Damage Detection in Carbon Fiber Composite Materials Using ...
1
Piezoelectric sensors would significantly reduce the cost of damage detection MAIN ACHIEVEMENT: Piezoelectric sensors produce Lamb waves that can be used to identify damages at a lower cost in the increasingly common carbon fiber composites. RESEARCH IMPACT STATUS QUO EXPERIMENT Damage Detection in Carbon Fiber Composite Materials Using Lamb Waves: FFT vs. PCA Sensor 1 Sensor 2 Hole Joshua Hardisty Dr. Arda Vanli Spandan Sharma High Performance Materials Institute – Florida State University 0 5 10 15 20 25 30 35 40 45 50 0.00 0.10 0.20 0.30 0.40 0.50 0.60 Percent Difference Hole size PCA Damage Index vs. Damage Severity 6 in 10 in FFT Chang (2011) Lecture Notes • 10 x 27 in. carbon fiber composite plate • Damages were introduced at varying distances (3 in. – 21 in.) and with different severity (1/8 in. – ½ in. hole) • Lamb waves were sent through the plate by two sensors located at the ends of the plate Signal and its Fast Fourier Transform due to a 13/32” hole at 3” away from Sensor 1 Fast Fourier Transform (FFT) Method • The time domain signal is converted to a frequency domain signal • The frequency domain signal can be integrated to find the energy of the signal Fast Fourier Transform (FFT) vs. Principle Component Analysis (PCA) • A comparison of the slopes of the FFT and PCA methods of analysis show that for every increase in damage size, there is a much larger drop in factor score/area • The PCA is much more sensitive Principal Component Analysis (PCA) Method • PCA is a method used to reduce the number of dimensions in an experiment • Principal components are found by projecting the data onto the principal axes of the covariance matrix y = -7.9234x + 8.4317 0 1 2 3 4 5 6 7 8 9 0 0.2 0.4 0.6 Area 3 inch 6 inch 10 inch 14 inch 18 inch 21 inch Area FFT Score vs. Damage distance y = -36.449x + 16.068 -15 -10 -5 0 5 10 15 20 0 0.2 0.4 0.6 F a c t o r S c o r e Damage size (inch) 3 inch 6 inch 10 inch 14 inch 18 inch 21 inch Factor Score PCA vs. Damage distance Economic Value • Structural Health Monitoring in the aerospace industry is worth over $9 billion. • Piezoelectric sensors in combination with the PCA method could reduce the cost of damage detection by 50%. To be able to accurately detect damages. •Build a model to characterize the location and severity of damages in carbon fiber composites END-OF-PHASE GOAL Example: What is the distance of a 13/32 in. damage from sensor 1 with a DI percentage difference of 45%? Solve for x using the equation from the PCA DI vs. Distance graph with y = 45%: The answer should be about 5 in. Carbon Fiber Composite Board Piezoelectric Sensor y = 0.01x 4 - 0.72x 3 + 14.99x 2 - 131.56x + 418.71 R² = 1.00 0 20 40 60 80 100 0 5 10 15 20 25 Percent Difference Distance from Sensor PCA Damage Index vs. Distance 13/32 in y = 219.19x 2 - 65.65x + 16.18 R² = 0.99 0 10 20 30 40 50 0.00 0.20 0.40 0.60 Percent Difference Hole size PCA Damage Index vs. Damage Severity 6 in Poly. (6 in) Sensors on airplane wing Future Inquiries with Principle Component Analysis (PCA) • “Tail” of the PCA vs. Distance graph increases when a decrease is expected • How sensitive is the measurement (How small are the damages this method can detect)? 8.1 8.3 7.8 8 9.3 9.6 9.9 8.7 8.7 9.9 0 5 10 15 20 25 2008 2009 2010 2011 2016 USD Billions Year Aviation Week "10-year Forecast" April 2011 – Market Value Heavy Maintenance and Modifications Line Maintenance Airbus 30% Boeing 61% Other 9% Fleet Distribution By Manufacturer - 2009 Market Value How does Principal Component Analysis add value? • Fewer sensors result in lower fixed and installation costs Why use piezoelectric sensors? • By using smarter sensor systems, maintenance time can be cut by over 40% • Most maintenance costs come from aircraft “down time” • Cutting maintenance time costs directly cuts total costs Possible Manufacturers International Air Transport Association
Transcript of Damage Detection in Carbon Fiber Composite Materials Using ...
Piezoelectric sensors would significantly reduce the cost of damage detection
MAIN ACHIEVEMENT:
Piezoelectric sensors produce Lamb waves that can be used to identify damages at a lower cost in the increasingly common carbon fiber composites.
RE
SE
AR
CH
IM
PA
CT
STA
TU
S Q
UO
E
XP
ER
IM
EN
T
Damage Detection in Carbon Fiber Composite Materials Using Lamb Waves: FFT vs. PCA
Sensor 1 Sensor 2
Hole
Joshua Hardisty Dr. Arda Vanli Spandan Sharma High Performance Materials Institute – Florida State University
0
5
10
15
20
25
30
35
40
45
50
0.00 0.10 0.20 0.30 0.40 0.50 0.60
Pe
rce
nt
Dif
fere
nce
Hole size
PCA Damage Index vs. Damage Severity
6 in 10 in
FFT Chang (2011) Lecture Notes
• 10 x 27 in. carbon fiber composite plate
• Damages were introduced at varying distances (3 in. – 21 in.) and with different severity (1/8 in. – ½ in. hole)
• Lamb waves were sent through the plate by two sensors located at the ends of the plate
Signal and its Fast Fourier Transform due to a 13/32” hole at 3” away from Sensor 1
Fast Fourier Transform (FFT) Method
• The time domain signal is converted to a frequency domain signal • The frequency domain signal can be integrated to find the energy of the signal
Fast Fourier Transform (FFT) vs. Principle Component Analysis (PCA) • A comparison of the slopes of the FFT and PCA methods of analysis show that for every increase in damage size, there is a much larger drop in factor score/area • The PCA is much more sensitive
Principal Component Analysis (PCA) Method
• PCA is a method used to reduce the number of dimensions in an experiment • Principal components are found by projecting the data onto the principal axes of the covariance matrix
y = -7.9234x + 8.4317
0
1
2
3
4
5
6
7
8
9
0 0.2 0.4 0.6
Are
a
3 inch
6 inch
10 inch
14 inch
18 inch
21 inch
Area FFT Score vs. Damage distance
y = -36.449x + 16.068
-15
-10
-5
0
5
10
15
20
0 0.2 0.4 0.6
F
a
c
t
o
r
S
c
o
r
e
Damage size (inch)
3 inch
6 inch
10 inch
14 inch
18 inch
21 inch
Factor Score PCA vs. Damage distance
Ec
on
om
ic
V
alu
e
• Structural Health Monitoring in the aerospace industry is worth over $9 billion. • Piezoelectric sensors in combination with the PCA method could reduce the cost of damage detection by 50%.
To be able to accurately detect damages.
•Build a model to characterize the location and severity of damages in carbon fiber composites
EN
D-O
F-P
HA
SE
G
OA
L
Example: What is the distance of a 13/32 in. damage from sensor 1 with a DI percentage difference of 45%? Solve for x using the equation from the PCA DI vs. Distance graph with y = 45%: The answer should be about 5 in.
Carbon Fiber Composite Board
Piezoelectric Sensor
y = 0.01x4 - 0.72x3 + 14.99x2 - 131.56x + 418.71
R² = 1.00
0
20
40
60
80
100
0 5 10 15 20 25
Pe
rce
nt
Dif
fere
nce
Distance from Sensor
PCA Damage Index vs. Distance
13/32 in
y = 219.19x2 - 65.65x + 16.18 R² = 0.99
0
10
20
30
40
50
0.00 0.20 0.40 0.60
Pe
rce
nt
Dif
fere
nce
Hole size
PCA Damage Index vs. Damage Severity
6 in
Poly. (6 in)
Sensors on airplane wing
Future Inquiries with Principle Component Analysis (PCA) • “Tail” of the PCA vs. Distance graph increases when a decrease is expected • How sensitive is the measurement (How small are the damages this method can detect)?
8.1 8.3 7.8 8 9.3
9.6 9.9 8.7 8.7
9.9
0
5
10
15
20
25
2008 2009 2010 2011 2016
USD
Bill
ion
s
Year
Aviation Week "10-year Forecast" April 2011 – Market Value
Heavy Maintenance and Modifications
Line Maintenance
Airbus 30%
Boeing 61%
Other 9%
Fleet Distribution By Manufacturer - 2009
Ma
rke
t V
alu
e
How does Principal Component Analysis add value? • Fewer sensors result in lower fixed and installation costs
Why use piezoelectric sensors? • By using smarter sensor systems, maintenance time can be cut by over 40% • Most maintenance costs come from aircraft “down time” • Cutting maintenance time costs directly cuts total costs
Possible Manufacturers
International Air Transport Association
