Professional Documents
Culture Documents
PII: S0263-8223(18)33401-9
DOI: https://doi.org/10.1016/j.compstruct.2018.12.024
Reference: COST 10487
Please cite this article as: Cao, M., Zhao, Y., Gu, B.H., Sun, B.Z., Tay, T.E., Progressive failure of Inter-woven
carbon-Dyneema fabric reinforced hybrid composites, Composite Structures (2018), doi: https://doi.org/10.1016/
j.compstruct.2018.12.024
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Progressive failure of Inter-woven carbon-Dyneema fabric reinforced hybrid
composites
Cao M.a, Zhao Y.*,b., Gu B.H.a, Sun B.Z.a, Tay T.E.b
* Corresponding author (Email: mpezhy@nus.edu.sg)
Affiliations
a
College of Textiles, Donghua University, Shanghai, China, 201620
b
Department of Mechanical Engineering, National University of Singapore,
Singapore 117576
Abstract
This study investigates the in-plane mechanical behavior of a twill 2,2
carbon-Dyneema fabric reinforced hybrid composite. A finite element (FE) model of a
meso-scale representative volume element (RVE) is developed for simulating the
in-plane behavior of the composite. The development of damage in both carbon and
Dyneema yarns is modeled through progressive damage models with linear softening
laws and the non-linear response of the Dyneema yarn is experimentally determined
and described by the Ramberg-Osgood equation. The epoxy resin is regarded as an
elasto-plastic material. The damage development in different constituents of the
composite is analyzed and correlated with experimental observations. In addition, the
role of RVE size in modeling the behavior of the hybrid composite is investigated. It
is found that a minimum of size of about five basic units of RVEs is necessary to
achieve acceptable predictive results due to shear lag effects.
1
1.Introduction
Carbon fiber reinforced plastics (CFRP) are gaining market share in various industries
such as aerospace, wind energy, and automotive applications because of their high
specific stiffness and strength. However, the high stiffness and strength of CFRPs
come at the expense of their limited toughness due to the low failure strain of carbon
fibers. Fiber hybridization is a promising strategy to toughen composite materials. By
combining high failure strain fibers such as polymer fibers with carbon fibers in a
hybrid system, pseudo-ductility can be introduced in the composite constitutive
behavior [1-4]. Ultra-high molecular weight polyethylene (UHMWPE) fibers have
several advantages including low density (0.97 g/cm3), reasonably high stiffness and
strength and biocompatibility[5, 6]. However, limitations of UHMWPE fibers include
their relatively poor compressive properties, poor adhesion to resin matrices, low heat
resistance and low creep resistance[7]. Therefore, hybridizing carbon fibers with
UHMWPE fibers can offer a better balance in tensile strength, compressive strength
and impact resistance.
Several studies have probed the mechanical behavior of carbon-UHMWPE hybrid
composites. Peijs et al. [8-10] investigated the tensile, fatigue and impact behavior of
such hybrid composites. The results showed positive hybrid effects through the
hybridization of carbon fibers and UHMWPE fibers (such as higher failure strain,
better structural integrity, better energy absorption ability, etc.). Li et al.[7] reported
improvements in compressive and flexural properties by incorporating a moderate
amount of carbon fibers into a UHMPEF SK66/epoxy 618 composite. Some
studies[11, 12] presented the ballistic properties of the hybrid composite, which
showed potential in bulletproof applications. However, most of these earlier studies
focus on the properties of the interlaminar hybrids – i.e. laminates made by stacking
plies of different fiber types. Research on intralaminar hybrids – i.e. laminates of plies
with interwoven yarns of different fiber types within each ply – is still relatively rare
[13, 14]. Recently, carbon-Dyneema intralaminar hybrid composites have received
considerable attention in applications such as sports equipment, bicycles frames and
2
automotive components. Carbon fiber is strong, light, and stiff, but is also relatively
brittle and poor vibration characteristics. Dyneema fibers, on the other hand, can
effectively damp vibration and resist shattering. Their hybrids could take advantage of
the stiffness of carbon fibers as well as the vibration-absorption and ductility of
Dyneema fibers.
Modeling hybrid fabric reinforced composites is challenging because of multiple
failure modes, their interactions, interlacing meso-structures and very different fiber
failure mechanics. The analysis of an RVE through FE is often a preferred route to
investigate the mechanical behavior of textile composites, as the meso-scale model is
able to account for the geometry of the yarns and different damage mechanisms of the
constituents [15-20]. In this paper, the tensile and shear behavior of the
carbon-Dyneema intraply hybrid composite is investigated experimentally and
numerically. A meso-scale RVE model is established to provide insights into the local
mechanical responses. The yarns are modeled as transversely isotropic with
progressive damage laws. The epoxy resin pocket is modeled as an elasto-plastic
material with isotropic damage. The damage model is implemented into a
user-defined material model (VUMAT) in the FE code Abaqus/Explicit. The
predictions of the proposed model are compared with experimental test results on both
macroscopic stress-strain responses and local strain distributions. The damage
development in different constituents of the composite is modeled and analyzed with
experimental observations. The damage mechanism is compared between the RVE
models of different sizes. The shear lag effect on modeling the damage evolution of
the hybrid composite is also discussed.
3
was infused through vacuum assisted resin transfer molding (VARTM). The laminates
were cured for 24 hours at room temperature and cut into specimen size by water jet.
The specimens were 200mm in length, 20mm and 25mm in width for tensile and
in-plane shear tests respectively. The specimen surfaces were speckled for DIC
analysis: they were first coated with a base layer of white paint and a black stochastic
pattern was subsequently applied onto the surface by lightly spraying a mist of flat
black paint.
Both tensile and shear tests were conducted using an INSTRON 8501 universal
testing machine with a 100 kN load cell. The specimens were loaded at 2mm/min
until failure. The gauge length was about 100mm for both tensile and shears tests.
Sand papers were used at the clamping region to prevent slippage of the grips. The
details of the tested specimens are listed in Table.1. The tensile test was conducted for
the hybrid composite and the pure Dyneema specimen following the ASTM
D3039-08 standard[21]. Cyclic tensile tests were also conducted on the pure Dyneema
specimens to characterize the nonlinear response of the Dyneema composite. The
shear specimens were fabricated at an angle of 45o relative to the warp direction and
loaded at 0o as illustrated in Fig.2. Two high elongation strain gauges were placed at
0o and 90o on the specimen surface. A digital image correlation system from the
4
Correlated Solutions Inc (CSI) was utilized to obtain full field strain maps on the
surface of the specimen.
Pure Dyneema 4
Carbon-Dyneema 3
Carbon-Dyneema 4
5
3. Development of meso-scale FE model
Parameter Value
Yarn width W (mm) 1.5
Yarn thickness H (mm) 0.135
Gap between parallel yarns G (mm) 0.5
RVE width and length D (mm) 8
RVE thickness T (mm) 0.3
Yarn volume fraction Vyarn 0.53
6
3.2 Constitutive modelling
3.2.1 Effective properties of yarns
The bridging model of Huang[23] is employed to determine the effective stiffness
properties of yarns. In this paper, the quantity with suffix (superscript or subscript) f
or m indicates the fiber or the matrix phase. The homogenized compliance matrix for
the yarn is defined as:
(1)
where [Sf] and [Sm] refer to the compliance matrices of the fiber and resin,
respectively, and Vf and Vm are the volume fractions of the fiber and resin matrix in
yarns, respectively. [A] is a bridging matrix and [I] is a unit matrix. The fiber volume
fraction in each yarn is assumed to be 75%. The elastic properties of carbon and
Dyneema fibers as well as epoxy resin are listed in Table 3, and the predicted
mechanical properties of carbon and Dyneema yarns from Eq. (1) are given in Table 4.
In this study, the strength of the carbon and Dyneema yarns is assumed to be the same
as the strength obtained from unidirectional laminate tests and Table 5 lists the
strength parameters of the components of the RVE.
0.26[17] 0.2[26]
ν13
0.44[17] 0.2[26]
ν23
Table.4 Material properties of carbon and Dyneema yarns calculated from bridging model
Properties Carbon yarns Dyneema yarns
E11(GPa) 169.49 87.72
7
E22= E33 (GPa) 10.18 3.21
G12= G13 (GPa) 5.32 2.47
G23(GPa) 3.78 0.60
ν12=ν13 0.28 0.2
b
The transverse shear toughness of Dyneema yarn is assumed to be same as carbon yarn
(2)
8
tensile results of Dyneema composites and the initial modulus of Dyneema yarns are
listed in Table.6.
9
400
350
secant modulus
300
Stress (MPa)
250
200
150
100
50
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Strain (%)
(3)
where and are respectively the yarn tensile and compressive strength in the
fiber direction. The Tsai–Wu[40] failure criterion is used to determine the initiation of
matrix damage for both carbon and Dyneema yarns:
(4)
(5)
(6)
(7)
where and are the tensile and compressive strength in the transverse direction,
and is the in-plane shear strength. <> is the Macaulay bracket, the matrix failure
due to transverse compressive stress is not considered in this paper.
10
Once damage initiates, degradation factors are used to control
the stiffness degradation process. Here and are the degradation factors of
the fiber tensile and the compressive failure respectively. is the degradation
factor for the matrix dominated failure. A linear softening law using the effective
stress and displacement (Table.7) based on an energy criterion is adopted to describe
the damage progression of carbon yarns:
(8)
Considering the plastic deformation of Dyneema yarns, the degradation factors are
given by:
(9)
(10)
satisfied, is the equivalent plastic displacement at the damage initial point and
(11)
Here, denotes the initiation damage equivalence stress. In this paper, the matrix
Fiber tension
Fiber compression
11
Matrix tension
The characteristic element length is defined as cube root of solid element volume[42].
In this study, the degradation factor of the carbon yarns is defined as:
(12)
(14)
Furthermore, fiber damage sometimes occurs together with or after matrix failure. The
combined effect is described by the following degradation schemes for carbon and
Dyneema yarns:
(15)
secant modulus (as mentioned in Fig.4) at the point of unloading while and
are the degradation factors for matrix and fiber dominated damage, respectively.
12
3.2.3 Constitutive and damage model for resin
In this work, the resin surrounding the yarns is considered as an elastic-plastic
material after Melro[43] . The paraboloidal yield criterion proposed by [44] is adopted
to control the non-linear behavior of the resin material. The criterion considers
different yield strengths in the tensile and compressive directions, as well as the
pressure dependence of the material. The yield surface is defined by:
(16)
where and denote the yield strengths in compressive and tensile direction,
and I1 = trσ is the first invariant of the stress tensor. A non-associative flow rule is
used, allowing for volumetric deformation in plasticity. The hardening laws are
defined following the experimental results from Fiedler et al[31]. The model is
implemented using a return mapping algorithm with an elastic predictor/plastic
corrector strategy. The isotropic damage of the matrix is modeled. The damage
activation function is defined as:
(17)
where is an internal variable controlled by the damage evolution law. The loading
function is defined by:
(18)
where Xc and Xt are the compressive and tensile strengths of the material while the
(19)
where Am is a constant calculated for each element, which can be obtained from the
regulated dissipated energy by the element’s characteristic length. The details of
implementing damage evolution law can be found in [43].
13
3.3 Finite element model
The model of the RVE (Fig.6) is meshed with linear tetrahedral elements C3D4 due to
the complex geometry. The mesh seed size is 0.2mm. The total number of elements in
the unit cell is 109070 (72726 for the matrix and 36344 for yarns). Note that the
material orientation of the elements in yarn structures is specified following fiber
directions.
Periodic boundary conditions are applied to ensure maintenance of the response of the
RVE as a repeatable element. Periodicity[45] is considered in two dimensions due to
the planar geometry of the twill weave:
(20)
coordinates shared by the corresponding points on the pair of faces (Fig.7). In this
14
Fig.7 Illustration of periodic boundary conditions
In this paper, we opt to use an explicit solution method in order to avoid possible
convergence issues. The application of load steps should be as gradual and smooth as
possible to avoid spurious stress waves. This is achieved through the Abaqus built-in
displacement step amplitude curve where the first and second derivatives are
continuous. No contact algorithm was used in the model. A loading rate of 0.24m/s is
adopted for this study and the sensitivity of the FE results to the loading rate is also
investigated (Appendix A). Five RVE models (Fig.8) were made by the assembly of
different numbers of unit cells, i.e., RVE1 (single unit), RVE3 (three units), RVE5
(five units), RVE7 (seven units) and RVE5×2 (five units by two rows). They are
chosen to investigate the effect of RVE size on the stress-strain response under tension.
For shear, the RVE-Shear is obtained through 45 degrees rotation of the tensile model
(Fig.8).
15
Fig.8 Different RVE models used in this paper
16
Fig.9 Influence of interlacing structure on strain field distribution:(a) DIC tensile strain contours and (b)
corresponding interlacing structure; (c) DIC shear strain contours and (d) corresponding interlacing structure
The strain distribution obtained from the finite element simulation is compared with
that from the DIC results (Fig.10). The strain profile along a chosen path (dotted line
in Fig.10) obtained from the RVE models are quantitatively compared to experimental
curves measured by DIC. The strain value against the distance along the path (dotted
line in Fig.10) is plotted. The strain distributions from the numerical simulation and
DIC results are very similar. However, the strain localizations found in the DIC
images are not as obvious as those in the FE simulation. This is mainly because the
features on the thin resin pockets are too fine to be accurately captured even with a
high resolution DIC system.
17
(a) tensile
0.016
0.006 FEM
0.004
0.002
0.000 carbon
0 2 4 6 8 10
Distance (mm)
Dyneema
(b) shear
0.0055
0.0050 DIC
FEM
0.0045
DIC
0.0040
0.0035
0.0030
FEM
0.0025
0 2 4 6 8 10 12
Loading
Distance (mm)
direction
-0.0015
DIC
-0.0020 FEM
DIC
-0.0025
-0.0030
-0.0035
FEM
-0.0040
-0.0045
0 2 4 6 8 10 12
Distance (mm)
carbon
Dyneema
Fig.10 Correlation between FEM and DIC results: (a)tensile at global strain of 0.86%; (b)shear at global strain of
0.75%
19
(a) 450 (b) 35
1unit
3units
400 5units loading 30
5units by 2
350 7units
exp3
250 20
200 exp1
15
150 exp2
10 exp3
100 exp4
FEM
50 5
0 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.0 0.5 1.0 1.5 2.0 2.5
Strain (%) Shear strain (%)
Fig.11 Stress-strain response comparison between experiment and finite element models: (a)tensile (b)shear
20
21
Fig.12 Damage mechanisms and evolution process at six different global strains for RVE1: (a)Stress-strain curve;
Local fiber compressive damage occurs in the fill Dyneema yarns at a strain level of
0.63% (Fig.12(b)). This is probably due to the constraining effects of the contacted
warp yarns at the interlacing area and the Dyneema yarns are vulnerable to
compressive loading. Matrix cracking (transverse tension damage) is also observed at
the interlacing area (Fig.12(c)). Despite the early damage in Dyneema yarns, the
model still exhibits a good load bearing capacity as indicated in the stress-strain
curves. At a global strain level of 1.25%, the carbon yarns approach complete failure,
leading to the first load drop off in the load-displacement curve. It is also found that
matrix cracking damage propagates to all the fill yarns indicating that fill yarns are
undergoing continuous damage during the tension test (Fig.12(c)). The breakage of
carbon yarns causes total transverse damage in the contacted Dyneema fill yarns
which is observed in experiments (Fig.12(c)). Furthermore, the resin pocket damage
also initiates at the carbon fiber fracture zone (Fig.12(d)). After the first load drop
caused by the carbon yarns breakage, the global stress gradually increases due to the
tensile load taken by the Dyneema yarns. In this RVE1 model, the Dyneema yarns
break at the global strain of 2.80% which is much higher than the experimental values.
Consider the limitations of the RVE1 model on predicting fracture process, several
22
RVE models in different sizes are established to compare the failure behavior of the
hybrid composites.
180
2.4
150
120
2.1
90
60 1.8
30
0 1.5
RVE1 RVE3 RVE5 RVE5×2 RVE7
RVE size
Fig.13 Effect of RVE size on the lowest point and total failure strain
Fig.14(a) shows the stress distribution of a fractured carbon yarn and a nearby
Dyneema yarn in the RVE1, RVE3, RVE5 and RVE7 models at the strain of 1.85%,
just before the Dyneema yarn breakage. Due to the shear lag effect, the broken carbon
yarn gradually recovers its load carrying capacity away from the fracture. Meanwhile,
the fracture of the carbon yarn causes stress concentration in nearby Dyneema yarns
which locally takes over the additional load. It is found that the stress concentration is
more pronounced with the increase of the model length due to shear lag effects.
Therefore, the global failure strain reduces with the increase of size of the RVE. The
stress distribution of a fracture carbon yarn in different RVEs is plotted Fig.14(b). The
recovery stress converges at the stress level of 1600 MPa at the distance
approximately 20mm away from the fractured end in the RVE5 and RVE7 models. It
23
can be inferred that the critical effective length of the carbon yarns is around 20mm in
this model.
RVE1
RVE3
S11 (MPa)
RVE7
1600 1600
1400 1400
1200 1200
S11 (MPa)
S11 (MPa)
1000 1000
800 800
600 600
400 400
200 200
0 0
0.0 1.5 3.0 4.5 6.0 7.5 9.0 0 5 10 15 20 25
distance (mm) distance (mm)
1800
RVE5 1800
RVE7
1600 1600
1400 1400
1200 1200
S11 (MPa)
S11 (MPa)
1000 1000
800 800
600 600
400 400
200 200
0 0
0 5 10 15 20 25 30 35 40 45 0 5 10 15 20 25 30 35 40 45 50 55 60
distance (mm) distance (mm)
Fig.14 Stress comparison between different RVE models: (a) stress distribution in carbon and Dyneema yarns in
24
different RVE models; (b) stress distribution of a fractured carbon yarn in different RVE models
Additionally, more than one fracture occurs in the carbon yarns of the RVE7 model.
Therefore, the increased length cannot contribute to the stress recovery in the
fractured yarns. Accordingly, the RVE5 model is the optimum size to characterize the
tensile behavior of the hybrid material.
5.Conclusion
The in-plane behavior of a twill 2,2 carbon-Dyneema fabric reinforced hybrid
composite is analyzed by a three-dimensional meso-scale representative volume
element model. In this model, the carbon and Dyneema yarns are regarded as a
transversely isotropic homogenized material. The non-linear behavior of the Dyneema
yarns in fiber direction is described by the Ramberg-Osgood equation. A combination
of the max-stress and Tsai–Wu failure criterion is used to predict the fiber-dominated
failure and matrix-dominated failure in both carbon and Dyneema yarns respectively.
A linear softening law using the effective stress and displacement based on fracture
energy is adopted to describe the damage progression. The resin pocket is modeled as
a non-associative elasto-plastic material with isotropic damage.
The proposed model is correlated and validated using local strain distributions and
global effective stress-strain responses. The predicted surface strain contours
correlated reasonably well with the DIC results. Besides, the failure mechanism of the
hybrid composite under tension load is analyzed. Both the experimental and FE
simulation curves show two peaks which indicates the non-simultaneous failure of
carbon and Dyneema yarns. In addition, the damage morphology of each constituent
observed in the experiment shows good agreement with the prediction of the FE
model.
The size of the RVE has a significant effect on the results of the simulation. With the
increase of the model size from RVE1 to RVE5, the progressive failure prediction
approaches the experimental curves. This phenomenon is due to the shear lag effect.
However, further increasing the model size from RVE5 to RVE7 shows no significant
25
change, which suggests that the RVE5 model is the optimum size for characterize the
tensile behavior of the hybrid composite.
Acknowledgements
The support of NUS through research grant R265-000-523-646 is gratefully
acknowledged. The support from the Special Excellent PhD International Visit
Program by DHU is also gratefully acknowledged.
Appendix A
The sensitivity of the FE results to the tensile loading rate is investigated at three
different loading rates, i.e., 2.4 m/s, 0.24 m/s and 0.024 m/s. The obtained stress-strain
curves of RVE1 are displayed in Fig.A1.
450
0.024m/s
400 0.24m/s
350 2.4m/s
Stress (MPa)
300
250
200
150
100
50
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Strain (%)
It is found that increasing the loading rate from 0.024 m/s to 0.24 m/s does not
significantly alter the stress-strain curve. However, when the rate is increased to 2.4
m/s, both stress and total strain to failure start to increase. Therefore, the loading rate
of 0.24 m/s is adopted in this study. Furthermore, as recommended in Abaqus
Analysis User’s Manual[42]: to perform a quasi-static analysis, the work done by the
external forces ( ) should be nearly equal to the internal energy ( ) of the system,
and the kinetic energy ( ) of the deforming material should not exceed a small
26
fraction (typically 5% to 10%) of its internal energy throughout most of the process.
The corresponding energy history curves using the loading rate of 0.24 m/s are plotted
in Fig. A2, which confirm that 0.24 m/s is an appropriate loading rate for the current
model.
18
EW
120 16
EI
14
Energy (mJ)
EK/EI (%)
90 12
10
60 8
6
30 4
2
0 0
0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010
Time (s) Time (s)
Fig.A2 Energy history curves at 0.24 m/s: (a) Comparison between the and ; (b)
Reference
2018;109:20-30.
[3] Hine PJ, Bonner MJ, Ward IM, Swolfs Y, Verpoest I. The influence of the hybridisation
configuration on the mechanical properties of hybrid self reinforced polyamide 12/carbon fibre
27
[4] Swolfs Y, Verpoest I, Gorbatikh L. Recent advances in fibre-hybrid composites: materials
[5] Ladizesky NH, Pang MKM. Mechanisms of failure of polymeric resins reinforced with
Technology. 1992;44:127-36.
[6] Porter RS, Kanamoto T. A review on the tensile-strength of polyethylene fibers. Polymer
[7] Li Y, Xian XJ, Choy CL, Guo ML, Zhang ZG. Compressive and flexural behavior of
[8] Peijs A, Dekok JMM. Hybrid composites based on polyethylene and carbon fibers. 6.
[9] Peijs A, Vanklinken EJ. Hybrid composites based on polyethylene and carbon fibers. 5.
1992;11:520-2.
[10] Peijs A, Venderbosch RW, Lemstra PJ. Hybrid composites based on polyethylene and
Composites. 1990;21:522-30.
[11] Ellis RL, Lalande F, Jia HY, Rogers CA. Ballistic impact resistance of SMA and spectra
[12] Larsson F, Svensson L. Carbon, polyethylene and PBO hybrid fibre composites for
28
2002;33:221-31.
[13] Zhao Y, Cao M, Lum WP, Tan VBC, Tay TE. Interlaminar fracture toughness of hybrid
2018;114:377-87.
[15] Bednarcyk BA, Stier B, Simon JW, Reese S, Pineda EJ. Meso- and micro-scale modeling
[16] Zhang C, Li N, Wang WZ, Binienda WK, Fang HB. Progressive damage simulation of
Structures. 2015;125:104-16.
[17] Wang L, Wu JY, Chen CY, Zheng CX, Li B, Joshi SC, et al. Progressive failure analysis of
[18] Wehrkamp-Richter T, De Carvalho NV, Pinho ST. A meso-scale simulation framework for
predicting the mechanical response of triaxial braided composites. Composites Part A-Applied
fabrics from finite element analyses and experimental tests. Textile Research Journal.
2011;81:992-1007.
29
[21] ASTM D3039-08. Standard test method for tensile properties of polymer matrix
composite materials2008.
[22] Zhou Y, Baseer MA, Mahfuz H, Jeelani S. Monte Carlo simulation on tensile failure
process of unidirectional carbon fiber reinforced nano-phased epoxy. Materials Science and
Engineering: A. 2006;420:63-71.
[23] Huang ZM. A unified micromechanical model for the mechanical properties of two
Materials. 2000;13:252-71.
[24] Melro AR, Camanho PP, Pires FMA, Pinho ST. Numerical simulation of the non-linear
[26] Chen XG, Zhou Y, Wells G. Numerical and experimental investigations into ballistic
[27] Liu PF, Zheng JY. Progressive failure analysis of carbon fiber/epoxy composite laminates
[28] Kromm FX, Lorriot T, Coutand B, Harry R, Quenisset JM. Tensile and creep properties of
[29] Camanho PP, Maimi P, Davila CG. Prediction of size effects in notched laminates using
[30] Kermanidis T, Labeas G, Sunaric M, Johnson AF, Holzapfel M. Bird strike simulation on a
30
2006;11:189-201.
[31] Fiedler B, Hojo M, Ochiai S, Schulte K, Ochi M. Finite-element modeling of initial matrix
failure in CFRP under static transverse tensile load. Composites Science and Technology.
2001;61:95-105.
[32] Bullegas G, Pinho ST, Pimenta S. Engineering the translaminar fracture behaviour of
[33] Burger D, de Faria AR, de Almeida SFM, de Melo FCL, Donadon MV. Ballistic impact
[34] Han G, Guan ZD, Li X, Du SY. Failure analysis of carbon fiber reinforced composite
subjected to low velocity impact and compression after impact. Journal of Reinforced Plastics
[35] Torsten R. Lässig FN, Werner Riedel, Michael May An assessment of experimental
techniques for measuring the mode I fracture toughness of UHMW-PE composites. ECCM17 -
[37] Liu HX, Huang W, Lian YS, Li LN. An experimental investigation on nonlinear behaviors of
synthetic fiber ropes for deepwater moorings under cyclic loading. Applied Ocean Research.
2014;45:22-32.
[38] Bogetti TA, Hoppel CPR, Harik VM, Newill JF, Burns BP. Predicting the nonlinear
31
response and progressive failure of composite laminates. Composites Science and
Technology. 2004;64:329-42.
[39] Ridha M, Wang CH, Chen BY, Tay TE. Modelling complex progressive failure in notched
composite laminates with varying sizes and stacking sequences. Composites Part A-Applied
[40] Tsai SW, Wu EM. A general theory of strength for anisotropic materials. Journal of
[43] Melro AR. Analytical and numerical modelling of damage and fracture of advanced
[44] Tschoegl NW. Failure surfaces in principal stress space. Journal of Polymer Science Part
[46] Stier B, Simon JW, Reese S. Comparing experimental results to a numerical meso-scale
32