Original Article

Journal of Reinforced Plastics

and Composites

Failure analysis of carbon fiber reinforced 0(0) 1–20

! The Author(s) 2016
Reprints and permissions:
composite subjected to low velocity sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0731684415627381
impact and compression after impact jrp.sagepub.com

Geng Han1, Zhidong Guan1, Xing Li2 and Shanyi Du1,3

Abstract
Composites are very sensitive to the load in the direction of thickness, especially out-plane low velocity impact (LVI).
In this article, the detailed failure mechanisms of composite laminates with four material systems (CCF300/Epoxy,
CCF300/Bismaleimide, CCF800/Epoxy, and CCF800/Bismaleimide) under the loading of LVI and compression after
impact (CAI) were studied by experiment and finite element analysis. In finite element model, a newly proposed
multi-scale failure criterion (MMF3) and cohesive elements are used to determine the intralaminar damage and inter-
laminar delamination, respectively. Through experiment and finite element analysis, it can be concluded that the damage
tolerance performance of epoxy resin composites are better than that of bismaleimide resin composites, and the damage
tolerance performance of CCF300 carbon fiber composites are better than that of CCF800 carbon fiber composites.
The proposed multi-scale failure analysis method can effectively simulate the impact and CAI damage process of
composite laminates.

Keywords
Carbon-fiber reinforced composite, low velocity impact, damage tolerance, finite element analysis, multi-scale failure
criterion (MMF3)

effects on the efficient application of composite

Introduction
structures.
Carbon fiber reinforced composites have been widely A great deal of works have been done on this issue
used in various fields since their birth because of their by many researchers, especially focused on carbon
specific stiffness and specific strength, strong heat resist- fiber1–3 or glass fiber composites,4–7 and other material
ance and corrosion resistance, designability, and many systems have also been researched, such as aluminum
other advantages. During the manufacture, mainten- alloy fiber metal laminates,8 GLARE fiber metal lamin-
ance, and service life, composites may suffer a variety ates,9 and stitched foam-core sandwich composites.10
of damages, of which low velocity impact (LVI) is con-
sidered as one of the most dangerous because it can
cause significant reduction of the structure strength;
1
however, the damage usually cannot be detected from School of Aeronautic Science and Engineering, Beihang University,
Beijing, China
the surface of the component obviously. Compression 2
Composite Materials and Structures Group, Beijing Aeronautical Science
after impact (CAI) strength is one of the main quotas & Technology Research Institute (BASTRI), Beijing, China
for damage tolerance properties of composite struc- 3
Center for Composite Materials and Structures, Harbin Institute of
tures, and it can give a quantitative characterization Technology, Harbin, China
of the residual intensity of composite induced by initial
damage such as LVI. So the ability to accurately predict Corresponding author:
Zhidong Guan, School of Aeronautic Science and Engineering, Beihang
the damage extent and CAI strength of composite University, Xue Yuan Road No. 37, Hai Dian District, Beijing 0086-
structures can potentially lead to the exploration of a 100191, China.
larger design space with low cost, and it has significant Email: [email protected]

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
2 Journal of Reinforced Plastics and Composites 0(0)

Abrate11 wrote an entire book on impact which approach to take into account different damage
explains how damage develops during impact, the modes by introducing different damage variables, of
effect of impact-induced damage on the mechanical which the Hashin criteria32,33 still remains one of the
behavior of structures, and methods of damage predic- most effective failure criteria.
tion and detection, and also numerous examples are Although the above researchers have analyzed the
included to illustrate these topics. Robin Olsson12 has failure modes of LVI and CAI through experiments
done a series of researches on impact, including impact and numerical simulations, the progressive damage
response of orthotropic composite plates predicted process based on composite constituents needs to be
from a one-parameter differential equation, analytical further researched. And the CDM is based on the fail-
prediction of large mass impact damage in composite ure modes of experience, in which the complex failure
laminates,13 engineering method for prediction of mechanisms and accurate failure envelope of resin
impact response and damage in sandwich panels,14 matrix have not been considered adequately. In this
closed form prediction of peak load and delamination article, the detailed failure mechanisms of composite
onset under small mass impact,15 and delamination laminates with four material systems (CCF300/Epoxy,
threshold load for dynamic impact on plates.16 CCF300/Bismaleimide, CCF800/Epoxy, and CCF800/
LVI and quasi-static indentation tests to laminates Bismaleimide) under the loading of LVI and CAI were
were made for various composite systems. Chen et al.17 studied by experimental method and finite element ana-
discovered that fiber breakage plays an important role lysis. In finite element model, a newly proposed multi-
in the permanent indentation formation. Belingardi and scale failure criterion (MMF3) is used to determine the
Vadori18 observed a linear relationship between contact intralaminar damage, and cohesive elements are used to
force and thickness for different graphite/epoxy lamin- determine the interlaminar delamination. The calcu-
ates. Caprino et al.19 observed that the maximum lated strength and failure modes can agree with test
contact force and penetration energy of carbon/epoxy results very well so that it can verify the validity of
laminates increased with increasing thickness. In this model.
numerical simulation, Wei He20 established a finite
element model to predict the permanent indentation
incorporating fiber failure based on the anisotropic Materials and experiment
elasto-plasticity theory, and delamination was also
Specimen and preparation
taken into account in the model using cohesive elem-
ents. Also there have been a lot of works to reveal the Out-plane LVI is the most sensitive loading form
failure mechanism and influence factors of CAI. Thi D. for composite laminates, which can cause a large
Dang21 addressed the problem of the impact and CAI number of invisible delamination damages in the inter-
behavior prediction of variable stiffness composite nal structure and severely reduces the residual compres-
laminates with emphasis on the effect of the interaction sive strength of composite laminates. Composite
between fiber orientations, matrix cracks, and delamin- systems with four different fiber/matrix combin-
ations. H. Yan22 conducted a detailed numerical ana- ations have been designed, which are CCF300/Epoxy,
lysis of the compression after impact response of a CCF300/Bismaleimide, CCF800/Epoxy, and CCF800/
woven fiber composite composed of E-glass fibers and Bismaleimide, respectively. The mechanical properties
vinyl ester resin, and failure mechanisms of matrix of these four composite systems are shown in Tables 1
cracking and fiber breakage were explicitly modeled and 2. The standard energy levels of LVI are 4.45 J/mm
based on a multi-scale computational method recently and 6.67 J/mm, respectively. The information of speci-
proposed by Oskay and coworkers.23,24 B. Vieille25 men layer and size, and the impact energy are shown in
further investigated the mechanisms leading to the com- Table 3. The stacking sequence of two kinds of CCF300
pressive failure of impact-damaged laminates and carbon fiber composite laminate is [45/0/–45/90]4s, and
analyzed the role of initial damage (delamination, post- single layer thickness is 0.125 mm. The stacking
impact cracks, permanent indentation, etc.) on the sequence of two kinds of CCF800 carbon fiber compos-
onset of buckling and cracks development during CAI ite laminate is [45/0/–45/90]3s, and single layer thickness
loadings by multi-instrumented CAI tests (Digital is 0.19 mm.
Image Stereo Correlation). LVI tests were conducted based on ASTM D713637
During the impact and compression process, various and using an instrumented drop-weight impact testing
damage modes may occur simultaneously, including device as shown in Figure 1a. The testing device con-
fiber breakage, matrix cracking and crush, delamin- sists of a drop-weight tower, a clamping fixture, and a
ation, buckling and post-buckling, which are very com- piezo-electric load cell together with the data acquisi-
plex. In previous researches, continuum damage tion system and an optical velocity indicator. The great-
mechanics (CDM)26–31 was proved to be an effective est vertical distance from the dent bottom induced by

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
Han et al. 3

Table 1. The mechanical properties of four composite systems.34,35

Mechanical properties CCF300/Epoxy CCF300/Bismaleimide CCF800/Epoxy CCF800/Bismaleimide

Longitudinal modulus, E1 (GPa) 123.91 145.0 156.7 170.7

Transverse modulus, E2 ¼ E3 (GPa) 9.72 9.75 7.19 8.89
The in-plane shear modulus, G12 ¼ G13 (GPa) 4.53 5.69 4.21 4.90
Transverse shear modulus, G23 (GPa) 2.56 3.38 2.48 3.00
The in-plane Poisson’s ratio, v12 ¼ v13 0.288 0.312 0.314 0.329
Transverse Poisson’s ratio, v23 0.347 0.444 0.451 0.482
Longitudinal tensile strength, XT (MPa) 1762.3 1858.0 2195.6 2262.6
Longitudinal compressive strength, XC (MPa) 1362.2 1318.0 1365.7 1450.7
Transverse tensile strength, YT (MPa) 71.1 31.8a 55.8 31.8
Transverse compressive strength, YC (MPa) 218.3 185.9a 223.3 185.9
The in-plane shear strength, S (MPa) 83.5 60.7a 87.7 60.7
a
The transverse tensile strength, transverse compressive strength and in plane shear strength of CCF300/Bismaleimide are chosen as the same with
CCF800/Bismaleimide in due to the same kind of matrix.

Table 2. Microscopic critical strength.36

Microscopic mechanical parameters CCF300/Epoxy CCF300/Bismaleimide CCF800/Epoxy CCF800/Bismaleimide

Fiber tensile strength Tf (MPa) 3091.8 3261.3 3868.3 3987.2

Fiber compressive strength Cf (MPa) 2440.6 2362.3 2452.4 2603.3
Matrix tensile strength Tm (MPa) 138.7 70.9 108.1 70.9
Matrix compressive strength Cm (MPa) 353.3 277.2 338.5 277.2
Matrix shear strength Sm (MPa) 145.1 103.1 148.9 103.1

Table 3. The test matrix of different material systems.

Nominal Impact energy

Material system Stacking sequence thickness (mm) Size (mm) Test standard (J/mm)

CCF300/Epoxy [45/0/–45/90]4s 4.0 150  100 ASTM D7136 4.45

ASTM D7137 6.67
CCF300/Bismaleimide [45/0/–45/90]4s 4.0 150  100 4.45
6.67
CCF800/Epoxy [45/0/–45/90]3s 4.56 150  100 4.45
6.67
CCF800/Bismaleimide [45/0/–45/90]3s 4.56 150  100 4.45
6.67

the impact to specimen surface which is not affected by loading process was recorded through the data acqui-
the influence of the dent is measured by dent measuring sition system, and the specimen strain in test section
device and the internal damage detection is conducted was acquired through strain gauge.
by using C-scan method as shown in Figure 1b. Also
layer-tear-off technique is used to research the layered
Test result
morphology of every monolayer caused by impact.
CAI tests were conducted according to ASTM LVI. The impact force–time curves of the four kinds of
D713738 and on WDW-200E 20T electronic static test- material under the impact energy of 4.45 J/mm
ing machine. The load–displacement curve during and 6.67 J/mm, respectively, are shown in Figure 2.

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
4 Journal of Reinforced Plastics and Composites 0(0)

Figure 1. (a) The drop-weight impact testing device. (b) The dent measuring device and C-scan equipment.

The peak load and duration of impact process can be carbon fiber composites are smoother compared to the
obtained from the response of contact force between CCF300 carbon fiber composites, with no more violent
the laminate and punch during impact, and damage fluctuations.
modes such as delamination and fiber fracture can Under the impact energy of 6.67 J/mm, the max-
also be determined to a certain extent. It can be seen imum contact force of CCF800 carbon fiber composites
from Figure 2 that under the impact energy of 4.45 J/mm, reaches about 10,000 N, but the maximum contact
the maximum contact forces of two kinds of CCF800 force of CCF300 carbon fiber composites is about
carbon fiber composite materials are about 8000 N, 6000–7000 N without significant improvement com-
which are significantly higher than that of the two pared to the impact energy of 4.45 J/mm. At this
kinds of CCF300 carbon fiber composites (about time, the impact process time of two kinds of
6000 N). It also shows that the stiffness of CCF800 CCF800 carbon fiber composites did not change obvi-
carbon fiber is higher than CCF300 carbon fiber. The ously compared to 4.45 J/mm, and the shortest impact
maximum contact force is larger, and the time of process time is still 0.005 s (CCF800/Epoxy).
impact process is shorter. The shortest impact process Meanwhile the impact process time of the two kinds
time is about 0.005 s (CCF800/Epoxy), and the longest of CCF300 carbon fiber composites increased signifi-
impact process time is about 0.0067 s (CCF300/Epoxy). cantly, and the longest impact process time reached
The contact force curves of the two kinds of CCF800 0.007 s (CCF300/Epoxy). Usually, the longer duration

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
Han et al. 5

Figure 2. (a) The impact force–time curves of four kinds of material under the impact energy of 4.45 J/mm. (b) The impact
force–time curves of four kinds of material under the impact energy of 6.67 J/mm.

of impact contact force indicates more serious damages images processed by Photoshop are shown in Figure 4a
(especially fiber damage) inside layers, and the above and b, respectively. It can be seen that the appearance
analysis also shows that fiber breakage occurred in two of delamination presents like ‘‘peanut,’’ and the angle
kinds of CCF300 carbon fiber composite laminates of delamination is related to the ply direction of the
under the impact energy of 6.67 J/mm. adjacent layer. With the middle plane of thickness dir-
Ultrasonic C-scan equipment (Figure 1b) was ection regarded as the boundary, delamination area of
used to detect the delamination of the four kinds of each layer in back of the punch is slightly larger than
laminates under the impact energy of 4.45 J/mm and that in front of the punch. Especially, a wide range of
6.67 J/mm, respectively, as shown in Figure 3, in matrix cracking and delamination damages occur in
which the results of CCF300/Bismaleimide were super- back of the laminate under the impact energy of
imposed from the delamination area of each layer 6.67 J/mm.
resulted from layer-tear-off technique through image Under the impact energy of 6.67 J/mm, a small
processing software. It can be clearly seen that delam- amount of fiber breakages occur within some mono-
ination area would increase under larger impact energy layers of CCF300/Bismaleimide composite laminates
for all kinds of materials. On the other hand, damage in back of the punch as shown in Figure 5. This also
morphology and area under the same impact energy proves the conclusion in the analysis of the impact
differ greatly for different materials. For the same force–time curve that fiber breakage leads to the stiff-
kind of carbon fiber, delamination areas of bismalei- ness decrease of laminate.
mide resin composite laminates are larger than those The dent depths and delamination areas of each
of epoxy resin composite laminates, which shows that material system under different impact energy are
the interlaminar toughness of bismaleimide resin is shown in Table 4. It can be concluded that the dent
weaker. For the same kind of resin, delamination depths of the two kinds of CCF800 carbon fiber
areas of CCF800 carbon fiber composite laminates composite laminates under different impact energy are
are larger than those of CCF300 carbon fiber composite relatively small. The dent depth of CCF300 carbon
laminates, and this is because the higher stiffness of fiber composite laminates under the impact energy of
CCF800 carbon fiber leads to the greater impact con- 4.45 J/mm is small, which under the impact energy of
tact force. So the interlaminar stress of CCF800 carbon 6.67 J/mm is quite large. The determination of struc-
fiber composite laminates is greater than that of tural design values need to ensure that the structure
CCF300 carbon fiber composite laminates, which bears ultimate force in the condition of Barely Visible
results in more severe interlaminar damage. Impact Damage (BVID). And the dent depth of impact
Layer-tear-off technique was used to obtain the clear is an important index of damage detection. If the dent
damage morphology of each layer of the CCF300 depth is quite large, we can detect the damage effect-
carbon fiber composite laminates under the impact ively so as to eliminate the security risks caused by
energy of 4.45 J/mm and 6.67 J/mm, respectively. The completely invisible damage.

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
6 Journal of Reinforced Plastics and Composites 0(0)

Figure 3. The C-scan results after impact of four kinds of composite laminates.

CAI. The CAI stress–displacement curves of the four other side is fiber compressive fracture, which is the
kinds of composite laminates under the impact energy same with CCF800/Bismaleimide composite laminates
of 4.45 J/mm and 6.67 J/mm are shown in Figure 6. under the impact energy of 4.45 J/mm. While for
Under the impact energy of 4.45 J/mm, the compressive CCF800/Bismaleimide composite laminates under the
stress–displacement curves of the four kinds of mater- impact energy of 6.67 J/mm, sublayer buckling both
ials are all linear basically. Only the curve of CCF300/ appears on the right and reverse face during compres-
Epoxy in the zone near damage shows some nonlinear sive failure. From C-scan delamination area in
characteristics. Under the impact energy of 6.67 J/mm, Figure 3, it can be seen that the internal delamination
composite laminates of CCF300/Epoxy and CCF800/ area of CCF800/Bismaleimide composite laminates
Bismaleimide both show certain nonlinearity during under the impact energy of 6.67 J/mm is very large, so
compressive loading process, but on the whole the com- the sublayer buckling is more obvious. This also proves
pressive stress–displacement curves of the four kinds of that the delamination damage of impact has an import-
materials all show quite linear characteristics. ant effect on the compressive fracture of composite
The compressive failure morphologies of two kinds laminates.
of CCF800 carbon fiber composites under different The CAI strengths of the four kinds of composite
impact energy are shown in Figure 7. It can be seen laminates are shown in Table 5. For the same kind of
that regardless of the impact energy of 4.45 J/mm or carbon fiber, the CAI strengths of epoxy matrix com-
6.67 J/mm, sublayer buckling both occurs during the posites and bismaleimide matrix composites have obvi-
compressive failure of two kinds of composite lamin- ous differences under the same impact energy. The CAI
ates. For CCF800/Epoxy composite laminates under strength of epoxy matrix composite laminate is signifi-
two kinds of impact energy, the compressive failure cantly greater than that of bismaleimide matrix com-
shows that one face is sublayer buckling, and the posite laminate. So it can be concluded that the damage

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
Han et al. 7

Figure 4. (a) The processed results of layer-tear-off technique of CCF300/Bismaleimide composite laminate under the impact energy
of 4.45 J/mm. (b) The processed results of layer-tear-off technique of CCF300/Bismaleimide composite laminate under the impact
energy of 6.67 J/mm.

tolerance performance of epoxy resin composite system these two stages to avoid errors brought by impact
are better than those of bismaleimide resin composite damage and obtain better simulation result.
system.
For the same kind of resin matrix, the CAI strengths
of CCF300/Epoxy and CCF800/Epoxy composite
FEM model
system differ not quite under the same impact energy. The established FEM model is shown in Figure 8.
The CAI strength of CCF300/Bismaleimide composite Because the analysis of LVI and CAI are conducted
laminates is significantly higher than that of CCF800/ in a same numerical model, so the size of the established
Bismaleimide composite laminates. laminate should contain the boundary conditions of
outer edge. According to the boundary condition of
CAI specimen in Specimen and preparation section,
Computational model the CAI specimen was put in the standard fixture
The impact problem and compression after impact were designed in terms of ASTM D7137.38 Based on the
simulated separately in previous research. In this art- size of the edge fixed end of CAI standard fixture, the
icle, finite element method (FEM) model is used to unsupported panel size of the CAI numerical model is
simulate the impact damage and the failure process of 130 mm  90 mm. According to the size of the base
CAI based on the intralaminar multi-scale failure ana- opening in drop-weight impact testing device in
lysis method and interlaminar cohesive element. The ASTM D7136,37 the minimum unsupported panel size
two stages of LVI and CAI are connected together, of the impact numerical model is 125 mm  75 mm. In
and stable process and damping are inserted between Figure 8, the boundary condition of LVI is specified at

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
8 Journal of Reinforced Plastics and Composites 0(0)

Figure 5. Fiber breakage of monolayer of CCF300/Bismaleimide composite laminate under the impact energy of 6.67 J/mm.

Table 4. The dent depth of impact and delamination area of layer was simplified in the model. Monolayers of the
different material systems. same angle were concentrated in one layer to reduce
the unit-cell number of the thickness direction without
Impact Dent
energy depth Delamination changing the thickness and layer ratio of composite
Material system (J/mm) (mm) area (mm2) laminate. This simplified method was also adopted in
the literature,41,42 and it has been proved to be reason-
CCF300/Epoxy 4.45 0.38 584 able by the experiment in the study by Wei.43
6.67 0.69 740 With CCF300 carbon fiber composite laminates as an
CCF300/Bismaleimide 4.45 0.16 727 example, the original stacking sequence of actual compos-
6.67 0.81 1394 ite layer [45/0/–45/90]4S was simplified as [454/04/–454/
CCF800/Epoxy 4.45 0.14 553 908/–454/04/454], total including seven monolayers and
6.67 0.32 1120 six interlaminar cohesive regions. The intralaminar
CCF800/Bismaleimide 4.45 0.22 1983 region is meshed with eight-node linear brick reduced
6.67 0.52 3885
integration elements (C3D8R) with hourglass control,
and MMF3 is used to determine the intralaminar failure
of fiber and matrix. The interlaminar region is meshed
with eight-node three-dimensional cohesive elements
the edge of green area, and the boundary condition of (COH3D8). The detailed descriptions of failure criterion
CAI is specified at the edge of gray area. and relevant material parameters are shown in sections
In order to reduce the computational cost, several Intralaminar failure criterion (MMF3) and
approaches have been proposed. The global–local Interlaminar failure criterion (cohesive element).
approach is a kind of simulation measure to reduce the Calculation results show that as the mesh is compacted
CPU time effectively, which is shown in literature.39,40 In to a certain extent, the failure load does not depend on the
this article, the stacking sequence of actual composite mesh density any longer. The approximate in-plane mesh

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
Han et al. 9

Figure 6. (a) The CAI stress–displacement curves of four kinds of material under the impact energy of 4.45 J/mm. (b) The CAI
stress–displacement curves of four kinds of material under the impact energy of 6.67 J/mm.

Figure 7. The compressive failure morphologies of CCF800 carbon fiber composites under different impact energy.

size is 2.5 mm  2.5 mm, and this model includes 13,104 the punch were constrained, and only the translational
solid elements and 11,232 cohesive elements totally. degree of freedom in the z direction was released as
Because the stiffness of steel punch is much higher shown in Figure 8a.
than that of composite, so the punch in the numerical Nominal thickness of CCF300 carbon fiber com-
model is established by using analytical rigid body. The posite laminates is 4 mm. For the impact energy of
diameter of spherical punch is 16 mm, and the lumped 4.45 J/mm and 6.67 J/mm, the initial velocities of the
mass 5.36 kg was specified in reference point of the punch are 2.577 m/s and 3.155 m/s, respectively. The
rigid. The moment of inertia is set to be a large value general contact44 is defined between the punch and
to prevent rotation of the punch. The translational and composite laminate, and hard contact is defined
rotational degrees of freedom in the x, y direction and for normal direction, ignoring the effect of tangential
the rotational degree of freedom in the z direction of friction force. In the analysis step of impact, the

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
10 Journal of Reinforced Plastics and Composites 0(0)

displacement in the z direction of the bottom edges in displacement of compressive reference point were rec-
the impact model was constrained, and four corners orded in the process of calculating. Fiber breakage and
were clamped to simulate the fixation of clamp to matrix damage were outputted through the form of a
laminate in experiment. In the analysis step of CAI, user-defined variable.
displacement in the z direction of the left and right
sides was constrained, and the bottom was clamped.
The translational degrees of freedom in the y, z direc-
Intralaminar failure criterion (MMF3)
tion and the rotational degrees of freedom in the three Guagliano and Aliabadi45 have proposed a lot of com-
directions of the upper surface were constrained, and posite damage models in their publication. In this art-
compressive displacement was applied in the x direc- icle, a stress-based multi-scale failure criterion is used to
tion, as shown in Figure 8b. The contact force between predict the intralaminar failure of fiber and matrix,
the punch and the laminate and the reaction force and which is newly proposed by Li Xing.46 This is a three-
dimensional failure criterion based on physical mech-
anism, in which damage initiation and evolution of
Table 5. The CAI strengths of four kinds of composite
laminates. fiber and matrix are determined at the microscopic
level. First, the meso-level stresses in fiber and matrix
Impact CAI are obtained through the square and hexagonal repre-
energy strength sentative volume elements (RVEs) by applying six
Material system (J/mm) (MPa) kinds of macro stress components on them, which are
CCF300/Epoxy 4.45 278.3 shown in Figure 9, and the transformation formula
6.67 247.2 from macro stresses to micro stresses can be written as
CCF300/Bismaleimide 4.45 208.8
 ¼ M  þ A T ð1Þ
6.67 173.5
CCF800/Epoxy 4.45 261.9
where r,  (6  1) are microscopic and macroscopic
6.67 257.2 stress vectors respectively. M (6  1) is the matrix of
CCF800/Bismaleimide 4.45 167.4 mechanical stress amplification factors caused by differ-
6.67 140.3 ent mechanical properties of fiber and matrix, and A
(6  1) is the matrix of thermal stress amplification

Figure 8. The FEM model of low velocity impact and CAI. (a) The boundary condition of low velocity impact. (b) The boundary
condition of CAI.

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
Han et al. 11

Figure 9. The stress distribution of RVEs under different macro stresses and the selection of reference points.

factors caused by their different thermal expansion 8 9 2 3

>  11 > A11
coefficients. T is the amount of temperature change. >
> >
> >
> > 6A 7
Because of the coupling relationship between the macro > 22 >
> 6 21 7
>
> >
> 6 7
and micro vertical, horizontal stress (11 , 22 , 33 ) and <  33 = 6 A31 7
 þ6
6
7T
7 ð2Þ
transverse shear stress (23 ), and the coupling relation- >  12 >
> > 6 0 7
ship between the two longitudinal shear stresses >
> >
> 6 7
>
>  13 >
> 4 0 5
>
> >
>
(12 , 13 ), equation (1) can be written as : ;
 23 A61
8 9 2 3
> 11 > M11 M12 M13 0 0 M16
>
> >
>
>
> 22 >> 6 M21 M22 M23 0 0 M26 7
>
> >
> 6 7 The failure criterion can be expressed as
< >
> = 6 7
33 6 M31 M32 M33 0 0 M36 7
¼6
60
7
> 
> 12 >
>
>
> 6 0 0 M44 M45 0 7 7
1. fiber tensile failure:
>
> > 6 7
> 13 >
> > 40
> 0 0 M54 M55 0 5
>
: >
; f1 4 Xft , f1 4 0 ð3Þ
23 M61 M62 M63 0 0 M66

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
12 Journal of Reinforced Plastics and Composites 0(0)

2. fiber compressive failure: the separation displacement between the top and
bottom faces of the element to the traction vector
f1 5  Xfc , f1  0 ð4Þ acting upon it.
The nominal traction stress vector, t, consists of
three components: tn, ts, and tt, which represents the
3. matrix dilatational failure: normal and two shear tractions, respectively.
The corresponding separations are denoted by dn, ds,
2 and dt. Denoted by T0, the original thickness of the
I1 þ VM 4 I1crit ð5Þ
cohesive element, the nominal strains can be defined as

4. matrix distortional failure: n s t

"n ¼ , "s ¼ , "t ¼ ð9Þ
T0 T0 T0
VM 4 VMcrit ð6Þ
In elastic-linear softening constitutive model, the
I1 ¼ 11 þ 22 þ 33 traction stress vector is defined as follows:
2
I2 ¼ 11 22 þ 22 33 þ 33 11  ð12 2
þ 13 2
þ 23 Þ ð7Þ 8 9 2 38 9
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi < tn >
> = Knn < "n >
> =
VM ¼ I21  3I2 6 7
t ¼ ts ¼ 4 Kss 5 "s ð10Þ
: >
> ; : >
> ;
tt Ktt "t

where f1 is the longitudinal stress of the fiber, and Xft , Kii ði ¼ n, s, tÞ is the stiffness coefficient corresponding
Xfc are the longitudinal tensile and compressive to the three stresses in cohesive element and
strengths of the fiber, respectively. I1 is the volumetric "i ði ¼ n, s, tÞ is the three strains.
stress invariant, i.e., the first stress invariant; VM is the Quadratic nominal stress criterion is used to simu-
deviatoric stress invariant, i.e., Von Mises equivalent late the initial damage, in which damage is assumed to
stress; and  is the correlation factor denoting the influ- initiate when a quadratic interaction function involving
ence of deviatoric stress on dilatational failure of the nominal stress ratios (as defined in the expression
matrix. The critical strength values of composite con- below) reaches a value of 1. This criterion can be rep-
stituents are shown in Table 2. resented as
When the failure criterion is satisfied, the Young’s
 2  2  2
modulus of the fiber and matrix is reduced as follows: htn i ts tt
þ 0 þ 0 ¼1 ð11Þ
t0n ts tt
Edf ðEdm Þ ¼ df ðdm ÞEf ðEm Þ ð8Þ 
tn tn 4 0
ht n i ¼ , t0i ði ¼ n, s, tÞ is the interface
Ef is elastic modulus of the fiber without damage. Edf is 0 tn 5 0
the elastic modulus after damaged. The coefficient df is strength parameters.
the degradation factor of fiber; here, we used a very
little value 0.01 but not zero in order to ensure the Damage evolution can be defined based on the
nonsingular of element stiffness. energy which is dissipated as a result of the damage
Em is elastic modulus of the matrix without damage. process, which is also called the fracture energy. The
Edm is elastic modulus after damaged. The coefficient dm dependence of the fracture energy on the mode mix can
is degradation factor of matrix which is chosen as 0.05. be defined based on a power law fracture criterion. The
The values of degradation factors of fiber and matrix power law criterion states that failure under mixed-
are chosen according to the study by Tsai.47 mode conditions is governed by a power law interaction
FEM analysis of the macroscopic model is con- of the energies required to cause failure in the individ-
ducted in ABAQUS/Explicit. VUSDFLD is used as a ual (normal and two shear) modes. It is given by
kind of user subroutine to realize the whole multi-scale      
failure analysis. Gn Gs Gt
þ c þ c ¼1 ð12Þ
Gcn Gs Gt
Interlaminar failure criterion (cohesive element)
 is a material parameter, which in this article is
Interlaminar delamination is simulated using cohesive selected to be 1.0. Gci ði ¼ n, s, tÞ is the critical fracture
element, the constitutive response of which is defined in energy required to cause failure in the normal, the first,
terms of a linear traction–separation law which relates and the second shear directions.

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
Han et al. 13

When the above condition is satisfied, the mixed- size will make it impossible to calculate. In order to
mode fracture energy Gc is shown as solve this problem, the formula for determining the
interfacial strength of the given mesh size has been
      1=
m1 m2 m3 given by Turon et al.48 The formula is as follows:
Gc ¼ 1= þ c þ c ð13Þ
c
Gn Gs Gt sffiffiffiffiffiffiffiffiffiffiffiffiffiffi

0 9EGc
Because the interfacial parameters and the critical  ¼ ð16Þ
32N0e le
energy release rates are very difficult to obtain and
now there is seldom effective experimental method
which can get these parameters accurately. In this art- where N0e is the desired number of elements in the cohe-
icle, the interfacial parameters and the critical energy sive zone whose value is greater than or equal to 2.
release rates were obtained according to the study by Finally, the interfacial strength is chosen as
Turon et al.48 The interface stiffness should be large  
enough to provide a reasonable stiffness but small
T ¼ min  0 ,  0 ð17Þ
enough to reduce the risk of numerical problems such
as spurious oscillations of the tractions in an element.
The recommended formula of cohesive stiffness coeffi- The critical energy release rates were also obtained
cient in the literature48 is as follows: according to Turon et al.48 and our previous
experiences.36
E3 The interfacial parameters’ values of CCF300/Epoxy
Knn ¼ ð14Þ
t are shown in Table 6 and those of CCF300/
Bismaleimide are shown in Table 7.
where  is a parameter much larger than 1. For values
of  greater than 50, the loss of stiffness due to the
presence of the interface is less than 2%, which is suf-
Analysis steps and the introduction of damping
ficiently accurate for most problems. E3 is the elastic The FEM analysis is conducted in ABAQUS/Explicit.
modulus of the composite layer in the thickness direc- The whole calculation process is divided into three
tion. t is the smallest thickness of composite material stages: (a) Impact stage. The total time is 0.008 s. The
layers connected by cohesion. The determination meth- impact damage process under a given energy is simulated
ods of Kss and Ktt are the same with Knn . in this stage, based on MMF3 to predict the fiber, matrix
One of the drawbacks in the use of cohesive zone damage, and interlaminar delamination, then to reduce
model is that very fine meshes are needed to assure a the stiffness of damaged material and keep the damage
reasonable number of elements in the cohesive zone. state. (b) The stable stage. The total time is 0.01 s.
The length of the cohesive zone lcz is one of the material
and structural properties which is defined as the dis-
tance from the crack tip to the point where the max- Table 6. The interfacial parameters’ values of CCF300/
imum cohesive traction is attained, and it is Epoxy.36,47
proportional to the release rate of the fracture energy
and is inversely proportional to the square of the inter- Knn (MPa) Kss (MPa) Ktt (MPa) t0n (MPa) t0s (MPa) t0t (MPa)
facial strength. The formula is defined as follows: 1E5 1E5 1E5 80 80 80

Gc GnC (J/m2) GsC (J/m2) GtC (J/m2)

lcz ¼ ME ð15Þ
ð 0 Þ2 556 1497 1497

where E is the Young modulus of the material, Gc is the

critical energy release rate,  0 is the maximum inter-
facial strength, and M is a parameter that depends on
Table 7. The interfacial parameters’ values of CCF300/
each cohesive zone model, usually between 0.2 and 1.0. Bismaleimide.36,47
For typical carbon fiber reinforced resin matrix com-
posites, the length of cohesive region is less than 1 mm; Knn (MPa) Kss (MPa) Ktt (MPa) t0n (MPa) t0s (MPa) t0t (MPa)
usually at least two elements should be guaranteed in 1E5 1E5 1E5 60 60 60
the direction of crack propagation for the requirement
2 2 2
of element division at the cohesive region. So the mesh GnC (J/m ) GsC (J/m ) GtC (J/m )
size of cohesive element must be less than 0.5 mm. But
280 494 494
in the analysis of large structures, the element of this

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
14 Journal of Reinforced Plastics and Composites 0(0)

Damping and mass amplification are introduced to

Numerical result and discussion
quickly stop vibration of laminate and to restore to the The numerical simulation is concentrated on composite
state near stationary. (c) CAI stage. The total time is 1 s. material system of CCF300/Epoxy and CCF300/
Quasi-static calculation method with mass amplification Bismaleimide, with two kinds of impact energy for
is used to simulate the failure process of impact- each kind of material system, 4.45 J/mm and
damaged laminate under compressive load. 6.67 J/mm. The obtained impact force–time curves by
In this article, Rayleigh damping is used to reduce numerical calculation in comparison with experimental
the dynamic oscillation of the model, which is shown as results are shown in Figure 10. For CCF300/Epoxy
composite system, the duration of impact force curves
C ¼ M þ K ð18Þ calculated by two kinds of impact energy are slightly
shorter compared to the test result which is probably
where C is the matrix of Rayleigh damping, M and K resulted from the differences of stiffness and boundary
are mass matrix and stiffness matrix of the element, conditions between specimen and finite element model,
respectively, and  and  are proportional damping while the maximum contact forces in the test results
coefficient of mass and proportional damping coeffi- and calculation results agree well. For CCF300/
cient of stiffness, respectively. From the numerical Bismaleimide composite system, the maximum contact
results with several groups of parameters, the values force and the impact duration of test results and calcu-
 ¼ 3000 and  ¼ 1 e8 had been chosen with little influ- lation results are all in good agreement. Under the
ence on the analysis results and computational cost. impact energy of 6.67 J/mm, fiber fracture occurs in

Figure 10. The comparison of calculated impact contact force curves with experimental results. (a) CCF300/Epoxy 4.45 J/mm;
(b) CCF300/Epoxy 6.67 J/mm; (c) CCF300/Bismaleimide 4.45 J/mm; and (d) CCF300/Bismaleimide 6.67 J/mm.

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
Han et al. 15

laminate, so stiffness decrease appears in the contact M44,m and M54,m are the components of the matrix
force curve of experiment when the load is large. mechanical stress amplification coefficient matrix,
Fiber fracture also occurs in numerical model under i ¼ 1, 2, 3 . . . represents the reference points in the
the impact energy of 6.67 J/mm, but there is no signifi- matrix phase of the square and hexagonal RVEs,
cant decrease in contact force curve, resulting in a cer- ij,f and ij,m are the microscopic stress of fiber and
tain error. matrix along all directions in RVE separately, (i) is the
reference point in square RVE and hexagonal RVE,
The monolayer stress state of the impact model was
KTf is used to characterize fiber tensile failure; KCf is
analyzed to study the most likely failure mode before
used to characterize fiber compressive failure; KTm is
various damages occurred. According to the MMF3,
used to characterize matrix tensile failure; KCm is used
the micro failure mode characterizations of fiber and
to characterize matrix compressive failure; and KSm is
matrix were defined as follows:
used to characterize matrix shear failure. As the value is
! ! closer to 1, the corresponding microscopic failure mode
ðiÞ ðiÞ is easier to happen.
11,f 11,f
KTf ¼ max , KCf ¼ max  With the impact problem of CCF300/Epoxy com-
Tf Cf
! posites as an example, time point is selected as
ðiÞ ðiÞ
22,m þ 33,m 0.28 ms. At this time point, all kinds of damage have
KTm ¼ max , not occurred, and the characterization of microscopic
Tm
! failure mode of each monolayer is shown in Figure 11.
ðiÞ ðiÞ ð19Þ
22,m þ 33,m As illustrated in the figure, fiber tensile failure most
KCm ¼ max 
Cm likely occurs in the monolayer away from the punch,
! and fiber compressive failure most likely occurs near
 2ðiÞ þ 13,m
2ðiÞ 2ðiÞ
þ 23,m ðiÞ ðiÞ
 22 33 the contact point of the laminate and punch. Matrix
KSm ¼ max 12,m 2
Sm tensile failure is easy to appear in each monolayer,
whose damage area is larger in the monolayer away
i ¼ 1, 2, . . .
from the punch, and matrix compressive and shear fail-
ure most likely occur in the monolayer near the punch,
Tf , Cf , Tm , Cm , and Sm are fiber tensile strength, fiber which is concentrated in the contact point of the punch
compressive strength, matrix tensile strength, matrix and laminate.
compressive, and shear strength, respectively whose Fiber damage and matrix damage in each monolayer
values have been shown in Table 2. In the study by of CCF300/Epoxy and CCF300/Bismaleimide compos-
Xing,36 a series of experiments and multi-scale failure ites under different impact energy predicted by MMF3
analysis method-MMF3 are used to obtain the micro- are shown in Figure 12. For the two kinds of composite
scopic strength parameters of composite. The tensile systems, matrix damage radius increases when improv-
and compressive strengths of fiber (Tf and Cf ) can be ing the impact energy, and fiber damages appear in
calculated by 0 longitudinal tensile and compressive monolayers of the two kinds of material under the
strength (XT and XC ) of unidirectional composite. impact energy of 6.67 J/mm (no fiber damages for
The tensile and compressive strengths of matrix (Tm the two kinds of material under the impact energy of
and Cm ) can be calculated by 90 transverse tensile 4.45 J/mm). Because the microscopic strength value
and compressive strength (YT and YC ) of unidirectional of bismaleimide resin is lower than that of epoxy
composite. The shear strength of matrix (Sm ) can be resin, under the same impact energy, matrix damage
calculated by shear strength (S) of unidirectional area of CCF300/Bismaleimide composite laminate is
composite. significantly larger than that of CCF300/Epoxy com-
posites, which also coincides with the experimental
The microscopic shear strength of matrix Sm can be
results.
obtained by the shear strength of single plate S (for
The typical damage morphology obtained by numer-
CCF300 composite laminate, the single layer thickness
ical simulation of compression after impact is shown in
is 0.125 mm, and for CCF800 composite laminate,
Figure 13. Fold occurs in the middle of laminate per-
the single layer thickness is 0.19 mm). The expression
pendicular to the loading direction, consistent with the
of Sm is
experimental observations from Figure 7. It can also be
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi seen from the profile that each monolayer is separated
 2  2
during the compressive failure and sublayer buckling
ði Þ ðiÞ
Sm ¼ max M44,m S þ M54,m S , i ¼ 1, 2, 3 . . . appears, which also proves that delamination damage
plays a key role during the failure of compression after
ð20Þ impact.

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
16 Journal of Reinforced Plastics and Composites 0(0)

Figure 11. Numerical simulation of microscopic failure mode characterization in each monolayer of CCF300/Epoxy composites
under impact load.

The simulation results of CAI failure morphology

for the two material systems are shown in Figure 14,
Conclusions
and the comparison of CAI failure strengths between Four kinds of different fiber, matrix matching compos-
calculation result and experimental result is shown in ite systems (CCF300/Epoxy, CCF300/Bismaleimide,
Table 8. In brief, the numerical simulation method CCF800/Epoxy, and CCF800/Bismaleimide) were
established in this article can effectively predict the designed in this article to conduct the impact
damage process and the ultimate failure strength of and CAI experiments under the impact energy of
composite impact and compression after impact and 4.45 J/mm and 6.67 J/mm separately and to compare
also proves the feasibility of the multi-scale failure ana- the impact contact force–time curve, dent depth,
lysis method. C-scan delamination area, and other characteristics.

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
Han et al. 17

Figure 12. Fiber and matrix damage of composite laminates under different impact energy. (a) CCF300/Epoxy and (b) CCF300/
Bismaleimide.

Figure 13. The damage morphology of compression after impact obtained by numerical simulation.

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
18 Journal of Reinforced Plastics and Composites 0(0)

fiber, the delamination area of bismaleimide matrix

composites is larger than that of epoxy matrix com-
posites under the same impact energy, and the dent
depth is also slightly larger, which shows that the
interlaminar properties of bismaleimide matrix are
significantly weaker than that of epoxy matrix.
Under the impact energy of 6.67 J/mm, fiber fracture
both occurs in CCF300/Epoxy and CCF300/
Figure 14. CAI failure morphologies of CCF300/Epoxy and
Bismaleimide composite laminates.
CCF300/Bismaleimide composite laminates.
4. For the same kind of carbon fiber, CAI strength of
bismaleimide matrix composites is significantly
lower than that of epoxy matrix composites under
Table 8. The comparison of CAI failure strengths between the same impact energy. So it can be concluded that
calculation result and experimental result. the damage tolerance performance of epoxy resin
composite system are better than that of bismalei-
CCF300/ CCF300/
Material system Epoxy Bismaleimide mide resin composite system. For the same kind of
matrix, the CAI strengths of CCF300/Epoxy and
Impact energy (J/mm) 4.45 6.67 4.45 6.67 CCF800/Epoxy composite system under the same
Experimental result (MPa) 278.3 247.2 208.8 173.5 impact energy differ not quite. The CAI strength of
Calculation result (MPa) 261.1 237.2 224.6 199.5 CCF300/Bismaleimide composite laminate is signifi-
Error (%) 6.18% 4.05% 7.57% 14.99% cantly higher than that of CCF800/Bismaleimide
composite laminate. So it can be concluded that
the damage tolerance performance of CCF300
Layer-tear-off technique was conducted for CCF300/ carbon fiber composites are better than that of
Bismaleimide composites, and CAI stress–displacement CCF800 carbon fiber composites.
curve, failure morphology, and residual strength were 5. Through comparison between calculations and
further compared. Then finite element analysis model experimental results, it proves that the proposed
of composite laminate was established and the whole multi-scale failure analysis method can effectively
impact-CAI damage process was analyzed based on the simulate the impact and CAI damage process of
MMF3 and cohesive element. The following conclu- composite laminates and can be used to analyze
sions can be obtained: the failure mechanisms of laminates in-depth.

1. For all kinds of composite materials, larger impact

Declaration of Conflicting Interests
energy can lead to higher maximum impact contact
force and shorter impact duration, and more serious The author(s) declared no conflict of interest with respect to
interlaminar damage and larger dent depth. the research, authorship, and/or publication of this article.
2. Under the same impact energy, the impact contact
forces of two kinds of CCF800 carbon fiber compos- Funding
ites are significantly higher than that of two kinds of The author(s) disclosed receipt of the following financial sup-
CCF300 carbon fiber composite materials, and the port for the research, authorship, and/or publication of this
former impact duration is shorter. This is because article: The financial support of the National Basic Research
Program (973) of China (under Grant No. 2010CB631103) is
the longitudinal modulus of CCF800 carbon fiber
acknowledged.
is higher than that of CCF300 carbon fiber. For
composites with same fiber and different matrix,
References
the impact contact force curves have no clear
differences. 1. Lopes CS, Seresta O, Coquet Y, et al. Low-velocity
3. For epoxy matrix and bismaleimide matrix, the impact damage on dispersed stacking sequence laminates.
Part I: experiments. Compos Sci Technol 2009; 69(7–8):
delamination area of CCF800 carbon fiber compos-
926–936.
ites is larger than that of CCF300 carbon fiber com- 2. Ghelli D and Minak G. Low velocity impact and compres-
posites under the same impact energy, and the dent sion after impact tests on thin carbon/epoxy laminates.
depth is shallower. This is due to the higher inter- Compos Part B-Eng 2011; 42(7): 2067–2079.
laminar stress resulted from the higher longitudinal 3. Saito H, Morita M, Kawabe K, et al. Effect of ply-
modulus of CCF800 carbon fiber under the same thickness on impact damage morphology in CFRP lamin-
impact energy. For CCF300 and CCF800 carbon ates. J Reinf Plast Compos 2011; 30(13): 1097–1106.

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
Han et al. 19

4. Mathivanan NR and Jerald J. Experimental investigation 21. Dang TD and Hallett SR. A numerical study on impact
of low-velocity impact characteristics of woven glass fiber and compression after impact behaviour of variable angle
epoxy matrix composite laminates of EP3 grade. Mater tow laminates. Compos Struct 2013; 96: 194–206.
Design 2010; 31(9): 4553–4560. 22. Yan H, Oskay C, Krishnan A, et al. Compression-after-
5. Evci C and Gülgec M. An experimental investigation on impact response of woven fiber-reinforced composites.
the impact response of composite materials. Int J Impact Compos Sci Technol 2010; 70(14): 2128–2136.
Eng 2012; 43: 40–51. 23. Oskay C and Fish J. Eigendeformation-based reduced
6. Pattofatto S, Zhao H and Tsitsiris H. Influence of the order homogenization for failure analysis of heteroge-
impact velocity on the perforation resistance of a glass- neous materials. Comp Meth Appl Mech Eng 2007;
reinforced polypropylene material. J Reinf Plast Compos 196(7): 1216–1243.
2011; 30(13): 1107–1114. 24. Crouch R and Oskay C. Symmetric meso-mechanical
7. Hassan A, Hassan AA and Rafiq MIM. Impact proper- model for failure analysis of heterogeneous materials.
ties of injection molded glass fiber/polyamide-6 compos- Int J Mult Comp Eng 2010; 8(5): 447–461.
ites: effect of testing parameters. J Reinf Plast Compos 25. Vieille B, Casado VM and Bouvet C. Influence of matrix
2011; 30(10): 889–898. toughness and ductility on the compression after impact
8. Fan J, Cantwell WJ and Guan ZW. The low-velocity behavior of woven-ply thermoplastic and thermosetting
impact response of fiber-metal laminates. J Reinf Plast composites: a comparative study. Compos Struct 2014;
Compos 2011; 30(1): 26–35. 110: 207–218.
9. Tsamasphyros GS and Bikakis GS. Dynamic response of 26. Lopes CS, Camanho PP, Gürdal Z, et al. Low velocity
circular GLARE fiber-metal laminates subjected to low impact damage on dispersed stacking sequence laminates.
velocity impact. J Reinf Plast Compos 2011; 30(11): Part II: numerical simulations. Compos Sci Technol 2009;
978–987. 69(7–8): 937–947.
10. Ma J, Yan Y, Liu YJ, et al. Compression strength of 27. Donadon MV, Iannucci L, Falzon BG, et al. A progres-
stitched foam-core sandwich composites with impact sive failure model for composite laminates subjected
induced damage. J Reinf Plast Compos 2012; 31(18): to low velocity impact damage. Comput Struct 2008;
1236–1246. 86(11–12): 1232–1252.
11. Abrate S. Impact on composite structures. England: 28. Faggiani A and Falzon BG. Predicting low-velocity
Cambridge University Press, 1998. impact damage on a stiffened composite panel. Compos
12. Olsson R. Impact response of orthotropic composite Part A-Appl Sci Manuf 2010; 41(6): 737–749.
plates predicted from a one-parameter differential equa- 29. Yokoyama NO, Donadon MV and de Almeida SFM.
tion. AIAA J 1992; 30(6): 1587–1596. A numerical study on the impact resistance of composite
13. Olsson R. Analytical prediction of large mass impact shells using an energy based failure model. Compos Struct
damage in composite laminates. Compos Part A-Appl 2010; 93(1): 142–152.
Sci Manuf 2001; 32(9): 1207–1215. 30. Raimondo L, Iannucci L, Robinson P, et al. A pro-
14. Olsson R. Engineering method for prediction of impact gressive failure model for mesh-size-independent FE ana-
response and damage in sandwich panels. J Sandw Struct lysis of composite laminates subject to low-velocity
Mater 2002; 4(1): 3–29. impact damage. Compos Sci Technol 2012; 72(5):
15. Olsson R. Closed form prediction of peak load and 624–632.
delamination onset under small mass impact. Compos 31. Batra RC, Gopinath G and Zheng JQ. Damage and fail-
Struct 2003; 59(3): 341–349. ure in low energy impact of fiber-reinforced polymeric
16. Olsson R, Donadon MV and Falzon BG. Delamination composite laminates. Compos Struct 2012; 94(2):
threshold load for dynamic impact on plates. Int J Solids 540–547.
Struct 2006; 43(10): 3124–3141. 32. Hashin Z and Rotem A. A fatigue failure criterion for
17. Chen P, Shen Z, Xiong J, et al. Failure mechanisms of fiber reinforced materials. J Compos Mater 1973; 7:
laminated composites subjected to static indentation. 448–464.
Compos Struct 2006; 75(1–4): 489–495. 33. Hashin Z. Failure criteria for unidirectional fiber com-
18. Belingardi G and Vadori R. Influence of the laminate posites. J Appl Mech 1980; 47(2): 329–334.
thickness in low velocity impact behavior of carbon- 34. Li H, Dai F and Du SY. Numerical and experimental
epoxy composite material plate. Compos Struct 2003; study on morphing bi-stable composite laminates actu-
61(1–2): 27–38. ated by a heating method. Compos Sci Technol 2012;
19. Caprino G, Lopresto V, Scarpani C, et al. Influence of 72(14): 1767–1773.
material thickness on the response of carbon-epoxy 35. Chen X. Handbook of polymer matrix composites. Beijing:
panels to low velocity impact. Compos Sci Technol Chemical Industry Press, 2004, pp.245–279.
1999; 59(15): 2279–2286. 36. Xing L. Research on multi-scale failure analysis method
20. He W, Guan Z, Li X, et al. Prediction of permanent and typical laminates’ failure law of composite materials.
indentation due to impact on laminated composites Beijing: Beihang University, 2014.
based on an elasto-plastic model incorporating fiber fail- 37. ASTM D7136/D7136M-12. Standard test method for
ure. Compos Struct 2013; 96: 232–242. measuring the damage resistance of a fiber-reinforced

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016
20 Journal of Reinforced Plastics and Composites 0(0)

polymer matrix composite to a drop-weight impact event. 43. Wei H. Study on issues concerning impact damage per-
West Conshohocken USA: ASTM International, 2012. formances of composite structures. Beijing: Beihang
38. ASTM D7137/D7137M-12. Standard test method for University, 2014.
compressive residual strength properties of damaged poly- 44. Johnson KL. Contact mechanics. England: Cambridge
mer matrix composite plates. West Conshohocken, USA: University Press, 1985.
ASTM International, 2012. 45. Guagliano M and Aliabadi MH. Fracture and damage of
39. Caputo F, De Luca A and Sepe R. Numerical study of composites. England: WIT Press, 2005.
the structural behaviour of impacted composite laminates 46. Li X, Guan ZD, Li ZS, et al. A new stress-based multi-
subjected to compression load. Compos Part B-Eng 2015; scale failure criterion of composites and its validation in
79: 456–465. open hole tension tests. Chin J Aeronaut Epub ahead of
40. Riccio A, De Luca A, Di Felice G, et al. Modelling the print 2014. DOI: 10.1016/j.cja.2014.03.002.
simulation of impact induced damage onset and evolu- 47. Tsai SW. Strength & life of composites. Stanford, CA:
tion in composites. Compos Part B-Eng 2014; 66: Composites Design Group, Department of Aeronautics
340–347. & Astronautics, Stanford University, 2008, pp.6-29–6-51.
41. Huang CH and Lee YJ. Experiments and simulation of 48. Turon A, Davila CG, Camanho PP, et al. An engineering
the static contact crush of composite laminated plates. solution for mesh size effects in the simulation of delam-
Compos Struct 2003; 61: 265–270. ination using cohesive zone models. Eng Fract Mech
42. Nettles AT, Douglas MJ and Estes EE. Scaling effects in 2007; 74(10): 1665–1682.
carbon/epoxy laminates under transverse quasi-static load-
ing. NASA/TM-1999-209103. Washington: NASA, 1999.

Downloaded from jrp.sagepub.com at UNIV CALIFORNIA SAN DIEGO on March 13, 2016