Spectral Finite Element Method for

Propagation of Guided Elastic Waves in

Wind Turbine Blades for SHM Purposes

A. AK
1
, M. KRAWCZUK
1,2
and W. OSTACHOWICZ
1,3


ABSTRACT

Certain results of numerical simulations obtained by the use of the spectral
finite element method in time domain are presented by the authors. They were
selected in order to show the effectiveness of the spectral finite element method for
investigation of problems associated with propagation of guided elastic waves in a
wind turbine laminated composite blade. Results of these simulations were
subsequently evaluated in order to detect and localise simulated damage in the form
of cracks or delaminations.


INTRODUCTION

Wind turbine blades made out of laminated composites are very sensitive to
fatigue cracks or impact damage. Specific environmental conditions together with
large costs of their failures results in intensive research that aims to ensure their
safe and longtime operation. An interesting approach used to increase the strength
and durability of wind turbine blades comes from the application of various
structural health monitoring (SHM) techniques. These techniques are meant to
provide necessary information about the state of the blades as well as the location,
size and extent of potential damage offering crucial information and help in order
to estimate the operational remaining life time of the blades [13].
A very promising SHM technique used nowadays for that purpose is the
technique based on the propagation of guided elastic waves and their coupled
interaction with damage. Guided elastic waves can propagate for relatively long
distances as wave packet signals and allow one not only to detect small scale
damage, but very often also to differentiate and classify damage thanks to the
conversion of propagating wave modes at structural discontinuities [46].
In the case of wind turbine blades numerical simulations of these complex
1
Polish Academy of Sciences, Fiszera 14, 80952 Gdansk, Poland
2
Gdansk University of

Technology Narutowicza 11/12, 80952 Gdansk, Poland
3
Gdynia Maritime University, Al. Zjednoczenia 3, 81345 Gdynia, Poland
6th European Workshop on
Structural Health Monitoring - Tu.2.D.2

Licence: http://creativecommons.org/licenses/by-nd/3.0

1
coupled interaction phenomena can be a source of very interesting and important
information that can help further development of new, and improvement of the
existing, SHM systems used in the field. It is worth to mention that many
traditional modelling approaches based on the use of the finite differences fail at
complex geometries, while at the same time the classical finite element method
breaks down for high frequency signals as are guided elastic waves. Contrary to
this the spectral finite element method in time domain, as originally proposed by
Patera in 1984 [7], is very well suited for such problems as is investigation of the
phenomena associated with wave propagation in structures of complex geometries.
The method originates from the application of spectral series for solution of partial
differential equations [8, 9], while at the same time its basic ideas are very similar
to the classical finite element approach The characteristic feature of the method is
the use of special approximation polynomials (based on orthogonal Legendre
polynomials) together with the GaussLobattoLegendre points integration
numerical rule. As a consequence of that the inertia matrix obtained in this spectral
approach is diagonal making the total cost of any numerical calculations much less
demanding. Additionally, thanks to the discrete orthogonality of the approximation
polynomials the spectral finite element method in time domain is characterised by
exponential convergence. Because of all these properties, robustness and flexibility
nowadays the spectral finite element method is frequently used to solve various
problems in fluid dynamics [10], heat transfer [11], acoustics [12], seismology [13],
etc.
Certain results of numerical simulations obtained by the use of this approach and
presented by the authors were related with propagation of guided elastic waves in a
scaled model of a laminated composite wind turbine blade, scaled down 10 times
for the purpose of numerical and experimental investigations, in order to detect and
localise damage in the form of small cracks or delaminations. Various damage and
excitation scenarios were considered by the authors in this study together with
signal processing techniques that can next be applied for damage visualization
purposes.


NUMERICAL SIMULATIONS

Numerical model

Figure 1 shows the geometry of the composite wind turbine blade under
investigation. The skin of the blade was assumed as made of 6 layers of glass
epoxy composite [14] with equal thickness and orientation [0/90/45/0/90].
The resulting total thickness of the blade was 3 mm, while the relative volume
fraction of the reinforcing glass fibres was assumed as 0.4.
Based on this geometry a numerical model of the structure was built. For this
model the total number of 2160 isoparametric shell spectral finite elements was
employed, previously developed by the authors (36node shell elements of 6
degrees of freedom per each node according to the classical shell theory [6]). This
resulted in the total number of 275,000 degrees of freedom of the numerical model.
As structural damage to the wind turbine blade two types of failures were
considered. As the first an open transverse fatigue throughcrack was taken into
2
account [6] located at the distance 30 mm from the left end of the blade, marked as
L
1
, modelled by separation of appropriate element nodes. Its total length was 15
mm. As the second a delamination of layers was assumed, located at the same
distance L
1
for the left end of the blade, position between the 3
rd
and 4
th
layer,
covering the total area of 210 mm
2
, and modelled by degradation of the elastic
properties within appropriate material layers that are associated with shear stresses.



Figure 1. Geometry of a laminated composite wind turbine blade.

Strong structural damping observed in the case of laminated composite materials
can drastically reduce the wave propagation distance. Due to this fact and based on
results of experimental observations it was decided to reduce the carrier frequency
of the excitation signal F(t) in order to minimise this undesired effect. As a
consequence as the signal excitation F(t) a transverse 15 kHz 5 period sine force
pulse modulated by the Hann window was considered. The amplitude of the
excitation was 1 N. Boundary conditions of free type were used.
The total computation time T was assumed as covering 1.5 ms and was divided
into 24,000 equal time steps. In order to solve the equation of motion the central
different method was chosen as the most effective time integration scheme due to
the diagonal form of the global inertia matrix.
3
Calculation programme

The main objective of the calculation programme was not only to demonstrate
the effectiveness of the spectral finite element method in time domain but also to
demonstrate the application of the RMS measure for damage detection based on
small disturbances in wave propagation patterns that can be observed. The
application of RMS maps seems particularly attractive in the case of wind turbine

No damage Damage fatigue crack

t = 0.3 ms

t = 0.3 ms

t = 0.6 ms

t = 0.6 ms

t = 0.9 ms

t = 0.9 ms

t = 1.2 ms

t = 1.2 ms

Figure 2. Wave propagation patterns obtained for a laminated composite wind turbine blade based
on the transverse displacement component.

blades as structures of very complex geometries, where classical SHM approaches
based on the use of phased arrays or pitchcatch techniques turned to be unsuitable
4
due to multimode characteristics of propagating signals as well as multiple mode
conversions and boundary reflections.
First results of numerical calculations are related with propagation of guided
elastic waves within the wind turbine blade with no damage as well as in the case
of the presence of the open transverse fatigue throughcrack. They are presented in
Fig. 2 as snapshots taken at various time instances for the transverse displacement
component only.

Damage fatigue crack (no baseline) Damage delamination (with baseline)

t = 0.3 ms

t = 0.3 ms

t = 0.6 ms

t = 0.6 ms

t = 0.9 ms

t = 0.9 ms

t = 1.2 ms

t = 1.2 ms

Figure 3. Evolution of RMS maps calculated for a laminated composite wind turbine blade based on
the transverse displacement component.

As can be seen from Fig. 2 the wave propagation patterns are only moderately
affected by the presence of the crack. This is most clearly visible at the time
5
instance t = 0.6 ms. This result, significantly different from similar results obtained
for structural elements of simpler geometries as plates or pipes, is strongly affected
by both directional material properties as well as geometrical complexity of the
blade itself. Due to the relatively high length of the blade in comparison with its
other dimensions, propagating elastic waves are strongly affected by structural
boundaries at the very early stage of the wave front development. For this reason
classical SHM approaches usually fail in such circumstances.
In the case of a known baseline state of the structure under investigation the
problem mentioned above can be avoided in a relatively straightforward manner, if
only appropriate signal processing techniques are employed in order to remove any
noise from the signals being subtracted. On the other hand RMS maps, because of
their white noise filtering properties, come as a good alternative.
The following examples of numerical calculations concern the evolution of the
RMS maps based on the calculated wave propagation patters for the case of the
wind turbine blade with the open transverse fatigue throughcrack as well as the
delamination. They are presented in Fig. 3 as snapshot taken at various time
instances also for the transverse displacement component. Detailed information
about various definitions and calculation routines used for RMS maps can be found
in [15]. From Fig. 3 it can be noted that both types of damage could be successfully
detected and localised, however, it should be reminded that damage in the form of
delamination presents a serious challenge due to very small amplitudes of signal
disturbances observed in the transverse displacement component.


CONCLUSIONS

The application of the spectral finite element method in time domain for analysis
of wave propagation patterns in a laminated composite wind turbine blade was
successfully demonstrated. Two different numerical examples were considered
showing the effectiveness and robustness of the approach proposed by the authors
and comprising wave propagation analysis in the blade with an open fatigue
transverse throughcrack as well as delamination.
Based on the results of numerical simulations it can be stated that both types of
damage could be successfully detected. However, in order to accomplish this task
application of appropriate signal processing techniques was required. As shown in
this work one of such signal processing techniques can be a technique be based on
calculation of RMS values of wave propagation signals registered over the whole or
selected area of the blade leading to socalled RMS maps. It should be mentioned
that practically such measurements can be performed on real structures of
complicated geometries by using laser scanning vibrometry. Thanks to the
application of RMS maps all undesired effects resulting from geometrical
complexity of the blade, directional properties of blade material and noise influence
may be easily overcome. In special cases additional information about the baseline
state of the structure under investigation can additionally enhance damage detection
sensitivity of the approach proposed by the authors. In SHM practice this type of
information is usually gathered during required planned periodical inspections of
the monitored structure enabling one to track the evolution of any previously
detected damage.
6
ACKNOWNLEDGEMENTS

The authors of this work would like to gratefully acknowledge the support for
this research provided by the Polish Ministry of Science and Higher Education
through the European Funds System under the Sectoral Operational Programme
Improvement of the Competitiveness of Enterprises via MONIT project
(Monitoring of Technical State of Construction and Evaluation of Its Lifespan) nr
POIG.01.01.0200013/08.

REFERENCES

1. Chang, F.K. 1998. Structural Health Monitoring: Current Status and Perspectives. CRC Press.
2. Adams, D.E. 2007. Health Monitoring of Structural Materials and Components: Methods and
Applications, John Wiley & Sons Ltd.
3. Giurgiutiu, V. 2008. Structural Health Monitoring with Piezoelectric Wafer Sensors, Elsevier
Inc.
4. Wilcox, P.D. 2003. Omnidirectional guided wave transducer arrays for the rapid inspection of
large areas of plate structures, IEEE Transactions on Ultrasonics, Ferroelectrics, and
Frequency Control, 50:699709.
5. Kampanis, V., Dougalis, V.A. and J.A. Ekaterinaris. 2008. Effective Computational Methods for
Wave Propagation, Chapman & Hall/CRC.
6. Ostachowicz, W., P. Kudela, M. Krawczuk and A. ak. 2011. Guided Waves in Structures for
SHM. The Timedomain Spectral Element Method, John Wiley & Sons Ltd.
7. Patera, A.T.. 1984 A spectral element method for fluid dynamics: Laminar flow in a channel
expansion, Journal of Computational Physics, 54:468488.
8. Doyle, J.F. 1997. Wave Propagation in Structures. Spectral Analysis Using Fast Discrete
Fourier Transforms, Springer.
9. Boyd, J.P. 1989. Cheybyshev and Fourier Spectral Methods, Springer.
10. Seriani, G. 1998 3D large scale wave propagation modelling by spectral element method on
Cray T3E multiprocessor, Computational Methods Applied in Mechanical Engineering,
164:235247.
11. Canuto, C., Hussaini, M.Y., Quarteroni, A. and T.A. Zang. 1988. Spectral Methods in Fluid
Dynamics, Springer.
12. Spall, R. 1995 Spectral collocation methods for one dimensional phase change problems,
International Journal of Heat Mass Transfer, 15: 27432748.
13. Dauksher, W. and A.F. Emery. 1996 The use of spectral methods in predicting the reflection
and transmission of ultrasonic signals through flaws, Review of Progress in Quantitative Non
destructive Evaluation: D.O. Thompson, D.E. Chimenti, eds. 15:97104.
14. Vinson, J.R. and R.L. Sierakowski. 2002. The Behavior of Structures Composed of Composite
Materials, Springer.
15. Radzieski, M. and W. Ostachowicz. 2011 Experimental studies of structure inspection and
damage detection based on elastic waves energy distribution, Structural Health Monitoring
2011, 22452251.
7