Asymptotic Theory For Rayleigh and Rayleigh-Type Waves: Julius Kaplunov, Danila A. Prikazchikov

CHAPTER ONE
Asymptotic Theory for Rayleigh

and Rayleigh-Type Waves
Julius Kaplunov1, Danila A. Prikazchikov
School of Computing and Mathematics, Keele University, Keele, United Kingdom
1
Corresponding author: e-mail address: j.kaplunov@keele.ac.uk
Contents
1. Introduction 2
2. Time-Harmonic Rayleigh Wave on an Elastic Half-Space 4
2.1 Elementary Derivation 4
2.2 Rayleigh Wave of Arbitrary Profile 8
3. Rayleigh-Type Interfacial and Edge Elastic Waves 12
3.1 Scho€lte–Gogoladze Wave 13
3.2 Stoneley Wave 15
3.3 Bending Edge Wave on a Thin Plate 16
4. Hyperbolic–Elliptic Model for the Rayleigh Wave Induced by Surface Stresses 19
4.1 Plane Strain Problem 19
4.2 3D Problem 24
4.3 Examples 29
5. Generalizations of the Hyperbolic–Elliptic Model 37
5.1 Coated Half-Space 37
5.2 Mixed Boundary Value Problems 41
5.3 Interfacial Waves 45
5.4 Bleustein–Gulyaev Wave 48
5.5 Effect of Anisotropy 50
6. Moving Load Problems 53
6.1 Plane Strain Steady-State Problem 53
6.2 Transient Plane Strain Problem 56
6.3 3D Steady-State Problem 65
6.4 3D Steady-State Problem for a Coated Half-Space 71
7. Parabolic–Elliptic Model for a Bending Edge Wave on a Thin Plate 81
7.1 Bending Edge Wave of Arbitrary Profile 81
7.2 Bending Edge Wave Induced by Prescribed Moments and Shear Forces 85
7.3 Plate on Elastic Foundation 91
7.4 Examples 93
8. Conclusion 96
Acknowledgments 97
Advances in Applied Mechanics, Volume 50 # 2017 Elsevier Inc. 1

ISSN 0065-2156 All rights reserved.
https://doi.org/10.1016/bs.aams.2017.01.001
2 Julius Kaplunov and Danila A. Prikazchikov
Appendix A. Near-Resonant Behavior of a Single Degree of Freedom Linear

Oscillator 97
Appendix B. Exact Analysis of Plane Moving Load Problems 98
References 101
Abstract
Explicit asymptotic formulations are derived for Rayleigh and Rayleigh-type interfacial
and edge waves. The hyperbolic–elliptic duality of surface and interfacial waves is
established, along with the parabolic–elliptic duality of the dispersive edge wave on
a Kirchhoff plate. The effects of anisotropy, piezoelectricity, thin elastic coatings, and
mixed boundary conditions are taken into consideration. The advantages of the devel-
oped approach are illustrated by steady-state and transient problems for a moving load
on an elastic half-space.
1. INTRODUCTION
A time-harmonic surface wave on a linearly elastic isotropic half-
space was discovered by Rayleigh (1885). This pioneering piece of work
was inspired by the needs of seismology, including earthquake prediction.
Later on, similar Rayleigh-type waves were found for solid–solid and fluid–
solid interfaces, see Stoneley (1924), Gogoladze (1948), and Sch€ olte
(1949), as well as for the edge of a thin Kirchhoff plate, investigated by
Konenkov (1960). We also mention a piezoelastic surface wave studied
by Bleustein (1968) and Gulyaev (1969), important for various applications,
see Campbell (1998) and references therein. Nowadays surface waves are
also widely used in the theory and practice of nondestructive evaluation,
see, e.g., Gudra and Stawiski (2000). Another recent application is associ-
ated with cloaking of surface waves and seismic metamaterials, see Brûle,
Javelaud, Enoch, and Guenneau (2014) and Colombi, Colquitt, Roux,
Guenneau, and Craster (2016).
Current trends in this area include taking into consideration inhomoge-
neity, anisotropy, and prestress with one of the main focuses on the existence
and uniqueness of localized time-harmonic eigensolutions, see references in
the introductory Sections 2 and 3, which also recall the standard derivations
of well-known dispersion relations along with elementary proofs of exis-
tence and uniqueness. However, in spite of a substantial interest in the topic,
the Rayleigh wave for a long time seemed to be somehow hidden within the
classical elasticity model. In particular, the related Lame wave potentials, e.g.,
see Achenbach (2012), govern the bulk waves but not the surface one. At the
Asymptotic Theory for Rayleigh and Rayleigh-Type Waves 3
same time the Rayleigh wave contribution often dominates in the overall
dynamic response, including the case of a resonant surface excitation. It also
usually prevails over a far-field zone near the surface. These observations
motivate the derivation of a specialized formulation oriented to the
Rayleigh wave.
In Section 4 we develop a multiscale perturbation procedure for an elas-
tic half-space subject to prescribed surface stresses. The peculiarities of the
procedure are clarified in Appendix A by a simple example of a single degree
of freedom linear oscillator. In Section 4.1 the 2D dynamic equations of the
plane strain problem are perturbed around the eigensolution for a surface
wave of arbitrary profile obtained in Chadwick (1976b), see also earlier
papers of Friedlander (1948) and Sobolev (1937), as well as more recent pub-
lications, including Achenbach (1998), Kiselev (2004), Parker and Kiselev
(2009), Kiselev and Parker (2010), Rousseau and Maugin (2010),
Prikazchikov (2013), Parker (2013), and Kiselev (2015), treating homoge-
neous Rayleigh and Rayleigh-type waves in a more general setup. It appears
that this eigensolution can be expressed in terms of a single harmonic func-
tion. As a result, we arrive at a hyperbolic–elliptic theory. This involves a
wave equation for one of the Lame potentials that governs propagation of
surface disturbances along with pseudo-static elliptic equations for calculat-
ing the Lame potentials over the interior. The derived model is extended to
the general 3D case in Section 4.2 using the integral Radon transform.
The proposed formulation reflects a duality of the Rayleigh wave.
Indeed, hyperbolicity stands for propagation along the surface, whereas
ellipticity may be associated with decay into the interior. At the same time,
it should be noted that hyperbolicity is characteristic only for one of the
Lame potentials along the surface, see Erbaş and Şahin (2016) for more
detail.
A similar theory is also established in Section 5 for several Rayleigh-type
waves, including interfacial waves and the surface wave on a coated half-
space. For the latter, the hyperbolic equation along the surface is singularly
perturbed by a pseudo-differential operator. The effects of anisotropy and
mixed boundary conditions are also addressed in this section, along with
the extension to piezoelastic surface waves.
The validity of the proposed models is tested in Section 6 by comparison
with the exact solution of steady-state and transient plane strain moving load
problems, given in Appendix B and Kaplunov, Nolde, and Prikazchikov
(2010), respectively. The near-resonant regimes of moving loads apparently
present the optimal framework for evaluating the accuracy of the asymptotic
approach. In this section we also obtain the explicit solutions of 3D moving

load problems for a homogeneous and coated half-space in terms of elemen-
tary functions.
Finally, in Section 7 we derive a parabolic–elliptic model for a dispersive
bending wave propagating along the edge of a thin elastic plate. In this case
the adapted asymptotic procedure also perturbs in slow time the eigen-
solution corresponding to an edge wave of general shape expressed through
a single plane harmonic function. This eigenfunction is obtained starting
from an implicit ansatz, for which the counterpart for the Rayleigh wave
is the classical wave equation.
2. TIME-HARMONIC RAYLEIGH WAVE ON AN ELASTIC

HALF-SPACE
In this section we present some basic results for surface waves, includ-
ing the derivation of the classical Rayleigh equation as well as analysis of a
surface wave of arbitrary profile, preceding the development of the explicit
hyperbolic–elliptic formulation for the Rayleigh wave field.
2.1 Elementary Derivation

Let us first derive the original surface wave equation discovered by Rayleigh
(1885). Consider a linearly elastic isotropic half-space
H3+ ¼ fðx1 ;x2 ;x3 Þ j ∞ < x1 < ∞, ∞ < x2 < ∞, 0 x3 < ∞g:
The equations of motion in linear elastodynamics are given by
σ ij, j ¼ ρui, tt , i, j ¼ 1, 2,3, (1)
e.g., see Achenbach (2012). Here σ ij and ui are the components of the
Cauchy stress tensor σ and the displacement vector u, respectively, ρ denotes
mass volume density, and comma indicates differentiation with respect to
spatial or time variables. Einstein’s summation convention is adopted throu-
ghout this chapter, unless otherwise stated. The stress–strain relations for an
isotropic solid are given by
σ ij ¼ λuk, k δij + 2μεij , i, j ¼ 1,2,3, (2)
with λ and μ being the Lame elastic moduli, δij standing for the Kronecker
delta, and the kinematic relations for the components of the strain tensor ε
being expressed as
1
εij ¼ ui, j + uj, i : (3)
2
On substituting the constitutive relations (2) into the equations of motion (1)
and taking into consideration the relations (3), we have the wave equation
ðλ + μÞuj, ji + μ ui, jj ¼ ρui, tt : (4)
Let us now decompose the displacement field according to the Helm-
holtz theorem as
u ¼ grad ϕ + curl Ψ, (5)
where ϕ and Ψ¼ ðψ 1 ,ψ 2 , ψ 3 Þ are scalar longitudinal and vector transverse
elastic potentials, respectively, allowing separation of extensional and shear
motions for an isotropic solid. In addition, the constraint
div Ψ ¼ 0, (6)
is required. This condition is not unique, e.g., see Miklowitz (1978). On
substituting the Helmholtz decomposition (5) into (4), we have
ϕ, tt c12 Δϕ ¼ 0, Ψ, tt c22 ΔΨ ¼ 0, (7)
where Δ is the 3D Laplace operator in x1, x2 and x3, and
rffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffi
λ + 2μ μ
c1 ¼ , c2 ¼ (8)
ρ ρ
are the longitudinal and transverse wave speeds, respectively.
The stress-free boundary conditions are imposed on the surface x3 ¼ 0, i.e.,
σ 3i ¼ 0, i ¼ 1,2,3, (9)
or, in terms of the elastic potentials,
2ϕ, 13 ψ 1, 12 + ψ 2, 11 ψ 2, 33 + ψ 3, 23 ¼ 0,
2ϕ ψ 1, 22 + ψ 1, 33 + ψ 2, 12 ψ 3, 13 ¼ 0, (10)
, 23
λ ϕ, 11 + ϕ, 22 + ðλ + 2μÞϕ, 33 + 2μ ψ 2, 13 ψ 1, 23 ¼ 0:
The solutions of (7) and (10) are now sought in the form of the traveling
harmonic wave
ðϕ,ψ 1 ,ψ 2 , ψ 3 Þ ¼ ðA1 , A2 , A3 ,A4 Þeikðx1 cosθ + x2 sin θctÞkqx3 , (11)
where Ai, i ¼ 1,4, are arbitrary constants, k is the wave number, q is the
attenuation factor to be determined and c is the sought for phase speed;
the condition RðqÞ > 0 is assumed, ensuring decay of the surface wave field
as x3 ! ∞. On substituting (11) into the equations of motion (7), we obtain
ϕ ¼ A1 eikðx1 cos θ + x2 sin θctÞkαx3 , (12)

and
ðψ 1 , ψ 2 ,ψ 3 Þ ¼ ðA2 ,A3 ,A4 Þeikðx1 cosθ + x2 sin θctÞkβx3 , (13)
with the attenuation orders
sffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffi
c2 c2
α ¼ 1 2, β ¼ 1 2: (14)
c1 c2
Then, on inserting (12) and (13) into the boundary conditions (10) and using
the constraint (6), we obtain a set of linear algebraic equations in respect of
the constants Ai, namely
2iαcos θA1 + sin θ cosθA2 ð cos 2 θ + β2 ÞA3 iβ sinθA4 ¼ 0,
2iα sinθA1 + ðsin 2 θ + β2 ÞA2 sinθ cosθA3 + iβ cosθA4 ¼ 0,
(15)
ð1 + β2 ÞA1 + 2iβ sinθA2 2iβ cos θA3 ¼ 0,
icos θA2 + i sinθA3 + βA4 ¼ 0:
The determinant of (15) equals zero provided that
ð1 + β2 Þ2 4αβ ¼ 0: (16)
The latter is the well-known Rayleigh equation. It is often presented as
RðrÞ ¼ 0, (17)
where
pffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffi
RðrÞ ¼ ð2 rÞ2 4 1 r 1 κ2 r , (18)
with
c2 c2
r ¼ 2, κ ¼ < 1: (19)
c2 c1
The first proof of the existence and uniqueness of the solution of (17) was
seemingly presented by Sobolev (1937), see also Babich and Kiselev (2014).
A similar proof is given below, showing that the Rayleigh equation (17) has
a unique solution for the phase speed c ¼ cR over the interval 0 < r < 1. First,
note that
Rð0Þ ¼ 0, Rð1Þ ¼ 1, (20)
while the derivative
R0 ð0Þ ¼ 2ðκ 2 1Þ < 0: (21)

In addition, the second derivative
ð1 κ2 Þ2
R00 ðrÞ ¼ 2 + (22)
½ð1 rÞð1 κ2 rÞ3=2
is positive for 0 < r < 1, so the function R(r) is concave upward over the
interval. Thus, the conditions (20) and (21) imply the existence of a zero
of R(r) in the interval 0 < r < 1, with (22) ensuring uniqueness of solution.

1 1
A typical behavior of R(r) is shown in Fig. 1 for κ ¼ ν ¼ .
3 4
It is obvious that the solution of (17) for a given Poisson ratio ν is a
constant, i.e., the Rayleigh wave is nondispersive. A variation of the scaled
Rayleigh wave speed cR/c2 vs the Poisson ratio is shown in Fig. 2.
It is also well known from the original paper of Rayleigh (1885) that (17)
may be transformed to the cubic equation

r 3 8ðr 1Þ r 2ð1 κ 2 Þ ¼ 0: (23)
0.8
0.6
0.4
R (r)
0.2
–0.2
–0.4
0 0.2 0.4 0.6 0.8 1
r
1
Fig. 1 Variation of R(r) along the interval 0 < r < 1 for ν ¼ .
4
0.96
0.95
0.94
0.93
0.92
cR
c2
0.91
0.9
0.89
0.88
0.87
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
n
Fig. 2 The Rayleigh wave speed vs the Poisson ratio ν.
The solution of this equation may be obtained through Cardano’s formula,

see, e.g., Malischewsky (2000) and Vinh and Ogden (2004). It should be
noted that existence and uniqueness of the Rayleigh wave have been also
proved for anisotropic materials, e.g., see Barnett and Lothe (1974) and
Kamotskii and Kiselev (2009). Obviously, the evaluation of the surface wave
speed becomes more difficult in anisotropic medium. In this case the
so-called surface–impedance matrix technique, leading to a matrix Riccati
equation, see Fu and Mielke (2002), proved to be robust. We also mention
contributions studying surface waves in media with more sophisticated
properties, see, e.g., Hayes and Rivlin (1961), Alenitsyn (1963), Agarwal
(1978), Ivanov (1988), Dowaikh and Ogden (1990), Rogerson (1996),
Destrade (2004), Destrade (2007), Steigmann and Ogden (2007), Kiselev
and Rogerson (2009), Dockrey et al. (2013), Vinh, Anh, and Thanh
(2014), and many others.
2.2 Rayleigh Wave of Arbitrary Profile

In the previous section the solution was sought in the form of a traveling
time-harmonic plane wave. Below we generalize it to a wave of general
time-dependence, relying on the key contribution of Chadwick (1976b),
see also the earlier works of Friedlander (1948) and Sobolev (1937). Follow-
ing these papers, we consider a plane strain problem, assuming
u2 ¼ 0, ui ¼ ui ðx1 , x3 , tÞ, i ¼ 1,3, (24)

for the half-plane
H2+ ¼ fðx1 ; x3 Þ j ∞ < x1 < ∞, 0 x3 < ∞g: (25)

In this case the displacement field is conventionally expressed through
the 2D elastic potentials ϕ and ψ as
@ϕ @ψ @ϕ @ψ
u1 ¼ , u3 ¼ + , (26)
@x1 @x3 @x3 @x1
satisfying the wave equations

ϕ, tt c12 ϕ, 11 + ϕ, 33 ¼ 0, ψ , tt c22 ψ , 11 + ψ , 33 ¼ 0: (27)
together with the boundary conditions

2ϕ, 13 + ψ , 11 ψ , 33 ¼ 0,
2 (28)
κ 2 ϕ, 11 + κ2 ϕ, 33 + 2ψ , 13 ¼ 0,
along the stress-free surface x3 ¼ 0, where, as before,

rffiffiffiffiffiffiffiffiffiffiffiffi
c2 1 2ν
κ¼ ¼ :
c1 2 2ν
Let us study the elastic potentials of the form
ϕ ¼ ϕðx1 ct,x3 Þ, ψ ¼ ψðx1 ct, x3 Þ, (29)
corresponding to a wave of arbitrary shape propagating at a speed c. Using
(29), the equations of motion (27) are reduced to the elliptic equations
ϕ, 33 + α2 ϕ, 11 ¼ 0, ψ , 33 + β2 ψ , 11 ¼ 0, (30)
where α and β are defined by (14). Thus, the eigensolutions for the elastic
potentials are
ϕ ¼ ϕðx1 ct, αx3 Þ, ψ ¼ ψðx1 ct, βx3 Þ, (31)
being plane harmonic functions. In the subsequent analysis we employ the
Cauchy–Riemann identities
1
f, 3 ¼ γf,*1 , f, 1 ¼ f,*3 , f ** ¼ f , (32)
γ
for a harmonic function f(x1, γx3), being a solution of the equation

f, 33 + γ 2 f, 11 ¼ 0, γ ¼ ðα, βÞ,
where the asterisk denotes a harmonic conjugate function, see, e.g.,
Titchmarsh (1939).
After straightforward manipulations involving the aforementioned
Cauchy–Riemann identities (32), the boundary conditions (28) at x3 ¼ 0
may be rearranged as
2α ϕ*, 11 + ð1 + β2 Þ ψ , 11 ¼ 0,
(33)
ð1 + β2 Þ ϕ, 11 + 2β ψ *, 11 ¼ 0,
Then, on taking conjugate of the first equation, we deduce from the solv-
ability of (33) that
ð1 + β2 Þ2 4αβ ¼ 0, (34)
which is the Rayleigh equation, see (16). Therefore c ¼ cR, hence, the sought
for harmonic eigenfunctions are
ϕ ¼ ϕðx1 ct,αR x3 Þ, ψ ¼ ψðx1 ct, βR x3 Þ, (35)
where
sffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffi
c2 c2
αR ¼ 1 R2 , βR ¼ 1 R2 : (36)
c1 c2
In addition, we obtain the relations between the potentials along the surface
x3 ¼ 0, following from (28), namely
2 1 + β2R
ψ,1 ¼ ϕ, 3 , ψ,3 ¼ ϕ, 1 : (37)
1 + β2R 2
Moreover, the maximum principle for harmonic functions implies a condi-
tion relating the potentials ϕ and ψ not only along the surface x3 ¼ 0, but
over the entire half-plane. Thus,
2αR
ψðx1 cR t, βR x3 Þ ¼ ϕ*ðx1 cR t, βR x3 Þ, (38)
1 + β2R
and
2βR
ϕðx1 cR t, αR x3 Þ ¼ ψ*ðx1 cR t, αR x3 Þ, (39)
1 + β2R
for more details see Chadwick (1976b). This allows expressing the displace-
ments in terms of a single plane harmonic function, say in terms of the
potential ϕ, as
1 + β2R
u1 ðx1 ,x3 ,tÞ ¼ ϕ, 1 ðx1 cR t, αR x3 Þ ϕ, 1 ðx1 cR t, βR x3 Þ,
2
(40)
2
u3 ðx1 ,x3 ,tÞ ¼ ϕ, 3 ðx1 cR t, αR x3 Þ ϕ, 3 ðx1 cR t, βR x3 Þ:
1 + β2R
The obtained representation (40) extends the class of decaying
eigensolutions, in particular, allowing a nonperiodic behavior along the sur-
face. In fact, the potentials ϕ and ψ generally may not decay at infinity. As an
example, we take
sR αR
ϕðsR , x3 Þ ¼ tan 1 , ψðsR , x3 Þ ¼ ln s2R + ðβR x3 + aÞ2 , (41)
α R x3 + a 1 + βR
2
where a > 0 is a parameter chosen in order to smooth discontinuities at the

origin and sR ¼ x1 cRt is a moving coordinate. The potentials (41) corre-
spond to the following decaying displacement components
αR x3 + a 1 + β2R β R x3 + a
u1 ðx1 , x3 ,tÞ ¼ 2 ,
sR + ðαR x3 + aÞ
2 2 sR + ðβR x3 + aÞ2
2
(42)
αR sR 2 β R sR
u3 ðx1 , x3 ,tÞ ¼ 2 2 + 2:
sR + ðαR x3 + aÞ 1 + β 2 2
R sR + ðβR x3 + aÞ
Variation of the displacements (42) on sR at several depths is shown in

1
Fig. 3 for a ¼ 0.1 and the Poisson’s ratio ν ¼ . An example of nondecaying
3
displacements is studied in Section 6.1.
It is clear that similar derivation may be performed for a surface wave
traveling in the opposite direction, when the argument x1 ct in (11) is rep-
laced by x1 + ct. Moreover, it is also possible to take into account simulta-
neously surface waves propagating along both directions. To this end, we
should assume an implicit traveling wave ansatz given by
ϕ, tt c 2 ϕ, 11 ¼ 0, ψ , tt c 2 ψ , 11 ¼ 0, (43)
where ϕ ¼ ϕ(x1, x3, t) and ψ ¼ ψ(x1, x3, t). It is easily verified that on
employing the last assumptions, the wave equations (27) become the elliptic
equations (30), with all the following derivations above being perfectly valid
and leading to the eigensolution of arbitrary profile (40).
To conclude this section, we note that the obtained representation of the
Rayleigh wave field in terms of a single harmonic function may be gener-
alized to the 3D case. More details on the subject may be found in the recent
contribution by Kiselev and Parker (2010). The same results also follow from
a more general analysis in Dai, Kaplunov, and Prikazchikov (2010).
A
1
x3 = 1
x3 =0.5
x3 = 0.3
0.5
u1
–0.5
–1
–3 –2 –1 0 1 2 3
sR
B
1.5
x3 = 1
x3 =0.5
1 x3 = 0.3
0.5
u3
–0.5
–1
–1.5
–3 –2 –1 0 1 2 3
sR
Fig. 3 The profiles of (A) horizontal and (B) vertical displacements (42) vs the moving
coordinate sR.
3. RAYLEIGH-TYPE INTERFACIAL AND EDGE

ELASTIC WAVES
In this section we discuss several extensions of the Rayleigh elastic
wave, including the waves localized at solid–fluid and solid–solid interfaces.
In addition, we derive the dispersion relation for the bending wave propa-
gating along the edge of a semiinfinite thin elastic plate. All of these waves
were originally referred to as Rayleigh-type waves, see Stoneley (1924),

Sch€
olte (1949), Gogoladze (1948), and Konenkov (1960).
€lte–Gogoladze Wave
3.1 Scho
Let us consider the half-plane H2+ composed of a linearly elastic isotropic
material, see (25), contacting with the half-plane
H
2 ¼ fðx1 ; x3 Þ j ∞ < x1 < ∞, ∞ < x3 0g,
occupied by an ideal fluid. We concentrate on the wave propagating along

the interface x3 ¼ 0 and decaying away from it. This wave was discovered
independently by Sch€ olte (1949) and Gogoladze (1948) and is usually named
after these authors.
The equations of motion for the elastic medium are given by (27),
whereas the fluid motion is governed by the wave equation

χ, tt cf2 χ , 11 + χ , 33 ¼ 0, (44)
where χ is the displacement potential and cf is the sound wave speed. The
interfacial conditions at x3 ¼ 0 are written as
σ 31 ¼ 0, u3 ¼ v, σ 33 ¼ pf , (45)
where v and pf are the vertical displacement and pressure in the fluid, respec-
tively, given by
v ¼ χ, 3, pf ¼ ρf χ , tt ,
with ρf denoting the fluid volume density.
The conditions (45) may be presented as
2ϕ, 13 ψ , 11 + ψ , 33 ¼ 0,
ϕ, 3 + ψ , 1 χ , 3 ¼ 0, (46)

μ κ2 2 ϕ, 11 + κ2 ϕ, 33 2ψ , 13 ρf χ , tt ¼ 0:
The solution for the potentials is found from the wave equations (27) and
(44) in the form of a traveling harmonic wave ensuring decay away from the
interface into the interior, namely

ðϕ,ψ, χ Þ ¼ A1 ekαx3 , A2 ekβx3 , A3 ekγx3 eikðx1 ctÞ , (47)
where, as before, k is the wave number, c is the sought for phase speed, the
attenuation orders α and β are defined by (14), and
sffiffiffiffiffiffiffiffiffiffiffi
c2
γ ¼ 1 2: (48)
cf
Substitution of (47) into (46) results in a set of linear algebraic equations in

respect of the constants A1, A2, and A3. The solvability of the latter implies
the related determinant being zero, giving
2 ρf αSG 2
1 + β2SG 4αSG βSG + 1 β2SG ¼ 0, (49)
ρ γ SG
where
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c2 c2 c2
αSG ¼ 1 SG , βSG ¼ 1 SG , γ SG ¼ 1 SG , (50)
c12 c22 cf2
with cSG denoting the Sch€ olte–Gogoladze wave speed. It is may be shown
that the sought for solution always exists in the interval 0 < cSG < cf provided
that cf < c2 < c1, see, e.g., Gogoladze (1948), and also Viktorov (1981), tackling
a more general setup. Indeed, let us denote the left-hand side of (49) by
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffi ρf 1 κ 2 r 2
SGðrÞ ¼ ð2 rÞ 4 1 r 1 κ2 r + qffiffiffiffiffiffiffiffiffiffiffiffiffiffi r
2
(51)
ρ 1 κ 2f r
where
κf ¼ c2 =cf > 1, (52)
with r and κ defined by (19). It is then clear that
SGð0Þ ¼ 0, lim SGðrÞ ¼ + ∞, SG0 ð0Þ ¼ 2ðκ 2 1Þ < 0, (53)

r!κ2
f
0
guaranteeing existence of the solution over the interval 0 < r < κ 2

f . It is also
known that leaky interfacial waves are possible, see, e.g., Zhu, Popovics, and
Schubert (2004) and references therein.
Similarly to the surface wave of arbitrary profile considered in
Section 2.2, the eigensolution for the Sch€ olte–Gogoladze wave of general
time dependence may be derived, as shown by Kiselev and Parker (2010)
and Parker (2012).
3.2 Stoneley Wave

Let us now consider the wave propagating along the interface of two elastic
half-planes H2 and H2 . This wave was first discovered by Stoneley (1924).
+
Unlike the Rayleigh wave, it only exists for a restricted range of material
parameters. The existence conditions for two bonded isotropic half-spaces
have been obtained by Sch€ olte (1947). It is also known from Barnett, Lothe,
Gavazza, and Musgrave (1985) that the Stoneley wave speed exceeds the
smaller of the Rayleigh wave speeds for the two half-spaces. The cited con-
tribution of Barnett et al. (1985) also contains rigorous results on existence of
the Stoneley wave on the interface of two anisotropic half-spaces. Similarly
to the Rayleigh wave, the surface–impedance matrix method may be used
for robust computations of the Stoneley wave speed, see Destrade and Fu
(2006). We also mention contributions dealing with Stoneley waves in
media with more sophisticated properties, see, e.g., Dowaikh and Ogden
(1991), Goda (1992), and Vinh and Seriani (2010).
For the sake of simplicity, we once again restrict ourselves to the frame-
work of the plane strain assumption. The equations of motion are written in
terms of the elastic wave potentials as

ϕn, tt c1n
2
ϕn, 11 + ϕn, 33 ¼ 0, ψ n, tt c2n
2
ψ n, 11 + ψ n, 33 ¼ 0, (54)
where the suffix n ¼ 1, 2 corresponds to the elastic media occupying H2+ and
H2 , respectively, and c1n and c2n denote the longitudinal and transverse wave
speeds in these media. The interfacial conditions at x3 ¼ 0 for perfectly
bonded half-spaces are given by
ϕ1, 1 ϕ2, 1 + ψ 1, 3 ψ 2, 3 ¼ 0,
ϕ1, 3 ϕ2, 3 ψ 1, 1 + ψ 2, 1 ¼ 0,

2μ1 ϕ1, 13 2μ2 ϕ2, 13 + μ1 ψ 1, 33 ψ 1, 11 μ2 ψ 2, 33 ψ 2, 11 ¼ 0,
λ1 ϕ1, 11 + ðλ1 +2μ1 Þϕ1, 33 λ2 ϕ2, 11 ðλ2 +2μ2 Þϕ2, 33 2μ1 ψ 1, 13 + 2μ2 ψ 2, 13 ¼ 0,
(55)
where λn and μn, n ¼ 1, 2, are the Lame elastic moduli.

As usual, the potentials are found from (54) in the form of a traveling
harmonic wave decaying away from the interface, namely

ðϕ1 ,ψ 1 , ϕ2 , ψ 2 Þ ¼ A1 ekα1 x3 ,A2 ekβ1 x3 ,A3 ekα2 x3 , A4 ekβ2 x3 eikðx1 ctÞ , (56)
where
sffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffi
c2 c2
αn ¼ 1 2 , βn ¼ 1 2 , n ¼ 1,2: (57)
c1n c2n
On substituting (56) into (55), we deduce the well-known equation

cS4 ðρ1 ρ2 Þ2 a1 a2 + 2cS2 m12 ðρ2 b1 ρ1 b2 Þ + m212 b1 b2 ¼ 0, (58)
in which cS is the Stoneley wave speed and
a1 ¼ ρ1 α2S + ρ2 α1S , a2 ¼ ρ1 β2S + ρ2 β1S ,
(59)
bn ¼ 1 αn βn , m12 ¼ 2ðμ1 μ2 Þ,
with
sffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffi
c2 c2
αnS ¼ 1 2S , βnS ¼ 1 2S , n ¼ 1,2: (60)
c1n c2n
An interfacial Stoneley wave of arbitrary profile have been investigated

by Chadwick (1976b), resulting in the representation of the displacement
field in terms of a single plane harmonic function, similarly to the Rayleigh
wave. These results have been recently generalized to three dimensions by
Kiselev and Parker (2010). The cited papers allow reduction of a vector
problem in linear elasticity for both surface and interfacial waves to a scalar
Dirichlet problem for the Laplace equation. Thus, the analysis of interfacial
waves turns out to be not much different compared to that of the
Rayleigh waves.
3.3 Bending Edge Wave on a Thin Plate

Consider now the Rayleigh-type bending wave on a semiinfinite elastic
plate of thickness 2h occupying the region ∞ < x1 < ∞, 0 x2 < ∞,
h x3 h. This wave has been originally discovered by Konenkov
(1960); however, clear hints may already be seen in the earlier work of
Ishlinsky (1954) studying stability of a thin plate. It is peculiar that this wave
was rediscovered several times, for more detail see Norris, Krylov, and
Abrahams (2000) and Lawrie and Kaplunov (2012).
Within the framework of the classical Kirchhoff plate theory, the deflec-
tion of the midplane W ¼ W(x1, x2, t) is governed by
DΔ2 W + 2ρhW, tt ¼ 0, (61)
where ρ is volume mass density, Δ is the Laplacian in the variables x1 and x2,
and the bending stiffness D is given by
2Eh3
D¼ , (62)
3ð1 ν2 Þ
with E and ν denoting the Young’s modulus and the Poisson ratio,
respectively.
In the absence of bending moments and modified shear forces the
boundary conditions at the edge x2 ¼ 0 are written as
W, 22 + νW, 11 ¼ 0,
(63)
W, 222 + ð2 νÞW, 112 ¼ 0:
The solution of (61) is found in the form of a traveling harmonic wave as
X
2
W ðx1 , x2 ,tÞ ¼ Aj eiðkx1 ωtÞkγ j x2 , (64)
j¼1

where the attenuation coefficients γ j are given by Rðγ j Þ > 0

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rffiffiffiffiffiffiffi ffi
2ρh ω
γ j ¼ 1 + ð1Þj , j ¼ 1,2: (65)
D k2
On substituting (64) into (63), we arrive at the set of linear algebraic

equations
2
γ 1 ν A1 + γ 22 ν A2 ¼ 0,
(66)
ð2 νÞγ 1 γ 31 A1 + ð2 νÞγ 2 γ 32 A2 ¼ 0:
Then, on employing the solvability condition, we deduce the dispersion

relation
Dk4 γ 4e ¼ 2ρhω2 , (67)

as first shown by Konenkov (1960). Here the coefficient
h pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
i1=4
γ e ¼ ð1 νÞ 3ν 1 + 2 2ν2 2ν + 1 (68)
depends on the Poisson’s ratio only, see Fig. 4. In view of (67), we have for
the attenuation coefficients
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
γj ¼ 1 + ð1Þj γ 2e , j ¼ 1,2: (69)
Thus, the bending edge wave is a dispersive analogue of the Rayleigh

surface wave. It should also be noted that such a wave may be considered
within the framework of refined plate theories, see, e.g., Zakharov
(2004), and may be generalized to a plate with a curved contour as in
Cherednichenko (2007). The effect of anisotropy may also be addressed,
see, e.g., Norris (1994) and Thompson, Abrahams, and Norris (2002).
The aspects of existence and uniqueness of bending edge waves in a gener-
ally anisotropic thin elastic plate were studied by Fu (2003) by using the
edge–impedance matrix. Recent developments in the theory of bending
edge waves involve incorporating the effect of an elastic foundation, see
Kaplunov, Prikazchikov, Rogerson, and Lashab (2014), Kaplunov and
Nobili (2015), and Kaplunov, Prikazchikov, and Rogerson (2016). The
recent contribution by Destrade, Fu, and Nobili (2016) on a closely related
problem of edge wrinkling opens the path for bending edge waves in pres-
tressed plates.
In addition to bending edge waves in thin elastic plates, we also mention
extensional waves, see Pichugin and Rogerson (2012), and related 3D edge
waves, see, e.g., Kaplunov, Prikazchikov, and Rogerson (2005), Zernov and
Kaplunov (2008), and Krushynska (2011).
0.998
0.996
ge
0.994
0.992
0.99
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Fig. 4 The coefficient γ e vs the Poisson’s ratio ν.

4. HYPERBOLIC–ELLIPTIC MODEL FOR THE RAYLEIGH

WAVE INDUCED BY SURFACE STRESSES
In this section we present the derivation of an asymptotic formulation
for the near-surface wave field in an elastic half-space, induced by prescribed
surface stresses. This formulation was first reported in Kaplunov and
Kossovich (2004), starting from the symbolic Lur’e method, e.g., see
Kaplunov, Kossovitch, and Nolde (1998) and references therein. The sub-
ject was further developed in Kaplunov, Zakharov, and Prikazchikov
(2006), using a slow time perturbation procedure applied to the self-similar
eigensolution in Chadwick (1976b), followed by extension to the 3D case in
Dai et al. (2010) and other contributions. Below we present an improved
perturbation scheme based on the implicit traveling wave ansatz (43).
4.1 Plane Strain Problem

In the plane strain setup the equations of motion of the elastic isotropic half-
plane H2+ are taken in the form (27), with the boundary conditions along the
surface x3 ¼ 0 given by
σ 31 ¼ Qðx1 , tÞ, σ 33 ¼ Pðx1 , tÞ, (70)
where P(x1, t) and Q(x1, t) are prescribed surface stresses. The conditions
(70) may be expressed in terms of the wave potentials as
Q
2ϕ, 13 + ψ , 11 ψ , 33 ¼ ,
μ (71)
2 2 P
κ 2 ϕ, 11 + κ ϕ, 33 + 2ψ , 13 ¼ ,
μ
where, as previously, μ is the Lame elastic shear modulus and κ ¼ c2/c1.
In what follows we consider separately the effects of vertical (Q(x1, t) ¼ 0)
and horizontal (P(x1, t) ¼ 0) loads. First, we analyze the near-resonant regimes
of the vertical load
P ¼ Pðx1 ctÞ, (72)
see Fig. 5, where the load speed c is close to the Rayleigh wave speed cR, i.e.,
c ¼ cR ð1 εÞ, (73)
with a small parameter 0 < ε ≪ 1.
O x1
x3
Fig. 5 An elastic half-plane subject to a vertical load.
We proceed with a multiple scale perturbation scheme, see, e.g., Kevorkian

and Cole (2013) and Nayfeh (2008), introducing the fast and slow time vari-
ables, τf and τs, respectively, as
τf ¼ t, τs ¼ εt: (74)
The concept of slow time is essential for modeling of the near-resonant

phenomena caused by the load of the form, see (72),
P ¼ Pðx1 cR τf ,τs Þ: (75)
The simplest example of a near-resonant excitation is presented in Appendix

A for a single degree of freedom linear oscillator.
On taking into account the symbolic identity
@ @ @
¼ +ε , (76)
@t @τf @τs
the equations of motion (27) become

ϕ, τf τf + 2εϕ, τf τs + ε2 ϕ, τs τs c12 ϕ, 11 + ϕ, 33 ¼ 0,
(77)
ψ , τf τf + 2εψ , τf τs + ε2 ψ , τs τs c22 ψ , 11 + ψ , 33 ¼ 0:
We now employ the implicit traveling wave ansatz (43) having
ϕ, τf τf cR2 ϕ, 11 ¼ 0, ψ , τf τf cR2 ψ , 11 ¼ 0: (78)
It is obvious that the load (75) satisfies (43) with respect to fast time, i.e.,
P, τf τf cR2 P, 11 ¼ 0: (79)
Then, Eq. (77) take the form

ε ε2
ϕ, 33 + α2R ϕ, 11 2 2 ϕ, τf τs 2 ϕ, τs τs ¼ 0,
c1 c1
(80)
ε ε2
ψ , 33 + β2R ψ , 11 2 2 ψ , τf τs 2 ψ , τs τs ¼ 0:
c2 c2
Next, we expand the potentials in asymptotic series as

ϕðx1 ,x3 , τf ,τs Þ ¼ ε1 ϕ0 ðx1 , x3 , τf , τs Þ + εϕ1 ðx1 , x3 , τf , τs Þ + ⋯ ,
(81)
ψðx1 ,x3 , τf ,τs Þ ¼ ε1 ψ 0 ðx1 , x3 , τf , τs Þ + εψ 1 ðx1 , x3 , τf , τs Þ + ⋯ ,
where the factor ε1 is due to a near-resonant excitation, see also a similar
asymptotic expansion in Appendix A.
Below we present in detail the perturbation procedure for the potential
ϕ, which, in view of the ansatz (78), at leading order satisfies the equation
ϕ0, 33 + α2R ϕ0, 11 ¼ 0, (82)
with the solution given by a plane harmonic function in the first two
arguments, i.e.,
ϕ0 ¼ ϕ0 ðx1 , αR x3 , τf ,τs Þ: (83)
At next order
2
ϕ1, 33 + α2R ϕ1, 11 ¼ ϕ , (84)
c12 0, τf τs
so the first-order correction can be written as
ϕ1 ¼ ϕ10 + x3 ϕ11 , (85)
where ϕ10 ¼ ϕ10(x1, αRx3, τf, τs) and ϕ11 ¼ ϕ11(x1, αRx3, τf, τs) are plane
harmonic functions. On substituting the solution (85) into (84) and using
the Cauchy–Riemann identities, we arrive at
1
ϕ11, 1 ¼ ϕ* , (86)
αR c12 0, τf τs
where, as previously, the asterisk denotes a harmonic conjugate function.
The derivation for the transverse potential ψ is very similar to that
presented above. Thus, the two-term expansions for the derivatives of
the potentials are

1 x3 *
ϕ, 1 ¼ ϕ0, 1 + ε ϕ10, 1 + ϕ +⋯ ,
ε αR c12 0, τf τs
(87)
1 x3 *
ψ , 1 ¼ ψ 0, 1 + ε ψ 10, 1 + ψ +⋯ :
ε βR c22 0, τf τs
It may be readily observed that the obtained expansion is essentially a slow
time perturbation of the eigensolution for the Rayleigh wave of arbitrary
profile discussed in Section 2.2.
On substituting (87) into (71) at Q ¼ 0 and making use of the
Cauchy–Riemann identities, we obtain at leading order

2αR ϕ0, 11 + 1 + β2R ψ *0, 11 ¼ 0,
(88)
1 + β2R ϕ0, 11 + 2βR ψ *0, 11 ¼ 0,
implying the Rayleigh equation

2
1 + β2R 4αR βR ¼ 0 (89)
as a solvability condition. In addition, we recover the relations (37) between

the leading order potentials ϕ0 and ψ 0.
At next order we deduce
ϕ* ψ*
2 2
2ϕ10, 113 + ψ 10, 111 ψ 10, 133 ¼ 0, 1τf τs + ,
α R c1
2 βR c22 0, 3τf τs
ϕ* ψ*
2 2 P, 1
ðκ2 2Þϕ10, 111 + κ2 ϕ10, 133 +2ψ 10, 113 ¼ 0, 3τf τs 0, 1τf τs + :
αR κ c1
2 2 β R c2
2 μ
(90)
Then, on using the Cauchy–Riemann identities along with (37), these equa-
tions may be simplified to

* 2 1 + β2R
2αR ϕ10, 111 + 1 + βR ψ 10, 111 ¼
2
ϕ0, 1τf τs ,
αR c12 βR c22
2
(91)
2 1 + β P
ð1 + β2R Þϕ10, 111 + 2βR ψ *10, 111 ¼ 2 2 2R ϕ0, 1τf τs :
, 1
c2 βR c2 μ
The solvability of the latter implies
ð1 + β2R ÞcR2
ϕ0, τf τs ¼ P, (92)
4μB
where
αR β
B¼ ð1 β2R Þ + R ð1 α2R Þ 1 + β4R (93)
βR αR
is a dimensionless constant. On taking into account the leading order approx-
imation ϕ ¼ ε1ϕ0 and the ansatz (78), the relation (92) may be rewritten as
ð1 + β2R Þ cR2 P
ϕ, τf τf + 2εϕ, τf τs cR2 ϕ, 11 ¼ : (94)
2μB
Now, employing the approximate symbolic formula
@2 @2 @2
¼ + 2ε + Oðε2 Þ, (95)
@t2 @τ2f @τs @τf
we restate (94) in terms of the original variables as

1 ð1 + β2R Þ P
ϕ, 11 ϕ ¼ : (96)
cR2 , tt 2μB
Thus, the asymptotic formulation for the Rayleigh wave involves a scalar
problem for the pseudo-static elliptic equation
ϕ, 33 + α2R ϕ, 11 ¼ 0, (97)
subject to the Dirichlet boundary condition at x3 ¼ 0 in the form of the 1D
wave equation (96). The transverse potential ψ may then be restored from (38).
We also note that the hyperbolic equation (96) can be transformed to an
equation for the horizontal displacement u1. Indeed, on differentiating (96)
with respect to x1 and using (40) at x3 ¼ 0, we get
1 1 β4R
u1, 11 u 1, tt ¼ P, 1 : (98)
cR2 4μB
Consider now the boundary conditions (71) for a horizontal load, when
P ¼ 0. A very similar procedure leads to the same two-term expansions (87).
The analysis of the boundary conditions gives the Rayleigh equation (89) at
leading order, along with the relations (37) between the potentials ϕ0 and
ψ 0. Finally, we arrive at a 1D wave equation for the potential ψ which
can be written as
1 ð1 + β2R Þ Q
ψ , 11 ψ ¼ : (99)
cR2 , tt 2μB
Its solution provides a Dirichlet boundary condition for the elliptic equation
given by
ψ , 33 + β2R ψ , 11 ¼ 0: (100)
In this case the potential ϕ follows from (39). Similarly to (98), we also have
1 1 β4R
u3, 11 u3 , tt ¼ Q, 1 : (101)
cR2 4μB
The asymptotic formulations (96), (97) and (99), (100) reveal a dual
hyperbolic–elliptic nature of the Rayleigh wave. Indeed, the elliptic equations
(97) and (100) characterize decay of a wide range of surface disturbances into
the interior, whereas the 1D wave equations (96) and (99) govern the wave
propagation along the surface with a finite speed cR. At the same time, it
should be emphasized that a hyperbolic wave-like behavior is only typical
for one of the potentials along the surface, see (96) and (99).
Finally, it should be noted that the applicability of the developed explicit
formulations for the surface wave field is not restricted only to near-resonant
loading as in (75). In fact, it may be shown, using the integral transform tech-
nique, that they approximate the contribution of the Rayleigh wave to the
overall dynamic response for an arbitrary surface load as well. For example,
in case of the vertical load P ¼ P(x1, t) the transformed solution of the prob-
lem (96) and (97) is given by
ð1 + β2R Þ cR2 eαR jpjx3 FL

ϕFL ¼ P , (102)
2μBðp2 + cR2 k2 Þ
where k and p are the Fourier and Laplace transform parameters, respec-
tively, and the superscript FL denotes Fourier–Laplace transforms. The for-
mula (102) coincides with the local behavior of the transformed exact
solution of the original plane problem near the Rayleigh wave poles, see
consideration in Section 4.3.1 ensuring identical surface wave patterns.
4.2 3D Problem
Let us generalize the analysis in the previous section to the 3D setup, in
which the equations of motion are taken in the form (4), with the boundary
conditions at x3 ¼ 0 written as
σ 3i ¼ Qi ðx1 ,x2 ,tÞ, σ 33 ¼ Pðx1 ,x2 , tÞ, i ¼ 1,2: (103)
4.2.1 Vertical Load

Consider a vertical load, see Fig. 6, when Q1 ¼ Q2 ¼ 0 in (103). First, we
specify the Radon integral transform
Z∞
ðαÞ
f ðχ,α, x3 , tÞ ¼ f ð χ cos α η sinα, χ sin α + η cos α,x3 ,tÞdζ, (104)
∞
where
χ ¼ x1 cosα + x2 sinα, η ¼ x1 sinα + x2 cosα,
with the angle α varying over the interval 0 α < 2π; here and below the
Radon transforms are supplied with the superscript (α). It is well known that
the Radon transform allows reduction of the original 3D problem in elas-
ticity to a 2D problem for the associated transforms, see Georgiadis and
Lykotrafitis (2001).
Let us also define transformed displacements in the Cartesian frame
ðχ, ηÞ as
ðαÞ ðαÞ ðαÞ ðαÞ
uχðαÞ ¼ u1 cosα + u2 sinα, uηðαÞ ¼ u1 sin α + u2 cos α, (105)
imposing the assumption
uðαÞ
η ¼ 0, (106)
meaning that antiplane motion does not induce surface waves.

On taking into account (106), the transformed equations of motion (4)
become, see also Dai et al. (2010),
O
x2
x1
x3
Fig. 6 An elastic half-space under a vertical load.
ðαÞ ðαÞ ðαÞ

ðλ + 2μÞuðαÞ
χ , χχ + μuχ , 33 + ðλ + μÞu3, χ3 ¼ ρuχ , tt ,
(107)
ðαÞ ðαÞ ðαÞ ðαÞ
ðλ + μÞuχ , χ3 + μu3, χχ + ðλ + 2μÞu3, 33 ¼ ρu3, tt :
The transformed boundary conditions (103) at Q1 ¼ Q2 ¼ 0 are written as


σ χ3 ¼ μ uχ , 3 + u3, χ ¼ 0,
(108)
σ 33 ¼ μuχðαÞ
,χ + ð λ + 2μ Þu3, 3 ¼ P :
Eqs. (107) and (108) are formally identical to those in the plane strain
problem. Therefore, we can introduce the conventional scalar wave poten-
tials ϕ(α) and ψ (α), having
@ϕðαÞ @ψ ðαÞ ðαÞ @ϕðαÞ @ψ ðαÞ

uðαÞ
χ ¼ , u3 ¼ + (109)
@χ @x3 @x3 @χ
and follow the multiple scale procedure developed in the previous section. In
this case the transformed implicit traveling wave ansatz, see (43),
ϕðαÞ 2 ðαÞ
, τf τf cR ϕ, χχ ¼ 0: (110)
corresponds to the 2D wave equation
ϕ, τf τf cR2 Δϕ ¼ 0, (111)
where Δ denotes the 2D Laplace operator in x1 and x2.

The analogues of the two-term asymptotic expansions (87) become

ðαÞ 1 ðαÞ ðαÞ x3 ðαÞ *
ϕ, χ ¼ ϕ0, χ + ε ϕ10, χ + ϕ +⋯ ,
ε αR c12 0, τf τs (112)
ðαÞ 1 ðαÞ ðαÞ x3 ðαÞ *
ψ , χ ¼ ψ 0, χ + ε ψ 10, χ + ψ + ⋯ :
ε βR c22 0, τf τs
On substituting these formulae into the boundary conditions (108), at lead-
ing order we obtain the Rayleigh wave equation and the relations between
the potentials mirroring (37), whereas at next order the solvability dictates
ðαÞ ð1 + β2R ÞcR2 ðαÞ

ϕ0, τf τs ¼ P , (113)
4μB
with the material constant B defined by (93). Finally, the explicit formula-
tion for the Radon transforms includes the hyperbolic equation
1 ðαÞ ð1 + β2R Þ P ðαÞ

ϕðαÞ
, χχ ϕ, tt ¼ , (114)
cR2 2μB
along the surface x3 ¼ 0, following from (113) and providing the boundary
condition for the elliptic equation
ðαÞ
ϕ, 33 + α2R ϕðαÞ
, χχ ¼ 0 (115)
over the interior. Now, the transformed potential ψ (α) satisfies the equation
ðαÞ
ψ , 33 + β2R ψ ðαÞ
, χχ ¼ 0, (116)
being related on the surface to ϕ(α) by the formulae

2 ðαÞ
ψ ðαÞ
,χ ¼ ϕ (117)
1 + β2R , 3
and
ðαÞ 1 + β2R ðαÞ

ψ ,3 ¼ ϕ, χ : (118)
2
ðαÞ ðαÞ
Next, we introduce a pair of the potentials ψ 1 ¼ ψ ðαÞ cosα and ψ 2 ¼
ðαÞ
ψ sin α in order to invert the transforms in (115)–(118). As a result, we get
the elliptic equations
ϕ, 33 + α2R Δϕ ¼ 0, ψ i, 33 + β2R Δψ i ¼ 0, i ¼ 1,2, (119)

governing behavior over the interior. The boundary conditions at x3 ¼ 0
include a 2D wave equation for the potential ϕ, i.e.,
1 ð1 + β2R Þ P
Δϕ ϕ ¼ , (120)
cR2 , tt 2μB
along with the relations
2 1 + β2R
ϕ, i ¼ ψ i, 3 , ϕ, 3 ¼ ψ 1, 1 + ψ 2, 2 , i ¼ 1,2: (121)
1 + β2R 2
The Helmholtz representation (5) for the displacement vector u may
now be written as
u ¼ gradϕ + curlΨ, (122)
where Ψ ¼ (ψ 2, ψ 1, 0), as observed in Kaplunov and Prikazchikov (2013).
In view of (121) and (122), at x3 ¼ 0
1 β2R
ui ¼ ϕ, i , i ¼ 1, 2: (123)
2
Hence, we deduce hyperbolic equations on the surface for the horizontal
displacements, u1 and u2, namely,
1 1 β4R
Δui ui, tt ¼ P, i , i ¼ 1,2, (124)
cR2 4μB
corresponding to (98) within the plane strain formulation.
4.2.2 Horizontal Load

Consider now a horizontal load, setting P ¼ 0 in (103), see Fig. 7. First, we
decompose the surface stresses using the Helmholtz theorem into the gradi-
ent and rotational parts, Qg(x1, x2, t) and Qr(x1, x2, t), respectively, as
Q1 ¼ Q g , 1 + Q r , 2 , Q2 ¼ Q g , 2 Qr , 1 : (125)
On applying the Radon transform (104) along with the assumption (106),
we express the boundary conditions (103) as

ðαÞ ðαÞ
μ 2ϕ, χ3 + ψ , χχ ψ , 33 ¼ QgðαÞ
ðαÞ
,χ ,
2 ðαÞ (126)
2 ðαÞ ðαÞ
κ 2 ϕ, χχ + κ ϕ, 33 + 2ψ , χ3 ¼ 0:
Thus, as might be expected, the rotational part of the load does not contrib-
ute to the excitation of the Rayleigh wave, see Ege, Erbaş, and Prikazchikov
(2015) for further details.
A slow time perturbation procedure similar to that developed in
Section 4.1, results in a boundary value problem for the elliptic equation
Q
O
x2
x1
x3
Fig. 7 An elastic half-space under a horizontal load.
ðαÞ
ψ , 33 + β2R ψ ðαÞ
, χχ ¼ 0: (127)
Predictably, the surface wave propagation along the boundary x3 ¼ 0 is

governed by the hyperbolic equation
1 ðαÞ 1 + β2R ðαÞ
ψ ðαÞ
, χχ ψ , tt ¼ Qg , χ : (128)
cR2 2μB
The transformed potential ϕ(α) is then found from (116) combined with any
of the relations (117) or (118).
On applying the inverse Radon transform to the formulae above, we
arrive at the same 3D pseudo-static elliptic equations
ϕ, 33 + α2R Δϕ ¼ 0, ψ i, 33 + β2R Δψ i ¼ 0, i ¼ 1,2, (129)
together with the relations between the potentials at x3 ¼ 0
2 1 + β2R
ϕ, i ¼ ψ i, 3 ϕ, 3 ¼ ψ 1, 1 + ψ 2, 2 , i ¼ 1,2: (130)
1 + β2R 2
The counterpart of (120) now involves a 2D vector hyperbolic equation
containing the gradient part of the in-plane load Qg, namely
1 1 + β2R
Δψ i ψ ¼ Qg, i , i ¼ 1,2: (131)
cR2 i, tt 2μB
We also write down the 2D wave equation for the vertical displacement u3
on the surface x3 ¼ 0, following from (121) and (122), complementing (101)
1 1 β4R
Δu3 u3, tt ¼ ΔQg : (132)
cR2 4μB
4.3 Examples
This section contains a few examples of dynamic surface loading demon-
strating the advantages of the explicit formulation for the Rayleigh wave
developed in Sections 4.1 and 4.2.
4.3.1 Comparison With 2D Exact Solution

Let us compare the approximate solution coming from the derived hyper-
bolic–elliptic formulation with the exact solution of the plane strain prob-
lem. For example, for the vertical load P ¼ P(x1, t) we get by applying the
double Fourier–Laplace integral transform in (27) and (71) at Q ¼ 0
ϕFL
, 33 k aR ϕ ¼ 0,
2 2 FL
ψ FL
, 33 k bR ϕ ¼ 0,
2 2 FL
(133)
subject to
,3 k ψ
2ikϕFL ψ FL
, 33 ¼ 0,
2 FL
P FL (134)
2 κ 2 k2 ψ FL + κ2 ϕFL + 2ikψ FL
¼ ,
, 33 ,3
μ
at x3 ¼ 0. In the above
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p2 p2
aR ¼ 1 + 2 2 , bR ¼ 1 + 2 2 , (135)
c1 k c2 k
where k and p are the Fourier and Laplace transform parameters,

respectively.
The solution of the problem (133) and (134) for the transformed poten-
tial ϕFL is given by
1 + b2R P FL eaR jkjx3

ϕFL ¼ , (136)
R μ k2
where
2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffisffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
p p2 p2 p2
R¼R 2 ¼4 1+ 2 2 1+ 2 2 2+ 2 2 : (137)
k c1 k c2 k c2 k
At the same time, within the framework of the hyperbolic–elliptic model

in Section 4.1, the associated transformed solution takes the form (102). It is
clear that the Rayleigh poles in (136) are given by p2 ¼ cR2 k2 . In this case
the quantities aR and bR defined by (135) become αR and βR, respectively.
Therefore, by expanding the denominator R in (136) as a Taylor series
around p2 ¼ cR2 k2 , we have
2
p p2
R R0 ðcR2 Þ 2 + cR2 ¼ 2B 1 + 2 2 , (138)
k cR k
where B is the same as in (93).
Thus, the approximation of (136) in the vicinity of the Rayleigh poles
coincides with (102), providing another justification of the validity of the
developed model. We remark that the consideration in this section is
restricted to the loads, which do not generate the poles that are close to
the Rayleigh ones, arising from PFL in the transformed solution (136).
The proposed approach brings in a significant simplification of the

Rayleigh wave analysis, reducing the original problem to a scalar problem
for the elliptic equation (97) together with a boundary condition in the form
of the hyperbolic equation (96).
Consider, for example, the Lamb problem, see Lamb (1904), for the ver-
tical point impulse P ¼ P0δ(x1)δ(t). In this case PFL ¼ P0 in (136). On eval-
uating the residues related to the Rayleigh poles and taking the inverse
Fourier transform, we get

ð1 + β2r ÞcR P0 x1 cR t x1 + cR t
ϕ¼ tan 1 tan 1 : (139)
4πμB αR x3 αR x3
The last formula also follows from (102). However, the derivation starting
from the Rayleigh wave model allows a more straightforward treatment,
which does not require double integral transforms. Indeed, we solve the
hyperbolic equation
1 ð1 + β2R ÞP0
ϕ, 11 ϕ ¼ δðx1 ÞδðtÞ, (140)
cR2 , tt 2μB
on the surface x3 ¼ 0, immediately having
ð1 + β2R ÞcR P0
ϕjx3 ¼0 ¼ ½Hðx1 cR tÞ Hðx1 + cR tÞ, (141)
4μB
which follows from the fundamental solution of the wave equation, e.g., see
Polyanin and Nazaikinskii (2015). Then, the Poisson formula, e.g., see
Courant and Hilbert (1989), enables the potential ϕ to be restored over
the interior, coinciding with (139).
Thus, for the studied 2D problem the approximate approach captures the
contribution of the Rayleigh wave, which is usually dominant in the far-
field zone, see also a practical example in Chouet (1985).
4.3.2 2D Near-Resonant Time-Harmonic Excitation

Next, investigate the time-harmonic plane strain problem for the vertical
load
P ¼ P0 eikðx1 ctÞ , (142)

where k is the wave number and c is the phase speed (73), being close to the
Rayleigh wave speed cR.
The solution for the potential ϕ within the approximate formulation (96)
and (97) is readily obtained in the form
ð1 + β2R ÞP0 cR2

ϕ¼ eikðx1 ctÞkαR x3 , (143)
2μBk2 c 2 cR2
demonstrating a resonant response as c ! cR.
For the sake of definiteness, below we set c ¼ (1 + ε)cR. Then, c 2 cR2
2εcR2 . Hence, the near-resonant behavior of (143) becomes
ð1 + β2R ÞP0 ikðx1 ctÞkαR x3

ϕ¼ e : (144)
4εμBk2
Now, let us study the exact 2D solution for the prescribed near-resonant
harmonic loading. The same potential ϕ, satisfying Eq. (27), subject to the
conditions (71) with P in the form of (142), is written as
P0 ð1 + β2 Þ ikðx1 ctÞkαx3
ϕ¼ e , (145)
μk2 RðcÞ
with
RðcÞ ¼ ð1 + β2 Þ2 4αβ (146)
and α and β defined by (14). On employing the asymptotic formulae

1 α2R 1 β2R
α αR 1 ε 2 , β βR 1 ε 2 , (147)
αR βR
we deduce

αR βR
RðcÞ ð1 + β2R Þ2 4αR βR + 4ε ð1 βR Þ +
2
ð1 αR Þ 1 + βR ¼ 4εB,
2 4
βR αR
(148)
with B defined by (93). Thus, the limiting behavior of (145) coincides with
(144), since at leading order α αR and β βR as c ! cR.
A comparison of the asymptotic and exact solutions, given by (144) and
(145), respectively, is demonstrated in Fig. 8, where the scaled potential
εμk2 ikðx1 ctÞ

ϕs ¼ e ϕðx1 , 0,tÞ (149)
P0
0.45
Asymptotic solution
0.4 Exact solution
0.35
0.3
0.25
fs
0.2
0.15
0.1
0.05
0
–0.06 –0.04 –0.02 0 0.02 0.04 0.06
e
Fig. 8 Comparison of exact and asymptotic results for the near-resonant regime.
is displayed. Here the solid line shows variation of the exact solution over a
range of near-resonant speeds, whereas the dashed line corresponds to the
asymptotic model, clearly matching at ε ¼ 0.
In addition to the consideration in Section 4.3.1 mainly oriented to the
far-field analysis, this example highlights another major application of the
proposed methodology, associated with a near-resonant excitation.
4.3.3 Axisymmetric and 3D Problems for Point Impulses

Let us now analyze the effect of a vertical point force, setting P ¼ P0δ(x1)
δ(x2)δ(t) and Q1 ¼ Q2 ¼ 0 in (103), see Fig. 9.
The longitudinal potential ϕ on the surface is found as the fundamental
solution of the wave equation, see, e.g., Polyanin and Nazaikinskii (2015). It
follows from (120), that
ð1 + β2R ÞP0 H ðcR t r Þ

ϕðx1 , x2 ,0,tÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , (150)
4πμBcR cR2 t2 r 2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where r ¼ x21 + x22 is the polar radius. In order to restore the potential ϕ
over the interior, we solve Eq. (119) with the boundary condition (150).
On applying the Hankel transform, we get
x2 x1
x3
Fig. 9 An axisymmetric vertical point load.
@ 2 ϕH
α2R p2 ϕH ¼ 0, (151)
@x23
where
Z∞
ϕH ¼ ϕðr,x3 ,tÞJ0 ðprÞrdr: (152)
0
Therefore,
Z∞
ð1 + β2R ÞP0
ϕðx1 , x2 ,x3 ,tÞ ¼ sin ðcR ptÞexpðαR px3 ÞJ0 ðprÞ dp,
4πμBcR
0
Z∞
ð1 + β2R ÞP0 (153)
¼ Im exp½ðαR px3 icR tÞ J0 ðprÞ dp,
4πμBcR
0
ð1 + β2R ÞP0 n 2 1=2 o
¼ Im r + ðαR x3 icR tÞ2 :
4πμBcR
Then, using the relations (121), we obtain the potentials ψ n, n ¼ 1, 2, in the

form
8 9
αR P0 < cosθðβR x3 icR tÞ =
ψ 1 ðx1 , x2 , x3 , tÞ ¼ Re :
:r r 2 + ðβ x ic tÞ2 1=2 ;
(154)
2πμcR B
R 3 R
and
8 9
αR P0 < sinθðβR x3 icR tÞ =
ψ 2 ðx1 ,x2 ,x3 ,tÞ ¼ Re
:r r 2 + ðβ x ic tÞ2 1=2 ;
, (155)
2πμBcR
R 3 R
where θ is the polar angle. It may also be verified that (153)–(155) satisfy
both (119) and (121).
The behavior of the scaled longitudinal potential
n 1=2 o
ϕ* ðr1 ,zÞ ¼ Im r12 + ðαR z iÞ2 ,
with
r x3
r1 ¼ , z¼ ,
cR t cR t
is illustrated in Fig. 10. The Rayleigh wave front propagating along the sur-
face is shown by dotted line at r1 ¼ 1. The associated discontinuity becomes
smoother with depth as seen from Fig. 10. Graphs for the scaled transverse
potential
8 9
< cos θðβ z iÞ =
ψ 1* ðr1 ,zÞ ¼ Re R
:r r 2 + ðβ z iÞ2 1=2 ;
,
1 1 R
4
z=0
z = 0.05
3.5
z = 0.1
z = 0.2
3
z = 0.5
2.5
f*
1.5
0.5
0
0.6 0.8 1 1.2 1.4
r1
Fig. 10 The scaled longitudinal potential ϕ* vs the dimensionless polar radius r1.
are plotted in Fig. 11 for θ ¼ 0. Obviously, we may expect a similar behavior

of the second component of the transverse potential ψ 2.
Next, we investigate the effect of a horizontal point impulse, see Fig. 12,
on substituting Q1 ¼ Q0δ(x1)δ(x2)δ(t), and Q2 ¼ P ¼ 0 into (103).
We restrict ourselves to analysis of the surface transverse displacement
only. First, we decompose the prescribed in-plane load (Q1, Q2) according
to (125), having
ΔQg ¼ Q0 δ0 ðx1 Þδðx2 ÞδðtÞ: (156)
4 z=0
z = 0.05
3.5 z = 0.1
z = 0.2
z = 0.5
3
2.5
1*
1.5
0.5
0
0.6 0.8 1 1.2 1.4
r1
Fig. 11 The scaled transverse potential ψ 1* vs the dimensionless polar radius r1.
Q1
O
x2
x1
x3
Fig. 12 A horizontal point load.
This allows a straightforward evaluation of the aforementioned displacement

governed by (132). Indeed, on using (156) along with the fundamental solu-
tion (150), we finally arrive at
cR Q0 ð1 β4R Þr cos θ H ðcR t r Þ
u3 ðx1 , x2 , 0,tÞ ¼ 3=2
: (157)
8πμB ðcR2 t2 r 2 Þ
5. GENERALIZATIONS OF THE HYPERBOLIC–

ELLIPTIC MODEL
5.1 Coated Half-Space
The asymptotic formulation derived in the previous section may be
extended to a coated half-space at the long-wave limit. Consider an elastic
half-space H3+ coated by an elastic layer occupying the region h x3 0,
see Fig. 13. Throughout this section we assume the thickness of the coating h
to be small in comparison with a typical wave length L, i.e., h ≪ L. Similarly
to Section 4.2, we impose the boundary conditions (103) on the upper face
of the coating x3 ¼ h. We also assume continuity of all displacements and
stresses along the interface x3 ¼ 0 and restrict ourselves to a vertical
load only.
A standard asymptotic technique, applied to a thin coating, for more
details see Dai et al. (2010) and references therein, results in effective bound-
ary conditions on the interface x3 ¼ 0, namely

σ 31 ¼ ρ0 h u1, tt c20
2
u1, 22 + 4 1 κ20 u1, 11 + 3 4κ20 u2, 12 ,

σ 32 ¼ ρ0 h u2, tt c20
2
u2, 11 + 4 1 κ20 u2, 22 + 3 4κ20 u1, 12 ,
(158)
and
σ 33 ¼ ρ0 h u3, tt + P:
where ρ0 is density of the coating, c10 and c20 are associated bulk wave speeds,
and κ0 ¼ c20/c10.
Thus, the original problem for a coated half-space is reduced to analysis
of the homogeneous half-space H3+ subject to the boundary conditions
(158) imposed along its surface x3 ¼ 0. In this case the transformed equations
ðαÞ 1 ðαÞ ðαÞ 1 ðαÞ
ϕðαÞ
, χχ + ϕ, 33 2 ϕ, tt ¼ 0, ψ ðαÞ
, χχ + ψ , 33 2 ψ , tt ¼ 0, (159)
c1 c2
following from (107) and (109) are accompanied by the boundary condi-
tions at x3 ¼ 0
P
–h
0 x1
x3
Fig. 13 A half-space coated by a thin elastic layer.
h i h

i
ðαÞ ðαÞ ðαÞ 2 ϕðαÞ ψ ðαÞ 4 1 κ 2 ϕðαÞ ψ ðαÞ
μ 2ϕ, χ3 + ψ , χχ ψ , 33 ¼ μ0 h c20 , χtt , 3tt 0 , χχχ , 3χχ
,
h ðαÞ i
ð160Þ
ðαÞ ðαÞ 2 ϕðαÞ + ψ ðαÞ P ðαÞ :
μ κ2 2 ϕ, χχ + κ 2 ϕ, 33 + 2ψ , χ3 ¼ μ0 hc20 , 3tt , χtt
A multiple scale perturbation procedure similar to that in Section 4.1 results

in a singularly perturbed hyperbolic equation on the surface x3 ¼ 0, which
can be written as
1 ðαÞ bh ðαÞ 1 + β2R ðαÞ

ϕðαÞ
, χχ ϕ , tt + ϕ ¼ P , (161)
cR2 αR , 3χχ 2μB
where the constant
μ0 ð1 β2R Þ
b¼ ð1 β2R0 ÞðαR + βR Þ 4βR ð1 κ20 Þ (162)
μ 2B
depends on the properties of both the substrate and the coating, with B
defined in (93).
In the original variables, we get from (161)
1 bh ð1 + β2R Þ P
Δϕ ϕ, tt + Δϕ, 3 ¼ , (163)
2
cR αR 2μB
which is a boundary condition for the elliptic equation (129), where, as

before, Δ is the 2D Laplace operator in the variables x1 and x2.
The perturbed hyperbolic equation (163) can also be presented in a
pseudo-differential form, i.e.,
1 pffiffiffiffiffiffiffiffi ð1 + β2R Þ P
Δϕ ϕ , tt bh Δ ðΔϕÞ ¼ : (164)
cR2 2μB
In the plane strain case it becomes

1 pffiffiffiffiffiffiffiffiffiffiffi ð1 + β2R Þ P
ϕ, 11 ϕ bh @ , 11 ϕ ¼ : (165)
cR2 , tt , 11
2μB
This can also be rewritten using the Hilbert transform as
1 ^ ð1 + β2R Þ P
ϕ, 11 ϕ bh Hϕ ¼ , (166)
cR2 , tt , 111
2μB
where
Z∞
^ ðxÞ ¼ 1 f ðξÞ
Hf dξ, (167)
π ξx
∞
denotes the Hilbert transform, e.g., see Erdelyi, Magnus, Oberhettinger, and
Tricomi (1954).
Thus, the presence of a coating inevitably leads to a pseudo-differential
or an integro-differential formulation.
The derived equation enables a straightforward approximation of the
exact dispersion relation, e.g., see Shuvalov and Every (2008). Naturally,
we deduce from (165) that
c b
vph ¼ ¼ 1 jkhj + ⋯ , (168)
cR 2
demonstrating that the Rayleigh wave speed cR is a local extremum over the
long-wave domain kh ≪ 1, where, as usual, k denotes the wave number, see
Fig. 14.
In order to illustrate the developed approach, let us consider an impulse
point load, setting P ¼ P0δ(x1)δ(t) in the R.H.S. of (165) and specify the
dimensionless variables as
x1 cR t
X¼ , τ¼ , (169)
L L
where L is a chosen linear scale. Then, on introducing the quantities
ϕ hjbj
Θ¼ , hL ¼ ≪1 (170)
B0 L
where
ð1 + β2R Þ cR P0
B0 ¼ , (171)
4μB
A B
v ph v ph
b<0 b>0
1 1
0 kh 0 kh
Fig. 14 Local minimum (A) or maximum (B) of the phase speed.
Eq. (165) becomes

pffiffiffiffiffiffiffiffiffiffiffiffi
Θ, XX Θ, ττ hL sgnðbÞ @XX Θ, XX ¼ δðXÞδðτÞ: (172)
Below we implement the method of matched asymptotic expansions,

see, e.g., Kevorkian and Cole (2013) and Nayfeh (2008). The inner coor-
dinate associated with the boundary layer occurring in the vicinity of the
Rayleigh wave front X ¼ τ is
τX
ζ ¼ pffiffiffiffiffi : (173)
hL
1=2
Thus, the width of the boundary layer is OðhL Þ. On substituting (173) into
the homogeneous equation (172), we obtain for the inner region at leading
order
hL pffiffiffiffiffiffiffiffiffiffiffi
,τ
Θinn sgnðbÞ @, ζζ Θinn , ζ ¼ 0:
(174)
2
The solution of (174) is written as
Z∞
1 hL jωjτ sgnðbÞ
Θ inn
¼ CðωÞ exp iω ζ 0 dω, (175)
2π 2
∞
where ζ 0 ¼ τ X and C(ω) has to be determined from matching with the

leading order outer expansion.
In the outer region (172) reduces to the wave equation
Θout
, XX Θ, ττ ¼ 2δðXÞδðτÞ,
out
(176)
from which we have

Z∞
1 1
Θ out
¼ H ðζ 0 Þ ¼ πδðωÞ + exp ðiωζ0 Þdω: (177)
2π iω
∞
Matching of the expansions (175) and (177) yields

Z∞
1 1 hL ωτ sgnðbÞ
Θ inn
¼ πδðωÞ + exp iω ζ0 dω, (178)
2π iω 2
∞
resulting in the uniform asymptotic behavior

1 1 ðX τÞsgnðbÞ
Θ ¼ sgnðbÞ I pffiffiffiffiffiffiffiffiffi , (179)
2 π 2hL τ
where
Z∞
sin ðt 2 + 2txÞ π 1
IðxÞ ¼ dt ¼ + sgnðxÞ½CðxÞ + SðxÞ C ðxÞ S ðxÞ ,
2 2
t 2 2
0
(180)
and C(x) and S(x) are Fresnel integrals, see, e.g., Prudnikov, Brychkov,
Marichev, and Romer (1988),
Zx π
Zx π
CðxÞ ¼ cos t2 dt, SðxÞ ¼ sin t2 dt: (181)

2 2
0 0
Numerical results illustrating the effect of the coating on smoothing

the wave front are displayed in Fig. 15 for τ ¼ 1 and hL ¼ 0.01. In case
b > 0 corresponding to the local maximum of the phase velocity at the Ray-
leigh wave speed, see Fig. 14A, we observe a receding front. Another case
b < 0 is associated with the minimum of the phase velocity leading to an
advancing front, see Fig. 14B. We also remark that similar graphs expressed
in terms of Airy functions were observed for receding and advancing fronts
in the problem of prestressed plate extension, see Kaplunov, Nolde, and
Rogerson (2000).
5.2 Mixed Boundary Value Problems

The formulation for the Rayleigh wave developed in Section 4.1 may also
be extended to mixed boundary value problems in linear elasticity. Consider
A 1.4
1.2
0.8
Q
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
X
B
1
0.8
0.6
0.4
Q
0.2
–0.2
–0.4
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
X
Fig. 15 Receding (A) and advancing (B) Rayleigh wave fronts.
a vertical stamp acting on the surface of the elastic half-plane H2+ , see (25). The
boundary conditions at x3 ¼ 0 include the normal stress P and vertical dis-
placements U3 prescribed along the disjoint parts of the surface S1 and S2,
ðS1 [ S2 ¼ Þ, respectively, see Fig. 16, along with zero tangential stress, i.e.,
P
U3
S2 S1 x1
x3
Fig. 16 A vertical rigid stamp.
σ 33 ¼ Pðx1 ,tÞ, at x1 2 S1 ,
u3 ¼ U3 ðx1 , tÞ, at x1 2 S2 , (182)
σ 31 ¼ 0, at x1 2 :
On employing (96) and (97) along with (38), we arrive at a scalar mixed
boundary value problem for the elliptic equation
ϕ, 33 + α2R ϕ, 11 ¼ 0 (183)
subject to the following mixed boundary conditions along the surface x3 ¼ 0
1 1 + β2R
ϕ, 11 ϕ ¼ P, at x1 2 S1 , (184)
cR2 , tt 2μB
and
1 + β2R
ϕ, 3 ¼ U3 , at x1 2 S2 , (185)
1 β2R
see Erbaş, Kaplunov, and Prikazchikov (2013) for more detail. The shear
potential ψ may then be determined through (38).
We stress that the approximate formulation above is meaningful only
provided that the contribution of the Rayleigh wave is dominant compared
to that of the bulk waves. As an example, consider the near-resonant regime
of a stamp, moving steadily at the constant speed c given by (73), setting in
the above formulae U3(x1, t) ¼ f(x1 ct) and also assuming for simplicity the
absence of normal stresses on S2, see Fig. 17.
On introducing the moving coordinate s ¼ x1 ct, the problem (183)–
(185) may be reduced to a standard mixed problem for the Laplace equation
for the scaled normal derivative
c
f
x1
x3
Fig. 17 Steady-state motion of a rigid stamp.
β2R 1
φ1 ðs,zÞ ¼ ϕ, 3 , (186)
β2R + 1
where z ¼ αRx3. Indeed, we have
φ1, zz + φ1, ss ¼ 0, (187)
with the mixed boundary conditions
φ1 ¼ f ðsÞ, at s 2 S20 (188)
and
φ1, z ¼ 0, at s 2 S10 , (189)
where and S10 and S20 are the traction free and constrained parts of the surface
z ¼ 0, respectively.
The vertical displacement is expressed in terms of the quantity φ1 as
1
u3 ðs,x3 Þ ¼ β2R + 1 φ1 ðs,αR x3 Þ 2φ1 ðs,βR x3 Þ : (190)
β2R 1
This result matches the limiting behavior of the exact solution (B.6), see also
(B.14) and (B.15), as c ! cR, leading to α ! αR and β ! βR. It is confirmed
by the numerical comparison in Erbaş et al. (2013) for a semiinfinite stamp of
an exponential shape f(s) ¼ beas, where a, b > 0. For the latter
pffiffiffiffiffiffiffiffi
φ1 ðs,zÞ ¼ bRe eaq 1 erf aq , (191)
where q ¼ s + iz, see Sveshnikov and Tikhonov (2005), and
Zq
2
ex dx
2
erf ðqÞ ¼ pffiffiffi
π
0
is the error function, e.g., see Abramowitz and Stegun (2012).

A resonant nature of the Rayleigh wave is also clearly seen from the
formula for the normal stress under the stamp

2μBα3R c 2 cR2
σ 33 ðs, 0Þ ¼ φ1, z ðs,0Þ at s 2 S2 , (192)
ðβ2R 1ÞcR2
demonstrating that the limit as c ! cR corresponds to asymptotically
vanishing stresses induced by displacements of finite magnitude.
The approach exposed in this section can also be adapted for high-speed
moving cracks and dislocations, see, e.g., important contributions by
Eshelby (1949) and Yoffe (1951), along with the substantial book by
Freund (1998). In particular, it would be interesting to apply the hyper-
bolic–elliptic model for surface waves to analysis of crack front waves, see
Willis and Movchan (1995), Morrissey and Rice (1998), Norris and
Abrahams (2007), and references therein.
5.3 Interfacial Waves

In this section we implement the proposed methodology to the
Sch€olte–Gogoladze and Stoneley interfacial waves introduced in
Section 3. Similarly to the Rayleigh wave, we obtain a scalar formulation
for an elliptic equation over the interior, subject to the condition in the form
of a hyperbolic equation along the interface.
5.3.1 Scho€lte–Gogoladse Wave

The equations of motion are given by (27) and (44), respectively. Let us
focus on the wave field arising from a vertical interfacial load P, assuming
the following contact conditions at x3 ¼ 0
σ 31 ¼ 0, u3 ¼ v, σ 33 pf ¼ Pðx1 ,tÞ: (193)
The relations (193) expressed in terms of the potentials ϕ, ψ, and χ become
2ϕ, 13 ψ , 11 + ψ , 33 ¼ 0,
ϕ, 3 + ψ , 1 χ , 3 ¼ 0, (194)

μ κ 2 2 ϕ, 11 + κ2 ϕ, 33 2ψ , 13 ρf χ , tt ¼ P:
Then, similarly to the consideration in Section 4.1, we establish a multi-
scale perturbation procedure resulting in an approximate hyperbolic–elliptic
formulation for the Sch€olte–Gogoladze wave. The behavior over the interior
is again governed by the elliptic equation
ϕ, 33 + α2SG ϕ, 11 ¼ 0, (195)
whereas the wave propagation along the interface is described by the hyper-
bolic equation
@ 2 ϕ 1 @ 2 ϕ 1 + β2SG
2 ¼ P, (196)
@x21 cSG @t 2 2μBSG
where
βSG αSG
BSG ¼ 1 α2SG + 1 β2SG 1 + β4SG
αSG βSG
(197)
ρf ð1 β2SG Þ2 γ 2SG α2SG 4α2SG γ 2SG
,
ρ 4αSG γ 3SG
with αSG, βSG, and γ SG defined by (50).
The potentials ψ and χ are related to ϕ as
2αSG
ψðx1 , βSG x3 , tÞ ¼ ϕ*ðx1 , βSG x3 ,tÞ, (198)
1 + β2SG
and
1 β2SG
χðx1 , γ SG x3 ,tÞ ¼ ϕðx1 , γ SG x3 ,tÞ: (199)
1 + β2SG
5.3.2 Stoneley Wave

In case of the wave propagating along a solid–solid interface the statement of
the problem includes the equations of motion (54), subject to
ϕ1, 1 ϕ2, 1 + ψ 1, 3 ψ 2, 3 ¼ 0,
ϕ1, 3 ϕ2, 3 ψ 1, 1 + ψ 2, 1 ¼ 0,

2μ1 ϕ1, 13 2μ2 ϕ2, 13 + μ1 ψ 1, 33 ψ 1, 11 μ2 ψ 2, 33 ψ 2, 11 ¼ 0, (200)
λ1 ϕ1, 11 + ðλ1 + 2μ1 Þϕ1, 33 λ2 ϕ2, 11 ðλ2 + 2μ2 Þϕ2, 33
2μ1 ψ 1, 13 + 2μ2 ψ 2, 13 ¼ P,
where P ¼ P(x1, t) is once again a prescribed vertical disturbance.
The asymptotic model for the Stoneley wave arising from the boundary
value problem (54) and (200), contains the elliptic equation
ϕ1, 33 + α21S ϕ1, 11 ¼ 0, (201)
governing the behavior over the interior. The rest of the wave potentials are
determined by
f4
ψ 2 ðx1 , β2S x3 , tÞ ¼ ϕ* ðx1 , β2S x3 , tÞ,
f1 β2S 1
f2
ϕ2 ðx1 , α2S x3 , tÞ ¼ ϕ1 ðx1 ,α2S x3 ,tÞ, (202)
f1
f3
ψ 1 ðx1 , β1S x3 , tÞ ¼ ψ 2 ðx1 ,β1S x3 ,tÞ,
f4
where
f1 ¼ ðm12 ρ1 cS2 Þb2 + ρ2 cS2 ð1 + α2S β1S Þ,

f2 ¼ ðρ2 cS2 + m12 Þb1 ρ1 cS2 ð1 + α1S β2S Þ,
(203)
f3 ¼ ρ2 cS2 ðα1S + α2S Þ m12 α1S b2 ,
f4 ¼ ρ1 cS2 ðα1S + α2S Þ m12 α2S b1 ,
with the problem parameters introduced in Section 3.2, see (59).

The hyperbolic equation for ϕ1 on the interface x3 ¼ 0 is written as
1 f1 P
ϕ1, 11 2 ϕ1, tt ¼ 2 , (204)
cS cS BS
where the constant BS is given by

BS ¼ 2cS2 ðρ1 ρ2 Þ2 a1 a2 m12 cS2 ðρ2 l2 ρ1 l1 Þ
m212 c4 (205)
ðb2 l1 + b1 l2 Þ S ðd1 a2 + d2 a1 Þ + 2m12 ðρ2 b1 ρ1 b2 Þ,
2 2
with
ρ2 ρ αnS β
dn ¼ + 12 , ln ¼ + nS2 , k ¼ 1,2:
αnS c1n βnS c2n
2 βnS c2n αnS c1n
2
It is remarkable that even though the models for the interfacial Stoneley
and Sch€ olte–Gogoladze waves contain rather cumbersome material con-
stants BSG and BS, the obtained formulations are not more difficult than that
for the Rayleigh wave, due to the relations (198), (199), and (202) allowing
reduction to a scalar elliptic problem for one of the potentials.
5.4 Bleustein–Gulyaev Wave

Next consider the antiplane motion of the transversely isotropic piezoelastic
half-plane H2+ , with the out of plane axis oriented along the direction of the
sixfold axis for a crystal in the symmetry class C6mm. The associated surface
wave was discovered independently by Bleustein (1968) and Gulyaev
(1969), see also Ikeda (1990). The governing equations of motion are
written as
u2, 11 + u2, 33 ce2 u2, tt ¼ 0, ψ e, 11 + ψ e, 33 ¼ 0, (206)
where u2 is the out of plane displacement, ψ e is a complementary function

related to the electric potential ϕe as
e15
ψ e ¼ ϕe u2 (207)
E11
with
1=2
c 44 e215
ce ¼ , c 44 ¼ c44 + : (208)
ρ E11
Here c 44 is the piezoelectrically stiffened elastic constant, ρ is the volume
mass density, ce is the low-frequency limit of the shear horizontal wave
speed, and c44, e15, and E11 are the elastic, piezoelectric, and dielectric
constants, respectively.
In what follows we study two types of boundary conditions along the
surface x3 ¼ 0, e.g., see Kaplunov et al. (2006), including a surface coated
by an infinitesimally thin perfectly conducting grounded electrode, for
which
σ 32 ¼ c 44 u2, 3 + e15 ψ e, 3 ¼ P, ϕe ¼ 0, (209)
and also a surface in contact with a vacuum modeled by
σ 32 ¼ P, ^ e,
ϕe ¼ ϕ ^ e, 3 :
e15 u2, 3 E11 ϕe, 3 ¼ ϕ (210)
^ e is the electric potential in a vacuum satisfying

In the above ϕ
^ e, 11 + ϕ
ϕ ^ e, 33 ¼ 0, x3 0, (211)
and P ¼ P(x1, t) denotes a prescribed mechanical load along the surface

x3 ¼ 0.
First, we derive the eigensolution of arbitrary time-dependence for a sur-

face coated with an electrode, see (209) at P ¼ 0. The traveling wave ansatz,
similar to (43), now takes the form
u2, tt c 2 u2, 11 ¼ 0, ϕe, tt c 2 ϕe, 11 ¼ 0, (212)
where c is the sought for phase speed. Then, (206) implies
u2, 33 + α2e u2, 11 ¼ 0, ψ e, 11 + ψ e, 33 ¼ 0, (213)
with
c2
α2e ¼ 1 2 : (214)
ce
The solution of (213) may be expressed in terms of plane harmonic func-
tions as
u2 ¼ u2 ðx1 , αe x3 , tÞ, ψ e ¼ ψ e ðx1 ,x3 , tÞ: (215)
On substituting the latter into (209) and employing the Cauchy–Riemann
identities, we have at x3 ¼ 0
αec 44 u2, 1 + e15 ψ e, 1 ¼ 0,
(216)
e15 u2 + E11 ψ e ¼ 0:
Hence, we get from the solvability
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c ¼ cBG ¼ ce 1 α2BG , (217)
coinciding with Bleustein (1968), where αBG is the electromechanical cou-
pling factor
e215
αBG ¼ : (218)
E11c 44
The eigenfunctions u2 and ψ e are related as
e15
u2 ðx1 ,αBG x3 , tÞ ¼ ψ ðx1 ,αBG x3 ,tÞ, (219)
c 44 αBG e
being a counterpart of (37) for the Rayleigh wave.
In the same manner, for a surface contacting with a vacuum we insert
(215) together with the implicit ansatz
^ e, tt c 2 ϕ
ϕ ^ e, 11 ¼ 0 (220)
into (210), having the following boundary conditions at x3 ¼ 0

αec 44 u2, 1 + e15 ψ e, 1 ¼ 0,
e15 u2 + E11 ðψ e ϕ ^ e Þ ¼ 0, (221)
^ e, 3 ¼ 0:
E11 ψ e, 3 + ϕ
These formulae lead to the same expression (217), with the coupling factor
e215
αBG ¼ : (222)
E11 ð1 + E11 Þ c 44
Finally, in addition to (219), the electrical potential in a vacuum is
given by
^ e ðx1 , x3 , tÞ ¼ E11 ψ e ðx1 , x3 , tÞ:
ϕ (223)
In case of a nonzero mechanical forcing P in (209) and (210) a hyper-

bolic–elliptic formulation extracting the contribution of the piezoelastic sur-
face wave to the overall dynamic response may be derived, for more detail
see Kaplunov et al. (2006). For a coated surface (209) it contains the elliptic
equation
u2, 33 + α2BG u2, 11 ¼ 0, (224)
where αBG is defined by (218), with the boundary condition at x3 ¼ 0 arising

from the hyperbolic equation
1 2α2BG
χ e, 11 χ e, tt ¼ P, 11 , (225)
2
cBG ρcBG
2
where χ e ¼ u2, 3. The function ψ e is then determined from (215). For a sur-
face contacting with a vacuum we get the same relations (224), (225), and
(215), with αBG now given by (222). The electrical potential in a vacuum ϕ ê
may be obtained from (223).
5.5 Effect of Anisotropy

The described methodology may also be adapted for anisotropic media. The
surface wave eigensolution in terms of a single harmonic function has been
recently derived in Parker (2013) for arbitrary anisotropy by means of the
Stroh formalism. Here we present briefly a more explicit result obtained
by Prikazchikov (2013) for an orthorhombic half-plane.
The equations of motion are written as

c11 u1, 11 + c55 u1, 33 + ðc13 + c55 Þu3, 13 ¼ ρ€
u1 ,
(226)
ðc13 + c55 Þu1, 13 + c55 u3, 11 + c33 u3, 33 ¼ ρ€
u3 ,
where c11, c13, c33, and c55 are the stiffness components satisfying the
conditions
c11 > 0, c11 c33 c13
2
> 0, c55 > 0, (227)
ensuring the positive definiteness of the strain-energy density, see Chadwick
(1976a). The stress-free boundary conditions along the surface x3 ¼ 0 of the
half-plane H2+ are expressed in terms of the displacements as
u1, 3 + u3, 1 ¼ 0, c13 u1, 1 + c33 u3, 3 ¼ 0: (228)
As in Section 2.2, we start from the implicit traveling wave ansatz
uj, tt c 2 uj, 11 ¼ 0, j ¼ 1,3, (229)
where, as usual, c is the sought for surface wave speed, see also (43). Substi-
tution of the latter into Eq. (226) gives

c11 ρc 2 u1, 11 + c55 u1, 33 + βu3, 13 ¼ 0,
(230)
βu1, 13 + c55 ρc 2 u3, 11 + c33 u3, 33 ¼ 0,
with β ¼ c13 + c55.
On eliminating one of the displacements, (230) may be transformed to
a fourth order partial differential equation. For example, we obtain in
terms of u1
u1, 3333 + δu1, 1133 + γu1, 1111 ¼ 0, (231)
where
2
c55 + c11 c33 β2 ðc33 + c55 Þρc 2 ðc11 ρc 2 Þðc55 ρc 2 Þ
δ¼ , γ¼ (232)
c33 c55 c33 c55
throughout this section.
It not difficult to verify that Eq. (231) is elliptic, with δ and γ coinciding
with the coefficients in the related secular equation for the attenuation
orders, see, e.g., Royer and Dieulesaint (1999). It may therefore be rewritten
in an operator form as

@33 + Λ21 @11 @33 + Λ22 @11 u1 ¼ 0, (233)
where
Λ21 + Λ22 ¼ δ, Λ21 Λ22 ¼ γ: (234)
Hence, assuming that Λn > 0, n ¼ 1, 2, the solution of (233) is
X
2
u1 ¼ Un ðx1 , Λn x3 , t Þ, (235)
n¼1
where Un are arbitrary plane harmonic functions in the first two arguments.
Then, on employing the Cauchy–Riemann identities, it is possible to
express the remaining displacement from (230) as
X
2
u3 ¼ T ðΛn , ρc 2 ÞUn* ðx1 , Λn x3 , tÞ, (236)
n¼1
with the asterisk, as before, denoting a harmonic conjugate and
c55 Λ2n c11 + ρc 2

T ðΛn ,ρc 2 Þ ¼ : (237)
βΛn
The solutions (235) and (236) are now substituted into the boundary
conditions (228), giving
X
2
T Λn , ρc 2 Λn Un, 1 ðx1 ,0,t Þ ¼ 0,
n¼1
(238)
X
2
c13 + c33 Λn T ðΛn , ρc 2 Þ Un, 1 ðx1 , 0,tÞ ¼ 0:
n¼1
The solvability dictates

2
c33 c55 ρ2 c 4 c11 ρc 2 c55 ρc 2 c33 c11 ρc 2 c132 ¼ 0, (239)
coinciding with the surface wave equation in an orthorhombic half-plane,

see again Royer and Dieulesaint (1999). Then, as in Section 2.2, we express
the displacements in terms of a single harmonic function. Indeed, as follows
from (238), the functions U1 and U2 are related as
U2 ðx1 ,Λ2 x3 , tÞ ¼ Y ðρcR2 Þ U1 ðx1 , Λ2 x3 , tÞ, (240)

where c ¼ cR is the unique real root of (239) and

T ðΛ1 ,ρcR2 Þ Λ1
Y ðρcR2 Þ ¼ :
T ðΛ2 ,ρcR2 Þ Λ2
Therefore, the representation of the surface wave field through the plane
harmonic function U1 is given by
u1 ðx1 , x3 , tÞ ¼ U1 ðx1 , Λ1 x3 , tÞ Y ðρcR2 Þ U1 ðx1 , Λ2 x3 ,tÞ,

u2 ðx1 , x3 , tÞ ¼ T ðΛ1 , ρcR2 ÞU1* ðx1 , Λ1 x3 , t Þ T Λ2 , ρcR2 Y ðρcR2 Þ U1* ðx1 ,Λ2 x3 , t Þ:
(241)
This eigensolution is a key preliminary step for deriving a hyperbolic–elliptic
model for the surface wave in anisotropic media.
6. MOVING LOAD PROBLEMS

This is seemingly the optimal setup for validating the asymptotic con-
siderations in Sections 4.1, 4.2, and 5.1. First of all, a near-resonant moving
load obeys the basic assumption (73). At the same time there are several
benchmark exact solutions available in literature, including the famous paper
by Cole and Huth (1958). In addition, a further insight into near-resonant
regimes of moving loads is inspired by the needs of modern high-speed
transport, e.g., see Dieterman and Metrikine (1996), de Hoop (2002),
Cao, Xia, and Li (2012), and also experimental data in Madshus and
Kaynia (2000).
6.1 Plane Strain Steady-State Problem

We begin with the plane strain problem for a steadily moving vertical line
force, see Fig. 18, using the hyperbolic–elliptic model derived in
Section 4.1, with P ¼ P0δ(x1 ct) in the wave equation (96) along the sur-
face x3 ¼ 0. Here c is a constant speed of the load, which is assumed to be
close to the Rayleigh wave speed, see (73).
Then, (96) reduces to
ð1 + β2R ÞcR2 P0
ϕ, ss ðs, 0Þ ¼ δðsÞ, (242)
2μBðcR2 c 2 Þ
where s ¼ x1 ct is a moving coordinate. Remarkably, the resonant effect at
c ¼ cR follows immediately from an elementary analysis of the surface behav-
ior (242). On integrating the last equation with respect to s, we deduce
P0 c
O x1
x3
Fig. 18 A line force traveling along the surface of a half-space.

ð1 + β2R ÞcR2 P0 1
ϕ, s ðs, 0Þ ¼ HðsÞ , (243)
2μBðcR2 c 2 Þ 2
where H denotes the Heaviside function and a constant of integration is cho-

sen because of symmetry.
Let us focus on the derivative φ ¼ ϕ,s, for which
φ, 33 + α2R φ, ss ¼ 0, (244)
following from (97), subject to the boundary condition (243). On

employing the Poisson formula, e.g., see Courant and Hilbert (1989), we
obtain
Z∞
ð1 + β2R ÞcR2 P0 αR x3 1
ϕ, s ðs,x3 Þ ¼ HðrÞ dr
2μBðcR2 c 2 Þ ðr sÞ2 + α2R x23 2
∞ (245)
ð1 + β2R ÞP0 cR2 s
¼ tan 1 :
2πμBðcR2 c 2 Þ αR x3
The derivative of the transverse potential ψ is restored through the relations

(38), giving
αR P0 cR2 2
ψ , s ðs,x3 Þ ¼ ln s + β 2 2
x : (246)
4πμBðcR c 2 Þ
2 R 3
On using the Cauchy–Riemann identities, the steady-state displacements

are written as

ð1 + β2R ÞP0 vR2 1 ξ 1 + β2R 1 ξ
ust1 ðξÞ ¼ tan tan ,
2μπBðvR2 v2 Þ αR 2 βR
(247)
ð1 + β2R ÞP0 vR2 αR 2 2 2
u2 ðξÞ ¼
st
ln ξ + αR
2
ln ξ + βR ,
2
4μπBðvR2 v2 Þ 1 + β2R
with the following dimensionless quantities

s c cR
ξ¼ , v¼ , vR ¼ :
x3 c2 c2
It is worth noting that the displacements (247) do not decay at infinity in

contrast to the Rayleigh wave eigensolutions in Section 2.2. At the same
time, it is possible to show that (247) is the leading order Taylor expansion
of the exact solution around the Rayleigh speed c ¼ cR presented in
Appendix B, see (B.11).
In Fig. 19 we present the graphs of the scaled stresses
πσ 33 x3
S33 ¼ , (248)
P0
1
against the speed v at ξ ¼ 0.1 and ν ¼ corresponding to vR 0.9194.
4
40
Approximate solution
30 Exact solution
20
10
S33
–10
–20
–30
–40
0.76 0.78 0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94
u
Fig. 19 The near-resonant stresses (249) and (250).
The exact solution

" #
α ð1 + β2 Þ2 4β2
S33 ¼ (249)
RðcÞ ξ2 + α2 ξ2 + β2
corresponding to the displacements (B.11) is shown by the solid line,

whereas the asymptotic formula

2αR β v2 αR βR
S33 ¼ 2 R2 R 2 + (250)
ðvR v ÞB ξ + α2R ξ2 + β2R
following from (247) is depicted by the dashed line. It may be easily shown
that over the near-resonant region, as the speed c is defined by (73), the lim-
iting behavior of (249) agrees with (250), since
vR v 2
α αR , β βR , RðcÞ 4εB, 2ε, (251)
vR2
with B defined in (93).
6.2 Transient Plane Strain Problem

Next, we address the associated transient problem. In this case the boundary
condition for (244) at x3 ¼ 0 is written as
1 ð1 + β2R Þ P0
ϕ, ss ϕ ¼ δðsÞ: (252)
cR2 , tt 2μB
On calculating the convolution of the moving impulse with the funda-
mental solution, see (141), we have for the potential ϕ along the surface, see
Kaplunov et al. (2010) for more detail,
Zt
ϕðs, 0,tÞ ¼ B0 ½H ðs + ðc cR Þr Þ H ðs + ðc + cR Þr Þdr, (253)
0
where B0 has been introduced in (171). Then, we immediately derive

from (253)
8 s s1
>
> B0 , 0 s < s1 ;
>
< cR c
ϕðs, 0,tÞ ¼ B s s2 , s < s < 0; (254)
>
> 0 2
>
: cR + c
0, otherwise,
and
8 cR s
>
> 2B0 2 2 , s1 s 0;
>
< c cR
ϕðs, 0,tÞ ¼ B s s2 (255)
> , s2 < s < s1 ;
>
>
0
cR + c
:
0, otherwise:
for the sub-Rayleigh (c < cR) and super-Rayleigh (c > cR) regimes, respectively.
In the above the values s1 and s2 are
s1 ¼ tðcR cÞ, and s2 ¼ tðcR + cÞ: (256)
For the resonant regime (c ¼ cR) we have
( s s2
B0 , s2 s 0;
ϕðs, 0,tÞ ¼ 2cR (257)
0, otherwise,
with s2 ¼ 2cRt.
Analysis of the 1D problem along the surface, resulting in the solutions
(254), (255), and (257), provides an immediate insight into the peculiarities
of the near-resonant transient phenomena. The plots of the function
ϕ(s, 0, t) at a fixed time instance t in Fig. 20 show that the resonant regime
is clearly distinctive from the two others. If c6¼cR, the solution in question is
continuous in s, see Fig. 20A and B. At the same time, the limiting resonant
solution in Fig. 20C demonstrates a discontinuity under the force (s ¼ 0),
which grows linearly in time. As a result, we should not expect a steady-state
regime at c ¼ cR. Thus, a rather trivial analysis of the 1D moving load prob-
lem (252) for an infinite string reveals the resonant phenomena associated
with the Rayleigh wave.
Once the potential ϕ is determined along the surface x3 ¼ 0, it may then
be restored over the interior through the Poisson formulae, as in the previ-
ous section. In the sub-Rayleigh and super-Rayleigh regimes the displacement
components are given by

2B0 vR ξ 1 + β2R ξ
u1 ðξ, τÞ ¼ arctan arctan
πc2 ðvR2 v2 Þ αR 2 βR

B0 ξ ξ2 1 + β2R ξ ξ2
arctan arctan (258)
πc2 ðvR + vÞ αR 2 βR

B0 ξ ξ1 1 + β2R ξ ξ1
arctan arctan ,
πc2 ðvR vÞ αR 2 βR
A
f (s,0, t)
- B0t
s2 0 s1 s
B
f (s,0, t)
2B0tcR
-
c + cR
s2 0 s
C f (s,0, t)
- B0t
s2 0 s
Fig. 20 The longitudinal wave potential ϕ vs the moving coordinate s along the surface
x3 ¼ 0. (A) sub-Rayleigh regime (c < cR); (B) super-Rayleigh regime (c > cR); (C) resonant
regime (c ¼ cR).
and
" #
B0 αR ðξ ξ2 Þ2 + α2R 2 ðξ ξ2 Þ2 + β2R
u3 ðξ,τÞ ¼ ln ln
2πc2 ðvR + vÞ ξ2 + α2R 1 + β2R ξ2 + β2R
" #
+ ln ln ,
2πc2 ðvR vÞ ξ2 + α2R 1 + β2R ξ2 + β2R
(259)
with
c2 t s1 s2
τ¼ , ξ1 ¼ ¼ ðvR vÞτ, ξ2 ¼ ¼ ðv + vR Þτ, (260)
x3 x3 x3
and s1 and s2 defined by (256). For the resonant regime we obtain

B0 αR τ 1 2β2R
u1 ðξ,τÞ ¼
πc2 ξ2 + α2R ð1 + β2R Þðξ2 + β2R Þ

β ξ ξ ξ2
+ arctan arctan (261)
2πc2 vR αR αR
2
B0 ð1 + βR Þ ξ ξ ξ2
arctan arctan ,
4πc2 vR βR βR
and

B0 αR ξτ 2 1
u3 ðξ,τÞ ¼
πc2 ð1 + β2R Þðξ2 + β2R Þ ξ2 + α2R
" #
+ ln ln ,
4πc2 vR ξ2 + α2R 1 + β2R ξ2 + β2R
(262)
with ξ2 ¼ 2vRτ.
The obtained displacements (258)–(262) are expressed in terms of ele-
mentary functions in contrast to the integral exact solution of the problem
presented in Appendix of Kaplunov et al. (2010). The approximate solution
also captures all of the key features of the studied problem. In particular, a
large time limiting behavior as τ ! ∞ is immediately deduced from the for-
mulae above. In the sub-Rayleigh regime we have
ui ðξ, τÞ u∞
i ðξ,τÞ, u∞
i ðξ,τÞ ¼ ui ðξÞ + ui ðτÞ, i ¼ 1,2,
st r
(263)
where usti are the steady-state displacements calculated in the previous sec-
tion, see (247), whereas
ur1 ðτÞ ¼ ur0

1, ur2 ðτÞ ¼ ur0
2 + u2 ðτÞ,
rτ
(264)
with

B0 v 1 β2R
ur0 ¼ ,
1
2c2 ðvR2 v2 Þ

B0 αR ð1 β2R Þ ln ðvR + vÞ ln jvR vj
u2 ¼
r0
+ ,
πc2 ð1 + β2R Þ vR + v vR v (265)
and
2B0 vR αR ð1 β2R Þ
2 ðτÞ ¼
urτ ln τ:
πc2 ðvR2 v2 Þð1 + β2R Þ
Here uri (i ¼ 1, 2) are the components of the rigid body motion of the half-
space. It is remarkable that the rigid body motion along the vertical axis
demonstrates a logarithmic growth in time according to (264) and (265),
which was first noted in Kaplunov (1986). This means that the steady-state
regime for the vertical displacement established in Section 5.1, cannot be
reached at a large time limit.
The formulae (263)–(265) are also valid for the super-Rayleigh case,
except the expression for the rigid body motion component along the hor-
izontal axis, which is now given by

B0 vR 1 β2R
u1 ¼
r0
: (266)
2c2 ðvR2 v2 Þ
In the resonant case the limiting behavior as τ ! ∞ is
ui ðξ,τÞ u∞
i ðξ, τÞ, i ¼ 1,2, (267)
with

B0 αR τ 1 2β2R
u∞
1 ðξ, τÞ ¼ , (268)
πc2 ξ2 + α2R ð1 + β2R Þðξ2 + β2R Þ
and

B0 αR ξτ 2 1
u∞
2 ðξ,τÞ ¼ 2 2
πc2 ð1 + βR Þðξ + βR Þ ξ + α2R
2 2
(269)
B0 αR ðβ2R 1Þ
+ ln τ:
4πc2 vR ðβ2R + 1Þ
Thus, the displacements exhibit a linear growth in time, except for the ver-
tical displacement at ξ ¼ 0, which increases as lnτ.
Numerical examples in Figs. 21–23 were produced taking the Poisson’s

1
ratio ν ¼ , which corresponds to vR 0.9194. We plot the dimensionless
4
quantities
πμui
Ui ¼ , i ¼ 1,2,
P0
A
0
–2
–4
–6
U1
–8
–10 t = 10
t = 50
–12 t = 100
t = 200
–14
t = 1000
–16 Large time limit
–4 –3 –2 –1 0 1 2 3 4
x
B
0
–10
–20
Ũ2
–30
t = 50
t = 100
–40
t = 200
t = 1000
–4 –3 –2 –1 0 1 2 3 4
x
Fig. 21 The sub-Rayleigh (A) horizontal and (B) vertical displacements (258) and (259)
and their large time limits for v ¼ 0.9.
–2
–4
U1
–6
t = 10
t = 50
–8
t = 100
t = 1000
–4 –3 –2 –1 0 1 2 3 4
x
B 30 t = 10
t = 50
25
t = 100
t = 1000
20
Large time limit
15
Ũ2
10
–5
–4 –3 –2 –1 0 1 2 3 4
x
Fig. 22 The super-Rayleigh (A) horizontal and (B) vertical displacements (258) and (259)
and the large time limits (263) for v ¼ 0.95.
and also
πμ
U~ 2 ¼ u2 ðξ, τÞ urτ
2 ðτÞ ,
P0
subtracting from the vertical displacement u2(ξ, τ) the function urτ
2 ðτÞ which
has a logarithmic growth in time, see (264) and (265). In this case we depict
A
8 t = 10
t = 30
6
t = 50
4 t = 100
0
U1
–2
–4
–6
–8
–10
–3 –2 –1 0 1 2 3
x
B 40 t = 10
t = 30
30
t = 50
20 t = 100
10
U2
–10
–20
–30
–40
–3 –2 –1 0 1 2 3
x
Fig. 23 The resonant (A) horizontal and (B) vertical displacements (261) and (262) for
v ¼ vR.
only a bounded in time component in order to achieve convergence at a

large time limit.
The sub-Rayleigh displacements of the half-space (258) and (259) are
plotted in Fig. 21 for v ¼ 0.9 and several values of time τ. Similar results
for the super-Rayleigh regime are given in Fig. 22 at v ¼ 0.95. The limiting
behaviors (263) are shown by solid lines. As might be expected, the transient
displacements tend to their large time values as time increases. The resonant
displacements (261) and (262) are displayed in Fig. 23 for τ ¼ 10, 30, 50, and
100. They demonstrate a linear growth in time according to the formulae
(261) and (262).
Fig. 24 illustrates a comparison of the horizontal displacement obtained
from the asymptotic model with that arising from the exact solution of the
associated plane strain problem presented in Appendix of Kaplunov et al.
1
(2010). The computations are performed for v ¼ 0.9, ξ ¼ 0, and ν ¼ . It
4
is readily observed from Fig. 24 that the use of the asymptotic solution is
justified once the effect of the bulk waves can be neglected, say, at τ 10. This
observation is in agreement with the validity of the formula (258), which
fails at small times, when the arguments of tan 1 become small. This results
in the absence of the pole v ¼ vR violating the original assumption of the
Rayleigh wave dominance.
Another interesting example is concerned with the resonant regime of a
moving semiinfinite strip, for which P ¼ P0H(x1 cRt), see Fig. 25. In this
case the asymptotic model for the Rayleigh wave recovers a rather technical
result in Gol’dshtein (1965), but with much less effort. Indeed, the derivative
of the step load P corresponds to a moving impulse. Thus, the sought for dis-
placements may be obtained by straightforward integration of (261) and (262).
1
Asymptotic model
Exact solution
0.5
-0.5
U1
-1
-1.5
-2
0 5 10 15 20 25 30 35 40 45
␶
Fig. 24 Comparison of approximate and exact solutions.
cR
P0
O x1
x3
Fig. 25 A moving step load.
In the latter case we have for the stresses at τ ! ∞

ð1 + β2R Þvr P0 2α2R + β2R 1 βR ð1 + β2R Þ
σ 11 2 τ,
4μπB ξ2 + α2R ξ + β2r

αR β R v r P0 βR αR
σ 22 τ, (270)
μπB ξ2 + β2R ξ2 + α2R
ð1 + β2R ÞαR vr P0 α2R β2R
σ 12 ξτ,
2μπB ðξ2 + β2R Þðξ2 + α2R Þ
for more detail see Kaplunov et al. (2010). Note that the expression for σ 11
coincides with the formula (2.4) presented in Gol’dshtein (1965).
6.3 3D Steady-State Problem

Let us now study the steady-state near-resonant regimes of a point vertical
force
P ¼ P0 δðx2 Þδðx1 ctÞ,
traveling along the Ox1 axis at a constant speed c, see Fig. 26. From the very
beginning we distinguish between the sub-Rayleigh and super-Rayleigh
regimes, having from (120) the elliptic equation
ð1 + β2R ÞP0
ϕ, 22 + E2 ϕ, ss ¼ δðsÞδðx2 Þ (271)
2μB
for the sub-Rayleigh case, and the hyperbolic equation
ð1 + β2R ÞP0
ϕ, 22 E2 ϕ, ss ¼ δðsÞδðx2 Þ (272)
2μB
x2 x1
x3
Fig. 26 A moving vertical point load.
for the super-Rayleigh one. Here and below in this section the small param-
eter E is defined as

c 2
E ¼ 1 2 1=2 : (273)
cR
Let us introduce the scaling

s x3
η1 ¼ , η2 ¼ x2 , η3 ¼ : (274)
E E
and rewrite Eqs. (271) and (272) as
ð1 + β2R ÞP0
ϕ, 22 + ϕ, 11 ¼ δðη1 Þδðη2 Þ (275)
2μBE
and
ð1 + β2R ÞP0
ϕ, 22 ϕ, 11 ¼ δðη1 Þδðη2 Þ (276)
2μBE
with the subscript i indicating differentiation along the variable ηi, i ¼ 1, 2, 3.
The elliptic equations (129) and relations (130) now take the form
ϕ, 33 + α2R ϕ, 11 ¼ 0, ψ i, 33 + β2R ψ i, 11 ¼ 0, i ¼ 1, 2: (277)
and
1 + β2R ð1 + β2R ÞE
ψ 1, 3 ¼ ϕ, 1 , ψ 2, 3 ¼ ϕ, 2 : (278)
2 2
6.3.1 Sub-Rayleigh Regime

For the sub-Rayleigh speeds c < cR the solution along the surface is
ð1 + β2R ÞP0
ϕðη1 ,η2 ,0Þ ¼ lnðη21 + η22 Þ, (279)
8πμBE
see, e.g., Polyanin and Nazaikinskii (2015). Then, we restore the potential ϕ
over the interior from (277), using the Poisson formula. The result is
ð1 + β2R ÞP0 2
ϕðη1 , η2 , η3 Þ ¼ ln η1 + ðαR η3 + jη2 jÞ2 : (280)
8πμBE
This solution is formally not differentiable with respect to η2 along the plane
η2 ¼ 0 due to the omitted O(ε2) terms in Eq. (277). Now, using (277) for the
transverse wave potentials ψ 1 and ψ 2 together with (278), we have
αR P0 β η + jη2 j
ψ 1 ðη1 ,η2 ,η3 Þ ¼ tan 1 R 3 , (281)
2πμBE η1
and
αR P0
ψ 2 ðη1 ,η2 , η3 Þ ¼ sgnðη2 Þ ln η21 + ðβR η3 + jη2 jÞ2 : (282)
4πμBE
The leading order displacements are found from (122) as

" #
ð1 + β2R ÞP0 η1 2 1 + β2R
u1 ¼ + , (283)
8μπBE2 η21 + ðαR η3 + jη2 jÞ2 η21 + ðβR η3 + jη2 jÞ2
" #
ð1 + β2R ÞP0 sgnðη2 Þ 2ðαR η3 + jη2 jÞ 1 + β2R ðβR η3 + jη2 jÞ
u2 ¼ + 2 ,
8μπBE η21 + ðαR η3 + jη2 jÞ2 η1 + ðβR η3 + jη2 jÞ2
(284)
and
" #
ð1 + β2R ÞP0 2αR ðαR η3 + jη2 jÞ 1 + β2R ðβR η3 + jη2 jÞ
u3 ¼ : (285)
8μπBE2 η21 + ðαR η3 + jη2 jÞ2 βR η21 + ðβR η3 + jη2 jÞ2
We note that the asymptotically secondary displacement u2 has a disconti-

nuity at η2 ¼ 0 coming from differentiation of (280) with respect to η2.
Numerical results for the scaled displacements

μ ui
Ui ¼ , i ¼ 1,2,3,
P0
are presented in Figs. 27–29. The graphs in these figures show no disconti-
nuities, with the displacement amplitudes increasing slightly while getting
closer to the surface.
h3 = 1
20 h3 = 0.5
h3 = 0
15
10
5
U1
-5
-10
-15
-20
-6 -4 -2 0 2 4 6
h1
Fig. 27 The sub-Rayleigh diplacement U1 vs η1 at jη2j ¼ 1 and η3 ¼ 0, 0.5, 1.0.
5
h3 = 1
h3 = 0.5
4 h3 = 0
3
U2
0
-10 -5 0 5 10
h1

-1
h3 = 1
-2 h3 = 0.5
h3 = 0
-3
U3
-4
-5
-6
-6 -4 -2 0 2 4 6
h1
6.3.2 Super-Rayleigh Regime

Inspection of the hyperbolic equation (276) along the surface η3 ¼ 0 imme-
diately suggests a Mach cone associated with the Rayleigh wave, see Fig. 30.
Then, on employing the causality principle, i.e., assuming absence of surface
disturbances in front of a traveling load, we obtain, see Erbaş and Şahin
(2016) for further detail,
ð1 + β2R ÞP0
ϕðη1 ,η2 ,0Þ ¼ Hðη1 Þ½Hðη2 η1 Þ Hðη2 + η1 Þ: (286)
2μBE
The potential ϕ over the interior is determined from (277), giving
ð1 + β2R ÞP0 η + jη2 j

ϕðη1 ,η2 ,η3 Þ ¼ cot 1 1 : (287)
4πμBE αR η3
The potentials ψ 1 and ψ 2 are then found from (277) and (278) as
P0 α2R
ψ 1 ðη1 , η2 , η3 Þ ¼ ln ðη1 + jη2 jÞ2 + α2R η23 (288)
4πμBβR E
h2
h
1 =-
h
2
O h1
=h 2
h1
Fig. 30 A Mach cone.
and
P0 α2R sgnðη2 Þ
ψ 2 ðη1 ,η2 ,η3 Þ ¼ ln ðη1 + jη2 jÞ2 + α2R η23 : (289)
4πμBβR
Finally, the leading order displacement components corresponding to

(287)–(289) are obtained from (122). They are
" #
ð1 + β2R ÞP0 αR η1 2 1 + β2R
u1 ¼ + ,
8μπBE2 ðη1 + jη2 jÞ2 + α3R η23 ðη1 + jη2 jÞ2 + β3R η23
(290)
" #
ð1 + β2R ÞP0αR η3 sgnðη2 Þ 2 1 + β2R
u2 ¼ + ,
8μπBE ðη1 + jη2 jÞ2 + α3R η23 ðη1 + jη2 jÞ2 + β3R η23
(291)
and
" #
ð1 + β2R ÞP0 ðη1 + jη2 jÞαR 2 1 + β2
u3 ¼ R
:
8μπBE2 ðη1 + jη2 jÞ2 + α3R η23 ðη1 + jη2 jÞ2 + β3R η23
(292)
The discontinuities along the surface η3 ¼ 0 associated with the Mach cone
η1 + jη2j ¼ 0, can be seen from the formulae (290)–(292).
Numerical results are displayed in Figs. 31–33. In contrast to the sub-
Rayleigh case, the analyzed super-Rayleigh displacements increase signifi-
cantly near the surface η3 ¼ 0. The discontinuities occurring on the surface
at η1 ¼ 1 are shown by solid line.
-50 h3 = 1
h3 = 0.5
-100
h3 = 0 .1
-150
-200
U1
-250
-300
-350
-400
-2 -1.5 -1 -0.5 0
h1
Fig. 31 The super-Rayleigh diplacement U1 vs η1 at jη2j ¼ 1 and η3 ¼ 1, 0.5, 0.1.
h3 = 1
-5
h3 = 0.5
-10 h3 = 0.1
-15
-20
U2
-25
-30
-35
-40
-2.5 -2 -1.5 -1 -0.5 0 0.5

h1
Fig. 32 The super-Rayleigh diplacement U2 vs η1 at jη2j ¼ 1 and η3 ¼ 1, 0.5, 0.1.
6.4 3D Steady-State Problem for a Coated Half-Space

Let us extend the consideration of the previous section to an elastic half-
space H3+ coated by a thin layer occupying the domain h x3 0, subject
to the moving point force P ¼ P0δ(x1 ct)δ(x2), see Fig. 34. The statement
h3 = 1
300
h3 = 0.5
h3 = 0.1
200
100
U3
-100
-200
-300
-3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1

h1
Fig. 33 The super-Rayleigh diplacement U3 vs η1 at jη2j ¼ 1 and η3 ¼ 1.0, 0.5, 0.1.
P
c
-h
0 x1
x2
x3
Fig. 34 A moving point load on a coated half-space.
of the problem includes now a pseudo-differential equation at x3 ¼ 0, see

(163), which may be written as

c2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ϕ, 22 + 1 2 ϕ, ss bh ð@, ss + @, 22 Þ ϕ, ss + ϕ, 22
cR
(293)
ð1 + βR ÞP0
2
¼ δðsÞδðx2 Þ,
2μB
where, as previously, s ¼ x1 ct is a moving coordinate. The formulated
problem has two small parameters, including E ≪ 1 defined by (273), and
h/L ≪ 1 over the long-wave domain, see Section 5.1. The degenerations at
E ¼ 0 and h ¼ 0 correspond to the critical speed of the load coinciding with
the Rayleigh wave speed and a homogeneous half-space, respectively. These
observations motivate the scaling
E2 s E3 s
ξ¼ , η¼ , (294)
bh bh
for more detail throughout this section see Erbaş, Kaplunov, Prikazchikov,
and Şahin (2017). Then, (293) takes the form of
pffiffiffiffiffiffiffiffiffiffiffi ð1 + β2R ÞP0
ϕ, ξξ + ϕ, ηη @, ξξ ϕ, ξξ ¼ δðξÞδðηÞ (295)
2μBE
or
pffiffiffiffiffiffiffiffiffiffiffi ð1 + β2R ÞP0
ϕ, ξξ ϕ, ηη @, ξξ ϕ, ξξ ¼ δðξÞδðηÞ (296)
2μBE
for the sub-Rayleigh and super-Rayleigh regimes, respectively.
First, we investigate the sub-Rayleigh case. On applying the Fourier
transform in (295), we have
ð1 + β2R ÞP0
ϕF, ηη k2 ð1 jkjÞϕF ¼ δðηÞ, (297)
2μBE
where
Z∞
ϕF ðk, η, 0Þ ¼ ϕðξ,η,0Þeikξ dξ: (298)
∞
The solution of this equation is piecewise-defined in the parameter jkj. Due

to symmetry in η along with decay at infinity, it may be written as
8 pffiffiffiffiffiffiffiffi
>
> ð1 + β 2
ÞP e jkj 1jkjjηj
>
> R 0
pffiffiffiffiffiffiffiffiffiffiffiffiffi , jkj < 1;
>
>
< 2μBE jkj 1 jkj
ϕF ðk, η, 0Þ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi
(299)
>
> jkj jkj jηj
>
> ð1 + βR ÞP0
2 sin 1
>
> pffiffiffiffiffiffiffiffiffiffiffiffiffi , jkj > 1:
: 2μBE
jkj jkj 1
Therefore, using (123), we have for the horizontal displacement u1 along the
plane x3 ¼ 0
2 pffiffiffiffiffiffi
Z1
ð1 + β2R ÞP0 εcR2 sgnðξÞ 4 ek 1kjξjm
u1 ðξ, η, 0Þ ¼ pffiffiffiffiffiffiffiffiffiffi sin ðkjξjÞdk
4πμBc22 bh 1k
pffiffiffiffiffiffiffiffiffiffi
0 3 (300)
Z∞
sin k k 1jξjm
pffiffiffiffiffiffiffiffiffiffi sin ðkjξjÞdk5,
k1
1

η
where m ¼ .
ξ
Let us study the far-field approximation jξj≫ 1. It can be shown that
the leading order asymptotic behavior of u1 is given by the contribution
of the stationary points arising from the second integral in (300). Changing
the variable k to t ¼ k 1, this integral takes the form
Z∞ pffiffiffiffiffiffiffiffiffiffi
sin k k 1jηj X2
pffiffiffiffiffiffiffiffiffiffi sin ðkjξjÞdk ¼ Gn ðjξj, mÞ, (301)
k1 n¼1
1
where
Z∞
n+1
Gn ðjξj, mÞ ¼ ð1Þ cos jξjðt 2 + 1Þðtm + ð1Þn Þ dt: (302)
0
It may be shown that only the integrand in G1 possesses stationary points,

namely
1 1 3m2 1
t* ¼ , 0 < m pffiffiffi , (303)
3m 3
pffiffiffi
which coincide along the line jξj ¼ 3jηj. In this case the uniform stationary
phase method, see, e.g., Borovikov (1994), yields
2π
G1 ðjξj, mÞ ¼ p
3
ffiffiffiffiffiffiffiffiffiffi
ffi cos ðp 1 jξj ÞAi p2 jξj2=3
, (304)
3mjξj
where
Z∞ 3
1 t
AiðzÞ ¼ cos + zt dt (305)
π 3
0
is the Airy function, see Abramowitz and Stegun (2012), with

2ð9m2 + 1Þ 3m2 1
p1 ¼ , p2 ¼ : (306)
27 ð3mÞ4=3
The resulting far-field approximation for the displacement u1 is given by
ð1 + β2R Þ P0 cR2 sgnðξÞ
u1 p ffiffiffiffiffiffiffiffiffiffi
ffi cos ðp 1 jξj ÞAi p2 jξj2=3
: (307)
2μBc22 bh 3 3jξjm
The numerical illustrations in Figs. 35 and 36 demonstrate comparisons
of the solution (300) with its far-field asymptotic approximation (307)
depicted by the solid and dotted lines, respectively. Fig. 35 shows the var-
iation of the scaled displacement
2bhμBc22
U1 ¼ u1 , (308)
ð1 + β2R ÞP0 cR2 E
over ξ for the fixed value of jηj ¼ 3, whereas Fig. 36 presents the perpen-
dicular cross-section of U1 at jξj ¼ 7. It may be observed from Figs. 35 and
36 that even though there is no Mach cone for the sub-Rayleigh regime,
1
there is still a region of oscillations associated with m < pffiffiffi. The period
3
of these oscillations decreases as m ! 0. The analyzed profiles suggest an
Fourier integral
0.6 Asymptotic behavior
0.4
0.2
U1
-0.2
-0.4
-0.6
2 4 6 8 10 12 14 16 18 20
x
Fig. 35 The longitudinal cross-section of the sub-Rayleigh horizontal displacement U1
at jηj ¼ 3.
0.6 Fourier integral

Asymptotic behavior
0.4
0.2
U1
-0.2
-0.4
2 3 4 5 6 7 8
x
Fig. 36 The transverse cross-section of the displacement profile U1 at jξj ¼ 7.
1
exponential decay over the region m > pffiffiffi. Next, we plot a 3D graph
3
of the scaled displacement U1 corresponding to (300), over the region
4 ξ 0 and 0 η 4, see Fig. 37.
The super-Rayleigh case is treated similarly. The pseudo-differential
equation (296) becomes
ð1 + β2R ÞP0
ϕF, ηη + k2 ð1 + jkjÞϕF ¼ δðηÞ, (309)
2μBE
from which
Z∞ pffiffiffiffiffiffiffiffiffiffi
ð1 + β2R ÞP0 sin k 1 + kjηj cos ðkξÞ
ϕðξ, η, 0Þ ¼ pffiffiffiffiffiffiffiffiffiffi dk: (310)
2πμBE k 1+k
0
The displacement u1 is then given by
ð1 + β2R ÞP0 cR2 E X

2
u1 ðξ,η,0Þ ¼ In ðξ, ηÞ, (311)
8πμBc22 bh n¼1
2
U1 1
0
0
1 -1
-2
0
2
h -1
3 -2
-3 x
4
-4
Fig. 37 A 3D profile of the longitudinal sub-Rayleigh displacement U1.
where
Z∞
cos ðjξjfn ðkÞÞ
In ðξ,ηÞ ¼ sgnðξÞ dk, n ¼ 1,2, (312)
gðkÞ
0
with
fn ðkÞ ¼ k gðkÞm + ð1Þn + 1 , gðkÞ ¼ 1 + k,

η
and, as previously, m ¼ .
ξ
The far-field asymptotic behavior of the oscillating integrals (312) as
jξj≫ 1, assuming m 1, may now be studied. It appears that the effect
of the first integral I1 is asymptotically minor, whereas I2 is dominated by
the contribution of the stationary point
2 1 3m2 + 3m2 + 1
k* ¼ (313)
9m 2
of the function f2(k).

The Mach cone observed in Section 6.3.2, corresponds to the limit h ! 0,

and is defined by jξj ¼ jηj or m ¼ 1. Moreover, on the contour of the Mach
cone, when m ¼ 1, the stationary point k* coincides with the lower limit of
the integral I2. Therefore, we again apply the uniform stationary phase
method, leading to
2 3
ijξjf Z 1
∞
e 00
I2 Re4 e2ijξjf2 ðk Þðkk Þ dk5,
2
(314)
g*
0
where
2 1 3m2 + 3m2 + 1 3m2 + 1 2

f* ¼ f2 ðk* Þ ¼ , (315)
27m2
and
1 + 3m2 + 1
g* ¼ gðk* Þ ¼ : (316)
3m
The displacement u1 becomes
ð1 + β2R ÞP0 εcR2 k* sgnðξÞ

u1 Fðjξj,mÞ, (317)
16πμBc22 bhg* a jξj1=2
where
n pffiffiffiffiffi
o n pffiffiffiffiffi
o
Fðjξj, mÞ ¼ cos f* jξj 1 2C a jξj sin f* jξj 1 2S a jξj ,
(318)
with
h pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiip ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rffiffiffiffiffiffiffiffiffiffiffiffi
ffi rffiffiffi
00
h ðk* Þ 2 3m 2
1 3m2 + 1 4 3m2 + 1
a ¼ k* ¼ h pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii , (319)
π π 3m 1 + 3m2 + 1
and S(x) and C(x) are the Fresnel functions (181). It is evident that the
derived uniform asymptotic formula is also valid at m > 1 when k* < 0 and
a takes imaginary values.
The interpretation of the formulae in this section, written in terms of
jξj and jηj, relies on the implementation of the causality principle, see also
b<0 b>0
0 x
Fig. 38 A Mach cone.
Erbaş and Şahin (2016). In absence of a coating, when h ¼ 0, Eq. (296)

degenerates into the wave equation. Hence, it seems to be logical to deal
with the Mach cone behind the load, i.e., for ξb > 0, see Fig. 38. In the
presence of a coating, we have to expect decay of the solution outside
the interior of the Mach cones predicted by the related degenerate non-
dispersive equation. The asymptotic behaviors of the Fresnel functions in
(318) at the large imaginary values of the argument show that the function
(318) is exponentially small at m 1 ≫jξj1.
Similarly to the sub-Rayleigh case, we analyze the longitudinal and
transverse cross-sections of the scaled horizontal displacement
8μπBc22 bh
U1 ¼ u1 , (320)
ð1 + β2R ÞP0 cR2 E
see Figs. 39 and 40. The results of numerical integration in (311) are depicted
by the solid line, with the dotted line corresponding to the far-field approx-
imation (317). Fig. 39 displays dependence of U1 on the transverse variable jηj
at jξj ¼ 3. Fig. 40 mirrors Fig. 39, showing the variation of U1 vs jξj at jηj ¼ 3.
It is clearly seen from both Figs. 39 and 40 that the dispersive effect of a
coating leads to smoothing of the discontinuities along the contours of the
Mach cone jξj ¼ jηj, arising in a homogeneous half-space, see Section 6.3.
The oscillations occurring inside the Mach cone decay outside of it. The
Fourier integral
1.5
Asymptotic behavior
0.5
0 h
U1
-0.5
-1
-1.5
1 2 3 4 5 6 7
Fig. 39 The super-Rayleigh displacement U1 at jξj ¼ 3.
1 Fourier integral
Asymptotic behavior
0.5
x
U1
-0.5
-1
2 4 6 8 10 12 14
Fig. 40 The super-Rayleigh displacement U1 at jηj ¼ 3.
period of the oscillations diminishes in both graphs as m ! 0, due to f* ! ∞,

as may be noticed from (315). Finally, we present a 3D graph of the scaled
displacement profile corresponding to the integral solution (311), over the
region 4 ξ 0 and 0 η 6 see Fig. 41.
5
U1
0
0
2
-5
0
h -1
4
-2
-3 x
6
-4
Fig. 41 A 3D profile of the horizontal super-Rayleigh displacement U1.
7. PARABOLIC–ELLIPTIC MODEL FOR A BENDING EDGE

WAVE ON A THIN PLATE
In this section we derive an asymptotic formulation for the bending
edge wave introduced in Section 3.3, complementing the hyper-
bolic–elliptic models for surface and interfacial waves considered in
Sections 4 and 5.
7.1 Bending Edge Wave of Arbitrary Profile

We begin with an eigensolution generalizing that of a sinusoidal profile, see
(64). Throughout this section we operate with the dimensionless variables
1=2
x1 x2 t E
ξ¼ , η¼ , τ¼ : (321)
h h h 3ρð1 ν2 Þ
Then, (61) becomes
Δ2 W + W, ττ ¼ 0, (322)
where Δ now denotes the 2D Laplace operator in the variables ξ and η.

Next, we proceed with the implicit ansatz
@4W @2W
γ 4e + ¼ 0, (323)
@ξ4 @τ2
where γ e is defined by (68). The latter is in fact a counterpart of the implicit
traveling wave ansatz (43) for the Rayleigh wave. It is clear that the sinusoi-
dal solution (64) satisfies (323), leading to the dispersion relation (67). At the
same time, in contrast to (43), (323) does not allow a functionally invariant
traveling wave solution.
Now, using (323), we eliminate the time derivative in (322), arriving, as
in Section 2.2, at the pseudo-static equation
@ 4W @4W @4W
1 γ 4e +2 + ¼ 0, (324)
@ξ4 @ξ2 @η2 @η4
which may be rewritten in an operator form as
Δ1 Δ2 W ¼ 0, (325)
where
Δj ¼ @ηη + γ 2j @ξξ , j ¼ 1,2 (326)
and
γ 2j ¼ 1 + ð1Þj γ 2e : (327)
Eq. (325) is elliptic, since γ 2j are both positive at 0 < γ e < 1. Its general
solution is therefore given by the sum of two arbitrary plane harmonic func-
tions in the variables ξ and γ jη, i.e.,
X
2
W¼ Wj ξ, γ j η, τ : (328)
j¼1
On substituting (328) into the homogeneous edge boundary conditions

(63), rewritten in terms of the dimensionless variables and employing the
Cauchy–Riemann identities, we obtain

ν γ 21 W1, ξξ + ν γ 22 W2, ξξ ¼ 0,
(329)
γ 1 γ 21 2 + ν W1*, ξξξ + γ 2 γ 22 2 + ν W2*, ξξξ ¼ 0,
with the asterisk, as previously, denoting a harmonic conjugate. These con-

ditions imply
γ 21 γ 22 + 2ð1 νÞγ 1 γ 2 ν2 ¼ 0, (330)

which coincides with the dispersion relation (67).
Hence, on using (329), the representation for the bending edge wave
through a single plane harmonic function is now established in the form

ν γ2
j
W ðx,y,t Þ ¼ Wj x, γ j y, t Wj ðx, γ n y, tÞ, 1 j 6¼ n 2, (331)
ν γ 2n
which is similar to (40) for the Rayleigh wave. On the edge η ¼ 0 the last
formula reduces to
γ 2j γ 2n
W ðx,0,t Þ ¼ Wj ðx, 0,tÞ, 1 j 6¼ n 2: (332)
ν γ 2n
As an example, we construct a nontime-harmonic eigensolution for the
bending edge wave mirroring that considered in Section 2.2. The sought for
functions Wj, j ¼ 1, 2 should satisfy the ansatz (323) along with the elliptic
equations
Wj, ηη + γ 2j Wj, ξξ ¼ 0: (333)
Let us specify the initial conditions

Wj τ¼0 ¼ Aj ðξ, γ j ηÞ, Wj, τ τ¼0 ¼ Bj ðξ, γ j ηÞ, (334)
where Aj and Bj are plane harmonic functions. Then, (331) implies
ν γ 2j
W jτ¼0 ¼ Aj ðξ, γ j ηÞ Aj ðξ, γ n ηÞ,
ν γ 2n
(335)
ν γ 2j
W, τ jτ¼0 ¼ Bj ðξ, γ j ηÞ Bj ðξ, γ n ηÞ,
ν γ 2n
The Fourier transforms of the functions Wj are given by
WjF ¼ wj ðk, τÞeγj jkjη , (336)
with the initial conditions for the quantities wj

wj τ¼0 ¼ AFj ðk,0Þ, wj, τ τ¼0 ¼ BFj ðk, 0Þ, (337)
where k is the Fourier transform parameter. Hence, the functions Wj satis-

fying (323) are
Z∞ " #
1 BFj ðk, 0Þ 2 2 2 2 γ jkjη + iξk
Wj ¼ sin γ e k τ + Aj ðk, 0Þ cos γ e k τ e j
F
dk,
2π γ 2e k2
∞
(338)
with the edge deflection following from (331).
Consider, for example, the functions Aj(ξ, γ jη) and Bj(ξ, γ jη) specified as
γ j ðη + aÞ
Aj ðξ, γ j ηÞ ¼
, Bj ðξ, γ j ηÞ ¼ 0, (339)
π γ 2j ðη + aÞ2 + ξ2
where a is a positive constant corresponding to a distributed delta-like initial

profile at the edge η ¼ 0; at the limit a ! 0 we get Wj(ξ, 0, 0) ¼ δ(ξ). On
inserting the Fourier transforms of (339) into (338), we arrive at
Z
+∞
1X 2
W ðξ, η, τÞ ¼ cos γ 2e k2 τ cos ðξkÞeγ j kðη + aÞ dk, (340)
π j¼1
0
decaying away from the edge η ¼ 0. The integrals in (340) may be evaluated
with the help of the formula, e.g., see Prudnikov et al. (1988)
Z
+∞ rffiffiffi !
1 π q
ep kqk dk ¼
2
Iðp, qÞ ¼ f pffiffiffi , (341)
2 p 2 p
0
2
where f ðzÞ ¼ ez erfcðzÞ, Re( p) ¼ 0, Im( p)6¼0, Re(q) 0, with
Z
+∞
2
et dt
2
erfcðzÞ ¼ pffiffiffi (342)
π
z
denoting the complementary error function, see Abramowitz and Stegun

(2012).
pffiffiffi
As a result, (340) is expressed in terms of the quantities ξ1 ¼ ξ= τ and
pffiffiffi
η1 ¼ ðη + aÞ= τ as

W ðξ, η, τÞ ¼ τ0:5 W ðξ1 , η1 Þ, (343)
where
" #
1 X 2 X2
W ðξ1 , η1 Þ ¼ pffiffiffi Re eiπ=4 f ðζ jm Þ , (344)
4γ e π j¼1 m¼1
eiπ=4 γ j η1 + ð1Þm iξ1

with ζjm ¼ :
2γ e
A numerical illustration of (344) is given in Fig. 42 for the Poisson’s ratio
ν ¼ 1/3. As might be expected, the amplitudes of the propagating distur-
bances decay away from the edge η1 ¼ 0.
7.2 Bending Edge Wave Induced by Prescribed Moments

and Shear Forces
Once the eigensolution is established, we proceed with the development of
an explicit model for the bending edge wave, aiming at extracting its con-
tribution to the overall dynamic response in a manner similar to that pres-
ented in Section 4 for the Rayleigh wave.
h1 = 1
0.5 h1 = 10
h1 = 50
0.4
0.3
0.2
W
0.1
-0.1
-0.2
-10 -5 0 5 10
x1

Fig. 42 The scaled deflection W vs ξ1 at ν ¼ 1/3.
Throughout this section we study bending edge waves induced by pre-

scribed bending moment M0 and shear force N0. The boundary conditions
along the edge x2 ¼ 0 are therefore given by
M0
W, 22 + νW, 11 ¼ ,
D (345)
N0
W, 222 + ð2 νÞW, 112 ¼ :
D
As before, we start from a multiple scale procedure, adapting it for plate
bending. Accordingly, the slow time τs ¼ ετ is introduced along with the fast
time τf ¼ τ, where ε ≪ 1 is a small parameter, characteristic of a near-
resonant edge excitation, see Section 7.4.2 and also Kaplunov et al. (2016).
First, we rewrite (322) in the form
Δ2 W + W, τf τf + 2εW, τf τs + ε2 W, τs τs ¼ 0: (346)
Then, the deflection W can be expanded into asymptotic series as
h2 P ð0Þ ð1Þ

W¼ W + εW + ⋯ : (347)
εD
Here and below in this section
P ¼ max ½M0 ðx, tÞ,hN0 ðx, tÞ: (348)
x, t
On substituting (347) into (346), we have at leading order
Δ2 W ð0Þ + W,ð0Þ
τf τf ¼ 0: (349)
In view of the ansatz (323), the last equation is transformed to the elliptic
equation
ð0Þ ð0Þ
W,ð0Þ
ηηηη + 2W, ξξηη + 1 γ e W, ξξξξ ¼ 0,
4
(350)
leading to
X
2
ð0Þ
W ð0Þ ¼ Wj ξ, γ j η, τf ,τs , (351)
j¼1
with the scaling factors γ j, j ¼ 1, 2, defined by (327).

At next order, we obtain from (346)
Δ2 W ð1Þ + W,ð1Þ ð0Þ

τf τf + 2W, τf τs ¼ 0: (352)
In view of the assumption (323) along with the superposition principle,

Eq. (352) may be rewritten as
ð1Þ
Δ1 Δ2 Wj ¼ 2W,ð0Þ
τf τs j ¼ 1,2, (353)
ð1Þ ð1Þ
with W ð1Þ ¼ W1 + W2 .
ð1Þ
Let us first study W1 . On employing the basic properties of harmonic
functions, we have
ð0Þ ð0Þ ð0Þ
Δ2 W1 ¼ ðγ 22 γ 21 ÞW1, ξξ ¼ 2γ 2e W1, ξξ : (354)
Then, due to (353) and (354), we infer

ð1Þ 1 ð0Þ
Δ1 Δ2 W1, ξξ ¼ Δ2 W1, τf τs , (355)
γe
2
from which
ð1Þ 1 ð0Þ
Δ1 W1, ξξη ¼ W : (356)
γ 2e 1, τf τs η
The solution of (355) is found in the form
ð1Þ ð1, 0Þ η ð0Þ
W1, ξξη ¼ Φ1, ξξη W , (357)
2γ 2e 1, τf τs

where Φ1 ¼ Φ1 ξ, γ 1 η, τf , τs is an arbitrary plane harmonic function in the
first two arguments.
ð1Þ
Similar derivations for W2 yield
ð1Þ ð1, 0Þ η ð0Þ
W2, ξξη ¼ Φ2, ξξη + 2
W2, τf τs , (358)
2γ e

where Φ2 ¼ Φ2 ξ, λ2 η, τf ,τs is also a plane harmonic function.
Thus, we obtain the following two-term asymptotic expansion
h2 P h 1 ð0Þ ð0Þ

ð1, 0Þ ð1, 0Þ
W, ξξη ¼ ε W1, ξξη + W2, ξξη + Φ1, ξξη + Φ2, ξξη
D
(359)
η ð0Þ ð0Þ

2 W1, τf τs W2, τf τs + ⋯ :
2γ e
Now, we proceed further with analysis of the nonhomogeneous edge
boundary conditions (345), beginning with a prescribed bending moment
(M06¼0, N0 ¼ 0). On rewriting (9) in terms of the dimensionless variables ξ

and η, we get
h2
W, ηη + νW, ξξ ¼ M0 ,
D (360)
W, ηηη + ð2 νÞW, ξξη ¼ 0:
Then, we insert the asymptotic expansion (359) into (360), having at

leading order
ð0Þ ð0Þ
ν γ 21 W1, ξξ + ν γ 22 W2, ξξ ¼ 0,
ð0Þ ð0Þ (361)
γ 1 γ 21 2 + ν W1, ξξξ + γ 2 γ 22 2 + ν W1, ξξξ ¼ 0:
It can be easily verified that these equations imply the dispersion relation
(330).
At next order, after straightforward transformations involving (355), the
boundary conditions (360) yield

ν ð1Þ ν ð1Þ ν ð0Þ ν ð0Þ 1
1 2 W1, ξξηη + 1 2 W2, ξξηη 2 2 W1, τf τs + 2 2 W2, τf τs ¼ M0, ξξ ,
γ1 γ2 γe γ1 γe γ2 P
ð1Þ ð1Þ ð1Þ ð1Þ

W1, ξξηηη + W2, ξξηηη + ð2 νÞW1, ξξξξη + ð2 νÞW2, ξξξξη ¼ 0:
(362)
On substituting (359) into (362), using the Cauchy–Riemann identities,

taking the harmonic conjugate of the second equation, and integrating with
respect to ξ, the analyzed boundary conditions may be rearranged to

ð1, 0Þ ð1, 0Þ 1 ν ð0Þ
ν γ 21 Φ1, ξξξξ + ν γ 22 Φ2, ξξξξ ¼ 2 1 + 2 W1, τf τs
2γ e γ1

1 ν ð0Þ 1
2 1 + 2 W2, τf τs M0, ξξ , (363)
2γ e γ2 P
ð1, 0Þ ð1, 0Þ γ ð0Þ γ ð0Þ
γ 1 ν γ 22 Φ1, ξξξξ + γ 2 ν γ 21 Φ2, ξξξξ ¼ 22 W2, τf τs 12 W1, τf τs :
γe γe
In contrast to (361), Eq. (363) is nonhomogeneous, with the determinant

vanishing due to (330). Then, the compatibility condition necessitates

γ 2 ðν2 γ 41 Þ ð0Þ ðν γ 21 Þðν + γ 22 Þ
γ 1 ðν γ 2 Þ
2
W1, τf τs + γ 2 ðν γ 2 Þ +
2
2γ 21 2γ 2
ð0Þ γ 2e γ 2 ðν γ 21 Þ

W2, τf τs ¼ M0, ξξ :
P
(364)
ð0Þ ð0Þ
In view of (331), it is possible to express W1 and W2 through W(0) on the
edge η ¼ 0, leading to
BK
W,ð0Þ
τ f τs ¼ M0, ξξ , (365)
2P
where
4γ 4e γ 2 ðν γ 21 Þ
BK ¼ , (366)
a1K ðν γ 21 Þ + a2K ðν γ 22 Þ
with
ðν + γ 22 Þðν γ 21 Þ γ 2 ðν2 γ 41 Þ
a1K ¼ + γ 2 ðν γ 22 Þ, a2K ¼ + γ 1 ðν γ 22 Þ:
2γ 2 2γ 21
After some rather tedious but straightforward algebra, (366) can be
reduced to
pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffi
1 γ 4e ν + 1 γ 4e
BK ¼ pffiffiffiffiffiffiffiffiffiffiffiffi , (367)
1 ν + 1 γ 4e
depending on the Poisson’s ratio only. Fig. 43 reveals a monotonic increase
of BK in ν.
Finally, on employing the leading order approximation
h2 P ð0Þ
W¼ W , (368)
εD
we present (365) as
3BK ð1 ν2 Þ
2εW, τf τs ¼ M0, ξξ : (369)
2Eh
Then, due to the ansatz (323), this equation becomes
3BK ð1 ν2 Þ
γ 4e W, ξξξξ + W, τf τf + 2εW, τf τs ¼ M0, ξξ , (370)
2Eh
0.2
0.15
BK
0.1
0.05
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
u
Fig. 43 The coefficient BK vs ν.
or, in the original variables to within the error O(ε2)
Dγ 4e W, 1111 + 2ρhW, tt ¼ BK M0, 11 : (371)
In this case the behavior over the interior is described by the elliptic
equations
Wj, 22 + γ 2j Wj, 11 ¼ 0, (372)
which should be solved in conjunction with the deduced parabolic equation

(371) and the relation (331). Thus, the solution of the dynamic equation
(371) is used together with (332) as a boundary condition for the
pseudo-static equations (372). The obtained plane harmonic function is then
substituted into (331) in order to restore the deflection of the plate.
Similarly to the Rayleigh wave, it may be shown that the derived para-
bolic–elliptic formulation provides a correct evaluation of the edge wave
contribution to the overall response arising from an arbitrary edge moment
M0. This is not surprising, since the developed procedure is, in fact, aimed at
accounting for edge wave poles, see Section 7.4.1.
A parabolic–elliptic formulation may also be established for a shear force
excitation, having N06¼0, M0 ¼ 0. However, instead of the deflection W, the
parabolic equation is now obtained in respect of the rotation angle θK ¼ W, 2

evaluated at the edge x2 ¼ 0. It is
Dγ 4e θK , 1111 + 2ρhθK , tt ¼ BK N0, 11 , (373)
with the constant BK defined by (367). In this case the parabolic–elliptic
model also contains the elliptic equations
θKj, 22 + γ 2j θKj, 11 ¼ 0, (374)
which are to be solved together with (373) and (331).
7.3 Plate on Elastic Foundation

Bending edge waves in plates on elastic foundations have recently been
investigated in Kaplunov et al. (2014), Kaplunov and Nobili (2015), and
Kaplunov et al. (2016). In this section we briefly address several peculiarities
of the bending edge wave on a thin Kirchhoff plate resting on a Winkler
foundation. In this case an additional term arises in the governing equation
(61), namely
DΔ2 W + 2ρhW, tt + βW W ¼ 0, (375)
where βW is the Winkler foundation modulus. The related dispersion rela-
tion is
Dk4 γ 4e + βW ¼ 2ρhω2 : (376)
The latter may be simplified to
K 4 ¼ Ω2 1, (377)
where
K ¼ kγ e ðD=βW Þ1=4 , Ω ¼ ωð2ρh=βW Þ1=2 : (378)

The presence of a Winkler foundation leads to the cut-off frequency
Ω0 ¼ 1, see Fig. 44. Another remarkable feature caused by a foundation
is the local minimum of the phase velocity
Ω Ω
V ph ¼ ¼ 1=4 (379)
K Ω 1
2
pffiffiffi pffiffiffi
V ph ¼ 2, occurring at Ω ¼ 2 or K ¼ 1, see Fig. 45. Moreover, at this
point V ph coincides with the group velocity
0.8
0.6
K
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
W0 W
Fig. 44 The dispersion curve (377).
1.6
V ph
1.4
1.2
V
Vg
0.8
1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8

W
Fig. 45 The phase and group velocities according to (379) and (380).
dΩ 2 2 3=4
Vg ¼ ¼ Ω 1 : (380)
dK Ω
It is also worth mentioning that the minimal value of the phase velocity
pffiffiffi
V ph ¼ V g ¼ 2, similar to that shown in Fig. 45, corresponds to the critical
speed of a moving load in the steady-state problem for a beam supported by a

Winkler foundation, e.g., see the well-known contribution of Timoshenko
(1926). Therefore, the same resonant effect may be expected for a moving
load on the edge of an elastically supported plate.
Finally, we present the parabolic–elliptic formulation for a plate on a
Winkler foundation, which may be derived in the same manner as in
Section 7.2, see also Kaplunov et al. (2016). The ansatz generalizing (323)
is now written as
βW h4
γ 4e W, ξξξξ + W, ττ + W ¼ 0: (381)
D
As a result, we arrive at the pseudo-static elliptic equation (372), subject to a
boundary condition on the edge x2 ¼ 0 given by the dynamic parabolic
equation
Dγ 4e W, 1111 + 2ρhW, tt + βW W ¼ BK M0, 11 , (382)
or the similar Eq. (374) with the boundary condition
Dγ 4e θK , 1111 + 2ρhθK , tt + βW θK ¼ BK N0, 11 : (383)
7.4 Examples
Consider now examples demonstrating the implementation of the formula-
tion developed in Section 7.2.
7.4.1 Comparison With Exact Solution

Let us apply the Laplace transform in time τ and the Fourier transform in the
longitudinal coordinate ξ in (322). Then, the transformed deflection W FL
satisfies the equation
4 2 FL
ηηηη 2k W, ηη + k + p W
W,FL ¼ 0,
2 FL
(384)
where k and p, as above, denote the parameters of the Fourier and Laplace
transforms, respectively. The solution of this equation is
W FL ¼ C1 eg1 η + C2 eg2 η , (385)
where Cn(k, p), n ¼ 1, 2, are arbitrary functions and
g1, 2 ¼ k2 ip: (386)
Consider an arbitrary bending moment M0 applied at the edge η ¼ 0.
In this case the boundary conditions (360) are transformed to
h2 FL
ηη νk W
W,FL ¼
2 FL
M ,
D 0 (387)
ηηη ð2 νÞk W, η ¼ 0,
2 FL
W,FL
where M0FL is the transformed moment M0. On substituting (385) into

(387), we determine Cn, having
1
W FL ¼ ðG1 eg1 η + G2 eg2 η Þ, (388)
G
where
h i h2 M FL
Gj ¼ ð1Þj gj gj2 ð2 νÞk2 0
, j ¼ 1,2
D (389)

G ¼ ðg1 g2 Þ g12 g22 + 2g1 g2 ð1 νÞk2 ν2 k4 :
Next, we note that the term in square brackets in the expression for G may
be rewritten as
pffiffiffiffiffiffiffiffiffiffiffiffiffi
k4 + p2 + 2ð1 νÞk2 k4 + p2 ν2 k4 : (390)
The poles
p2 ¼ γ 4e k4 , (391)
correspond to those of the dispersion relation (330). It is worth mentioning

that at the poles (391) Eq. (384) is identical to the transformed elliptic equa-
tion (325) (or (372)), while (388) is related to the parabolic equation (371) at
the edge η ¼ 0. In particular, on taking into account (389), we deduce
from (388)
pffiffiffiffiffiffiffiffiffiffiffiffiffi
h2 M0FL k4 + p2 + νk2
W FL jη¼0 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi
: (392)
D k + p + 2ð1 νÞk k + p ν k
4 2 2 4 2 2 4
Near the poles (391) the latter becomes

pffiffiffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffi
h2 M0FL p2 1 γ 4e 1 γ 4e + ν
W jη¼0 pffiffiffiffiffiffiffiffiffiffiffiffi
FL
, (393)
D 1 γ 4e + 1 ν γ 4e k4 + p2
which is indeed the transformed solution of Eq. (371) expressed in the vari-
ables ξ and τ. Thus, the formulation (371) and (372) captures the
contribution of the poles of the bending edge wave induced by a moment of

a general shape.
7.4.2 Near-Resonant Excitation

Let a prescribed bending edge moment be
M0 ¼ AeiðkxωtÞ , (394)
with no shear edge force assumed. Then, the solution of the plate bending
equation (61), subject to the nonhomogeneous boundary conditions (360),
can be written as
W ðx,y,tÞ ¼ V ðyÞeiðkxωtÞ : (395)

As a result, (61) becomes

d 4V 2
2d V 2ρhω2
2k + k
4
V ¼ 0, (396)
dy4 dy2 D
leading to
X
2
W ðx, y, tÞ ¼ Cn eiðkxωtÞkχ n y , (397)
n¼1
with
2ρhω2
χ 21 + χ 21 ¼ 2, χ 21 χ 22 ¼ 1 : (398)
Dk4
It is easily verified that the attenuation orders χ n, n ¼ 1, 2, coincide with
γ n defined by (327), provided that the frequency ω and the wave number k
satisfy the dispersion relation (67). The constants Cn may be determined
from the boundary conditions (360). The exact solution at the edge is then
given by
A χ1χ2 + ν
W ðx,0, tÞ ¼ eiðkxωtÞ : (399)
Dk χ 1 χ 2 + 2ð1 νÞχ 1 χ 2 ν2
2 2 2
Compare this formula with that obtained from the approximate formu-
lation derived in Section 7.2 and given by the solution of (371). The latter is
ABK k2
W ðx, 0,tÞ ¼ eiðkxωtÞ , (400)
Dk4 γ 4e 2ρhω2
with BK defined by (367). It may be observed that both (399) and (400)
exhibit a resonant behavior when the frequency ω and the wave number
k satisfy the dispersion relation (67).
Let the excitation frequency be close to that of the bending edge
wave, i.e.,
ω ¼ ω0 + εω1 , jεj≪1, (401)
sffiffiffiffiffiffiffiffiffiffiffiffi
Dγ 2e k4
where ω0 ¼ , see (67). This is exactly the setup of bending edge
2ρh
motion evolving in the slow time τs εt, which is in line with the asymp-
totic theory developed in Section 7.2.
First, we obtain
2ρh εω0 ω1
χ1χ 2 γ1γ2 : (402)
Dk4 γ 1 γ 2
On substituting the latter into (399) and making use of (330), we arrive at
A ðγ 1 γ 2 + νÞeiðkxωtÞ
W ðx, 0,tÞ
Dk2 4ρhεω0 ω1 1ν
½γ 1 γ 2 + 2ð1 νÞγ 1 γ 2 ν
2 2 2 1+
Dk4 γ1γ2
ABK k2 eiðkxωtÞ
¼ :
4ρhεω0 ω1
(403)
This expression coincides with the leading order behavior of (400). Indeed,
on inserting (401), we have
ABK k2 eiðkxωtÞ ABK k2 eiðkxωtÞ

W ðx,0,tÞ ¼ ,
Dk4 γ 4e + β 2ρhω20 4ρhεω0 ω1 4ρhεω0 ω1
(404)
which coincides with (403).
8. CONCLUSION
Explicit hyperbolic–elliptic and parabolic–elliptic models drastically
simplify treatment and understanding of dynamic phenomena involving
Rayleigh and Rayleigh-type waves. In particular, they enable a straightfor-
ward insight into various impact and moving load problems, including sur-
prisingly simple asymptotic formulae for the 3D displacement fields caused
by a point moving load. Another important advantage of these models is the

separation of the original exact formulations into two parts, namely hyper-
bolic or parabolic dynamic problems along the surface or edge together with
pseudo-static elliptic problems over the interior. The solutions of dynamic
problems provide boundary conditions for pseudo-static ones; in doing so,
sometimes it is enough to restrict ourselves to the dynamic problems only. In
addition, the possibility of dealing with elliptic equations over the interior
instead of hyperbolic ones appears to be very useful for optimizing numerical
computations.
The developed methodology allows a number of extensions. Among
them, we mention finite and curved bodies as well as half-spaces with a more
general vertical inhomogeneity. Modeling of the surface wave induced by
internal sources is also of obvious interest. Another important area is con-
cerned with interaction of surface and bulk waves. This may occur when
an elastic structure, e.g., a rod, is attached to the surface or in the case of
a rigid stamp. Also, the well-known phenomenon of the conversion of plate
bending and extensional waves into surface ones at a short-wavelength limit
may open new prospects for deriving refined plate theories. Analysis of some
of the aforementioned problems is already in progress.
ACKNOWLEDGMENTS
The material in this chapter originates from the publications: Kaplunov and Kossovich (2004),
Kaplunov, Kossovich, and Zakharov (2004), Kaplunov et al. (2006), Dai et al. (2010),
Kaplunov et al. (2010), Erbaş et al. (2013), Kaplunov and Prikazchikov (2013), Kaplunov,
Prikazchikov, Erbaş, and Şahin (2013), Prikazchikov (2013), Erbaş et al. (2017), Kaplunov
et al. (2014), Ege et al. (2015), Kaplunov and Nobili (2015), and Kaplunov et al. (2016).
We express our sincere gratitude to all the coauthors in the publications mentioned above,
and also acknowledge fruitful discussions with P. Chadwick, Y. Fu, A.P. Kiselev,
D.F. Parker, and A. Pichugin. We also thank P. Wootton for reading the final version of
the manuscript and making several valuable comments.
APPENDIX A. NEAR-RESONANT BEHAVIOR OF A

SINGLE DEGREE OF FREEDOM LINEAR
OSCILLATOR
Consider forced vibrations of a single degree of freedom linear oscil-
lator governed by
d 2x
+ ω20 x ¼ FðτÞeiω0 t , (A.1)
dt 2
where ω0 is the natural frequency, and τ ¼ εt is the slow time. Here the
parameter ε ≪ 1 is associated with a small deviation of the excitation
frequency from the natural one. For example, the forcing FðτÞ ¼ eiω1 τ pre-
scribes the excitation frequency of ω ¼ ω0 + εω1.
Let us now develop a multiple scale perturbation procedure, see, e.g.,
Kevorkian and Cole (2013) and Nayfeh (2008), expanding
xðt, τÞ ¼ ε1 ðx0 ðt,τÞ + εx1 ðt,τÞ + ⋯ Þ, (A.2)

and making use of the symbolic identity
d2 @2 @2 2@
2
¼ + 2ε + ε : (A.3)
dt 2 @t 2 @t@τ @τ2
Then, on substituting (A.2) into (A.1), we have at leading order
x0, tt + ω20 x0 ¼ 0: (A.4)

As before, the comma in the subscript denotes a partial derivative. Therefore
x0 ¼ AðτÞeiω0 t : (A.5)
At next order we obtain
x1, tt + ω20 x1 + 2x0, tτ ¼ FðτÞeiω0 t , (A.5)
or
x1, tt + ω20 x1 ¼ ðFðτÞ + 2iA0 ðτÞÞeiω0 t : (A.6)

In order to exclude a secular term, we impose
FðτÞ + 2iA0 ðτÞ ¼ 0: (A.7)
Then the sought for leading order solution becomes
Z
ieiω0 t
x0 ðt, τÞ ¼ FðτÞdτ: (A.8)
2
APPENDIX B. EXACT ANALYSIS OF PLANE MOVING

LOAD PROBLEMS
In this appendix we discuss steady-state problems for a vertical force or
stamp moving along the surface of an elastic half-plane.
First, we study dynamic response of the elastic half-plane H2+ , see (25),
subject to a vertical distributed force P, moving steadily at a constant speed c.
The boundary conditions at x3 ¼ 0 are specified as
σ 13 ¼ 0, σ 33 ¼ PðsÞ, (B.1)
where, as previously, s is a moving coordinate, i.e., s ¼ x1 ct. The
governing equations for the elastic potentials written in the moving coordi-
nate frame (s, x3), become

c2
ϕ, 33 + 1 2 ϕ, ss ¼ 0,
c1
(B.2)
c2
ψ , 33 + 1 2 ψ , ss ¼ 0:
c2
Then, the boundary conditions (B.1) take the form
2ϕ, s3 + ψ , ss ψ , 33 ¼ 0,
PðsÞ (B.3)
ðκ2 2Þϕ, ss + κ 2 ϕ, 33 + 2ψ , s3 ¼ :
μ
Let us focus on the subsonic regime (c < c2 < c1), when Eq. (B.2) are
elliptic. In this case the solution is expressed through plane harmonic func-
tions in line with the consideration in Section 2.2. Thus,
ϕ ¼ ϕðs, αx3 Þ, ψ ¼ ψðs, βx3 Þ, (B.4)
with α and β defined by (14). The only difference of solution (B.4) from (31)
is that, instead of the phase speed, c now denotes the speed of a moving load.
Not surprisingly, the elastic potentials ϕ and ψ may be related to each
other using the first homogeneous boundary condition (B.3) along with
the Cauchy-Riemann identities, resulting in
2α
ψ¼ ϕ*: (B.5)
β +1
2
Hence, the displacement components are expressed in terms of a single plane

harmonic function, similarly to Section 2.2, as
2αβ
u1 ðs, x3 Þ ¼ ϕ, s ðs,αx3 Þ ϕ, ξ ðξ, βx3 Þ,
1 + β2
(B.6)
2
u3 ðs, x3 Þ ¼ ϕ, 3 ðs, αx3 Þ ϕ, 3 ðs,βx3 Þ:
1 + β2
Also, using (B.5), the second boundary condition (B.3) becomes
ð1 + β2 Þ PðsÞ
ϕ, ss jx3 ¼0 ¼ , (B.7)
μRðcÞ
where R(c) is defined by (146).

dp
Let us assume that PðsÞ ¼ P0 . Then, on employing the Poisson for-
ds
mula, see, e.g., Courant and Hilbert (1989), the derivative ϕ, s is given by
Z∞
ð1 + β2 Þ P0 αx3 pðrÞ
ϕ, s ðs,αx3 Þ ¼ dr, (B.8)
πμRðcÞ ðr sÞ2 + α2 x23
∞
enabling a straightforward calculation of the displacement field through

(B.6). For example, for the point load P(s) ¼ P0δ(s), e.g., see the well-known
paper of Cole and Huth (1958), integration in (B.7) gives

ð1 + β2 Þ P0 1
ϕ, s jx3 ¼0 ¼ HðsÞ : (B.9)
μRðcÞ 2
This solution is determined to within an arbitrary constant, which cannot be
1
found from the steady-state formulation. The value in (B.9) is chosen in
2
order to have symmetry.
On satisfying the boundary condition (B.9), the harmonic function
ϕ,s(s, αx3) is obtained as
ð1 + β2 Þ P0 s
ϕ, s ðs, αx3 Þ ¼ tan 1 : (B.10)
πμRðcÞ αx3
Finally, the sought for displacements are given by

ð1 + β2 Þ P0 1 s 2αβ 1 s
u1 ðs,x3 Þ ¼ tan tan ,
πμRðcÞ αx3 1 + β2 βx3
(B.11)
αð1 + β2 Þ P0 2 2 2
u3 ðs,x3 Þ ¼ ln s + α x3
2 2
ln s + β x3 :
2 2
2πμRðcÞ 1 + β2
This result is identical to that presented in the aforementioned paper by Cole
and Huth (1958), up to a rigid body motion component of the horizontal
displacement.
The same methodology may be applied to mixed boundary value prob-
lems. In particular, consider the steady-state problem for a rigid stamp mov-
ing along the boundary of the elastic half-plane H2+ at a constant speed c < c2.
The equations of motion in the moving coordinate frame (s, x3) are again
taken in the form (B.2), whereas now the boundary conditions along the
surface x3 ¼ 0 become
σ 33 ¼ 0, s 2 S1 ;
u3 ¼ f ðsÞ, s 2 S2 ; (B.12)
σ s3 ¼ 0, s 2 ,
where S1 [ S2 ¼ . Due to the last condition, the relation (B.5) holds true.
On introducing the auxiliary function
β2 1
ϕ1 ¼ ϕ, 3 (B.13)
β2 + 1
and also the scaling z ¼ αx3, it is possible to reduce the first equation of (B.2)
with the boundary condition (B.12) to a conventional mixed boundary
value problem for the Laplace equation. Thus, we have
ϕ1, zz + ϕ1, ss ¼ 0 (B.14)
subject to
ϕ1 ¼ f ðsÞ, ξ 2 S2 ;
(B.15)
ϕ1, s ¼ 0, ξ 2 S1 ,
along the surface z ¼ 0.
REFERENCES
Abramowitz, M., & Stegun, I. A. (2012). Handbook of mathematical functions: With formulas,
graphs, and mathematical tables. New York: Dover Publications.
Achenbach, J. (1998). Explicit solutions for carrier waves supporting surface waves and plate
waves. Wave Motion, 28(1), 89–97.
Achenbach, J. (2012). Wave propagation in elastic solids. Amsterdam: Elsevier.
Agarwal, V. (1978). On surface waves in generalized thermoelasticity. Journal of Elasticity,
8(2), 171–177.
Alenitsyn, A. (1963). Rayleigh waves in a nonhomogeneous elastic half-space. Journal of
Applied Mathematics and Mechanics, 27(3), 816–822.
Babich, V., & Kiselev, A. (2014). Elastic waves: High-frequency theory. St-Petersburg: BKhV-
Peterburg.
Barnett, D., & Lothe, J. (1974). Consideration of the existence of surface wave (Rayleigh
wave) solutions in anisotropic elastic crystals. Journal of Physics F: Metal Physics, 4(5),
671–686.
Barnett, D., Lothe, J., Gavazza, S., & Musgrave, M. (1985). Considerations of the existence
of interfacial (Stoneley) waves in bonded anisotropic elastic half-spaces. Proceedings of the
Royal Society of London A: Mathematical, Physical and Engineering Sciences, 402(1822),
153–166.
Bleustein, J. L. (1968). A new surface wave in piezoelectric materials. Applied Physics Letters,
13(12), 412–413.
Borovikov, V. V. A. (1994). In Uniform stationary phase method (Vol. 40). London: Inst of
Engineering & Technology.
Brûle, S., Javelaud, E., Enoch, S., & Guenneau, S. (2014). Experiments on seismic
metamaterials: Molding surface waves. Physical Review Letters, 112(13), 133901.
Campbell, C. (1998). Surface acoustic wave devices for mobile and wireless communications. San Diego:
Academic Press.
Cao, Y., Xia, H., & Li, Z. (2012). A semi-analytical/FEM model for predicting ground
vibrations induced by high-speed train through continuous girder bridge. Journal of
Mechanical Science and Technology, 26(8), 2485–2496.
Chadwick, P. (1976a). The existence of pure surface modes in elastic materials with ortho-
rhombic symmetry. Journal of Sound and Vibration, 47(1), 39–52.
Chadwick, P. (1976b). Surface and interfacial waves of arbitrary form in isotropic elastic
media. Journal of Elasticity, 6(1), 73–80.
Cherednichenko, K. (2007). An asymptotic expansion of the boundary-layer type for flexural
waves along the curved edge of a Kirchhoff-Love elastic plate. Journal of Mathematical
Sciences, 142(6), 2682–2688.
Chouet, B. (1985). Excitation of a buried magmatic pipe: A seismic source model for volcanic
tremor. Journal of Geophysical Research: Solid Earth, 90(B2), 1881–1893.
Cole, J. D., & Huth, J. (1958). Stresses produced in a half plane by moving loads. Journal of
Applied Mechanics, 25(4), 433–436.
Colombi, A., Colquitt, D., Roux, P., Guenneau, S., & Craster, R. V. (2016). A seismic
metamaterial: The resonant metawedge. Scientific Reports, 6, 27717.
Courant, R., & Hilbert, D. (1989). In Methods of mathematical physics (Vol. 2). New York:
John Wiley & Sons.
Dai, H.-H., Kaplunov, J., & Prikazchikov, D. (2010). A long-wave model for the surface
elastic wave in a coated half-space. Proceedings of the Royal Society of London A: Mathemat-
ical, Physical and Engineering Sciences, 466(2122), 3097–3116.
de Hoop, A. T. (2002). The moving-load problem in soil dynamics—The vertical displace-
ment approximation. Wave Motion, 36(4), 335–346.
Destrade, M. (2004). Surface acoustic waves in rotating orthorhombic crystals. Proceedings of
the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 460(2042),
653–665.
Destrade, M. (2007). Seismic Rayleigh waves on an exponentially graded, orthotropic half-
space. Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering
Sciences, 463(2078), 495–502.
Destrade, M., & Fu, Y. (2006). The speed of interfacial waves polarized in a symmetry plane.
International Journal of Engineering Science, 44(1), 26–36.
Destrade, M., Fu, Y., & Nobili, A. (2016). Edge wrinkling in elastically supported pre-
stressed incompressible isotropic plates. Proceedings of the Royal Society of London A: Math-
ematical, Physical and Engineering Sciences, 472(2193), 20160410.
Dieterman, H., & Metrikine, A. (1996). The equivalent stiffness of a half-space interacting
with a beam. Critical velocities of a moving load along the beam. European Journal of
Mechanics Series A Solids, 15, 67–90.
Dockrey, J., Lockyear, M. J., Berry, S., Horsley, S., Sambles, J. R., & Hibbins, A. P. (2013).
Thin metamaterial Luneburg lens for surface waves. Physical Review B, 87(12), 125137.
Dowaikh, M., & Ogden, R. (1990). On surface waves and deformations in a pre-stressed
incompressible elastic solid. IMA Journal of Applied Mathematics, 44(3), 261–284.
Dowaikh, M., & Ogden, R. (1991). Interfacial waves and deformations in pre-stressed elastic
media. Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering
Sciences, 433(1888), 313–328.
Ege, N., Erbaş, B., & Prikazchikov, D. A. (2015). On the 3D Rayleigh wave field on an
elastic half-space subject to tangential surface loads. ZAMM-Journal of Applied Mathematics
and Mechanics/Zeitschrift f€ur Angewandte Mathematik und Mechanik, 95(12), 1558–1565.
Erbaş, B., Kaplunov, J., & Prikazchikov, D. A. (2013). The Rayleigh wave field in mixed
problems for a half-plane. IMA Journal of Applied Mathematics, 78, 1078–1086.
Erbaş, B., Kaplunov, J., Prikazchikov, D. A., & Şahin, O. (2017). The near-resonant regimes
of a moving load in a three-dimensional problem for a coated elastic half-space. Math-
ematics and Mechanics of Solids, 22, 89–100.
Erbaş, B., & Şahin, O. (2016). On the causality of the Rayleigh wave. Journal of Mechanics of
Materials and Structures, 11(4), 449–461.
Erdelyi, A., Magnus, W., Oberhettinger, F., & Tricomi, F. (1954). In Table of integral trans-
forms (Vol. 2). New York: McGraw-Hill.
Eshelby, J. (1949). Uniformly moving dislocations. Proceedings of the Physical Society Section A,
62(5), 307–314.
Freund, L. B. (1998). Dynamic fracture mechanics. Cambridge: Cambridge University Press.
Friedlander, F. (1948). On the total reflection of plane waves. The Quarterly Journal of Mechan-
ics and Applied Mathematics, 1(1), 376–384.
Fu, Y. (2003). Existence and uniqueness of edge waves in a generally anisotropic elastic plate.
Quarterly Journal of Mechanics and Applied Mathematics, 56, 605–616.
Fu, Y. B., & Mielke, A. (2002). A new identity for the surface–impedance matrix and its
application to the determination of surface-wave speeds. Proceedings of the Royal Society
of London A: Mathematical, Physical and Engineering Sciences, 458(2026), 2523–2543.
Georgiadis, H., & Lykotrafitis, G. (2001). A method based on the Radon transform for three-
dimensional elastodynamic problems of moving loads. Journal of Elasticity and the Physical
Science of Solids, 65(1–3), 87–129.
Goda, M. A. (1992). The effect of inhomogeneity and anisotropy on Stoneley waves. Acta
Mechanica, 93(1), 89–98.
Gogoladze, V. (1948). Rayleigh waves on the interface between a compressible fluid medium and
a solid elastic half-space. Trudy Seismologicheskogo Instituta Academii Nauk SSSR, 127, 27–32.
Gol’dshtein, R. (1965). Rayleigh waves and resonance phenomena in elastic bodies. Journal of
Applied Mathematics and Mechanics, 29(3), 608–619.
Gudra, T., & Stawiski, B. (2000). Non-destructive strength characterization of concrete
using surface waves. Ndt & E International, 33(1), 1–6.
Gulyaev, Y. V. (1969). Electroacoustic surface waves in solids. ZhETF Pisma Redaktsiiu,
9, 63.
Hayes, M., & Rivlin, R. S. (1961). Surface waves in deformed elastic materials. Archive for
Rational Mechanics and Analysis, 8(1), 358–380.
Ikeda, T. (1990). Fundamentals of piezoelectricity. Oxford: Oxford University Press.
Ishlinsky, A. (1954). A particular limit transition in the theory of the stability of rectangular
elastic plates. Doklady Akademii Nauk SSSR, 95(3), 474–479.
Ivanov, T. P. (1988). On the propagation of thermoelastic Rayleigh waves. Wave Motion,
10(1), 73–82.
Kamotskii, I., & Kiselev, A. (2009). An energy approach to the proof of the existence of
Rayleigh waves in an anisotropic elastic half-space. Journal of Applied Mathematics and
Mechanics, 73(4), 464–470.
Kaplunov, J. (1986). Transient dynamics of an elastic half-plane subject to a moving load. Moscow:
Institute for Problems in Mechanics.
Kaplunov, J., Kossovich, L., & Zakharov, A. (2004). An explicit asymptotic model for the
Bleustein–Gulyaev wave. Comptes Rendus Mecanique, 332(7), 487–492.
Kaplunov, J., & Nobili, A. (2015). The edge waves on a Kirchhoff plate bilaterally supported
by a two-parameter elastic foundation. Journal of Vibration and Control. https://doi.org/
10.1177/1077546315606838.
Kaplunov, J., Nolde, E., & Prikazchikov, D. (2010). A revisit to the moving load problem
using an asymptotic model for the Rayleigh wave. Wave Motion, 47(7), 440–451.
Kaplunov, J., Nolde, E., & Rogerson, G. (2000). A low-frequency model for
dynamic motion in pre-stressed incompressible elastic structures. Proceedings of the
2589–2610.
Kaplunov, J., & Prikazchikov, D. (2013). Explicit models for surface, interfacial and edge
waves. In R. V. Craster & J. Kaplunov (Eds.), Dynamic localization phenomena in elasticity,
acoustics and electromagnetism (pp. 73–114). Wien: Springer.
Kaplunov, J., Prikazchikov, D., Erbaş, B., & Şahin, O. (2013). On a 3D moving load prob-
lem for an elastic half space. Wave Motion, 50(8), 1229–1238.
Kaplunov, J., Prikazchikov, D., & Rogerson, G. (2005). On three-dimensional edge waves
in semi-infinite isotropic plates subject to mixed face boundary conditions. The Journal of
the Acoustical Society of America, 118(5), 2975–2983.
Kaplunov, J., Prikazchikov, D., & Rogerson, G. (2016). Edge bending wave on a thin elastic
plate resting on a Winkler foundation. Proceedings of the Royal Society of London A: Math-
ematical, Physical and Engineering Sciences, 472(2190), 20160178.
Kaplunov, J., Prikazchikov, D. A., Rogerson, G. A., & Lashab, M. I. (2014). The edge wave
on an elastically supported Kirchhoff plate. The Journal of the Acoustical Society of America,
136(4), 1487–1490.
Kaplunov, J., Zakharov, A., & Prikazchikov, D. (2006). Explicit models for elastic and
piezoelastic surface waves. IMA Journal of Applied Mathematics, 71, 768–782.
Kaplunov, J. D., Kossovitch, L. Y., & Nolde, E. (1998). Dynamics of thin walled elastic bodies.
San Diego: Academic Press.
Kaplunov, Y. D., & Kossovich, L. Y. (2004). Asymptotic model of Rayleigh waves in the far-
field zone in an elastic half-plane. Doklady Physics, 49(4), 234–236.
Kevorkian, J., & Cole, J. D. (2013). In Perturbation methods in applied mathematics (Vol. 34).
Berlin: Springer Science & Business Media.
Kiselev, A., & Parker, D. (2010). Omni-directional Rayleigh, Stoneley and Sch€ olte waves
with general time dependence. Proceedings of the Royal Society of London A: Mathematical,
Physical and Engineering Sciences, 466, 2241–2258.
Kiselev, A. P. (2004). Rayleigh wave with a transverse structure. Proceedings of the Royal Society
of London A: Mathematical, Physical and Engineering Sciences, 460(2050), 3059–3064.
Kiselev, A. P. (2015). General surface waves in layered anisotropic elastic structures. Zapiski
Nauchnykh Seminarov POMI, 438, 133–137.
Kiselev, A. P., & Rogerson, G. A. (2009). Laterally dependent surface waves in an elastic
medium with a general depth dependence. Wave Motion, 46(8), 539–547.
Konenkov, Y. K. (1960). A Rayleigh-type flexural wave. Soviet Physics Acoustics, 6, 122–123.
Krushynska, A. (2011). Flexural edge waves in semi-infinite elastic plates. Journal of Sound and
Vibration, 330(9), 1964–1976.
Lamb, H. (1904). On the propagation of tremors over the surface of an elastic solid. Philo-
sophical Transactions of the Royal Society of London. Series A, Containing papers of a mathemat-
ical or physical character, 203, 1–42.
Lawrie, J. B., & Kaplunov, J. (2012). Edge waves and resonance on elastic structures: An
overview. Mathematics and Mechanics of Solids, 17(1), 4–16.
Madshus, C., & Kaynia, A. (2000). High-speed railway lines on soft ground: Dynamic behav-
iour at critical train speed. Journal of Sound and Vibration, 231(3), 689–701.
Malischewsky, P. G. (2000). Comment to “A new formula for the velocity of Rayleigh
waves” by D. Nkemzi [Wave Motion 26 (1997) 199–205]. Wave Motion, 31(1), 93–96.
Miklowitz, J. (1978). The theory of elastic waves and waveguides. Amsterdam: North Holland.
Morrissey, J. W., & Rice, J. R. (1998). Crack front waves. Journal of the Mechanics and Physics of
Solids, 46(3), 467–487.
Nayfeh, A. H. (2008). Perturbation methods. New York: John Wiley & Sons.
Norris, A. (1994). Flexural edge waves. Journal of Sound and Vibration, 171(4), 571–573.
Norris, A., Krylov, V., & Abrahams, I. (2000). Flexural edge waves and comments on “A new
bending wave solution for the classical plate equation”[J. Acoust. Soc. Am. 104,
2220–2222 (1998)]. The Journal of the Acoustical Society of America, 107(3), 1781–1784.
Norris, A. N., & Abrahams, I. D. (2007). A multiple-scales approach to crack-front waves.
Journal of Engineering Mathematics, 59(4), 399–417.
Parker, D. (2012). Evanescent Sch€ olte waves of arbitrary profile and direction. European Jour-
nal of Applied Mathematics, 23(02), 267–287.
Parker, D. (2013). The Stroh formalism for elastic surface waves of general profile. Proceedings
of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 469(2160),
20130301.
Parker, D., & Kiselev, A. (2009). Rayleigh waves having generalised lateral dependence. The
Quarterly Journal of Mechanics and Applied Mathematics, 62, 19–30.
Pichugin, A., & Rogerson, G. (2012). Extensional edge waves in pre-stressed incompressible
plates. Mathematics and Mechanics of Solids, 17(1), 27–42.
Polyanin, A. D., & Nazaikinskii, V. E. (2015). Handbook of linear partial differential equations for
engineers and scientists. Boca Raton, FL: CRC Press.
Prikazchikov, D. (2013). Rayleigh waves of arbitrary profile in anisotropic media. Mechanics
Research Communications, 50, 83–86.
Prudnikov, A. P., Brychkov, Y. A., Marichev, O. I., & Romer, R. H. (1988). Integrals and
series. (AAPT)
Rayleigh, L. (1885). On waves propagated along the plane surface of an elastic solid. Proceed-
ings of the London Mathematical Society, 1(1), 4–11.
Rogerson, G. (1996). On the existence of surface waves and the propagation of plate waves in
pre-stressed fibre-reinforced composites. Journal of Mechanics and Physics of Solids, 46,
1581–1612.
Rousseau, M., & Maugin, G. A. (2010). Rayleigh surface waves and their canonically asso-
ciated quasi-particles. Proceedings of the Royal Society of London A: Mathematical, Physical and
Engineering Sciences, 467, 495–507.
Royer, D., & Dieulesaint, E. (1999). Elastic waves in solids II: Generation, acousto-optic interac-
tion, applications. Berlin: Springer Science & Business Media.
Sch€olte, J. (1947). The range of existence of Rayleigh and Stoneley waves. Geophysical Journal
International, 5(s5), 120–126.
Sch€olte, J. (1949). On true and pseudo Rayleigh waves. Proceedings of Koninklijke Nederlandse
Akademie van Wetenschappen Series-B, 52, 652–653.
Shuvalov, A. L., & Every, A. G. (2008). On the long-wave onset of dispersion of the surface-
wave velocity in coated solids. Wave Motion, 45(6), 857–863.
Sobolev, S. (1937). Some problems in wave propagation. In P. Frank & R. von Mises (Eds.),
Differential and integral equations of mathematical physics. Russian translation (pp. 468–617).
Moscow-Leningrad: ONTI.
Steigmann, D., & Ogden, R. (2007). Surface waves supported by thin-film/substrate inter-
actions. IMA Journal of Applied Mathematics, 72(6), 730–747.
Stoneley, R. (1924). Elastic waves at the surface of separation of two solids. Proceedings of the
Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character,
106(738), 416–428.
Sveshnikov, A. G., & Tikhonov, A. N. (2005). The theory of functions of a complex variable.
Moscow: Phyzmatlit.
Thompson, I., Abrahams, I. D., & Norris, A. N. (2002). On the existence of flexural edge waves
on thin orthotropic plates. The Journal of the Acoustical Society of America, 112(5), 1756–1765.
Timoshenko, S. (1926). Method of analysis of statistical and dynamical stresses in rail.
In Proceedings of second international congress for applied mechanics, Zurich (pp. 12–17).
Titchmarsh, E. C. (1939). The theory of functions. Oxford, USA: Oxford University Press.
Viktorov, I. (1981). In Surface acoustic waves in solids. Moscow: Nauka.
Vinh, P. C., Anh, V. T. N., & Thanh, V. P. (2014). Rayleigh waves in an isotropic elastic
half-space coated by a thin isotropic elastic layer with smooth contact. Wave Motion,
51(3), 496–504.
Vinh, P. C., & Ogden, R. (2004). On formulas for the Rayleigh wave speed. Wave Motion,
39(3), 191–197.
Vinh, P. C., & Seriani, G. (2010). Explicit secular equations of Stoneley waves in a non-
homogeneous orthotropic elastic medium under the influence of gravity. Applied Math-
ematics and Computation, 215(10), 3515–3525.
Willis, J., & Movchan, A. (1995). Dynamic weight functions for a moving crack. I. Mode
I loading. Journal of the Mechanics and Physics of Solids, 43(3), 319–341.
Yoffe, E. H. (1951). LXXV. The moving Griffith crack. The London, Edinburgh, and Dublin
Philosophical Magazine and Journal of Science, 42(330), 739–750.
Zakharov, D. D. (2004). Analysis of the acoustical edge flexural mode in a plate using refined
asymptotics. The Journal of the Acoustical Society of America, 116(2), 872–878.
Zernov, V., & Kaplunov, J. (2008). Three-dimensional edge waves in plates. Proceedings of the
301–318.
Zhu, J., Popovics, J. S., & Schubert, F. (2004). Leaky Rayleigh and Scholte waves at the
fluid–solid interface subjected to transient point loading. The Journal of the Acoustical Soci-
ety of America, 116(4), 2101–2110.

Asymptotic Theory For Rayleigh and Rayleigh-Type Waves: Julius Kaplunov, Danila A. Prikazchikov

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Asymptotic Theory For Rayleigh and Rayleigh-Type Waves: Julius Kaplunov, Danila A. Prikazchikov

Uploaded by

Copyright:

Available Formats

CHAPTER ONE

Asymptotic Theory for Rayleigh

Advances in Applied Mechanics, Volume 50 # 2017 Elsevier Inc. 1

Appendix A. Near-Resonant Behavior of a Single Degree of Freedom Linear

approach. In this section we also obtain the explicit solutions of 3D moving

2. TIME-HARMONIC RAYLEIGH WAVE ON AN ELASTIC

2.1 Elementary Derivation

ϕ ¼ A1 eikðx1 cos θ + x2 sin θctÞkαx3 , (12)

while the derivative

R0 ð0Þ ¼ 2ðκ 2 1Þ < 0: (21)

The solution of this equation may be obtained through Cardano’s formula,

2.2 Rayleigh Wave of Arbitrary Profile

u2 ¼ 0, ui ¼ ui ðx1 , x3 , tÞ, i ¼ 1,3, (24)

H2+ ¼ fðx1 ; x3 Þ j ∞ < x1 < ∞, 0 x3 < ∞g: (25)

together with the boundary conditions

along the stress-free surface x3 ¼ 0, where, as before,

for a harmonic function f(x1, γx3), being a solution of the equation

where a > 0 is a parameter chosen in order to smooth discontinuities at the

Variation of the displacements (42) on sR at several depths is shown in

3. RAYLEIGH-TYPE INTERFACIAL AND EDGE

were originally referred to as Rayleigh-type waves, see Stoneley (1924),

occupied by an ideal fluid. We concentrate on the wave propagating along

Substitution of (47) into (46) results in a set of linear algebraic equations in

with r and κ defined by (19). It is then clear that

SGð0Þ ¼ 0, lim SGðrÞ ¼ + ∞, SG0 ð0Þ ¼ 2ðκ 2 1Þ < 0, (53)

guaranteeing existence of the solution over the interval 0 < r < κ 2

3.2 Stoneley Wave

where λn and μn, n ¼ 1, 2, are the Lame elastic moduli.

On substituting (56) into (55), we deduce the well-known equation

An interfacial Stoneley wave of arbitrary profile have been investigated

3.3 Bending Edge Wave on a Thin Plate

where the attenuation coefficients γ j are given by Rðγ j Þ > 0

On substituting (64) into (63), we arrive at the set of linear algebraic

Then, on employing the solvability condition, we deduce the dispersion

Dk4 γ 4e ¼ 2ρhω2 , (67)

Thus, the bending edge wave is a dispersive analogue of the Rayleigh

Fig. 4 The coefficient γ e vs the Poisson’s ratio ν.

4. HYPERBOLIC–ELLIPTIC MODEL FOR THE RAYLEIGH

4.1 Plane Strain Problem

We proceed with a multiple scale perturbation scheme, see, e.g., Kevorkian

The concept of slow time is essential for modeling of the near-resonant

P ¼ Pðx1  cR τf ,τs Þ: (75)

The simplest example of a near-resonant excitation is presented in Appendix

the equations of motion (27) become

We now employ the implicit traveling wave ansatz (43) having

ϕ, τf τf cR2 ϕ, 11 ¼ 0, ψ , τf τf cR2 ψ , 11 ¼ 0: (78)

Then, Eq. (77) take the form

implying the Rayleigh equation

as a solvability condition. In addition, we recover the relations (37) between

The solvability of the latter implies

we restate (94) in terms of the original variables as

ð1 + β2R Þ cR2 eαR jpjx3 FL

4.2.1 Vertical Load

χ ¼ x1 cosα + x2 sinα, η ¼ x1 sinα + x2 cosα,

imposing the assumption

meaning that antiplane motion does not induce surface waves.

ðαÞ ðαÞ ðαÞ

The transformed boundary conditions (103) at Q1 ¼ Q2 ¼ 0 are written as

ðαÞ ðαÞ ðαÞ

@ϕðαÞ @ψ ðαÞ ðαÞ @ϕðαÞ @ψ ðαÞ

corresponds to the 2D wave equation

where Δ denotes the 2D Laplace operator in x1 and x2.

ðαÞ ð1 + β2R ÞcR2 ðαÞ

P ¼ Pðx1 cR τf ,τs Þ: (75)