JCOMA 631

Composites: Part A 31 (2000) 335351

www.elsevier.com/locate/compositesa

Stiffness degradation in cross-ply laminates damaged by transverse

cracking and splitting
M. Kashtalyan, C. Soutis*
Department of Aeronautics, Imperial College of Science, Technology and Medicine, Prince Consort Road, London SW7 2BY, UK
Received 15 March 1999; received in revised form 30 July 1999; accepted 1 September 1999

Abstract
In contrast to the few existing theoretical models (Highsmith and Reifsnider, ASTM STP 1986;907:233251; Hashin, Trans ASME J Appl
Mech 1987;54:872879; Daniel and Tsai, Comp Eng 1991;1(6):355362; Tsai and Daniel, Int J Solid Structures 1992;29(24)32513267;
Henaff-Gardin et al., Comp Structures 1996;36:113130; 1996;36:131140), based on the consideration of a repeated laminate element
defined by the intersecting pairs of transverse cracks and splits, the new approach for evaluating the stiffness degradation in [0m/90n]s
laminates due to matrix cracking both in the 908 (transverse cracking) and 08 (splitting) plies employs the Equivalent Constraint Model
(Fan and Zhang, Composites Science and Technology 1993;47:291298). It also uses an improved 2-D shear lag analysis (Zhang et al.,
Composites 1992;23(5):291298; 1992;23(5):299304) for determination of stress field in the cracked or split lamina and In-situ Damage
Effective Functions for description of stiffness degradation. Reduced stiffness properties of the damaged lamina are found to depend
explicitly upon the crack density of that lamina and implicitly upon the crack density of the neighbouring lamina. Theoretical predictions
for carbon and glass fibre reinforced plastic cross-ply laminates with matrix cracking in the 908 ply revealed significant reduction in the
Poissons ratio and shear modulus due to additional damage (splitting) in the 08 ply. q 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Cross-ply composite laminates

1. Introduction
Matrix cracking has long been recognised as the first
damage mode observed in composite laminates under static
and fatigue tensile loading. It does not necessarily result in
the immediate catastrophic failure of the laminate and therefore can be tolerated. However, its presence causes stiffness
reduction and can be detrimental to the strength of the
laminate. It also triggers the development of other harmful
resin-dominated damage modes, such as edge and local
delaminations, which can cause fibre-breakage in the
primary load-bearing plies.
Since the early 1970s transverse cracking in composite
laminates has been the subject of extensive research, both
theoretical and experimental. A number of theories have
appeared in an attempt to predict initiation of transverse
cracking and describe its effect on the stiffness properties
of the laminate, among them theories based on the selfconsistent method [1], variational principles [2,3],
* Corresponding author. Tel.: 144-0171-594-5070; fax: 144-0171-5848120.
E-mail address: c.soutis@ic.ac.uk (C. Soutis).

continuum damage mechanics [4,5], shear lag [611],

approximate elasticity theory solutions [12] and stress transfer mechanics [13,14]. Most of the models developed were
confined, however, to cross-ply laminates under uniaxial
tensile loading. More recently, the focus of investigation
into stiffness reduction due to matrix cracking has shifted
towards transverse cracking in unbalanced um =90n s
laminates under general in-plane loading [1518], matrix
cracking in the off-axis plies [19], multilayer matrix
cracking of angle-ply and quasi-isotropic laminates
[2022], and transverse cracking interacting with edge
and local delaminations [2327].
Shear-lag-based models remain the most commonly used
ones for calculating the reduced stiffness properties of transversally cracked composites. They are being modified and
generalised to enable better description of wider classes of
laminates. Thus, the modified 1-D shear-lag approach [28],
suitable for cross-ply laminates of various stacking
sequences, is based on the assumption that longitudinal
displacement is independent of the 08 ply thickness and
width and is a power function of the thickness co-ordinate
and indeterminate function of the length co-ordinate in the
908 piles. While in most of existing shear-lag models

1359-835X/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved.
PII: S1359-835 X( 99)00 077-9

336

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

Fig. 1. Cross-ply laminate damaged by transverse cracking and splitting.

longitudinal displacement of the 908 piles in a linear or

parabolic function of the thickness co-ordinate, in Ref.
[28] the exponent in power function was taken equal to
1.1, the value thought to be reasonable for many laminates.
The generalised shear-lag approach [29], aimed at better
description of laminates with thick 08 layers, assumes an
indeterminate variation of the longitudinal displacement
across the thickness of 08 layer, which is then chosen
empirically, by best fitting finite element results, and proved
to be an exponential function of cracking density in the 908
ply. The modified shear-lag model [30] assumes existence
of a thin interlaminar adhesive layer between neighbouring
layers, able to transfer not only interlaminar shear stress but
also interlaminar normal stress.
When a cross-ply laminate is subjected to biaxial tensile
loading (Fig. 1), matrix cracking may occur both in the
908 (transverse cracking) and in the 08 piles (splitting).
Transverse cracking and splitting in CFRP cross-ply laminates may also be observed under uniaxial tension [31].
Interestingly, multilayer matrix cracking of cross-ply laminates (i.e. transverse cracking combined with splitting) has
been the subject of a very small number of studies. The
reason for this lies perhaps in the problem itself, which is
mathematically more complex than that with just one
cracked ply, due to complicated interaction between two
damage modes. Highsmith and Reifsnider [32] were apparently the first who examined cross-ply laminates damaged
by matrix cracking and splitting. Their study was concerned
with evaluation of stresses and involved extensive numerical analysis. Hashin [2] treated the problem of stiffness
reduction and stress analysis of orthogonally cracked
cross-ply laminates under uniaxial tension by the variational
method based on the principle of minimum complementary
energy. He obtained a strict lower bound for the Youngs
modulus and an approximate value of Poissons ratio.
Hashins analysis revealed an increase in Poissons ratio
with increasing cracking and splitting density in orthogonally cracked laminates with thin 908 layer. Reduction of the
Youngs modulus proved to occur mainly due to transverse
cracking, with splitting having an insignificant effect.
Experimental data and theoretical predictions of axial stiffness reduction due to cracks in both layers were obtained by
Daniel and Tsai [33], who later also predicted and verified
experimentally the reduction of shear modulus due to

transverse cracking and splitting [34]. A finite difference

iteration method was used to solve a system of governing
equations that involves only in-plane displacements in both
layers. Although the interlaminar-shear-analysis, from
which a system of governing equations were derived
based on equilibrium, continuity and boundary conditions,
did not start with a usual assumption of the classical shear
lag approach that the interlaminar shear stresses are proportional to the displacement different between two layers, it
has shown agreement with this assumption. The initial value
for the iterative procedure, provided by superposition of
solutions for a single set of cracks, was thought to be
more than adequate for calculation of shear modulus reaction, with the difference between initial and final value less
than 1%. Experimental results for graphite/epoxy cross-ply
laminates appear to be in a good agreement with the analytical predictions. Henaff-Gardin et al. [35,36] examined
cross-ply composite laminates, damaged by doubly periodic
matrix cracking (i.e. transverse and longitudinal, or splitting), both under general in-plane and thermal biaxial loading. They assumed that the in-plane displacements in each
lamina vary parabolically through the lamina thickness in
the direction normal to the crack plane and are constant in
the other direction. While evaluating elastic constants of the
damaged laminate, it was further assumed that the elastic
constants of a cracked lamina depend only on the crack
spacing in that lamina, but not on the crack density in the
adjacent piles. Stresses in the direction perpendicular to
the crack plane were found to be almost independent of
the other in-plane co-ordinate. This observation was made
earlier by Highsmith and Reifsnider [32] in their experimental work and used as a basic assumption by Hashin [2].
It appears therefore that at present there are few theoretical models that can predict reduction of all in-plane elastic
properties (E, G, v) of cross-ply laminates due to transverse
cracking and splitting under general in-plane loading. None
of the existing models seems to be simple enough to allow
any feasible generalisation, with a purpose to describe, for
instance, delaminations growing from the crack/split tips.
The objective of the present investigation is to evaluate
efficiently the reduced stiffness properties of [0m/90n]s
cross-ply laminates damaged by transverse cracking and
splitting.

2. Theoretical modelling
In the current model, transverse cracks and splits in a
[0m/90n]s laminate are assumed to be spaced uniformly and
to span the full thickness and width of the 90 and 08 plies.
The assumption of the uniform spacing, crucial for theoretical modelling as it allows to solve the problem via analysis
of a representative element, was shown by many researches
to be justified from an engineering point of view. A
schematic of the cross-ply laminate containing bi-directional cracks is shown in Fig. 1. Spacings between splits

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

337

Fig. 2. Equivalent constraint model (ECM) of a damaged lamina: (a) initial laminate; (b) ECMk.

and transverse crack are denoted, respectively, 2s1 and 2s2.

A global set of Cartesian co-ordinates with the origin in the
centre of the laminate is introduced, with the x1-axis coinciding with the fibre direction in the 908 lamina (ply group),
the x2-axis parallel to the fibre direction in the 08 lamina and
the x3-axis is directed through the laminate thickness. The
laminate is subjected to biaxial tension (s 11 and s 12 ) and
shear loading s 12 :
2.1. Application of the Equivalent Constraint Model
In contrast to the existing analytical models [3236],
based on consideration of a repeated laminate element
defined by the intersecting pairs of transverse cracks and
splits, the present model uses a different approach and
employs the Equivalent Constraint Model (ECM) of the
damaged lamina [37]. It is now increasingly accepted that

Fig. 3. Representative segments of the two equivalent constraint models of

a laminate damaged by transverse cracking and splitting: (a) ECM1, (b)
ECM2.

a crack lamina behaves within a laminate in a different

manner compared to an infinite effective medium containing
many cracks. In Ref. [1], while evaluating stiffness and
compliance changes of a cracked laminate, it was suggested
to replace a cracked lamina with an effective medium
containing many cracks, following an assumption that interaction between the cracked lamina and the neighbouring
layers is limited to the vicinity of the crack tip and, therefore, has minor effect of the lamina stiffness. It was revealed
in Ref. [12] that the laminate-independent approach [1]
resulted in significant overestimation of the changes in
damaged lamina compliances. The stiffness of the damaged
lamina was clearly shown to be strongly influenced by the
laminate in which it is contained. In order to take into
account the effect of the in-situ constraint on the stiffness
of a particular cracked lamina, in Ref. [37] the ECM of the
damaged lamina was introduced. In the ECM (Fig. 2), all the
laminae below and above the damaged lamina under consideration are replaced with homogeneous layers (I and II)
having the equivalent constraining effect. The in-plane stiffness properties of the equivalent constraining layers can be
calculated from the laminated plate theory, provided stresses and strains in them are known. Although theoretically
the ECM does not impose any restrictions onto the laminate
lay-up, it may be less efficient if the equivalently constraining layers are anisotropic. Besides that, in its present form
the model does not reflect hierarchical constraining mechanisms, which may be acting in a multi-directional composite
laminate.
Thus, instead of considering the damaged laminate
configuration shown on Fig. 1, the following two equivalent
constraint models (ECMs) will be analysed. In ECM1 (Fig.
3(a)), the 08 lamina (layer 1) contains damage explicitly,
while the 908 lamina (layer 2), damaged by transverse
cracking, is replaced with the homogeneous layer with
reduced stiffness properties. Likewise, in ECM2 (Fig.
3(b)), the 908 lamina (layer 2) is damaged explicitly,
while the split 08 lamina is replaced with the homogeneous
layer with reduced stiffness. ECM1 and ECM2 represent
two particular cases of the general model (i.e. ECMk, 1 # k #
N; where 2N is the number of piles in the laminate) shown
on Fig. 2(b), namely when the kth layer is either central or
outer ply. All the quantities associated with the 08 lamina

338

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

(layer 1) will be henceforth denoted by a sub- or superscript

(1) and those associated with the 908 lamina (layer 2) with a
sub- or superscript (2).
The purpose of the analysis of the ECMm (i.e. ECM1 if
m 1 and ECM2 if m 2) to determine the reduced stiffness properties of the m th layer damaged by transverse
cracking or splitting. Since the reduced elastic properties
of the equivalently constraining layer k , k m; used in
the analysis of the ECMm , are determined from the analysis
of the ECMk , the problems for ECM1 and ECM2 are interrelated. In the absence of splitting, the analysis of ECM2
coincides with that performed earlier in Ref. [6].
2.2. Stress analysis of the ECMm
Due to periodicity of the damage configuration in the
ECMm , only their representative segments (Fig. 3), containing either two pairs of splits or a single pair of transverse
cracks, need to be considered. As the representative
segments are symmetric with respect to the mid-plane and
their material and geometry are noteworthy uniform in
direction perpendicular to the xm 0x3 plane, the analysis
can be further restricted to one-quarter of the representative
segments.
The three-dimensional stress field in the ECMm is
assumed to comply with the equilibrium equations
2 m;k
s
0;
2xj ij

i 1; 2; 3; k 1; 2; m 1; 2

where sijm;k denotes stress components in the kth layer of

the ECMm . It is assumed that the equivalent constraining
lamina(e) in the ECMm are homogeneous orthotropic, and
the constitutive equations for both layers can be written as
{sm;m} C^ m {em;m }

k; m 1; 2 k m

{sm;k } C^ k {em;k };

where C^ m denotes the stiffness matrix of the explicitly

damaged m th layer, (a circumflex (^) is used for representing the elastic properties of the undamaged material), and
C k ; k m denotes the stiffness matrix of the homogeneous orthotropic material of the equivalent constraining
k th layer. The equilibrium equations, Eq. (1), and constitutive equations, Eq. (2), can be averaged across the layer
thickness and the width of ECMm as indicated below
1 Z Z m;k
f
dx3 dxk ; k m
3
f~m;k
2wm hk hk wm
in order to make a transition to the in-plane microstresses
and microstrains. The equilibrium equations in terms of
microstresses become
m

tj
d m;k
s~ jm 1 21k
0;
hk
dxm
m 1; 2; j 1; 2; k 1; 2

where s~ ijm;k are the in-plane microstress in the kth layer of

the ECMm and tjm are the interface shear stresses at the
(0/90) interface of the EMCm in the jth direction. The
in-plane microstresses are related to the total stresses s ij
applied to the laminate by the following equilibrium
equations:

xs~ ijm;1 1 s~ ijm;2 1 1 xs ij ;

i; j 1; 2; x h1 =h2 5

The constitutive equations in terms of microstresses and

microstrains are
8
9 2
3 8 m ; m 9
m
m
>
Q^ 12
s~ 1m;m >
0 >
Q^ 11
>
> e~ 1 >
>
>
>
>
>
<
<
= 6
=
7>
6 ^ m ^ m
7 m ; m
m; m

;
6
7
~
s
~
0
e
Q
Q
2
22
2
4 12
5>
>
>
>
>
>
>
>
>
>
>
>
m : m ; m ;
: s~ m;m ;
e~ 6
0
0
Q^ 66
6
8
8
9 2
3 m;k 9
m; k >
^ k Q^ k
>
>
6
s
~
0
Q
>
> e~ 1 >
>
>
1
12
>
>
>
<
<
= 6 11
=
7>
6 ^ k ^ k
7 m;k
0 7
Q12 Q22
s~ 2m;k > 6
e~ 2 >;
4
5>
>
>
>
>
>
>
>
>
>
: s~ m;k ;
0
0
Q^ k : e~ m;k ;
66

m; k 1; 2; k m
where the components of the in-plane stiffness matrix are
related to the elastic moduli of the orthotropic material as
Q^ ijm C^ ijm 2

m ^ m
C^ i3
Cj3
;

C^ m
33

Qijk Cijk 2

k k
Ci3
Cj3
k
C33

km
In doing so, it is assumed that s~ 3m;k 0; k; m 1; 2: The inplane constitutive equations can also be written in the inverse
form as
8
9 2
38 m;m 9
m ; m >
^ m S^ m
>
>
~
e
0
S
>
> s~ 1 >
>
>
1
11
12
>
>
>
<
<
= 6
=
7>
6
7
m ; m
m
m
m;m
^
^
6 S12 S22
;
7 s~ 2
~2
e
0
5>
>
> 4
>
>
>
> m ; m >
> m;m >
>
>

:
:
;
;
e~ 6
s~ 6
0
0 S^ 66
8
8
9 2
9
3
m ; k >
m ; k >
k
k
>
>
8
~
e
s
~
S
0
S
>
>
>
>
1
1
11
12
>
>
> 6
>
<
=
=
7<
6 k
7
k
0 7
e~ 2m;k > 6
s~ 2m;k >;
4 S12 S22
5>
>
>
>
>
>
>
k >
: e~ m;k >
: s~ m;k >
;
;
0
0
S
66
6
6

k; m 1; 2; k m
The boundary conditions on the stress-free crack/split
surfaces are

s~ ijm;m 0;

i; j 1; 2

To determine the in-plane microstresses from the equilibrium equations, Eq. (4), the interface shear stresses tjm
have to be expressed in terms of the in-plane displacements
ujm;k ; j 1; 2: This can be done by averaging the out-ofplane constitutive equations across the lamina thickness and
making an assumption about the variation of either the outof-plane shear stresses or the in-plane displacements. Here,

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

Fig. 4. Variation of out-of-plane stresses in the improved 2-D shear lag

analysis.

it is assumed that the out-of-plane shear stresses s~ j3m;k ; j

1; 2 vary linearly with x3, which corresponds to a parabolic
variation of the in-plane displacements. Besides that, it is
assumed that in the 08-lamina linear variation of out-ofplane shear stresses s~ j3m;1 ; j 1; 2; is restricted to the region
of the shear layer. The thickness of the shear layer depends
on the cracked layer thickness and must be roughly proportional to it. For laminates with thick 08-layer this appears to
offer a more reasonable description of the cracked laminate
behaviour. For such laminates it was shown [29], by means
of a finite element (FE) analysis, that the assumption of
parabolic variation of the in-plane displacement across the
thickness of the whole 08-layer provides a very poor approximation to the distribution of the longitudinal stress across
the laminate thickness. This approximation becomes even
poorer as the transverse crack density increases. Thus, the
out-of-plane shear stresses are assumed to vary as follows
(Fig. 4):
m;1
s~ j3

m;2
s~ j3

tjm
hs

h2 1 hs 2 x3 h2 # ux3 u # h2 1 hs

After some mathematical calculations and equation rearrangements (see Appendix A.1), the interface shear stresses are
obtained as

tj m

Kjm u~ jm;1

u~jm;2

11

where the shear lag parameters Kj are functions of ply

properties
Kj

^ 2
3G^ 1
j3 Gj3
^ 2
h2 G^ 1
j3 1 1 1 1 2 h=2hh1 Gj3

(^)), therefore, they are the same for ECM1 and ECM2.
When the thickness of the shear layer is equal to the outer
layer thickness, i.e. when hs h1 ; or h 1; Eq. (12) is
reduced to the expression derived in Ref. [12].
The equilibrium equations, Eq. (4), along with expressions for the interface shear stresses, Eq. (10), the laminate
equilibrium equations, Eq. (6), and constitutive equations,
Eq. (8), provide a full set of equations, which are required
for determining the in-plane microstresses s~ jmm;m j; m 1; 2
in the representative segment of the EMCm . For instance,
s~ 1;1
11 can be found from the following set of eight equations
with respect to eight variables
ds~ 1;1
t1
11
2 1 0
dx1
h1

13a

~11;1 2 u~1;2
t1
1 K1 u
1

13b

x1 s~ 1;1
~ 1;2
~ 11
11 1 x2 s
11 s

13c

x1 s~ 1;1
~ 1;2
~ 22
22 1 x2 s
22 s

13d

8
1;1
>
>
< du~1
dx1
>
>
: e 1;1

9
>
>
=

8
1;2
>
>
< du~1
dx1
>
>
: e 1;2

9
>
>
=

10

tjm

x ux u # h2 ; j 1; 2
h2 3 3

12

h hs =h1 ; hs ms t; j 1; 2
Here, G^ k
j3 ; k 1; 2 are the out-of-plane shear moduli of the
kth layer, hs is the thickness of the shear layer, ms is the
number of plies in the shear layer ms hs =t; and t is the
ply thickness, Fig. 4. The presence of aligned microcracks
does not affect the value of the out-of-plane shear moduli
(this fact is emphasised by marking them with a circumflex

339

>
>
;

>
>
;

2
4

S^1
11
S^1
12

2
4

S2
11
S2
12

9
38
1;1 =
<
~
s
S^ 1
12
11
5
;
^S1 : s~ 1;1 ;

13e; f

9
38
< s~ 1;2
=
S2
12
11
5
S2 : s~ 1;2 ;

13g; h

22

22

22

22

After some rearrangement, this and other similar sets of

equations can be reduced to the single differential equations
m; m
d2 s~ mm
m
m ; m
2 L1m s~ mm
1 V11m s 11 1 V22
s 22 0;
dxm

14a

m 1; 2
d2 s~ 12m;m
m ; m
2 L2m s~ 12
1 V12m s 12 0
dxm

14b

m
m
m
; V22
and V12
are the laminate
where L1m ; L1m ; V11
constants depending on the layer compliances S^ ijm ;
Sijk ; k m; shear lag parameters Kj and the layer thickness ratio x h1 =h2 : In detail, they are presented in
Appendix A.2. Given the boundary conditions, Eq. (9),
at the crack/split surfaces, solutions to Eq. (14) are
0
1
q
m
x
cosh
L

1 B
m C
1
m
m
m ; m
q AV11
s~ mm
m @1 2
s 11 1 V22
s 22

L1
cosh L1 sm

15a

340

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

m;m
s~ 12

1
q
m
cosh
L

x
1 B
m C
2
q AV12m s 12
m @ 1 2

L2
cosh L2 sm

15b

Once the in-plane microstresses, Eq. (15), in the explicitly damaged m th layer of the ECMm are known, the
laminate macrostresses can be found as
1 Zsu
s jmm;m
s~ m;m dxm
16
2sm 2 sm jm
2.3. Reduced stiffness properties of a cracked/split layer
The reduced stiffness properties of the layer m , damaged
by transverse cracking or splitting, can be determined by
applying the laminate plate theory to the ECMm after replacing the explicitly damaged layer with an equivalent homogeneous one. The constitutive equations for the
homogeneous layer equivalent to the explicitly damaged
m th layer are
{s m;m } Qm {e m;m }
where the macrostrains are assumed to be
1 Zsu
e j m;m e jm;k e j
e m;k dxm ;
2sm 2 sm j

17

{W} was found to have only three non-zero components

[1]. The number of non-zero additional compliances was
assumed to be further reduced to two if the cracked material
were loaded only in the plane parallel to the fibres [37]. In
the co-ordinate system chosen as shown in Fig. 1, these nonzero additional compliances due to matrix cracking (in the
1
1
Voigt notations) will be W11
0; W66
0 for the layer 1
2
2
(fibres directed along the x2-axis) and W22
0; W66
0
for the layer 2 (fibres directed along the x1-axis). Inversion
of the in-plane compliance matrix Sm S^ m 1 W m
and subsequent extraction of the stiffness matrix of the
undamaged material Q^ m yields the in-plane stiffness
Qm in the form given by Eq. (19), if the following
notations are introduced
m
L22

m ^ m
Qmm
Wmm
;
m ^ m
1 1 Wmm Qmm

The in-plane reduced stiffness matrix [Q (m )] of the homogeneous layer equivalent to the m th layer of the ECMm is
related to the in-plane stiffness matrix Q^ m of the undamaged layer via the In-situ Damage Effective Functions
m
as
(IDEFs) L22m ; L66
2 m m
3
m
R^ 11 L22 Q^ 12m L22
0
6
7
6
7
m
m m
^ 22
Qm Q^ m 2 6 Q^ 12m L22
7 19a
R
L
0
22
4
5
m m
^
0
0
Q66 L66
^ 1 2
^ 2 2
^ 1 ; R^ 1 Q12 ; R^ 2 Q12 ; R^ 2 Q^ 2

Q
R^ 1
11
11
22
11
22
22
Q^ 1
Q^ 2
11
22
19b
m
; L66m was introduced in Ref.
The concept of the IDEFs L22
[37] on the basis of results acquired by the general theory of
inhomogeneous media [1,38]. It was proved [38], by means
of the self-consistent method, that the tensor of elastic
compliance {S} of a linearly elastic brittle anisotropic
solid containing microcracks can be represented as a sum
^ of
of two tensors: the tensor of elastic compliance tensor {S}
an undamaged solid and the tensor of additional
compliances {W}, dependent on the configuration and
distribution of microcracks. For a unidirectional fibrous
composite regarded as an effective homogenous orthotropic
medium and aligned microcracks regarded as elliptic
cylindrical cavities, the tensor of additional compliances

m ^ m
W66
Q66
;
m ^ m
11W Q
66

66

20

m 1; 2
From Eq. (19) and the constitutive equations for the m th
m
m
layer of the ECMm , Eq. (17), the IDEFs L22
; L66
can be
expressed as

L1
22 1 2

s 1;1
1
;
1;1
1;1

e
1
Q^ 1
Q^ 1
11 1
12 e 2

L2
22 1 2

s 2;2
2
;
Q^ 2 e 2;2 1 Q^ 2 e 2;2

18

k m; j 1; 2; 6

m
L66

12

m
L66
12

22

21

s 6m;m
m m ; m
Q^ 66
e 6

On substituting macrostresses, calculated from Eqs. (15)

and (16), and macrostrains, calculated from Eqs. (8), (15)
and (18), into Eq. (21), the closed from expressions for
IDEFs are obtained. They represent IDEFs as functions of
damage parameters Dmc
m hm =Sm ; associated with transverse cracking/splitting, the layer compliances S^ijm ; Sijk ;
k m; shear lag parameters Kj and the layer thickness
ratio x , i.e.
m
m
^ m k
Lqq
Lqq
Dmc
m ; Sij ; Sij ; Kj ; x

22

In detail, the closed form expressions for the IDEFs for the
m th layer of the ECMm are
" m #
Dmc
l
m
1 2 m tanh 1mc
D
l1
m
m
" m # ;
L22
12
mc
D
l
m
1 1 a1m m tanh 1mc
Dm
l1
23
" m #
Dmc
l2
m
1 2 m tanh
Dmc
l2
m
m
"
#;
L66
12
m 1; 2
mc
D
l2m
m m
1 1 a2 m tanh
Dmc
l2
m

0.02
0.05
0.1
0.2
0.33
0.5
1.0
2.0

Damage
parameter Dmc
2
Dmc
1 0

0.990
0.976
0.953
0.910
0.859
0.813
0.775
0.770

0.992
0.981
0.963
0.928
0.889
0.851
0.801
0.780

0.999
0.993
0.980
0.957
0.928
0.894
0.832
0.795

0.980
0.951
0.907
0.830
0.745
0.661
0.548
0.524

Ref. [2]

Ref. [2]

Refs.
[35,39]

GFRP [0/903]s

GFRP [0/90]s

ECM/2-D
shear lag
approach

Table 1
Youngs modulus reduction ratio for transversally cracked laminates

0.985
0.963
0.929
0.867
0.799
0.727
0.613
0.553

ECM/2-D
shear lag
approach
0.999
0.981
0.952
0.899
0.837
0.770
0.637
0.571

Refs.
[35,39]

0.999
0.997
0.994
0.989
0.982
0.975
0.971
0.970

Ref.
[2]

CFRP [0/90]s

0.999
0.998
0.995
0.991
0.986
0.980
0.974
0.971

ECM/2-D
shear lag
approach

0.999
0.999
0.998
0.995
0.991
0.986
0.978
0.973

Refs.
[35,39]

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

341

0.02
0.05
0.1
0.2
0.33
0.5
1.0
2.0

Damage
parameter
Dmc
2
Dmc
1 0

0.990
0.975
0.951
0.907
0.853
0.804
0.762
0.757

0.992
0.980
0.961
0.925
0.882
0.841
0.787
0.765

0.999
0.993
0.980
0.957
0.928
0.894
0.832
0.795

0.980
0.951
0.906
0.829
0.743
0.658
0.542
0.516

Ref. [2]

Refs.
[35,39]

Ref. [2]

ECM/2-D
shear lag
approach

GFRP [0/903]s

GFRP [0/90]s

Table 2
Youngs modulus reduction ratio for transversally cracked and split laminates

0.985
0.962
0.927
0.863
0.792
0.718
0.602
0.541

ECM/2-D
shear lag
approach
0.999
0.981
0.952
0.897
0.835
0.766
0.639
0.559

Refs.
[35,39]

0.999
0.997
0.994
0.988
0.982
0.981
0.974
0.968

Ref. [2]

CFRP [0/90]s

0.999
0.997
0.995
0.990
0.984
0.979
0.972
0.969

ECM/2-D
shear lag
approach

0.999
0.999
0.997
0.994
0.990
0.986
0.977
0.971

Refs.
[35,39]

342
M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

where the constants l1m ; a1m ; i 1; 2 (Appendix A.3)

depend solely on the layer compliance S^ijm ; Sijk ; k m;
shear lag parameters Kj and the layer thickness ratio x .
The modified compliances Sijk ; k m of the equivalently
constraining k th layer of the ECMm are determined from
the analysis of the ECMk and therefore are functions of the
k
k
; L66
: Thus, the IDEFs for the m th layer depend
IDEFs L22
implicitly on the damage parameters Dmc
k hk =sk
associated with the layer k .
The IDEFs for both layers form a system of simultaneous
nonlinear algebraic equations
1
mc ^ 1 2
mc ^ 2
2
L1
qq Lqq D1 ; Sij ; Sij D2 ; Sij ; Lqq ; x;
2
mc ^ 2 1
mc ^ 1
1
L2
qq Lqq D2 ; Sij ; Sij D1 ; Sij ; Lqq ; x;

24

q 2; 6
This system is solved computationally by a direct iterative
procedure, carried out in such a way that the newly calculated IDEFs Lqqm are used to evaluate the reduced stiffness
of the equivalently constrained k th layer repeatedly until
the difference between two iterative steps meets the
prescribed accuracy. As a result, all four IDEFs Lk
qq ; q
2; 6; k 1; 2 are determined as functions of damage paramc
meters Dmc
1 ; D2 : If interaction between damage modes in
different laminae are neglected, IDEFs associated with the
m th layer will depend only on damaged parameters for that
layer.

3. Verification of the model

Before predicting the reduced elastic properties of crossply laminates damaged by transverse cracking and splitting
using the new ECM/2-D shear lag approach, testing it
against other recent theoretical models and experimental
data seems to be worthwhile. As mentioned in Section 1,
there are few theoretical models, which can describe the
stiffness loss in cross-ply laminates due to matrix cracking
both in the 90 and 08 piles [3236].
Tables 1 and 2 show Youngs modulus reduction in
GFRP and CFRP laminates considered earlier in Ref. [2].
Table 1 contains data for laminates damaged by transverse
cracking without splitting, while Table 2 for laminates
damaged by transverse cracking and splitting. The material
properties used are given in Table 3. Since Hashin evaluates
Youngs modulus of a cracked laminate on the basis of the
principle of minimum complementary energy, his predictions are supposed to provide the rigorous lower bound for
the reduced Youngs modulus value. It may be seen that the
present ECM/2-D shear lag approach delivers results, which
not only comply with this expectation, but also are closer to

343

the lower bound than the results [39] based on the model by
Henaff-Gardin et al. [35]. Predictions for CFRP system are
the closest ones, while for GFRP results for the [0/90]s,
laminate are closer than those for the lay-up with a thicker
908 layer, i.e. for [0/903]s.
However, the situation is quite different for the Poissons
ratio. For transverse cracking without splitting (Fig. 5(a)),
the ECM/2-D shear lag approach predicts much greater
reduction in Poissons ratio than Hashins calculations.
This prediction is similar to that in Ref. [39] based on the
model in Ref. [35], though for small values of damage
parameter the results of model [35] are close to those in
Ref. [2]. For transverse cracking and splitting (Fig.
5(b)), Hashin predicts an increase of Poissons ratio
for a [0/90]s lay-up and asymptotic decrease to some
non-zero value for a [0/903]s lay-up, while, according to
ECM/2-D shear lag approach predictions, the Poissons
ratio should decrease almost to zero. The model [35]
predicts decrease in the Poissons ratio, yet the asymptotic value appears to be dependent upon the lay-up: non-zero
for the [0/90]s laminate and zero for the [0/90]s laminate
[39].
The Poissons ratio reduction due to transverse cracking
predicted by the present model has been also compared with
some other, recently developed, theories [40,41]. The properties of unidirectional material (E-glass/epoxy [2]) used for
comparison were taken from Table 3, and the ply thickness
was 0.203 mm. The theory of Pagano and Schoeppner [41]
employs the variational theorem by Reissner [42] to predict
the stress fields in flat laminates. McCartneys theory [40] is
based on the generalised plane strain model of stress transfer, which has been shown to lead to the stationary values of
the Reissner energy functional. It also uses an assumption
that the direct stresses in the 0 and 908 piles are independent
of the through-thickness co-ordinate [40]. The comparison
of results for [0/90]s GFRP laminate [41,43], which was
considered earlier in Refs. [2,32], reveals an excellent
agreement between all three models (Fig. 6). Together
with Fig. 5(a), it indicates that the source of discrepancy
between the present ECM/2-D shear lag approach and
Hashins model is presumably in the latter rather than in
the former one. Indeed, let us consider Eq. (60) in Ref. [2]
for the case of transverse cracking only. Then function c is
equal to zero and the rest of the equation yields that the
reduction of the Poissons ratio is proportional to the
reduction in the Youngs modulus, which is obviously
incorrect.
As far as reduction of shear modulus is concerned, the
present model can be compared to those of Tsai and Daniel
[34], who especially developed it for the description of the
cracked cross-ply laminates under shear loading, and
Henaff-Gardin et al. [35]. It is worth noting that the model
of Tsai and Daniel [34] and the present ECM/2-D shear lag
approach yield exactly the same analytical expression for
the shear modulus reduction ratio due to transverse cracking, if the thickness of the shear layer in the ECM/2-D shear

344

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

Table 3
Properties of unidirectional materials
Material

Source

EA (GPa)

ET (GPa)

EA (GPa)

ET (GPa)

nA

nT

t (mm)

GFRP (E-glass/epoxy)
CFRP
CFRP (AS4/3501-6)
GFRP (E-glass/epoxy)
CFRP (XAS/914)

Ref. [2]
Ref. [2]
Ref. [34]
Ref. [44]
Ref. [44]

41.7
208.3
145.
40.
145.

13.0
6.5
10.6
10.
9.5

3.40
1.65
6.9
5.
5.6

4.58
2.30
3.7
3.52
3.35

0.300
0.255
0.27
0.3
0.3

0.420
0.413

0.155
0.125

mc
mc
mc
Fig. 5. Poissons ratio as a function of damage parameter Dmc
2 : (a) D1 0 (transverse cracking without splitting); (b) D2 D1 (transverse cracking and
splitting).

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

345

Fig. 6. Poissons ratio reduction ratio as a function of transverse cracking density C2 2s2 21 for transversally cracked [0/903]s GFRP laminate.

lag approach is taken equal to that of the 08 lamina, i.e. if

hs h1 : In notations of the present paper this expression is
"
#21
GA
Dmc
l2;2
2
2
rG ;
1 1 2;2 tanh mc
D2
G^ A
l2

25

rG 1 1 x

Dmc
l1;1
1
tanh 2mc
1;1
D1
l2

"

Dmc
l1;1
1 Dmc
l2;2
1
2
2
2
tanh
1
tanh
mc
mc
2;2
D
x
D
l1;1
l
1
2
2
2
#
21
Dmc
l1;1
l2;2
2
2
2
tanh
tanh
Dmc
Dmc
l2;2
1
2
2

11x

For transverse cracking combined with splitting, Tsai and

Daniel [34] suggested a semi-empirical expression for the
shear modulus reduction ratio based on the superposition
of solutions for a single set of cracks as
"

08 lamina, yields an expression

#
"
Dmc
Dmc
l1;1
l2;2
1
2
p
2
2
rG 1 2 1;1 2;2 tanh mc tanh mc
D1
D2
l2 l2

#21
1 Dmc
l2;2
2
1
tanh 2mc
x l2;2
D2
2
26

In fact, it was obtained from Eq. (25) by adding one

more term to the expression within the brackets. Then
Tsai and Daniel calculated the shear modulus reduction
ratio from the work done by the external shear loading.
The shear stresses along the boundary of the block (i.e.
the laminate element between two transverse and two
longitudinal cracks) were obtained by the finite difference iteration procedure, used to solve the general
system of governing equations of the interlaminarshearstress analysis. The value of the shear modulus
reduction ration obtained by the finite difference iteration appeared to be within 1% of the value given by Eq.
(26). The present ECM/2-D shear lag model, if the
interaction between transverse cracks and splits is
neglected and the shear layer has the thickness of the

Dmc
1
l1;1
2

27

It may be seen from Eqs. (26) and (27) that the two
expressions differ by the underlined terms and rpG # rG :
In absence of splitting Dmc
1 0 they are both reduced
to Eq. (25).
Predictions, based on the semi-empirical expression,
Eq. (26), the ECM/2-D shear lag approach (with the shear
layer having the thickness of one ply and neglecting the
interaction between transverse cracks and splits), Eq. (27),
and Henaff-Gardin et al. [35,39] are presented in Fig. 7 for a
AS4/3506-1 [03/903]s laminate. Elastic properties of the
unidirectional material are given in Table 3. When Dmc
1 0
(splitting without transverse cracking), the results differ due
to the fact that in the ECM/2-D shear lag model the shear
layer (Fig. 4) is assumed to be of one ply thickness. In most
cases, predictions by Tsai and Daniel are within 10% of
those by the ECM/2-D shear lag approach. However, in
some cases the error of the semi-empirical expression [34]
can be as big as 20%. Predictions based on Henaff-Gardin et
al. [35] model appear to be in better agreement with those of
the ECM/2-D shear lag approach. The same slope of the
corresponding curves is particularly noticeable. Limited
experimental data acquired by Tsai and Daniel [34] are in

346

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

mc
Fig. 7. Shear modulus reduction ratio as a function of damage parameter Dmc
2 D1 for transversally cracked and split [03/903]s CFRP laminate.

an acceptable agreement with all predictions. Yet, further

experimental work is needed to validate the analytical
models.

4. Prediction of stiffness degradation

All results on the loss of stiffness in graphite/epoxy and
glass/epoxy [0m/90n]s cross-ply laminates due to transverse
cracks and splitting in this section were obtained taking into
account the interaction between transverse cracks and splits.
Up to 12 iterations were required to solve a system of simultaneous non-linear algebraic equations, Eq. (24), with the
accuracy of 10 29. The number of iterations increased with
increasing crack/splitting density. Two lay-ups were
analysed: [0/90]s and [0/903]s. The properties of unidirectional materials [44] used in the analysis are given in Table
1.
Fig. 8(a) shows reduction in the axial modulus, shear
modulus and Poissons ratio in a CFRP [0/90], laminate
with and without splitting as a function of transverse
crack density. Splitting density is taken as C1 2s1 21
10 cracks/cm. It may be seen that the axial modulus of the
transversally cracked and split laminate is almost the same
as that of the laminate damaged by transverse cracking only.
As one would expect, splitting does not affect its value.
However, it causes further reduction in the shear modulus
and Poissons ratio, by 1518% and 2122%, respectively,
for a given splitting density. The effect of splitting, more
pronounced on the Poissons ratio than on the shear
modulus, slightly increases with an increase in transverse
cracking density. Fig. 8(b) shows analogous predictions
for a CFRP [0/903]s laminate. Splitting density is again

C1 2s1 21 10 cracks/cm. Additional reduction in the

shear modulus and Poissons ratio due to splitting about 10
and 11%, for a given splitting density.
Predictions for GFRP [0/90]s and [0/903]s laminates are
shown in Fig. 9(a) and (b), respectively. Again, there is
almost no further reduction in the Youngs modulus value
due to splitting, although reduction due to transverse cracking in GFRP laminates is greater that in CFRP ones. As far
as shear modulus and Poissons ratio are concerned, reduction in their values due to splitting is approximately the
same, respectively, 1722% and 1823% in [0/90]s and
1015% and 1114% in [0/903]s for a given splitting
density. The effect of splitting is also observed to increase
with an increase in the transverse cracking density.
Fig. 10(a) and (b) illustrate reduction in the axial, transverse and shear moduli of CFRP and GFRP [0/90]s laminates when the extent of damage, described by the damage
parameter Dmc
i hi =si ; is the same in both layers. Such
situation appears when a laminate is subjected to biaxial
loading with the biaxiality ratio close to 1. Results for transmc
versally cracked and split Dmc
2 Dl laminates are shown
in comparison with the data for laminates, which contain
damage in the 908 layer only Dmc
2 0: The maximum
value of damage parameter corresponds to crack density
of 37.8 cracks/cm in CFRP and 32.3 cracks/cm in GFRP
laminates.

5. Conclusions
While stiffness loss due to matrix cracking in the 908 plies
of cross-ply [0m/90n]s laminates has been the subject of
numerous studies in the literature, matrix cracking occurring

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

347

Fig. 8. Elastic moduli reduction ratios as functions of transverse crack density C2 2s2 21 (cm 21) for transversally cracked CFRP laminates with (solid lines)
and without (hatched lines) splitting: (a) [0/90]s; (b) [0/903]s. Splitting density 10 cm 21.

both in the 908 (transverse cracking) and 08 (splitting) plies

has received considerably less attention. Existing theoretical
models [3236], on the one hand, do not describe reduction
of all stiffness components, and on the other hand, appear to
be complicated enough to allow any further extension or
generalisation, with the aim to involve into consideration
other damage modes, e.g. local delaminations that grow
from the crack/split tips. The new approach, developed to
overcome the above-mentioned limitations, is based on the
Equivalent Constraint Model (ECM) of the damaged
lamina. In the ECM, only one damaged layer contains
damage explicitly, while all neighbouring layers, damaged
or undamaged, are replaced with two (one) homogeneous
orthotropic layers with reduced stiffness properties. An

improved 2-D shear lag analysis is then performed to determine the in-plane microstresses in the explicitly damaged
layers of the two ECMs, which are considered instead of the
original one. Closed form expressions are obtained for the
IDEFs, which characterise reduced stiffness properties of
the damaged layers. IDEFs for a given layer are represented
as explicit functions of the damage parameters (relative
crack/split density) associated with that layer and implicit
functions of the damage parameters associated with the
neighbouring plies. Thus, interaction between all damage
modes is taken into account in the new approach.
The new ECM/2-D shear lag based approach is in an
excellent agreement with the models [2,35] as far as axial
stiffness reduction of the transversally cracked and split

348

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

Fig. 9. Elastic moduli reduction ratios as functions of transverse crack density C2 2s2 21 (cm 21) for transversally cracked GFRP laminates with (solid lines)
and without (hatched lines) splitting: (a) [0/90]s; (b) [0/903]s. Splitting density 10 cm 21.

laminate is concerned. For the Poissons ratio, significant

qualitative and quantitative scattering of predictions is
observed, the source of which appears to lie in the Hashins
approach. Comparison with estimations [34,35] of the shear
modulus reduction in transversally cracked and split laminates shows that all three models predict the same trend,
although some quantitative discrepancy (1020%) between
the present approach and that in Ref. [34] is noticeable.
The effects of transverse cracking and splitting on the
axial, transverse and shear moduli as well as Poissons
ratio of typical CFRP and GFRP laminates were examined
on the basis of the new ECM/2-D shear lag approach. The
axial modulus value has appeared to be almost not

affected by splitting. The effect of splitting on the Poissons

ratio is more pronounced than on the shear modulus, and it
slightly increases with an increase in the transverse
cracking density. For [0/90]s laminates, transversally
cracked and split under biaxial loading, the total reduction
in the transverse modulus is slightly larger than in the axial
one.

Acknowledgements
Financial support of this research by the Royal Society,
UK, Engineering and Physical Sciences Research Council

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

349

mc
mc
Fig. 10. Elastic moduli reduction ratios as functions of damage parameter Dmc
2 for transversally cracked [0/90]s laminates with (solid lines, D1 D2 ) and
mc
without (hatched lines, D1 0) splitting: (a) CFRP; (b) GFRP.

(EPSRC/GR/L51348) and Ministry of Defence is gratefully

acknowledged. The authors would like to thank Professor
Glyn Davies, Imperial College of Science, Technology and
Medicine, London, UK, Professor Paul Smith, University of
Surrey, Guilford, UK, and Dr Neil McCartney, National
Physical Laboratory, Teddington, UK, for helpful discussions. Also our special thanks to Dr McCartney and Dr
Catherine Henaff-Gardin, ENSMA, Futuroscope, France,
for providing numerical results of their models for comparison purposes.

Appendix A
A.1. Derivation of shear lag parameters
Assuming that 2u3m;k =2x1 2u3m;k =2x2 0; k 1; 2;
the out-of-plane constitutive equations are

gj3m;k <

2ujm;k
1 m;k
k sj3
;
^
2x3
Gj3

k 1; 2

A1

350

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

where G^ k
j3 are the out-of-plane shear moduli of the kth
layers. As was already noted, these elastic constants are
not undergoing reduction due to matrix cracking. For the
inner layer, substitution of Eq. (10) into Eq. (A1) and
repeated integration with respect to x3 across the layer
thickness yields
u~jm;2

ujm;2 ux3 h2

tjm h2
2
3G^ 2

A2

V1
11

K1
2
1 1 xS2
11 1 b1 S12
h1

V1
22

K1
2
1 1 xS2
12 1 b1 S22
h1

V1
12

K2
1 1 xS2
66
h1

L2
1

K2 1
^2
^2
S 1 a1 S1
12 1 xS22 1 a1 S12
h1 22

L2
2

K1 1
S 1 xS^ 2
66
h1 66

V2
11

K2
1
1 1 xS1
12 1 a1 S11
h1

V2
22

K2
1
1 1 xS1
22 1 a1 S12
h1

V1
12

K1
1 1 xS1
66
h1

j3

For the outer layer, substitution of Eq. (10) into Eq. (A1) and
integrating with respect to x3 across the thickness of the
shear layer (Fig. 4) leads to
ujm;l 2 ujm;l ux3 h2 2

tjm
hs G^ 2
j3

h2 1 hs x3 2 h2

2 x23 2 h22 =2;

h2 # x 3 # h2 1 hs
A3

Integrating Eq. (A3) across the thickness of the shear layer

again yields
ujm;l 2 ujm;l ux3 h2

tjm hs
3Gl
j3

j 1; 2

A4

ujm;l

are the displacements averaged across the thickwhere

ness of the shear layer. In the part of the outer layer h2 1
hs # x3 # h2 1 h7 ; free from the out-of-plane shear, Fig. 4,
m;l
the displacements uj are constant across the thickness and
can be found from Eq. (A3) by putting x3 h2 1 hs as
m;l

uj

2 ujm;l ux3 h2

tjm hs
2Gl
j3

A5

The displacements, averaged across the whole thickness of

the outer (constraining) layer are then
u~jm;l

1
m;l
h u m;l 1 h1 2 hs uj
hl s j

A6

Finally, the continuity of displacements at the interface due

to the perfect bonding between the layers in the considered
region should be taken into account, i.e.
ujm;l ux3 h2 ujm;l ux3 h2

L1
1

K1 ^ 1
2
2
S 1 b1 S^ 1
12 1 xS11 1 b1 S12
h1 11

L1
2

K2 ^ 1
S 1 xS2
66
h1 66

^ 2
S1
12 1 xS12
;
1
S11 1 xS^ 2
11

b1

2
S^ 1
12 1 xS12
;
1
S^ 22 1 xS2
22

h1
h2

K1, K2 are the shear lag parameters, Eq. (12).

A.3. Laminate constants
q
k
lk

h
k L1 ; k 1; 2
1

q
k
lk

h
k L2 ; k 1; 2
2

2
2
^ 1 2
^ 1 2
a1
1 xQ11 S11 1 b1 S12 1 Q12 S12 1 b1 S22

a2
1

1 ^ 2 1
1
^ 2 1
Q S 1 a1 S1
12 1 Q12 S12 1 a1 S11
x 22 22

^ 1 2
a1
2 xQ66 S66 ;

a2
2

1 ^ 2 1
Q S
x 66 66

A7

Combining Eqs. (A2), (A4)(A7) together yields Eq. (11)

for the interface shear stresses tjm ; with the shear lag
parameter Kj given by Eq. (12).
A.2. Laminate constants

a1

References
[1] Dvorak GJ, Laws N, Hejazi M. Analysis of progressive matrix cracking in composite laminates I. Thermoelastic properties of a ply with
cracks. Journal of Composite Materials 1985;19:21634.
[2] Hashin Z. Analysis of orthogonally cracked laminates under tension.
Transations of the ASME: Journal of Applied Mechanics
1987;54:8729.
[3] Hashin Z. Analysis of cracked laminate: a variational approach.
Mechanics of Materials 1985;4:12136.
[4] Talreja R. Transverse cracking and stiffness reduction in composite
laminates. Journal of Composite Materials 1985;19:35575.
[5] Talreja R. Transverse cracking and stiffness reduction in cross-ply
laminates of different matrix toughness. Journal of Composite
Materials 1992;26(11):164463.

M. Kashtalyan, C. Soutis / Composites: Part A 31 (2000) 335351

[6] Zhang J, Fan J, Soutis C. Analysis of multiple matrix cracking in
^um =90n s composites laminates Part 1: in-plane stiffness properties.
Composites 1992;23(5):2918.
[7] Zhang J, Fan J, Soutis C. Analysis of multiple matrix cracking in
^um =90n s composite laminates Part 2: development of transverse
ply cracks. Composites 1992;23(5):299304.
[8] Highsmith AL, Reifsnider KL. Stiffness reduction mechanisms in
composite laminates. ASTM STP 1982;115:10317.
[9] Ogin SL, Smith PA, Beaumont PWR. Matrix cracking and stiffness
reduction during the fatigue of a (0/90)s GFRP laminates. Composites
Science and Technology 1985;22:2331.
[10] Han YM, Hahn HT. Ply cracking and property degradations of
symmetric balanced laminates under general in-plane loading.
Composites Science and Technology 1989;35:37797.
[11] Lee JW, Daniel IM. Progressive cracking of crossply composite laminates. Journal of Composite Materials 1990;24:122543.
[12] Nuismer RJ, Tan SC. Constitutive relations of a cracked composite
lamina. Journal of Composite Materials 1988;22:30621.
[13] McCartney LN. Theory of stress transfer in a 0890808 cross-ply
laminate containing a parallel array of transverse cracks. Journal of
Mechanics and Physics of Solids 1992;40(1):2768.
[14] McCartney LN. The prediction of cracking in biaxially loaded crossply laminates having brittle matrices. Composites 1993;24(2):8492.
[15] McCartney LN. Stress transfer mechanics for ply cracks in general
symmetric laminates. NPL Report CMT(A)50, 1996.
[16] Yu H, Xingguo W, Zhengneng L, Qingzhi H. Property degradation of
anisotropic composite laminates with matrix scracking. Part 1: development of constitutive relations for (u m/90n)s cracked laminates by
stiffness partition. Journal of Reinforced Plastics and Composites
1996;15(11):1491160.
[17] Hua Y, Wang X, Li Z, He Q. Property degradation of anisotropic
laminates with matrix cracking. Part 2: determination of resolved
stiffness and numerical study of stiffness degradation. Journal of
Reinforced Plastics and Composites 1997;16(5):47886.
[18] Zhang J, Herrmann KP. Application of the laminate plate theory to the
analysis of symmetric laminates containing a cracked mid-layer.
Computational Materials Science 1998;13(13):195210.
[19] McCartney LN. Stress transfer mechanics for angle-ply laminates. In:
Proceedings of the Seventh ECCM-7. London, vol. 2, 1996. p. 235
40.
[20] Adolfsson E, Gudmundson P. Matrix crack induced stiffness reduction in 0m =90n = 1 up = 2 uq s M composite laminates. Composite
Engineering 1995;5(1):10723.
[21] Tong J, Guild FJ, Ogin SL, Smith PA. On matrix crack growth in
quasi-isotropic laminatesI. Experimental investigation. Composites Science and Technology 1997;57(11):152735.
[22] Tong J, Guild FJ, Ogin SL, Smith PA. On matrix crack growth in
quasi-isotropic laminatesII. Finite element analysis. Composites
Science and Technology 1997;57(11):152735.
[23] Lu TJ, Chow CL. Constitutive theory of matrix cracking and interply
delamination in orthotropic laminated composites. Journal of Reinforced Plastics and Composites 1992;11(5):494536.
[24] Zhang J, Soutis C, Fan J. Effects of matrix cracking and hygrothermal
stresses on the strain energy release rate for edge delamination in
composite laminates. Composites 1994;25(1):2735.
[25] Zhang J, Soutis C, Fan J. Stain energy release rate associated with

[26]

[27]

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]
[40]

[41]

[42]
[43]
[44]

351

local delamination in cracked composite laminates. Composites

1994;25(9):85162.
Xu LY. Interaction between matrix cracking and edge delamination in
composite laminates. Composites Science and Technology
1994;50(4):46978.
Zhang J, Fan J, Herrmann KP. Delaminations induced by constrained
transverse cracking in symmetric composite laminates. International
Journal of Solids and Structures 1999;36:81346.
Xu LY. Influence of stacking sequence on the transverse matrix cracking in continuous fiber crossply laminates. Journal of Composite
Materials 1995;29(10):133758.
Berthelot JM. Analysis of the transverse cracking of cross-ply laminates: a generalised approach. Journal of Composite Materials
1997;31(18):1780805.
Zhang C, Zhu T. On inter-relationships of elastic moduli and strains in
cross-ply laminated composites. Composites Sciences and Technology 1996;56(2):13546.
Gyekenyesi A, Hemann J, Binienda W. Crack development in carbon
polyimide cross-ply laminates under uni-axial tension. SAMPE Journal 1994;30(3):1728.
Highsmith AL, Reifsnider KL. Internal load distribution effects
during fatigue loading of composite laminates. ASTM STP
1986;907:23351.
Daniel IM, Tsai CL. Analytical/experimental study of cracking in
composite laminates under biaxial loading. Composite Engineering
1991;1(6):35562.
Tsai CL, Daniel IM. Behavior of cracked cross-ply composite laminate under shear loading. International Journal of Solids and Structures 1992;29(24):325167.
Henaff-Gardin C, Lafarie-Frenot MC, Gamby D. Doubly period
matrix cracking in composite laminates Part 1: general in-plane loading. Composite Structures 1996;36:11330.
Henaff-Gardin C, Lafarie-Frenot MC, Gamby D. Doubly period
matrix cracking in composite laminates Part 2: thermal biaxial loading. Composite Structures 1996;36:13140.
Fan J, Zhang J. In-situ damage evolution and micro/macro transition
for laminated composites. Composites Science and Technology
1993;47:10718.
Horii H, Nemat-Nasser S. Overall moduli of solids with microcracks:
load induced anisotropy. Journal of Mechanics and Physics of Solids
1983;31(2):15571.
Henaff-Gardin C. Private communiation, 19981999.
McCartney LN. A recursive method of calculating stress transfer in
multiple-ply cross-ply laminates subject to biaxial loading. NPL
Report DMM(A)150, 1995.
Pagano NJ, Schoeppner GA. Some transverse cracking problems in
cross-ply laminates. CTP AIAA/ASME/ASCE/AHS/ASC Structures,
Structural Dynamics and Materials Conference 1997, vol. 3, 1998. p.
203240.
Reissner E. On a variational theorem in elasticity. Journal of Mathematical Physics 1950;29:905.
Mccartney LN. Private communication, 1998.
Smith PA, Wood JR. Poissons ratio as a damage parameter in the
static tensile loading of simple cross-ply laminates. Composites
Science and Technology 1990;38:8593.