Composites Science and Technology 5S (1995) 187-193

Elsevier Science Limited

Printed in Northern Ireland
ELSEVIER 0266-3538(95)00100-X 0266-3538/95/$09.50

IN-PLANE A N D I N T E R L A M I N A R S H E A R PROPERTIES OF
CARBON/EPOXY LAMINATES

G. Zhou, E. R. Green & C. Morrison

Department of Engineering, University of Leicester, University Road, Leicester LE1 7RH, UK

(Received 7 April 1995; revised version received 13 July 1995; accepted 14 July 1995)

Abstract methods can be found for in-plane shear properties 1"2

Carbon~epoxy laminates of three different lay-ups have and for interlaminar shear (ILS) properties. 3 The
been tested to measure both in-plane shear and Iosipescu shear method 2'4 has recently emerged as a
interlaminar shear (ILS) properties by the Iosipescu versatile and reliable test, which can measure both
shear method. A reasonably pure state of shear strain is shear strength and stiffness. It also has potential for
achieved in the region between the roots of two measuring shear properties in all three material
V-notches in all the laminates. However, the assurance planes. The use of this method to measure in-plane
of proper shear failure modes is extremely difficult for shear properties has been explored extensively, 5-t°
some lay-ups on certain shear planes. It is found in the while efforts to measure ILS properties have rarely
tests of ILS properties that delamination or ILS been reported, 6"11"12 and in particular the ILS
cracking initiates at load levels much lower than the behaviour of the laminates used in the present
ultimate so that uncertainty arises as to which stress investigation is largely unknown, although a prelimi-
should be used as the measure of the material property. nary study to measure their ILS properties has been
It is shown that the ILS stiffnesses deteriorate gradually c a r r i e d o u t . 13
in a non-linear fashion even after the onset of Meanwhile, it is of considerable importance to
delamination. It is also found that the in-plane shear understand fully the ILS behaviour of composite
properties increase with increase in number of laminates whose reinforcement is often two-
reinforcing fibre directions. The ultimate ILS stresses of dimensional and in-plane. This lack of reinforcement
specimens with fibres in the length direction (if ultimate in the thickness direction inevitably results in
loads are used) seem to be independent of material relatively low ILS strengths compared to the in-plane
lay-up while the ILS moduli steadily decrease with tensile and compressive strengths. Consequently,
increase in number of reinforcing fibre directions. these composite laminates are prone to delamination
especially when they are relatively thick and subjected
Keywords: in-plane shear, interlaminar shear, to transverse bending loading. The desire of
carbon/epoxy laminate, Iosipescu shear method, de- examining and analysing delamination to improve
gradation of interlaminar shear stiffness laminate performance requires both ILS strength and
stiffness of given materials to be known. In particular,
it is well established 14-17 that delaminated laminates
can still carry much higher loads before reaching their
1 INTRODUCTION load-bearing capabilities so that the prediction of
post-delamination behaviour requires some physical
insight into how the ILS stiffnesses of laminates
The characterization of the shear behaviour of deteriorate and the effect of such deterioration on
composite laminates has attracted considerable atten- post-delamination structural behaviour. Also, the
tion in recent years because of the highly anisotropic three-dimensional elasticity model for the transient
nature of these materials. Many test methods have response of composite laminates ~s needs both in-plane
been developed and each method attempts to measure shear and ILS moduli for input data. It is the
either shear strength or shear stiffness, or ideally both, collection of these needs which motivates the present
in a single shear plane for a specific type of composite. experimental investigation.
However, all the methods have difficulty, to varying In the present paper, the behaviour of both in-plane
degrees, not only in producing a state of pure shear in shear and ILS has been examined for carbon/epoxy
a specimen but also in ensuring proper failure modes. laminates of three different lay-ups with emphasis
Comprehensive reviews and discussions of test placed on the ILS behaviour. Particular attention is
187
188 G. Zhou, E. R. Green, C. Morrison

paid to the load level at the onset of delamination and 50 mm and a height of 12.7 mm on a surface grinder
the ILS stiffness deterioration beyond the onset of with 80 grit wheel. Individual specimens were sliced
delamination. from the block with a diamond-impregnated wheel
saw, and all the surfaces were ground in each
specimen. Finally, all specimens were held in a vice on
2 MATERIALS A N D SPECIMEN a milling machine and positioned at an angle of 45 ° to
PREPARATION a silicon carbide end mill with a right-angle tip. The
90 ° notches with a depth of 20% of the laminate
The carbon/epoxy laminates were fabricated from
thickness were cut at an end mill rotational speed of
A C G T700/LTM45 prepregs, and cured in an
3500 r.p.m. The notches on the other side were cut by
autoclave following the manufacturer's recommended
the same procedure. The notch roots were observed to
cycle. The ply thickness was 0.15 mm. A total of 32
be circular under a low-powered microscope, with a
plies was employed, giving thicknesses of laminates
radius of about 0.16 mm.
varying from 4.1 to 4.4 mm. The laminates were made
The preparation of the in-plane shear specimens
in three lay-ups, namely unidirectional (UD), crossply
was straightforward, and the techniques of notch
(04/904)2s, and quasi-isotropic ( + 45/90/0)4s. The fibre
cutting and surface machining were the same as
volume fraction of the prepreg was 0.6, but for the
above.
cured laminates which lost a few percent of resin in
The notch distance of each specimen was measured
the curing process, the fibre volume fraction is not yet
accurately with a Shadowmaster optical device. The
known.
region on one side of the specimen between the two
In the preparation of ILS specimens a thickness of
V-notches was polished, and a 2 m m 90 ° biaxial
12.7 mm was required to fit in a test device based on
rosette (TML-FCA-2-11) was bonded at the midpoint
the original Wyoming design 2 (see Fig. 1 for details),
between the roots of two notches with two elements
whereas the laminates were nominally only 4 r a m
oriented in the +45 ° directions for selected specimens
thick. Thus, dummy tabs (woven fabric prepregs) were
to reduce costs. Plastic sleeves were slid onto the
bonded on the top and the bottom of the laminates to
strain gauge wires to prevent them from touching the
make up the thickness for a complete specimen. The
metal test device.
whole block was then machined down to a length of
Some recent results 9'~° have shown that the effect of
out-of-plane torsion can cause error in the determina-
tion of in-plane shear modulus for certain lay-ups
P
unless the average value of back-to-back strain gauge
Loadingfixture readings is used while other studies le conclude that
such effects are not significant. Thus, in the present
study no effort is made to examine this issue.

3 TEST METHOD A N D P R O C E D U R E
x

The loading fixture of the Iosipescu shear test and the

specimen dimensions are shown schematically in Fig.
1. In this arrangement, two counteracting force
couples were applied to a specimen, one on each side

P t
(a)
of the notch (as shown in Fig. l(a) of Ref. 2). A state
of constant shear force was induced in the midplane of
the specimen (as shown in Fig. l(b) of Ref. 2).
Induced moments cancel each other out at the
midsection thereby producing a state of pure shear in
the region between the two notches (as shown in Fig.
450 l(c) of Ref. 2). It is known 9'~°'~2 that the shear stress
Tab ~ Tab distribution in the region between the two notch roots
12.7 4[ Laminate is not uniform so that, ideally, correction factors
Tab Tab should be used to take that into account in calculating
moduli. However, the estimation of correction factors
Straingaugerosette by using a finite element computer program depends
on not only the use of uncorrected material properties
(b) but also the size of the strain gauge rosettes. Such an
Fig. 1. (a) The Iosipescu loading fixture and (b) specimen exercise is undoubtedly useful but has not been
geometry (dimensions in mm). carried out in the present investigation. Therefore, the
 In-plane and interlaminar shear properties of carbon~epoxy laminates 189

average shear stress is defined conventionally by the 4 EXPERIMENTAL RESULTS AND

ratio of an ultimate force, P, as measured by the DISCUSSION
testing machine, to the cross-sectional area between
the two notches:
A minimum of five tests were conducted for each
shear plane and the mean results for shear stresses
P and moduli, with standard deviations, are summarized
~: = - - (1)
wt in Table 1 in which the in-plane shear strength is
calculated from the ultimate load while the ILS
where w is the distance between the two notches and t stress-1 and stress-2 are calculated on the basis of the
is the thickness of the specimen. It will be discussed ultimate load and the load corresponding to the onset
later that in the determination of ILS strengths, of delamination, as will be discussed later. For
delamination or ILS cracking may initiate at much quasi-isotropic laminates from which the mean
lower load levels than the ultimate. If the shear strain ultimate ILS stress, r23, is demonstrated to be identical
is pure, both tensile and compressive strains should be to v13 with both failing in ILS, G23 is not measured (to
equal and opposite in sign, which will be checked later reduce costs) and is assumed to be the same as G13 as
with strain gauge responses. The shear strain is has been proved for the symmetric crossply laminate.
calculated by: Figure 2(a) and (b) shows typical single stress/strain
curves from UD, crossply, and quasi-isotropic
")/xy = IE--45 - - E + 4 5 (2) laminates in both in-plane shear and ILS (the sign of
the compressive strain is changed to positive for the
purpose of comparison). It can be observed that both
where E_45 is tensile while •+45 is compressive, so the
tensile and compressive strain values from the ILS
shear strain is positive. Thus, the shear modulus can
be determined from the slope (secant) of the shear tests are very close for a given load level so that the
presence of a reasonably pure field of shear strain
stress/strain curve up to 0-5% strain level as:
between the two V-notches is confirmed. From the
in-plane shear tests, the pure shear strain is again
Gxy = ! (3) observed for the crossply laminate while an obvious
"Yxy imbalance between tensile and compressive strains
exists for quasi-isotropic and U D laminates, but this
For any given material there are six possible disappears gradually for U D as loading increases. The
orientations within the specimen, corresponding to six in-plane shear behaviour, as shown in Fig. 3, is
shear stress components (as defined in Fig. 3(a)-(f) of dominated by significant non-linearity as expected for
Ref. 13). In the present investigation, only the U D and crossply laminates when shear failure occurs.
specimens on the 1-2, 1-3, and 2-3 shear planes are However, for quasi-isotropic laminates whose be-
tested as the specimens for the 2-1, 3-1, and 3-2 haviour is seemingly linear, specimens actually suffer
shear planes are not recommended for use. 2"3 crushing failure rather than shear owing to the
In each test, the specimen is placed in the fixture, presence of the ±45 ° fibres preventing the develop-
ensuring proper alignment, and is loaded monotoni- ment of shear state as shown in Fig. 4. This indicates
cally to failure at a crosshead speed of 1 mm/min. The that the Iosipescu shear method may not be
data are recorded with a data logging system at the appropriate in testing the in-plane shear properties of
sampling rate of one point per second, and are then a quasi-isotropic lay-up, especially since the shear
transfered to a computer disk for post-processing. strain field has already been seen in Fig. 2(a) to be far

Table 1. Summary of in-plane shear and interlaminar shear test results (with standard deviations) for 32-ply
carbon/epoxy laminates using the losipescu shear method

Material Shear plane 1-2 Shear plane 1-3 Shear plane 2-3
lay-up
Strengtha Modulus Stress-1 Stress-2 Modulus Stress-1 Stress-2 Modulus

UD 84 ± 2 4"6 ± 0"2 71 ± 0.9 41 ± 3 5.2 ± 0.3 15 ± 1h -- 3"0 ± 0"4

(04/904)2s 104 ± 9 4.7 ± 0.14 74 ± 1 29 ± 0.4 4.5 ± 0.3 70 ± 3 41 ± 3 4.3 ± 0.05
( ± 45/90/0)4s 204 ± 7e 17.9 ± 0.3 73 ± 1 29 ± 2 3.9 ± 0.2 73 ± 1 39 ± 3 3.9 + 0.2 d

All strengths are in MPa and moduli in GPa. Stress-1 and stress-2 are calculated on the basis of the ultimate load and
the load corresponding to the onset of delamination, respectively. " Denotes ultimate strength, h Specimens failed in
transverse tension, c Specimens failed in crushing. ~ Modulus is assumed to be the same as that in the 1-3 shear plane.
190 G. Zhou, E. R. Green, C. Morrison

200
175 Compressivestrain
q / / - - in quasi-isotropiclaminate
150 "~ H---Tensile strainin
125 ~ ] / quasi-isotropiclaminate
-~ [ / Tensilestrainin Compressivestrainin

1 0 0 5 c°rspyl 0 7'2 5 1 omp

/ r ! !~! l ! i ! inate

Fig. 4. Typical crushing failure in in-plane shear test

25 q l / / " Ten2s.fie"str.ain specimens of quasi-isotropic carbon/epoxy laminate.
0 in UDlaminate
0 7000 14000 21000 28000 35000
from pure. It is also found that the ultimate failure
Microstrain
load in those specimens which failed in shear is not
(a) always clearly defined so that the near levelling-off, if
not dropping-off, has to be taken as an indication of
90 - the ultimate failure.
80
Crossplylaminate Typical ILS stress/strain curves are shown in Figs
70 5-7 for UD, crossply, and quasi-isotropic laminates.
The UD specimens in the 2-3 shear plane where no
.. 60 fibres are present in the length direction all failed
catastrophically in transverse tension with the crack
40 propagating in the ±45 ° direction from the notch root
F Quasi-isotropic
laminate
as shown in Fig. 8. It is not surprising that both the
~ 30 ultimate stress and modulus, as given in Table 1, are
20 much lower that those of specimens measured in the
10 1-3 shear plane. The validity of the tests in this shear
configuration is thus questionable. Figure 6 shows the
' I ' I ' I ' I ' I ' I expected behaviour with significant non-linearity
6000 12000 18000 24000 30000 36000 measured in the 1-3 shear plane of both UD and
Microstrain quasi-isotropic specimens. They all failed in ILS in the
(b) end as shown in Fig. 9(a) and (b) with delaminations
propagating away from the notch. The difficulty of
Fig. 2. Stress/strain curves of UD, crossply, and quasi- precise identification for the ultimate failure load is
isotropic laminates in (a) in-plane shear tests and (b) again encountered.
interlaminar shear tests in the 1-3 shear plane.
What is most interesting is that unlike the
catastrophic ILS failure when delamination occurs in

14
210
180 I Quasi-isotropielaminate 12

150 ~ 1o

120 8

90 Crossply laminate
o •~ shearplane
60 4

30
0 I ' I ' I ' i ' I 0 ' I ' I b [ I

0 15000 30000 45000 60000 75000 0 2000 4000 6000 8000

In-plane shear mierostxain Interlaminar shear microstrain
Fig. 3. In-plane shear stress/strain curves of UD, crossply, Fig. 5. Interlaminar shear stress/strain curve of a UD
and quasi-isotropic laminates using Iosipescu shear method. laminate.
 In-plane and interlaminar shear properties of carbon/epoxy laminates 191

90

80 -

70
o~ 60 / ~ ^ i n ) h e 1-3 s h e a r p l a n e
- / J ~ Quasi-lsotropic laminate
o~ 50 _ ~ / ~ " in the 1-3 s h e a r plane

"~ 40 \

• 30 Audiblecrackingsound
20-
~ 10 "
0 ' I ' I ' I ' I

0 20000 40000 60000 80000

lnterlaminar microstrain Fig. 8. Typical transverse tensile failure in interlaminar
shear test specimens of UD carbon/epoxy laminate in the
Fig. 6. Interlaminar shear stress/strain curves of UD and 2-3 shear plane.
quasi-isotropic laminates.

and important question as to which stress should be

short-beam shear tests, ~9 delamination or ILS cracking used as the measure of material property. On the
here actually initiates at a load level much less than other hand, in some individual tests as shown in
the ultimate value, often with an audible cracking Figs 6 and 7 for UD, crossply, and quasi-isotropic
sound, and continues until specimens fail structurally. laminates there is no obvious load drop-off, but there
Such delamination could be clearly seen with a
magnifying lens from an in situ specimen during test
soon after the sound. However, no additional abrupt
change other than typical non-linearity can be
observed from the stiffnesses of these curves. This
seems to suggest that the deterioration of ILS stiffness
is a gradual process with the salient feature of
non-linearity. Since it has been recognized 14-17 that the
delaminated composite plates can still carry much
higher loads beyond the onset of delamination, the
present observation has a particularly important
implication to the prediction of post-delamination
behaviour of laminated plates. Moreover, the fact that
a delamination or ILS crack initiates at a load level
much lower than the ultimate prompts an unavoidable (a)

90 -

80 " Crossply laminate in the 2-3 s h e a r p l a n e

g, - \
60

50

• ne

i ° 30

10

' ' I ' I ' I ' I

15000 30000 45000 60000 75000
(b)
Interlaminarmicroslrain
Fig. 9. Typical interlaminar shear failure in ILS test
Fig. 7. Interlaminar shear stress/strain curves of a crossply specimens of (a) UD carbon/epoxy laminate in the 1-3
laminate. shear plane and (b) quasi-isotropic carbon/epoxy laminate.
192 G. Zhou, E. R. Green, C. Morrison

is some local stiffness reduction even though the

cracking sound is fairly audible when delamination
occurs. That could leave a misleading impression, if
the ILS stress/strain curves are used as sole evidence
for identifying ILS failure, that ILS occurred at the
ultimate load, which would correspond unbelievably
to about 6.5% of ILS strain (UD). If the loads
corresponding to the onset of delamination identified
by audible cracking sound and load drop-off, if
present, are used, the ILS stresses as shown in Table 1
are much lower than that calculated on the basis of
the ultimate loads. It seems inevitable that if ILS
strength data obtained using the Iosipescu shear
method are expected to be used for design analysis, an
(a)
independent test method such as the short-beam
shear 19 test should be used to confirm the results.
As can be seen from the 'stress-2' values of Table 1
as well as Fig. 7 for crossply (04/904)2s laminates,
additional complexity arises due to the fact that
different fibre orientations of plies where the V-notch
roots are located actually lead to different ILS
behaviour. For crossply and quasi-isotropic specimens
on the 1-3 shear plane, the notch roots are located in
the plies of either transverse or the ±45 ° fibres so that
the transverse tensile failure always precedes and also
induces ILS as illustrated in Fig. 10. The load drop-off
at only about 27 MPa in Fig. 7 must be due to the
transverse tensile failure or mixed mode between
shear and tensile failure. However, for UD specimens
on the 1-3 shear plane, and crossply and quasi-
(b)
isotropic specimens on the 2-3 shear plane, the notch
roots sit in the plies of longitudinal fibres so that ILS Fig. 11. Typical interlaminar shear failure in ILS test
cracking is initiated at the load level of about 40 MPa. specimens of crossply carbon/epoxy laminates in (a) the 1-3
shear plane and (b) the 2-3 shear plane.
Nevertheless, their ultimate ILS stresses are about the
same. Two failed specimens are shown in Fig. ll(a)
and (b). an abrupt change of ILS stiffness. Nevertheless, this
It becomes clear that the determination not only of aspect merits further investigation.
ILS, but also in-plane shear strengths by the Iosipescu For UD laminates whose mechanical behaviour is
shear method is far less straightforward than expected. often considered to be transversely isotropic, this
Complexity involves various damage mechanisms due simplification for shear stiffnesses (without considering
in part to stress concentrations at the notch roots correction factors) contains an error of about 14% for
when they are located in plies with orientations other the present carbon/epoxy laminate. This could be due
than longitudinal, and due in part to different stress to the notch sensitivity because the notch distance in
levels for the initial delamination and for the ultimate the in-plane shear specimens is three times greater
failure, respectively. It is tempting to use the ultimate than that in the ILS specimens so that the latter may
ILS stresses (stress-I) when the occurrence of the just fail at slightly lower load. As can also be seen
initial delamination does not result in a load drop and from Table 1, the crossply laminate shows that
G I 3 = G23. It has higher in-plane shear strength values
1-3 shear plane I 2-3 shear plane than UD and the inclusion of the +45 ° fibres in the
I
...... D u m m y tab
quasi-isotropic lay-up significantly increases both its
0-degree fibres strength and moduli as expected. The ultimate ILS
' ~' ~ / ~1~.~!i~i~i~.~i~.!~!~.~.!~{~i~i~y~i~.!~.i~.~i~.~.i:.i~i!~.
90-degree
~ fibres stresses of specimens with fibres oriented in the length
direction seem to be independent of material lay-ups
Transverse tensile I ILS crack
failure Symmetric axis while their ILS moduli decrease steadily from UD to
Fig. 10. Illustration of notch root locations and damage quasi-isotropic laminates due certainly to the corres-
initiation in interlaminar shear test specimens of crossply ponding reduction of the number of fibres in the
laminate in the 1-3 and 2-3 shear planes. length direction.
 In-plane and interlaminar shear properties of carbon~epoxy laminates 193

shear test results for a carbon fibre-reinforced epoxy

5 CONCLUSIONS composite. Composites, 16 (1985) 220-4.
6. Adams, D. F. & Walrath, D. E., In-plane and
The 32-ply carbon/epoxy laminates of three different interlaminar Iosipescu shear properties of various
lay-ups have been tested for both in-plane shear and graphite fabric/epoxy laminates. J. Comp. Technol.
ILS properties using the Iosipescu shear method. The Res., 9 (1987) 88-94.
results of both stresses and moduli have shown small 7. Lee, S. & Monro, M., Evaluation of testing techniques
scatter. However, the validity of the in-plane shear for the Iosipescu shear test for advanced composite
materials. J. Comp. Mater., 24 (1990) 419-40.
tests of quasi-isotropic lay-up and the ILS tests of U D 8. Wilson, D. W., Evaluation of the V-notched beam shear
in the 2-3 shear plane are questionable due to test through an interlaboratory study. J. Comp. Technol.
undesired failure modes. In particular, in the Res., 12 (1990) 131-8.
determination of ILS strengths the occurrence of 9. Morton, J., Ho, H. & Tsai, M. Y., An evaluation of the
delamination at a load level much lower than the Iosipescu specimen for composite materials shear
property measurement. J. Comp. Mater., 26 (1992)
ultimate leads to uncertainty on which stress should be
708-50.
used as the measure of material property. In the 10. Pierron, F. & Vautrin, A., Accurate comparative
meantime, it is found that the ILS stiffnesses determination of the in-plane shear modulus of
deteriorate gradually in a non-linear fashion even T300/914 by the Iosipescu and 45° off-axis tests. Comp.
after the onset of delamination. Sci. Technol., 52 (1994) 61-72.
11. Hodgkinson, J. M., Ayache, S. & Matthews, F. L.,
The in-plane shear properties increase with the In-plane and out-of-plane property measurements on
increase of the number of reinforcing fibre directions. thick woven glass/polyester laminates. Proc. ECCM-
The ultimate ILS stresses of specimens with fibres in CTS 1, Amsterdam, 1992.
the length direction (if the ultimate loads are used) 12. Gipple, K. L. & Hoyns, D., Measurement of the
seem to be independent of material lay-up while the out-of-plane response of thick section composite
materials using the V-notched beam specimen. J. Comp.
ILS moduli steadily decrease with the increase of the Mater., 28 (1994) 543-72.
number of reinforcing fibre directions. 13. Zhou, G., Green, E. R. & Morrison, C., Interlaminar
shear properties measurement in carbon/epoxy lamin-
ates using Iosipescu shear method. Report 94-29,
REFERENCES Department of Engineering, University of Leicester,
1994.
1. Munjal, A. K., Test methods for determining design 14. Lee, S.-W. R. & Sun, C. T., A quasi-static penetration
allowables for fibre reinforced composites. Test Methods model for composite laminates. J. Comp. Mater,, 27
for Design Allowables for Fibre Composites, ASTM (1993) 251-71.
STP 1003, 1989, pp. 93-110. 15. Elber, W., The effect of matrix and fiber properties on
2. Walrath, D. E. & Adams, D. F., The Iosipescu shear impact resistance. Tough Composite Materials: Recent
test as applied to composite materials. Experim. Mech. Development, Noyes Publishers, NJ, 1985.
March (1983) 105-10. 16. Lesser, A., & Filippov, A. G., Kinetics of damage
3. Broughton, W. R. & Sims, G. D., An overview of mechanisms in laminated composites. 36th Int. SAMPE
through-thickness test methods for polymer matrix Syrup., 36 (1991) 886-900.
composites. NPL Report DMM (A) 148, National 17. Zhou, G., Static behaviour and damage of thick
Physical Laboratory, Teddington, 1994. composite laminates. Comp. Struct. (submitted).
4. Adams, D. F. & Walrath, D. E., Current status of the 18. Green, E. R., Transient impact response of a fibre
Iosipescu shear test method. J. Comp. Mater., 21 (1987) composite laminate. Acta Mech., 86 (1991) 153-65.
494-507. 19. Zhou, G. & Davies, G. A. O., Experimental
5. Swanson, S. R., Messick, M. & Toombes, G. R., determination of interlaminar shear strength of thick
Comparison of torsion tube and Iosipescu in-plane GFRP. Proc. ECCM-CTS 2, Hamburg, Germany, 1994.