You are on page 1of 14

Article

Mathematics and Mechanics of Solids


1–14
Lamb’s problem: a brief history Ó The Author(s) 2019
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/1081286519883674
journals.sagepub.com/home/mms

Mohamad Emami and Morteza Eskandari-Ghadi


School of Civil Engineering, College of Engineering, University of Tehran, Tehran, Iran
Received 21 September 2019; accepted 29 September 2019

Abstract
In this review note, a historical scientific investigation is presented for Lamb’s problem in the mathematical theory of elas-
ticity. This problem first appeared in 1904 in the pioneering paper of Professor Sir Horace Lamb (Lamb, H. On the propa-
gation of tremors over the surface of an elastic solid. Philos Trans R Soc Lon 1904; 203: 1–42). Of special interest here are
the analytical studies of the three-dimensional version of Lamb’s problem, which consists of a semi-infinite, homogeneous,
isotropic elastic solid that is set in motion by the exertion of a dynamical point force applied suddenly on the surface of
the domain. The objective of this paper is to offer a comprehensive introduction to Lamb’s problem for the reader, along
with discussing its mathematical complexities. An account is given of the history of this ever-significant problem from its
earlier stages to the more recent investigations via outlining and discussing different rigorous approaches and methods of
solution that have been hitherto suggested. The limitations of different methods, if they exist, are also discussed.
Eventually, various solution methods are compared considering their nature, advantages, and restrictions.

Keywords
Lamb’s problem, Cagniard, Pekeris, Chao, Eason, Gakenheimer, Mooney, Johnson, Richards, Kausel, Feng and Zhang

1. Introduction
Because of its applications, engineers are very interested in studying wave propagation in elastic solids.
Analytical studies of wave propagation in the theory of elasticity are often involved with many mathe-
matical challenges, and thus mathematicians are also very interested in this subject. Even in problems
with simple geometries, such as a full-space or a half-space, in the classical theory of linear, homoge-
neous, isotropic elastic solids, researchers may face staggering barriers of mathematical complexities that
have to be overcome before useful solutions can be obtained. This quality, while being unfavorable from
a pragmatic point of view, has frequently attracted the attention of applied mathematicians and, in turn,
has led to the invention or use of new mathematical tools and further progression and enrichment of the
theory. On the other end of the spectrum lie applications in both science and engineering, from which
one may mention studying direct wave propagation, implementing the time-domain boundary element
method, detailed analyses of seismology, detecting anomalies in an elastic material via applying the
inverse method, investigating time-domain noises due to moving loads, detecting the problems of high-
speed trains, and investigating the real moving loads passing on bridges, as examples.
There are many good examples of such challenging problems, among which Lamb’s problem, abbre-
viated hereafter as LP, is especially noteworthy. The problem was first introduced by the influential
English mathematician and engineer, Sir Horace Lamb (1849–1934), in his 1904 memoir on the propaga-
tion of forced tremors over the surface of semi-infinite elastic solids [1]. Since then, there has been peren-
nial interest in this problem, and researchers have repeatedly studied its different aspects and its various
special cases or generalizations for decades. However, not all studies in this context were originally

Corresponding author:
Morteza Eskandari-Ghadi, School of Civil Engineering, College of Engineering, University of Tehran, P.O. Box 11165-4563, Tehran, Iran.
Email: ghadi@ut.ac.ir
2 Mathematics and Mechanics of Solids 00(0)

presented as an explicit contribution to LP; there are papers that investigate other related problems in
elastodynamics, but their general methods have also contributed to the solution of LP. Accordingly, it
might appear at first sight that the history of LP may not be particularly coherent and continuous. Yet,
by considering other related contributions, one may shed light on the progress of development of LP
and gain better insight.
One immediate application of LP in seismology is in the interpreting of surface or interior ground
motions during seismic events. From the various modern applications, one may mention the use of the
solutions of LP as Green’s functions in boundary integral formulation of more complicated elastody-
namic problems and in numerical solutions by the boundary element method. Theoretically, one is able
to find the elastodynamic motions for a half-space for loads with arbitrary distribution and time-
dependence by implementing appropriate convolution integrals with respect to spatial and temporal
variables. Moreover, the results of LP can be employed for solving inverse problems: for example, for
finding anomalies inside an elastic half-space. The results of the original LP are also useful for determin-
ing transient ground motions due to impact loading, for instance, during pile driving and soil compac-
tion operations in civil engineering. In addition, generalizations in many aspects of LP has been made,
each with interesting applications. As an example, the problem of a point load moving on the free sur-
face of a half-space has apparent applications in determining ground noises due to movement of high-
speed trains. The solution of a traveling point load may also be used for inventing a cheap means for
counting the number of heavy vehicles passing over an existing bridge, which can in turn be used for
fatigue analysis of the bridge.
As will be discussed shortly, the methods used for finding the solutions of LP in the time domain can
be roughly classified into two groups. The first group of methods involves the use of Fourier superposi-
tion (Fourier synthesis) over the time-harmonic solutions. This was the original procedure followed by
Lamb [1] and requires the evaluation of multiple integrals. To circumvent the complexities introduced
by multiple integrations, other, more sophisticated, methods were later developed, which attempt to find
the time-domain solutions directly without the use of Fourier synthesis. Such methods constitute the sec-
ond group of methods and are mainly rooted in the ideas of Cagniard [2], Pekeris [3–5], and de Hoop
[6].
Here, we aim to give a reasonably comprehensive account of the history of LP in a semi-chronological
order. It is surely impractical and not very helpful to try to cite every paper related to LP, since the rele-
vant literature spans over more than a century. Hence, only those researches that seem to be of greater
importance and relevance are made reference to in this paper. Following the history of LP and its devel-
opment, one inevitably notices a great diversity of formulations, notations, solution methods, etc. At
times, different approaches were being followed by different authors simultaneously and the methods
utilized seem to have changed over time. Fluctuations of interest can also be observed in this regard that
may be confounding to the new reader. Another important objective of the present paper is, therefore,
trying to explain in reasonable detail those various approaches, their similarities, and their discrepancies.
In order for the exposition to be more organized, the discussion is divided into a number of sections as
follows.

2. Definition of the problem


For future reference, it is appropriate first to give a brief description of LP. Originally, the three-
dimensional case of the problem proposed by Lamb [1] investigated the surface motions (displacement
components) of a linear elastic, isotropic, homogeneous half-space due to a dynamical point load, acting
vertically (its direction being perpendicular to the boundary plane) either upon its free surface or at a
finite depth within the medium. Here, however, by LP we shall mean a slightly generalized form of the
mentioned problem in which the load can be applied at an arbitrary direction having both vertical and
horizontal components. In addition, the full set of elastodynamic solutions of LP, consisting of both the
surface and interior motions, that is, displacement components for boundary and interior points of the
half-space, respectively, are herein considered; see Figure 1. Obviously, the two-dimensional versions of
LP can be defined likewise by merely replacing the point load by a line load.
When seeking the elastodynamic solutions, one should distinguish clearly between the cases of mono-
chromatic (also called steady-state and time-harmonic) and transient (anharmonic) motions. In the
Emami and Eskandari-Ghadi 3

Figure 1. The semi-infinite elastic solid disturbed by a dynamical point load. Point sets O and ∂O denote the half-space and its
boundary, respectively. (x1 , x2 , x3 ) and (r, u, z) represent, respectively, the Cartesian and cylindrical coordinates. l, m, and r are the
Lamé elastic constants and the mass density of the medium. p(t) is the time-dependent point load, being exerted at an arbitrary
depth h ø 0, and through arbitrary angles u0 and u0 with respect to the x3 - and x1 -axes.

former, the point load is considered to have trigonometric sine and/or cosine time-dependence or
equivalently the function eivt , where v is the angular frequency, and it is further assumed to be initiated
so far back in time that all the ephemeral effects have been geometrically attenuated beforehand and a
dynamical steady state has been reached. Hence, the initial conditions do not enter the formulation of
the problem. Mathematically speaking, one seeks in this case only the solutions that vary with time har-
monically [7]. On the other hand, in the latter the time-dependence of the point load is an arbitrary
function of time, and it is assumed to be initiated at a certain instant, so requiring the assumption of ini-
tial conditions. In this case, the motions develop over time and they may or may not reach an eventual
steady state.
Accordingly, the displacement equations of motion [8] along with the boundary conditions imposed
by the excitation constitute a boundary value problem (BVP) in terms of the components of the displa-
cement vector u(x, t) for the time-harmonic case, whereas in the transient case, one has to deal with the
corresponding initial-boundary value problem (IBVP). Naturally, owing to the added mathematical
complexities, the transient solutions of LP for both interior and boundary points are more difficult to
obtain than their time-harmonic counterparts.

3. Motivations, Lamb’s work, and early investigations


In the late 19th century, John William Strutt (Lord Rayleigh), prominent English physicist and later
Nobel Prize laureate, discovered a completely new form for the solutions of wave propagation problem
in an elastic half-space [9]. These solutions manifested the existence of free plane harmonic waves that
can propagate along the surface of the solid, the amplitude of which diminishes exponentially as the
depth increases, and the speed of which is lower than those of both the longitudinal (irrotational) and
transverse (equivoluminal) waves [9]. By free waves we refer to waves that can be sustained in the
medium without the need for an external excitation. Mathematically, these are the solutions of a BVP
or an IBVP in elastodynamics, which satisfy the homogeneous equations of motion and the homoge-
neous boundary conditions. The surface wave thus discovered by Rayleigh is essentially different from
the well-known body waves that propagate in the interior of solid media [10]. This proved that the
4 Mathematics and Mechanics of Solids 00(0)

existence of a traction-free boundary in an elastic solid alters the nature of its wave motions far more
seriously than had been previously expected [1].
It was, perhaps, Rayleigh’s remarkable discovery [9] that motivated Lamb to take further steps in this
direction and investigate the elastodynamic solutions of a half-space when it is in a state of forced
motion from a partial differential equation (PDE) point of view, that is, the case in which the motions
of the medium are caused by external forces. Lamb assumed a concentrated force to act as either a sur-
face traction on the plane boundary of the half-space or a body force buried within the interior of the
half-space. Attention was focused, exclusively, upon the motions of the free surface, that is, the bound-
ary solutions of the considered problem, in sensible hope that the results may offer better understanding
of seismic phenomena as they manifest themselves over the surface of the earth. In Lamb’s own words:

Although the circumstances of actual earthquakes must differ greatly from the highly idealized state of things
which we are obliged to assume as a basis of calculation, it is hoped that the solution of the problems here con-
sidered may not be altogether irrelevant [1].

These theoretical and practical incentives motivate well the initiation of Lamb’s investigations and other
following researches.
Lamb’s ultimate goal was to take into account the effect of an impulsive force [1], and so acquiring
solutions of the LP in their true transient form. For the purpose of analysis, however, he first obtained
monochromatic solutions corresponding to the effect of a time-harmonic force, passing then to the case
of transient solutions via Fourier synthesis [1]. The larger portion of Lamb’s paper is devoted to studies
of the two-dimensional versions of LP, that is, the motions of a half-space due to a line load. Most fully
discussed among these two-dimensional problems is the case of vertical surface force; however, cases of
horizontal surface force and buried force are also studied to some extent [1].
Lamb, with the use of Helmholtz decomposition accompanied with Fourier integral transform,
obtained his two-dimensional monochromatic solutions as improper integrals consisting of integrands
having singularities on the integration path. Utilizing clever methods of contour integration in the com-
plex plane and taking the Cauchy principal values thereof as the physically significant solutions, he
managed to study further the properties of his formulae. Doing so, he rigorously showed that the sur-
face motions are marked by three distinct epochs agreeing with the arrival of longitudinal, transverse,
and surface wavefronts, respectively. Therefore, Lamb proved the existence of forced surface waves on
the boundary of a half-space, which are somewhat similar to, but more complicated and more general
than Rayleigh’s free surface waves. Unsurprisingly, the speed of propagation of Lamb’s forced surface
wave turned out to be exactly equal to Rayleigh’s free surface wave [1,9].
Lamb then proceeded to the axisymmetric three-dimensional version of his problem, in which an elas-
tic half-space is excited by a vertical point force, acting either as a surface traction or a buried source.
Again, the time-harmonic surface solutions were first determined in a cylindrical coordinate system via
generalization and extension of the methods that were utilized for the two-dimensional cases. Utilizing
Helmholtz decomposition, these solutions are expressed in the form of improper integrals with Bessel
functions of the first kind multiplied by some algebraic expressions. In more modern terminology, these
integrals can be regarded as inverse Hankel integral transforms, and indeed, one could find them by
using the theory of Hankel transforms ab initio [11]. As before, forced surface waves analogous to those
found by Rayleigh were shown to exist as cylindrical wavefronts traveling away from the point of appli-
cation of the force [1].
Naturally, in order to find the transient solutions arising from arbitrary anharmonic excitations by
Fourier synthesis, one requires another integration of Fourier type to be carried out over the angular
frequency. Consequently, the transient axisymmetric surface solutions of LP as given in Lamb’s paper
[1] should be represented as double integrals if written out explicitly. Lamb did not evaluate the transi-
ent solutions, but studied their general properties in some detail, based on which he presented rough
time-history diagrams for surface motions of the half-space due to an impulsive source of excitation. As
expected, the consecutive arrivals of longitudinal, transverse, and surface wavefronts were apparent in
his diagrams [1]. It should be mentioned that the asymmetric three-dimensional case of LP, that is, the
case due to a horizontal or an inclined force, was altogether omitted in Lamb’s paper. Moreover, the
Emami and Eskandari-Ghadi 5

whole set of transient solutions for interior points of the half-space still remained to be found until
much later.
In the decades following the publication of Lamb’s paper [1], many of researchers focused on various
aspects of the two-dimensional versions of LP [12–14]. They were often interested in ascertaining clearly the
properties and the behavior of different wavefronts that travel along the surface of the half-space. In 1925,
Nakano [12], for instance, revisited the version of LP involving a buried line source and used an analysis
procedure similar to that of Lamb [1]. He then carried out approximate evaluation of the integrals using
the methods of stationary phase and steepest descent [12] (for these methods see, for instance, [15]). Among
other important contributions to the two-dimensional versions of LP, the papers of Lapwood [13] and of
Garvin [14] may be mentioned. Of special significance is Garvin’s 1956 paper [14], as it was the first depar-
ture from the Fourier synthesis method used by Lamb and others for the two-dimensional versions of LP.
The procedure that Garvin [14] used involved the utilization of Laplace transform with respect to time, and
was based on Cagniard’s method [2], which will be discussed in the next section.

4. Later investigations: toward direct formulae for transient solutions


4.1. A breakthrough by Cagniard
As mentioned earlier, the transient solutions of LP can be obtained via Fourier synthesis of its time-
harmonic solutions, as was proposed by Lamb [1] and others that followed. However, the process is
complicated and not computationally efficient. Direct formulae for transient solutions, if possible to be
acquired, are advantageous for both theoretical and practical purposes.
Taking the next major steps for solving the transient case of LP was facilitated by the invention of
new mathematical procedures by Louis Cagniard, French mathematician and geophysicist, who exten-
sively studied the reflection and refraction of elastic waves at the interface of two fully connected elastic
half-spaces in his doctoral thesis in 1939 [2] (see also its English version translated and revised by Flinn
and Dix [16]). In that work, Cagniard first proposed the application of Laplace transform with respect
to the temporal variable, by which, indeed, the IBVP can be reduced to a BVP in the Laplace-trans-
formed domain. Then, the Laplace-transformed BVP can be solved by conventional methods, either the
classical methods proposed by Lamb and others or the more modern use of integral transforms. The
solutions thereof are obtained in the form of improper integrals, in addition to which one would then
normally need to take the inversion integral of the Laplace transform [11]. In contrast, Cagniard’s inge-
nious method proposed that the integral solution of the Laplace-transformed BVP can be transmuted,
through certain changes of variables and alteration of integration paths in the complex plane, to the
form of Laplace transform integrals. Proceeding thus, the final time-domain solutions will be obtainable
by mere inspection [2]. In short, Cagniard’s method involves solving a certain class of integral equations
via complex-analytic techniques, an outline of which will be presented later. Typically, by utilizing
Cagniard’s method, one may obtain the transient solutions of LP, and many other time-domain wave
propagation problems, in the form of single integrals, instead of the double integrals found by using
Fourier synthesis on the time-harmonic solutions [1].
In Cagniard’s thesis [2], the propagation of transient elastic waves in a bi-material full-space, caused by
a compressional point source, was investigated. In that problem, the point source was assumed to be
imbedded in one of the half-spaces. A slightly simplified version of the problem considered by Cagniard is
the problem of investigating the transient motions of an elastic half-space due to a buried impulsive point
source, which closely resembles LP for a buried point load. This problem was studied thoroughly by
Pinney in 1954 [17] for both dilatational and torsional point sources, the latter having an axis either perpen-
dicular or parallel to the free surface. The dilatational point source problem was later revisited in 1961 by
de Hoop [18] as well, who used his modified version of Cagniard’s procedure for solving the problem.

4.2. Pekeris’ axisymmetric transient surface solutions


Although Cagniard’s seminal work paved the way for many later researches, it did not directly study
the transient solutions of LP. It was through two important papers [3,4] by Chaim Leib Pekeris, an
Israeli-American physicist and mathematician, that some special cases of transient solutions of the
problem were first published in 1955. In his first paper [3], Pekeris considered transient surface solutions
6 Mathematics and Mechanics of Solids 00(0)

of LP for the case in which the load is vertically applied to the surface (the seismic surface pulse prob-
lem), and in his second paper [4], he treated the same problem for the vertical load applied at a finite
depth (the seismic buried pulse problem) in the half-space. The numerical evaluation of the solutions pre-
sented in Pekeris’ second paper [4] was given in a subsequent paper published two years later in 1957 by
Pekeris and Lifson [19]. In both cases, the time-dependence of the load is taken to be the Heaviside unit
step function [3,4]. Pekeris’ transient surface solutions showed that after the successive arrival of longi-
tudinal and transverse wavefronts, the Rayleigh–Lamb surface wavefront arrives, the amplitude of
which is infinite if the load is exerted on the surface [3] and finite if the load is applied at an interior
point [4,19]. Although Pekeris’ method [3,4] is quite general for the calculation of transient solutions, he
made the assumption that the half-space is composed of a Poisson material, meaning that the Lamé
elastic constants of the material are equal: l = m. The restriction of solutions to Poisson materials in
Pekeris’ first 1955 paper [3] was later removed by Mooney [20]. Pekeris’ operational solution formulae
in integral form were taken as a starting point in Mooney’s paper [20] in 1974. By means of changing
the branch cuts and the integration paths, as implemented in the Cagniard–Pekeris method, Mooney
managed to generalize the work of Pekeris on the seismic surface pulse problem to general materials
having arbitrary Poisson’s ratios [20]. The solutions found by Mooney are in the integral form, and they
were numerically calculated and plotted. Via convolution, Mooney [20] also studied the effect of chang-
ing the time-dependence of the point force, so extending the solutions of Pekeris [3] to arbitrary source
waveforms.
In essence, the methods used by Pekeris in his papers are equivalent to the methods devised by
Cagniard (translating the problem to a certain class of integral equations that may be solved via
complex-analytic techniques), one formal difference being that Pekeris utilized the notions of opera-
tional calculus instead of the Laplace transform. Interestingly, as Pekeris himself pointed out [3], he had
solved the surface pulse problem much earlier, almost simultaneously with the publication of Cagniard’s
work [2]. This is supported by the fact that some of Pekeris’ results were already present in one of his
earlier papers (published in 1940) on the numerical solution of integral equations [21]. For this reason,
Cagniard’s method is sometimes also referred to as the Cagniard–Pekeris method. A modern and more
lucid interpretation of Pekeris’ method in his first 1955 paper [3], based on the use of the Laplace trans-
form, is presented in Achenbach’s treatise [8].

4.3. Chao’s asymmetric transient surface solutions and epicentral solutions


Soon after Pekeris, Chao in 1960 extended the transient surface solutions of LP for the case of a hori-
zontal point load [22]. This indeed is an asymmetric case of the three-dimensional LP and is much more
difficult to deal with as compared with the axisymmetric case due chiefly to the lack of symmetry and
the number of the equations of motion. Chao used Pekeris’ method [3,4] for finding the transient solu-
tions. One of the innovations presented in Chao’s paper was in considering also a very special case of
interior solutions. He found that obtaining the transient solutions for the interior points that lie directly
beneath the point of application of the surface load (the epicentral solutions) is quite simply carried out
by a change of variables in the integrals representing the solutions in the Laplace-transformed domain
[22]. In his paper, Chao [22] first formulates the problem for the case in which the horizontal load is bur-
ied. However, then, stating that the asymmetric solutions for the buried load are exceedingly difficult to
deal with, he carried on with the special case in which the depth of application of load is equal to zero,
so giving only the solutions for the surface load case [22].

4.4. Eason’s early attempts toward transient interior solutions


With the increased interest in the transient interior solutions of LP, Eason, in 1966, followed up with an
interesting paper in which he investigated the axisymmetric transient interior solutions of LP [23]. In
addition to a vertical surface point load, Eason also studied the case in which axisymmetric patch loads
were applied on the surface of the half-space [23]. The method applied by Eason [23] was a departure
from the Cagniard–Pekeris procedure in the sense that the inversion process of the Laplace transform
was based on the direct Laplace inversion theorem [11], rather than transmuting the Laplace-domain
integral solutions into the form of Laplace transform integrals and extracting the time-domain solutions
Emami and Eskandari-Ghadi 7

by solving integral equations [2–4]. Eason’s approach involved, instead, elaborate contour integrations
along paths encircling singular points and circumventing branch cuts in the complex plane of the
Laplace transform parameter [23].
As mentioned earlier, Pekeris considered a Poisson material for presenting his transient axisymmetric
solutions to LP, and so followed Chao in his paper [22]. The time-history diagrams of surface transient
solutions of Pekeris and Chao [3,4,22] are thus restricted to Poisson materials. On the other hand, fol-
lowing his own approach, Eason was able to present solutions that were not thus restricted to special
materials [23]. The final solutions that Eason [23] presented were in the form of quite complicated finite
integrals and, probably due to difficulties in the numerical evaluation, his time-history diagrams for
non-epicentral interior solutions were calculated and plotted only for times greater than the instant of
arrival of the transverse wavefront. Eason’s diagrams showed lucidly that the amplitude of the
Rayleigh–Lamb traveling surface wavefront diminishes rapidly to finite values as the depth of point
within the half-space increases [23].

4.5. Modifications of Cagniard’s procedure and the Cagniard–de Hoop method


Due to its intricate and sophisticated nature, the method devised by Cagniard [2] has been the subject of
a number of researches that tried to explain it or modify it into simpler, more elegant, or more generally
applicable procedures. Among such papers the explanatory works of Dix in 1954 and 1958 [24,25] and
the more recent papers by Abramovici [26], Murrell and Ungar [27], and Ben-Hador and Buchen [28]
may be cited. More popular among the later researchers of transient wave propagation problems than
Cagniard’s original approach is de Hoop’s modification of Cagniard’s method. This is often referred to
as the Cagniard–de Hoop method and was first presented in 1960 by de Hoop [6] (see also de Hoop’s doc-
toral thesis [29]), who deemed it more straightforward and simpler than Cagniard’s original approach [2].
In the Cagniard–de Hoop method, the wave propagation problems are first formulated in the
Cartesian coordinates, in lieu of the cylindrical coordinate system used in the Cagniard–Pekeris method,
so requiring the use of either two-dimensional or three-dimensional spatial Fourier transforms instead
of Hankel integral transforms. Yet again, the Laplace transform is taken with respect to the temporal
variable. Then, following the line of reasoning essentially similar to that of Cagniard [2] and Pekeris
[3,4] and a special set of variable changes, the time-domain solutions are found in integral or closed
form [6], depending upon the type of the problem. For more detailed discussions of this method, [30]
and [6] may be referred to.

4.6. Gakenheimer’s traveling point load problem and axisymmetric transient interior solutions for LP
Not long after the publication of Eason’s work [23], the next contribution to LP came through the doc-
toral thesis of Gakenheimer [31], the ensuing paper of Gakenheimer and Miklowitz [32], and another
paper by Gakenheimer [33]. What Gakenheimer studied was the problem of finding the transient
motions of an elastic half-space due to a suddenly applied normal point force that also suddenly started
moving rectilinearly and with constant speed upon the free surface of the half-space. In this problem,
the speed of the traveling point force has a great impact on the form and nature of solutions, as the
force can travel at subsonic, transonic, or supersonic speeds with respect to the speed of either P- or SV-
waves propagating within the half-space. The analytical results of Gakenheimer’s thesis [31], being in
integral form, were afterwards summarized in the paper by Gakenheimer and Miklowitz [32].
Indeed, the limit case in which the speed of the traveling point force is zero is exactly the axisymmetric
transient case of LP, which was then investigated exclusively by Gakenheimer in a subsequent paper [33]
in which he numerically evaluated his solutions for the case of l = m (namely Poisson materials) at the
interior points of the half-space. Unlike the results presented in Eason’s paper [23], the time-history dia-
grams of Gakenheimer [33] showed the complete process of wave propagation from the instant of the
point force exertion in the form of the Heaviside step function until after the passing of Rayleigh wave-
front and the eventual approach of the motions within the medium to a statical steady state. These
results [33] completed the diagrams of Eason [23] in the case of a Poisson material. Yet, the transient
interior solutions of LP for horizontal forces were still not known.
8 Mathematics and Mechanics of Solids 00(0)

4.7. Johnson’s concluding paper: Green’s functions for Lamb’s problem


Eventually in 1974, Johnson [34], utilizing the Cagniard–de Hoop method, presented a full set of transi-
ent solutions of LP for both interior and boundary points, involving vertical and horizontal point loads
applied either within the half-space or on its free surface. In Johnson’s paper [34], in addition to the
final set of transient solutions of LP, the spatial derivatives of the solutions were presented and the
time-history diagrams that he presented for the solutions completed the works of many earlier research-
ers; for instance [1, 3, 4, 22, 23]. For the process of solution, Johnson used three-dimensional bilateral
Laplace transforms with respect to the Cartesian coordinates and the unilateral Laplace transform with
respect to time, and proceeded with the notions and techniques of the Cagniard–de Hoop procedure.
Unfortunately, the details of the process of the solution were not presented in his paper [34], but only
briefly outlined. Because of this, his paper received less attention in the subject. Nevertheless, Johnson’s
classical paper [34] may be regarded as the complete and concluding account to the transient solutions
of LP for both interior and boundary points, as far as the final solutions in integral form are of interest.

5. Other remarks and more recent contributions


5.1. Integral equations encountered in and the solution steps of the Cagniard–Pekeris method
As mentioned earlier, the solutions of a certain class of integral equations are used for transient wave
propagation problems if the Cagniard–Pekeris procedure is utilized. The papers by Pekeris [5] and
Longman [35] may be mentioned in this subject. In addition, in a recent paper by Emami and
Eskandari-Ghadi a slightly generalized version of such integral equations has been studied [36]. In the
Cagniard–de Hoop method [6], another form of integral equations needs to be solved, for which the
classical paper by de Hoop [6] and also the comprehensive work of Pao and Gajewski [30] may serve
the reader well as excellent references.
In order to explain the essence of the Cagniard–Pekeris method, as presented in [36], an outline of
the steps of the solution is presented as follows. For the Cagniard–de Hoop method, the reader is best
referred to [6].
Step 1: taking the initial conditions into account, Laplace transform with respect to the temporal
variable t is implemented to reduce the IBVP to a corresponding BVP in the Laplace domain. Instead
of using the Laplace transform, Pekeris [3,4] used the equivalent notions of operational calculus [11].
The parameter s of the Laplace transform is here deliberately assumed to be real and positive; see, for
instance, [8].
Step 2: the resulting BVP is solved by suitable methods. Typically, this involves the use of a suitable
set of potential functions jointly with integral transforms taken with respect to some or all of the spatial
variables. By utilization of the method of potential functions, one is able to decouple the system of
PDEs governing the displacement components into separate PDEs governing the potential functions,
by which the displacement components can be derived later.
In the Cagniard–Pekeris method, the problem is formulated in cylindrical coordinates (r, u, z) ab initio;
see Figure 1. If the problem is axisymmetric, Hankel transforms, having a real and positive parameter j,
with respect to the radial coordinate r can reduce the PDEs governing the potential functions to ordinary
differential equations in terms of the vertical coordinate z, the general solutions of which can be obtained
by classical methods [37]. If the problem is non-axisymmetric, Hankel transform with respect to r is
jointed with Fourier series expansion (equivalently, finite Fourier transform with an integer parameter m)
with respect to the angular coordinate u to obtain the aforementioned set of ordinary differential equa-
tions. In the Cagniard–de Hoop method, the problem is primarily formulated in terms of the Cartesian
coordinates (x1 , x2 , x3 ); see Figure 1. Then, typically, in lieu of Hankel transform or joint Hankel–Fourier
transform, double Fourier transform is taken with respect to x1 and x2 , or a triple Fourier transform with
respect to x1 , x2 , and x3 . As mentioned earlier, Johnson [34] employed triple bilateral Laplace transform
instead of triple Fourier transform in his version of the Cagniard–de Hoop method.
By taking suitable integral transforms, the boundary conditions are also translated into the Hankel–
Fourier–Laplace or Fourier–Laplace-transformed domain, and the satisfaction of boundary conditions
ultimately results in determination of the solutions of the BVP in the transformed domain.
Emami and Eskandari-Ghadi 9

Step 3: suitable inverse transforms with respect to all transformed spatial variables are taken. In the
case of the Cagniard–Pekeris method this means a Hankel–Fourier inversion, and in the case of the
Cagniard–de Hoop method a double or triple Fourier inversion should be taken. With these inversions
the formal Laplace-domain solution of the wave propagation problem will be at hand in the form of a
joint infinite Fourier series and a Hankel inversion integral in the case of the Cagniard–Pekeris method
or multiple integrals, should the Cagniard–de Hoop method be used. If the source is concentrated, that
is, for example a point load as considered in LP, the finite Fourier transform can be analytically inverted,
as only a few of the terms in the corresponding infinite series are non-zero. As a result, the Laplace-
domain solution of the problem will be in the form of finite linear combinations of single inverse Hankel
transform integrals, each of which physically represents a Laplace-domain wavefront being propagated
in a given direction.
Step 4: the change of variables j = sz is implemented, that is, the Hankel transform parameter is
scaled with the Laplace transform parameter. This change of variables is perhaps rooted in the work of
Lamb [1], as he used a mathematically analogous change of variables in his time-harmonic solutions.
As s and j are both real and positive, so will be the new variable z. As a result of this, the mentioned
integrals take the following typical form
ð‘
~
Fn ðr, z, sÞ = zn + 1 gðzÞesh(z;z) Jn ðsrzÞdz, ð1Þ
0

by inverting of which, that is, finding their counterpart in the time-domain, one will then be able to con-
struct the time-domain displacement components that constitute the final solution of the problem. In the
above equation, F ~ n ðr, z, sÞ denotes the Laplace transform of some function Fn ðr, z, tÞ that physically repre-
sents a specific wavefront, n is an integer taking the values 0, 1, or 2 (in the case of a point load excitation),
Jn (  ) is the Bessel function of the first kind and of integer order n [38], g(z) is an algebraic expression of the
body wave characteristic radicals (to be defined shortly), and, finally, h(z; z) represents a linear combination
of the wave characteristic radicals with positive coefficients,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi one of which contains the variable z. The wave
characteristic radicals are functions of the form z2 + s 2 in the case of an isotropic medium, in which s is
the slowness (reciprocal of speed of propagation) of either longitudinal or transverse waves.
Step 5: to invert the mentioned integrals into the time domain, the Laplace transform should be
inverted, that is, the function Fn ðr, z, tÞ should be obtained. To do this, instead of the typical use of the
inversion formula for the Laplace transform [11,39], one may rewrite Equation (1) as the following inte-
gral equation
ð‘ ð‘
st
Fn ðr, z, tÞe dt = zn + 1 g ðzÞesh(z;z) Jn ðsrzÞdz, ð2Þ
0 0

in terms of the function Fn ðr, z, tÞ. To proceed further, we fix n = 0 and solve the above integral equation
for that special case. The solution for cases n = 1 and n = 2 can then be constructed from the case n = 0,
as fully explained in [36]. Afterwards, the function J0 (  ) will be replaced by its Mehler–Sonine integral
representation [38], by which and after a change in the order of integration, one can rewrite the above
equation as
ð ‘ ð‘ 
2
~ 0 ðr, z, sÞ = = 1 st
F pffiffiffiffiffiffiffiffiffiffiffiffiffi dx zg ðzÞe dz , ð3Þ
p 1 x2  1 0

where =(  ) denotes the imaginary part of a complex-valued function, and we have made the change of
variables
t = hðz; zÞ  irxz, ð4Þ

in the right-hand side of (3). The purpose of doing so is to try to transmute the right-hand side of the
integral equation into the form of a Laplace transform integral (similar to the integral on the left-hand
side of (2)). This, in turn, enables us to find the function F0 ðr, z, tÞ easily by inspection. As z varies from
zero to infinity as a real variable, the variable t will take on complex values. This, however, is undesirable
10 Mathematics and Mechanics of Solids 00(0)

as t should represent time and hence should be real and positive. Solving Equation (4) for z, one can
have this variable in terms of t, that is, z = z(t). Now, assuming t to be real and positive, z = z(t) may be
interpreted as a parametric representation of a curve G in the complex z-plane. For example, in the case
of transient interior solutions of LP for a surface point load, G is a curve that is composed of a linear seg-
ment attached to and followed by a hyperbolic semi-branch; see [4] or, for more details, [36]. It follows
then from Cauchy’s integral theorem [40] that the original path of integration of the inner integral in the
right-hand side of (3), which is the positive real semi-axis in the complex z-plane, can be deformed into
the path G on which t is real and positive. Therefore, the real integral just mentioned is transmuted into
a contour integral in the complex z-plane, or a contour parametrized by a positive parameter t. Doing so
and changing again the order of integration in the right-hand side of (3), one may write
ð‘ ð‘ ð  0  !
st 2 ‘ = zðtÞz ðtÞg ðzðtÞÞ
F0 ðr, z, tÞe dt = pffiffiffiffiffiffiffiffiffiffiffiffiffi dx est dt, ð5Þ
0 0 p 1 2
x 1

in which z(t) is the solution of Equation (4) for z, and z0 (t) is the partial derivative of z(t) with respect to
t. From the above equation and in view of the uniqueness theorem for the Laplace transform [39], one
simply obtains
ð  0 
2 ‘ = zðtÞz ðtÞgðzðtÞÞ
F0 ðr, z, tÞ = pffiffiffiffiffiffiffiffiffiffiffiffiffi dx: ð6Þ
p 1 x2  1
This is the solution technique of the integral equation (2) for the case n = 0. To simply construct the
solutions for the cases n = 1 and n = 2 from the solution of the case n = 0, the reader is referred to [36].
The procedure roughly described outlines the ideas behind the Cagniard–Pekeris method, as slightly
modified and generalized by Emami and Eskandari-Ghadi [36]. The original methods of Cagniard [2]
and Pekeris [3,4] are different in certain details, yet equivalent to the above in essence.

5.2. Closed-form solutions and removing the restriction on Poisson’s ratio


One point of interest in the investigations related to LP has been the possibility of expressing closed-form
formulations for the transient solutions. Many authors have attempted this and have been successful to
various degrees. Pekeris [3], for instance, by considering only the case of Poisson materials (l = m or
Poisson’s ratio equal to 0:25) was able to express his surface solutions for a vertical surface force in terms
of the well-known elliptic integrals. For the surface solutions due to a buried vertical load, however,
Pekeris [4] was unable to further reduce his integral-form solutions into closed formulae, and thus had to
do numerical calculations with Lifson for the purpose of finding time-history diagrams [19]. Chao [22],
following the work of Pekeris [3], was also able to represent his surface transient solutions of LP for a
horizontal surface load, again for the case of Poisson materials. Strictly speaking, such expressions con-
taining elliptic integrals may not be regarded as closed-form expressions, as they contain non-elementary
functions expressed as integrals or infinite series. Even so, this loose terminology is common amongst
authors, perhaps owing to the fact that such elliptic integrals are well known and the values of which can
be looked up in tables or easily acquired by the use of modern computer software.
The restriction of Poisson’s ratio of the half-space to the value of 0:25 in many papers is in part due
to its usefulness for practical purposes in seismology, as the Lamé elastic constants of Earth material are
often assumed to be equal. However, this restriction also leads to certain simplifications of Rayleigh’s
equation, that is, the algebraic equation by which the speed of propagation of surface waves can be
found. This, in turn, facilitates the reduction of integral solutions into closed-form expressions. The
restriction on Poisson’s ratio of the solid is implemented in Pekeris’ papers [3,4] and Chao’s work [22]
for the reason just mentioned. Moreover, in the numerical calculations carried out by others, such as
Gakenheimer [33] and Johnson [34], this restriction is placed deliberately in order for the time-history
diagrams of the solutions to be comparable to those presented by earlier references.
Interestingly, in a paper published in 1979 studying the propagation of cracks in elastic solids by
Richards [41], closed-form expressions for the boundary solutions of LP can be found that are not
restricted to Poisson materials. However, as Kausel mentioned in a recent paper [42], no derivation of
Emami and Eskandari-Ghadi 11

the formulae is presented in Richards’ paper [41]. Kausel in his paper [42] revisiting the same version of
LP and utilizing the Cagniard–de Hoop method and algebraic manipulations, gave the necessary deriva-
tion steps that had been previously missing in Richards’ work [41]. In this regard, Kausel’s work [42]
includes the full set of closed-form surface transient solutions of LP due to a surface point load exerted
in an arbitrary direction that are valid for the whole range of the values of Poisson’s ratio. This indeed
complements and completes the classical contributions of Pekeris [3], Chao [22], and Richards [41] on
the boundary solutions of LP.
In addition, in a recent paper by Feng and Zhang [43], the authors have managed to reduce the sur-
face solutions given by Johnson [34] for a buried load of arbitrary direction into closed-form expressions.
This can be regarded as a generalization of Kausel’s study [42], in which the load has been assumed to
be buried instead of being exerted upon the surface. However, the closed-form expressions found by
Feng and Zhang [43] are only applicable for materials having their Poisson’s ratio in the range from zero
to 0:2631, instead of being valid for the full admissible range 1 to 0:5. Accordingly, a non-restricted set
of closed-form surface solutions of LP due to a buried load is still an open question, as is the case for
interior solutions due to buried loads.

5.3. Recent, more detailed derivations of the transient interior solutions of LP


As mentioned earlier, Johnson’s classical paper [34], although including the most general form of the
solutions for LP as integral-form expressions, gave no detail as to the derivation of the final solutions,
but only briefly outlined the steps of the Cagniard–de Hoop procedure, which are by no means trivial.
This lack has been sensed over the past decades and, for this reason, Emami and Eskandari-Ghadi [36]
have studied in great detail the transient interior solutions of LP for a surface load of arbitrary inclina-
tion. The method of solution in this paper [36] is influenced by, but somewhat different from, the origi-
nal ideas of Cagniard [2] and Pekeris [3,4], instead of the more common use of the Cagniard–de Hoop
method. Following their newly devised method, Emami and Eskandari-Ghadi [36] have succeeded to
give a full set of transient interior solutions of LP for surface loads as integral-form expressions, which
are not restricted in the Poisson’s ratio of the solid and are more elegant and brief in form than those
found in many earlier works. In this paper, the obtaining of solutions in the Laplace domain is greatly
aided by the use of a special case of potential functions of Eskandari-Ghadi, originally introduced as a
complete solution in x3 -convex domains for transversely isotropic solids [44]. It has been shown that the
potential functions just mentioned can be applied to a broad spectrum of elasticity problems for investi-
gating elastostatics and elastodynamics, from isotropic to transversely isotropic media [45].

5.4. Generalized versions of LP in stratified and anisotropic media


In relation to the problem of reflection and refraction at the interface plane of two distinct elastic solids,
as considered in Cagniard’s thesis [2], the earlier papers by Stoneley [46] and Muskat [47] may be cited.
In particular, Stoneley [46] extended Rayleigh’s work [9] for an infinite elastic medium composed of two
different materials that are welded at their interface plane, totally denoted by a bi-material full-space.
Stoneley showed that a surface time-harmonic wave similar to Rayleigh’s surface wave exists at the inter-
face plane of the bi-material solid [46].
When elastic media composed of two or several different materials are being considered, the wave
propagation problem becomes increasingly complicated. In particular, when instead of infinite or semi-
infinite media one introduces finite layers (strata), the mathematical solutions of transient wave propa-
gation take exceedingly intricate forms owing to the complications introduced by the mode changes and
reverberations of elastic waves within the strata. The researches in this category, being of great signifi-
cance in theoretical seismology, have been going on for a long time. In order for the transient solution
of problems involving stratified media to be found, methods such as the Ray Theory have been devel-
oped, for a comprehensive introduction to which, and for further references, we cite the excellent work
by Pao and Gajewski [30] (see also [48]). The investigations in stratified media and Ray Theory were
also intently followed by Pekeris and his colleagues, and among their many papers, [49–51] may be
cited. Moreover, among important works studying wave propagation in stratified media, the series of
12 Mathematics and Mechanics of Solids 00(0)

Table 1. An overview of some papers presenting exact transient solutions for the three-dimensional version of Lamb’s problem
(LP). The time-dependence of the point load is assumed to be Heaviside’s unit step function.

Paper containing exact Load orientation Load placement Solution type Solution form Restrictions on
transient solutions of Poisson’s
the three-dimensional ratio (s)a
version of LP
Vertical Horizontal Surface Buried Boundary Interior Integral- Closed-
solutions solutions form form

Pekeris [3] P P P P s = 0:25


Pekeris [4] P P P P s = 0:25
Chao [22] P P P P s = 0:25
Eason [23] P P P P No restriction
Gakenheimer [33] P P P P s = 0:25
Mooney [20] P P P P No restriction
Johnson [34] P P P P P P P No restriction
Richards [41] P P P P P No restriction
Kausel [42] P P P P P No restriction
Feng & Zhang [43] P P P P P 0\s\0:2631
Emami & Eskandari- P P P P P No restriction
Ghadi [36]
a
The full admissible range of Poisson’s ratio is 1 \ s \ 0:5 [60].

papers by de Hoop and Van der Hijden [52–54] and the monograph by Van der Hijden [55] are also
worth mentioning.
Because of increasing applications in engineering and sciences, studying LP for anisotropic solids is
another branch of interest that has been followed by some researches. For instance, in most seismic
problems, Earth is better modeled as a transversely isotropic solid having a vertical axis of isotropy
instead of a perfectly isotropic solid. Owing to the added number of elastic constants in anisotropic
solids, the equations of motion take on more intricate forms, and new forms of body waves can propa-
gate through the medium that cannot be classified strictly as either longitudinal and transverse waves;
see, for example, [44]. In fact, the speed of wave propagation is direction-dependent in general anisotro-
pic solids. These considerations add more mathematical difficulties to the already complicated solution
of LP. Among the investigations of anisotropic forms of LP, the papers by Burridge [56], Suh et al. [57],
Ben-Menahem and Sena [58], and Wang and Achenbach [59] may be mentioned.

6. Conclusions
We have seen that the elastodynamic problem first set forth by Lamb and later revisited, complemented,
and completed by many others has contributed tremendously to the mathematical theory of elasticity
and our better understanding of the phenomena related to the propagation of stress waves in semi-
infinite media. Apart from the direct practical uses of this problem in interpreting surface ground
motions during seismic events, as considered by Lamb in his paper [1], LP and its generalizations have
various important engineering applications including, but not restricted to, the following:

 implementing the time-domain boundary element method for direct wave propagation problems;
 detecting anomalies in elastic media by employing inverse methods;
 investigating transient noises in elastic media due to impulsive and/or moving loads.

As an overviewing conclusion and a quick guide for the reader, Table 1 has been prepared regarding
a number of papers presenting transient solutions of the three-dimensional version of LP. It should be
mentioned that the epicentral solutions, although being given in some of the papers included in Table 1,
are not considered in this table for brevity. Hence, only the papers presenting the full set of solutions
for interior points (not only for points lying on the epicenter of the load) are indicated in Table 1 as
containing the interior solutions. The dynamical point load is, as always, assumed to vary with time as
Heaviside’s unit step function.
Emami and Eskandari-Ghadi 13

Declaration of conflicting interests


The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this
article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article:
This work was partially supported by the University of Tehran through 27840/1/09 to ME-G.

ORCID iD
Morteza Eskandari-Ghadi https://orcid.org/0000-0002-7008-0654

References
[1] Lamb, H. On the propagation of tremors over the surface of an elastic solid. Philos Trans R Soc Lon 1904; 203: 1–42.
[2] Cagniard, L. Réflexion et re´fraction des ondes se´ismiques progressives. Paris: Gauthier-Villars, 1939.
[3] Pekeris, CL. The seismic surface pulse. Proc Natl Acad Sci 1955; 41: 469–480.
[4] Pekeris, CL. The seismic buried pulse. Proc Natl Acad Sci 1955; 41: 629-639.
[5] Pekeris, CL. Solution of an integral equation occurring in impulsive wave propagation problems. Proc Natl Acad Sci 1956;
42: 439–443.
[6] de Hoop, AT. A modification of Cagniard’s method for solving seismic pulse problems. Appl Sci Res B 1960; 8: 349-356.
[7] Eringen, AC, and Suhubi, ES. Elastodynamics. New York: Academic Press, 1975.
[8] Achenbach, JD. Wave propagation in elastic solids. Amsterdam: North-Holland, Elsevier Science Publishers, 1975.
[9] Rayleigh, L. On waves propagated along the plane surface of an elastic solid. Proc Lon Math Soc 1885; 1: 4–11.
[10] Love, AEH. A treatise on the mathematical theory of elasticity. New York: Dover Publications, 1944.
[11] Sneddon, IN. Fourier transforms. New York: McGraw-Hill (reprinted in 1995 by Dover Publications), 1951.
[12] Nakano, H. On Rayleigh waves. Jpn J Astrophys Geophys 1925; 2: 1-94.
[13] Lapwood, ER. The disturbance due to a line source in a semi-infinite elastic medium. Philos Trans R Soc Lon Ser A 1949;
242: 63–100.
[14] Garvin, WW. Exact transient solution of the buried line source problem. Proc R Soc Lon Ser A Math Phys Sci 1956; 234:
528-541.
[15] Ablowitz, MJ, and Fokas, AS. Complex variables: introduction and applications (second edition). Cambridge: Cambridge
University Press, 2003.
[16] Flinn, EA, and Dix, CH. Reflection and refraction of progressive seismic waves. New York: McGraw-Hill, 1962.
[17] Pinney, E. Surface motion due to a point source in a semi-infinite elastic medium. Bull Seismol Soc Am 1954; 44: 571-596.
[18] de Hoop, AT. Theoretical determination of the surface motion of a uniform elastic half-space produced by a dilatational,
impulsive, point source. Proc Colloque International du CNRS 1961; 111: 21-31.
[19] Pekeris, CL, and Lifson, H. Motion of the surface of a uniform elastic half-space produced by a buried pulse. J Acoust Soc
Am 1957; 29: 1233–1238.
[20] Mooney, HM. Some numerical solutions for Lamb’s problem. Bull Seismol Soc Am 1974; 64: 473–491.
[21] Pekeris, CL. A pathological case in the numerical solution of integral equations. Proc Natl Acad Sci USA 1940; 26:
433–437.
[22] Chao, CC. Dynamical response of an elastic half-space to tangential surface loadings. J Appl Mech 1960; 27: 559-567.
[23] Eason, G. The displacements produced in an elastic half-space by a suddenly applied surface force. IMA J Appl Math
1966; 2: 299-326.
[24] Dix, CH. The method of Cagniard in seismic pulse problems. Geophysics 1954; 19: 722-738.
[25] Dix, CH. Small changes in Cagniard’s method. J Geophys Res 1958; 63: 613-617.
[26] Abramovici, F. A generalization of the Cagniard method. J Computat Phys 1978; 29: 328-343.
[27] Murrell, HC., and Ungar, A. From Cagniard’s method for solving seismic pulse problems to the method of the differential
transform. Comput Math Appl 1982; 8: 103–118.
[28] Ben-Hador, R., and Buchen, P. A new approach to Cagniard’s problem. Appl Math Lett 1999; 12: 65-72.
[29] de Hoop, AT. Representation theorems for the displacement in an elastic solid and their application to elastodynamic
diffraction theory. Doctoral Thesis, Delft University of Technology, Delft, Netherlands, 1958.
[30] Pao, YH, and Gajewski, RR. The generalized ray theory and transient responses of layered elastic solids. Phys Acoust
1977; 13 183-265.
[31] Gakenheimer, DC. Transient excitation of an elastic half-space by a point load traveling on the surface. Doctoral Thesis,
California Institute of Technology, Pasadena, California, 1969.
14 Mathematics and Mechanics of Solids 00(0)

[32] Gakenheimer, DC, and Miklowitz, J. Transient excitation of an elastic half space by a point load traveling on the surface.
J Appl Mech 1969; 36: 505-515.
[33] Gakenheimer, DC. Numerical results for Lamb’s point load problem. J Appl Mech 1970; 37: 522-524.
[34] Johnson, LR. Green’s function for Lamb’s problem. Geophys J R Astronom Soc 1974; 37: 99–131.
[35] Longman, IM. Solution of an integral equation occurring in the study of certain wave-propagation problems in layered
media. J Acoust Soc Am 1961; 33: 954-958.
[36] Emami, M, and Eskandari-Ghadi, M. Transient interior analytical solutions of Lamb’s problem. Math Mech Solid 2019;
24: 3485–3513.
[37] Coddington, EA. An introduction to ordinary differential equations. New York: Dover Publications, Inc., 1989.
[38] Watson, GN. A treatise on the theory of Bessel functions. Cambridge: Cambridge University Press, 1922.
[39] Churchill, RV. Operational mathematics. New York: McGraw-Hill, 1972.
[40] Ahlfors, LV. Complex analysis. New York: McGraw-Hill, 1966.
[41] Richards, PG. Elementary solutions to Lamb’s problem for a point source and their relevance to three-dimensional studies
of spontaneous crack propagation. Bull Seismol Soc Am 1979; 69: 947-956.
[42] Kausel, E. Lamb’s problem at its simplest. Proc R Soc Lon Ser A 2012; 469: 462–499.
[43] Feng, X, and Zhang, H. Exact closed-form solutions for Lamb’s problem. Geophys J Int 2018; 214: 444–459.
[44] Eskandari-Ghadi, M. A complete solution of the wave equations for transversely isotropic media. J Elastic 2005; 81: 1–19.
[45] Eskandari-Ghadi, M, and Pak, RYS. Elastodynamics and elastostatics by a unified method of potentials for x3-convex
domains. J Elastic 2008; 92: 187–194.
[46] Stoneley, R. Elastic waves at the surface of separation of two solids. Proc R Soc Lon A 1924; 106: 416–428.
[47] Muskat, M. The theory of refraction shooting. Physics 1933; 4: 14-28.
[48] Pao, YH, Gajewski, RR, and Thau, SA. Analysis of ground wave propagation in layered media. Washington, D.C.: Defense
Nuclear Agency, 1971.
[49] Pekeris, CL, and Longman, IM. Ray-theory solution of the problem of propagation of explosive sound in a layered liquid.
J Acoust Soc Am 1958; 30: 323-328.
[50] Pekeris, CL, Alterman, Z, Abramovici, F, et al. Propagation of a compressional pulse in a layered solid. Rev Geophys
1965; 3: 25–47.
[51] Pekeris, CL, Alterman, Z, and Abramovici, F. Propagation of an SH-torque pulse in a layered solid. Bull Seismol Soc Am
1963; 53: 39-57.
[52] de Hoop, AT, and Van der Hijden, JH. Generation of acoustic waves by an impulsive line source in a fluid/solid
configuration with a plane boundary. J Acoust Soc Am 1983; 74: 333-342.
[53] de Hoop, AT, and Van der Hijden, JH. Generation of acoustic waves by an impulsive point source in a fluid/solid
configuration with a plane boundary. J Acoust Soc Am 1984; 75: 1709–1715.
[54] de Hoop, AT, and Van der Hijden, JH. Seismic waves generated by an impulsive point source in a solid/fluid configuration
with a plane boundary. Geophysics 1985; 50: 1083–1090.
[55] Van Der Hijden, JH. Propagation of transient elastic waves in stratified anisotropic media. New York: Elsevier, 2016.
[56] Burridge, R. Lamb’s problem for an anisotropic half-space. Q J Mech Appl Math 1971; 24: 81-98.
[57] Suh, SL, Goldsmith, W, Sackman, JL, et al. Impact on a transversely anisotropic half-space. Int J Rock Mech Mining Sci
Geomech Abs 1974; 11: 413–421.
[58] Ben-Menahem, A, and Sena, AG. The elastodynamic Green’s tensor in an anisotropic half-space. Geophys J Int 1990; 102:
421–443.
[59] Wang, CY, and Achenbach, JD. Lamb’s problem for solids of general anisotropy. Wave Motion 1996; 24: 227-242.
[60] Gurtin, ME. The linear theory of elasticity. In: Fhügge, S (ed.) Handbuch der physik, vol. via/2, mechanics of solids II.
Berlin, Heidelberg, New York: Springer, 1972.

You might also like