Experimental and Numerical Analysis

of Damage in Woven GFRP Composites

Under Large-deflection Bending

Himayat Ullah, Andy R. Harland &

Vadim V. Silberschmidt

Applied Composite Materials

An International Journal for the Science
and Application of Composite Materials

ISSN 0929-189X

Appl Compos Mater

DOI 10.1007/s10443-011-9242-7

1 23
Your article is protected by copyright and
all rights are held exclusively by Springer
Science+Business Media B.V.. This e-offprint
is for personal use only and shall not be self-
archived in electronic repositories. If you
wish to self-archive your work, please use the
accepted author’s version for posting to your
own website or your institution’s repository.
You may further deposit the accepted author’s
version on a funder’s repository at a funder’s
request, provided it is not made publicly
available until 12 months after publication.

1 23
 Author's personal copy
Appl Compos Mater
DOI 10.1007/s10443-011-9242-7

Experimental and Numerical Analysis of Damage in Woven

GFRP Composites Under Large-deflection Bending

Himayat Ullah & Andy R. Harland &

Vadim V. Silberschmidt

Received: 16 September 2011 / Accepted: 16 December 2011

# Springer Science+Business Media B.V. 2011

Abstract Textile-reinforced composites such as glass fibre-reinforced polymer (GFRP) used

in sports products can be exposed to different in-service conditions such as large bending
deformation and multiple impacts. Such loading conditions cause high local stresses and strains,
which result in multiple modes of damage and fracture in composite laminates due to their
inherent heterogeneity and non-trivial microstructure. In this paper, various damage modes in
GFRP laminates are studied using experimental material characterisation, non-destructive
micro-structural damage evaluation and numerical simulations. Experimental tests are carried
out to characterise the behaviour of these materials under large-deflection bending. To obtain in-
plane shear properties of laminates, tensile tests are performed using a full-field strain-
measurement digital image correlation technique. X-ray micro computed tomography (Micro
CT) is used to investigate internal material damage modes – delamination and cracking. Two-
dimensional finite element (FE) models are implemented in the commercial code Abaqus to
study the deformation behaviour and damage in GFRP. In these models, multiple layers of
bilinear cohesive-zone elements are employed to study the onset and progression of inter-ply
delamination and intra-ply fabric fracture of composite laminate, based on the X-ray Micro CT
study. The developed numerical models are capable to simulate these features with their
mechanisms as well as subsequent mode coupling observed in tests and Micro CT scanning.
The obtained results of simulations are in agreement with experimental data.

Keywords Textile composites . Micro computed tomography . Fracture . Delamination .

Finite-element models

1 Introduction

Woven-fabric composite laminates offer a number of attractive mechanical properties

compared to their unidirectional-tape counterparts such as good resistance to fracture and

H. Ullah (*) : A. R. Harland : V. V. Silberschmidt

Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University, Ashby Road,
Loughborough, Leicester-shire LE11 3TU, UK
e-mail: H.Ullah@lboro.ac.uk
 Author's personal copy
Appl Compos Mater

transverse rupture due to weaving resistance, and high impact strength [1, 2]. These
properties have attracted the sports industry to incorporate woven GFRP laminates in the
design of sports products. Such products could be subjected to large-deflection bending and
multiple impacts in service conditions. This type of quasi-static and dynamic loads generate
high local stresses and strains leading to complex damage modes due to heterogeneity and
anisotropy of composite laminates. In textile composites, damage begins at micro-scale with
matrix cracking, fibre matrix debonding and fibre failure within the ply. This is followed by
meso-scale damage such as intra-yarn cracking and inter-ply delaminations. On the macro-
scale, composite failure is characterised by a strong interaction of intra-ply cracking and
inter-ply delamination and ultimate fabric rupture [3]. Traditionally, these intralaminar and
interlaminar damage mechanisms have been treated separately in damage analysis of
composite components and structures. Delamination is normally analysed using methods
based on linear-elastic fracture mechanics, such as the virtual crack closure technique
(VCCT) [4], or using cohesive-zone models (CZMs) [5–9]. The intra-ply damage mecha-
nisms are normally predicted using ply-based strength failure criteria coupled with contin-
uum damage mechanics (CDM) [10–14]. Since the interlaminar and intralaminar damage
modes are strongly interacting energy-dissipation mechanisms in composite structures,
continuum damage modelling is incapable of capturing such interaction between these
discrete damage mechanisms. A homogenization process inherent in CDM models, leads
to the loss of key information on multiple-damage coupling at the macroscopic scale and
thus may result in inaccurate prediction of a crack path [12, 15]. To accurately predict
interaction between these discrete failure modes, it is essential to have an explicit kinematic
representation of all damage mechanisms in global composite structure models [16]. Re-
cently, this interactive damage modelling approach has been studied in [16–19]. Hallet et al.
[17] simulated fracture of quasi-isotropic laminate tension tests using CZMs to represent
both inter-ply delamination and intra-ply splitting damage mechanisms and their interaction.
Khokhar et al. [18] developed a method of stochastic cohesive-zone elements (CZE) to study
the interaction of matrix cracking and delamination in CFRP cross-ply laminates. Okabe et
al. [19] applied CZEs to model transverse cracking and delamination, and truss elements to
represent fibre breakage in cross-ply glass-epoxy laminates loaded in tension. The numerical
models reproduced the complex progressive damage mechanisms in GFRP very well.
However, research on damage interaction in woven laminates subjected to large bending
deformation is limited. Damage models based on CZEs have the capability to simulate
initiation, propagation and coupling between inter-ply and intra-ply failure mechanisms.
Although CZMs have been widely used in simulation of composite damage, their applica-
tion is limited by the fact that they require a priori knowledge of a potential crack path; and
the need to introduce CZEs at every single interface where a crack may develop [20]. An
alternative approach is to use improved numerical methods that allow for arbitrary crack
initiation and propagation in continua such as extended finite element method (X-FEM).
However, X-FEM still has the difficulty in simulating interaction among the multiple crack
mechanisms [16].
In this paper, damage initiation, progression and interaction in GFRP woven laminates
subjected to large-deflection bending is studied. Flexural and tensile tests were carried out to
characterise the material behaviour. Location of discrete inter-ply and intra-ply damage
mechanisms was identified by micro-structural examination of tested GFRP laminates using
X-ray micro computed tomography (Micro CT) technique. Two-dimensional finite-element
(FE) models are developed in Abaqus to study the behaviour of multiple delaminations,
fabric fracture and their interaction using CZEs at crack locations identified with X-ray
Micro CT.
 Author's personal copy
Appl Compos Mater

2 Material Characterisation

In this study, two types of fabric-reinforced GFRP materials were used. The main one,
used in our flexural tests, Micro CT study and numerical simulations, is a 1 mm-thick
four-ply on-axis 2D balanced woven laminate. Due to requirements of ASTM D3518
[19] for measuring in-plane shear properties that asks for a minimum thickness of
2 mm, another off-axis [+45/–45]2s 2D woven laminate consisting of 8 plies (that are
the same as in the main material) was employed.

2.1 Flexural Tests

Two types [0°]4 and [90°]4 GFRP specimens were tested under three-point bending con-
ditions to characterise the material behaviour and validate our numerical models. These
specimens were fabricated from a composite with a 2/2 twill glass-fibre fabric reinforcement
in thermoplastic polyurethane (TPU) polymer matrix, with a fibre volume fraction of 45%;
the fabric has the same number of yarns in the warp and weft directions. Water-jet cutting
was used to prepare rectangular specimens of 80 mm length, 25 mm width and 1 mm
thickness, each laminate having four plies of 0.25 mm thickness. Flexural tests were carried
out at indenter speed of 100 mm/min, equivalent to a strain rate of 0.042 s−1, using the
Instron 5569 machine in accordance with the ISO 178 standard. Five samples per orientation
were tested under large-deflection bending until their ultimate fracture.
Both 0° and 90° samples exhibited a quasi-brittle failure response in flexure as shown in
Fig. 1. The tests resulted in the flexural modulus of 21 GPa and 21.5 GPa, and ultimate flexural
strengths of 650 MPa and 668 MPa, for 0° and 90°, respectively. The similarity in flexural
moduli was due to the symmetry of fibres in both warp and weft directions in both types of
samples. Stiffness degradation due to internal cracks and delamination occurred in both 0° and
90° samples at about 70% of the ultimate load. The specimens underwent matrix cracking and
interlaminar damage before the structure lost its load-carrying capacity; the development of
such intra-ply and inter-ply damage was reflected by flattening of the stress-strain curve at about
3% strain, where the curve became nonlinear. The elastic flexural moduli such as E11 and E22
were calculated from the mechanical tests of 0° and 90°, respectively, and were listed in Table 1.

2.2 In-plane Shear Tests

Tensile tests were conducted in accordance with ASTM D3518 [21] to determine the in-
plane shear modulus, in-plane Poisson’s ratio and shear strength of ±45° GFRP. Although
the standard outlines the procedure for using strain gauges in longitudinal and transverse
Fig. 1 Stress-strain diagram from
flexural tests of twill 2/2 textile
GFRP Laminates
 Author's personal copy
Appl Compos Mater

Table 1 Material properties of GFRP used in FE models

E11 (GPa) E22 (GPa) E33 (GPa) G12 (GPa) G13 0G23 (GPa) ν12

21 21.5 8 3.1 3.6 0.11

directions, they only provide the local strain data. Therefore, an ARAMIS digital image
correlation (DIC) system was used to obtain the full-field in-plane displacement and strain
data. DIC technique has already been used to characterise the material behaviour of textile
laminates e.g.,[22]. An Instron 5569 machine was used in those tests; while two high-speed
1 megapixel digital cameras took photographs of the specimen at frequency of 50 frames per
second during loading. The specimens were sprayed with white paint, and then black
speckles were marked on them to provide contrast necessary for image correlation. The
Instron machine had an integrated load cell, and load measurements were synchronised to
each photograph at each time interval.
Specimens of [+45/−45]2s GFRP laminates were tested at crosshead speed of 10 mm/min
equivalent to a strain rate of 0.0014 s−1. Rectangular specimens of 200 mm length, 25 mm
width and 2 mm thickness were prepared. Nonlinear shear stress-strain behaviour of off-axis
GFRP specimen is presented in Fig. 2. The stress-strain curve becomes nonlinear at shear
strain of approximately 0.01, which is due to matrix cracking and fibre trellising as explained
in [23]. Major (axial), minor (transverse) and shear strain contour plots from digitally
imaged tension tests of GFRP samples in the linear elastic range are shown in Fig. 3. The
variation in the strain contour plots might be due to more compliant regions such as resin
pockets between the tows, or tows aligned along the loading directions (fibre trellising). The
GFRP in-plane shear modulus G12 and Poisson’s ratio ν12 were determined from the
obtained data in the linear elastic range and listed in Table 1. The rest of elastic properties
such as E33, G13 and G23 in Table 1 were taken from [24], where a similar GFRP woven
laminate was studied.

3 X-ray Micro Computed Tomography for Damage Characterisation

Damage such as delamination and matrix cracking in composites laminates usually

occur inside the laminates and is barely visible. These damage mechanisms need
examination for reliable and accurate modelling and analysis of composite laminates.
Non-destructive evaluation method such as micro computed tomography (Micro CT)
was used to visualise the 3D internal deformation and damage behaviour of tested

Fig. 2 Shear stress-strain

diagram from digitally imaged
tensile test of [+45/−45]2s twill
2/2 GFRP laminate
 Author's personal copy
Appl Compos Mater

Fig. 3 Digital images of (a)

major; (b) minor; and (c) shear
strain from [+45/−45]2s GFRP
tensile test

composite laminates. This technique has been used to investigate micro cracking and
delamination in composite laminates at micron-range level [25, 26]. In this study, X-
ray Micro CT measurements were performed using an XT H 225 X-ray scanner. The
system consists of an X-ray detector and an electronic X-ray source, creating 2D
cross-sections of the object. The source is a sealed X-ray tube operating at 25–225 kV
with a 3 μm spot size. The specimen was positioned by an object manipulator with
two translations and one rotation, rotating and raising/lowering the sample to a
specific region of interest for adjustment of the sample magnification and acquisition
of tomographic data. Following acquisition, a software program builds a precise 3D
map from 2D radiograph images by ‘stacking’ the individual slices one on top of the
other; this process is known as reconstruction. As denser materials absorb more X-
rays than voids and air, this attenuation contrast allows detection and characterisation
of cracks and flaws in tomographic images. High scan resolution is required to obtain
maximum internal details of damage in the composite laminate; as the resolution is
increased, the field of view of the sample is reduced. However, samples must remain
within the field of view to obtain the radiographs of the region of interest, thus there
is a tradeoff. A small sample of size 6.5 mm×5 mm was prepared from the damaged
region of the tested GFRP laminate to meet these requirements. Data for the sample
was collected at 75 kV and 80 μA. Transmission X-ray images were acquired from
3600 rotation views over 360° of rotation (0.1° rotation step) for 3D reconstruction.
These settings resulted in tomographs with resolution of 6.1 μm.
The reconstructed 2D and 3D images of the transversely fractured GFRP specimen are
shown in Fig. 4. Figure 5 demonstrates inter-ply and intra-ply damage mechanisms in the
sample at various locations across its width. Dark grey regions in images represent cracks
and damage whereas light grey regions represent higher density material i.e. glass-fibre
yarns. Realisation of damage mechanisms at outer edge, 25%, 50% and 75% of the sample
width is shown in Fig. 5a, b, c and d, respectively. It was shown that before ultimate fracture,
the laminate exhibited matrix cracking and then delaminations and tow debondings. Matrix
cracks developed in the weak resin-rich pockets around the tows. Inter-ply delamination and
intra-ply delamination such as tow debonding can also be observed. These delaminations
normally appeared near the matrix cracks area, which suggested that the formation of the
cracks initiated delamination. In the fibre-rich regions the damage was apparently associated
with debonding at the fibre/matrix interface. At the time of fabric fracture, the analysis of
 Author's personal copy
Appl Compos Mater

Fig. 4 Reconstructed 2D and 3D

images of fractured twill 2/2
GFRP specimen (resolution 6.1
μm): (a) top surface; (b) 3D view;
(c) bottom surface

internal structure showed that almost every ply was delaminated. All the tomographs showed
that delamination and subsequently transverse ply fracture are the prominent failure modes.
This examination formed a basis for placement of CZEs at the locations of interlaminar and
intralaminar damage in our computational models.

4 Finite-element Analysis

Failure prediction in fibre-reinforced composite laminates is complicated by their heteroge-

neous nature, which gives rise to various types of multiple cracks, interacting strongly as
failure progresses. Hence, finite-element method (FEM) is capable of simulating the entire
damage process from its initiation through evolution to complete failure of a composite
structure. However, the development of a proper numerical model representing the physics
of damage mechanisms is a challenging task [12]. In this study, cohesive interface elements
 Author's personal copy
Appl Compos Mater

Fig. 5 Damage mechanisms in reconstructed slices across width of the sample (resolution 6.1 μm): (a) edge;
(b) 25%; (c) 50%; (d) 75% of the width

are implemented in FE models to characterise the onset, progression and coupling of damage
for analysis of composite structures without any pre-cracks. The study involves monitoring
of a particular type of parameter such as stiffness degradation for prediction and analysis.

4.1 Modelling Strategy

Finite-element models are developed in the commercial FE package ABAQUS/Explicit to

investigate large-deflection bending of tested composite laminates and the resulting inter-ply
and intra-ply damage. Two FE models - A and B - as shown in Fig. 6 are developed representing
 Author's personal copy
Appl Compos Mater

Fig. 6 FE models for damage

simulation in three-point bending

the bending tests on an 80 mm long, 25 mm wide and 1 mm thick GFRP laminate. Model A
contains three cohesive layers - one above the beam’s neutral axis (NA), the second coinciding
with it, and the third below the NA to simulate multiple delamination scenarios. These layers are
at interfaces between plies of the studied laminate. Model B has the same three layers with one
additional transverse cohesive layer at the laminate centre to simulate coupling between
delamination and ply fracture. The cohesive layer above the NA is referred to as top cohesive
layer (TCL), the cohesive layer on the NA is referred to as mid cohesive layer (MCL), the one
below the NA is referred to as bottom cohesive layer (BCL) and the through-thickness centre
layer is referred to as centre cohesive layer (CCL). These cohesive layers are included in the
model since the location of damage initiation is a priori known from our Micro CT analysis.
Obviously, if the damage location is not known in advance, the modelling effort, complications
due to the calibration of penalty stiffness and computational times may increase, and the
solution convergence becomes rather complicated. Further, the number of cohesive layers in
FE models should not affect the load-bearing capacity of the specimen and its global stiffness
before the damage starts in the actual laminate. In three-point bending, the layers above and
below the NA experience compression and tension respectively due to normal bending stress,
whereas the mid layers are subjected to shear. Therefore, the cohesive layers at these locations
are defined to simulate both single and mixed-mode fracture mechanisms.
The laminate has a considerable width in z-direction, thus, two-dimensional FE models
based on plane-strain elements are developed to represent out-of-plane bending behaviour in
a computationally cost-effective manner. Composite laminas are meshed with plane-strain
reduced-integration and hourglass-control CPE4R elements capable of eliminating shear
locking in bending problems, using a structured meshing technique. Interlaminar and intra-
laminar cohesive layers are meshed with two-dimensional COH2D4 elements. Indenter load
application is represented by a circular rigid arc at the centre of the beam that is also laterally
supported by two other circular rigid arcs. Surface-to-surface explicit contact is defined
between the rigid arcs and the laminate’s top and bottom surfaces. All composite laminas are
assigned elastic properties listed in Table 1.

4.2 Damage Modelling Using Cohesive-zone Elements

Computational models with the capability to predict the initiation and progression of damage
can reduce the number of costly experimental tests and can lead to improved designs.
 Author's personal copy
Appl Compos Mater

Cohesive-zone elements (CZE) have the ability to capture the onset and propagation of
delamination [7–9]. The cohesive behaviour assumes that failure of elements is characterized
by progressive degradation of the material stiffness, which is driven by a damage process.
Inter-ply and intra-ply failures in composite laminates initiate and propagate under the
combined influence of normal and shear stresses. The nominal quadratic stress criterion is
used for damage initiation. Damage propagation is usually predicted by criteria established
in terms of the energy release rates and fracture toughness under mixed-mode loading. This
study is based on the criterion proposed by Benzeggah and Kenane [27] :
 n
GS
GC ¼ GIC þ ðGIIC  GIC Þ ; ð1Þ
GT
where GT is the work done by the interface tractions; GS/GT is the fraction of cohesive
energy dissipated by shear tractions; GS is the work done by the shear components of
interface tractions; GIC and GIIC are critical energy release rates in modes I and II,
respectively, and, η is the material mode-mixity parameter.
The interlaminar normal σI0 and shear σII0 strengths of 27.3 MPa and 10.6 MPa,
respectively, are used as typical strength values for the resin matrix in GFRP laminate.
The fracture toughness values in mode I GIC and mode II GIIC of 1.21 N/mm and 4.55 N/
mm, respectively, as well as the value of mode-mixity parameter η01.53 are taken from [28],
where fracture tests are reported for a composite configuration similar to that under
investigation in this study. The interface element stiffness should be large enough to
avoid relative displacements between the connected ply elements but also not too
large to cause numerical problems such as spurious oscillations in interfacial traction
of the cohesive element [29]. Turon et al. [29] proposed that elastic properties of the
composite would not be affected if the interface stiffness is defined as K ¼ aE33 =tp ,
where E33 is the material’s through-thickness stiffness, tp is the thickness of ply
connected by the cohesive element, and α is a non-dimensional parameter, which
should be greater than 50 for accurate simulation of various problems. Based on this
relation, the value of interface stiffness 6×106 N/mm3 is used in this study. The
application of CZE requires a fine spatial discretization at the cohesive zone to
capture the damage growth properly. The number of elements Ne in the cohesive-
zone is determined according to Ne ¼ lcz =le , given in [29], where lcz is the length of
cohesive-zone and le is the mesh size in the direction of crack propagation. Before
performing our simulations, a mesh convergence study with different lengths of cohesive
elements was performed and is summarized in Fig. 7. The results indicate that by decreasing
the element length, the damage zone along the laminate length increases and solution

Fig. 7 Damage sensitivity to

cohesive element size
 Author's personal copy
Appl Compos Mater

convergence is achieved. Mesh 3 with interface element length of 0.1 mm and 2 μm

thickness was selected for computationally effective simulations of damage in numer-
ical models A and B.

4.3 Boundary Conditions and Solution

Simply-supported boundary conditions are applied at the reference points of the rigid
supports below the laminate representing the test fixture. A displacement-controlled load
is applied at the centre rigid arc representing the indenter, which is in contact with the top ply
of the laminate, for better convergence of the solution. Boundary conditions are applied at
rigid surfaces instead of constraining the ply nodes, as the local stresses due to the constraint
edge effects disperse over greater distances of the specimen because of the composite’s
anisotropy. As elaborated in [30] for anisotropic composite materials, the application of St.
Venant’s principle for plane elasticity problems involving anisotropic materials is not
justified in general. Quasi-static analysis is carried out for large-deflection effects by
applying the load gradually in small time increments of 10−8 s to capture the damage
process, keeping the amount of kinetic energy small with respect to the total energy. The
final FE model B contained a total of 12,832 elements and it took 4 h on a dual core machine
with two 2.7 GHz processors each. The computational cost is a direct consequence of a fine
mesh coupled with the highly nonlinear behaviour of interface damage elements.

5 Results and Discussion

Numerical results of simulations of the large-deflection bending behaviour of GFRP lami-

nates and comparison with experimental tests and Micro CT examination are presented in
this section. Experimental flexural and in-plane shear tests characterised the material behav-
iour. The initiation and progression behaviours of multiple delaminations in FE model A is
shown in Fig. 8. Damage is represented by normalised length Ld/L of the cohesive layer
along the beam axis against the normalised displacement loading δ/δf, where δf is the
displacement at the ultimate failure of the test specimen. Damage initiates earlier, at about
5% of the applied load in the overhang regions L (edge) of the top, mid and bottom cohesive

Fig. 8 Damaged zone at the edge

and middle of the beam—Model
A
 Author's personal copy
Appl Compos Mater

layers, than in the mid-regions L (mid). This may be due to a free-edge effect caused by
mismatch of the elastic material properties between two adjacent dissimilar layers, due to a
combination of interlaminar normal and shear stresses. This combination in the overhang
region initiates the mixed-mode fracture. After a sudden initiation, the damage propagates at
a slower rate in the beam edge up to 70% of the applied load. After this, a faster stable
delamination growth is predicted in the overhang edges until the ultimate failure. As studied
by the authors [8, 9], in three-point bending, the maximum through-thickness interlaminar
normal stress is compressive near the loading and support rollers and is negligible near the
mid plane. Similarly the maximum shear stresses occur in the vicinity of loading and support
rollers due to stress concentration, and through-thickness high interlaminar shear stress
(ILSS) at the mid plane of the laminate due to the shear force associated with bending.
These interlaminar shear stresses are responsible for damage initiation in the mid-region.
Therefore, as the applied load reaches 50%, where the interlaminar shear stress equals to the
interface shear strength σII0, mode-II type delamination initiates in the specimen’s mid-
region and grows more rapidly in the MCL than TCL and BCL because of high ILSS at the
mid plane. The TCL above the NA is in the state of compressive stresses, and thus damage
initiates at higher loads.
The initiation and progression behaviours of multiple inter-ply delaminations and intra-
ply fracture in model B are demonstrated in Fig. 9. Here, too, delamination initiates earlier in
the overhang regions of the top, mid and bottom cohesive layers than in the mid-region. The
mid-region exhibits a faster damage growth of mode-II type than the overhang region, which
shows mixed-mode fracture. The MCL is more damaged as the ILSS increases toward the
middle, i.e., the NA of the beam. In this case, the entire mid interface layers are totally
damaged at 95% of the applied load where the ply fracture starts as shown by damage in
CCL. The damage dissipation energy density in both models A and B is presented in Fig. 10.
Since model A represents only the multiple inter-ply delamination process, the maximum
damage dissipation energy is 2 J/mm3. In model B, this energy abruptly increases to 37 J/
mm3 where the ultimate laminate fracture occurs at 98% of the applied load. The sequence of
the inter-ply delamination and intra-ply fabric breakage is illustrated in the contour plots of
deformed damage shown in Fig. 11. Here, the interface layers are fully delaminated at 95%
of the applied load before the fabric breakage initiates, as represented by the laminate mid-
region in Fig. 11a. The first ply fracture occurs at 96% as shown in Fig. 11b followed by the

Fig. 9 Damaged zone at the

edge, middle and ply-crack of the
beam—Model B
 Author's personal copy
Appl Compos Mater

Fig. 10 Damage dissipation

energy density predicted in
models A and B

second and third ply fracture illustrated in Fig. 11c and d, respectively, at higher loads. In
damaged areas, the ply elements move relative to each other as in mode-II fracture and the
mismatch of element boundaries shows delamination and ply fracture. The fourth ply is left
intact, since in bending it is less damaged because of compressive stresses on top of the
beam. The damage then proceeds on both sides of the cut plies along the delaminated TCL in
the form of mode-II type fracture. Variation of flexural stress σxx along the beam thickness in
model B is presented in Fig. 12. Although the laminate is fully delaminated in Fig. 12a, the
flexural stress contour is uniform. This stress is reduced as the first and second plies are
fractured (Fig. 12b and c, respectively). Stress concentration at the locations of interlaminar
crack tips is apparent in Fig. 12b and c. In all the numerical models, delamination initiates
earlier at the laminate edges and then propagates at a higher rate. Damage suddenly spreads
after attainment of 60% of the failure load, especially in the middle section of the beam.
Although micro-cracking such as matrix cracks observed in Micro CT tomographs is not
represented by the meso-scale modelling approach, still the models are capable to predict
delamination, fabric fracture and their interaction.
Comparison of load-deflection behaviour obtained numerically for two different models
and experimentally from three-point bend tests of GFRP woven laminates is presented in
Fig. 13, which shows a good agreement. It can be observed that the development of such

Fig. 11 Interaction of inter-ply

and intra-ply damage in model—
B at different stages of three-point
bending (scaling factor 0.5)
 Author's personal copy
Appl Compos Mater

Fig. 12 Contour plot of bending stress σxx in Model B at different stages of three-point bending test (scaling
factor 0.5)

interlaminar damage did not affect noticeably the force-deflection curves till the stiffness
degradation occurred at points P, Q, and T (Fig. 13). Delamination damage occurs at these
points, characterised by a drop in load, followed by instantaneous loss of structure’s load-
carrying capability. Both the numerical models indicate the delamination initiation at 60% of
the applied load as can also be observed in Figs. 8 and 9. During delamination propagation,
both the FE models A and B show nonlinear progressive damage behaviour. The ultimate
fabric fracture in model B at 98% of the ultimate load is identified by a quick load drop.
Although the numerical models show earlier stiffness degradation due to delamination which

Fig. 13 Numerical and

experimental load-deflection
response of GFRP laminates
under bending
 Author's personal copy
Appl Compos Mater

may be because of the lower interface properties than the test results, the main features of
behaviour are the same.

6 Conclusions

Damage in GFRP textile composites under large-deflection bending was studied using
expeirmental tests, Micro CT scanning and numerical simulations. Flexural as well as in-
plane shear tests using DIC technique were carried out to characterise the behaviour of the
woven GFRP laminate. Micro CT images were obtained at 6.1 μm resolution to characterise
the internal structural damage of the laminate. Based on the damage behaviour identified in
the scanned images, meso-scale 2D FE models were developed to simulate some of the
interlaminar and intralaminar damage in woven composites. A series of simulations were
performed to study the onset, progression and interaction of inter-ply and intra-ply damage
processes under mixed-mode large-deflection bending by employing multiple layers of
CZEs in the FE models of the originally undamaged specimen. The numerical results were
close to the experimental ones, and the numerical models have the capability to reproduce
the damage sequence and pattern observed experimentally in composite laminates. Numer-
ical simulations showed that damage initiation and growth was sensitive to the mesh size of
cohesive-zone elements. The load level at which the load drops occurred, caused by
delamination damage, also showed good correlation as did the predicted fibre failure load.
However, the calibration of numerical models based on interface layers proved to be highly
mesh- and stiffness-sensitive and would certainly represent a critical issue in the application
of the approach to real-world components and structures. It is also important for all interface
parameters to be calibrated and specified correctly in order to avoid long computational
times, solution oscillations or even premature termination and to obtain better convergence.
Therefore, a complete investigation of mesh-sensitivity of the results, identification of
interface stiffness and strength levels through experiments, and a comparison of the results,
obtained by applying different constitutive laws, are required to develop more reliable and
robust 3D FE models for 3D damage growth and interaction of failure mechanisms in
composites. Further, Micro CT may be employed to acquire both qualitative and quantitative
data of the developed damaged zones in composite laminates under various loading con-
ditions to validate our FE models.

Acknowledgements The authors acknowledge the support of Dr Tim Lucas and Dr Dan Price from adidas
AG, Herzogenaruch, Germany, for provision of composite samples.

References

1. Ernst, G., Vogler, M., Hühne, C., Rolfes, R.: Multiscale progressive failure analysis of textile composites.
Compos. Sci. Tech. 70(1), 61–72 (2010)
2. Ansar, M., Xinwei, W., Chouwei, Z.: Modeling strategies of 3D woven composites: a review. Compos.
Struct. 93(8), 1947–1963 (2011)
3. Daggumati, S., Voet, E., Van Paepegem, W., Degrieck, J., Xu, J., Lomov, S.V., Verpoest, I.: Local strain in
a 5-harness satin weave composite under static tension: Part I—Experimental analysis. Compos. Sci.
Tech. 70, 1926–1933 (2010)
4. Krueger, R.: Virtual crack closure technique: history, approach, and applications. Appl. Mech. Rev. 57(2),
109–143 (2004)
5. Yang, Q., Cox, B.: Cohesive models for damage evolution in laminated composites. Int. J. Fract. 133(2),
107–137 (2005)
 Author's personal copy
Appl Compos Mater

6. Allix, O., Blanchard, L.: Mesomodeling of delamination: towards industrial applications. Compos. Sci.
Tech. 66(6), 731–744 (2006)
7. Turon, A., Camanho, P.P., Costa, J., Renart, J.: Accurate simulation of delamination growth under mixed-
mode loading using cohesive elements: definition of interlaminar strengths and elastic stiffness. Compos.
Struct. 92(8), 1857–1864 (2010)
8. Ullah, H., Harland, A.R., Lucas, T., Price, D., Silberschmidt, V.V.: Finite-element modelling of bending of
CFRP laminates: multiple delaminations. Comput. Mater. Sci. 52(1), 147–156 (2012)
9. Ullah, H., Harland, A.R., Lucas, T., Price, D., Silberschmidt, V.V.: Analysis of nonlinear deformations
and damage in CFRP textile laminates. J. Phys. Conf. 305, 12045 (2011)
10. Maimí, P., Camanho, P.P., Mayugo, J.A., Dávila, C.G.: A continuum damage model for composite
laminates: Part II—Computational implementation and validation. Mech. Mater. 39(10), 909–919 (2007)
11. Thollon, Y., Hochard, C.: A general damage model for woven fabric composite laminates up to first
failure. Mech. Mater. 41(7), 820–827 (2009)
12. Van Der Meer, F.P., Sluys, L.J.: Continuum models for the analysis of progressive failure in composite
laminates. J. Compos. Mater. 43(20), 2131 (2009)
13. Silberschmidt, V.V.: Matrix cracking in cross-ply laminates: effect of randomness. Compos. Appl. Sci.
Manuf. 36(2), 129–135 (2005)
14. Silberschmidt, V.V.: Crack propagation in random materials: computational analysis. Comput. Mater. Sci.
26, 159–166 (2003)
15. Wisnom, M.R.: Modelling discrete failures in composites with interface elements. Compos. Appl. Sci.
Manuf. 41(7), 795–805 (2010)
16. Fang, X.J., Zhou, Z.Q., Cox, B.N., Yang, Q.D.: High-fidelity simulations of multiple fracture processes in
a laminated composite in tension. J. Mech. Phys. Solid (2011). doi:10.1016/j.jmps.2011.04.007
17. Hallett, S.R., Jiang, W.G., Khan, B., Wisnom, M.R.: Modelling the interaction between matrix cracks and
delamination damage in scaled quasi-isotropic specimens. Compos. Sci. Tech. 68(1), 80–89 (2008)
18. Khokhar, Z.R., Ashcroft, I.A., Silberschmidt, V.V.: Interaction of matrix cracking and delamination in
cross-ply laminates: simulations with stochastic cohesive zone elements. Appl. Compos. Mater. 1–14
(2011).
19. Okabe, T., Nishikawa, M., Takeda, N.: Numerical modeling of progressive damage in fiber reinforced
plastic cross-ply laminates. Compos. Sci. Tech. 68(10–11), 2282–2289 (2008)
20. Maimi, P., Mayugo, J.A., Camanho, P.P.: A three-dimensional damage model for transversely isotropic
composite laminates. J. Compos. Mater. 42(25), 2717 (2008)
21. ASTM D3518/D3518M-94: Standard Test Method for In-Plane Shear Response of Polymer Matrix
Composite Materials by Tensile Test of a ±45° Laminate (2007).
22. Lomov, S.V., Ivanov, D.S., Verpoest, I., Zako, M., Kurashiki, T., Nakai, H., Molimard, J., Vautrin, A.:
Full-field strain measurements for validation of meso-FE analysis of textile composites. Compos. Appl.
Sci. Manuf. 39(8), 1218–1231 (2008)
23. Ullah, H., Harland, A.R., Blenkinsopp, R., Lucas, T., Price, D., Silberschmidt, V.V.: Analysis of nonlinear
shear deformations in CFRP and GFRP textile laminates. Appl. Mech. Mater. 70, 363–368 (2011)
24. Johnson, H.E., Louca, L.A., Mouring, S., Fallah, A.S.: Modelling impact damage in marine composite
panels. Int. J. Impact. Eng. 36(1), 25–39 (2009)
25. Awaja, F., Nguyen, M.-T., Zhang, S., Arhatari, B.: The investigation of inner structural damage of UV and
heat degraded polymer composites using X-ray micro CT. Compos. Appl. Sci. Manuf. 42(4), 408–418
(2011)
26. Badel, P., Vidal-Sallé, E., Maire, E., Boisse, P.: Simulation and tomography analysis of textile composite
reinforcement deformation at the mesoscopic scale. Compos. Sci. Tech. 68(12), 2433–2440 (2008)
27. Benzeggagh, M.L., Kenane, M.: Measurement of mixed-mode delamination fracture toughness of
unidirectional glass/epoxy composites with mixed-mode bending apparatus. Compos. Sci. Tech. 56(4),
439–449 (1996)
28. Dharmawan, F., Simpson, G., Herszbergand, S.: Mixed mode fracture toughness of GFRP composites.
Compos. Struct. 75(1–4), 328–338 (2006)
29. Turon, A., Davila, C.G., Camanho, P.P., Costa, J.: An engineering solution for mesh size effects in the
simulation of delamination using cohesive zone models. Eng. Fract. Mech. 74(10), 1665–1682 (2007)
30. Horgan, C.O., Simmonds, J.G.: Saint-Venant end effects in composite structures. Compos. Eng. 4(3),
279–286 (1994)