Simulation of Leaky Rayleigh Wave at Air-Solid Cylindrical Interfaces by Finite Element Method

http://www.paper.edu.
cn
Ultrasonics 44 (2006) e1169e1172
Simulation of leaky Rayleigh wave at airsolid cylindrical interfaces

by nite element method
Yan Zhao, Zhonghua Shen *, Jian Lu, Xiaowu Ni
School of science, Nanjing University of Science and Technology, Xiaolingwei 200#, Nanjing 210094, PR China
Available online 9 June 2006
Abstract
The nite element method is used to simulate the laser-excited leaky Rayleigh wave at airsolid cylindrical interfaces. A whole arithmetic of uidsolid interaction is presented, which includes a coupling matrix that describing the process of the interaction between uid
and solid, the Arbitrary LagrangianEulerian (ALE) formulation for treating the variation of uid domain, which results from the Rayleigh wave propagating on the cylindrical interface, etc. Typical calculation is executed and the results show that the polarity of leaky
Rayleigh waveform gradually changes as it propagates on the airsolid cylindrical interface.
2006 Elsevier B.V. All rights reserved.
Keywords: Airsolid cylindrical interface; Finite element method; Leaky Rayleigh wave
1. Introduction
In the last few decades, leaky waves at uidsolid plane
interface (leaky Rayleigh and leaky Lamb wave) have
attracted extensive attention for their potential application
in nondestructive test and characterization of materials [1
3]. The development of laser ultrasonic technique provides
an eective solution to investigate the leaky waves on the
uidsolid interface. In recent years, some researchers
begin to study the leaky waves by laser ultrasonic method.
For example, Desmet et al. detected the laser-induced leaky
Rayleigh waves on watermetal interfaces, and they realized all-optical excitation and detection of leaky Rayleigh
waves by means of the laser ultrasonic technique [3]. However, these works mainly concentrated on the plane structure, and less attention was attached to the leaky
Rayleigh waves on uidsolid cylindrical interfaces [13].
As a result, the main objective of this paper is to investigate
the laser-induced leaky Rayleigh waves on the airsolid
cylindrical interface. Except for experiment, numerical cal*
Corresponding author. Tel.: +86 2584315075; fax: +86 2584315699.

E-mail addresses: zhaoyan7906@sina.com (Y. Zhao), shenzh@mail.
njust.edu.cn (Z. Shen).
culation is another eective solution to solve the problem

of acoustic wave, such as nite element method, nite difference method, and so on.
In this research, the nite element method is introduced
to simulate the laser-excited leaky Rayleigh wave on air
solid cylindrical interface because it has two prominent
advantages: one is the ability to model laser-induced ultrasonic under any conditions. The other is the convenience to
take the temperature dependence of physical parameters
into account.
2. Finite element model
2.1. Conguration
Consider a homogenous and isotropic cylinder of radius
a, and density q, which is placed in air. As depicted in Fig. 1,
the symmetric axis of the cylinder coincides with the z-axis
of our cylindrical coordinates (r, h, z). Using a cylindrical
lens, the pulse laser beam was focused onto the surface of
solid cylinder to form a line source, whose orientation is
along the z direction. Owing to the symmetry imposed by
the source shape, this problem shows invariance along the
z direction. As a result, the nonzero components of the
0041-624X/$ - see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.ultras.2006.05.177
http://www.paper.edu.cn
e1170
Y. Zhao et al. / Ultrasonics 44 (2006) e1169e1172
a
Air
Solid
cylinder
Laser
beam
Fig. 1. Conguration for the laser-induced excitation of leaky Rayleigh

waves at the airsolid cylindrical interface.
2.4.1. Fluidsolid interaction

The generation of leaky Rayleigh wave can be considered as the result of the uid and solid interaction at the
interface between solid cylinder and uid. The interaction
causes the acoustic pressure to exert a force applied to
the solid and the solid motions produce an eective uid
load. The governing nite element matrix equations then
become
displacement vector only depends on two spatial variables r,

h and on time t.
g H solid fU g fF solid g RfP g;

M solid fU
g;
M air fP g H air fP g fF air g q0 RT fU
2.2. Heat conduction
[R] is a coupling matrix that represents the eective surface area associated with each node on the uidsolid interface. The coupling matrix [R] also takes into account the
direction of the normal vector dened for each pair of coincident uid and solid element faces that comprises the
interface surface. The positive direction of the normal vector, in our calculation, is dened to be outward from the
uid mesh and in towards the solid. Both the solid and
uid load quantities that are produced at the uid-structure
interface are functions of unknown nodal degrees of freedom. Placing these unknown load quantities on the left
hand side of the equations and combining the two equations into a single equation produces the following:

( )

U
H solid R
F solid
M solid 0
U
;
F air
P
q0 RT M air
0
H air
P
For laser-induced heat transfer, convection can be

neglected [4], the classical heat conduction equation for
nite elements can be expressed as [5]
KfT g CfT_ g fpq g fpQ g;
where {T} is the temperature vector, fT_ g is the temperature

rate vector, [K] is the conductivity matrix, [C] is the heat
capacity matrix, {pq} is the heat ux vector and
R {pTQ} is
}
is
N qdS,
the heat source
vector.
For
an
element,
{p
q
S
R
T
and {pQ} is V N QdV , where S and V are the area and
volume of an element, respectively; [N]T is the transpose
of the shape function; q is the heat ux and Q is the heat
source.
2.3. Wave equation
g is the accelerawhere {U} is the displacement vector, fU

tion vector, [Msolid] is the mass matrix, [Hsolid] is the stiness matrix and {Fsolid} is the external force vector. For
ultrasonic propagations in elastic uid medium, the governing equation can be written as
M air fP g H air fP g fF air g;
For the generation and propagation of ultrasonic, ignoring damping, the NavierStokes equation can be expressed
as
g H solid fU g fF solid g;
M solid fU
where {P} is the acoustic pressure vector.

2.4. Solution schemes
A lumping scheme [5] coupled with the central-dierence
integration method, which is most applicable to problems
of large size and is a compromise between accuracy and
eciency, is used to solve the equations for the uid and
solid domains independently of each other. Then, uid
forces and heat uxes and solid displacements, velocities,
and temperatures are transferred across the uidsolid
interface. The algorithm continues to loop through the
solid and uid analyses until convergence is reached for
that time step (or until the maximum number of stagger
iterations is reached).
This equation implies that nodes on a uid-structure interface have both displacement and pressure degrees of
freedom.
2.4.2. Interface determination
Ultrasonic waves propagating in the surface of solid
make its surface deform, which lead to the uid domain
change with time. As a result, the nite element mesh must
move to satisfy the boundary conditions in the nite element calculation. The Arbitrary LagrangianEulerian
(ALE) formulation is taken to solve this problem. The general concept of the ALE formulation is that an arbitrary
referential domain is dened for the description of motion
that is dierent from the material (Lagrangian) and spatial
(Eulerian) domains.
In a pure Lagrangian system, the mesh deforms with the
material being modeled so that there is no material ow
between elements. The main disadvantage of the Lagrangian approach is that problems develop in physical situations that involve highly deformed surfaces. Other
disadvantages of the Lagrangian approach are that only
one material can be modeled in each element and that
new surfaces cannot be created.
In an Eulerian based formulation, the mesh is stationary, the material ows through the mesh, and new surfaces
are automatically created. However, the greatest disadvantage of the Eulerian approach is that a ne mesh is required
to capture the material response, making the method very

computationally expensive.
The Arbitrary LagrangianEulerian (ALE) approach is
a very eective alternative. In its most basic sense, the
ALE method denes that the mesh motion is independent
of the motion of the material being analyzed. Although
the mesh motion may be arbitrary, it typically deforms
with the material in near Lagrangian ow elds. The
greatest advantage of the ALE method is that it allows
smoothing of a distorted mesh without performing a complete remesh. This smoothing allows the free surface of the
material to be followed automatically without encountering the distortional errors of the Lagrangian approach.
The main diculty of the ALE method is the path dependent behavior of the plastic ow being modeled. Due to the
path dependence, the relative motion between the mesh
and the material must be accounted for in the material constitutive equations. In addition, the ALE method does not
allow new surfaces to be created and is limited to geometries where the material ow is relatively predictable.
2.4.3. Finite element mesh and time step
The selection of nite element size and time step are critical for the stability and accuracy of nite element calculation. In our calculation, the nite element sizes Le for both
solid and uid domain and time step are determined by the
criterions suggested by Schubert et al. [6]
1 C
;
10 fmax
Le
Dt 6 p ;
C 3
Le 6
e1171
the radius of the laser pulse spot on the interface are

10 ns and 300 lm, respectively. The radius of aluminum
cylinder is taken to be 4 mm, and the physical parameters
of air and aluminum used in the calculation are listed in
Table 1.
3.2. Typical results
Fig. 2 shows the leaky Rayleigh waveform recorded at
the interface 90 away from the source. Compared this calculated waveform with the Rayleigh waveform on the free
surface of a solid, as expected, very good agreements are
observed in the time, shape, and relative amplitude. LR1
and LR2 represent the rst leaky Rayleigh pulse and the
second one which propagate respectively 1/4 and 3/4 turn
on the cylindrical interface. Leaky Rayleigh wave on a
uidsolid interface is an elastic wave corresponding to
the Rayleigh wave on the free surface of a solid that
becomes damped in its direction of propagation because
of seepage of energy in the liquid. In addition, leaky whispering gallery modes (LWG) corresponding to the whispering gallery modes on the free surface of a solid are also very
obvious.
Thanks to the advantage of nite element method, the
whole eld data can be obtained in a simulation, we also
obtained the waveforms of leaky Rayleigh wave obtained
at dierent angles h between the excited source and the
recording point, as shown in Fig. 3. Compared those waveforms in Fig. 3, one can easily nd an interesting result that
8
6
LR1
where C represents the highest wave speed of the medium,

fmax is the highest frequency in the ultrasonic wave, which
can be evaluated by [7]
p
2 2C
;
9
fmax
pr0
LR2
Normal displacement (nm)
where r0 is the radius of the laser pulse spot.

3. Numerical simulation and results
3.1. Laser and material parameters
LWG
-2
-4
Using above nite element arithmetic of uidsolid

interaction, we have simulated the laser-induced leaky
Rayleigh wave at the cylindrical interface between air and
aluminum cylinder. The parameters of laser and material
were chosen as following: the laser pulse rise time t0, and
-6
0
Time (s)
Fig. 2. Leaky Rayleigh waveform recorded at the airaluminum cylinder

cylindrical interface 90 away from the source.
Table 1
Physical parameters of air and aluminum used in the nite element calculation (T = 300 K)
Al
Air
Thermal
diusivity (K1)
Specic heat
(J kg1 K1)
Heat conductivity
(W m1 K1)
Absorption
coecient (m1)
Density
(kg m3)
Velocity of
longitudinal wave (m s1)
Velocity of
transverse wave (m s1)
71.8 106
37.2 103
896
1005
237
0.0261
103
103
2700
1.21
6260
343
3080
e1172
tive). Then, the leaky Rayleigh waveform changes, it

becomes bipolar at h = 90 and monopolar (positive) at
h = 180. The change of polarity results from the dispersion eect of cylindrical interface. In this gure, the pulses
for the clockwise and counterclockwise propagation merge
into a single peak at h = 180 so that the amplitude is
increased twofold.
30
0.3
0.0
-0.3
-0.6
0
45
0.3
0.0
-0.3
-0.6
0
4. Conclusions
60
0.3
0.0
-0.3
-0.6
0
Normal displacement (nm)
0.6
0.3
0.0
-0.3
-0.6
90
0.6
0.3
0.0
-0.3
120
0.6
0.3
0.0
-0.3
Acknowledgement
This work is supported by the Young Scholars Foundation of the Nanjing University of Science and Technology
(Grant NJUST200503).
135
0.6
0.3
0.0
-0.3
150
1.2
0.6
0.0
-0.6
References
180
A nite element model is developed to simulate the leaky

Rayleigh wave at the airsolid cylindrical interface induced
by pulse laser. The whole process of uidsolid interaction
has been formulated by nite element equations. According
to nite element calculation, typical leaky Rayleigh waveforms are obtained and they indicate that the polarity of
leaky Rayleigh waveform gradually changes as it propagates on the airsolid cylindrical interface.
Time (s)
Fig. 3. Leaky Rayleigh waveforms recorded on the airaluminum cylinder

cylindrical interface versus angle h from 30 to 180.
the polarity of leaky Rayleigh waveform gradually changes

as it propagates on the airsolid cylindrical interface. Near
the source, the leaky Rayleigh pulse is monopolar (nega-
[1] C. Mattei, L. Adler, Ultrasonics 38 (2000) 570.

[2] V. Gusev, C. Desmet, W. Lauriks, C. Glorieux, J. Thoen, J. Acoust.
Soc. Am. 100 (3) (1996) 1514.
[3] C. Desmet, V. Gusev, W. Lauriks, C. Glorieux, J. Thoen, Opt. Lett. 22
(2) (1997) 69.
[4] J.B. Spicer, Laser ultrasonics in nite structures: comprehensive
modeling with supporting experiment. Ph.D. thesis, Johns Hopkins
University, 1991.
[5] R.D. Cook, D.S. Malkus, M.E. Plesha, Concepts and Applications of
Finite Element Analysis, third ed., John Wiley, 1989.
[6] F. Schubert, B. Koehler, A. Peier, J. Comput. Acoust. 9 (2001) 1127.
[7] Y. Sohn, S. Krishnaswamy, Ultrasonics 39 (2002) 543.

Simulation of Leaky Rayleigh Wave at Air-Solid Cylindrical Interfaces by Finite Element Method

Uploaded by

Copyright:

Available Formats

Simulation of Leaky Rayleigh Wave at Air-Solid Cylindrical Interfaces by Finite Element Method

Uploaded by

Document Information

Original Description:

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

Simulation of Leaky Rayleigh Wave at Air-Solid Cylindrical Interfaces by Finite Element Method

Uploaded by

Copyright:

Available Formats

http://www.paper.edu.

Ultrasonics 44 (2006) e1169e1172

Simulation of leaky Rayleigh wave at airsolid cylindrical interfaces

Corresponding author. Tel.: +86 2584315075; fax: +86 2584315699.

culation is another eective solution to solve the problem

Y. Zhao et al. / Ultrasonics 44 (2006) e1169e1172

Fig. 1. Conguration for the laser-induced excitation of leaky Rayleigh

2.4.1. Fluidsolid interaction

displacement vector only depends on two spatial variables r,

g H solid fU g fF solid g RfP g;

2.2. Heat conduction

For laser-induced heat transfer, convection can be

where {T} is the temperature vector, fT_ g is the temperature

g is the accelerawhere {U} is the displacement vector, fU

where {P} is the acoustic pressure vector.

to capture the material response, making the method very

the radius of the laser pulse spot on the interface are

where C represents the highest wave speed of the medium,

Normal displacement (nm)

where r0 is the radius of the laser pulse spot.

Using above nite element arithmetic of uidsolid

Fig. 2. Leaky Rayleigh waveform recorded at the airaluminum cylinder

Y. Zhao et al. / Ultrasonics 44 (2006) e1169e1172

tive). Then, the leaky Rayleigh waveform changes, it

Normal displacement (nm)

A nite element model is developed to simulate the leaky

Fig. 3. Leaky Rayleigh waveforms recorded on the airaluminum cylinder

the polarity of leaky Rayleigh waveform gradually changes

[1] C. Mattei, L. Adler, Ultrasonics 38 (2000) 570.

You might also like