See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228807543

Leaky Rayleigh and Scholte waves at the fluid-solid interface subjected to

transient point loading

Article  in  The Journal of the Acoustical Society of America · January 2004

DOI: 10.1121/1.1791718

CITATIONS READS
118 1,642

3 authors:

Jinying Zhu John S Popovics

University of Nebraska at Lincoln University of Illinois, Urbana-Champaign
90 PUBLICATIONS   1,301 CITATIONS    176 PUBLICATIONS   3,622 CITATIONS   

SEE PROFILE SEE PROFILE

Frank Schubert
Fraunhofer Institute for Ceramic Technologies and Systems IKTS
92 PUBLICATIONS   1,081 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Monitoring Stress using Acoustoelastic Effect View project

Thermal modulation of Nonlinear ultrasonic waves View project

All content following this page was uploaded by Frank Schubert on 06 June 2014.

The user has requested enhancement of the downloaded file.

Leaky Rayleigh and Scholte waves at the fluid–solid interface
subjected to transient point loading
Jinying Zhu and John S. Popovicsa)
Department of Civil and Environmental Engineering, University of Illinois at Urbana, Illinois 61801

Frank Schubert
Lab EADQ, Fraunhofer Institute for Nondestructive Testing, D-01326 Dresden, Germany

共Received 6 February 2004; revised 9 July 2004; accepted 18 July 2004兲

The analysis of acoustic waves generated by a transient normal point load applied on a fluid–solid
interface is presented. The closed-form exact solution of the wave motion is obtained by using
integral transform techniques. The obtained analytical solution provides necessary theoretical
background for optimization of fluid-coupled ultrasonic and acoustic wave detection in experiments.
Numerical simulation 共elastodynamic finite integration technique兲 is performed to verify the
obtained analytical solution. Detailed descriptions of leaky Rayleigh and Scholte wave solutions are
presented. A simplified solution to isolate the contributions of leaky Rayleigh and Scholte waves
generated by a transient point load is proposed, and closed-form formulations for displacement and
stress components are then presented. The simplified solution is compared to the exact solution for
two configurations: water/concrete and air/concrete. The excitation effectiveness of leaky Rayleigh
waves for the air/concrete configuration is studied, which has practical significance to air-coupled
sensing in civil engineering structures. © 2004 Acoustical Society of America.
关DOI: 10.1121/1.1791718兴
PACS numbers: 43.35.Pt, 43.20.El, 43.35.Zc 关YHB兴 Pages: 2101–2110

I. INTRODUCTION tally by Glorieux et al. 共2001兲. The propagation of leaky

Rayleigh waves under the influences of viscous damping and
The propagating leaky Rayleigh wave that emanates heat conduction at the fluid–solid interface was studied by
from a fluid–solid interface has been used as an effective Qi 共1994兲, who concluded that the effect of viscosity can be
means for surface and subsurface defect detection. With re- neglected for fluids with Reynolds number larger than 2500.
cent improvements in instrumentation, air-coupled transduc- For common fluids such as water and air at normal driving
ers have been used for detection of leaky surface 共Zhu et al., frequencies 共⬍10 MHz兲, the viscous effect of the fluid can
2001兲 and leaky guided waves 共Castaings et al., 2001兲. With be neglected because the Reynolds numbers are far above the
the advantage of noncontact sensing, air-coupled transducers critical value of 2500.
provide an opportunity for quick scanning and imaging of The response of leaky waves owing to transient point
large civil engineering structures by detecting the leaky Ray- loading is of great practical interest to nondestructive evalu-
leigh wave. Therefore, detailed study of leaky surface waves ation 共NDE兲 researchers, especially in civil engineering,
for this case is needed. where an impulse hammer or a point impactor applied to the
Extensive studies and applications of leaky surface surface of the solid is often used as a transient wave source.
waves have been reported during the past 40 years. A com- Gusev et al. 共1996兲 provided detailed theoretical analyses of
prehensive study of Rayleigh waves and leaky Rayleigh laser-induced wave motions at the fluid–solid interface,
waves has been given by Viktorov 共1967兲, where leaky Ray- which include Scholte, leaky Rayleigh, and lateral waves.
leigh waves at the interfaces of a solid half-space with both a General solutions for interface wave motion were given us-
fluid layer and a fluid half-space were investigated in great ing a 2-D formulation. 2-D and 3-D analytical solutions for a
detail. Leaky Rayleigh waves propagate with a velocity fluid–solid configuration subjected to implosive line and
slightly higher than the ordinary Rayleigh wave, and attenu- point sources in the fluid have been given by de Hoop et al.
ate more intensively with distance due to continuous energy 共1983, 1984兲. However, for the case of the ‘‘Lamb’’ problem
radiation into the fluid. It was initially believed that the leaky in fluid–solid configuration, where a normal transient point
Rayleigh wave exists when the fluid wave velocity c F is load is applied at the fluid–solid interface, no closed-form
smaller than the leaky Rayleigh wave velocity c LR . How- exact solution has been reported so far.
ever, Mozhaev and Weihnacht 共2002兲 showed the actual In this paper, Laplace and Hankel integral transforms are
threshold phase velocity for leaky Rayleigh wave existence employed to derive the full analytical solution to the
was 1.45% lower than the acoustic wave velocity of the ‘‘Lamb’’ problem in a fluid–solid half-space system, where a
fluid. The character and existence conditions of leaky Ray- point load, varying with time as a step function, is applied at
leigh and Scholte waves were also investigated experimen- the interface. The obtained step response solution is shown in
integral format, which can be calculated numerically. Im-
a兲
Author to whom correspondence should be addressed. Electronic mail: pulse responses are then obtained by differentiating the step
[email protected] responses. Therefore, for any transient impact loading that

J. Acoust. Soc. Am. 116 (4), Pt. 1, October 2004 0001-4966/2004/116(4)/2101/10/$20.00 © 2004 Acoustical Society of America 2101
 ⳵␸ 2 ⳵ ␺ ⳵␸ 2 1 ⳵ 共 r ␺ 兲
u 2⫽ ⫺ , w 2⫽ ⫹ , 共2兲
⳵r ⳵z ⳵z r ⳵r
where the subscripts 1 and 2 represent quantities related to
the fluid and solid, respectively.

B. Continuity and initial condition

A point load is applied at the origin as a Dirac delta
function in space and varies as a Heaviside step function in
FIG. 1. A transient point load applied at the interface of a fluid–solid half- time, which can be expressed as QH(t) ␦ (r)/2␲ r. Because
space system. ideal fluid and shear-free interface conditions are assumed,
only the normal stress and vertical component of the dis-
has arbitrary temporal variation and spatial distribution, the placement are continuous at the interface. The continuity
responses can be obtained by convolving the impulse re- conditions at z⫽0 are
sponse in time and space domains. The obtained analytical w 1 ⫽w 2 ,
solutions are verified by EFIT 共elastodynamic finite integra-
tion technique兲 numerical simulation, which is a powerful ␦共 r 兲 ␦共 r 兲
␶ zz2 ⫽ ␶ zz1 ⫹ 共 ⫺Q 兲 H 共 t 兲 ⫽⫺ P⫺QH 共 t 兲 , 共3兲
tool for elastodynamic wave field analysis. Then a simplified 2␲r 2␲r
analytical formulation for the pressure field in the fluid is
derived and illustrated. ␶ zr2 ⫽ ␶ zr1 ⫽0,
where ␶ zz and ␶ zr are the normal and shear components of
stress, and P the pressure in the fluid.
II. COMPLETE FORMULATION
Assuming the system is at rest prior to t⫽0, we have
Consider a fluid–solid half-space system as shown in
␸ 1 共 r,z,0兲 ⫽ ␸¨ 1 共 r,z,0兲
Fig. 1. The solid half-space is given by z⬎0, and the fluid by
z⬍0. The properties of the fluid are given by the Lamé con- ⫽ ␸ 2 共 r,z,0兲 ⫽ ␸¨ 2 共 r,z,0兲
stant ␭ 1 and mass density ␳ 1 , and those of the solid by the
Lamé constants ␭ 2 and ␮ and density ␳ 2 . The interface be- ⫽ ␺ 2 共 r,z,0兲 ⫽ ␺¨ 2 共 r,z,0兲 ⫽0. 共4兲
tween the fluid and solid half-space is subjected to a normal
point load of magnitude QH(t), where H(t) is the Heaviside C. Integral transform
step function. Because the wave motion in the fluid and solid
One-sided Laplace and Hankel transforms are used to
generated by the point load is axially symmetric about the z
obtain solutions to the equations. The Laplace and nth-order
axis, cylindrical coordinates are employed, where the origin
Hankel transforms are defined respectively as
is located at the load point on the interface.
A. Fluid–solid half-space f̄ 共 p 兲 ⫽ 冕
0
⬁
f 共 t 兲 e ⫺pt dt, f H n共 ␰ 兲 ⫽ 冕0
⬁
f 共 r 兲 J n 共 ␰ r 兲 r dr, 共5兲
Introducing displacement potential functions ␸ 1 in the
fluid and ␸ 2 and ␺ in the solid, the governing equations for where p and ␰ are variables of the Laplace and Hankel trans-
the fluid and the solid half-spaces are forms, respectively.
Applying the Laplace transform to Eq. 共1兲 with respect
⳵ 2␸ 1 1 ⳵␸ 1 ⳵ 2 ␸ 1 1 to time t and the zeroth- and first-order Hankel transform
⫹ ⫹ ⫽ 2 ␸¨ 1 ,
⳵r2 r ⳵r ⳵z2 cF with respect to the radial distance r yields
H
⳵ 2␸ 2 1 ⳵␸ 2 ⳵ 2 ␸ 2 1 d 2 ¯␸ 1 0
⫹ ⫹ ⫽ 2 ␸¨ 2 , 共1兲 H
⫺ ␣ 21 ¯␸ 1 0 ⫽0,
⳵r2 r ⳵r ⳵z2 cP dz 2

共6兲
⳵ 2␺ 1 ⳵ ␺ ⳵ 2␺ ␺ 1 H
d 2 ¯␸ 2 0 d 2 ¯␺ H 1
⫹ ⫹ 2 ⫺ 2 ⫽ 2 ␺¨ , H
⫺ ␣ 22 ¯␸ 2 0 ⫽0, ⫺ ␤ 2 ¯␺ H 1 ⫽0,
⳵r 2 r ⳵ r ⳵z r cS 2
dz dz 2
where c F2 ⫽␭ 1 / ␳ 1 is the acoustic wave velocity in the fluid, where
and c 2P ⫽(␭ 2 ⫹2 ␮ )/ ␳ 2 and c 2S ⫽ ␮ / ␳ 2 are the P- and S-wave
velocities in the solid. The double dots represent a double ␣ 21 ⫽ ␰ 2 ⫹s F2 p 2 , ␣ 22 ⫽ ␰ 2 ⫹s 2P p 2 , ␤ 2 ⫽ ␰ 2 ⫹s 2S p 2 , 共7兲
derivative with respect to time. The displacements are related
to potential functions by and s F , s P , s S are P- and S-wave slowness in the fluid and
the solid. Only choosing the terms which lead to finite re-
⳵␸ 1 ⳵␸ 1 sponses for large values of 兩 z 兩 , we obtain the solutions to Eq.
u 1⫽ , w 1⫽ ,
⳵r ⳵z 共6兲,

2102 J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 Zhu et al.: Leaky Rayleigh and Scholte waves
 ¯␸ H 0
⫽⌽ 1 共 ␰ ,p 兲 e ␣ 1 z , Q 1
关共 s 2S p 2 ⫹2 ␰ 2 兲 2 e ⫺ ␣ 2 z
H0
1 ¯␶ zz2 ⫽⫺
2␲ p
¯␸ H
2
0
⫽⌽ 2 共 ␰ ,p 兲 e ⫺ ␣ 2 z , ¯␺ H 1 ⫽⌿ 共 ␰ , p 兲 e ⫺ ␤ z , 共8兲
1
where ⌽ 1 ( ␰ ,p), ⌽ 2 ( ␰ , p), and ⌿( ␰ , p) are functions of p ⫺4 ␣ 2 ␤ ␰ 2 e ⫺ ␤ z 兴 ,
D H 共 ␰ ,p 兲
and ␰ that need to be determined. From Eq. 共2兲 and the
displacement–stress relationships, the transformed displace- H1 Q 1 ⫺␣2z ⫺␤z
2 ␣ 2 ␰ 共 s 2S p 2 ⫹2 ␰ 2 兲
ments and stresses 共pressure for the fluid兲 in the fluid are ¯␶ zr2 ⫽ 关 ⫺e ⫹e 兴 .
2␲ p D H 共 ␰ ,p 兲
H
H
d ¯␸ 1 0 H H H
w̄ 1 0 ⫽ , ū 1 1 ⫽⫺ ␰␸
¯ 1 0, P̄ H 0 ⫽⫺ ␳ 1 p 2 ¯␸ 1 0 , 共9兲 III. CHARACTERISTIC EQUATION FOR LEAKY
dz
RAYLEIGH AND SCHOLTE WAVES
and in the solid are
D H ( ␰ ,p)⫽0 is the characteristic equation for leaky
H
H
d ¯␸ 2 0 H d ¯␺ H 1 Rayleigh waves at the interface of fluid–solid half-spaces.
w̄ 2 0 ⫽ ⫹ ␰ ¯␺ H 1 , ū 2 1 ⫽⫺ ¯H
⫺ ␰␸ 2
0
, Introducing the substitution ␥ ⫽i ␰ /ps S and notations q
dz dz
⫽s L /s S and u⫽s F /s S , the equation D H ⫽0 can be changed
H0
¯␶ zz2 冋 H
⫽ ␮ 共 s 2S p 2 ⫹2 ␰ 2 兲 ¯␸ 2 0 ⫹2 ␰
d ¯␺ H 1
dz
册, 共10兲
to the following in terms of ␥:
␳1 冑␥ 2 ⫺q 2

冋 册
共 1⫺2 ␥ 2 兲 2 ⫺4 ␥ 2 冑␥ 2 ⫺1 冑␥ 2 ⫺q 2 ⫹ ⫽0.
H1
H
d ¯␸ 2 0 ␳2 冑␥ 2 ⫺u 2
¯␶ zr2 ⫽⫺ ␮ 2 ␰ ⫹ 共 s 2S p 2 ⫹2 ␰ 2 兲 ¯␺ H 1 . 共15兲
dz
Applying the same integral transforms to the continuity Equation 共15兲 is the same as that given by Viktorov 共1967兲,
conditions Eq. 共3兲, then substituting Eqs. 共9兲 and 共10兲 into it, which differs from the regular Rayleigh equation in the third
generates a group of linear equations in terms of ⌽ 1 ( ␰ ,p), term due to the pressure of the fluid. Equation 共15兲 produces
⌽ 2 ( ␰ ,p), and ⌿( ␰ , p). Solving the equations yields eight Riemann sheets owing to the square roots. The roots
have the physical meaning of normalized slowness with re-
Q ␣ 2 s 2S p 2 Q s 2S p 2 ⫹2 ␰ 2 spect to s S . When leaky Rayleigh waves exist, we can obtain
⌽ 1⫽ , ⌽ 2 ⫽⫺ ,
2 ␲ ␮ p ␣ 1 D H共 ␰ , p 兲 2 ␲ ␮ p D H 共 ␰ ,p 兲 six roots, which include two pairs of complex conjugate
pairs corresponding to the leaky Rayleigh wave 关 Re(␥R2)
Q 2 ␣ 2␰ ⬍u 2 兴 , and two opposite real roots corresponding to the
⌿⫽⫺ , 共11兲
2 ␲ ␮ p D H共 ␰ , p 兲 Scholte wave ( ␥ sch 2
⬎u 2 ).
The two pairs of complex conjugate roots corresponding
where
to the leaky Rayleigh wave take the form of ⫾ 关 Re(␥)
␳1 4 4 ␣2 ⫾i Im(␥)兴, in which only two roots are acceptable. Because p
D H 共 ␰ ,p 兲 ⫽ 共 s 2S p 2 ⫹2 ␰ 2 兲 2 ⫺4 ␰ 2 ␣ 2 ␤ ⫹ p sS . 共12兲
␳2 ␣1 is the Laplace transform variable, Re(p) must be negative to
have a finite time domain response. Therefore, only the roots
Substituting Eq. 共11兲 into Eqs. 共9兲 and 共10兲, we obtain the which lead to Re(p)⬍0 are acceptable, where p⫽i ␰ / ␥ s S . A
displacements and pressure in the fluid, quick analysis shows that the two acceptable roots take the
Q 1 ␣ 2 s 2S p ␣ z form of ␥ R1,R2 ⫽⫾Re(␥)⫺i Im(␥). Here, we denote them as
H
w̄ 1 0 ⫽ e 1, ␥ R1 , ␥ R2 , and note ␥ R2 ⫽⫺ ¯␥ R1 , where the top bar indicates
2 ␲ ␮ D H共 ␰ , p 兲
complex conjugate.
Q 1 ␣ 2 s 2S p ␰ Analysis also shows that 兩 Re(␥R1,R2 ) 兩 ⬍s R /s S , where s R
e ␣1z,
H
ū 1 1 ⫽⫺ 共13兲 is the slowness of Rayleigh wave in free surface solid half-
2 ␲ ␮ ␣ 1 D H共 ␰ , p 兲
space. This result confirms that the slowness of the leaky
Q ␳ 1 ␣ 2 s 4S p 3 Rayleigh wave is smaller than s R , i.e., in a fluid–solid half-
P̄ H 0 ⫽⫺ e ␣1z, space system, the leaky Rayleigh wave will travel faster than
2 ␲ ␳ 2 ␣ 1 D H共 ␰ , p 兲
the pure Rayleigh wave in the same free surface solid. This
and the displacements and stresses in the solid conclusion is reasonable considering the influence of fluid
Q 1 pressure applied on solid surface. The imaginary part of ␥ R
关共 s 2S p 2 ⫹2 ␰ 2 兲 e ⫺ ␣ 2 z
H
ū 2 1 ⫽ accounts for the energy leaked into fluid. The larger the value
2␲ ␮p
of 兩 Im(␥R1 ) 兩 , the more energy leaks into the fluid.
␰ The Scholte wave velocity can be obtained from the real
⫺2 ␣ 2 ␤ e ⫺ ␤ z 兴 , roots ⫾ ␥ sch , where ␥ sch ⬎u and s sch ⫽ ␥ sch s S . Scholte
D H共 ␰ , p 兲
waves always exist for any fluid–solid combination case. For
Q 1 the lighter fluids lying on stiffer solids case, s sch is only
关共 s 2S p 2 ⫹2 ␰ 2 兲 e ⫺ ␣ 2 z
H
w̄ 2 0 ⫽
2␲ ␮p slightly smaller than the acoustic wave velocity of the fluid,
and the Scholte wave contribution is relatively small. The
␣2 property of Scholte waves will be discussed in more detail in
⫺2 ␰ 2 e ⫺ ␤ z 兴 , 共14兲
D H共 ␰ , p 兲 Sec. VII.

J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 Zhu et al.: Leaky Rayleigh and Scholte waves 2103
IV. SOLUTION OF PRESSURE IN THE FLUID

The solution of fluid pressure in the space–time domain

is derived because pressure in the fluid is the most often
measured quantity in practical application. By taking the in-
verse zeroth-order Hankel transform to Eq. 共13兲 and intro-
ducing the substitution ␰ ⫽p ␩ , we have
Qs 4S ␳ 1
冕 冑冑␩␩ ⫹s 2P
⬁ 2
p
P̄⫽⫺
2␲ ␳2 0 2
⫹s F2 D共 ␩ 兲

⫻e pz 冑␩
2 ⫹s 2
F J 0共 p ␩ r 兲 ␩ d ␩ , 共16兲

where D( ␩ ) is defined as
FIG. 2. Complex v plane and the modified integral path. The indices denote:
D 共 ␩ 兲 ⫽ 共 2 ␩ 2 ⫹s 2S 兲 2 ⫺4 ␩ 2 冑␩ 2 ⫹s 2P 冑␩ 2 ⫹s 2S P→leaky P wave, S→leaky S wave, LR→leaky R wave, f→fluid acoustic
wave, SCH→Scholte wave.
␳1 冑␩ 2 ⫹s 2P 4
⫹ s . 共17兲
␳2 冑␩ 2 ⫹s F2 S where G P (t) is defined as the Green’s function for pressure.
The variable v is related to ␥ in Eq. 共15兲 by v ⫽⫺ ␥ s S .
The inverse Laplace transform is evaluated by the According to the analysis in the previous section, there are
Cagniard–de Hoop method, as described by Achenbach two poles in the right half complex v plane, which corre-
共1973兲. Introducing the following representation of J 0 (x), spond to leaky Rayleigh and Scholte wave arrivals. The in-

冕冑
2 ⬁ e ixs tegral path along the negative imaginary axis can be de-
J 0共 x 兲 ⫽ Im ds, 共18兲 formed to path ⌫ as shown in Fig. 2. The new integration
␲ s 2 ⫺1 path ⌫ is defined by the equation ␶ ⫽ v r⫺z 冑s F2 ⫺ v 2 , which
1

and substituting it into Eq. 共16兲 yields can be solved for v to yield

P̄⫽⫺
Qs 4S ␳ 1
␲2 ␳2
Im 冕
0
⬁ 冑␩ 2 ⫹s 2P ␩
冑␩ 2 ⫹s F2 D 共 ␩ 兲
d␩ v共 ␶ 兲 ⫽
␶ r⫹z 冑s F2 R 2 ⫺ ␶ 2
R2
, ␶ is real and 0⭐ ␶ ⭐⬁,
共23兲
e ip ␩ rs⫹pz 冑␩
冕
2 ⫹s 2
⬁ F
⫻ p ds. 共19兲
1 冑s 2 ⫺1 where R⫽ 冑r 2 ⫹z 2 . The benefit of deforming the original
integration path to ⌫ is obvious, because there is no pole
Introducing the substitution t⫽⫺i ␩ rs⫺z 冑␩ 2 ⫹s F2 and ␩ along the new path. The Green’s function integration along
⫽i v yields the new path is

P̄⫽
Qs 4S ␳ 1
␲2 ␳2
Im 冕 0
⫺i⬁ 冑s 2P ⫺ v 2 v
冑s F2 ⫺ v 2 D 共 i v 兲
dv
冉
G P 共 t 兲 ⫽ Im 冕 t 冑s 2P ⫺ v共 ␶ 兲 2 v共 ␶ 兲
t p 冑s F ⫺ v共 ␶ 兲 2 D 关 i v共 ␶ 兲兴
2

⫻ 冕 1
⬁
p
e ⫺pt
冑s 2 ⫺1
ds, 共20兲
⫻
r⫺ ␶ z/ 冑s F2 R 2 ⫺ ␶ 2
冊
d␶ , 共24兲
R 2 冑关 t⫺ ␶ ⫹ v共 ␶ 兲 r 兴 2 ⫺ v共 ␶ 兲 2 r 2
where

L ⫺1 冉冕 1
⬁
p 2
冑s ⫺1
e ⫺pt
ds ⫽
d
dt 冊 冉冑 H 共 t⫹z 冑s F2 ⫺ v 2 ⫺ v r 兲
共 t⫹z 冑s F2 ⫺ v 2 兲 2 ⫺ v 2 r 2
冊 where t p ⫽rs P ⫹ 兩 z 兩 冑s F2 ⫺s 2P is the leaky P wave arrival time.
G P (t)⫽0 when t⬍t p .
However, when the receiver is located on the interface,
共21兲 i.e., z⫽0, the integral path ⌫ is along the real axis, and
and L represents the Laplace transform. Therefore, the in- passes by the Scholte pole. In this case, the principal value of
verse Laplace transform of Eq. 共20兲 is Eq. 共24兲 must be taken.

冉
The integrand in Eq. 共24兲 has a square root singularity at
P共 t 兲⫽
Qs 4S ␳ 1 d
␲ 2 ␳ 2 dt
Im 冕 0
⫺i⬁ 冑s 2P ⫺ v 2 v
冑s F2 ⫺ v 2 D 共 i v 兲
end point ␶ ⫽t, which increases the difficulty of numerical
integration. This problem can be solved by further introduc-

冊
ing the following transformation,
H 共 t⫹z 冑s F2 ⫺ v 2 ⫺ v r 兲
⫻ dv ␶ ⫽t 1 cos2 共 ␪ 兲 ⫹t 2 sin2 共 ␪ 兲 , 共25兲
冑共 t⫹z 冑s F2 ⫺ v 2 兲 2 ⫺ v 2 r 2
where t 1 , t 2 are the lower and upper limits of the integration.
Qs 4S ␳ 1 d P
⫽ G 共 t 兲, 共22兲 Then the integration interval t 1 ⭐ ␶ ⭐t 2 is mapped to the
␲ 2 ␳ 2 dt fixed interval 0⭐␪⭐␲/2, while

2104 J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 Zhu et al.: Leaky Rayleigh and Scholte waves
FIG. 3. Green’s function G P (t) for pressure in the fluid at position r
⫽1.5 m, z⫽⫺0.05 m. Material parameters for the fluid ␳ 1 ⫽1000 kg/m3 ,
FIG. 4. Absolute value of pressure 共in fluid兲 and stress ␶ zz 共in solid兲 field
c F ⫽1500 m/s; for the solid, ␳ 2 ⫽2400 kg/m3 , c P ⫽4000 m/s, v ⫽0.25.
snapshot at t⫽0.72 ms given by EFIT analysis. The interface is subject to a
transient point load that varies with time as f (t)⫽sin2 (␲t/T) with T
⫽200 ␮ s.
d ␶ ⫽2 共 t 2 ⫺t 1 兲 sin共 ␪ 兲 cos共 ␪ 兲 ,
V. VERIFICATION BY NUMERICAL SIMULATION
冑关 t 2 ⫺ ␶ ⫹ v共 ␶ 兲 r 兴 2
⫺ v共 ␶ 兲 r
2 2
To verify the obtained analytical solutions, numerical
⫽ 冑共 t 2 ⫺t 1 兲 cos共 ␪ 兲 冑共 t 2 ⫺t 1 兲 cos 共 ␪ 兲 ⫹2 v共 ␶ 兲 r.
2
共26兲 analyses were performed to simulate the response for fluid–
solid half-space cases. The elastodynamic finite integration
technique 共EFIT兲 is a numerical time-domain scheme to
Another square root singularity at point ␶ ⫽t f ⫽s F R can be model elastic wave propagation in homogeneous and hetero-
processed similarly. A similar technique was also used by de geneous, dissipative and nondissipative, as well as isotropic
Hoop 共1984兲. and anisotropic elastic media 共Fellinger et al., 1995兲. EFIT
The vertical component of displacement in the fluid is uses a velocity-stress formalism on a staggered spatial and
also obtained in the same way: temporal grid complex. The starting point of EFIT is the
integral form of the linear governing equations, i.e., the

w 共 t 兲 ⫽⫺
Q s 4S
␲2 ␳2
Im 冕冑t

tp
s 2P ⫺ v共 ␶ 兲 2 v共 ␶ 兲
D 关 i v共 ␶ 兲兴 R 2
Cauchy equation of motion, and the equation of deformation
rate. EFIT performs integrations over certain control vol-
umes V, and the surfaces of these cells S, assuming constant
r⫺ ␶ z/ 冑s F2 R 2 ⫺ ␶ 2 velocity and stress within V and on each S. This method
⫻ d␶. 共27兲 requires staggered grids and leads to a very stable and effi-
冑关 t⫺ ␶ ⫹ v共 ␶ 兲 r 兴 2 ⫺ v共 ␶ 兲 2 r 2 cient numerical code, which also allows easy and flexible
treatment of various boundary conditions. In recent years,
Figure 3 shows the Green’s function G P (t) at a near- EFIT has been successfully used for a wide variety of appli-
surface position r⫽1.5 m, z⫽⫺0.05 m in the fluid. The ma- cations, especially in the field of nondestructive testing
terial configuration simulates a concrete/water system. The 共Schubert and Marklein, 2003兲.
arrival times of all wave types are marked. It is noticed that In the present case of a transient point load at a fluid–
the slope is discontinuous at positions t⫽t p , t s , and t f , solid interface, we used a special axisymmetric EFIT code in
which correspond to the leaky P-, leaky S-, and fluid acoustic cylindrical coordinates 共Schubert et al., 1998兲. The water/
waves in water. From the mathematics viewpoint, these dis- concrete case shown in Fig. 3 was studied. The material pa-
continuities come from the branch points along the integral rameters are for water, ␳ 1 ⫽1000 kg/m3 , c F ⫽1500 m/s, and
path ⌫. The poles corresponding to the leaky Rayleigh and for concrete ␳ 2 ⫽2400 kg/m3 , P wave velocity c P
Scholte wave arrivals contribute to the large smooth peaks. ⫽4000 m/s, and Poisson’s ratio ␯⫽0.25. The vertical tran-
When the receiving position is very close to the interface, the sient point load varies with time as function f (t)
integral path ⌫ in Fig. 2 will bend to the real axis, and a ⫽sin2(␲t/T), where the force duration is T⫽200 ␮ s. A grid
sharp peak will appear nearby the Scholte wave arrival in spacing of ⌬r⫽⌬z⫽2.5 mm and a time step of 0.44 ␮s are
Fig. 3. used in order to guarantee stability as well as sufficient dis-
The impulse response to a point loading on the interface cretization of the shortest wavelengths. The dimensions of
between a fluid and a solid half-space can be obtained from the model are 2 m in radial and 3 m in axial direction, re-
the corresponding step response solution by taking differen- sulting in 800⫻1200 grid cells. At the outer boundaries of
tiation with respect to time. Then for any transient loading the model, highly effective absorbing boundary conditions
that has arbitrary temporal variation and spatial distribution, based on the perfectly matched layer 共PML兲 are used in order
the response can be obtained by convolving the impulse re- to suppress interfering reflections 共Liu, 1999兲.
sponse in both time and space domains. Figure 4 shows the cross-sectional snapshot image of

J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 Zhu et al.: Leaky Rayleigh and Scholte waves 2105
 leigh wave effects in a free surface half-space by Chao et al.
共1961兲. Achenbach 共1973兲 reproduced Chao’s results in de-
tail. The simplified solution provides an easy and quick way
to estimate leaky Rayleigh and Scholte wave effects.

A. Displacements and stresses in the fluid

Applying the inverse Hankel and Laplace transforms to
the transformed solution in Eq. 共13兲, the pressure in fluid can
be expressed as

P 共 r,z,t 兲 ⫽⫺
Q s 4S ␳ 1
2␲ 2␲i ␳2
冕
0
⬁
J 0共 ␰ r 兲 ␰

FIG. 5. Comparison of the analytical and the numerical solutions for pres-
⫻ 冕 ⑀ ⫹i⬁

⑀ ⫺i⬁
␣2 p3
␣ 1 D H 共 ␰ ,p 兲
e ␣ 1 z⫹pt d p d ␰ . 共28兲
sure in the fluid at r⫽1.5 m, z⫽⫺0.05 m. The interface is subject to a point
load that varies with time as f (t)⫽sin2 (␲t/T) with T⫽200 ␮ s. Material Considering the integrand term, the contributions from the
parameters for the fluid ␳ 1 ⫽1000 kg/m3 , c F ⫽1500 m/s; for the solid ␳ 2 leaky Rayleigh poles are
⫽2400 kg/m3 , c P ⫽4000 m/s, v ⫽0.25.

pressure field 共absolute value兲 as generated by EFIT, at time

I 1⫽
1
2␲i
冕 ⬁

0
J 0共 ␰ r 兲 ␰ 冕 ⑀ ⫹i⬁

⑀ ⫺i⬁
␣2 p3
␣ 1 D H 共 ␰ ,p 兲
e ␣ 1 z⫹pt dp d ␰

冕
t⫽0.72 ms. Only the region with r⬎0 was calculated, but ⬁
for better illustration the reversed region with r⬍0 is also ⫽ J 0共 ␰ r 兲
shown here. The half-circle in the upper half-plane repre- 0

sents the acoustic wave front in water, and the inclined lines
represent the leaky Rayleigh wave fronts, which are tangent
to the half circle at the leaky angle direction. In this case, the
⫻␰ 冋 ␣2 p3
␣ 1 ⳵ D H 共 ␰ ,p 兲 / ⳵ p
e ␣ 1 z⫹pt 册 p⫽i ␰ /s S ␥ R1 ,i ␰ /s S ␥ R2
d␰,

leaky Rayleigh angle determined by Snell’s law is ␪⫽43.8°,

共29兲
measured from the vertical axis. The leaky Rayleigh wave
front is separable from the subsequent Scholte and fluid where the expression 关 ( ␣ 2 / ␣ 1)
acoustic wave fronts at larger values of radial distance r. The ⫻关 p 3 / ⳵ D H ( ␰ ,p)/ ⳵ p 兴兴 p⫽i ␰ /s S ␥ R1 ,i ␰ /s S ␥ R2 represents a pair of
3-D shape of the combined leaky Rayleigh and fluid acoustic complex conjugate constants, denoted as A 1 and Ā 1 .
wave fronts looks like a domed cone. The leaky Rayleigh Introducing

冋冑 册
wave in concrete is also seen, which behaves similarly to the
i z t
ordinary Rayleigh wave, and attenuates exponentially with m 1,2⫽⫺ u 2 ⫺ ␥ R1,R2
2
⫹ 共30兲
increasing depth in the solid. In the near-interface region, the ␥ R1,R2 r s Sr
Scholte wave effect is strong in both the fluid and the solid. generates
Figure 5 shows the analytical and numerical time do-
main near-surface response of the pressure in water, at posi- 关 ␣ 1 z⫹ pt 兴 p⫽i ␰ /s S ␥ R1,R2 ⫽⫺ ␰ rm 1,2 , 共31兲
tion r⫽1.5 m, z⫽⫺0.05 m. Very good agreement between where m 1 ⫽m̄ 2 . Thus, Eq. 共29兲 can be expressed as
analytical and numerical responses is observed.

VI. SIMPLIFIED LEAKY RAYLEIGH WAVE RESPONSE

I 1 ⫽A 1 冕0
⬁
J 0 共 ␰ r 兲 ␰ e ⫺ ␰ m 1 r d ␰ ⫹Ā 1 冕 ⬁

0
J 0共 ␰ r 兲 ␰ e ⫺␰m2r d ␰

In Sec. IV, the complete analytical solution of pressure

and displacement in the fluid are obtained. However, the in- ⫽2 Re 冕 ⬁

0
A 1e ⫺␰m1rJ 0共 ␰ r 兲 ␰ d ␰

冋 册
tegral form solution is not always convenient to use. In this
and the next sections, the authors provide a simplified 2 A 1m 1
closed-form solution to the same problem by considering ⫽ Re , Re共 m 1 兲 ⬎0, 共32兲
r 2
共 1⫹m 21 兲 3/2
only Rayleigh and Scholte pole contributions. This simplifi-
cation is acceptable when measurements are taken at a large where we use the zeroth-order Hankel transform formula

冕
distance from the source.
⬁ a
According to the previous analysis, the wave field ex- J 0 共 ␰ r 兲 ␰ e ⫺a ␰ d ␰ ⫽ , Re共 a 兲 ⬎0. 共33兲
cited by a normal point load at the interface includes contri- 0 共 r 2 ⫹a 2 兲 3/2
butions from leaky P, S, Rayleigh, fluid acoustic, and Scholte
From Eq. 共28兲, the pressure P is expressed as

冋 册
waves. Analysis shows that, at large horizontal distances
from the source, disturbances near the interface are domi- Q ␳1 1 4 A 1m 1
nated by leaky Rayleigh and Scholte wave contributions, P⫽⫺ s S Re . 共34兲
␲ ␳2 r 2
共 1⫹m 21 兲 3/2
which can be obtained from the residues at corresponding
poles. The similar idea was already used to investigate Ray- Similarly, by introducing

2106 J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 Zhu et al.: Leaky Rayleigh and Scholte waves
 A 2⫽ ␰ 冋 ␣2p
⳵ D H共 ␰ , p 兲 / ⳵ p 册 p⫽i ␰ /s S ␥ R1
, ␶ zz2 ⫽⫺
Q 1
␲ r2
Re 冋
B 5 n p1
2 3/2
共 1⫹n p1 兲
⫺
B 6 n s1
共 1⫹n s12 3/2
兲
, 册
冋 册 冋 册
共35兲
␣2 p␰2 Q 1 ⫺B 7 B7
A 3⫽ , ␶ zr2 ⫽ Re ⫹ ,
␣ 1 ⳵ D H共 ␰ , p 兲 / ⳵ p p⫽i ␰ /s S ␥ R1 ␲ r 2 2
共 1⫹n p1 兲 3/2 2 3/2
共 1⫹n s1 兲
the vertical and radial components of the displacement in the where the coefficients are

冋 册
fluid are obtained:
s 2S p 2 ⫹2 ␰ 2

w 1⫽
Q s 4S 1
␲ ␳2 r
Re
A2
冋, 册 B 1⫽
p ⳵ D H 共 ␰ ,p 兲 / ⳵ p
␣ 2␰
p⫽i ␰ /s S ␥ R1
,

冋 册
共 1⫹m 21 兲 1/2
2 ␰ 3␣ 2

冉冋 册冊
共36兲 B 2⫽ ,
Q 1s 2S m1 p ⳵ D H 共 ␰ ,p 兲 / ⳵ p
u 1 ⫽⫺ Re 1⫺ A3 , p⫽i ␰ /s S ␥ R1
␲ ␮ r

冋 册
共 1⫹m 21 兲 1/2
共 s 2S p 2 ⫹2 ␰ 2 兲
Re共 m 1 兲 ⬎0. B 3⫽ ␰2 , 共40兲
p ⳵ D H 共 ␰ ,p 兲 / ⳵ p p⫽i ␰ /s S ␥ R1

冋 册
Care should be taken with the square root when calcu-
lating m 1,2 from Eq. 共30兲. To have bounded results, only the 2 ␰ 2␣ 2␤
B 4⫽ ,
branch that gives Re(m1)⬎0 should be selected. p ⳵ D H 共 ␰ ,p 兲 / ⳵ p p⫽i ␰ /s S ␥ R1

B. Displacement and stress in the solid half-space

The response of the leaky Rayleigh wave in the solid
B 5⫽ 冋 共 s 2S p 2 ⫹2 ␰ 2 兲 2
p ⳵ D H 共 ␰ ,p 兲 / ⳵ p
册 p⫽i ␰ /s S ␥ R1
,
can be obtained in a similar way. Introducing

冋 册
B 6 ⫽2B 4 , B 7 ⫽2B 1 .
1 z
n p1,2⫽ ⫾ 冑␥ R1,R2
2
⫺q 2 ⫺i ␶ , The expressions in Eq. 共39兲 are only valid in the region
␥ R1,R2 r
where Re(np1,p2 )⬎0 and Re(ns1,s2 )⬎0. Other stress compo-

冋 册
共37兲
1 z nents in the solid can be derived from the following stress–
n s1,2⫽ ⫾ 冑␥ R1,R2
2
⫺1⫺i ␶
␥ R1,R2 r displacement relations:
yields
关 ⫺ ␣ 2 z⫹ pt 兴 p⫽i ␰ /s S ␥ R1,R2
␶ rr2 ⫽ 共 ␭ 2 ⫹2 ␮ 兲
⳵u2
⳵r
⫹␭ 2
u2 ⳵w2
r
⫹
⳵z
, 冉 冊
冉 冊
共41兲

冋 册
u2 ⳵u2 ⳵w2
r z ␶ ␪␪ 2 ⫽ 共 ␭ 2 ⫹2 ␮ 兲 ⫹␭ 2 ⫹ .
⫽⫺ ␰ ⫾ 冑␥ R1,R2
2
⫺q 2 ⫺i ␶ ⫽⫺ ␰ rn p1,2 , r ⳵r ⳵z
␥ R1,R2 r
共38兲
关 ⫺ ␤ z⫹pt 兴 p⫽i ␰ /s S ␥ R1,R2 C. Attenuation and dispersion of leaky Rayleigh
waves

⫽⫺ ␰
r
␥ R1,R2
2
冋
⫾ 冑␥ R1,R2
z
⫺1 ⫺i ␶ ⫽⫺ ␰ rn s1,2 .
r 册 In addition to the geometric decay due to the effect of
point loading, which varies as 1/冑r along the interface for
The real part of Eq. 共38兲 must be negative to have bounded the Rayleigh wave, there is another type of attenuation
responses, therefore the real parts of n p1 , n p2 , n s1 , n s2 must caused by continuous radiation 共leakage兲 of energy into the
be positive. Using the similar argument for dealing with fluid. In frequency domain, the solutions are exponential
m 1,2 , only those results of n p1 , n p2 , n s1 , n s2 that have functions of (⫺ ␰ r), where ␰ has the physical meaning of
positive real parts are acceptable. In addition, n p1 , n p2 and wavenumber. According to Eq. 共32兲, higher frequency 共larger
n s1 , n s2 should be complex conjugate pairs. ␰兲 contents give more attenuation during propagation. There-
Applying the inverse Hankel and Laplace transforms to fore the waveform generated by a transient loading becomes
Eq. 共14兲 and calculating the residues at Rayleigh poles yields wider with increasing distance, i.e., it shows dispersion prop-
the displacements and stresses in the solid, erty due to leakage-induced attenuation, although the phase

冋 册
velocity of leaky Rayleigh waves does not vary with fre-
Q s 2S 1 B1 B2 quency.
w 2⫽ Re ⫺ ,
␲ ␳2 r 冑 1⫹n 2p1 冑1⫹n s1
2

u 2⫽
Q s 2S 1
␲ ␳2 r
Re B 3 1⫺
n p1
冋冉
冑1⫹n 2p1 冊 VII. SCHOLTE WAVE RESPONSE

The real roots of the Scholte equation correspond to

⫺B 4 1⫺ 冉 n s1
冑1⫹n s1
2 冊册 ,
Scholte wave propagation along the interface. The Scholte
wave solutions can be obtained by calculating the residues at
the poles ␥ ⫽⫾ ␥ sch . For common cases of light fluids lying
共39兲
J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 Zhu et al.: Leaky Rayleigh and Scholte waves 2107
on stiff solids, the root 兩 ␥ sch 兩 ⫽s sch /s S is slightly larger than
u⫽s F /s S . Therefore, Eqs. 共31兲 and 共38兲 are changed to
关 ␣ 1 z⫹ pt 兴 p⫽i ␰ /⫾s S ␥ sch ⫽⫺ ␰ rm 1,2
sch
,

关 ⫺ ␣ 2 z⫹pt 兴 p⫽i ␰ /⫾s S ␥ sch ⫽⫺ ␰ rn sch

p1,2 , 共42兲

关 ⫺ ␤ z⫹ pt 兴 p⫽i ␰ /⫾s S ␥ sch ⫽⫺ ␰ rn s1,2

sch
,
where

sch
m 1,2 ⫽⫺
1
␥ sch 冉冑␥ sch
2
z
⫺u 2 ⫾i
r
t
s Sr
, 冊
p1,2⫽
n sch
1
␥ sch 冉冑 ␥ sch
2
z
⫺q 2 ⫾i
r
t
s Sr
,冊 共43兲

sch
n s1,2 ⫽
1
␥ sch 冉冑 ␥ sch
2
z
⫺1 ⫾i
r
t
s Sr冊.

The solutions for the leaky Rayleigh wave are also valid for
the Scholte wave by substituting m sch , n sch sch
p , and n s for m,
n p , and n s . The first terms in the expressions of Eq. 共43兲
represent the real parts, which are positive and result in the
decay in z direction. It can be seen that the Scholte wave
decays exponentially in z direction in both the fluid and the
solid. For lighter fluid cases, i.e., where the acoustic imped-
ance of the fluid is less than that of the solid, 冑␥ sc2
⫺u 2 is
much smaller than 冑␥ sc 2
⫺q 2 and 冑␥ sc
2
⫺1. This indicates
that the Scholte wave attenuates much more quickly in the
solid than in the fluid. Therefore, in contrast to the leaky
Rayleigh wave, most of Scholte wave energy is localized in
the fluid 共Gusev et al., 1996兲. The Scholte wave generation
efficiency increases with increasing acoustic impedance of
the fluid. For example, it is much easier to generate Scholte
waves in the water/concrete configuration than the air/ FIG. 6. Comparison of the exact and simplified solutions for water/concrete
concrete configuration. In fact, almost no Scholte wave effect case. Pressure in the fluid at position 共a兲 r⫽1.5 m, z⫽⫺0.05 m, and 共b兲 r
⫽1.5 m, z⫽⫺0.5 m. The interface is subject to a point load that varies with
can be observed in the air/concrete case. With increasing time as f (t)⫽sin2(␲t/T) with T⫽200 ␮ s. Material parameters for the fluid
impedance of the fluid, Scholte waves have deeper penetra- ␳ 1 ⫽1000 kg/m3 , c F ⫽1500 m/s; and for the solid ␳ 2 ⫽2400 kg/m3 , c P
tion depth in the solid. This property provides the possibility ⫽4000 m/s, v ⫽0.25.
for NDT application of Scholte waves, which was studied
experimentally by Glorieux et al. 共2001兲. Because there is no
leakage during Scholte wave propagation along the radial simplified and exact analytical solutions decreases. Generally
direction, the decay in the radial direction is only attributed speaking, the simplified solution provides better estimation
to the geometrical effect. In 2-D cases, the Scholte wave of pressure for larger r and smaller 兩 z 兩 cases, where the con-
travels without attenuation along the propagation direction tribution of body waves is negligible.
共Glorieux et al., 2001兲. Figures 7共a兲 and 共b兲 show the comparison of the exact
and simplified solutions for the air/concrete case. The re-
ceiver positions are r⫽1.0 m, z⫽⫺0.05 m and r⫽1.0 m, z
VIII. COMPARISON OF THE EXACT AND SIMPLIFIED
⫽⫺0.5 m, respectively. Good agreement is observed near
SOLUTIONS
the leaky Rayleigh wave arrival time for both positions,
Figure 6 shows the comparison of the exact and simpli- while obvious differences can be seen near the Scholte and
fied analytical solutions for fluid pressure for the water/ fluid acoustic wave arrival times. The reason is that the leaky
concrete case. In Fig. 6共a兲, when the receiver position is Rayleigh wave is well separated from the fluid acoustic
close to the interface (r⫽1.5 m, z⫽⫺0.05 m), good agree- wave, and Scholte wave contribution is very small compared
ment is observed around the leaky Rayleigh and Scholte to the acoustic wave contribution in the fluid for air/concrete
wave arrival times. The small yet noticeable differences be- case, even in the near-interface region. Therefore, for the
fore the leaky Rayleigh and Scholte wave arrivals are due to air/concrete configuration, the pressure field in the fluid is
the absence of leaky body waves and fluid acoustic waves in dominated by leaky Rayleigh and fluid acoustic waves. In
the simplified solution. When the receiver is away from the air-coupled sensing, the leaky Rayleigh wave is usually the
interface, as shown in Fig. 6共b兲 for receiver position r component in which we are interested. The acoustic wave
⫽1.5 m, z⫽⫺0.5 m, the degree of agreement between the contribution in the fluid can be separated by increasing mea-

2108 J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 Zhu et al.: Leaky Rayleigh and Scholte waves
 Hoop method. Simplified formulations are also derived,
which provide an easy and quick way to estimate leaky Ray-
leigh and Scholte wave contributions. The following conclu-
sions can be drawn based on the analysis:
共1兲 A transient point load applied to the interface is an ef-
fective way to generate leaky Rayleigh waves in the
fluid. For air-coupled wave detection in concrete, the ex-
citation effectiveness of leaky Rayleigh waves is around
0.1–1.0 Pa/kN, depending on the impact force duration.
共2兲 For the light fluid/heavy solid case, the leaky Rayleigh
wave is separable from Scholte and acoustic waves in
the fluid when distance r is large enough, where r de-
pends on velocities of leaky Rayleigh, Scholte, and
acoustic waves, vertical distance 兩 z 兩 , and force duration.
For the air–concrete configuration shown in Fig. 7共a兲,
where 兩 z 兩 ⫽0.05 m and r⬎0.2 m, the difference in arrival
time between leaky Rayleigh and acoustic waves is
⬎362 ␮s. Therefore the received signals will be domi-
nated by leaky Rayleigh waves, which provide important
material information of the underlying solid.
共3兲 Simplified solutions are obtained when contributions
from leaky Rayleigh waves and Scholte waves poles
only are considered. Equations 共34兲–共36兲 and 共39兲 give
the solution to responses in the fluid and solid, respec-
tively. The simplified solution accurately simulates the
transient pressure field response for the air/concrete case
when the fluid acoustic wave contribution is removed or
separated.
共4兲 The Scholte wave contribution is prominent in the near-
interface region for the water/concrete case. Because
most of the energy of Scholte waves is localized in the
FIG. 7. Comparison of the exact and simplified solutions for air/concrete fluid, however, Scholte wave properties are not very sen-
case. Pressure in the fluid at position 共a兲 r⫽1.0 m, z⫽⫺0.05 m, and 共b兲 r
⫽1.0 m, z⫽⫺0.5 m. The interface is subject to a point load that varies with
sitive to the variation of the underlying solid materials,
time as f (t)⫽sin2(␲t/T) with T⫽200 ␮ s. Material parameters for the fluid which limits the NDE application of Scholte waves for
␳ 1 ⫽1.21 kg/m3 , c F ⫽343 m/s; for the solid ␳ 2 ⫽2400 kg/m3 , c P the common light fluid/heavy solid cases.
⫽4000 m/s, ␯ ⫽0.25.

suring distance r between receivers and the source, or elimi- ACKNOWLEDGMENTS

nated by employing the directional property of the air- This work was carried out in the course of research
coupled sensor 共Zhu et al., 2001兲. sponsored by the National Science Foundation under Grant
The excitation effectiveness of leaky Rayleigh waves No. 0223819. The authors also wish to thank Dr. Nelson Hsu
induced by an impact point load can be inferred from Fig. 7. from National Institute of Standards and Technology for his
For a typical impulse force f (t)⫽sin2(␲t/T) with a modest helpful suggestions.
peak value of 1 kN and duration T⫽200 ␮ s, the output pres-
sure of the leaky Rayleigh wave is 0.1–0.15 Pa, which is
Achenbach, J. D. 共1973兲. Wave Propagation in Elastic Solids 共North-
approximately equivalent to a sound pressure level of 75 dB. Holland, Amsterdam, The Netherlands兲, Chap. 7, pp. 262–321.
Such a pressure is large enough to be detected readily by an Castaings, M., and Hosten, B. 共2001兲. ‘‘Lamb and SH waves generated and
air-coupled sensor, even when material attenuation effects detected by air-coupled ultrasonic transducers in composite material
plates,’’ NDT & E Int. 34, 249–258.
are considered. The excitation effectiveness of leaky Ray- Chao, C. C. 共1961兲. ‘‘Surface waves in an elastic half space,’’ J. Appl. Mech.
leigh waves is dependent on the impact force duration. 28, 300–301.
Shorter force durations give higher output pressure of leaky de Hoop, A. T., and van der Hijden, J. 共1983兲. ‘‘Generation of acoustic
Rayleigh waves. For example, when the duration is de- waves by an impulsive line source in a fluid/solid configuration with a
plane boundary,’’ J. Acoust. Soc. Am. 74, 333–342.
creased to T⫽50 ␮ s, the output peak pressure of the leaky de Hoop, A. T., and van der Hijden, J. 共1984兲. ‘‘Generation of acoustic
Rayleigh wave will increase to 1.2 Pa 共95 dB兲. Most impac- waves by an impulsive point source in a fluid/solid configuration with a
tors used for concrete testing will generate transient forces plane boundary,’’ J. Acoust. Soc. Am. 75, 1709–1715.
with duration between 50 and 200 ␮s. Fellinger, P., Marklein, R., Langenberg, K. J., and Klaholz, S. 共1995兲. ‘‘Nu-
merical modelling of elastic wave propagation and scattering with EFIT—
IX. CONCLUSIONS Elastodynamic finite integration technique,’’ Wave Motion 21, 47– 66.
Glorieux, C., and Van de Rostyne, K. 共2001兲. ‘‘On the character of acoustic
The exact analytical solution to the ‘‘Lamb’’ problem in waves at the interface between hard and soft solids and liquids,’’ J. Acoust.
a fluid/solid half-space system is derived by the Cagniard–de Soc. Am. 110, 1299–1306.

J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 Zhu et al.: Leaky Rayleigh and Scholte waves 2109
Gusev, V., Desmet, C., Lauriks, W., Glorieux, C., and Thoen, J. 共1996兲. gration Technique 共EFIT兲,’’ in Proc. IEEE Ultrasonics Symp., Munich,
‘‘Theory of Scholte, leaky Rayleigh, and lateral wave excitation via the Germany, 8 –11 October, 2002, Article 5G-5, 778 –783 共on CD兲.
laser-induced thermoelastic effect,’’ J. Acoust. Soc. Am. 100, 1514 –1528. Schubert, F., Peiffer, A., Koehler, B., and Sanderson, T. 共1998兲. ‘‘The elas-
Liu, Q. H. 共1999兲. ‘‘Perfectly matched layers for elastic waves in cylindrical todynamic finite integration technique for waves in cylindrical geom-
and spherical coordinates,’’ J. Acoust. Soc. Am. 105, 2075–2084.
etries,’’ J. Acoust. Soc. Am. 104, 2604 –2614.
Mozhaev, V. G., and Weihnacht, M. 共2002兲. ‘‘Subsonic leaky Rayleigh
Viktorov, I. A. 共1967兲. Rayleigh and Lamb Waves 共Plenum, New York兲.
waves at liquid-solid interfaces,’’ Ultrasonics 40, 927–933.
Qi, Q. 共1994兲. ‘‘Attenuated leaky Rayleigh waves,’’ J. Acoust. Soc. Am. 95, Zhu, J., and Popovics, J. S. 共2001兲. ‘‘Non-contact detection of surface waves
3222–3230. in concrete using an air-coupled sensor,’’ in Review of Progress in Quan-
Schubert, F., and Marklein, R. 共2003兲. ‘‘Numerical Computation of Ultra- titative Nondestructive Evaluation 共American Institute of Physics,
sonic Wave Propagation in Concrete using the Elastodynamic Finite Inte- Melville, NY兲, Vol. 20B, 1261–1268.

2110 J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 Zhu et al.: Leaky Rayleigh and Scholte waves

View publication stats