Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Characterization of the interlaminar fracture toughness

of a laminated carbon/epoxy composite


Mary J. Mathews, Stephen R. Swanson
*
Department of Mechanical Engineering, University of Utah, Salt Lake City, UT 84112, USA
Received 21 July 2005; received in revised form 11 July 2006; accepted 20 July 2006
Available online 6 October 2006
Abstract
Delamination between layers is an important problem in applications of ber reinforced composite laminates. Tests were carried out
to determine the interlaminar fracture toughness of AS4/3501-6 (carbon/epoxy) composite laminates using mixed-mode bending tests.
Analysis of the test specimens in terms of mode I and mode II energy release rates showed good agreement between methods based
on beam equations, compliance measurements, and detailed nite element analyses. The results showed that the critical mode I energy
release rate for delamination decreased monotonically with increasing mode II loading. This is in contrast to some results in the liter-
ature. Various analytic representations of the mode interaction from the literature were compared, and shown to t the data with rea-
sonable accuracy.
2006 Elsevier Ltd. All rights reserved.
Keywords: C. Delamination; B. Fracture toughness; B. Mixed-mode fracture
1. Introduction
Interlaminar delamination is a common and potentially
dangerous mode of failure in composite structures. Delam-
ination often results in the loss of stiness and strength,
which may lead to safety and reliability problems. Unde-
tected subsurface delamination can lead to catastrophic
failures without any external signs. This makes delamina-
tion a major obstacle in achieving the full weight saving
potential of advanced composite materials [1,2].
Delamination in composites is often a mixed-mode frac-
ture. Both interlaminar tensile and shear stresses can be
present at the delamination front. Interlaminar tensile
stresses give rise to mode I fracture while interlaminar
shear stresses result in mode II fractures. Using fracture
mechanics to characterize the onset and growth of delami-
nation has become a generally accepted practice. It is well
established that the fracture toughness of materials can be
measured by the critical energy release rate G
Ic
for pure
mode I and G
IIc
for pure mode II. While mode I fracture
dominates the failure in isotropic materials, the interaction
of shear and tensile fracture complicates the mechanism of
failure in composite materials. Delamination resistance of a
composite laminate under mode I fracture is dierent from
the delamination resistance under mode II fracture. The
mechanism of fracture also varies with the ratio of the dif-
ferent modes of fractures involved [1,35]. This interaction
of shear and tensile fractures complicates the mechanism of
failure in composite materials [2,68].
Because of the typical involvement of both modes I
and II, a rule for relating failure under combined modes
to that of the more readily available individual modes is
required. Several empirical relationships have been sug-
gested as delamination criterion in the literature [912].
Russell and Street [9] derived a failure criterion based
on the energy release rate calculated using the so-called
super ellipse stress intensity criterion suggested by Spencer
and Barnby [13]. Several authors [10,14] have shown that
the coecients of the super ellipse change with the mate-
rial. Reeder and Crews [4] have shown that a linear form
0266-3538/$ - see front matter 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compscitech.2006.07.035
*
Corresponding author. Tel.: +1 801 581 6407; fax: +1 801 585 9826.
E-mail address: stephen.swanson@utah.edu (S.R. Swanson).
www.elsevier.com/locate/compscitech
Composites Science and Technology 67 (2007) 14891498
COMPOSITES
SCIENCE AND
TECHNOLOGY
of this criterion is adequate for AS4/PEEK composites.
Experimental results by Rhee [15] also indicated a linear
G
I
G
II
behavior. However, Hashemi et al. [11,16]
obtained a nonlinear failure locus for delamination prop-
agation in the above material. Failure loci for glass/epoxy
developed by Ducept et al. [17], for carbon /epoxy derived
by Hashemi et al. [11], Kinloch et al. [18], Reeder [12],
and Singh and Patridge [19] are nonlinear. A delamina-
tion criterion was developed by Hashemi et al. [11,16]
using a modes III interaction parameter. These parame-
ters appear to be more accurate for the PEEK (tough)
composites rather than the epoxy (brittle) composites,
especially when the mode I to mode II ratio is high. Ree-
der [12] has evaluated the parameters for dierent delam-
ination criteria using his experimental results. The criteria
he considered are power law criterion (super ellipse),
exponential hackle criterion, exponential K
I
/K
II
criterion
and interaction criterion. However, it should be noted
that the data that Reeder used for this evaluation had
G
I
values higher than G
Ic
for some mixed-mode fracture
conditions. Further discussion of this aspect will be given
below. An additional form for the mode interaction has
been presented by Benzeggagh and Kenane [5] which will
also be discussed further below.
As indicated above, attention has been given in the lit-
erature to mixed mode delamination fracture. However,
the review of the literature given above shows some dis-
crepancies between the results of dierent investigators,
both with regard to data reduction and to trends of the
interaction between mode I and mode II fracture. The
objective of this paper was to shed further insight into
these problem areas. To carry this out, combined loading
delamination fracture tests were performed on AS4/3501-
6 (carbon/epoxy) composite laminates under pure mode
and mixed mode loading conditions. The fracture tough-
ness thus obtained was used to develop failure loci using
dierent failure criteria and to evaluate the parameters
involved. Further assessments of data analysis procedures
are also considered.
2. Fracture toughness measurements of AS4/3501-6
composite laminate
Delamination fracture was investigated for pure mode I
and mode II, and for combinations of modes I and II.
Double cantilever beams (DCB) were used to determine
the mode I fracture toughness while end-notch exure
(ENF) beam as well as end-loaded split (ELS) beam cong-
urations were used for the measurement of the mode II
fracture toughness. Toughness was evaluated for dierent
ratios of mode I and mode II fractures using a mixed-mode
bending (MMB) conguration. The experimental materi-
als, tests, and data analysis procedures are described as
follows.
Specimen preparation and equipment the current prac-
tice for measuring the fracture toughness is to use unidirec-
tional laminates to obtain the material toughness for
dierent modes of fracture in ber reinforced laminated
composites. The specimens were prepared from 150 mm
(6 in.) wide AS4/3501-6 carbon/epoxy prepreg tapes.
Twenty-four unidirectional plies were assembled to form
a 150 by 150 mm panel. A thin Teon lm (thickness
<13 lm) was inserted at the mid plane to form an initial
delamination. The panels were then cured using a hot plate.
Specimens were cut out of each panel using a diamond
rotary cutter. No additional precracking beyond that of
the teon insert was used before testing.
An Universal Testing Machine (Z250/SN5S) with a load
cell capacity of 250,000 kg/100 kg was used for the mode I
and mode II tests, whereas a Universal Testing Machine
with a load cell capacity of 5000 kg was used for all the
mixed-mode fracture tests.
3. Experimental techniques
3.1. Mode I fracture
The conguration of a double cantilever beam (DCB)
specimen indicated by the ASTM standard D 5528-94a
[20] was used to determine the mode I fracture toughness.
Specimens with piano hinges were used, with the hinges
bonded to the specimens after sand blasting the contact
surfaces. The edges of the specimens were coated with type-
writer correction uid to observe the propagation of the
delamination. Marks were made at every 5 mm from the
tip of the delamination to observe the crack propagation
and record the corresponding load point displacement.
Six test specimens were used for the mode I fracture
tests. The specimens were tested in a displacement mode
with a constant displacement rate of 0.5 mm/min. The load
and the load point displacements were recorded by the data
acquisition system. The extension of the delamination front
was observed and the crack opening displacement corre-
sponding to every 5 mm extension was recorded. The fail-
ure load for each case was selected as the highest load
point of the linear section of the load displacement plot
and used to determine the mode I critical energy release
rate G
Ic
.
3.2. Mode II fracture
Two methods were used for characterizing mode II frac-
ture toughness, which were the end-notched exure (ENF)
[21,22] and the end-loaded split beam (ELS) [23]. Five spec-
imens each were tested using the two congurations to
evaluate G
IIc
. The specimens were tested in a displacement
mode with a constant displacement rate of 0.5 mm/min.
The delaminated sections were wedged using a needle to
eliminate the friction of the interfaces for most of the spec-
imens, although some were tested without this wedge for
comparison. The load and the load point displacements
were recorded by the data acquisition system. The failure
load for each case was selected as the load at the onset of
1490 M.J. Mathews, S.R. Swanson / Composites Science and Technology 67 (2007) 14891498
the delamination growth at the insert to determine the
mode II critical energy release rate.
3.3. Mixed-mode fracture
The toughness for dierent ratios of mode I and mode
II was evaluated using the mixed-mode bending (MMB)
conguration developed by Reeder and Crews [4]. The
ratio of the modes can be altered by changing the dis-
tance c to the load application point as shown in
Fig. 1. The spans of the beams were approximately
100 mm with a 25 mm delamination length. Five speci-
mens were tested using the MMB beam congurations
to evaluate G
II
and G
I
for each mode ratio. The mode
ratios G
I
/G
II
used in these experiments were approxi-
mately 8/100, 25/100, and 80/100. The specimens were
tested in a displacement mode with a constant displace-
ment rate of 0.5 mm/min. As in the mode II specimens,
the delaminated sections of the specimens for low mode
I ratios were wedged to eliminate the friction of the inter-
faces. The load and the load point displacements were
recorded by the data acquisition system. The failure load
for each case was selected as the load at the onset of the
delamination growth to determine the mode I and mode
II energy release rate. Since the loading arm is reasonably
heavy compared to the load, this value and the distance
to the centre of gravity of the loading arm is also
included in the calculations.
4. Analysis to determine the energy release rates
The data analysis methods used for the determination of
the mode I and mode II fracture toughness are described
below. A modied beam theory, compliance calibration,
and the nite element method were used to analyze the
mode I and mode II test data. The material stiness prop-
erties of Youngs modulus, Poissons ratio and the shear
modulus given in [24] for AS4/3501-6 were used in the anal-
ysis. These properties are listed in Table 1.
4.1. Modied beam theory for DCB [20]
Beam theory in general over estimates the energy release
rate for mode I fractures since the theory uses a model that
is perfectly clamped at the delamination front. In practice a
rotation may occur at the delamination front. A correction
for this may be applied using the cube root of compliance
plotted against the delamination length. The corrected
energy release rate can be expressed as
(a) Load configuration for a double cantilever beam.
(b) Load configuration for an end notch flexure beam.
(c) Mixed-mode bending specimen mounted on the loading apparatus.
P
P
c
hinges
specimen
P
P
Fig. 1. Schematic representation of mode I, mode II and mixed-mode fracture test congurations.
Table 1
Properties of the orthotropic composite material AS4/3501-6 [24]
Youngs modulus,
E
xx
GPa (Msi)
Youngs modulus,
E
zz
GPa (Msi)
m
xz
Shear modulus,
G
xz
GPa (Msi)
0 ply 126.8 (18.4) 11.0 (1.60) 0.29 6.55 (0.95)
M.J. Mathews, S.R. Swanson / Composites Science and Technology 67 (2007) 14891498 1491
G
I

3Pd
2ba jDj
1
where P is the failure load, d the load point deection, b the
width of the beam, and a is the length of the delamination.
D is the correction for the delamination length as suggested
in ASTM standard D 5528-94a [20] and can be obtained as
the x-intercept of the linear least-squares t of the cube
root of the compliance plotted against the delamination
length. Compliance is dened as d/P.
4.2. The compliance calibration method for DCB [20]
In this method the corrected energy release rate is
expressed as
G
I

nPd
2ba
2
where a is the delamination length, C the corresponding
compliance of the beam, and n is a dimensionless exponent
taken as the slope of a least-square linear t of log(C) ver-
sus log(a). Details of this method can be obtained in ASTM
standard D 5528-94a [20].
4.3. Beam analysis for ENF beam and ELS beam
Based on the strain energy in the body and the compli-
ance of the structure, the expressions for the energy release
rates for ENF beam [22,25] and ELS beam [23] can be
given as
G
II

9P
2
a 0:42vh
2
16E
1
b
2
h
3
3
where
v
E
1
11G
13
3 2
C
C 1
_ _
2
_ _ _ _
1=2
; C 1:18

E
1
E
2
p
G
13
and
G
II

9P
2
a 0:42vh
2
4E
1
b
2
h
3
4
for ENF and ELS beams, respectively.
4.4. Beam analysis for MMB beam
Based on the strain energy in the body and the compli-
ance of the structure, the expressions for the energy release
rates for MMB beam [22,25] can be given as
G
I

12P
2
1
E
1
b
2
h
3
a vh
2
5
and
G
II

9P
2
2
a 0:42vh
2
16E
1
b
2
h
3
6
where
P
1
P
3c L
4L
_ _
and
P
2
P
c L
L
_ _
In the above equations E
1
and E
2
are the Youngs modulus
in the direction of the ber and the transverse directions,
respectively, G
13
the shear modulus, h the thickness of
the delaminated section of the beam, c the length of the
loading lever, 2L the span of the beam, and a is the length
of the delaminated section that is illustrated in Fig. 1.
It should be noted that while compliance measurements
could be considered for the analysis of the MMB specimen,
as discussed in [22] it has been found that better accuracy
can be achieved by using the formulas above directly.
4.5. Finite element analysis
The nite element method has become an useful tech-
nique for determining strain energy release rates for linear
elastic fracture problems. The method usually used is based
on Irwins contention that if a crack extends by a small
amount Dc, the energy release rate can be calculated
through the crack closure integral [26]. Rybicki and Kanni-
nen [27] modied this theory and developed a method to
determine the energy release rates for mode I and mode
II fractures using nite element analysis. It is assumed that
if the elements are small enough, the force at the crack tip
node is approximately equal to the force that is needed to
close the nodes at D and E behind the crack tip node C
as illustrated in Fig. 2. In the gure, D
a
is the distance
between nodes C and E. The energy release rates G
I
and
G
II
can be expressed as the limiting values given by
G
I
lim
Da!0
F
v
v
d
v
e

2bD
a
7
and
G
II
lim
Da!0
F
h
u
d
u
e

2bD
a
8
respectively, where F
h
and F
v
are the horizontal and verti-
cal forces at the crack tip node C, u
d
and u
e
are the horizon-
tal displacements of the nodes D and E behind the crack
tip, v
d
and v
e
are vertical displacements of the nodes D
and E behind the crack tip.
The failure load obtained from the experimental study is
used in a nite element analysis to determine the mode I
and mode II fracture toughness of the material. Dierent
specimens were analyzed using the nite element method
to obtain the energy release rates G
I
for tensile fractures
and G
II
for shear fractures. The two-dimensional element
42 of the nite element code ANSYS 6.0 was used to model
each of these cases. Element size in the vicinity of the crack
tip was 0.2 by 0.2 mm, which is small compared to the
delamination length of 25.4 mm. The width of the beam
1492 M.J. Mathews, S.R. Swanson / Composites Science and Technology 67 (2007) 14891498
is relatively small compared to the length and therefore, the
axial strain through the width was not constrained in the
analysis. Hence, a plane stress analysis was adopted here.
The contact element 40 was used along the delaminated
surface to prevent overlapping of the surfaces. These ele-
ments were used in such a way that they would take com-
pression and detach under tension. A previous study of the
convergence of the FEA solutions by the authors [28] has
shown that this technique produces very accurate
solutions.
5. Experimental results
The typical load displacement behavior of DCB speci-
mens is displayed in Fig. 3. The failure load for each case
was selected as the highest load point of the linear section
of the loaddisplacement plot, following the recommenda-
tion given in [22]. It can be observed from the loaddis-
placement behavior of double cantilever beams that the
loads increase slightly and stay steady as the delamination
propagates. This indicates a stable crack growth under
mode I fracture. The failure loads for the DCB specimens
were estimated from the experimental results and used to
determine the critical energy release rate for mode I frac-
ture. The results obtained using the modied compliance
method, the modied beam theory and the nite element
analysis are listed in Table 2. The modied compliance
method and the modied beam theory use the failure load
as well as the compliance of the global system as the delam-
ination propagates to determine the energy release rate.
The nite element method uses only the failure load and
the corresponding stress eld in the vicinity of the crack
tip. It can be seen in Table 2 that the results from all three
analysis methods are in good agreement.
Two types of experimental techniques were used in this
investigation to evaluate mode II toughness of AS4/3501-6.
Typical loaddisplacement plots for ENF and ELS beam
specimens are given in Fig. 4. Unlike the double cantilever
beams, the load drops at the onset of the delamination
propagation in the case of ENF and ELS beam specimens.
This is a characteristic of unstable crack growth. While the
literature indicates that ELS specimens can give stable
crack growth if the delamination size/specimen length is
suciently large, the delamination in the specimens tested
here, which had a ratio of 1/4, was unstable. The failure
loads measured from these experiments were used to deter-
mine the mode II fracture toughness of the material. The
X
Y
Coarser
mesh
Coarser
mesh
Finer mesh
E C
D

a
Size of the crack tip element.
(same as the crack extension size)

a
Fig. 2. A schematic representation of a typical model for a beam with edge delamination.
0
5
10
15
20
25
30
35
0 5 10 15 20 25
Displacement (mm)
L
o
a
d


(
N
)
Fig. 3. The loaddisplacement behavior of a DCB specimen.
Table 2
The critical energy release rate calculated in three dierent ways for mode
I fracture in AS4/3501-6
Modied
compliance
method
Modied
beam
theory
Finite element
analysis
G
I
(J/m
2
) (mean value) 111 107 116
Coecient of variation (%) 14 13 15
M.J. Mathews, S.R. Swanson / Composites Science and Technology 67 (2007) 14891498 1493
results are listed in Table 3. The results from the two test
types agree within 2%, giving condence in the results.
Typical loaddisplacement plots from the MMB tests
are shown in Fig. 5. The response is shown for each of
the mode ratios used in this study. For lower G
I
/G
II
ratios,
the load drops soon after the onset of the delamination
propagation indicating an unstable crack growth as in an
ENF beam. For higher G
I
/G
II
ratios, the loaddisplace-
ment is similar to DCB tests.
The mixed-mode failure response can be illustrated
using a G
I
vs. G
II
plot. The energy release rates obtained
using FE models and analytical models are shown using
such a plot in Fig. 6. The data from the pure mode I and
mode II (both ENF and ELS) are also given in Fig. 6. It
can be seen that the method of data analysis used does
not signicantly aect the results. The energy results G
Ic
from FEA and beam analysis varied by 57% while the
results for the mode II and mixed-mode fractures were less
than 3%. The results from the DCB, ENF, ELS, and MMB
tests, reduced using the nite element analysis, are summa-
rized in Table 4.
The ENF and ELS tests reported in Tables 3 and 4 all
used the experimental technique of wedging open the dela-
minated sections with a needle to minimize the friction at
the interfaces. For comparison, additional tests were per-
formed at two dierent mode ratios without the wedge.
No signicant dierence was obtained between the results,
indicating that this technique did not aect the results and
was thus unnecessary.
6. Discussion
6.1. Comparison with data in the literature
The primary objective of the present paper is to explore
the interaction between the values of G
I
and G
II
in mixed-
mode fracture. Reeder [12] has published results that seem
to indicate that G
I
rst increases, and then decreases with
increasing G
II
, and has indicated that a bi-linear rela-
tionship describes the mode I and mode II interaction. Ben-
zeggagh and Kenane [5] also reported a similar interaction
behavior between G
I
and G
II
for glass epoxy. The mecha-
nism that could account for this is not explained, and this
initial increase of G
I
with G
II
seems to the present authors
to be suspect. Since the present data and that of Reeder [12]
are both for AS4/3501-6 carbon/epoxy laminates, it is
interesting to compare the results. This comparison is
shown in Fig. 7, along with additional data from [2933].
0
50
100
150
200
250
300
350
400
450
0 2 4 6 8 10 12 14 16 18
Displacement (mm)
L
o
a
d


(
N
)
ENF
ELS
Fig. 4. The typical loaddisplacement behavior of an ENF beam specimen
and an ELS beam specimen.
Table 3
Mode II energy release rate calculated from experimental data for AS4/
3501-6 using beam theory and nite element analysis
Test specimen Modied beam
theory G
IIc
(J/m
2
)
Finite element
analysis G
IIc
(J/m
2
)
ENF 683 (cov = 12%) 651 (cov = 12%)
ELS 659 (cov = 11%) 642 (cov = 15%)
10
60
110
160
210
260
310
360
410
0
Displacements, mm
L
o
a
d
,

N
Load, N (GI/GII =8/100)
Load, N (GI/GII =27/100)
Load, N (GI/GII =83/100)
1 2 3 4 5
Fig. 5. Loaddisplacement for mixed-mode bending beam with dierent
mode ratios.
0.000
0.020
0.040
0.060
0.080
0.100
0.120
0.140
0.160
0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 0.800
G
II
(kJ/m
2
)
G
I


(
k
J
/
m
2
)
FEA solutions
Analyticalsolutions [22]
Fig. 6. Energy release rates calculated using the analytical and the FE
models.
1494 M.J. Mathews, S.R. Swanson / Composites Science and Technology 67 (2007) 14891498
Fig. 7 shows that the present data compare well with
that of Reeder, with the exception that the G
Ic
values of
Reeder are lower than both the present data and the G
Ic
values from the literature [22,31,33], and thus seem anom-
alous. This can be seen further in Table 5, where the values
of G
Ic
given in the literature are compared with the mean
values of the present work, and that of Reeder. If these
low G
Ic
values of Reeder are omitted, the overall trend of
all of the data is that G
I
decreases monotonically with G
II
in the G
I
vs. G
II
interaction curves for AS4/3501-6 car-
bon/epoxy laminates.
To illustrate the importance of this dierence in results,
OBrien [3] has noted that many practical problems of
delamination in real structures involve relatively low ratios
of G
II
/G
I
. Thus the question of whether G
I
increases or
decreases as G
II
increases from zero is of crucial impor-
tance. The comparisons shown above would seem to indi-
cate that G
I
does decrease monotonically with increasing
G
II
.
6.2. Mixed-mode delamination criteria
As noted previously above, it is necessary to be able to
predict delamination fracture under conditions of com-
bined loading. It is convenient to have this criterion in
the form of an analytic expression that interpolates
between the pure mode I and mode II loadings. As noted
in the introduction, a number of dierent forms have been
developed for this purpose. These criteria have been
applied to various material systems such as brittle thermo-
set resins such as the AS4/3501-6 investigated at present, to
toughened thermoset and thermoplastic resin systems, as
well as both carbon and glass bers.
The failure criteria, in general, are expressed as func-
tions of G
Ic
and G
IIc
, the critical energy release rates for
pure mode I and mode II fractures and G
I
and G
II
, the
energy release rate corresponding to fracture under
mixed-mode loading. The linear criterion [10,12] states that
G
I
and G
II
normalized by G
Ic
and G
IIc
add up to 1 as given
by
G
I
G
Ic
_ _

G
II
G
IIc
_ _
1 9
The power law criterion [12] takes the form of what is
termed a super ellipse. The failure envelope for this power
law criterion is dened by
G
I
G
Ic
_ _
a

G
II
G
IIc
_ _
b
1 10
Other criteria of this type are the exponential hackle crite-
rion [12], the exponential K
I
/K
II
criterion shown by Reeder
[12], and the interaction criterion developed by Hashemi
et al. [11]. In general all of these criteria were found to give
reasonable representations of the present experimental
-0.02
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0 0.2 0.4 0.6 0.8 1
Present data
Reeder [12]
Cairns [31,32]
Bradley et al. [29,30]
Hooper and Subramanian [33]
G
I


(
k
J
/
m
^
2
)
G
II
(kJ/m^2)
Fig. 7. Comparison of present data with that of Refs. [12,2933] for AS4/
3501-6 mixed-mode delamination.
Table 4
Summary of present interlaminar fracture results for AS4/3501-6 carbon/
epoxy, calculated from the experiments with nite element analysis
Test G
I
J/m
2
G
II
(J/m
2
)
DCB
136.2 0
99.0 0
119.6 0
135.4 0
105.2 0
106.7 0
110.6 0
ENF
0 729
0 555
0 584
0 712
0 674
ELS
0 777
0 681
0 510
0 677
0 595
0 614
MMB
78.9 111
72.5 102
73.9 104
77.4 110
90.9 128
66.0 290
44.6 196
49.3 217
58.2 258
62.5 278
49.9 222
25.9 412
28.7 454
28.1 437
28.1 434
25.4 394
M.J. Mathews, S.R. Swanson / Composites Science and Technology 67 (2007) 14891498 1495
data. However, the linear and power law forms of Eqs. (9)
and (10) are representative of this class of representations
and are convenient to work with, and the others above will
not be displayed further.
A somewhat dierent approach to represent the mode
interaction is suggested by Benzeggagh and Kenane [5].
They recommend an approach where the total energy
release rate G = G
I
+ G
II
is expressed as a function of the
ratio G
II
/G, and further suggest that the interaction be
modeled by the expression
G G
Ic
G
IIc
G
Ic

G
II
G
_ _
m
11
This criterion is also recommended by ASTM standards
[25] as the preferred interaction envelope for mixed-mode
delamination.
To completely characterize the failure criteria described
above, the interaction parameters must be determined, as
well as the mode I and mode II energy release rates G
Ic
and G
IIc
. To accomplish this, mean values for G
Ic
and G
IIc
were used, which are given in Table 5. The parameters a
and b of Eq. (10) were evaluated by a least squares tting
to the experimental data, by minimizing the sum of the
squared errors between measured values of G
I
and the
value of G
I
predicted by the failure envelope for each mea-
sured value of G
II
using a grid search, and are listed in
Table 6. The error formula is given by
Error; % 100

GIcurveGI
data
GI
data
_ _
2
N
_

_
_

_
1=2
12
Failure envelopes based on these models are illustrated in
Fig. 8. As can be observed from Fig. 8, both the linear
and power law criteria considered give a reasonable repre-
sentation of the experimental delamination fracture data
for AS4/3501-6 carbon/epoxy.
The criteria with more free parameters obviously have
an advantage when being tted to any particular data set.
For example, the linear interpolation formula between
G
Ic
and G
IIc
has no free parameters, while the others have
one or two free parameters, thus permitting a better t in
the least squares procedure. In view of the scatter in the
data, this advantage may not be very signicant. The crite-
rion of Benzeggagh and Kenane [5] of Eq. (11) is demon-
strated in Fig. 9, where the total energy release rate G is
plotted against G
II
/G. The least squares error used is simi-
lar to that of Eq. (12), and is given by
Table 5
Comparison of present G
Ic
and G
IIc
interlaminar fracture toughness for
AS4/3501-6 carbon/epoxy with literature values
Material G
Ic
(J/m
2
) G
IIc
(J/m
2
) Reference
AS4/3501-6 116 646 Present data
136 740 Cairns [31,32]
111 814 Hooper and Subramanian [33]
69 560 Reeder [12]
81 554 Adams et al. [22]
Table 6
Estimated parameters for dierent mode I and mode II interaction criteria
Delamination failure criterion Free parameters Root mean
square error %
Linear, Eq. (9) None 39.7, Eq. (12)
Power law, Eq. (10) a = 1.227 b = 0.478 17.3, Eq. (12)
Benzeggagh and
Kenane, Eq. (11)
m = 4.78 15.7, Eq. (13)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Linear, Eqn 9
Power law, Eqn 10
Present data
G
I

(
k
J
/
m
2
)
G
II
(kJ/m
2
)
Fig. 8. Comparison of curve-t models with present mixed-mode delam-
ination data for AS4/3501-6.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
-0.2 0 0.2 0.4 0.6 0.8 1 1.2
Present data
Eqn 11
Eqn 10
G

k
J
/
m
2
G
II
/ G
Fig. 9. Comparison of Benzeggagh and Kenane criterion, Eq. (11), power
law criterion, Eq. (10), and present data when total energy release rate is
plotted as a function of G
II
/G mode ratio.
1496 M.J. Mathews, S.R. Swanson / Composites Science and Technology 67 (2007) 14891498
Error; % 100

GcurveGI
data
GII
data

GI
data
GII
data
_ _
2
N
_

_
_

_
1=2
13
The parameter m determined in this way has a value of
m = 4.78, and is also listed in Table 6.
Fig. 9 also includes the power law criteria for compari-
son plotted in the same form, using the parameters of Table
6. Adams et al. [22] show that this is a convenient way to
present and compare the fracture toughness and mode
interaction for a variety of carbon ber and polymer matrix
systems, and demonstrates the smooth increase in G with
increasing G
II
/G ratio. This had led to its widespread cur-
rent usage and inclusion in the ASTM standard [25].
As noted in [22] and can be seen in the present work, the
more brittle thermoset matrix composites show a large sen-
sitivity to the mode ratio, i.e. G
IIc
is much larger than G
Ic
.
For these systems, presenting the toughness interaction in
the form of Fig. 9 tends to mask the interaction between
mode I and mode II for low values of mode II. For exam-
ple, replotting the data of Reeder [12] in the form of Fig. 9
hides the bilinear interaction between G
I
and G
II
. It
should also be noted that while Eq. (11) appears to t
the data in the form of Fig. 9, when replotted as G
I
as a
function of G
II
it given a much poorer t to the data, as
illustrated in Fig. 10. It appears that this is inherent in
Eq. (11), as various values for m all give similar behavior
in that G
I
appears to be underpredicted for low (but not
zero) values of G
II
.
An important question not addressed at present is
whether the mode interaction could be predicted, based
on a model of the fracture process. OBrien [3] gives some
idea of this approach, noting that mode II fracture is a
complicated process, involving small scale tensile fractures
in the direction of maximum principal stress, followed by
failure of the ligaments between these tensile fractures.
A nal point to be commented on is the relatively good
agreement between the various forms of beam equations
and the nite element analysis in obtaining results for
energy release rates in the various specimens. This appears
to be in contrast with results given by Hashemi et al. [11],
who obtained dierent results when considering the global
response from beam equations and when considering the
detailed stress distribution near the delamination. More
detail on the present nite element calculations has been
given previously [28], where it is shown that good agree-
ment is possible if proper techniques are employed.
7. Summary and conclusions
Experimental measurements have been presented for the
delamination fracture toughness of AS4/3501-6 (carbon/
epoxy) composite laminates. A double cantilever beam
conguration was used to determine mode I fracture
toughness while ENF beam and ELS beam congurations
were used to determine mode II fracture toughness. Addi-
tional experiments were conducted to characterize the
behavior of the material under mixed-mode fracture, using
the MMB test. Analysis of the specimens using various
forms of beam equations as well as compliance measure-
ments agreed with each other, and with detailed nite ele-
ment analysis of the specimens, thus giving condence in
the data analysis procedures. The results showed a mono-
tonic decrease of G
I
from the pure mode I value, as G
II
was increased. This is in contrast to previous results in
the literature that showed rst an increase and then a
decrease in G
I
with increasing G
II
. Several analytic repre-
sentations from the literature were used to represent the
mixed-mode delamination fracture toughness results, and
these were found to all represent the data with reasonable
accuracy.
Acknowledgements
Authors thank Prof N. S. Choi of Hanyang university,
Ansan, Korea for his help with the experimental work.
This work was supported in part by the National Science
Foundation through East Asia Pacic Summer Institute
program.
References
[1] Garg AC. Delaminations a damage mode in composite structures.
Eng Fract Mech 1988;29(5):55784.
[2] OBrien TK. Mixed-mode strain-energy release rate eects on edge
delamination of composites. Eects of defects in composite materials.
ASTM STP, vol. 836. American Society for Testing and Materials;
1984. p. 125142.
[3] OBrien TK. Composite interlaminar shear fracture toughness, G
IIc
:
shear measurement or sheer myth? In: Bucinell RB, editor. Composite
materials: fatigue and fracture seventh volume. ASTM STP, vol.
1330. American Society for Testing and Materials; 1998. p. 318.
[4] Reeder JR, Crews Jr JH. Nonlinear analysis and redesign of the
mixed-mode bending delamination test. NASA Technical Memoran-
dum 102777; 1991.
-0.02
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
-0.2 0 0.2 0.4 0.6 0.8
Present data
Eqn 11
Eqn 10
G
I

k
J
/
m
2
G
II
kJ/m
2
Fig. 10. Comparison of Benzeggagh and Kenane criterion, Eq. (11),
power law criterion, Eq. (10), and present data when G
I
is plotted as a
function of G
II
.
M.J. Mathews, S.R. Swanson / Composites Science and Technology 67 (2007) 14891498 1497
[5] Benzeggagh ML, Kenane M. Measurement of mixed-mode delami-
nation fracture toughness of unidirectional glass/epoxy composites
with mixed-mode bending apparatus. Compos Sci Technol
1996;56:43949.
[6] OBrien TK. Characterization of delamination onset and growth in a
composite laminate. In: Reifsnider KL, editor. Damage in composite
materials. ASTM STP, vol. 775. American Society for Testing and
Materials; 1984. p. 14067.
[7] Law GE. A mixed-mode fracture analysis of (25/90
n
)
s
graphite/
epoxy composite laminates. Eects of defects in composite materials.
ASTM STP, vol. 836. American Society for Testing and Materials;
1984. p. 143160.
[8] Lee SM. Failure mechanisms of edge delaminations of composites. J
Compos Mater 1990;24:120012.
[9] Russell AJ, Street KN. Moisture and temperature eects on the
mixed-mode delamination fracture of unidirectional graphite/epoxy.
In: Johnson WS, editor. Delamination and debonding of materials.
ASTM STP, vol. 876. American Society for Testing and Materials;
1985. p. 34970.
[10] Wu EM. Application of fracture mechanics to anisotropic plates. J
Appl Mech 1967;34:96774.
[11] Hashemi S, Kinloch AJ, Williams G. Mixed-mode fracture in ber
polymer composite laminates. In: OBrien TK, editor. Composite
materials: fatigue and fracture. ASTM STP, vol. 1110. American
Society for Testing and Materials; 1991. p. 14368.
[12] Reeder JR. A bilinear failure criterion for mixed-mode delamination.
In: Camponechi Jr ET, editor. Composite materials: testing and
design. ASTM STP, vol. 1206. American Society for Testing and
Materials; 1993. p. 30322.
[13] Spencer B, Barnby JT. The eect of notch and ber angles on
crack propagation in ber reinforced polymers. J Mater Sci
1976;11:838.
[14] Jurf RA, Pipes RB. Interlaminar fracture of composites. J Compos
Mater 1982;16:38694.
[15] Rhee KY. Characterization of delamination behavior of unidirec-
tional graphite/PEEK laminates using cracked lap shear (CLS)
specimens. Compos Struct 1994;29:37982.
[16] Hashemi S, Kinloch AJ, Williams JG. The eects of geometry, rate
and temperature on the mode I, mode II and mixed-mode I/II
interlaminar fracture of carbonber/poly(etherether ketone) com-
posites. J Compos Mater 1990;24:91856.
[17] Ducept F, Davies P, Gamby D. An experimental study to validate
tests used to determine mixed-mode failure criteria of glass/epoxy
composites. Compos: Part A 1997;28A:71929.
[18] Kinloch AJ, Wang Y, Williams JG, Yayla P. The mixed-mode
delamination of Fibre composite materials. Compos Sci Technol
1993;47:22537.
[19] Singh S, Partridge IK. Mixed-mode fracture in an interleaved
carbonber/epoxy composite. Composites Science and Technology
1995;55:31927.
[20] Standard Test Method for Mode I Interlaminar Fracture Toughness
of Unidirectional Fiber Reinforced polymer Matrix Composites,
ASTM Standard D 5528-94a, ASTM Annual Book of Standards,
15.03; 1994. p. 27281.
[21] Russell AJ, Street KN. Factors aecting the interlaminar fracture
energy of graphite/epoxy laminates. In: Hayashi T, Kawata K,
Umekawa S, editors. Progress in science and engineering of compos-
ites. ICCM-IV. Tokyo: ASM International; 1985. p. 27986.
[22] Adams DF, Carlsson LA, Pipes RB. Experimental characterization of
advanced composite materials. third ed. Boca Raton: CRC Press;
2003.
[23] Davies P et al. Measurement of G
Ic
and G
IIc
in carbon/epoxy
composites. Compos Sci Technol 1990;39:193205.
[24] Swanson SR. Introduction to design and analysis with advanced
composite materials. New Jersey: Prentice Hall; 1997.
[25] Standard Test Method for Mixed Mode I Mode II Interlaminar
Fracture Toughness of Unidirectional Fiber Reinforced polymer
Matrix Composites. ASTM Standard D 6671-01, ASTM Annual
Book of Standards, 15.03; 2001. p. 392403.
[26] Irwin GR. Fracture. Handbuch der Physik 1958;6:551.
[27] Rybicki EF, Kanninen MF. A nite element calculation of stress
intensity factors by a modied crack closure integral. Eng Fract Mech
1977;9:9318.
[28] Mathews MJ, Swanson SR. A numerical approach to separate the
modes of fracture in interface crack propagation. J Compos Mater
2005;39:24764.
[29] Hibbs MF, Tse MK, Bradley WL. Interlaminar fracture toughness
and real-time fracture mechanism of some toughened graphite/epoxy
composites. In: Johnston NJ, editor. Toughened composites, ASTM
STP vol. 937. Philadilphia; 1987. p. 11530.
[30] Hibbs MF, Bradley WL. Correlations between micromechanical
failure processes and the delamination toughness of graphite/epoxy
systems. In: Masters JE, Au JJ, editors. Fractography of modern
engineering materials: composites and metals, ASTM STP, vol. 948.
Philadilphia; 1987. p. 6897.
[31] Cairns DS. Static and dynamic mode II strain energy release rates in
toughened thermosetting composite laminates. J Compos Technol
Res 1992;14:3742.
[32] Long BJ, Swanson SR. Ranking of laminates for edge delamination
resistance. Composites 1994;25:1838.
[33] Hooper SJ, Subramanian R. Eects of water and jet fuel absorption
on mode i and mode ii delamination of graphite/epoxy, composite
materials: fatigue and fracture, fourth volume, ASTM STP, vol. 1156.
Philadelphia; 1993. p. 31840.
1498 M.J. Mathews, S.R. Swanson / Composites Science and Technology 67 (2007) 14891498

You might also like