investigation of incident lamb wave parameters on ...rjha/research/smasis2010-3780.pdf · lamb...
TRANSCRIPT
Copyright © 2010 by ASME
1
Proceedings of the ASME 2010 Conference on Smart Materials, Adaptive Structures and Intelligent Systems SMASIS2010
September 28 - October 1, 2010 Philadelphia, PA, USA
SMASIS 2010-3780
INVESTIGATION OF INCIDENT LAMB WAVE PARAMETERS ON DETECTION OF COMPOSITE DELAMINATION
Ratneshwar Jha Inho Kim Dulip Widana-Gamage Department of Mechanical and Aeronautical Engineering
Clarkson University, Potsdam, New York 13699-5725 ABSTRACT
A preliminary investigation of excitation signal parameters
on the detection of delamination in composite plates is
presented. Composite plates typical of aerospace
applications are used and excitation is provided through
integrated PZT actuator. A scanning laser vibrometer is
used for recording structural responses. Experimental
results at three frequencies (15, 18, and 20 kHz) and two
pulse widths (3.5 and 5.5 cycles) show the significant
effects of incident Lamb waves on damage signature.
Numerical studies using a commercial finite element code
(ABAQUS) show good correlation with experimental
results with appropriate level of structural damping.
INTRODUCTION
The use of composite materials has many benefits for
engineering structures; however, the damage mechanisms
of composites are very different from those of metals. One
particular damage mechanism, namely delamination cracks,
can be initiated by loading, impact, or manufacturing
defects. The crack lengths can reach critical value before
visual inspection. Lamb waves (fundamental symmetric and
asymmetric modes) have shown particular promise for
damage diagnosis of composite structures1-7
. The
fundamental idea behind Lamb wave propagation based
diagnostics is that different types of damages interact
differently with waves. Therefore, based on the measured
time history of the propagated wave, the traveling time,
speed reduction, and wave attenuation parameters are
extracted and used as the damage identification variables.
Further processing of the measured signals (e.g., using
wavelet transform, Hilbert-Huang transform, etc.) helps
damage recognition.
The incident Lamb waves depend on the excitation signal
parameters (pulse shape, amplitude, frequency and number
of cycles to be sent during each pulse period) and may have
significant impact on damage detection. Published papers
discuss the effects of actuation pulse parameters in a
general way, but specifics are seldom reported. Kessler1
considered the actuation pulse parameters, but presented
results for 15 kHz 3.5 cycles only. Several authors suggest
higher frequencies, but Diamanti and Soutis7 consider
frequencies below 50 kHz to be particularly sensitive for
composite diagnostics. The fundamental anti-symmetric
Lamb mode A0 is generated at frequencies below 50 kHz
which has much lower phase velocity than the symmetric S0
mode and thus a smaller wavelength making it more
sensitive to damage detection. Delamination sizes well
below the wavelength of the propagating mode were
successfully detected by Diamanti and Soutis7.
This paper presents a preliminary investigation of excitation
signal parameters on the detection of delamination in
composite plates. We use composite plates typical of
aerospace applications and provide excitation using
integrated PZT actuator. A scanning laser vibrometer is
used for recording structural responses. Experimental
results at three frequencies (15, 18, and 20 kHz) and two
pulse widths (3.5 and 5.5 cycles) show significant effect on
damage signature. Numerical studies using a commercial
finite element code (ABAQUS) show good correlation with
experimental results with appropriate level of structural
damping.
LAMB WAVE DISPERSION CURVES
Lamb waves result from the superposition of guided
longitudinal and transverse (shear) waves. Lamb waves
travel in thin plates with unconstrained boundaries and have
the capability of traveling long distances with little
attenuation. Due to their propagation characteristics, Lamb
waves can be used as a means to detect both superficial and
Copyright © 2010 by ASME
2
internal flaws in a structure. Local stiffness degradation,
crack or delamination (for composite structures) causes
reflection, dispersion, attenuation and mode shape change.
These changes in wave characteristics can be used to
diagnose structural defects.
Lamb waves are highly dispersive, exhibiting dependency
of phase/group velocity on wave excitation frequency.
Development of dispersion curves, which map modal
phase/group velocities with respect to propagation
frequency, provide an important tool in examining the
propagation characteristics of Lamb waves. Due to the
dependency of longitudinal and transverse wave velocity on
the properties of the medium, dispersive characteristics vary
with material properties and geometry. The dispersive
characteristics of the cross-ply composite plates used in our
experiments (carbon-epoxy, AS4/3501-6, [0/90]2s) are
shown in Figure 1.
(a)
(b)
Figure 1. Analytical dispersion curves for AS4/3501-6 [0/90]2s: (a) Phase velocity and
(b) Group velocity
NUMERICAL SIMULATIONS
A finite element model to simulate Lamb wave propagation
in carbon-fiber epoxy laminate using finite element code
ABAQUS has been developed. The model is used to
simulate transient dynamic response for healthy and
delaminated cases. The AS4/3501-6 carbon-epoxy
composite laminate model has dimensions of 250 x 126 x
1.25 mm (10 x 5 x 0.05 in) and consists of 8 cross-ply
[0/90]2s layers similar to the experimental plate. The plate
is meshed with 15,750 C3D8I solid elements to ensure at
least 10 nodes per wavelength1. It is restricted in all degrees
of freedom at nodes along the two ends as shown in Figure
2. In order to simulate the piezoelectric excitation similar to
the experimental studies, actuating piezoelectric transducer
(PZT-5H) has been modeled using 164 piezoelectric
elements (C3D8E in ABAQUS). Perfect bonding between
plate and PZT is assumed and an input of 30V is applied to
PZT.
(a)
(b)
Figure 2. (a) Finite element model of composite plate
with PZT; (b) isoparametric view of the PZT bonded to the plate
A representative delamination between the 4th
and the 5th
layers of the laminate is introduced by disconnecting nodes
in the region. The delamination area is10 mm wide (shown
red in Figure 2) located 80 mm from the center of PZT. The
through-width delamination covers 4% of the plate area.
The electrical excitation given to the PZT has the shape of a
tone burst obtained using a Hanning window. The dynamic
analysis is performed using ABAQUS/ Standard solver.
Copyright © 2010 by ASME
3
Solutions are obtained for 3.5 cycles of the input tone burst
at carrier frequency of 15 kHz. Out of plane velocity
responses are recorded at several sensor points. A series of
out of plane velocity fields for healthy plate are shown in
Figure 3 for illustrating wave propagation.
100 µs
220 µs
Figure 3. Lamb wave propagation and scattering at different time instants
Response of the plate for the input of 15 kHz 3.5 cycles for
healthy and delamination cases are shown in Figure 4. The
responses for the two cases show clear difference in the
wave amplitude near delamination area with delaminated
plate showing significantly larger amplitude.
Responses at sensor location just before the delamination
(Figure 5a) show significant differences in the wave packets
for healthy and delaminated cases. The amplitude of the
first wave packet is increased due to reflections from
delamination. Further observation at this sensor location
reveals formation of an extra wave packet right after the
incident wave. For the sensor location on top of the
delamination (Figure 5b) the increase in wave amplitude for
the first wave packet is smaller and the second wave packet
has much reduced amplitude. Similar trends were observed
15 kHz, 5.5 cycles input as well.
(a)
(b)
Figure 4. Response at time instance 480 µs for (a) healthy plate (b) plate with delamination
Copyright © 2010 by ASME
4
(a)
(b)
Figure 5. Responses recorded at sensor location (a) Just before delamination (b) On top of delamination
Structural damping has a significant effect on wave
propagation, including initial amplitude and attenuation.
Rayleigh linear damping model available in ABAQUS/
Standard is used to introduce damping to the plate. In
Rayleigh damping, it is assumed that damping matrix is a
linear combination of mass and stiffness matrices.
��� � ���� � ��� (1)
where � and � are user defined constants. For a given
mode the fraction of critical damping, �� , can be
expressed in terms of the damping factors � and � as:
�� �
����
��
� (2)
where � is the natural frequency of a mode. This implies
the mass proportional damping parameter, � , damps the
lower frequencies and the stiffness proportional damping ,
�, damps the higher frequencies. In our studies, the most
appropriate values for � and � are found to be 0 and 4x10-7
through comparisons with experimental data. Figure 6
shows a comparison of FE results with experimental data at
a point with x=150 mm, y=64 mm.
(b)
Figure 6. FE vs. experimental response of healthy case for 15 kHz 3.5 cycles input
The FE responses match well with experimental results for
the first two wave packets. Subsequent responses show
some deviations possibly due to the differenc
and experimental conditions such as boundary, damping,
bonding between plate and PZT (assumed to be perfect
FE), etc. Since the first few wave packets contain most
important information, the FE analysis with PZT modeling
and damping values will be used in future for simulations
with other input excitations.
(a)
Figure 7. AS4/3501-6 pre-preg cross ply composite plate (a) schematic diagram with delamination area (b) healthy
5
experimental response of healthy case for 15 kHz 3.5 cycles input
The FE responses match well with experimental results for
Subsequent responses show
the differences between FE
and experimental conditions such as boundary, damping,
assumed to be perfect in
Since the first few wave packets contain most
important information, the FE analysis with PZT modeling
will be used in future for simulations
EXPERIMENTAL INVESTIGATIONS
The composite plate used in this study
AS4/3501-6 pre-preg using vacuum bagging and oven
curing technique. (Dimensions
numerical simulations.) Hi-temp mold release wax from
PARTALL was applied between 4
the delaminated plate. A piezoelectric (PZT) actuator
(diameter 13.5 mm and thickness
onto the composite plate using epoxy as shown in Figure
(b)
cross ply composite plate (a) schematic diagram with delamination area (b) healthy
composite plate with PZT actuator
Copyright © 2010 by ASME
experimental response of healthy case for 15 kHz 3.5 cycles input
EXPERIMENTAL INVESTIGATIONS
in this study was fabricated from
preg using vacuum bagging and oven
. (Dimensions and layup are given under
temp mold release wax from
PARTALL was applied between 4th
and 5th
layers to make
A piezoelectric (PZT) actuator
diameter 13.5 mm and thickness 0.22 mm) was affixed
site plate using epoxy as shown in Figure 7.
(b)
cross ply composite plate (a) schematic diagram with delamination area (b) healthy
Copyright © 2010 by ASME
6
A National Instruments PXI 6339 and a BNC-2110 board
were used to generate signals and a QuickPack® power
amplifier was used to amplify the actuation signal. Tone
burst excitations at 3 different frequencies (15, 18 and 20
kHz) and 2 different cycle numbers (3.5 and 5.5 cycles)
were used to generate A0 Lamb waves. A Scanning Laser
Doppler Vibrometer (SLV) was employed to acquire the
signals at designated locations. The experimental set up is
shown in Figure 8.
(a)
(b)
Figure 8. (a) Experimental setup (b) Schematic diagram
Table 1 shows time of flight (TOF) for each signal at a
location 60 mm away from the center of PZT and the phase
velocity calculated using TOF. Similar values for TOF and
phase velocity are obtained for 5.5 cycles. With these TOF,
the reflected wave packets from boundaries can be
identified. Based on Lamb wave dispersion curve,
calculated phase velocities indicate that the signals are
Lamb wave A0 mode.
Table 1. TOF and Phase Velocity measured at 60 mm from center of PZT
15kHz,
3.5Cycle
18kHz,
3.5Cycle
20kHz,
3.5Cycle
TOF(ms) 0.0977 0.0928 0.0910
Phase
velocity(km/s) 0.614 0.646 0.659
(a)
(b)
Figure 9. Acquired signal at 80 mm away from PZT (a) Healthy plate (b) Delaminated plate
0 0.5 1 1.5 2 2.5-1
-0.5
0
0.5
1
Time(ms)
Norm
aliz
ed a
mplit
ude
0 0.5 1 1.5 2 2.5-1
-0.5
0
0.5
1
Time(ms)
Norm
aliz
ed a
mplit
ude
Copyright © 2010 by ASME
7
Figures 9(a) and 9(b) show the 15 kHz 3.5 cycle acquired
signals at 80 mm away from PZT (which is 8 mm away
from the beginning of delamination area). For convenient
visualization, obtained signals are normalized with the
maximum value of the first wave packet at 20 mm from the
PZT actuator. For the healthy plate, two boundaries are
constrained (upper and bottom of plate) and left and right
sides are free. For delaminated plate, there is an additional
boundary due to the delamination area. The first wave
packet for healthy plate is the incident wave and other wave
packets are boundary reflections. With TOF of 0.278 ms
(Figure 4a), the second packet travels a distance of 170.9
mm. This distance indicates that the second wave packet is
the reflected wave from bottom of the plate since this
sensor location is 85 mm away from bottom boundary. For
the delaminated plate (Figure 4b), there are combined
packets between first and second waves. These combined
packets indicate delamination as they include reflected
waves from the delamination area (based on TOF analysis).
(a) (b)
(c)
Figure 10. Acquired signal from healthy and delaminate plates at 80 mm away from PZT:
(a) 15 kHz, 3.5 Cycle (b) 18 kHz, 3.5 Cycle (c) 20 kHz, 3.5 Cycle
0 0.5 1 1.5 2 2.5-1
-0.5
0
0.5
1
Time(ms)
Nom
aliz
ed a
mplit
ude
Healthy plate
Delaminated plate
0 0.5 1 1.5 2 2.5-1
-0.5
0
0.5
1
Time(ms)
Nom
aliz
ed a
mplit
ude
Healthy plate
Delaminated plate
0 0.5 1 1.5 2 2.5-1
-0.5
0
0.5
1
Time(ms)
Nom
aliz
ed a
mplit
ude
Healthy plate
Delaminated plate
Copyright © 2010 by ASME
8
The effect of incident (excitation) wave frequency is shown
in Figure 10. Signals from both healthy and delaminated
plates are presented for 3.5 cycle case with 15, 18, and 20
kHz frequencies. The shape of the wave packets change as
expected due to different velocities for the three
frequencies. All three cases show that delaminated plate has
larger wave amplitude due to reflected wave packet from
delamination area. Among these frequencies, 18 kHz seems
to indicate delamination most clearly. Wavelet transforms
of the signals at 18 kHz show appearance of additional
frequency components after 1.4 ms for the delaminated
plate (Figure 11). Number of cycles of the incident wave is
another important parameter. Comparison between 3.5 and
5.5 cycle signals show that 5.5 cycle case gives a better
indication of delamination (Figure 12). Further studies will
include other sensor locations, frequencies, and
quantification of energy of the reflected wave.
(a)
(b)
Figure 11. Wavelet transform of signals at 80 mm away from PZT: (a) Healthy plate (b) Delaminated plate
Time (µs)
Frequency (kHz)
Time (µs)
Frequency (kHz)
Copyright © 2010 by ASME
9
Figure 12. Acquired signal from healthy and delaminate plate at 80 mm away from PZT: (a) 18 kHz, 3.5 Cycle (b) 18 kHz, 5.5 Cycle
CONCLUSIONS
This study has presented a presents a preliminary
investigation of excitation signal parameters on the
detection of delamination in composite plates. We used
composite plates typical of aerospace applications and
provided excitation using integrated PZT actuator. A
scanning laser vibrometer was used for recording structural
responses. Experimental results at three frequencies (15, 18,
and 20 kHz) and two pulse widths (3.5 and 5.5 cycles)
showed significant effect on damage signature. Numerical
studies using a commercial finite element code (ABAQUS)
showed good correlation with experimental results with
appropriate level of structural damping. Further studies will
include other sensor locations, frequencies, and
quantification of energy of the reflected wave.
REFERENCES
1. S. S. Kessler. 2002. Piezoelectric-Based In-Situ Damage
Detection of Composite Materials for Structural Health
Monitoring Systems, Ph.D. Thesis, MIT, Cambridge, MA.
2. P. S. Tua, S. T. Quek and Q. Wang. 2004. Detection of
Cracks in Plates using Piezo-actuated Lamb Waves, Smart
Materials and Structures, 13:643-660.
3. W. Lestari and P. Qiao. 2005. Damage Identification for
Carbon/Epoxy Laminated Composite Structures Based on
Wave Propagation Analysis, 46th AIAA/ASME/ASCE/
AHS/ASC Structures, Structural Dynamics & Materials
Conference, Austin, TX.
4. S. Banerjee, F. Ricci, E. Monaco, L. Lecce and A. Mal.
2007. Autonomous Impact Damage Monitoring in a
Stiffened Composite Panel, Journal of Intelligent Material
Systems and Structures, 18(6):623-633.
5. Y.-H. Kim, D.-H. Kim, J.-H. Han and C.-G. Kim. 2007.
Damage Assessment in Layered Composites using Spectral
Analysis and Lamb Wave, Composites Part B: Engineering,
38: 800-809.
6. A. Raghavan and C. E. S. Cesnik. 2007. Review of
Guided-wave Structural Health Monitoring, The Shock and
Vibration Digest, 39(2):91-114.
7. Diamanti K, Soutis C. Structural health monitoring
techniques for aircraft composite structures. Prog
Aerospace Sci (2010), doi:10.1016/j.paerosci.2010.05.001.
0 0.5 1 1.5 2 2.5-1
-0.5
0
0.5
1
Time(ms)
Nom
aliz
ed a
mplit
ude
Healthy plate
Delaminated plate
0 0.5 1 1.5 2 2.5-1
-0.5
0
0.5
1
Time(ms)
Nom
aliz
ed a
mplit
ude
Healthy case
Delaminated case
(a) (b)