Computational and Experimental Studies On Aircraft Structural Health Monitoring Systems
Computational and Experimental Studies On Aircraft Structural Health Monitoring Systems
Computational and Experimental Studies On Aircraft Structural Health Monitoring Systems
12 de Novembro de 2007
Abstract
Structural health monitoring (SHM) is a research field that has been growing in the last
years. It has the ultimate goal of guaranty the safety of the aircrafts but with the minimal costs
as possible. To achieve that, the aircrafts will be subjected to nondestructive tests (NDT) that
don’t need them to stop their activity. The sensors used in the tests will be ideally embedded
in the structure and will be interrogated when the aircrafts stop to regular inspection between
flights or even when in the air. Structural flaws represent changes in effective thickness and
local material properties. The Lamb waves can be used to detect those changes.
This thesis will focus on the generation and reception of Lamb waves to detect damage
in a aluminum plate. Piezoelectric wafer active sensors (PWAS) were bonded on the plate
to evaluate its capability in Lamb wave generation and detection. The Lamb waves group
velocity was predicted theoretically and compared with the experimental results. The response
to changes in the excitation frequency was also studied. Cuts were made in the plate in order
to see if the Lamb waves could detect them. Numerical simulations of the wave propagation
process and interaction with damage were also performed using the commercially available
finite-element code ANSYS.
1 Introduction
SHM (Structural Health Monitoring) is an emerging technology leading to the development of
integrated systems capable of continuously monitoring structures. The ultimate goal of SHM is
to increase reliability, improve safety, enable light-weight design and reduce maintenance costs for
all kinds of structures. Actual NDT (Non-destructive tests) systems are stationary and ground
based which imply that the aircraft is fully stopped whenever it needs to be inspected. Aircrafts
have regular mandatory inspections and every minute that an aircraft is on the ground is an
extra cost to the company due to the non operation. With the new SHM systems the mandatory
periodic procedures will be reduce which will result in a reduction in the maintenance costs.
The development of integrated vehicle health monitoring could also reduce or eliminate a number
of present design constrains which will make possible new structures with less redundancy and
therefore lighter.
Damage detection using Lamb waves is a new promising research field. This thesis will focus on
that area. Structural flaws represent changes in effective thickness and local material properties,
and therefore measurements of variations in Lamb wave propagation can be employed to assess the
integrity of these structures. In this work piezoelectric wafers are used to transmit and received
the Lamb waves.
1
1.1 Lamb Waves
Lamb waves are elastic waves that propagate across thickness of thin wall structures with free
boundaries parallel to the mid-surface. Lamb waves can be either symmetric or anti-symmetric
across the material thickness (Sn and An , respectively, where n represents the number of inflection
points across the material thickness).
The entire thickness of the plate can also be interrogated by various Lamb modes, affording the
possibility of detecting internal damage as well as that on surface. Since the wave speed varies with
frequency, the propagation of Lamb waves is essentially dispersive. For a given frequency multiple
modes can exist and therefore the received signals are a complex mixture from different modes and
difficult to evaluate. The analytical dispersion curves give an idea of the various existing modes
and its velocities for each frequency of excitation. Therefore, it becomes necessary to plot the
dispersion curves for this case to choose the optimal frequency excitation.
2 Experimental Setup
The experimental setup is composed by an aluminum plate (1.5mx1.5m) of 2mm thickness, 3
piezoelectrics wafers, a function generator, an oscilloscope and a GPIB board to connect the
computer to the function generator and the oscilloscope. In the computer the ”‘Fluke AnyWave”’
program was used to send the signal to the generator and to receive the response signals from the
oscilloscope.
2
Figure 3: Experimental Setup Figure 4: Actuators/Sensors in the plate
Figure 5: Actuation signal (left top), zoom of the actuation signal (left bottom) and the power
spectrum (right)
The power spectrum of the hanning windowed tone burst (excitation signal used) shows a wider
main frequency peak, and there is almost no side band peak. In this case, the frequency component
is concentrated around the main peak, hence reduce the frequency spread range.
The group velocity represents the speed with which Lamb-wave packs are sent and received
along the thin-wall plate. With the dispersion curves presented one can know the velocity of the
waves for the actuation signal in question i.e. 100kHz. In this case the velocity of the S0 is 5440
m/s and the A0 velocity is 2300 m/s. Knowing the velocity and the distances between the sensors
one can calculate the time.
3
2.3 Results
In the next table are the results predicted theoretically and the experimental results obtained in
the laboratory. It can be seen that the errors are very small in the order of microseconds.
Figure 8: Wave generated by actuator 1 and received by sensor 2 (blue) and sensor 3 (orange)
In figure 8 the R represents a primary reflection (the wave reflects in one boundary)while the
r represents a secondary reflection (the wave reflects in two boundaries before it arrives to the
sensor).
The sensor 2 is closer from the actuator 1, so the wave arrives first to sensor 2 than to sensor 3
which is visible in the results. The first wave to arrive is S0 that as seen previously in the dispersion
curves is faster than the A0 . In such a way that the reflections from the boundaries from the S0
wave arrive first than the direct A0 wave.
Figure 9: Influence of the frequency in the Figure 10: Influence of the frequency in the
received waves (actuator 2 to sensor 1) received waves (actuator 2 to sensor 3)
The S0 mode is received at a time that is almost independent of frequency which is consistent
with the calculated dispersion curves for this plate. As the frequency increases the waves show less
dispersion.
They were also performed tests with higher frequencies than 200kHz. But the amplitude of the
results is in the order of the noise amplitude and it is impossible to distinguish anything.
4
2.5 2D Visualization of the Results
Lamb waves propagates as a circle, across thickness. In the results saved by the sensor with the
oscilloscope one can only see a 1D plot of the waves arriving at that sensor. To give some visibility
to the results, that 1D results felt by the sensor were rotated 360 around the actuator. The case
presented here is a wave generated in actuator 2 at the instant that arrives at sensor 1 (figure 11)
and at sensor 3 (figure 12).
Figure 11: Wave arriving at sensor 1 Figure 12: Wave arriving at sensor 3
3 Damage detection
As explained before a cut or a mass in the plate will reflect the Lamb waves. The cuts (20mm)
were made with a saw to make the cut as clean as possible.
Figure 13: Cuts in the plate Figure 14: Picture of the cut
Figure 15: Actuator 1 to Sensor 2 with dam- Figure 16: Actuator 1 to Sensor 3 with dam-
age (orange) and without (blue) age (orange) and without (blue)
Comparing the case damaged with the case undamaged one can see by figures 15 and 16 that
at visible eye there cannot be seen any difference. These can be due to the orientation of the cut.
The second cut was made with an angle of 45 with the horizontal. This was done in order to
be sure that the reflections from the damage will be received by the sensors.
Figures 17 and 18 show the reflection from the damage. It is also visible the reflection from
the third sensor. Each sensor is acting like a strange presence that is felt by the other sensors.
5
Figure 17: Actuator 1 to Sensor 2 with two Figure 18: Actuator 1 to Sensor 3 with two
cuts cuts
Figure 19: Plate without Figure 20: Plate with a 20mm Figure 21: Plate with a 25mm
damage cut cut
Figure 22: Plate with a 30mm Figure 23: Plate with a 35mm Figure 24: Plate with a 40mm
cut cut cut
The results in the figure 19 refer to the undamaged case. In the figure 20 the plate has a
20mm cut and there is visible the reflection from that cut. It was drawn a reference pink line in
the plots to be easier to compare them. In figure 20 the reflection from the damage is below that
pink line. As the cut is increased the reflection from the damage is also slowly increasing until the
case in figure 24 (40mm) where the reflection from the damage pass the pink line.
One can conclude that the reflection from the damage increases with the increase of the dam-
age.
4 Numerical Simulations
Numerical simulations of the wave propagation process were performed using the commercially
available finite-element code ANSYS. The plate is square so the grid is regular with square four
node elements (SHELL63).
6
4.1 Natural Frequencies
The first thing calculated in ANSYS was the natural frequencies of the plate and the vibration
modes.
Figure 25: First mode Figure 26: Second mode Figure 27: Third mode
Figure 28: Fourth mode Figure 29: Fifth mode Figure 30: Sixth mode
Figures 25 untill 30 represent the vibration modes of the plate. The colors represent the
displacement where the blue represents the minor displacement and the red the maximum. In the
table below are the natural frequencies that are very low and far away from the frequencies used
in this work.
Mode Natural Frequency (Hz) Maximum displacement (m)
1 7.9875 0.706248
2 16.290 0.697349
3 16.290 0.697332
4 24.017 0.640984
5 29.205 0.693357
6 29.343 0.86885
It was also studied the grid convergence. The grid chosen is an agreement between the compu-
tational effort and good results. The final choice was a grid of 22500 elements.
Figure 31: Actuator 1 to Sensor 2 without Figure 32: Actuator 1 to Sensor 2 without
damage (experimental results) damage (ANSYS)
There is a good agreement between the time of arrival of the first pack of waves and the arrival
of the first boundary reflection. In the laboratory the reflection from the south and east boundaries
7
arrive at the same time. However, in the numerical simulations the wave that is created is not
exactly circular and so, the reflections from the two boundaries don’t arrive at the same time.
In figure 33 it can be seen that the amplitude of the first pack to arrive is constant in all the
three cases but the amplitude of the second pack in slowly increasing as the number of clamped
nodes increases. As the crack length increases, the amplitude of the reflection increases.
The next step was to simulate the damages made in the laboratory.
Figure 34: Actuador 1 para Sensor 2 Figure 35: Actuador 1 para Sensor 2 (se-
(primeiro corte) gundo corte)
In figure 34 the amplitude of the cut reflection is in the order of the signal noise of the laboratory.
In the simulations there is no noise and one can see the reflections, but it the experimental results
it’s very difficult to distinguish the reflection from the noise. In the case of the second cut (figure
35)the amplitudes from the reflections are bigger than in the case of the first one and the damage
can be detected. As observed in the experimental results.
5 Conclusion
From the performed experiments it can be concluded that the piezoelectric sensors/actuators are
efficient in the generation and detection of Lamb waves. There is a good agreement between the
theoretical time of flight and the results obtain in the lab (in the order of 99%). Even with the
8
material available good results were obtained and a damage was detected. The first cut was more
difficult to detect but with the right positioning of the sensors in order to eliminate that ”dead
zones” it could be more effective. The results from the numerical simulations also had a good
agreement with the experimental ones.
References
[1] Purekar, A.; Pines, D. J.;” Interrogation of Beam and Plate Structures Using Phased Array Concepts”
ICAST 2001
[2] Hillger, W.; Pfeiffer, U.; ”Structural Health Monitoring Using Lamb Waves” Institute of Composite
Structures and Adaptive Systems, Braunschweig, Germany, ECNDT 2006
[3] Nieuwenhuis, J. H.; Neumann, J. J.; Greve, D. W.; Oppenheim, I. J; ”Simulation and Testing of
Transducers for Lamb Wave Generation”
[4] Giurgiutiu, V; Bao, J. and Zhao, W. ”Piezoelectric Wafer Active Sensor Embedded Ultrasonics in
Beams and Plates” March 2003
[5] Edalati, K.; Kermani, A.; Seiedi, M; Movafeghi M.; ”Defect Detection in Thin Plates by Ultra-
sonic Lamb Wave Techniques” The 8th International Conference of the Slovenian Society for Non-
Destructive Testing ”Application of Contemporary Non-Destructive Testing in Engineering” Septem-
ber 1-3, 2005, Portoro, Slovenia, pp. 35-43
[6] Gomez-Ullate, Y.; Espinosa, F.; Reynolds, P.; Mould, J.; ”Selective Excitation of Lamb Wave Modes
in Thin Aluminium Plates using Bonded Piezoceramics: Fem Modelling and Measurements” ECNDT
2006 - Poster 205
[7] IFFM, USHF, ”Elastic Waves propagation modelling” 1st Annual Meeting, December 19th 2005,
Lisbon, Portugal
[8] Giurgiutiu, V; Bao, J. ”Embedded Ultrasonic Structural Radar With Piezoelectric Wafer Active
Sensors for the NDE of Thin-Wall Structures” 2002 ASME International Mechanical Engineering
Congress November 17-22, 2002, New Orleans, LA
[9] Park, G.; Sohn, H.; Farrar C. R. and Inman, D. J.; ”Overview of Piezoelectric Impedance-Based
Health Monitoring and Path Forward” The Shock and Vibration Digest, vol 35, no6, November 2003
[10] Wang, X. D.; Huang, G. L; ”Wave Propagation in Electromechanical Structures: Induced by Surface-
Bonded Piezoelectric Actuators” Journal of Intelligent Materials Systems and Structures, vol. 12 -
February 2001
[11] Worden, K.; Dulieu-Barton, J. M.; ”An Overview of Intelligent Fault Detection in Systems and
Structures” Structural Health Monitoring
[12] Purekar, A.; Pines, D. J.; ”A Phased Sensor/Actuator Array for Detecting Damage in 2-D Structures”
AIAA-2002-1547 Alfred Gessow Rotorcraft Center - Department of Aerospace Engineering - University
of Maryland - College Park, MD 20742
[13] IFFM, USHF, ”Piezotransducer network design” 1st Annual Meeting, December 19th 2005, Lisbon,
Portugal
9
[14] Wang, X. D.; Huang, G. L; ”Study of Elastic Wave Propagation Induced by Piezoelectric Actuators
for Crack Identification” International Journal of Fracture 126: 287-306, 2004
[15] Van der Auweraer, H. and Peeters, B.; ”International Research Projects on Structural Health Moni-
toring: An Overview” Structural Health Monitoring
[16] Farrar, C. R. and Lieven, N. A. J.; ”Damage Prognosi: the future of structural health monitoring”
The Royal Society, December 2006 Structural Health Monitoring
[17] http://www.ndt-ed.org/ NDT Resource Center
[18] Chambers, J. T.; Wardle B. L.; Kessler, S. S.; ”Durability Assessment of Lamb Wave-Based Structural
Health Monitoring Nodes” AIAA
[19] Purekar, A. S.; ”Piezoelectric Phased Array Acousto-Ultrasonic Interrogation of Damage in Thin
Plates” Dissertation submitted to the Faculty of the Graduate School of the University of Maryland,
College Park in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2006
[20] Hillger, W.; Ullmann, T; ”NDT-Verfahren zur Qualittskontrolle und fr in-service Inspektionen von
Composite- und integralen Flugzeugstrukturen” Werkstoffkolloquium 2006, 05. 12. 2006, DLR Kln
[21] Nieuwenhuis, J. H.; Neumann, J. J.; Greve, D. W.; Oppenheim, I. J; ”Generation and Detection of
Guided Waves using PZT Wafer Transducers”
[22] Pea, J.; Kawieck, G.; Ullate, Y.G.; Freijo, F.M.E.; Melguizo, C.P.; ”Low Cost, Low Frequency Phased
Array System for Damage Detection in Panels”
[23] http://ccrma.stanford.edu/ The Stanford University Center for Computer Research in Music and
Acoustics
[24] Su, Z.; Ye, L.; Lu, Y.; ”Guided Lamb waves for identification of damage in composite structures: a
review” Journal of Sound and Vibration, 2006, Australia
[25] Doebling, S. W.; Farrar, C. R. and Prime M. B.; ”A Summary Review of Vibration-Based Damage
Identification Methods” Engineering Analysis Group, Los Alamos National Laboratory
[26] Doebling, S. W.; Farrar, C. R. and Prime M. B.; ”An Overview of Modal-Based Damage Identification
Methods” Engineering Analysis Group, Los Alamos National Laboratory
[27] Bao, J.; ”Lamb Wave Generation and Detection with Piezoelectric wafer Active Sensors” College of
Engineering and Information Technology - University of South Carolina 2003
[28] Seagar, A.; ”Basic Principles of Ultrasound Imaging System Design” March 25, 2002
[29] Taha, M. M.; Noureldin, A.; Lucero J. and Baca, T.;”Wavelet Transform for Structural Health Mon-
itoring: A compendium of uses anf features” Structural Health Monitoring
[30] Rocha, B.; Silva, C.; Suleman, A.; ”Vibration-based Health Monitoring of Aircraft Structures”
[31] AGFA - ”Ultrasonic Array Probes For Nondestructive Testing Online or in the Field”
[32] Betz, D. C.; Thursby, G.; Culshaw, B.; Staszewski, W. J.; ”Identification of structural damage using
multifunctional Bragg grating sensors: I. Theory and implementation” Institute of Physics Publishing
2006
[33] Greve, D.W.; Oppenheim I.J.; Sohn H.; Yue C.P. ”Structural health monitoring with an inductively
coupled (wireless) Lamb wave transducer” Paper presented at the Third European Workshop on Struc-
tural Health Monitoring, Granada, Spain, July 2006
[34] Lamb, Horace; ”The Dynamical Theory of Sound” London, Edward Arnold, 1910
10