You are on page 1of 8

Composites: Part A 31 (2000) 733740

www.elsevier.com/locate/compositesa

A method for predicting the fracture toughness of CFRP laminates


failing by fibre microbuckling
C. Soutis a,*, P.T. Curtis b
a

Department of Aeronautics, Imperial College of Science, Technology & Medicine, Prince Consort Road, London SW7 2BY, UK
Structural Materials Centre, Mechanical Sciences Sector, Defence Evaluation & Research Agency, Farnborough, Hants GU14 0LX, UK
Received 12 July 1999; received in revised form 13 December 1999; accepted 22 December 1999

Abstract
Cohesive zone models have had notable success in the prediction of damage from notches in engineering materials loaded in tension. They
have also been used to determine the growth of fibre microbuckling from a hole in a composite laminate under compression, (Soutis C, Fleck
NA, Smith PA. Failure prediction technique for compression loaded carbon fibre/epoxy laminate with an open hole. J Compos Mater,
1991;25(11):14761498). The usual strategy is to replace the inelastic deformation associated with plasticity or microbuckling with a linecrack and to assume some form of stress-displacement bridging law across the crack faces. For instance, in the Dugdale analysis of plastic
deformation in metals from the root of a notch (Dugdale DS. Yielding of steel sheets containing slits, J Mech Phys Solids, 1960;8:100104),
it is assumed that the bridging normal traction across the crack faces equals the tensile yield strength of the solid. The material response
elsewhere in the cracked structure is assumed to be linear elastic. In this paper, a new method is presented for the calculation of the fracture
energy Gc associated with fibre microbuckling, a parameter required in the Soutis et al. linear softening cohesive zone model for the
prediction of the open hole compression strength. Theoretical results are found to be in good agreement with experimental data for several
carbon fibreepoxy laminates. Crown copyright 2000 Published by Elsevier Science Ltd. All rights reserved.
Keywords: Carbon fibreepoxy composites; Compression; Microbuckling; Crack bridging model

1. Background
Failure in compression of fibre composite laminates with
an open hole is by the initiation and growth of a microbuckle
from the edge of the hole, Soutis and co-workers [15].
They find that failure is governed by microbuckling in the
0 plies. The geometric inhomogeneity induces fibre rotation under increasing applied load; deformation localises
within a band and a microbuckle is initiated, Fig. 1. The
microbuckle then propagates in a stable manner for 23 mm
and the component fails at a higher load than the initiation
load. This process has been modelled with varying degrees
of sophistication. Early models assumed that failure
occurred when the stress at a characteristic length ahead
of the cut-out reaches a critical value equal to the unnotched
strength of the laminate [6,7]. The characteristic distance is

Published with the permission of the Defence Evaluation and Research


Agency (DERA) on behalf of the controller of HMSO, UK.
* Corresponding author. Tel.: 44-171-594-6070; fax: 44-171-5848120.
E-mail address: c.soutis@ic.ac.uk (C. Soutis).

used as a free parameter to be fixed by best fitting the


experimental data.
Soutis et al. [1,2] compared the microbuckled zone at the
edges of the hole in carbonepoxy and carbonPEEK laminates to an equivalent crack containing cohesive stresses.
This is a crack bridging analysis that predicts the size of the
buckled region as a function of the applied load, with the
local stress supported by the buckled fibres decreasing linearly with the closing displacement of the microbuckle
(linear softening cohesive zone law). The model is able to
predict successfully the effects of hole size and lay-up upon
the compressive strength and has recently been incorporated
into a user friendly computer program [8]. Soutis et al. use
coupon tests to measure both the unnotched strength and the
compressive toughness of the laminate which are required
as the models input. From a design point of view it is
desirable to predict these laminate properties from the
mechanical properties of the fibres and the matrix and
from the lay-up geometry.
In this paper, a brief description of this model is presented
in order to develop a better understanding of the two parameters (unnotched strength, s un and fracture energy, Gc)

1359-835X/00/$ - see front matter Crown copyright 2000 Published by Elsevier Science Ltd. All rights reserved
PII: S1359-835 X( 00)00 003-8

734

C. Soutis, P.T. Curtis / Composites: Part A 31 (2000) 733740

Fig. 1. A schematic showing the geometry of the 0 buckled fibres at the edge of a hole. b boundary orientation, f inclination angle and w kink band
width.

required in the analysis. The unnotched strength and fracture energy is estimated from a micromechanics model for
fibre microbuckling. Predictions of this approach are
compared with the measured values and results for the
notched compressive strength of six multidirectional
T800/924C laminates are discussed.
2. Linear softening cohesive zone model [2]
Consider compressive failure of a finite width, multidirectional composite panel, which contains a central circular hole. It is assumed that microbuckling initiates when the
local compressive stress parallel to the 0 fibres at the hole
edge equals the unnotched strength of the laminate s un, that
is
k t s s un

where kt is the stress concentration factor and s is the


remote axial stress. Damage development (by microbuckling of the 0 plies, delamination and by plastic deformation
in the off-axis plies, Fig. 2(a)) is represented by replacing
the damage zone by an equivalent crack, loaded on its faces
by a crack normal traction T, Fig. 2(b), which decreases

linearly with the crack closing displacement (CCD), 2v.


The crack bridging law is shown schematically in Fig.
2(c). It is assumed that the length of the equivalent crack
` represents the length of the microbuckle. When the
remote load is increased the equivalent crack grows in
length, thus representing microbuckle growth. The evolution of microbuckling is determined by requiring that the
total stress intensity factor at the tip of the equivalent crack
equals zero,
K tot K KT 0

where K is the Mode I stress intensity factor due to the


remote stress s and KT is the stress intensity factor due to
the local bridging traction T across the faces of the equivalent crack. When this condition is satisfied, stresses remain
finite everywhere [9].
The equivalent crack length from the circular hole is
deduced as a function of remote stress s using the following algorithm. For an assumed length of equivalent crack `,
we solve for s and for the crack bridging tractions by
matching the crack displacement profile from the crack
bridging law to the crack profile deduced from the elastic
solution for a cracked body [2]. The cracked body is

Fig. 2. (a) A schematic of the damage zone; (b) softening cohesive zone analogy for compressive damage; and (c) normal traction T decreases linearly with the
crack closing displacement 2v. Gc is the critical value for the energy release rate [4].

C. Soutis, P.T. Curtis / Composites: Part A 31 (2000) 733740

735

The model contains two parameters, which can be


measured independently or predicted analytically: the
unnotched strength s un and the critical CCD vc, which is
related to the area Gc (fracture energy) under the assumed
linear tractioncrack displacement curve.
2.1. Unnotched compression strength
It is now well established that the unnotched strength s c,
of unidirectional carbon fibre-epoxy laminates is governed
by fibre microbuckling which is associated with non-linear
shear of the polymer matrix initiating from regions of preexisting fibre waviness (of magnitude only few degrees).
For a rigid-perfectly-plastic body Budiansky [10] showed
that

Fig. 3. Stable microbuckling growth and failure load for an orthotropic


T800/924C laminate with a 5 mm diameter hole.

subjected to a remote stress s and crack face tractions T.


The crack is discretised into a number n of elements and the
loading T on the crack flanks is represented by piecewise
constant loading packets, each of magnitude Ti. As the crack
advances, the number of elements increases. The linear
stressdisplacement relationship in the crush zone, Fig.
2(c), allows direct calculation of the local tractions Ti
from the local crack surface displacements vi using the
expression,


Ti
3
v i v c Ii
s un
where Ii 1 for i 1; 2; n; and n is the number of
segments into which the cohesive zone is divided and vc is
the critical crack displacement. The normal displacement vi
of an element ofcrack surface is calculated by adding the
displacement vis due to remote stress s and v T due to
local stress acting over the buckled zone,

v i vsi vTi

Eq. (4) is combined with Eqs. (2) and (3) to give an expression for the applied compressive stress as a function of
microbuckling length `, unnotched strength s un, critical
CCD vc, laminate elastic properties, E and geometry (plate
width, W and hole radius, R), i.e.

n
X

b i Ti f `; s un ; vc ; E; R; W

i1

Detailed expressions for the functions b i and Ti are given in


Ref. [2]. At a critical length of equivalent crack `cr ; the
remote stress s attains a maximum value, designated
s cr, and catastrophic failure occurs, Fig. 3.

t y
f0 f

where f 0 is the assumed fibre misalignment angle in the


kink band, f is the additional fibre rotation in the kink band
under a remote stress s , and
"
# 12
!2
s Ty

2
t y ty 1
tan b :
7
ty
Here t y and s Ty are the in-plane shear and transverse yield
strengths of the composite, respectively. The band orientation angle b and fibre rotation f are shown in Fig. 1. The
critical stress s s c is achieved at f 0 in Eq. (6). In the
case of b 0 Eqs. (6) and (7) simplify to Argons result
[11]
ty
sc
8
f0
By using the above kinking theory, the unnotched strength
of the unidirectional laminate can be predicted in terms of
the shear properties of the composite and the initial fibre
misalignment. More recently, Berbinau and Soutis [12]
modelled the 0 fibre as an Euler slender column on a
non-linear foundation and obtained s c as a function of
fibre bending stiffness, fibre volume fraction, initial waviness and shear modulus that varies with shear stress/strain,
G dt=dg:
Once the failure stress of the 0-ply is known, the
compressive strength of any multidirectional laminate
containing 0 layers can be determined on a ply-by-ply
analysis using the laminate plate theory with discrete failure
criteria (e.g. maximum stress failure criterion). For 0-dominated lay-ups the strength could be accurately estimated by
the stiffness ratio method,

s un

N
sc X
n k Exk
u
NE1 k1

where s un is the unnotched laminate strength, s c is the


strength of the 0 lamina, N is the total number of the laminae in the laminate, E11 is the 0 ply stiffness in the fibre

736

C. Soutis, P.T. Curtis / Composites: Part A 31 (2000) 733740

Table 1
Unnotched compressive strength of T800/924C laminates (L1, [(^45/
04)2]s; L2, [(^45/02)3]s; L3, [(0/902/0)3]s; L4, [^45/02/902/02/902/02]s; L5,
[(^45/0/90)3]s; L6, [(^45)2/0/(^45)2/0/^45]s)

strain with the angle u . Average failure strains for each of


the four angles u (30, 45, 60 and 75) examined were scattered between 0.9 and 1%.

Lay-up

% 0-plies

s exp (MPa)

e f (%)

s th1a (MPa)

s th2b (MPa)

2.2. Fracture energy, Gc

L0
L1
L2
L3
L4
L5
L6

100
67
50
50
50
25
17

1485
1010
810
670 c
820
568
428

1.10
1.04
1.1
0.96
1.05
1.07
1.35

1145
912
858
916
619
448

1400
902
705
909
918
567
436

Rice [14] has shown that the work done to advance the
crack by unit area Gc equals the area under the crack traction
versus crack displacement curve,
Zn
10
Gc 2 sv dv s un vc
0

th1, predicted strength, assuming elastic laminate plate theory and max
stress sc 1485 MPa:
b
th2, predicted strength, using the CCSM program [9] and Budiansky
model [10].
c
Premature failure due to out-of-plane microbuckling of the 0 surface
plies.

direction, n is the number of plies of a given orientation u ,


and Exu is the modulus of a ply of orientation u in the
loading direction (x).
The unnotched strength of six different T800/924C multidirectional lay-ups predicted by the maximum stress criterion using the measured value for the unidirectional strength
s c 1485 MPa or the value obtained from Budianskys
model [10], assuming ty 60 MPa; f0 3 and b 25;
are presented in Table 1. Theoretical results are in good
agreement with the test data; the difference is within 5
13%, except for L3 lay-up that failed prematurely due to
out-of-plane fibre microbuckling of the 0 surface plies. The
experimental strength values quoted in Table 1 are based on
the average of five specimens and the coefficient of variation
is 10%. The failure strain e f 1% is almost independent
of lay-up and comparable to the failure strain of the unidirectional laminate. This supports the concept that the critical
event in these laminates is microbuckling of the 0 plies,
which leads to final failure by the nucleation and growth of a
kink band at an angle b 525: Similar observations were
made recently by Berbinau et al. [13] who studied the effect
of off-axis ply orientation on 0 fibre microbuckling developed in T800/924C [(^u /02)2]s laminates. The tests did not
reveal any obvious trend as to the variation of the failure
Table 2
Compressive fracture toughness of T800/924C laminates. ( )experimental value
Lay-up

Exx (GPa)

E (GPa)

p
Kc (MPa m)

L1
L2
L3
L4
L5
L6

115.2 (109)
91.8 (88)
85.6 (78)
91.3 (84)
61.7 (58)
44.7 (41)

73.4
73.3
42.5
58.4
61.7
59.1

50.5
46.5
40.6
40.0
42.5
35.0

a
b

0a

Effective modulus obtained from Eq. (11).


Gc is the experimental fracture energy, Eq. (11).

Gc (kJ/m )
34.7
29.5
38.8
27.3
29.3
20.7

where vc is the critical CCD on the crack tractioncrack


displacement curve, as shown in Fig. 2, which is analogous
to the crack opening displacement in tension. For an orthotropic plate in plane stress, the fracture energy Gc is related
to fracture toughness Kc, by (Paris and Sih [15]),
#1=2
! 1=2 "
!1=2
1
Exx
Exx
nxy
Kc2 11
Gc
Eyy
2Gxy
2Exx Eyy
where E and G are the laminate in-plane extensional and
shear moduli, respectively, and n is Poissons ratio (x
loading axis, y transverse axis). This is similar to the
fracture mechanics relationship for an isotropic elastic
plate in plane stress, Gc K 2 =E: It is assumed that the
toughness Gc represents the total energy per unit projected
area dissipated by fibre microbuckling, matrix plasticity in
the off-axis plies and delamination.
In earlier works [2,4], Gc was obtained from a separate
compressive kink propagation
test, wherein the fracture
p
toughness Kc Y s pa of a laminate containing a sharpened long slit ( 2a) is measured, Table 2. The compressive fracture toughness concept is meaningful, provided the
damage zone at the onset of crack advance is much smaller
than other specimen dimensions. The measured in-plane
fracture toughness is shown for the six T800/924C multidirectional laminates in Fig. 4. The observation that the
measured value of Kc was independent of notch length
supported the concept of compressive fracture toughness.
The alternative method is to estimate the critical CCD vc,
Eq. (10).
The crack overlap displacement d 2v represents the
end-shortening associated with fibre rotation in the microbuckled band and can be estimated by using the straightforward kinematic expression

d 2v w1 cos f

12

where f is the fibre rotation given by f 2b; and b is the


orientation of the microbuckle band as shown in Fig. 5. The
condition f 2b is a statement that fibres rotate until the
volumetric strain in the band vanishes and then lock-up,
see Fleck et al. [16,17]. However, once fibre fracture has
occurred, the kink inclination angle f in the crush zone can
increase rapidly f ! 90 if the load is not removed. The
broken fibre segments (debris) do not lock-up, as shown in
Fig. 6, but are free to move into the bore of the hole, Fig.

C. Soutis, P.T. Curtis / Composites: Part A 31 (2000) 733740

737

Fig. 4. Fracture toughness versus a=w ratio for 50 mm wide T800/924C plates [4].

7(a), or delaminations that have been created between


neighbouring plies in order to permit the buckled fibres to
undergo both compressive and shear deformation; Fig. 7(b)
illustrates the broken fibre segments in the crush zone.
Guynn et al. [18] suggested that the microscopic
processes within the damage zone could be represented by
the schematic shown in Fig. 6(a), with a possible constitutive relationship of Fig. 6(b). Step 1 represents the region far
from the hole and step 3 the region adjacent to the open hole.
Fig. 6(b) assumes that fibre buckling reduces the load bearing capacity of the material in the crush zone, giving load
redistribution to material adjacent to this zone. Once the
crush zone has grown across the width of the specimen,
then the load on the crush zone again increases because
there is no longer adjacent material of greater stiffness to
which the load can be redistributed. The increased load on
the damaged region will lead to steps 4 and 5 in Fig. 6(a) and
increasing stiffness in the crush zone. However, such constitutive relationship would lead to a stress distribution that

Fig. 5. Kinematics of the fibre microbuckle. Lock-up occurs at f 2b [17].

does not allow a monotonic increase in load with increasing


crush zone length and would predict the notched compressive strength incorrectly. In contrast, the crack bridging law
shown in Fig. 2(c) implies that the broken fibres do not lockup and the rubble stiffness is reduced to zero, which appears
to be more appropriate and supported by extensive experimental measurements and observations [15,19].
Following the suggestion that the fibres in the cohesive
zone rotate completely, the critical crack overlap displacement d c from Eq. (12) is found equal to the kink band width
w, which Budiansky [10] related in his microbuckling
analysis explicitly to fibre diameter and fibre volume fraction by
! 1=3
pdf Vf Ef
13
w 2v c
4
2ty
where df is the fibre diameter, Ef is the fibre elastic modulus
and t y is the in-plane shear yield strength of the composite.
Jelf and Fleck [20] examined the results of six experimental
studies that appeared in the composites literature and found
that the kink width w satisfies the following empirical
expression
!0:37
Vf E f
14
w 0:68df
ty
which is in good agreement with the prediction of Budianskys Eq. (13). For the T800/924C system, using Eq. (14) and
material data vf 0:65; Ef 294 GPa; ty 60 MPa a
width of about 14 fibre diameters ( 80 mm) is obtained,
which is representative of observed kink band widths
[12,21]. In Table 3, the fracture energy for six different
T800/924C lay-ups is presented, using the measured or
theoretical unnotched strength (see Table 1) and the linear
softening crack bridging curve with vc 40 mm; the experimental values are also shown for comparison purposes. The
difference between the estimated Gc value, Eq. (10), and the
measured one is 425%, depending on lay-up and the

738

C. Soutis, P.T. Curtis / Composites: Part A 31 (2000) 733740


Table 3
Compressive fracture energy of T800/924C laminates
Lay-up

Gexp
(KJ/m 2)
c

a
Gth0
(KJ/m 2)
c

Gth1
c

L1
L2
L3
L4
L5
L6

34.7
29.5
38.8
27.3
29.3
20.7

40.4
32.4
34.0
32.8
23.0
17.1

45.8
36.5
34.3
36.6
24.8
18.0

a
b
c

(KJ/m 2)

c
Gth2
(KJ/m 2)
c

36.1
28.2
36.4
36.7
22.7
17.5

exp
Gth0
c theoretical value predicted by Eq. (10) using s un and vc 40 mm:
th1
Gth1
theoretical
value
predicted
by
Eq.
(10)
using
s
c
un and vc 40 mm:
th2
Gth2
c theoretical value predicted by Eq. (10) using s un and vc 40 mm.

method of calculating the unnotched strength. It will be


shown later that this discrepancy results to only 212%
error in the calculation of the notched strength. This is
not a psurprising
result, since the notched strength s n varies

with G c :

Fig. 6. (a) A schematic of the microscopic processes within the damage


zone; and (b) local constitutive relationship assuming lock-up [18].

Fig. 7. Scanning electron micrographs showing 0 fibre microbuckling in a


[(^45/02)3]s T800/924C laminate with a 5 mm hole: (a) the 0 fibres buckle
into the bore of the hole at 8085% of the failure load (100); and (b) the
buckle zone grows from the edge of the hole across the specimen width
(625).

3. Notched compressive strength results


Experimental works [15] have shown that the failure
strength of T800/924C notched laminates is approximately
half that of the unnotched material. However, the experimental data for the six lay-ups examined in Ref. [4] lay
above the ideally brittle behaviour (maximum stress criterion), due to the development of sub-critical damage in the
form of fibre microbuckling accompanied by matrix cracking of the off-axis plies, matrix plasticity and delamination
between the plies. This reduces the stress concentration at
the edge of the hole and delays final failure to higher applied
stresses.
Microbuckling of the 0 plies nucleates at the sides of the
hole at 80% of the failure load and propagates like a linecrack into the interior of the specimen. The length of the
buckled zone increases with increasing applied load, propagating stably until it reaches a critical length. Then unstable
growth begins and the microbuckle transverses the specimen completely. The linear softening model of Soutis et al.
[1,2] predicts that the applied stress attains a maximum at a
microbuckle length of 23 mm, depending on lay-up and a
softening behaviour thereafter. It is accurate to within 10%
for 0 dominated laminates, but is less accurate for laminates composed mainly of ^45 plies. For laminate L6 the
damage is diffuse in nature, and a cohesive zone representation of damage becomes less appropriate.
In Table 4, strength predictions, based on estimated and
measured Gc values, for the six T800/924C notched laminates d=w 0:1 studied in Ref. [4] are presented. The
difference is between 2 and 12% giving confidence to the
new approach of calculating the fracture energy. In these
calculations, the critical crack overlap displacement was
assumed constant and equal to 40 mm for all lay-ups. Taking
vc 33 mm (corresponding to a kink band width w 12df

C. Soutis, P.T. Curtis / Composites: Part A 31 (2000) 733740

739

Table 4
Notched compressive strength of T800/924C laminates d=w 0:1
Lay-up

% 0-plies

s nexp (MPa)

s th0a (MPa)

s th1b (MPa)

s th2c (MPa)

L1
L2
L3
L4
L5
L6

67
50
50
50
25
17

505
440
358 d
365
335
309

503.1
438.0
375.3
402.6
362.3
286.6

574.4
489.7
403.2
458.9
355.9
280.9

486.1
410.9
421.7
459.6
335.2
275.6

a
b
c
d

s nth0 predicted by the Soutis et al. model [2] using measured values for s un and Gc.
th1
s nth1 predicted by the Soutis et al. model [2] using as an input s un
and Gth1
c .
th2
s nth2 predicted by the Soutis et al. model [2] using as an input s un
and Gth2
c .
The 0 surface plies buckle out-of-plane leading to a lower strength.

would result in lower Gc values and therefore more conservative notched strength predictions (10% lower). The
stacking sequence and the supporting ply angle u may
have an effect on the geometry of the 0 fibre microbuckling
and therefore vc, but from the results presented in Table 4 for
the T800/924C carbon fibreepoxy system, it appears to be
relatively small.
Recent studies by Fleck and co-workers [17,22] on
single-edge crack unidirectional T800/924C CFRP specimens have shown that the overall kink band width of an
arrested microbuckle increases from about 70 mm at the
microbuckle tip to about 800 mm near the notch root.
During propagation, the microbuckles grows for 10
16 mm across the specimen width and broadens in the
fibre direction by the formation of multiple kink bands.
This confirmed that a growing microbuckle propagates in
a crack-like manner and the authors [22] modelled the
microbuckling in this unidirectional material as a bridged
compressive Mode I crack. The crack was ascribed a tip
toughness Gc of 17 kJ/m 2 and a constant normal bridging
stress s b of the order of 100150 MPa (1/10 of the
unidirectional strength) across its flanks; attempts to predict
microbuckling toughness are described in Refs. [23,24]. The
constant crack bridging stress is associated with the
phenomenon of steady state band broadening. In multidirectional material the bridging stress was associated with two
mechanisms: band-broadening in the 0 layers and delamination crack growth from the ends of microbuckles. It was
found that the evolution of microbuckle length with applied
stress for both unidirectional and multidirectional material
is similar and the response is remarkably independent of
lay-up.
Whilst this large scale crack bridging model with a finite
tip toughness gave accurate predictions for microbuckle
growth from a pre-existing sharp notch, it failed for the
case of a blunt notch such as a hole. The model erroneously
predicts that an infinite remote stress is required for initiation of microbuckling in the absence of an initial flaw. The
authors [22] recognised that the Soutis et al. linear cohesive
zone model [1,2] is more appropriate for the open hole case,
predicting that the applied stress attains a maximum at a
microbuckle length of 23 mm, which is in agreement

with experimental observations [4]; for such microbuckle


lengths the width of the kink band is in the region of 80 mm,
which further justifies the value of 40 mm for the critical
crack overlap displacement vc employed to predict the open
hole compression strengths presented in Table 4.

4. Conclusions
The compressive failure of current unidirectional and
multidirectional carbon fibreepoxy laminates is controlled
by fibre microbuckling. Its initiation depends on material
imperfection, such as resin rich regions, voids and fibre
misalignment (waviness). The fibres break at two points to
create a kink band inclined at an angle b 525 to the
transverse direction. Scanning electron micrographs of
arrested microbuckles reveal that the fibres within the
band rotate by an angle f 30 from the initial fibre direction but f can increase rapidly if the applied load is not
immediately removed. The kink band width w is approximately equal to 6080 mm (1214 fibre diameters).
In multidirectional laminates, failure is always by microbuckling of the 0 plies, and is accompanied by delamination between the off-axis and 0 layers, and by plastic
deformation in the off-axis plies. The scatter in strength is
10% and the failure strain e f 1% is almost independent
of lay-up and comparable to the failure strain of the unidirectional laminate. Existing plastic buckling analyses of
microbuckling are able to account for most of the experimental observations. They correctly predict that the shear
properties and fibre waviness are the most important parameters affecting the compressive strength of the composite.
Although the initial fibre misalignment can be quite arbitrary, for current CFRP systems values of f0 23
produce strength predictions for unidirectional and multidirectional laminates within 1015% of the measured data.
The linear softening cohesive zone model of Soutis et al.
[1,2] successfully predicts the effects of hole size and lay-up
upon the compressive strength and buckle zone size at failure. In the analysis, the inelastic deformation associated
with fibre microbuckling and matrix plasticity is mathematically replaced with a line-crack loaded across its faces by a

740

C. Soutis, P.T. Curtis / Composites: Part A 31 (2000) 733740

bridging normal traction that decreases linearly with the


overlap displacement of the microbuckle. The model takes
as its input the laminate unnotched compressive strength s un
and the critical crack overlap displacement, vc or fracture
energy Gc. In this work, vc is related to the kink band width
w and the fracture energy associated with the crush zone
predicted by the new method is in good agreement with
experimental Gc values obtained in earlier work [4] from a
separate compressive kink band propagation test. The theoretical notched strengths for six different lay-ups vary by
only 212% when the theoretical Gc value is used in the
fracture model instead of the measured one.
The next step under investigation is to examine the effect
of the supporting ply angle u on the magnitude of the critical
CCD vc.

[8]

[9]
[10]
[11]
[12]

[13]

[14]

[15]

Acknowledgements
[16]

This work was carried out as part of Technology Group 4


of the Ministry of Defence (MoD, UK) Corporate Research
programme.
References
[1] Soutis C. Compressive failure of notched carbon fibreepoxy panels,
PhD Thesis, Cambridge University, Engineering Department (UK),
1989.
[2] Soutis C, Fleck NA, Smith PA. Failure prediction technique for
compression loaded carbon fibreepoxy laminate with an open
hole. J Compos Mater 1991;25:147698.
[3] Soutis C, Fleck NA. Static compression failure of carbon fibre T800/
924C composite plate with a single hole. J Compos Mater
1990;24:53658.
[4] Soutis C, Curtis PT, Fleck NA. Compressive failure of notched carbon
fibre composites. Proc R Soc London A 1993;440:24156.
[5] Soutis C. Damage tolerance of open-hole CFRP laminates loaded in
compression. Compos Engng 1994;4(3):31727.
[6] Nuismer RJ, Labor JD. Applications of the average stress failure
criterion. Part 2. Compression. J Compos Mater 1979;13:4969.
[7] Rhodes MD, Mikulas MM, McGowan PE. Effects of orthotropy and

[17]

[18]

[19]

[20]
[21]

[22]

[23]

[24]

width on the compressive strength of graphiteepoxy panels with


holes. AIAA J 1984;22(9):128392.
Xin XJ, Sutcliffe PPF, Fleck NA, Curtis PT. Composites compressive
strength modeller. A users manual, Cambridge University Engineering Department, CUED MAT/TR139, 1995.
Dugdale DS. Yielding of steel sheets containing slits. J Mech Phys
Solids 1960;8:1004.
Budiansky B. Micromechanics. Comput Struct 1983;16(1):312.
Argon AS. Fracture of composites. Treatise of materials science &
technology, vol. 1. New York: Academic Press, 1972.
Berbinau P, Soutis C, Guz IA. Compressive failure of 0 unidirectional carbon fibre-reinforced plastic laminates by fibre microbuckling. J Compos Sci Technol 1999;59(9):14515.
Berbinau P, Soutis C, Goutas P, Curtis PT. Effect of off-axis ply
orientation
on
0
fibre
microbuckling.
Compos
A
1999;30(10):1197207.
Rice JR. In: Liebowitz H, editor. Mathematical analysis in the
mechanics of fracture, vol. 2. New York: Academic Press, 1968
(Chap. 3).
Paris PC, Sih GC. Stress analysis of cracks, Fracture toughness and
applications, STP-381. ASTM, 1969. p. 3083.
Fleck NA, Budiansky B. In: Dvorak GJ, editor. Compressive failure
of fibre composites due to fibre microbuckling, Proc. IUTAM Symp,
Troy, NY, 29 May1 JuneBerlin: Springer, 1990.
Sivashanker S, Fleck NA, Sutcliffe MPF. Microbuckle propagation in
a unidirectional carbon fibreepoxy matrix composite. Acta Mater
1996;44(7):258190.
Guynn EG, Bradley WL, Elber W. A detailed investigation of the
micromechanisms of compressive failure in open hole composite
laminates. J Compos Mater 1989;23:479504.
Smith FC. The effect of constituents properties on the mechanical
performance of fibre-reinforced plastics, PhD Thesis, Centre for
Composite Materials, Imperial College, London, UK, April 1998.
Jelf PM, Fleck NA. Compression failure mechanisms in unidirectional composites. J Mater Sci 1994;29:30805.
Soutis C. Compressive strength of unidirectional composites:
measurement and prediction. In: Hooper SJ, editor. Composite materials: testing & design, vol. 13. ASTM STP 1242, 1997. p. 16876.
Fleck NA, Sivashanker S, Sutcliffe MPF. Compressive failure of
composites due to microbuckle growth. Eur J Mech Solids
1997;16:6582 (Special issue).
Sutcliffe MPF, Fleck NA, Xin XJ. Prediction of compressive toughness for fibre composites. Proc R Soc London Series A
1996;452:244365.
Budiansky B, Fleck NA, Amazigo JC. On compressive kink-band
propagation. J Mech Phys Solids 1998;46(9):163753.

You might also like