Materials and Design 31 (2010) 3825–3834

Contents lists available at ScienceDirect

Materials and Design

journal homepage: www.elsevier.com/locate/matdes

Recent developments on damage modeling and finite element analysis

for composite laminates: A review
P.F. Liu *, J.Y. Zheng **
Institute of Chemical Machinery and Process Equipment, Zhejiang University, Hangzhou 310027, China

a r t i c l e i n f o a b s t r a c t

Article history: Under complex environments such as continuous or cyclic loads, the stiffness degradation for the lami-
Received 19 January 2010 nated composites such as the carbon fiber reinforced polymer matrix composites is an important physical
Accepted 18 March 2010 and mechanical response to the damage and failure evolution. It is essential to simulate the initial and
Available online 21 March 2010
subsequent evolution process of this kind of damage phenomenon accurately in order to explore the
mechanical properties of composite laminates. This paper gives a comprehensive review on the general
Keywords: methodologies on the damage constitutive modeling by continuum damage mechanics (CDM), the vari-
B. Laminates
ous failure criteria, the damage evolution law simulating the stiffness degradation, and the finite element
E. Mechanical
H. Failure analysis
implementation of progressive failure analysis in terms of the mechanical response for the variable-stiff-
ness composite laminates arising from the continuous failure. The damage constitutive modeling is dis-
cussed by describing the evolvement of damage tensors and conjugate forces in the CDM theory. The
failure criteria which interpret the failure modes and their interaction are compared and some advanced
methods such as the cohesive theory which are used to predict the damage evolution properties of com-
posites are also discussed. In addition, the solution algorithm using finite element analysis which imple-
ments progressive failure analysis is summarized and several applicable methods which deal with the
numerical convergence problem due to singular finite element stiffness matrices are also compared in
order to explore the whole failure process and ultimate load-bearing ability of composite laminates.
Finally, the multiscale progressive failure analysis as a popular topic which associates the macroscopic
with microscopic damage and failure mechanisms is discussed and the extended finite element method
as a new finite element technique is expected to accelerate its practical application to the progressive fail-
ure analysis of composite laminates.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction fiber principal orientations into play in order to suit the anticipated
load-bearing distributions [1–3].
Currently, the carbon fiber reinforced polymer matrix compos- The design of the laminated composites associates the physical
ites have been increasingly employed in various fields including and mechanical properties of materials with the geometry shapes
the airplane, fuel cell vehicle, electricity generation and communi- of structure. Before the structure optimization which aims to
cation power systems due to their advantages such as high achieve the perfect matching between the mechanical perfor-
strength/stiffness-to-weight ratio, excellent resistance to fatigue mance and weight [4], an initial exploration of the structural
and corrosion as well as satisfactory durability. In general, the car- load-carrying capacity which is represented by the structural fail-
bon fiber composites are used to manufacture the laminated com- ure and damage behavior is required for reliable and economical
posites by deploying the carbon fibers in several unidirectional design of composite laminates [5–7]. Various failure modes may
layers and all fibers have a common orientation in each layer, be provoked under complex loading and these failure modes may
and these layers are stacked in angled orientations to achieve high continue to evolve with external loads, and the damage evolution
stiffness and strength in different directions. Since the stiffness and behavior of composite laminates depends on these failure modes
strength of an individual layer are much higher in the fiber direc- to some extent. Thus, the knowledge of the damage mechanisms
tion than in the transverse direction, the designable composite lay- of composite laminates plays an important role in instructing the
ers sufficiently bring the mechanical performance in the carbon practical design of composite structures [8–10].
The stiffness degradation phenomenon of fiber-reinforced com-
posite (FRC) laminates under continuous or cyclic loads is an
* Corresponding author. Tel.: +86 571 87953370; fax: +86 571 87953393.
** Corresponding author. Tel.: +86 571 87953370; fax: +86 571 87953393. important response to the damage and failure evolution of com-
E-mail addresses: pfliu1980@yahoo.com (P.F. Liu), jyzh@zju.edu.cn (J.Y. Zheng). posite laminates and the corresponding analysis is called as the

0261-3069/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.matdes.2010.03.031
3826 P.F. Liu, J.Y. Zheng / Materials and Design 31 (2010) 3825–3834

‘‘progressive failure analysis” [11–13]. In order to account for the where 0 6 D < 1. D = 0 represents the perfect materials and D = 1
continuous stiffness degradation and to predict the final failure denotes the completely damaged materials.
strength of composites, the methodologies which implement the The Helmholtz free energy per unit mass w for elastic–plastic
progressive failure analysis is significant. Since Kachanov [14] first materials under isothermal conditions is written as
applied the continuum damage mechanics (CDM) to study the
creep rupture of metals, progressive failure analysis using CDM qw ¼ qwe ðeij  epij ; Dij Þ þ qwp ðjÞ þ qwd ðjÞ ð2Þ
has been proved to be an excellent method to acquire the damage where we, wp and wd are the free energy which represents the elas-
initiation and accumulation information of composite laminates. tic, plastic deformations and damage hardening, respectively. j is
The CDM replaces the mechanical properties of the damaged mate- an internal variable and q is the density.
rials with those of the homogenous materials by associating the The thermodynamic conjugate forces ðY ij ; B; RÞ corresponding to
damage mechanisms with their effects on the mechanical parame- the internal variables ðDij ; j; aÞ are expressed as
ters of materials. A large number of theoretical models for FRC lam-
inates had already been established by defining the damage @w @w @w @w
rij ¼ q ; Y ij ¼ q ; B¼q ; R¼q ð3Þ
tensors and their damage evolution laws based on the CDM theory, @ eij @Dij @j @a
in which the internal state variables are introduced to simulate the
The thermodynamic formulation gives the thermodynamic
damage evolution in FRCs. These variables are related to the
forces and the dissipation inequality equation but no information
mechanical aspects of damage mechanisms and the dissipation en-
about the evolution laws for internal variables. The only restriction
ergy required for the evolution of the damaged states. Thus, differ-
imposed by the continuum thermodynamics on the evolution laws
ent damage mechanisms may have distinct internal variables to
is that the Clausius–Duhem dissipation inequality must be fulfilled,
track the damage evolution.
which takes the following form under isothermal conditions
Currently, various theories and failure criteria have been pro-
posed to predict the damage initiation, the progressive failure c ¼ qw_ þ rij e_ ij P 0 ð4Þ
properties and the ultimate failure strengths of composite lami-
nates. Already, there are continuous efforts in developing failure where c is the power of dissipation due to damage and a dot de-
criteria to predict the initial failure and microcracking of composite notes the rate with respect to time.
laminates such as the maximum stress, Hashin, Hoffman, Yamada- From Eqs. (2)–(4), the dissipation inequality is expressed as
Sun, Tsai-Hill and Tsai-Wu criteria [15–23]. These failure criteria c ¼ rij epij þ Y ij D_ ij  Bj_  Ra_ P 0 ð5Þ
based on the stress or strain components specify the different fail-
ure modes with respect to various failure mechanisms such as the If the dissipation c reaches the maximum, the damage evolution
fiber breakage, fiber buckling, matrix cracking and shear failure. law is given by
With the rapid development of the calculation performance of d
computer, the finite element analysis has become a powerful tool @F @F d @F p @F p
D_ ij ¼ k_ d ; j_ ¼ k_ d ; e_ pij ¼ k_ p ; ap ¼ k_ p ð6Þ
to deal with the complex boundary value problems. Zhang and @Y ij @B @ rij @R
Yang [24] and Orifici et al. [25] gave comprehensive reviews where F d ðY ij ; D; BÞ and F p ðrij ; D; RÞ are the damage potential func-
respectively on the damage constitutive relationships, the failure tions. k_ d P 0 and k_ p P 0 are called the consistency parameters
modes and failure criteria, but did not gave no information on and they are assumed to obey the Kuhn–Tucker consistency
the damage evolution and the solution algorithms using finite ele- requirements
ment analysis. After introducing the CDM theory and the failure
criteria, this paper gives a literature review on the damage consti- k_ d P 0; F d 6 0; F_ d ¼ 0
tutive modeling, the failure criteria and the finite element imple- p
ð7Þ
k_ p P 0; F 6 0; F_ p ¼ 0
mentation in the progressive failure analysis which predicts the
stiffness degradation and failure strengths of composite laminates. In the associative plastic flow criterion, the functions Fd and Fp
Specially, the methodologies that solve the numerical convergence in Eq. (7) are equivalent to the corresponding potential function
problems due to the loss of element stiffness in the finite element Gd and Gp. Only after the activation functions F d ðY ij ; D; BÞ and
analysis are discussed. Besides, recent developments about several F p ðrij ; D; RÞ are determined can the damage evolution information
new finite element techniques such as the cohesive theory and the be acquired. Then, as the consistency parameters kd and kP are de-
extended finite element method are reviewed. rived, the damage tensor and internal variable as well as the con-
sistent constitutive equations are obtained.
Based on the CDM theory, Schapery [29], Murakami and Kamiya
2. Damage constitutive modeling of composite laminates using [30], Hayakawa et al. [31], Tang et al. [32], Brünig [33], Olsson and
CDM Ristinmaa [34], Basu et al. [35] and Maimí et al. [36,37] proposed
the isotropic/anisotropic stiffness degradation models and damage
Over the past five decades, the CDM theory has been widely evolution models respectively by introducing a two-order or four-
used to predict the isotropic/anisotropic stiffness degradation order damage tensor. In these models, the relationships between
and damage evolution of composite laminates by introducing a the damage dissipation potential F, the conjugate force Y and the
phenomenological damage tensor D relevant to those failure damage tensor D were addressed, which were explained by differ-
modes above. In the CDM theory, the loss of stiffness can be phys- ent damage evolution laws. Besides, the micromechanics theory
ically considered the macroscopic representation of a series of dis- which assumed a representative volume element (RVE) was asso-
tributed microcracks and microvoids. According to Lemaitre and ciated with the CDM to describe the evolvement of damage prop-
Chaboche [26,27], Ristinmaa and Ottosen [28], Schapery [29], erties. Specially, Matzenmiller et al. [38], Kwon and Liu [39],
Murakami and Kamiya [30], and Hayakawa et al. [31], a brief re- Schipperen [40], Maa and Cheng [41], Camanho et al. [42], Barbero
view of progressive failure analysis using CDM is given below.For and Vivo [43] established the thermodynamic models to describe
the damaged composite laminates, the Cauchy stress tensor r the progressive failure properties and to interpret the stiffness deg-
can be substituted by the nominal stress tensor r radation of composite laminates. In their models, the various fail-
r ¼ ð1  DÞr ð1Þ ure modes were assumed and the relationships between the
damage tensor, conjugate forces and internal stresses/strains were
 P.F. Liu, J.Y. Zheng / Materials and Design 31 (2010) 3825–3834 3827

Fig. 1. (a) 2D composite laminates and (b) aluminium–carbon fiber/epoxy composite cylindrical laminates.

further formulated. However, these thermodynamic models were in the complex failure modes which may potentially interact in a
all limited to the plane structures, as shown in Fig. 1a. For the alu- unique pattern. The failure of the first layer represents the initia-
minium–carbon fiber/epoxy composite cylindrical laminates tion of the failure and damage evolution of the whole composite
shown in Fig. 1b, Perreux et al. [44] and Ferry et al. [45] derived laminates, but does not represent the ultimate catastrophic dam-
the constitutive relationships of the damaged materials using the age of structures since the residual load-bearing ability of compos-
CDM and fracture mechanics. However, these models cannot ex- ite laminates still remains, which is reflected by continuous layer
plain different failure modes of FRCs because they defined only failure and macroscopic stiffness degradation [52].
the individual thermodynamic conjugate force and damage tensor. Several phenomenological failure criteria on the initial failure of
Further, Perreux et al. [46], Liu and Zheng [47] described the dam- composite laminates with fiber principal orientations had been
age evolution process by defining three independent damage ten- developed in terms of their properties with regard to the compli-
sors to explain three kinds of failure modes: fiber breakage, cated nature of laminates over the past five decades. The popular
matrix cracking and shear failure. failure criteria are the maximum stress, Hashin, Hoffman, Yam-
However, it should be noted that the elastic/damage coupling ada-Sun, Tsai-Hill and Tsai-Wu criteria [15–23]. Among all failure
damage constitutive model may be insufficient in order to accu- criteria, the most famous failure criterion for composite laminates
rately acquire the damage initiation and evolution information of is the polynomial stress tensor criterion proposed by Tsai and Wu
composite laminates. Therefore, the damage/plasticity coupling [15]. For anisotropic composite laminates, the generalized Tsai-Wu
nonlinear models develop to describe the interactive effect of the failure criterion can be written as
plastic deformations on the damage properties of composite mate-
F i ri þ F ij ri rj þ F ijk ri rj rk P 1; i; j; k ¼ 1; 2; . . . ; 6 ð8Þ
rials in terms of the dissipative energy concept according to the
CDM. This is generally realized by introducing the thermodynamic where F is strength coefficient tensor.
conjugate forces into the damage surfaces or plasticity potential Tay et al. [53], Davila et al. [54], Sleight [55], Wu and Stachurski
functions. Chow and Yang [48] outlined the nonlinear constitutive [56], and Cuntze and Freund [57] performed extensive literature
models for the mechanical response due to damage for inelastic surveys on the existing failure criteria and they classified the exist-
composite materials. The damage/plasticity model formulation ing failure criteria for composite laminates into two types: the
was based on the concept of damage surface and on the assump- non-interactive failure criteria and interactive failure criteria. The
tion that the change of material behavior is independent of loading non-interactive failure criteria, sometimes also called independent
path over which the damage state of material develops, but is failure criteria, assume no interactions between the stress or strain
dependent on the current stress state and the energy dissipated. components and compare the individual stress or strain compo-
Lin and Hu [49] proposed a nonlinear elasticity–plasticity/damage nents with the corresponding material allowable strength values.
coupling constitutive model together with a mixed failure criterion The maximum stress criteria belong to this category. The failure
to simulate the behavior of composite laminates under uniaxial surfaces for these criteria are rectangular in stress and strain space,
tension. Boutaous et al. [50] established an elastoplastic damage respectively. In contrast, the interactive failure criteria including
model by applying a homogenization method at the microscopic the Hoffman, Yamada-Sun, Tsai-Hill and Tsai-Wu criteria involve
scale. It is modeled by an anisotropic yield criterion which takes interactions between the stress and strain components. The inter-
into account isotropic and nonlinear kinematic hardening. How- active failure criteria generally use the polynomial tensor expres-
ever, plasticity develops only in the matrix and its flow is blocked sions based on the material strengths to describe a failure
in the direction of the fibers. Barbero [51] proposed a damage/plas- surface and to separate each mode of failure. The maximum stress,
ticity coupling model by combining the CDM and classical thermo- Tsai-Hill, and Hashin criteria are generalized for either tensile or
dynamic theory to predict the inelastic effects as the reduction of compressive stresses and the corresponding (tensile or compres-
stiffness and the increments of damage and unrecoverable defor- sive) strength value must be chosen based on the sign of the ap-
mation. In summary, further work is expected to describe the dam- plied stress. The Yamada-Sun criterion assumed only the
age/plasticity potential function coupled with the plastic flow rule combining effect of fiber principal stress and shear stress. In con-
to explore the effect of the damage evolution on the equivalent trast, the Tsai-Wu criterion is designed for use in all quadrants of
plastic strain rate. the stress plane and 3D problems and may be directly used with-
out modification for different stress signs. The Tsai-Wu criterion
3. Damage initiation and evolution criteria of composite requires a biaxial test to experimentally determine the interaction
laminates term F12, which is the only difference between the Tsai-Wu and
Hoffman criteria and is found to be insignificant for the most part
Essentially, the failure of composite laminates is a complex by Narayanaswami and Adelman [21]. Both the Tsai-Hill and Tsai-
problem and the modes of failure depend on the loading, the Wu criteria allow quadratic stress interactions, but Tsai-Hill is a
geometry size, the physical and mechanical properties of materials purely quadratic criterion without the linear stress term included
as well as the specimen defects. The composite laminates may fail by the Tsai-Wu criterion. Although the Hoffman, Yamada-Sun,
3828 P.F. Liu, J.Y. Zheng / Materials and Design 31 (2010) 3825–3834

Tsai-Hill and Tsai-Wu criteria identify a layer failure, but they can where n is the number of node sets on the crack closure surface ex-
not identify the modes of failure. In this case, Liu and Zheng [47], cept the crack tip node. S is the area across which the crack propa-
Padhi et al. [58] separated three stress components in Eq. (8) to gates. F is the crack closure force at those nodes on two separated
identify the detailed failure modes of the failed composite layers. surfaces before crack propagation and Du is the crack opening dis-
In addition, several new failure theories are developing such as placement for two separated nodes after crack propagation.
the strain invariant failure theory proposed by Tay et al. [59,60] However, the VCCT cannot be competent for the prediction of
and Li et al. [61], the multicontinuum theory proposed by Mayes the failure initiation and subsequent evolution for the cracking
and Hansen [62,63] and the micromechanics-based failure theory composite laminates since the self-similar crack propagation
proposed by Zhu et al. [64] and Ha et al. [65]. However, these mode is assumed. In this case, Dugdale [74] and Barenblatt [75]
new criteria seem to be less widely used than the criteria above originally proposed a cohesive zone conception that described
though they are similar to the Tsai-Wu criterion based on the discrete fracture as a material separation across as surface. Later,
stress or strain components. This may be the case that the inte- Hutchinson and Suo [76], Tvergaard and Hutchinson [77], Allen
grated formula for these criteria judging the failure modes are and Searcy [78], Camanho et al. [79], Xie and Waas [80], Turon
much more complicated than the criteria like the Tsai-Hill et al. [81] developed the cohesive theory which assumes there
criterion. is a process zone in front of the crack tip whose fracture proper-
Currently, there is also an evolving trend to develop the fracture ties which consists of upper and lower surfaces controlled by the
energy-based failure criteria to judge the initial failure and damage cohesive traction-displacement discontinuity relationships, and
evolution of composite laminates. However, the accurate predic- allows non-self-similar crack propagation. Sun and Jin [82] stud-
tion of fracture properties which is based on the stress and strain ied the fracture properties of composites deeply by linking the
calculations requires further technique. For complex 3D laminated cohesive theory to bridging toughening model. Bao and Suo [83]
structure, the conventional methods such as the J-integral pro- gave a comprehensive review on the crack bridging mechanisms
posed by Rice [66] in linear-elastic fracture mechanics (LEFM) can- by proposing the bridging constitutive relationship concepts. Fur-
not be used to calculate the energy release rate equivalently again ther, Clegg [84] and Kruzic [85] summarized the crack bridging
since it is limited to only the case for plane structure. Besides, the toughening mechanisms which were considered to play a domi-
calculation of J-integral makes high request for the mesh quality at nating role in affecting toughening mechanisms under static or
the crack tip using finite element method. In contrast, the virtual fatigue load for FRCs. Fig. 3 shows the relationship between the
crack closure technique (VCCT) initially proposed by Rybicki and fiber bridging and cohesive zone. It can be easily recognized that
Kanninen [67,68] based on the Irwin’s crack tip energy analysis the cohesive theory is essentially consistent with the bridging
[69] for the linear-elastic materials can calculate the fracture toughening mechanisms, i.e., the bridging constitutive relation-
parameters at the crack tip for the mixed fracture modes more ship in toughening FRCs. Cox and Yang [86] performed a compre-
effectively than the J-integral due mainly to two points: (1) the hensive review on the advantages of the cohesive theory over the
VCCT can be applicable to 3D structure, (2) the mesh requirement VCCT and the conventional LEFM. They pointed out the critical
using the VCCT is not high [70]. Xie and Biggers [71], Leski [72] and conceptual limitation of LEFM is representing all the material
Orifici et al. [73] pointed out that the finite element calculations nonlinearity during crack extension as a point process with a con-
using the VCCT are simple since the sole assumption that the en- stant fracture toughness resisting crack propagation which yet
ergy required for the crack propagation length Da is equal to that deviates from the realistic nonlinear finite-size fracture zone. In
for closuring two separate crack surfaces with the crack length contrast, the cohesive damage modeling describes the develop-
Da can be easily realized. Fig. 2 shows the crack growth from point ment of the nonlinear fracture zone explicitly. While it has been
i to point j for layered materials and the energy release rate G using successfully used to predict the delamination failure mechanisms
VCCT is calculated as of dissimilar material interfaces, the cohesive theory extends rap-
idly to predict the failure initiation and damage evolution in the
1 X n
 
G¼ F ix Duix þ F iy Duiy þ F iz Duiz ð9Þ progressive failure analysis of composite laminates. The overall
2S k¼1 technique in the cohesive zone model for FRCs is that employing
the fracture energy based, i.e., the critical energy release rate in
LEFM based damage evolution criteria. Shin and Wang [87] pro-
posed an improved cohesive zone model to predict the damage
evolution properties and residual notched strength of anisotropic
composite laminates with different orthotropic lay-ups. Lapczyk
and Hurtado [88] proposed a progressive damage model which
used the Hashin’s failure criterion as the damage initiation crite-
rion and cohesive fracture energy-based damage evolution crite-
ria for the fiber, matrix tensile and compressive failure modes
respectively in the aluminium–glass fiber/epoxy composite lami-
nates. Rami [89] proposed a cohesive micromechanical model
based on a 3D representative unit cell model with incremental
damage formulations to perform the progressive failure analysis
of laminated composite materials and structures. These cohesive
fracture energy-based damage criteria can further gain insight
into the microscopic damage evolution mechanisms of compos-
ites though they are to be validated by experiments for different
cases. However, the numerical convergence problems may arise
from the cohesive modeling due to an assumed softening proper-
ties which required to be solved appropriately as it is imple-
mented using finite element analysis. This poses a large
challenge to the practical application of cohesive theory in the
Fig. 2. Schematic representation of virtual crack closure technique (VCCT). damage evolution modeling of composite laminates.
 P.F. Liu, J.Y. Zheng / Materials and Design 31 (2010) 3825–3834 3829

Fig. 3. (a) Fiber bridging/cohesive zone model and (b) a linear cohesive zone model.

4. Finite element implementation of progressive failure analysis metrically nonlinear static problems sometimes involve collapse
behavior, where the load–displacement response often shows neg-
In recent years, the exploration for the methodologies of pro- ative stiffnesses and negative eigenvalues. In this case, the conven-
gressive failure analysis of composite laminates using finite ele- tional Newton–Raphson algorithm can not further track the load
ment analysis has attracted much attention. In order to interpret path and becomes incapable because there is no static equilibrium
the damage constitutive relationship and stiffness degradation path that connects the structure states at the start and end of a
properties of composite laminates under continuous or fatigue load increment since the integrated structural stiffness matrices
loading, developing a set of effective finite element methods is cru- at the final failure point are singular. Fig. 5 shows two typical
cial. How to implement the initiation and damage evolution mod- unstability cases including snap-through and snap-back effects.
els for different failure modes using finite element analysis is an At this time, the structure must release strain energy to remain
important topic. Ochoa and Reddy [90] gave an excellent overview in equilibrium. How to predict the failure strength of composite
of the basic steps for performing the progressive failure analysis of laminates accurately is a challenging task which requires some
composite laminates. In the following, the finite element imple- special manipulation mathematically.
mentation for progressive failure analysis of composite laminates In order to solve this problem, Tay et al. [53,59,60] proposed
is particularly addressed in detail. A typical finite element equation the element failure method for the progressive failure analysis
which accounts for the geometrically nonlinear behavior is given of composite laminates so that more rapid convergence and
by computational robustness are achieved. This method essentially
manipulates the nodal forces of finite elements directly to simu-
K T ðuÞu ¼ P ð10Þ late the effect of damage while leaving the material stiffness val-
ues unchanged. However, this method cannot acquire the
where u is the displacement vector, KT is the tangent stiffness ma-
information on the damage evolution due to stiffness degradation
trix which depends on the material properties and the unknown
effectively. In the following, three applicable methods are pro-
displacement u and P is the load vector. Generally, the well-known
posed to solve the softening problems.
Newton–Raphson iterative algorithm can be widely used to solve
the nonlinear equations due to its rapid convergence speed.
(1) The finite element equation in the dynamic algorithm is
By using these nonlinear equations, the flow chart for the pro-
given by
gressive failure analysis of composite laminates using finite ele-
ment method is illustrated in Fig. 4, which is summarized by four €¼P
K T u þ Mu ð11Þ
points: (1) For each load step, the finite element analysis is per-
formed and the on-axis stresses/strains at each element are ob- where M is the mass matrix. When KT becomes singular or
tained. (2) The stresses/strains at each element are compared non-positive after the failure point, the adaptable mass ma-
with the material allowable values and used to determine whether trix M is introduced to prevent singularity in the iterative
some elements have failed according to a certain failure criteria. If solution. For stable nonlinear calculations, the inertia effect
no failure is detected, then the applied load is increased and the represented by the matrix M should be small enough not to
analysis continues. (3) If some elements fail, the stiffness constants affect the calculation accuracy. This method can be accom-
are degraded according to the proposed stiffness degradation mod- plished by restarting the terminated static procedure after
els (damage evolution laws) using the CDM. At this time, as the ini- Newton–Raphson solution and switching to the dynamic
tial nonlinear solution no longer corresponds to an equilibrium analysis using implicit direct integration.
state, the equilibrium of composite structure requires to be re-
established using the modified mechanical properties for the failed (2) The finite element equation in the viscous regularization
elements while maintaining the current load level. This adjustment algorithm is expressed as
accounts for the material nonlinearity due to local stiffness degra-
K T u þ C u_ ¼ P ð12Þ
dation in the finite element analysis. (4) The calculations are per-
formed again and again under the same load P until there is no where C is the damping tensor. As KT becomes singular after
failed element anymore. (5) This iterative process repeats to obtain the failure point, the elements in C increase to obtain the
nonlinear equilibrium solutions after each load step until the solutions for u in such a way that the viscous forces intro-
appearance of a sudden catastrophic failure. duced are sufficiently large to prevent instantaneous collapse
However, as the incremental load increases to some values, the but small enough not to affect the behavior significantly
local/global buckling or unstability, i.e., the numerical convergence while the problem is stable. It is more efficient and accurate
problem for composite laminates may appear, which are reflected in most cases to relate the viscous regularization algorithm
by two points: (1) After some elements repeatedly fail, the corre- with the multiframe restart analysis in the finite element
sponding elastic constants degraded and several small eigenvalues analysis. Besides, the convergence problem can also be over-
in the element stiffness matrix approach zero, leading to the come by introducing a viscous constant into the damage ten-
appearance of ‘‘ill-conditioned” finite element equations. (2) Geo- sor D by its evolution rate [91]
3830 P.F. Liu, J.Y. Zheng / Materials and Design 31 (2010) 3825–3834

Start

Finite element modeling including

the material properties, boundary Macroscopic
conditions and load P scale

Calculate the off-axis

and on-axis stress/ Progressive
strain of each ply (or stiffness
element) in composite degradation
laminates
Mesoscopic/
Increase the Microscopic
load P=P+ΔP scale
Is there any Yes
failed ply
(element)?

No

No Does
the structure
fail?
Macroscopic
scale
Yes

End

Fig. 4. Flow chart of progressive failure analysis.

D_ ¼ ðD  D0 Þ=g ð13Þ linear for the viscous regularization algorithm. Although it can be
competent for both local and global instability problem with few
where g is the viscosity coefficient representing the relaxa- limitations, the viscous regularization algorithm cannot capture
tion time of the viscous system. Similarly, Gao and Bower the negative-slope portion of the load–displacement curve for global
[92] revised the Xu and Needlman’s cohesive model [93] by instability. Those three algorithms above should be appropriately
introducing a viscous constant into the constitutive equa- chosen in such a way that the theoretical results are in good agree-
tions so that the convergence problem was successfully ment with the experimental results. By the way, Germain et al.
solved. [97], Drabek and Böhm [98] used the nonlocal method which estab-
lished a precise link between the discrete and continuum formula-
(3) The finite element equation in the arc-length algorithm tions of the fracture energy to deal with the numerical
which was proposed by Riks [94], Ramm [95] and Crisfield convergence and mesh sensitivity problems at strain softening stage
[96] is written as by performing the internal integration over the weighted damage
tensor D. Further, Fredriksson [99] used the strain gradient theory
K T u ¼ kF ð14Þ to deal with these problems by introducing a variable characteristic
where k is a load factor within [1,1] which changes with the length into the anisotropic material constitutive relationship though
stiffness matrix KT to ensure an accurate solution of u. The it evolves initially in order to solve the multiscale problems.
arc-length algorithm imposes another constraint, which is Alternatively, the progressive failure analysis can be imple-
stated as mented using one of famous finite element software ABAQUS or
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ANSYS [100–103]. Besides, both the codes ABAQUS/UMAT (User
Material subroutine) and ANSYS-UPFs (User-Programmable Fea-
Du2 þ k2 ¼ R ð15Þ
tures) provide interfaces to simulate the stiffness degradation
where Du is the displacement increment and R is the arc- and damage evolution according to the proposed damage constitu-
length radius. tive models and damage evolution rates in the quasi-static analy-
It can be found from numerical calculations that the arc-length sis. In addition, ABAQUS/EXPLICIT provides a dynamic solution
algorithm can ensure highest solution precision among three algo- module on progressive failure analysis of composite laminates, in
rithms though the calculations are relatively time-consuming. In which the Hashin failure criterion is used and the damage evolu-
contrast, the dynamic algorithm exhibits more rapid convergence tion law stems from the cohesive energy consumption. In general,
velocity than other two algorithms since it shields off the ardous for the damage and failure problems of composite laminates, the
solution of inverse matrix and approximately substitutes differenti- eigenvalue buckling analysis is required to be initially performed
ation in the finite element method with the finite difference. Besides, to provide preliminary information about the structural cata-
the prediction could be inaccurate when the problem is highly non- strophic failure, followed by the subsequent progressive failure
 P.F. Liu, J.Y. Zheng / Materials and Design 31 (2010) 3825–3834 3831

CDM and homogenization theories. These microcracks and microv-

oids stems mainly from the uneven loading-bearing due to statis-
tical distributions of fiber strength and imperfect fiber/matrix
interface bonding properties. This shows the progressive failure
analysis should further directly considered the microscopic failure
properties of composites. Yet, current computers cannot be compe-
tent for such large number of calculations if accurate finite element
models from the fiber/matrix microscopic scale (for example, the
diameter of a carbon fiber is only 7 lm) to the macroscopic scale
of whole composite laminates (generally dozens of mm) would
be established. In this case, there is a trend for developing multi-
scale finite element methods in recent years to perform the pro-
gressive failure analysis of composite laminates which replaces
the evolvement of phenomenological damage tensor D with the
practical microscopic damage and failure mechanisms from the
point of view of composite micromechanics. This method over-
comes the disadvantage of exploring the failure problem of com-
posite laminates merely by continuum modeling. Bednarcyk and
Arnold [106] presented a framework for the multiscale stochastic
progressive failure analysis of composite laminates based on the
micromechanics and developed the finite element codes to predict
the nonlinear mechanical response for composites. They showed
that incorporating the fiber strength randomness on multiple
scales improves the quality of the simulation by enabling failure
at the microscopic locations other than that at the structural level
stress. Laurin et al. [107] proposed a multiscale progressive failure
approach for composite materials which allows predicting the fail-
ure of a lamina from the thermo-mechanical properties of the indi-
vidual unidirectional layers. This kind of approach takes into
account the effects of layer failures on the macroscopic behavior
Fig. 5. Newton–Raphson iterative algorithm becomes unavailable for two typical and the microscopic progressive failure criterion based on physical
unstability cases: (a) snap-through effect and (b) snap-back effect. principles. Van Der Meer and Sluys [108] examined the perfor-
mance of continuum material models for microlevel modeling of
progressive failure in composite laminates and proposed two dif-
analysis. As the finite element analysis for progressive failure anal- ferent continuum models: a continuum damage model and a soft-
ysis of polymer composite materials generally relates to the update ening plasticity model. They showed that a limitation of the
of damage tensor D for different failure modes and stiffness matri- continuum approach in the representation of matrix failure is
ces after each iteration, a large number of calculations require encountered which was explained as a consequence of the homog-
high-performance computer to perform parallel calculations in or- enization that is inherent in continuum models irrespective of the
der to improve calculation velocity and efficiency. applied failure criteria and material degradation laws. Yet, the
After performing the stiffness degradation on the failed com- transforming representation between the macroscopic-micro-
posite elements, the stresses redistribute and the stresses of the scopic damage and failure mechanisms were not really established.
failed elements are concentrated on the neighbouring elements. Pineda et al. [109] proposed a novel multiscale progressive damage
When the stress concentrations develop to some extent, the com- and failure model for fiber reinforced laminated composites which
posite structures completely lose their load-carrying capacity and used the generalized method of cells to resolve the lamina level
ultimately collapse. The tensile or compressive stress–strain curves strains into the constituent stresses/strains level and determined
for plane composite laminates and the internal pressure-radial the constituent failure by evaluating failure criteria at the micro-
strain curves for cylindrical composite laminates can lead to the fi- scopic fiber/matrix level. Further, Robbins and Reddy [110] pro-
nal failure strengths. posed an adaptive kinematics-based multiscale progressive
Reddy et al. [104], Akhras and Li [105] performed stiffness deg- damage and global failure model for in FRC laminates in which a
radation on the anisotropic elastic constants of composite materials hierarchical finite element method is developed to reduce the com-
after identifying the failure modes. In their work, the elastic modu- putational effort. Cox and Yang [1] emphasized the research on
lus and Poisson’s ratio of composite materials were multiplied by an hierarchical and multiscale models had offered formulations for
average value D within 0 and 1 regardless of the damage evolution formally linking top-down and bottom-up approaches. These
laws, which yet neglected the evolution of damage tensor D. There- developments focused on the mathematical issues of how to
fore, the stiffness degradation models should be chosen in such a embed the calculations representing fine-scale phenomena into
way that the practical damage and failure mechanisms of compos- the calculations representing larger scale phenomena. The mathe-
ite laminates can be perfectly represented instead of adopting sim- matical challenges include devising hierarchical meshing strate-
ple constant damage tensors. For different type of failure modes in gies that are coarse enough at the largest scales to cover perhaps
composite structures, how to determine the related coefficients in an entire structure, while cascading down through finer meshes,
the damage evolution laws accurately to match the experimental and associating the CDM with micromechanics of composites at
data is also worthy of further discussion and research. different scales by properly solving the various boundary-match-
As the stiffness degradation properties of composite laminates ing problems. Therefore, it had to be recognized the CDM still oper-
are essentially the macroscopic representation of the evolvement ates well in engineering design of composites since the multiscale
of microcracks and microvoids, enabling it insufficient to explore failure analysis using finite element method costs more than the
the failure performance of composites merely by employing the CDM in terms of calculation time and efficiency.
3832 P.F. Liu, J.Y. Zheng / Materials and Design 31 (2010) 3825–3834

At the same time, the extended finite element method (XFEM) tschko [120] used the XFEM to model the fracture properties of the
as a new finite element technique proposed by Belytschko and fiber/matrix composites by treating fracture with meshes that are
coworkers [111–114] develops which allows a crack to propagate independent of fiber/matrix interfaces and crack morphology. All
at any a plane in a composite element, as shown in Fig. 6. This discontinuities and near-tip enrichments based on the framework
method defines solid continuum elements with enriched element of local partition of unity are modeled and the level sets are used to
freedoms in which a general displacement discontinuity is intro- describe the geometry of the interfacial cracks so that no explicit
duced if some failure criterion is satisfied in the element at any representation of either the cracks or the material interfaces are
time during a simulation. Further, the displacement discontinuity needed. Further work is expected to concentrate on the methodol-
can be applied to a cohesive crack model with an evolving traction. ogies for the progressive failure analysis by combining the CDM,
Abdelaziz and Hamouine [115] presented an overview and recent cohesive theory and XFEM to investigate the multiscale failure
progress of the XFEM in the analysis of crack growth and showed properties for composite laminates that may be simultaneously
the XFEM makes it possible the accurate solution of engineering cracked in the interlamination and intralamination during the
problems in complex domains that may be practically impossible damage evolution process.
to solve using the classical finite element method. Guidault et al.
[116] proposed a multiscale XFEM for crack propagation based
on a domain decomposition method to account for the crack’s glo- 5. Conclusions
bal and local effects efficiently. Nistor et al. [117] presented
numerical implementation of the XFEM to analyze crack propaga- This paper presents a review on the recent developments for the
tion in a structure under dynamic loading and Giner et al. [118] methodologies in progressive failure analysis of composite lami-
introduced an implementation of the XFEM for fracture problems nates including the damage constitutive modeling based on the
within the finite element module ABAQUS-UEL (User element sub- CDM theory, the failure criteria, the damage evolution laws and
routine) including single or multiple cracks, as shown in Fig. 7. The the finite element implementation for predicting the variable-stiff-
nodes on those elements that crack passes through include two ness failure analysis of laminated structures. The methods for solv-
additional degrees of freedom (DOFs) by multiplying the shape ing numerical convergence problem using finite element analysis
functions by the introduced Heaviside functions, and those nodes due to stiffness degradation are also discussed. It should be noted
at the crack tip elements are assumed to have eight additional that the multiscale finite element analysis for progressive failure of
DOFs by introducing crack tip functions. Specially, Yan and Park laminated structures is an important topic and proper selection for
[119] applied the XFEM to the simulation of near-interfacial crack finite element algorithms are also crucial problems which deserve
propagation in a metal-ceramic layered structure. Huynh and Bely- further research. However, the phenomenological modeling based
on the CDM still bears responsibility to describe some specific fea-
tures of the damage evolution of composite laminates compared
with the multiscale progressive failure analysis which yet meets
large challenges in incorporating discrete fiber/matrix microscopic
modeling and failure mechanisms into the damage and failure
properties of the whole composite laminates. In addition, several
advanced finite element techniques such as the cohesive theory
which can be used to predict the damage evolution properties of
composite laminates and the extended finite element method that
allows cracks to penetrate into an element are expected to acceler-
ate their application to the progressive failure analysis of compos-
ite laminates.

Acknowledgements

This research is supported by the national natural science fund-

ing of China (Number: 50905197), the high-technology research
and development program (863 key program) of China (Number:
Fig. 6. Schematic representation of crack passing through elements in XFEM [111].
2009AA044801), the key project Chinese universities scientific
funding of Zhejiang University and the key project of national pro-
grams for fundamental research and development (973 program)
of China (Number: 2007CB209706).

References

[1] Cox BN, Yang QD. In quest of virtual tests for structural composites. Science
2006;314:1102–7.
[2] Young KS. Advanced composites storage containment for hydrogen. Int J
Hydrogen Energy 1992;17(7):505–7.
[3] Chapelle DD, Perreux D. Optimal design of a Type 3 hydrogen vessel: part I-
analytic modelling of the cylindrical section. Int J Hydrogen Energy
2006;31(5):627–38.
[4] Liu PF, Xu P, Zheng JY. Artificial immune system for optimal design of
composite hydrogen storage vessel. Comput Mater Sci 2009;47(1):261–7.
[5] Zheng JY, Liu PF. Elasto-plastic stress analysis and burst strength evaluation of
Al–carbon fiber/epoxy composite cylindrical laminates. Comput Mater Sci
2008;42(4):453–61.
Fig. 7. Enriched nodes in the XFEM using finite element analysis. Circles: nodes [6] Xu P, Zheng JY, Liu PF. Finite element analysis of composite hydrogen storage
with two additional DOFs. Squares: nodes with eight additional DOFs [118]. vessel. Mater Des 2009;31(7):2295–301.
 P.F. Liu, J.Y. Zheng / Materials and Design 31 (2010) 3825–3834 3833

[7] Liu PF, Zheng JY. Review on methodologies of progressive failure analysis of [43] Barbero EJ, Vivo LD. A constitutive model for elastic damage in fiber-
composite laminates. In: Koppel A, Oja J, editors. Continuum mechanics. New reinforced PMC laminae. Int J Damage Mech 2001;10(1):73–93.
York: Nova Science Publishers; 2009 [chapter 11]. [44] Perreux D, Thiebaud F. Damaged elasto-plastic behaviour of [+u,u] fibre-
[8] Effendi RR, Barrau JJ, Guedra-Degeorges D. Failure mechanism analysis under reinforced composite laminates in biaxial loading. Compos Sci Technol
compression loading of unidirectional carbon/epoxy composites using 1995;54(3):275–85.
micromechanical modeling. Compos Struct 1995;31(2):87–98. [45] Ferry L, Perreux D, Rousseau J, Richard F. Interaction between plasticity and
[9] Phillips EA, Herakovich CT, Graham LL. Damage development in composites damage in the behaviour of [+u,u] fibre reinforced composite pipes in
with large stress gradients. Compos Sci Technol 2001;61(5):2169–82. biaxial loading (internal pressure and tension). Composites Part B
[10] Bussiba A, Kupiec M, Ifergane S, Piat R, Böhlke T. Damage evolution and 1998;29(6):715–23.
fracture events sequence in various composites by acoustic emission [46] Perreux D, Robinet D, Chapelle D. The effect of internal stress on the
technique. Compos Sci Technol 2008;68(5):1144–55. identification of the mechanical behaviour of composite pipes. Compos Part A
[11] Zhang DX, Ye JQ, Lam D. Ply cracking and stiffness degradation in cross-ply 2006;37(4):630–5.
laminates under biaxial extension, bending and thermal loading. Compos [47] Liu PF, Zheng JY. Progressive failure analysis of carbon fiber/epoxy composite
Struct 2006;75(1–4):121–31. laminates using continuum damage mechanics. Mater Sci Eng A 2008;485(1–
[12] Bouazza M, Tounsi A, Benzair A, Adda-bedia EA. Effect of transverse cracking 2):711–7.
on stiffness reduction of hygrothermal aged cross-ply laminates. Mater Des [48] Chow CL, Yang F. Three-dimensional inelastic stress analysis of center
2007;28(4):1116–23. notched composite laminates with damage. Int J Damage Mech
[13] Spootswood MS, Palazotto AN. Progressive failure analysis of a composite 1997;6(1):23–50.
shell. Compos Struct 2007;53(1):117–31. [49] Lin WP, Hu HT. Nonlinear analysis of fiber-reinforced composite laminates
[14] Kachanov LM. Time of the rupture process under creep conditions. IVZ Akad subjected to uniaxial tensile load. J Compos Mater 2002;36(12):1429–50.
Nauk SSR Otd Tech Nauk 1958;8:26–31 [in Russian]. [50] Boutaous A, Peseux B, Gornet L, Bélaidi A. A new modeling of plasticity
[15] Tsai SW, Wu E M. A general theory of strength for anisotropic materials. J coupled with the damage and identification for carbon fibre composite
Compos Mater 1971;5(1):58–80. laminates. Compos Struct 2006;74(1):1–9.
[16] Hoffman O. The brittle strength of orthotropic material. J Compos Mater [51] Barbero EJ. An inelastic damage model for fiber reinforced laminates. J
1967;1(2):200–6. Compos Mater 2002;36(8):, 941–962.
[17] Hinton MJ, Soden PD. Predicting failure in composite laminates: the [52] Reddy YSN, Pandey AK. A first-ply failure analysis of composite laminates.
background to the exercise. Compos Sci Technol 1998;58(7):1001–10. Comput Struct 1987;25(3):371–93.
[18] Soden PD, Hinton JM, Kaddour AS. A comparison of the predictive capabilities [53] Tay TE, Liu G, Tan VBC, Sun XS, Pham DC. Progressive failure analysis of
of current failure theories for composite laminates. Compos Sci Technol composites. J Compos Mater 2008;42(18):1921–66.
1998;58(7):1225–54. [54] Davila CG, Camanho PP, Rose CA. Failure criteria for FRP laminates. J Compos
[19] Theocaris PS. Weighing failure tensor polynomial criteria for composites. Int J Mater 2005;39(4):323–45.
Damage Mech 1992;1(1):4–46. [55] Sleight DW. Progressive failure analysis methodology for laminated
[20] Rotem A, Hashin Z. Failure modes of angle ply laminates. J Compos Mater composite structures. NASA report; 1999. p. 48–63.
1975;9(2):191–206. [56] Wu RY, Stachurski Z. Evaluation of the normal stress interaction parameter in
[21] Narayanaswami R, Adelman HM. Evaluation of the tensor polynomial and the tensor polynomial strength theory for anisotropic materials. J Compos
Hoffman strength theories for composite materials. J Compos Mater Mater 1984;18(5):456–63.
1977;11(4):366–77. [57] Cuntze RG, Freund A. The predictive capability of failure mode concept-based
[22] Hashin Z. Failure criteria for unidirectional fiber composites. J Appl Mech strength criteria for multidirectional laminates. Compos Sci Technol
1980;47:329–34. 2004;64(3–4):343–77.
[23] Yamada SE, Sun CT. Analysis of laminated strength and its distribution. J [58] Padhi GS, Shenoi RA, Moy SSJ, Hawkins GL. Progressive failure and ultimate
Compos Mater 1978;12:275–84. collapse of laminated composite plates in bending. Compos Struct
[24] Zhang YX, Yang CH. Recent developments in finite element analysis for 1997;40(3–4):277–91.
laminated composite plates. Compos Struct 2009;88(1):147–57. [59] Tay TE, Tan SHN, Tan VBC, Gosse JH. Damage progression by the element-
[25] Orifici AC, Herszberg I, Thomson RS. Review of methodologies for composite failure method (EFM) and strain invariant failure theory (SIFT). Compos Sci
material modelling incorporating failure. Compos Struct 2008;86(1– Technol 2005;65(6):935–44.
3):194–210. [60] Tay TE, Liu G, Yudhanto A, Tan VBC. A micro-macro approach to modeling
[26] Lemaitre J, Chaboche JL. Mechanics of solid materials. Cambridge: Cambridge progressive damage in composite structures. Int J Damage Mech 2008;17(1):
University Press; 1990. 5–28.
[27] Lemaitre J. A course on damage mechanics. Berlin: Springer-Verlag; 1992. [61] Li R, Kelly D, Ness R. Application of a first invariant strain criterion for matrix
[28] Ristinmaa M, Ottosen NS. Viscoplasticity based on an additive split of the failure in composite materials. J Compos Mater 2003;37(22):1977–2000.
conjugated forces. Euro J Mech A – Solids 1998;17(2):207–35. [62] Mayes JS, Hansen AC. Composite laminate failure analysis using
[29] Schapery RA. A theory of mechanical behavior of elastic media with growing multicontinuum theory. Compos Sci Technol 2004;64(3–4):379–94.
damage and other changes in structure. J Mech Phys Solids [63] Mayes JS, Hansen AC. A comparison of multicontinuum theory based failure
1990;38(2):215–53. simulation with experimental results. Compos Sci Technol 2004;64(3–
[30] Murakami S, Kamiya K. Constitutive and damage evolution equations of 4):517–27.
elastic–brittle materials based on irreversible thermodynamics. Int J Mech Sci [64] Zhu H, Sankar BV, Marrey RV. Evaluation of failure criteria for fiber
1997;39(4):473–86. composites using finite element micromechanics. J Compos Mater
[31] Hayakawa K, Murakami S, Liu Y. An irreversible thermodynamics theory for 1998;32(8):766–82.
elastic–plastic-damage materials. Euro J Mech A – Solids 1998;17(1):13–32. [65] Ha SK, Jin KK, Huang Y. Micro-mechanics of failure (MMF) for continuous
[32] Tang XS, Jiang CP, Zheng JL. Anisotropic elastic constitutive relations for fiber reinforced composites. J Compos Mater 2008;42(18):1873–95.
damaged materials by application of irreversible thermodynamics. Theor [66] Rice JR. A path independent integral and the approximate analysis
Appl Fract Mech 2002;38(3):211–20. of strain concentration by notches and cracks. J Appl Mech 1968;35(2):
[33] Brünig M. An anisotropic ductile damage model based on irreversible 379–86.
thermodynamics. Int J Plast 2003;19(10):1679–713. [67] Rybicki EF, Kanninen MF. A finite element calculation of stress intensity
[34] Olsson M, Ristinmaa M. Damage evolution in elasto-plastic materials factors by a modified crack closure integral. Eng Fract Mech 1977;9(4):931–8.
response due to different concepts. Int J Damage Mech 2003;12(2):115–39. [68] Rybicki EF. An energy release rate approach for stable crack growth in the
[35] Basu S, Waas AM, Ambur RD. Prediction of progressive failure in free-edge delamination problem. J Compos Mater 1977;11(4):470–87.
multidirectional composite laminated panels. Int J Solids Struct [69] Irwin GR. Fracture dynamics, fracturing of metals, vol. 8. Cleveland: American
2007;44(9):2648–76. Society for Metals; 1948. p. 147–66.
[36] Maimí P, Camanho PP, Mayugo JA. A continuum damage model for composite [70] Krueger R. Virtual crack closure technique: history, approach, and
laminates: part I – constitutive model. Mech Mater 2007;39(10):897–908. applications. Appl Mech Rev 2004;57(2):109–43.
[37] Maimí P, Camanho PP, Mayugo JA. A continuum damage model for composite [71] Xie D, Biggers SB. Progressive crack growth analysis using interface element
laminates: part II – computational implementation and validation. Mech based on the virtual crack closure technique. Finite Elem Anal Des
Mater 2007;39(10):909–19. 2006;42(11):977–84.
[38] Matzenmiller A, Lubliner J, Taylor RL. A constitutive model for anisotropic [72] Leski A. Implementation of the virtual crack closure technique in engineering
damage in fiber-composites. Mech Mater 1995;20(2):125–52. FE calculations. Finite Elem Anal Des 2007;43(3):261–8.
[39] Kwon YW, Liu CT. Study of damage evolution in composites using damage [73] Orifici AC, Thomson RS, Degenhardt R, Bisagni C, Bayandor J. A finite element
mechanics and micromechanics. Compos Struct 1997;38(1–4):133–9. methodology for analysing degradation and collapse in postbuckling
[40] Schipperen JHA. An anisotropic damage model for the description of composite aerospace structures. J Compos Mater 2009;43:3239–63.
transverse matrix cracking in a graphite–epoxy laminate. Compos Struct [74] Dugdale DS. Yielding of steel sheets containing slits. J Mech Phys Solids
2001;53(3):295–9. 1960;8(2):100–4.
[41] Maa RH, Cheng JH. A CDM-based failure model for predicting strength of [75] Barenblatt I. Mathematical theory of equilibrium cracks in brittle fracture.
notched composite laminates. Composites Part B 2002;33(6):479–89. Adv Appl Mech 1962;7:55–129.
[42] Camanho PP, Maimí P, Dávila CG. Prediction of size effects in notched [76] Hutchinson JW, Suo Z. Mixed mode cracking in layered materials. Adv Appl
laminates using continuum damage mechanics. Compos Sci Technol Mech 1992;29:63–191.
2007;67(13):2715–27.
3834 P.F. Liu, J.Y. Zheng / Materials and Design 31 (2010) 3825–3834

[77] Tvergaard V, Hutchinson JW. Effect of strain-dependent cohesive zone model [100] Huang ZM. Failure analysis of laminated structures by FEM based on
on predictions of crack growth resistance. Int J Solids Struct 1996;33(20– nonlinear constitutive relationship. Compos Struct 2007;77(3):270–9.
22):3297–308. [101] Fredriksson P, Gudmundson P, Mikkelsen LP. Finite element implementation
[78] Allen DH, Searcy CR. Numerical aspects of a micromechanical model of a and numerical issues of strain gradient plasticity with application to metal
cohesive zone. J Reinf Plast Compos 2000;19(3):240–8. matrix composites. Int J Solids Struct 2009;46(22–23):3977–87.
[79] Camanho PP, Davila CG, Moura MF. Numerical simulation of mixed-mode [102] Hibbit, Karlsson, Sorensen. ABAQUS user’s manual. Version 6.8; 2007.
progressive delamination in composite materials. J Compos Mater [103] Swanson J. ANSYS user’s manual. Version 11.0; 2007.
2003;37(16):1415–38. [104] Reddy YSN. Non-linear progressive failure analysis of laminated composite
[80] Xie D, Waas AM. Discrete cohesive zone model for mixed-mode fracture using plates. Int J Non-linear Mech 1995;30(5):629–49.
finite element analysis. Eng Fract Mech 2006;73(13):1783–96. [105] Akhras G, Li WC. Progressive failure analysis of thick composite plates using
[81] Turon A, Dávila CG, Camanho PP, Costa J. An engineering solution for mesh the spline finite strip method. Compos Struct 2007;79(1):34–43.
size effects in the simulation of delamination using cohesive zone models. [106] Bednarcyk BA, Arnold SM. A framework for performing multiscale stochastic
Eng Fract Mech 2007;74(10):1665–82. progressive failure analysis of composite structures. In: 2006 ABAQUS users’
[82] Sun CT, Jin ZH. Modeling of composite fracture using cohesive zone and conference.
bridging models. Compos Sci Technol 2006;66(10):1297–302. [107] Laurin F, Carre‘re N, Maire JF. A multiscale progressive failure approach for
[83] Bao G, Suo Z. Remarks on crack-bridging concepts. Appl Mech Rev composite laminates based on thermodynamical viscoelastic and damage
1992;45:355–66. models. Composites Part A 2007;38(1):198–209.
[84] Clegg WJ. Controlling cracks in ceramics. Science 1999;286:1097–9. [108] Van Der Meer FP, Sluys LJ. Continuum models for the analysis of progressive
[85] Kruzic JJ. Predicting fatigue failures. Science 2009;325:156–8. failure in composite laminates. J Compos Mater 2009;43(20):2131–56.
[86] Cox BN, Yang QD. Cohesive models for damage evolution in laminated [109] Pineda EJ, Waas AM, Bednarcyk BA, Collier CS, Yarrington P. Progressive
composites. Int J Fract 2005;133(2):107–37. damage and failure modeling in notched laminated fiber reinforced
[87] Shin CS, Wang CM. An improved cohesive zone model for residual notched composites. Int J Fract 2009;158(2):125–43.
strength prediction of composite laminates with different orthotropic lay- [110] Robbins DH, Reddy Jr JN. Adaptive hierarchical kinematics in modeling
ups. J Compos Mater 2004;38(9):713–36. progressive damage and global failure in fiber-reinforced composite
[88] Lapczyk I, Hurtado JA. Progressive damage modeling in fiber-reinforced laminates. J Compos Mater 2008;42(2):143–72.
materials. Composites Part A 2007;38(1):2333–41. [111] Belytschko T, Black T. Elastic crack growth in finite elements with minimal
[89] Rami HA. Cohesive micromechanics: a new approach for progressive damage remeshing. Int J Numer Meth Eng 1999;45(5):601–20.
modeling in laminated composites. Int J Damage Mech 2009;18(8):691–719. [112] Moës N, Dolbow J, Belytschko T. A finite element method for crack growth
[90] Ochoa O, Reddy JN. Finite element analysis of composite without remeshing. Int J Numer Meth Eng 1999;46(1):131–50.
laminates. Dordrecht: Kluwer Academic Publishers; 1992. [113] Moës N, Belytschko T. Extended finite element method for cohesive crack
[91] Duvaut G, Lions JL. Inequalities in mechanics and physics. Berlin: Springer; growth. Eng Fract Mech 2002;69(7):813–33.
1976. [114] John Dolbow, Nicolas Moës, Ted Belytschko. Discontinuous enrichment in
[92] Gao YF, Bower AF. A simple technique for avoiding convergence problems in finite elements with a partition of unity method. Finite Elem Anal Des
finite element simulations of crack nucleation and growth on cohesive 2000;36(3–4):235–60.
interfaces. Model Simul Mater Sci Eng 2004;12(3):453–63. [115] Abdelaziz Y, Hamouine A. A survey of the extended finite element. Comput
[93] Xu XP, Needleman A. Numerical simulations of fast crack growth in brittle Struct 2008;86(11–12):1141–51.
solids. J Mech Phys Solids 1994;42(9):1397–434. [116] Guidault PA, Allix O, Champaney L, Cornuault C. A multiscale extended finite
[94] Riks E. An incremental approach to the solution of snapping and buckling element method for crack propagation. Comput Meth Appl Mech Eng
problems. Int J Solids Struct 1979;15(7):529–51. 2008;197(5):381–99.
[95] Ramm E. Strategies for tracing the nonlinear response near limit points. In: [117] Nistor I, Pantalé O, Caperaa S. Numerical implementation of the extended
Nonlinear finite element analysis in structural mechanics. New finite element method for dynamic crack analysis. Adv Eng Software
York: Springer-Verlag; 1981. 2008;39(7):573–87.
[96] Crisfield MA. An arc-length method including line searches and accelerations. [118] Giner E, Sukumar N, Tarancón JE, Fuenmayor FJ. An Abaqus implementation
Int J Numer Meth Eng 1983;19(8):1269–89. of the extended finite element method. Eng Fract Mech 2009;76(3):347–68.
[97] Germain N, Besson J, Feyel F. Composite layered materials: anisotropic [119] Yan YH, Park SH. An extended finite element method for modeling near-
nonlocal damage models. Comput Meth Appl Mech Eng 2007;41– interfacial crack propagation in a layered structure. Int J Solids Struct
44(196):4272–82. 2008;45(17):4756–65.
[98] Drabek T, Böhm HJ. Micromechanical finite element analysis of metal matrix [120] Huynh DBP, Belytschko T. The extended finite element method for fracture in
composites using nonlocal ductile failure models. Compos Mater Sci composite materials. Int J Numer Meth Eng 2009;77(2):214–39.
2006;37(1–2):29–36.
[99] Huang ZM. Inelastic and failure analysis of laminate structures by ABAQUS
incorporated with a general constitutive relationship. J Reinf Plast Compos
2007;26(11):1135–81.