Simulation of Lamb waves using the spectral cell method

S. Duczeka , M. Joulaianb , A. Düsterb and U. Gabberta

a Otto-von-Guericke-University Magdeburg, Universitätsplatz 2, 39106 Magdeburg, Germany;
b Hamburg University of Technology, Schwarzenbergstraße 95c, 21073 Hamburg, Germany.

ABSTRACT
Today a steadily growing interest in on-line monitoring of structures is seen. Commonly referred to as structural
health monitoring (SHM), the basic idea of this technique is to decrease the maintenance costs based on a
continuous flow of information concerning the state of the structure. With respect to the aeronautic industry
increasing the service time of airplanes is another important goal. A popular approach to SHM is to be seen
in ultrasonic guided wave based monitoring systems. Since one focus is on typical lightweight materials elastic
waves seem to be a viable means to detect delimitations, cracks and material degradation. Due to the complexity
of such structures efficient numerical tools are called for. Several studies have shown that linear or quadratic
pure displacement finite elements are not appropriate to resolve wave propagation problems. Both the mesh
density and the spatial resolution needed to control the numerical dispersion are prohibitively large. Therefore,
higher order finite element methods (p-FEM, SEM) are considered by the authors.
One important goal is to simulate the propagation of guided ultrasonic waves in carbon/glass fiber reinforced
plastics (CFRP, GFRP) or sandwich materials. These materials are typically deployed in aeronautical and
aerospace application and feature a complex micro-structure. This micro-structure, however, needs to be resolved
in order to capture effects like transmission, reflection and conversion of the different wave modes. It is known
that using standard discretization techniques it is almost impossible to mesh the aforementioned heterogeneous
materials without accepting enormous computational costs. Therefore, the authors propose to apply the finite
cell method (FCM) and extend this approach by using Lagrange shape functions evaluated at a Gauss-Lobatto-
Legendre grid. The latter scheme leads to the so called spectral cell method (SCM). Here, the meshing effort is
shifted towards an adaptive integration technique used to determine the cell matrices and load vectors. Hence,
a rectangular Cartesian grid can be used, even for the most complex structures.
The functionality of the proposed approach will be demonstrated by studying the Lamb wave propagation in a
two-dimensional plate with a circular hole. The perturbation is not symmetric with respect of the middle plane
in order to introduce mode conversion. In the paper, an efficient method to simulate the elastic wave propagation
in heterogeneous media utilizing the finite or spectral cell method is presented in detail.
Keywords: p-FEM, SEM, FCM, SCM, Fictitious Domain Method, Lamb Waves, SHM.

1. INTRODUCTION
Reliable damage detection plays a vital role for structural health monitoring (SHM) systems. Regarding thin-
walled structures, techniques based on the propagation of ultrasonic guided waves seem to be a viable choice. 1–3
Since the aeronautic industry is a driving force behind developments concerning SHM the focus in research
activities has shifted from relatively simple aluminum to more complex composite and sandwich plates. 4–6 To
compute the wave propagation in these highly heterogeneous structures is a rather demanding task. Both, the
spatial as well as the temporal discretization have to be fine in order to resolve the high frequency Lamb waves
and to capture the small wavelength. A guideline on how to choose the optimal polynomial degrees and the
corresponding finite element size to reduce the numerical costs has been published by Willberg et al.7 In this
article the authors deal with three different higher order finite element approaches, namely the spectral ele-
ment method8 (SEM), the p-version of the finite element method 9, 10 (p-FEM) and the isogeometric analysis 11
(IGA). It is shown that when using higher order finite element methods significant savings in memory storage
Send correspondence to S. Duczek and A. Düster
S. Duczek: E-mail: sascha.duczek@st.ovgu.de, Telephone: +49 (0)391 67 12754
A. Düster: E-mail: alexander.duester@tu-harburg.de, Telephone: +49 (0)40 42878 6083

Health Monitoring of Structural and Biological Systems 2013, edited by Tribikram Kundu, Proc. of SPIE
Vol. 8695, 86951U · © 2013 SPIE · CCC code: 0277-786X/13/$18 · doi: 10.1117/12.2009983

Proc. of SPIE Vol. 8695 86951U-1

requirements and numerical costs can be made compared to commercially available FEM packages based on the
h-version.
The widely accepted use of the spectral element method in dynamic applications is due to the fact that mass
lumping can be easily achieved. Dauksher and Emery 12, 13 demonstrated that a row-summing procedure can
result in accurate solutions with spectral elements. This row-summing procedure is only performed on the global
mass matrix. Nonetheless, the solution characteristics are relatively unaffected by the diagonalized mass matrix
but the numerical costs for the solution of the equation system are drastically reduced when applying an explicit
time-integration scheme. The authors opt for deploying row-summed mass matrix spectral elements with central
time differencing to solve the wave equation.
Despite their favourable numerical properties the aforementioned methods (p-FEM, SEM, IGA) cannot be rec-
ommended without restrictions. Considering heterogeneous materials with a complex micro-structure, such as
sandwich panels with a hollow sphere or foam core, the user is still faced with the problem of mesh generation.
It is widely known that generating conforming finite element grids from complex CAD based geometrical models
is rather computationally expensive and is hard, if at all possible, to fully automate. Normally the produced
mesh has to be adjusted manually by the user. Since immersed boundary methods do not demand body-fitted
meshes14 but divide the domain in a regular grid of Cartesian cells the geometric complexity does not cause
difficulties during the discretization process. That is to say, that the desired hexahedral meshes can be generated
on the basis of measurements given by X-ray tomography. 15–17 Thus, no user interaction is required and the
procedure can be fully automated. Consequently, to circumvent the time consuming discretization the authors
propose to use the finite cell method to solve Lamb wave propagation problems.
The FCM was first proposed by Parvizian et al.15 and then successfully applied to different classes of static and
eigenvalue problems.14, 16–19 The simple yet effective idea behind this method is to extend the partial differential
equation beyond the physical domain of computation up to the boundaries of an embedding domain, which can
be meshed without difficulty. Therefore, the finite element mesh can be replaced by structured cells embedding
the whole domain. By doing so the manual effort of discretizing a structure is shifted to the numerical costs of
an adaptive integration scheme. Noteworthy applications so far include the modal analysis of the cover plate of a
violin,19 the homogenisation of cellular and foam structures 16 and the analysis of a femur bone 18 among others.
The main objective of this paper is to extend the finite cell method to structural dynamics, especially the sim-
ulation of elastic guided waves in thin-walled structures used in aerospace and automotive applications is of
interest to the authors. To this end, the so called spectral cell method (SCM) is introduced. The term is chosen
analogously to the differentiation between FEM and SEM. We introduce Lagrange shape functions on a Gauss-
Lobatto-Legendre (GLL) grid in conjunction with the GLL quadrature for uncut cells and row-summing for cut
cells. Thus, we retain all advantages of higher order finite element methods (accuracy, convergence rate) and the
ease of discretization connected to a fictitious domain method.
To this end, the current contribution is structured as follows. In the 2 nd section the fundamentals of the spectral
element method are briefly recalled. Thereafter, the 3 rd section features the basic principles of finite cell method
and its extension to the spectral cell method. The 4 th section is concerned with their application to a simple
two-dimensional wave propagation problem in a perturbed structure. Finally, the paper is concluded and an
outlook to on going research is given in the 5 th section.

2. SPECTRAL ELEMENT METHOD

In the following section the fundamentals of the finite and spectral element method are briefly recalled. Both
methods differ only in the choice of Ansatz functions. All fundamental principles are identical and thus we do
not promote the notion to differentiate between those two methods based on the different shape functions that
are being deployed.

2.1 Fundamental Equation

The point of departure to develop finite elements is the variational formulation corresponding to Navier’s equa-
tions, namely Hamilton’s principle. It states that the motion of the system within the time interval [ t1 , t2 ] is
such, that under infinitesimal variations of the displacements the Hamiltonian action S vanishes, meaning that
the motion of the system takes the path of the stationary action. 20 After some calculus the following equation

Proc. of SPIE Vol. 8695 86951U-2

is obtained, describing the equation of motion in a discrete form

MÜ + KU = f . (1)

Here, the mass matrix M, the stiffness matrix K and the load vector of the external forces f , are introduced.
The displacement field u(x, t) is approximated by the product of the space-dependent shape function matrix
N(x) and the time-dependent vector of unknowns U(t)

u(x, t) = N(x) ∙ U(t). (2)

For further explanations the reader is referred to standard text books on this subject by Zienkiewicz, 21–23 Bathe24
and Hughes,25 for instance.

2.2 Spectral Element Method

Spectral finite elements are based on a set of Lagrange polynomials utilizing a specific nodal distribution on the
interval [−1, +1]. The (p + 1) one-dimensional basis functions are formally defined by
p+1
Y ξ − ξj
NnLagrange, p (ξ) = , n = 1, 2, . . . , p + 1. (3)
ξn − ξj
j=1, j6=n

The nodal distribution ξj with j = 1, . . . , (p + 1) that is applied in the current article is defined as

 −1 if j=1
ξj = ξ Lo,p−1 if 2 ≤ j < p + 1 , (4)
 0,j−1
+1 if j =p+1
Lo,p−1
where ξ0,a with a = 1, . . . , (p − 1) denotes the roots of the (p − 1)th -order Lobatto polynomial

1 dp+1 h 2
p i
Lop−1 (ξ) = ξ − 1 . (5)
2p p! dξ p+1
This nodal distribution is generally referred to as Gauss-Lobatto-Legendre (GLL) grid. Spectral elements em-
ploying the GLL-grid have been proposed by Komatitsch and Tromp, 26 Zak et al.,27 Kudela et al.,28 Peng et
al.29 and Ha and Chang30 in order to mention just a few.

3. THE SPECTRAL CELL METHOD

In this section the concept of the finite cell method (FCM) is briefly introduced. In the literature various articles
dealing with the so called p- and B-spline versions of this method, cf. 15–18 can be found. In this contribution
we extend this approach to the spectral cell method. The developed method is then referred to as SCM. The
general concept is the same for all versions and is explained in the following.

3.1 The fictitious domain concept

As depicted in Fig. 1, the physical domain Ω, on which the problem to be solved is defined in terms of a variational
formulation, is embedded in a fictitious domain Ω f ict resulting in the embedding domain Ω e . The variational
principle describing linear elastic problems, whether static or dynamic, is the same as for standard FEM (cf.
Eq. (1)).
The embedding domain is now discretized in a mesh using rectangular cells which are independent of the original
domain. These “finite elements” of the embedding domain do not necessarily fulfill the usual geometric properties
of elements for the physical domain Ω, as they may be intersected by the boundary of Ω. To distinguish them
from classical elements they will be called finite cells. It is simpler and more advantageous to apply cells which
are rectangular quadrilaterals/hexahedrals in two/three dimensions, resulting in a constant Jacobian matrix of
the cell-wise mapping function.

Proc. of SPIE Vol. 8695 86951U-3

 ˉt
Ωf ict = Ωe \Ω
ΓN

Ω + = Ωe

ΓD
Figure 1: The physical domain Ω is extended by an so called fictitious domain Ω f ict . This allows for an easy
discretization of the whole embedding domain Ω e .

Fig. 2 illustrates the situation for a two-dimensional setting. Here, also the role of the indicator function α
is indicated. This parameter is used to control the influence of the fictitious domain on the accuracy of the
numerical results. Generally, the weak form of equilibrium is given by

Be (u, v) = Fe (v) (6)

where the bilinear form is Z

T
Be (u, v) = [D v] Ce [D u] dΩ (7)
Ωe

in which

Ce = αC (8)

is the elasticity matrix of the embedding domain, with


1.0 ∀ x ∈ Ω
α(x) = (9)
0.0 ∀ x ∈ Ωe \ Ω .

α  1.0
Nok
No Pm α = 1.0

MEIN
α  1.0

`NE
IINIMMEr
Figure 2: Finite cell discretization. To control the Figure 3: The embedding domain Ω e is discretized
influence of the fictitious domain on the accuracy of using rectangular cells.
the results an indicator function α is introduced.

Inserting Eq. (8) and (9) into Eq. (7), the bilinear functional can be expressed as
Z
T
Be (u, v) = [D v] αC [D u] dΩ
Ωe
Z Z
T T
= [D v] C [D u] dΩ + [D v] 0 [D u] dΩ = B(u, v) . (10)
Ω Ωe \Ω

Proc. of SPIE Vol. 8695 86951U-4

The union of all nc cells which are not completely located outside the physical domain form the embedding
domain (Fig. 3), where Ωc is the domain represented by a cell
nc
[
Ωe = Ωc . (11)
c=1

At the discretized level, the weak form Eq. (7) turns into
nc Z
X T
Be (u, v) = [D v] αC [D u] dΩ . (12)
c=1 Ω
c

To resolve the the boundary of the physical domain an adaptive integration scheme has been proposed by Düster
and co-workers.17, 31 Fig. 4 depicts a schematic sketch of the deployed integration technique. The Gaussian
quadrature assumes smoothness of the integrands which is disturbed by introducing the indicator function α
within the finite cells. Consequently, the finite cell method uses a composed Gaussian quadrature to improve
the integration accuracy in cells cut by geometric boundaries. To this end, the original cells are divided for
integration purposes by means of a quadtree/octree space partitioning scheme set of sub-cells. 15, 31

k=4

k=3

k=2

k=1

k=0
discontinuity cut cell

Figure 4: Sub-cells for the adaptive integration scheme. The original finite cell is refined if it is cut by the
boundary of the actual structure. Several refinement levels are required until the boundary is appropriately
resolved. This is checked by computing the area of the domain we are interested in. If a certain relative error is
reached the refinement is stopped.

Several refinement levels are required until a sufficient accuracy is ensured. Starting from the original finite
cell of level k = 0, each sub-cell of level k = i is first checked whether it is cut by a geometric boundary. If true,
the relevant cell is replaced by 4 equally spaced cells of level k = i + 1. In Fig. 4 a finite cell, cut by the physical

Proc. of SPIE Vol. 8695 86951U-5

boundary, is depicted. Depending on the sub-division level k the sub-cells, only used for integration purposes,
are filled in with gray color. Deploying this technique the effort of meshing complex structures is shifted to the
numerical integration of the system matrices and load vectors. In order to avoid problems with ill-conditioning of
the resulting equation system, the indicator function is set to very small (positive) values α ≈ 10−10 . . . 10−5 for
integration points inside the fictitious domain, cf. Fig. 2. To increase the efficiency of the adaptive quadrature
scheme, the number of integration points can be adjusted for each sub-cell individually as shown in the article by
Abedian et al.32 The adaptive element integration is performed deploying a Gauss-Legendre quadrature. Since
the Jacobian of the mapping function is constant, the integrands can be exactly evaluated. After the assembly
process the system matrices can be lumped utilizing the row-summing technique. 12, 13 Hence, efficient and fast
explicit solver like the central difference method (CDM) can be used for time marching.

4. NUMERICAL STUDIES
This section demonstrates the capabilities of the finite and spectral cell method. Important wave propagation
characteristics such as the multi-modal behaviour of Lamb waves as well as their dispersive nature are to be
resolved. To this end a simple two-dimensional benchmark problem with a circular hole is chosen.
4.1 Two-Dimensional Benchmark Problem
The first simple benchmark problem is a two-dimensional plate with a circular hole. The hole is eccentrically
located with respect to the mid-plane of the structure, cf. Fig. 5. This geometric perturbation causes mode
conversion and thus highlights an important characteristic of Lamb waves. 33
In Fig. 6 contour plots of the travelling wave are depicted. At the left boundary of the plate a pure S0 -mode
is excited, cf. Fig. 6a. After interacting with the local perturbation a converted A0 -mode can be observed
additionally to the transmitted S0 -wave packet, cf. Fig. 6b. For the sake of minimizing the numerical costs
we assume infinite dimensions in x3 -direction and consequently utilize the plain strain assumption. This poses
no limitation on the received results as all key features of guided wave propagation can be studied using the
described setup. The boundary conditions, material properties as well as the dimensions of the plate are given
in Fig. 5. The time history of the displacement field is saved at the measurement points P1 to P4 . The results
obtained using the finite cell method and the spectral cell method are later compared to reference solutions
gained by applying the p-version of FEM.

F1 (t)
l1 l2 l3 l4
x2

x1 P3 P1 P2 P4 t
d e

F2 (t)

Figure 5: 2d aluminum plate (E = 70 000 N/mm2 ; ν = 0.33, ρ = 2700 kg/m3 ) with a circular hole. Loads and
boundary conditions are marked in the figure. Two point forces F1 (t) and F2 (t) = aF1 (t) are applied, with a = 1
for the excitation of a purely symmetric Lamb wave mode (S0 ) and a = −1 if the anti-symmetric mode (A0 ) is
considered. The dimensions of the plate are: l1 = 100 mm, l2 = 50 mm, l3 = 152 mm, l4 = 298 mm, tp = 5 mm,
d = 2 mm, e = 2 mm. The signal parameters are: F̂ = 1 N , ω = 2πf , f = 200 kHz and n = 5.

(a) Incident S0 wave packet. (b) Converted A0 - and transmitted S0 -modes.

Figure 6: Contour plots of the travelling waves in an aluminum plate with a circular hole highlighting the effect
of mode conversion at anti-symmetrical perturbations of the structure under investigation.

Proc. of SPIE Vol. 8695 86951U-6

 The ultrasonic guided waves are excited using a pair of collocated point forces on the top and the bottom
surfaces of the plate. They act in the x2 -direction and their time-dependent amplitudes follow a sine burst signal.
The amplitude signal (see Fig. 7a) is given by
 
ωt n
F (t) = F̂ sin ωt sin 2
, 0≤t≤ .
2n f

This kind of pulse has the advantage that the frequency content is narrow-banded and thus limiting the physical
dispersion of these waves to a minimum. The number of cycles within the signal n determines the width of the
excited frequency band around the central frequency f (see Fig. 7b). For a narrow frequency bandwidth, n has
to be chosen relatively high.

1 7E-6
6E-6
0.5 5E-6
4E-6

A [−]
A [−]

0
3E-6

-0.5 2E-6
1E-6
-1 0E+0
0 0.5 1 1.5 2 2.5 0E+0 1E+5 2E+5 3E+5 4E+5
t [10 −5
s] f [Hz]
(a) Time domain signal (b) Frequency content
Figure 7: Sine-burst.

Applying concentrated loads at the top and bottom surfaces allows us to exploit the advantages of a mono-modal
excitation, which means that only a single mode is present at a time for f ∙ tp < 1.5 M Hzmm (see Fig. 8). In
order to generate a signal containing only the A0 -mode both forces have to act in the same direction, meaning
they have to be in-phase. If the two loads are out-of-phase a pure S0 -mode is generated.

10
5 S0
8 A1
4 0.4 1
S0
cp [km/s]
cg [km/s]

6
3

2 A0 4 0.4 1

2 A0
1 A1

0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
f d [M Hzmm] f d [M Hzmm]
(a) Group velocity (b) Phase velocity
Figure 8: Group and phase velocity dispersion diagrams for an aluminum plate.

The discretization parameter are chosen according to the guidelines published in Willberg et al.7 Accordingly,

Proc. of SPIE Vol. 8695 86951U-7

for a polynomial degree template (px1 = 3, px2 = 4) an element size of

be,max = 0.003 m

is derived and the time step Δt should not exceed

Δtmax = 1e-7 s.

4.2 Results
As a reference solution, to evaluate the performance of the finite cell method and the spectral cell method, we
choose the results obtained by applying the p-version of FEM (nelem = 404, ndof = 11288, be = 3 mm) with
the optimal polynomial degree template and the corresponding finite element size determined in the previous
section 4.1. The geometrical features such as the circular hole are described exactly using the blending function
method.9, 34–36 These blending elements have been proposed as a remedy to circumvent the geometrical approx-
imation error. In this case the corresponding blending functions are sin- and cos-functions. Consequently, the
error of the geometry representation is equal to zero. The finite element discretization is depicted in Fig. 9. In
the finite and spectral cell method the mesh does not necessarily match the geometry and therefore the mesh
generation is straightforward. Here the domain is simply discretized by using a Cartesian grid of cells, cf. Fig.
10. For the current example 400 quadrilateral high order cells are utilized. The polynomial degree and the cell
dimensions are similar to the p-FEM grid. To account for the circular hole in the 4 cut cells (cells, which exhibit
a discontinuous integrand) an adaptive integration is carried out. Fig. 10 depicts the sub-cells, that are used
during the integration process.
In order to judge the accuracy of the FCM and SCM they are compared to the reference solution (p-FEM) of
the displacement time history at the points P1 to P4 , cf. Fig. 11. It is worth mentioning that even though the
cells do not resolve the circular hole geometrically, the FCM and SCM are able to produce results with a similar
accuracy as the p-FEM even at the points located at the boundary of the hole, cf. Fig. 11a and 11b.
The different curves are virtually coincident for all numerical approaches and thus highlight the ability of ficti-
tious domain methods to capture the discontinuity of the structure.
As can be seen the reflection and mode conversion phenomena are well resolved. Only the displacements in x1 -
direction are depicted. From the displacement field in x2 -direction no new insights can be gained, consequently
these figures have been omitted.
Due to the extension of the higher order fictitious domain method to utilizing spectral shape functions the nu-
merical efficiency for structural dynamics applications has been vastly improved. Hence, we can sum up, that
both the FCM and SCM are viable tools to investigate the propagation of ultrasonic guided waves in thin-walled
structures.

Figure 9: Blending elements used for the geometrically Figure 10: Sub-cells generated by the adaptive integra-
exact discretization. tion algorithm.

All results for the SCM have been obtained using a consistent mass matrix. During the course of the
investigation the authors observed that the diagonalized mass matrix featured negative masses on the principal
diagonal. This led to a divergence of the central differencing scheme. Note, that the negative masses are only
related to cut cells. Therefore, if only a few cells are used to discretize the perturbation in the structure the
following approach is suggested:

Proc. of SPIE Vol. 8695 86951U-8

 • Numerical integration of all uncut cells deploying a GLL-quadrature rule. This results in diagonalized
elemental mass matrices.
• Numerical integration of all cut cells utilizing the described adaptive quadrature scheme. This results in a
consistent (fully populated) elemental mass matrix.
The inversion of the assembled mass matrix results nonetheless in a sparsely populated matrix and thus facilitates
the use of an explicit time integration scheme. First studies show that the results are comparable to the fully
consistent mass matrix approaches.
2E-13 1E-13
p-FEM p-FEM
1E-13 FCM 6E-14 FCM
5E-14 SCM SCM
2E-14

u1 [m]
u1 [m]

0E+0
-2E-14
-5E-14
-1E-13 -6E-14
-2E-13 -1E-13
0E+0 2E-5 4E-5 6E-5 8E-5 1E-4 1E-4 0E+0 2E-5 4E-5 6E-5 8E-5 1E-4 1E-4
t [s] t [s]
(a) u1 -displacement signal at P1 (b) u1 -displacement signal at P2

8E-14 8E-14
p-FEM p-FEM
FCM FCM
4E-14 SCM 4E-14 SCM
u1 [m]
u1 [m]

0E+0 0E+0
-4E-14 -4E-14
-8E-14 -8E-14
0E+0 2E-5 4E-5 6E-5 8E-5 1E-4 1E-4 0E+0 2E-5 4E-5 6E-5 8E-5 1E-4 1E-4
t [s] t [s]
(c) u1 -displacement signal at P3 (d) u1 -displacement signal at P4
Figure 11: Comparison of the time history of the displacements for the different numerical approaches, namely
p-FEM, FCM and SCM. An excellent agreement of the numerical results is to be observed.

5. OUTLOOK
The obtained results demonstrate the abilities of the FCM and SCM to deal with ultrasonic guided wave prop-
agation in thin-walled structures. Since lightweight designs are often made of carbon fiber reinforced plastic
(CFRP) plates or sandwich panels with honeycomb or foam cores these complex, heterogeneous materials need
to be addressed as well in future studies.

`sA

(a) Hollow sphere core (b) Honeycomb core

Figure 12: Aluminum sandwich panels with different core layers.

We think that using the finite cell and/or spectral cell method is the only viable solution for the described
problem. These algorithms retain the accuracy of higher order finite element methods while simplifying the

Proc. of SPIE Vol. 8695 86951U-9

mesh generation dramatically. Thus, we are able to discretize geometrically complex structures with a coarse
but regular grid of cells. In future applications this enables us to compute sandwich panels - cf. Fig. 12 - with a
manageable amount of degrees of freedom.
Ongoing research is dealing with the aluminum hollow sphere plate shown in Fig. 12a. A CT-scan (computed
tomography) and LSDV (laser scanning Doppler vibrometer) measurements revealed damages such as debond-
ings within the plate. These measured data are used to validate the proposed spectral cell method for wave
analysis problems. The related investigations are in progress and will be published in a forthcoming paper. Also
phenomena like the continuous mode conversion (observed in CFRP plates with plys made of woven fabric),
discovered and described in Willberg et al.,37 are studied without simplifying the geometry. Overall, with the
FCM and/or SCM we have an efficient and effective tool at our hand to study the Lamb wave propagation
in heterogeneous materials and geometrically complex structures. To improve the efficiency further alternative
mass lumping procedure have to be investigated and developed, such that zero or negative masses due to the cut
cells are avoided.

ACKNOWLEDGMENTS
The authors would like to thank the German Research Foundation (DFG) and all project partners (PAK 357)
for their support (GA 480/13-2). Furthermore, the support received under grant DU 405/4-1 is also gratefully
acknowledged.

REFERENCES
[1] Staszewski, W. J., [Health Monitoring for Aerospace Structures ], John Wiley & Sons (2003).
[2] Giurgiutiu, V., [Structural Health Monitoring with Piezoelectric Active Wafer Sensors: Fundamentals and
Applications ], Elsevier (2008).
[3] Pavlopoulou, S., Soutis, C., and Staszewski, W. J., “Cure monitoring through time-frequency analysis of
guided ultrasonic waves,” Plastics, Rubber and Composites 41, 180–186 (2012).
[4] Weber, R., Hosseini, S. M. H., and Gabbert, U., “Numerical simulation of the guided lamb wave propagation
in particle reinforced composites,” Composite Structures 94, 3064–3071 (2012).
[5] Hosseini, S. M. H. and Gabbert, U., “Numerical simulation of the lamb wave propagation in honeycomb
sandwich panels: a parametric study,” Composite Structures (2012).
[6] Hosseini, S. M. H., Kharaghani, A., Kirsch, C., and Gabbert, U., “Numerical simulation of lamb wave
propagation in metallic foam sandwich structures: a parametric study,” Composite Structures (2012).
[7] Willberg, C., Duczek, S., Vivar Perez, J. M., Schmicker, D., and Gabbert, U., “Comparison of different
higher order finite element schemes for the simulation of Lamb waves,” Computational Methods in Applied
Mechanics and Engineering 241-244, 246–261 (2012).
[8] Ostachowicz, W., Kudela, P., Krawczuk, M., and Zak, A., [Guided Waves in Structures for SHM: The
Time-Domain Spectral Element Method ], John Wiley & Sons (2011).
[9] Szabó, B. and Babuška, I., [Finite Element Analysis ], John Wiley and Sons (1991).
[10] Szabó, B. and Babuška, I., [Introduction to Finite Element Analysis: Formulation, Verification and Valida-
tion ], John Wiley and Sons (2011).
[11] Cottrell, J. A., Hughes, T. J. R., and Bazilevs, Y., [Isogeometric Analysis: Toward Integration of CAD and
FEA ], John Wiley & Sons (2009).
[12] Dauksher, W. and Emery, A. F., “An evaluation of the cost effectiveness of Chebyshev spectral and p-finite
element solutions to the scalar wave equation,” International Journal for Numerical Methods in Engineer-
ing 45, 1099–1113 (1999).
[13] Dauksher, W. and Emery, A. F., “Accuracy in modeling the acoustic wave equation with Chebyshev spectral
finite elements,” Finite Elements in Analysis and Design 26, 115–128 (1997).
[14] Schillinger, D., Dede, L., Scott, M. A., Evans, J. A., Borden, M. J., Rank, E., and Hughes, T. J. R.,
“An isogeometric design-through-analysis methodology based on adaptive hierarchical refinement of nurbs,
immersed boundary methods, and t-spline cad surfaces,” tech. rep., University of Texas at Austin (2012).

Proc. of SPIE Vol. 8695 86951U-10

[15] Parvizian, J., Düster, A., and Rank, E., “Finite cell method: h- and p-extension for embedded domain
problems in solid mechanics,” Computational Mechanics 41, 121–133 (2007).
[16] Düster, A., Sehlhorst, H.-G., and Rank, E., “Numerical homogenization of heterogeneous and cellular
materials utilizing the finite cell method,” Computational Mechanics 50, 413–431 (2012).
[17] Düster, A., Parvizian, J., Yang, Z., and Rank, E., “The finite cell method for three-dimensional problems
of solid mechanics,” Computer Methods in Applied Mechanics and Engineering 197, 3768–3782 (2008).
[18] Schillinger, D., Ruess, M., Zander, N., Bazilevs, Y., Düster, A., and Rank, E., “Small and large deformation
analysis with the p- and B-spline versions of the finite cell method,” Computational Mechanics online
available, 1–34 (2012).
[19] Rank, E., Kollmannsberger, S., Sorger, C., and Düster, A., “Shell finite cell method: A high order fictitious
domain approach for thin-walled structures,” Computer Methods in Applied Mechanics and Engineering 200,
3200–3209 (2011).
[20] Hamel, G., [Theoretische Mechanik ], Springer (1967).
[21] Zienkiewicz, O. C. and Taylor, R. L., [The Finite Element Method: Volume 1 The Basis ], Butterworth
Heinemann (2000).
[22] Zienkiewicz, O. C. and Taylor, R. L., [The Finite Element Method: Volume 2 Solid Mechanics ], Butterworth
Heinemann (2000).
[23] Zienkiewicz, O. C. and Taylor, R. L., [The Finite Element Method: Volume 3 Fluid Dynamics ], Butterworth
Heinemann (2000).
[24] Bathe, K. J., [Finite-Elemente-Methoden ], Springer Verlag (2002).
[25] Hughes, T. J. R., [The Finite Element Method: Linear Static and Dynamic Finite Element Analysis ],
Prentice-Hall (1987).
[26] Komatitsch, D. and Tromp, J., “Spectral-element simulations of global seismic wave propagation - I valida-
tion,” International Journal of Geophysics 149, 390–412 (2002).
[27] Zak, A., Krawczuk, M., and Ostachowicz, W., “Propagation of in-plane waves in an isotropic panel with a
crack,” Finite Elements in Analysis and Design 42, 929–941 (2006).
[28] Kudela, P., Krawczuk, M., and Ostachowicz, W., “Wave propagation modelling in 1D structures using
spectral elements,” Journal of Sound and Vibration 300, 88–100 (2007).
[29] Peng, H., Meng, G., and Li, F., “Modeling of wave propagation in plate structures using three-dimensional
spectral element method for damage detection,” Journal of Sound and Vibration 320, 942–954 (2009).
[30] Ha, S. and Chang, F.-K., “Optimizing a spectral element for modeling PZT-induced lamb wave propagation
in thin plates,” Smart Materials and Structures 19, 1–11 (2010).
[31] Abedian, A., Parvizian, J., Düster, A., Khademyzadeh, H., and Rank, E., “Performance of different inte-
gration schemes in facing discontinuities in the finite cell method,” International Journal of Computational
Methods accepted for publication (2012).
[32] Abedian, A., Parvizian, J., Düster, A., and Rank, E., “The finite cell method for the j2 flow theory of
plasticity,” Finite Elements in Analysis and Design under review (2013).
[33] Ahmad, Z. A. B. and Gabbert, U., “Simulation of lamb wave refelections at plate edges using the semi-
analytical finite element method,” Ultrasonics 52, 815–820 (2012).
[34] Királyfalvi, G. and Szabó, B., “Quasi-regional mapping for the p-version of the finite element method,”
Finite Elements in Analysis and Design 27, 85–97 (1997).
[35] Gordon, W. J., “Blending-function methods of bivariate and multivariate interpolation and approximation,”
Journal on Numerical Analysis 8, 158–177 (1971).
[36] Düster, A., Bröker, H., and Rank, E., “The p-version of the finite element method for three-dimensional
curved thin walled structures,” International Journal for Numerical Methods in Engineering 52, 673–703
(2001).
[37] Willberg, C., Koch, S., Mook, G., Gabbert, U., and Pohl, J., “Continuous mode conversion of Lamb waves
in CFRP plates,” Smart Materials and Structures 21, 9pp (2012).

Proc. of SPIE Vol. 8695 86951U-11