http://jim.sagepub.com/
Structures
Journal of Intelligent Material Systems and
http://jim.sagepub.com/content/17/12/1059
The online version of this article can be found at:
DOI: 10.1177/1045389X06064889
2006 17: 1059Journal of Intelligent Material Systems and Structures
Michaud
Hyuk-Jin Yoon, Daniele Marco Costantini, Hans Georg Limberger, René Paul Salathé, Chun-Gon Kim and Veronique
In situ Strain and Temperature Monitoring of Adaptive Composite Materials
Published by:
http://www.sagepublications.com
can be found at:Journal of Intelligent Material Systems and StructuresAdditional services and information for
http://jim.sagepub.com/cgi/alertsEmail Alerts:
http://jim.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://jim.sagepub.com/content/17/12/1059.refs.htmlCitations:
What is This?
- Dec 5, 2006Version of Record >>
at EPFL Scientific information and libraries on June 20, 2012jim.sagepub.comDownloaded from
In situ Strain and Temperature Monitoring of Adaptive
Composite Materials
HYUK-JIN YOON,
1
DANIELE MARCO COSTANTINI,
2
HANS GEORG LIMBERGER,
2
RENE
´
PAUL SALATHE
´
,
2
CHUN-GON KIM
1
AND VERONIQUE MICHAUD
3,
*
1
Smart Structures & Composites Laboratory, KAIST, Korea
2
Advanced Photonics Laboratory, EPFL, Switzerland
3
Laboratoire de Technologie des Composites et Polyme
`
res (LTC)
Ecole Polytechnique Fe
´
de
´
rale de Lausanne (EPFL), Switzerland
ABSTRACT: An optical fiber sensor is designed to simultaneously measure strain and
temperature in an adaptive composite material. The sensor is formed by splicing two fiber
Bragg gratings (FBGs) close to each other, which are written in optical fibers with different
core dopants and concentrations. Their temperature sensitivities are hence different. The
sensor is tested on an adaptive composite laminate made of unidirectional Kevlar-epoxy
prepreg plies. Several 150 mm diameter prestrained NiTiCu shape memory alloy (SMA) wires
are embedded in the composite laminate together with one fiber sensor. Simultaneous
monitoring of strain and temperature during the curing process and activation in an oven
is demonstrated.
Key Words: fiber Bragg grating, fiber sensor, adaptive composite, shape memory alloy, cure
monitoring.
INTRODUCTION
W
HEN considering the development of composite
materials in the past 40 years, an evolution is
clearly observed: the initial search was for very high
specific properties alone, driven by aerospace applica-
tion; the need to maintain high properties while reducing
manufacturing time and production costs was then
driven by automotive and other large scale applications;
more recently, the desire to integrate additional func-
tionality in the composite parts emerged. Since conven-
tional structural composite materials cannot fulfill this
last requirement alone, adaptive or smart composite
materials, which integrate actuators and sensors, may
hence well represent a next step in the development of
composite materials. This evolution is driven in part by
the need of automotive, sport or aerospace applications
to gain efficiency not only by reducing structural weight,
but also by directly integrating functions into the
structure. Examples span from civil engineering where
fiber optic sensors monitor bridges or dams (Bugaud
et al., 2000), to sports equipments including skis or
tennis rackets which are actively damped by piezo-
electric fiber systems, to health monitoring or active
flutter reduction of airplanes (Balta et al., 2001a;
Simpson and Boller, 2002). The evolution of composite
materials is also driven by the fact that it is now possible
to integrate both actuators and sensors directly into
the composite material. Actuators and sensors have
currently achieved a high enough degree of miniaturiza-
tion not to disrupt the structural integrity of the
composite part, both during processing and in service.
Fiber reinforced polymer composites containing thin
shape memory alloy (SMA) wires as actuating elements
(Wei et al., 1998a,b; Roytburd et al., 2000; Boller, 2001)
represent an important class of ‘smart materials’. The
SMAs have been available for 40 years and have
already found applications as actuators (Gandhi and
Thompson, 1992), but they have only recently been
manufactured as high quality wires with diameters
below 0.2 mm. Thin SMA wires may be integrated in
the host composite materials without affecting their
integrity. In comparison to other actuating technologies,
SMAs provide the following advantages: high reversible
strains (up to 6%), high damping capacity, large
reversible change of mechanical and physical character-
istics, and most importantly, the ability to generate
high recovery stresses. As a result, composite materials
with SMA wires demonstrate added functional effects
such as a shape change, a controlled overall thermal
expansion or a shift in the natural vibration frequency
*Author to whom correspondence should be addressed.
E-mail: veronique.michaud@epfl.ch
Figures 1 and 5–7 appear in color online: http://jim.sagepub.com
JOURNAL OF INTELLIGENT MATERIAL SYSTEMS AND STRUCTURES, Vol. 17—December 2006 1059
1045-389X/06/12 1059–9 $10.00/0 DOI: 10.1177/1045389X06064889
ß 2006 SAGE Publications
at EPFL Scientific information and libraries on June 20, 2012jim.sagepub.comDownloaded from
upon activation. The working principle is that
prestrained martensitic SMAs tend to recover their
initial shape when heated above their transformation
temperature. If the wires alone are clamped externally
or if they are embedded into a stiff host material, the
wire strain recovery is limited, and instead large recovery
stresses are generated, which may lead to the effects
cited previously (Michaud et al., 2002; Schrooten et al.,
2002a,b; Michaud, 2004).
The main drawback of these materials is that their
transformation is induced by temperature, which greatly
limits their response time controlled by heat transfer
kinetics. Also, if a smart material is to be produced,
it is necessary to locally monitor the strain and the
temperature inside the structure. A sensor is thus
needed, which is able to simultaneously measure strain
and temperature and which can also be embedded into
the host composite without modifying its properties and
functions.
Currently, the most promising candidates are fiber
optic sensors, which can be embedded in composite
materials to locally measure strain, both during cure and
during use (Measures, 1992; Hadjiprocopiou et al., 1996;
Guemes and Menendez, 2002). In the last 10 years,
researchers have been working on the principles and
techniques to measure strain (e.g., by using fiber Bragg
gratings, or other types of sensors such as Fabry-Perot
cavities), temperature or other parameters such as
humidity (Ferreira et al., 2000; Chi et al., 2001; Allsop
et al., 2002; Kronenberg et al., 2002; Lai et al., 2002;
Rao et al., 2002; Sivanesan et al., 2002; Shu et al., 2002;
Frazao et al., 2003; Han et al., 2003). Moreover, the
process of fiber embedding was investigated to ensure
reliability and precision of the measurement (Measures,
1992; Bao et al., 2002; Guemes and Menendez, 2002).
Combined with SMA actuators, as described earlier,
optical fiber sensors have been shown to provide a
composite with actuating and sensing capabilities
for shape or stress control (Balta et al., 2005; Yoon
et al. 2005).
However, the use of fiber Bragg grating (FBG)
sensors is limited by their simultaneous dependence on
strain and temperature. In a previous work (Balta et al.,
2005), a thermocouple was pasted on the structure to
monitor the temperature and back-calculate the strain.
To overcome this cross sensitivity using only embedded
optical fibers, a number of techniques have been
proposed, most of them relying on the deconvolution
of two simultaneous measurements. These methods
include the dual-wavelength superimposed gratings
(Xu et al., 1994), the use of first- and second-order
diffraction grating wavelengths (Echevarria et al., 2001),
FBGs in optical fibers with different dopants (Cavaleiro
et al., 1999; Guan et al., 2000), hybrid Bragg grating/
long period gratings (Patrick et al., 1996), dual-diameter
FBGs (James et al., 1996), FBGs combined with EDFAs
(Jung et al., 1999), FBG/EFPI combined sensors
(Zeng and Rao, 2001; Kang et al., 2003), FBGs in
high-birefringence optical fibers (Ferreira et al., 2000),
the employment of strain-free FBGs (Song et al., 1997;
Guan et al., 2002), and a combination of FBGs of
different ‘type’ (Shu et al., 2002; Frazao et al., 2003).
The use of a strain-free reference grating turns out to
be the most efficient way to discriminate strain and
temperature. On the other hand, it is not easy to
implement this technique when sensors must be
embedded into a host composite, since it requires
placing a grating into a small capillary tube or another
envelop protecting it from strain.
In the present article, a method to simultaneously
monitor the strain and temperature in an adaptive
composite laminate using a FBG-based sensor has been
proposed. The technique relying on the combination of
the two FBGs written in optical fibers with different
core dopants was improved for the present application.
A suitable couple of optical fibers were selected, and
their respective temperature and strain sensitivities
were measured before integration. Finally, strain and
temperature inside a Kevlar epoxy composite laminate
with SMA wires were measured during curing and
activation, to demonstrate the operation of the fiber
sensor.
SENSOR DEVELOPMENT
Principle
A fiber Bragg grating consists of a periodic change of
the core refractive index of an optical fiber and reflects
light around the resonance peak wavelength defined by
the following phase matching condition:
B
¼ 2n
e
ð1Þ
where
B
is the Bragg wavelength, n
e
is the core effective
refractive index, and is the period of the grating. The
shift of the Bragg wavelength
B
due to a temperature
change T and an external axial strain e can be
expressed for a bare fiber Bragg grating, which is neither
bonded nor embedded, as:
B
¼
B
þ ðÞT þ 1 p
e
ðÞ"½: ð2Þ
where is the coefficient of thermal expansion of the
fiber material, is the thermooptic coefficient and p
e
is the strain-optic constant (Kersey et al., 1997). The
relative Bragg wavelength shift
B
/
B
may simply be
written as a function of the temperature sensitivity
K
T
¼ þ and the strain sensitivity K
e
¼ 1p
e
of the
fiber Bragg grating:
B
B
¼ K
T
T þ K
"
": ð3Þ
1060
H.-J. YOON ET AL.
at EPFL Scientific information and libraries on June 20, 2012jim.sagepub.comDownloaded from
It is assumed that the thermal and applied strain
responses of the grating are independent. This has been
shown to be valid, if the sensing length is not too long,
or the temperature and strain variations are not too
large. Otherwise, cross-sensitivity terms need to be taken
into account (Farahi et al., 1990). In this work,
this assumption has been adopted, as the temperature
variation does not exceed 120
C, and the strain is lower
than 10
3
. The response of the two fiber Bragg gratings
having different temperatures and/or strain sensitivities
can be combined into a linear system of two equations,
which can be solved to obtain simultaneously the
temperature change and strain:
T
"
¼
K
T1
K
T2
K
"1
K
"2
1
B1
B1
B2
B2
2
6
4
3
7
5
¼ K
1
y: ð4Þ
where the subscripts 1, 2 indicate the specific FBGs,
K is the 2 2 matrix of the coefficients, and y is the
vector of relative wavelength shifts. Equation (4) can be
written in vector form as x ¼ K
1
y, where x is the vector
formed by the temperature and strain components. In
order to efficiently discriminate the temperature and the
strain contributions, the matrix of coefficients K must be
well-conditioned (Sivanesan et al., 2002). The relative
error x in temperature and strain is related to the Bragg
wavelength change vector y as follows:
x
kk
x
kk
CðK Þ
y
y
: ð5Þ
where CðKÞ¼ K
kk
K
1
is the condition number of
the matrix of coefficients K (Strang, 1988). In order to
reduce the error in the simultaneous evaluation of strain
and temperature, a small condition number is desired.
Fiber Sensor Design
The thermooptic coefficient of an optical fiber
depends on the core dopants and their relative
concentrations (Cavaleiro et al., 1999; Guan et al.,
2000; Oh et al., 2000). Based on this principle, a panel of
fibers, with different core dopants and concentrations,
was investigated, in order to select a couple that
provided a low condition number. Two fiber Bragg
gratings, one written in each fiber, and spliced to each
other as shown in Figure 1 formed the strain and
temperature sensor. The fiber specifications and the
fabrication parameters are listed in Table 1. The gratings
were fabricated in these optical fibers with the phase-
mask technique using a 193 nm ArF excimer laser. The
same phase mask with a period of 1058.5 nm was used.
The length of the FBG was of 7 mm and their reflectivity
was around 50%. Since the maximum curing tempera-
ture of the Kevlar epoxy composite material is of 140
C,
the FBG sensors were pre-annealed at 160
C during
48 h, in order to ensure the Bragg wavelength stability
(Kannan et al., 1997).
The temperature and strain sensitivities of the gratings
were separately measured. The wavelength shift of
the Bragg gratings as a function of temperature was
measured in reflection with a wavelength resolution
of 0.1 pm by a tunable laser (Tunics 1550, Nettest), a
photo-detector (MA9305B, Anritsu), and a wavelength
meter (WA-1500, EXFO Burleigh). The relative
wavelength shifts were acquired for each Bragg grating
by increasing the temperature from 20 to 150
Cin
increments of 10
C. Each grating was heated in an oven
and the temperature was measured with a resolution
of 0.1
C by a thermocouple connected to a voltmeter
(FLUKE 52 k/J). The temperature sensitivities of the
FBGs calculated with the linear fit are listed in Table 2.
The thermal expansion coefficient of the fiber material
is nearly of an order of magnitude lower than
the thermooptic coefficient (Jewell et al., 1991). Thus
the temperature sensitivity is mainly affected by the
thermooptic coefficient, which, in turn, depends on the
concentrations of GeO
2
and B
2
O
3
in the core. This
explains the difference in temperature sensitivity of
the gratings written in the GeO
2
series and GeO
2
-B
2
O
3
co-doped silica fibers, and the higher temperature
sensitivity with increasing GeO
2
concentration. The
G22 and PS1500 fibers which exhibit a large difference
in temperature sensitivity were selected as the combina-
tion of fibers for the strain–temperature sensor. The G25
fiber was not selected because its cladding was 80 mmin
Table 1. Fiber Bragg gratings fabrication parameters
(core dopants and concentrations as provided by the
manufacturer, F
p
: fluence per pulse, F
T
: total fluence and
k
B
: Bragg wavelength).
Fiber
label
Fiber
supplier
Core
dopants
GeO
2
B
2
O
3
(mol%)
F
p
(mJ/cm
2
)
F
T
(J/cm
2
)
k
B
(nm)
SMF-28e Corning 3 – 90.9 4080 1532
G9 Cabloptic 9 – 150 230 1532
G18 Cabloptic 18 – 68.8 26.68 1536
G22 CSEM 22 – 68.8 1.1 1547
G25 CSEM 25 – 68.8 1.31 1551
PS1500 Fibercore 10 14–18 68.8 5.5 1532
Photosil Spectran 30 Unknown 100 58.1 1531
Gauge length
Splice
Optical fiber
Fiber Bragg grating
Figure 1. Design of the FBG-based sensor to simultaneously
measure temperature and strain.
In situ Strain and Temperature Monitoring 1061
at EPFL Scientific information and libraries on June 20, 2012jim.sagepub.comDownloaded from
diameter, and slightly elliptic, which prevented to ensure
a reproducible splicing operation.
The solid and dotted lines in Figure 2 represent
the linear fitting to the experimental data for the G22
and PS1500 fibers, respectively. The G22 fiber has a
temperature coefficient K
T1
of 7.48 0.02 10
6
C
1
,
while the PS1500 has a coefficient K
T2
of
6.00 0.01 10
6
C
1
. A deviation from linearity is
observed in the figure, and the response of the FBGs
is shown to rather follow a quadratic behavior. The
residual wavelength difference for the linear and the
second-order polynomial fits are given in Figure 3. This
nonlinear temperature response was already modeled
by Ghosh (Ghosh, 1995; Pal et al., 2004). However, in
order to use Equation (4) for a simultaneous recovering
of strain and temperature, the temperature response of
the sensors was fitted to a linear curve, which can lead
to a maximum error of 1.3% in the temperature range
of operation of 20–150
C.
Figure 4 shows the relative Bragg wavelength change
for the two gratings when the applied strain is between
0 and 2000 me. The strain was induced by suspending
the calibrated weights to the optical fiber, and assuming
for the silica fiber a Young modulus of 72.5 GPa
(Kersey et al., 1997) and a cladding diameter of
125 mm. The experimental data obtained for the G22
and PS1500 grating sensors were linearly fitted. Their
strain sensitivities were similar and equal to
0.75 0.01 10
6
me
1
and 0.77 0.01 10
6
me
1
,
respectively.
(a)
(b)
20 40 60 80 100 120 140
−20
−10
0
10
20
30
Residual wavelength (pm)
Temperature (°C)
FBG in PS1500 fiber
Linear fit
2nd order polynomial fit
Figure 3. Residual wavelength shift for linear and second-order
polynomial fit as a function of temperature change for both gratings
written: (a) in the G22 and (b) in the PS1500 optical fibers.
0 20 40 60 80 100 120 140
0.0
2.0×10
−4
4.0×10
−4
6.0×10
−4
8.0×10
−4
1.0×10
−3
∆λ / λ
∆T (°C)
G22: K
T1
=7.48×10
−6
°C
−2
PS1500: K
T2
=6.00×10
−6
°C
−1
Figure 2. Relative Bragg wavelength shift versus temperature
change for both the gratings written in the G22 and PS1500 optical
fibers.
0 500 1000 1500 2000
0.0
1.6 × 10
−3
1.2 × 10
−3
8.0 × 10
−4
4.0 × 10
−4
2.0 × 10
−3
∆λ / λ
Strain (µε)
G22: K
ε1
= 0.75·10
−6
µε
−1
PS1500: K
ε2
= 0.77·10
−6
µε
−1
Figure 4. Relative Bragg wavelength shift versus applied strain for
both the gratings written in the G22 and PS1500 optical fibers.
Table 2. Temperature sensitivities of FBGs.
Fiber label K
T
(10
6
C
1
)
SMF-28e 6.92 0.02
G9 6.97 0.02
G18 7.19 0.02
G22 7.48 0.02
G25 7.60 0.02
PS1500 6.00 0.01
Photosil 6.39 0.02
1062 H.-J. YOON ET AL.
at EPFL Scientific information and libraries on June 20, 2012jim.sagepub.comDownloaded from
