Article

International Journal of Damage

Mechanics
0(0) 1–16
Progressive failure analysis ! The Author(s) 2017
Reprints and permissions:
of fiber-reinforced laminated sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1056789517715088
composites containing a hole journals.sagepub.com/home/ijd

Hadi Bakhshan1, Ali Afrouzian2, Hamed Ahmadi2

and Mehrnoosh Taghavimehr3

Abstract
The present work aims to obtain failure loads for open-hole unidirectional composite plates under tensile
loading. For this purpose, a user-defined material model in the finite element analysis package, ABAQUS,
was developed to predict the failure load of the open-hole composite laminates using progressive failure
analysis. Hashin and modified Yamanda-Sun’s failure criteria with complete and Camanho’s material
degradation model are studied. In order to achieve the most accurate predictions, the influence of failure
criteria and property degradation rules are investigated and failure loads and failure modes of the com-
posites are compared with the same experimental test results from literature. A good agreement
between experimental results and numerical predictions was observed.

Keywords
Progressive failure, user-defined material model, failure criteria, ABAQUS, fiber-reinforced composite

Introduction
Fiber-reinforced composite materials’ properties such as high stiffness and strength, relative light
weight, and also long lifetimes compared to metallic materials make them a proper selection for
using in the joints in aerospace structures (Afrouzian et al., 2017). In particular, the mechanical
joints need fastener hole that creates local damage and stress concentration, which contributes to
loss of structural strength. In addition, designing of composite joints with fastener hole, failure, and
damage analysis around the hole are crucial. As a result, particular attention must be paid to
investigate the failure strength and failure modes of these kinds of structure.

1
School of Automotive Engineering, Iran University of Science and Technology, Iran
2
Impact and Composite Laboratory, Faculty of Mechanical Engineering, Tarbiat Modares University, Iran
3
Polymer Engineering Group, Faculty of Chemical Engineering, Tarbiat Modares University, Iran
Corresponding author:
Hamed Ahmadi, Impact and Composite Laboratory, Department of Mechanical Engineering, Tarbiat Modares University, Nasr
Bridge, Tehran 1411713116, Iran.
Email: h_ahmadi@modares.ac.ir
2 International Journal of Damage Mechanics 0(0)

The joint design has a special importance in fiber-reinforced composite structures for two
main reasons: (1) joints are often the weakest parts in a composite structure and (2) the compos-
ite materials do not possess the forgiven features of ductile metals, namely, their capability to
redistribute local high stresses by yielding (Khashaba et al., 2013). Despite conventional metallic
materials, failure pattern in composite structures is more complicated and consists of local damage
such as matrix failure, fiber breakage, fiber matrix de-bonding, and delamination, which eventually
cause the ultimate structure failure (Ubaid et al., 2014). The first three failure modes are called
intra-laminar and the last is called inter-laminar failure (Mokhtari et al., 2015). The mechanically
fastened joints fail by three basic mechanisms: net-tension, shear out, and bearing failure (Khashaba
et al., 2013).
Failure accumulation and propagation (the combination of matrix cracking, cleavage, delamin-
ation, and fiber failure) in the open-hole composite plates are extremely complicated that
made analytical methods pointless. Hence, finite element solution is an appropriate selection.
Progressive failure modeling as a finite element solution by realistic simulation of failure initiation
and propagation using only lamina properties appears to be a suitable method for failure analysis
and strength determination of composite joints (Camanho and Matthews, 1999; Chang and Chang,
1987a, 1987b; Chen et al., 2014; Kozlov and Sheshenin, 2016; O’Higgins et al., 2004; Shokrieh and
Lessard, 1996; Tan, 1991; Tserpes et al., 2002). Different numerical methods had been utilized to
analyze the progressive failures in both unidirectional and woven composites (Taheri-Behrooz and
Bakhshan, 2016).
Failure propagation models are mainly categorized into two groups: ply-discounting material
degradation approach and models based on a continuum damage mechanics approach using
internal state variables. Comprehensive researches studied the property degradation rules of unidir-
ectional composites (Garnich and Akula, 2009; Knight and Reeder, 2006).
Two-dimensional (2D) finite element method based on progressive failure is performed
by Kweon et al. (2007). They used various failure criteria (maximum stress, Tsai and Wu, 1971;
Yamada and Sun, 1978) in order to investigate strength of the pinned-joint by considering stiffness
degradation. Also, they conducted some tests on unidirectional and woven composite laminated
joints with different geometries to verify their numerical results and found that Tsai–Wu and
Yamada–Sun criteria are more precise than maximum stress criterion.
Three-dimensional (3D) finite element model (FEM) is conducted to predict failure
progression of the mechanically fastened joints of carbon fiber-reinforced plastic that is subjected
to bearing, tensile, and compression loadings. The effect of damage on material elastic
properties were considered using a 3D failure criterion and on a constitutive equation. A good
agreement between experimental result and numerical prediction was observed (Camanho and
Matthews, 1999).
The effects of variation of geometry on the failure modes and failure loads of the multi-
pin joints (three-pin and four-pin joints) have been performed experimentally and numerically
by Kishore et al. (2009). The variation of pitch-to-diameter ratio (P/D), side width-
to-diameter (S/D), and edge-to-diameter (E/D) are the main geometrical changes in the specimens.
They used 2D Tsai–Wu criteria to predict damage loads and failure modes. Regarding the
numerical consideration, failure loads analogous to the results were obtained by experimental
procedure.
A finite element code was applied in order to predict the strength of composite structures,
exhibiting different levels of complexity (un-notched plates, open-hole plates), which subjected to
complex loadings (membrane or bending loadings) (Tserpes et al., 2001).
Bakhshan et al. 3
A progressive failure model for open-hole compression that is based on continuum shell and cohe-
sive elements is used to model in-plane damage and delamination, respectively. Tsai–Wu and
Hoffman failure criteria put on to determine the initiation of matrix damage. It is shown that
based on open-hole compression strength prediction and failure modes simulation, Tsai–Wu criter-
ion exhibit better accuracy (Zhou et al., 2016).
The influence of the stacking sequence on the failure modes and loads of the pinned-joint glass
fiber-reinforced composite was investigated experimentally and numerically (Khashaba et al., 2013).
They obtained mechanical features of the single lamina by conducting some experimental tests which
are obligatory for the FEM.
3D failure criteria of unidirectional fiber-reinforced composites were defined in terms of quadratic
stress polynomials (Hashin, 1980). Three-dimensional Hashin’s criterion was studied by considering
degradation and progressive damage of the four distinct failure modes such as tensile and compres-
sive fiber and matrix modes. They found that composites with [0/90]2S lay-up has the highest
ultimate strength. Moreover, the minimum bearing strength was observed in [30/60]2S laminates.
As a result, they investigated that layers with 90 laminates play an important role to increase the
displacement of the pin and energy absorption.
Karami et al. (2015) investigated the influence of fiber orientation of thin-walled steel square
section vessels wrapped by glass fibers on failure mechanisms of these structures subjected to an
internal pressure. Failure mechanisms of these structures were inquired numerically using the pro-
gressive damage models (Hashin’s criteria and energy-based damage). They observed that damage
evolution was highly affected by fiber orientation and found that their numerical simulation was in a
good agreement with their experimental data.
FEM that considered contact at the pin–hole interface, progressive damage, large deform-
ation theory, and non-linear shear stress–strain behavior has been presented by Dano et al.
(2007). Their goal was to determine the effect of the failure criteria (Hashin and maximum
stress), embodiment of the non-linear shear behavior on the strength of the structure, and load-
pin displacement curve. They reported in the linear part of stress–strain curve that maximum stress
criterion predicts a realistic form of the composite behavior and bearing strength compared with
Hashin’s criterion.
The objective of the current research is to model 3D progressive failure of unidirectional com-
posite plate with a central notched (hole) subjected to tensile loading. For analyzing the initiation of
the failure, two criteria, Hashin and modified Yamada–Sun, have been used for understanding the
interactive effect of failure criteria in composites. The failure propagation has been modeled using
complete (Tserpes et al., 2001) and Camanho and Matthews’s (1999) degradation rules based on the
reduction of the elastic material stiffness coefficients. Predicting strength by numerical methods can
provide a better sight to apply reasonable parameters for underloaded structures. Therefore, the
challenge is to predict damage initiation and growth, so a catastrophic failure will be inhibited and
structure provides a reliable safety in designing. Finally, numerical results are compared with the
study by Zhang et al. (2014).

Problem description
A unidirectional composite plate with a central hole subjected to a uniform tensile
loading is assumed. Failure load determination of the plate by means of progressive failure
analysis (PFA) searching adequate failure criteria and property degradation rules are the aims
of this manuscript. Also, the other goal is to analyze the failure behavior of the plies in the
4 International Journal of Damage Mechanics 0(0)

Figure 1. Schematic of the open-hole plate.

Table 1. Material properties (Zhang et al., 2014).

Properties E1 ðGPaÞ E2 ðGPaÞ E3 ðGPaÞ G12 ðGPaÞ G13 ðGPaÞ G23 ðGPaÞ #12 #13 #23

Mean 195 8.58 8.58* 4.57 4.57* 2.9* 0.33 0.33* 0.48*
Properties XT ðMPaÞ Xc ðMPaÞ YT ðMPaÞ Yc ðMPaÞ ZT ðMPaÞ Zc ðMPaÞ S12 ðMPaÞ S23 ðMPaÞ S13 ðMPaÞ
Mean 3071 1747 88 271 88* 271* 143 143* 143*
Properties GIc ðJ=m2 Þ GIIc ðJ=m2 Þ T
Yise ðMPaÞ STise ðMPaÞ T
Yiso ðMPaÞ STiso ðMPaÞ __ __ __
Mean 420 1480 157 305 100 216 __ __ __

laminate under the applied load. Schematic configuration of a tensile sample is indicated in
Figure 1.
According to Zhang et al. (2014), composite laminates were prepared using carbon/epoxy X850
with a common and symmetric lay-up of [45/0/-45/0/90/0/45/0/-45/0]s configuration. The laminate
specimen is then cut into required dimensions. The standard elastic and strength properties of the
laminate are given in Table 1. The marked sample with (*) are based on transversely isotropic
assumptions and also engineering experiences.

Progressive failure analysis

Failure criterion is one the effective parameters in PFA, which creates a motivation to offer various
failure criteria. According to the researchers, none of the criteria could perfectly predict all the
failure mechanisms for a selective structure. Likewise, a certain criterion is favorable for a specific
structure with specified failure mechanism. Failure criteria of Yamada–Sun and Hashin were chosen
for this study due to the extreme applications in analyzing the mechanical joints, which are described
in Table 2.
Bakhshan et al. 5
Table 2. Hashin and modified Yamanda-Sun failure criterion.

Modified Yamada–Sun
Failure modes (Chang and Chang, 1987b) Hashin (1980)
 2  2  2  2 
1 12 1 1 2
Fiber tension ð1 4 0Þ XT þ SC ¼1 XT þ S2 12 þ 13 ¼ 1
12
 2  2
1 12
Fiber compression ð1 5 0Þ XC þ SC ¼1  X1C ¼1
 2  2 ð2 þ 3 Þ2 12
2 2
þ 13 2
þ 23  2 3
2 12
Matrix tension ð2 þ 3 4 0Þ YT þ SC ¼1 2
þ 2
¼1
YT S12
 2  2 #
" "  #
2 YC 2 1 YC 2 ð2 þ 3 Þ2
þ 1 1 ð2 þ 3 Þ þ
SC 2SC YC YC 2S23 4S223
Matrix compression  2
ð2 þ 3 4 0Þ 12 1 2 1 2 2
þ ¼1 þ ð  2 3 Þ þ 2 ð12 þ 13 Þ¼1
SC S223 23 S12

 2  2
1 12
Fiber–matrix shear XT þ SC ¼1 ________

Table 3. Camanho’s and complete material degradation rules.

Comanhos Complete
Failure modes (Camanho and Matthews, 1999) (Tserpes et al., 2001)
E1 ¼ E2 ¼ E3 ¼ 0, G12 ¼ G13 ¼ G23 ¼ 0,
Fiber tension ð 1 4 0Þ Ed1 ¼ Dd1 E1
#12 ¼ #13 ¼ #23 ¼ 0
E1 ¼ E2 ¼ E3 ¼ 0, G12 ¼ G13 ¼ G23 ¼ 0,
Fiber compression ð1 5 0Þ ðEd1 ¼ Dc1 E1 Þ
#12 ¼ #13 ¼ #23 ¼ 0
Ed2 ¼ DT2 E2 , Gd12 ¼ DT4 G12 ,
Matrix tension ð2 þ 3 4 0Þ E2 ¼ 0, E3 ¼ 0, #12 ¼ 0
Gd23 ¼ DT4 G23
Ed2 ¼ Dc2 E2 , Gd12 ¼ Dc4 G12 ,
Matrix compression E2 ¼ 0, E3 ¼ 0, #12 ¼ 0
ð2 þ 3 4 0Þ Gd23 ¼ Dc4 G23

Ed2 ¼ Dc2 E2 , Gd12 ¼ Dc4 G12 ,

Fiber–matrix shear G12 ¼ 0, #12 ¼ 0
Gd23 ¼ Dc4 G23

The other critical parameter is the complete material degradation model. The subject of material
degradation should be studied precisely in damage mechanics. Based on the created damage mode,
an appropriate coefficient is used for simulation of structure–behavior to decrease its stiffness.
Various coefficients have been suggested in Tserpes et al. (2002). It is necessary to find a desirable
coefficient for troubleshooting the divergence problem of numerical solutions. In this study, both
complete and Camanho’s degradation rule were applied in numerical simulations according to the
Table 3.
6 International Journal of Damage Mechanics 0(0)
Constitutive model
The constitutive model of a linearly elastic orthotropic material in three dimensions and applicable
in ABAQUS software is written as follows:
8 9 2 0 38 9
> 11 > C11 C012 C013 0 0 0 > "11 >
>
> >
> >
> >
> >
> > 6 C0 C0 C023 0 7 >" > >
>
> 22 >
> 6 0 0 7>
> 22 >
>
7> >
21 22
< >
> = 6 > >
33 6 C 0
C 0
C033 0 0 0 7 "33 =
<
f g ¼ ¼6 31 32 7
> 12 >
> > 6 6 0 0 0 C044 0 0 7 > 12 >
7> > ð1Þ
>
> >
> 6 7>
> >
>
>
>  >
> 4 0 0 0 0 C055 0 > 5>
>  >
>
>
> 13 >
> > 13 >
>
: ; : ;
23 0 0 0 0 0 C066 23
 0 1  0
¼ S f"g ¼ C f"g

The stiffness coefficients, C0ij in equation (1) using elastic material constants are defined as:

ð1  23 32 ÞE11

C011 ¼ ;

ð1  13 31 ÞE22
C022 ¼

0 ð1  12 21 ÞE33
C33 ¼ ;

ð12 þ 13 32 ÞE22 ð21 þ 31 23 ÞE11
C012 ¼ C021 ¼ ¼ ; ð2Þ
 
ð13 þ 12 32 ÞE33 ð31 þ 21 23 ÞE11
C013 ¼ C031 ¼ ¼ ;
 
ð23 þ 13 21 ÞE33 ð32 þ 31 12 ÞE22
C023 ¼ C032 ¼ ¼ ;
 
C044 ¼ G12 ; C055 ¼ G13 ; C066 ¼ G23
 ¼ 1  12 21  23 32  13 31  221 32 13

In 3D solution, laminate is modeled in the direction of thickness by 3D solid elements. In each

solid element, integration points are evaluated to determine the strain state and then, the stress state
can be calculated.

User-defined material model

The mechanical constitutive behavior of a material can be discussed by user subroutine User-defined
material model (UMAT). For progressive analysis, stresses and solution-dependent state variables
will be updated at the end of the increment. In other words, at each load step, non-linear analysis is
accomplished until a converged solution is obtained. An updated stiffness matrix is executed to
specify the deformation and stress states in the composite. As a result, UMAT can give the material
Jacobian matrix for the mechanical constitutive model. User subroutine USDFLD can be used in
conjunction to redefine any field variables before they are passed in. In some cases, it is better to set
up the FORTRAN environment and control interactions with external data files that are used in
conjunction with user subroutines. For instance, some history-dependent quantities need to be
Bakhshan et al. 7
computed externally within the analysis, once per increment, or accumulated output quantities of
multiple elements in COMMON block variables during user subroutines can be written to external
files at the end of a converged increment for post-processing. User subroutine UEXTERNALDB is
able to handle these operations and exchanges data with another code, allowing for ‘‘stagger’’
between ABAQUS and other codes.
User subroutines should be used carefully. Following the rules and guidelines below is necessary
to warrant accurate implementation. Each user subroutine must do the commands. After the argu-
ment list, at first statement, the file ABA_PARAM.INC is installed on the system using the
ABAQUS installation procedure. Both IMPLICIT REAL*8 (A-H, O-Z) and IMPLICIT REAL
(A-H, O-Z) are defined by ABA_PARAM.INC, which are for double precision machines and for
single precision machines, respectively. Figure 2 shows the performance method of ABAQUS, which
compiles and makes a connection between the user subroutine and the rest of ABAQUS and will

Figure 2. Flowchart of the interaction between ABAQUS and UMAT.

8 International Journal of Damage Mechanics 0(0)

include the ABA_PARAM.INC file automatically. ABAQUS can find the exact location of this file,
so there is no need to copy it to any specific directory.
The specified function must be executing user subroutines without overwriting other parts of
ABAQUS. Specially, just those identified variables in this chapter as ‘‘variables to be defined’’ should
be redefined by user and Redefining ‘‘variables passed in for information’’ will have unknown results.

(1) Before applying user subroutines in production analysis work, investigate them thorough smaller
examples while they are expanding because it is the only complicated aspect of the model.
If needed, debug output can be written to FORTRAN unit 7 to appear in the message (.msg)
file or to FORTRAN unit 6 to appear in the data (.dat) file; these units should not be opened by
the user’s routines since they are already opened by ABAQUS.
(2) Other user-specified information is read or written by FORTRAN units 15 through 18 or units
greater than 100. The use of other FORTRAN units may interfere with ABAQUS file oper-
ations. User must open these FORTRAN units; and because of the use of scratch directories, the
full pathname for the file must be used in the OPEN statement.
(3) Solution-dependent state variables are values that can be defined to evolve with the solution of
an analysis.

Both the equation, of motion and the constitutive law, constitute a system that includes an
initial—boundary problem and an ordinary differential equation. A finite element package
(ABAQUS) and ordinary differential equations can solve the equation of the motion and constitu-
tive law, respectively. A user with the aid of subroutine UMAT will enter the relevant constitutive
information to ABAQUS. So ABAQUS can run an (incremental) loading with the time increment
t and an initial guess "00 for the strain increment, which is started from an equilibrium at time tn.
New Cauchy stress tensor s (t þ t) should be supplied to ABAQUS with the user subroutine
UMAT which receives the data from constitutive law and also derivative of stress according to
the strain increment. Therefore, a new guess for the strain increment is obtained and the whole
process should be repeated until convergence. To access fast convergence in Newton-type repetition,
the exact data on the Jacobian are exigent.

Finite element model

The 3D simulations of the specimen were developed using software ABAQUS with C3D8 elements.
Mesh type and applied boundary conditions are shown in Figure 3.
The nodes on section A were fixed in all three directions and section B is only fixed in directions
vertical to the load direction. Since the distributed tension is only focused around the central hole,
radial meshes were applied around the hole. In order to reduce the time of calculations, the meshes
become coarser with increase in the distance from hole elements.
The stacked approach is used for modeling the composite because of the existence of enormous
differences to provide appropriate answer. For this analysis, the layered approach is not a suitable
method and the mixed approach has numerous errors. It can be claimed that the existence of one or
a few elements through thickness of laminate.

Mesh density
The mesh density can be very effective in analytical results of modeling; therefore, it is crucial to
choose an appropriate mesh density in order to guarantee the accuracy of results. It should be noted
Bakhshan et al. 9

Figure 3. Finite element model of open-hole laminate.

Figure 4. Mesh density around hole.

that the mesh density for the areas under high stress concentration is different from the other areas.
Three different mesh densities around the hole are shown in Figure 4.
The results of previous figures (Figure 4) show the convergence of failure loads from M2 to M3.
According to Table 4, it can be concluded that there is negligible difference between M2 and M3
models, which is almost close to 1%. So this amount was used for comparing the models, since
failure loads are the desirable parameters of the analysis. As mentioned earlier, the number of
existing elements in thickness is the same as the layers of laminate which is equal to 20 in this
analysis, so just the in-plane mesh density was analyzed. The concept of the symbols and the number
of the elements on the edge of specimen have been shown in Table 5 and Figure 5, respectively.

Results and discussion

The both failure criteria of Yamada–Sun and Hashin and the both models of complete material and
Camanho’s degradation will be used in modeling of progressive failure. Then, in the latter analysis,
the most convenient model has been selected to evaluate the strength of an open-hole laminate.
The solution dependence of state variables presents the damage mode in the mentioned UMAT,
which is defined by STATEV. The load–displacement curves of experimental results are plotted in
Figure 6. The average of failure load resulted from experiments is 90.9 KN and the variation
coefficient is 3.15%.
10 International Journal of Damage Mechanics 0(0)

Table 4. Mesh density comparison.

Mesh Number of Failure

Model density elements load Difference

1 M1 35,760 112.6 0
2 M2 47,600 89.3 20.6
3 M3 62,480 88.1 21.7

Table 5. Marking.

Concept Symbol Concept Symbol

Fiber failure in tension SDV1 Maximum failure modes SDV6

Fiber failure in compression SDV2 Hashin criterion—Complete degradation H-T
Matrix failure in tension SDV3 Hashin criterion—Camanho’s degradation H-C
Matrix failure in compression SDV4 Sun criterion—Complete degradation Y-T
Fiber–matrix failure in shear SDV5 Sun criterion—Camanho’s degradation Y-C

Figure 5. Element numbers in M2 mesh density.

The results of predicted numerical modeling are compared with experimental results in a
column chart as shown in Figure 7. The complete material degradation model cannot perfectly
forecast the criteria of both Yamada–Sun and Hashin, while the Camanho’s degradation rule has
a better prediction. It can be said that the less prediction ability of the complete material deg-
radation model is due to additional reduction in stiffness and strength of the elements adjacent to
the hole, which leads to elimination of these elements and therefore faster divergence. In laminates
for each failure mode of complete degradation model, all properties of an element will be
decreased equally. For instance if the first failure of an element is the fiber tension mode, as
illustrated in Table 3, the whole properties of that element will be declined, and so it cannot carry
the transversal and shear loads.
H-C model with the error of 2.5% presents the best prediction; hence, it will be applied in the next
analysis. Load–displacement curves are shown in Figure 8 for four numerical models. Linear elastic
behavior assumption is applied. It is clear that in linear step, the type of failure criteria and material
degradation model have no impact on initial stiffness; therefore, all of the curves follow a linear
behavior and after stacking a certain amount of failure, the material behavior becomes non-linear.
Amount of stacking in complete degradation is more than Camanho’s model in a shorter period of
time, which leads to reduction in load-carrying capacity of the composite structure in complete
Bakhshan et al. 11

Figure 6. Load-displacement curves for specimens (Zhang et al., 2014).

Figure 7. The comparison of numerical modeling with experimental results.

compared with Camanho’s model. In numerical analysis, divergence time of complete material
degradation model was lesser than Camanho’s model.
Figure 9 indicates further investigation of the load–displacement curve of H-C for each step,
while Figure 10 shows that the steps of load–displacement curves for Y-T, H-C, and Y-T were
selected to compare the effect of the both criteria and both material degradation models. The initial
load of fiber failure for H-C is the same as Y-T. Since the failure initiation is a dependent parameter
to failure criteria, the reason of this dependency should be traced in comparing the both failure
criteria. In equation, which is related to tensile fiber failure mode of Hashin criteria, the out-of-plane
12 International Journal of Damage Mechanics 0(0)

Figure 8. Load-displacement curves of numerical modeling.

Figure 9. Load–displacement curve of H-C.

Bakhshan et al. 13

Figure 10. Load–displacement curves for Y-T.

shear tension has insignificant impact on failure index, which eliminates in the relation of tensile
fiber failure mode for Yamada–Sun. This argument is applied to express the difference of initial
failure load of matrix between two criteria. As can be resulted from figures, the matrix of Y-T fails in
lower initial loads that the initial loads for Y-T and H-C are 63.2 KN and 71.8 KN, respectively.
In the equation of tensile matrix failure mode for Hashin, the existence of out-of-plane tensions
suggest the following view that the tensions increase the failure index and thus it arrives at unit
earlier and failure starts at lower loads. It should be considered that the negative amount of trans-
versal tensions decrease the latter equation and increases the index.
Initial fiber failure started in the 10th layer with 0 orientation for both H-C and Y-T model. It is
observed that tensile fiber failure occurs at the layer orientation of 0 because the longitudinal
tension has the maximum amount.
However, the reason of failure initiation from the 0 -oriented inner layer could be explained in
different aspects. First, the layer’s thickness is doubled due to putting two layers together so the
exerted force increases twofold and decreases the tension. Therefore, a late failure is expected from
the inner layers but the result does not match accurately. The existence of 45 layer around the 0
central layers leads to the creation of a severe in-plane shear tension, which has a great influence on
Yamada–Sun and Hashin’s equations of tensile fiber failure; this means that it should be considered
in designing viewpoints.
In Figure 10, tensile matrix failure initiation for Y-T takes place on the 5th and 15th layers with
the orientation of 90 as expected for the 90 layers to fail. While in H-C, tensile matrix failure
occurs for the oriented layers because of the impact of shear tensions on Hashin failure index for
matrix tension.
14 International Journal of Damage Mechanics 0(0)

Figure 11. The progress of failure index, for SDV1 and SDV3.

In maximum of tensile fiber failure for Y-T, damage occurs at the entire layers because of
decreasing all of the material properties in complete material degradation model. In addition to
the matrix failure mode, damage of 0 layers is inconsiderable, so there is no damage for the both
outer layers of 0 orientation.
As presented in Table 3 for tensile fiber failure mode, the complete material degradation model
only changes the stiffness in transversal direction so the 0 layers remains intact.
For H-C in maximum load, damage does not occur for two 90 layers because in this failure
mode, the Camanho’s coefficient only reduces the fiber stiffness. For tensile matrix failure, damage
happened for all the orientations of layer due to large maximum load.
Figure 11 shows the progress of failure index, for SDV1 and SDV3, for both of 0 and 90 layers
in Y-T and H-C models. The figures are for four time steps. Tensile fiber failure is indicated for 0
layers and tensile matrix failure is presented for 90 layers.
The schematics of progressive failure for both types of layers and in both models are almost the
same. The main difference of two models is at the form of defect propagation.
In Y-T, damage is progressing on net-tension failure plane until a specific step and then extends
near the edge. This procedure is not observed in H-C, and the failure mode goes to the end of net-
tension failure plane, nevertheless the Camanho’s degradation rule presents the more realistic
results. The figure also shows the net-tension failure mode for all layers just like the experimental
results. In complete material degradation model, a remarkable growth of damage is observed at the
Bakhshan et al. 15
final time step which is consistent with coefficient behavior of this model. The same results are
obtained from the figure for 90 layers.

Conclusion
The PFA of notched composite plate under tensile loading is performed in a finite element-based
simulation and compared with experiment in the literature. Two sets of failure criteria in combin-
ation with two sets of material property degradation rules were used in this paper to examine the
effect of failure analysis and material property degradation on the predicted load–displacement
curve and failure load of the joint. The load–displacement curves are obtained through the finite
element analysis. Failure loads from numerical simulation and experiment were compared. Results
showed that despite Camanho’s coefficients, the complete material degradation model cannot per-
fectly predict both the modified Yamada–Sun and Hashin criteria. Reduction of stiffness and
strength in vicinity of the hole leads to faster divergence and less accurate results in complete
material degradation model. Numerical predictions demonstrate that the overall material response
is sensitive to the fiber and matrix properties such as Young modulus and stiffness. Therefore, it is
exigent in the design of fiber-reinforced composites to carefully choose proper material properties of
the matrix and fibers for optimization between the reinforcement effect and fiber breaking resistance
(Ju and Wu, 2016).
Moreover, initial fiber failure load is the same for both H-C and Y-T concepts, while some
differences could be observed in the matrix failure load. This difference is established from eliminat-
ing out-of plane shear effect in the Yamada–Sun criterion. Also, failure is observed on net-tension
mode in the Y-T concept and a considerable damage growth happened using complete material
degradation model.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or pub-
lication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.

References
Afrouzian A, Movahhedi Aleni H, Liaghat G, et al. (2017) Effect of nano-particles on the tensile, flexural and
perforation properties of the glass/epoxy composites. Journal of Reinforced Plastics and Composites.
1 January. DOI: 0731684417694753.
Camanho P and Matthews F (1999) A progressive damage model for mechanically fastened joints in composite
laminates. Journal of Composite Materials 33: 2248–2280.
Chang F-K and Chang K-Y (1987a) A progressive damage model for laminated composites containing stress
concentrations. Journal of Composite Materials 21: 834–855.
Chang F-K and Chang K-Y (1987b) Post-failure analysis of bolted composite joints in tension or shear-out
mode failure. Journal of Composite Materials 21: 809–833.
Chen X, Li Z and Wang H (2014) Progressive failure analysis of an open-hole composite laminate by using the
S-version finite-element method. Mechanics of Composite Materials 50: 279–294.
Dano M-L, Kamal E and Gendron G (2007) Analysis of bolted joints in composite laminates: Strains and
bearing stiffness predictions. Composite Structures 79: 562–570.
16 International Journal of Damage Mechanics 0(0)

Garnich MR and Akula VM (2009) Review of degradation models for progressive failure analysis of fiber
reinforced polymer composites. Applied Mechanics Reviews 62: 010801.
Hashin Z (1980) Failure criteria for unidirectional fiber composites. Journal of Applied Mechanics 47: 329–34.
Ju J and Wu Y (2016) Stochastic micromechanical damage modeling of progressive fiber breakage for longi-
tudinal fiber-reinforced composites. International Journal of Damage Mechanics 25: 203–227.
Karami P, Tabatabaei SA, Zangaraki R, et al. (2015) Experimental and numerical analyses of progressive
damage in non-circular metal/composite hybrid vessels under internal pressure. International Journal of
Damage Mechanics 24: 1261–1279.
Khashaba U, Sebaey T, Mahmoud F, et al. (2013) Experimental and numerical analysis of pinned-joints
composite laminates: Effects of stacking sequences. Journal of Composite Materials 47: 3353–3366.
Kishore AN, Malhotra S and Prasad NS (2009) Failure analysis of multi-pin joints in glass fibre/epoxy com-
posite laminates. Composite Structures 91: 266–277.
Knight NF Jr and Reeder JR (2006) User-defined material model for progressive failure analysis. NASA
technical report server, NASA Langley Research Center; Hampton, VA, US.
Kozlov M and Sheshenin S (2016) Modeling the Progressive Failure of Laminated Composites. Mechanics of
Composite Materials 51: 695–706.
Kweon J-H, Shin S-Y and Choi J-H (2007) A two-dimensional progressive failure analysis of pinned joints in
unidirectional-fabric laminated composites. Journal of Composite Materials 41: 2083–2104.
Mokhtari A, Ould Ouali M and Tala-Ighil N (2015) Damage modelling in thermoplastic composites reinforced
with natural fibres under compressive loading. International Journal of Damage Mechanics 24: 1239–1260.
O’Higgins R, Padhi G, McCarthy M, et al. (2004) Experimental and numerical study of the open-hole tensile
strength of carbon/epoxy composites. Mechanics of Composite Materials 40: 269–278.
Shokrieh MM and Lessard LB (1996) Effects of material nonlinearity on the three-dimensional stress state of
pin-loaded composite laminates. Journal of Composite Materials 30: 839–861.
Taheri-Behrooz F and Bakhshan H (2016) Characteristic length determination of notched woven composites.
Advanced Composite Materials. 28 September. DOI: 10.1080/09243046.2016.1232007.
Tan SC (1991) A progressive failure model for composite laminates containing openings. Journal of Composite
Materials 25: 556–577.
Tsai SW and Wu EM (1971) A general theory of strength for anisotropic materials. Journal of Composite
Materials 5: 58–80.
Tserpes K, Labeas G, Papanikos P, et al. (2002) Strength prediction of bolted joints in graphite/epoxy com-
posite laminates. Composites Part B: Engineering 33: 521–529.
Tserpes K, Papanikos P and Kermanidis T (2001) A three-dimensional progressive damage model for bolted
joints in composite laminates subjected to tensile loading. Fatigue & Fracture of Engineering Materials &
Structures 24: 663–675.
Ubaid J, Kashfuddoja M and Ramji M (2014) Strength prediction and progressive failure analysis of carbon
fiber reinforced polymer laminate with multiple interacting holes involving three dimensional finite element
analysis and digital image correlation. International Journal of Damage Mechanics 23: 609–635.
Yamada S and Sun C (1978) Analysis of laminate strength and its distribution. Journal of Composite Materials
12: 275–284.
Zhang J, Liu F, Zhao L, et al. (2014) A progressive damage analysis based characteristic length method for
multi-bolt composite joints. Composite Structures 108: 915–923.
Zhou S, Sun Y, Chen B, et al. (2016) Progressive damage simulation of open-hole composite laminates under
compression based on different failure criteria. Journal of Composite Materials 51: 1239–1251.

View publication stats