Escolar Documentos
Profissional Documentos
Cultura Documentos
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
Integral Transform
Techniques for Green’s
Function
123
Kazumi Watanabe
Department of Mechanical Engineering
Yamagata University
Yonezawa
Japan
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
xi
xii Contents
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.
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
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 Jn ðzÞ is the n-th order Bessel function of the first kind.
4 1 Definition of Integral Transforms and Distributions
f ð xÞ ) Oðjxjm Þ; m [ 1 ð1:1:13Þ
x!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.
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,
0 0 0 0
x=a x=a
6 1 Definition of Integral Transforms and Distributions
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
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
8 1 Definition of Integral Transforms and Distributions
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
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
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
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.
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
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
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
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 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 /
14 2 Green’s Functions for Laplace and Wave Equations
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
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
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.
16 2 Green’s Functions for Laplace and Wave Equations
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,
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
Zþ1
d 2 /# #
expðþinxÞdx ¼ n2 / ð2:2:9Þ
dx2
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
18 2 Green’s Functions for Laplace and Wave Equations
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.
x<0
(ϖ + iε ) / c
Re(ξ )
−(ϖ + iε ) / c
x>0
C (−)
2.2 1D Time-Harmonic Source 19
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
2p n2 ðx=cÞ2 2p n2 ðx=cÞ2
n¼x=c
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
2p n ðx=cÞ 2ðx=cÞ
1
ð2:2:19Þ
x<0
−(ϖ − iε ) / c
Im(ξ )
(ϖ − iε ) / c
x>0
C (−)
Zþ1 " #
1 P 2pi Pðn x=cÞ
expðinxÞdn ¼ expðinxÞ
2p n2 ðx=cÞ2 2p n2 ðx=cÞ2
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
2p n ðx=cÞ 2ðx=cÞ
1
Similarly, when we employ the upper loop C ðþÞ in the case of x\0,
Zþ1 " #
1 P 2pi Pðn þ x=cÞ
expðinxÞdn ¼ expðinxÞ
2p n2 ðx=cÞ2 2p n2 ðx=cÞ2
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
2p n ðx=cÞ 2ðx=cÞ
1
Unifying two Eqs. (2.2.21) and (2.2.23), we have for the frequency with the
negative imaginary part,
2.2 1D Time-Harmonic Source 21
Zþ1
1 P iP
2 2
expðinxÞdn ¼ expðixjxj=cÞ ; x ¼ - ie
2p n ðx=cÞ 2ðx=cÞ
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.
22 2 Green’s Functions for Laplace and Wave Equations
Im(ξ )
Integration path
exp(+iω t)
+ω / c +∞
−∞
Re(ξ )
−ω / c
Integration path
Im(ξ )
exp(−iω t)
+ω / c +∞
−∞
Re(ξ )
−ω / c
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
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
¼ S expðjnjjyjÞ
/ ð2:3:9Þ
2jnj
24 2 Green’s Functions for Laplace and Wave Equations
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
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
2.3 2D Static Source 25
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.
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
26 2 Green’s Functions for Laplace and Wave Equations
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
/ðnÞ ¼ /ðxÞ expðþinxÞdx; /ðxÞ ¼ /ðnÞ expðinxÞdn ð2:4:5Þ
2p
1 1
Zþ1 Zþ1
~ 1 ~
/ðgÞ ¼ /ðyÞ expðþigyÞdy; /ðyÞ ¼ /ðgÞ expðigyÞdg ð2:4:6Þ
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
2.4 2D Impulsive Source 27
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
/ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp jyj n2 þ ðs=cÞ2 ð2:4:10Þ
2 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
28 2 Green’s Functions for Laplace and Wave Equations
o2 / o 2 / 1 o2 / cP Hðct rÞ
þ ¼ PdðxÞdðyÞdðtÞ ) /¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2:4:19Þ
ox2 oy2 c2 ot2 2p
ðctÞ2 r 2
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
2.5 2D Time-Harmonic Source 29
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.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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.
30 2 Green’s Functions for Laplace and Wave Equations
In order to evaluate the inversion integral of Eq. (2.5.12), we define the integral
I as
Zþ1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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)
branch cut
Fig. 2.7 Branch cuts in the case of the negative imaginary part of the frequency
2.5 2D Time-Harmonic Source 31
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
Zþ1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
32 2 Green’s Functions for Laplace and Wave Equations
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
Zþ1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp inx y n2 ðx=cÞ2 dn
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
2.5 2D Time-Harmonic Source 33
Zþ1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp inx y n2 ðx=cÞ2 dn
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.
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
34 2 Green’s Functions for Laplace and Wave Equations
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
2.6 3D Static Source 35
Zþ1 Zþ1
1
/ðnÞ ¼ /ðxÞ expðþinxÞdx; /ðxÞ ¼ /ðnÞ expðinxÞdn ð2:6:3Þ
2p
1 1
Zþ1 Zþ1
~ 1 ~
/ðgÞ ¼ /ðyÞ expðþigyÞdy; /ðyÞ ¼ /ðgÞ expðigyÞdg ð2:6:4Þ
2p
1 1
Zþ1 Zþ1
^ 1 ^ expðifzÞdf ð2:6:5Þ
/ðfÞ ¼ /ðzÞ expðþifzÞdz; /ðzÞ ¼ /ðfÞ
2p
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 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
36 2 Green’s Functions for Laplace and Wave Equations
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
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
2.7 3D Impulsive Source 37
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
and 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
ð2:7:3Þ
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
/ðnÞ ¼ /ðxÞ expðþinxÞdx; /ðxÞ ¼ /ðnÞ expðinxÞdn ð2:7:5Þ
2p
1 1
Zþ1 Zþ1
~ 1 ~
/ðgÞ ¼ /ðyÞ expðþigyÞdy; /ðyÞ ¼ /ðgÞ expðigyÞdg ð2:7:6Þ
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
38 2 Green’s Functions for Laplace and Wave Equations
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 transform parameter ‘‘s’’ is included only in the argument of the exponential
function. We remember the simple Laplace inversion formula for the delta
function,
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,
40 2 Green’s Functions for Laplace and Wave Equations
o/ o/ o/
/jpffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
ð2:8:2Þ
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
2.8 3D Time-Harmonic Source
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
QdðxÞdðyÞ expðþixtÞ /ðx; y; tÞ ¼ iQ r ¼ x2 þ y 2
4 H0 ðxr=cÞ expðþixtÞ ;
2 2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
QdðxÞdðyÞdðzÞ expðþixtÞ /ðx; y; z; tÞ ¼ 4pR expfþixðt R=cÞg ; R ¼ x2 þ y2 þ z2
41
42 2 Green’s Functions for Laplace and Wave Equations
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
Watson GN (1966) A treatise on the theory of bessel functions. Cambridge University Press,
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,
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.
3.1 2D Impulsive Source 45
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
ox pxffiffiffiffiffiffiffiffiffi
2 þy2 !1 oy pffiffiffiffiffiffiffiffiffi
x2 þy2 !1
are assumed.
46 3 Green’s Dyadic for an Isotropic Elastic Solid
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
3.1 2D Impulsive Source 47
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
(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 >
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
48 3 Green’s Dyadic for an Isotropic Elastic Solid
8 9
Z1 >
< qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 >
=
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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 >
=
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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,
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp c x2 þ a2 cosðbxÞdx ¼ K0 a b2 þ c2 ð3:1:19Þ
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,
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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),
3.1 2D Impulsive Source 49
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),
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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.
50 3 Green’s Dyadic for an Isotropic Elastic Solid
(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
þ1 ; x [ 0
sgnðxÞ ¼ ð3:1:31Þ
1 ; x\0
The next Fourier inversion integral with respect to the parameter n
Zþ1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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)
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
3.1 2D Impulsive Source 51
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
8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi93
> >
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
52 3 Green’s Dyadic for an Isotropic Elastic Solid
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
8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi9
> >
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
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
ð3:2:4Þ
54 3 Green’s Dyadic for an Isotropic Elastic Solid
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
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ipa ð2Þ
x2 a2 expðixzÞdx ¼ þ H ðazÞ
2z 1
a
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
ð1Þ
where Hn ð:Þ is the n-th order Hankel function of the first kind, defined by
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~
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
semi-infinite integral,
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
is applied to the far right integral in Eq. (3.3.9). We have for Ixx
3.3 2D Static Source 57
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
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,
58 3 Green’s Dyadic for an Isotropic Elastic Solid
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 .
3.3 2D Static Source 59
(4) Inversion of ~
I xy
The Fourier inversion integral with respect to the parameter g is reduced to the
semi-infinite integral,
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
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
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
3.4 3D Impulsive Source 61
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
¼ 2 dðxÞdðyÞdðzÞdðtÞ
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
¼ 2 dðxÞdðyÞdðzÞdðtÞ
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
and the convergence condition at infinity,
oui pffiffiffiffiffiffiffiffiffiffiffiffiffiffi oui pffiffiffiffiffiffiffiffiffiffiffiffiffiffi oui pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ui jpxffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 þy2 þz2 !1 ¼ j ¼ j ¼ j ¼0
ox x2 þy2 þz2 !1 oy x2 þy2 þz2 !1 oz x2 þy2 þz2 !1
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 ^uj expðifzÞdf
2p
1 1
are applied to the displacement equations (3.4.2). The simple algebraic equations
for the transformed displacement are
62 3 Green’s Dyadic for an Isotropic Elastic Solid
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
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
3.4 3D Impulsive Source 63
1
I0 ðx; y; z; tÞ ¼ dðt R=cs Þ ð3:4:13Þ
4pc2s R
where the 3D radial distance is defined by
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
As for the second group of the inversion in Eq. (3.4.12), we can learn that the
necessary inversion formula is
64 3 Green’s Dyadic for an Isotropic Elastic Solid
!
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
3.4 3D Impulsive Source 65
ð3:4:25Þ
The integration formula (2.4.13)
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exp c x2 þ a2 cosðbxÞdx ¼ K0 a b2 þ c2 ð3:4:26Þ
x 2 þ a2
0
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r ¼ y2 þ z2 ð3:4:28Þ
The last Fourier inversion integral with respect to the parameter n is given by
Zþ1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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,
66 3 Green’s Dyadic for an Isotropic Elastic Solid
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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,
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
3.4 3D Impulsive Source 67
1 y2 y2
I12 ðx; y; z; tÞ ¼ 2 2 dðt R=cd Þ þ 2 2 dðt R=cs Þ
4pR cd R cs R
2
ð3:4:38Þ
3y t
þ 1 2 Hðt R=c d ÞHðR=c s tÞ
R R2
1 z2 z2
I13 ðx; y; z; tÞ ¼ 2 2 dðt R=cd Þ þ 2 2 dðt R=cs Þ
4pR cd R cs R
2
ð3:4:39Þ
3z t
þ 1 2 Hðt R=c d ÞHðR=c s tÞ
R R2
^
(2) Invesion of ~
I 21
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
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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)
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
x x2 þ b2 K1 a x2 þ b2 sinðcxÞdx
0 ð3:4:47Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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,
3.4 3D Impulsive Source 69
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
4pR c2d R2 c2s R2 R4
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
4pR c2d R2 c2s R2 R4
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
4pR c2d R2 c2s R2 R4
ð3:4:54Þ
70 3 Green’s Dyadic for an Isotropic Elastic Solid
Px xz xz 3xzt
uz ¼ dðt R=c d Þ dðt R=c s Þ þ Hðt R=cd ÞHðR=c s tÞ
4pR c2d R2 c2s R2 R4
Py yz yz 3yzt
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
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
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
3.5 3D Time-Harmonic Source 71
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
oy ox oy oz ox oy oz
ð3:5:2Þ
1 o2 uy Qy
¼ 2 dðxÞdðyÞdðzÞ expðixtÞ
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
¼ 2 dðxÞdðyÞdðzÞ expðixtÞ
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
72 3 Green’s Dyadic for an Isotropic Elastic Solid
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
3.6 3D Static Source
> >
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
i 1 xi xj ð2Þ 2xi xj cd ð2Þ
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
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
oy ox oy oz ox oy oz
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
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.
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
ox oy oz ot ox oy oz
orxz oryz orzz ovz ovz ovz ovz
þ þ þ qBz ¼ q þ vx þ vy þ vz
ox oy oz ot ox oy oz
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
¼ q vx þ vy þ vz þ
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
2
o
p l o ovx ovy ovz o vx o2vx o2vx x
e þ þ þ þl þ þ þ qB
ox 3 ox ox oy oz ox2 oy2 oz2
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
oy 3 oy ox oy oz ox2 oy oz
ð4:2:5Þ
oVy oVy oVy oVy
¼ þq0 þ Vx þ Vy þ Vz
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
¼ þq Vx þ Vy þ Vz
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
ot ox ox oy oy oz oz
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
oz 3 oz ox oy oz ox2 oy2 oz2
ð4:2:7Þ
oVz oVz oVz oVz
¼ þq0 þ Vx þ Vy þ Vz
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
¼ þq Vx þ Vy þ Vz
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
ot ox ox oy oy oz oz
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
þ þ þ þl þ 2 þ 2
ox 3 ox ox oy oz ox2 oy oz
ð4:2:11aÞ
oVx oVx oVx oVx
¼ þq0 þ Vx þ Vy þ Vz
ot ox oy oz
2
op0 l o oVx oVy oVz o V y o 2 Vy o 2 V y
þ þ þ þl þ 2 þ 2
oy 3 oy ox oy oz ox2 oy oz
ð4:2:11bÞ
oVy oVy oVy oVy
¼ þq0 þ Vx þ Vy þ Vz
ot ox oy oz
2
op0 l o oVx oVy oVz o Vz o2 Vz o2 Vz
þ þ þ þl þ 2 þ 2
oz 3 oz ox oy oz ox2 oy oz
ð4:2:11cÞ
oVz oVz oVz oVz
¼ þq0 þ Vx þ Vy þ Vz
ot ox oy oz
2
op l o ovx ovy ovz o vx o2vx o2vx x
þ þ þ þl þ þ þ q0 B
ox 3 ox ox oy oz ox2 oy2 oz2
oVx oVx oVx
¼ þq Vx þ Vy þ Vz
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
ot ox ox oy oy oz oz
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
oy 3 oy ox oy oz ox2 oy oz
oVy oVy oVy
¼ þ
q Vx þ Vy þ Vz
ox oy oz
ovy ovy oVy ovy oVy ovy oVy
þ q0 þ q0 Vx þ vx þ q0 Vy þ vy þ q0 Vz þ vz
ot ox ox oy oy oz oz
ð4:2:13bÞ
2
op l o ovx ovy ovz o vz o2vz o2vz z
þ þ þ þl þ þ þ q0 B
oz 3 oz ox oy oz ox2 oy2 oz2
oVz oVz oVz
¼ þ
q Vx þ Vy þ Vz
ox oy oz
ovz ovz oVz ovz oVz ovz oVz
þ q0 þ q0 Vx þ vx þ q0 Vy þ vy þ q0 Vz þ vz
ot ox ox oy oy oz oz
ð4:2:13cÞ
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
þ þ þ þl þ þ þ q0 B
ox 3 ox ox oy oz ox2 oy2 oz2
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
þ þ þ þl þ þ þ q0 B
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
op l o ovx ovy ovz o vz o2vz o2vz z
þ þ þ þl þ þ þ q0 B
oz 3 oz ox oy oz ox2 oy2 oz2
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
ot ox oy oz ox oy oz
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,
84 4 Acoustic Wave in a Uniform Flow
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.
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
86 4 Acoustic Wave in a Uniform Flow
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
f ðsÞ ¼ f ðtÞ expðstÞdt ð4:4:3Þ
0
Fourier transform:
Z1 Z1
f ðnÞ ¼ 1 f ðnÞ expðinxÞdn
f ðxÞ expðþinxÞdx; f ðxÞ ¼ ð4:4:4Þ
2p
0 0
Fourier transform:
Z1 Z1
~f ðgÞ ¼ 1 ~f ðgÞ expðigyÞdg
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
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 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,
88 4 Acoustic Wave in a Uniform Flow
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,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
C
−iα M x D
−iα
4.4 Wave Radiation in a Uniform Flow 89
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z1 exp jyj 1 M 2 12 þ a2
Q expðþaMx xÞ 1 x
p ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cosð1xÞd1 ð4:4:18Þ
2 1 Mx2 p 1 2 þ a2
0
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cQ2p 1 Mx2
p ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð4:4:24Þ
fð1 Mx2 Þct þ Mx xg2 fx2 þ ð1 Mx2 Þy2 g
ð4:4:26Þ
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,
Flow Vx ct
x
source
Substituting the impulsive response of Eq. (4.4.24), not the final one, we have
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
92 4 Acoustic Wave in a Uniform Flow
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
Erdélyi A (ed) (1954) Tables of integral transforms, vol I. McGraw-Hill, New York
Watson GN (1966) A treatise on the theory of bessel functions. Cambridge University Press,
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.
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,
ow
wjt¼0 ¼ ¼ 0 ð5:1:5Þ
ox t¼0
and the convergence condition at infinity,
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
f ðsÞ ¼ f ðtÞ expðstÞdt ð5:1:7Þ
0
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.
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
96 5 Green’s Functions for Beam and Plate
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
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
1 Þðn n2 Þðn n2 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
Zþ1
1 expfinðx VtÞg
dn
2p ðn nþ n
1 Þðn
þ
1 Þðn n2 Þðn n2 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
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
98 5 Green’s Functions for Beam and Plate
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
2
Q exp ð1=aÞ x X þ iX ðx VtÞ þ ixt
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 ; x Vt [ 0
4a
x X2 x X2 iX
ð5:2:9aÞ
n
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
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
and the double Fourier transform with respect to the space variables ðx; yÞ,
Zþ1 Zþ1
f ðnÞ ¼ 1 f ðnÞ expðinxÞdn
f ðxÞ expðþinxÞdx; f ðxÞ ¼ ð5:3:5Þ
2p
1 1
Zþ1 Zþ1
~f ðgÞ ¼ 1 ~f ðgÞ expðigyÞdg
f ðyÞ expðþigyÞdy; f ðyÞ ¼ ð5:3:6Þ
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
100 5 Green’s Functions for Beam and Plate
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
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
is applied to Eq. (5.3.17). The Green’s function for the dynamic plate deflection is
thus given by
2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Q r
w¼ 2
si 2
; r ¼ x2 þ y2 ð5:3:20Þ
4pb 4b t
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
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffi pffiffiffiffi
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),
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
pffiffiffiffi pffiffiffiffi
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.
104 5 Green’s Functions for Beam and Plate
C
arg( z ) = +π
z = +i ω / β
z = − ω /β
F
D E A B
arg( z ) = 0
branch cut
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
4
Q
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
106 5 Green’s Functions for Beam and Plate
References
Erdélyi A (ed) (1954) Tables of integral transforms, vol I. McGraw-Hill, New York
Watson GN (1966) A treatise on the theory of bessel functions. Cambridge University Press,
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
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.
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
μ, ρ
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
f ðsÞ ¼ f ðtÞ expðstÞdt ð6:1:8Þ
0
With aid of the quiescent and convergence conditions, the displacement equation
(6.1.6) and Hooke’s law (6.1.3) are transformed to
110 6 Cagniard-de Hoop Technique
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
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
6.1 2D Anti-Plane Deformation 111
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
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
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
Zþ1 n qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
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
114 6 Cagniard-de Hoop Technique
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,
Zþ1 n qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
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.
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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),
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,
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
σ yx = Pxδ ( x) δ (t ) x
y=0
y=0
λ , μ, ρ
y
116 6 Cagniard-de Hoop Technique
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
f ðsÞ ¼ f ðtÞ expðstÞdt ð6:2:7Þ
0
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
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
118 6 Cagniard-de Hoop Technique
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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,
120 6 Cagniard-de Hoop Technique
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’’,
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
6.2 2D In-Plane Deformation 121
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
1saddle ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:2:31Þ
x2 þ y2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
122 6 Cagniard-de Hoop Technique
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
( )
ðþÞ ðÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
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
6.2 2D In-Plane Deformation 123
where
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
The saddle point at t ¼ x2 þ y2 =cs is
ðsÞ icx
1saddle ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:2:44Þ
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
124 6 Cagniard-de Hoop Technique
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 ∞
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
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,
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
2p
1
Z1 ð þÞ ð Þ
1 ðsÞ ðþÞ d1s ðsÞ ðÞ d1s
¼ Fxx 1s Fxx 1s expðstÞdt
2p pffiffiffiffiffiffiffiffiffi dt dt
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
126 6 Cagniard-de Hoop Technique
tends to zero, but the line integral along the branch cut must be discussed. The
pffiffiffiffiffiffiffiffiffiffiffiffiffi
radical for the dilatational wave, 12 þ 1, has different arguments, depending on
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
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
2p
1
Z1 ( )
ðþÞ ðÞ
1 ðsÞ ðþÞ d1s ðsÞ ðÞ d1s
¼ Fxx ð1s Þ Fxx ð1s Þ expðstÞdt
2p pffiffiffiffiffiffiffiffiffi dt dt
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
Z n pffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
1 ðsÞ
þ Fxx ð1Þ exp ðs=cd Þ y 12 þ c2 þ i1x d1
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
6.2 2D In-Plane Deformation 127
Z n pffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
1 ðsÞ
Fxx ð1Þ exp ðs=cd Þ y 12 þ c2 þ i1x d1
2p
AB
Z n pffiffiffiffiffiffiffiffiffiffiffiffiffiffi o
1 ðsÞ
þ Fxx ð1Þ exp ðs=cd Þ y 12 þ c2 þ i1x d1
2p ð6:2:50Þ
DE
ffiffiffiffiffiffiffi
p cx
Zx2 þy2
1
¼ fxxðsÞ ðgÞ expfðs=cd Þðbs y þ xgÞgdg
2p
1
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
128 6 Cagniard-de Hoop Technique
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
2p
1
Z1 ( )
ðþÞ ðÞ
1 ðsÞ ðþÞ d1s ðsÞ ðÞ d1s
¼ Fxx ð1s Þ Fxx ð1s Þ expðstÞdt
2p pffiffiffiffiffiffiffiffiffi dt dt
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
ðþÞ ðÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
ðþÞ ðÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðsÞ ðþÞ d1s ðsÞ ðÞ d1s
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
ðþÞ ðÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðsÞ ðþÞ d1s ðsÞ ðÞ d1s
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
6.2 2D In-Plane Deformation 129
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
ðsÞ ðþÞ d1s ðsÞ ðÞ d1s
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
130 6 Cagniard-de Hoop Technique
c st =r
y = γ 2 −1 | x |
(γ | x | = r )
cd t = r
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffi
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.
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
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
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
and the convergence condition at infinity
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
f ðsÞ ¼ f ðtÞ expðstÞdt ð6:3:7Þ
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
6.3 3D Dynamic Lamb’s Problem 133
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
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
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
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
6.3 3D Dynamic Lamb’s Problem 135
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
ðn; g; z; sÞ ¼ expðas zÞ
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
ðn; g; z; sÞ ¼ expðas zÞ
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
ðn; g; z; sÞ ¼ expðas zÞ
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.
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
6.3 3D Dynamic Lamb’s Problem 137
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
Ixx ðx; y; z; sÞ ¼ 2
expfsðbd z þ irpÞgdpdq
ð2prÞ Rðp; qÞ
1 1
ð6:3:24Þ
where
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
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
6.3 3D Dynamic Lamb’s Problem 139
Re ( p )
−∞ +∞
pd( − ) −i q 2 + 1/ cd2
∞
−i q 2 + 1/ cs2 pd( + )
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
pffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
140 6 Cagniard-de Hoop Technique
t t = R q 2 + 1/ cd2
R / cd
q = (t / R ) 2 − 1 / cd2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
6.3 3D Dynamic Lamb’s Problem 141
ðÞ
where pd are defined by Eq. (6.3.31). The integrands are given by
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
~
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
Ixx ðx; y; z; sÞ ¼ 2
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,
142 6 Cagniard-de Hoop Technique
Z1 Zþ1
ðsÞ ðsÞ
Ixx ðx; y; z; sÞ ¼ dq Fxx ðp; qÞ expfsðbs z þ irpÞgdp ð6:3:45Þ
0 1
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
t2 ðq2 þ 1=c2s ÞR2
irt z
¼ pðÞ
s ð6:3:49Þ
R2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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.
6.3 3D Dynamic Lamb’s Problem 143
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( + )
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
pffiffiffiffiffiffiffiffiffiffiffiffi
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
144 6 Cagniard-de Hoop Technique
(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
Fxx ðp; qÞ expfsðbs z þ irpÞgdp
1
Z1 ( )
ðþÞ ðÞ
ðsÞ ðþÞ dps ðsÞ ðÞ dps
¼ Fxx
ðps ; qÞ Fxx ðps ; qÞ expðstÞdt
pffiffiffiffiffiffiffiffiffiffiffiffi
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
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
On DE: p ¼ i1 ; q2 þ 1=c2d \1\ Rr q2 þ 1=c2s
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi
2
bd ¼icd
bs ¼cs
q þ1=cd
pffiffiffiffiffiffiffiffiffiffiffiffiffi
ðr=RÞ q2 þ1=c2s
Z
¼ fxxðsÞ ð1; qÞ expfsðcs z þ r1Þgd1
pffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi
2
q þ1=cd
and
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
146 6 Cagniard-de Hoop Technique
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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,
pffiffiffiffiffiffiffiffiffiffiffiffiffi
ðr=RÞ q2 þ1=c2s
Z
fxxðsÞ ð1; qÞ expfsðcs z þ r1Þgd1
pffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi
2
q þ1=cd
pffiffiffiffiffiffiffiffiffiffiffiffi
ffi ð6:3:62Þ
2 2
R
Zq þ1=cs
d1H
¼ fxxðsÞ ð1H ; qÞ expðstÞdt
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
Fxx ðp; qÞ expfsðbs z þ irpÞgdp
1
Z1 ( )
ðþÞ ðÞ
ðsÞ ðþÞ dps ðsÞ ðÞ dps
¼ Fxx ðps ; qÞ Fxx ðps ; qÞ expðstÞdt
pffiffiffiffiffiffiffiffiffiffiffiffi
ffi dt dt
R 2
q þ1=cs 2
pffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi
2
R
Zq þ1=cs
d1H
þ fxxðsÞ ð1H ; qÞ expðstÞdt
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi pffiffiffiffiffiffiffiffiffiffiffiffi
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
Fxx ðp; qÞ expfsðbs z þ irpÞgdp
1
Z1 ( )
ðþÞ ðÞ
ðsÞ ðþÞ dps ðsÞ ðÞ dps
¼ Fxx ðps ; qÞ Fxx ðps ; qÞ expðstÞdt
pffiffiffiffiffiffiffiffiffiffiffiffi
ffi dt dt
R 2
q þ1=cs 2
pffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R
Zq þ1=cs
d1
þH ðr=cs Þ2 ðR=cd Þ2 qz fxxðsÞ ð1H ; qÞ H expðstÞdt
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi pffiffiffiffiffiffiffiffiffiffiffiffi
ffi dt
z 2 2
1=cs 1=cd þr 2 2
q þ1=cd
ð6:3:64Þ
6.3 3D Dynamic Lamb’s Problem 147
t t = R q 2 + 1/ cs2
R / cs
q = (t / R ) 2 − 1 / cs2
pffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi
2
Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R
Zq þ1=cs
d1H
þ H ðr=cs Þ2 ðR=cd Þ2 qz dq fxxðsÞ ð1H ; qÞ expðstÞdt
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi pffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
ðt=RÞ 2
Z1 Z1 Z1 Z 1=cs
dq gðt; qÞdt¼ dt gðt; qÞdq ð6:3:66Þ
0
pffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi
2
Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R
Zq þ1=cs
H ðr=cs Þ2 ðR=cd Þ2 qz dq hðt; qÞdt
0
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi pffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
ffi
2 0 R=cs
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
r=cd þz 1=cs 1=cd ðt=RÞ 1=c2s
ð6:3:67Þ
148 6 Cagniard-de Hoop Technique
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,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi dt
r=cd þz 2
1=cs 1=cd 2 0
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
R2
1=c2s 1=c2d
1
ðtz 1=c2s 1=c2d Þ ðr=cd Þ2
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
6.3 3D Dynamic Lamb’s Problem 149
ðsÞ
Ixx ðx; y; z; tÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z 1=cs ( )
2 2
ðt=RÞ
ðþÞ ðÞ
ðsÞ ðþÞ dps ðsÞ ðÞ dps
¼ Hðt R=cs Þ Fxx ðps ; qÞ Fxx ðps ; qÞ dq
dt dt
0
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
1
ðtz 1=c2s 1=c2d Þ ðr=cd Þ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r
Z
d1H
þ H t r=cd z 1=c2s 1=c2d HðR=cs tÞ fxxðsÞ ð1H ; qÞ dq
dt
0
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
ðtz 1=c2s 1=c2d Þ ðr=cd Þ2
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,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
1
ðtz 1=c2s 1=c2d Þ ðr=cd Þ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r
Z
ðsÞ d1H
þ H t r=cd z 1=c2s 1=c2d HðR=cs tÞ fij ð1H ; qÞ dq
dt
0
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
ðtz 1=c2s 1=c2d Þ ðr=cd Þ2
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
150 6 Cagniard-de Hoop Technique
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 ⋅ |
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
( )
c2s 12 þ q2 4icd cs c2 12 þ q2 þ 4icd 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
c2s 12 þ q2 þ4icd cs ð12 q2 Þ ðc2 12 þ q2 Þ 4icd cs ð12 q2 Þ
ð6:3:72aÞ
xy 12 þ q 2
fyxðsÞ ð1; qÞ ¼ ðiÞ 2
2ðpr Þ 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 ðc2 12 þ q2 Þ 4icd cs ð12 q2 Þ
xy ðy1Þ2 þðxqÞ2
fyyðsÞ ð1; qÞ ¼ ðiÞ
2ðpr Þ2 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 ðc2 12 þ q2 Þ 4icd cs ð12 q2 Þ
y
fyzðsÞ ð1; qÞ ¼ 2 1cd cs
p
(r )
1 1
2 þ 2
c2s 12 þ q2 þ4icd cs ð12 q2 Þ ðc2 12 þ q2 Þ 4icd cs ð12 q2 Þ
ð6:3:72bÞ
152 6 Cagniard-de Hoop Technique
x
fzxðsÞ ð1; qÞ ¼ ðþiÞ 2 1 c2s 12 þ q2
( 2p r )
1 1
2 2
c2s 12 þ q2 þ4icd cs ð12 q2 Þ ðc2 12 þ q2 Þ 4icd cs ð12 q2 Þ
y
fzyðsÞ ð1; qÞ ¼ ðþiÞ 2 1 c2s 12 þ q2
( 2p r )
1 1
2 2
c2s 12 þ q2 þ4icd cs ð12 q2 Þ ðc2 12 þ q2 Þ 4icd cs ð12 q2 Þ
cd ð12 q2 Þ
fzzðsÞ ð1; qÞ ¼ 2
(p )
1 1
2 þ 2
c2s 12 þ q2 þ4icd cs ð12 q2 Þ ðc2 12 þ q2 Þ 4icd cs ð12 q2 Þ
ð6:3:72cÞ
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,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
t ¼ z p2 þ q2 þ 1=c2s þ irp ð6:3:76Þ
ð6:3:81Þ
The order of integration is also interchanged as
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
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
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.
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
is satisfied, we apply the double Fourier transform with respect to two space
variables as defined by
Z1 Z1
f ðnÞ ¼ 1 f ðnÞ expðinxÞdn
f ðxÞ expðþinxÞdx; f ðxÞ ¼ ð7:1:6Þ
2p
1 1
Z1 Z1
~f ðgÞ ¼ 1 ~f ðgÞ expðigyÞdg
f ðyÞ expðþigyÞdy; f ðyÞ ¼ ð7:1:7Þ
2p
1 1
The displacement equations are transformed to algebraic equations for the trans-
formed displacement components,
Qx n2 þ bg2
~
ux ¼ þ
C22 ðg2 þ p21 n2 Þðg2 þ p22 n2 Þ
Qy ðc þ 1Þng
C22 ðg2 þ p21 n2 Þðg2 þ p22 n2 Þ
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
C22 ðg2 þ p21 n2 Þðg2 þ p22 n2 Þ
7.1 2D Static Green’s Dyadic for an Orthotropic Elastic Solid 159
~
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
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),
160 7 Miscellaneous Green’s Functions
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
η = +ip2 | ξ |
η = +ip1 | ξ |
y<0
Re(ξ )
y>0 η = −ip1 | ξ |
η = −ip2 | ξ |
7.1 2D Static Green’s Dyadic for an Orthotropic Elastic Solid 161
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
I2j ¼ exp pj jnjjyj expðinxÞdn
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
I2j ¼ exp pj jnjjyj expðinxÞdn
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
162 7 Miscellaneous Green’s Functions
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
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,
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
7.1 2D Static Green’s Dyadic for an Orthotropic Elastic Solid 163
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
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.
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
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
166 7 Miscellaneous Green’s Functions
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
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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,
7.2 2D Static Green’s Dyadic for an Inhomogeneous Elastic Solid 167
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
c lð0Þ 2p ðn ix1 Þðn þ ix1 Þðn ix2 Þðn þ ix2 Þ
1
Zþ1
By 1 c2 n 2 þ g
2 þ j2
þ 2
expðinxÞdn
c lð0Þ 2p ðn ix1 Þðn þ ix1 Þðn ix2 Þðn þ ix2 Þ
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
168 7 Miscellaneous Green’s Functions
+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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
~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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
þ 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi coshðjpxÞ sinhðjpjxjÞ exp jxj g2 þ j2 q2
4c lð0Þ g2 þ j2 q2
jp
ð7:2:23bÞ
7.2 2D Static Green’s Dyadic for an Inhomogeneous Elastic Solid 169
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
I1 ðx; yÞ ¼ g2 þ j2 q2 igy dg
exp jxj ð7:2:26Þ
2p
1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
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
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,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
7.2 2D Static Green’s Dyadic for an Inhomogeneous Elastic Solid 171
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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 ,
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
172 7 Miscellaneous Green’s Functions
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r ¼ x2 þ y 2 ð7:2:39Þ
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
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
ox pffiffiffiffiffiffiffiffiffi
x2 þy2 !1 oy pffiffiffiffiffiffiffiffiffi
x2 þy2 !1
SH-source
Vt
Q
x
x=0 x=l
Moving boundary
174 7 Miscellaneous Green’s Functions
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
7.3 Reflection of a Transient SH-Wave at a Moving Boundary 175
Zþ1 Zþ1
~f ðgÞ ¼ 1 ~f ðgÞ expðigyÞdg ð7:3:11Þ
f ðyÞ expðþigyÞdy; f ðyÞ ¼
2p
1 1
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
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,
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
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.
7.3 Reflection of a Transient SH-Wave at a Moving Boundary 177
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
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
178 7 Miscellaneous Green’s Functions
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
icsη
γs Re( s )
branch no singular point
−icsη s( p)
γ s − i∞ p = γ p − i∞
A
7.3 Reflection of a Transient SH-Wave at a Moving Boundary 179
p ¼ s as V ð7:3:35Þ
its inverse is given by
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
180 7 Miscellaneous Green’s Functions
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
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
7.3 Reflection of a Transient SH-Wave at a Moving Boundary 181
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
pffiffiffiffiffiffiffiffiffiffiffiffiffi
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
182 7 Miscellaneous Green’s Functions
at
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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
ðz þ lÞ2 þðmyÞ2
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
Zþ1 pffiffiffiffiffiffiffiffiffiffiffiffiffi
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
7.3 Reflection of a Transient SH-Wave at a Moving Boundary 183
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,
Zþ1 pffiffiffiffiffiffiffiffiffiffiffiffiffi
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
184 7 Miscellaneous Green’s Functions
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,
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
7.3 Reflection of a Transient SH-Wave at a Moving Boundary 185
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
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
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.
186 7 Miscellaneous Green’s Functions
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
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