Theoretical and Applied Fracture Mechanics 126 (2023) 103967

Contents lists available at ScienceDirect

Theoretical and Applied Fracture Mechanics

journal homepage: www.elsevier.com/locate/tafmec

Extrinsic/intrinsic cohesive zone modelling and experiments on interface

failure of multi-walled carbon nanotube reinforced adhesively bonded
joints under mode II loading
P.J. Saikia , N. Muthu *
Department of Mechanical Engineering, IIT Guwahati, North Guwahati 781039, Assam, India

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

Keywords: The present work is primarily concerned with determining mode II cohesive zone parameters for eliminating the
Interface fracture crack tip stress-field singularity for constant, linear, and parabolic extrinsic cohesive zone models (CZMs). A
CNT semi-analytical approach is used to estimate the stress intensity factor (SIF) associated with the cohesive zone
CZM
ahead of the physical crack tip. Simultaneously, the obtained SIF is compared with the popular Interaction in­
Mode II crack propagation
Interaction integral
tegral technique results considering the influence of crack face traction. The influence of the order of the traction-
separation law (TSL) on the global load–displacement responses is numerically investigated using a crack
propagation methodology based on the nullification of SIF. The effectiveness of the proposed approach has been
demonstrated through a series of experiments involving the failure of epoxy-based adhesively bonded modified
compact tension specimens (CTS) subjected to pure mode II loading. The adhesives are reinforced with different
multi-walled carbon nanotubes (MWCNTs) reinforcement of 0.1 wt%, 0.2 wt%, and 0.3 wt%, and its effect on the
global load vs displacement responses are recorded. Specific cohesive zone parameters are obtained for the
experimental problem to model the failure behaviour for a broad spectrum of cases considered using extrinsic
and intrinsic TSLs.

1. Introduction plastic zone’s size. A cohesive law characterizes the cohesive zone with
three critical parameters; the cohesive stiffness Kc , the cohesive strength
Adhesively bonded joints have been widely used in structural, σ c , and the fracture energy GC . The cohesive stiffness defines the elastic
automobile, and aerospace applications due to their excellent strength- response of the cohesive till it reaches the damage initiation stress σ c .
to-weight ratio, uniform stress distribution, superior fatigue resistance, The fracture energy GC accounted for the energy accumulated in the
and capabilities to join dissimilar materials [1,2]. However, the per­ cohesive zone prior to its failure. The CZM is of two types — intrinsic
formance of the adhesive bonding depends on the geometry of the joints, with zero traction at zero separation and extrinsic with non-zero finite
the surface preparation, the types of adhesive used, and their curing traction at zero separation, as shown in Fig. 1. Although a cohesive law
conditions for a specific joining process [3]. A design engineer must can be of various shapes, such as triangular, polynomial [6], or expo­
consider all these processing parameters to obtain a reliable joint for any nential [7], accurate parameter identification with appropriate, cohe­
specific application. sive law is equally essential for a specific material failure behaviour.
The failure of adhesively bonded joints can be accurately predicted Additionally, the strength of stress singularity also needs to be elimi­
using the mechanics of material and conventional damage/fracture nated at the tip of the cohesive zone, which is one of the primary ob­
mechanics approach. Cohesive zone modelling (CZM) is an efficient jectives of the CZM.
methodology under mesoscale modelling of damage mechanics and can The SIF is one of the most important parameters that define the
be coupled with finite element (FE) analysis for failure estimation of strength of stress singularity under linear elastic fracture mechanics
bonded joints. Barenblatt [4] initially used the concepts of CZM for (LEFM). There are different domain integral procedures [8–12] for the
predicting the failure of brittle material. Later, Dugdale [5] extended his estimation of SIF within the context of computational mechanics. The
idea to investigate the yielding of thin steel sheets and approximate the characteristic length of the cohesive zone and the cohesive strength

* Corresponding author.
E-mail address: [email protected] (N. Muthu).

https://doi.org/10.1016/j.tafmec.2023.103967
Received 15 February 2023; Received in revised form 8 May 2023; Accepted 7 June 2023
Available online 11 June 2023
0167-8442/© 2023 Elsevier Ltd. All rights reserved.
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

plays a pivotal role that affects the SIF of the cohesive zone. It is standard tests with different modes of loadings [22–24]. Kulkarni et al.
important to mention that only the cohesive law with non-zero traction [24] used the intrinsic bilinear CZM technique to estimate the brittle
at its initial separation displacement can be used to eliminate the SIF at fracture of dissimilar-metal friction stir welded joints. Pang and Seetoh
the tip of the cohesive zone. Although Jin and Sun [13] have provided a [25] have developed a compact mixed-mode (CMM) fracture adhesively
closed-form analytical expression for evaluating the cohesive zone pa­ bonded specimen to evaluate the fracture toughness under pure mode I,
rameters for eliminating the crack tip singularity, there is no straight­ mode II, and mixed-mode loading. A parametric study was also con­
forward method available for estimating the size of the cohesive zone ducted numerically to investigate the effect of adhesive layer thickness,
due to the complex nature of the problem. Karihaloo [14] obtained the crack length ratio, and fracture mode angle on the failure behaviour of
pure mode I asymptotic stress and displacement field expression for the joints. Choupani [26] used a modified Arcan specimen to investigate
quasi-brittle material at the tip of the cohesive zone. Saikia and Muthu the mixed-mode cohesive fracture behaviour of adhesively bonded
[15] recently proposed a semi-analytical methodology for estimating the joints. The fracture toughness values for a range of substrates were also
cohesive zone size for eliminating the strength of stress singularity for evaluated with different mixed-mode angles, and a relationship between
any generalized cohesive law. The proposed procedure was validated on stress intensity factor (SIF) and fracture toughness was derived. Noury
pure mode I case studies of neat adhesively bonded joints. et al. [27] have performed mixed-mode fracture tests on Polyvinyl
One of the significant drawbacks of adhesive joining is their low Chloride (PVC) foam using a compact-tension-shear test loading device.
fracture resistance and susceptibility to aggressive environments. After obtaining the mode I and mode II fracture toughness experimen­
Incorporating nano-sized filler in the adhesive is an efficient method for tally, the fracture angle for a mixed mode loading condition was accu­
improving the mechanical properties of the joints. Several nanofillers (e. rately evaluated using Richard’s criteria.
g., graphene, carbon nanotubes, silicon carbide) have been widely used The failure behaviour of material ranges from ductile to brittle
in joining structural elements and the interfaces of composite materials. depending upon its geometry, material properties, types of loadings, and
Carbon nanotubes are popular reinforcements due to their superior boundary conditions. Carpenteri and Colombo [28] studied the influ­
mechanical, thermal, and electrical properties. Therefore, numerous ence of strength, fracture toughness, and geometric dimensions on the
studies have been carried out on the influence of carbon nanotubes on global load–displacement behaviour of the specimen. Based on the
the mechanical response of bonded joints [16–18]. Gojny [19] has dimensional analysis, they found that the global failure behaviour of any
investigated the mechanical properties of epoxy-based composites under specimen depends upon the ratio between fracture energy and strength
the influence of different carbon nanotubes (single, double, and multi- along the failure interface, popularly known as the brittleness number.
walled). A significant increase in fracture toughness (up to 43 %) was For a specimen with low fracture toughness and high strength, the
observed with the addition of 0.5 wt% amino-functionalized DWCNT brittleness number will be less, and failure behaviour will be brittle.
(double-walled carbon nanotube). Hsiao et al. [20] prepared a single lap Identically the crack propagation path will be gradual, and failure
joint specimen to examine different weight proportions of MWCNTs behaviour will be ductile for a material with a higher value of brittleness
(multi-walled carbon nanotubes) on the average shear strength of number. Carpenteri [29] obtained a range of load–displacement re­
graphite fiber-reinforced composites. An improvement of 45.6 % in the sponses with the variation in brittleness number using node-released
average shear stress was observed with the addition of 5 wt% of techniques within the context of the FE method. Using the concepts of
MWCNTs in the epoxy resin. Tabaei et al. [21] examined the lap shear brittleness number, Moës and Belytschko [30] used a novel solution
strength and thermal stability of bisphenol A/ epoxy novolac adhesives methodology for modelling the cohesive crack growth in pure mode I
with a series of MWCNTs, nano-Al2 O3 , and nano-SiO2 . It was found that and mixed-mode using three and four-point bending test configurations,
the lap shear strength was enhanced by almost 70 % and 50 % with the respectively. However, it is observed that the global load–displacement
addition of nano-Al2 O3 and MWCNTs up to a certain level. Additionally, response also depends on the shape of the cohesive laws in addition to
a combination of 0.75 wt% of MWCNTs and 0.75 wt% of nano-Al2 O3 the values of the brittleness number. So, one of the objectives of the
produced superior thermal stability of epoxy novolac adhesives at high present work is to estimate the failure behaviour for a set of brittleness
temperatures. numbers with three different extrinsic cohesive laws — constant, linear,
Although mode II failure behaviour of adhesively bonded joint can and parabolic. Additionally, to the author’s knowledge, no literature
be characterized by standard lap shear test, additional design consid­ illustrates the failure response of an adhesive bonded structure within
erations must be considered for crack initiation and propagation. the context of SIF nullification under the pure mode II loading with
Several authors have investigated the fracture aspects of adhesively different types of reinforcement. This has given another motivation for
bonded/welded joints and evaluated the fracture toughness under the present study. The modified CTS specimen is numerically modelled

Fig. 1. Schematic of (a) intrinsic and (b) extrinsic CZM.

2
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

in ABAQUS® software. The failure response is analyzed for cohesive law weight relative to the combined weight of resin and hardener
with constant traction, linear, and parabolic softening. In order to utilize mixer for each variation of MWCNTs.
this approach practically, a modified CTS is fabricated with pristine and (b) The resin mixer is sonicated in a tip sonicator SONICS of 20 kHz
three different proportions of 0.1 wt%, 0.2 wt%, and 0.3 wt% of frequency range and 500-watt power capacity. The amplitude of
MWCNTs. The global load–displacement response obtained for all four the sonicator is kept at 40 % and 20 % for the sonication time 5
variations is validated with linear softening extrinsic cohesive zone min and 1hr 55 min, respectively, with 5 min on and 2 min off
methodology. In addition, cohesive zone parameters for an intrinsic bi- cycle. The solution container is kept in an ice bath to reduce the
linear CZM are also evaluated and compared with those of the extrinsic temperature of the solution generated during the sonication
CZM. process.
The outline of this paper is as follows. The experimental section on (c) The solution is mechanically stirred for another 2 h at 60 ◦ C and
interfacial fracture studies with different wt.% of the MWCNTs is added 600 rpm, and the hardener is added afterward with a weight ratio
first in Section 2. Section 3 describes a theoretical framework to deter­ of 3:10 to that of the resin.
mine the cohesive zone parameters for removing the stress singularity
for any generalized extrinsic cohesive law. This formulation is adopted The machined surface of the aluminum substrate is initially wiped
for simulating the failure response, i.e., the load vs displacement with a cloth wetted with acetone solution. The substrates were subjected
response of adhesively bonded interface fracture under mode II loading. to etching and anodizing with H2 SO4 solution as per the standard ASTM
This is followed by Section 4, where the numerical procedure is D2651 [34]. The adhesives are applied to the surfaces of the substrate;
described to determine the cohesive zone length to eliminate the sin­ bonded by placing it properly in a suitable fixture. All the specimens are
gularity. In Section 5, the proposed approach is applied using the finite cured for a day in the furnace at a temperature of 80 ◦ C, and the excess
element to predict the experimentally observed load vs displacement materials from the specimen are removed using abrasive paper. The
response. In the passing, we have also used the intrinsic CZM and prepared specimen for all wt.% variation of MWCNTs is shown in Fig. 3.
simulated the same load vs displacement response. Some comparative The thickness of the adhesive layer is in the range of 0.1+.01 − .01 mm,
comments were made at the end, in Section 6, highlighting the advan­ measured using a Vernier caliper. Therefore, the thickness of the ad­
tages and limitations of both types of CZM. hesive can be considered uniform for all the specimens throughout the
process of experimentation. Five specimens are prepared for each
2. Experimental procedure – mode II interface crack nanofiller variation to confirm the repeatability of the results.
propagation In order to obtain pure mode II loading, the specimens are connected
to the cross-head of the UTM (universal testing machine) through
Aluminium samples are joined with the adhesives mixed with vary­ Richard’s [35] fixture. The fixture is made of steel with a thickness of 4
ing proportions of nanofillers – MWCNTs. The modified compact tension mm. All other geometric dimensions of the fixture and the specimen are
specimens (CTS) are fabricated with pristine, 0.1 wt%, 0.2 wt%, and 0.3 shown in Fig. 4(a) and (b), respectively. The specimens are connected to
wt% of MWCNTs. All the specimens are tested in Instron 8801 100 kN the fixture with a 13 mm diameter pin made of mild steel. The material
dynamic testing system under quasi-static loading conditions. The properties of the fixture are already mentioned in Table 1.
following subsections illustrate the experimental methodology in detail. The strength of the bonded joints is found to vary with the changes in
strain rate. Machado et al. [36] investigated the fracture toughness of
2.1. Material carbon fiber reinforced composites (CFRP) and found that a higher
strain decreases the mode II fracture toughness considerably. Therefore,
The half CTS-shaped Aluminium-6061 substrates are joined by a bi- the cross-head speed is kept constant to maintain a strain rate of 1 mm/
component adhesive epoxy resin Epofine-1564 and hardener FINE­ min. The Richard’s fixture is connected to the upper and lower part of
HARD-3486–2. The material properties of the aluminium substrate are the cross-head of the UTM, as shown in Fig. 4 (c). The cross-head applied
provided in Table 1. The specimens are cut out from an aluminium alloy the tensile force to the bonded surface through the fixture. The
plate of dimension 500 × 500 × 10 mm3 using a wire-cut EDM (electrical displacement data are simultaneously recorded till the fracture happens
discharge machine). The Young’s modulus and the Poisson’s ratio of the along the interface.
epoxy-based adhesive are 3.1 GPa and 0.29, respectively, as provided by
the manufacturer. The MWCNTs had a diameter of 30 nm, length of 10
μm, and elastic modulus of 400 GPa supplied by Ad nano technologies, 2.3. Experimental results
India. The fixture is made of steel.
The average experimental load–displacement responses for all the
2.2. Specimen fabrication and test procedure variations of MWCNTs are shown in Fig. 5(a). Initially, the load data is
recorded for each common displacement value for every category of
In order to achieve a bonded joint with superior interface properties, specimens (five specimens each for pristine and all other variations of
the nanofillers must be effectively dispersed in the epoxy resin. The MWCNTs). The average load is obtained by summing the respective load
MWCNTs of different weight proportions are dispersed in the epoxy of the specimens and dividing by the no of samples tested for every wt.%
resin using the ultra-sonication process used by most researchers of MWCNTs. The failure response is brittle for all the specimens; the load
[31–33]. The detailed dispersion procedures of MWCNTs in the epoxy linearly increased with the displacement, followed by a drastic drop in
resin are described and schematically shown in Fig. 2. the load-bearing capacity due to the unstable crack propagation. From
the observations, as shown in Fig. 5(b), it is found that the maximum
(a) The MWCNTs are first manually blended with the resin and failure load is obtained for 0.2 wt% of MWCNTs. The load-bearing ca­
stirred for 5 min. The amount of MWCNTs used is fixed by their pacity for 0.1 wt% and 0.2 wt% MWCNTs are 8.1 % and 21.2 % higher,
respectively, compared to pristine bonded joints. The failure load for
Table 1 0.3 wt% of MWCNTs is relatively less as compared to 0.1 wt% and 0.2 wt
Material properties of specimen and fixture. % of MWCNTs variation (approximately 1.5 % less compared with the
pristine samples). This is possibly due to the agglomeration of MWCNTs
Material Young’s modulus (E) in GPa Poisson’s ratio (ν)
that reduces the bonding ability of the adhesives with the aluminium
Aluminium 70 0.33 substrate [19,37]. The fracture surfaces of the tested specimens are
Steel 210 0.30
shown in Fig. 6. Visual observation revealed that the fracture occurred

3
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

Fig. 2. Dispersion of MWCNTs in Resin (a) manual mixing (b) ultra-sonication (c) Magnetic stirring.

Fig. 3. Fabricated specimens for fracture tests (a) Pristine (b) 0.1%, (c) 0.2% and 0.3% of MWCNTs.

between the aluminium substrate and the adhesive, and the failure mode where τs (ζ) is the distribution of shear stress at any location ζ in the
is primarily interfacial. cohesive zone along the crack axis. The crack tip stress singularity can be
eliminated when
3. Theoretical formulation
Knet = Kapplied + Kcohesive = 0 (3)
This section illustrates a theoretical formulation to estimate the CZM From Eq. (1) and Eq. (2), we can write
parameters towards simulating the experimentally observed load vs √̅̅̅ ∫ ρ
displacement behaviour for different wt. % of MWCNTs for any gener­ √̅̅̅̅̅
τ∞ π a =
2 1
τs (ζ)ζ− 2 dζ (4)
alized extrinsic cohesive law. The primary objective of this section is to π 0

obtain the characteristic length associated with the cohesive zone to For the constant traction separation law, the cohesive traction is
nullify the stress singularity. constant, i.e., τs (ζ) = τs for the entire cohesive zone length ρ. So, Eq. (4)
can be rewritten as
3.1. Crack tip stress singularity and the cohesive zone √̅̅̅
√̅̅̅̅̅ 2 √̅̅̅
τ∞ π a = 2 τs ρ (5)
Consider a semi-infinite elastic body of crack length 2a at the center π
subjected under remote shear traction τ∞ on the boundary surfaces, as However, for other extrinsic cohesive laws, i.e., linear/parabolic or
shown in Fig. 7. A cohesive zone of length ρ = b − a is developed ahead higher-order softening, the variation of cohesive traction along the
of the crack tip due to the external loading. The local coordinate system length of the cohesive zone is unknown.
x and y – axis is oriented along and perpendicular to the crack front, A semi-analytical approach involving both the FEM and theoretical
respectively. The SIF generated due to applied loading is given as expression for separation has been used to calculate the SIFs. Parallelly,
√̅̅̅̅̅ domain-independent interaction integral formulation is adopted to
Kapplied = τ∞ πa (1)
numerically obtain the SIF in the presence of a cohesive zone. The SIF
Similarly, the SIF generated due to the existence of a cohesive zone variation is recorded for eight different characteristic lengths. From this
[38] is data, one can infer the value of ρ for which the SIF tends to be zero.
√̅̅̅ ∫ ρ
2
(2)
1
Kcohesive = − τs (ζ)ζ − 2 dζ
π 0

4
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

Fig. 4. Schematic dimensions of the (a) Fixture and (b) Specimen. (c) Experimental setup.

Fig. 5. Load displacement response of (a) Pristine and all weight proportions of MWCNTs (b) Peak load with their standard deviations.

3.2. Interface cohesive law where τc is the peak cohesive traction. δs and δfs are the instantaneous
and final separation in the cohesive zone. The cohesive stress is initially
It is already mentioned that for removing the strength of the crack tip τc ; it decreases gradually and approaches zero once the cohesive sepa­
stress singularity, a cohesive zone should have non-zero traction at its ration δs reaches the critical cohesive separation δfs . The exponent N is
zero separation displacement. To satisfy the requirements, a generalized used to define the shape of the cohesive law. τs represents the instan­
extrinsic cohesive law is used to define the behaviour of the interface. taneous cohesive traction.
The constitutive response of the cohesive law can be defined as Three distinct values of N = 0, 1 and 0.5 are considered, corre­
⎧ ( )N sponding to constant traction, linear, and parabolic softening law,
⎨ τc 1 − δs , δs < δf
⎪
respectively. The variation of the TSLs for different values of N is shown
(6)
s
τs (δs ) = δfs
⎪
⎩ in Fig. 8.
f
0, δs ≥ δs

5
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

Fig. 6. Fracture surfaces illustrating the interface fracture of (a) Pristine, (b) 0.1 wt%, (c) 0.2 wt%, (d) 0.3 wt% of MWCNTs reinforced adhesively bonded joints.

Fig. 7. Center cracked plate subjected to remote shear loading.

Fig. 9. Crack tip with a cohesive zone (the coordinate system x and y are
oriented parallel and perpendicular to the crack axis, respectively).
Fig. 8. A generalized extrinsic cohesive law.
conditions around the cohesive crack are defined as
3.3. Cohesive zone crack tip displacement and stress field ⃒ ⃒ ⃒ ⃒ ⃒
σ xy ⃒θ=π = σ xy ⃒θ=− π ∕
= 0; σyy ⃒θ=π = σyy ⃒θ=− π = 0; σyy ⃒θ=π = σ xx |θ=0
⃒
The cohesive zone with the crack tip local coordinate system is = 0 ⃒and u| = 0
θ=0 (7)
shown in Fig. 9. Here, the physical and mathematical tip is defined at the Substituting the above boundary conditions in Muskhelishvili’s [39]
end of the cohesive zone with maximum and zero openings, respec­ stress and displacement formulations, one can obtain asymptotic
tively. For a pure mode II crack problem, the stress field boundary displacement and stress field expression with a CZM. Muskhelishvili’s

6
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

asymptotic stress and displacement field are mentioned in the appendix [ ( ) ( ) ( )]

K θ θ 3θ
section. The mode II asymptotic displacement field in the presence of II
σ yy = √̅̅̅̅̅̅̅ sin cos cos
CZM can be obtained as 2π r 2 2 2
∑ 2n+3 ( )[ ( ) ( )]
δs = ds r 2 (8) K II
σ xy = √̅̅̅̅̅̅̅
θ
cos 1 − sin
θ
sin
3θ
(11)
n=0 2π r 2 2 2
where, The singularity-dominated region shown in Fig. 7 is modelled in the
k+1
(
2n + 3
) FE package ABAQUS® 2017. The material behaviour and failure criteria
ds = b2n sin π (9) of the cohesive are defined using the user subroutine UMAT. The global
μ 2
coordinate system is located at the center of the geometry, i.e., at the
μ and k are the shear modulus and Kolosov’s constant, respectively. mathematical tip. A magnified view of representative FE mesh at the tip
For a problem of planar elasticity k = 3− ν
1+ν
and k = 3 − 4ν for plane stress of the cohesive zone is shown in Fig. 10. The cracked geometry is
and strain condition, respectively. b2n represents the value of the real meshed using four-node quadrilateral elements with two-dimensional
coefficients. s and n represent the integer values with s = n + 1, where cohesive elements placed (COH2D) along the interface. Two different
n = 0, 1, 2, 3⋯ and so on. The unknowns ds are determined directly from mesh geometry are considered: one square domain covering the tip of
the FE solutions. After evaluating the displacement field, the stress field the cohesive zone with a circular domain surrounding it. The radius of
can be estimated for a particular cohesive law mentioned in the previous the circular domain is r = 100 mm with an initial crack length of a =
section. r − ρ. The bulk material is aluminium with the material properties:
Young’s modulus (E) = 70 GPa and Poisson’s ratio (ν) = 0.33. A KII
√̅̅̅̅̅̅̅̅
4. Numerical methodology to estimate the CZM parameters value of 100 MPa mm is considered for the subsequent analysis. The
thickness of the cohesive is kept as t = 0.1 mm.
This section describes the numerical procedure to estimate the
characteristic zone length ρ considering the singularity-dominated re­ 4.2. Domain-independent interaction integral
gion (SDZ) as depicted in Fig. 7. The obtained results are compared with
the analytical and semi-analytical results for constant traction law and A domain-independent interaction integral formulation is used to
linear-softening law, respectively. The procedure can be easily extended estimate the SIF. The interaction integral formulation assumes zero body
to any higher-order laws as described in Eq. (6). force with linear elastic behaviour and is applied on a square contour
(consisting of 8 finite parent elements) surrounding the tip of the
cohesive zone shown in Fig. 11. In the presence of crack face traction
4.1. Model geometry and the boundary conditions terms; the interaction integral formulation is expressed as
∫[
Since the zone close to the crack is K – dominated, a circular region, 1(
IS = σij uaux aux
j,1 + σ ij uj,1 − σ jk εaux
as shown in Fig. 7, of a finite radius with the mathematical tips as the
jk
2
A
center, is chosen to determine extrinsic cohesive zone parameters. The ) ] ∫
traction field developed at the periphery of the K – dominated region + σaux
jk εjk δ1,i q,i dA − (tj uaux
j,1 )qdC (12)
under pure mode II loading conditions can be defined as C+ +C−

[ ] [ ][ ] where uj,1 , σij and uaux

j,1 , σ ij defines the displacement and stress vector
aux
tx σ σ xy nx
= xx (10) due to the actual and auxiliary fields, respectively. tj represents the
ty σ xy σ yy ny
cohesive traction acting on the crack surfaces C+ and C− , as shown in
where nx and ny are the direction cosines of the unit normal vector to Fig. 12. q is a smoothly increasing function with values of zero in C1 and
the boundary with nx = cosθ and ny = sinθ, respectively. σ xx , σ yy , and σxy unity in Γ. The auxiliary and actual stress and displacement fields are
are the near-tip stress field under pure mode II, which is given as given in the appendix section.
[ ( ) ( )] The assumed material properties for the cohesive are:
KII θ 3θ
σ xx = − √̅̅̅̅̅̅̅ sinθ 2 + cos cos
2π r 2 2 KII2
τc = 50 MPa; GIIC = = 0.1428MPa − mm. (13)
E

Fig. 10. FE mesh geometry of the singularity-dominated region.

7
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

magnitude and the variation of cohesive stress along the length of the
CZM is the only deciding factor that affects the net SIF (Knet ).

4.3. Comparison between FEA results with the semi-analytical studies

For a constant traction separation law, the SIF developed due to the
presence of a cohesive zone at the crack tip can be easily estimated
analytically from Eq. (5) as mentioned in the earlier section 3.1. The net
SIF obtained using the analytical expression and the interaction integral
is shown in Fig. 14. It is important to note that the numerical results
approach close to the analytical results with the reduction in thickness of
the cohesive zone. The effect of cohesive zone thickness is not consid­
ered in the present work. For any other generalized TSLs, straightfor­
ward analytical expressions for evaluating the SIF are unavailable due to
the complexity of the cohesive profile. Therefore, a semi-analytical
procedure is adopted to estimate the SIF around the tip of the cohe­
sive zone. A linear-softening cohesive law is used for this purpose.
Fig. 11. Elements surrounding the mathematical tip.
For a linear softening cohesive law, the displacement profile is
numerically obtained for a particular value of cohesive zone length ρ
with the specific cohesive properties and far-field loading mentioned in
Section 4.1 and 4.2. The numerical displacement profile can be fitted
with the analytical separation expression mentioned in Eq. (8) with
constant d1 = 2.802,d2 = -7.680 , d3 = 10.245, d4 = -6.257 , d5 = 1.425
for ρ = 1.5 mm as shown in Fig. 15, where d1 , d2 , d3 , d4 and d5 represents
the actual values of ds . After obtaining the displacement profile, one can
obtain the stress profile τs (δs ) for linear softening cohesive law using Eq.
(6) with exponent N = 1.
The comparison between the semi-analytical cohesive stress profile
with the numerical results from the FE analysis along the length of the
cohesive zone is shown in Fig. 16. The SIF can be evaluated with the
semi-analytical stress field using expression Eq. (2) and Eq. (4). Fig. 17
represents the SIF obtained for semi-analytical and interaction integral
with the changes in cohesive zone length for a linear-softening cohesive
Fig. 12. Crack path for the implementation of Interaction integral procedure. law. The numerical results agree with the semi-analytical formulation.

The length of the cohesive zone ρ is considered from 1 mm to 2.75 5. Results and discussions – extrinsic and intrinsic cohesive
mm with an increment of 0.25 mm. The cohesive energy and the zone model
cohesive strength are kept constant, mentioned in Eq. (13), while
varying ρ. The SIF is evaluated at the tip of the cohesive zone using In this section, both intrinsic and extrinsic CZM parameters are used
interaction integral for all three cohesive laws, i.e., constant traction, to simulate the load–displacement response of MWCNTs reinforced
linear and parabolic softening. It is observed from Fig. 13 that the adhesively bonded joints. A correlation between extrinsic and intrinsic
reduction in the SIF in constant traction cohesive law is more rapid parameters is also established.
compared to linear and parabolic softening. In the case of constant Depending upon the behaviour of the TSLs and implementation, the
traction cohesive law, the magnitude of the Kcohesive increases rapidly, as CZM can be broadly categorized into intrinsic and extrinsic cohesive
evident from Eq. (5), resulting in a rapid reduction in the net SIF. So, the zone models. Before the computational simulation, cohesive elements
are inserted into the potential failure path in intrinsic CZMs. However,

Fig. 13. Variation of SIFs with changes in cohesive zone length using the
Interaction integral. Fig. 14. Comparison of SIFs for a constant traction cohesive law.

8
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

the cohesive elements are incrementally added in the case of the

extrinsic CZMs. For both the CZMs, the cohesive element is deleted once
the separation of the cohesive δs approaches critical separation δcs under
pure mode II loading conditions. The detailed implementation proced­
ure of intrinsic and extrinsic CZM is outlined in the flowchart shown in
Fig. 18. Here a and a′ are the distance from the crack end to the physical
and mathematical crack tip, respectively. da represents the incremental
cohesive zone length.

5.1. Numerical simulation of extrinsic CZM for three-point bending test

The three-point bending test, shown in Fig. 19, is mainly used for
investigating crack propagation under the pure mode I condition. The
crack propagation methodology is similar to Moës and Belytschko’s [30]
analysis with a linear softening law for defining the cohesive behaviour.
However, to the author’s knowledge, not a single work on pure mode II
has been carried out within the framework of removing the crack tip
stress singularity with extrinsic CZM procedures. Therefore, the stated
Fig. 15. Displacement profile behind the mathematical crack tip (ρ = 1.5 mm).
mode I problem can be considered the benchmark for estimating crack
propagation under pure mode II conditions.
The length of the beam is l = 600 mm. The cross-section of the beam
is square with b = t = 150 mm; an initial crack length of a = 45 mm,
shown in Fig. 19, is considered for the analysis. t and b represented the
width and depth of the beam, respectively. The load is applied on the
mid-length of the beam on the top surface ahead of the crack plane. The
finite element mesh consists of quadrilateral elements with a mesh size
of 6 mm × 6 mm and two-dimensional interface cohesive elements with
a mesh size of 0.15 mm × 0.1 mm, as shown in Fig. 20 inserted ahead of
the initial crack length a. The plane stress condition is considered with
an incremental cohesive zone length da = 15 mm. The properties of the
bulk material are E = 36500 MPa, ν = 0.1 with material properties at the
cohesive interface σc = 3.19 MPa. E, ν, σc , and GC are Young’s modulus
and Poisson’s ratio, mode I cohesive strength, and fracture energy,
respectively.
The dimensionless load–displacement behaviour is plotted for two
different brittleness number values, SE = 0.00208 and SE = 0.000627,
respectively. The brittleness number SE can be defined as [29]
GC
SE = (14)
σc b
Fig. 16. Stress profile in the cohesive zone (ρ = 1.5 mm).
The peak values of the dimensionless load P
σ c b2 for the work of Car­
penteri and Colombo [28] and the present analysis are 0.1857 and
0.1832 for Fig. 21, whereas 0.1582 and 0.1571 for Fig. 22, respectively.
P represents the magnitude of the peak load under load displacement
response. It is observed that the failure behaviour using the present
crack propagation methodology is in close agreement with the results of
Carpenteri and Colombo [28]. The same approach is extended to
analyzing mode II failure for the adhesively bonded interface.

5.2. Parametric studies for the extrinsic cohesive laws

In this section, the failure response of a structure is evaluated for

three different extrinsic cohesive laws under the pure mode II loading
condition. The dimensions and material properties of the specimen and
fixture are the same as that mentioned in the experimental section. The
extrinsic crack propagation methodology is already mentioned in
Fig. 18. The depth term (d) is considered as the width of the CTS ge­
ometry (w) in the direction of the applied load. The same terminology is
followed in the subsequent sections.
Two reference points (RP-1 and RP-2) are provided in the fixture to
Fig. 17. Comparison of SIFs for a linear softening cohesive law.
apply the displacement-based boundary conditions, as shown in Fig. 23.
To eliminate the rigid body motion, RP-2 is allowed to move only in the
x-direction. The displacement of the other reference point RP-1 is con­
strained in both x and y-directions. The thickness of the cohesive is set at
0.1 mm with cohesive strength as τc = 200 MPa .

9
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

Fig. 18. Implementation procedures for intrinsic and extrinsic CZM.

One square region is considered in the specimen geometry for eval­ load–displacement responses for all three cohesive laws (constant,
uating the SIF at the tip of the cohesive zone. The specimen and fixtures linear, and parabolic) are shown in Fig. 24. It is noticed that depending
are modelled with 4-noded bilinear quadrilateral elements with 3640 upon the shape of the cohesive laws, the peak load changes and the
and 1723 elements in the specimen and fixture, respectively. To start the failure behaviour of material changes from ductile to brittle. Also, the
simulation with extrinsic CZM methodology, da is set at 5 mm. The failure response with parabolic softening cohesive law is more ductile
cohesive elements with an element size of 0.1 mm thickness are placed with reasonable strain hardening. For linear softening cohesive behav­
ahead of the crack tip. The out-of-plane thickness is considered 1 mm, iour, a softening load–displacement response is observed once it reaches
and the plane stress condition is assumed throughout the analysis. the peak load. However, the failure response with the constant traction
Firstly, a brittleness number of SE = 0.00175 is considered, corre­ cohesive law is unstable.
sponding to fracture energy GIIC = 35 MPa-mm from Eq. (14). The It is evident that for the same values for fracture energy and

10
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

Fig. 19. Schematics of three-point bending (TPB) test.

Fig. 20. Mesh geometry of TPB specimen with a magnified view around the crack tip.

maximum cohesive stress, the parabolic softening cohesive law will have
the highest values of critical separation δfs compared to the other two
cohesive laws. It prolonged the crack propagation leading to ductile
failure behaviour. So, a parabolic softening extrinsic TSL will result in
ductile types of fracture compared to constant and linear softening TSLs
for the same cohesive strength and fracture energy. The peak load and
the corresponding displacement for constant traction cohesive law are
the highest among the three laws. Due to the stress profile of the con­
stant TSLs, the specimen must be subjected to higher values of applied
displacement to remove crack tip stress singularity.
The size effect of the specimen is further investigated by three
different values of brittleness number shown in Fig. 25. The failure
response changes from ductile to brittle with the reduction in brittleness
number from 0.0020 to 0.0015 with linear softening behaviour for the
cohesive interfaces. A perfect brittle specimen with a sudden drop in the
load after the peak value can be modelled using the information of the
brittleness number. In this case, it lies between 0.0015 and 0.00175 for a
linear TSL.
Fig. 21. Dimensionless load–displacement behaviour for SE = 0.00208.
5.3. Intrinsic cohesive zone parameters

One of the most popular ways to estimate the CZM parameters

(cohesive strength, stiffness, and fracture energy) is to compare the re­
sults obtained from the experiment with the numerical simulation.
Cohesive strength (τc ) and fracture energy (GC ) are directly obtained
from the peak load (Fmax ), and failure displacement is obtained for each
MWCNTs variations. It is found that any increase in the cohesive stiff­
ness (Kc ) increases the slope of the load vs displacement response in the
linear elastic regime. Accordingly, based on iterations, the Kc is chosen
in such a way that the elastic responses obtained from the numerical
study correlated well with the experimental observations.
The experiments described in Section 2 showed that the specimen is
subjected to pure shear loading, leading to uniform and maximum shear
stress distribution across the interface. Therefore the interface shear
stress obtained from the experiment can be assumed as a cohesive
strength τc in the numerical simulation. For a 50 mm interface length
with 10 mm thickness, the cross-section of the adhesive bond across the
Fig. 22. Dimensionless load–displacement behaviour for SE = 0.000627. interface is 500 mm2 . The maximum load obtained for each proportion
of MWCNTs divided by the interface bond area will give the values of the

11
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

Fig. 23. FE mesh of the specimen and fixture.

cohesive strength τc . Numerically, the cohesive strength:

Fmax
τc = (15)
A
where A is the sheared area, and Fmax is the maximum load under
load–displacement behaviour obtained from the experiment shown in
Fig. 5(a).
To investigate the influence of cohesive stiffness, the CTS specimen is
modelled under pure mode II with a triangular TSL. The cohesive stiff­
ness is varied from 10 to 30 MPa/mm, and the global load–displacement
behaviour is shown in Fig. 26. As the failure response of the adhesive is
perfectly brittle, it can be considered that the adhesive undergoes rapid
damage evolution. Therefore, the critical separation corresponding to
the damage initiation δcs is equal to the displacement at the failure δfs of
the cohesive, as shown in Fig. 27. In that scenario, the mode II fracture
energy is estimated as:

Fig. 24. Load-displacement response for constant, linear, and parabolic soft­ 1
GIIC = τc δfs (16)
ening law. 2
The intrinsic cohesive zone parameters for all proportions of
MWCNTs are provided in Table 2.

5.4. Numerical results

The crack propagation methodology for extrinsic cohesive zone

methodology is already described in Fig. 18. The governing parameters

Fig. 25. Load-displacement response for linear-softening cohesive law to

examine the size effect.

Fig. 26. Effect of cohesive stiffness on the load–displacement response (τc =

20 MPa , GIIC = 20 MPa-mm).

12
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

for 0.2 wt % of MWCNTs is the highest. However, as seen in Eq. (16), the
fracture energy (GIIC ) of the intrinsic CZM depends on the displacement
at failure in addition to the cohesive strength (τc ). Therefore, among all
the MWCNTs variations, 0.1 wt% MWCNTs sustain maximum failure
separation and, therefore, exhibit maximum fracture energy (GIIC ). On
the other hand, the load vs displacement response of the extrinsic CZMs
is dependent upon the ratio of fracture energy to the product of the
width of the specimen and the cohesive strength for any specific TSL.
The ratio is almost constant at approximately 0.0017 for pristine and all
other variations of MWCNTs with the assumed linear-softening extrinsic
TSLs. This supports the conclusions made at the end of Section 5.2. It
implies that for a set of bulk material properties, geometric dimensions,
and type of test, the brittleness number remains constant for simulating
a particular manner of load vs displacement response, e.g., a perfectly
Fig. 27. Triangular CZM. brittle response. The cohesive fracture energy and strength can change
with the type of adhesives, but they get tuned to a specific brittleness
for the extrinsic cohesive zone modelling are only the cohesive strength number to obtain a particular load vs displacement response.
and fracture energy. A linear softening cohesive law is considered It is observed that there is a good correlation between the cohesive
throughout the analysis. The cohesive stress is estimated for all pro­ strength (τc ) of the intrinsic CZM with the τc .t (cohesive strength times
portions of MWCNTs to correlate well with the elastic failure response the thickness) of the extrinsic CZM. A higher displacement at failure
obtained from the experiment. The removal of SIF at the tip of the observed during the experiment requires an increase in extrinsic cohe­
cohesive zone was also sequentially carried out. sive strength (τc ) to nullify the SIF. Consequently, the displacement at
It is noticed that for higher values of brittleness number, the global failure in the experiments shown in Fig. 28 and the extrinsic cohesive
failure behaviour of the specimen is mainly influenced by the continuum strength (shown in Table 2) follow a similar trend. In summary, the
element placed ahead of the cohesive zone. The cohesive zone length intrinsic CZM strength solely depends on the experimental peak load.
incrementally varied from da = 5 mm to 45 mm, and their However, the extrinsic cohesive strength relies upon the displacement at
load–displacement response is shown in Fig. 28. The estimated extrinsic the failure of the load–displacement response.
CZM properties are provided in Table 2.
Intrinsic CZM is an easy and straightforward implementation. 6. Conclusions
Additionally, depending upon the fracture energy values, one can obtain
a wide variety of load–displacement behaviour for pure mode II loading, In this work, the load vs displacement response of adhesively bonded
ranging from ductile to brittle. The problems of removing the crack tip joints using extrinsic and intrinsic CZM is simulated under pure mode II
stress singularity still need to be addressed, even though implementing loading for various wt.% of MWCNTs nanofillers. The stress and
intrinsic CZM is computationally more convenient than the extrinsic displacement fields in the cohesive zone are obtained semi-analytically
CZM. The extrinsic CZM parameters – cohesive strength and fracture for linear extrinsic TSL that are used to determine the extrinsic cohesive
energy reported in Table 2- differ considerably from the intrinsic parameters that nullify the stress singularity. This approach could be
parameters. extended to any generalized extrinsic cohesive law. A domain-
Ideally, the presented extrinsic crack propagation methodology is independent interaction integral procedure incorporating crack face
initially developed for continuous specimens with no bonded interface. traction has been used to compute the mode II SIF for three extrinsic
As the interfaces are significantly weaker in adhesively bonded joints, cohesive laws, i.e., constant, linear, and parabolic softening. The SIF
the out-of-plane thickness (t) is set to match the stiffness of the experi­ obtained using semi-analytical formulation agrees well with those ob­
mental load vs displacement response. Accordingly, the t for the pristine, tained using the interaction integral. Finally, the crack propagation
0.1 %, 0.2 %, and 0.3 wt% of MWCNTs are set at 1.36, 1.08, 1.42, and along the bonded interface is simulated using the SIF nullification
1.03, respectively. Any other combination will not give the expected criteria using the extrinsic CZM to obtain the load vs displacement
response. The comparison between experimental and numerical (both response. Whereas, in the case of intrinsic CZM, the crack propagates
intrinsic and extrinsic CZM) results is shown in Fig. 28(a)-28(d). The automatically when the cohesive energy reaches the fracture energy.
numerical results agreed well with the experimental observation. On the The modified compact tension specimens (CTS) are fabricated with
other hand, for the intrinsic CZM, the out-of-plane thickness t = 10, MWCNTs reinforced epoxy-based adhesive. The strength of the joints is
which is the same as that of the specimen thickness used in the experi­ maximum for 0.2 wt% of MWCNTs. Any further increase in MWCNTs
mental Section. concentration leads to deteriorated strength. Some specific conclusions
The intrinsic CZM parameters, i.e., cohesive strength and fracture from the current work are:
energy, are precisely calculated using Eq. (15), and Eq. (16), as
mentioned in Section 5.3. As the cohesive strength depends entirely on • For the same value of the brittleness number, the SIF decreases
the peak load of the load–displacement response, the cohesive strength quickly for constant TSL with cohesive zone length, followed by

Table 2
Estimated CZM parameters for MWCNTs reinforced bonded joints.
wt. % of Intrinsic CZM parameters Extrinsic CZM parameters
MWCNTs Stiffness (Kc in MPa/ Strength (τc in Fracture energy (GIIC in MPa- Strength (τc in Strength × Fracture energy (GIIC in MPa-
mm) MPa) mm) MPa) thickness mm)
(τc .t in MPa-
mm)

Pristine 15.2 14.40 6.82 158 214.88 26.70

0.1 % MWCNTs 12.2 15.62 10 215 232.20 37
0.2 % MWCNTs 18 17.41 8.41 170 241.40 28.80
0.3 % MWCNTs 11.5 14.31 8.90 206 212.18 34.80

13
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

Fig. 28. Numerical validation of load–displacement response for (a) Pristine (b) 0.1 wt.% (c) 0.2 wt.% (c) 0.3 wt.% of MWCNTs.

linear and then parabolic TSL. So, lower-order TSL is recommended analysis, Validation, Visualization, Investigation, Writing – original
to nullify the SIF if a small cohesive zone length is considered. draft, Writing – review & editing. N. Muthu: Conceptualization, Su­
• The experimental peak load is the only factor influencing the pervision, Writing – review & editing, Project administration, Funding
intrinsic CZM strength. However, the displacement at the failure of acquisition.
the load–displacement response is used to find the extrinsic cohesive
strength. The extrinsic CZM, though computationally cumbersome, Declaration of Competing Interest
gives good predictions based on the SIF nullification criteria, which
is more reasonable. On the other hand, the intrinsic CZM laws are The authors declare that they have no known competing financial
popular due to the ease of modelling but do not truly capture the interests or personal relationships that could have appeared to influence
essence of the cohesive zone, i.e., null singularity. the work reported in this paper.
• Various load–displacement responses are presented for linear soft­
ening cohesive law for different brittleness number values. It is Data availability
observed that the failure behaviour of the adhesively bonded struc­
ture does not solely depend on the values of the brittleness number Data will be made available on request.
but is also affected by the order of extrinsic cohesive laws used.
Perhaps a parameter that considers cohesive strength, cohesive en­ Acknowledgement
ergy, and order of the cohesive law could be used as a better measure
for predicting the same load vs displacement response. The work was supported by the Science and Engineering Research
Board, India under the scheme ‘Early Career Research Award’, No: ECR/
CRediT authorship contribution statement 2018/001638.

P.J. Saikia: Conceptualization, Methodology, Software, Formal

14
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

Appendix I:

For a general planar mixed mode (mode I + mode II) problem, the stresses, and displacements are expressed in terms of complex functions ϕ(z) and
χ(z).
∑ ∑
ϕ(z) = An zλn = An rλn eiλn θ
n=0 n=0

∑ ∑
χ(z) = Bn zλn +1 = Bn rλn +1 ei(λn +1)θ (A1.1)
n=0 n=0

where An and Bn are the complex coefficients with An = a1n +ia2n and Bn = b1n + ib2n . a1n , a2n , b1n and b2n are real coefficients. λn is the Eigenvalue
expressed in terms of integer n. where n = 0, 1, 2⋯.
The asymptotic stress and the displacement field in the presence of CZM can be defined as:
∑
σ xx = rλn − 1 {2λn [a1n cos(λn − 1)θ − a2n sin(λn − 1)θ ] − λn (λn − 1)[a1n cos(λn − 3)θ − a2n sin(λn − 3)θ ] − λn (λn + 1)[b1n cos(λn − 1)θ − b2n sin(λn − 1)θ]}
n=0

(A1.2)
∑
σ yy = rλn − 1 {2λn [a1n cos(λn − 1)θ − a2n sin(λn − 1)θ ] + λn (λn − 1)[a1n cos(λn − 3)θ − a2n sin(λn − 3)θ ] + λn (λn + 1)[b1n cos(λn − 1)θ − b2n sin(λn − 1)θ]}
n=0

(A1.3)
∑
τxy = rλn − 1 {λn (λn − 1)[a1n sin(λn − 3)θ + a2n cos(λn − 3)θ ] + λn (λn + 1)[b1n sin(λn − 1)θ + + b2n cos(λn − 1)θ ]} (A1.4)
n=0

∑rλn
u= {k(a1n cosλn θ − a2n sinλn θ) + λn [ − a1n cos(λn − 2)θ + a2n sin(λn − 2)θ ] + (λn + 1) (− b1n cosλn θ + b2n sinλn θ) } (A1.5)
n=0
2μ

∑rλn [ ]
v= {k(a1n sinλn θ + a2n cosλn θ) + λn a1n sin(λn − 2)θ + a2n cos(λn − 2)θ + (λn + 1) (b1n sinλn θ + b2n cosλn θ) } (A1.6)
n=0
2μ

Appendix II:

The auxiliary displacement and stress field for Interaction integral are as follows
1 [ aux I ]
σ aux aux II
ij = √̅̅̅̅̅̅̅ KI gij (θ) + KII gij (θ) (A2.1)
2π r
√̅̅̅̅̅
r [ aux I ]
uaux
i = K f (θ) + KIIaux fiII (θ) (A2.2)
2π I i

k − cosθ θ 2 + k + cosθ θ
f1I (θ) = cos , f1II (θ) = sin (A2.3)
2μ 2 2μ 2

k − cosθ θ 2 − k − cosθ θ
f2I (θ) = sin , f2II (θ) = cos (A2.4)
2μ 2 2μ 2

3 θ 1 5θ 7 θ 1 5θ
gI11 (θ) = cos + cos , gII11 (θ) = − sin − sin , (A2.5)
4 2 4 2 4 2 4 2
5 θ 1 5θ 1 θ 1 5θ
gI22 (θ) = cos − cos , gII22 (θ) = − sin + sin , (A2.6)
4 2 4 2 4 2 4 2

gI12 (θ) = gI21 (θ) = gII22 (θ), gII12 (θ) = gII21 (θ) = gI11 (θ) (A2.7)

References [4] G. Barenblatt, The formation of equilibrium cracks during brittle fracture. General
ideas and hypotheses. Axially-symmetric cracks, J. Appl. Math. Mech. 23 (1959)
622–636, https://doi.org/10.1016/0021-8928(59)90157-1.
[1] F. Cavezza, M. Boehm, H. Terryn, T. Hauffman, A review on adhesively bonded
[5] D.S. Dugdale, Yielding of steel sheets containing slits, J. Mech. Phys. Solids. 8
aluminium joints in the automotive industry, Metals (Basel) 10 (2020) 730,
(1960) 100–104.
https://doi.org/10.3390/met10060730.
[6] A. Needleman, Continuum model for void nucleation by inclusion debonding, Am.
[2] M. Linke, R. Lammering, On the calibration of the cohesive strength for cohesive
Soc. Mech. Eng. 54 (1987) 525–531.
zone models in finite element analyses, Theor. Appl. Fract. Mech. 124 (2023),
[7] A. Needleman, An analysis of tensile decohesion along an interface, J. Mech. Phys.
103733, https://doi.org/10.1016/j.tafmec.2022.103733.
Solids 38 (1990) 289–324, https://doi.org/10.1016/0022-5096(90)90001-K.
[3] H. Khoramishad, R.S. Ashofteh, M. Mobasheri, F. Berto, Temperature dependence
[8] M. Stern, E.B. Becker, R.S. Dunham, A contour integral computation of mixed-
of the shear strength in adhesively bonded joints reinforced with multi-walled
mode stress intensity factors, Int. J. Fract. 12 (1976) 359–368, https://doi.org/
carbon nanotubes, Eng. Fract. Mech. 199 (2018) 179–187, https://doi.org/
10.1007/BF00032831.
10.1016/j.engfracmech.2018.05.032.

15
P.J. Saikia and N. Muthu Theoretical and Applied Fracture Mechanics 126 (2023) 103967

[9] C.F. Shih, B. Moran, T. Nakamura, Energy release rate along a three-dimensional [24] S.S. Kulkarni, V. Gupta, A. Ortiz, H. Das, P. Upadhyay, E. Barker, D. Herling,
crack front in a thermally stressed body, Int. J. Fract. 30 (1986) 79–102, https:// Determining cohesive parameters for modelling interfacial fracture in dissimilar-
doi.org/10.1007/BF00034019. metal friction stir welded joints, Int. J. Solids Struct. 216 (2021) 200–210, https://
[10] F.Z. Li, C.F. Shih, A. Needleman, A comparison of methods for calculating energy doi.org/10.1016/j.ijsolstr.2021.01.023.
release rates, Eng. Fract. Mech. 21 (1985) 405–421, https://doi.org/10.1016/ [25] H.L.J. Pang, C.W. Seetoh, A Compact Mixed Mode (CMM) fracture specimen for
0013-7944(85)90029-3. adhesive bonded joints, Eng. Fract. Mech. 57 (1997) 57–65, https://doi.org/
[11] H. Deng, B. Yan, Y. Zhu, A new path-independent interaction integral for the SIFs 10.1016/S0013-7944(97)00014-3.
of interfacial crack, Theor. Appl. Fract. Mech. 120 (2022), 103389, https://doi. [26] N. Choupani, Mixed-mode cohesive fracture of adhesive joints: experimental and
org/10.1016/j.tafmec.2022.103389. numerical studies, Eng. Fract. Mech. 75 (2008) 4363–4382, https://doi.org/
[12] H.H. Zhang, S.M. Liu, S.Y. Han, L.F. Fan, T-stress evaluation for multiple cracks in 10.1016/j.engfracmech.2008.04.023.
FGMs by the numerical manifold method and the interaction integral, Theor. Appl. [27] P.M. Noury, R.A. Shenoi, I. Sinclair, On mixed-mode fracture of PVC foam, Int. J.
Fract. Mech. 105 (2020), 102436, https://doi.org/10.1016/j.tafmec.2019.102436. Fract. 92 (1998) 131–151, https://doi.org/10.1023/A:1007495223740.
[13] Z.H. Jin, C.T. Sun, Cohesive fracture model based on necking, Int. J. Fract. 134 [28] A. Carpinteri, G. Colombo, Numerical analysis of catastrophic softening behaviour
(2005) 91–108, https://doi.org/10.1007/s10704-005-7864-1. (snap-back instability), Comput. Struct. 31 (1989) 607–636, https://doi.org/
[14] Q. Xiao, B.L. Karihaloo, Asymptotic fields at frictionless and frictional cohesive 10.1016/0045-7949(89)90337-4.
crack tips in quasibrittle materials, J. Mech. Mater. Struct. 1 (2006) 881–910. [29] A. Carpinteri, Notch sensitivity in fracture testing of aggregative materials, Eng.
[15] P.J. Saikia, N. Muthu, Extrinsic cohesive zone modelling for interface crack growth: Fract. Mech. 16 (1982) 467–481, https://doi.org/10.1016/0013-7944(82)90127-
numerical and experimental studies, Eng. Fract. Mech. 266 (2022), 108353, 8.
https://doi.org/10.1016/j.engfracmech.2022.108353. [30] N. Moës, T. Belytschko, Extended finite element method for cohesive crack growth,
[16] A. Esmaeili, C. Sbarufatti, R. Casati, A. Jiménez-Suárez, A. Ureña, A.M. Eng. Fract. Mech. 69 (2002) 813–833, https://doi.org/10.1016/S0013-7944(01)
S. Hamouda, Effective addition of nanoclay in enhancement of mechanical and 00128-X.
electromechanical properties of SWCNT reinforced epoxy: strain sensing and crack- [31] H. Miyagawa, A.K. Mohanty, L.T. Drzal, M. Misra, Nanocomposites from biobased
induced piezoresistivity, Theor. Appl. Fract. Mech. 110 (2020), 102831, https:// epoxy and single-wall carbon nanotubes: synthesis, and mechanical and
doi.org/10.1016/j.tafmec.2020.102831. thermophysical properties evaluation, Nanotechnology 16 (2004) 118.
[17] H. Ejaz, A. Mubashar, E. Uddin, Z. Ali, N. Arif, Influence of MWCNTs on strength [32] N. Yu, Z.H. Zhang, S.Y. He, Fracture toughness and fatigue life of MWCNT/epoxy
properties of high viscous epoxy adhesive and fracture behaviour of adhesively composites, Mater. Sci. Eng. A. 494 (2008) 380–384, https://doi.org/10.1016/j.
bonded joints, Theor. Appl. Fract. Mech. 120 (2022), 103412, https://doi.org/ msea.2008.04.051.
10.1016/j.tafmec.2022.103412. [33] X.-H. Zhang, Z.-H. Zhang, W.-J. Xu, F.-C. Chen, J.-R. Deng, X. Deng, Toughening of
[18] M.S. Konsta-Gdoutos, P.A. Danoglidis, S.P. Shah, High modulus concrete: effects of cycloaliphatic epoxy resin by multiwalled carbon nanotubes, J. Appl. Polym. Sci.
low carbon nanotube and nanofiber additions, Theor. Appl. Fract. Mech. 103 110 (2008) 1351–1357.
(2019), 102295, https://doi.org/10.1016/j.tafmec.2019.102295. [34] A. Alloys, C.A. Treatment, Standard guide for preparation of metal surfaces for
[19] F. Gojny, M. Wichmann, B. Fiedler, K. Schulte, Influence of different carbon adhesive bonding 1, Society 90 (2000) 1–6, https://doi.org/10.1520/D2651-
nanotubes on the mechanical properties of epoxy matrix composites – a 01R16.2.
comparative study, Compos. Sci. Technol. 65 (2005) 2300–2313, https://doi.org/ [35] M. Sander, H. Richard, Experimental and numerical investigations on the influence
10.1016/j.compscitech.2005.04.021. of the loading direction on the fatigue crack growth, Int. J. Fatigue. 28 (2006)
[20] K.-T. Hsiao, J. Alms, S.G. Advani, Use of epoxy/multiwalled carbon nanotubes as 583–591, https://doi.org/10.1016/j.ijfatigue.2005.05.012.
adhesives to join graphite fibre reinforced polymer composites, Nanotechnology 14 [36] J.J.M. Machado, E.A.S. Marques, R.D.S.G. Campilho, L.F.M. da Silva, Mode II
(2003) 791–793, https://doi.org/10.1088/0957-4484/14/7/316. fracture toughness of CFRP as a function of temperature and strain rate, Compos.
[21] M. Mansourian-Tabaei, S.H. Jafari, H.A. Khonakdar, Lap shear strength and Part B Eng. 114 (2017) 311–318, https://doi.org/10.1016/j.
thermal stability of diglycidyl ether of bisphenol a/epoxy novolac adhesives with compositesb.2017.02.013.
nanoreinforcing fillers, J. Appl. Polym. Sci. 131 (2014) n/a-n/a, https://doi.org/ [37] A.R. Alian, S. El-Borgi, S.A. Meguid, Multiscale modelling of the effect of waviness
10.1002/app.40017. and agglomeration of CNTs on the elastic properties of nanocomposites, Comput.
[22] X.T. Miao, Q. Yu, C.Y. Zhou, J. Li, Y.Z. Wang, X.H. He, Experimental and numerical Mater. Sci. 117 (2016) 195–204, https://doi.org/10.1016/j.
investigation on fracture behaviour of I-II mixed mode crack for commercially pure commatsci.2016.01.029.
Titanium, Theor. Appl. Fract. Mech. 96 (2018) 202–215, https://doi.org/10.1016/ [38] J.R. Rice, G.C. Sih, Plane problems of cracks in dissimilar media, J. Appl. Mech. 32
j.tafmec.2018.04.012. (1965) 418–423, https://doi.org/10.1115/1.3625816.
[23] P. Jin, Z. Liu, X. Wang, X. Chen, Three-dimensional analysis of mixed mode [39] N. Muskhelishvili, Some basic problems of the mathematical theory of elasticity, n.
compact-tension-shear (CTS) specimens: stress intensity factors, T-stresses and d.
crack initiation angles, Theor. Appl. Fract. Mech. 118 (2022), 103218, https://doi.
org/10.1016/j.tafmec.2021.103218.

16