =lÞ

Aðn; sÞ ¼
Rðn; sÞ
ð6:2:16Þ
as ða2s þ n2 ÞðPx =lÞ 2inad as ðPy =lÞ
Bðn; sÞ ¼
Rðn; sÞ
where Rðn; sÞ is the Rayleigh equation, defined by
Rðn; sÞ ¼ ða2s þ n2 Þ2 4ad as n2 ð6:2:17Þ

Substitution of Eqs. (6.2.16) into Eq. (6.2.12) gives the displacement in the double
transformed domain. The transformed displacement can also be expressed as a
sum of the dilatational and shear wave contributions,
Px nðdÞ o P n
y ðdÞ
o
ux ¼
ðsÞ
Ixx ðn; y; sÞ þ Ixx ðn; y; sÞ þ ðsÞ
Ixy ðn; y; sÞ þ Ixy ðn; y; sÞ
l l
P n o P n o ð6:2:18Þ
x ðdÞ y ðdÞ
uy ¼
ðsÞ
Iyx ðn; y; sÞ þ Iyx ðn; y; sÞ þ ðsÞ
Iyy ðn; y; sÞ þ Iyy ðn; y; sÞ
l l
ðkÞ
where the superscripts ðdÞ and ðsÞ for k in Iij stand for the dilatational and shear
wave contributions, respectively. The first subscript i stands for the direction of the
displacement component and the second j for the component of the load. More
explicit expressions for each of the contributions are
ðdÞ
Ixx 2n2 as
ðn; y; sÞ ¼ expðad yÞ
Rðn; sÞ
ðsÞ
Ixx as ða2s þ n2 Þ
ðn; y; sÞ ¼ expðas yÞ
Rðn; sÞ
ð6:2:19Þ
2 2
ðdÞ
Ixy inðas þ n Þ
Rðn; sÞ
ðsÞ
Ixy 2inad as
Rðn; sÞ
x , and
for u
ðdÞ
Iyx 2inad as
Rðn; sÞ

ðsÞ in a2s þ n2
Iyx ðn; y; sÞ ¼ expðas yÞ
Rðn; sÞ
ð6:2:20Þ
ðdÞ
Iyy ad ða2s þ n2 Þ
Rðn; sÞ
ðsÞ
Iyy 2n2 ad
Rðn; sÞ
y .
for u
We have obtained explicit expressions for the transformed displacement

components. Our next task is to invert the displacement into the real domain
ðx; y; tÞ. However, it might be impossible to invert the displacement as a whole.
The inversion must be carried out for each wave contribution one by one. As all
ðdÞ
dilatational wave contributions given by Iij have the same exponential function
ðsÞ
expðad yÞ and the shear wave contributions Iij have the function expðas yÞ, we
invert these two wave contributions separately.
(1) Inversion of the dilatational wave contribution
As a representative for the dilatational wave contribution, we consider the
ðdÞ
inversion of Ixx . The formal Fourier inversion integral with respect to the
parameter n is given by
Zþ1
ðdÞ 1 2n2 as
Ixx ðx; y; sÞ ¼ expðad y inxÞdn ð6:2:21Þ
2p Rðn; sÞ
1
In order to eliminate the Laplace transform parameter s in the integrand except the
exponential function, the variable transform from n to the new variable 1 is
introduced as
n ¼ ðs=cd Þ1 ð6:2:22Þ
The inversion integral is rewritten as
Zþ1 n pffiffiffiffiffiffiffiffiffiffiffiffiffi o
ðdÞ 1 ðdÞ
Ixx ðx; y; sÞ ¼ Fxx ð1Þ exp ðs=cd Þ y 12 þ 1 þ ix1 d1 ð6:2:23Þ
2p
1
where
ðdÞ 212 12 þ c 2
Fxx ð 1Þ ¼ ð6:2:24Þ
Rð 1 Þ
and its denominator, which is called the Rayleigh equation/function, is redefined by
2 pffiffiffiffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Rð1Þ ¼ 212 þ c2 412 12 þ 1 12 þ c2 ð6:2:25Þ
Note that the Rayleigh equation Rð1Þ differs from Rðn; sÞ in Eq. (6.2.17).
Subsequently, we consider the complex integral U whose integrand is the same
as that in Eq. (6.2.23),
I n pffiffiffiffiffiffiffiffiffiffiffiffiffi o
1 ðdÞ
U¼ Fxx ð1Þ exp ðs=cd Þ y 12 þ 1 þ ix1 d1 ð6:2:26Þ
2p
C ðÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
The closed loops C ðÞ are discussed here. The two radicals 12 þ 1 and 12 þ c2
in the integrand have each two branch points at 1 ¼ i; ic respectively. Thus,
two branch cuts are introduced along the imaginary axis in the complex 1-plane as
shown in Fig. 6.5. Only the lower half plane is shown since the path and singular
point are symmetric about the real axis. The Rayleigh equation has two pure
imaginary roots, 1 ¼ icR , where cR is the velocity ratio of the dilatational wave
cd to the Rayleigh wave cR ,
cR ¼ cd =cR ð6:2:27Þ
This velocity ration cR is always greater than c, since the following inequality
holds for the isotropic media:
cR \cs \cd ð6:2:28Þ
As to the argument of the exponential function, we hope to transform it to the
product of the new (time) variable t and the transform parameter s. So, we
introduce the new variable t as
pffiffiffiffiffiffiffiffiffiffiffiffiffi
t ¼ y 12 þ 1 þ ix1 =cd ð6:2:29Þ
Solving for the variable 1, we have two solutions, the so-called ‘‘Cagniard’s path’’,
C ( − ) for x > 0 saddle point

−ix
ς =
Im(ς ) x + y2
2
Re(ς )
−∞ +∞
Cagniard ’s path
ς = −i
ς d( + )
ς d( − ) ∞
ς = −i γ
branch cut ς = −i γ R
Fig. 6.5 Closed loop C ðÞ and Cagniard’s path for the dilatational wave contribution
ðÞ
ixðcd tÞ y ðc d t Þ2 ðx2 þ y 2 Þ
1d ¼ ð6:2:30Þ
x2 þ y2
ðÞ
These solutions 1d take the form of two symmetrical semi-hyperbolas in the
complex 1-plane as the real parameter t varies from x2 þ y2 =cd to infinity. The
two hyperbolas are connected to each other at the saddle point
ðdÞ ix
x2 þ y2
where the parameter t starts from t ¼ x2 þ y2 =cd . This connected hyperbola is
called Cagniard’s path as shown in Fig. 6.5. Fortunately, the saddle point is not on
any branch cut. Then, we can determine the closed loop in the complex 1-plane.
When the Cagniard’s path is in the lower plane, the closed loop is composed of
an infinite line along the real axis, Caginiard’s path and two large arcs which
connect the line to the Cagniard’s path. When the Caginiard’s path is in the upper
plane, the closed loop is composed of the similar ones in the lower plane and is
symmetric with the lower loop about the real axis. The lower loop CðÞ and the
branch cuts and points are shown in Fig. 6.5. The closed loop CðÞ is employed
when the space variable x is positive, and the loop CðþÞ when it is negative, due to
the convergence at the large arc. In any case, positive or negative x, i.e. the upper
or lower circuit in the complex plane, the closed loop does not include any singular
point. Then we can apply Cauchy’s integral theorem to the complex integral U.
Since the integral along the large arc vanishes, we can convert the integral along
the real axis to that along the Cagniard’s path. That is
Zþ1 n pffiffiffiffiffiffiffiffiffiffiffiffiffi o
ðdÞ 1 ðdÞ
Ixx ðx; y; sÞ ¼ Fxx ð1Þ exp ðs=cd Þ y 12 þ 1 þ ix1 d1
2p
1
Z1 ( )
d1ðþÞ ðÞ
1 ðdÞ ðþÞ d ðdÞ ðÞ d1d
¼ Fxx 1d Fxx 1d expðstÞdt
2p pffiffiffiffiffiffiffiffiffi dt dt
x2 þy2 =cd
ð6:2:32Þ
If we understand that the integration variable t as the real time variable, the
integral has just the form of a Laplace transform integral, but with the shifted
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðdÞ
starting time x2 þ y2 =cd . As the Laplace inversion of Ixx ðx; y; sÞ is the original
function before Laplace transform, the inversion is carried out by inspection on
Eq. (6.2.32). Utilizing the step function, the original function is given by
( )
h i ðþÞ ðÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 ðdÞ 1 ðdÞ ðþÞ d1d ðdÞ ðÞ d1d
L Ixx ðx; y; sÞ ¼ Fxx 1d Fxx 1d H t x2 þ y2 =cd
2p dt dt
ð6:2:33Þ
ðdÞ
Consequently, the double inversion for Ixx ðn; y; sÞ is given by
( )
ðþÞ ðÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðdÞ 1 ðdÞ ðþÞ d1d ðdÞ ðÞ d1d
Ixx ðx; y; tÞ ¼ Fxx 1d Fxx 1d H t x2 þ y2 =cd
2p dt dt
ð6:2:34Þ
The same Cagniard-de Hoop technique can be applied to the other dilatational
ðdÞ
wave contributions Iij . Thus, we have the unified expression for the dilatational
wave contributions as
( )
ðdÞ 1 ðdÞ ðþÞ d1d ðdÞ ðÞ d1d
Iij ðx; y; tÞ ¼ Fij 1d Fij 1d H t x2 þ y2 =cd
2p dt dt
ð6:2:35Þ
ðdÞ
where Fij ð1Þ
are given by
ðdÞ 212 12 þ c2 ðdÞ i1ð212 þ c2 Þ
Fxx ð1Þ ¼ ; Fxy ð1Þ ¼
Rð1Þ Rð1Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:2:36Þ
ðdÞ 2i1 12 þ 1 12 þ c2 ðdÞ ð212 þ c2 Þ 12 þ 1
Fyx ð1Þ ¼ ; Fyy ð1Þ ¼
Rð1Þ Rð1Þ
(2) Inversion of the shear wave contribution

We consider the double inversion for the shear wave contribution, which has
the exponential function expðas yÞ. As a representative, we consider the inversion
of

ðsÞ as a2s þ n2
Ixx ðn; y; sÞ ¼ expðas yÞ ð6:2:37Þ
Rðn; sÞ
Its formal Fourier inversion is given by
Zþ1 ( )
ðsÞ 1 as a2s þ n2
Ixx ðx; y; sÞ ¼ expðas y inxÞdn ð6:2:38Þ
2p Rðn; sÞ
1
Introducing the variable transform defined by Eq. (6.2.22), the inversion integral is
rewritten as
Zþ1 n pffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
ðsÞ 1 ðsÞ
Ixx ðx; y; sÞ ¼ Fxx ð1Þ exp ðs=cd Þ y 12 þ c2 þ i1x d1 ð6:2:39Þ
2p
1
where
ðsÞ ð212 þ c2 Þ 12 þ c2
Fxx ð1Þ ¼ ð6:2:40Þ
Rð1Þ
and the Rayleigh equation Rð1Þ is defined by Eq. (6.2.25).
We shall consider the complex integral U whose integrand is the same as that in
Eq. (6.2.39),
I n pffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
1 ðsÞ
U¼ Fxx ð1Þ exp ðs=cd Þ y 12 þ c2 þ i1x d1 ð6:2:41Þ
2p
C ðÞ
The closed loop C ðÞ is discussed here. Four branch points at 1 ¼ i; ic are
found from the two radicals; and corresponding branch cuts are introduced along
the imaginary axis in the 1-plane. The denominator, i.e. the Rayleigh equation, has
two symmetric zeros at 1 ¼ icR . These zeros give the simple poles for the
integrand and the poles are on the imaginary axis (on the branch cut), but is greater
in magnitude than any of the branch points. The branch points, cuts and poles are
shown in Figs. 6.6 and 6.7.
As to the Cagniard’s path, we introduce a new parameter t, defined by
t ¼ ðy 12 þ c2 þ i1xÞ=cd ð6:2:42Þ
Its inversion gives the Cagniard’s path in the complex 1-plane,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ixðc d tÞ y ðcd tÞ2 c2 ðx2 þ y2 Þ
1ðÞ
s ¼ ð6:2:43Þ
x2 þ y2
The saddle point at t ¼ x2 þ y2 =cs is
ðsÞ icx
x2 þ y2
Comparing the magnitude of the saddle point with those of two branch points, we
find that the saddle point is always smaller than the branch points 1 ¼ ic of the
shear wave, but greater than the branch points 1 ¼ i of the dilatational wave, if
the inequality
cjxj pffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi [ 1 ) y\jxj c2 1 ð6:2:45Þ
x2 þ y2
holds. Thus, we will have two different closed loops for the complex integral.
When the saddle point is smaller in magnitude than the branch points 1 ¼ i,
the Cagniard’s path does not cross any branch cut; the closed loop is then com-
posed of the line along the real axis, the Cagniard’s path, and two large arcs which
connect the line with the Cagniard’s path. This closed loop is similar to that in the
γx
Cagniard ’s path I: < 1 for x > 0
x2 + y 2
Im(ς )
Re(ς )
−∞ Cagniard’s path I +∞
ς s( − ) ς = −i
ς s( + )
ς = −iγ ∞
branch cut ς = −iγ R
Fig. 6.6 Closed loop and Cagniard’s path I
γx
Cagniard’s path II : > 1 for x > 0
x + y2
2
Im(ς ) Re(ς )
−∞ H
I
C ε →0 G
ς = −i
B D
Cagniard ’s path II
A E
ς (−) ς s( + ) ∞
s
ς = −iγ
F
J branch cut ς = −iγ R
Fig. 6.7 Closed loop and Cagniard’s path II

case of the dilatational wave contribution. We denote this Cagniard’s path as the
path I shown in Fig. 6.6.
On the other hand, when the saddle point is greater in magnitude than the
branch point of the dilatational wave, i.e. the inequality of Eq. (6.2.45) holds, the
Cagniard’s path crosses the branch cut. The Cagniard’s path has to be deformed
along the imaginary axis, so that the closed loop does not include any branch
points and cuts. This deformed Cagniard’s path is shown as the path II in Fig. 6.7,
where the regular Cagniard’s path is deformed by two short lines along the dila-
tational branch line and a small circle around the dilatational branch point. Thus,
the closed loop for this case is composed of the line along the real axis, the
deformed Cagniard’s path II, and two large arcs. Then, two closed loops are
considered for the complex integral U and its choice depends on the inequality in
Eq. (6.2.45).
When the inequality of Eq. (6.2.45) does not hold, we employ the closed loop
with the Cagniard’s path I, and the Cauchy’s integral theorem is applied to the
complex integral U in Eq. (6.2.41). The integral along the real axis is converted to
that along the Cagniard’s path I, since no singular point is included in the loop and
the integral along the large arc vanishes. That is,
ðsÞ 1 ðsÞ
Ixx ðx; y; sÞ ¼ Fxx ð1Þ exp ðs=cd Þ y 12 þ c2 þ i1x d1
2p
1
Z1 ð þÞ ð Þ
1 ðsÞ ðþÞ d1s ðsÞ ðÞ d1s
¼ Fxx 1s Fxx 1s expðstÞdt
x2 þy2 =cs
ð6:2:46Þ
Then, the Laplace inversion is carried out by inspection and the double inversion
for Ixx
ðsÞ
is given by
ðsÞ
Ixx ðx; y; tÞ
!
1 ðþ Þ ðÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi cj x j
ðs Þ ðþÞ d1s ðsÞ ðÞ d1s
¼ Fxx 1s Fxx 1s H t x þ y =cs H 1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
2p dt dt x2 þ y2
ð6:2:47Þ
where the last step function is the conditional for the inequality.
When the inequality of Eq. (6.2.45) holds, we have to employ the loop with the
Cagniard’s path II. In addition to the regular hyperbolic path, the deformed
Cagniard’s path II has one small circle around the branch point 1 ¼ i and two
short lines along the branch cut. We perform the complex integral along the closed
loop shown in Fig. 6.7. The integral around the small circle vanishes as its radius
tends to zero, but the line integral along the branch cut must be discussed. The
radical for the dilatational wave, 12 þ 1, has different arguments, depending on
the side of the cut, but the radical 12 þ c2 for the shear wave does not change its
argument since the saddle point is smaller in magnitude than the branch points
ic. The sign and argument of the two radicals on the deformed Cagniard’s path II
are summarized in Table 6.1, where the new radicals are introduced as
pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
bd ¼ g2 1; bs ¼ c2 g2 ð6:2:48Þ
Using the new notations, the complex integral U in Eq. (6.2.41) is carried out
along the closed loop with the deformed Cagniard’s path II. Since no singular
point is included in the loop and the integral along the large arc vanishes, the line
integral along the real axis is converted to that along the Cagniard’s path II. Since
the deformed Cagniard path is composed of two lines along the imaginary axis
with different argument for the radicals and the regular Cagniard’s path, the
Fourier inversion integral, i.e. the integral along the real axis, is converted to the
three integrals
ðsÞ 1 ðsÞ
2p
1
Z1 ( )
ðþÞ ðÞ
¼ Fxx ð1s Þ Fxx ð1s Þ expðstÞdt
x2 þy2 =cs
Z n pffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
1 ðsÞ
þ Fxx ð1Þ exp ðs=cd Þ y 12 þ c2 þ i1x d1
2p
AB
1 ðsÞ
2p
DE
ð6:2:49Þ
The last two integrals along the branch line are rearranged further. With use of
Table 6.1, the two branch line integrals are unified as
Table 6.1 Argument and magnitude of the radical and Rayleigh functions
Radicals Path AB: 1 ¼ ig Path DE: 1 ¼ ig
pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi
12 þ 1 þi g2 1 ¼ þibd i g2 1 ¼ ibd
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
12 þ c2 þ c2 g2 ¼ þbs þ c2 g2 ¼ þbs
Rð1Þ ðc2 2g2 Þ2 þ 4ig2 bd bs ðc2 2g2 Þ2 4ig2 bd bs
pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
bd ¼ g2 1; bs ¼ c2 g2
1 ðsÞ
Fxx ð1Þ exp ðs=cd Þ y 12 þ c2 þ i1x d1
2p
AB
1 ðsÞ
2p ð6:2:50Þ
DE
ffiffiffiffiffiffiffi
p cx
Zx2 þy2
1
¼ fxxðsÞ ðgÞ expfðs=cd Þðbs y þ xgÞgdg
2p
1
where the new integrand is
8g2 ðc2 2g2 Þbd b2s

fxxðsÞ ðgÞ ¼ ð6:2:51Þ
ðc2 2g2 Þ4 þ 16g4 b2d b2s
In order to transform the last integral in Eq. (6.2.50) to the form of a Laplace
transform integral, we introduce a variable transform from g to the new variable t
as
t ¼ ðbs y þ xgÞ=cd ð6:2:52Þ
Solving for g, we have
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
xðcd tÞ y c2 ðx2 þ y2 Þ ðcd tÞ2
gH ¼ ð6:2:53Þ
x2 þ y2
The last integral in Eq. (6.2.50) is transformed to that of the Laplace transform,
pffiffiffiffiffiffiffi
cx
Zx2 þy2
1
fxxðsÞ ðgÞ expfðs=cd Þðbs y þ xgÞgdg
2p
1
pffiffiffiffiffiffiffiffiffi ð6:2:54Þ
x2 þy2 =cs
Z
1 dgH
¼ fxxðsÞ ðgH Þ expðstÞdt
2p ffi
pffiffiffiffiffiffiffi dt
xþy 2 c 1 =cd
Finally, we have converted the Fourier inversion integral to the form of a Laplace
transform integral as
ðsÞ 1 ðsÞ
2p
1
Z1 ( )
ðþÞ ðÞ
¼ Fxx ð1s Þ Fxx ð1s Þ expðstÞdt
x2 þy2 =cs
pffiffiffiffiffiffiffiffiffi
x2 þy2 =cs
Z
1 dgH
þ fxxðsÞ ðgH Þ expðstÞdt
2p pffiffiffiffiffiffiffiffi dt
xþy c2 1 =cd
ð6:2:55Þ
Then, the Laplace inversion is carried out by inspection. The branch line integral
of Eq. (6.2.54) appears only when the conditional of Eq. (6.2.45) holds. Using the
step function for this conditional, the Laplace inversion is finally expressed by
ðsÞ
Ixx ðx; y; tÞ
!" ( )
cx 1
ðsÞ ðþÞ d1s ðsÞ ðÞ d1s
¼ H pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 Fxx ð1s Þ Fxx ð1s Þ H t x2 þ y2 =cs
x2 þ y2 2p dt dt
pffiffiffiffiffiffiffiffiffiffiffiffiffi!#
1 dg pffiffiffiffiffiffiffiffiffiffiffiffiffiffi x þ y c2 1
þ fxxðsÞ ðgH Þ H H x2 þ y2 =cs t H t
2p dt cd
ð6:2:56Þ
Comparing this expression with that of Eq. (6.2.47), we see that the above
equation includes the full conditional and thus, the final result for the double
inversions of the shear wave contribution is given by
ðsÞ
Ixx ðx; y; tÞ
( )
1
¼ Fxx ð1s Þ Fxx ð1s Þ H t x2 þ y2 =cs
2p dt dt
! pffiffiffiffiffiffiffiffiffiffiffiffiffi!
1 ðsÞ dgH cx pffiffiffiffiffiffiffiffiffiffiffiffiffiffi x þ y c2 1
þ fxx ðgH Þ H pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 H x2 þ y2 =cs t H t
2p dt 2
x þy 2 cd
ð6:2:57Þ
The same inversion technique is applied to another double inversion for the
shear wave contribution and the double inversion is expressed as
ðsÞ
Iij ðx; y; tÞ
( )
1
¼ Fij ð1s Þ Fij ð1s Þ H cs t x 2 þ y2
2p dt dt
!
1 ðsÞ dg cjxj pffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi
þ fij ðgH Þ H H pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 H x2 þ y2 cs t H cd t jxj y c2 1
2p dt x2 þ y2
ð6:2:58Þ
where the integrands are given by

pffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðsÞ ð212 þ c2 Þ 12 þ c2 ðsÞ 1 2 þ 1 1 2 þ c2
2i1
Fxx ð1Þ ¼ ; Fxy ð1Þ¼
Rð1Þ Rð1Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:2:59Þ
ðsÞ i1ð212 þ c2 Þ ðsÞ
2 2
21 1 þ 1
Fyx ð1Þ ¼ ; Fyy ð1Þ ¼
Rð1Þ Rð1Þ
8g2 ðc2 2g2 Þbd b2s 4gðc2 2g2 Þ2 bd bs

fxxðsÞ ðgÞ ¼ ; fxyðsÞ ðgÞ ¼
RH ðgÞ RH ðgÞ
ð6:2:60Þ
8g3 ðc2 2g2 Þbd bs 4g2 ðc2 2g2 Þ2 bd
fyxðsÞ ðgÞ ¼ ; fyyðsÞ ðgÞ ¼
RH ðgÞ RH ðgÞ
and
RH ðgÞ ¼ ðc2 2g2 Þ4 þ 16g4 b2d b2s ð6:2:61Þ

Green’s Dyadic
We have just inverted the two wave contributions and thus the displacement
response is given by
Px n ðdÞ ðsÞ
o P n
y ðdÞ ðsÞ
o
ux ¼ Ixx ðx; y; tÞ þ Ixx ðx; y; tÞ þ Ixy ðx; y; tÞ þ Ixy ðx; y; tÞ
l l
Px n ðdÞ o P n o ð6:2:62Þ
ðsÞ y ðdÞ ðsÞ
uy ¼ Iyx ðx; y; tÞ þ Iyx ðx; y; tÞ þ Iyy ðx; y; tÞ þ Iyy ðx; y; tÞ
l l
When we express the displacement in terms of Green’s dyadic Gij ðx; y; tÞ,
ux ¼ Px Gxx ðx; y; tÞ þ Py Gxx ðx; y; tÞ

ð6:2:63Þ
uy ¼ Px Gyx ðx; y; tÞ þ Py Gyy ðx; y; tÞ
the dyadic is given by

1 n ðdÞ ðsÞ
o
Gij ðx; y; tÞ ¼ Iij ðx; y; tÞ þ Iij ðx; y; tÞ ð6:2:64Þ
l
A more detailed expression is
Gij ðx; y; tÞ
"( )
ðþÞ ðÞ
1 ðdÞ ðþÞ d1d ðdÞ ðÞ d1d
¼ Fij ð1d Þ Fij ð1d Þ Hðcd t rÞ
2pl dt dt
( )
ðþÞ ðÞ ð6:2:65Þ
þ Fij ð1s Þ Fij ð1s Þ Hðcs t rÞ
dt dt
ðsÞ dg pffiffiffiffiffiffiffiffiffiffiffiffiffi
þ fij ðgH Þ H Hðcjxj rÞHðr cs tÞH cd t jxj y c2 1
dt
von Schmidt wave region cd t −x

y=
γ 2−1
c st =r
y = γ 2 −1 | x |
(γ | x | = r )
cd t = r
Fig. 6.8 von Schmidt wave region
where the radial distance r ¼ x2 þ y2 is introduced.
It should be noticed that the step function Hðcd t rÞ shows the disturbed
circular region produced by the dilatational wave and Hðcs t rÞ shows the region
produced by the shear wave. These two circular waves is emanating from the
source point. The triple product of the step functions
Hðcjxj rÞHðr cs tÞH cd t jxj y c2 1
gives the region disturbed by the shear wave which is induced by the precursor
dilatational wave. We call this region a ‘‘von Schmidt wave’’. Its wave region is
shown in Fig. 6.8.
6.3 3D Dynamic Lamb’s Problem
The transient response of an fully 3D elastic half space is discussed here. We take
the 3D coordinate system ðx; y; zÞ and denote the surface with z ¼ 0, the interior
with z [ 0 as shown in Fig. 6.9. The 3D deformation of the isotropic elastic solid
is governed by equations of motion,
6.3 3D Dynamic Lamb’s Problem 131
Fig. 6.9 A suddenly applied σ zz z =0

= Pzδ ( x)δ ( y ) H (t ) x
load Pi on the surface of a
semi-infinite elastic solid
σ zy z =0
= Pyδ ( x)δ ( y ) H (t ) z=0
σ zx z =0
= Pxδ ( x) δ( y ) H (t )
λ, μ, ρ
orxx oryx orzx o2 ux

þ þ ¼q 2
ox oy oz ot
orxy oryy orzy o2 uy
þ þ ¼q 2 ð6:3:1Þ
ox oy oz ot
orxz oryz orzz o2 uz
þ þ ¼q 2
ox oy oz ot
and Hooke’s law for the isotropic elastic solid,

oux ouy ouz oux
rxx ¼ k þ þ þ 2l
ox oy oz ox

oux ouy ouz ouy
ryy ¼ k þ þ þ 2l
ox oy oz oy

oux ouy ouz ouz
rzz ¼ k þ þ þ 2l
ox oy oz oz
ð6:3:2Þ
oux ouy
rxy ¼ ryx ¼ l þ
oy ox

ouz ouy
ryz ¼ rzy ¼ l þ
oy oz

oux ouz
rzx ¼ rxz ¼ l þ
oz ox
We assume that a point load is suddenly applied at the coordinate origin on the
surface z ¼ 0,
rzx jz¼0 ¼ Px dðxÞdðyÞHðtÞ

rzy z¼0 ¼ Py dðxÞdðyÞHðtÞ ð6:3:3Þ
rzz jz¼0 ¼ Pz dðxÞdðyÞHðtÞ
where Pj ; j ¼ x; y; z are components of the load. Note that the loading time
function is a Heaviside’s unit step function, not a Dirac’s delta function. The
quiescent condition at an initial time

oui
ui jt¼0 ¼ ¼ 0 ; i ¼ x; y; z ð6:3:4Þ
ot t¼0

oui oui oui
ui jpffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ ¼ ¼ ¼0;
x2 þy2 þz2 !1 ox pxffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 þy2 þz2 !1 oy pxffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 þy2 þz2 !1 oz pxffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 þy2 þz2 !1
i ¼ x; y; z
ð6:3:5Þ
are employed.
Substituting Hooke’s law of Eq. (6.3.2) into the equation of motion (6.3.1), the
displacement equations are given by
2
o2 ux o2 ux o2 ux o uy o2 uz 1 o2 ux
c2 2 þ 2 þ 2 þ ðc2 1Þ þ ¼ 2 2
ox oy oz oxoy oxoz cs ot
2 2 2 2 2
o uy 2 o uy o uy 2 o ux o uz 1 o2 uy
þ c þ þ ðc 1Þ þ ¼ ð6:3:6Þ
ox2 oy2 oz2 oxoy oyoz c2s ot2
2
o 2 uz o 2 u z 2
2 o uz 2 o ux o2 uy 1 o2 uz
þ þ c þ ðc 1Þ þ ¼
ox2 oy2 oz2 oxoz oyoz c2s ot2
The present 3D elastodynamic problem is also discussed by the integral transform
method. With use of the quiescent and convergence conditions, we apply Laplace
transform with respect to the time, defined by
Z1
0
and the double Fourier transform with respect to two space variables x and y,
defined by
Zþ1 Zþ1
~
f ðn; gÞ ¼ f ðx; yÞ expðþinx þ igyÞdxdy
1 1
ð6:3:8Þ
Zþ1 Zþ1
1 ~
f ðn; gÞ expðinx igyÞdndg
f ðx; yÞ ¼
ð2pÞ2
1 1
to the displacement equations (6.3.6) and Hooke’s law of Eq. (6.3.2). The dis-
placement equations are transformed to the coupled ordinary differential
equations,
ux n 2 2
d2 ~
o
2 ~ d ~u
2
c n þ g 2
þ ðs=c s Þ ux ngðc2 1Þ~uy inðc2 1Þ z ¼ 0

dz dz
uy n 2
d2 ~
o
2 ~ d ~uz
n þ c 2 2
g þ ðs=c s Þ
u y ngðc 2
1Þ ~

u x igðc 2
1Þ ¼ 0 ð6:3:9Þ
dz2 dz
u n
d2 ~
o d~u
d ~uy
c2 2z n2 þ g2 þ ðs=cs Þ2 ~ uz inðc2 1Þ x igðc2 1Þ
¼0
dz dz dz
The stress components to be used for the boundary condition are transformed as
follows:
1 du
~
zx ¼ x in~
~
r uz

l dz
1 d~uy

~
zy ¼
r uz
ig~
ð6:3:10Þ
l dz
1 d~ u

~
r ux igðc2 2Þ~uy
zz ¼ c2 z inðc2 2Þ~

l dz
The boundary condition on the surface is also transformed, to
1 Pj
~
zj ¼
r ; j ¼ x; y; z ð6:3:11Þ
l z¼0 ls
The general solution for the displacement equations (6.3.9) is obtained as
in
ux ¼
~
Aðn; g; sÞ expðad zÞ þ Bðn; g; sÞ expðas zÞ
ad
ig
~y ¼ Aðn; g; sÞ expðad zÞ þ Cðn; g; sÞ expðas zÞ

u ð6:3:12Þ
ad
i
~
uz ¼ Aðn; g; sÞ expðad zÞ fnBðn; g; sÞ þ gCðn; g; sÞg expðas zÞ

as
The stress components are
1 1 2
~zx ¼ 2inAðn; g; sÞ expðad zÞ

r ðas þ n2 ÞBðn; g; sÞ þ ngCðn; g; sÞ expðas zÞ
l as
1 1
r~zy ¼ 2igAðn; g; sÞ expðad zÞ ngBðn; g; sÞ þ ða2s þ n2 ÞCðn; g; sÞ expðas zÞ

l as
1 a2s þ n2 þ g2
~
¼
r Aðn; g; sÞ expðad zÞ þ 2ifnBðn; g; sÞ þ gCðn; g; sÞg expðas zÞ
l zz ad
ð6:3:13Þ
where Aðn; g; sÞ; Bðn; g; sÞ and Cðn; g; sÞ are unknown coefficients to be deter-
mined by the boundary condition. The radicals are defined by
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
aj ¼ n2 þ g2 þ ðs=cj Þ2 ; Reðaj Þ [ 0 ; j ¼ d; s ð6:3:14Þ
where the conditional Reðaj Þ [ 0 guarantees the convergence condition at infinity.

Applying the boundary condition (6.3.11) to the stresses of Eq. (6.3.13), we obtain
the following simple algebraic equations for the coefficients, Aðn; g; sÞ; Bðn; g; sÞ
and Cðn; g; sÞ,
1 2
Px
2inAðn; g; sÞ ðas þ n2 ÞBðn; g; sÞ þ ngCðn; g; sÞ ¼
as ls
1
P y
2igAðn; g; sÞ ngBðn; g; sÞ þ ða2s þ n2 ÞCðn; g; sÞ ¼ ð6:3:15Þ
as ls
a2s þ n2 þ g2 Pz
Aðn; g; sÞ þ 2ifnBðn; g; sÞ þ gCðn; g; sÞg ¼
ad ls
The coefficients are determined as
2inad as Px 2igad as Py ad ða2s þ n2 þ g2 Þ Pz

Aðn; g; sÞ ¼ ð6:3:16Þ
sRðn; g; sÞ l sRðn; g; sÞ l sRðn; g; sÞ l
2 2
n ðas þ n2 þ g2 4ad as Þ 1 Px
Bðn; g; sÞ ¼
sas Rðn; g; sÞ sas l
ð6:3:17Þ
ngða2s þ n2 þ g2 4ad as Þ Py 2inad as Pz
þ þ
sas Rðn; g; sÞ l sRðn; g; sÞ l
ngða2s þ n2 þ g2 4ad as Þ Px
Cðn; g; sÞ ¼
sas Rðn; g; sÞ l
2 2 2 2
ð6:3:18Þ
g ðas þ n þ g 4ad as Þ 1 Py 2igad as Pz
þ þ
sas Rðn; g; sÞ sas l sRðn; g; sÞ l
where the Rayleigh equation is defined by
Rðn; g; sÞ ¼ ðn2 þ g2 þ a2s Þ2 4ad as ðn2 þ g2 Þ ð6:3:19Þ

Then, the transformed displacement components are given by
Px n~
I ðdÞ ~ ðsÞ ~ ðSHÞ
o
ux ¼
~
ðn; g; z; sÞ þ
I ðn; g; z; sÞ þ I ðn; g; z; sÞ
l xx xx xx
Py ~n o
þ I ðdÞ ðn; g; z; sÞ þ ~I ðsÞ
xy ðn; g; z; sÞ ð6:3:20aÞ
l xy
Pz n~ I ðdÞ ~
ðsÞ
o
þ ðn; g; z; sÞ þ I ðn; g; z; sÞ
l xz xz
Px n~I ðdÞ ~ ðsÞ

o
uy ¼
~
ðn; g; z; sÞ þ I ðn; g; z; sÞ
l yx yx
Py ~n o
þ I ðdÞ ~
ðsÞ ~ðSHÞ ðn; g; z; sÞ
yy ðn; g; z; sÞ þ I yy ðn; g; z; sÞ þ I yy ð6:3:20bÞ
l
Pz n~ I ðdÞ ~
ðsÞ
o
þ ðn; g; z; sÞ þ I ðn; g; z; sÞ
l yz yz
P nðdÞ o
~z ¼ x ~

u I zx ðn; g; z; sÞ þ ~ I ðsÞ
zx ðn; g; z; sÞ
l
Py ~n o
þ I ðdÞ ~
ðsÞ
zy ðn; g; z; sÞ þ I zy ðn; g; z; sÞ ð6:3:20cÞ
l
Pz n~ðdÞ o
þ I zz ðn; g; z; sÞ þ ~ I ðsÞ
zz ðn; g; z; sÞ
l
where
~
I ðdÞ 2n2 as
xx ðn; g; z; sÞ ¼ expðad zÞ
sRðn; g; sÞ
~ ðsÞ
I xx n2 ða2s þ n2 þ g2 4ad as Þ
ðn; g; z; sÞ ¼ expðas zÞ
sas Rðn; g; sÞ
~ ðSHÞ
I xx 1
sas
~
I ðdÞ 2ngas
xy ðn; g; z; sÞ ¼ expðad zÞ ð6:3:21aÞ
sRðn; g; sÞ
~ ðsÞ
I xy ngða2s þ n2 þ g2 4ad as Þ
sas Rðn; g; sÞ
~
I ðdÞ inða2s þ n2 þ g2 Þ
xz ðn; g; z; sÞ ¼ expðad zÞ
sRðn; g; sÞ
~ ðsÞ
I xz 2inad as
sRðn; g; sÞ
~I ðdÞ ðn; g; z; sÞ ¼ 2ngas

yx expðad zÞ
sRðn; g; sÞ
2 2 2
~I ðsÞ ðn; g; z; sÞ ¼ ngðas þ n þ g 4ad as Þ expða zÞ
yx s
sas Rðn; g; sÞ
2
~I ðdÞ ðn; g; z; sÞ ¼ 2g as expða zÞ
yy d
sRðn; g; sÞ
2 2 2 2
~I ðsÞ ðn; g; z; sÞ ¼ g ðas þ n þ g 4ad as Þ expða zÞ ð6:3:21bÞ
yy s
sas Rðn; g; sÞ
~I ðSHÞ ðn; g; z; sÞ ¼ 1 expða zÞ
yy s
sas
2 2 2
~I ðdÞ ðn; g; z; sÞ ¼ igðas þ n þ g Þ expða zÞ
yz d
sRðn; g; sÞ
~I ðsÞ ðn; g; z; sÞ ¼ 2igad as expða zÞ
yz s
sRðn; g; sÞ
~
I ðdÞ 2inad as
zx ðn; g; z; sÞ ¼ expðad zÞ
sRðn; g; sÞ
~
I ðsÞ inða2s þ n2 þ g2 Þ
zx ðn; g; z; sÞ ¼ expðas zÞ
sRðn; g; sÞ
~
I ðdÞ 2igad as
zy ðn; g; z; sÞ ¼ expðad zÞ
sRðn; g; sÞ
ð6:3:21cÞ
~
I ðsÞ igða2s þ n2 þ g2 Þ
zy ðn; g; z; sÞ ¼ expðas zÞ
sRðn; g; sÞ
~
I ðdÞ ad ða2s þ n2 þ g2 Þ
zz ðn; g; z; sÞ ¼ expðad zÞ
sRðn; g; sÞ
2 2
~
I ðsÞ 2ad ðn þ g Þ
zz ðn; g; z; sÞ ¼ expðas zÞ
sRðn; g; sÞ
In the above equations, the superscript k in ~ I ðkÞ
ij indicates the wave type. The
superscript d stands for the dilatational wave, s for the vertically polarized shear
wave (SV-wave), and SH for the horizontally polarized shear wave (SH-wave).
The meaning of the subscripts i and j are the same as those in the 2D problem.
Note that the SH-wave contribution appears only in the horizontal components and
that no SH-wave contribution is included in the vertical component which is
normal to the surface.
Now, we shall invert each wave component separately.
(1) Inversion of the dilatational wave contribution: ~I ij ðn; g; z; sÞ

ðdÞ
As a representative, we consider the inversion of ~I xx ðn; g; z; sÞ. Its formal

ðdÞ
Fourier double inversion with respect to the parameters n and g is given by

Zþ1 Zþ1
ðdÞ 1 2n2 as
Ixx ðx; y; z; sÞ ¼ 2
expðad z inx igyÞdndg ð6:3:22Þ
ð2pÞ sRðn; g; sÞ
1 1
We introduce the variable transform from ðn; gÞ to the new variables ðp; qÞ defined
by
n ¼ ðs=rÞðxp yqÞ; g ¼ ðs=rÞðyp þ xqÞ ð6:3:23Þ
where the horizontal distance defined by r ¼ x2 þ y2 is introduced. This variable
transform corresponds to a rotation of the axes ðn; gÞ. The equivalent area element
is given by dndg ¼ s2 dpdq. The double integral in Eq. (6.3.22) is converted to
another double integral,
Zþ1 Zþ1
ðdÞ 1 2ðx2 p2 þ y2 q2 2xypqÞbs
expfsðbd z þ irpÞgdpdq
ð2prÞ Rðp; qÞ
1 1
ð6:3:24Þ
where
Rðp; qÞ ¼ ðb2s þ p2 þ q2 Þ2 4bd bs ðp2 þ q2 Þ ð6:3:25Þ

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
bj ¼ p2 þ q2 þ 1=c2j ; Reðbj Þ [ 0 ; j ¼ d; s ð6:3:26Þ
Note that the Rayleigh equation Rðp; qÞ is different from the former Rayleigh
equation Rðn; g; sÞ.
The integrand in Eq. (6.3.24) includes both odd and even functions of the new
variable q. As the integration with respect to q is carried out over the whole range
from 1 to þ1, the integral whose integrand has an odd power (or the anti-
symmetric function of q) vanishes. Eq. (6.3.24) is then slightly simplified, to
Z1 Zþ1
ðdÞ ðdÞ
Ixx ðx; y; z; sÞ ¼ dq Fxx ðp; qÞ expfsðbd z þ irpÞgdp ð6:3:27Þ
0 1
where the new notation for the integrand is introduced as
ðdÞ ðx2 p2 þ y2 q2 Þbs

Fxx ðp; qÞ ¼ ð6:3:28Þ
ðprÞ2 Rðp; qÞ
The inner integral in Eq. (6.3.27) is in a convenient form for the application of the
Cagniard-de Hoop technique; thus we apply the technique to the inner integral
with respect to the variable p. Now, we shall consider the complex integral U
whose integrand is the same as that in Eq. (6.3.27),
Z
ðdÞ
U ¼ Fxx ðp; qÞ expfsðbd z þ irpÞgdp ð6:3:29Þ
C
In order to transform the argument of the exponential function to the simple

product of the new variable t and the Laplace transform parameter s, the new
variable t is introduced as
t ¼ bd z þ irp ð6:3:30Þ
Solving for p, the Cagniard’s path is given by
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðÞ irt z t2 ðq2 þ 1=c2d ÞR2
pd ¼ ð6:3:31Þ
R2
and its gradient is
ðÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dpd zt ir t2 ðq2 þ 1=c2d ÞR2
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:3:32Þ
dt R2 t2 ðq2 þ 1=c2d ÞR2
where the radial distance R is defined by
R ¼ x2 þ y2 þ z2 ð6:3:33Þ
The integrand has four branch points, at
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p ¼ i q2 þ 1=c2d ; i q2 þ 1=c2s ð6:3:34Þ
From these branch points, four branch cuts are introduced along the imaginary axis
in the complex p-plane. Since the saddle point of the Cagniard’s path at
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
t ¼ R q2 þ 1=c2d ,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðdÞ ir
psaddle ¼ q2 þ 1=c2d ð6:3:35Þ
R
is smaller in magnitude than any of the branch points, the Cagniard’s path does not
cross any cut. In addition, the Rayleigh poles p ¼ i q2 þ 1=c2R , which are
derived from Rðp; qÞ ¼ 0, are on the imaginary axis and is greater in magnitude
than the saddle point and all branch points. Then, the closed loop C is composed of
the infinite line along the real p-axis, the Cagniard’s path, and two large arcs which
connect the Cagniard’s path to the line on the real axis.
The closed loop C is shown in Fig. 6.10. The complex integral U in
Eq. (6.3.29) is carried out along the closed loop. Since no singular point is
included in the loop and the integrals along the large arc vanishes, the integral
along the real axis is converted to that along the Cagniard’s path, i.e.
Zþ1
ðdÞ
Fxx ðp; qÞ expfsðbd z þ irpÞgdp
1
Z1 ( ) ð6:3:36Þ
ðþÞ ðÞ
ðdÞ ðþÞ dp ðdÞ ðÞ dp
¼ Fxx ðpd ; qÞ d Fxx ðpd ; qÞ d expðstÞdt
pffiffiffiffiffiffiffiffiffiffiffiffi
ffi dt dt
R 2
q þ1=cd 2
Substituting the above integral into the inner integral of Eq. (6.3.27), we have
Closed loop C and Cagniard’s path

Im ( p )
Re ( p )
−∞ +∞
pd( − ) −i q 2 + 1/ cd2
∞
−i q 2 + 1/ cs2 pd( + )
branch cut −i q 2 + 1 / cR2
Fig. 6.10 Closed loop and Cagniard’s path for the complex integral U
Z1 Z1 ( )
ðþÞ ðÞ
ðdÞ ðdÞ ðþÞ dp ðdÞ ðÞ dp
Ixx ðx; y; z; sÞ ¼ dq Fxx ðpd ; qÞ d Fxx ðpd ; qÞ d expðstÞdt
ffi dt dt
0 R 2
q þ1=cd 2
ð6:3:37Þ
As the inner integral is just in the form of a Laplace transform integral, we
exchange the order of integration so that the outer integral can be in the form of a
Laplace transform. In discussing the supporting region for the double integral as
shown in Fig. 6.11, the exchange is symbolically carried out as
ðt=RÞ2 1=c2d
Z1 Z1 Z1 Z
dq gðt; qÞ expðstÞdt¼ expðstÞdt gðt; qÞdq
0
p ffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi
2 R=c 0
R q þ1=cd d
ð6:3:38Þ
where gðt; qÞ is an arbitrary non-singular function. Then, the double integral in
Eq. (6.3.37) is converted to the Laplace transform integral
Fig. 6.11 Supporting region t

for the double integral in Supporting region
Eq. (6.3.38)
t t = R q 2 + 1/ cd2
R / cd
q = (t / R ) 2 − 1 / cd2
ðt=RÞ2 1=c2d ( )
Z1 Z ðþÞ ðÞ
ðdÞ ðdÞ ðþÞ dpd ðdÞ ðÞ dp
Ixx ðx; y; z; sÞ ¼ expðstÞdt Fxx ðpd ; qÞ Fxx ðpd ; qÞ d dq
dt dt
R=cd 0
ð6:3:39Þ
We have just arrived at the form of a Laplace transform integral, but its integrand
is a finite integral with respect to the variable q. The Laplace inversion is carried
out by inspection and it results in the following integral:
ðdÞ
Ixx ðx; y; z; tÞ ¼
ðt=RÞ2 1=c2d ( )
Z ðþÞ ðÞ
ðdÞ ðþÞ dpd ðdÞ ðÞ dpd
Hðt R=cd Þ Fxx ðpd ; qÞ Fxx ðpd ; qÞ dq
dt dt
0
ð6:3:40Þ
We can easily find that the disturbed region by the dilatational wave is given by the
operation of the step function Hðt R=cd Þ and that it is the semi-sphere R cd t
with radius cd t.The triple inversion for the dilatational wave contribution has been
carried out successfully. The other dilatational wave contributions can be inverted
by the same technique developed here. The unified expression for the dilatational
wave contribution is given by
ðdÞ
Iij ðx; y; z; tÞ ¼
ðt=RÞ2 1=c2d ( )
Z ðþÞ ðÞ
ðdÞ ðþÞ dpd ðdÞ ðÞ dpd
Hðt R=cd Þ Fij ðpd ; qÞ Fij ðpd ; qÞ dq
dt dt
0
ð6:3:41Þ
ðÞ
where pd are defined by Eq. (6.3.31). The integrands are given by
ðdÞ ðx2 p2 þ y2 q2 Þbs

Fxx ðp; qÞ ¼
ðprÞ2 Rðp; qÞ
ðdÞ xyðp2 q2 Þbs
Fxy ðp; qÞ ¼ ð6:3:42aÞ
ðprÞ2 Rðp; qÞ
ðdÞ ixpðb2s þ p2 þ q2 Þ
Fxz ðp; qÞ ¼
2p2 rRðp; qÞ
ðdÞ xyðp2 q2 Þbs

Fyx ðp; qÞ ¼
ðprÞ2 Rðp; qÞ
ðdÞ ðy2 p2 þ x2 q2 Þbs
Fyy ðp; qÞ ¼ ð6:3:42bÞ
ðprÞ2 Rðp; qÞ
ðdÞ iypðb2s þ p2 þ q2 Þ
Fyz ðp; qÞ ¼
2p2 rRðp; qÞ
ðdÞ ixpbd bs
Fzx ðp; qÞ ¼
p2 rRðp; qÞ
ðdÞ iypb b
Fzy ðp; qÞ ¼ 2 d s ð6:3:42cÞ
p rRðp; qÞ
ðb2s þ p2 þ q2 Þbd
FzzðdÞ ðp; qÞ ¼
2p2 Rðp; qÞ
(2) Inversion of SV wave contribution: ~

I ij ðn; g; z; sÞ ðsÞ
We consider the inversion of the typical SV-wave contribution,
~
I ðsÞ n2 ða2s þ n2 þ g2 4ad as Þ
xx ðn; g; z; sÞ ¼ expðas zÞ ð6:3:43Þ
sas Rðn; g; sÞ
The formal double Fourier inversion is given by
Zþ1 Zþ1
ðsÞ 1 n2 ða2s þ n2 þ g2 4ad as Þ
expðas z inx igyÞdndg
ð2pÞ sas Rðn; g; sÞ
1 1
ð6:3:44Þ
The variable transform defined by Eq. (6.3.23) is introduced. Examining the odd
and non-symmetric nature with respect to the variable q, we obtain the following
simpler form for the double integral,
Z1 Zþ1
ðsÞ ðsÞ
Ixx ðx; y; z; sÞ ¼ dq Fxx ðp; qÞ expfsðbs z þ irpÞgdp ð6:3:45Þ
0 1
where the integrand is given by
ðsÞ ðx2 p2 þ y2 q2 Þðb2s þ p2 þ q2 4bd bs Þ

Fxx ðp; qÞ ¼ ð6:3:46Þ
2ðprÞ2 bs Rðp; qÞ
Now, we consider the complex integral U whose integrand is the same as that of
the inner integral in Eq. (6.3.45),
Z
ðsÞ
U ¼ Fxx ðp; qÞ expfsðbs z þ irpÞgdp ð6:3:47Þ
C
In order to determine the closed loop C, the Cagniard’s path is examined. Intro-
ducing the time variable t as
t ¼ bs z þ irp ð6:3:48Þ
the Cagniard’s path is given by
t2 ðq2 þ 1=c2s ÞR2
irt z
¼ pðÞ
s ð6:3:49Þ
R2
and its saddle point at the time t ¼ R q2 þ 1=c2s is
rq 2
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðsÞ
psaddle ¼ i q þ 1=c2s ð6:3:50Þ
R
The integrand has the four branch points given by Eq. (6.3.34); and branch cuts are
introduced along the imaginary axis as shown in Figs. 6.12 and 6.13. Comparing
the magnitude of the branch points with that of the saddle point, we see that there
are two cases: the saddle point is on the branch cut or it is not; in other words, the
Cagniard’s path crosses the branch cut or it does not.
When the saddle point is smaller in magnitude than the branch point of the
dilatational wave,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r
q2 þ 1=c2s \ q2 þ 1=c2d ð6:3:51Þ
R
the Cagniard’s path does not cross any branch cut. Then, the closed loop C is
composed of the infinite line along the real axis, the Cagniard’s path, and two large
arcs which connect the line to the Cagniard’s path. This closed loop is similar to
that in the case of the dilatational wave and is shown in Fig. 6.12. Employing this
closed loop, we apply Cauchy’s theorem to the complex integral U in Eq. (6.3.47).
Since the loop does not include any singular point, the integral along the real axis
is converted to that along the Cagniard’s path, i.e.
r
Cagniard ’s path I: q 2 + 1/ cs2 < q 2 + 1/ cd2
R
Im ( p )
Re ( p )
−∞ Cagniard ’s path I +∞
ps( − ) −i q 2 + 1/ cd2
∞
−i q 2 + 1/ cs2 ps( + )
branch cut −i q 2 + 1 / cR2
Fig. 6.12 Closed loop and Cagniard’s path I
Zþ1
ðsÞ
Fxx ðp; qÞ expfsðbs z þ irpÞgdp
1
Z1 ( ) ð6:3:52Þ
ðþÞ ðÞ
ðsÞ ðþÞ dps ðsÞ ðÞ dps
¼ Fxx ðps ; qÞ Fxx ðps ; qÞ expðstÞdt
ffi dt dt
R 2 q þ1=cs2
When the saddle point is larger in magnitude than the dilatational wave branch
point,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r
q þ 1=cs [ q þ 1=cd ) q\ ðr=cs Þ2 ðR=cd Þ2 =z
2 2 2 2 ð6:3:53Þ
R
the Cagniard’s path crosses the branch cut. In order to avoid the crossing, the path
is deformed along the cut, such as the Cagniard’s path II in Sect. 6.2. Our
deformed Cagniard’s path is composed of two symmetric semi-hyperbolas, two
short lines along the cut, and a small circle around the dilatational wave branch
point. We name the deformed path the Cagniard path II, too. Figure 6.13 shows the
Cagniard path II and the closed loop C. The closed loop C has two large arcs
which connect the line with the Cagniard path II. Then, we apply Cauchy’s the-
orem to the complex integral U with this closed loop.
No singular point is included in the loop. The integral along the small circle
(BCD) vanishes as its radius tends to zero; and those along the large arcs (IJ and
FG) also vanish as the radius tends to infinity. Then, the integral along the real axis
(IHG) is converted to the sum of two integrals. One is that along the regular
Cagniard’s path (AJ and EF) and the other is the sum of two line integrals along
the branch cut (AB and DE). Thus the integral along the real axis is converted to
two integrations as
Zþ1
ðsÞ
1
Z1 ( )
ðþÞ ðÞ
¼ Fxx
ðps ; qÞ Fxx ðps ; qÞ expðstÞdt
ffi dt dt
R q2 þ1=c2s
0 1
Z Z
B C ðsÞ
þB
@ þ CF ðp; qÞ expfsðbs z þ irpÞgdp
A xx
! !
AB DE
ð6:3:54Þ
We have to examine the argument of the radicals along the branch cut, AB and
DE in Fig. 6.13. Due to the radiation condition Reðbj Þ [ 0, the argument of the
radical bd on AB, which is the left side of the cut, is þp=2 and that on the right
side DE is p=2. The argument of the radicals and the integration variable along
the cut are summarized as follows:
γ |x|
Cagniard ’s path II : >1
x2 + y 2
Im (ς ) Re (ς )
−∞ H +∞
I
C ε →0 G
ς = −i
B D
A E
p s( − ) p s( + ) ∞
ς = −iγ
F
J branch cut ς = −iγ R
Fig. 6.13 Closed loop and Cagniard’s path II

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
On AB: p ¼ i1 ; q2 þ 1=c2d \1\ Rr q2 þ 1=c2s
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
bd ¼ þi 12 ðq2 þ 1=c2d Þ ; bs ¼ þ q2 þ 1=c2s 12 ð6:3:55Þ
On DE: p ¼ i1 ; q2 þ 1=c2d \1\ Rr q2 þ 1=c2s
bd ¼ i 12 ðq2 þ 1=c2d Þ ; bs ¼ þ q2 þ 1=c2s 12 ð6:3:56Þ
The two line integrals along the branch cut are calculated and are unified as
follows:
0 1
Z Z
B C ðsÞ
B þ CF ðp; qÞ expfsðbs z þ irpÞgdp
@ A xx
! !
AB DE
ðr=RÞ q2 þ1=c2s
Z

¼ ðsÞ
Fxx ðp; qÞ p¼i1 expfsðcs z þ 1rÞgðid1Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffi bd ¼þicd
bs ¼cs
q2 þ1=c2d
pffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:3:57Þ
ðr=RÞ q2 þ1=c2s
Z

þ ðsÞ
Fxx ðp; qÞ p¼i1 expfsðcs z þ 1rÞgðid1Þ
2
ffi
2
bd ¼icd
bs ¼cs
q þ1=cd
ðr=RÞ q2 þ1=c2s
Z
¼ fxxðsÞ ð1; qÞ expfsðcs z þ r1Þgd1
2
ffi
2
q þ1=cd
where a new notation for the integrand is introduced as

8 9
< =
fxxðsÞ ð1; qÞ ¼ ðiÞ Fxx
ðsÞ
ðp; qÞ p¼i1 Fxx
ðsÞ
ðp; qÞ p¼i1 ð6:3:58Þ
: bd ¼icd bd ¼þicd ;
bs ¼cs bs ¼cs
and
cd ¼ 12 ðq2 þ 1=c2d Þ ; cs ¼ q2 þ 1=c2s 12 ð6:3:59Þ
In order to reduce the last integral in Eq. (6.3.57) to the form of a Laplace
transform, the variable transform from 1 to the time t is introduced as
t ¼ cs z þ 1r ð6:3:60Þ
Its inverse is
rt z ðq2 þ 1=c2s ÞR2 t2
1H ¼ ð6:3:61Þ
R2
The integral along the branch cut is thus converted to that of a Laplace transform,
ðr=RÞ q2 þ1=c2s
Z
fxxðsÞ ð1; qÞ expfsðcs z þ r1Þgd1
2
ffi
2
q þ1=cd
ffi ð6:3:62Þ
2 2
R
Zq þ1=cs
d1H
¼ fxxðsÞ ð1H ; qÞ expðstÞdt
ffi pffiffiffiffiffiffiffiffiffiffiffiffi
ffi dt
z 2 2
1=cs 1=cd þr 2 2
q þ1=cd
Substituting this equation into Eq. (6.3.54), we have the inner integral in the form
of a Laplace transform integral,
Zþ1
ðsÞ
1
Z1 ( )
ðþÞ ðÞ
ffi dt dt
R 2
q þ1=cs 2
2
ffi
2
R
Zq þ1=cs
d1H
þ fxxðsÞ ð1H ; qÞ expðstÞdt
ffi dt
z 2 2
1=cs 1=cd þr 2 2
q þ1=cd
ð6:3:63Þ
Comparing this equation with Eq. (6.3.52), we readily see that the last integral in
the above equation appears only when the conditional of Eq. (6.3.53) holds. Using
the step function, we can express the inner integral in a unified form as
Zþ1
ðsÞ
1
Z1 ( )
ðþÞ ðÞ
ffi dt dt
R 2
q þ1=cs 2
2
ffi
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R
Zq þ1=cs
d1
þH ðr=cs Þ2 ðR=cd Þ2 qz fxxðsÞ ð1H ; qÞ H expðstÞdt
ffi dt
z 2 2
1=cs 1=cd þr 2 2
q þ1=cd
ð6:3:64Þ
Fig. 6.14 Supporting region t

for the double integral in
Eq. (6.3.66) Supporting region
t t = R q 2 + 1/ cs2
R / cs
q = (t / R ) 2 − 1 / cs2
Further, we substitute the above equation into Eq. (6.3.45) to get

Z1 Z1 ( )
ðþÞ ðÞ
ðsÞ ðsÞ ðþÞ dps ðsÞ ðÞ dps
Ixx ðx; y; z; sÞ ¼ dq Fxx ðps ; qÞ Fxx ðps ; qÞ expðstÞdt
ffi dt dt
0 R 2q þ1=cs 2
2
ffi
2
Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R
Zq þ1=cs
d1H
þ H ðr=cs Þ2 ðR=cd Þ2 qz dq fxxðsÞ ð1H ; qÞ expðstÞdt
ffi dt
0 z 2 2
1=cs 1=cd þr 2 2
q þ1=cd
ð6:3:65Þ
To perform the Laplace inversion, the order of integration must be interchanged.
Discussing the supporting region for the double integral such as in Fig. 6.14, we
exchange the order of the integrations in the first double integral as
2
ðt=RÞ 2
Z1 Z1 Z1 Z 1=cs
dq gðt; qÞdt¼ dt gðt; qÞdq ð6:3:66Þ
0
2
ffi
2 R=c 0
R q þ1=cs s
The supporting region for the second double integral is also shown in Fig. 6.15.
After some examinations, we obtain the exchange formula
2
ffi
2
Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R
Zq þ1=cs
H ðr=cs Þ2 ðR=cd Þ2 qz dq hðt; qÞdt
0
2
2 2
ffi
2
z 1=cs 1=cd þr q þ1=cd
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
1
r ðtz 1=c2s 1=c2d Þ ðr=cd Þ2 R2
1=c2s 1=c2d
1
r ðtz 1=c2s 1=c2d Þ ðr=cd Þ2
ZR=cs Z z Z Z
¼ dt hðt; qÞdq þ dt hðt; qÞdq
2
ffi
2 0 R=cs
2
r=cd þz 1=cs 1=cd ðt=RÞ 1=c2s
ð6:3:67Þ
t
R2
1/ cs2 − 1/ cd2
z
t = R q 2 + 1/ cs2
t
R / cs
t
t = r / cd + z 1/ cs2 − 1/ cd2
r / cd + z 1/ cs2 − 1/ cd2
q
q1 q2
q=0
1
q= (r / cs ) 2 − ( R / cd ) 2
z
q1 = (t / R) 2 − 1/ cs2 q2 =
1
r (t − z 1/ c 2
s )
− 1/ cd2 − (r / cd ) 2
Fig. 6.15 Supporting region for the double integral in Eq. (6.3.67)
where gðt; qÞ and hðt; qÞ in Eqs. (6.3.66 and 6.3.67) are arbitrary non-singular
integrands. Applying the exchange formulas to the double integral in Eq. (6.3.65),
we have the Laplace transform integral for the Laplace transformed component,
Z 1=cs ( )
2 2
Z1 ðt=RÞ
ðþÞ ðÞ
ðsÞ ðsÞ ðþÞ dps ðsÞ ðÞ dps
Ixx ðx; y; z; sÞ ¼ expðstÞdt Fxx ðps ; qÞ Fxx ðps ; qÞ dq
dt dt
R=cs 0
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
ðtz 1=c2s 1=c2d Þ ðr=cd Þ2
1
ZR=cs r
Z
ðsÞ d1H
þ expðstÞdt fxx ð1H ; qÞ dq
ffi dt
r=cd þz 2
1=cs 1=cd 2 0
ffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
R2
1=c2s 1=c2d
1
z Z r
Z
d1H
þ expðstÞdt fxxðsÞ ð1H ; qÞ dq
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dt
R=cs 2
ðt=RÞ 1=c2s
ð6:3:68Þ
Finally, the Laplace inversion is carried out by inspection; and the SV-wave
contribution is given by
ðsÞ
Ixx ðx; y; z; tÞ
Z 1=cs ( )
2 2
ðt=RÞ
ðþÞ ðÞ
¼ Hðt R=cs Þ Fxx ðps ; qÞ Fxx ðps ; qÞ dq
dt dt
0
1
r
Z
d1H
þ H t r=cd z 1=c2s 1=c2d HðR=cs tÞ fxxðsÞ ð1H ; qÞ dq
dt
0
1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi r
Z
R2 d1H
þ H ðt R=cs ÞH 1=c2s 1=c2d t fxxðsÞ ð1H ; qÞ dq
z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dt
2
ðt=RÞ 1=c2s
ð6:3:69Þ
Examining the operation of the step functions ahead of the integral, the first term
with Hðt R=cs Þ shows a semi-spherical region disturbed by SV-wave. The
second
term which has the product of the two step functions,
H t r=cd z 1=c2s 1=c2d HðR=cs tÞ, shows a region of von-Schmidt
2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
wave. The last term with H ðt R=cs ÞH Rz 1=c2s 1=c2d t does not show a
wave nature, but appears only in the case of the 3D deformation (this non-wave
front is discussed by Gakenheimer and Miklowitz (1969)). These wave regions are
denoted by regions A and B in Fig. 6.16.
To the other SV-wave contributions, the same inversion technique is applied;
and we obtain the unified expression for SV-wave contributions as
ðsÞ
Iij ðx; y; z; tÞ
Z 1=cs ( )
2 2
ðt=RÞ
ðþÞ ðÞ
ðsÞ dps ðsÞ dps
¼ Hðt R=cs Þ Fij ðpðþÞ
s ; qÞ Fij ðpðÞ
s ; qÞ dq
dt dt
0
1
r
Z
ðsÞ d1H
þ H t r=cd z 1=c2s 1=c2d HðR=cs tÞ fij ð1H ; qÞ dq
dt
0
1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi r
Z
R2 ðsÞ d1H
þ H ðt R=cs ÞH 1=c2s 1=c2d t fij ð1H ; qÞ dq
z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dt
2
ðt=RÞ 1=c2s
ð6:3:70Þ
ðÞ
where ps and 1H are given by Eqs. (6.3.49) and (6.3.61) respectively and
Wave fronts and von Schmidt wave region
Region B Region A
ef v = t + | γ 4 − 3
| = t γ 4 −3
eu v = T
ef v = T
T 4 γ 4 − 3 = ef v ⋅ |
Fig. 6.16 Wave fronts and disturbed regions A and B
ðsÞ ðx2 p2 þ y2 q2 Þðb2s þ p2 þ q2 4bd bs Þ

Fxx ðp; qÞ ¼
ðsÞ xy ðp2 q2 Þðb2s þ p2 þ q2 4bd bs Þ
Fxy ðp; qÞ ¼ ð6:3:71aÞ
ðsÞ x ipbd bs
Fxz ðp; qÞ ¼
p2 r Rðp; qÞ
ðsÞ xyðp2 q2 Þðb2s þ p2 þ q2 4bd bs Þ

Fyx ðp; qÞ ¼
ðsÞ ðy2 p2 þ x2 q2 Þðb2s þ p2 þ q2 4bd bs Þ
Fyy ðp; qÞ ¼ ð6:3:71bÞ
ðsÞ y ipbd bs
Fyz ðp; qÞ ¼
p2 r Rðp; qÞ

ðsÞ x ip b2s þ p2 þ q2
Fzx ðp; qÞ ¼ 2
2p r Rðp; qÞ
2
ðsÞ y ip bs þ p2 þ q2
Fzy ðp; qÞ ¼ 2 ð6:3:71cÞ
2p r Rðp; qÞ
2 2
b ðp þ q Þ
FzzðsÞ ðp; qÞ ¼ d 2
p Rðp; qÞ
ðx1Þ2 ðyqÞ2
fxxðsÞ ð1; qÞ ¼ ðiÞ
2ðprÞ2 cs
( )
c2s 12 þ q2 4icd cs c2 12 þ q2 þ 4icd cs
2 2
c2s 12 þ q2 þ4icd cs ð12 q2 Þ ðc 12 þ q2 Þ 4icd cs ð12 q2 Þ
2
xy 12 þ q 2
fxyðsÞ ð1; qÞ ¼ ðiÞ
2ðpr Þ2 cs
( )
2 2
c2s 12 þ q2 þ4icd cs ð12 q2 Þ ðc2 12 þ q2 Þ 4icd cs ð12 q2 Þ
x
fxzðsÞ ð1; qÞ ¼ 2 1cd cs
p r
( )
1 1
2 þ 2
ð6:3:72aÞ
xy 12 þ q 2
fyxðsÞ ð1; qÞ ¼ ðiÞ 2
2ðpr Þ cs
( )
2 2
xy ðy1Þ2 þðxqÞ2
fyyðsÞ ð1; qÞ ¼ ðiÞ
2ðpr Þ2 cs
( )
2 2
y
fyzðsÞ ð1; qÞ ¼ 2 1cd cs
p
(r )
1 1
2 þ 2
ð6:3:72bÞ
x
fzxðsÞ ð1; qÞ ¼ ðþiÞ 2 1 c2s 12 þ q2
( 2p r )
1 1
2 2
y
fzyðsÞ ð1; qÞ ¼ ðþiÞ 2 1 c2s 12 þ q2
( 2p r )
1 1
2 2
cd ð12 q2 Þ
fzzðsÞ ð1; qÞ ¼ 2
(p )
1 1
2 þ 2
ð6:3:72cÞ
(3) Inversion of SH wave contribution: ~

I ii ðSH Þ
ðn; g; z; sÞ
The two SH-wave contributions are identical
~ I yyðSH Þ ðn; g; z; sÞ ¼ 1 expðas zÞ

I xxðSH Þ ðn; g; z; sÞ ¼ ~ ð6:3:73Þ
sas
The formal double Fourier inversion is given by
Zþ1 Zþ1
ðSH Þ 1 1
Iii ðx; y; z; sÞ ¼ expðas z inx igyÞdndg ð6:3:74Þ
ð2pÞ2 sas
1 1
Now, we introduce the variable transform defined by Eq. (6.3.23),

Z1 Zþ1 !
1 1 n qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
ðSH Þ
Iii ðx; y; z; sÞ ¼ 2 dq pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp s z p2 þ q2 þ 1=c2s þ irp dp
2p P2 þ q2 þ 1=c2s
0 1
ð6:3:75Þ
The Cagniard-de Hoop technique is applied to the inner integral. Introducing the
variable transform from p to the new variable t,
t ¼ z p2 þ q2 þ 1=c2s þ irp ð6:3:76Þ
the Cagniard’s path is given by

irt z t2 ðq2 þ 1=c2s ÞR2
pðÞ
s ¼ ð6:3:77Þ
R2
We have exact expressions for the radical and the gradient,

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
tz ir t2 ðq2 þ 1=c2s ÞR2
p2 þ q2 þ 1=c2s ðÞ ¼ ð6:3:78Þ
p¼ps R2
ðÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dps tz ir t2 ðq2 þ 1=c2s ÞR2
¼ ð6:3:79Þ
dt R2
Following the former discussion for the Cagniard-de Hoop technique, we apply
Cauchy’s integral theorem to the complex integral,
Z !
1 1 n qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
U¼ 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp s z p2 þ q2 þ 1=c2s þ irp dp
2p p2 þ q2 þ 1=c2s
C
ð6:3:80Þ
and find that the integral along the real axis is converted to that along the Cagn-
iard’s path. The double inversion integral in Eq. (6.3.75) is transformed to
Z1 Z1 !
ðSHÞ 1 1
Iii ðx; y; z; sÞ ¼ 2 dq pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi expðstÞdt
p pffiffiffiffiffiffiffiffiffiffiffiffiffi t2 ðq2 þ 1=c2s ÞR2
0 R q2 þ1=c2s
ð6:3:81Þ
The order of integration is also interchanged as
Z1 ðt=RÞ 2
Z 1=cs
2 !
ðSHÞ 1 1
Iii ðx; y; z; sÞ ¼ 2 expðstÞdt pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dq
p t2 ðq2 þ 1=c2s ÞR2
R=cs 0
ð6:3:82Þ
It is very lucky that the inner integral with respect to the variable q can be
evaluated exactly as
ðt=RÞ 2
Z 1=cs
2 !
1 p
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dq ¼ ð6:3:83Þ
2 2
t ðq þ 1=cs ÞR 2 2 2R
0
Thus, the SH-wave contribution is reduced to the simple Laplace transform

integral,
Z1
ðSHÞ 1
Iii ðx; y; z; sÞ ¼ expðstÞdt ð6:3:84Þ
2pR
R=cs
The Laplace inversion is carried out and we have the final result,
ðSHÞ 1
Iii ðx; y; z; tÞ ¼ Hðt R=cs Þ ; i ¼ x; y ð6:3:85Þ
2pR
Consequently, we have obtained the SH-wave contribution in the closed form.
Green’s Dyadic
The three wave components have just been inverted by the Cagniard-de Hoop
technique. The displacement response is expressed in terms of the Green’s dyadic
Gij ðx; y; z; tÞ as
ux ¼ Px Gxx ðx; y; z; tÞ þ Py Gxy ðx; y; z; tÞ þ Pz Gxz ðx; y; z; tÞ

uy ¼ Px Gyx ðx; y; z; tÞ þ Py Gyy ðx; y; z; tÞ þ Pz Gyz ðx; y; z; tÞ ð6:3:86Þ
uz ¼ Px Gzx ðx; y; z; tÞ þ Py Gzy ðx; y; z; tÞ þ Pz Gzz ðx; y; z; tÞ
where the dyadic components are given by

1 n ðdÞ ðsÞ ðSHÞ
o
Gxx ðx; y; z; tÞ ¼ Ixx ðx; y; z; tÞ þ Ixx ðx; y; z; tÞ þ Ixx ðx; y; z; tÞ
l
1 n ðdÞ ðsÞ
o
Gxy ðx; y; z; tÞ ¼ Ixy ðx; y; z; tÞ þ Ixy ðx; y; z; tÞ ð6:3:87aÞ
l
1 n ðdÞ ðsÞ
o
Gxz ðx; y; z; tÞ ¼ Ixz ðx; y; z; tÞ þ Ixz ðx; y; z; tÞ
l
1 n ðdÞ ðsÞ
o
Gyx ðx; y; z; tÞ ¼ Iyx ðx; y; z; tÞ þ Iyx ðx; y; z; tÞ
l
1 n ðdÞ ðsÞ ðSHÞ
o
Gyy ðx; y; z; tÞ ¼ Iyy ðx; y; z; tÞ þ Iyy ðx; y; z; tÞ þ Iyy ðx; y; z; tÞ ð6:3:87bÞ
l
1 n ðdÞ ðsÞ
o
Gyz ðx; y; z; tÞ ¼ Iyz ðx; y; z; tÞ þ Iyz ðx; y; z; tÞ
l
1 n ðdÞ ðsÞ
o
Gzx ðx; y; z; tÞ ¼ Izx ðx; y; z; tÞ þ Izx ðx; y; z; tÞ
l
1 n ðdÞ ðsÞ
o
Gzy ðx; y; z; tÞ ¼ Izy ðx; y; z; tÞ þ Izy ðx; y; z; tÞ ð6:3:87cÞ
l
1 n ðdÞ o
Gzz ðx; y; z; tÞ ¼ Izz ðx; y; z; tÞ þ IzzðsÞ ðx; y; z; tÞ
l
Exercises
(6.1) If the applied load is a semi-infinite distribution such as

p ; x[0
ryz y¼0 ¼ dðtÞ 0 ðaÞ
0 ; x\0
the transformed boundary condition of Eq. (6.1.14) is replaced with

yz ¼ p0 dþ ðnÞ
r ðbÞ
y¼0
where dþ ð:Þ is Heisenberg’s delta function. Verify the above equation (b).
(6.2) When the load is distributed uniformly in a quarter region ðx [ 0; y [ 0Þ on
the surface, how do you change the mathematical expression for the
boundary condition (6.3.3)?
References
Cagniard L (1962) Reflection and refraction of progressive seismic waves. (trans: Flinn ED, Dix
CH). McGraw-Hill, New York
De-Hoop AT (1961) The surface line source problem. Appl Sci Res B 8:349–356
Achenbach JD (1973) Wave propagation in elastic solids. North-Holland, New York
Fung YC (1970) Foundation of solid mechanics. (Japanese trans: Ohashi Y et al (1965) Bai-Fu-
Kan, Tokyo). Prentice-Hall, New Jersey
Graff KF (1975) Wave motion in elastic solids. Clarend Press, Oxford
Miklowitz J (1978) The theory of elastic waves and guides. North-Holland, New York
Gakenheimer DC, Miklowitz J (1969) Transient excitation of an elastic half space by a point load
traveling on the surface. J Appl Mech (Trans ASME, Ser E) 36:505–515
Chapter 7
Miscellaneous Green’s Functions
This last chapter presents three Green’s functions. The first and second sections
consider the 2D static Green’s dyadic for an orthotropic elastic solid, and for an
inhomogeneous elastic solid. The third section is concerned with Green’s function
for an SH wave which is reflected at a moving boundary. All the Green’s functions
are obtained by applying the method of integral transform. Especially, in the last
section which discusses wave reflection at a moving edge, a conversion formula
between two different Laplace inversion integrals is developed so that we can treat
the moving boundary problem.
7.1 2D Static Green’s Dyadic for an Orthotropic Elastic

Solid
An anisotropic elastic solid whose orthogonal Young’s moduli differ from each
other is called an ‘‘orthotropic’’ solid. In 2D in-plane deformation, Hooke’s law for
the orthotropic elastic solid is given by
2 3 2 32 3
rxx C11 C12 0 exx
4 ryy 5 ¼ 4 C12 C22 0 54 eyy 5 ð7:1:1Þ
rxy 0 0 C66 exy
where Cij are elastic moduli. The strain components are defined by

oux ouy 1 oux ouy
exx ¼ ; eyy ¼ ; exy ¼ þ ð7:1:2Þ
ox oy 2 oy ox
where we have assumed the state of plane strain. The equilibrium equations are
orxx orxy
þ ¼ Qx dðxÞdðyÞ
ox oy
ð7:1:3Þ
orxy oryy
þ ¼ Qy dðxÞdðyÞ
ox oy

158 7 Miscellaneous Green’s Functions
where Qi is the magnitude of a body force.

Substituting Hooke’s law into the equilibrium equations, we obtain the dis-
placement equations as

o2 ux C66 o2 ux C66 o2 uy
C11 2 þ þ C12 þ ¼ Qx dðxÞdðyÞ
ox 2 oy2 2 oxoy
ð7:1:4Þ
C66 o2 ux C66 o2 uy o2 uy
C12 þ þ þ C22 2 ¼ Qy dðxÞdðyÞ
2 oxoy 2 ox2 oy
Since we are concerned with the Green’s dyadic, a particular solution corre-
sponding to the body force is explored. Assuming that the medium is of infinite
extent and that the convergence condition at infinity,

pffiffiffiffiffiffiffiffiffi oui oui
ui j x2 þy2 !1 ¼ ¼ ¼ 0 ; i ¼ x; y ð7:1:5Þ
2 þy2 !1
is satisfied, we apply the double Fourier transform with respect to two space
variables as defined by
Z1 Z1
2p
1 1
Z1 Z1
2p
1 1
The displacement equations are transformed to algebraic equations for the trans-
formed displacement components,
fC11 n2 þ ðC66 =2Þg2 g~

ux þ ðC12 þ C66 =2Þng~uy ¼ Qx

ð7:1:8Þ
2 2
ðC12 þ C66 =2Þng~
ux þ fðC66 =2Þn þ C22 g g~uy ¼ Qy

Explicit expressions for the displacement components are obtained as
Qx n2 þ bg2
~
ux ¼ þ

C22 ðg2 þ p21 n2 Þðg2 þ p22 n2 Þ
Qy ðc þ 1Þng

ð7:1:9Þ
Qx ðc þ 1Þng
~
uy ¼

C22 ðg þ p21 n2 Þðg2 þ p22 n2 Þ
2
Qy an2 þ g2
þ
7.1 2D Static Green’s Dyadic for an Orthotropic Elastic Solid 159
where the eigenvalues, pj ðj ¼ 1; 2Þ, are

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffi
p1 ab cðc þ 2Þ þ 2 bc ab cðc þ 2Þ 2 bc
¼ ð7:1:10Þ
p2 4b 4b
and the elastic modulus parameters, a; b; c, are introduced as

C11 C22 C12
a¼ ; b¼ ; c¼ ð7:1:11Þ
C66 =2 C66 =2 C66 =2
Inspecting the transformed displacement in Eq. (7.1.9), we find that two inversion
formulas are necessary for the inversion. They are
~
I 1 ðn; g; a; bÞ ¼ an2 þ bg2 ~
I 2 ðn; gÞ ¼ ng
;
ðg2 þ p21 n2 Þðg2 þ p22 n2 Þ ðg2 þ p21 n2 Þðg2 þ p22 n2 Þ
ð7:1:12Þ
If we get the inversions for these two expressions, the displacement in the actual
space can be expressed as
Qx Qy
ux ¼ þ I1 ðx; y; 1; bÞ ðc þ 1ÞI2 ðx; yÞ
C22 C22
ð7:1:13Þ
Qx Qy
uy ¼ ðc þ 1ÞI2 ðx; yÞ þ I1 ðx; y; a; 1Þ
C22 C22
(1) Fourier inversion with respect to the parameter g

The formal Fourier inversion with respect to the parameter g is given by the
integrals
Zþ1
I1 ðn; y; a; bÞ ¼ 1 an2 þ bg2
expðigyÞdg
2p ðg2 þ p21 n2 Þðg2 þ p22 n2 Þ
1
ð7:1:14Þ
Zþ1
I2 ðn; yÞ ¼ 1 ng
expðigyÞdg
2p ðg þ p1 n Þðg2 þ p22 n2 Þ
2 2 2
1
These two integrals can be evaluated by the application of some formulas, but, in
order to show the application of Jordan’s lemma, we consider the complex inte-
grals in the g-plane.
Let us consider the complex integral whose integrand is the same as that of the
corresponding integral in Eq. (7.1.14),
I
1 an2 þ bg2
U1 ¼ expðigyÞdg
2p ðg2 þ p21 n2 Þðg2 þ p22 n2 Þ
L
I ð7:1:15Þ
1 ng
U2 ¼ expðigyÞdg
2p ðg þ p1 n Þðg2 þ p22 n2 Þ
2 2 2
L
The integrand has four simple poles at g ¼ ipj jnj; ðj ¼ 1; 2Þ as shown in

Fig. 7.1. The closed loop L is composed of a straight line along the real axis and a
semi-circle with infinite radius. Due to the convergence on the semi-circle, we
have to choose the lower loop when y [ 0 and the upper loop when y\0. Since no
other singularity exists in the loop, the integral along the real axis is converted to
the sum of the residues at the two poles. Thus, we have for Ij ; j ¼ 1; 2
I1 ðn; y; a; bÞ ¼ a bp21 1 a bp22 1

2 2
exp ð p 1 jnjjyj Þ þ 2 2
expðp2 jnjjyjÞ
2p1 ðp1 p2 Þ jnj 2p2 ðp1 p2 Þ jnj

I2 ðn; yÞ ¼ þ sgnðyÞ i i
expðp1 jnjjyjÞ expðp2 jnjjyjÞ
2 2
2ðp1 p2 Þ n n
ð7:1:16Þ
where the sign function is defined as

þ1; y [ 0
sgnðyÞ ¼ ð7:1:17Þ
1; y\0
Fig. 7.1 Closed loop L for Im(ξ )

Fourier inversion integral
η = +ip2 | ξ |
η = +ip1 | ξ |
y<0
Re(ξ )
y>0 η = −ip1 | ξ |
η = −ip2 | ξ |
(2) Fourier inversion with respect to the parameter n

Applying the formal Fourier inversion integral with respect to the parameter n
to Eqs. (7.1.16), we have
Zþ1
a bp21 1 1
I1 ðx; y; a; bÞ ¼ 2 2
expðp1 jnjjyj inxÞdn
2p1 ðp1 p2 Þ 2p jnj
1
Zþ1
a bp22 1 1
þ 2 2
expðp2 jnjjyj inxÞdn
2p2 ðp1 p2 Þ 2p jnj
1
8 ð7:1:18Þ
Zþ1
sgnðyÞ < 1 i
I2 ðx; yÞ ¼ þ expðp1 jnjjyj inxÞdn
2ðp1 p2 Þ :2p
2 2 n
1
9
Zþ1 =
1 i
expðp2 jnjj yj inxÞn
2p n ;
1
Inspecting the above equations, the two necessary inversion integrals are extracted
as
Zþ1
1 1
I1j ¼ exp pj jnjjyj expðinxÞdn
2p jnj
1
ð7:1:19Þ
Zþ1
1 i
2p n
1
The second integral I2j is reduced to a semi-infinite integral and can be evaluated
by the application of the formula (Erdélyi 1954, p. 72, 2),
Z1
1 b
expðaxÞ sinðbxÞdx ¼ tan1 ð7:1:20Þ
x a
0
That is
Zþ1
1 i
2p n
1
Z1
1 1 ð7:1:21Þ
¼ exp pj njyj sinðnxÞdn
p n
0

1 1 x
¼ tan
p pj jyj
The first integral I1j has a singularity at n ¼ 0, and thus this integral cannot be
evaluated in this form. The singular behavior is the same as that in the 2D static
plane problem in Chap. 3. We differentiate the integral with respect to the variable
x and jyj, respectively, to obtain
Zþ1
oI1j 1 in
¼ exp pj jnjjyj expðinxÞdn;
ox 2p jnj
1
ð7:1:22Þ
Zþ1
oI1j 1 pj jnj
¼ exp pj jnjjyj expðinxÞdn;
ojyj 2p jnj
1
Reducing the above integrals to semi-infinite ones, we get

Z1
oI1j 1
¼ exp pj njyj sinðnxÞdn;
ox p
0
ð7:1:23Þ
Z1
oI1j pj
¼ exp pj njyj cosðnxÞdn
ojyj p
0
and then we apply the formulas (Erdélyi 1954, pp. 14 and 72), to get
Z1
pj jyj
exp pj njyj cosðnxÞdn ¼
x2 þ p2j y2
0
ð7:1:24Þ
Z1
x
exp pj njyj sinðnxÞdn ¼
x2 þ p2j y2
0
The two integrals with different derivative are expressed in terms of algebraic
functions,
oI1j 1 x oI1j p2j jyj

¼ 2 ; ¼ ð7:1:25Þ
ox p x þ p2j y2 ojyj p x2 þ p2j y2
In order to obtain a formula for I1j , we integrate the derivatives with respect to
each space variable,
1
1

I1j ¼ log x2 þ p2j y2 þ Cx ðyÞ; I1j ¼ log x2 þ p2j y2 þ Cy ðxÞ
2p 2p
ð7:1:26Þ
Since the two expressions for I1j must be same, the integration ‘‘constant’’ func-
tions Cx and Cy must be a simple constant, without any space variable. Thus, we
have for I1j
1

I1j ¼ log x2 þ p2j y2 þ const: ð7:1:27Þ
2p
Summarizing the above discussion for the inversion integrals, we obtain the simple
formulas,
Zþ1

1 1 1
I1j ¼ exp pj jnjjyj expðinxÞdn ¼ log x2 þ p2j y2
2p jnj 2p
1
ð7:1:28Þ
Zþ1
1 i 1 1 x
I2j ¼ exp pj jnjjyj expðinxÞdn ¼ tan
2p n p pj jyj
1
where the arbitrary constant for the integral I1j is neglected since it does not
produce any stress and strain. Substituting these formulas into Eqs. (7.1.18), we
obtain closed expressions for the double Fourier inversion for I1 and I2 as
a bp21 a bp22
I1 ðx; y; a; bÞ ¼ þ 2 2
log x2 þ p21 y2 log x2 þ p22 y2
4pp1 ðp1 p2 Þ 4pp2 ðp21 p22 Þ

sgnðyÞ 1 x 1 x
I2 ðx; yÞ ¼ þ tan tan
2pðp21 p22 Þ p1 jyj p2 jyj
ð7:1:29Þ
Thus, the displacement components in Eq. (7.1.13) are expressed in the closed
form as

Qx 1 bp21 2 2 2
1 bp22 2 2 2

ux ðx; yÞ ¼ þ log x þ p1 y log x þ p2 y
4pC22 ðp21 p22 Þ p1 p2

Qy x x
ðc þ 1Þ tan1 tan1
2pC22 ðp21 p22 Þ p1 y p2 y

Qx 1 x 1 x
uy ðx; yÞ ¼ ðc þ 1Þ tan tan
2pC22 ðp21 p22 Þ p1 y p2 y

Qy a p21 2 2 2
a p22 2 2 2

þ log x þ p1 y log x þ p2 y
4pC22 ðp21 p22 Þ p1 p2
ð7:1:30Þ
Furthermore, if we introduce the notation of the Green’s dyadic Gij , the dis-
placement can be rewritten as
ux ðx; yÞ ¼Qx Gxx ðx; yÞ þ Qy Gxy ðx; yÞ

ð7:1:31Þ
uy ðx; yÞ ¼Qx Gyx ðx; yÞ þ Qy Gyy ðx; yÞ
where the dyadic components are given by


1 1 bp21 2 2 2
1 bp22 2 2 2

Gxx ðx; yÞ ¼ þ log x þ p 1 y log x þ p 2 y
4pC22 ðp21 p22 Þ p1 p2

1 x x
Gxy ðx; yÞ ¼ Gyx ðx; yÞ ¼ ðc þ 1Þ tan1 tan1
2pC22 ðp21 p22 Þ p1 y p2 y

1 a p21 2 2 2
a p22 2 2 2

Gyy ðx; yÞ ¼ þ log x þ p 1 y log x þ p 2 y
4pC22 ðp21 p22 Þ p1 p2
ð7:1:32Þ
It should be noticed that the symmetry Gxy ¼ Gyx holds in spite of the orthotropic
nature. The logarithmic singularity at the source point and at infinity is same as
that for isotropic media in Chap. 3. These singular behaviors are inherent in 2D
plane elasticity.
7.2 2D static Green’s Dyadic for an Inhomogeneous

Elastic Solid
Materials whose material parameters, such as elastic moduli and density, are
varying with space point are called ‘‘inhomogeneous’’ materials. Many mathe-
matical models for the inhomogeneity are proposed. The simplest model for the
inhomogeneity is a uniaxial exponential function for the material parameters. This
model is mathematically simple and tractable, but not fully correct since the
material parameters vanish or diverge at infinity. In spite of the unrealistic nature
of the exponential model, this type of the inhomogeneity is frequently used for
stress analysis due to its simplicity.
Let us consider the 2D plane deformation of an inhomogeneous isotropic elastic
solid. Hooke’s law for an inhomogeneous solid is the same as that for a homo-
geneous one, but its elastic moduli ðk; lÞ are functions of the space variables,
oux ouy
rxx ¼ ðk þ 2lÞ þk
ox oy
oux ouy
ryy ¼ k þ ðk þ 2lÞ ð7:2:1Þ
ox oy

oux ouy
rxy ¼ ryx ¼ l þ
oy ox
As we are concerned with Green’s dyadic corresponding to a point source, the 2D
equilibrium equations with a point body force are given by
orxx oryx
þ ¼ Bx dðxÞdðyÞ
ox oy
ð7:2:2Þ
orxy oryy
þ ¼ By dðxÞdðyÞ
ox oy
7.2 2D Static Green’s Dyadic for an Inhomogeneous Elastic Solid 165
where the body force acts at the coordinate origin ð0; 0Þ. The inhomogeneity which
we adopt here is a uniaxial exponential function and thus, Lame’s constants ðk; lÞ
are assumed to be exponential functions with one space variable y,
kðyÞ ¼ k0 expfkðy=hÞg; lðyÞ ¼ l0 expfkðy=hÞg ð7:2:3Þ
Here ðk0 ; l0 Þ are moduli at the coordinate origin, k is an inhomogeneity parameter,
and h is a reference length. Since we assume that two Lame’s constants have the
same exponential function, the Poisson ratio m is constant throughout the medium.
Substituting Hooke’s law of Eq. (7.2.1) with the inhomogeneity of Eq. (7.2.3)
into the equilibrium Eqs. (7.2.2), we obtain displacement equations with constant
coefficients,
2
2 o ux o2 ux 2 o2 uy k oux ouy Bx
c 2
þ 2 þ ðc 1Þ þ þ ¼ dðxÞdðyÞ
ox oy oxoy h oy ox lðyÞ

o2 ux o2 uy o2 uy k oux ouy By
ðc2 1Þ þ 2 þ c2 2 þ ðc2 2Þ þ c2 ¼ dðxÞdðyÞ
oxoy ox oy h ox oy lðyÞ
ð7:2:4Þ
where the constant material parameter c is defined by
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
k þ 2l k0 þ 2l0 2ð1 mÞ
c¼ ¼ ¼ ð7:2:5Þ
l l0 1 2m
In order to obtain a particular solution corresponding to the nonhomogeneous body

force term, we apply the double Fourier transform with respect to two space
variables,
Zþ1 Zþ1
f ðxÞ expðþinxÞdx; f ðxÞ ¼
2p
1 1
ð7:2:6Þ
Zþ1 Zþ1
f ðyÞ expðþigyÞdy; f ðyÞ ¼
2p
1 1
The convergence condition at infinity is also imposed on the displacement,

pffiffiffiffiffiffiffiffiffi oui oui
ui j x2 þy2 !1 ¼ ¼ ¼ 0 ; i ¼ x; y ð7:2:7Þ
2 þy2 !1
With aid of the integration formulas which include the delta function,
Zþ1 Zþ1
dðyÞ 1
dðxÞ expðþinxÞdx ¼ 1 ; expðþigyÞdy ¼ ð7:2:8Þ
lðyÞ lð0Þ
1 1
the displacement Eqs. (7.2.4) are transformed to algebraic equations for the dis-
placement components,
Bx
fc2 n2 þ g2 þ igðk=hÞg~
ux þ fðc2 1Þng þ inðk=hÞg~uy ¼

lð0Þ
ð7:2:9Þ
By
ux þ fn2 þ c2 g2 þ igc2 ðk=hÞg~uy ¼
fðc2 1Þng þ inðc2 2Þðk=hÞg~

lð0Þ
where
lð0Þ ¼ l0 ð7:2:10Þ
Solving Eq. (7.2.9) for the transformed displacement components, we get
Bx n2 þ c2 ð
g2 þ j 2 Þ By ðc2 1Þg ijðc2 3Þ
~
ux ¼ þ
n ð7:2:11aÞ
lð0Þ 2
c Dðn; gÞ lð0Þ c2 Dðn; gÞ
Bx ðc2 1Þ
g þ ijðc2 3Þ By c2 n2 þ g2 þ j2
~
uy ¼
nþ ð7:2:11bÞ
lð0Þ 2 gÞ
c Dðn; lð0Þ c2 Dðn; gÞ
where the new inhomogeneity parameter j is introduced as
j ¼ k=ð2hÞ ð7:2:12Þ
gÞ and
and the denominator Dðn; g are given by
Dðn; g2 þ j2 Þ2 þ 4ð1 2=c2 Þj2 n2

gÞ ¼ ðn2 þ ð7:2:13Þ
g ¼ g þ ij
ð7:2:14Þ
In order to apply the inversion integral, we factorize the denominator as
gÞ ¼ ðn ix1 Þðn þ ix1 Þðn ix2 Þðn þ ix2 Þ
Dðn; ð7:2:15Þ
where the eigenvalues xj ; j ¼ 1; 2 are
x1 ¼ g2 þ j2 q2 jp; x2 ¼
g2 þ j2 q2 þ jp ð7:2:16Þ
and
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p¼ 1 2=c2 ; q¼ 2ð1 1=c2 Þ ð7:2:17Þ
The formal Fourier inversion with respect to the parameter n is given by the
integrals,
Zþ1
Bx 1 n2 þ c2 ð
g2 þ j2 Þ
ux ¼ þ 2
~ expðinxÞdn
c lð0Þ 2p ðn ix1 Þðn þ ix1 Þðn ix2 Þðn þ ix2 Þ
1
Zþ1
By 1 ðc2 1Þg ijðc2 3Þ
n expðinxÞdn
c2 lð0Þ 2p ðn ix1 Þðn þ ix1 Þðn ix2 Þðn þ ix2 Þ
1
ð7:2:18aÞ
Zþ1
Bx 1 ðc2 1Þg þ ijðc2 3Þ
uy ¼ 2
~ n expðinxÞdn
1
Zþ1
By 1 c2 n 2 þ g
2 þ j2
þ 2
expðinxÞdn
1
ð7:2:18bÞ
The above integrals can be evaluated by the complex integral theory. So, we apply
Cauchy’s theorem (Jordan’s lemma) to the integral in the complex n-plane. As an
example, the first integral in Eq. (7.2.18a) is discussed. We represent it by the
complex integral Uxx ,
Z
1 n2 þ c2 ð
g2 þ j2 Þ
Uxx ¼ expðinxÞdn ð7:2:19Þ
2p ðn ix1 Þðn þ ix1 Þðn ix2 Þðn þ ix2 Þ
C
where the closed loop C is composed of a semi-circle with infinite radius and a
straight line along the real axis (see Fig. 7.2). In order to guarantee the conver-
gence on the large semi-circle, we have to employ the lower loop for x [ 0, and
the upper for x\0. The integrand has four poles; two of them are in the upper
plane and other two are in the lower plane. Each closed loop includes two poles.
Applying the Jordan’s lemma to the complex integral Uxx in Eq. (7.2.19), the
integral along the real axis is converted to the sum of two residues. When we
employ the lower loop for x [ 0, the complex integral yields
Zþ1
1 n2 þ c2 ð
g2 þ j2 Þ
Uxx ¼ expðinxÞdn
2p ðn ix1 Þðn þ ix1 Þðn ix2 Þðn þ ix2 Þ ð7:2:20Þ
1
¼2piResðix1 Þ þ 2piResðix2 Þ
The detailed calculation for the residues is
Fig. 7.2 Closed loop C for Im(ξ )

the complex integral Uxx
+iω 2
+iω1 x<0
−∞ +∞ Re(ξ )
−∞ +∞
x>0
−iω1
−iω 2
2piResðix1 Þ þ 2piResðix2 Þ

2pi n2 þ c2 ð
g2 þ j2 Þ
¼ expðinxÞ
2p ðn ix1 Þðn ix2 Þðn þ ix2 Þ n¼ix1 ð7:2:21Þ

2pi 2 2 2
n þ c ð g þj Þ2
þ expðinxÞ
2p ðn ix1 Þðn ix2 Þðn þ ix1 Þ n¼ix2
Arranging the above equation, the inversion integral along the real axis can be
evaluated exactly as
Zþ1
1 n2 þ c2 ð
g2 þ j2 Þ
expðinxÞdn
2p ðn ix1 Þðn þ ix1 Þðn ix2 Þðn þ ix2 Þ
1 ð7:2:22Þ
( )
c2 1 c2 þ 1 pffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2 2
¼ sinhðjpxÞ þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi coshðjpxÞ ex g þj q
4jp 4 2
g þj q 2 2
Similarly, the other integrals in Eq. (7.1.18) can be evaluated and we have
( )
Bx c2 þ 1 c2 1
~x ¼ þ 2
u pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi coshðjpxÞ þ sinhðjpjxjÞ exp jxj g2 þ j2 q2
4c lð0Þ g2 þ j2 q2
jp
By igðc2 1Þ þ jðc2 3Þ
þ 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sinhðjpjxjÞ exp jxj g2 þ j2 q2
4c lð0Þ jp g2 þ j2 q2
ð7:2:23aÞ
Bx igðc2
1Þ jðc2 3Þ
uy ¼ þ
~ 2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sinhðjpjxjÞ exp jxj g2 þ j2 q2
4c lð0Þjp 2
g þj q 2 2
( )
By c2 þ 1 c2 1
þ 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi coshðjpxÞ sinhðjpjxjÞ exp jxj g2 þ j2 q2
4c lð0Þ g2 þ j2 q2
jp
ð7:2:23bÞ
Our next task is to evaluate the inversion integral with respect to the parameter g.
The formal inversion of the displacement is expressed in terms of the fundamental
inversion integrals Ii ðx; yÞ; i ¼ 0; 1; 2,
2
B x c2 1 c þ1 sinhðjpjxjÞ
ux ¼ þ expðjyÞ coshðjpxÞI 0 ðx; yÞ þ I1 ðx; yÞ
4lð0Þ c2 c2 1 jp

By c2 1 sinhðjpjxjÞ c2 3
þ 2
expðjyÞ I2 ðx; yÞ þ j 2 I0 ðx; yÞ
4lð0Þ c jp c 1
ð7:2:24aÞ

Bx c2 1 sinhðjpjxjÞ c2 3
uy ¼ þ expðjyÞ I 2 ðx; yÞ j I0 ðx; yÞ
4lð0Þ c2 jp c2 1

B y c2 1 c2 þ 1 sinhðjpjxjÞ
þ expðjyÞ 2 coshðjpxÞI0 ðx; yÞ I1 ðx; yÞ
4lð0Þ c2 c 1 jp
ð7:2:24bÞ
where the fundamental integrals Ii are defined by

Zþ1 exp jxj g 2 þ j2 q2 i g y
1
I0 ðx; yÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dg ð7:2:25Þ
2p g2 þ j2 q2

1
Zþ1
1
I1 ðx; yÞ ¼ g2 þ j2 q2 igy dg
exp jxj ð7:2:26Þ
2p
1

Zþ1 exp jxj g2 þ j2 q2 igy
1
I2 ðx; yÞ ¼ g
i pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dg ð7:2:27Þ
2p g2 þ j2 q2

1
and the variable g

is defined by Eq. (7.2.14).
Observing and examining the above three equations (7.2.25–7.2.27), we find the
relation among the fundamental integrals as
oI0 ðx; yÞ oI0 ðx; yÞ
I1 ðx; yÞ ¼ ; I2 ðx; yÞ ¼ ð7:2:28Þ
ojxj oy
Since the two integrals I1 and I2 can be derived from I0 , it is enough to evaluate the
only one integral I0 . For the integral I0 in Eq. (7.2.25), we introduce the variable
transform from g to the new variable 1,
1¼
g ¼ g þ ij ð7:2:29Þ
Eq. (7.2.25) is transformed to the integral along the complex line from 1 þ ij to
þ1 þ ij,

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z
þ1þij
exp jxj 12 þ j2 q2 i1y
1
I0 ðx; yÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d1 ð7:2:30Þ
2p 12 þ j 2 q2
1þij
The above integral is very similar to the formula

Zþ1
1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp a x2 þ c2 expðibxÞdx ¼ 2K0 c a2 þ b2 ; ð7:2:31Þ
x2 þ c2
1
where K0 ð:Þ is the modified Bessel function of zeroth order. However, this formula
cannot be applied directly since the integration path for our integral of Eq. (7.2.30)
is not real. In order to apply the formula, we need to transform the integral to one
along the real path.
Let us consider the complex integral U along the rectangular closed loop
ABCDA shown in Fig. 7.3,

Z exp jxj 12 þ j2 q2 i1y
1
U¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d1 ð7:2:32Þ
2p 12 þ j2 q2
ABCDA
The integrand has two branch points at 1 ¼ ijq. Two branch cuts along the
imaginary axis are also introduced as shown in the figure. Fortunately, neither of
the branch cuts cross the integration lines DA and BC, especially the line BC since
qð¼ 1=ð1 mÞÞ [ 1. No other singular point is included in the closed loop
ABCDA and the integrals along the straight lines, AB and CD, vanish at infinity,
Im(ς )
branch cut
+iκ q
−∞ + iκ +iκ +∞ + iκ
C B
Re {ς 2
+ (κ q) 2
}> 0
D A
−∞ +∞ Re(ς )
branch cut
−iκ q
Fig. 7.3 Transform of integration path from the complex line to the real line
since the real part of the radical is positive. Thus, the integral along the complex
! !
line CB is converted to one along the real axis DA . That is

Z exp jxj 12 þ j2 q2 i1y Z exp jxj 12 þ j2 q2 i1y
1 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d1 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d1
2p 12 þ j2 q2 2p 12 þ j2 q2
! !
CB DA
ð7:2:33Þ
Consequently, the integral (7.2.30) along the complex line can be converted to that
along the real path,

Zþ1 exp jxj 12 þ j2 q2 i1y
1
I0 ðx; yÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d1 ð7:2:34Þ
2p 12 þ j2 q2
1
and we can apply the integration formula Eq. (7.2.31) to the above integral. Then,
we have the exact expression for the integral I0 ,
1
I0 ðx; yÞ ¼ K0 jjjq x2 þ y2 ð7:2:35Þ
p
Substituting this Eq. (7.2.35) into Eq. (7.2.28), the other two integrals, I1 and I2 ,
are exactly evaluated as
jjjqjxj
I1 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi K1 jjjq x2 þ y2
p x2 þ y2

pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð7:2:36Þ
jjjqy
I2 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi K1 jjjq x2 þ y2
p x2 þ y2
Finally, we substitute Eqs. (7.2.35) and (7.2.36) into Eq. (7.2.24) and rewrite the
displacement in terms of Green’s dyadic Gij ,
ux ¼Bx Gxx ðx; yÞ þ By Gxy ðx; yÞ

ð7:2:37Þ
uy ¼Bx Gyx ðx; yÞ þ By Gyy ðx; yÞ
where the exact expressions for the dyadic are
c2 1
Gxx ðx; yÞ ¼ expðjyÞ
4pc2 lð0Þ
2
c þ1 x sinhðjpxÞ
2 coshðjpxÞK0 ðjjjqr Þ þ jjjq K1 ðjjjqr Þ
c 1 r jp
ð7:2:38aÞ

c2 1 sinhðjpxÞ y c2 3
Gxy ðx; yÞ ¼ expðjyÞ jjjq K1 ð jjjqr Þ þ j K0 ðjjjqr Þ
4pc2 lð0Þ jp r c2 1
ð7:2:38bÞ

c2 1 sinhðjpxÞ y c2 3
Gyx ðx; yÞ ¼ expðjyÞ jjjq K1 ðjjjqr Þ j 2 K0 ðjjjqr Þ
4pc2 lð0Þ jp r c 1
ð7:2:38cÞ
c2 1
Gyy ðx; yÞ ¼ expðjyÞ
4pc2 lð0Þ
c2 þ 1 x sinhðjpxÞ
2 coshðjpxÞK0 ðjjjqr Þ jjjq K1 ðjjjqr Þ
c 1 r jp
ð7:2:38dÞ
Here the radial distance r from the source is defined as
r ¼ x2 þ y 2 ð7:2:39Þ
2D Kelvin’s Solution for Homogeneous Media
Inspecting the dyadic in Eq. (7.2.38a), we find that there is only one parameter for
the inhomogeneity, i. e. jð¼ k=ð2hÞÞ. When this parameter vanishes, the medium
becomes homogeneous. Thus, taking the limit j ! 0 as
sinhðjpxÞ
x; coshðjpxÞ 1 ð7:2:40Þ
jp
1
K0 ðjjjqr Þ logðjjjqr Þ; jjjqK1 ðjjjqr Þ ð7:2:41Þ
r
we have the dyadic for the homogeneous medium, i.e. the 2D Kelvin’s solution,

1 2 2
x2
Gxx ðx; yÞ ¼ c þ 1 logð r Þ þ c 1 ð7:2:42aÞ
4pc2 l0 r
c2 þ 1 xy
Gxy ðx; yÞ ¼ Gyx ðx; yÞ ¼ ð7:2:42bÞ
4pc2 l0 r 2

1 2 2
y2
Gyy ðx; yÞ ¼ c þ 1 logðr Þ þ c 1 ð7:2:42cÞ
4pc2 l0 r
where l0 is a uniform rigidity throughout the medium. The reader will find that
this dyadic is the same as the 2D static dyadic of Eq. (3.3.33) in Sect. 3.3.
7.3 Reflection of a Transient SH-Wave at a Moving Boundary 173
7.3 Reflection of a Transient SH-Wave at a Moving

Boundary
As an example of a moving boundary problem, wave reflection at a moving edge is

discussed. The simplest single SH-wave is considered and a conversion formula
between two Laplace transforms is developed so that the integral transform
method can be applicable to the moving boundary problem.
Let us consider a semi-infinite elastic solid and define Cartesian coordinates
ðx; yÞ as shown in Fig. 7.4. The initial position of a moving edge of the half space is
at x ¼ l and the edge moves toward the positive x-direction with velocity V. A point
and stationary source of SH-wave is placed at the coordinate origin and is assumed
as an impulsive body force with magnitude Q. The SH-wave field produced by the
impulsive source is governed by the nonhomogeneous displacement equation,
o2 uz o2 uz 1 o2 uz Q
þ ¼ 2 dðxÞdðyÞdðtÞ ð7:3:1Þ
ox2 oy2 c2s ot2 cs
where uz is the anti-plane displacement and cs is the SH-wave velocity. The stress
components are derived from Hooke’s law as
ouz ouz
rzx ¼ l ; rzy ¼ l ð7:3:2Þ
ox oy
where l is the shear rigidity.We assume a stress-free condition at the moving edge
x ¼ Vt þ l,
rxz jx¼Vtþl ¼ 0 ð7:3:3Þ
The radiation condition at infinity

p ffiffiffiffiffiffiffiffiffi ouz ouz
uz j x2 þy2 !1 ¼ ¼ ¼0 ð7:3:4Þ
x2 þy2 !1 oy pffiffiffiffiffiffiffiffiffi
x2 þy2 !1
Fig. 7.4 Moving edge of a y t=0 t>0

semi-infinite elastic solid
SH-source
Vt
Q
x
x=0 x=l
Moving boundary
and the quiescent condition at an initial time

ouz
uz jt¼0 ¼ ¼0 ð7:3:5Þ
ot t¼0
are also assumed to hold. Then, Eqs. (7.3.1–7.3.5) constitute the present elasto-
dynamic problem.
The wave field can be decomposed into two parts: the incident wave contri-
bution uinc and the reflected wave contribution urefl: . The total displacement (wave
field) is the sum of the two,
uz ðx; y; tÞ ¼ uinc ðx; y; tÞ þ urefl ðx; y; tÞ ð7:3:6Þ
where uinc and urefl are a particular and the general solution of the non-homoge-
neous wave Eq. (7.3.1), respectively.
(1) Incident wave
The incident wave is generated by the impulsive source and radiates from the
source point. The incident wave is given as the particular solution of the nonho-
mogeneous wave equation for the displacement,
o2 uinc o2 uinc 1 o2 uinc Q

þ ¼ 2 dðxÞdðyÞdðtÞ ð7:3:7Þ
ox2 oy2 c2s ot2 cs
The particular solution has already been obtained in Sect. 2.4, i. e.

2 2
Q H cs t x þ y
uinc ðx; y; tÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð7:3:8Þ
2pcs
ðcs tÞ2 ðx2 þ y2 Þ
However, we do not use this solution for the latter analysis, instead, the Fourier
transformed solution is employed. We apply a triple integral transform to
Eq. (7.3.7): Laplace transform with respect to the time,
Z1 Z
1
f ðsÞ ¼ f ðtÞ expðstÞdt; f ðtÞ ¼ f ðsÞ expðþstÞds ð7:3:9Þ
2pi
0 BrðsÞ
where ‘‘BrðsÞ’’ in the inverse integral denotes the Bromwich line in the complex s-
plane, and the double Fourier transform with respect to the space variables x and y,
Zþ1 Zþ1
f ðnÞ ¼ 1 f ðnÞ expðinxÞdn ð7:3:10Þ
f ðxÞ expðþinxÞdx; f ðxÞ ¼
2p
1 1
Zþ1 Zþ1
~f ðgÞ ¼ 1 ~f ðgÞ expðigyÞdg ð7:3:11Þ
f ðyÞ expðþigyÞdy; f ðyÞ ¼
2p
1 1
The nonhomogeneous wave Eq. (7.3.7) is reduced to a simple algebraic equation

for the transformed displacement,
Q
fn2 þ g2 þ ðs=cs Þ2 g~
uinc ¼
ð7:3:12Þ
c2s
and the particular solution in the transformed domain is then given by
Q 1
uinc ¼
~
ð7:3:13Þ
cs n þ g þ ðs=cs Þ2
2 2 2
Its formal Fourier inversion with respect to the parameter n is given by

Zþ1
Q 1
uinc
~ ¼ expðinxÞdn ð7:3:14Þ
2pc2s 2
n þ g2 þ ðs=cs Þ2
1
and is easily evaluated by the application of the formula (2.1.22), to give

Q
uinc ¼
~ expðas jxjÞ ð7:3:15Þ
2c2s as
where the radical as is defined by
as ¼ g2 þ ðs=cs Þ2 ; Reðas Þ [ 0 ð7:3:16Þ
Furthermore, the formal Laplace inversion for the incident wave of Eq. (7.3.15) is
given by the Bromwich integral,
Z
Q 1 1
uinc ¼ 2
~ expðas jxj þ stÞds ð7:3:17Þ
2cs 2pi as
BrðsÞ
where BrðsÞ denotes the Bromwich line from cs i1 to cs þ i1 in the complex s-

ðincÞ
plane. The transformed stress r
~xz produced by the incident wave is also obtained
in the form of a Laplace inversion integral,
Z
ðincÞ Q 1
~xz ¼ 2 sgnðxÞ
r expðas jxj þ stÞds ð7:3:18Þ
2cs 2pi
BrðsÞ
where sgn(x) is the sign function.

In the subsequent discussion for the stress-free condition, the integral form of
Eq. (7.3.18) is convenient since we develop the conversion formula for two
different Laplace transforms. Therefore, we shall stop the discussion for the
incident wave.
(2) Reflected wave
The reflected wave is the general solution of the homogeneous displacement
equation,
o2 urefl o2 urefl 1 o2 urefl

þ ¼ ð7:3:19Þ
ox2 oy2 c2s ot2
The reflected wave is generated at the moving edge and runs toward the negative
x-direction. So, we transform the Eq. (7.3.19) to the moving coordinate systems
ðz; y; tÞ where the moving coordinate z is defined as
z ¼ Vt þ l x ð7:3:20Þ
Due to this coordinate transform, the displacement of the reflected wave has a
different set of three variables, as
urefl: ðx; y; tÞ urefl: ðz; y; tÞ ð7:3:21Þ
Thus, the wave equation (7.3.19) is transformed to
o2 urefl 2M o2 urefl 1 o2 urefl o2 urefl

m2 2 þ ¼0 ð7:3:22Þ
oz2 c2 ozot cs ot2 oy2
where Mach number M and its related parameter m are defined by
M ¼ V=cs ; m ¼ 1 M 2 ð7:3:23Þ
In order to solve the differential equation (7.3.22), we apply the double integral
transform: Laplace transform with respect to the time variable in the moving
coordinate system defined by
Z1 Z
1
f ð pÞ ¼ f ðtÞ expðptÞdt; f ðtÞ ¼ f ð pÞ expðptÞdp ð7:3:24Þ
2pi
0 Br ð pÞ
and Fourier transform with respect to the space variable y,

Zþ1 Zþ1
f ð yÞ expðþigyÞdy; f ð yÞ ¼ ð7:3:25Þ
2p
1 1
It should be noted that the Laplace transform is not same as that in (7.3.9), but that
the Fourier transform is same as that in Eq. (7.3.11). This is because we have
introduced the moving coordinate system along the x-axis, not along the y-axis.
Thus, it should be understood that the time variable t in the moving coordinate
system is slightly different from the time in the fixed coordinate system since the
variable z includes the time variable.Using the quiescent condition at the initial
time and the convergence condition at y ! 1, the wave equation (7.3.22) is
transformed to the ordinary differential equation,
u
d2 ~ u
d~ n o
2
m2 refl
2 ð p=c s ÞM refl
g 2
þ ð p=c s Þ ~u
refl ¼ 0 ð7:3:26Þ
dz2 dz
and the solution which satisfies the convergence condition at z ! þ1 is given by

Mp z
urefl ¼ Aðg; pÞ exp bs þ
~ ð7:3:27Þ
mcs m
where Aðg; pÞ is an unknown coefficient and the radical bs is defined by
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2ffi
p
b s ¼ g2 þ ; Reðbs Þ [ 0 ð7:3:28Þ
mcs
Then, the formal Laplace inversion of the reflected wave is given by the Bromwich
integral,
Z
1 Mp z
urefl ¼
~ Aðg; pÞ exp bs þ expðptÞdp ð7:3:29Þ
2pi mcs m
Br ð pÞ
where Br ð pÞ denotes the Bromwich line from cp i1 to cp þ i1 in the complex

p-plane. In order to obtain the stress, we change the moving coordinate z back to
the original one, z ¼ Vt þ l x, to get
Z
1 Mp Vt þ l x
urefl ¼
~ Aðg; pÞ exp bs þ þ pt dp ð7:3:30Þ
2pi mcs m
Br ð pÞ
We then substitute this expression into the first of Eq. (7.3.2). The stress produced
by the reflected wave is thus given by
Z
ðreflÞ 1 1 Mp 1 Mp
~xz ¼
r b Aðg; pÞ exp bs þ ðVt þ l xÞ þ pt dp
2pi m s mcs m mcs
Br ð pÞ
ð7:3:31Þ
The unknown coefficient is determined by the stress-free condition on the
moving edge. The stress is also the sum of two wave contributions, the incident
and the reflected waves. Thus, the Fourier transformed boundary condition is given
by the sum of each wave contribution,
~ðxzincÞ þ r
~xz ¼ r
r ~ðxzreflÞ ¼ 0; at x ¼ Vt þ l ð7:3:32Þ
Substituting Eqs. (7.3.18) and (7.3.31) into the above condition, we obtain
Z Z
Q 1 1 1 Mp
2 expfas ðVt þ lÞ þ stgds þ b Aðg; pÞ expðptÞdp ¼ 0
2cs 2pi 2pi m s mcs
Br ðsÞ Br ð pÞ
ð7:3:33Þ
This is the integral equation for the unknown coefficient Aðg; pÞ. The second
integral on the left-hand side is in the form of a Laplace inversion integral with
respect to the parameter p, however, the first integral is not (in the form of a
Laplace inversion integral). We thus have two different forms of the Laplace
inversion integral in a single equation and it is not possible to apply any one of the
Laplace transforms to reduce it to an algebraic equation for the unknown. Thus,
the inversion integral in the first term must be converted to that with respect to the
parameter p, i.e. the matching of the inversion integral.
Let us consider the complex integral UA whose integrand is the same as that in
the first integral in Eq. (7.3.33),
Z
1
UA ¼ expðas lÞ expfðs as V Þtgds ð7:3:34Þ
2pi
C
where the radical as is defined by Eq. (7.3.16). The integration loop C is discussed
now. The integrand has two branch points s ¼ ics g and corresponding two
branch cuts are introduced along the imaginary axis in the complex s-plane. These
are shown in Fig. 7.5. If we introduce the variable transform from s to the new
variable p as
Fig. 7.5 Closed loop C for Im( s ) B

the complex integral UA in
Eq. (7.3.34) γ s + i∞ p = γ p + i∞
icsη
γs Re( s )
branch no singular point
cut inside of the loop

C
−icsη s( p)
γ s − i∞ p = γ p − i∞
A
p ¼ s as V ð7:3:35Þ
its inverse is given by
1
s=cs ¼ 2 p=cs M ðmgÞ2 þðp=cs Þ2 ð7:3:36Þ
m
In order to determine the multiple sign ðÞ, we examine Eq. (7.3.35). We have the
following asymptotic relation between two variables:
p
s ; j pj ! 1 ð7:3:37Þ
1M
This relation must be hold for the inverse of the variable transform, i. e. we have to
choose the positive sign in Eq. (7.3.36). Then, the suitable inversion for the var-
iable transform is
1
s=cs ¼ 2 p=cs þ M ðmgÞ2 þðp=cs Þ2 ð7:3:38Þ
m
When the new variable p varies along the Bromwich line Br ð pÞ in the complex p-
plane, i.e.
Br ð pÞ : cp i1 ! cp þ i1 ð7:3:39Þ
where cp is a proper constant, the inverse s given by Eq. (7.3.38) moves along the
bumped curve ACB in Fig. 7.5, but it is almost straight at infinity. At infinity, the
inverse function takes the asymptotic form
( c þi1
p
; p ! cp þ i1
s c1M
p i1
ð7:3:40Þ
1M ; p ! cp þ i1
So, if we take the real part of the Bromwich line Br ðsÞ from cs i1 to cs þ i1 as
cp
cs ¼ ð7:3:41Þ
1M
the bumped curve can be connected to the edge of the Bromwich line Br ðsÞ and
then, these two curves constitute a closed loop ACBA as shown in the figure. Thus,
we employ this closed loop C for the complex integral UA in Eq. (7.3.34). For-
tunately, the real part of the two edges is sufficiently large and the two Bromwich
lines
Br ð pÞ : cp i1\p\cp þ i1
ð7:3:42Þ
Br ðsÞ : cs i1\s\cs þ i1
have sufficiently large real parts so that all singular points of the integrand are on
the left side of the Bromwich line Br ðsÞ. Then, Cauchy’s integral theorem can be

applied to the complex integral. Since the closed loop C ACBA does not include
any singular point, the integral along Br ðsÞ is converted to a parametric integral
along the bumped curve. That is,
s þi1
cZ cp þi1
Z
1 1 ds
expðas lÞ expfðs as V Þtgds ¼ expðas lÞ expðptÞdp
2pi 2pi dp s¼sð pÞ
cs i1 cp i1
ð7:3:43Þ
where sð pÞ is the inverse given by Eq. (7.3.38).
Consequently, we were able to convert the Laplace inversion integral with
respect to the parameter s to that with respect to the parameter p and thus,
Eq. (7.3.33) is rewritten in the form of a single Laplace inversion integral with
respect to the parameter p, i. e.
Z
1 1 Mp
bs Aðg; pÞ expðptÞdp
2pi m mcs
Br ð pÞ
Z ð7:3:44Þ
Q 1 ds
¼ expðas lÞ expðptÞdp
2c2s 2pi dp s¼sð pÞ
Br ð pÞ
Both sides of the above equation have just the same form of a Laplace inversion
integral with respect to the single parameter p. Applying the Laplace transform
with respect to the time t to both sides, we obtain a simple algebraic equation for
the unknown coefficient,

1 Mp Q ds
b Aðg; pÞ ¼ 2 expðas lÞ ð7:3:45Þ
m s mcs 2cs dp s¼sð pÞ
Thus, the coefficient is determined as

Q m ds
Aðg; pÞ ¼ exp ð a s l Þ ð7:3:46Þ
2c2s bs ðMpÞ=ðmcs Þ dp s¼sð pÞ
Using the relations derived from Eqs. (7.3.35) and (7.3.38), the radical and the
gradient are given by
1
as ¼ fM ðp=cs Þ þ mbs g ð7:3:47Þ
m2
dsð pÞ M ðp=mcs Þ þ bs
¼ ð7:3:48Þ
dp m2 bs
The coefficient is then rewritten in the explicit form,

Q 1 bs þ M ðp=mcs Þ M ðp=cs Þ þ mbs
Aðg; pÞ ¼ 2 exp l ð7:3:49Þ
2cs mbs bs M ðp=mcs Þ m2
Lastly, substituting Eq. (7.3.49) into Eq. (7.3.27), we find that the double trans-
formed displacement produced by the reflected wave is given by

Q 1 bs þ M ðp=mcs Þ Mðp=cs Þ þ mbs Mp z
urefl ¼ 2
~ exp l exp bs þ
2cs mbs bs M ðp=mcs Þ m2 mcs m
ð7:3:50Þ
We have thus obtained the displacement produced by the reflected wave, but in the
transformed domain. As the incident wave has been obtained in explicit form by
Eq. (7.3.8), our next task is to perform the double inversion for the displacement in
Eq. (7.3.50). The double inversion is carried out by applying the Cagniard-de Hoop
technique. The formal Fourier inversion with respect to the parameter g is given by
Zþ1
Q 1 bs þ M ðp=mcs Þ M ðp=cs Þðl zÞ bs ðz þ lÞ
u
~ refl ¼ exp igy dg
4pc2s mbs bs M ðp=mcs Þ m2 m
1
ð7:3:51Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Since the radical bs ¼ g2 þ ðp=mcs Þ2 has the Laplace transform parameter p, we
introduce a variable transform from g to the new variable 1 as
g ¼ ðp=mcs Þ1 ð7:3:52Þ
Eq. (7.3.51) is rewritten as
Zþ1 pffiffiffiffiffiffiffiffiffiffiffiffiffi
Q 1 12 þ 1 þ M M ðl zÞ z þ l pffiffiffiffiffiffiffiffiffiffiffiffi
ffi
~u
refl ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi exp ðp=mcs Þ þ 12 þ 1 þ i1y d1
4pmc2s 12 þ 1 12 þ 1 M m m
1
ð7:3:53Þ
Now, let us consider the complex integral UB whose integrand is the same as
that of Eq. (7.3.53),
Z pffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1 12 þ 1 þ M M ðl zÞ z þ l pffiffiffiffiffiffiffiffiffiffiffiffi
ffi
UB ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi exp ðp=mcs Þ þ 12 þ 1 þ i1y d1
2p 12 þ 1 12 þ 1 M m m
C
ð7:3:54Þ
The closed loop C is discussed here. The integrand has two branch points at
1 ¼ i. Two branch cuts are thus introduced along the imaginary axis in the
complex 1-plane. The vanishing point of the denominator, 12 þ 1 M ¼ 0,
gives two poles at 1 ¼ im on the imaginary axis; but they are smaller in mag-
nitude than the branch points.The Cagniard’s path is determined by the variable
transform from 1 to the new variable t as

1 M ðl zÞ z þ 1 pffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi
t¼ þ 1 þ 1 þ i1y ð7:3:55Þ
mcs m m
Solving for 1, the Cagniard’s path is given by

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n o

imy m2 cs t M ðl zÞ ðz þ lÞ fm2 cs t M ðl zÞg2 ðz þ lÞ2 þðmyÞ2
1ðsÞ ¼
ðz þ lÞ2 þðmyÞ2
ð7:3:56Þ
The saddle point of the Cagniard’s path is
my
1saddle ¼ i qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð7:3:57Þ
at
1
t¼ M ð l z Þ þ ðz þ lÞ2 þðmyÞ2 ð7:3:58Þ
m2 cs
If we take the closed loop composed of the Cagniard’s path, the straight line
along the real axis and two large arcs which connect the straight line with the
Cagniard’s path, the pole 1 ¼ im is inside of the loop when
mj yj zþl
m\ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ) j yj [ ð7:3:59Þ
M
but outside of it when

m j yj zþl
m [ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ) j yj\ ð7:3:60Þ
M
The location of the pole depends on this inequality. The contribution from the pole
at 1 ¼ im is included only when the inequality of Eq. (7.3.59) holds. Thus, we
determine the closed loop C as ABCDOEA shown in Fig. 7.6. Applying Cauchy’s
theorem to the complex integral UB in Eq. (7.3.54), the line integral along the real
axis is converted to one along the Cagniard’s path and the residue at the pole gives
1 1 12 þ 1 þ M M ðl zÞ z þ l pffiffiffiffiffiffiffiffiffiffiffiffi ffi
pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi exp ðp=mcs Þ þ 12 þ 1 þ i1y d1
2p 12 þ 1 12 þ 1 M m m
1
2( p ffiffiffiffiffiffiffiffiffiffiffiffi
ffi ) ( pffiffiffiffiffiffiffiffiffiffiffiffiffi ) 3
Z 1
1 1
4 pffiffiffiffiffiffiffiffiffiffiffiffi 12 þ 1 þ M 1 12 þ 1 þ M 5ept dt
¼ ffi pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi
2p pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 12 þ 1 12 þ 1 M 1¼1ðþÞ 12 þ 1 12 þ 1 M 1¼1ðÞ
s s
M ðlzÞþ ðzþlÞ2 þðmyÞ2
m2 cs
" pffiffiffiffiffiffiffiffiffiffiffiffiffi #
2pi 1 þ im 12 þ 1 þ M M ðl zÞ z þ l pffiffiffiffiffiffiffiffiffiffiffiffi
ffi zþl
pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi exp ðp=mcs Þ þ 12 þ 1 þ i1y H j yj
2p 12 þ 1 12 þ 1 M m m M
1¼im
ð7:3:61Þ
Im(ς )
−iy
ς=
( z / m) 2 + y 2
Re(ς )
E O D
−∞ +∞
B
ς ( + ) (u )
ς ( − ) (u )
ς = −i
pole
ς = − im
A
y>0
C
branch cut
Fig. 7.6 Closed loop C for the complex integral UB
Rearranging the second residue term in the right hand side, we obtain the little bit
simpler form,
1 1 12 þ 1 þ M M ðl zÞ z þ l pffiffiffiffiffiffiffiffiffiffiffiffi ffi
pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi exp ðp=mcs Þ þ 12 þ 1 þ i1y d1
2p 12 þ 1 12 þ 1 M m m
1
2( p ffiffiffiffiffiffiffiffiffiffiffiffi
ffi ) ( pffiffiffiffiffiffiffiffiffiffiffiffiffi ) 3
Z 1
1 1
4 pffiffiffiffiffiffiffiffiffiffiffiffi 12 þ 1 þ M d1 1 12 þ 1 þ M d1 5ept dt
¼ ffi pffiffiffiffiffiffiffiffiffiffiffiffiffi p ffiffiffiffiffiffiffiffiffiffiffiffi
ffi p ffiffiffiffiffiffiffiffiffiffiffiffi
ffi
2p pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 12 þ 1 12 þ 1 M dt 1¼1ðþÞ 12 þ 1 12 þ 1 M dt 1¼1ðÞ
s s
M ðlzÞþ ðzþlÞ2 þðmyÞ2
m2 cs

2M p 2Ml zþl
þ exp þ my H j yj
m mcs m M
ð7:3:62Þ
Substituting this equation into Eq. (7.3.53), we obtain the Laplace transformed
displacement as the sum of the Laplace transform integral and the simple expo-
nential function,
2 pffiffiffiffiffiffiffiffi 3
2 þ1þM
p 1 ffi p1ffiffiffiffiffiffiffi
ffiffiffiffiffiffiffi ffi d1
Z1
6 12 þ1 12 þ1M dt ðþ Þ 7
Q 6 1¼1s 7 pt
u
refl ¼ 6 p ffiffiffiffiffiffiffiffi 7e dt
4pmc2s 4
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffi 1 2 þ1þM
1 ffi pffiffiffiffiffiffiffiffi d1 5
M ðlzÞþ ðzþlÞ2 þðmyÞ2 12 þ1 12 þ1M dt ðÞ
ð7:3:63Þ
m2 cs
1¼1s

MQ p 2Ml zþl
þ exp þ my H j yj
m2 c2s mcs m M
The first term is just in the form of a Laplace transform integral and is easily
inverted by inspection. The second term is the exponential function and the simple
Laplace inversion formula,
L1 fexpðapÞg ¼ dðt aÞ ð7:3:64Þ

is applied. Thus, the Laplace inversion is easily carried out and we get the final
form for the displacement produced by the reflected wave as
0 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
Q M ðl zÞ þ ðz þ lÞ2 þðmyÞ2
urefl ¼ H @t A
4pmc2s m2 cs
2( pffiffiffiffiffiffiffiffiffiffiffiffiffi ) ( pffiffiffiffiffiffiffiffiffiffiffiffiffi ) 3
1 2
1 þ 1 þ M d1 1 2
1 þ 1 þ M d1
4 pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi 5
12 þ 1 12 þ 1 M dt 1¼1ðþÞ 12 þ 1 12 þ 1 M dt 1¼1ðÞ
s s

MQ 1 2Ml zþl
þ 2 2d t þ my H jyj
m cs mcs m M
ð7:3:65Þ
Wave Fronts
Disturbed regions and wave fronts are easily obtained. They are given by the
conditionals of the step and delta functions in Eq. (7.3.65). Returning from the
moving coordinate to the original one, we find that the step function attached to the
first term in Eq. (7.3.65) gives the disturbed region by the reflected wave:
0 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1

M ðl zÞ þ ðz þ lÞ2 ðmyÞ2 2lm 2 2l 2 2
@
Wrefl : H t A ) cs t 2 [ x 2 þy
m2 cs m m
ð7:3:66Þ
This is a circular region, but not a full circle since the edge is moving and the
conditional for x is imposed as x
Mcs t þ l. This reflected wave is shown in
Fig. 7.7.
The second term included a product of delta and step functions.

1 2Ml zþl
Wflat :d t þ my H j yj
mcs m m
ð7:3:67Þ
2Ml
) d cs t 2 y H ðM j yj x Mcs t 2lÞ
m
l ⎛ 2 ⎞ y
cs t = ⎜ + 1⎟
M ⎝ m2 ⎠ M
Flat wave
B C
A Q Moving edge
cs t
2l / m 2 x
O
r = 3l / M
Reflected wave
Incident wave
xedge = 2l (1 + 1 / m 2 )
2
Fig. 7.7 Wave fronts for incident and reflected waves at the time cs t ¼ Ml m2 þ1
Examining the product, we find that the delta function gives a line singular wave
and the step function restricts its line segment BC (in the figure). Thus, we denotes
the wave as follows:
Flat wave:
2Ml
Wflat : cs t ¼ y þ ð7:3:68Þ
m2
Supporting region:
x 2l
j yj [ þ cs t ð7:3:69Þ
M
The incident wave has already been given by Eq. (7.3.8) and it is the circular
region. Its front is given by

Winc : H cs t x2 þ y2 ) ðcs tÞ2 [ x2 þ y2 ; x\M ðcs tÞ þ l ð7:3:70Þ
Figure 7.7 shows a typical wave front shape for the incident and reflected waves.
Exercise
(7.1) Decompose the integrand in Sect. 7.1, as
!
n2 1 1 1
2
2 2 ¼ 2 2

2 2 2
g þ p1 n g p 2 n 2 2 2
p1 p2 g þ p1 n g þ p22 n2
2
! ðaÞ
g2 1 p21 p22
2
2
¼ þ 2 2

g2 þ p21 n g2 p22 n 2
p1 p2 g2 þ p21 n g2 þ p22 n2
and carry out the inversion integral (7.1.14) with aid of the tabulated inte-
gration formula.
Reference
Erdélyi A (ed) (1954) Tables of integral transforms, Vol I. McGraw-Hill, New-York

Index
A path II, 125, 126, 144

Achenbach, J. D., 155 Cartesian coordinate, 43, 108
Acoustic fluid, 86 system, 43
Acoustic wave, 58, 77, 79, 83, 86 Cauchy’s (integral) theorem, 9, 19, 89, 111,
Adiabatic change, 78 113, 121, 179, 153
Amplitude function, 16, 28, 29, 33 Calculus, 6, 24
Anisotropic elastic solid, 157 Complex frequency, 18, 21, 96
Anti-plane, 45, 108, 109, 173 Complex integral, 9, 19, 30, 31, 88, 96,
deformation, 108 102–104, 112, 119, 125, 126, 137,
displacement, 45, 109, 173 159, 167, 178, 181, 182
shear load, 108 Continuity equation, 79, 81, 83, 84
Convergence condition, 4, 13, 22, 25, 26, 34,
37, 45, 55, 61, 94, 96, 101, 110, 132,
B 158, 165, 177
Beam Convolution integral, 39, 42, 53, 71, 91
1D response, 104 Critical velocity, 98
Bending rigidity, 99
Bessel function, 3, 27, 48, 49, 65, 89, 99, 100,
104, 170 D
Body force Delta function, 5, 6, 70, 96, 184
potential, 83, 86 Deflection, 101
Boundary condition, 110, 117, 133, 134, 155, divergence, 98
177 equation, 101, 93–95
Branch cut, 30, 31, 103, 125, 142, 145 response, 93
Branch line, 31, 125, 127, 128 dynamic plate, 101
Branch point, 29, 30, 103, 112, 125, 142 Deformation, 45, 60, 72, 108, 115, 149, 157,
Bromwich integral, 175, 177 164
Bromwich line, 174, 175, 177, 179 De-Hoop‘ A. T., 108
Bumped curve, 179 Denominator, 96, 119, 123, 166, 181
Density, 43, 78, 79, 81, 83, 99, 108, 116, 164
Dilatational wave, 52, 119, 120, 122, 125, 130,
C 136, 140, 142, 144
Cagniard-de Hoop technique, 107, 108 Dirac’s delta function, 5, 9, 40
Cagniard, L., 111, 120, 123, 124 Dispersive, 95
Cagniard’s, 111, 120, 123, 125, 126, 138, 142, Displacement, 45, 116, 132, 158
144, 152, 181 equation, 44, 46, 52, 54, 55, 60, 61, 70,
path, 120, 123, 138, 142, 152, 181 109, 133, 165, 176
path I, 125 Distribution, 5, 6

DOI: 10.1007/978-3-319-00879-0, Ó Springer International Publishing Switzerland 2014
188 Index
Doppler effects, 77 Green’s dyadic, 43, 44, 52, 54, 60, 70, 71, 129,
Double transform, 29, 94, 95 154, 157, 163, 164, 171
Dyadic, 129, 154, 157, 163, 165, 171 Green’s function, 11, 53
2D, 11, 33
3D, 11, 34, 36, 39, 40
E
Eigenvalues, 96, 159, 166
Elastic H
beam, 93 Hankel function, 32, 33, 54, 55, 92, 103
moduli, 164 Hankel transform, 3
modulus, 159 Heaviside’s (unit), 6
solid, 45, 115, 116, 131, 157, 164, 173 Helmholtz equation, 28
Elasticity equations, 44 Heisenberg’s delta function, 6, 8
Elastodynamic Hooke’s law, 44, 108, 109, 116, 117, 131, 157,
equations, 108 158, 164, 173
problem, 108 Hydro-static pressure, 78, 173
3D, 34, 36, 39, 42, 37, 62, 132 Static hydro-pressure, 79
Erd’elyi, A., 9, 14, 15, 27, 32, 36, 38, 49–51, Hyperbola
67, 68, 89, 94, 95, 101, 161, 162 semi, 112
Equation of motion, 108, 132 Hyperbolic curve
Equation of state, 78, 79, 81 semi, 111
Equilibrium equation, 157, 158, 164, 165
2D, 157
Ewing, W. M., 29 I
Ichimatsu, S., 9
Impulsive
F body force, 173
Fixed coordinate system, 176 point load, 9, 32, 99, 115
Fourier response, 53, 91, 95, 99
cosine series, 2 (point) source, 11, 45, 60, 173, 174
cosine transform, 33 2D response, 25, 45, 53
inversion, 6, 11, 14, 17, 18, 20, 21, 23, 26, Incident wave, 174, 175, 181, 185
31–33, 47, 50, 67, 94, 100, 107, 110, Incoming wave, 21
114, 122, 127, 141, 159, 161, 166, Inhomogeneity, 164
181 Inhomogeneous (isotropic elastic) solid, 164
sine series, 3 Inhomogeneous media, 164
sine transform, 3 Initial condition, 12, 25, 37, 93, 108, 132, 177
Fourier transform, 3 In-plane
complex, 3 deformation, 115
finite, 2 displacement, 45
double, 22, 26, 28, 45, 86, 132, 158, 165 Isotropic, 43, 164
triple, 26, 34, 37, 46, 61, 86, 99, 174
Fresnel integral, 95
Fundamental functions, 46, 51 J
Fung, Y. C., 155 Jardetzky, W. S., 29
Jordan’s lemma, 19, 96, 167
G
Gakenheimer, D. C., 149 K
Gradshteyn, I. S., 7, 9, 31, 33, 56, 59 Kelvin’s solution
Graff, K. F., 155 2D, 172
Index 189
Kernel, 1, 2, 12 Q
Kronecker’s delta, 51 Quiescent condition, 11, 12, 25, 37, 45, 61, 93,
108, 116, 132, 173
L
Lamb’s problem, 116 R
Lame’s constants, 43, 165 Radiation condition, 112, 145, 117, 174, 173
Laplace Radiation wave, 21
equation, 22, 34, 36 Rayleigh
inversion, 14, 27, 39, 66, 68, 94, 107, 121, equation/function, 118–120, 123, 134
125, 128, 147, 154, 157, 177, 178, poles, 138
183 wave, 120
transform, 1, 26, 37, 44, 87, 90, 94, 107, Reference
109, 116, 119, 121, 137, 139, 173, length, 85, 165
180 state, 78, 79, 81, 82, 86
Laplacian Reflection, 77, 173
operator, 84 Reflected wave, 174, 176, 177, 180, 183
Logarithmic singularity, 103, 164 Refraction, 77
Residue, 19, 96, 103, 160, 167, 182
Reynolds number, 85
M Rigid motion, 60
Mach number, 176, 86 Ryzhik, I. M., 7, 9, 31, 33, 56, 59
Magnus, W., 9
Miklowitz, J., 149
Moriguchi, K., 9 S
Moving Saddle point, 112, 123, 125, 126, 138, 142,
boundary, 157, 173 144, 182
coordinate, 176, 177, 184 Shear rigidity, 108, 173
edge, 173, 176 Shear wave, 52, 61, 109, 118, 119, 122, 126,
128, 136
SH-wave, 136, 152, 153, 173
N Sign function, 24, 50
Navier equations, 44 Sine integral, 101
Navier-Stokes equations, 79 Specific heat, 78
Newtonian fluid, 78 Step function, 4, 8, 40, 52, 57, 115, 128, 130,
184
Strain, 43, 45, 60, 163
O Stress, 43, 44, 78, 110, 117, 133, 163, 164,
Orthotropic, 157, 164 173, 175, 177
Out-going wave, 21, 34 Stress-free, 173, 175, 177
Subsonic, 89
SV-wave, 141, 148, 149
P
Parametric integral, 113, 180
Particle velocity, 78 T
Plane strain, 116 Time-harmonic
Plate, 93 function, 40, 53
Poisson ratio, 43, 44, 72, 165 Green’s function, 11, 53
Pole, 17, 19, 24, 29, 182 (point) load, 97
Press, F., 29 response, 21, 40, 53, 54, 71, 72, 92, 98
Pulse, 5 solution, 55
190 Index
Time-harmonic (cont.) W
source, 15, 33, 39, 40, 42, 91 Water waves, 77
vibration, 15, 39, 70, 96, 104 Watson, G. N., 9, 32, 33, 54, 89, 92, 100, 103
wave, 104 Wave equation
Titchmarsh, E. C., 9 2D, 25, 28
Triple integral transform, 26, 46, 86, 99, 174 3D, 37, 62
Transient response Wave front, 91, 115, 185
2D, 115 Wave nature, 18, 21, 95
Wave reflection, 157, 173
Wave source, 11, 79, 86
U
Udagawa, K., 9
Uniform flow, 83 Y
Young’s moduli, 157
Young’s modulus, 43
V
Velocity
gradient, 83
potential, 83–85
ratio, 116, 120
Viscosity, 78, 83
Viscous fluid, 85
von Schmidt wave, 130

Integral Transform Techniques For Green S Function

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Integral Transform Techniques For Green S Function

Enviado por

Direitos autorais:

Formatos disponíveis

Lecture Notes in Applied and Computational Mechanics 71

For further volumes:

ISSN 1613-7736 ISSN 1860-0816 (electronic)

Library of Congress Control Number: 2013940095

Ó Springer International Publishing Switzerland 2014

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

This book intends to show how to apply integral transforms to partial

transform method. In addition to the tabulated integration formulas, the inversion

Yonezawa, Japan, January 2013 Kazumi Watanabe

1 Definition of Integral Transforms and Distributions ........... 1

2 Green’s Functions for Laplace and Wave Equations . . . . . . . . . . . 11

3 Green’s Dyadic for an Isotropic Elastic Solid . . . . . . . . . . . . . . . . 43

4 Acoustic Wave in a Uniform Flow . . . . . . . . . . . . . . . . . . . . . . . . 77

4.4 Wave Radiation in a Uniform Flow . . . . . . . . . . . . . . . . . . . . . 85

5 Green’s Functions for Beam and Plate . . . . . . . . . . . . . . . . . . . . . 93

6 Cagniard-de Hoop Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

7 Miscellaneous Green’s Functions . . . . . . . . . . . . . . . . . . . . . .... 157

1.1 Integral Transforms

we call this integration pair an ‘‘integral transform.’’ The function f ðxÞ is an

K. Watanabe, Integral Transform Techniques for Green’s Function, 1

where the integration path from c i1 to c þ i1 is called the ‘‘Bromwich line.’’

where the index ‘‘n’’ is an integer.

(a3) Fourier sine series:

(b) Complex Fourier transform: f ð xÞ; x 2 ½1; 1

(c) Fourier cosine/sine transform: f ðxÞ; x 2 ð0; 1

(c2) Fourier sine transform:

(d) Hankel transform: f ð xÞ; x 2 ð0; 1

In order to guarantee the application of the integral transform, each integral

for the original function. However, if we employ the Hankel transform, it is

1.2 Distributions and Their Integration Formulas

For modeling engineering phenomena, many mathematical functions are used.

Fig. 1.1 Heaviside’s unit H ( x − a)

H ð0þ Þ ¼ þ1; H ð0 Þ ¼ 0 ð1:2:2Þ

Using this simple integration formula, we can obtain an integral representation

Fig. 1.2 Schematic (a) (b) ∞

We remember the integration formula that gives 1=2 in each semi-infinite x-

As a byproduct, we can obtain a new formula. Two operations, addition and

1.3 Comments on Inversion Techniques and Integration

Explain why the infinite sequence?

2.1 1D Impulsive Source

K. Watanabe, Integral Transform Techniques for Green’s Function, 11

and Fourier transform with respect to the space variable x : 1\x\ þ 1,

We define the Laplace transform of the unknown function as

The time-derivative term is integrated by parts,

is applied. Then, we have an ordinary differential equation for the function / in

The Fourier transform integral is applied to each term as

The integration formula for the delta function

can be applied. Then, Eq. (2.1.21) yields to

Fig. 2.1 1D expanding wave

2.2 1D Time-Harmonic Source

When the source is time-harmonic, the nonhomogeneous term in Eq. (2.1.1) is

/ðx; tÞ ¼ /# ðxÞ expðþixtÞ ð2:2:3Þ

Defining the Fourier transform of the amplitude function as

the space derivative term in Eq. (2.2.6) is integrated by parts as

The nonhomogeneous term is evaluated as

Fig. 2.3 Integration path in Im(ξ )

Applying Cauchy’s integral theorem (Jordan’s lemma) to the complex integral

Then, in the case of negative x, we have

(2) Small negative imaginary x ¼ - ie

Fig. 2.4 Integration path in Re(ξ )

Fig. 2.5 Path deformation for the positive frequency

Fig. 2.6 Path deformation for the negative frequency

We remember the integration formula that gives 1=2 in each semi-infinite x-

is applied. Then, we have an ordinary differential equation for the function / in

L1 ½expðasÞ ¼ dðt aÞ ð2:7:18Þ

L1 ½expðasÞ ¼ dðt aÞ ð3:4:34Þ