Thin–Walled Structures 161 (2021) 107455

Contents lists available at ScienceDirect

Thin-Walled Structures
journal homepage: http://www.elsevier.com/locate/tws

Experimental and numerical investigation of open-hole tensile properties

and damage mechanisms of 3D woven composites under weft-loading
Qiwei Guo a, c, Yifan Zhang a, b, *, Diansen Li d, Qingtao Lv a, b, Xiaolun Sun a, b, Ming Ma a,
Li Chen a, b, **
a
Key Laboratory of Advanced Textile Composite Materials, Ministry of Education, TianGong University, Tianjin, 300387, China
b
School of Material Science and Engineering, TianGong University, Tianjin, 300387, China
c
School of Textile Science and Engineering, TianGong University, Tianjin, 300387, China
d
School of Chemistry, Beijing University of Aeronautics and Astronautics, Beijing, 100191, China

A R T I C L E I N F O A B S T R A C T

Keywords: For 3D woven composites used in the structural applications, the open-hole mechanical performance is of critical
3D woven composites concern for the components in service. In the present study, the circular notched 3D woven composite (N-3DWC)
Open-hole tensile properties plates with three different apertures were prepared, and the effects of notch size on the tensile properties were
Macro-meso coupling model
investigated by experimental and numerical methods. Digital image correlation technique was used to record the
Damage propagation
CT scanning
full-field strain distributions, and the micro computed tomography was applied to evaluate the failure fractures.
Based on the developed macro-meso coupling model of N-3DWC, the open-hole tensile strength and failure
mechanisms were predicted and analyzed, and the model was validated with the experimental results. The results
showed that the stress-strain curves of all N-3DWC specimens were basically linear in the initial loading stage.
The notch size had little effect on the tensile stiffness of notched plates, while the tensile strength decreased with
the increasing aperture. The stress concentration mostly occurred at the hole edge, and the distribution width
was approximately equal to the notch diameter. As the aperture increased, the main failure mode of weft yarn
changed from fiber pull-out to breakage, and the contour of damaged region changed from to convex to concave.

1. Introduction The mechanical behaviors of open-hole laminated composites have

been intensively studied during the past decades. The effect of nothch
Three-dimensional (3D) woven compoistes are increasingly used as size and shape, stacking sequence and ply thickness on the open-hole
the structural materials in aerospace, aeronatutic and automotive fields strength [18–25], as well as the progressive failure analysis and dam­
due to their attractive advantages of excellent damage tolerance, eve­ age mechanisms of notched composite laminates [26–33] have been
valated fracture toughness and near-net-shape capacity with respect to reported in literatures. A review on the analytical methods of stress
the conventional laminates [1–11]. In structural applications, notches analysis for notched laminates can be found in Ref. [34]. Zitoune et al.
and bolt connections are inevitable in mechanical fastening and as­ [35] made a comparative study of mechanical performance between the
sembly [12,13]. However, such discontinuities may introduce stress composites made from unidirectional prepreg with drilled and moulded
concentration around the hole edge, and finally lead to strength or life holes, and found that the tensile strenght of specimens with moulded
reduction of composite structures [14,15]. Moreover, when the notched holes was 28% higher than their drilled-hole counterparts. Mohammadi
compoiste components are subjected to external loads, damage mecha­ et al. [36] studied the damage behavior of open-hole laminated com­
nisms such as fiber breakage, matrix cracking and fiber/matrix posites subjected to the tensile tests, and quantified damage mechanisms
debonding are vulnerable to occur at the hole edge during the service using acoustic emission and FEM. Recently, Almeida et al. [37] pre­
period [16,17]. Consequently, the understanding of mechanical prop­ sented a comprehensive set of designed, optimized, manufactured and
erties of notched composites is of particular interest for composite tested open-hole laminated composite specimens under tension, and
design and safety assessment, especially in the aircraft industry. developed a progressive damage model (PDM) to reveal the failure and

* Corresponding author. Key Laboratory of Advanced Textile Composite Materials, Ministry of Education, TianGong University, Tianjin, 300387, China.
** Corresponding author. Key Laboratory of Advanced Textile Composite Materials, Ministry of Education, TianGong University, Tianjin, 300387, China.
E-mail addresses: [email protected] (Y. Zhang), [email protected] (L. Chen).

https://doi.org/10.1016/j.tws.2021.107455
Received 18 October 2020; Received in revised form 15 December 2020; Accepted 9 January 2021
Available online 25 January 2021
0263-8231/© 2021 Elsevier Ltd. All rights reserved.
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 1. 3D woven composites: (a) weave pattern, (b) surface morphology, reconstructed 3D model of (c) 6 mm, (d) 10 mm, and (e) 14 mm notches, (f) schematic of
notched specimens.

damage mechanisms. fully understood yet.

Concerning the open-hole mechanical behaviors of 3D woven com­ The main purpose of the present work is to reveal the influence of the
posites, only limited research work is available [38,39]. Dai et al. [40] notch size on the tensile performance and failure mechanisms of 3D
examined the open-hole tensile behavior of 3D orthogonal and woven composites (3DWC) subjected to the weft loading by experi­
angle-interlock woven composites, and found that the notched tensile mental and numerical methods. The tensile behavior of notched and
strength was less than 17% lower than the unnotched counterpart. Xu unnotched 3DWC specimens were experimentally characterized, and the
et al. [41] experimentally evaluated mechanical properties of 3D full-field strain distributions were recorded using the digital image
orthogonal woven composites with drilled and moulded-in holes. The correlation (DIC). Based on the meso structures of notched plate, a
results showed that the specimens with moulded-in holes presented macro-meso coupling model was proposed for the strength prediction
higher tensile and flexural strengths. Saleh et al. [42] studied the and damage propagation analysis, wherein the failure fractures were
notched response of off-axis 3D woven composites by open-hole tensile obtained by micro-computed tomography (Micro-CT).
tests. Results showed that the reduction of notched strength was less The paper is organized as follows: Section 2 provides the experi­
than 10% compared to the unnotched strength, and the final failure was mental details, including specimen preparation and experimental setup;
progressive failure rather than catastrophic brittle fracture. Song et al. Section 3 describes the establishment of finite element (FE) model;
[43] investigated the fatigue behavior and damage mechanisms of Section 4 introduces the PDM; Section 5 presents the results and dis­
centrally notched 2.5D woven composites at room and elevated tem­ cussion, including experimental analysis and numerical investigation,
peratures. However, the effect of notch size on the open-hole mechanical and conclusions are drawn in Section 6.
properties and damage propagation of 3D woven composites is still not

Table 1
Mechanical properties of component yarn and matrix.
Components Modulus Poisson’s ratio Strength Density

T800 carbon fiber Ef1 Ef2 Gf12 νf12 F1T

294 GPa 19.3 GPa 43.5 GPa 0.29 5500 MPa 1.8 g/cm3
TDE-86 epoxy resin Em Gm νm Xm,T Xm,C
3.55 GPa 1.33 GPa 0.33 80 MPa 241 MPa 1.22 g/cm3

2
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Table 2 lathe machine (She Hong Industrial Co. Ltd), as shown in Fig. 2(a).
Specific parameters of 3D woven composites. The specially developed drill bits made of cemented carbide were
Parameter Value applied for the hole-drilling with the nominal diameters of 6 mm, 10 mm
and 14 mm respectively. The drill bit of 6 mm with a point angle of 110◦
Number of weft layers 5
Warp density (ends/cm) 4 was depicted in Fig. 2(b and c), and the diameter and length of first step
Weft density (picks/cm) 4 were 4.3 mm and 6 mm respectively. For the case of drill bits of 10 mm
Binder density (tows/cm) 4 and 14 mm, the first step diameters were 7.3 mm and 9.8 mm respec­
Thickness (mm) 4 tively, while the first step lengths were 12 mm and 16 mm respectively.
Fiber volume fraction (%) 55
Dry drilling with vacuum cleaning was adopted in the drilling process.
The spindle speed of 5000 rpm and feed rate of 100 mm/min were
2. Materials and experimental details chosen based on the manufacturer’s recommendations, which were
collected from their internal tests for the balance between drilling effi­
2.1. Materials and specimen preparation ciency and hole quality. During the drilling process, two small glass/
epoxy plates were used to sandwich the samples to minimize the damage
The preforms were made of 3D angle-interlock woven reinforcement induced by machining process around the hole rims. The ultimate holes
with a warp-reinforced woven characteristic, which consists of three were inspected, and only minor fiber splitting and short matrix cracks
yarn systems, as shown in Fig. 1(a). The warp and the weft yarns were were observed on the outermost interfaces near the hole edge, as shown
respectively orientated along the X-axis and Y-axis, while the binder in Fig. 1(c)–(e). The rest of the interfaces did not exhibit any noticeable
yarns passed through the in-plane layers and interlaced with the weft damage. It could be assumed that the hole-drilling damage had little
yarns in the Z-axis. The fibers used in this paper was supplied by the effect on the following tests and analysis. Afterwards, aluminum grips of
Shanxi Gangke Carbon Materials Co. Ltd, among which the warp and 80 mm in length were attached to the clamped ends of each specimen.
weft yarns were carbon TG800X–12K × 2 (1000tex), and the binder Fig. 1(f) displays the configurations of N-3DWC specimens.
yarns were TG800HXC-6K (250tex). The preforms were fabricated on
the weaving loom in the Institute of Composite Materials, Tiangong 2.2. Mechanical characterization
University. Afterwards, the preforms were impregnated with TDE-86
epoxy resin following the resin transfer molding (RTM) process. Fig. 3 shows the experimental setup for the tensile tests. The static
Detailed information of the fiber and matrix is listed in Table 1. The final tensile tests in the weft direction were performed on a SHIMADZU AG-
composite plate is shown in Fig. 1(b), and the specific parameters are 250KNE universal testing machine with a 250 kN load cell. The constant
summarized in Table 2. loading rate applied to all specimens was 2 mm/min, and five specimens
Tensile specimens were cut from the moulded composite plates along were tested for each data family. The full-field strain distributions were
the weft direction with the dimension of 300 mm × 36 mm, following recorded using the ARAMIS (GOM, Germany) digital image correlation
ASTM D5766. The thickness for both unnotched 3DWC (UN-3DWC) and (DIC) system (see Fig. 3(a)), and all specimens were painted with the
centrally notched 3DWC (N-3DWC) specimens were 4 mm. random speckle pattern (see Fig. 3(b)).
The drilling operation was performed on a Hartford CNC-500LG

Fig. 2. (a) Drilling lathe machine, (b) physical photo for the drill bit of 6.0 mm, (c) specifications of the drill bit.

3
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 3. Experimental setup: (a) testing machine, and (b) tensile specimens.

exposure time for each radiograph was 2s, and the images were collected
from 1600 radiographs over 360◦ . The damage in the reconstructed
volumes was qualitatively analyzed.

3. Finite element simulation

3.1. Geometric model of 3DWC

Fig. 4 illustrates the cross sections of 3DWC obtained by CT scanning.

Results revealed that the cross-sections of the warp (WP) and binder (B)
yarns were similar to rectangle. The cross-section of inner weft yarn
(IWF) exhibited as hexagon, while that of surface weft yarn (SWF)
approached to rectangle, taking into account the compaction in the
molding process. IPC and SPC respectively denoted the inner and surface
angles of the inclined segment of binder yarn with respect to the X-axis.
Due to the composites molding consolidation, the cross-section of SWF
Fig. 4. CT slices of (a) warp and (b) weft yarn cross-sections.
was flatter than that of IWF, and SPC was slightly larger than IPC. The
measurement results of different yarn parameters are summarized in
Table 3.
2.3. Micro-CT measurement
Based on the yarn trajectories and cross-section statistics reported in
Table 3, a full thickness representative unit cell (RUC) of 3DWC was
The meso structure of 3DWC and failure morphologies of all speci­
developed, as shown in Fig. 5(a). In this model, the 3DWC architecture
mens were examined by Micro-CT using a diondo-d2 X-ray machine
was simplified into a combination of rein matrix (Fig. 5(b)) and yarn
(diondo GmbH, Germany). The source voltage and current were 100 kV
system (Fig. 5(c)). Each unit cell contained 8 warp yarns (Fig. 5(d)), 10
and 100 μA, respectively. Tomogram voxel size was 10 μm/voxel. The

Table 3
Statistical results of characterizing different cross-sections.
No. Height (mm) Width (mm) No. Height (mm) Width (mm) No. Angle ( )
◦

WP1 0.38 1.63 SWF1 0.21 1.69 SPC1 70

WP2 0.44 1.81 SWF2 0.34 1.72 SPC2 60
WP3 0.46 1.77 SWF3 0.35 1.87 SPC3 62
WP4 0.52 1.59 SWF4 0.22 1.84 SPC4 68
Average 0.45 1.70 Average 0.28 1.78 Average 65

B1 0.36 0.78 IWF1 0.60 1.51 IPC1 59

B2 0.42 0.85 IWF2 0.54 1.62 IPC2 63
B3 0.40 0.84 IWF3 0.64 1.56 IPC3 58
B4 0.46 0.77 IWF4 0.70 1.67 IPC4 64
Average 0.41 0.81 Average 0.62 1.59 Average 61

4
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 5. Geometric model of (a) full thickness unit cell of 3DWC, (b) matrix, (c) yarn system, (d) warp yarn, (e) weft yarn, and (f) binder yarn.

Fig. 6. FE-model of (a) quarter of full-thickness unit cell, (b) matrix, (c) warp yarn, (d) weft yarn, and (e) binder yarn.

weft yarns (Fig. 5(e)) and 8 binder yarns (Fig. 5(f)). Additionally, the 3.2. Finite element model of RUC
full-thickness unit cell model exhibited periodic geometric characteris­
tics in the plane. On the basis of the symmetrical characteristic of the The model was generated in the ANSYS environment. Based on the
unit cell, a quarter of geometric model was chosen in the simulation. aforementioned RUC model, a swept method was used to generate the
FE-model of 3DWC. For the sake of clarity, a brief description of the

5
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 7. FE-model of (a) quarter of N-3DWC plate, (b) meso-structured part of D6mm sample, (c) meso-structured part of D10mm sample, and (d) meso-structured
part of D14mm sample.

Table 4
Statistics in the FE-models of the unit cell and different notched specimens.
Number Diameter (mm) Yarn Matrix Homogenized mesh Total

Warp Weft Binder

Element (Solid185) Unit cell 14,344 11,424 7512 20,848 – 54,128

D6 96,690 76,732 50,271 140,255 5400 369,348
D10 88,870 71,144 46,601 129,293 5400 341,308
D14 191,450 62,890 41,053 113,509 5400 300,793
Node Unit cell 17,712 16,905 10,665 25,794 – 48,475
D6 117,588 108,543 68,303 169,789 6954 323,318
D10 108,056 100,647 63,411 156,616 6954 299,145
D14 193,815 88,944 55,861 137,765 6954 264,250

Fig. 8. Boundary conditions of the notched plates.

methodology was herein presented. Firstly, the geometric features of the

unit cell were projected into the plane along the weft direction, gener­
ating the planar meshes. Secondly, the solid meshes were created by
sweeping the planar meshes along the weft direction, and assigned to
generate the mesh set within a quarter of unit cell. Finally, the quarter of Fig. 9. The coordinate system of a strand yarn.
the unit cell model was symmetrically copied and the nodes of symmetry
plane were merged to build the FE-model in ANSYS. 3.3. Finite element model of notched specimen
Fig. 6 shows the FE-model for a quarter of full-thickness unit cell,
wherein the triangular prism elements (Solid185 with six nodes) were Owing to the symmetrical geometry, only a quarter of the notched
adopted to fill in the inclined segment of binder yarn, and the hexahe­ specimen was considered, as depicted in Fig. 7(a). In order to reach
dral elements (Solid185 with eight nodes) were used to mesh the rest of balance between the calculation accuracy and efficiency, the FE-model
the elements. of the notched specimen was divided into two parts: the remote-zone

6
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

homogenized part and the hole-edge meso-structured part. Moreover, ( )2 ( )2 ( )2

the homogenization models were exactly the same regardless of the hole φ1T =
σ1
+α
σ12
+α
σ31
≥1 (1)
dimeter, while the meso-structured models were variable between the
S S
F1T F12 F31
different notched specimens. Yarn compressive failure in direction 1 (σ1 < 0)
Fig. 7(b)–(d) show the FE-models of hole-edge parts for different N-
( )2
3DWC specimens. Due to the complex meso-structures around the hole σ1
φ1C = ≥1 (2)
edge, the mesh stitching and modification technologies in HyperMesh F1C
software were adopted to ensure the mesh quality. The meshing process
Yarn tensile and shear failure in direction 2 and 3 (σ2 + σ 3 ≥ 0),
was described as follows. The array operation was conducted to create 2
unit cells in the length direction, and 3.5 unit cells in the width direction ( )2
(σ2 − σ2 σ3 )
respectively. Meanwhile, the nodes were merged. Subsequently, the φ2(3)T = σ2F+Tσ3 + 23 S 2
(F23 )
(3)
2
unexpected meshes within the hole-region were deleted, followed by the ( )2 ( )2
mesh modification to generate the stepwise hole-edge meshes. + Fσ12S + Fσ31S ≥1
Compared to the traditional methods, the FE meshes in this paper could
12 31

effectively improve the convergence and accuracy of the calculation, Yarn compressive and shear failure in direction 2 and 3 (σ2 + σ 3 < 0),
and precisely analyze the mechanical behaviors of N-3DWC specimens. [( )2 ] ( )
Table 4 gives the numbers of elements and nodes for the unnotched and 1 F2C σ2 + σ3 2
φ2(3)C = C − 1 (σ 2 + σ3 ) +
notched specimens of 6 mm, 10 mm and 14 mm. It could be noted that
S S
F2 2F23 2F23
the total numbers of both elements and nodes decreased with the in­ ( 2 ) ( )2 ( )2 (4)
crease of aperture. + (
σ23 − σ2 σ3
)2 +
σ12
+
σ 31
≥1
S S
FS F12 F31
23
3.4. Boundary conditions
where σij (i, j = 1, 2, 3) are the stress components in the material co­
As shown in Fig.8, in view of the symmetric characteristics, the FE- ordinate system, FiT and FiC (i = 1, 2, 3) denote the tensile and
model was imposed four boundary conditions throughout the simula­ compressive strengths in the 1, 2 and 3 direction of yarn, respectively.
tion: symmetry constraints, a fixed constraint at one vertex of X-axis and S
F12 S
, F23 S
and F31 express 12, 23 and 31 shear strength, respectively. α is
Z-axis, node coupling at the contact face between the remote-zone ho­ the contribution factor. In this paper, α is taken as 1, being consistent
mogenized part and hole-edge meso-structured part, and normal tensile with that in Ref. [47].
load at end face. The criteria of matrix damage initiation based on von Mises criterion
are given by:

( )2 ( )2 ( )2 ( )
σm,11 − σm,22 + σ m,22 − σ m,33 + σ m,11 − σm,33 + 6 σ 2m,12 + σ2m,13 + σ2m,23
φm,T = ≥ 1, ​
2
2Xm,T (5)

σm,11 + σm,22 + σ m,33 ≥ 0

( )2 ( )2 ( )2 ( )
σm,11 − σm,22 + σm,22 − σ m,33 + σ m,11 − σ m,33 + 6 σ 2m,12 + σ2m,13 + σ2m,23
φm,C = ≥ 1,
2
2Xm,C (6)

σm,11 + σm,22 + σ m,33 < 0

4. Progressive damage model

where φm is the loading function of resin matrix, and σm,i is the stress of
In order to predict the notched strength of N-3DWC, a progressive matrix. Xm,T and Xm,C are the tensile and compressive strengths of ma­
damage model is used and written as a subroutine in ANSYS linked to the trix, respectively.
software package in this paper. Since Hashin failure criteria have been
demonstrated to work well with the 3D textile composites [44–46],
Hashin failure criteria are introduced to estimate the damage initiation 4.2. Damage evolvement model
of impregnated tows, and the material coordinate system referred to is
shown schematically in Fig. 9. In addition, von Mises criterion is When the yarn damage is initiated, the development of damage in­
implemented on the pure resin matrix. dicator for woven composite material will be controlled by a damage
evolution law, which is related to the element strains, material fracture
4.1. Element initial failure criteria energy density and element characteristic length. In order to reduce the
mesh dependence caused by local damage, the crack band model pro­
Yarn tensile failure in direction 1 (σ 1 ≥ 0), posed by Bažant and Oh [48] is applied. The element characteristic
length l is defined as the cube root of element volume. When the com­
ponents of woven composites are damaged, the element dissipated en­
ergy is equal to its elastic energy, which can be expressed as [49].

7
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 10. Loading stress-strain response of different specimens: (a) unnotched specimen, (b) D6mm specimen, (c) D10mm specimen, and (d) D14mm specimen.

1 If
εI,f σ I,f l3 = GI l2 (7) Ii
where Xeq and Xeq are the equivalent displacement of the initiation and
2
full damage, respectively. Moreover, according to the crack band theory,
where GI ,εI,f , σ I,f are the fracture energy density, equivalent peak stress the energy dissipated per unit volume gM can be determined by Refs.
and equivalent peak strain of failure mode I, respectively. [50,51]:
Then, the element equivalent displacement can be defined by
gM = GlM ​
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√ ∫ ∞ ∂G
˙ (10)
√ ∑ 3 = ∂dM d M dt
(8)
0
Xeq = l√ε2ii + αi,k
i,k
ε2ij (i = 1, 2, 3; k = T, C) (M=1T, ​ 1C, ​ 2T, ​ 2C, ​ 3T, ​ 3C;m,T;mC)
j=1,j∕
=i
where GM and G are the energy release rate and Helmholtz free energy,
Here, the value of αi,k is equal to that of Eq. (1). respectively. The Helmholtz free energy for yarn and matrix can be
The damage evolution law of damage variable dI is then expressed by written as:
the following equation:
1 n!
If
(XeqI − XeqIi )
X
Gyarn = σ : ε
dI = XeqI 2 r!(n − r)!
(9)
If
eq (Xeq − Xeq )
Ii

(I = 1T, 1C, 2T, 2C, 3T, 3C) σ 21 σ 22 σ 23 ν12

= + + − σ1 σ2
2(1 − d1 )E1 2(1 − d2 )E2 2(1 − d3 )E3 E1

ν13 ν23 σ 212 σ223 σ231

Table 5 − σ1 σ3 − σ2 σ3 + + +
E1 E2 2(1 − d4 )G12 2(1 − d5 )G23 2(1 − d6 )G31
Comparison of experimental and predicted results.
(11)
Property Diameter (mm) Experiment Simulation Error
( ) ( )
Average C.V Gm = Gm εem , dm + Gpm εpm
e

Stiffness (GPa) 0 67.67 4.33% 66.97 1.04%

6 66.18 5.04% 68.79 3.94% σ 2m,11 + σ 2m,22 + σ2m,33
νm ( )
10 66.78 3.56% 68.90 3.17% = − σ m,11 σ m,22 + σ m,22 σm,33 + σ m,33 σm,11 (12)
2(1 − dm )Em Em
14 66.89 4.54% 70.42 5.28%
Strength (MPa) 0 841.93 5.20% 827.02 1.77% ( )
6 520.30 7.84% 551.10 5.92% (1 + νm ) σ 2m,12 + σ2m,23 + σ2m,31 ( )
10 472.45 6.83% 487.18 3.12% + + Gpm εpm
2(1 − dm )Em
14 461.87 1. 96% 440.62 4.60%
In Eq. (12), Gem and Gpm represent the matrix elastic and plastic

8
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 11. Variation of Y-direction strain field of UN-3DWC.

contributions, respectively. well with the experimental curves. The catastrophically abrupt failure
was also obtained in the curves of N-3DWC specimens, as shown in
Fig. 10(b)–(d), indicating the brittle failure mode. As the aperture
4.3. Damage model
increased, the peak values and the nonlinear characteristics of N-3DWC
specimens decreased, and the deviation of the experimental curves
Since the yarn and resin matrix are regarded as transversal isotropic
reduced as well. This was because that the more serious stress concen­
and isotropic, respectively, the constitutive equation for yarn can be
tration emerged in the notched specimens of smaller aperture, which
characterized by:
involved much more critical damage.
σ = C(D)ε (13) Comparing the stress-strain curves of UN-3DWC and N-3DWC spec­
imens, it could be found that the curves of unnotched specimen
where C(D) is the stiffness matrix of damaged configuration, and can be exhibited smoother characteristic and smaller deviation near the peak
expressed as follows: point, which were marked by the blue dotted line in Fig.10. With respect
⎡ ⎤ to the curves of different notched specimens, they were all basically
⎢ b21 Q11 b1 b2 Q12 b1 b3 Q13 0 0 0 ⎥ linear in the initial loading stage. Afterwards, the experimental stress-
⎢
⎢
⎥
⎥ strain response followed the nonlinear increase, which was also
⎢ b22 Q22 b2 b3 Q23 0 0 0 ⎥
⎢ ⎥ observed from simulated results. This was due to the fiber and matrix
⎢ b23 Q33 0 0 0 ⎥
C(D) = ⎢ ⎥ (14) damages emerged in the loading process. Besides, the nonlinear char­
⎢ ⎥
⎢
⎢
sym b12 Q44 0 0 ⎥
⎥ acteristics of UN-3DWC specimens were not clear in comparison to the
⎢
⎢ b13 Q55 0 ⎥
⎥ N-3DWC ones, which also proved the instantaneous fracture mentioned
⎣ b23 Q66 ⎦ above.
Table 5 summarizes the results of stiffness and strength for different
specimens. The maximum coefficient of variation (C.V) of the experi­
where mental values for stiffness and strength was 5.04% and 7.84% respec­
⎧ tively, demonstrating excellent repeatability. The maximum prediction
⎨ bi = 1 − dik
⎪
( ( )) errors of stiffness and strength were 5.28% and 5.92% respectively,
2(1 − dik ) 1 − djk 2 ​ (i ∕
= j; ​ i, j = 1, 2, 3; ​ k = T, C) (15)
⎪
⎩ bij = indicating that the proposed model could essentially predict the open-
2 − dik − djk
hole tensile properties.
Regarding the tensile stiffness of different notched specimens, it was
Qij (i, j = 1, 2, 3) is the component of undamaged stiffness matrix.
found that the hole size had little effect on the stiffness of N-3DWC, and
the difference between the maximum and minimum stiffness was less
5. Results and discussions than 5%. It could be concluded that the stiffness did not change with the
aperture. This was mainly because that the interlacing structure of the
5.1. Tensile stress-strain response material was not changed, so the stiffness remained unchanged after the
drilling operation. However, when the notch diameter increased from 6
Fig. 10 illustrates the tensile stress-strain response of different mm to 14 mm, the average tensile strength decreased nonlinearly from
specimens obtained by the experimental and predicted methods. It was
520.30 MPa to 461.87 MPa, reducing by 11.2%, which was a result of
clear from Fig. 10(a) that the tensile stress increased linearly to the peak the reduction in the effective load-bearing area of N-3DWC specimens.
point, and a sudden drop was observed when the maximum stress was
achieved. During the entire loading process, the simulated curve agreed

9
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 12. Variation of Y-direction strain field of D6mm specimen.

5.2. Variations of strain field concentration. Here, δx and δy were introduced to describe the hole-
edge damage before the final fracture. It could be seen from Figs. 12
Figs. 11–14 present the tensile strain distributions of the UN-3DWC (e), 13(e) and 14(e) that δx(D6) > δx(D10) > δx(D14), while δy(D6) < δy
and different N-3DWC specimens, and significant differences of the (D10) < δy(D14). Thus, it could be inferred that the stress concentration
strain distribution could be observed between them. As shown in Fig.11, area increased while the stress concentration degree decreased as the
the tensile strain field of unnotched specimen changed uniformly aperture size enlarged.
throughout the loading process, accompanied with some high-strain The UN-3DWC specimen broke suddenly under the continuous
spots at the weaving points due to the stress concentration. In loading, and the release of strain energy was manifested as the fiber
contrast, obvious stress concentration effect near the hole-edge could be fractured at the geometric boundary. It was noteworthy that the surface
clearly noted in Figs. 12–14. Furthermore, the strain at the weaving high-strain spots disappeared at the moment just before the specimen
points within the hole-edge meso-structured part were higher than those catastrophically fractured. Although the initial damage occurred at the
of the remote-zone part, and the low-strain spots sporadically distrib­ free edge, no obvious damage propagation process was observed on the
uted along the length-wise direction. specimen surface. For N-3DWC specimens (Figs. 12–14), the stress
As the aperture increased, the low-strain regions became wider and concentration effect always existed at the hole edge. During the loading
shorter (Figs. 12–14), and their distribution widths were approximately process, the damage initially happened on the left and right sides of the
equal to the hole diameters. This was because that the lager aperture central hole, and accumulated gradually near the hole edge up to the
induced the smaller geometric gradient, and resulted in the lower stress final failure.
gradient accordingly. The stress concentration for three N-3DWC spec­
imens were vertically distributed at the left and right sides of the hole,
where the initial cracks happened. With the increase of tensile loads, 5.3. Initial equivalent stress field distribution
these cracks propagated further and finally broke at the points of stress
Fig.15 shows the initial equivalent stress field distribution of

Fig. 13. Variation of Y-direction strain field of D10mm specimen.

10
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 14. Variation of Y-direction strain field of D14mm specimen.

different specimens subjected to the weft loading, wherein the loading direction due to the yarns’ interlacing structure, which induced the
strain was 0.1%. In Fig.15(a), it could be seen that the stress distribution stress concentration within a small region. It was inferred that the
of unit cell of 3DWC was not uniform. The weft yarn was the main load- centrally circular notch had an influence on the mechanical performance
bearing component, and the maximum equivalent stress was 193.01 of N-3DWC. Moreover, it could also be seen that the initial stress con­
MPa. For the N-3DWC specimens shown in Fig. 15(b)–(d), the maximum centration points were distributed at the left and right sides of the holes
equivalent stress was about 1343.12 MPa of D6mm, which was nearly along the loading direction, being consistent with the DIC measurements
seven times that of unit cell of 3DWC. The maximum equivalent stresses shown in Figs. 12–14.
of D10mm and D14mm were 1067.73 MPa and 905.14 MPa respec­ As shown in Fig. 15(e), the maximum equivalent stress of N-3DWC
tively, proving that the stress intensity decreased with the increasing specimens decreased nonlinearly with the increasing aperture. This was
notch size. Besides, the hole-edge stress concentration was merely mainly because that as the aperture increased, the stress gradient in the
distributed on the weft yarn, and the stress level in the weft yarn was stress concentration region diminished, but the stress concentration area
clearly greater than the other components. This was because that the enlarged (see Fig. 15(f)).
weft yarn was the load-carrying component. The hole-edge stress con­
centration was restricted within one weft’s width along the weft

Fig. 15. Initial equivalent stress field distribution with the loading strain of 0.1%: (a) unit cell of 3DWC without matrix, (b) D6mm, (c) D10mm, (d) D14mm, (e)
relationship between the aperture and maximum equivalent stress, and (f) schematic diagram of stress concentration degree and area.

11
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 16. Progressive damage processes of unit cell of 3DWC.

Fig. 17. Progressive damage processes of D6mm specimen.

12
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 18. Progressive damage processes of D10mm specimen.

Fig. 19. Progressive damage processes of D14mm specimen.

5.4. Damage propagation process analysis transverse damage occurred at the edge of warp and binder yarns
simultaneously. The damage localization of each constituent yarn could
The progressive damage process of unit cell of 3DWC (remote-zone be divided into two regions: the intersections of three yarn systems and
as well) under the weft loading are presented in Fig. 16. As the primary the weaving points between binder and weft yarns. This was mainly
load-bearing component, weft yarn primarily experienced the longitu­ because that these zones were the stress concentration areas due to the
dinal damage, while that of warp and binder yarns mainly suffered from intrinsic characteristics of 3DWC architecture. Moreover, the damage in
the transverse damage. During the whole damage propagation process, matrix region was found mainly surrounding the yarn damage areas,
only a small amount of transverse damage occurred at the warp edge, rather than existed separately. This indicated that the damage of matrix
while the matrix around the warp was almost completely damaged. The and yarn were coupled. In other words, when one constituent material
interlacement of binder and weft yarns were responsible for the critical was damaged, the strain energy release within the region would transfer
transverse damage. This was because that the binder and weft yarns to another adjacent material, inducing the coupled damage.
were interwoven to form an integral interlock architecture, and were In the process of damage propagation, it could be clearly seen that all
less susceptible to debonding in comparison to the warp yarn. kinds of damage propagated along the weft and warp directions at the
At the beginning of loading, the damage initiated at the region of same time. After the damage reaching a certain level, the damaged el­
high internal stress, and then propagated along the weft direction, being ements remained stable up to the final fracture.
independent of the macro geometry. It could be noted that slight lon­ When the final damage was achieved, the longitudinal damage
gitudinal damage appeared at the weft edge, while a small amount of penetrated all over the weft yarn, and the whole architecture completely

13
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

damaged region compared to the other two specimens. It could be seen

that the damaged region of D6mm specimen accounted for more than
half of the total area of its homologous meso-model, while that of
D14mm specimen was clearly less than half of the total area (see
Fig. 21), indicating that the damage of notched specimens diminished
with the increase of aperture.
Moreover, for the different N-3DWC specimens, all damaged ele­
ments started at the hole-edge stress concentration region and then
propagated in the weft and warp directions. The involved damage
modes were longitudinal damage of weft yarn, transverse damage of
warp and binder yarns and matrix cracks. However, the damage prop­
agation paths of three specimens were obviously variable. As shown in
Fig. 21, when the aperture increased from 6 mm to 14 mm, the contour
of damaged region changed from convex to concave owing to the
decreasing stress concentration effect. In the case of D6mm specimen,
the severe stress concentration induced the convex-shaped damage re­
gion with longer propagation path. However, as the aperture increased,
the stress concentration weakened, and the contours of damaged region
for D10mm and D14mm specimens tended to be concave-shaped.
Fig. 20. Schematic diagram of the tip points.
5.5. Fracture morphologies analysis
lost the load-bearing capacity. The stress-strain curve showed a sudden
drop after the maximum stress (see Fig. 10 (a)). Meanwhile, the trans­ Fig. 22 depicts the tensile fractures of unnotched and notched
verse damage of binder yarn and matrix damage were spread specimens. The failure modes mainly included the weft yarn pull-out
throughout the cross-sections, however, the transverse damage of warp and longitudinal fracture, warp yarn transversal debonding and
yarn rarely developed. In addition, the damage mode of unit cell of cracks, binder yarn transversal cracks and matrix cracks. For the
3DWC exhibited periodic characteristics, which was consistent with the unnotched specimen, the principal damage modes were fiber pull-out
interweaving structure. and longitudinal rupture at the fracture sites. This was because that
Figs. 17–19 present the damage propagation for N-3DWC specimens the weft yarn was principal load-bearing component under the weft
of different apertures subjected to the weft loading. The initial damage loading, and there was none local stress concentration caused by the
characteristic observed was yarn longitudinal damage at the hole-edge inherent woven structure or micro defects. For the notched specimens,
stress concentration areas, accompanied by the surrounding matrix the circular hole resulted in obvious stress concentration at the hole
cracks. As the tensile load continued, the damaged elements of N-3DWC edge. It could be found that the tensile fractures appeared at the left and
specimens gradually increased, and extended to the plate free-edge. right sides of hole for three N-3DWC specimens, being consistent with
During the damage evolution process, the damaged elements of weft the DIC and simulation results mentioned above. As the tensile load
longitudinal damage reduced with the increase of aperture, while those increased, the stress concentration became intensive, and induced the
of matrix damage decreased too. In the view of energy release, the hole-edge damage such as fiber pull-out. Besides, from the DIC strain
smaller aperture resulted in more obvious stress concentration, which maps and fracture images, it could be found that the fiber fracture of UN-
enabled the energy release to be achieved by the damaged elements of 3DWC specimen happened at the geometric boundary. However, the N-
high strain energy. Conversely, in the case of large aperture, only low 3DWC specimens of 6 mm, 10 mm and 14 mm notches did not exhibit
strain energy was required. Due to the hole drilling, the cut-off warp and such fracture behavior. The main cause was that the strain energy of
weft yarns produced tip points at the hole edge as shown in Fig. 20, notched specimens generated during the loading process was released
where it was likely to generate the stress concentration region. The by the hole-edge stress concentration.
initial damage basically occurred at the hole-edge stress concentration Comparing the experimental fracture morphologies of different
region, but they were deviated from the maximum stress point. notched specimens (Fig. 22(b)–(d)), it could be noted that the weft yarn
With respect to the ultimate failure, it could be seen that the damage pull-out, transversal cracking of warp and binder yarns within the hole-
of D6mm specimen was the severest, and that of D14mm was the least. edge region were severer than the remote-zone. This damage behavior
The main cause was that the D6mm specimen retained more complete was much more visible in the case of smaller aperture, demonstrating
load-carrying yarns, and presented more damaged elements and larger the severe stress concentration. Additionally, as the aperture increased,

Fig. 21. Schematic illustration of damage propagation path for different notched specimens.

14
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 22. The experimental fractures of different specimens.

Fig. 23. Comparison of simulated and experimental fractures of unnotched specimen.

15
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

Fig. 24. Comparison of simulated and experimental fractures of D6mm specimen.

Fig. 25. Comparison of simulated and experimental fractures of D10mm specimen.

Fig. 26. Comparison of simulated and experimental fractures of D14mm specimen.

16
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

the fracture surface became increasingly uniform, and the failure mode the work reported in this paper.
of weft yarn changed from fiber pull-out to breakage due to the
decreasing stress concentration degree. Acknowledgments
The comparison of simulated and experimental failure modes of UN-
3DWC and N-3DWC specimens are shown in Figs. 23–26. The main This work was supported by the National Natural Science Foundation
predicted characteristics of all specimens were longitudinal fracture of of China (grant numbers 11802203,11802204), Tianjin Major Science
weft yarn, transverse cracking of warp and binder yarns and matrix and Technology Project (grant number 18ZXJMTG00190), Major Sci­
cracks. As shown in Fig. 23, apart from the weft pull-out at the specimen ence and Technology Project of Shanxi Province (grant number
edge, fracture surface of UN-3DWC specimen was relatively smooth. In 20181102022), Tianjin Municipal Natural Science Foundation (grant
the simulated results, the weft longitudinal fracture was parallel to the numbers18JCQNJC73200, 18JCYBJC87300) and Program for Innova­
loading direction. The warp yarn was barely damaged and merely tion Team in Colleges of Tianjin (grant number TD13-5043).
exhibited tiny transverse cracking on the yarn edge, while the binder
yarn induced extensive transverse fragmentation. From the simulated References
and experimental results, it could be seen that the fractures of matrix
were almost in the same position. [1] S.V. Lomov, A.E. Bogdanovich, D.S. Ivanov, D. Mungalov, M. Karahan, I. Verpoest,
A comparative study of tensile properties of non-crimp 3D orthogonal weave and
Figs. 24–26 present the simulated and experimental fractures of multi-layer plain weave E-glass composites. Part 1: materials, methods and
different N-3DWC specimens. Different from the damage mode in UN- principal results, Composites Part A 40 (8) (2009) 1134–1143.
3DWC, the weft damage in N-3DWC was almost in the domain-wide [2] M. Ansar, W. Xinwei, Z. Chouwei, Modeling strategies of 3D woven composites: a
review, Compos. Struct. 93 (8) (2011) 1947–1963.
scale, especially for the weft yarn near the hole edge. Moreover, the [3] A. Hallal, R. Younes, F. Fardoun, S. Nehme, Improved analytical model to predict
warp yarn was extensively damaged, and a small range of longitudinal the effective elastic properties of 2.5D interlock woven fabrics composite, Compos.
damage occurred due to the extrusion. Struct. 94 (10) (2012) 3009–3028.
[4] N. Vernet, F. Trochu, Analysis and modeling of 3D Interlock fabric compaction
behavior, Compos. Appl. Sci. Manuf. 80 (2016) 182–193.
6. Conclusions [5] T. Gereke, C. Cherif, A review of numerical models for 3D woven composite
reinforcements, Compos. Struct. 209 (2019) 60–66.
[6] T. Zheng, L. Guo, J. Huang, G. Liu, A novel mesoscopic progressive damage model
This work focused on investigating the tensile performance and
for 3D angle-interlock woven composites, Compos. Sci. Technol. 185 (2020)
damage mechanisms of centrally notched 3D woven composites under 107894.
the weft loading. A macro-meso coupling model of the N-3DWC was [7] W. Cao, J. Zhang, B. Sun, B. Gu, X-ray tomography and numerical study on low-
proposed, which could be applied to predict the tensile properties and velocity impact damages of three-dimensional angle-interlock woven composites,
Compos. Struct. 230 (2019) 111525.
damage propagation process of the N-3DWC with different notch sizes. [8] Q. Guo, Y. Zhang, R. Guo, M. Ma, L. Chen, Influences of weave parameters on the
Tensile strength and stiffness were experimentally evaluated and pre­ mechanical behavior and fracture mechanisms of multidirectional angle-interlock
dicted by finite element simulation, wherein the full-field strain distri­ 3D woven composites, Mater. Today Commun. 23 (2020) 100886.
[9] Z.G. Li, D.S. Li, H. Zhu, Z.X. Guo, L. Jiang, Mechanical properties prediction of 3D
butions were recorded with the DIC system, and fracture morphologies angle-interlock woven composites by finite element modeling method, Mater.
were obtained by Micro-CT. Based on the experimental and numerical Today Commun. 22 (2020) 100769.
investigations, the following conclusions can be drawn: [10] G. Sun, X. Guo, S. Li, D. Ruan, Q. Li, Comparative study on aluminum/GFRP/CFRP
tubes for oblique lateral crushing, Thin-Walled Struct. 152 (2020) 106420.
[11] Z. Wu, S. Li, Z. Pan, X. Hu, Off-axial angle sensitivity of 3D orthogonal woven
(1) The stress-strain curves of all specimens presented to be nonlinear composite failure behavior subjected to in-plane compressive loading, Thin-Walled
in general, however, they followed the linear increase in the Struct. 157 (2020) 107149.
[12] A. Haeger, M. Grudenik, M.J. Hoffmann, V. Knoblauch, Effect of drilling-induced
initial loading stage. The notch size had little effect on the tensile damage on the open hole flexural fatigue of carbon/epoxy composites, Compos.
stiffness of N-3DWC. However, the tensile strength decreased Struct. 215 (2019) 238–248.
with the increase of aperture. [13] U.A. Khashaba, A.I. Khdair, Open hole compressive elastic and strength analysis of
CFRE composites for aerospace applications, Aero. Sci. Technol. 60 (2017) 96–107.
(2) The tensile strain field of UN-3DWC distributed uniformly
[14] M.A. Caminero, M. Lopez-Pedrosa, C. Pinna, C. Soutis, Damage monitoring and
throughout the loading process, while the stress concentration of analysis of composite laminates with an open hole and adhesively bonded repairs
N-3DWC mostly occurred at the hole edge, and the distribution using digital image correlation, Compos. B Eng. 53 (2013) 76–91.
width was approximately equal to the notch diameter. [15] R. Mounien, C. Fagiano, P. Paulmier, B. Tranquart, F.X. Irisarri, Experimental
characterization of the bearing behavior of 3D woven composites, Compos. B Eng.
(3) The initial equivalent stress of N-3DWC was about seven times 116 (2017) 369–376.
that of UN-3DWC, and the hole-edge stress concentration was [16] O.J. Nixon-Pearson, S.R. Hallett, P.J. Withers, J. Rouse, Damage development in
restricted within one weft’s width along the weft direction. open-hole composite specimens in fatigue. Part 1: experimental investigation,
Compos. Struct. 106 (2013) 882–889.
(4) The fiber pull-out and longitudinal fracture were the principal [17] M. Azadi, H. Sayar, A. Ghasemi-Ghalebahman, S.M. Jafari, Tensile loading rate
damage modes in UN-3DWC. For N-3DWC, the flatness of frac­ effect on mechanical properties and failure mechanisms in open-hole carbon fiber
tured surface was improved as the aperture increased. Accord­ reinforced polymer composites by acoustic emission approach, Compos. B Eng. 158
(2019) 448–458.
ingly, the main failure mode of weft yarn changed from fiber pull- [18] B.G. Green, M.R. Wisnom, S.R. Hallett, An experimental investigation into the
out to breakage, and the contour of the damaged region changed tensile strength scaling of notched composites, Composites Part A 38 (3) (2007)
from to convex to concave. 867–878.
[19] G.H. Erçin, P.P. Camanho, J. Xavier, G. Catalanotti, S. Mahdi, P. Linde, Size effects
on the tensile and compressive failure of notched composite laminates, Compos.
Credit author statement Struct. 96 (2013) 736–744.
[20] M. Ridha, C.H. Wang, B.Y. Chen, T.E. Tay, Modelling complex progressive failure
in notched composite laminates with varying sizes and stacking sequences,
Qiwei Guo: Conceptualization, Methodology, Formal analysis,
Composites Part A 58 (2014) 16–23.
Investigation, Writing – original draft; Yifan Zhang: Supervision, [21] Z. Su, C. Xie, Y. Tang, Stress distribution analysis and optimization for composite
Writing – review & editing; Qingtao Lv: Resources, Data curation; laminate containing hole of different shapes, Aero. Sci. Technol. 76 (2018)
466–470.
Xiaolun Sun: Visualization; Ming Ma: Visualization; Li Chen: Valida­
[22] T. Shafighfard, E. Demir, M. Yildiz, Design of fiber-reinforced variable-stiffness
tion, Supervision. composites for different open-hole geometries with fiber continuity and curvature
constraints, Compos. Struct. 226 (2019) 111280.
Declaration of competing interest [23] H. Liu, B.G. Falzon, S. Li, W. Tan, J. Liu, H. Chai, B.R.K. Blackman, J.P. Dear,
Compressive failure of woven fabric reinforced thermoplastic composites with an
open-hole: an experimental and numerical study, Compos. Struct. 213 (2019)
The authors declare that they have no known competing financial 108–117.
interests or personal relationships that could have appeared to influence

17
Q. Guo et al. Thin-Walled Structures 161 (2021) 107455

[24] S. Zhou, J. Zhang, Y. Sun, K. Tian, Experimental and numerical investigation of [38] Y. Zhang, Q. Guo, X. Chen, J. Xie, L. Chen, Effect of apertures on tensile property of
open hole carbon fiber composite laminates under compression with three different warp-reinforced 2.5D woven composites notched plates, Compos. Struct. 252
stacking sequences, J. Mater. Res.Technol. 8 (3) (2019) 2957–2968. (2020) 112693.
[25] G. Sun, L. Wang, D. Chen, Q. Luo, Tensile performance of basalt fiber composites [39] S. Dai, P.R. Cunningham, Multi-scale damage modelling of 3D woven composites
with open circular holes and straight notches, IJMS 176 (2020) 105517. under uni-axial tension, Compos. Struct. 142 (2016) 298–312.
[26] H. Bao, G. Liu, Progressive failure analysis on scaled open-hole tensile composite [40] S. Dai, P.R. Cunningham, S. Marshall, C. Silva, Open hole quasi-static and fatigue
laminates, Compos. Struct. 150 (2016) 173–180. characterisation of 3D woven composites, Compos. Struct. 131 (2015) 765–774.
[27] R. Higuchi, T. Okabe, T. Nagashima, Numerical simulation of progressive damage [41] F. Xu, L. Sun, L. Zhu, S. Yang, D. Hui, Y. Qiu, X-ray 3D microscopy analysis of
and failure in composite laminates using XFEM/CZM coupled approach, Compos. fracture mechanisms for 3D orthogonal woven E-glass/epoxy composites with
Appl. Sci. Manuf. 95 (2017) 197–207. drilled and moulded-in holes, Compos. B Eng. 133 (2018) 193–202.
[28] P.L. Rosendahl, P. Weissgraeber, N. Stein, W. Becker, Asymmetric crack onset at [42] M.N. Saleh, Y. Wang, A. Yudhanto, A. Joesbury, P. Potluri, G. Lubineau, C. Soutis,
open-holes under tensile and in-plane bending loading, IJSS 113 (2017) 10–23. Investigating the potential of using off-Axis 3D woven composites in composite
[29] H. Sayar, M. Azadi, A. Ghasemi-Ghalebahman, S.M. Jafari, Clustering effect on joints ’ applications, ApCM 24 (2) (2017) 377–396.
damage mechanisms in open-hole laminated carbon/epoxy composite under [43] J. Song, W. Wen, H. Cui, Y. Wang, Y. Lu, W. Long, L. Li, Warp direction fatigue
constant tensile loading rate, using acoustic emission, Compos. Struct. 204 (2018) behavior and damage mechanisms of centrally notched 2.5D woven composites at
1–11. room and elevated temperatures, Compos. Sci. Technol. 182 (2019) 107769.
[30] A. Khechai, A. Tati, B. Guerira, A. Guettala, P.M. Mohite, Strength degradation and [44] H. Zhu, D.-S. Li, L. Jiang, Mesoscale progressive damage and strength analysis of
stress analysis of composite plates with circular, square and rectangular notches three-dimensional braided composites under tension, EnFM 237 (2020) 107221.
using digital image correlation, Compos. Struct. 185 (2018) 699–715. [45] F. Guo-dong, L. Jun, W. Bao-lai, Progressive damage and nonlinear analysis of 3D
[31] S. Zhou, Y. Sun, B.Y. Chen, T.E. Tay, Progressive damage simulation of open-hole four-directional braided composites under unidirectional tension, Compos. Struct.
composite laminates under compression based on different failure criteria, 89 (1) (2009) 126–133.
J. Compos. Mater. 51 (9) (2017) 1239–1251. [46] C. He, J. Ge, B. Zhang, J. Gao, S. Zhong, W.K. Liu, D. Fang, A hierarchical
[32] B. Aidi, S.W. Case, Experimental and numerical analysis of notched composites multiscale model for the elastic-plastic damage behavior of 3D braided composites
under tension loading, ApCM 22 (6) (2015) 837–855. at high temperature, Compos. Sci. Technol. 196 (2020) 108230.
[33] J. Koord, J.L. Stüven, E. Petersen, O. Völkerink, C. Hühne, Investigation of exact [47] Z. Lu, Y. Zhou, Z. Yang, Q. Liu, Multi-scale finite element analysis of 2.5D woven
analytical solutions for circular notched composite laminates under tensile loading, fabric composites under on-axis and off-axis tension, Comput. Mater. Sci. 79 (2013)
Compos. Struct. 243 (2020) 112180. 485–494.
[34] R. Sevenois, S. Koussios, Analytic methods for stress analysis of two-dimensional [48] Z.P. Bažant, B.H. Oh, Crack band theory for fracture of concrete, Matériaux et
flat anisotropic plates with notches: an overview, ApMRv 66 (2014), 060802. Construct. 16 (3) (1983) 155–177.
[35] R. Zitoune, L. Crouzeix, F. Collombet, T. Tamine, Y.H. Grunevald, Behaviour of [49] G. Fang, J. Liang, Q. Lu, B. Wang, Y. Wang, Investigation on the compressive
composite plates with drilled and moulded hole under tensile load, Compos. Struct. properties of the three dimensional four-directional braided composites, Compos.
93 (9) (2011) 2384–2391. Struct. 93 (2) (2011) 392–405.
[36] R. Mohammadi, M.A. Najafabadi, M. Saeedifar, J. Yousefi, G. Minak, Correlation of [50] C. He, J. Ge, D. Qi, J. Gao, Y. Chen, J. Liang, D. Fang, A multiscale elasto-plastic
acoustic emission with finite element predicted damages in open-hole tensile damage model for the nonlinear behavior of 3D braided composites, Compos. Sci.
laminated composites, Compos. B Eng. 108 (2017) 427–435. Technol. 171 (2019) 21–33.
[37] J.H.S. Almeida, L. Bittrich, A. Spickenheuer, Improving the open-hole tension [51] J. Ge, C. He, J. Liang, Y. Chen, D. Fang, A coupled elastic-plastic damage model for
characteristics with variable-axial composite laminates: optimization, progressive the mechanical behavior of three-dimensional (3D) braided composites, Compos.
damage modeling and experimental observations, Compos. Sci. Technol. 185 Sci. Technol. 157 (2018) 86–98.
(2020) 107889.

18