Smart Materials and Structures

You may also like

- Tensile and fatigue testing of impacted
New concept for modeling the electromechanical smart CFRP composites with embedded
PZT transducers for nonlinear ultrasonic
behavior of unidirectional carbon-fiber-reinforced monitoring of damage evolution
Christos Andreades, Michele Meo and
Francesco Ciampa
plastic under tensile loading
- Effect of interface and matrix strength on
transverse tensile strength of unidirectional
To cite this article: Jae Beom Park et al 2003 Smart Mater. Struct. 12 105 CFRP composite
Pengfei Bu, Wenjun Ruan and Jiangong
Liu

- Inhibiting the Oxygen Reduction Reaction

Kinetics on Carbon Fiber Epoxy
View the article online for updates and enhancements. Composites Through Diazonium Surface
Modification-Impacts on the Galvanic
Corrosion of Coupled Aluminum Alloys
Isuri N. Dammulla and Greg M. Swain

This content was downloaded from IP address 185.76.131.70 on 18/04/2024 at 08:24

INSTITUTE OF PHYSICS PUBLISHING SMART MATERIALS AND STRUCTURES
Smart Mater. Struct. 12 (2003) 105–114 PII: S0964-1726(03)56159-6

New concept for modeling the

electromechanical behavior of
unidirectional carbon-fiber-reinforced
plastic under tensile loading
Jae Beom Park1 , Tomonaga Okabe2,3 and Nobuo Takeda1
1
Graduate School of Frontier Sciences, The University of Tokyo, c/o Takeda Laboratory,
Komaba Open Laboratory (KOL), 4-6-1 Komaba, Meguro-ku, Tokyo 153-8904, Japan
2
Smart Structure Research Center, AIST, 1-1-1 Umezono, Tsukuba, Ibaraki 305-8568, Japan

Received 5 February 2002, in final form 4 September 2002

Published 10 January 2003
Online at stacks.iop.org/SMS/12/105
Abstract
In this paper, the electromechanical behavior of carbon-fiber-reinforced
plastic (CFRP) composites under tensile loading is investigated
experimentally and analytically. We review the measurement method of the
electrical resistance change in CFRP and discuss the effect of the electrode
configuration on the measured values. For this purpose, the measurement
results of the different electrode configurations, DC two- and four-probe
methods, are compared and the internal voltage distribution is investigated.
Using the selected test procedure, the electrical resistance change curve of
the specimen with different fiber volume fractions is obtained and the
electrical ineffective length, which was previously proposed as a new
concept for the electromechanical modeling of CFRP, is reviewed. Then, a
Monte Carlo simulation, which combines the fiber fragmentation process
and the concept of the electrical ineffective length, is conducted and
compared with the experimental and the analytical results. The new concept
and methodology for the modeling of the electromechanical behavior of
CFRP composites is established concretely.

1. Introduction and electrical anisotropy is inevitable. In our previous

study [6], we proposed a new concept of electrical ineffective
As shown in previous studies [1–5], the electrical conductivity length δec , over which the broken fibers cannot function as
of carbon-fiber-reinforced plastic (CFRP) composites can be a conducting path. Introducing this new concept makes it
a very useful property for the detection of internal damage. possible to put in order the internal conducting network using
Under mechanical loading, the conductive carbon fibers in the average distance between adjacent contact points, and the
CFRP are strained and broken gradually. This mechanical electrical resistance change can be calculated by an analytical
damage causes an increase of the electrical resistance, form.
thus the ‘integrity’ of CFRP composite structures can be The object of this study is to provide a quantitative
estimated by monitoring the resistance change. However, to evaluation of the electrical resistance change measurement
establish an effective structure health monitoring technique, method and to prove the validity of the previously proposed
the experimental and the numerical analytical tools should be concept of electrical ineffective length. For this purpose, we
developed concretely. In particular, considering the contact review several types of electrical DC measurement procedure
behavior between conductive carbon fibers, which is caused of the resistance change in CFRP specimen and suggest
by the misalignment of the carbon fibers included in CFRP, the the optimum procedure. Using the suggested measurement
internal conducting path of CFRP shows strong randomness, technique, the resistance change curve for the specimen with
different fiber volume fractions are obtained, and the electrical
3 Author to whom any correspondence should be addressed. ineffective length δec is also estimated. Then, Monte Carlo

0964-1726/03/010105+10$30.00 © 2003 IOP Publishing Ltd Printed in the UK 105

J B Park et al

pressure Table 1. Material properties of the T700S carbon fiber.

epoxy resin Properties T700S

(Epikote828/TETA) Young’s modulus, E (GPa) 230
Elongation, εult. (%) 1.9
Weibull scale factor, σ0 (MPa) 4100
Weibull shape factor (m) 4.76
Gage factor α 5.14
Characteristic length, L 0 (mm) 15
CFRP pre-preg
hot press

epoxy matrix composite were used. The material properties of

pressure T700S carbon fiber are listed in table 1. The composite was
processed by prepreg molding in a hot press (2 h at 130 ◦ C
Figure 1. Schematic representation of the manufacturing process of and 40 GPa). To prepare the specimens with different fiber
CFRP laminates with different volume fractions. volume fractions, the epoxy resin, which had been made of
Epikote 828 epoxy resin and TETA curing agent (11 wt% of
epoxy resin), was inserted between layers during the lay-up
process (figure 1). As shown in figure 2, the laminates obtained
have different fiber volume fractions and the distance between
fibers varies with fiber volume fraction. Therefore, the contact
situation of the embedded fibers is different for each laminate
and this makes it possible to prepare the specimen with
different contact characteristic. The coupon-type specimens
as shown in figure 3 were prepared from these laminates for
the experimental investigations.

3. Review on resistance measurement methods

3.1. DC two- and four-probe methods

The general procedure to evaluate the resistivity of the
electrically conductive solid materials is to measure the voltage
difference between the electrodes under constant electrical
current condition. However, several types of configuration
Figure 2. Micrograph showing the cross sections of CFRP exist for the arrangement of the electrodes [7–9] and the
laminates. (a) V f = 52%. (b) V f = 27 %. (c) V f = 16%. measurement results of each electrode configuration are
different in absolute values and even in tendencies [10]. Wang
110mm et al [10] carried out an experimental study on the effect of
electrode configuration on the measured resistance values and
explained that the difference in absolute values and tendency
in DC two- and four-probe methods stems from the contact
5mm resistance of the two-probe method. However, their result is
limited to the case of very small strain below the 0.1% level
and, at this strain level, the breakage of conductive carbon
fiber, which is the major mechanism of the electrical resistance
t change of CFRP (and to which much attention has been
paid), cannot occur. Therefore, experimental and theoretical
Figure 3. Coupon-type specimen used for the experimental investigations on the effect of the electrode configurations on
investigation. The specimen thickness varies according to the
amount of inserted epoxy resin. the electrical resistance change of CFRP due to fiber breakage
are needed.
The resistance change of CFRP under mechanical tensile
simulation using the concept of the fiber fragmentation process
loading was measured using the DC two- and four-probe
and the concept of δec is conducted. The experimental and the
methods. Figure 4 shows the specimen shape and the electrode
numerical results show that the proposed concept of δec has
configurations of the two- and four-probe methods. For
concrete physical meaning as a unique parameter representing
the DC two-probe method, two electrodes were set on the
the internal conducting network of CFRP.
polished surface of the specimen using silver paste (figure 4(a))
2. Materials and the input of current and measurement of voltage were
conducted using the same electrodes. For the DC four-probe
For the CFRP specimen, carbon-fiber/epoxy unidirectional method, we considered two types of configurations denoted
laminates made of a Toray T700S-type carbon-fiber/2500 as configuration (b) and (c), where the current electrodes were

106
 Modeling the electromechanical behavior of unidirectional CFRP

(a) (b)
silver paste polished surface
: electrode for current input polished surface
: electrode for voltage measurement

GFRP tab
15mm 15mm

(c) polished surface

15mm

Figure 4. Electrode configurations for the resistivity measurement. (a) DC two-probe method. (b) DC four-probe method with current
electrodes at edges of the specimen. (c) DC four-probe method with current electrodes at surface of the specimen.

100 absolute values of the measured resistance, possibly due to the

contact resistance, the overall tendency of the resistance rate
: DC 2 probe (a)
change curves shows a similar behavior. A linear increase at
80 : DC 4 probe (b) lower strain level and a nonlinear steep increase at higher strain
: DC 4 probe (c) level can be observed in both configurations. In particular,
∆ R/Ro(%)

60 the results of the DC four-probe methods show the same

result despite the different electrode configurations. In our
40 results, the decrease of the resistance at lower strain level
for the case of the DC four-probe method as shown in the
results of Wang et al [10] cannot be identified. It is not
20
certain why this difference exists between ours and their
results. However, the manufacturing process and the prepreg
0 of ours are different from theirs and the fiber misalignment
0 0.005 0.01 0.01 .02 angle, which was mentioned by them as the main reason for
strain (ε ) the resistance reduction at the small strain level, is strongly
affected by the prepreg situation and the flow of the epoxy
Figure 5. Electrical resistance change curves measured by three resin during the curing process. Therefore, the difference of the
different electrode configurations for the specimen with a 52% fiber fiber alignment condition is a possible reason for the different
volume fraction.
electrical resistance change at the low strain level, especially
located either on the cut edge of the specimen or on its polished below the 0.1% strain. Compared to the results of the DC four-
surface. The electrodes for the voltage measurement were probe methods, the nonlinear increase of the DC two-probe
located on its surface for both configurations. method starts at lower strain level and the measured resistance
values are higher. These differences might originate from
differences in the two measurement methods since the steep
3.2. Measurement results
nonlinear increase in resistance correlates to the carbon fiber
Figure 5 shows the measurement results of the specimen with breakage for tested specimens manufactured from the same
52% of the fiber volume fraction. Despite the difference in the materials and process. To investigate the reasons why these

107
J B Park et al

x=0 Table 2. Top and bottom voltage measurement results of the CFRP
specimen with 52% of fiber volume fraction (I = 1 mA).
Electrode position x (mm) Top (mV) Bottom (mV)

10 0.17 0.09
top view 20 0.34 0.18
30 0.51 0.26
40 0.68 0.34
50 0.90 0.42
60 1.11 0.48
70 1.38 0.53
bottom view 80 1.69 0.59

z and further from Vtop

   
x Vtop b π D ρz 1/2
(ρz ρx )1/2
≈ tanh ,
Figure 6. Schematic representation of the attachment of the 2I sin[π(x2 − x1 )/2L] L ρx
contacts and the coordinate system of the CFRP specimen used for (5)
the investigation of the internal voltage distribution. where x1 and x2 denote the coordinates of the voltage
electrodes.
differences occur, we introduced a linear anisotropy resistivity
model and analyzed the internal voltage distribution, which 3.4. Measurement methods using internal voltage distribution
can provide good information in explaining the experimental To produce the internal voltage distribution map, the top
results. and the bottom voltage differences of the CFRP specimen
with 52% of fiber volume fraction were measured. Based
3.3. Linear anisotropic resistivity model on the measured voltage data as listed in table 2 and using
equations (3)–(5), the voltage distribution map was plotted as
To take account of the voltage distribution in CFRP composite,
shown in figure 7(a). Figure 7 provides important information
we proceed with the linear anisotropic resistivity model as
about the experimental results of figure 5. Considering that
shown in the studies of the Busch et al [11] and Gonzalez the electric current flows in the direction perpendicular to
et al [12]. In the quasistatic limit, the following differential the equipotential line, the internal current distribution of
equation for the two-dimensional voltage distribution V (x, z) configuration (c) is very similar to the case of configuration (b),
must be satisfied: in which the electrical current flows parallel to the x direction
1 ∂2V 1 ∂2V (figure 7(b)). This can explain the same resistance change
div j = + = 0, (1) curve of the two DC four-probe methods despite the difference
ρx ∂x 2 ρz ∂z 2
in the current electrodes’ position. Moreover, as the position
where ρx and ρz are the resistivities which correspond to the x of the voltage electrode approaches the position of current
and z coordinates of the coordinate system included in figure 6, electrode, the equipotential line approaches the surface of the
respectively. An approximate solution is given by an expansion specimen. Therefore, for the case of the DC two-probe method
in which the voltage and the current electrodes share the

   1/2 
nπ x ρz nπ z same position, the measured voltage values might be strongly
V (x, z) = Vn sin cosh affected by the condition of the specimen surface. However,
n=1,3,5,...
L ρx L
as mentioned in section 3.1, the surface of the specimens was
(2)
polished, thus the strength of carbon fibers at the specimen
where L is the specimen length. By applying the boundary
surface decreased considerably. To evaluate the effect of the
conditions and approximating V (x, z) by the lowest n = 1
surface polishing on the fiber strength, acoustic emission (AE)
term alone, the voltage distribution of CFRP composite with
evaluation was carried out. As shown in figure 8, the AE
configuration of figure 4(c) can be obtained as given by
signals during the tests were detected by a piezoelectric sensor
I sin(π x/L) (type R15) attached to the sample by silicon grease and an
V (x, z) ≈ − (ρz ρx )1/2
b sinh[(π D/L)(ρz /ρx )1/2 ] adhesive tape. Its operating frequency range was between 10
  1/2  and 2000 kHz. The signals were amplified by a pre- and a
π z ρz
× cosh . (3) post-amplifier with a total gain of 40 dB. The threshold of the
L ρx
AE system was 45 dB. AE test were carried out for two types
The detailed development of equation (3) is in appendix A. of specimen: with polished surface and with non-polished
The resistivity terms in equation (3) can be substituted by the surface. Figure 9 shows a comparison of the AE measurement
measurable values of the top and the bottom surface voltages. results. Due to the drop in the strength of the carbon fiber
From the ratio of the top and the bottom surface voltages, we which is located at the surface of CFRP, the event number of
obtain  1/2 the polished specimen which corresponds to the fiber breakage
ρz L shows higher values than that of the unpolished specimen.
≈ arccos h(Vtop / Vbot ) (4)
ρx πD Therefore, the differences between the resistance measurement

108
 Modeling the electromechanical behavior of unidirectional CFRP

(a) (b)
z
0.5
0.7 I
0.6
0.4
0.4 0.5 x
0.4
Configuration (b)
position z (mm)

0.3
0.3

0.2 0.3 V

0.2
0.2
0.1 I
z
0.1
0.1

x
0.0
55 10
10 15
15 20
20 25
25 30
30 335 40 Configuration (c)

position x (mm)

Figure 7. Comparison of the configuration (b) and (c) using the linear anisotropic resistivity model. (a) Equipotential lines of the
configuration (c) calculated using the measurement results of table 2 and equation (3). (b) Schematic representation showing the internal
current distribution.

(a) (b)
ε

load cell

15mm
grip
0.5mm

specimen

AE sensor
AE sensor strain gage
(R15)
preamplifier
20mm

bridge box strain amplifier

Figure 8. Schematic representation showing the specimen and the test system of AE measurement. (a) Test specimen with AE sensor.
(b) Test system (Physical Acoustics Corp.) with AE sensor and preamplifier.

results of the DC two- and four-probe method, i.e. faster start each of length equal to the average electrical ineffective length
of the steep increase and higher resistance values, might be δec [6]. The resistance change of the proposed DC circuit due
caused by the effect of the surface polishing and not by the to the mechanical tensile loading is given by
differences in the strength of the CFRP specimen. This can
be shown later quantitatively using the electrical ineffective R (1 + αε)
=   δec  E f ε m − 1 (6)
length, which demonstrates the internal contact condition of R0 exp − L 0 σ0
the CFRP composites.
where R/R0 is the normalized resistance change, E f is the
4. Fiber volume fraction effect on behavior Young modulus of the carbon fiber, α is the gage factor of
carbon fiber, ε is the applied strain, and m and σ0 are the Weibull
4.1. Electrical ineffective length δec shape and scale parameters, respectively, corresponding to a
reference length, L 0 . Equation (6) indicates that the electrical
Figure 10 shows a micrograph of the CFRP (T700S carbon resistance change of CFRP composites is governed by δec
fiber/2500 epoxy resin) under tensile loading and the DC circuit which represents the internal electrically conductive network
model consisting of a serial array of discrete parallel cells, due to the contacts between conductive carbon fibers. As

109
J B Park et al

2500 : DC 2 probe (a), : DC 4 probe (c)

100

2000
Cumulative event counts

80
polished surface
δec= 4mm, 0.77σ o
1500

∆ R/Ro(%)
60
unpolished surface
1000 δec= 4mm, 1.00σ o
40

500
20

0
0 0.004 0.008 0.012 0.016 0
Strain (ε ) 0 0.005 0.01 0.015 0.02

Figure 9. Comparison of AE cumulative event counts of the strain (ε )

polished and unpolished specimens.
Figure 12. Estimation of the electrical ineffective length δec by
comparison between experimental and calculated results
ε (V f = 52%).

the electrically ineffective length over which the conductive

δec carbon fiber cannot function as an electrically conductive path.
Therefore, the loss of the conductive path due to the fiber
fiber breakage
breakage is restricted to the distance between adjacent contact
points; thus a carbon fiber sustains its function of a conductive
path until every fiber segment divided by the contact point
contact between
fibers experiences breakage. Figure 10 and equation (6) describe
this situation very effectively by introducing the average
distance between adjacent contact points δec . Figure 12 shows
the comparison between the experimental and the calculated
results of the resistance change under tensile loading for the
case of specimen with 52% of fiber volume fraction. For
the case of the DC four-probe method, we used the material
ε properties as shown in table 1. However, for the case of the DC
two-probe method, we used the adjusted strength of the carbon
Figure 10. Schematic representation of the equivalent DC electrical fiber, 77% of the original strength, which was estimated based
circuit of CFRP laminates. The DC circuit consists of serial arrays
on the AE measurement results (appendix B). The estimated
of discrete parallel cells of length δec .
values of δec for both measurement results show the same value
i: electric current
of 4 mm. This shows the validity of δec as the parameter
representing the internal contact situation and the discussion
of the measurement methods in the previous section.

4.2. Effect of the conductive carbon fiber volume fraction

i
As mentioned in section 4.1, the resistance change behavior of
: electrical node CFRP is governed by the internal fiber contact situations. Since
it is natural that the fiber contact situation depends on the fiber
volume fraction, we investigated the resistance changes of the
specimens with different fiber volume fractions under tensile
i loading. The DC four-probe method with configuration (c)
was used since this method has the advantage of maintaining
0
sound contact between specimen and electrode compared to the
method of configuration (b), and it can eliminate the effect of
electrical ineffective length
the surface polishing as discussed in section 3. The measured
Figure 11. Schematic representation of the electrical ineffective resistance change curves show strong dependence on the fiber
length originated from the contact behavior of conductive carbon volume fraction. As shown in figure 13, the nonlinear increase
fibers. of the resistance change with lower volume fraction starts
more quickly than that of CFRP with higher volume fraction.
shown in figure 11, after the fiber breakage occurs, the These experimental results can be quantitatively explained by
existence of the contact points on the carbon fibers represents a volume-fraction-dependent value of δec and equation (6).

110
 Modeling the electromechanical behavior of unidirectional CFRP

: V = 52%
f
: V = 27%
f
: V = 16%
f
distance between contact points has the same length. However,
120 considering the real situation, the contact points are distributed
randomly, thus the length of the inter-contact points distance
varies considerably. In this section, we conduct a Monte Carlo
90 simulation to consider the random distribution of the contact
δec 50 mm
points and compare the simulated results to the analytical
∆ R/Ro(%)

results of equation (6). For this purpose, we consider a long

60 δec 20 mm conductive wire as shown in figure 14, over which electrical
nodes are randomly distributed, as a substitute for the electrical
δec 4 mm circuit with random contact points. In this way, it is possible
30 to apply the fiber fragment process of which the numerical
and the analytical solutions have been established by various
researches [13, 14].
0
0 0.005 0.01 0.015 0.02
5.2. Monte Carlo simulation procedure
Strain (ε )
Under the tensile strain and its subsequent fiber elongation
Figure 13. Experimental results of resistance change for different and breakages, the resistance change of the conductive wire
fiber volume fraction and fitted values of δec using equation (6). with randomly distributed electrical nodes, through which the
electric current can freely flow in and out, can be given by
As the fiber volume fraction decreases, it is obvious that the
R (1 + αε)
number of fiber contacts decreases and hence that δec increases, = −1 (7)
as found in figure 13. The increase of δec implies that the length R0 le f f /l0
of the resistors in the equivalent circuit for specimen with lower where l0 is the initial length of conductive wire and le f f is
volume fraction is longer than that of higher volume fraction, the remaining effective conductive path at the strain level of
and the longer fibers break at lower strains. This explains the ε (figure 15). Therefore, the term of le f f /l0 can represent the
experimental data showing the earlier start of the nonlinear survival rate of the electrical path at strain ε and is equivalent to
increase in resistance for lower fiber volume fractions. Note the denominator of equation (6), which represents the survival
that all other material properties are held fixed in making the rate of the conductive carbon fiber within the parallel cell of
predictions in figure 13. length δec . To estimate the value of le f f at each strain level, we
consider a carbon fiber with length l0 embedded in an epoxy
5. Monte Carlo simulation resin matrix under a unidirectional tensile strain. The fiber is
represented as a series of n jointed small elements with length
5.1. Considerations for the general contact situation l and the strength of each fiber element was assigned by the
Weibull equation:
As identified by the schematic diagram of figure 10 showing
the equivalent DC circuit, equation (6) assumes that every     1/m
L0 1
σult,i = σ0 ln (8)
l 1− F
(a) (b)

: conductive carbon fiber, : contact between fibers

Figure 14. Schematic representation of the transformation of the parallel cell of resistors to a long conductive wire. (a) Unit parallel cell
with random contact points. (b) Transformation of the unit cell to a long conductive wire over which the electrical nodes are distributed
randomly.

111
J B Park et al

a b c d e : Eq. (6) : simulation

120

Estimated δec (mm)

lo = a + b + c + d + e, leff = b + d + e 90 V (%) sim. Eq. (6)

f

Figure 15. Schematic diagrams showing the calculation scheme of 52 3 < 4

∆ R/Ro(%)
the effective conductive path. 27 11 < 15
60 16 27 < 50
120

: simulation 30

90 : V = 52%
f
: V = 27%
f 0
∆ R/Ro(%)

: V = 16% 0.0 0.0050 0.010 0.015 0.020

f
60
Strain (ε )

Figure 17. Comparisons between the best fit lines of the Monte
30 Carlo simulation and the analytical calculation using equation (6).

different. Considering the Weibull–Poisson statistics [16], the

0
0 0.005 0.01 0.015 0.02 possibility of the occurrence of a fiber breakage is proportional
to the fiber length. Therefore, the increase of the resistance
Strain (ε ) change due to the decrease of effective conductive path for
the case of random contact points starts at lower strain level
Figure 16. Monte Carlo simulation results for different fiber volume
fraction (l = 100 µm, l0 = 2 m). compared to that of equal-distance contact points, due to the
existence of the relatively long inter-contact-point distances
(figure 18(a)). In contrast, the existence of very short distances
where σult,i is the strength of the i th element and F is the
between contact points in the random case requires a very
random number from 0 to 1. For the calculation of the stress
high level of strain for a carbon fiber to lose its function
in the fiber at a distance x from the broken fiber end, we used
as an electrically conductive path. These differences can
the Kelly–Tyson model [15] as given by
be identified in figure 18(b), which shows the comparisons
x between two results using the same value of δec over the strain
σ f = 2τ y (9)
r range up to 3%. The magnification of the difference in δec for
the case of lower volume fraction gives support to the above
where τ y is the shear stress at the fiber/matrix interface and explanations.
r is the fiber radius. Whether the breakage of fiber element
occurs or not was judged by comparing the calculated stress
σ f to the assigned strength of each element σult,i , and the 6. Conclusion
effective conductive path was assessed by checking the mutual
relationship between the positions of the fiber breakages and We reviewed the measurement method of the resistance change
the electrical nodes. in CFRP composites under tensile loading and checked the
previously proposed concept of the electrical ineffective length
using the experimental and the numerical investigations. For
5.3. Comparisons and discussions
the measurement method, we found that the DC four-probe
For the simulation, the material parameters of carbon fiber as method with configuration of figure 4(c) can be recommended
listed in table 1 and Young’s modulus of the epoxy matrix for the assessment of the internal damage of CFRP composites.
and the interfacial shear stress, 3.17 GPa and 45.4 MPa, In addition to the higher values of the measured resistance
respectively, were used. We adjusted the contact situation data due to the electrode contact resistance, the measurement
to obtain the best fit line for the experimental results. results of the DC two-probe method, which shows very
Figure 16 shows the comparisons between the simulated and sensitive response compared to that of the DC four-probe
the experimental results. As mentioned in section 4, the methods, were affected by the surface condition such as
experimental lines for the different volume fractions can be polishing, thus the fall in fiber strength due to the surface
predicted effectively by adjusting the electrical ineffective polishing should be considered effectively in analyzing the
length δec . However, as shown in figure 17, the estimated measurement results. The experimental results of the specimen
values of δec for each volume fraction using the Monte with different fiber volume fractions and the Monte Carlo
Carlo simulation are different from the values obtained by simulation results strongly suggested that the concept of
equation (6). This might originate from the differences in the electrical ineffective length is valid, although the analytical
schemes for arranging the position of contact points. As shown model of equation (6) using the equal inter-contact-point
in figure 18, for the carbon fiber with randomly distributed distance is too ideal to describe the random distribution of
contact points, the length of distances between contact points is contact points in CFRP composites. Based on the results of

112
 Modeling the electromechanical behavior of unidirectional CFRP

(a) (b)
1500
: simulation
l

: Eq. (6)
1000

∆ R/Ro(%)
Pf ,l (ε ) = 1 − exp[−( l Lo )( E f ε σ o ) m ]

a b c d e 500

a<d <b<e<c
Pf , a (ε ) < Pf , d (ε ) < Pf ,b (ε ) < Pf ,e (ε ) < Pf ,c (ε ) 0
0.005 0.01 0.015 0.02 0.02 .03

Strain (ε )

Figure 18. Explanations of the difference in δec between the Monte Carlo simulation and analytical calculation using equation (6) (see text
for details). (a) Fracture probability P f of the each line segment of the Monte Carlo simulation varies according to the distance between
contact points. (b) Comparison between calculation results of the two methods for the case of the same value of δec (δec = 4 mm).

this study, we will embark on our next study on the correlation 2500
between electrical anisotropy and δec , which is indispensable polished surface
for CFRP composites that are to be used as a smart patch sensor. 2000
Cumulative event numbers

unpolished surface

Acknowledgments unpolished surface

1500 (adjusted by 0.77σo)

This research was conducted as a part of the ‘R&D

for Smart Materials/Structure System’ project within the 1000
Academic Institutions Centered Program supported by
NEDO (New Energy and Industrial Technology Development 500
Organization), Japan.
0
Appendix A 0 0.005 0.01 0.015
Strain (ε )
Given

Figure A.1. Estimation of the adjusted Weibull scale parameter for
V (x, z) = Vn sin(nπ x/L) cosh[(ρz /ρx )1/2 nπ z/L], the polished surface specimen based on the ratio of the AE results of
n=1,3,5,... the surface-polished and -unpolished specimens.
the current density Jx (x, z) can be obtained through Jx =
−(1/ρx )(∂V (x, z)/∂x), i.e. Appendix B


Jx (x, z) = Vn (nπ/L) cos(nπ x/L) According to the Weibull–Poisson model, the mean number of
n=1,3,5,...
defects per unit length with strength less than or equal to σ is
× cosh[(ρz /ρx )1/2 nπ z/L]. given by [15]
We may approximate V (x, z) by the lowest n = 1 term λ(σ ) = (1/L 0 )(σ/σ0 )m
alone. From the condition that the applied current I must
equal the integrated current density in the midplane x = 0 of where m and σ0 are the Weibull shape and scale parameters,
the specimen, i.e. respectively, for the strength of fibers of length L 0 . It is certain
that the AE cumulative event count number Nc is proportional
D to the mean number of flaws, thus the ratio of the Nc for the
I =b V1 (π/L) cosh[(ρz /ρx )1/2 π z/L] dz polished and the unpolished specimens is given by
0

where b and D are the width and thickness of the specimen, (Nc,up /Nc, p ) = (σ0, p /σ0,up )m
respectively. The coefficient V1 is then uniquely determined
as given by where Nc,up , Nc, p , σ0,up and σ0, p are the AE cumulative event
number and the Weibull scale parameter of the unpolished
I (ρx ρz )1/2 and the polished specimens, respectively. Based on the
V1 = − .
b sinh[(ρz /ρx )(π D/L)] experimental results of figure 8 and the above relationship,
we found the value of 0.77σ0,up as the adjusted Weibull

113
J B Park et al

scale parameter for the polished surface CFRP specimen [8] Logan B F, Rice S O and Wick R F 1971 Series for computing
(figure A.1). current flow in a rectangular block J. Appl. Phys. 42
2975–80
[9] Wan Y M, Lemberger T R, Hebboul S E and Garland J C 1996
References Transport properties of high-Tc Bi2 Sr2 CaCu2 O8+δ crystals
near the superconducting transition Phys. Rev. B 54 3602–9
[1] Prabhakaran R 1990 Damage assessment through electrical [10] Wang S and Chung D D L 2000 Piezoresistivity in continuous
resistance measurement in graphite fiber reinforced carbon fiber polymer–matrix composite Polym. Compos. 21
composites Exp. Tech. 14 16–20 13–9
[2] Schulte K and Baron Ch 1989 Load and failure analysis of [11] Busch R, Ries G, Werthner H, Kreiselmeyer G and
CFRP laminates by means of electrical resistivity Saemann-Ischenko G 1992 New aspects of the mixed state
measurements Comput. Sci. Technol. 36 63–76 from six-terminal measurements on Bi2 Sr2 CaCu2 Ox single
[3] Chung D D L 2001 Structure health monitoring by electrical crystals Phys. Rev. Lett. 69 522–5
resistance measurement Smart Mater. Struct. 10 624–36 [12] Gonzalez J L, Espinoza Ortiz J S and Baggio-Saitovitch E
[4] Abry J C, Bochard S, Chateauminois A, Salvia M and 1999 Influence of the anisotropy in the c-axis resistivity
Giraud G 1999 In situ detection of damage in CFRP measurements of high-Tc superconductors Physica C 315
laminates by electrical resistance measurements Comput. 271–7
Sci. Technol. 59 925–35 [13] Okabe T and Takeda N 2001 Estimation of strength
[5] Muto N, Yanagida H, Nakatsuji T, Sugita M and Ohtsuka Y distribution for a fiber embedded in a single-fiber
1993 Preventing fatal failure in composite: experiments and statistical simulation based on
carbon-fiber-glass-fiber-reinforced plastic composites by the elasto-plastic shear-lag approach Comput. Sci. Technol.
monitoring change in electrical resistance J. Am. Ceram. 61 1789–800
Soc. 76 185–8 [14] DiBenedetto A T and Gurvich M R 1997 Statistical simulation
[6] Park J B, Okabe T, Takeda N and Curtin W A 2002 of fiber fragmentation in a single-fiber composite Comput.
Electromechanical modeling of unidirectional CFRP Sci. Technol. 57 543–55
composites under tensile loading condition Composites A [15] Kelly A and Tyson W R 1965 Tensile properties of
33 267–75 fiber-reinforced metals: copper/tungsten and
[7] Montgomery H C 1971 Method for measuring electrical copper/molybdenum J. Mech. Phys. Solids 13 329–50
resistivity of anisotropic materials J. Appl. Phys. 42 2971–5 [16] Weibull W A 1952 A statistical distribution function of wide
applicability ASME J. 293–7

114